ALAN S.

FOUST
LEONARD A. WENZEL
CURTIS W CLUMP
LOUIS MAUS
L. BRYCE ANDERSEN
Principles of
Unit Operations
Board of Advisors,
Engineering

A. H-S, Ang Civil Engineering — Systems

University of Illinois and Probability

Donald S. Berry Transportation Engineering

Northwestern University

James M. Gere Civil Engineering and

Stanford University Applied Mechanics

J. Stuart Hunter Engineering Statistics

Princeton University

T. William Lambe Civil Engineering — Soil

R. V. Whitman Mechanics
Massachusetts Institute of
Technology

Perry L. McCarty Environmental Engineering

Stanford University

DonT. Phillips Industrial Engineering

Texas A& M

Dale Rudd Chemical Engineering

University of Wisconsin

Robert F. Steidel, Jr, Mechanical Engineering

University of California
Berkeley

R. N. White Civil Engineering — Structures

Cornell University
PRINCIPLES
OF UNIT
OPERATIONS
Second Edition

Alan S. Foust
Professor Emeritus. Lehigh University

Leonard A. Wenzel
Lehigh University

Curtis W. Clump
Lehigh University

Louis Maus
Science Center,
Rockwell International

L. Bryce Andersen
New Jersey Institute of Technology

JOHN WILEY & SONS New York •Chichester* Brisbane •Toronto 'Singapore
Copyright © 1960. 1980, by John Wiley & Sons, Inc.

All rights reserved. Published simultaneously in Canada.

Reproduction or translation of any part of

thiswork beyond that permitted by Sections
107 and 108 of the 1976 United States Copyright
Act without the permission of the copyright
owner is unlawful. Requests for permission
of further information should be addressed to
the Permissions Department, John Wiley & Sons.

Library of Congress Cataloging in Publication Data:

Foust, Alan Shivers, 1908—

Principles of unit operations.

Includes bibliographical references and index.

1. Chemical processes. I. Title.

TP155.7.F67 1979 660.2'842 78-12449

ISBN 0-471-26897-6

Printed in the United States of America

10 9 8 7 6 5
Preface

After a lapse of years, even the most basic textbook in a dynamic field must be
either updated or allowed to fade away. This fact alone dictates an updating to allow
inclusion of new technology, new analyses, and new concepts. Added to the expected
advances of the technology of chemical engineering, the move toward metrication
imposed by the federal policy to change over to a decimal dimension system forces further
bilinguality upon chemical engineers. This will not be any imposition on chemical
engineers now, but industry will require some significant shifting of calculation habits and
values by the next generation. A mixture of SI (Systeme International) and English
dimensions appears probable for some years; hence, both systems are used in this book.
Two new chapters have been added, written in a fashion that will permit the
elimination of one or both without interrupting the continuity. Chapter 8 includes a
presentation of the basics of multicomponent, multistage separations formulated in a

fashion easily adaptable to computer evaluation, but written with every intention that the
computer aspect should not befog the basics of the separation process. No attempt is

made to incorporate sparse matrix techniques that permit considerable shortening of

computer time for such actual evaluations. Chapter 1 1 presents basic equations of change
for those who wish to cover that topic more thoroughly.
Covering the entire text will probably require more course credits than most
schools will allot to the instruction in unit operations. Therefore, as in the first edition,
the material is presented so that the instructor can omit chapters or even use only one or
two of the three major parts of the book.
We have consciously minimized the inclusion of data in the appendices if the
information is available in Chemical Engineers Handbook.
The subject matter is arranged as in the first edition to emphasize similarities and to
offer a possible partition among three-semester courses. These might represent fluid
mechanics, heat transfer, and mass transfer for which the following organization has been
found satisfactory:
1. Fluid mechanics: Chapters 1, 9, 10 (fluid mechanic portions), Chapters 11, 13,
14,20,21,22.
2. Heat transfer: Chapters 9, 10, 11, 12, 13, 14 (heat transfer portions). Chapter
15.

3. Mass transfer: Chapters 2, 3, 4, 5, 6, 7, 8; brief review of Chapters 9, 10, 1 1, 12,

1 3, 1 4; also Chapters 1 6, 1 7, 1 8, 1 9.

The book has been used quite successfully in student paced instruction courses.
We believe this edition, like the first one, will satisfy instructors who appreciate the
educational advantages offered by a treatment of the unit operations that emphasizes
their common fundamentals and their interrelationships. The pedagogical effectiveness of
this approach has now been demonstrated.

Alan S. Foust Louis Maus

Leonard a. Wenzel l Bryce Andersen
Curtis W. Clump
Preface
to First Edition

The treatment of unit operations given in this book emphasizes the scientific
principles upon which the operations are based, and groups those with similar physical
bases so that they may be considered together. The development usually begins with an
analysis of the physical behavior of a system and the establishment of a simplified
physical model. A basic mathematical relation is written using the model and is solved.
The resulting general expression is then applied to the specific unit operations. In order to
maintain a clarity of presentation at an elementary level, refinements of the physical
models and the resulting elaborate mathematics necessary for rigorous treatment of
complex situations are generally omitted, and, in order to emphasize underlying
similarities among the various unit operations, descriptions of equipment and specialized
calculation methods are condensed. However, since visualization of equipment helps to
add significance to the theoretical treatment and since the young engineer should be
familiar with major equipment types, the important pieces of processing equipment are
shown in line drawings and photographs and are discussed briefly. The more important of

the specialized calculation methods necessary for process design are considered after the
underlying principles have been fully developed.
The traditional concept of unit operations has been a major factor in the
phenomenal success of chemical engineers and chemical engineering in the last fifty years.

We believe that the unification presented here is the next logical step in the evolution of
the concept of unit operations. This treatment is it is more
offered in the belief that
more economical in time, more adequate in its presentation of the
efficient in teaching,

fundamentals, and more effective in training toward the definition and solution of broad
problems in chemical processing. This book should serve as a basis for advanced work in
the more specialized theory and practice of the individual unit operations. The engineer
educated in this approach may not be as immediately adept in the manipulation of a
given specialized procedure of calculation, but he should be firmer in his understanding
of the fundamental principles, more aware of the similarities among many of the unit
operations, and more flexible and original in his solution of new processing problems. In
short, he should be more readily adaptable to change and progress.
The continuing rapid extensions of knowledge of unit operations has created a
serious problem in the coverage of this information within the time properly allotted in

an undergraduate program. The increasing number of process steps that might be included
as unit operations and the breadth of their applicability dictate that instruction be
systematized and be made more adaptable to newer operations. This treatment, with its

unification of the principles of similar operations, makes it possible to maintain a realistic

balance between the unit operations and other vital facets of chemical engineering
education.

vu
VU1

Many of the formalized calculations that have occupied chemical engineers in the
past will very shortly be done by electronic computers. Multicomponent distillation
calculations and multiple-effect evaporator calculations already have been programmed
for machine computation. Work is going forward in the use of computers to design
chemical reactors, to determine the dynamics of a system during start-up, to predict and
optimize the response of systems to automatic control, and to help in many other
applications. The use
computer to do in minutes what previously took man-months
of a
of engineering time has opened up several avenues of development. One avenue is that
complex engineering problems, which previously have been solved only crudely and
perhaps only qualitatively, can now be answered with high accuracy. Many of these
applications have been in the field of process economics, as for example in optimizing the
products from a petroleum refinery. Another avenue may be that rigorous but
cumbersome calculation procedures may become preferable to approximate, short-cut
methods. This has already happened to some extent in the field of multicomponent
distillation. For work of this sort to be effective, the chemical engineer with thorough
understanding of the mechanism of the process must cooperate with the mathematician
who understands the mathematical possibilities and limitations of the computer. Thus,
the increasing need is for a chemical engineer with firm mastery of the fundamental
characteristics of process operations and in addition with the mathematical background
necessary to attack the problem of describing these operations by a mathematical model.
Traditional training in the details of a calculation method is thus important only as
mental discipline.
With the increasing understanding of the fundamental principles, it is possible to
classify the unit operations into groups based upon similar principles. In this book two
major groups are considered: the stage operations and the rate operations. The stage
operations are considered by using a generalized model which is applied to all of the
mass-transfer operations. The rate operations are introduced with a thorough coverage of
the principles of molecular and turbulent transport. After the fundamental similarities of
each group of operations have been considered, the principles are applied to the analysis
of the more common operations in each group.
A generalized treatment of the mass-transfer stage operations is presented in Part I.

A generalized method of calculation based upon the physical model of an equilibrium

stage is developed without regard to the nature of the particular phases in contact.
Specific examples are taken from the various mass-transfer operations. The stage

operations have been placed first because they are based on simple stoichiometric and
equilibrium concepts and thus follow logically from the stoichiometry course which
usually precedes the unit operations course. Full coverage of Part I requires at least two
semester hours.
The fundamental principles of the rate operations are developed in Part II. Included
are the operations in which a property of a phase diffuses or is transferred under the

influence of a potential gradient. Molecular and turbulent transport of heat, mass, and
momentum are considered in detail. Turbulence is explained at the junior-year level

without the rigorous and abstruse concepts of more complete advanced consideration.
a

Full coverage of all the material in Part II requires at least three semester hours. Our
experience is that coverage is possible in three semester hours only if sections of Chapter

13 are covered superficially and then reviewed when the corresponding operations are
studied in Part III.

In Part III the principles introduced in Parts I and II are applied to the calculations
involved in process design of equipment for the various operations. Our objective is to
make the transition from principle to practice without obscuring the principles with an
excess of practical details and special methods. Because Part III is dependent on Part II
and to a minor extent on Part I, we recommend that Part III be introduced only after a
thorough coverage of the other parts. All the material in Part III may be covered in four
semester hours. The order of presentation of the major topics in Part 1 1 1 is flexible, and it

may be rearranged according to the wishes of the instructor. For example, momentum
 L\

transfer (Chapters 20 through 22) may be covered before heat and mass transfer
(Chapters 15 through 19). With this flexibility the instructor is free to omit subjects at his
discretion.
Appendix A discusses dimensions, units, and dimensional analysis. These subjects
are included in an appendix to avoid disruption of the major development of the
principles. Knowledge of dimensions and units as discussed in Appendix A is necessary
for Part II.The concepts of dimensional analysis are fundamental to the principles
developed in Chapter 13. If these topics have not been studied earlier, they should be
introduced when needed. Supplementary material beyond that given in this appendix may
be introduced if the instructor desires.
Appendix B deals with the measurement and description of small particles. This
material is relevant to any operation involving the presence of a particulate solid phase
and especially to the operations discussed in Chapters 18, 19, and 22. Other aspects of
small particle technology are omitted because they are so frequently available in
specialized courses.
The entire book may be covered in nine or ten semester hours. Judicious selection
of material and abbreviated consideration of subjects considered by the instructor to be
of more limited utility makes an eight-hour course quite feasible.
Part II might serve as a nucleus for a basic three-hour course in transport operations
for all drawn from Part III. Such a basic course is in
engineers, with selected applications
agreement with the recommendations of the American Society for Engineering Education
for a more unified and fundamental coverage of heat, mass, and momentum transfer.
This book is the result of several years of teaching the unified approach. The
preliminary draft has been used for over two years as the textbook for junior students in

the chemical engineering curriculum at Lehigh University. The material has undergone
several revisions based upon the experiences gained from using the preliminary draft.
Every effort has been made to insure that the material included in this book can be
taught successfully to junior-year engineering students. Our experience has been that the
generalized approach is briefly more bewildering to the average student than is the
traditional approach; however, after a short period, the material becomes clear, and the
student ultimately gains a greater understanding of the unit operations.
The integration of unit operations with the important fields of kinetics,

thermodynamics, and economics can be accomplished in a subsequent design course. The

unit operations are among the most important tools of the chemical engineer, but they
must not be allowed to crowd out other important subjects in a chemical engineering
curriculum. Although human relations is seldom formally taught, it is no less important
to the chemical engineer than is his background in the physical sciences and economics.
Fortunate indeed is the young engineer whose education has been broad enough to
impress upon him the importance of all three aspects of the triad of engineering: physical
science, economics, and human relations.

We wish to express our appreciation to the administration of Lehigh University for

its cooperation in the testing of this book. We also wish to thank the several classes of
students on whom the developing versions of this book were tested. Their forbearance,
cooperation, and suggestions have been of great help in preparing the final version.

Alan S. Foust
Leonard A. Wenzel
Curtis W. Clump
Louis Maus
L. Bryce Andersen

Bethlehem, Pennsylvania
December 1959
Contents

1 Unit Operations In Chemical Engineering 1

Part 1 Stage Operations 1

2 Mass Transfer Operations 13

3 Phase Relations 29
4 Equilibrium Stage Calculations 48
5 Countercurrent Multistage Operations 60
6 Countercurrent Multistage Operations with Reflux 75
7 Simplified Calculation Methods 96
8 Multicomponent State Operations 122

Part 2 Molecular and Turbulent Transport 159

9 Molecular Transport Mechanism 161
10 Differential Mass, Heat, and Momentum Balances 189
11 Equations of Change 212
12 Turbulent-Transport Mechanism 232
13 Fundamentals of Transfer Mechanisms 246
14 Interphase Transfer 302

Part 3 Applications to Equipment Design 325

1 Heat Transfer 327
16 Mass Transfer 384
17 Simultaneous Heat and Mass Transfer — Humidification 420
18 Simultaneous Heat and Mass Transfer — Drying 456
19 Simultaneous Heat and Mass Transfer — Evaporation and Crystallization 494
20 The Energy Balance Flow Systems
in 541
21 Fluid Motive Devices 580
22 Particulate Solids Flow and Separation Through Fluid Mechanics 61

APPENDIX A Dimensions and Units, Dimensional Analysis, and Model Theory 686
APPENDIX B Description of Particulate Solids 699
APPENDIX C Equipment Design Data 716
APPENDIX D Physical Data 727
APPENDIX E Equations of Change 754

INDEX 755

XI
Principles of
Unit Operations
 This ethylene oxide plant is typical of currently operating EO reactor, C0 2 absorber, CO, stripper. The plant is operated by
petrochemical plants. Ethylene is oxidized to ethylene oxide as a BASF-Wyandotte, Development Co., and
licensed by Shell built by
first step toward many compounds. The major towers, from the left, Foster-Wheeler Energy Co. Reproduced with permission.
are: ethylene oxide (EO) stripper, EO absorber, coolant separator.

XIV

Unit Operations in Chemical Engineering
Chemical engineering is defined as ". the applica- . .
tion of the principles of the physical sciences,
together with the principles of economics and
human relations, to fields that pertain directly to
processes and process equipment in which matter is
treated to effect a change in state, energy content,
or composition..." (1).* This very vague defi-
nition is intentionally broad and indefinite as to the
extent of the field. It is probably as satisfactory a
definition as any chemical engineer
practicing
would give. It should be noted that considerable
emphasis is placed on the process and process
equipment. The work of many chemical engineers
would better be called process engineering.
The process may be any collection of steps
involving changes in chemical composition or
involving certain physical changes in material being
prepared, processed, separated, or purified. The
work of many chemical engineers involves choosing
the appropriate steps in the appropriate order to
formulate a process for accomplishing a chemical
manufacturing operation, a separation, or a purifi-
cation. Since each of the steps constituting a
process is subject to variations, the process
engineer must also specify the exact conditions
under which each step is to be carried out.
As the process evolves and equipment must be
designed, the work of the chemical engineer merges
with that of the mechanical and civil engineer. The
transfer of primary responsibility from the process
engineer to the mechanical engineer can take place
satisfactorily at various stages of the design, so it is
•References are collected alphabetically at the end of each
chapter; parenthetical numbers refer to references.
One
impossible to define a fixed extent to which the
responsibility should be called that of a chemical
engineer or a stage at which the mechanical
engineer should take over responsibility for equip-
ment.
At the time the definition quoted above was
presented, the physical sciences referred to were
primarily chemistry and some classical physics. As
the understanding of mathematical models of
chemical processes advances, the treatment of the
chemistry and physics of the process is expressed
in considerably more mathematical form.
Increasing use of thermodynamics, fluid dynamics,
and mathematical techniques such as probability
and statistics, matrix manipulation, and complex
variables is characteristic of modern chemical
engineering practice. In most processes being
carried out on a large scale, however, the chemistry
has been previously worked out, and the physical
changes incident to preparation and purification of
the reaction mixtures demand considerably more
study than does the chemical reaction. Frequent
application of the principles of physics and of
physical chemistry is required in the processing
steps that produce physical changes, such as vapor-
ization, condensation, or crystallization. As a
process evolves into a plant and the work merges
with that of mechanical designers, the science of
mechanics becomes increasingly important. The
chemical engineers who specialize in equipment
must have thorough and extensive grounding in
mechanics of materials.
All of an engineer's work must be quantita-
tive, and mathematicsis therefore a fundamental
tool the engineer. Unfortunately, our under
of
standing of mathematics is largely restricted to the
 UNIT OPERATIONS IN CHEMICAL ENGINEERING

domain of linear mathematics, and, equally SOME BASIC CONCEPTS

unfortunately, chemical molecules seldom behave
Before attempting to describe the operations that
in accordance with linear mathematical rules.
comprise a chemical process, it is necessary to
Energy and material balance calculations, which
introduce several basic concepts that must be
are fundamental to any process study, can usually
understood before a description of the operations
be expressed with confidence and precision in
is meaningful.
terms of linear mathematics, as long as we omit
atomic and nuclear processes from our considera-
tions. In the economic study to determine the
Equilibrium

most profitable operating conditions — and in There exists for all combinations of phases
a con-
accounting for the sales receipts and distribution of dition of zero interchange
of properties
net
income to profits and costs, including replacement (usually mass or energy in chemical processing)
of the plant — mathematical calculations are called equilibrium. For all such combinations not
universal. at equilibrium, the difference in concentration of
The existence or contemplation of a process some property between one in the existing con-
implies that a material is to be produced for which dition and one that would exist at the equilibrium
customers must be delivered in a quan-
will pay. It condition is a driving force, or a potential differ-

tity, of a quality, and at a price which are accept- ence, tending to alter the system toward the
able to the customer. Simultaneously, it must pay equilibrium condition. The tendency of thermal
for materials, labor, and equipment used in the energy to flow from a region of high concentra-
manufacture and return a profit over and above all tion — hot body — to a region of low concentra-
costs. Many materials produced by the chemical tion — cold body — is universally familiar.
industry are planned and plants built before the Similarly, the tendency of electrical energy to flow
real market potential has been developed. For a from a region of high potential to one of low
completely new product, some estimate of the size potential in accordance with Ohm's law (/ = E/R)
of the market must be made, and the plant should is well known. The tendency of acetic acid to flow
be scaled in proportion. from an acetic acid-water solution into an ether
The human-relations aspect of engineering phase in contact with widely known. The
it is less

practice is not usually emphasized in under- description of this equilibrium is considerably

graduate training because of the great quantity of
technical information and techniques the student
must learn. That this may be a fallacious course is
implied by the fact that failures of young engineers
because of personnel problems are at least five
times as frequent as failures because of inadequate
technical training. All engineers must realize that
the industry in which they are working requires
more complicated than the statement of equality
of temperatures, which describes the equilibrium
of energy of molecules. Material will flow from a
region of high concentration (activity) to one of
low concentration (activity), just as heat and
electricity flow from high- to low-concentration
regions in the situations mentioned above.
The expression of the equilibrium condition is

team effort of personnel. Valuable information

all familiar to connection with electrical and
all in

can be obtained from operators of limited thermal energy. The concentration of such energy
educational background who
have observed similar is expressed directly as a voltage potential or a
processes. The person who
has "lived" with an temperature. Accordingly, two bodies at the same
operation has probably observed actions and electrical potential, or at the same temperature,
effects and has learned methods of detailed control will be in equilibrium with regard to that particular
that cannot be approached by formal theory alone. kind of energy. For the equilibrium between a
The best engineering job can be done only with liquid and its vapor, the vapor-pressure curve is
proper regard for all available facts regardless of reasonably familiar. The curve expresses in pressure
their source. A new process or the technical units the concentration of vapor that is in equilib-
improvement of an existing one designed without rium with the pure liquid when both are at a
due regard for the operators is usually destined to specified temperature. In case of a liquid mixture,
failure. The start-up of a new plant or the installa- equilibrium must exist between the liquid phase
tion of a technical change is likely to be much and the vapor phase in regard to each and every
smoother and the cost of it much less if the operat- constituent present. For a binary mixture, the
ing personnel understand the objectives and are relation is a relatively simple one describing the
convinced of their soundness. concentration or partial pressure of each constitu-
 UNIT OPERATIONS IN CHEMICAL ENGINEERING

ent in the vapor phase that is in equilibrium with two phases that are not in equilibrium are brought
a liquid of one composition at the
particular together, a transfer analogous to that for electrical
specified temperature. Obviously, the vapor will be and thermal energy will occur. The
result will be a
of different composition when it is in equilibrium transfer of isopropyl the water-acid
ether into
with different liquid mixtures. The expressions for phase and the transfer of both water and acid into
equilibrium in multicomponent mixtures between the ether phase until the potential of each constit-
the liquid phase and its vapor or between two uent is identical in the two phases. There is no
liquid phases having partial solubilities become convenient and simple expression for the chemical
more involved. In every case, the condition must potential; hence, the amount per unit volume, or
be satisfied that the potential for each constituent concentration, of mass in such a phase is
is identical in all equilibrium phases of a particular commonly so designated. Mass concentration is not
system. a rigorous definition, but the more accurate and
more complex functions of activity, fugacity, and
Driving Force Gibbs free energy demand more knowledge of
When two substances or phases not at equilibrium physical chemistry than is expected at this time. In
are brought into contact, there is a tendency for a the preceding example the mass concentration of a
change to take place that will result in an approach component is different in each phase at equilib-
toward the equilibrium condition. The difference rium.
between the existing condition and the equilibrium In all cases discussed above, the potential
condition the driving force causing this change.
is (concentration) of an existing substance or mixture
The difference can be expressed in terms of when compared with the potential at the equilib-
concentrations of the various properties of sub- rium condition yields a difference in potential that
stances. For example, water of low energy
if liquid is a driving force, tending to change the conditions

concentration — that low temperature — is

is,
brought in contact with water vapor of high energy
concentration — that is, high temperature — energy
will be transferred from the vapor phase to the
liquid phase until the energy concentration is the
same in both phases. In this particular case, if the
amount of liquid is large in comparison with the
vapor, both phases become one by the condensa-
tion of the vapor as its energy is transferred to the
of the system toward the equilibrium. The driving
forces, or differences in the potential of energy or
matter, will tend to produce a change at a rate
directly proportional to the difference from the
equilibrium potential. The rate at which the system
changes toward equilibrium is one of the major
topics to be covered in this book.
Separations

cold water. The final mixture will be an increased Obviously, the separation of a solution, or other
amount of liquid water at a higher temperature physically homogeneous mixture, requires prefer-
than initially and a decreased amount of water ential transfer of a constituent to a second phase
vapor. This combination reaches equilibrium very that may be physically separated from the residual
quickly, at a temperature such that the vapor mixture. Illustrations are the dehumidification of
pressure of the water equals the pressure of the air by condensing or by freezing a part of the
vapor phase. A similar line of reasoning can be moisture, or the use of a liquid solvent that is

followed in the case of two electrical condensers insoluble in the unextracted material. Any two
charged to different concentrations (i.e., voltage). phases that exhibit preferential distribution of con-
If they are brought into electrical contact, the stituents and that can be easily separated may be
electrical energy will flow from the region of involved in a separation operation. Two solid
higher concentration to that of lower. Both con- phases may be very difficult to separate; a liquid
densers will be charged to the same voltage when and a gas or solid usually may be easily separated;
equilibrium is reached. two liquids of approximately equal density and no
A less familiar type of driving force exists interfacial tension may resist all practicable separa-
when a solution of acetic acid and water is brought tion means short of altering one of the phases
in contact with isopropyl ether. The three
materials will usually separate into two liquid Flow Patterns
phases, each containing some quantity of all three In many
of the operations for transferring energy
components. The concentration of each of the or from one phase to another, it is
material
three substances in each of the two phases must be necessary to bring two streams into contact to
known to describe the equilibrium condition. If permit a change toward equilibrium of energy or of
 UNIT OPERATIONS IN CHEMICAL ENGINEERING
material, or both. The transfer may be accom-
plished with both streams flowing in the same
direction (i.e., cocurrent flow). If cocurrent flow is
used, the limit in amount of transfer that can occur
is firmly set by the equilibrium conditions that will
be reached between the two streams being contacted.
If, however, the two streams being contacted are
made to flow in opposite directions, transfer of
material or energy in considerably greater amounts
is possible. Such a flow pattern is known ascounter-
current flow.
As an illustration, if a stream of hot mercury
and a stream of cold water are allowed to reach
thermal equilibrium, the temperature attained can
be predicted by a heat balance that recognizes the
relative quantities of the streams, their initial
temperatures, and their heat capacities. If the
streams flow simultaneously from the same inlet
point to the same outlet point, the equilibrium
temperature is definite, and the path is as indicated
in Figure 1.1a. If the streams are made to flow in
opposite directions, as by letting the mercury flow
downward through an upflowing stream of water,
it is possible for the entering hot-mercury stream
to raise the temperature of the leaving cool-water
stream to a temperature above that to which the
mercury stream is lowered as it leaves the contact-
ing equipment, as indicated in Figure 1.1/?. The
counterflow principle is used in many chemical
engineering operations in order to permit greater
transferof a property than would be indicated
merely by the attainment of a single equilibrium
between the leaving streams.
Continuous and Batch Operation
In the majority of chemical processing operations,
it more economical to maintain continuous and
is
steady operation of equipment, with a minimum of
disturbances and shutdowns. This is not always
practical in some small-scale operations, in opera-
H,0„ Hg„
1 :
i i
H2 out
-t -
.
1 L
.
„ Hg
(a)
Figure 1.1. Flow and temperature
current
tions where extremely corrosive conditions force
frequent repairs, and in others for various specific
reasons. Because of the greater productivity of
continuously operating equipment and the resul-
tant lower unit cost, it is usually advantageous to
operate equipment continuously. This means that
time is not a variable in the analysis of such a
process, except during the rather brief start-up and
shutdown periods. The time rate of transfer or of
reaction is important in fixing the necessary size
and capacity of equipment, but the performance is
expected to be the same today, tomorrow, or next
year if the operating conditions remain the same.
Conditions are not constant throughout a system
at any time, but those at a particular point are
constant with time.
When small quantities of material are to be
processed, it is often more convenient to charge
the entire quantity of material to the equipment,
process it in place, and remove the products. This
is called a batch operation.
An operation that is variant with time is
spoken of as a transient or unsteady state, in con-
trast with that spoken of as steady state, in which
conditions are invariant with time. Quenching a
steel part for heat treating and freezing ice cubes
in a domestic refrigerator are illustrations of
unsteady-state operations. In batch operations,
almost the entire cycle is a start-up transient and a
shutdown transient. In a continuous operation, the
time during which the start-up transient exists may
be extremely small in comparison with the steady
state operation. Analysis of transient or batch
operations is usually more complex than of steady-
state operation. Because of the greater simplicity
and the wide occurrence throughout chemical
processing of steady-state operations, the intro-
ductory treatment is in terms of conditions that do
not vary with time. Analysis of a transient opera-
tion is different from the steady state only in the
H,0
Temperature H 2 0„ *— ' ' * Hg. Temperature
(b>
in a contactor, (a) Concurrent, (b) Counter-
 UNIT OPERATIONS IN CHEMICAL ENGINEERING

introduction of the additional variable of time. tions, or steps in a manufacture that could be used
This variable complicates the analysis but does not without alteration in a variety of
significant
fundamentally change it. processes. This the basis of a terminology of
is

'unit operations,' which now offers us a list of

techniques, all of which cannot be covered in a
UNIT OPERATIONS
reasonable text.
Chemical processes may consist of widely varying Very frequently chemical changes occur in a
sequences of steps, the principles of which are material being distilled or heated. In such cases the
independent of the material being operated upon physical operation is the primary concern, and if a
and of other characteristics of the particular chemical change occurs simultaneously, it is
system. In the design of a process, each step to be commonly handled by a modification of the
used can be studied individually if the steps are physical properties of the material. When chemical
recognized. Some of the steps are chemical rates and equilibria are known, these may be
reactions, whereas others are physical changes. The mathematically modeled into the unit operation
versatility of chemical engineering originates in calculations.
training to the practice of breaking up a complex The typical chemical manufacturing operation
process into individual physical steps, called unit involves a few chemical steps that are probably
operations, and into the chemical reactions. The straightforward and well understood. Extensive
unit-operations concept in chemical engineering is equipment and operations are usually needed for
based on the philosophy that the widely varying refining or further preparing the often complex
sequences of steps can be reduced to simple opera- mixture for use as an end product. The result is
tions or reactions, which are identical in funda- that the work of the typical process engineer is
mentals regardless of the material being processed. much more concerned with physical changes than
This principle, which became obvious to the with chemical reactions. The importance of the
pioneers during the development of the American chemical reactions must not be overlooked because
chemical industry, was first clearly presented by of the economic importance of small improve-
A. D. Little in 1915: ments in percentage yield from chemical reactions.
In many cases a relatively small percentage im-
Any chemical process, on whatever scale conducted, may provement in yield may economically justify con-
be resolved into a coordinated series of what may be siderably more extensive processing operations and
termed "unit actions," as pulverizing, mixing, heating, equipment.
roasting, absorbing, condensing, lixiviating, precipitating, All unit operations are based upon principles
crystallizing, filtering, dissolving, electrolyzing and so on.
of science that are translated into industrial appli-
The number of these basic unit operations is not very large
cations in various fields of engineering. The flow of
and relatively few of them are involved in any particular
fluids, for instance, has been studied extensively in
process. The complexity of chemical engineering results
theory under the name of hydrodynamics or fluid
from the variety of conditions as to temperature, pressure,
etc., under which the unit actions must be carried out in
mechanics. It has been an important part of the
different processes and from the limitations as to materials work of civil engineers under the name of hydrau-
of construction and design of apparatus imposed by the lics and is of major importance in sanitary engineer-

physical and chemical character of the reacting sub- ing. Problems of water supply and control have
stances. (2) been met by every civilization.
Heat transfer has been the subject of many
The original listing of the unit operations theoretical investigations by physicists and mathe-
quoted above names twelve actions, not all of which maticians; it has played a major part in the gener-
are considered unit operations. Additional ones have ation of power from fuels, as developed by mechan-
been designated since then, at a modest rate over ical engineers. Dissipation of heat in electrical
the years but recently at an accelerating rate. Fluid equipment ismajor limitation on the power out-
a
flow, heat transfer, distillation, humidification, gas put of such machinery. Pyrometallurgy and the
absorption, sedimentation, classification, agitation, heat treatment of materials of construction and
and centrifugation have long been recognized. In tools represent additional major applications.
recent years increasing understanding of new tech- Throughout industry, one finds examples of
niques — and adaptation of old but seldom used most of the unit operations in applications that are
separative techniques — has led to a continually in the province of other engineering fields. The
increasing number of separations, processing opera- chemical engineer must carry out many unit oper-
 UNIT OPERATIONS IN CHEMICAL ENGINEERING
ations on materials of widely varying physical and
chemical properties under extremes of conditions
such as temperature and pressure. The unit oper-
ations used to separate mixtures into more or less
pure substances are unique to chemical engineer-
ing. The materials being processed may be natu-
rally occurring mixtures or they may be the prod-
ucts of chemical reactions, which virtually never
yield a pure substance.
INTEGRATION OF THE UNIT OPERATIONS
Early studies of the unit operations as independent
steps laid an important part of the foundation for
the phenomenal growth of the chemical industry.
In a complex manufacture, the interactions of the
steps force engineers to consider the total process,
or system, as an entity. Simultaneously, clearer
understanding of the interrelations of the funda-
mental principles leads to a grouping of operations
to fit the same mathematical expression of actions,
or model, permitting valuable generalizations.
The traditional presentation of unit operations
has been the collection of appropriate theoretical
and practical information about each unit oper-
ation as a package. In previous textbooks each
operation has been presented independently from
the others. It is seldom obvious in introductory
presentations that several of the unit operations
overlap in their foundations and are quite intri-
cately related each other. The interrelations
to
become more obvious in monographs on several of
the unit operations because of the impossibility of
presenting the theory on any one of the operations
completely without regard for the influence of
others. Specifically, heat transfer in a flowing sys-
tem cannot be completely presented without con-
sideration of the fluid mechanics; mass transfer
cannot be divorced from heat transfer and fluid
mechanics.
With increasing information has come broader
recognition of the basic similarities. Conversely,
recognition and exploitation of the similarities
have contributed to a broader understanding of
each operation. It now appears that the compart-
mentalization of information by unit operation
leads to unnecessary repetition and waste of time
and that study of basic principles common to a
group of the operations will lead to a better under-
standing of all of them.
This book presents under single headings those
operations having similar fundamentals, using
generalized nomenclature and concepts. This pre-
sentation has been found to result in time econ-
omy in learning and is believed to contribute a
greater breadth of understanding of all the oper-
ations when the interrelations are understood.
Analysis of the Unit Operations
The unit operations may be analyzed and grouped
using any one of three possible methods. A unit
operation may be analyzed using a simple physical
model that reproduces the action of the operation;
it may be analyzed by considering the equipment
used for the operation; or it may be analyzed
starting with a mathematical expression that de-
scribes the action and is tested using experimental
process data.
The physical model is established by careful
study of the basic physical mechanism. The model
is then applied to a real situation either through
mathematical expression or by physical descrip-
tion. Because the model is idealized, some correc-
tions are necessary in its application to real oper-
ations. This approach develops an understanding of
the basic similarities among the principles of the
various unit operations.
The grouping could be made in terms of those
operations accomplished in similar equipment or in
which a similar function occurs. For most of the
operations, the art preceded a scientific under-
standing, and equipment was built and operated on
the basis of woefully incomplete basic knowledge.
Some improvements and refinements came, as ex-
pected, purely from the art and the equipment.
Grouping on the basis of equipment and its func-
tioning exposes one to the risk of only perpetu-
ating the mistakes of the past. A thorough under-
standing of the basic operation seems much more
likely to yield improvements in operations.
The operations could also be grouped in the
light of similarity of the basic mathematical form-
ulation of the operation. This method of grouping
is unsatisfactory because of the perversity of mole-
cules in their disregard of mathematics. Because of
the nonlinearities involved and because boundary
conditions of one phase usually respond to changes
occurring in an adjacent phase, it is frequently
impossible to formulate the boundary conditions
for solution of a mathematical expression in man-
ageable terms.
Each of the three modes of grouping could be
used as a basis. The physical model of the funda-
mental operation is the most satisfactory approach
and is used in this presentation. Wherever possible,
the physical model is described mathematically,
and the performance is expressed in mathematical
relations derived from the fundamental principles.

This formulation gives the best basis for under-
standing and refining those operations in which the
art is ahead of theory. This is true in spite of the
fact that the models are oversimplified and that the
mathematical formulation of the behavior of the
model cannot be transposed perfectly into an ex-
pression of the behavior of the prototype.
It should be obvious that there is no universal
criterion dictating a particular choice of method of
analysis and that allcontributing factors should be
recognized in deciding upon a particular mode.
Any grouping requires some arbitary choice and
always leaves one with some of the operations that
fit poorly into the general scheme. Such operations
must be studied individually.
Two Major Physical Models
One widely applicable model for unit operations is
a device in which two streams, or phases, are
brought together, allowed to reach equilibrium,
then separated and withdrawn. It is assumed that
the leaving streams are at equilibrium, and this
model is called an equilibrium stage. Evaluation of
the changes in the streams that must be accom-
plished to attain equilibrium establishes a measure
of ultimate performance. Real equipment is eval-
uated by expressing the changes accomplished in it
as a fraction or percentage of the changes that
would occur in an equilibrium stage. In another
possible model for transfer of a property between
two streams, we visualize the carriers of the prop-
erty, evaluate theirnumber and rate of migration,
and arrive at an expression of the rate of transfer
between the two streams in continuous contact.
This rate of transfer multiplied by the time of
contact yields an expression for the amount of
transfer accomplished. The equilibrium-stage
model may be expressed mathematically in a fi-
nite-difference equation relating entering con-
centrations of any property with the equilibrium
concentrations of the property in the leaving
streams. Graphical techniques frequently can be
used more conveniently than the finite-difference
equation. The mathematical expression for the
rate-of-transfer model is a differential equation
that can sometimes be integrated rigorously but
more frequently must be handled in terms of aver-
age conditions. Since a large number of chemical
processing operations are actually carried out
continuous contact, these two
either in staged or in
models are widely applicable for the analysis of
unit operations.
Most of the unit operations can be studied on
either of the two bases. Many of them are carried
UNIT OPERATIONS IN CHEMICAL ENGINEERING
out sometimes in continuous-contact equipment
and sometimes in staged equipment. In some oper-
ations, the advantage of one or the other mode of
analysis may be obvious. In many others, the
choice is dictated by availability of the necessary
data and constants. Equilibrium data are a part of
the stock in trade of the physical chemist and are
available for a large number of substances under
various conditions. To some extent, the con-
venience in analysis is related to the work of earlier
investigators, in that their results may have been
interpreted in a fashion thatmakes one or the
other analysis more convenient. Choice of one
method of analysis does not necessarily restrict the
actual operation to the same model.
The Stage Operations
Operations in which staged contacting is frequently
used will be considered first. The model is the
device in which the two incoming streams interact
to attain equilibrium between the streams as they
leave the stage. The model is known as an equi-
librium stage and is assumed always to yield two
product streams in equilibrium with each other.
The generalized treatment does not require a speci-
fication of the property being transferred or of the
nature of the phases being contacted. The practical
analysis is based on the fraction of transfer accom-
plished in the actual stage as compared to the
equilibrium stage. Presentation will be in as com-
pletely general terms as possible, without regard tc
the particular nature of the phases in a particular
case.
Staged contacting may be illustrated using the
mercury and water streams discussed earlier. As
shown in Figure 1.1, the mercury and water
streams are in continuous contact, and heat is
transferred continuously from the hot stream to
the cold stream. For stage contacting, the equfp-
ment is modified as described below. If the hot-
mercury and cold-water streams used above as an
illustration are mixed intimately and then fed to a
settler where the phases are separated, the out-
flowing streams will be at practically the same
temperature. The equilibrium temperature can be
predicted by a material and an energy balance.
Suppose now that two mixer-settlers are provided,
one of which receives the hot mercury and the
other of which receives the cold water. The mer-
cury leaving the warmer mixer-settler flows to the
cooler mixer-settler, and the water leaving the cool-
er mixer-settler flows to the warmer mixer-settler.
The two mixer -settlers will accomplish the transfer
of more heat than the one. If the number of
 UNIT OPERATIONS IN CHEMICAL ENGINEERING

mixer-settlers is increased to n, even more energy equation for average conditions rather than solved
can be taken from the mercury. In this case, the as a differential.
mercury would be passed through them in the In the simplest cases, quite unusual in chemical
order 1, 2, 3, . . . ,n and the water n 3,2,1. processing, the rate of transport is constant with
The introduction of additional stages decreases the time and position within the system. The driving
heat transferred per stage because the potential force may be assumed to be constant and distri-
difference from equilibrium becomes less, but the buted over a path of fixed length and of constant
total transfer is increased. No one would do this area. The physical properties of the path may be
particular operation in the manner described, but constant so that the proportionality factor 6 may
many transfer operations use stage contacting. be assumed to be constant. These assumptions are
Stage contacting is a common way of extracting the simplifications that have been introduced in
one component from a liquid mixture by prefer- arriving at Ohm's law in its form as usually pre-
ential solution of that compound or group of com- sented in introductory physics courses. The
pounds such as in the removal of sludge-forming counterpart in chemical transport becomes
components from lubricating oils.
driving force/unit distance
Rate of transport =
The Rate Operations resistance/unit of path area
The unit operations involving continuous contact- 1.2)
ing depend upon the rate of transfer, and are
therefore called rate operations. The transfer of a Chemical substances seldom fit nice mathe-
large number of properties of a material — such as matical equations and chemical equilibrium is con-
electric, magnetic, thermal, mass, and momentum stantly upsetting neat formulations of boundary
concentrations — follows the same basic mathe- conditions that would permit rigorous solutions of
matical expression of rate of transfer as a function Equation 1.1; therefore, various averages and ap-
of concentration gradient proximations must be used in arriving at an answer
in an economical length of time. The simplifi-

— = 5^
30 3x 2
where Y = concentration of the property to be
transferred
8 = time
x = distance measured in direction of
transport
6 = proportionality constant for a system
This equation is frequently called the diffusion
equation. It is a general expression that reduces to
Ohm's law for electrical flow for specified con-
ditions. The broad study of electric and magnetic
transport is the important "field theory" of the
electrical engineer. These two phenomena follow
well-established laws and involve
stant proportionality factors (such
above equation). Since boundary conditions can
usually be evaluated, analytical solutions are fre-
quently possible for engineering calculations.
Chemical substances are less well behaved mathe-
matically, and the proportionality "constants" are
seldom really constant. The boundary conditions
are more elusive; hence, chemical engineers are
seldom in position to apply mathematically elegant
and rigorous solutions of the diffusion equation. In
order to arrive at a solution of the diffusion equa-
tion, it may be simplified into a finite-increment
(1.1
cations usually approach Equation 1.2 much more
closely than the rigorous diffusion equation.
For the rate operations, analysis must be based
upon the driving force causing a change, the time
during which a driving force is allowed to act, and
the quantity of material upon which it acts. The
diffusion equation above expresses the transient
behavior of a large number of properties under the
influence of a driving force for transport of the
property. In chemical engineering, mass, momen-
tum, and thermal energy are the three properties
whose transport is the most frequently involved.
As mentioned above, it is universal that these three
properties, along with a number of others with
which chemical engineers are less frequently con-
relatively con- cerned, tend to flow from regions of high concen-
as 5 in the tration to regions of low concentration. Accurate
prediction of the amount of the property that
flows from a donor region (source) to a receiver
region (sink) can be made if the driving force, the
area of the path, and the unit resistivity of the path
(the proportionality constant used in Equation
1.1) are accurately known. Throughout the study
of the rate operations, the importance of a clear
understanding of the meaning of concentration
cannot be overstressed. In every case, the con-
centration expresses the amount of property per
unit volume of the phase being processed. The

amount being transferred can usually be expressed
in some absolute unit measuring that quantity,
such as British thermal units or pound moles. It
can also be expressed in terms of the decrease in
concentration of the property in a known amount
of phase having a known capacity for this prop-
erty. For example, a quantity of energy leaving a
system as heat can be expressed in terms of the
number of Btu's or calories of energy. It can also
be expressed in terms of the decrease of temper-
ature of a known amount of the phase. These
generalizations will become more meaningful as
different operations are analyzed and the trans-
ported quantities are expressed in terms of the
various possible units. Because the basic principles
of transport are identical for the three properties,
an analysis will be offered in completely general
terms before specification of the particular prop-
erty in specific operations.
Unsteady-State Operation
The diffusion equation, Equation 1.1, is applicable
for transfer one direction only and that is a
in
function of time. However, transfer may occur in
more than one direction. Rigorous solution requires
knowledge of boundary conditions and interaction
of variables. Computer techniques make solutions
quite accessible, and they usually employ finite-
difference techniques.
GENERAL CONSIDERATIONS
Understanding the basic physical principles of an
operation and formulating these principles into a
mathematical expression are the first requirements
for applying the principles of the unit operations.
In engineering practice, however, numerical values
must always be incorporated and a practical answer
obtained.
The same problem may be met by the design
engineer in specifying equipment, by the operating
engineer checking the performance of installed
in
equipment, or by any engineer in seeking improve-
UNIT OPERATIONS IN CHEMICAL ENGINEERING
ment in quality or quantity. It is therefore neces-
sary that mathematical and/or graphical techniques
be available to permit the prediction of any un-
known answer for a particular system regardless of
whether the unknown is a composition, quantity,
temperature, or number of stages required to ac-
complish a specified amount of enrichment of any
chosen property.
Although this book is devoted exclusively to
the principles of the unit operations of chemical
engineering, it should be emphasized to the prospec-
tive chemical engineer that unit operations are only
one sector of chemical engineering. The real objec-
tive is the engineering of the most economical pro-
cess. The unit operations are techniques in arriving
at this process, but they must not be allowed to
crowd out of consideration the other important
scientific principles that must be recognized.
The best process can be designed only with
proper regard for the basic chemistry, kinetics, and
thermodynamics, with adequate recognition of the
limitations imposed by materials of construction
and auxiliaries to the plant. The equipment design
will involve work by engineers trained in disciplines
normally not covered by the chemical engineer.
The ultimate object of the engineering is the ac-
cumulation of a profit from the operation. The
largest return of profit, after all costs are account-
ed for, demands full exploitation of all the tech-
nical factors involved, favorable human relations
within the producing team, and accurate know-
ledge of the amount of product that can be sold
for maximum eventual profit.
REFERENCES
1. Constitution of the American Institute of Chemical
Engineers.
2. Little, A. D., Report to the Corporation of M.I.T., as
quoted in Silver Anniversary Volume, A. I.Ch.E. (1933),
p. 7.

Stage Operations
Nearly chemical processes require the separation
all
of a solid, a liquid, or a gas phase into its con-
stituent components. Such separations are often
made by contacting the phase with another phase
so that some of the components are transferred
from one phase to the other. The* two phases are
then separated and processed further.
Raw materials, intermediate streams, and final
products are frequently purified using a separation
process. For example, in the manufacture of dry
ice, carbon dioxide is removed from a mixture with
air by contacting the gas with a liquid (such as a
diethanolamine solution) that dissolves carbon di-
oxide but not air. The pure carbon dioxide is then
recovered by simply heating the amine solution.
The term separation processes includes those
unit operations involving separation of components
by transfer of mass between phases. The separation
of whole phases, as in filtration, may also be in-
cluded. Here we deal only with those operations in
which component transfer between phases occurs,
leaving the separation of heterogeneous, multi-
phase systems to Part III. Within this class of pro-
cesses the choice of unit operation to be used
depends upon the physical properties of the mix-
ture to be separated. Obviously, the separation is
easier if it is based upon a physical property that
varies widely among the components to be sepa-
rated.
The equipment used in mass-transfer oper-
ations depends upon the phases to be processed, as
discussed in Chapter 2. In some cases, the phases
are repeatedly brought together and then separated
as they pass through the equipment. The equi-
librium stage is the physical model used to analyze
mass-transfer operations in which the phases are
alternately contacted and separated a number of
11
Part One
times. Hence, the term stage operations refers to
those separation processes that can be described by
the model of an equilibrium stage. These opera-
tions are the subject of the next seven chapters.
In other types of equipment, the two phases
are continuously contacted so that the extent of
mass transfer depends explicitly upon the rate of
transfer and upon the time of contact. In this case,
the physical model must include the factors of rate
of mass transfer and time of contact. These rate
operations will be considered in Parts II and III. In
Chapter 16 the mass-transfer operations are ana-
lyzed using a continuous-contact model.
The basic calculations required to design
equipment for the various staged operations are
based upon identical concepts. This section con-
siders the general concepts involved in the design
of multistage separation processes. Where the vari-
ous stage operations differ, each is discussed in
detail, but wherever possible the operations are
developed in a unified manner. The calculation
procedures are illustrated with examples from spe-
cific stage operations.
Chapter 2 introduces the various mass-transfer
operations and describes typical equipment used in
the operations. The phase relationships upon which
the separation processes are based are considered in
Chapter 3. Chapters 4 through 6 develop a general
method of calculation of steady-state staged oper-
ations. In Chapter 7, certain simplified methods are
discussed, and Chapter 8 develops calculation
methods for multicomponent staged operations.
A complete tabulation of the general notation
and its application to specific staged operations is
given at the end of Part I. The reader is urged to
refer to it often and study it carefully while reading
the following chapters.
 Two

Mass-Transfer Operations

When two phases of different compositions are riving at a choice of separation processes. The
brought into contact, a transfer of components energy requirements, the cost and availability of
may occur from one phase to the other, and vice process and construction materials, and the inte-
versa. This is the physical basis of mass-transfer gration step in the overall chemical process all
operations. If the two phases are allowed to remain contribute to determining which separation process
in contact for a sufficient time, they will reach an is economically most attractive.

equilibrium condition where there is no further net Unit operations is concerned with those sep-
transfer of components between phases. In most aration processes that depend upon differences in
cases of interest in mass-transfer operations, the physical properties, rather than chemical behavior.
two phases are only partially miscible, so that at Such processes depend either upon a difference in
equilibrium there two phases that can be
still exist composition of phases at equilibrium or upon a
separated from each Usually, these two
other. difference in the rate of mass transfer of con-
phases have compositions different from each stituents of a mixture. In the following discussion,
other and also different from the compositions of separation processes using both of these physical
the two phases that were initially contacted. As a bases are discussed. In this section, primary interest
result, the relative amounts of components trans- is centered on the separations based upon differ-
ferred between phases are different, so that a sepa- ences in composition of equilibrium phases. The
ration is achieved. Under appropriate conditions, principles of separations based upon transfer rates
repeated contacting and separation of phases can are discussed again in more detail in Part III.

lead to an almost complete separation of com-

ponents. The dissimilar compositions of equi- Distillation

librium phases are the physical bases for the separ- The most widely used separation process in the
ation processes utilizing multistage equipment, chemical industry is distillation. This unit oper-
such as that shown in Figure 2.1 ation is also referred to as fractionation or frac-
tional Separation of constituents is
distillation.
Separation Processes based upon differences in volatility. In distillation,
When faced with the problem of separating com- a vapor phase contacts a liquid phase, and mass is
ponents out of a homogeneous mixture, the engineer transferred both from the liquid to the vapor and
utilizes differences in the properties of the con- from the vapor to the liquid. The liquid and vapor
stituents of the mixture to effect the separation. The generally contain the same components but in dif-
various chemical and physical properties of the ferent relative quantities. The liquid is at its bubble
constituents of the mixture are examined point,* and the vapor in equilibrium is at its dew
to determine which properties offer the greatest point.* Mass is transferred simultaneously from the
difference among components, because a greater liquid by vaporization and from the vapor by con-
difference in a property will generally permit an
easier, more economical separation. Of course, the
engineer must consider many other factors in ar- 'Defined in Chapter 3.

13
14 STAGE OPERATIONS

relationship exists. This is discussed more fully in

Chapter 3. Distillation is widely used to separate
liquid mixtures into more or less pure components.
Because distillation involves vaporization and con-
densation of the mixture, large quantities of energy
are required.
A great advantage of distillation is that no
additional component need be added to effect the
separation. As seen in the following discussion,
many other separation processes require the addi-
tion of another component, which must then be
removed later in another separation step. The tem-
perature and the volume of materials being boiled
depend on the pressure. Elevated pressure may be
used to decrease volumes and/or to increase tem-
peratures to facilitate condensation; decreased
pressures may be needed to lower the boiling point
below the point of thermal decomposition.
Applications of distillation are tremendously
diverse. Pure oxygen, for use in steel-making, in
rockets, and for medical applications, is produced
by the distillation of air that has been liquefied.
Crude petroleum is initially separated into a num-
ber of fractions (such as light gases, naphtha, gas-
oline, kerosene, fuel oils, lubricating oils, and as-
phalt) in large distillation columns. These fractions
are further processed into finished products, and
distillation is frequently used in the intermediate
steps in the manufacture of the final products.
Figure 2.1 Separation processes in a petroleum refinery.
Distillation is frequently carried out in multi-
Furfural, an organic liquid made from oat hulls, selectively
stage equipment; continuous-contact equipment is
dissolves undesirable components of lubricating oil. In the
also used.
pictured process, lubricating oil stock flows through the
large pipe entering from the left into the extraction column
Gas Absorption and Desorption
between the two columns left of center.
taller distillation
Furfural is fed to the top of the column and leaves at the Gas absorption involves the transfer of a soluble
bottom with the undesirable components. The purified component of phase into a relatively non-
a gas
lubricating oil leaves at the top. The remaining columns volatile liquid absorbent. Desorption is the reverse
pictured are for distillation and stripping, to recover fur- process, removal of a component of liquid by con-
fural solvent, and to further purify the product —the de- tact with a gas phase.
sired lubricating oil. (Courtesy Foster Wheeler Energy
In the simplest case of gas absorption, none of
Corp.)
the liquid absorbent vaporizes, and the gas contains
only one soluble constituent. For example, am-
monia is absorbed from an air-ammonia mixture by
densation. The net effect is concen-
an increase in contacting the gas with liquid water at room tem-
tration of the morevolatile component in the vapor perature. Ammonia is soluble in water, but air is
and of the less volatile component in the liquid. almost insoluble. The water does not vaporize to
Vaporization and condensation involve the latent an appreciable extent at room temperature. As a
heats of vaporization of the components, and heat result, the only mass transfer is of ammonia from
effects must therefore be considered in distillation the gas phase to the liquid. As ammonia is trans-
calculations. In an ideal solution (such as a mixture ferred to the liquid, its concentration increases
of benzene and toluene), the volatility can be re- until the dissolved ammonia is in equilibrium with
lated directly to the pure-component vapor pres- that in the gas phase. When equilibrium is reached,
sure of each component. In nonideal solutions there is no further net mass transfer.
(such as a mixture of ethanol and water), no simple In more complex cases of absorption, several
 MASS-TRANSFER OPERATIONS 15

components may be absorbed, and part of the extracted component must be sufficiently in-
absorbent may vaporize. Some of these cases are soluble to produce two phases that can be sepa-
covered in Chapter 8. rated.
In absorption equipment, the liquid absorbent Liquid— liquid extraction is also referred to as
is below its bubble point and the gas phase is well solvent extraction. The principles are illustrated in
above its dew point. A further difference between the following chapters using three component sys-
distillation and gas absorption is that the liquid and tems: one solute, one solvent, and a second solvent.
gas phases usually do not contain all of the same An oversimplified presentation of the principles is
components. The heat effects absorption are due
in appropriate here with clarification later. The sepa-
to the heat of solution of the absorbed gas, in ration of one component from a homogeneous
contrast to the heats of vaporization and con- solution is accomplished by adding another and
densation involved in distillation. insoluble constituent, the solvent, in which the
Absorption involves the addition of a com- desired constituent from the solution, the solute, is

ponent to the system (i.e., the liquid absorbent). In preferentially soluble, and to which it will diffuse
many cases, the solute must be removed from the at a characteristic rate until the equilibrium con-
absorbent. This removal may require a distillation centrations of the solute have been reached in each
column, a desorber, or some other separation pro- phase. For example, the solute acetic acid may be
cess. separated from aqueous solution by contact with
Desorption, or stripping, is the opposite of isopropyl ether solvent. Although water is slightly
absorption. In this case, the soluble gas is transfer- soluble in the ether, it is essentially the un-
red from the liquid to the gas phase, because the extracted raffinate component.
concentration in the liquid is greater than that in Solvent extraction is remove un-
used to
equilibrium with the gas. For example, ammonia desirable components from and
lubricating
oil

can be stripped from an aqueous solution by bub- other crude oil fractions, to separate niobium from
bling fresh air through the solution. The entering tantalum, to produce concentrated phosphoric
air contains no ammonia and the liquid does, so acid, and many other applications. Both multistage
transfer is from the liquid to the gas. and continuous-contact equipment is used. In most
Absorption and stripping are widely used in cases, it is necessary to separate the resulting ex-
the chemical industry. Hydrochloric acid is pro- tract into its components.
duced by the absorption of hydrogen chloride gas
in water. Aerobic fermentation of sewage sludge Solid— Liquid Extraction
requires the absorption of Carbonation of soft
air. Components of a solid phase can be separated by
drinks involves the absorption of carbon dioxide; selectively dissolving the soluble part of the solid
some desorption occurs as the bottle is opened and with an appropriate solvent. This operation is also
the pressure is reduced. called leaching or washing. The solid must be finely
Both absorption and stripping are carried out divided so that the liquid solvent contacts all of it.
in multistage equipment and to a lesser extent in Usually the desirable component is soluble, and the
continuous-contact equipment. remainder of the solid is insoluble. The solute must
then be recovered from the extract solution in an
Liquid— Liquid Extraction additional separation step.
A liquidmixture can sometimes be separated by An everyday example of solid— liquid extrac-
contacting it with a second solvent liquid. The tion is making coffee. Here the soluble constituents
components of the mixture are soluble to varying of the ground coffee are separated from the in-
extents in the solvent liquid. Ideally, the com- soluble grounds by solution in hot water. If
ponent to be extracted is soluble in the solvent, allowed to perk long enough, the coffee solution
and the other components are insoluble. Then the would reach an equilibrium with the remaining
solute is the only component transferred from the solid. The resultant solution is separated from the
initial mixture to the solvent phase. The initial grounds.
mixture becomes the raffinate as it is stripped of Solid— liquid extraction is also used indus-
solute. The solvent phase becomes the extract as it triallyin the manufacture of instant coffee to re-

picks up solute. In practice, all components are cover the soluble coffee from the grounds. Other
likely to be soluble to some extent, and a sepa- industrial applications include the extraction of oil
ration is possible only when the solubilities are from soybeans using hexane as a solvent and the
sufficiently different. In any event, the un- recovery of uranium from low-grade ores by ex-
16 STAGE OPERATIONS
traction with sulfuric acid or sodium carbonate
solutions. Because one of the phases is a solid that
does not flow like a fluid, special types of equip-
ment are required for solid-liquid extraction.
Adsorption
Adsorption involves the transfer of a constituent of
a fluid to the surface of a solid phase. To complete
the separation, the adsorbed constituent must then
be removed from the solid. The fluid phase
either a gas or a liquid. If several constituents are
adsorbed to varying extents, it is often possible to
separate them into relatively pure components.
Many solid adsorbents are used. Strictly
speaking, the term adsorbent is usually applied to a
solid that holds the solute to its surface by physical
forces. An example would be the adsorption of
organic vapors by charcoal. The lighter fractions of
natural gas are separated industrially using a
moving bed of adsorbent. Most other industrial
processes use fixed beds and batch or cyclical pro-
cesses rather than multistage equipment because of
the difficulty of moving the solid.
Molecular sieves (4) are special synthetic
adsorbents that can be used to separate mixtures
by differences in molecular size, polarity, or car-
bon-bond saturation. For example, water mole-
cules are relatively small and highly polar, so they
are preferentially adsorbed by molecular-sieve
adsorbents. Thus, these adsorbents are useful in
drying gases and liquids. Molecular sieves are being
used to separate normal paraffins from hydro-
carbon streams in an oil refinery for use in the
manufacture of biodegradable detergents.
Closely related to adsorption are separation
processes that retain solutes in solids by various
means. One such process is ion exchange, where
the solute is retained by a chemical reaction with
the solid ion-exchange resin (1,6). As the name
implies, ions in solution may be removed by this
process. It is widely used to produce highly pure
water. Other applications include recovery of anti-
biotics from fermentation broths and separation of
rare earths.
Clathration is a physical phenomenon
been applied to separate mixtures on the basis of
molecular shape. The solid clathrate compound at-
tracts and retains molecules of a particular shape.
For example, an industrial process separates para-
xylene from meta-xylene by trapping the former in
a clathrate compound. These two compounds have
nearly identical physical and chemical properties so
that conventional separation processes cannot be
used. The clathrate compound might be considered
as a special type of adsorbent.
Ion-exchange resins, molecular sieves, and
clathrate compounds are relatively expensive com-
pared to conventional adsorbents such as charcoal
and silica gel. Hence, these special adsorbents are
used in separations that cannot be accomplished by
more conventional means. These special separation
processes are carried out in batch processes in
fixed-bed adsorbers, rather than in multistage and
continuous-flow equipment.
may be The mass-transfer operations involving trans-
fer to a solid phase are sometimes referred to as
sorption processes, thereby including all types of
solid attractive forces.
A mixture of several adsorbable solutes can be
separated using a fixed bed of adsorbent. First, a
small volume of the mixture is fed to the bed
where completely adsorbed near the entrance.
it is
Then continuous flow of solvent is passed
a
through the bed. Part of the solute mixture is
desorbed and passes out with the solvent. The rate
at which each solute passes out of the bed depends
upon its equilibrium adsorption on the solid. If a
component is strongly held by the adsorbent, it
passes slowly through the bed, whereas a weakly
adsorbed component moves quickly through and
out of the bed. As a result, the components flow
out of the bed at different times and can thereby
be separated. This separation process is called
chromatography (8). Chromatographic separation
is a widely used method of chemical analysis of
small samples. Large-scale industrial applications
are limited. The ion-exchange separation of rare
earths mentioned earlier is an example of chroma-
tographic separation.
Membrane Separations
A number of separation processes involve the trans-
fer of mass through a thin plastic membrane (1).
Although these operations have relatively limited
application, they show promise for special sepa-
ration problems. The theory of membrane separa-
tions is not fully developed, but as a first approxi-
mation, the separation can be considered a matter
of molecular size. Smaller molecules more easily
that has pass through the small pores of the membrane. As
a result, if some driving force is used to "push" the
molecules through the membrane, the smaller
molecules are selectively separated. Some larger
molecules also pass through, so that the separation
is not perfect.
Dialysis is a separation process in which mass
is transferred across a membrane by a con-
centration gradient driving force (13). In the larg-
est industrial application, sodium hydroxide is
separated from an aqueous solution with hemi-

cellulose in the manufacture of viscose rayon. The
solution flows past one side of the membrane and
pure water flows on the other side. The sodium
hydroxide is forced through the membrane because
of the concentration difference between the solu-
tion and the pure water, but the hemicellulose is
in colloidal suspension in the solution and is too
large to pass through the pores in the membrane.
Dialysis is also used to recover sulfuric acid from
copper refinery waste liquors and in the purifi-
cation of blood in the artificial kidney.
Electrodialysis utilizes an electrical potential
difference as a driving force for ions
tions (5). Positive pass through an appro-
ions
priate membrane they are attracted to the neg-
as
ative terminal, and negative ions travel in the oppo-
site direction through a membrane to the positive
terminal. In this way, both positive and negative
ions can be removed.
Electrodialysis is being used to recover waste
acid and to remove salt from brackish well water.
It may also be useful in seawater purification, if
membrane and power costs can be reduced.
Dialysis and electrodialysis both involve mass
transfer across a membrane from one phase to a
second phase. The two phases are completely mis-
cible, so that the membrane also serves to keep
them from mixing together.
A more recently developed membrane separ-
ation process uses a pressure difference to force
molecules of solvent through a membrane, while
molecules of solute remain behind. This process
does not involve mass transfer between phases. It is
more analogous to filtration, where suspended
solid particles are separated from a liquid by for-
cing the mixture through a porous filter medium so
that the liquid passes through and the solid remains
behind. The use of a membrane permits a similar
"filtration" ofhomogeneous solutions that cannot
be separated by an ordinary filter. For this reason
the process is called ultrafiltration. The pressure
applied must exceed the osmotic pressure; the pro-
cess is sometimes referred to as reverse osmo-
sis (10).
In addition to large molecules, low-molecular-
weight ionic salts can be separated if appropriate
ionic membranes are used. Then the membrane
charge serves to prevent passage of charged ions, so
that only the solvent passes through. Reverse
osmosis is being used as a means of removing pure
water from seawater on a large scale.
Gaseous diffusion is a separation process that
bears a superficial resemblance to ultrafiltration,
although the physical mechanism is different. In
gaseous diffusion, a gas mixture is forced through a
MASS-TRANSFER OPERATIONS 17
porous metal barrier (2). The gas component with
a lower molecular weight flows more rapidly
through the barrier, so that the gas beyond the
barrier is richer in that component. Only a fraction
of the gas flowing to the barrier passes through.
The remainder, which is depleted in the lighter
component, is removed for further processing.
Power costs for gas compression are very high, so
gaseous diffusion has only one major industrial
application: the separation of uranium-235 from
uranium-238. Over a billion dollars worth of gas-
eous diffusion plants make this separation to
in solu- produce enriched U 235 for nuclear reactor fuel.
Other Separation Processes
A number of important unit operations involve the
separation of components of a mixture, but the
operations are not customarily grouped with the
usual separation processes.
Evaporation, crystallization, and drying in-
volve simultaneous heat and mass transfer. In evap-
oration a liquid solution is concentrated by vapor-
izing part of the solvent. The heat requirements are
large because the latent heat of vaporization of the
solvent must be supplied. Evaporation has many
industrial applications wherever solutions must be
concentrated. What is commonly referred to as
seawater distillation is in fact an evaporation pro-
cess to recover potable water.
If a solution is evaporated until it becomes
saturated with solute, further evaporation or cool-
ing results in precipitation of solid crystals. This is
the physical basis for crystallization a unit oper-
,
ation used to separate solutes from solution.
Crystallization is used industrially in the manu-
facture of many inorganic salts. It can also be used
to separate mixtures of salts by fractional crystal-
lization. Evaporation and crystallization are dis-
cussed in detail in Chapter 19.
Drying separates a liquid from a solid by
vaporizing the liquid. This important unit oper-
ation is covered in Chapter 18.
Dehumidification separates a vapor com-
ponent of a gas phase by cooling the gas until the
vapor condenses. This mass-transfer operation is
the basis for air-conditioning; it is covered in
Chapter 17.
Multiphase mixtures may be separated into
two or more fractions by a variety of operations.
In this case, the phase mixture must be hetero-
geneous; that is, the phases must be physically
distinct from each other, even though they may be
finely divided and intimately mixed together. Mix-
tures of solid particles can be separated into differ-
ent sizes by screening, elutriation, jigging, or c/assi-
18 STAGE OPERATIONS
fication. Solid be separated from
particles can
liquids by filtration, or sediment-
centrifugation , equipment, the equilibrium stage is a useful physi-
ation. Immiscible liquid phases may also be sepa-
rated by centrifugation, as in the typical cream
separator. Solids can be separated from gases or
liquids, and liquids can be separated from gases by
cyclone separators. These separation operations do
not involve mass transfer in the usual definition of
the term; therefore, they are not mass-transfer
operations. The components are already physically
distinct, so that aprocedure rather
mechanical
than mass transfer is required to separate them.
The mechanical separation operations are mention-
ed here to emphasize the wide variety of physical
principles upon which separations are based.
of them
are discussed in Chapter 22.
This section has briefly described the unit
operations that separate components of a mixture.
Those separations that can be described by the
physical model of an equilibrium stage are dis-
cussed in the next several chapters. Many of the
other separation processes are discussed in Part III
of this book.
The Stage Concept
A stage may be defined as a unit of equipment in
which two dissimilar phases are brought into inti-
mate contact with each other and then are mechan-
ically separated. During the contact, various dif-
fusing components of the mixture redistribute
themselves between the phases. The resultant two
phases have approached equilibrium and therefore
have compositions different from the initial phases.
By successive contact and separation of dissimilar
phases (a multistage operation), large changes
the compositions of the phases are possible. In an
equilibrium stage the two phases are well mixed for
a time sufficient to allow establishment of thermo-
dynamic equilibrium between the phases leaving
the stage. At equilibrium no further net change of
composition of the phases is possible for a given set
of operating conditions. In actual industrial equip-
ment, it is usually not practical to allow sufficient
time with thorough mixing to attain equilibrium.
Therefore, an actual stage does not accomplish as
large a change in composition as an equilibrium
stage. The stage efficiency is defined as the ratio of
a composition change in an actual stage to that
an equilibrium stage. Stage efficiencies for indust-
rial equipment range between a few percent and
about 100 percent. Because an equilibrium stage
gives the greatest composition change possible for
given set of operating conditions, it is
to as an ideal or theoretical stage. Even though
equilibrium conditions are rarely obtained in actual
cal model in the analysis of stage operations. As is
often the case in chemical engineering, the physical
model is an idealized version of an actual unit of
equipment.
Calculating equipment requirements for in-
dustrial multistage
operations usually involves
determining the number of equilibrium stages fol-
lowed by applying stage efficiency to give the
number of actual stages required. The stage effici-
ency is a "correction factor" to carry the physical
model of an equilibrium stage into the practical
application of actual stages in industrial equip-
Some ment. Stage efficiencies depend upon the rate of
mass transfer and upon the time and area of con-
tact in a single stage. These factors are complex
functions of the geometry of the stage and the
flow rates of phases. Evaluation of stage efficien-
cies is discussed in Chapter 16.
Equipment for Stage Operations
Equipment for stage operations varies greatly in
size and in construction details, but there are many
fundamental similarities. In general, each stage of
the equipment mixes the incoming two phases
thoroughly so that the material can be transferred
as rapidly as possible from one phase to the other.
Each stage then must separate the resultant two
phases as completely as possible and pass them on
to the adjacent stages. Some industrial equipment
may consist of a single stage, but more often multi-
stage units are employed with countercurrent flow
of the two Multistage operation permits
phases.
in greater changes in the composition of the two
phases than can be accomplished in one stage.
The equipment for mixing and separating the
two phases depends upon the physical character-
istics of the phases themselves. The L-phase may be
a liquid or a finely divided solid; the V-phase may
be a gas, vapor, or liquid, depending on the sepa-
ration process. Gases and liquids flow easily from
stage to stage, but solids are more difficult to
transfer. For this reason, stage operations where
the solid moves countercurrent to a fluid phase are
less common. Very finely divided solids can be
made to flow like fluids, so this technique is used
in to move solids. In other cases, the solid must be
mechanically transferred from stage to stage.
For economic reasons, stages must accomplish
the contact and separation as simply as possible.
a For example, the force of gravity is used to separ-
also referred ate phases and promote flow whenever possible, to
avoid the need for expensive equipment such as
 MASS-TRANSFER OPERATIONS 19

pumps or centrifuges. On the other hand, where distillationand absorption columns. The bubble
the separation is difficult, expensive equipment cap (Figure 2.2) is designed to disperse the gas
may be unavoidable. phase as fine bubbles in the liquid. It also prevents
liquid from flowing down through the gas passages
Gas— Liquid Contacting at low gas rates. Bubble caps are manufactured in
The between gases and
large difference in density many sizes and shapes.
liquidsmakes phase separation relatively easy. The
two phases may be mixed by bubbling the gas up
through the liquid. The gas bubbles separate from Plan view of top plate

the liquid into a continuous gas phase when they

reach the liquid surface. Stages are built in a verti-
cal column so that can flow by gravity
liquid
downward from The gas flows up-
stage to stage.
ward because of a pressure difference maintained
by a compressor or blower or by a boiler. Actual
stages in a vertical column are referred to as plates
or trays.
Many different plate types have been suggested;
only a few are considered here. For many years the
bubble-cap plate was the most widely used for Downcomer

s**+
Liquid flow

S^4

l¥ll

:£WW\

Figure 2.2. Exploded view of a bubble cap. Gas flow

The riser shown bottom of the picture
at the
is attached to the plate. Gas flows up Figure 2.3. Cross-flow bubble-cap plates for contacting two
through the riser into the cap and out of the fluid phases. The gas flow is indicated by the light arrows.
vertical slots, where it is dispersed in the The liquid flow is shown by heavy arrows. The bubble caps
liquid phase, which covers slots in the bub- disperse the gas in the liquid. They are designed to mini-
ble cap. (Vulcan Mfg. Co.) mize leakage of the liquid through the gas channels.
20 STAGE OPERATIONS

A typical flow pattern in a bubble-cap column

is shown Figure 2.3. Here the liquid flows across
in

the plate and into the downcomer to the plate

below. The gas flows upward through the bubble
caps into the liquid. Figure 2.3 is a simplified
schematic representation of a bubble-cap column.

Plan view of top plate

Figure 2.4. Bubble-cap plate with cross-flow. The sheet-

metal dams (or weirs) that run along the left and right sides
of the plate pictured maintain the liquid level high enough
to cover the slots in the bubble caps. The liquid flows
downward from the plate above to the left side of the plate
pictured. It then flows over the weir across the plate to the

weir and downcomer on the right side. The inlet weir

maintains a liquid level covering the lower end of the
downcomer, so that the gas cannot short circuit up through Downcomer
the downcomer. The plate shown is six feet in diameter and
is constructed of copper for use in a 20-plate distillation

column in a winery. (Courtesy Vulcan Mfg. Co.)

Figure 2.6. Disk-and-doughnut sieve trays

Figure 2.5. The view is of the
Sieve tray with cross-flow. for contacting two fluid phases. The light

underside of the tray, with the downcomer attached at the arrows indicate the gas flow, and the heavy
bottom. The grooves running from top to bottom are for arrows show the liquid flow. The gas is dis-
structural strength, as are the support struts running across persed in the liquid on each plate by passage
the plate. (Courtesy Vulcan Mfg. Co.) through the small holes in the plate.
 MASS-TRANSFER OPERATIONS 21

An actual bubble-cap plate is shown in Figure 2.4.
Bubble-cap columns can be operated over a wide
range of phase flow rates with high stage effici-
ency. Little liquid can leak through the gas pass-
ages, and the height of liquid on each tray can be
maintained at the desired level. As a result, gas-
liquid contact is good over a wide range of flow
rates, so relatively constant efficiencies are poss-
ible. Because of their relatively high cost, bubble-
cap trays have given way to newer types of trays,
described below.
Perforated plates have become very popular in
recent years because of their simplicity and low
cost. The most common type of perforated plate is

&«&&::

(a)

.y-.y/;.y:y;. y :yy.yy;s.y~

Figure 2.7. Valve caps — two types are shown. As gas flow is reduced, the caps
drop back to the surface of the tray to seal the opening against liquid flow,
(a) Model A-1. (b) Model V-1. (Courtesy Glitsch, Inc.)
22 STAGE OPERATIONS

the sieve tray, such as the cross-flow tray shown in rates. In addition, the caps direct the gas horizon-
Figure 2.5. Another flow pattern for sieve trays is tally into the liquid, thereby giving more thorough
shown in Figure 2.6. As the name implies, the sieve mixing than sieve trays, where the gas passes
tray is a metal sheet perforated with hundreds of straight upward through the liquid. Perforations in
round holes. The holes are usually 1/8 to 1/2 inch valve trays are often larger than in sieve trays — as
in diameter. The liquid flows across the plate and large as VA inches in diameter.
the gas bubbles up through the holes. At high gas A typical column with cross-flow valve trays is

rates, the gas flow prevents liquid from flowing shown Figure 2.8. Valve trays can operate over a
in

through the perforations, but at low gas rates wider range of flow rates than do sieve trays. Their
serious leaking may occur. This leaking, or weeping, cost between that for sieve trays and bubble-cap
is

may significantly reduce the stage efficiency. This trays. Stage efficiencies are high and nearly con-
is a limitation of sieve trays that must be carefully stant over a wide range of gas rates. Sieve trays and
considered in design. valve trays have replaced bubble-cap trays in many
A more recent development in perforated applications.
plates the valve tray. The perforations in a valve
is The design of gas— liquid contactors involves
tray are covered with liftable caps, such as those complex problems in fluid mechanics. References
shown in Figure 2.7. The caps are lifted as gas flows 7, 9, and 12 give detailed design procedures for
upward through the perforations, but they fall into multistage columns. Here only a brief summary of
place over the perforation as the gas rate decreases. the problems is given. The primary function of a
In this way they greatly reduce weeping at low gas tray is to give intimate contact between all of the
gas and all of the liquid. The space immediately
above the tray should be filled with a gas— liquid
mixture in violent agitation. There must then be
become disengaged.
space for the two phases to
The tray diameter is determined primarily by
the volume of the gas phase passing upward
through the tower. A high gas velocity through the
trays is undesirable because it yields a high gas-
phase pressure drop. A large pressure drop may
require greater distance between trays for the
hydraulic needed to give liquid downflow
head
from tray to tray. Gas-phase pressure drop must
have an especially low value in columns operating
under a vacuum.
The liquid flow across the tray must give a
sufficient depth to cover the gas ports adequately
at all gas rates. This may require complex flow
patterns with several inlets and outlets on large
trays. Figure 2.9 shows a valve tray with split flow,
and Figure 2.10 shows terracing on a 40-foot tray
necessary to maintain reasonably uniform liquid
depth. The liquid depth is governed by the liquid
flow rate and the exit weir height. The gas-phase
pressure drop also depends upon the height of
liquid above the gas ports. Therefore, the liquid
height must be adequate to cover the ports — but it
cannot be excessive, because it may cause undue
pressure drop.
Entrainment is the carryover of liquid droplets
in the gas phase to the plate above. Such carryover
is undesirable because it effectively lowers the
Figure 2.8. A cutaway view of a column stage efficiency. Entrainment occurs at high gas
equipped with cross-flow valve trays. (Cour- flow rates, so that the column diameter must be
tesy Glitsch, Inc.) adjusted to give a gas velocity (for a specified mass
 MASS-TRANSFER OPERATIONS 23

Figure Z9. A valve tray with a split-flow pattern. The liquid flows from the plate
above through downcomers to the two blank rectangular areas on the plate shown.
It then flows both ways to downcomers across the diameter and at each edge

(delineated by weirs). (Courtesy Glitsch, Inc.)

Figure 2.10. A terraced valve tray. This tray has such a large diameter (40 ft) that
it must be terraced to ensure uniform liquid depth. The tray section in the
foreground have not been installed. The downcomer to the tray below is at the
central diameter, beyond which three terraces of valve caps can be seen. The wire
mesh, partially installed in the right foreground, aids in removing liquid droplets
from the gas flowing up from below. Trays of this size are used in crude oil
distillation. (Courtesy, Glitsch, Inc.)
24 STAGE OPERATIONS

throughput) that will minimize entrainment. En- contacting. Either the light or heavy liquid may be
trainment lowers stage efficiency and therefore re- dispersed. To good dispersion at the sieve tray,
give
quires additional stages to yield the specified sepa- the liquid that does not wet the tray should be the
ration. Avoiding entrainment may require a column one dispersed. The liquid that preferentially wets
of greater diameter. Therefore, it is necessary to the tray should be the continuous phase. In this
strike an economic balance between number of way, the dispersed phase breaks away from the
trays and column diameter when determining to tray quickly to give small bubbles. A section of a
what extent entrainment should be eliminated. typical column with the phase dispersed is
light
Although a froth or foam on the plate gives shown in Figure 2.11a. This column resembles a
large gas— liquid surface and resultant high
area gas— liquid sieve-tray column, where the light liquid
rates of mass transfer, excessive foaming can fill is analogous to the gas. A column where the heavy

the space between trays with foam and give too phase is dispersed (Figure 2.116) is basically an
much entrainment. inverted light-phase-dispersed column. The down-
Flooding occurs when there is excessive entrain- comers pass the phase upward, and the heavy
light
ment or when there is excessive liquid backup in phase is dispersed below each tray. Perforations in
the downcomer. Flooding results when an attempt sieve trays range from 1/8 to 1/4 inch in diameter.
is made to pass too much liquid or too much gas The number of holes in a tray is determined by the
through the column. Avoidance of flooding is a total flow rate of dispersed phase. For adequate
primary factor in choosing a column diameter. dispersion, the liquid must have a significant veloc-
Tray spacing must be sufficient to give a clear ity as it passes through the holes. The height of
space above the liquid froth so that entrainment is liquid on the tray is determined by the pressure
minimized. In large columns, tray spacing must be needed to give the desired flow rate through the
at least 2 feet to permit maintenance people to perforations. For example, in Figure 2.116, the
crawl between the trays. A greater tray spacing height of heavy phase above the tray must be
yields a taller tower with attendant higher costs. sufficient to push the phase through the perfora-
Again, it is necessary to make an economic balance tions at the rate required to give good dispersion.
between tray spacing and entrainment. A smaller
tray spacing yields a shorter, cheaper column, but
may give entrainment that forces addition of trays, Light Light
liquid
height, and cost to the column. liquid
out out
In sieve trays, the size and number of holes must
t

be carefully selected. Larger holes give low pressure

drop, but poor gas dispersion. A large total hole Heavy
-
liquid Heavy
area gives low pressure drop but also low gas veloc-
liquid
ities that can lead to excessive weeping. -_„,-,—- T

Liquid— Liquid Contacting '-

In solvent extraction, two immiscible liquids must

be thoroughly mixed and then separated. In gas-
liquid contact, the large difference in density be-
*t
tween the two phases makes it possible to use the
force of gravity to promote flow and disperse the
gas in the liquid; in extraction, the difference in
4
density of the two liquid phases is small, so that *~ Light Light
r~ liquid liquid
gravity is much less effective for phase dispersion.
Bubble-cap trays and valve trays are not used for
liquid— liquid extraction because the gravity-
induced flow does not give adequate phase disper- *
I
sion. The lighter phase flowing up through the tray Heavy Heavy
liquid
is not divided into sufficiently fine bubbles as it liquid
out out
passes through the bubble caps or valve caps.
Sieve trays with downcomers are widely used Figure 2.11. Sieve tray columns for liquid-
for liquid— liquid contacting, although the column liquid extraction, (a) light phase dispersed;

construction is different from that for gas— liquid (b) heavy phase dispersed.
 MASS-TRANSFER OPERATIONS 25

V -phase out •* 1

*
/.-phase -»
Motor-driven
m

L-phase
impeller

in

(heavy phase)
~~~-^ji

V
II

^JL
nr>-V-phase

-\
in (light

V-phase
phase)
_-f 1
|
-\

V-phase
in
^~
1_ »-
|

V \ *
/.-phase ,

II

L-phase out L- phase out

(a) (b)

Figure 2.12. Mixer settlers for liquid— liquid extraction. The two entering phases
are thoroughly mixed by the impeller. The mixture flows into the settling tank,
where the two phases are allowed to separate under the influence of gravity,
(a) Single-stage mixer settler, (b) Multistage mixer settler arrangement.

The downcomer area is chosen to give a continuous stage to stage. Mixer— settlers are popular because
phase flow rate low enough to prevent dispersed- the thorough mixing leads to high stage efficiency
phase droplets from being carried through the (90—100 percent). However, the use of motor-
downcomer. Figure 2.1 shows two large multistage driven pumps and separate settling tanks increases
extraction columns. the cost per stage, when compared with sieve-trays.
Mechanical mixers are used to disperse the The design of liquid— liquid contacting equip-
phases more thoroughly than is possible in a sieve- ment is discussed in References 7 and 1. 1

tray column. After thorough mixing, the two-phase

mixture is passed to a settling tank where the Solid-Fluid Contacting
phases are allowed to separate by gravity settling. The contact of a granular solid with a liquid or gas
These units are called mixer— settlers. A schematic in continuous multistage system is complicated
a
diagram of a mixer— settler is shown in Figure by the problems involved in transporting the solid
2.12a. Mixer— settlers are used in series to give
multistage separation (Figure 2.126). Impeller shaft

The degree independently variable

of mixing is

by changing the speed and shape of the impeller.

Although thorough mixing is desirable, an emul-
sion that cannot be separated by gravity settling
may form if mixing is too vigorous. The size of the
settler tank is determined by the flow rates and the
rate at which the dispersed phase settles out. If the
phase densities are nearly equal, settling is very
slow, so that a centrifuge may be needed. The
centrifuge supplies a centrifugal force much greater
than the force of gravity. Emulsions can also be Settling end Mixed end Settling end

separated in centrifuges (see Chapter 22).

Several flow arrangements have been used in Figure 2.13. "Pump-mix" mixer-settler.
This end view shows three adjacent stages.
mixer— settlers. The stages can be arranged verti-
The two phases are mixed by a pump impel-
cally for gravity flow of both phases. A stepped
lerand then are allowed to separate at the
arrangement, such as that shown in Figure 2.126,
opposite end of the stage. To simplify the
may be used for gravity flow with units too large flow pattern, the impellers are at opposite
for vertical arrangement. Horizontal arrangements ends of adjacent stages. The impeller not
require pumping of one or both phases. The only mixes the phases, it also causes phase
"pump-mix" mixer— settler (Figure 2.13) uses the flow between stages. [Courtesy B. L. Coplan
impeller both to mix and to promote flow from et al., Chem. Eng. Prog., 50, p. 403 (1954).)
26 STAGE OPERATIONS

Intermediate^^/*
extract Air to atmosphere
Final extract 50-100 ppm CS 2
20 ppm H 2 S
Figure 2.14. Basket type of extractor. The
Bollman extractor is widely used in separ-
ating oil from oil-seeds (such as soybeans). Dust
collector
The ground seeds are fed to perforated bas-
kets that rotate on a continuous belt. The
baskets are dumped at the top of the cycle
just before they are refilled. The fresh sol-
vent flows countercurrent to the upward-
Carbon _
moving baskets on the left.The intermediate elevator
extract is collected at the bottom of the left
side and pumped to the top of the right side,
where it is sprayed onto the descending bas-
kets. Flow on the right is concurrent. The
final extract is filtered to remove solid part-

icles. An extractor of this type is used to

process 405 tons/day of soybeans. It con-

tains 38 baskets each 6.5 ft long, 2.3 ft wide,
and 1.8 ft deep (3).

phase from stage to stage. The solid may be moved

in baskets, as shown in Figure 2.14. The resulting
flow pattern is in stages, but countercurrent in
only half of the system. A finely divided solid may
be made to flow like a fluid under proper con-
ditions, as described in Chapter 22. This generally
involves some gas or liquid flow through the solid Carbon conveyor
 MASS-TRANSFER OPERATIONS 27

Figure 2.15. A five-stage countercurrent adsorption oper- 9. Smith, B. D., Design of Equilibrium Stage Processes,
ation. This adsorber recovers carbon disulfide from the McGraw-Hill, New York (1963).
exhaust air of a viscose plant. Viscose is made by treating 10. Sourirajan, S., Reverse Osmosis, Academic Press, New
cellulose with sodium hydroxide and carbon disulfide York (1970).
(CS2).The air in the plant contains 1000 parts per million 11. Treybal, R. E., Liquid Extraction, 2nd ed., McGraw-
CS2 by volume, and this concentration is reduced to 50— Hill, New York (1968).
100 ppm by adsorption on fluidized carbon. Process data 12. Van Winkle, M., Distillation, McGraw-Hill, New York
are given in the accompanying simplified flow dia- (1967).
gram. 13. Vromen, B. H., Eng. Chem., 54, No.
Ind. 6, p. 20
In the photograph, the tall black tank in the left (1962).
side of the framework is the carbon storage tank. The
large-diameter light-colored column below the top of the
PROBLEMS
framework is the adsorption section. Each of the stages is
38 ft in diameter, and is covered with 2—3 inches of
2.1. What is the primary objective when fluid phases
fluidized carbon. The equipment to the right of the carbon
are contacted in multistage equipment?
storage vessel uses high-temperature steam to strip the car-
2.2. Which of the following arrangements can be
bon from the carbon. The carbon disulfide is
disulfide
considered a stage? Describe any mass transfer that occurs.
recovered from the steam by condensing and then mechan-
ically separating the immiscible liquids. (a) Three streams of differing compositions are fed
The tall stack to the left of the framework is used to to a large tank where they are thoroughly mixed. The
exhaust the plant air when the adsorber is not in operation. resulting mixture splits into two phases that are withdrawn
A process flow sheet is given below. (Courtaulds, Ltd., separately from the tank.
Coventry, England.) (b) Water and solid salt are added separately to an
agitated tank. The salt dissolves completely, and the result-
ing solution is withdrawn at the same rate that water and
to "fluidize" it. The principle of fluidized solids salt are added.
has been applied to the design of a number of (c) Crushed soybeans are fed to an agitated tank. An
solid— liquid contactors. A large industrial adsorber organic solvent is also fed to the tank, where it dissolves oil
uses a gas-fluidized adsorbent on each of five stages from the beans. The oil solution flows from the top of the
(Figure 2.15). For large-scale liquid— solid sepa- tank, and the oil-free beans are removed at the bottom.
rations such as those occurring in ore processing, (d) Air bubbles up through a tank of hot water. The
continuous thickeners are used (Chapter 22). The air emerging from the surface of the water is saturated with
solid passes from the stage as a slurry with water. water vapor.
In concluding a discussion of multistage equip- 2.3. In your own words describe the difference be-
ment, it is necessary to reiterate that only staged tween gas absorption and distillation.
equipment has been discussed here. Many sepa- 2.4. What is the difference between gas absorption
ration processes may also use continuous-contact and gas adsorption?
equipment, which is discussed in Chapter 16. 2.5. Describe the equipment that might be used to
accomplish each of the following transfers of mass. Also
REFERENCES give the name of each unit operation involved.
(a) Oil is to be removed from crushed flax seed by
washing the seed with an organic solvent.
1. AI.Ch.E., Chem Engr. Prog. Symp. Series, No. 24
(1955).
(b) A solution of methyl alcohol and water is to be

2. Benedict, and T. H. Pigford, Nuclear Chemical

M., separated into relatively pure components by multistage

Engineering, McGrawHill, New York (1957). contact of liquid and vapor containing the components.
3. Kenyon, R. L, N. F. Kruse, S. P. Clark, Ind. Eng. (c) Benzene is removed from hydrogen by contacting
Chem, 40, p. 186 (1948). the gaseous mixture with a nonvolatile hydrocarbon sol-
4. Li, N. N., Recent Developments in Separation Science, vent.
CRC Press, Cleveland (1972). (d) Acetic acid is to be removed from an aqueous
5. Mintz, M. S., Ind. Eng. Chem., 55, No. 6, p. 18 (1963). solution by contacting the solution with an organic liquid
6. Nachod, F. C, and V. Schubert, eds., Ion Exchange
that preferentially dissolves the acetic acid. The organic
Technology, Academic Press, New York (1960).
liquid has nearly the same density as the aqueous solution.
7. Perry, R. H., and C. H. Chilton, Chemical Engineers
(e) Sulfur compounds are removed from lubricating
Handbook, McGrawHill, New York (1973).
oil by contacting the oil with liquid phenol.
8. Schoen, H. M. ed., New Chemical Engineering Sepa-
ration Techniques, Wiley Interscience, New York (f) Hydrogen sulfide is removed from water by con-
(1962). tacting the solution with warm air.
28 STAGE OPERATIONS

2.6. In recent years perforated plates have nearly 2.7. A perforated-plate distillation column operating
replaced bubble-cap plates in most distillation applications. at 200° F is in full operation when it is suddenly shut down.
To what would you attribute this major change? What What happens to the liquid on the plate? (Figure 2.6). What
possible disadvantages do you see in perforated plates, happens to the vapor between the plates?
compared to bubble-cap plates? What advantages?
 Three

Phase Relations

As discussed in Chapter 2, the choice of mass where

transfer operation to be used to make a separation = chemical potential; the subscript refers
/i
from a multicomponent mixture depends upon
to the component; the superscript refers
factors of operating feasibility, applications of
to the phase
physical-chemical theory, and economics. The re-

quired size of the operating unit, and hence its
and operability, can be calculated from a know-
ledge of the phase equilibrium limitations and the
rate of transfer obtainable. This chapter considers
the basic theoretical framework used in predicting
the composition of phases in equilibrium, presents
some of the empirical relations used to correlate
phase equilibrium data, and shows some typical
cost Gibbs defined chemical potential in terms of
fundamental thermodynamic functions (11). Quali-
tatively, it expresses the tendency of the com-
ponent to leave the phase or chemical state being
described. Thus, it is a measure of component
instability.
Workers later showed that the fugacity (de-
fined by the relation n, = RT \r\ f, + 6 where f, r

examples of the available data. From this sort of approaches p, asp, approaches zero, and where d is
information the design calculations can progress. a function of temperature alone) could be sub-
These calculations often require not only the stituted for the chemical potential; hence,
equilibrium compositions, but also the change in
the equilibrium values with changing temperature U' = U" = f, (3.2)
and pressure.
The fugacity has several advantages over the
chemical potential, not the least of which is that,
in concept at least, it can be related to measurable
GENERAL BASIS OF PHASE EQUILIBRIUM physical properties.
the number of variables in a system of
If
Before the phase relationships for specific systems
severalcomponents in equilibrium among several
are discussed, the general basis of phase equilibria
phases is related to the number of equations
will be summarized. The quantitative analysis of
relating them, as outlined above, the result is
phase equilibria is based upon the work of J.
Willard Gibbs, which is presented in all basic F=C+2-P (3.3)
thermodynamic texts (1, 11, 13). He showed that
equilibria between phases required a balance of all where
potentials, including not only P and T but also the
chemical potential of each component in each
F= number of intensive variables that can
be varied independently
phase. Thus, at constant P and T, the equilibrium
restriction for a single component is that C = number of components in the system
P = number of phases in the system

Mi =Mi = Mi (3.1 A phase may be defined as a physically distinct and

29
30 STAGE OPERATIONS

homogeneous portion of a system, either a solid, a where

liquid, or a gas. Several solid and liquid phases may
coexist; but, because gases are totally miscible with
ya = mole* fraction of a in the vapor

each other, there can be only one gas phase. An = ratio of the fugacity of a to the total
intensive variable is independent of the total
y—J
P a pressure
quantity of the phase. For example, the temper- P= total pressure
ature, pressure, and composition of a phase are
intensive variables. On the other hand, the total The liquid-phase fugacity will be equal to the
volume of the phase depends upon the quantity gas-phase fugacity at the vapor pressure of pure a,
and is therefore an extensive variable. Similarly, since equilibrium exists at that pressure. Consider-
the overall composition of a system of several ing that the mole fraction relates pure component
phases depends upon the extent of each phase and fugacity to the fugacity of the component in a
is an extensive variable. For phase equilibria, the mixture of "ideal" liquids, and defining an activity
number of components of a phase may be defined coefficient as the ratio of the fugacity of a in a
ya
as the least number of chemical species necessary real solution to that of a in an ideal solution, the
to prepare the phase. liquid fugacity is:
The phase rule is useful in predicting the L -
= Pa *a la (3.46)
number of intensive variables that may be varied
independently any system. For example, con-
in where
sider liquid water. Since it is pure, C= 1. With the
Pa = vapor pressure of a at the equilibrium
single liquid phase, P='\, and, by Equation 3.3,
temperature
F 2. It is therefore possible to vary both the
=
x a = mole fraction of a in the liquid
temperature and pressure of a single pure liquid
phase. Consideration of an equilibrium mixture of Combining Equations 3.2a, 3.4a, and 3.46, the
water and steam shows that C=1, P=2, and general vapor-liquid phase equilibrium relation is
F= 1. Thus, only one intensive property can be obtained.
varied independently. If the temperature is speci-
fied,the pressure is automatically set.
If the phase rule is applied to a vapor-liquid o. Py a =PaXaJa (3.4)

equilibrium mixture of ethanol and water, it is

found that C=2, P = 2, and F = 2. Thus, two Chemical engineering thermodynamicists have
variables be set. The two may be any
may devoted much attention to efforts to apply the
combination of the temperature, pressure, and knowledge available through physical chemistry to
phase concentration. For example, if the com- predict the fugacity coefficient (f/P) a and the
position and pressure of a phase are specified, the activity coefficient ya to fundamental molecular
temperature is set. Further application of the phase properties. This effort has been quite successful for
rule w'ill be useful in considering specific systems simple nonassociating components. Where liquid
discussed in the remainder of this chapter. phase nonidealities can be ignored, for example,
Calculations for distillation and gas absorp- with nonpolar components belonging to homol-
tion require knowledge of gas-liquid equilibrium. ogous series (light hydrocarbons are the best
Thus, Equation 3.2 becomes examples), ya = 1. The fugacity coefficient then
must be determined. Because this parameter is a
3 function of gas phase properties alone, it can be
fj - fJ- (3.2a)
determined from a gas phase equation of state.
Precise equations of state have been developed, so
The gas-phase fugacity could be replaced by the that this is done quite successfully (1, 13). A
partial pressure of the gas atthe pressure of particular resulting advantage is that vapor-liquid
interest the behavior is
if ideal at very low pressure. equilibrium can be expressed analytically and
It is convenient to put in a correction term relating written into computer programs.
actual gas behavior to this ideal behavior. Then

'The unit of molar quantity is the gram-mole in the

traditional metric system. In SI. the corresponding quantity \smole.

-©. PVa (3.4a) abbreviated mol. Strict adherence to

confusion with gram-mole and pound mole.
that abbreviation avoids
 PHASE RELATIONS 31

Where liquid- phase non idealities occur, it is where

necessary to determine the activity coefficient
from intermolecular force-field information or to
pa = the partial pressure of a in the vapor
H= Henry's law constant, experimentally
correlate experimental data empirically. Both
measured
approaches have been used. A combined approach
c a = concentration of a in the liquid
that has been successful at low temperatures has
been presented by Zellner (14). The theoretical In a binary (two-component) system, the
bases for this and for other recent attempts have component with the higher vapor pressure at a
been presented by Prausnitz (11). given temperature is referred to as the "more
If the gas phase behaves as an ideal gas, volatilecomponent," whereas that with the lower
(f/P) a = 1, and if the liquid phase is an ideal vapor pressure is called the "less volatile com-
solution, ya = 1. In this case Raoult's law results. ponent." By convention, the composition of a
binary mixture will be expressed as the concen-
Pv, = Pa*i (3.5) tration of the more volatile component.
For binary systems where a is the more
These equations indicate that the vapor evolved
volatile component and b is the less volatile
from a liquid mixture will be a mixture of the same
component, assuming Raoult's law is valid,
components as the liquid. The vapor will normally
be richer in the component having the higher vapor
(3.8)
pressure at the temperature of the vaporization. A *a Yb Pb
pure component will never be evolved from a
liquid mixture, although in the limiting case the
Since yb = 1 - ya and x b = 1 -x L

vapor pressure of one component may be so low as 1-x,

to make the component
Raoult's
practically nonvolatile.
law is accurate only in predicting
®[ 1-KaJ = <*ab (3.9)

vapor-liquid equilibrium for an ideal solution in Equation 3.9 defines the relative volatility (ocab of )

equilibrium with an ideal gas mixture. Solutions component a relative to b. For systems that do not
thatshow negligible deviation from ideality include follow Raoult's law, the relative volatility is de-
those whose components have similar structures fined as ocab = Ka /Kb The relative volatility is
.

and physical properties, such as benzene-toluene, constant when either Henry's or Raoult's law
propane-butane, and methanol-ethanol. Raoult's holds. In other cases it varies with composition.
law shows that the compositions in an equilibrium Figure 3.1 shows vapor-liquid composition data for
mixture depend upon the total pressure of the
system and upon the vapor pressures of the
components. The vapor pressures vary with
temperature but not with composition or total
pressure.
In engineering calculations, it is often con-
venient to express phase equilibrium by the
relation

- K~x. (3.6)

where Ka anis experimentally determined

constant. Kaoften only weakly dependent upon
is

concentration, even for systems that do not obey

Raoult's law; customarily called the "/C-value."
it is

In many
systems, the vapor pressure of the
solute in a very dilute solution can be used in a
Raoult's law formulation as a constant applicable 0.2 0.4 0.6 0.8 10
xa mole fraction more volatile component in liquid
over a significant range of concentrations. This is ,

the basis of Henry's law, which is most often Figure 3.1. Typical vapor-liquid equilibrium curves at
written: atmospheric pressure (5). (By permission of Amer.
Chem. Soc; copyright © 1949.) (9). (By permission
Pa = Hc a (3.7) of Amer. Inst, of Chem. Eng.; copyright © 1953.)
32 STAGE OPERATIONS

240 Since only butane and pentane are present in the

~ Superheated vapor
liquid and vapor, x B + x P = 1, and y
B + y P = 1.
There are now four equations and four unknowns.
:.-:
Solving,

\ ^v
-
yB +y P = } = 1.24x 5 + 0.39x^
:::
= 1.24x fi +0.39(1 -x fi ),etc.

f
x B = 0.72 y B = 0.89
: do
:

. :::
x P =0.28 / / = 0.11
,

\ p
(b) Experimental values of K at 300 kN/m 2
C

20c
and 40 C are for butane, KB = 1.2, and, for
\sr.-«>: V-- pentane, KP = 0.38 (see Appendix D-2). Therefore,
—
\ V
90
y B = 1.2xt and y P = 0.38 x P
* P = 0.76 Yb =0.91
\ x P = 0.24 y P = 0.09
:s:
Subcooied liquid

»:
The small difference between values in parts (a)
and (b) within the precision with which the
is

170 charts for vapor pressure and K can be read.

2 0.8 1.0
Therefore, mixtures of butane and pentane follow

:« fraction benzene in the vapor

Raoult's law at 3 atm pressure. In general, devi-
ations from Raoult's law are greater at higher
Figure 3.2. Temperature-composition diagram for liquid- pressures.
vapor equilibrium of benzene and toluene at atmospheric
pressure. Example 3.2. What is the relative volatility
of butane to pentane at 40°C and 465 psia?

two systems where the relative volatility is con- Solution. Assuming Raoult's law holds at
stant and also for two systems where the relative elevated pressures, from Equation 3.9, where
volatility varies with composition. The systems P B = 373 kN m2 and PP = 1 17 kN m2
shown in Figure 3.1 are at constant pressure, but
the temperature varies with composition. The 373 x 103
variation of temperatures is shown in Figures 3.2
^- =
Ti77^
= 3 18
-

and 3.3.
Using experimental values of K at 465 psia

Example 3.1. A mixture of butane and pen- (Appendix D-2), KB = 0.27, KP = 0.10
tane is at equilibrium at 300 kNm
2
pressure and
40~C. Calculate the compositions of the liquid 0.27
&B = 2.70
and vapor: (a) using Raoult's law, and (b) using 0.10
experimental values of Ka and Equation 3.6.
This indicates that at 465 psia, butane-pentane
Solution,Vapor pressures are found from
(a) mixtures deviate from Raoult's law by about 18
the table of physical data (see Appendix D-1 percent
C
Vapor pressure of pentane at 40 C =117 kN/m 2
;
Vapor pressure of butane at 40 C = 373 kN/m 2 A mixture does not boil at a single temper-
= 300 kN/m 2 ature for a constant total pressure, in contrast to
Total pressure P
the behavior of a pure liquid. The temperature at
373
For butane: yB = — 5
300
x B = 1.24x e
which a mixture of a given composition
liquid
begins to vaporize as the temperature is increased is
called the bubble point. Conversely, the temper-

For pentane: y P = —
Pp

P
Xp =
117
^rzx P
300
= 0.39x F
P
ature at which a vapor mixture first begins to
condense on cooling is called the dew point. For a
 PHASE RELATIONS 33

100

0.4 0.5 0.6 0.7 0.8 0.9 1.0

Mole fraction isopropanol

(a)

150

0.4 0.6 Oi
*a . >'a m °le fraction acetone

W
Figure 3.3. Temperature-composition diagram for mixtures forming an azeo-
trope at atmospheric pressure (a) Isopropanol-propylene chloride system at

mospheric pressure (5); (b) Acetone-chloroform (5). (By permission of Amer.

Chem. Soc; copyright © 1949.)

pure liquid, the bubble point and dew point are 0.2 x 10 5
identical and equal to the boiling point; a pure For benzene: y B = 5
xB = 0.2; xB =
1 x 10 Pb
component vaporizes or condenses at one temper-
ature. Dew points and bubble points of ideal 0.3 x 10 5
Pt
mixtures can be calculated from Raoult's law. For For toluene: y T = xT = 0.3; xT =
other mixtures, /C-values can be used. 1 x 10 5 Pj

Px 0.5 x 10 5
For xylene: yx = x x = 0.5;xx =
Example 3.3. Calculate the dew point of a 1 x 10 5 Px
gaseous mixture containing 20 mole percent ben-
zene, 30 mole percent toluene, and 50 mole At the correct temperature, x B + x T + x x = 1.0.

percent o-xylene at 100 kN/m 2 pressure. 5 5

0.2 x 10 5 0.3 x 10 0.5j<_10
= 1
Solution.At this moderate pressure for these P>
similar compounds, Raoult's law will be suffici-
The vapor pressures of the three
ently accurate. Now, if temperature is fixed, the vapor pressures

components depend upon temperature, which is to may be obtained from Appendix D-1, and the left
be determined. A trial-and-error procedure is neces- side of the equation can be calculated. If the left
sary, sincethe vapor pressures cannot be expressed side does not equal 1, a new temperature must be
as simple mathematical functions of temperature. tried.
34 STAGE OPERATIONS

Assume 7"=40°C: PB = 24.13 x 10 3 N/m 2 ;

If bubble point determination is to be made for a
a

P T = 7.58 x 10 3 N/m 2 ; Px = 2.96 x 10 3 N/m 2 .

ternary mixture that obeys Raoult's law, the
combination of the material balance equation with
0.2 x 10 5 0.30 x 10 5 Raoult's law and the Antoine equation gives
+
0.24 x 10 5 0.076 x 10 5 Xa [10) A » + B ' llC » +T) + Xb ^0) A b + B bUCb + T) +
5
0.5 x 10
= 21.67 > 1 Xc C\0) A c + B cHC c + T) _ p
0.0296 x 10 5
The vapor pressure constants are known, together
Therefore, the first trial T = 40°C is too low.
of with the initial composition and total
liquid
2
Assume T= 130°C: PB = 365.4 x 10 3 N/m ;
pressure, and this equation can therefore be solved
P T = 158.6 x 10 3 N/m 2
Px =72 .4 x 10 N/m
;
3 2 .
for the bubble-point temperature, the only un-

0.2 x 10 5 0.3x10 5 known. A systematic convergence procedure, such

+ as Newton's method, would be needed to complete
3.65 x10 5 1.586 x10 5
the iterative solution.
0.5 x 10 5
= 0.93 < 1
0.724 x 10 5
The dew point and bubble point of a binary
Therefore, the temperature than 130°C.
is less mixture are functions of its composition, as shown
3 2
Assume T= 128°C: PB = 351.6 x 10 N/m ;
in Figure 3.2 for the system benzene-toluene. The
2 3 2
P T = 151.7 x 10 3 N/m Px = 67.2 x 10 N/m
; .
curves are calculated from vapor-pressure data,

0.2 x 10

3.516 x 10 5
5
+
0.3 x 10 5

1.517 x 10 5
+ —0.5 x 10 5
0.672 x 10 5
— = 1 .00
assuming Raoult's law is valid for this system. The
dew point and bubble point become identical at
x a<Va~§ ar, d x a ,y a ~ 1-0, because these com-
positions represent pure toluene and pure benzene,
The check 128°C is sufficiently close to consider
at respectively. The single values of temperatures at
this the dew
point of the mixture. these compositions are the boiling points of the
At the dew point, the first liquid formed has a pure components.
composition such that the pressure exerted by each If a liquid mixture containing 0.4 mole
component of the liquid is equal to the partial fraction benzene is heated (as represented by
pressure of the component in the vapor. The moving up the vertical line on Figure 3.2), it will
composition of the first equilibrium liquid formed start to vaporize at its bubble point (points),
as condensation begins at 128°C is which is 203.5° F. The first equilibrium vapor
5 formed is also at 203.5°F, and its composition (b)
0.2 x 10
= 0.057 is 0.625 mole fraction benzene. As the temperature
*B "
3.516 x 10 5 is raised (from a to c), more liquid evaporates until

0.3 x 10 5 the last liquid vaporizes at 215°F {d). During the

Xr " = 0.198 vaporization process, the liquid composition varies
1.517x10 5 from 0.4 to 0.215 and the vapor from 0.625 to 0.4
5 mole fraction benzene, but the overall composition
0.5 x 10
= 0.744
*x E of the liquid-vapor mixture remains constant at 0.4
0.672 x 10
mole fraction benzene.
Phase equilibrium calculations, such as those of The region below the bubble-point curve
Example 3.3, can be easily done on a digital represents the liquid phase. A liquid at a temper-
computer. However, the phase equilibrium expres- ature below bubble point may be referred to as
its

sion must be supplied in an analytical form. If a subcooled liquid, whereas one at its bubble point
Raoult's law can be used, the vapor pressure is called a saturated liquid. Similarly, a vapor at its

relation must be so stated. The Antoine equation dew point is called a saturated vapor. Between the
has been found to fit most vapor pressure data. bubble-point and dew-point curves is a region of
log Pa =A a +B a /(Ca + T) (3.10) two phases where a saturated liquid and a saturated
vapor coexist. The relative quantities of liquid and
where vapor in the two-phase region can be calculated by
A a Ba
, , and Ca = empirical constants. (See a material balance (see Chapter 4). Above the

Appendix D-2e for values dew-point curve lies the region of superheated
for some common liquids.) vapors.

Nonideal solutions may have temperature-
composition curves substantially different from
those of Figure 3.2. For example, Figure 3.3
presents two nonideal systems, isopropanol-
propylene chloride and acetone-chloroform. Each
system forms an azeotrope — a mixture with an
equilibrium vapor of the same composition as the
liquid. At the azeotropic composition, the dew
point and bubble point are equal, and the mixture
vaporizes at a single temperature. This conclusion
can be reached by applying the phase rule. In a
binary liquid— vapor system there can be two
independent variables. Usually the pressure is
specified and so there remains one variable. The
definition of the azeotrope sets the compositions
of the two phases as equal to each other, and this
restriction eliminates the final independent vari-
able. Therefore, the temperature is set at an unique
value. For example, in the isopropanol-propylene
chloride system, the equilibrium liquid and vapor
compositions become equal at 0.68 mole fraction
isopropanol. Since the dew point and bubble point
are also equal at this composition, the mixture will
vaporize at a single temperature, as does a pure
liquid. For this reason, azeotropes are often called
constant-boiling mixtures. This azeotrope boils at a
temperature lower than the boiling point of either
pure isopropanol or pure propylene chloride.
Systems of this type are referred to as minimum-
boiling mixtures. On the other hand, the system
acetone-chloroform has a maximum-boiling azeo-
trope at 0.34 mole fraction acetone (Figure 3.3/?).
The presentation of x, y, T information at a
constant pressure, as in Figure 3.2, is frequently a
useful choice; however, in phase-relation material
balance computations done at a specific pressure,
the temperature may not be specifiable. For these
calculations, a plot of x versus y as in Figure 3.1 is
much more useful. If necessary a T versus x or y
locus can be added.
In distillationand other mass-transfer process
calculations, energy balances as well as mass
balances are required. Thus, a measure of the
system energy is needed. This is usually the
enthalpy, defined as the internal energy plus the
PV product (see Chapter 20). Enthalpy-concentra-
tion data are given for ethanol— water in Figure 3.4
and for ammonia— water and nitrogen— oxygen in
Appendix D. The molar enthalpy is plotted on the
vertical axis and the mole fraction of the more
volatile component on the horizontal axis. Any
point on the diagram defines both the enthalpy and
the composition of a binary mixture. Similar plots
can be made using mass-fraction compositions.
PHASE RELATIONS 35
The enthalpy of a binary mixture is relative to
some arbitrary reference condition chosen for the
pure components. For example, in Figure 3.4 the
enthalpies of pure liquid ethanol at 32° F and pure
liquid water at 32°F were chosen equal to zero.
Mixtures of ethanol and water may have enthalpies
different from zero at 32° F because of enthalpy
changes on solution and dilution. The changes are
small for the ethanol— water system. Liquid mix-
tures at temperatures other than 32° F will have
enthalpies different from zero because of the enth-
alpy change on heating, as determined by the heat
capacity and the temperature change. If a mixture
is vaporized, its enthalpy will increase by an
amount equal to the latent heat of vaporization.
All enthalpy changes are considered in constructing
an enthalpy-composition diagram for a binary sys-
tem using experimental data on heat capacities,
heats of solution, and heats of vaporization.
On Figure 3.4, the curve labeled H represents
vapors at the dew point (saturated vapors), and the
curve labeled h represents liquids at the bubble
point (saturated liquids). The vertical distance
between the curves is therefore the latent heat of
vaporization of a mixture. Between the two curves
is a liquid-vapor two-phase region. To determine
points on the saturated vapor and liquid curves,
which are in equilibrium with each other, the
equilibrium curve on the lower part of the diagram
is used. A value of the vapor composition y a may
be chosen on the vapor enthalpy curve (H). From
the equilibrium curve, the value of x a in equili-
brium with the chosen y a may be determined. The
value of x a may then be plotted on the liquid
enthalpy curve (h). A tie line then can be drawn
connecting the two points representing composi-
tions that are in equilibrium. The temperature-
labeled lines in the two-phase region are tie lines.
If the equilibrium temperature is given, the
vapor and liquid compositions are obtained from a
temperature-composition plot, such as Figure 3.3a
for the C3H7OH— 3 H 6 CI 2 system. The region be-
low the saturated- liquid curve represents subcooled
liquids. Lines of constant temperature (isotherms)
are given in this region. In the ethanol-water sys-
tem, the heats of solution and dilution are small, so
that the isotherms are nearly straight lines, indi-
cating constant enthalpy. Above the saturated-
vapor curve is the region of superheated vapors.
Typically, increasing the pressure increases
the temperature at which vapor-liquid equilibria
occur, and also decreases the sharpness of separa-
tion between vapor and liquid compositions. This
is illustrated for the ethanol-water system in Figure
36 STAGE OPERATIONS

c: 4 0.6 : B

i„ , mole fraction ettianol in liquid

Figure 3.4. Enthalpy-composition diagram for ethanol-water mixtures at 1 atm

pressure (3). Reference state: liquid water at 32° F, liquid ethanol at 32 . (By
permission of John Wiley; copyright © 1950.)

-
 PHASi: RLLATIONS 37

3.5. The vapor composition in equilibrium with a

liquid containing 0.2 mole fraction ethanol would
contain 0.53 mole fraction ethanol if the system
pressure were atmospheric, but would be 0.42
mole fraction ethanol at a system pressure of 300
psia. Also, the composition of the azeotrope shifts,
becoming richer in ethanol as the pressure de-
creases.

Example3.4. What is the composition of a

vapor equilibrium with a saturated liquid of 0.5
in

mole fraction ethanol, 0.5 mole fraction water?

Solution. On Figure 3.6, a vertical line is

0.2 04 06 0.8 drawn downward from the liquid enthalpy curve at

Mole fraction ethanol in liquid x a = 0.5. The equilibrium value of y a is determined
by the intersection of the vertical with the equili-
Figure 3.5. Vapor-liquid equilibrium for ethyl
brium curve and may be read from the right-hand
alcohol-water system at various constant pressures
(9). (By permission of Amer. Inst, of Chem. Eng.; y a = 0.66. To transpose this value to
vertical axis as

copyright ©1953.)
the horizontal axis, a horizontal is drawn from the
equilibrium curve to the diagonal, and then a
vertical is drawn to the saturated-vapor enthalpy
curve. The value on the H curve may be connected
ya '- = 66 with the original value on the h curve by a tie line.
H 1

The temperature of the equilibrium mixture can be

/ /
20.000
/ / /
determined by reference to Figure 3.4. It is approxi-
mole mately 176°F.
"o o o
-
it,
Btu/lb

•
<t
/
/ <rj/
c/ /
/ co Example. 3.5. Determine the compositions
~ c\i
vapor.

"o o o
7 £>/
°Ol
y
<"/
*~/ /
/
/ of the phases present under the following condi-
tions:
o

liquid
1 /
ti- (a)Ethanol— water at 177°F and 1 atm. The
/">
bulk concentration is 50 percent ethanol.
/
/ /^ _
of oo o
/ Xa = o5a (b) Ammonia— water at 100 psia, x a = 0.3.
£ h .
\r
140*
(c) Ethanol— water at 100°F and 1 atm.
- F
JS 180 F-* '
. :

c
(d) Ammonia-water at 100 psia, H = 10,000,
60° F

*^- 20* F 1.0

overall mole fraction NH 3 = 0.5.

-c

,
,
y^ (e) Ammonia-water at 100 psia, H = 20,000,
-5,000 1
0.8 overall mole fraction ammonia = 0.4.

Solutions,(a) The compositions are most

ya = 0.66 easily determined from Figure 3.4. The tie-line at
3.6
177°F intersects the bubble-point and dew-point
lines at the equilibrium compositions x a = 0.42,

04 Ya = 0.63.
(b) From the equilibrium curve in Appendix
D atx a = 0.3, y a = 0.93.
0.2 >> (c) Since the composition is not given, the
mixture cannot be located exactly on Figure 3.4.
-

However, at 100°F a/1 mixtures of ethanol and

water are subcooled liquids, as indicated by the
0.2 0.4 0.6 0.8 1.0
100°F isotherm. Therefore, only a liquid phase will
,mole fraction ethanol in liquid
be present at 100°F and 1 atm, and there is no
jr a

Figure 3.6. Solution to Example 3.4. equilibrium vapor.

38 STAGE OPERATIONS

id) The point (H = 10,000 Btu/lb mole, for the solute. Where the equations are inaccurate,
z a = 0.5) is located on the diagram in Appendix D. data may be tabulated or plotted on an equilibrium
Since the point lies in the two-phase region, there diagram. For example, in the absorption of ammo-
will be a liquid and vapor in equilibrium. It is nia from an air-ammonia mixture into water, the
necessary to determine a tie line through the point. slight solubility of air may be neglected, and the
A first approximation is made by drawing a tie line concentration of ammonia in
the liquid may be
through the point with a slope intermediate be- tabulated as a function of the partial pressure of
tween those of the 250 and 200° F isotherms. The ammonia in the gas (Appendix D).
values of x a and y a obtained by the intersections
of the tie line with the enthalpy curves are checked
UQUID-LIQUID EQUILIBRIA
on the lower part of the equilibrium diagram. The
correct values are x a = 0. 1 7, y a = 0.74. In liquid extraction, one component of a solution
(e) The point (H = 20,000 Btu/lb mole, is transferred to another liquid phase that is

z a = 0.4) is located on the diagram in Appendix D. relatively insoluble in the first solution. In the
Since the point is in the superheated-vapor region, most simple case, the solute is partitioned between
there will be only a vapor phase of composition two insoluble liquid phases. Equilibrium data for
V a = 0.4. this case can be recorded as weight ratios of solute
to solvent in each of the phases at equilibrium. An
Gas-absorption calculations often require data example is the partition of uranyl nitrate between
on equilibrium between the gas dissolved in a nitric acid solution and an organic solvent, as
liquid phase and the gas phase. The equilibria often shown in Figure 3.7.

can be expressed by Henry's law or Raoult's law In many cases, the two solvents are partially

05
c 'hno,= 3.0 20 0.5
^^-~
0.0

0.4

03

11//
11/'/
7 / //
/
If
O
\ 02
o /

c o.i
/
tl / /

0.1 0.2 0.3 0.4 5 H6 07

Concentration ot I'CMN 1
""
1
) in agueoub phase, g moles/ liter HO
Figure 3.7. Distribution of uranyl nitrate between aqueous nitric acid solutions and
an organic solvent at 25°C (4). Organic solvent: 30 volume percent tributyl
phosphate in kerosene Chno = 9 moles nitric acid per liter of water. (By
permission of Amer. Chem. Soc; copyright © 1958.)
 PHASE RELATIONS 39

0.4 0.6 10
xa y
, , mass fraction acetic acid

0.2 0.4 0.6 10

mass fraction acetic acid in water layer

Figure 3.8. Isopropyl ether-acetic acid-water system at 20°Cand atmospheric pres-

sure (8). (By permission of Amer. Chem. Soc; copyright © 1951 .)
40 STAGE OPERATIONS

soluble each other; in addition, the concentra-

in of unextracted raffinate component (x b or y b )

tion of the solute may influence the mutual
solubility of the solvents. In such cases, equilibrium
data must be plotted on a three-component
diagram. Application of the phase rule to two-
phase, three-component systems shows that there
can be three independent variables. In this case,
there are four possible variables: temperature,
pressure, and the concentrations of two compo-
nents. The concentration of the third component is
not independent because it is set by stipulating the
other two. But the concentrations of the two
components are not entirely independent because
need not be plotted. The diagonal side of the
triangle represents mixtures where x b = or
yb = 0. The coordinate axis for component b
actually is a perpendicular to the diagonal side of
the triangle from the right angle. Since the third
coordinate axis is not used, it is omitted.
Any point within the triangle represents a
three-component mixture. Certain overall composi-
tions exist as one liquid phase, whereas others may
split into two liquid phases. The curve nopqr
(Figure 3.8) separates the two-phase region (to the
left) from the one-phase region (to the right) and is

it was stipulated that the two phases are in
equilibrium. Therefore, if temperature, pressure,
and one concentration are specified, the other two
concentrations are set.
Data for ternary systems are often reported
on equilateral-triangular diagrams; however, for
engineering calculations a right triangle is more
convenient for reporting the data, since ordinary
rectangular-coordinate graph paper may be used in
constructing the diagram. If the calculation in-
volves only relatively low solute concentrations,
only that part of the diagram should be plotted,
with an expanded horizontal scale for greater
graphical accuracy.
The ternary system isopropyl
acid-water (Figure 3.8) is representative of systems
called the phase envelope. Compositions of phases
in equilibrium must lie on the phase envelope. A
mixture on the lower part of the envelope (such as
q) contains a high concentration of the un-
extracted component and is called the raffinate
phase. A mixture on the upper part of the envelope
(such as o) is rich in the solvent and is referred to
as the extract. Compositions of phases which are in
equilibrium may be connected by a tie line, such as
o—q. Alternatively, equilibrium compositions may
be determined by use of the equilibrium curve on
the lower part of the diagram. At the plait point
(p) the tie line has been reduced to a point where
the extract and raffinate are identical. Note that
ether-acetic the equilibrium curve ends at the plait point at
xa =y a -

in which one pair of liquids is partially miscible The system diethylene glycol-styrene-ethyl-
and the other two pairs are completely miscible. In benzene is representative of systems in which two
Figure 3.8 each vertex of the triangle represents a of the three pairs of liquids are partially soluble
pure component: the right angle, pure water; the (Figure 3.9). In this system the extract and
upper vertex, pure isopropyl ether; and the right- raffinate phases are represented by independent
hand vertex, pure acetic acid. In general, the right curves, and there is no plait point. The line for the

angle represents a pure phase consisting of the extract phase on the diagram is inconveniently
major component of the raffinate that is not short for stage calculations. This situation can be
extracted (b)* the upper vertex, pure solvent used remedied by replotting the data on a solvent-free
in extraction (c); and the right-hand vertex, pure basis (Figure 3.10). On the horizontal axis is

solute (a) to be extracted from the raffinate by the plotted

solvent. The horizontal boundary of the triangle
mass of solute
represents mixtures of solute and unextracted Xa i v a, ~r 7~
raffinate component, with no solvent present. The mass of solute
vertical boundary represents binary mixtures of the + unextracted raffinate component
unextracted raffinate component and solvent.
On the vertical axis is plotted
Since x a + x + x c = 1 it is necessary to plot the
,

concentrations of only two components. The third mass of solvent

can always be found by difference. On the right- c '
mass of solute
triangular diagram, a mixture is most easily plotted + unextracted raffinate component
by the mass-fraction solute (x a or y a and the )

mass-fraction solvent (x c or y c ). The mass fraction The denominator of each of these quantities is the
mixture to be separated, one component identified
as the solute and the other as the component that
'Hereafter called the unextracted raffinate component will not be extracted to any significant extent. The
 PHASE RELATIONS 41

1.0 -

: i
^»

o
at

| 0.6

—
•
a
c

-
N

\
\ \^

02 0.4 0.6 0.8 1.0

. yu • m ass fraction styrene

*02

r . -ass •'-•action styrene in ethylbenzene layer

Figure 3.9. Diethylene glycol-styrene-ethylbenzene system at 25°C and atmospheric

pressure (2). (By permission of Amer. Chem. Soc.; copyright © 1951.)
42 STAGE OPERATIONS

mass solute
30 - Ya =
mass of solute + unextracted component

0.10
- = 0.59
60 -
1 a>|
.^- Extract phase
0.10 + 0.07
1 - /

- -
1 */
*~(
mass solvent
1

— Yr =
40 mass of solute + unextracted component
l/> O)

E >, 0.83
= = 4.88
20 - - 0.10 + 0.07

^Raffinate phase
- The point (Va =0.59, Yc = 4.88) may then be
plotted as a point on the extract curve of Figure
( D 0.2 0.4 0.6 08 1 3.10. Additional points may be calculated in this
manner, and the extract curve may then be
'

,
plotted. On the raffinate curve of Figure 3.9 at
1
xa = 0.8, xc = 0.022 and x b = 0.178. Therefore,

0.8 0.022
0.8 - Xa = = 0.817; Xr = = 0.0225
0.8 + 0.178 0.8 + 0.178
/A* This point may be plotted on the raffinate curve of
0.6 - Figure 3.10. Additional points on the raffinate
curve may be plotted in the same way. Since the
solvent concentration of the raffinate is low, the
ro 0.4 curve nearly coincides with c
= X
for this system.
The two parts of Figure 3.9 require two parts
as in Figure 3.10 for representation on a solvent-
- - free basis. The concentration of the solute in the
0.2
//
nonsolvent material is the horizontal axis of each
part of the diagram. The vertical axis of one part is
the concentration of solute in the extract, and the
0.2 0.4 0.6 0.8 1.0
axis of the other is the amount of solvent per unit
mass ratio in raffmate
of nonsolvent material.
Figure 3.10. Diethylene glycol-styrene-ethylbenzene The
parallels of solvent in liquid-liquid extrac-
system at 25° Con solvent-free basis. tion,of solid adsorbent in adsorptive separation,
and of enthalpy in distillation will be explored in
basis determination of these composition
for
Chapter 5, and the differences and similarities will
measures thus solvent-free. This nonsolvent
is
be indicated.
material can be characterized in terms of the
The propane-oleic acid-cottonseed
oil system

concentration of one component — logically the

(Figure demonstrates how the solubility
3.11)
characteristics of a system may vary with temper-
solute done for Xa and Ya Any amount of
— as is .

ature. In this case, an increase in temperature

solvent be added per unit of nonsolvent
can
material. With these definitions, Xa + Xb = 1 and
decreases mutual solubilities, contrary to the more

Xc may have any value. Similarly, Ya + Yb = 1, common case where an increase in temperature
and Yc may have any value. For example, for pure causes an increase in solubility. Systems that form

solvent Yc is infinite.
two phases at lower temperatures may be com-
pletely miscible at a higher temperature. Figure
3.1 1 also demonstrates the use of a partial triangle
Example 3.6. Calculate the data of Figure and a change of scale, which are often desirable for
3.10 from that given in Figure 3.9. accuracy in graphic calculations. None of the data
Solution. On the extract curve of Figure 3.9, for the system is beyond x a = 0.5; the diagram
at Ka =0.10, yc =0.83. Therefore, y b = 0.07. beyond this point is thus eliminated and the
Basis: 1 lb of this extract. remaining diagram is increased in scale.
 PHASE RELATIONS 43

0.2 0.3 04 0.5

xa .
ya . mass fraction oleic acid

0.20

\ 8b°C
2
\3420 kN/m

0.15 / \

/
/

1/
y 0.10 /
/

^s*^
0.05
^-^98.5 °C
/
^<^ 4310 kN/m
2

/ ^^
0.1 0.2 0.3 0.4 0.5
ia , mass fraction oleic acid in cottonseed oil layer

Figure 3.11. Ternary equilibrium diagram for propane-oleic acid-cottonseed oil (6).

(By permission of Amer. Inst. Chem. Eng.; copyright © 1942.)

44 STAGE OPERATIONS

800

i+

0.6 0.8 I ]

f . ya < lc
|
m oles
propane/lb mole
proppne and propylene

0.02 0.04 0.06 0.08 0.10 0.12

Xa , lb adsorbate/lb silica gel

Figure 3.12. Adsorption isotherms of hydrocarbons

of silica gel (7). (By permission of Amer. Chem. Soc.;
copyright ©1950.)

SOLID-GAS EQUILIBRIA
Adsorption, the commonest form of gas-solid
equilibrium, has been observed to occur by both
physical and chemical mechanisms. Physical 2 4 6 :o
A" lb moles propane/ lb mole propane
adsorption occurs when the intermolecular forces
.

+ propylene in adsorbate
of attraction between the fluid molecules and the
Figure 3.13. Simultaneous adsorption of propane
solid surface are greater than the attractive forces
and propylene on silica gel at 25°C (7). (By per-
between molecules of the fluid itself. The mole-
mission of Amer. Chem. Soc; copyright © 1950.)
cules of fluid adhere to the surface of the solid
adsorbent, and an equilibrium is established be-
tween the adsorbed fluid and that remaining in the
fluid phase. (L) has values of X c greater than zero, in contrast
Experimental physical adsorption isotherms to an analogous diagram for liquid extraction, such
for a number of pure hydrocarbon vapors on silica as Figure 3.9.
gel are given in Figure 3.12. Adsorption of a Data for physical adsorption can often be
mixture of gases results in an adsorbate of com- expressed by an empirical equation. Common
position different from that in the gas. Therefore, equations for adsorption isotherms include
it is possible to separate gaseous mixtures by Vn
selective adsorption. The equilibrium compositions
Fruendlich: X=k y p (3.11)

of gas phase and adsorbate for propane-propylene k2P

mixtures are given in Figure 3.13. This equilibrium Langmuir: X= (3.12)
1 + k3p
diagram closely resembles those for distillation
= k 2 p for small values of p
(Figure 3.4) and liquid extraction (Figure 3.10).
An analogy can be drawn among the adsorbent in where
adsorption, the enthalpy in distillation, and the
weight of adsorbate
solvent in extraction. It should be noted that the X=
gas phase (V) occurs at Yc = 0, and the solid phase unit weight of adsorbent
 PHASE RELATIONS 45

p = partial pressure of adsorbed gas in

gas phase

n, ky,k2,kz ~ empirical constants

Chemical adsorption, or chemisorption, in-

volves chemical interaction between the adsorbed

fluid and the adsorbent solid. In many cases the
adsorption is irreversible, and it is difficult to
separate the adsorbate from the adsorbent. For this
reason only physical adsorption lends itself to
continuous staged operation.
Adsorption is employed to process liquids as
well as gases. Common adsorbents include silica
gel, charcoal, alumina, synthetic zeolites, and
various clays.
0.2 0.4 0.6 0.8 1 o
An
even simpler form of gas-solid equilibrium x a ,y a , mass fraction solute

occurs on the sublimation of a solid into a gas

Figure 3.14. Underflow compositions for solid-
phase, or the direct crystallization of a solid from a
liquid extraction.
gas. Here the concept of vapor pressure is used in
mass adhering solvent
exactly the forms applied in Equations 3.4—3.7. Curve 1 = Constant ratio of
Usually, however, only one component is crystal- mass inerts

lizing subliming,
or and hence x=1.0 for the mass adhering solution
=
Curve 2 = Constant ratio of 1
condensed phase. mass inerts
exchange is another solid-liquid oper-
Ion Curve 3 = Experimentally determined curve for
ation, but one that resembles chemical adsorption. halibut livers and ether 1 2). (

The solid ion-exchange resin is manufactured to (Courtesy Amer. Chem. Soc.)

include an ion that will be replaced by a particular
component ion of the fluid phase. A common
underflow may be considered a mixture of inert
example of ion exchange is the "zeolite" water
solids and solution. The right-angle vertex repre-
softener, where sodium ions in the resin are
sents pure inert solids, and the hypotenuse of the
exchanged for calcium ions in the water. Industrial
ternary diagram represents clear solutions. Any line
ion-exchange operations are batch or continuous-
joining a clear-solution composition on the hypo-
contact instead of staged, so they will be con-
tenuse with the right angle intersects the underflow
sidered in a later chapter.
locus at the underflow composition in equilibrium
with the clear solution. Therefore, the portion of
the line from the hypotenuse to the underflow
SOLID-LIQUID EQUILIBRIA
locus is an equilibrium tie line. The underflow
In solid-liquid extraction, a liquid solvent is used locus determines the relative quantity of solution
to dissolve a soluble solid from
an insoluble adhering to the inert solids. The data of Figure
solid. Some of the resultant solution adheres to the 3.13 are representative of underflow loci. When
insoluble residue. An "equilibrium stage" can be experimental data are available, they should be
defined as one where the liquid adhering to the used. Where data are limited, curves such as 1 and
solids leaving has the same composition as the 2 may be used as approximations.
liquid-extract phase leaving. It is then necessary to
determine experimentally the mass of adhering
REFERENCES
liquid per unit mass of insoluble solids. Such data
can be plotted on a ternary diagram, as shown in
1. Balzhiser,R. E., M. R. Samuels, and J. D. Eliassen,
Figure 3.14. The resulting curves are referred to as
Chemical Engineering Thermodynamics, Prentice Hall,
underflow loci and do not represent true equi-
Englewood Cliffs (1972).
librium conditions. Included are curves for con- 2. Boobar, M. G., et al., Ind. Eng. Chem., 43, p. 2922
stant ratios of solvent to inerts and solution to (1951).
inerts and an experimentally determined curve for 3. Brown, G. G., et al., Unit Operations, John Wiley &
halibut livers with ether as the solvent. The Sons, New York (1950).
46 STAGE OPERATIONS

4. Codding. J. W., W. O. Haas, and F. K. Hermann, Ind. (d) The liquid and vapor phases for a mixture of
Eng. Chem., 50, 145 (1958). p. ethanol and water with an overall composition of 60 mole
5. Fordyce, C. R., and D. R. Simonson, Ind. Eng. Chem., percent ethanol at 175 F and atmospheric pressure.
41, p. 104(1949).
3.4. Choose the two correct words (one from each
6. Hixson, A. W., and J. B. Bocklemann, Trans. AI.Ch.E.,
pair), and explain your choice.
38, p. 891 (1942).
(a) "For a given liquid mixture, the dew point is
7. Lewis, W. K., et al., J. Am. Chem. Soc, 72, p. 1153
(usually/always) (higher/lower) than the bubble point."
(1950).
(b) "In an ideal solution a relative volatility (aa _
8. Othmer, D. F., R. E White, and E. Trueger, Ind. Eng. 6 )

Chem., 33, p. 1240(1951). greater means pure component a boils

than 1.0 at a

9. Otsuki, H., and F. C. Williams, Chem. Eng. Progr. (higher/lower) temperature than pure component b."
Symposium Ser., Vol. 49, No. 6, 55 (1953). (c) "Considering an ideal binary mixture, the 'more
10. Perry, J. H., ed. Chemical Engineers' Handbook, 5th volatile component' boils at a (higher/lower) temperature
ed., McGraw-Hill, New York (1974). than the less volatile component, when each is in its pure
11. Prausnitz, John M. r
Molecular Thermodynamics of state."
Fluid-Phase Equilibria, Prentice Hall, Englewood Cliffs
3.5. A vapor containing 35 mole percent propane
(1969).
and 65 mole percent ethane is being compressed at 0°C.
12. Ravenscroft, E. A., Ind. Eng. Chem., 28, p. 851 (1936).
C VanNess, Introduction (a) What pressure must be applied to condense
13. Smith, J. M., and H. to
Chemical Engineering Thermodynamics, 3rd ed., one-half of the vapor (on a molar basis)?

McGrawHill, New York (1975). (b) What is the composition of the liquid formed in

14. Zellner, M. G., L C Claitor, J. M. Prausnitz, Ind. Eng. (a) if the final liquid and vapor are at equilibrium?
Chem. Fundamentals, 9, pp. 549-563 (1970). (c) At what pressure would the last bubble of vapor
disappear?
3.6. A liquid mixture of 25 mole percent benzene,
75 mole percent toluene is heated at constant pressure of
PROBLEMS 1 atm to a final temperature of 21 7° F. At this temperature
and pressure, a two-phase mixture results. What is the
3.1. Calculate the dew point and the bubble point of composition of each phase, and how much of the initial

the following mixture at a total pressure of 300 kN/m material was vaporized?
3.7. A liquid mixture containing 30 mole percent
n- butane 35 mole percent
isopropanol and 70 mole percent propylene chloride is
n-pentane 35 mole percent
heated in a closed container to a final temperature of 83°C
n-hexane 30 mole percent
at atmospheric pressure. What is the composition of each of
What is the composition of liquid and vapor phases, and the the phases at the final equilibrium temperature, and what

temperature, when half the total moles are liquid? fraction of the original liquid is vaporized?
3.2. An equilibrium mixture of liquid and vapor has an 3.8. Calculate the relative volatility of acetone to
overall composition of 0.35 mole fraction benzene and 0.65 chloroform at atmospheric pressure for liquid-phase com-
mole fraction toluene at 14.7 psia and 210°F. positions containing 0.1 to 0.9 mole fraction acetone at

(a) Using vapor-pressure data and assuming Raoult's intervals of 0.1 mole fraction.

law is valid, calculate the compositions of liquid and vapor 3.9. One pound mole of a mixture of 60 mole
fractions and the amounts of each. Check th" composition percent ethanol, 40 mole percent water is at 60° F and
values with those of Fig. 3.2. 14.7 psia.
(b) What percentage of the vapor will condense if the (a) What are the enthalpy and physical state?
total pressure is increased to 29.4 psia? (b) How much heat must be added to raise the
3.3. Determine the compositions of the phases speci- temperature to 1 75° F? What is the state of the system?
fied for each of the following vapor-liquid equilibria: (c) What temperature and physical state would be
(a) The vapor phase in equilibrium with a liquid of 21 reached if 20,000 Btu were added to the original mix?
mole percent O2 and 79 mole percent N2 at atmospheric 3.10. Two pound moles of a saturated vapor of
pressure. ammonia and water exist at 250 F and atmospheric
(b) The liquid phase in equilibrium with a vapor of 50 pressure.
mole percent a and 50 mole percent 6, where the relative (a) What is the vapor composition and the com-
volatility of a to b is 1 .9. position of the first liquid droplets that condense from it as

(c) The liquid and vapor phases for a mixture of cooling begins?
acetone and chloroform at 135°F and atmospheric pres- (b) How much heat must be removed to cool the
sure. system to 200° F?
 PHASE RELATIONS 47

(c) What will be the composition of the equilibrium solution contains HNO3 at a concentration of 1.0 g mole
phases present at 200°F? per liter of water.
(d) At what temperature will the system be totally (b) The isopropyl ether-rich phase in equilibrium
condensed? with a water-rich phase containing 40 mass percent acetic
3.11. Calculate the relative volatilities of N2 to O2 at acid at 20°C.
5 atm pressure and at several compositions spanning the (c) The extract phase in equilibrium with a raffinate
entire composition range. phase containing 0.6 mass of styrene/mass of styrene plus
3.12. Solute a is equally soluble in solvents b and c ethylbenzene on a glycol-free basis at 25° C.
when solvents b and c are in contact with each other, but (d) Adsorbed phase concentration in equilibrium
solvents b and c are completely immiscible at all concen- with a gas phase containing 50 mole percent propane, 50
trations of a. mole percent propylene. The adsorbent is silica gel at 25° C.

(a) Construct the triangular diagram and equilibrium (e) An underflow sludge in equilibrium with a

curve for this system. solution containing 30 mass percent solute and 70 mass
(b) Could a be recovered from a mixture of a —b percent solvent. The sludge contains 1.5 lb of solution per
using solvent c in multistage equipment? Explain. pound of insoluble material.
Which of the following mixtures will form two
3.13. How much enthalpy would have to be removed
3.15.
phases, and what are the compositions of the phases? to cool 60 percent O2, 40 percent N2 (mole basis)
a

(a) 25 mass percent acetic acid, 30 mass percent mixture that is originally a vapor at its dew point at 5 atm
isopropyl ether in water at 20°C and atmospheric pressure. pressure to such a temperature that no vapor would be
(b) 40 mass percent acetic acid, 40 mass percent present upon dropping the pressure to 1 atm?
isopropyl ether, 20 mass percent water at 20°C and 3.16. Construct a temperature-composition diagram
atmospheric pressure. similar to Fig. 3.2 for the system benzene-o-xylene at
2
30 mass percent acetic acid, 5 mass percent water
(c) 100 kN/m pressure.
in isopropyl ether at 20°C and atmospheric pressure. 3.17. Convert the following compositions to a sol-
(d) 60 mass percent styrene, 20 mass percent diethy- vent-free basis and make a sketch showing where each
lene glycol, and 20 mass percent ethylbenzene at 25°C composition is plotted on Fig. 3.10:
atmospheric pressure. (a) xa = 0.2, xb = 0.7, xc = 0.1
(e) 10 mass percent styrene, 70 mass percent diethyl-
(b) xa = 0.2, xb = 0.8, x c = 0.0
ene glycol, and 20 mass percent ethylbenzene at 25°
atmospheric pressure. (c) xa = 0, xb = 0, xc = 1.0

(f) 70 mass percent styrene, 10 mass percent diethy- 3.18. Four hundred pounds of silica gel are mixed
lene glycol, and 20 mass percent ethylbenzene at 25°C with 2 lb moles, of a gas containing half propane and half
atmospheric pressure. propylene at 25° C.
(g) 30 mass percent 60 mass percent
oleic acid, (a) Determine the composition of the resultant gas
propane, and 10 mass percent cottonseed oil at 85°C and phase.
3420 kN/m 2 pressure. (b) Determine the mole fraction of propane in the
Same composition
(h) as in part (g) but at 98.5°C, gas adsorbed on the silica gel.

4310 kN/m 2 .
(c) How much gas is adsorbed?
0.25 g moles of U02(N03) 2 in one liter of water
(i) 3.19. Convert the isopropyl ether-acetic acid-water
containing 3 g moles of HNO3 and one liter of 30 volume diagram of Fig. 3.8 to a solvent-free basis.
percent tributyl phosphate in kerosene. 3.20. Plot the following underflow loci by calcu-
3.14. Determine the composition of the phases speci- lating four points on each line:

fied for each of the following phase equilibria: (a) Constant ratio of mass of adhering solvent to
(a) The organic solvent phase in equilibrium with an mass of inerts= 0.75.
aqueous solution of uranyl nitrate of concentration 0.3 g (b) Constant ratio of mass of adhering solution to
mole U0 2 (N0 3 )2 per liter H 2 at 25°C. The aqueous mass of inerts= 0.75.
 Four

Equilibrium Stage Calculations

A first step in the analysis of multistage operations where two phases result The question of whether
isthe consideration of a single equilibrium stage. In or not the mixture I actually exists in the stage is

this chapter, procedures are developed for cal- of no concern; it is a useful calculational concept.
culating the unknown phase flow rates and com- Similarly, zmay be defined as the overall compo-
positions for a single stage from available data. mixture 3.
sition of the
Methods are then developed for calculating multi- Mass balances may be written in terms of
stage systems. conventional mass units, or in molar units, for a
A prerequisite for the analysis of an equi- ternary system containing components a, b, andc:
librium stage knowledge of the appropriate phase
is

equilibrium for the system being processed. Be-

Total mass balance: L + V2 = £ (4.1)

cause these equilibrium data are often presented Components balance: L x a0 + v^/^ = ^ za
graphically, the use of graphical methods of calcu- (4.2)
lation is convenient. The procedures involved in
graphical calculations can also be applied directly
Component c balance: L x cQ + V2 y C 2 ~ -^c
(4.3)
to calculations on the digital computer. Many of
the examples and problems in the next several In these equations, L and V2 are the total masses
chapters are too simple to warrant computer sol- of the entering streams, x, and y j2 are the
ution. In more complex problems in stage oper- respective mass fractions, and I and z are the total
ations, such as multicomponent distillation con- mass and overall composition of the resulting
sidered in Chapter 8, computer solution is almost mixture. An equation for component b might also
indispensable. Calculation on the digital computer be written, but it would not be independent
requires that the equilibrium data be available in a because x b = 1 — x a — x c If the quantities and
.

form that can be easily fed to the computer, such compositions of the entering streams are known,
as constants for empirical equations. both the mass (I) and the compositions (za ,zc )

The following discussion applies generally to can be calculated for the sum.
calculations on ternary, enthalpy-composition, and Mass is transferred between the phases in the
solvent-free-basis diagrams.The procedures are de- mixture I until equilibrium is established. The
veloped using a ternary diagram and are then ex- phases are then separated to flow from the stage as
tended to the other diagrams. Li and V, The compositions of the phases leaving
.

are different from those entering. The compo-

sitions of the two phases leaving (x n and yn are )
SINGLE EQUILIBRIUM STAGE
Two streams enter a single equilibrium stage, and
two streams leave the stage (Figure 4.1). The 1 Equilibrium
Sta § e
entering streams are thoroughly mixed together to T

give a sum that may be designated Z. To be useful

in a separation process, 1 must be in a region Figure 4.1. Single equilibrium stage.

48
 EQUILIBRIUM STAGE CALCULATIONS 49

determined by the phase equilibrium; the quan- Solving the two equations yields Ly =65.4, \A, =
tities(Z.! and V-\ axe determined by mass bal-
) 34.6. So 34.6 percent of the original liquid is
ances: vaporized.

2= /.! + Vy (4.4)
Example What
are the compositions and
4.2.
!Lz a = Lyx ay + V,y a ,
(4.5) quantities phases leaving an equilibrium
of the
stage into which the following two phases are fed?
2zc = Z.!X cl + ^Kd (4.6)

/.-phase: 100 kg acetic acid, 20 kg isopropyl

The compositions of the phases leaving an equi-
ether, 80 kg water.
librium stage depend not only upon the equilib-
V-phase: 15 kg acetic acid, 82 kg isopropyl
rium relationships of the system, but also upon the
ether, 3 kg water.
compositions and relative quantities of the two
phases entering the stage. These latter quantities Solution. From the data given
determine the mixture composition z, which then
Z. = 200,x a0 = 0.5,x 60 = 0.4,x c0 =0.1
yields the exit compositions when coupled with
the equilibrium relationship. Mass-balance equa- V2 = 100, y a2 = 0.15, y b2 = 0.03, y c2 = 0.82
tions similar to Equations 4.1—4.6 can be written
By Equation 4.1, 2 = L + V2 = 200+ 100 =
for systems of any number of components. For
300 and by Equation 4.2,
two components, as in binary distillation, Equa-
tions 4.3 and 4.6 are not needed, because there is
no component c. Generally, for a system of n za = —
(200)(0.5) + 100(0.15)
= oooo
0.383
components, n mass balances for components and 300
total quantities are required to define the system. Similarly,

200 0.1 +100 0.82 „

zc = = 0.34
Example 4.1. An equimolar liquidmixture 300
of benzene and toluene is initially at 250°F and
100 psia. The pressure reduced to 30 psia while
is and z b = 1 -
0.383 - 0.34 = 0.277. The compo-
the temperature is maintained at 250°F. Calculate sition of mixture can be located in the
this
the fraction of liquid that is vaporized. two-phase region of the ternary diagram (Figure
4.2). The mixture splits into two phases, as
Solution. Raoult's law may be used to calcu- indicated by the tie line through z. The correct tie
latethe compositions of the final liquid and vapor. line through z is determined by trial and error, as
From Appendix D-1 at 250° F, the vapor pressures described in Chapter 3. The compositions of the
are Pb = 43 psia and P t = 20 psia. By the method phases leaving the stage may be read from the
of Example 3.1, diagram: x a1 =0.45, x 61 =0.45 x c = 0.10; i

43 20
ya, =0.32, y M
=0.11,}/ c1 =0.57.
The quantities of L^ and V^ are determined
by Equations 4.4 and 4.5
So xb = 0.435 300 = /.! + V,
43 (300X0.383) = /.! (0.45) + V, (0.32)
and Vb= Xb
Yo
Solving, /.,= 146 kg 1/,= 154 kg.
= 0.623

In Equations 4.4 and 4.5, Z may be used to Graphical Calculations

identify the original so that z b = 0.5. L^
liquid,
The numerical calculations of Example 4.2 may
and V-i will be the quantities of final liquid and
also be carried out graphically. For many prob-
final vapor, respectively. Choosing 2 = 100 lb
lems, graphical procedures are more rapid than
moles and applying Equations 4.4 and 4.5 gives
numerical calculation; they are also subject to the
inherent inaccuracy of reading graphs. If the
100=/.! + V y
equilibrium data are presented graphically, calcu-
100(0.5) = 0.435/. ! +0.623^ lations can be made directly on the equilibrium
50 STAGE OPERATIONS

Rearrangement yields

Lo_z_ Va7
"
(4.7)
V2 x a0

Equation 4.7 relates the composition of the

resultant mixture to the compositions and masses
of the original two mixtures. The compositions are
expressed terms of component a.
in
Similarly, combining Equations 4.1 and 4.3
gives for component c

Lo_
=
Zc - Vc7
(4.8)
V2 x c0

Combining Equations 4.7 and 4.8 and rearranging

gives

x c0 ~ z c Vcl
(4.9)
x a0 — Za Val

which will be used in the following geometric

argument.
The compositions representing L V2 and I , ,

can be plotted on a ternary diagram, as shown in

Figure 4.3. The composition z of I is shown in
Figure 4.3 to be at some point not necessarily on a
straight line between x and y, because first it must
0.2 0.4 06 08
be shown that angle a is equal to angle j3 before
x a mass
. fraction acetic acid in water layer
plotting x, y, and z on a straight line is justified.
Figure 4.2. Solution to Example 4.2. The tangent of a is (z c — y C 2)Kz a — y a2 and the ),

tangent of is (x c0 - z c )/(x a0 - z a Therefore, ).

diagram. The error involved in reading the graph from Equation 4.9, tan a = tan and point
0,
may be no greater than the experimental error in (z a , zc ) must lie on a straight line between points
the equilibrium data. (x a ,x c ) and iy a ,y c ), as shown in Figure 4.4. The
Probably the most important graphical tech- exact location ,zc is determined
of by the
(z a )

nique is the determination of the composition of a relative quantities of the L and V phases. Equa-
mixture resulting from the addition of two mix- tions 4.7 and 4.8 each define the location. They
tures. The following derivation develops the equa-
tions needed to justify graphical addition on
equilibrium diagrams. It is shown that the compo-
sition of a mixture resulting from the addition of
two mixtures lies on a straight line between the
_ *<-o r^o
compositions of the original two mixtures. This is
called the graphical addition rule. A relationship is
then developed to determine the location of the
composition of the resultant mixture on the V
fa
>
straight line: the inverse lever-arm rule.
If 2 in Equation 4.2 is replaced by the _/a
expression of Equation 4.1, V2

^-o*ao + V2 y a2 = (l-o + V2 )z a
Collecting terms in L and V2 gives
Mass or mole fraction a
L (x a0 -z a )
= V2 (z a -y a2 ) Figure 4.3. Proof of the addition rule.
 EQUILIBRIUM STAGE CALCULATIONS 51

Example 4.3. Calculate Example 4.2 using

graphical methods.

u The compositions of L and V2

Solution.
E*cO |^0 given Example 4.2 are plotted in Figure 4.2. The
in
u
•9
composition z of the mixture 2 must lie on the
o>
o straight line between x and y 2 Its exact location .

E
V is determined by Equation 4.10.
o Zc

TO z - y2 200
2 = 2
yc2 fv2 100
*o

Therefore the distance from y 2 to z is twice the

ya 2 2a *a0 distance from x to z, and the point z may be
Mass or mole fraction a located, as shown. The same conclusion results

Figure 4.4. Graphical addition.

from the application of Equation 4.1 1.
2-y-i = 2
x - y2 3
are both expressions of the inverse lever-arm rule.
Therefore, the point z lies two-thirds of the
The rule can also be stated as follows: if a mixture
distance from y 2 to x and from Figure 4.2,
added to another mixture V2 the compo-
,

L is ,
z a - 0.38, z b = 0.28, z c - 0.34. These values check
sition of the resultantmixture 2 lies at a point on
Example 4.2 within the accuracy of the graph.
a straight line between the compositions of L and
To the quantities of Z.-| and V-\
find the
V2 such that the ratio L /V2 is equal to the ,

inverse lever-arm
rule is applied to the tie line
distance from V2 to 2 divided by the distance
through z which connects x^ and /1 on the phase
from L to 2. That is,
envelope. Equation 4.10 may be rewritten for the
splitting of mixture 2 into two phases.
Z-o
=
*/2 z-Y2
(4.10)
V2 x^z Xq-Z hi -
z ~ K1

The distances between L V 2 and 2 may be , , The right-hand side of this equation can be
measured along the line connecting them, but it is evaluated by measuring the distances, or more
more convenient to use one of the coordinate easily by substituting the coordinates. For a,
scales as indicated by Equation 4.10. The inverse
feature of the lever-arm rule is indicated in its
0.38 - 0.32
= 0.86
definition.For example, a very large quantity of 0.45-0.38
phase L added to a small quantity of phase V2 or for c,
would produce a mixture 2 of a composition
0.34-0.57
nearly the same as L Therefore, the composition
. = 0.96
z of 2 would lie very close to the composition x of 0.10-0.34
L Q and on the straight line between x and /. The error of graphical reading causes the answers
Alternate forms of the lever-arm rule can be to disagree. The value for c is more accurate
derived.
because the differences are larger and less sensitive

to a graphical reading error. Therefore

L z-y 2
(4.11)
2 X -/2 — = 0.96

V2 x -z
and
2 x -y 2
(4.12)
Z.! +Vi =2 = 300

300
and Vy = = 153 lb
These equations are particularly useful in locating 1.96
the point z. Equations 4.10, 4.11, and 4.12 are
/., =300-153= 1471b
written without component subscripts. They apply
to each of the three components. These results check with Example 4.2.
52 STAGE OPERATIONS

Subtraction of mixtures can also be carried x a =0.0 to x a = 1.0. The graphical addition rule
out graphically. The difference (A) can be expres- then states that when two mixtures of a and b are
sed by added, the composition of the resultant mixture
lies on the straight line running through x
a
= 0.0
A = L - V for the total mass (4.13) and x a = 1.0. The inverse lever-arm rule may be
applied to determine where on the line the
Ax A = Lx - Vy for any of the components (4. 1 4)
composition lies.

If Equation 4.13 is rearranged, A+V=L, it The graphical methods developed in terms of

becomes clear that graphical subtraction is equiv- ternary systems also apply to other systems involv-
alent to graphical addition and that the inverse ing conserved properties. For example, in distil-
lever-arm rule applies. In this case the composition lation, both mass and enthalpy are conserved, so
of the difference on a straight line drawn
x^ lies that graphical calculation on an enthalpy-compo-
through x and y but extended beyond x from y. sition diagram is possible. In addition to the
Examination of Figure 4.5 shows the similarity inherent enthalpy of the streams entering and
between graphical addition and subtraction. leaving an equilibrium stage, an enthalpy balance
The separation of the mixture - into L^ and must also include heat directly added to or
Vy discussed earlier may be considered as a removed from the stage. In addition to the mass-
subtraction of Vy from I to give L^ (or as a balance equation (Equations 4.1 and 4.2), it is
subtraction of Z-i from I to give Vi ). Either of the necessary to write an enthalpy balance for the
two parts of Figure 4.5 can be applied to deter- addition of two streams:
mine the masses of L^ and V A reexamination of
V2 H 2 +q
.
y
L hQ + = ^h z (4.15)
Example 4.3 suggests the procedure for solving
subtraction problems graphically by treating them where h is the enthalpy of stream L 2 is the , H
as addition problems. enthalpy of V2 and h z is the average enthalpy per
,

Graphical procedures can be applied to two- unit mass of mixture -. All enthalpies are ex-
component systems, wherein all possible compo- pressed per unit mass of the stream. When dis-
sitions fall on a single straight line running from cussing distillation, the stream masses are expressed
in pound-moles, so that enthalpy is in Btu per
pound-mole (or in Joules/mole). The heat added to
the mixture is q in Btu (or in Joules). If heat is
removed, q has a negative numerical value, but
Equation 4.15 still holds. In enthalpy-composition
diagrams, only two components are involved, so
Lx * Equation 4.3 is not needed. The addition rule and
the inverse lever-arm rule are written in terms of
composition and enthalpy. Because addition of
l + v=i heat does not affect the mass balance, Equations
4.10 to 4.12 still hold for component a or b.
Therefore, inverse lever-arm equations are valid
along the composition axis.
Inverse lever-arm relationships for the en-
Subtrdcbofl
thalpy axis are complicated by the heat-addition
term q. If q is zero, Equation 4.15 is directly
*e.* comparable to Equation 4.3, and the derivation of
L-V=A
the inverse lever-arm relationships is identical in
>- » A. x
form to that of the ternary system. In this sense,
L. x
enthalpy in distillation can be considered anal-
ogous to solvent c in extraction. In extraction, a
solvent c is used to separate a more soluble
component a from a one b. In
less soluble
distillation, "used" to separate a more
enthalpy is

volatile component a from a less volatile one b.

Figure 4.5. Comparison of graphical addition and subtrac- If q is not zero, the enthalpy of the mixture is

tion. increased by heat added to the system. Equation

 EQUILIBRIUM STAGE CALCULATIONS 53

4.1 5is the correct energy balance, but it is difficult Example 4.4. A saturated liquid containing
to derive addition and lever-arm rules because of 50 mole percent ethanol and 50 mole percent
the term q. Only enthalpy terms per unit mass water is heated in a closed and evacuated tank.
(e.g., Btu/lb mole) can be plotted on the enthalpy- Liquid is vaporized until the temperature of the
composition diagram; but q is simply heat (Btu). remaining liquid reaches 180°F at 1 atm pressure.
Therefore, q must be expressed in terms that will (a) How much of the original liquid was
permit it to be represented graphically. Suppose vaporized?
the mixing process is imagined to be split into two (b) How much heat per pound mole was
steps: added to the system to vaporize the liquid?

1. L and V2 are mixed to give 2. Solution, (a) Since the liquid was vaporized
2. q units of heat are added to 2. in a closed tank, the vapor formed is still in contact
with the liquid, and equilibrium conditions will
The enthalpy balance for step 1 is
exist. Therefore, the liquid is at its bubble point,
L h + V2 H 2 =2/^ (4.16) which is 180°F. Its composition must lie on the
saturated-liquid enthalpy curve of Figure 4.6 at
where /)S i is the enthalpy of the mixture before 180°F. Thus, the remaining liquid contains 26.5
heat is added. The enthalpy balance for step 2 is mole percent ethanol. The vapor must lie on the
+qr= saturated-vapor curve at 180°F, so that the mole
2/7 21 2A/ :
(4.17)
fraction ethanol in the vapor is 0.565.
where the final enthalpy after q is added. The
/? 2 is From this point, the problem may be solved
sum two equations gives Equation 4.15, as
of these either as an addition or as a subtraction. For
expected. The second equation may be rearranged addition, the original liquid is 2 and the final
to give liquid and vapor are L and V, respectively, so that
2 = L + V. For subtraction, the initial liquid is L,
q/Z = (h z -/> £1 ) (4.18)
the final vapor V, and the remaining liquid A. That
is, vapor is subtracted from the original liquid to
The term (h x —h
z -[) is simply the heat added to
give a final liquid (L — V=A). Addition is used
the mixture per pound-mole of mixture, as shown
here, and subtraction is left as an exercise (see
on the left side of Equation 4.18. This heat
Problem 4.18).
addition is given a new symbol:
The problem must be restated as an addition:
Qz =<7/2 = (/7 -r/ s1 (4.19) the equilibrium liquid L and vapor V are mixed;
z )

then heat is removed until the total mixture

Rearranging gives becomes a saturated liquid. This is just the reverse
/7 21 =/7 L -Q Z (4.20) of the stated process. The heat removed in the
reverse process is exactly equal to that added in the
The term /?
s1 may be regarded as a fictitious
enthalpy, defined by Equation 4.20.
The addition and inverse lever-arm rules may H
be applied to Equation 4.16 to evaluate h z Since ^
. 20,000

Equation 4.16 is in the same form as Equation 4.3,

'-*— (2,A
the inverse lever-arm relationships (Equations £ ,)
4.10—4.12) can be written in terms of enthalpy. 5 15.000 1/
Sf/
For example, h Q H 2 and h^ all lie on a straight
, , y

line; Equation 4.12 becomes

10.000
hn-h SI
(4.21!
2 h -H-,
- Uo.>ioN /
This equation may be used to evaluate /)v then \ ,

Qz or /?£ may be determined, depending upon Mz.Ai)

which is known.
The heat addition could be defined per pound 2 4 6 8 1

x„. mole fraction ethanol

mole L or V, instead of 2. Other useful
of \\, .

equations then result (see Example 4.4). Figure 4.6. Solution to Example 4.4.
54 STAGE OPERATIONS

forward process. Now the equations derived using and b is A prime (') on the symbols for
considered.
mass and enthalpy balances for addition may be mass used to denote that the stream is being
is

used to solve the problem. By inverse lever-arm
rule (Equation 4.12):
V2 length of linex z
2 length of linex K2
or by Equation 4. 1 2 for
considered on a solvent-free basis. Thus, L' is the
mass of a and b in stream L, and A' is the net flow
of a and b. Mass ratios are used here to report
compositions. Thus, X
c is the mass of c per unit
= 0.78
mass of a +b. (Note the use of capital letters for
mass ratio.)
component a: The notation and relationships then become:

V2 _ xa o ~ ?a 0.265 - 0.50 = mass

= 0.78 L' of two components (a and b)
~
2 x a0 -y a2 0.265 -0.565 disregarding the solvent (c), in phase L.
\/'=mass of two components (a and b)
Therefore, 78 mole percent of the original liquid disregarding the solvent (c), in phase V.
was vaporized. X = mass of any component (a, b, ore) per
(b) Basis: One pound mole of the original unit mass of L'.
liquid (2). Q L is the required quantity. This is V = mass of any component {a, b, or c) per
most easily calculated from
Equation 4.19, pro- unit mass of V'.
viding /7 S1 and /?v can be determined. Equation
4.16 is used to locate h z That equation requires
, .
With this notation,

that h H 2 and /j^ lie on a straight line. The

, ,
Xa + Xb = 1, Xc can have any value
inverse lever-arm rule may be used to determine
where on the straight line h z lies, but it is easier ^
Ya + Yb = 1, Yc can have any value
to note that step 2 (Equation 4.17) involved no Yc is now
the mass of solvent per unit mass of
addition of mass, so that the composition after solvent-free material (a plus b), and it may have
step 1 is the same as after step 2. As a result, /? £ ^ any value between and °°. When pure solvent is
is located on the straight line between (x h ) and , fed, V c (/v + 1 = oo, and the inverse lever-arm ex-
)

(y 2 ,H 2 ), at z a = 0.5. From Figure 4.6, /7 S1 = pressions involving Vc(/V + 1 )

are indeterminate. In
17,000. The mixture 2 is a saturated liquid, so that this case it is necessary to carry out additions
/7v = 3,500. Therefore, by Equation 4.19 numerically rather than graphically.
Qv =3,500- 17,000 = -13,500 Btu/lb mole
Example 4.5. 1000 kg of pure c is added to
This Qv is the heat added for the reverse process.
2000 kg of a mixture containing 50 percent a and
Hence, the heat added for the forward process is
50 percent b. Determine the mass and compo-
13,500 Btu/lb mole.
sition of the resulting mixture, on a total mass
basis and on a solvent-free basis.
Solvent-Free Basis
Ternary equilibrium data for extraction can be
Solution. On a total mass basis:
presented on a solvent-free basis, as described in
Chapter 3; the resulting diagram can be used for Z. = 2000, x a =0.5,x 6 =0.5
graphical calculations. The addition rule and the
V= 1000, y c = 1.0
inverse lever-arm rule apply, but the latter relation-
ship is sometimes indeterminate. In using the and
solvent-free basis, parallels may be drawn between
the solvent in extraction and enthalpy in distil-
L + V=2 = 3000

lation. Each can be viewed as the "agent" that from a mass balance
produces the separation. Graphical calculations on
z a = 0.333, z b = 0.333, zc = 0.333
the equilibrium diagram plotted on a solvent-free
basis parallel those for an enthalpy-composition This point z could be located by the inverse
diagram. lever-arm rule on the line between x and / (Figure
The is convenient for calcu-
solvent-free basis 4.7). On the other hand, on a solvent-free basis:
lations, perhaps confusing at first contact.
but it is

When considering the mass of a stream, the solvent

Z.' = 2000, X a = 0.b,Xb =0.5

is ignored, so that only the mass of components a V' = 0, Yc = oo(Ya and Yb are indeterminate)
 EQUILIBRIUM STAGE CALCULATIONS 55

V= 1000

Stage Stage Stage

L
1 2 3

Figure 4.8. Multiple-stage contacting.

stages, such as that shown Figure 4.8, permits a

in
greater recovery of the desired component than is
possible in a single stage. For example, the under-
flow leaving a stage in solid-liquid extraction may
be contacted with fresh solvent again to remove
additional solute. The underflow from this second
stage may again be washed with fresh solvent in a
third stage, and so on. As stages are added, a
greater recovery of solute is achieved, but the
concentration of the solute in the extract becomes
more and more dilute. If the solute or solvent must
be recovered from the extract, the removal of large
quantities of solvent would be expensive. The
following chapters develop methods to give high
recovery and concentrated solutions. Of course, if
the solute is worthless and there is no need to
recover it, and if it is not an objectionable
pollutant, the dilute extracts are not a problem. An
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
example would be the washing of an insoluble
paint pigment to remove traces of the reactant
Figure Comparison of addition on a
4.7.
total mass basis and on a solvent-free basis
solution from which the pigment was originally
precipitated. Here the traces of reactants are not
(a) total mass basis, (b) solvent-free basis.
worth recovering, but it is necessary to remove
them from the pigment. In this situation the
pigment might be washed with three successive
Then U + V = 2' = 2000 lb of a and b and batches of free water, to give the arrangement

-,
Za = —
1000
2000
nr _
- = 0.5; Zb =
1000
2000
= 0.5; Zc =
1000
2000
= 0.5
shown in Figure 4.8. The wash water might then be
discarded.

When Yc = oo the addition rule holds, even though

the inverse lever-arm rule cannot be used. Yc can be Example 4.6. One thousand pounds of a

considered to be at infinity, so that a straight line roasted copper ore containing 10 percent CuS0 4 ,

from any other composition to /will be a vertical. 85 percent insoluble gangue, and 5 percent mois-
Thus, the composition of the sum Z lies on a ture is to be extracted by washing it three times
vertical line above X
(that is, on a straight line with 2000- lb batches of fresh water. It has been
between X and Y where Y is at °°) (Figure 4.7). found that the solids retain 0.8 lb of solution per
pound of gangue. What is the composition of the
final underflow sludge after three washings?
MULTIPLE EQUILIBRIUM STAGES

Numerical and graphical calculations can be ap- Solution. The underflow locus is located on
plied to a series of equilibrium stages. A series of Figure 4.9 from the following two points (basis:
56 STAGE OPERATIONS

-*0 y3 y.

0.02 0.04 0.06 0.08 0.10

i„, ya , mass fraction CuS0 4

Figure 4.9. Solution to Example 4.6.

1 lb gangue) sition on the diagonal, since it contains no

lies

solids. L = 1000 and V = 2000, the

Since
At x a = 0, xc = —
0.8

1.8
= 0.445, composition of Ej = L + V is located by the
inverse lever-arm rule. Now, St = L-\ + V-j, and L-\
consists of insoluble gangue and solution of
xb = —1
= 0.555 composition y Therefore, the mixture L^ + Vy
y
.

can be considered to consist of gangue and

Atx a =0.1,x c = — -0.1
0.8
=0.345, solution. Because the gangue is represented by the
right angle (x a = 0, x b = 1.0, x c = 0.0), a straight
line through x b = 1.0 and z^ will locate Ki where
Xb = — = 0.555 it intersects the diagonal.
Since Ly = 2^ — V-\, the composition x^ of
The process is by Figure 4.8. The
represented Li is located at the intersection of the underflow
compositions of the incoming ore (x and the fresh )
locus with the straight line through z z and y^ In
.

solvent (y are located on Figure 4.9

)
the second stage, L-\ is mixed with V and the
procedure outlined above is repeated.
x a0 =0.10,x c0 = 0.05; K a0 = 0.0, yc0 = 1.0.

In the first stage L Q and V are mixed, and Ly and I2 = ^-1 + V = L2 + V2 , etc. as above
Vt leave the stage in equilibrium. For leaching,
First, it is necessary to calculate L by a balance
equilibrium occurs when the solution adhering to :

around the first stage.

the gangue has the same as the composition
solution leaving in the extract. Therefore, the
L y y^zy 0.0465-0.0333
underflow consists of gangue, which is represented = 1.07
by the right angle, and solution, whose compo- V y
zyxy 0.0333-0.0210
 EQUILIBRIUM STAGE CALCULATIONS 57

/.!+ Vy = 3000 lb, and thus Z. =1550lb and 1

lies on a straight line that runs between the compositions of
V-\= 1450 lb. For the second stage 2 2 is located the two streams mixed (x and y). See Figure 4.5, but do
by the inverse lever-arm rule with L-\ = 1550 and not use the addition rule to prove this "subtraction rule."
VQ = 2000. Then the quantity of L 2 is determined. 4.5. Would the addition rule and inverse lever-arm
rule apply to the combination of mixtures containing four
L2 y2 z2 0.99 - 0.74 components? Explain.
= 0.83 4.6. Derive an addition rule and an inverse-arm rule
V2 z2 x 2 0.74 - 0.44
expressed in terms of enthalpy-composition diagrams.
4.7. Perform the indicated calculation graphically for
Since L 2 + V2 = 3550, L 2 = 1615 lb, V2 = 1935 lb. the following three mixtures. Use a triangular diagram with
For stage 3, 2 3 = L 2 + V = L 3 + V3 , etc. From x a on the horizontal axis and x c on the vertical axis:
Figure 4.9, the composition of the underflow from
Q: 2000 lb: xa = 0.7, xb = 0.2, xc = 0.1.
the third stage is 0.16 per cent CuS0 4 The mass of .

the underflow (L 3 ) is 1555 1b. Therefore, the /?: 1000 lb: xa = 0.3, xb = 0.5, xc = 0.2.
percentage CuS0 4 not recovered is
S: 3000 lb: xa = 0.1, xb = 0.2, xc = 0.7.
(0.0016)(1555)
— — = n
2.5 percent (a) Add Q to S, then subtract /?.
(0.10)(1000)
(b) Subtract R from S, then add Q.

To recover the copper sulfate from the product

(c) Subtract R from Q, then add S.
(d) Add Q, R, and S.
solutions in Example 4.6, the large quantity of
4.8. Calculate graphically the composition of the
water would have to be evaporated. The quantity
final mixture resulting from the following processes. Use a
of water required for a fresh wash to each stage is
triangular diagram where water is plotted on the vertical
large, and the resultant solutions are dilute. An
axis, salt is plotted on the horizontal axis, and sand is on
arrangement that gives a more concentrated extract
the diagonal axis.
and requires less solvent for a given recovery of
(a) 100 lb of water is evaporated from 1000 lb of the
solute is discussed in the next chapter.
following mixture: water, 35 percent; salt, 20 percent; sand,
45 percent.
(b) 2000 lb of water, 5000 lb of salt, and 3000 lb of
PROBLEMS sand are mixed together.
(c) The following two mixtures are mixed together
and then 250 lb of sand is strained out and 150 lb of salt is

4.1. A vapor at 90°C consists of 30 mole percent

added. Initial mixtures:
propane, 30 mole percent n-butane, and 40 mole percent
n-pentane. As the vapor is compressed, 1000 lb: 40 percent sand, 10 percent water,
(a) At what pressure does condensation begin? 50 percent salt.

(b) At what pressure are half of the original moles of

200 lb: 10 percent sand, 70 percent water,
vapor condensed?
(c) At what pressure is all the vapor condensed? 20 percent salt.

4.2. A liquid containing 40 mole percent n-pentane, (d) 300 lb of the solution in the following mixture is

60 mole percent n-hexane is at 180°F and 100psia. The

poured off of the insoluble sand.
temperature is increased until 75 percent of the original
1000 lb: 35 percent sand, 55 percent water,
moles of liquid is vaporized. What is the final temperature?
4.3. Thirty moles of n-heptane are mixed with 50 10 percent salt

moles of n-octane at 70°C. Determine the composition(s)

(e) 245 lb of a mixture (70 percent sand, 20 percent
of the phase(s) present at a pressure of
water, 10 percent salt) is removed from the following mix-
(a) 5,000 N/m 2 . ture:

400 40 percent sand, 10 percent water, 50 percent

(b) 20,000 N/m 2 .
lb:

salt.
(c) 40,000 N/m 2 .

The following two mixtures are to be added

4.9.

4.4. Prove that the composition (x A ) of a difference together:

(A): 125 kg:x a = 0.52, x b = 0.36, x c = 0.12

A = L- V 82 kg: x a = 0.25, x b = 0.05, x c = 0.70

58 STAGE OPERATIONS

Determine the composition of the mixture graphically using 4.15. Calculate the minimum quantity of diethylene
the inverse lever-arm rule: glycol that must be added to 100 kg of
a solution of 80
(a) By measuring along the line between the two mole percent styrene and 20 mole percent ethylbenzene to
compositions. produce two phases. How much more diethylene glycol
(b) By using the x a coordinate scale for measure- must be added to produce one phase again?
ment. 4.16. The net effect of the following two-step pro-
(c) By using the x^ coordinate scale for measure- cess is equivalent to the process described by Equation
ment. 4.15:
4.10. The following two phases are mixed together:
1. Q Btu of heat are added to liquid L.
200 lb: 20 percent oleic acid, 80 percent cottonseed 2. The resulting mixture is added to vapor V to give
oil
mixture £.
100 lb: pure propane.
(a) Using this two-step procedure, develop a calcu-

Determine graphically the composition of the mixture: lation method to evaluate the final enthalpy and compo-
(a) Using Equation 4.10. sition of 2.

(b) Using Equation 4.1 1. (b) Using a diagram similar to Figure 4.6, locate Q^,
(c) Using Equation 4.12. the heat added per pound-mole of original liquid.

4.11. Calculate graphically the composition of the 4.17. Solve Example 4.4 using the method developed
mixture resulting when the following three solutions are in Problem 4.6.

mixed: 4.18. Solve Example 4.4 using the graphical subtrac-

tion method.
250 kg: 15 percent acetic acid, 85 percent water.
4.19. Derive a graphical addition rule and an inverse
50 kg: pure isopropyl ether.
lever-arm rule for use with solvent-free coordinates. Define
100 kg: 20 percent acetic acid, 78 percent water, 2
all terms carefully.
percent isopropyl ether.
4.12. Forty pound-moles of water at 1 atm pressure 4.20. A seashore sand contains 85 percent insoluble

and 60°F is mixed with 30 lb moles of ethanol at its dew sand, 12 percent salt, and 3 percent water. 1000 lb of this

mixture is to be extracted so that after drying it will

point at 1 atm. The resultant mixture is heated to 185°F at
1 atm.
contain only 0.2 percent salt. How many washings with
(a) How much heat was added to the mixture? 2000-lb batches of pure water are required to give the
(b) What are the compositions and quantities of the desired purity? The sand retains 0.5 lb of water per pound
resultant phases? of insoluble sand.

4.13. The following pairs of phases are brought into 4.21. One thousand pounds of a solution of 35 mass
contact and are allowed to come to equilibrium in a single percent acetic acid in water is to be extracted with 2000 lb

stage. Calculate graphically the mass and composition of of pure isopropyl ether. Calculate the percentage removal

the resulting two phases. of the acetic acid from the aqueous phase for each of the

(a) 15 moles of liquid: 0.65 mole fraction nitrogen, following proposals.

0.35 mole fraction oxygen at saturation at 1 atm. (a) The ether is split into four 500-lb fractions. The
25 moles of vapor: 80 mole percent C>2, 20 mole aqueous solution is mixed with one 500-lb fraction, the

percent N2 at its dew point at 1 atm. ether phase is separated, and the remaining aqueous phase is

(b) 250 kg of pure liquid propane at 98.5°C and 625 washed with the second 500-lb ether fraction. The process
psia. is continued until all four fractions are used.
100 kg of a liquid: 30 mass percent oleic acid, 70 (b) The same as part (a) but the ether fractions are
mass percent cottonseed oil at 98.5 C and 625 psia. successively 1000, 500, 500 lb.

(c) 400 tons of a roasted copper ore containing 12 (c) The same as part (a) but only three washes of 667
mass percent CUSO4 and 88 mass percent insoluble inerts. lb are used.

1000 tons of water (d) A single wash of 2000 lb of ether is used.

(The inerts will retain 1.3 tons of water per ton of inerts. 4.22. How much propane must be added to 1 kg of
Of course, some CUSO4 is dissolved in the 1.3 tons of pure cottonseed oil to give two phases? How much more
water.) propane must be added before the mixture returns to one
(d) 1 lb mole of a gas of 40 percent propane, 60 phase?
percent propylene at 25° C. 4.23. One pound-mole of a gas containing 50 mole
300 lb silica gel at 25° C. percent propane and 50 mole percent propylene is allowed
4.14. Derive Equations 4.1 1 and 4.12, and show with to come to equilibrium with 300 lb of silica gel at 25°C and
a sketch the utility of the equations in graphical calcu- 1 atm. The resultant gas is withdrawn and contacted with a
lations. fresh batch of 300 lb of silica gel. What is the final compo-
 EQUILIBRIUM STAGE CALCULATIONS 59

sition of the gas after three contacts with successive batches Ten moles of a saturated vapor (79 percent N 2/
4.26.
of 300 lb of silica gel? 21 percent O2) is mixed with 20 moles of a saturated liquid

4.24. One hundred pounds of a 30 percent solution (15 percent N 2 85 percent O2) at normal atmospheric
,

of acetic acid in water is mixed thoroughly with 100 lb of pressure.

pure isopropyl ether at 20 C. The resulting two phases are (a) Calculate the quantities and compositions of the
separated, and the aqueous phase is mixed with a second resulting equilibrated phases.

batch fo 100 lb of isopropyl ether. Two phases again result. (b) What is the temperature of the resulting equilib-
(a) What percentage of the acetic acid in the original rium phases?
solution remains in final aqueous solution? (c) How much energy would have to be added to
What mass of water was
(b) transferred to the ether convert the resulting liquid phase to a saturated vapor?
phase in the two contacts? 4.27. Write a computer program to solve Example
4.25. Ten pound-moles of a 50 percent ethanol- 4.5.

water solution is heated from 60° F to a temperature at (a) The program should calculate the recovery for
which half of the original mixture has vaporized (on a any preset number of washings.
molar basis). Assume that the vapor formed remains in (b) The program should calculate the number of
contact with the liquid remaining. stages required for any preset recovery of CUSO4.
(a) To what temperature is the mixture heated?
(b) What is the composition of the vapor formed?
(c) How much heat is added?
 Five

Countercurrent Multistage Operations

In designing a separation process, the engineer is through all of the stages, so that regardless of the
concerned with separating the constituents of a number only one fresh solvent
of stages, there is

mixture into more or less pure fractions. The re- feed (to the lastand only one ultimate
stage)
quired product recovery and purity depend on the extract product (from the first stage). Because the
end use of the product; the engineer must be V phase now flows in a direction opposite to the L
prepared to design a mass-transfer operation that phase, the arrangement is called countercurrent
will give the required product at the lowest cost. A flow. This arrangement always used in multistage
is

single stage is rarely sufficient to give high product equipment. The driving force for mass transfer is
recovery and purity. The greatest separation pos- maintained because the entering fresh solvent con-
sible in one stage is achieved when the two phases tacts the most dilute raffinate phase, whereas the
leaving are in equilibrium with each other; even concentrated extract contacts the entering concen-
this separation is usually inadequate. trated raffinate phase. Countercurrent multistage
To obtain the desired separation, several operation permits high recovery and high purity.
stages in must be used. One possible series
series Absorption, extraction, and adsorption utilize
arrangement was discussed in Chapter 4 (Figure the simple countercurrent flow pattern described
4.8). As illustrated in Example 4.5, this arrange- here. An example of an industrial multistage ex-
ment leads to high percentage recoveries of the traction operation is shown in Figure 5.1. Al-
desired component. Additional stages in series though countercurrent flow is used in distillation,
would lead to even higher recovery, if this were the flow pattern is more complex. It is discussed in
required. Because fresh solvent is added at each the following chapter. For purposes of illustration,
stage, the recovered solute is in a very dilute much of the following discussion is in terms of
solution. In the next step in the process, the solute extraction, but the concepts developed are appli-
might be recovered from the solvent by evap- cable to all stage operations. The terminology for
oration. Such a recovery would be expensive be- the various operations is summarized on page 157.
cause of the large quantity of heat required to Calculations for multistage operations are
vaporize the solvent. In addition, in many cases the usually applied to problems in either synthesis or
cost of the solvent is high, so that it is desirable to analysis. In the design of a new industrial process,
use a smaller amount. the engineer is called upon to synthesize a stage
To conserve solvent and also to produce a operation to make the specified separation. For
more concentrated, higher-purity final extract prod- simple countercurrent operations, this would in-
uct, the extract from the last stage (V volve calculating the number of stages required for
3 in Figure
4.8) can be used as the feed solvent to the next- a specified product recovery and purity. Such cal-
to-the-last stage. The extract from the last culations require information on phase equilibrium
stage is very dilute, so that it is not too different and phase flow rates. After the number of equi-
from the pure solvent. Then the extract from the librium stages is determined, a stage efficiency is
next-to-the-last stage is used as the feed solvent to applied to calculate the number of actual stages to
the previous stage. This arrangement is continued be built. Then the flow rates of both phases are

60
 COUNTERCURRENT MULTISTAGE OPERATIONS 61

used to calculate the physical size of each actual

stage. Numerous mechanical problems must also be
considered.
In the analysis of an existing piece of equip-
ment, the engineer may be asked to determine if
product recovery or purity can be improved, if
more material can be processed, or if the equip-
ment can be used in a different application. In this
case, the number of actual stages is fixed, so the
engineer calculates what product recovery and
purity can be obtained with the new operating
conditions. It may be possible to vary phase flow

rates, but only within the limitations of the equip-

ment. Changes in temperature or pressure may
favorably affect the phase equilibrium to obtain
the desired performance. A new solvent, adsorbent,
or absorbent may improve recovery and make it
possible to use the existing equipment. In addition
to the principles considered in this chapter, the
synthesis and analysis of industrial equipment re-
quire a knowledge of fluid mechanics and mechan-
ical design, as well as a great deal of previous
experience.

OVERALL BALANCES
Mass and enthalpy balances around an entire series
cascade) of stages are referred to as overall
(or
balances. Only the streams entering or leaving the
Figure 5.1. Liquid extraction of lubricating oils. The two process as a whole are involved in an overall
towers to the right of center are rotating-disk contactors in
balance.
which a solvent such as propane is used to separate lubri-
Overall balances may be taken around the
cating oilfrom undesirable asphalt components using
generalized countercurrent cascade shown in Fig-
liquid-liquid extraction. The solvent is fed to the bottom of
ure 5.2. The symbols L, x, and h refer to the mass,
the towers, which operate in parallel, and the lubricating oil
composition, and enthalpy of the L phase at any
stock is fed to the top. The solvent selectively dissolves the
point; and the symbols V, y, and H refer to mass,
lubricating oil, leaving the asphalt to flow out of the bot-
tom. composition, and enthalpy of the V phase. If the
The extract from the top is fed to the stripping column mass is expressed in pounds (or kilograms), the
to the left of center in the picture, where the solvent is composition must be in mass fraction and the
flashed off in a reboiled stripping operation. (Courtesy enthalpy in Btu/lb (or joules/kg). If the mass is in
Foster Wheeler Energy Co.) pound-mole (or mole) units, the composition is

ff] Hz Hv tf.v+i

y\ *> yN y.v+i

Vl V2 vN *w,
< 1 i-r
Stage Stage Stage Stage
1 2 J"L
>i >
2V-1 N
L„
^0 Z-1 L.\- /
—> '- .,
1 \

*o x\ .a x \-i xN
*0 hi *A '!
\-i hN

Figure 5.2. Countercurrent multistage contact. Normally the desired solute is carried
in stream Lq; therefore it is thought of as the feed stream, F.
62 STAGE OPERATIONS

in mole fraction, and the enthalpy is in Btu/lb

mole (or joules/mole). The numerical subscripts

(0, 1, . N) refer to the stage from which the
. . ,

stream is flowing. Where it is necessary to distin-

guish between components, a subscript (a, b, ore)
is used.
An overall balance may be written for any
conserved property in the cascade. These typically *<••*

include the total mass, the mass of any one com-

ponent, and the enthalpy. At steady state, which is
assumed here, the quantity entering must exactly
equal the quantity leaving:

Total mass balance:

L + \Z/v + 1
= Z = LN + V y I 5.1)

Component mass balance:

Figure 5.3. Graphical determination of the
L xq + ^/v-hK/v-m = 2z £ = L N x N + V,yi overall mass balance.

(5.2)
Enthalpy balance: 5.5. The equations are simply analytical expres-
sions of the addition rule developed in Chapter 4.
L hQ + l//v + 1
HN + i
= 2/? r = LN hN + V :
H^ Examination of the group on the right side of
(5.3)
Equation 5.4 shows that it is simply the sum of the
masses of one component divided by the sum of
This enthalpy balance assumes that no heat is the total masses of the two streams L N and V-\
added or removed through the walls of the cascade. Similarly, Equation 5.5 gives the sum of the enthal-
A component balance (Equation 5.2) maybe pies of the two streams divided by the total masses
written for each of the components present. The of the streams. Calculations utilizing Equations 5.4
left-and right-hand expressions in Equations 5.1, and 5.5 are often more easily accomplished graph-
5.2, and 5.3 define the overall balances. The center ically. The procedure is shown in Figure 5.3. For
term in each equation is a definition of mass (2), example, perhaps the masses and composition of
composition (z 2 ), and enthalpy (/? 2 of the ficti- ) L and V/v + are known. It is then possible to
1

tious sum 2. The sum is fictitious because, for locate z £/ as shown. Then, if x N is known, y can y

example, L never actually added to VN +

is in a y be determined by drawing a straight line between
physical sense. This sum is useful in solving overall x N and z z and extending it to intersect the extract
balances graphically, as discussed later. curve. The masses of L N and V-\ can be determined
The composition of the fictitious sum is by applying the inverse lever-arm rule to the line
defined by solving Equation 5.2 for z r and then X/vZv /i Only Equation 5.4 is needed to make the
.

substituting for I from Equation 5.1. overall mass balance for the three-component sys-
tem of Figure 5.3. On the other hand, for calcu-

Zz — ——V
^-0*0 :

Lo +
+ ^/v + iK/v +

N+ ,
i
L N xN +

L„ +
\/ 1

\A
/ 1

(5.4)
lation on an enthalpy-composition diagram, both
Equations 5.4 and 5.5 are relevant because both
mass and enthalpy balances must be made.
Similarly, the enthalpy of 2 is obtained by com-
bining Equations 5.1 and 5.3.
STAGE-TO-STAGE CALCULATIONS

h-z
L h
;
—— + VN + ,/-/„ + ! L h + V H,
= N N y
(5.5)
In one type of calculation for countercurrent flow,
Lq + V//v + i U
+ V, enough information is available to calculate the
overall mass balance, as outlined above. After an
The composition and enthalpy of the fictitious overall mass balance has been used to determine
sum can be represented by a point on a diagram the end conditions of the cascade, stage-to-stage
showing the phase equilibrium of the system (such calculations may be made to determine the number
as Figure 3.4).Such a sigma point may be defined of equilibrium stages required to give the desired
using the coordinates derived in Equations 5.4 and end conditions.
 COUNTERCURRENT MULTISTAGE OPERATIONS 63

Calculation may proceed from either end of useful in calculating from stage to stage, as out-
the cascade. If, for example, y and Hy are known, y
lined in the following discussion.
use of the equilibrium relationship will give x^ and Because the net flow is defined as being
/?1 . It is now necessary to develop an interrelation- positive to the right, a net flow to the left has a
ship between x^ , h-\ and y 2l H2 . The two streams negative numerical value. For example, if VN + y
is

L-\ and V2 between stages 1

are passing each other greater than L Nl by Equation 5.6 the total net
and 2. In order to calculate the quantity and flow A has a negative value, so that the total net
composition of V2 from those of Ly the new flow is actually to the left. Physically, this means
concept of net flow must be introduced. Since simply that there is a greater flow of mass to the
there is no accumulation of a constituent, a left than there is to the right.
property, or a total quantity in the cascade, the net Equation 5.7 defines the net flow of one
flow is constant throughout the cascade. The net component. For example, if copper sulfate is to be
flow is a fictitious stream; it is the difference extracted from a roasted ore (Z. by water )

between two streams that are not in reality sub- (

V//v +1 )
there will be a net flow of insoluble
,

tracted one from the other. gangue and copper sulfate to the right and a net
Net flow to the right in the cascade shown in flow of water to the left. The term x A is a fic-
Figure 5.2 is defined as the difference between the titious composition of A that would result if V
flow to the right and the flow to the left, or were actually subtracted from L. Combining
Total net flow: Equations 5.6 and 5.7 gives

A = L -V, =L n - l/„ + =L N -VN + Ax A ^ L q xq - V,y, L n x n - Vn + yn + - -

XA =
i [

1 =
(5.6) A L -V, Ln -Vn + }

Component net flow:

L-n><n — ^/v + iK/v + 1

Ax A = L x - V^y, = L n x n - Vn + ^y n + -
[ L-n — \//v + i

= L N x N - V/v-nK/v + 1 (5.7) net flow of component

I5.9)
Enthalpy net flow: total net flow

A/7 A =L h -V,H =L n h n -Vn + ,Hn + The numerators of Equation 5.9 are the net flow
: ,

of any component, and the denominators are the

= Z./v/7/v — \//v + i ^/v + 1 (5.8)
total net flow. Therefore, if the net flow of a
Equations 5.6, 5.7, and 5.8 state that the net flow component is in a direction opposite to the total
is constant at any point in the cascade. For net flow,xA will have a negative value for that
example, Equation 5.6 may be proved by the component.
following brief development. The total net flow Application of the graphical-addition rule to
(A) of mass at the left end of the cascade is equal Equations 5.6 and 5.7 for the entire cascade shows
to the mass flowing to the right (L minus that ) that x A should lie on the straight line y-\x Q
flowing to the left Vy ). Similarly, at the right end
( extended and also on the line y N + N extended. -\

of the cascade, the total net flow (A) is equal to
Ln — Wv + 1 By Equation 5.1 •
^-o
— V-\
Therefore, the total net flow is the same at either
end of the cascade. The total net flow at any stage
(L n — Vn _
is easily shown to be equal to that at
1 )
either end by writing a mass balance between
either end and stage n. Thus, the total net flow is
constant at any point in the cascade. Equations 5.7
and 5.8 can be justified in a similar manner. The
concept of net flow is useful because the net flows
are constant through the cascade. Once they are
evaluated at either end of the cascade, the values
can be used at any point. There is usually no need
to determine the actual values of net flows in
typical graphical calculations, but the concept is
This is shown in Figure 5.4. For the case shown,
x Aa and x Ab are negative, and x Ac is positive and
greater than 1.0. This indicates that the net flows
= Ln — V/v + i
of the solute and major raffinate component are in
a direction opposite to the total net flow, but the
net flow of solvent is in the same direction as the
total net flow and is larger than the total net flow.
Even for the fictitious stream A, x a + x b + x c must
equal 1.0. Therefore, if x a and x b are negative, x c
must be greater than 1.0. The delta point located
by using the concentrations at the ends of the
cascade will be used to calculate from stage to
stage, as shown in the following discussion.
Equations 5.6, 5.7, and 5.8 also define the net
flow at any stage n in the cascade. It can be seen
that the component net flow is constant through
all stages at steady state, so that the delta point
64 STAGE OPERATIONS

which may be used on enthalpy-composition dia-

grams together with Equation 5.9 to locate the
delta point Calculation of the overall material and
enthalpy balances and calculation from stage to
stage on an enthalpy-composition diagram follows
the procedures outlined previously for ternary
composition diagrams. The enthalpy may be con-
sidered to replace the solvent.
The graphical procedure for calculating the
number of stages required, as outlined in previous
paragraphs, is used in problems of synthesis where
sufficient data are available on recovery and exit
compositions. It may be summarized as follows:
*c -\

1. Overall balances are used to determine un-

known exit compositions.
2. The delta point (x^ ) is located at the
intersection of Ki and x N y N +
x ,

3. Starting at y^ stages are stepped off using

,

the equilibrium tie-line relationship to find anyx„

from the corresponding y n and using the delta
Figure 5.4. Graphical determination of delta point to find any y„ + from the corresponding x„.
1

point and stage-to-stage calculation. Stepping-off of stages is complete when x„ equals

or is less than x N Stages may be stepped off
.

(x A ) is any point in the cascade. Therefore,

valid at starting at x N by reversing the above procedure
since A = L — V2 and Ax s = L^x — V2 y 2 the
y : ,
(i.e., find y n from x n and then find x„_i from
value of y 2 es on tne straight line x A x^ It must
l' .

y n ), and continuing until y n equals or exceeds/! ).

also be located on the phase envelope because V2
problems of analysis where the number of
In
is flowing from an equilibrium stage. The determina-

tion of the compositions of the phases in succes- stages fixed and the recovery is to be deter-
is

sive stages may now proceed by alternate use of the

mined, the graphic procedure is by trial and error.
equilibrium data and the delta point. The location
The steps are as follows:

of Xy from y by tie-line data and of y 2 from x^

: The sigma point (2 = L Q + V/v +
1. is loca- i )

by the delta point is shown in Figure 5.4. When y 2 ted, but neither y^ norx/v'isknown, so the overall
has been located, x 2 can be found, and so forth, by balance is not completed. But, y-i 2v and x N are , ,

alternate application of equilibrium data and the known to lie on a straight line.
delta point, until the composition x n of the raf- 2. An assumed, so that y y also
x/v is is
finate leaving a stage equals or surpasses the final located (by the addition rule). Stages are stepped
desired composition, x/y The stage-to-stage calcu-
.
off from /t to xN .

lation can also be made in the opposite direction, 3. The number

of stages is checked
calculated
from x/v to y The choice of direction depends
y
.
against the correct number. they are not equal, a
If
upon the original information given. In some cases new X/v is assumed, a new y^ is located, and stages
the percentage recovery in the extract of the solute are again stepped off and checked. This procedure
originally supplied is specified. It is possible to is repeated until an x N is assumed so that it gives
calculate the exit compositions from this infor- exactly the correct number of stages.
mation. 4. The recovery is calculated from the correct
Combination of Equations 5.6 and 5.8 gives exit compositions.
the enthalpy coordinate of the delta point,
A number of variations on these two basic
_ ^-o^o — Vi Hy _ L n h n — Vn + yHn + y
procedures are possible, but the variations involve
*~ Lo-V, Ln -V„ + the same basic relationships.
i

t-Nh N — l//v + i #/v + 1

Examples. 1. Acetic acid is to be extracted

LN '/V + 1 from 2000 kg/hr of a 40-percent aqueous solution
(5.10) by countercurrent extraction with 3000 kg/hr of
 COUNTERCURRENT MULTISTAGE OPERATIONS 65

pure isopropyl ether. The concentration of acetic envelope. x 2 is now determined as before from
acid is to be reduced to 3 percent in the exit equilibrium data, and the construction proceeds
raffinate. How many equilibrium stages are re- untilxn equals or is less than x N This occurs
.

quired to reduce the acetic acid concentration to 3 between x-l3 and x 14 . Therefore, approximately
percent? 13.5 equilibrium stages are required. The dashed
lines of Figure 5.5 are equilibrium tie lines for the
Solution. The complete solution is given in
stages of this example. The construction lines to
Figure 5.5. In this case there are three components
x A have been omitted two to avoid
after the first
to consider. Therefore, a three-component (ter-
obscuring the stage tie lines.
nary) diagram is used. y\ is located by a n overal l

The coordinates of the delta point can be read

mass balance. First, z z is located on line xny/v + i-
from Figure 5.5 or calculated from Equation 5.9 as
by the inverse lever-arm rule.
x a = -0.012, x b = -0.45, x c = 1.46. The negative
LQ _ 2000 _ //v + 1 -2V _ 2 values for x a and xb indicate that the net flows of
" 3' acetic acid and water are
in a direction opposite to
r
/V + 1
3000 *o
the total net flow. In this case the total net flow is
Then y^ must ie on the phase envelope on the
l
to the left. Therefore, the net flow of acetic acid
extension of \\nex N z z since L N + V-\ = 2.
,
and water is to the right. Even for this ficti-
The delta point is now located on the exten- tious composition, x a + x b + x c = —0.01 2 + (—0.45)
sions of the straight lines xq/i ar| d x N Yn + 1 The •
+ 1 .46 = 1
calculation of stages may start at either y^ or x N .

Starting with the final extract [yy ), Xy must be in

equilibrium with y^ and the value of x^ is deter-
DESIGN AND OPERATING VARIABLES
,

mined from the equilibrium curve as shown by the There are four important variables to consider in
construction. Since L^ — V2 = A, y 2 lies at the simple countercurrent operations:
intersection of the straight linex A x-| and the phase
1. Product purity. The concentration of com-

ponent a should be high, so that further purifi-

cation will not be difficult. For example, in
extraction the final extract should be rich in solute
(high y a1 ), so that further recovery of solute from
the extract is inexpensive.
2. Percentage recovery. It is desirable to re-
cover as much of component a as possible. A large
Vi Ki a or a small Z./vX/v a gives high recovery.
3. Number of stages. The more stages re-

quired, the greater is the cost of equipment.

4. VI L ratio. A high VI L ratio leads to dilute
V-phase products but high recovery.

Specifying any two of these four variables

fixes the other two. The choice of optimum values
for the variables requires an economic analysis con-
sidering cost of equipment, processing materials,
and further separations, if necessary. This discus-
sion has presumed that the desired component is in
the V-phase product. A similar analysis is possible
for desired L-phase products. In many cases, both
phases contain valuable products that must be re-
covered. It is rarely possible to achieve high re-
covery as well as purity of both products in simple
counterflow.

Example Develop and plot the relation-

5.2.
ships among the
following variables for the feed
Figure 5.5. Solution to Example 5.1. mixture and solvent of Example 5. 1
66 STAGE OPERATIONS

(a) Extract purity (the concentration of sol- gave more than five stages, and Trial 2 gave less

ute the extract / 1a ).

in than five stages. Trial 3 is correct, because it gives
(b) Solute recovery in the extract (percentage exactly five stages, as shown. Therefore, y 1a =0.148.
of solute entering that leaves in the extract). To find the recovery, V-\ must be determined. By
(c) Number of stages. the inverse lever-arm rule along the line y N + 1 Zv x :

(d) V/L ratio. 0.667 -

I/, = 6000 = 5030
This problem might be worked in a
Solution. 0.795-0
number of ways. The solution yields a graph of
Then,
extract purity and solute recovery as functions of
V/L, with number of stages as an independent 5030 x 0.148
parameter. As an example, the calculations for one Recovery = = 93%
2000 x 0.40
value of V/L are shown in detail. Assume
V/L = 2.0, and then assume that N = 5 stages. The Next, other values of N for V/L = 2 are calculated;
trial-and-error solution is shown in Figure 5.6. The additional values of V/L are then chosen. The final
compositions x and K/v + i are plotted and Zv is results are given in Figure 5.7. Examination of this
located.With V/L = 2.0, z,_ is 2/3 of the distance figure shows that a maximum purity of 25 percent
from x to K/v + 1 ( '* IS most easily measured on the is possible for this system, but this occurs at lower
vertical axis where z^ c
= 0.667). A straight line is The percentage recovery increases with
recoveries.
drawn through z z so that it intersects both the V/L, even though the product purity decreases
extract curve and the raffinate curve. The inter- with V/L. At large values of V/L, the purity and
section with the extract curve is y-\ and the inter-
section with the raffinate curve is x^. These
points, together with x and y^ , are used to locate
xA , and stages are stepped off starting at y-[. The
first straight line (labeled Trial 1 in Figure 5.6)

Trial 1
v*i

0.05 0.1 0.15 02 025 03 035 0.4

Figure 5.6. Solution to Example 5.2. V/L = 2.0; N = 5. Figure 5.7. Results of Example 5.2.
 COUNTERCURRENT MULTISTAGE OPERATIONS 67

recovery tend to approach asymptotic values, with
the recovery approaching 100 percent and the
purity approaching percent as V/L approaches
infinity. In actual operation, an economic balance
must be made between recovery and purity. A high
extraction recovery is desirable, but the solute then
must be recovered from a dilute extract — by distil-
lation, for example. This secondary recovery is
more difficult with dilute solutions. Obviously, it is
impossible to obtain both high recovery and high
purity in simple countercurrent flow for this sys-
tem. The next chapter considers techniques for
obtaining both high recovery and high purity.
Areduction in V/L moves the delta point
farther from the triangular diagram; an increase
shifts it closer. The minimum V/L may be deter-
mined by extending the tie lines in the direction of
the delta point. The extended tie line on which the
delta point first falls as it moves away from the dia-
gram determines the minimum V/L. It is difficult
to predict generally exactly which tie line will
determine the minimum V/L. It may be a tie line
through the feed composition, through the exit
raff inate composition, or at any intermediate point,
as demonstrated in the following example.
The fraction unrecovered of component a is
given by L N x Na IFz Fa so that the fraction re-
,

Minimum V/L covered is 1 — L N x Na IFzp a The terms F andz Fa .

An arbitrary choice of two of the four process do not change for a specified feed, but both L N
variables cannot always be made. There are limi- and X/va vary as the V/L ratio is varied. The prod-
tations imposed by the physical properties of the uct L N x Na must be held constant as V/L is re-
system being processed. For example, examination duced to find the minimum V/L. This is not easy
of Figure 5.7 shows that it is impossible to attain a to do. In some cases it may be easier to choose
60 percent recovery of solute at a V/L of 0.5, several values of V/L and determine for each the
because even an infinite number of stages gives recovery at infinite stages. If these recoveries are
only a 31 percent recovery. If a 60 percent re- plotted as a function of V/L, the V/L for the
covery is specified, the minimum V/L that gives desired recovery can be read from the graph, as in
this recovery is about 0.65, as shown in Figure 5.7. Figure 5.7. The procedure is shown in the follow-
The line for N = °o gives the minimum V/L for any ing example.
specified recovery and L.
In a typical extractionproblem, as the solvent Example 5.3. Determine the recovery ob-
flow (V) is deceased, more equilibrium stages are tained with an infinite number
of stages for (a)
required to give the specified recovery of solute.
V/L = 2.0 and (b) V/L = 1.0 using the feed and
Finally, a value of solvent flow
reached where an
is
solvent ofExample 5.1.
infinite number of stages is required to give the
recovery. This is referred to as the minimum V- Solution, (a) The sigma point (zv for V/L = )

phase flow or the minimum V/L. Although the 2.0 is located in Figure 5.8a. The procedure now is
minimum V/L has no direct practical use, it is a search for the tie line on which the A-point first
useful as a limiting value for the actual flow. Often falls as the mass balance line through 2 is pivoted.
the actual V/L will be taken as some arbitrary First, a number of tie lines are located and ex-
factor times the minimum. tended toward the upper left, because the delta
The minimum V/L occurs at the highest value point is likely to lie toward the upper left. Next, a
of V that gives a delta point lying on any extended trial X/v^v/, line is drawn (labeled Trial 1), and the
tie line.Such a coincidence results in no change of resulting Trial 1 delta point is located. Because it
composition from stage to stage and requires an lies to the left of any possible tie line in the
infinite number of stages to give a finite change in system, Trial 1 does not require infinite stages. The
composition. Because the lines representing the line TfJT^Y\ slowly rotated clockwise until the
' s
stages are crowded together, the point at which the delta point on an extended tie line. This
falls
infinite number of stages occurs is referred to as finally occurs at x Na =0.0, the left-hand vertical
the pinch. At values of V/L lower than the mini- axis of the diagram. With x Na = 0.0, the recovery
mum, the desired separation cannot be made, even of a is obviously 100 percent. This agrees with the
with an infinite number of stages. Many of the N = o° line in Figure 5.7.
subminimum values of V/L give delta points that (b) The sigma point for V/L = 1.00 is located
lie on an extended tie line, and it should be em- in Figure 5.8b, and the procedure outlined in part
phasized that the largest value of V/L giving a delta (a) is repeated. In this case, the delta point first
point on any extended tie line is the correct mini- falls on an intermediate tie line, as shown in the
mum V/L. final trial. From Figure 5.8, y 1a =0.235, and by
68 STAGE OPERATIONS

Final x

Trial 1:*

(a)

Figure 5.8. Solution to Example 5.3. (a) V/L = 2.0.

the inverse lever-arm rule I/-, = 2910, so that range from x to x Nl and they are extended to
intersect the line X/v Yn + 1 extended. The trial delta
(2910X0.235)
Recovery = = 85.5% point is the correct minimum delta point only if it
(2000X0.400) coincides with the extended tie line that intersects
The procedure used in the solution of the x n Yn + 1 nearest y N + -\. The explanation for this
previous example is complex because both L N and follows.
x N vary as the recovery remains constant. the If At a large value of V/L, a small number of
exit composition x N is held constant instead of the stages yields the desired exit composition, and the
reco very, the procedure is simplified, because the delta point is relatively near Yn + :- As V/L is
line X/vK/v + i
is now fixed. In this case, the mini- decreased, more stages are required, and the delta
mum V/L required to give the specif ied x N can be point moves away from the diagram. As V/L is
determined directly. First the line x N y N + is ex- -
[
decreased further, the delta point eventually co-
tended above the diagram. The minimum delta incides with an extended tie line. This delta point
point li es somewhere on this line. Next, a trial represents the minimum value of V/L that requires
x a/^l/i line is drawn through x N This locates a . an infinite number of stages to give the desired x N .

trial z z and y so that the trial delta point can be

y , If V/L is reduced below this minimum value, the
located. Now several tie lines are determined in the desired x N cannot be obtained.
 COUNTERCURRENT MULTISTAGE OPERATIONS 69

Trial 1:

Final xA

Figure 5.8. Solution to Example 5.3. (b) V/L

It
is not really necessary to make several librium stages. The only limitation is that the tie
trials, because the correct minimum x A can be lines must be in a composition range that actually
located mmediat ely at the intersection of the tie
i
exists in the desired separation; that is, the tie lines
line with x N y N + :that falls nearest //v + i, as illus- should lie between x and x N .

trated in the following example. The ratio V/L is It is instructive to start with a delta point that
evaluate d by apply ing the inverse lever-arm rule to is close to the equilibrium diagram, so that it gives
the line /w + 1 z 2 xo. a number of stages. Then the delta
small finite
It should be emphasized that any set of tie point can be shifted away from the diagram, and a
lines may be extended to find the minimum V/L. larger number of stages are stepped off. The delta
These tie lines do not represent specific equi- point is again shifted away and stages calculated.
70 STAGE OPERATIONS

As this procedure is continued, more and more gram gives the minimum V/L. All other extended
stages are required until the stages are so close tie lines intersect the extended line x/vV/v + 1 fa-
together that graphic procedures are mechanically ther from the diagram. The value of y^ may be
impossible. This procedure gives a close approach determined by drawing the line x A xn, as shown.
to the true minimum V/L. The quantity of solvent (V/v + i) is now deter-
mined by an overall material balance. Zv is located
Example 5.4. Determine the minimum sol-
at the intersection of x K/v + i
and *nV\ and the
<

for the feed and x N of Example 5.1 inverse lever-arm rule is applied.
vent rate

Solution. The location of the delta point in V/V + 1

x zz 13.6 length units
= 1.31
Figure 5.5 is for a solvent rate of 3000 Ib/hr. The 10.4 length units
^-o ^s//v + i

minimum solvent rate will give a delta point farther

away. Therefore, several tie lines are extended VN +i = 1.31 /.o =(1.31) (2000) = 2620 lb

toward the delta point in Figure 5.9. The line

Therefore, the minimum solvent flow rate is
*/vK/v + is the same as in Example 5.1, since the
i
2620 Ib/hr.
compositions in Example 5.1 also apply to this
example. The tie line that, when extended, inter-
sects theextended line X/v V/v + 1 nearest the dia- SOLVENT-FREE BASIS AND ENTHALPY-
CONCENTRATION DIAGRAMS
There is an important parallel between distillation
and liquid-liquid extraction. In distillation, adding
enthalpy (i.e., energy) to a liquid mixture generates

X& or (h^ x s
. )

Vwior-

0.1 0.2 0.3 0.4

Xa or x a

Figure 5.9. Solution to Example 5.4. Figure 5.10. Analogy between extraction and distillation.
 CO INTERCURRENT MULTISTAGE OPERATIONS 71

an upward- flowing vapor stream that carries the "\ + i O2/N2 equilibrium

more volatile component with it. In extraction, 8000 i „ c atm r

H-5 t\
1

adding a less-dense solvent generates an upward-

flowing stream that carries the solute with it.
Enthalpy in distillation and solvent inextraction
therefore serve a similar function — generating a
V-phase that is richer in component a than the
L-phase in equilibrium with it.
Calculation of two-component distillation on
an enthalpy-composition diagram closely resembles
extraction calculations on a solvent-free basis, as
shown in Figure 5.10. The diagram could apply to
a hypothetical case in either extraction or distil-
lation. Therefore, the diagram is labeled both for
extraction and for distillation (in parentheses) to
show the analogy. The solvent feed is shown to
contain a small quantity of the major raffinate
component, so that YN + ^ does not lie at infinity.
The vapor feed in distillation that corresponds to
this point would be a superheated vapor composed
of pure, less-volatile component. The raffinate feed
is shown to contain a small quantity of solvent

(X ). Analogously, the liquid feed in distillation

"contains" a small quantity of enthalpy (h ).
Distillation calculations utilizing the enthalpy-
composition diagram are usually referred to as the
Ponchon-Savarit method (1, 2).
The "adsorbent-free" diagram given in Figure
3.13 may also be used in calculation in the manner
outlined above.

Example The following two streams are

5.5. i

contacted simple countercurrent sieve tray

in a
column equivalent to four equilibrium states:
Figure 5.11. Solution to Example 5.5.

200 moles saturated 79 mole percent N2

liquid: 21 mole percent 2 close enough. From the figure, X/v=0.31,
100 moles superheated 20 mole percent N2 yy = 0.93, and by the inverse lever-arm rule on
vapor: 80 mole percent 2 x N z^y-\, V, =135, L N = 165. Note that four stages
/7=8000J/mol accomplished nearly as much as an infinite number
could (y, = 0.94), but even an infinite number of
Calculate the quantities and compositions of the
stages does not produce high-purity oxygen
resulting phases.
(x/v = 0.30).
The points for VN + and L are
Solution. i

plotted and the 2 point is located 2/3 of the REFERENCES

distance from \//v + to L The line connecting the
i
.

points for V^ and L N passes through the 2 point, Ponchon, M., Tech. moderne 20 (1921).
1. 13, p.
but too little information is given to plot either 2. Savarit, R., Arts et metiers, pp. 65ff. (1922).
outlet stream. The is trial and error, where
solution
(y-i.H-i) is and (x Nl h N
selected located. The
)

delta point is determined and stages are stepped off PROBLEMS

to see if exactly four fit between the two outlet
compositions. Figure 5.11 shows a trial solution 5.1. In a countercurrent extraction system, is xa 2
requiring slightly less than four stages, but it is likely to be less than or greater than x a 4? Explain.
72 STAGE OPERATIONS

5.2. If A is a fictitious quantity, why is it useful? (a) Calculate the number of equilibrium stages re-

Explain the physical meaning of x Aa in your own words. quired, without using graphic methods.

5.3. Develop an expression for Z^ using an overall (b) Calculate the number of equilibrium stages re-

material balance for stages 1 through n; repeat the develop- quired, using graphic methods.

ment for stages n + 1 to N. Are the two Zv's related? (c) What is the flow rate of fresh water to the cas-

Explain. Carry out the same development and comparison cade?

forx A .
(d) Calculate the total net flow at each stage.
5.4. Consider a typical countercurrent extraction cas- (e) What are the coordinates of the delta point?

cade. What is the usual direction for the net flow of (f) In what direction is the net flow of water? Of
component a? b? c? Under what exceptional conditions inerts? Of CuS0 4 ?
might one or more of these net flows be in the opposite 512. Halibut liver (700 kg/hr) is to be extracted in
direction? Can one predict the direction of the total net a countercurrent cascade with ether to recover the oil. The
flow, using the earlier conclusions about component net ether, which has been partially purified, contains 2 percent

flows? Can the total net flow be zero? What happens to the oil. The fresh livers contain 20 percent oil and are to be
coordinates of the delta point if the total net flow is zero? extracted to a composition of 1 percent oil (on a solvent-
5.5. Show with a diagram and equations that the free basis). 500 kg of solvent is to be used.
following statement is true: "A reduction in V/L will move (a) What percentage of the oil entering in the livers is

the delta point farther from the diagram; increase shifts it recovered in the extract?
closer." (b) How many equilibrium stages are required?
5.6. In your own words, describe the procedure re- (c) Calculate the mass and direction of the total and
quired to find a minimum V/L. What is the physical signifi- component net flows.
cance of this value? (d) Calculate the coordinates of the delta point.
5.7. Consider the countercurrent extraction cascade Underflow data are given in Figure 3.14.
shown in Figure 5.2. Suppose that another product is 5.13. It is proposed to separate liquid a from its

withdrawn from the extract phase at some stage, /, located solution with b by contacting the solution with
liquid

between 2 and N— 1. How does this additional withdrawal solvent c. Liquids b and c are completely immiscible with
affect the net flow? The A point? Develop a procedure to each other at all concentrations of a. Liquid a is soluble in b
calculate the number of stages required for such a situation. and c at equilibrium such that the concentration of a in
Call the extra product V/ and withdraw it from the extract liquid c is always equal to that of a in b.

flowing between stages / and / — 1 An existing mixer-settler unit is available for the
5.8. Derive Equations 5.7 and 5.8. separation. The unit is equivalent to five equilibrium stages.
5.9. From the equations defining the delta point, The feed solution is 20 percent a and 80 percent b and
derive must be processed at the rate of 1000 kg/hr. The fresh
solvent rate is limited to 1500 kg/hr by the design of the
equipment.
V y
x xA
(a) Plot the ternary diagram for the system a—b—c.
5.10. Copper sulfate in a roasted ore is to be leached Label the extract and raffinate loci and show typical tie
out with water in a continuous countercurrent extraction lines.

cascade. 1000 Ib/hr of ore containing 10 percent CuS0 4 ,

(b) What percentage of the component a in the feed
85 percent insoluble gangue, and 5 percent moisture will be can be recovered in the leaving extract, using the maximum
extracted with water. The final underflow sludge will con- permitted solvent flow rate?
tain only 0.15 percent CUSO4. The underflow sludge re- (c) What solvent flow rate would give 80 percent
tains 0.8 lb of solution per pound of gangue. Calculate the recovery of a?
number of equilibrium stages required for an entering sol- (d) Determine the coordinates of the delta point in

vent flow of 2000 Ib/hr, (b) 4000 Ib/hr. Compare the

(a) parts (b) and (c). In which direction are the total and
results with Example 4.4. component net flows for each case?
5.11. Copper sulfate in roasted ore is to be leached 5.14. A gaseous mixture of 50 mole percent
out with water in a continuous countercurrent extraction propane, 50 mole percent propylene is to be separated by
cascade. 100 tons/day of the ore containing 9 percent passing it countercurrent to silica gel in a rotating-plate
CUSO4, 86 percent insoluble gangue, and 5 percent moisture column. The gas leaving the top of the unit is to be 95 mole
will be extracted with water. The strong extract solution percent propane.
will contain 10 percent CuS0 4 and 90 percent water. Since (a) How many equilibrium stages would be required if

CUSO4 is the valuable constituent, 95 percent of it must be 350 lb of silica gel is supplied for each pound mole of gas?
recovered in the extract. The gangue retains 1.5 tons of (b) What minimum mass of silica gel per pound mole
water per ton of gangue. of gas will give a product with 95 percent propane?
 COUNTERCURRENT MULTISTAGE OPERATIONS 73

5.15. A countercurrent extraction cascade is The tung meal is so finely divided that some of it goes
processing an aqueous solution of acetic acid with pure out suspended in the overflow solution. This amounts to
isopropyl ether. The following data are available: 0.05 kg of solids per kg of solution. The underflow must
have a concentration no more than 2 mass percent oil.
Entering raff inate: 1000kg/hr
(a) How many equilibrium stages are required?
35 percent acetic acid,
(b) What is the percentage recovery of tung oil ex-
65 percent water
tracted from the meal?
Final extract: 2500 kg/hr
5.19. Acetic acid is to be extracted from aqueous
12 percent acetic acid
solution by countercurrent extraction with isopropyl ether.

To how many equilibrium stages is the cascade equivalent? The feed is 10,000 Ib/hr of 28 percent acetic acid.

5.16. The following information is available on a (a) Determine the number of equilibrium stages

multistage codliver oil extraction unit: required for a solvent feed rate of 30,000 Ib/hr and an
extract composition of 8 percent acetic acid.
Entering livers: 1000lb/hr (b) Determine the minimum solvent/feed ratio that
32.6 mass percent oil will give the same raffinate composition as in part (a).
67.4 mass percent inerts 5.20. A liquid containing 40 percent styrene and 60
Entering solvent: 2000 Ib/hr percent ethylbenzene is to be extracted in a countercurrent
1 mass percent oil cascade by diethylene glycol solvent. The raffinate is to
99 mass percent ether contain 10 percent styrene on a solvent-free basis.
Exit underflow: 1180 Ib/hr (a) Determine the minimum solvent to feed ratio.
1.14 mass percent oil (b) Using 1.3 times the minimum solvent to feed
41.0 mass percent ether ratio, calculate the number of equilibrium stages required
57.9 mass percent inerts on ternary coordinates.
Exit extract: 1820 Ib/hr (c) Repeat parts (a) and (b) using coordinates plotted
18.3 mass percent oil on a solvent-free basis.
81.7 mass percent ether 5.21. A mixture of 50 percent styrene, 50 percent
ethylbenzene is to be extracted in a continuous counter-
Calculate the three coordinates of the delta point.
current mixer-settler using pure dietheylene glycol as a
5.17. Isopropyl ether is to be used to extract acetic
solvent. Evaluate certain design limitations by answering
acid from an aqueous solution. An existing countercurrent
the following questions:
extraction column which is equivalent to four theoretical If an extract composition of 9 percent styrene
(a) is
stages is to be used.
specified, what is the maximum possible percentage
(a) Determine the flow rate of pure isopropyl ether
recovery of styrene in the extract?
solvent required for the following conditions:
(b) What is the maximum possible concentration of
styrene in the extract?
Feed flow rate: 1000 Ib/hr
(c) For an extract composition of 9 percent and 90
Feed composition: 35 percent acetic acid,
percent recovery of styrene in the extract, what is the flow
65 percent H2O
of fresh solvent for 100 kg/hr of feed solution? How many
Final extract composition: 1 percent acetic acid
stages are required?

(b) What percentage recovery of acetic acid is

5.22. A mixture of 50 mole percent ethanol, 50 mole
percent water contains a very small quantity of a nonvola-
achieved?
tile impurity that discolors the solution. It has been pro-
5.18.Tung meal containing 55 mass percent oil is
posed to concentrate and purify the ethanol by stripping
to be extracted at a rate of 3000 kg/hr, using 12,000 kg/hr
the ethanol from solution with an available supply of super-
of solvent. The solvent contains 98 percent n-hexane and 2
heated steam. The solution will flow downward in a bubble-
percent tung oil. The solution adhering to the insoluble
meal cap column countercurrent to the upward-flowing steam.
in the underflow was determined experimentally, as
tabulated below:
The solution is at 70° F and the steam at 350° F and 1 atm.
(a) What is the maximum ethanol vapor concen-
Composition of Solution Quantity tration that can be obtained with an infinite number of
Adhering, of Solution, equilibrium stages?
mass fraction oil kg solution/kg inerts (b) What ethanol concentration and recovery in the

0.0 vapor is obtained with a column equivalent to five equi-

2.0
0.2 2.5 librium stages with a steam to ethanol feed ratio of 1 : 2

0.4 3.0 mole/mole?

0.6 3.5 5.23. Consider a sieve-tray column into which is fed
74 STAGE OPERATIONS

saturated pure liquid ethanol at the top and saturated pure Your answer should include any required calculations

water vapor at the bottom. Molar flow rates of ethanol and and a careful account of the reasoning used in arriving at

water are equal. your conclusions.

(a) How many equilibrium stages are required to give
a product vapor composition of 80 percent ethanol?
SOLUBILITY DATA FOR THE SYSTEM DOCOSANE-
(b) What is the maximum vapor composition attain- DIPHENYLHEXANE-FURFURAL AT 45 °C.
able? Briggs. S. W., and E. W. Comings, Ind. Eng. Chem. Vol. 35,
(c) Calculate the coordinates of the delta point for No. 4, pp. 41 1-417 (1943).
part (a).

(d) Does this appear to be an economically feasible Phase Envelope

process for producing 80 percent ethanol solution? Mass Fraction Mass Fraction Mass Fraction
5.24. The following two streams are fed to a simple Furfural Diphenylhexane Docosane
countercurrent multistage contactor at normal atmospheric
pressure: 0.040 0.0 0.960
0.050 0.110 0.840
Saturated liquid: 100 moles/sec of 79 mole 0.070 0.260 0.670
percent N2, 21 mole percent C"2- 0.100 0.375 0.525
Saturated vapor: 100 moles/sec of 30 mole 0.200 0.474 0.326
percent N2, 70 mole percent C"2- 0.300 0.487 0.213
0.400 0.468 0.132
(a) What are the outlet compositions for a contactor 0.500 0.423 0.077
equivalent to four equilibrium stages? 0.600 0.356 0.044
(b) What are the outlet compositions for an infinite 0.700 0.274 0.026
number of stages? 0.800 0.185 0.015
(c) If the vapor flow rate is changed to 200 moles/ 0.900 0.090 0.010
0.993 0.0 0.007
sec, what are the outlet compositions for an infinite num-
ber of stages? Tie Lines
5.25. A new solvent extraction unit has just been
(Furfural phase 0.891 0.098 0.011 in equi-
|
installed, and it is not giving the desired separation. The (Docosane phase 0.048 0.100 0.852 librium
j

unit was designed to extract 10,000 Ib/hr of a 30 percent

) 0.736
Furfural phase 0.242 0.022) in equi-
diphenylhexane— 70 percent docosane mixture with
(Docosane phase 0.065 0.245 0.690 j librium
20,000 Ib/hr of solvent. It was desired to produce an ex-
[ Furfural phase 0.523 0.409 0.068| in equi-
tracted docosane phase with only a 1 percent concentration
\ Docosane phase 0.133 0.426 0.439/ librium
of diphenylhexane, but the actual concentration during the
test run was much higher. Since a diphenylhexane concen-
tration in the extracted docosane above 1 percent is un-
5.26. The equations describing a simple leaching
acceptable, you must make a recommendation so that the
process (such as that included in Problem 5.11) can be
operating engineer can make the desired separation. Of
easily solved using a digital computer. From the following
course, it is necessary to maintain the feed rate of
given information, prepare a calculational procedure to
10,000 Ib/hr, and no more than 20,000 Ib/hr of solvent can
solve for the number of stages (N) required for the specified
be used, since the unit can handle no more. The solvent
separation. Your procedure should list all equations used,
used in the test was 98 percent furfural and 2 percent
solved explicitly for each variable needed. In addition,
diphenylhexane, recycled from the solvent recovery system.
include a step-by-step description of how the procedure
The unit has 15 actual stages and the overall efficiency is
a computer
should be organized complete enough so that
about 30 percent.
programmer could program and run the problem, even
(a) Utilizing the principles and calculations of liquid-
without a background in chemical engineering.
liquid extraction, determine why the desired separation was
Known quantities:
not accomplished in the test run.
(b) Suggest a change in operating conditions to give
1-0. *0- V/V+1.//V+1. K1
the desired separation with no modification in the extrac-
tion equipment. Tell how you would calculate to determine Underflow: m lb solvent/lb inerts.

that your suggestion is valid before it is tried. Find: N; percentage recovery.

 Six

Countercurrent Multistage
Operations with Reflux

In many separation processes, simple counter- the old feed? The term reflux arises from the fact
current flow cannot give as complete a separation that the new L is produced from the \Z, product,
as required. The best possible separation in simple as indicated in Figure 6.1, so that part of the
counterflow is obtained when an infinite number product is returned to the cascade. The unit C may
of stages is used, as discussed in Chapter 5. Even be called the V-phase refluxer, in general terms.
with infinite stages, it is usually not possible to This unit is not an equilibrium stage as it is shown
obtain both high purity and high recovery in sim- in Figure 6.1. Generally it removes most of the
ple counterflow, as illustrated in Example 5.2. In- separative component (component c) of the sys-
finite stages may result from the coincidence of a tem, and returns a mixture consisting mostly of
delta point and a tie line at any point in the component a with a small quantity of b. Thus, the
composition range from x to x N If, for example,
. stream Z, contains the same components as the
the coincident tie line is the one through x ,then feed, but with a in a higher concentration. The
with an infinite number of stages, the composition higher concentration makes it possible to produce
of V-\ is that in equilibrium with L This gives the
. a Vi with a higher concentration of a. The V-phase
highest purity \/-phase product (V^) attainable; product is withdrawn as stream D so that it has
because L is rarely highly concentrated, neither is exactly the same composition and properties as L .

V) If the coincidence between the delta point and

. The feed must be added to the cascade at a
a tie line occurs at some intermediate point in the new location. The best place is at a stage where the
x —x N composition range, the purity of V-\ is composition and properties of the stream flowing
lower than in the case discussed previously. to the stage most closely resemble those of the
It would be possible to increase the purity of feed. That is, it is best to mix the feed with a
Vi by using an L feed that was of greater purity, stream that is identical with it. In that way neither
but, of course, the feed composition is fixed by the feed nor the stream is diluted. The feed usually
other considerations. The use of reflux involves does not exactly match a stream flowing between
replacing the impure L feed with a more concen- stages, so the feed is added to the stream whose
trated stream, so that the V-\ product can be more composition and properties are most like those of
concentrated. Two questions immediately arise. the feed. The addition of the feed is shown in
What is the source of the new L ? What happens to Figure 6.1, where F is the feed stage.

Enriching
section

D A-^*- F A-

Figure 6.1. Multistage cascade with /.-phase reflux.

75
76 STAGE OPERATIONS

The cascade can now be split into two sec- flux in extraction Equipment designed to
is useless.
tions, one from stage and the other
1 to stage F— 1, would only serve to mix
reflux part of the raffinate
from stage F to stage N. Each section by itself is a the fresh solvent with the refluxed raffinate. Such
simple countercurrent cascade. Because the section simple mixing is no different from the mixing that
from F— 1 to 1 effectively enriches the V-phase in takes place in stage N.
component a, it is called the enriching section. The single reflux of Figure 6.1
is rarely used

Similarly, the from F to N strips com-

section in distillationbecause it requires the addition of an
ponent a from the /.-phase and is referred to as the extraneous vapor stream, VN + The single reflux-
[
.

stripping section. Any stage in the enriching section is occasionally applied when water is to be sep-

is designated n, and any stage in the stripping arated from a more volatile component; VN + -\ is

section is labeled m. As shown below, the net flows then steam. In distillation, reflux is employed at
are different in the two sections. both ends of the cascade, as indicated in Fig-
ure 6.2, in order to produce distillate (D) and bot-
Applications of Reflux toms (B) of high purity. The vapor reflux is pro-
Reflux is in liquid extraction and
most widely used duced by adding heat in the reboiler (S) to part of
although the principle is applicable to
distillation, the liquid flowing from the cascade to vaporize it.
any countercurrent multistage operation. The liquid reflux is produced by removing heat
In liquid-liquid extraction, extract reflux (L ) from the vapor product to condense it. All of the
can be employed to obtain a greater recovery of vapor may be condensed in the condenser (C), but
solute in the extract. In this case, solvent is re- only part of it is used as the liquid reflux.
moved from v7 ! in the solvent separator (C) to In distillation, heat (q s ) is added at the still

produce a liquid (L c of low solvent concen- ) (S) to vaporize part (L S - \

) of the liquid flow (L N )

tration, part of which is extract product (D) and from the bottom stage. The vapor reflux (Vs)
part extract reflux (L ). The solvent separator formed in the still rises through the stages, approach-
usually is a distillation column. Extract reflux is ing equilibrium with the liquid downflow at each
particularly beneficial for ternary liquid systems stage. Ultimately, the vapor from the top stage
with two pairs of partially miscible components, (V-i is condensed in the total condenser (C); part
)

such as the diethylene glycol-styrene-ethyl benzene isremoved as distillate product (D) and the re-
system (Figure 3.9). In such systems it is possible mainder (Lq) is the liquid reflux. In binary distilla-
to obtain separation into an extract containing a
negligible amount of unextracted raffinate com-
ponent and containing a negligible quan-
a raffinate -Qc Condenser
tity of solute. On the other hand, a system with
only one partially miscible pair, such as isopropyl
ether-acetic acid-water (Figure 3.8), always yields
one product phase with considerable quantities of
both solute and unextracted raffinate component.
Enriching
In this case, reflux is not so beneficial because
section
separation becomes increasing difficult as the plait
point is approached.
In liquid extraction, solvent (W/v + i ) is added
at the last stage (N). The solvent added passes
through the stages, coming to equilibrium with the
raffinate at each stage. Ultimately, the extract Vy (

from the first stage is sent to the solvent separator, Stripping

where the solvent is removed. Part (D) of the section

remaining solute-rich liquid is the extract product,

and the remainder (L Q is the extract reflux. With
)

this arrangement, the concentration of solute in

the V-phase is increased over that in equilibrium
with the feed in stages F— 1 to 1. Therefore, the Reboiler
(still pot)
resultant extract product may have a much higher
concentration of solute than is possible in simple Figure 6.2. Multistage cascade with reflux at
counterflow (Figure 5.2, feed = L ). Raffinate re- both ends for distillation.
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 77

tion, the succession of stages above the feed extraction, so must the heat contents (enthalpies)
enriches the vapor in the more
component, volatile of the liquid and vapor be considered in distil-
and the stages below the feed strip the more volatile lation.
component from the liquid downflow. Reflux may constitute a large fraction of the
The reflux rates of streams L and Vs as well , material reaching the ends of the cascade. The
as the number of equilibrium stages, determine the products would then be but a small fraction of the
degree of separation of the two components pres- phases circulating internally within the cascade.
ent in a given feed. In extraction there is a net Reflux may be employed in other stage oper-
flow of solvent carrying the extracted solute from ations where it proves beneficial. For example,
stage N to stage 1. Similarly, in distillation there is certain separations in gas absorption and adsorp-
a net flow of heat that might be thought of as tion are improved by using reflux. An industrial
"carrying" the more volatile component from stage separation process using reflux is shown in Figure
N to stage 1. Just as the solvent distribution be- 6.3.
tween extract and raffinate must be considered in In distillation there is enough difference in den-
sities of the L- and V-streams to establish gravi-
tational counterflow, but in liquid-liquid extrac-
tion no such predictable pattern exists. It is not
unusual in liquid-liquid extraction that interstage
pumping is required. The drawing of cascades in
horizontal (Figure 6.1) may approx-
orientation
imate reality in extraction in some cases, but distil-
lation cascades are oriented vertically (Figure 6.2).

STAGE CALCULATIONS WITH REFLUX

In a typical design calculation the quantity and
composition of the feed and products may be
specified. the reflux ratio (L /D) has been
If

chosen, the required number of equilibrium stages

for the separation may be calculated. Another sit-
uation often encountered involves prediction of
the operation of existing equipment in the separa-
tion of a new system. Usually the construction of
the cascade allows the variation of reflux ratio over
a limited range. Then the possible product com-
may be calculated at the
61,6
N- B^Wlfe^j Ojg^n^ positions for a given feed
allowable reflux ratios.
The graphical procedure will be developed first
for liquid extraction with a single reflux and then
for distillation with two refluxes. In either case,
the cascade with reflux may be considered as two
Figure 6.3. Extraction of aromatics from other hydro- simple countercurrent cascades joined at the feed.
carbons. The tall twin columns in the center of the picture Consideration of the net flows yields delta points
are used to separate aromatics (benzene, toluene, and for both sections of the cascade. Stage-to-stage
xylenes) from other hydrocarbons using liquid-liquid calculation in either section is identical to the
extraction. The solvent (such as diethylene glycol in water)
procedures developed in Chapter 5. The delta point
is fed to the top of the columns and the hydrocarbon feed
determined by net-flow considerations in the en-
is fed at an intermediate point. Hydrocarbon reflux is fed
riching section is used for stage-to-stage calcu-
at the bottom.
lations from stage 1 until the feed composition is
The barely visible tower behind the right-hand extrac-
tion column is a stripper that removes aromatics from the
reached. Then a switch is made to the delta point
solvent. The other towers near the extraction columns are for the stripping section; calculations are continued
for further purification of extract and raffinate by extrac- until the final /.-phase composition is reached. In
tion and distillation. (Courtesy Universal Oil Products addition to the stage-to-stage procedures of Chap-
Company.) ter 5, it is necessary to develop procedures for
78 STAGE OPERATIONS

locating the two delta points from the data avail- evaluated. On the other hand, the net flows can be
able on streams flowing to and from the cascade. determined from the compositions and quantities
of the external streams.
The change of net flow that occurs at the feed
Liquid-Liquid Extraction
requires the use of anew delta point. In the enrich-
The number of equilibrium stages required for a ing section, the delta point implied by Equation
given separation can be calculated using the con- 6.1 must be used to calculate from stage to stage.
cept of net flow. For extraction, the net flow in For the stripping section, the delta point implied
the enriching section in Figure 6.1 is
by Equation 6.2 should be used. Equation 6.5
A= Z. -l/i =L„- Vn + , (6.1) indicates that the compositions of A, A, and F lie
on a straight line. Calculations may be carried out
where n is any stage in the enriching section.
on ternary or solvent-free coordinates. The pro-
Similarly, the net flow in the stripping section is
cedure is shown on ternary coordinates in Figure
A = LN - l//v + 1
=L m - Vm+ , (6.2) 6.4.
The two delta points may be located on the
where m is the stripping section. The
any stage in
ternary diagram using Equations 6.1, 6.2, and 6.5
introduction of feed changes the net flow between if the reflux ratio is known, but first it is necessary
stages F— 1 and F.
to locate the compositions of various streams on
In order to relate the net flow A to known
the diagram. The entering solvent V/v + i
may con-
data on the streams flowing from the cascade, a tain a small quantity of solute (a) if the solvent is
mass balance is made around the extract refluxer being reused after removal from the solvent sep-
(C):
arator. Similarly, the solvent {V c removed from )

V,=VC + Lc the extract in the solvent separator may contain

small quantities and b as impurities. The
of a
Because Lc = L + D
compositions and yc shown on Figure 6.4
//v + i

V, = VC + L +D indicate that V c and \//v + are impure solvents.

i

Frequently the Vc stream is recirculated and used

or L - V =-(/?+ Vc
y )

as the fresh solvent (V/v-n). But since a small

So that quantity of solvent (component c) is lost in the
exit raffinate, additional fresh solvent must be add-
(D+ Vc ) (6.3)
ed to Vc to give VN + so that y c and K/v +
-
[
,
are 1

An overall mass balance around the cascade in not necessarily identical. The feed F may contain
Figure 6.1 gives solvent, but it is shown by z F on Figure 6.4 as

F= VC +D + L N - VN+ , (6.4)

Combining Equations 6.2 and 6.3 with 6.4 gives

5 = A + F (6.5)

Equation 6.5 relates the net flows in the two

sections of the cascade. Addition of the feed in-
creases the net flow by an amount equal to F. The
feed may consist of a mixture of /.-phase and
\/-phase, such as liquid and vapor in distillation. In
this situation, the feed will split between the L-
(solvent)
phase and lAphase when it is introduced into the
cascade, but Equations 6.4 and 6.5 are still correct.
It should be emphasized that the magnitude of

the internal circulation (L and V) bears no direct

relation to the quantities of F, D, and B. It is
therefore impossible to predict the values of L and
V solely from the values of feed and products, in V.»/)>
contrast to the case for a simple countercurrent (solute)

cascade. It is necessary to know the ratio of reflux Figure 6.4. Graphical calculation of liquid extraction
to product before the internal circulation can be with reflux.
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 79

containing no solvent (z Fc = 0). In general, the
feed may be /.-phase or l/-phase or a mixture.
Therefore, its composition is denoted by z instead
of x or y. Because x N is the composition of a
stream flowing from an equilibrium stage, it must
lie on the phase envelope, as shown in Figure 6.4.
The compositions x Dl x c and x are identical,
since L c is simply split into D and L If the
solvent separation is not complete, L c contains a
small quantity of solvent, as indicated by x c on
Figure 6.4. By a mass balance around the solvent
separator, V-\ = L c + VCl and y^ may be located at
the intersection of the straight line / c x c and the
phase envelope, since V } is flowing from an equi-
librium stage.
The delta point (x A shown by Equations 6.
) is
and 6.3 to be on line x /i or Wne x D y c which are
rene, and the two products are to contain 90
percent and 5 percent styrene (on a solvent-free
basis). The ratio of extract reflux to product is to
be 10/1. The solvent from the solvent separator
contains a negligible quantity of styrene and ethyl-
benzene. How many equilibrium stages are re-
, quired? What is the solvent feed rate per 1000 kg
. of fresh feed?
Solution. The extraction cascade with extract
reflux is shown in Figure 6.1. The solution to the
problem is given in Figure 6.5, which is a replotting
of the equilibrium diagram of Figure 3.10. The
solvent-free coordinates are more convenient than
the triangular coordinates (see Problem 6.3). All
compositions and masses must be given on a
solvent-free basis.
,

coincident. The exact location is given by ZFa =0.40 ZFc =0

XDa =0.90 XDc =0
(6.6) XNa =0.05 XNc = equilibrium value
D x -y y
I /
= 0.008 (from Figure 3.9)

(when y Ca = 0) (6.6a) Because Vc and V/v + are pure solvents, i

Vc '
= 0, V/v + i' = 0, Y c = <» and YN+ c =co ^ -

These expressions can be derived from Equation Equation 6.6a may be rewritten for solvent-free
coordinates.
6.1, together with the equivalent expression for
component net flow. Y^-XA -

(6.6b)
The coordinates of the delta point (x A are )
D' XD -Y,
determined with Equation 6.6 by substituting the
known values of reflux ratio {L /D), x 0/ A, Equation 6.6b is easily derived by rewriting
y : ,

and D and solving for x A If y Ca = 0, Equation .

Equations 6.1 and 6.3 in terms of solvent-free
variables. The "total" net flow is
6.6a is used with (L /D), y 1a x 0a and x Da The , , .

other delta point (x A is located by use of Equa-

)
A' = Z. '-VV (6.1a)
tions 6.2 and 6.5, w hich sho that x A lies at the w
intersection of line x N y N + i and line x A z F Be- .
The component net flow is

cause the values of x N x A //v + i, and z F are, , A'X A = -

L 'X - V,'Y :

known, the two be drawn and x A

lines may
located at their intersection. After the two delta Because Vc is pure solvent, Vc =0, and by'
Equa-
tion 6.3,
points are located as shown on Figure 6.4, they
may be used along with equilibrium data to cal- A' = -D' (6.3a)
culate the number of stages required. x A is used to
Eliminating Vy between the ' first two of these
calculate from x D to z F and x A from z F to x N
, .
three equations and substituting — D' for A' gives
The method outlined above also holds for cal-
Equation 6.6b directly.
culations on solvent-free coordinates. The appropri-
The ratio L /D is given as 10, so that
ate equations may be rewritten in terms of solvent-
L '/D'= 10, and X A can be evaluated from Equa-'

free masses and mass ratios. The primary difference

tion 6.66, if Y can be determined. By a mass
:
involves a pure solvent stream with a composition
balance around the solvent separator,
of Yc = oo
l//= Vc + L c ' '
= Vc '
+ Lo' + D'

Example 6.1. A mixture of styrene and ethyl- Now X Ca = X Da = X0a = 0.9, so t hat Y y is located
benzene is to be separated by extraction with pure at the intersection of the line Y C XC with the
diethylene glycol in a countercurrent cascade with extract curve. Note that the coordinates of Y c are
extract reflux. The feed contains 40 percent sty- (Y Ca = 0.0, Y Cc = oo). This point lies infinitely far
80 STAGE OPERATIONS

Xc , ic

Figure 6.5. Solution to Example 6.1

 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 81

above the phase diagram, so any line drawn from Net flow of solvent = (34.2M-410) = -14,030 kg.
the phase diagram to it would be parallel to the Because there is no solvent in Z. or D,
vertical axis. Because Y Cc =°°, a vertical line is
Vc = 14,030 kg per 1000 kg feed
drawn at Xc , where it intersects the extract curve
lies /,.
After Y^ is located, application of Equation Distillation

6.66 gives Calculations for distillation are made on the en-

XA >- Y y
= 10(/, -XD )
thalpy-composition diagram. Heat added to a system
is defined as q, according to the usual convention
so that XA -
is — XD located by measuring (V^ )
in thermodynamics. Therefore, the heat added at
and then measuring off ten times that distance the reboiler, or still, is the reboiler duty (q s ), and
above V-i This delta point is valid for calculating
.
the heat removed at the condenser is the condenser
stages from 1 to F— 1, but the feed changes the net duty (-q c ).
flow for stages F to N.
The net flows of mass are defined by Equa-
Since S - A' + F (6.5a) tions 6.1 and 6.2:

the points X A ZF - , , and XA -

lie on a straight line.
Enriching: A=L - V, = L„ - l/„ +1 (6.1

In addition, Stripping: A = LN - Vs = Lm - Vm+} (6.2)

A' = ZV- V/v+i' (6.2a)
A mass balance around the condenser (Figure 6.6)
But \//v + i' = 0, and so X Aa = XNa , and XA is gives
located at the intersection of the line XA ZF and a
vertical line through XN
shown. Stages are now , as V =L Q
X
+D
stepped off using X A from stage 1 to the feed and
- or
XA -, from the feed to the end of the cascade. L - V, = -D
Figure 6.5 shows eleven stages in the enriching
section and about twelve stages in the stripping
Combining this expression with Equation 6.1 gives
section. A = -D (6.7)
required solvent flow rate can now be
The
determined, with a basis of 1000 kg of fresh feed. A similar analysis around the reboiler gives
Vc
and V/v-m' are zero, so an overall mass
'
A = B (6.8)
balance gives F' = L N + D'. The inverse lever-arm '

rule is applied to this mass balance: Combining the overall mass balance

F=D + B (6.9)
LN
'

=
XD 'Zr _ 0.9 - 0.4
= 0.59
~ with Equations 6.7 and 6.8 yields
F XD -XN ~ 0.9 - 0.05

Therefore A = A + F (6.10)

ZV=590kg This equation, which is identical to Equation 6.5,

relates the net flow in the two sections of the
With Equation 6.2a
cascade.
A' = L N = 590 kg '
The net flow of enthalpy may be defined for
From Figure 6.5, the enriching section.

XA C -
= -23.8 A/? A =L hQ - V,H y
(6.11

Therefore, the net flow of solvent = (-23.8) (590) =

-14,030 kg. The solvent in the raffinate is (590/
0.992)0.008 = 4.7 kg. Therefore, the solvent feed
rate is

YW+1 = 4.7 - (- 1 4,030) = 1 4,035 kg per 1 000 kg feed

L v,
At the solvent separator, since A' = A' + F to V,

A' = -410 kg
Figure 6.6. Enthalpy and mass nomenclature
X Ac =34.2 around the condenser.
82 STAGE OPERATIONS

where h A equal to the net flow of enthalpy per

is and an enthalpy balance around the condenser
unit of total mass net flow. For use in graphic gives
calculations h A may be related to q c by an en-
L h + D(h D -Q CD )
= V^^-l^
thalpy balance around the condenser (Figure 6.6).
The enthalpies of the streams entering must equal Therefore, L h +Dh A =V :
H^
the enthalpies of the streams leaving. The con-
L _ hA -H,
denser is a special type of heat exchanger, where (6.17)
the heat removed from the vapor in condensation D
is transferred to a coolant liquid through a metal Equation 6.17 is useful in determining
h A from the
wall. An enthalpy balance for the coolant gives reflux ratio. The and h may
values of L /D, Hu
—q c = enthalpy of coolant leaving be substituted into Equation 6.17 to evaluate h A ,

or h A may be located graphically using the equa-

— enthalpy of coolant entering
tion as a guide. For a total condenser, h = h D and
An enthalpy balance around the condenser then
gives A similar analysis of the stripping section
+Dh D yields the following relationships:
V :
H, =L h +(-qr c ) (6.12)

The quantity of heat added at the condenser per A = LN -Vs = L m -Vm+ , =B (6.18)
unit mass of distillate Q CD can be defined
net flow of heat
h\= ho — Qeo =
Q ~-
QCD' (6.13) total net flow of mass
D (6.19)
The enthalpy balance becomes where

V,H =L h y
+ D(h D ~Q CD ) (6.14) Q
QsB
"^
~ (6.20)
B
Combining this equation with Equation 6.11 gives
X& = Xg (6.21)
Ah A = L h -Vyh y
= -D(h D - Q CD )
Vs hN -
-h A
(6.22)
Solving for the enthalpy coordinate of the delta B Hs -h N
point gives
The direction of the net flows in the distil-
-D(h D -Q CD ) lation column should be considered. Distillation
"a : nD — Ucd (6.15)
columns are built vertically with the condenser at
the top and the reboiler at the bottom. Therefore,
net flow of heat
net flow is defined as positive in a downward
total net flow of mass direction in a vertical distillation column (Fig-
ure 6.2). Equation 6.7 shows that the total net
The x-coordinate of the delta point is obtained
flow of mass in the enriching section is upward and
from
hence has a negative numerical value (-D). Equa-
Ax A = LqX - 7
v Ki = !
-Dx D tion 6.8 shows that the total net flow in the

stripping section downward with a positive

is
-D
XA ~ .
XD ~ XD (6.16) numerical value Because heat is added at the
(B).
A reboiler and removed at the condenser, the net
flow of heat is upward through both the stripping
The two coordinates (h A ,x A of the delta point, )
and enriching sections and hence has a negative
as defined by Equations 6.15 and 6.16, can be used
numerical value. Therefore, h± is always positive
to locate the delta point when distillate quantity
and h A is always negative.
and composition and condenser duty are known. If
the reflux ratio is known, h A may be determined
The use of the equations derived previously
for graphical calculations is shown in Figure 6.7.
from it, analogous to the derivation of Equation 6.6
For the illustration, the composition and thermal
for extraction. A mass balance around the con-
denser gives condition of the feed, bottoms, and distillate are
assumed to be known. In addition, either the re-
LQ + D = V, boiler or condenser duty is known. The feed is
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 83

stepping-off of stages is continued using (x A ,/? A )

until zF is reached. Then the construction switches

to the delta point for the stripping section
(*a,^a)- The calculation is completed when the
liquid composition equals or exceeds the bottoms
composition x B The calculation of stages across
.

the feed stage as outlined is not rigorous, consider-

ing the enthalpy and phase characteristics of the
feed, but it is a sufficiently accurate approximation
and is generally used. About 4.6 stages are required

(x B , V in
in
the enriching section and about 4.2 are required
the stripping section. The steps in this calcu-
lation are identical to those outlined in Example
6.1 for extraction. The analogy between the two
operations should be carefully studied.
Figures 6.2 and 6.6 show a total condenser in
which all the vapor {V,) is condensed and then
split into distillate (D) and reflux (L ). In a partial
condenser, on the other hand, only the reflux is
condensed and the distillate is drawn off as a
vapor. Since the liquid reflux and vapor distillate
are in contact, they may approach equilibrium and
give as much as one additional equilibrium stage, as
shown in Figure 6.8. In actual practice, the liquid
and vapor may not have sufficient contact to equi-

Figure 6.7. Distillation calculations with vapor and

liquid reflux.
ne
shown as a subcooled liquid (z F , HF ). The distil-

late is bubble point (x D h D ), and the

a liquid at its ,

bottoms is its bubble point (x B ,h B ).

a liquid at
The reflux is also a liquid at its bubble point lA tv2
(x ,h ). Since h A = h D — Q CD and x A = x =x Di
the delta point for the enriching section can be
determined from the condenser duty. The delta
point for the stripping section is located by the line
x&zp extended and the vertical line x B = x A .

Stages may be stepped off from either end with use

of the appropriate delta point. In Figure 6.7 the
stages are stepped off from the top of the column
downward, by alternate use of the equilibrium data
h.H
and the delta point. The column has a total con-
denser /-, = x = x D and the point (/i H-^ can be
, , )

located. This point represents the vapor leaving the

The liquid leaving the first stage is in
first stage.

equilibrium with this vapor. Therefore, its compo-

sition (xt is determined from the equilibrium
)

curve, and its enthalpy is determined from its i h

composition and the saturated liquid enthalpy

curve. To locate (y 2 ,H 2 ), a line is drawn from *o • >a

(x^hi) to (x A ,/? A ). From (y 2 H 2 the equilib-, ) Figure 6.8. Partial condenser equivalent to one equili-
rium relation is applied to find (x 2 ,h 2 ). This brium stage.
84 STAGE OPERATIONS

librate, so that a partial condenser functions as
than one equilibrium stage.
Figures 6.2 and 6.7 show the bottoms with-
drawn before the reboiler, so that the stream enter-
ing the reboiler is totally vaporized. In most distil-
lation towers, the bottoms are withdrawn from the
less or water. The choice of medium is based upon the
intended operating temperatures. For example,
liquid nitrogen (— 300° F, 40 psig) is used as a
cooling fluid in the condenser of a methane purifi-
cation column in a liquid hydrogen plant.
In practice, the partial reboiler is used much

still, as shown in Figure 6.9. This results in a more frequently than the total reboiler, because
reboiler that may be equivalent to as much as the former contributes to the separation of com-
one equilibrium stage. In this book a partial con- ponents, and total vaporization of a liquid leads to
denser or a partial reboiler is assumed to be equal problems in heat transfer. Partial condensers are
to exactly one equilibrium stage. used where it is difficult to condense all the vapor
Reboilers and condensers are heat exchangers from the column because some of the components
of special design. The principles of design of heat are too volatile. Partial condensation might also be
exchangers are considered in later chapters. Steam used where a vapor product was required. On the
is frequently used as a heating fluid in reboilers. other hand, if a liquid product is specified, a total
The steam temperature must be sufficiently higher condenser would be used.
than the bubble point of the bottoms to give a
reasonable rate of heat transfer in the reboiler. Example 6.2. One hundred pound-moles per
Water is frequently the cooling fluid in con- hour of a 30 mole percent ethyl alcohol— 70 mole
densers. The water temperature must be suf- percent water solution at 140°F is to be distilled at
ficiently lower than the bubble point of the distil-
normal atmospheric pressure in a column with a
late to give adequate heat transfer to the cooling
total condenser and a partial reboiler. The distillate
water. Various heating and cooling media are used
is to have a composition of 80 mole percent ethyl
for temperatures that are not appropriate for steam
alcohol and it is to contain 95 percent of the
alcohol in A reflux ratio (L /D) of 2.2 is
the feed.
to be used. How many equilibrium stages are re-
quired? Where is the feed stage? What is the re-
boiler duty? The condenser duty?
Solution. F=100lb moles/hr, z F = 0.3,
xD = 0.8.Alcohol in the distillate is 95 percent of
that in the feed: (0.95) (.3) (100) = 28.5 lb moles

D =
28.5
= 35.6, B = 64.4, xB = —-
1.5
= 0.023
0.8 '
" 64.4
The points (x D ,h D ), (z F ,h F ), and (x B ,h B are )

located on Figure 6.10. The feed is a subcooled

liquid, as shown. The point (x A/ /j a ) is located

from Equation 6.17:
H.h
D Hi -h
So
H, =2.2(H, -h )

With Equation 6.1 0, hj is located at the inter-

section of the line h±h F and the vertical line at
x A -x B . The stages are now stepped off beginning
with the reboiler. h& is used from x B to z F and ,

h A from z F to x D Seven stages (plus the reboiler)

.

are required. The feed is to the second stage above

the reboiler.
Figure &9. Partial reboiler equivalent to equilibrium The reboiler duty is calculated using Equa-
stage. tions 6.19 and 6.20. From Figure 6.10, Q SB =
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 85

<*i.\*> 1,936,600 Btu/hr. The reboiler duty is somewhat

greater than the condenser duty because of the
subcooled feed.

DESIGN AND OPERATING VARIABLES

The following discussion is primarily concerned
with distillation, but many of the principles also
apply to liquid extraction or to any other separ-
ation process using reflux. The application to
extraction is mentioned after distillation is dis-
cussed.
For a feed of given composition and flow
rate, the separation at a given pressure into a distil-
late product and a bottoms product is dependent
upon the variables h F x B x D — q c q s L /D, and
, , , , ,

N. When any four of these values are specified, the

other three are thereby fixed and may be evalu-
25.000
ated. Of the many possible combinations of four
specified variables, a few represent most of the
20.000 common engineering problems.
the design of new equipment, the most
In
common case involves knowledge of h F x B and , ,

x D Lq/D is then chosen to give the most eco-

.

nomical combination of equipment costs and energy

costs.Equipment costs are proportional to the
number of stages N, and energy costs are directly
fxH ,V or
(x n ,h ) related to the reboiler duty (95), because it repre-
sents steam supplied to the reboiler. As the number
of stages increases, the reboiler duty decreases, so
that a minimum total cost of equipment plus
energy will exist.
the operation of existing equip-
In evaluating
ment or planning a new use for it, N is usually
in
fixed, and L /D, q s and — q c may not exceed
,

certain values determined by the size of the equip-

ment. The enthalpy and composition of the feed
are usually known, and the possible compositions
of the products are to be evaluated. Some flexi-
bility exists, because L /D, q s and — q c may be ,

-20.000 adjusted over a range below their maximum values.

Reflux
-25.000
0.2 0.4 0.6 0.8 As the ratio of reflux (L ) to distil lated product
Xa
(D) is decreased, more stages are required for a
specified separation. Examination of Equation

Figure 6.10. Solution to Example 6.2. 6.17 shows that a reduction in L /D for fixed
values of h and H causes the delta point to move
y

closer to Hy As the delta point moves toward H^

.
,

/?e
= 31,500 Btu/lb mole and g, = (64.4)
-^a a smaller change in composition between stages is
(31,500) = 2,028,600 Btu/hr. For the condenser obtained, that more states are required. For
so
duty, by Equation 6.15, — Q =h A D = CD —h extraction, Equation 6.6 shows that as L /D de-
54,400 Btu/lb mole (from Figure 6.10). Using creases for fixed values of x Ql x D and y the delta , y ,

Equation 6.13 yields -q c - (54,400) (35.6) = point approaches y, more stages are required.
;
86 STAGE OPERATIONS

If L /D is reduced to a point where an ex- points must lie on a straight line through z Fl so

tended tie line passes through either delta point, a this feed composition is used as a pivot-point for a
pinch occurs, and an infinite number of stages is line that intersects the verticals at x A = x D and
required to accomplish the specified separation. x a =x b- A very large L /D gives delta points far
This value of L /D is referred to as the minimum from the phase diagram and few stages (Figure
reflux ratio. An L /D below the minimum value 6.11). As L Q /D is reduced, the line through z F
cannot give the specified separation even with an pivots clockwise to give delta points nearer the
infinite number of stages. A reflux ratio higher diagram. Finally, the pivoting line intersects a tie
than the minimum requires a finite number of line at either x A =x D orx A = x B The first tie line
.

stages for the specified separation. intersected as x A and x A

approach the diagram
The minimum reflux ratio is determined by determines the minimum Lq/D because it specifies
the following systematic procedure. First, a num- the lowest L /D that gives the desired separation.
ber of tie lines are located in the composition range In the case shown in Figure 6.11, the tie line
from x D to x B (Figure 6.1 1). It is useful to choose through the feed determined the minimum L /D.
the one tie line that passes exactly through z F , as In general, any tie line may be determining, de-
well as several others on either side of the feed. pending upon the shape of the equilibrium curve.
The tie lines from z F to x D are extended upward As the reflux is increased, the number of
to intersect the vertical line at x A =x D The tie . stages required for a given separation decreases. At
lines from z F to x B are extended downward to total reflux no products are withdrawn (D = 0,
intersect the vertical line at x A =x B The two delta
. B= 0), no feed is supplied (F = 0), and a minimum
number of stages is required for a specified separa-
tion. such a case, Lq/D = °° and Vs /B = °°.
In
Under this condition both delta points lie at infin-
ity, and construction lines to the delta point are
vertical; in other words, y„ + = x n and ym +\ =
1

Distillation columns actually can be operated

at total reflux. Sometimes a column is started up
by charging sufficient liquid feed to fill all the
trays to the column. The reboiler vaporizes the
liquid it flows to
as it, and the resultant vapor
passes up through the column until it reaches the
condenser. The resulting condensed liquid flows
downward from tray to tray. Eventually, all the
trays are filled with liquid. If no more feed is
added and no bottoms or distillate withdrawn, the
column will operate at total reflux. With the num-
ber of stages known, it is possible to calculate the
steady-state composition of the top and bottom
The data at total reflux
liquid, as outlined earlier.
tell the operator what is the best separation pos-
sible in the column. Once the column has reached a
steady state at total reflux, distillate and bottoms
are withdrawn and feed is supplied at increasing

rates until the desired production is achieved. As

Figure 6.11. Determination of
distillate and bottoms are withdrawn, the product
purity falls off.
The concept of total reflux as defined here
has no meaning where reflux is used at only one
end, as in liquid extraction, because there is no
way to reflux the i-phase product.
The concepts of minimum reflux and mini-
mum stages at total reflux are useful in estimating
minimum L /D. the difficulty of separation, in setting the actual
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 87

operating costs for a specified separation. A typical

evaluation is shown in Figure 6.13. The optimum
design occurs at minimum total cost.

Example 6.3. Determine the minimum reflux

ratio and the minimum number of stages for the
separation of Example 6.2.

Solution. From Example 6.2, z F = 0.3,

x B =0.023, x D =0.8 are plotted on Figure 6.14.
To find the minimum reflux, tie lines in the enrich-
ing section are extended to intersect the vertical
ftX line at x D = 0.8. The tie line that intersects farthest
D
from the diagram determines h A for the minimum
Figure 6.12. Variation of number of stages
with reflux ratio for a specified separation.

operating reflux ratio, and in analyzing column

performance. Figure 6.12 shows how the number
of stages varies as reflux ratio is changed. The
minimum reflux and minimum stages serve as
asymptotes for the curve. Distillation columns are
normally designed to avoid the asymptotic regions
of the curve.
A higher reflux in distillation requiresfewer 20,000
stages but greater reboiler and condenser duties.
Because the heat supplied to the reboiler costs
money, an economic balance between operating 15,000
costs and initial equipment costs will determine the
best reflux ratio. Similarly, in extraction a higher
reflux ratio gives fewer stages but necessitates a 10,000
higher solvent rate and a greater load on the sol-
vent separator, thereby increasing operating costs. h,H
An economic evaluation can be made to determine 5,000
the optimum combination of equipment and

^*mm

-5,000

-10,000 -

-15,000

-20,000

-25,000
Number of stages

Figure 6.13. Determination of optimum de-

sign in distillation. Figure 6.14. Solution to Example 6.3.
88 STAGE OPERATIONS

ratio (see Figure 6.14). This/7 A yields (Lo/D)™^ = operate — if possible, choosing a pressure that will
0.93 by Equation 6.1 7. A check of the lines in the give reboiler and condenser temperatures in a range
stripping section shows that the tie line already that will permit the use of low-cost heating and
identified in the enriching section determines the cooling media. Low-pressure steam is often used
minimum reflux ratio. For the minimum number for heating in the reboiler, and air or water is used
of stages, both delta points are at infinity. Stepping for the cooling fluid in the condenser. These are
of stages from x B = 0.023 to x D = 0.8 yields relatively available and inexpensive. If a pressure is
/V min =5.2 (Figure 6.14). chosen so that the column operates below ambient
The occurrence of the pinch at the feed point, temperature, then the need for refrigeration to
as in Example 6.3, is characteristic of ideal solu- supply a cooling fluid will greatly increase energy
tions. For many the equilibrium
real systems, consumption and operating costs. Sometimes use
relationships pinch at other compo-
produce a of refrigeration is unavoidable, as in the distillation
sitions. This will be considered again in Chapter 7. of highly volatile gases, such as natural gas or air.
In setting the pressure, the designer must also con-
Subcooled Reflux
sider the thermal sensitivity of the materials being
A total condenser may not only condense, but also
distilled. Sometimes operation below atmospheric
cool the liquid reflux below its bubble point. Such
pressure is necessary to provide a temperature low
a reflux is referred to as a subcooled reflux. When a
enough to avoid thermal decomposition. For
subcooled reflux is fed into the distillation column,
example, in the distillation of crude oil, the
it causessome vapor to condense and thereby column processing the least volatile materials oper-
produces downflow greater than the reflux
a liquid
ates under a vacuum.
supplied. The ratio L/V is often referred to as the
internal reflux ratio. For the same value of the
Control
external reflux ratio subcooled reflux
(L /D), a
Distillation columns require careful control to en-
will produce a larger internal reflux ratio than will
sure stable operation and desired product compo-
a reflux at its bubble point. All the derived equa-
sitions. Industrial distillation columns are control-
tions hold for subcooled reflux, providing h is
led automatically using electric or pneumatic
properly plotted below the saturated-liquid en-
systems. Such control systems are often extremely
thalpy curve. It should be noted that subcooled
reflux does not change the net flow. A subcooled
complex and require a minimum of human super-
reflux in distillation is analogous to an extract
by an operator in a central control
vision, usually

reflux in extraction that contains a quantity of

room Control systems must be able to cope with
solvent less than would give an overall compo- unexpected fluctuations in process variables that
often occur, even inprocesses operating at a steady
sition lying on the phase envelope.
state. For example, the feed composition or
Entrainment thermal condition may change
because of fluctu-
Entrainment results when the heavier /.-phase is ation somewhere the chemical process. A sudden
in
not completely separated from the lighter \/-phase rainstorm can lower the ambient temperature sur-
before each leaves the equilibrium stage. The carry- rounding an uninsulated tower, thereby changing
over of /.-phase results in an effective V-phase the heat loss from the tower. Additional control
composition different from the equilibrium value. problems arise in the start-up and shutdown of a
If the ratio of entrained /.-phase to V-phase is chemical process containing distillation units. In
known, a new effective V-phase locus can be deter- such cases, more direct operator control may be
mined and used to calculate the required number necessary.
of equilibrium stages. Examples of entrainment of A
simple control scheme for a distillation
the /.-phase in the \/-phase include carry-over of column is shown in Figure 6.15. Many more com-

liquid with the vapor in a bubble-cap column in plex systems are possible and are frequently used.
distillationand the carry-over of insoluble solids A detailed discussion of control of distillation
with the extract in leaching. In liquid-liquid extrac- columns is included in Reference 1.

tion both the extract and the raffinate may be

entrained in the other phase.
INTERMEDIATE STREAMS
Pressure
In any mass-transfer operation, it is possible to
The designer of a distillation column has some have more than one feed or more than two product
choice of the pressure at which the column will streams. In distilling crude petroleum, for example,
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 89

propane-butane, gasoline, naphtha, kero-

light gas,
sene, gas
oil, lubricating-oil stock, fuel oil, and
asphalt may all be withdrawn from the column at
different points. Calculation of the distillation be-
havior of such multicomponent mixtures is ex-
tremely complex. A simple case of an intermediate
stream for binary distillation or extraction may be
handled by the methods already covered. An
example of this, with two feeds, is shown in Figure
6.16. The introduction or removal of an inter-
mediate stream (/) changes the net flow. Therefore,
a new delta point must be established. In Figure
@ Flow control
6.16 there are three different net flows and there-
Feed
(fixed rate)
@ Level control
fore three delta points: A for stages 1 through M,
® Pressure control A for stages / through FA, and A from stages F
@ Temperature control through N.
The equations for the total net flows are
A = Lo-V^=L n -Vn+ ,=-{D+Vc ) (6.1)

Steam A = L N -VN+: =L m - Vm + , (6.2)

Reboiler 2 = /./- /+ i (6.23)

Bottoms £. Condensate The subscript / any stage in the

refers to inter-
product
mediate section of the column between stage / and
stage F-1.
Figure 6.15. Distillation column control. The The net flow at the left end of the cascade is
system shown is a relatively simple scheme for
A. After the intermediate stream (/) is added, the
maintaining column stability and desired
net flow becomes
product The pressure of the column is
purities.
maintained at the desired value by controlling
I=A+/ (6.24)
the temperature of the cooling water in the
condenser. An increase in cooling water flow When the feed (F) is added, the net flow is

effectively increases the rate of heat transfer in

the condenser and lowers the condensation
A=A+F=A+/+F (6.25)
temperature, which in turn causes the pressure
The stream / may have either a positive value (a
to decrease.
feed) or a negative value (a product). Additional
The distillate purity is maintained by
measuring the temperature on a plate in the
intermediate streams may be calculated by defining
enriching section. For a binary mixture, the
additional net flows.
temperature uniquely determines the compo- There are three different net flows, and there
sition. If the temperature becomes too high, must be three delta points. The determination of
there is too much of the less-volatile component
in the enriching section, and the tempera-
ture controller increases the flow of reflux to shell surrounding the tubes. Liquid flows into
improve the separation and reduce the con- the bottom of the reboiler, and a mixture of
centration of less-volatile component on the liquid and vapor flows back to the column. The
plate in the enriching section. steam flow rate is controlled at a preset value
Liquid from the condenser is accumu- that effectively fixes the vapor reflux knowing
lated in a reflux drum. When the liquid level the feed flow rate. Alternatively, the steam
exceeds a preset value, the level controller flow could be controlled by the temperature on
opens the distillate product valve to a point a plate in the stripping section to control bot-
that will maintain the desired liquid level. toms product purity more closely.
At the bottom of the column, a liquid It is assumed here that the feed flow rate,

level must be maintained to ensure proper func- composition, and thermal condition are con-
tioning of the reboiler. In the thermosyphon trolled at some other point in the chemical

reboiler shown, liquid boils inside the tubes, process, although the control shown could cope
which are heated by steam condensing in the with minor fluctuations in feed conditions.
90 STAGE OPERATIONS

-9c or Is

1 ^n^ I- 1 1 F_1 N S
-J''U

Figure 6.16. Intermediate stream for distillation. For liquid-liquid extraction,

disregard unit S and add solvent as V/v+ 1 , Ka/ + 1 i nto N.

v v
•
c •
.v+

*6->»=0

Figure 6.17. Graphical determination of the delta points

for liquid extraction with an intermediate product
stream.

Figure 6.18. Crude oil distillation. This atmospheric

distillation column makes the first separation of crude
the three delta points is shown in Figure 6.17 for a
petroleum into several product streams. The feed crude oil
typical case. If the reflux ratio is known, x A may
is heated in a furnace shown to the left of the base of the
be located on the line Xn/i by use °f Equation
column. The feed enters the column
partially vaporized
6.6a. Once x A has been located, x A rnay be deter- through the large somewhat above the concrete support
line
mined by reference to Equations 6.2 and 6.25. ring. Gasoline and volatile gases leave the column at the
Equation 6.2 implies that x A will be on the line top through the large line and are conducted to a partial
x nYn + \. so this line is drawn on the diagram. condenser at the base of the column. Reflux is pumped
Equation 6.25 can be utilized to construct another back to the top of the column. Piping at various points
line that intersects x N y N + at x A Since there are
-
.
along the height of the column indicates intermediate
{

three terms on the right-hand side of Equation
6.25, the addition must be carried out in two steps.
First, / and F are graphically added to give the sum
2. The 2 and A may be added to give A (i.e., x A
will lie on the straight line x A zv
of this line with x N y N + i isx A .
The sum I may be considered as a "fictitious
feed." The sections of the cascade between 1 and /
and between F and N behave
streams, including naptha, kerosine, light diesel fuel, and
heavy dieselfuel. The heaviest components (tars and/or
asphalt) are withdrawn from the bottom of the column,
barely visible below the concrete support ring.
The column, which is 197 ft tall and 28 ft in diameter,
). The intersection processes 200,000 barrels per day of crude oil. Columns of
thistype typically use valve trays, as shown in Figure 2.10.
The scale of the picture can be appreciated by identifying
two maintenance workers on a catwalk about halfway up
as if there were a the column. (Courtesy Foster Wheeler Energy Co.)
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 91

single feed 2 = / + F at some point between / and and higher product purity than can simple countercurrent
F. The net flows in these two sections may be flow.

determined by combining / and F into the fic- 6.9. Convert Figure 6.4 to solvent-free coordinates
titious feed (2) and treating the cascade as the and outline the graphical calculation procedures required
usual case of one feed. Because the stream / was a for reflux using solvent-free coordinates. Write all of the
product in the illustration of Figure 6.16, the equations upon which the graphical calculation is based.
numerical value of / is negative and Zv lies beyond 6.10. With a feed of composition Xa = 0.3, Xb = 0.7,

z F on the line z t z F as in the usual case of subtrac-

,
and with pure entering solvent:
tion. The concept of the "fictitious feed" is aban- (a) What is the maximum attainable extract concen-
doned in order to determine x%, which is an ex- tration of a (on a solvent-free basis) for the system shown
pression of the net flow from / to F. The third in Figure 3.8?
delta point is located by inference from Equations
6.24 and 6.25 at the intersection of the lines x A z t 1. Without reflux.

and x^Zf. 2. With extract reflux.

The stages may now be stepped off using the

appropriate delta point in each section of the (b) Repeat (a) with the system of Figure 3.9 (or
cascade. 3.10).
Intermediate streams in distillation and other (c) Compare the results of (a) and (b).

operations are calculated in a similar manner. An 6.11. Outline the graphical calculations required to
intermediate stream in distillation may consist of determine the number of stages in ethanol-water distillation
only the addition or removal of heat. An example where only distillate reflux is used. Instead of using a
of distillation with intermediate streams is shown reboiler, pure saturated steam is fed into the bottom of the
in Figure 6.18. column countercurrent to the bottoms product. Develop
any necessary new equations.

REFERENCE 6.12. What is the maximum ethanol product concen-

tration attainable starting with an aqueous feed of 30 mole
percent ethanol. Feed is a saturated liquid:
1. Luyben, W. L, Process Modeling, Simulation, and Con- (a) Without reflux.
trol forChemical Engineers, McGraw-Hill, New York With
(b) distillate reflux.
(1973).
(c) With distillate and bottoms reflux.
(d) What is the advantage of (c) over (b)?
PROBLEMS 6.13. A solution of 95 mole percent ethanol, 5 per-
cent water, is fed to a distillation column that is equivalent

6.1. Evaluate the total and component net flows in to a large number of equilibrium stages. If the column is

Example 6.1. operated at a high reflux ratio, estimate the distillate and

6.2. Using the same rates and compositions of sol- bottoms compositions.
vent and feed as in Example 6.1, determine the maximum 6.14. Is it possible to operate a distillation column at

possible concentration of styrene in the extract product of total reflux? Why would one wish to run a column at total

a simple countercurrent cascade. reflux? Is it possible to run column at minimum reflux?

6.3. Locate the delta points of Example 6.1 on a Explain.

ternary diagram. Attempt to step off the required number 6.15. In a typical distillation column with reflux,

of stages on this diagram. what direction is the net flow of heat? Of the more volatile
R4. Prove that L N 'lF' = Xd Z f IXD XN component? Of the less volatile component? In what direc-
6.5. Derive Equation 6.66. tion is the total net flow? Be sure to consider both stripping

6.6. Derive Equations 6.18 through 6.22. and enriching sections.

6.7. Show that for a given feed composition it is
6.16. Why is h& negative while h A is positive in

possible to obtain a greater recovery of solute a and a more Figure 6.7?

concentrated extract product through the use of extract 6.17. Make a sketch of how a reboiler might be
reflux (that is, compare extraction with reflux to extrac- constructed:
tion without reflux and show the limiting concentrations (a) So that it is equivalent to an equilibrium stage.
that are possible with a large number of stages; this can be (b) So that it does not contribute any fraction of a

most easily done on a solvent-free basis). stage.

6.8. Show by use of a diagram that reflux at both 6.18. (a) The following data are known about a

ends of a distillation column can lead to greater recovery three-component liquid extraction system:
92 STAGE OPERATIONS

Extract curve: Xc = 4.0 — Xa . (c) Determine the minimum reflux ratio and the
Raffinate curve: Xc = 1.0 — Xa . minimum number of stages.
Equilibrium curve: Ya = 2Xa /(1 + Xa ). 6.22. A gaseous mixture of propane and propylene
can be separated by fractional adsorption using silica gel at
Determine the number of stages required to make the
1 atm pressure. The silica gel will be fed into the top of the
following separation:
column and will pass countercurrent to the gas. At the
Feed: 1000lb/hr; Xa = 0.4, Xb = 0.6 bottom of the column all the adsorbed gas will be stripped
Extract product: Xa = 0.8, Xb = 0.2 from the silica gel. Part of the gas will be withdrawn as
Bottoms product: Xa = 0.05, Xb = 0.95 product, and part will be fed back into the column as
Entering solvent: Ya = 0, Ye = 4.0, V„ +1 = 500 Ib/hr reflux.The feed contains 60 mole percent propane, and the
Extract reflux ratio (LqID) = 3.0. product streams are to contain 90 percent and 5 percent
(No raffinate reflux) propane.
(a) What is the minimum reflux ratio?
(b) The following data are known about an ideal
(b) How many equilibrium stages are required at
binary liquid mixture:
twice the minimum reflux ratio?
Saturated vapor enthalpy curve: H= 4000 — 1000x a (c) How much silica gel per pound mole of feed is

Saturated liquid enthalpy curve: h = 1000 - 1000x a required at twice the minimum reflux ratio?

Relative volatility = 3.0 6.23. A flow of 100 lb moles/hr of a vapor contain-

ing 0.40 mole fraction NH3 and 0.6 mole fraction H2O is

Determine the number of stages required to make the to be enriched in ammonia in a distillation tower that
following separation: consists of an enriching section and a total condenser. The
feed into the bottom of the column is a saturated vapor (at
Feed: 1000 lb moles/hr; x a = 0.4, h F =
its dew point) at 100 psia, the operating pressure. The
Distillate: x a = 0.8
distillate is withdrawn from a total condenser and has a
Bottoms: x a = 0.05
composition of 0.90 mole fraction NH3. Part of the liquid
Distillate reflux ratio (L /D) = 3.0
condensed in the total condenser is withdrawn as the distil-
There is no bottoms reflux. Pure saturated vapor b is late product and part is returned to the column as reflux;
fed into the bottom of column at a rate of 500 lb moles/hr. 85 percent of the ammonia charged must be recovered in

6.19. Calculate the liquid and vapor flow rates at the distillate.

each stage in Problem 6.18(b). What generalization might (a) Calculate the number of equilibrium stages re-

be inferred from the result? quired and the reflux ratio (Lq/D).
6.20. Five hundred Ib/hr of a mixture of 50 mass (b) What is the condenser duty (heat removed per
percent styrene, 50 mass percent ethylbenzene is to be hour from the condenser)?
separated into products with compositions 90 and 10 mass (c) After the column has been built and is operating,
percent styrene (solvent-free basis) by extraction with pure a change in the process changes the feed to 0.2 mole
diethylene glycol solvent Extract reflux will be used. A fraction NH3 at its dew point. Can the existing column of
mixer-settler system equivalent to ten equilibrium stages is part (a) be used? The reflux ratio can be adjusted for this
available. new feed, but the number of theoretical stages is fixed. The
(a) Determine a feasible reflux ratio, assuming that distillate composition and percentage recovery of NH3
the solvent leaving the solvent separator is pure. must be unchanged.
(b) For the conditions of (a), determine the fresh 6.24. A liquid ethanol-water feed is to be stripped
solvent feed rate. of ethanol in distillation column consisting of a stripping
621. Two thousand Ib/hr of a cottonseed oil— oleic section and a reboiler. The liquid will be fed to the top
acid solution containing 30 percent acid is to be extracted plate and vapor reflux will be furnished by a reboiler that
with propane in a continuous countercurrent cascade at takes liquid from the bottom plate and totally vaporizes it.

98.5 C using extract reflux. The extract product is to The bottoms product is also withdrawn from the bottom
contain 85 percent acid, and it is to have a negligible plate.
propane content. The raffinate product should contain 3 (a) Calculate the number of equilibrium stages re-

percent acid. The extract reflux ratio is to be 3.5. The fresh quired for the following conditions:
solvent and the recovered solvent are pure propane. moles/hr, 0.2 mole fraction
Feed: 100 lb
(a) Calculate the number of equilibrium stages re- mole fraction water
ethanol, 0.8 at
quired on the triangular diagram. 100°F and 1 atm
(b) Calculate the number of equilibrium stages re- Vapor product: 0.5 mole fraction ethanol
quired on solvent-free coordinates. Bottom product: 0.03 mole fraction ethanol
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 93

(b) What is the reboiler duty (Btu/hr added at ethanol and a bottoms of 2 mole percent ethanol. The total
reboiler)? condenser has duty of 1,000,000 Btu/hr. The bottoms are
a

(c) Can a vapor product of 0.7 mole fraction ethanol withdrawn from the bottom plate. The feed is a liquid at
be obtained with a stripping unit like this? (You may use 140°F.
more stages and higher reflux ratio but use the same feed (a) How many equilibrium stages are required?
and bottoms composition.) Explain. (b) What is the reboiler duty?
6.25. Several schemes for producing relatively pure (c) Plot the number of stages required for this separ-
oxygen from air are to be examined. Air is available as a ation as a function of reflux ratio. The plot should extend
liquid at its bubble point (79 mole percent N2 , 21 mole from the minimum reflux to total reflux.

percent O2). It is desired to produce liquid oxygen of 99 (d) How many stages would be required if the feed
mole percent purity. Pressure is 101.3 kN/m 2 . were half liquid and half vapor?
(a) Suppose the nitrogen is stripped out of the feed 6.28. Oxygen and nitrogen are to be produced by the
in a simple countercurrent column (without reflux) using fractional distillation of air at a pressure of 101.3 kN/m 2 .

an O2/N2 vapor at its dew point as a stripping agent. What The feed (79 mole percent nitrogen, 21 mole percent oxy-
is the minimum mole fraction of O2 in this vapor that will gen) is 75 percent vapor. The composition of the distillate
permit recovery of a liquid of the purity specified above? product,a vapor withdrawn from the condenser, is to be 95
Comment on the practicality of this process. mole percent nitrogen. The bottoms product, a liquid with-
(b) Suppose a reboiler is added to the countercurrent drawn from the bottom plate before the reboiler, is to have
cascade of part (a) to provide the necessary vapor stream a composition of 2 mole percent nitrogen.
for stripping the nitrogen out of the feed. The desired (a) What percentage of the oxygen in the feed will be
oxygen product is withdrawn from the reboiler and a waste recovered in the bottoms product?
vapor escapes from the top of the cascade, where the liquid (b) What are the minimum reflux ratio and the mini-
air is fed. What is the minimum waste vapor rate that will mum number of equilibrium stages required for this separ-
permit the separation, for 100 moles/sec of liquid feed? ation?
What percentage of the oxygen in the feed is recovered in
(c) What reflux ratio would be required if a column
the liquid product? with seven equilibrium stages were used?
(c) Suppose a partial condenser is added to the cas-
(d) For 100 moles/sec of feed, determine the liquid
cade of part (b) to permit greater recovery of the oxygen in
and vapor flow rates in the enriching section and in the
feed. With infinite stages and the feed properly located, stripping section for the column of part (c).
what is the maximum possible recovery of oxygen? If the
(e) What is the temperature of the reboiler liquid? Of
desired recovery is only 95 percent, what is the minimum
the condenser liquid?
reflux from the condenser that will accomplish the separ-
(f) What is the reboiler duty? The condenser duty?
ation?
629. Fifty pound-moles/hr of an aqueous ethanol
(d) What are the comparative advantages and dis-
solution containing 23 mole percent ethanol is to be frac-
advantages of schemes (b) and (c)?
tionated at 1 atm in a distillation column equipped with a
6.26. Seventy-two pound-moles/hr of an ammonia-
total condenser and a reboiler from which the bottoms are
water mixture at 100 psia and 70° F, containing 25 mole
withdrawn. The distillate is to have a composition of 82
percent ammonia, is to be fractionated in a distillation
mole percent ethanol and the bottoms 3 mole percent
tower into an overhead product of 95 percent ammonia and
ethanol. The feed is at its bubble point.
a bottoms of 4 percent ammonia. The overhead product is
(a) At a reflux ratio (Lq/D) of 3, how many equilib-
withdrawn as a vapor in equilibrium with the reflux from a
rium stages are required?
partial condenser, and the bottoms are withdrawn from the
reboiler. The reboiler duty is 700,000 Btu/hr. (b) Plot the number of stages required for this separ-

What ation as a function of the reflux ratio from minimum reflux

(a) is the condenser duty (Btu/hr)?
to total reflux.
(b) What is the reflux ratio (L Q /D)7
How many (c) Plot the reboiler duty as a function of reflux ratio
(c) equilibrium stages are required?
(d) What is the net flow of heat in the column? from minimum reflux to total reflux.

(e) What is the minimum reflux ratio for this separ- (d) How many stages would be required if the reflux

ation? ratio [Lq/D) is 3 but if the reflux is subcooled 30 F below

(f) What is the minimum number of stages required its bubble point?
at total reflux? 6.30. A distillation column with a total condenser
6.27. One hundred pound-moles/hr of an aqueous and a partial reboiler is to be designed to produce a distil-

solution containing 20 mole percent ethanol is to be frac- late containing 95 percent of the ethanol charged with a

tionated at 1 atm to produce a distillate of 80 mole percent distillate composition of 80 mole percent ethanol. The feed
94 STAGE OPERATIONS

is 100 1b moles/hr of a liquid at its bubble point; its (d) What are the condenser and reboiler duties for

composition is 25 mole percent ethanol. (O?

(a) What is the minimum reflux ratio? Label the 6.36. In the separation of high-purity oxygen from
critical tie line. air, a double distillation column is used, as shown in the
(b) How many stages are required at a condenser accompanying figure. The condenser of the high-pressure

duty of 1,500,000 Btu/hr? column serves as the reboiler for the low-pressure column,
(c) Determine rigorously the rate of liquid flowing thereby avoiding the need for supplying a low-temperature
from the third stage from the top for the conditions of (b). condenser coolant from an external source. The boiling
6.31. An existing distillation system equivalent to liquid in the reboiler of the low pressure column serves as

eight equilibrium stages is to be used to separate a saturated the coolant to condense the vapor for reflux and distillate
liquid feed containing 40 mole percent ethanol, 60 mole product in the high-pressure column. The low-pressure
percent water. The saturated liquid distillate is to have a column has no condenser, and the high-pressure column has
composition of 80 mole percent ethanol and the saturated no reboiler. The columns operate at 1 atm and 5 atm
liquid bottoms, 10 mole percent ethanol. The system oper- pressure.

ates at 1 atm pressure with no heat loss at each stage. Partially liquefied air is fed to the high-pressure

(a) Determine the reboiler duty for this separation. column, which effectively has only an enriching section.
(b) What is the percentage recovery of ethanol in the The vapor in the feed becomes the upward-flowing vapor in
distillate? the column while the liquid in the feed joins the liquid
6.3Z A saturated liquid feed containing 30 percent downflow in the column to become the bottoms, which,
ethanol, 70 percent water is to be separated to give a after an adiabatic expansion, becomes the feed to the
distillate of composition 80 percent ethanol. Ninety per- low-pressure column.
cent of the feed ethanol is to be recovered in the distillate. The distillate product from the high-pressure column
(a) Determine the minimum reflux ratio and the is subcooled in a heat exchanger and then passes through an
minimum number of stages for the separation. adiabatic expansion valve before it becomes the liquid

(b) At a reflux ratio 20 percent greater than the downflow (in place of reflux) in the low-pressure column.
minimum, how many stages are required? This subcooling is necessary to provide sufficient liquid
(c) What is the reboiler duty in (b)? downflow after the expansion.

6.33. A liquid mixture at 80° F contains 30 percent

ethanol, 70 percent water. It is to be separated into two ^N
1%0
2

products of compositions 5 percent and 75 percent 2

sat. vap.
ethanol. An existing distillation column is available. It is

equivalent to seven equilibrium stages and its reboiler can

supply up to 1,000,000 Btu/hr.
c
How much liquid feed can be processed in one hour
using the existing column?
6.34. An aqueous solution of ammonia containing 27
mole percent ammonia is to be fractionated at 100 psia in a
tower equivalent to five equilibrium stages. It is desired to
recover 95 percent of the ammonia in a distillate with a
Subcooler
concentration of 98 mole percent ammonia. The feed solu-
tion is 20°F below its bubble point. What reflux ratio
(L Q ID) should be used? -^99°g0 2
sat. liq.
6.35. Air is to be separated into relatively pure
5=^N 2
oxygen and nitrogen by low- temperature distillation at nor- 2%0 2
mal atmospheric pressure (101.3 kN/m 2 ). One hundred sat liq.

moles/sec of feed (79 mole percent N2 , 21 mole percent

O2) is half liquid and half vapor. The distillate product (97
mole percent N2) is a vapor withdrawn from a partial
condenser. The bottoms product (98 mole percent O2) is
sat. liq.
withdrawn from the reboiler.

(a) What is the minimum number of stages at total

79=v N-
reflux?
21% 2
(b) What is the minimum reflux ratio? 20% liq.
(c) At a reflux ratio of 1.8, how many equilibrium
80% vap.
stages are needed? Flow diagram for Problem 6.36.
 COUNTERCURRENT MULTISTAGE OPERATIONS WITH REFLUX 95

Because of the coupling of the columns, the con- (a) How much entering raffinate can be processed
denser duty in the high-pressure column is equal to the with the available solvent?
reboiler duty in the low-pressure column. (b) Determine the mass of the exit extract and exit
For a reflux ratio of 1.5 in the high-pressure column raffinate.

and 100 moles/sec of feed to that column: (c) Determine the number of equilibrium stages re-

(a) Calculate the flow rates of the final oxygen-rich quired.

and nitrogen-rich products. (d) At what stage should the impure solvent be fed?
(b) Determine the enthalpy of the distillate product (Data for this system are in Chapter 5, Problem 5.13.)
from the high-pressure column after it has been subcooled 6.41. In Problem 6.27, how many equilibrium stages
and expanded. (The expansion does not change its en- are required if the liquid entrainment amounts to 10 mole
thalpy.) Assume no heat losses through the column walls. percent of the vapor leaving each stage?
(c) Determine the number of equilibrium stages re- 6.42. Derive Equations 6.24 and 6.25.
quired in each column. 6.43. The oil from halibut livers is to be extracted
6.37. A simple countercurrent extraction has an in- using ethyl ether as a solvent in a simple countercurrent
termediate extract product stream. Develop necessary equa- cascade. There are available two sources of halibut livers:
tions and outline the calculation procedure for this 200 Ib/hr of livers of 10 percent oil, and 90 percent
situation. insoluble; 300 Ib/hr of livers of 35 percent oil and 65
6.38. Consider a distillation column with two feeds, percent insoluble. The entering ethyl ether solvent contains
a total condenser, and a total reboiler. Develop expressions 2 percent halibut liver oil. The final extract is to contain 75
for the net flow in each section of the column. Outline the percent oil and 25 percent ether, and 95 percent of the oil
graphical calculation for this situation, assuming compo- charged in the livers is to be recovered in the extract.
sitions of feeds, distillate, and bottoms are known together (a) Calculate the number of equilibrium stages re-
with feed flow rates. The top reflux ratio may be specified quired when each liver source is fed separately into the
and the number of stages calculated. cascade at the best location. Underflow data are in Figure
6.39. A mixture is to be fractionated into a distillate, 3.14.
a bottoms, and an intermediate stream between the feed (b) Calculate the number of equilibrium stages re-

and the bottoms. quired when the two sources of livers are mixed and fed
(a) Draw a flow sheet for the fractionating column, together into the cascade. Use the same solvent rate and
and label streams. recovery as in part (a).

(b) Sketch a typical enthalpy-concentration diagram 6.44. A mixture of 100 lb moles/hr of 40 mole
and show how the location of the required delta points is percent a and 60 mole percent b is to be separated into a
determined. Assume that Lq/D is set and that the quan- distillate of composition 90 mole percent a and a bottoms
tities, compositions, and enthalpies of F, D, B, and / are of 5 mole percent a. Instead of using a conventional re-

known. boiler or condenser, the following proposal has been made.

6.40. Diphenylhexane is to be separated from doco- Each theoretical plate in the enriching section will
sane by extraction with furfural in a simple countercurrent have a cooling coil that will remove 100,000 Btu/hr from
cascade at 45 C. There are two sources of furfural solvent, the liquid on the plate, and each theoretical plate in the
one pure and one containing a small amount of diphenyl- stripping section will have a heating coil that will supply
hexane and docosane. Since it is desired to process a maxi- 100,000 Btu/hr to the liquid on the plate.

mum quantity of feed, both solvents must be used. It will be (a) Calculate the number of equilibrium stages re-

desirable to feed the impure solvent separately from the quired to give the desired separation.
pure solvent. Relative volatility: aa -b = 3.0
Mass Mass Enthalpy: H= 9000 - 5000x; h = 1000*
Flow Mass Mass Fraction
where x = mole fraction a
Rate, Fraction Fraction Diphenyl-
(b) Is there any advantage to this arrangement over
Stream Ib/hr Furfural Docosane hexane
the conventional condenser and reboiler?

Entering raffinate 0.00 0.70 0.30

Entering solvent 1 5000 1.00
Entering solvent 2 3000 0.92 0.05 0.03
Exit extract 0.15
Exit raffinate 0.01
 Seven

Simplified Calculation Methods

The methods developed in the three previous chap-
ters can be applied to the calculation of multistage
separation processes when sufficient physical data
are available for the system to be separated. In
many cases alternative approximate calculation
methods may be used. The approximate methods
are usually more rapid than those considered pre-
viously, and they often require less physical data.
But the methods are based upon simplified physical
models of the systems to be separated, so that the
methods are applicable only to systems that more
or less the physical models. Some of the approxi-
fit
equilibrium stage, but a new relationship replaces
the delta point in calculating from stage to stage.
The relationship is obtained from a mass balance
over a section of the cascade. The following para-
graphs develop the relationship and then indicate
under what conditions it is useful in calculations of
specific mass-transfer operations.
A mass balance around any stage n and the
left end of the simple countercurrent cascade in
Figure 5.2 gives the following.
Total mass balance:

mate methods may be applied to systems where L + Vn+: =L n + V, (7.1:

the physical data are insufficient for the more
rigorous procedures of the earlier chapters. In such Component balance:
cases, the system may deviate greatly from the
assumed underlying physical model, so that the *-o*o + Vn + 1 yn + :
= L n xn + V-i J/, (7.2)

results may be inaccurate but still useful for

Solving Equation 7.2 for y„ +1 gives
approximate engineering estimates.
A number of approximate calculation methods Ln V y - Lqx
y :
=
,

and their limitations are considered in this chapter. Kn + 1 xn + (7.3)

Simplified graphical procedures are developed for Wi Vn ^
simple countercurrent flow and for countercurrent
Equation 7.3 relates the compositions of a lAphase
flow with reflux. All the concepts covered in the
-[) and an /.-phase (x„) flowing past each other
(y n +
preceding chapters are redeveloped using the simpli-
between stages. This relationship was precisely the
fied graphical procedures. In addition, certain
purpose of the delta point defined in an earlier
analytical procedures and other topics based upon
chapter. If the line represented by Equation 7.3 is
these simplified methods are considered.
plotted on x-y coordinates it will be the locus of all
possible values of (x n , y„+i). It is usually referred

GRAPHICAL CALCULATIONS ON THE EQUILIBRIUM to as the operating line. If the composition (x n )

of the /.-phase is known for any point in the
DIAGRAM
cascade, (y„+i) of the V-phase
the composition
In certain cases of stage operations, the equilibrium flowing the opposite direction at the same point
in
x-y diagram may
be used alone for a simplified can be determined from the plot of Equation 7.3.
calculation number of equilibrium stages
of the If in addition the equilibrium curve is plotted on

required for a given separation. The equilibrium the same x-y diagram, it is possible to calculate
curve continues to relate compositions leaving an from stage to stage graphically.

96
 SIMPLIFIED CALCULATION METHODS 97

A point on the equilibrium curve represents distillation column are influenced by four factors:
two streams leaving a single stage in equilibrium;
for example, streams leaving stage 3 would be 1 The molar heat of vaporization of mixtures.
represented by a point on the equilibrium curve 2. The heat of mixing in the vapor and liquid.
with coordinates (x 3/ y 3 ). On the other hand, the 3. The increase in sensible heat with increase
coordinates of a point on the operating line repre- in temperature through the column.
sent the compositions of two streams flowing past 4. Heat losses from the column walls.

each other between two stages. For example, the If the molar heats of vaporization of all mixtures
point (x 3 y 4 represents the compositions of the L
,
)
are constant and if the other factors are negligible
and V
streams flowing between stage 3 and stage 4. in comparison, for every mole of any liquid vapor-
The horizontal and vertical construction lines that ized one mole of vapor is condensed, and L and V
connect points on the equilibrium curve and oper- are constant through the cascade. Constant molar
ating line to effect have no
give the "staircase" heats of vaporization give parallel vapor and liquid
physical meaning by themselves; they merely help lines on the enthalpy-molar composition diagram.
locate points with physical meaning. If x 3 is Under this condition it can be shown geometrically
known, y 2 is found by drawing a vertical at x 3 to that L and V do not vary. It is possible to have a
the equilibrium curve. Then a horizontal is drawn straight operating line even if all four factors vary
at y 3 to intersect with the operating line at x 4 The .
and are influential, as long as the net effect gives
graphical calculation may continue in this steplike
constant liquid and vapor flow. Many binary
manner, alternately using the operating line and systems give sufficiently constant molar flow rates
the equilibrium curve. Further details of the calcu- to permit use of the simplified calculation
lation procedure are given in following sections, methods. When applied to distillation, the simpli-
but first it is necessary to examine the physical fied procedures are often referred to as the
model to which Equation 7.3 can be conveniently McCabe-Thiele method (3).
applied in graphical calculations. When the two solvent components (b and c)
Equation 7.3 yields a straight operating line if
in liquid-liquid extraction are completely immis-
L and V are constant through the cascade. The cible at the concentrations of solute (a) under
terms L V-\, x and y^ have unique values, and
, ,
consideration, can be reported as concen-
data
therefore if V is constant, the second term of
tration (ya ) phase V in equilibrium with
of a in
Equation 7.3 is a constant. If in addition L is
concentration (x a of a in phase L. Because com-
)

constant, the equation has the form of the stan-

ponents o and c are mutually insoluble, a ternary
dard slope-intercept equation of a straight line. The diagram is unnecessary, and complete equilibrium
subscripts on L and V may be dropped, since the
data may be presented on a simple x-y diagram.
terms are constant. With mutual insolubility of solvents b and c, solute
a is simply split between the two solvent phases,
_L - Lx Q
Vy - }

Ym+*- Xm + (7.3a) analogous to the split of the more volatile com-

v ponent a between the vapor and liquid phases in
distillation. The total mass flow rate of raffinate
The slope of the operating line is LIV, and the
(L) decreases as the solute is transferred from it to
/-intercept the last term of Equation 7.3a. The
is
the extract phase (V), and the extract increases in
line representing 7.3a is easily plotted if one point
mass. Thus, L and V vary, and Equation 7.3 does
and the slope are known, or if two points are not describe a straight line. The mass of each of the
known. If L and V are not constant, the line is not two solvents does not vary from stage to stage, so
straight, and more detailed calculations are re-
Equation 7.3 yields a straight line if the symbols
quired to plot it.
are redefined in terms of mass ratios instead of
For most purposes the simplified method is mass fractions:
more convenient only if Equation 7.3 describes a
straight line. For this reason, it is of interest to
examine several stage operations to determine
Yn + ,= —X n + '— °-
(7.3b)

under what circumstances L and Vare constant. where

In distillation if the heat effects in the column L' = mass of unextracted raffinate com-
do not change the molar flow rate of liquid or ponent {b), not including solute (a)
vapor from stage to stage, Equation 7.3 describes a V' = mass of extract solvent (c), not includ-
straight line. The liquid and vapor flows in a binary ing solute (a)
98 STAGE OPERATIONS

/„ +1 = mass of solute (a) per unit mass

Xn
solvent (c)

= mass of solute (a) per unit mass of LI

unextracted raffinate component (b)
The extraction equilibrium data must be replotted
on this basison X—Y
coordinates. The equations
derived subsequently in the dimensions of Equa- Multi-
stage
tion 7.3 may also be written in the dimensions of
cascade
Equation 1.3b.
In many cases of gas absorption, a single gas is

absorbed from another nonabsorbed gas into a

relatively nonvolatile liquid. An example is the
absorption of ammonia from air— ammonia mix-
tures by water. Equation 7.3b can be used, where
now
TT
L' = moles of liquid absorbent (b)
Figure 7.1. Calculations on the equilibrium diagram for
V = moles of the nonabsorbed constituent
simple countercurrent flow with transfer from the /.-phase
of the gas (c) to the V-phase.
V„ + 1
= mole ratio: moles of solute (a) in gas
per mole of nonabsorbed gas (c)
Even though the mathematical expression for the
Xn = mole moles of solute (a)
ratio: in
and
operating line is valid for any values of y„ + 1
liquid per mole of liquid absorbent (b)
x„, it has physical meaning only for compositions
Where a gas is adsorbed by a solid from a that actually occur in the cascade. That is, the
nonadsorbed gas, Equation 7.3b may be used. An actual operating line extends from the point
example of this would be the adsorption of water (*o,Ki) at one en d of the cascade to the point
vapor by silica gel from humid air. Equilibrium the other end, as shown in Figure 7.1.
(x/v, //v + i) at
data must be expressed in the appropriate units. The equilibrium curve can be plotted in the
If both phases are dilute in components, the same range of x and y, as shown in Figure 7.1 The .

changes in L and V are small through the cascade. procedure for stepping off stages may begin at
It is then possible to use Equation 7.3 in its either end of the cascade. For example, if the
original form with mole-fraction or mass-fraction calculation is begun with the composition (y-\ of )

units for adsorption, extraction, and absorption, the V-phase leaving the cascade, the composition
by assuming that L and V are constant at an (x-i) of the /.-phase leaving the first stage is deter-
average value. Serious errors can result if L and V mined by drawing a horizontal at y-\ on the
are not constant. equilibrium diagram. The horizontal intersects the
equilibrium curve at x-\ as shown. Next, the com-
,

GRAPHICAL CALCULATIONS FOR SIMPLE COUNTER- position y 2 must be determined fromx! by use of
CURRENT OPERATIONS the equation for the operating line. This equation
written specifically for the flow between stages 1
The graphical determination of the number of
equilibrium stages involves alternate use of the and 2 is
operating line and the equilibrium curve. An - Lx
example of stage calculations for simple counter- V-i
2
= ~L Xi
1
,

+
Vy,
(7.3a)
V V
current operation is shown in Figure 7.1. The
operating line may be plotted either by knowing all Because the operating line is a plot of the general
four of the compositions at both ends of the form of this equation, the value of y 2 is deter-
cascade or by knowing three compositions and the mined by the intersection of a vertical through Xt
slope (LIV) of the operating line. Equation 7.3a with the operating line, as shown. Nowx 2 is deter-
can be rearranged to give mined by the intersection of a horizontal at y 2
with the equilibrium curve, and this stepwise calcu-
L(x n -x Q )= V(yn+ , -y A ) (7.3c)
lation is continued until (x N //y/ + is reached. As )
, i
or
shown inFigure 7.1, slightly over four stages are
(7.3d) required. The location of the operating line below
V i *0 the equilibrium curve indicates that the net mass
 SIMPLIFIED CALCULATION METHODS 99

transfer is from the /.-phase to the IZ-phase. On the operating line through U/v,//v + i) increases until
other hand., an operating line above the equilibrium the operating line first touches the equilibrium
curve shows that the net mass transfer is from the curve at a single point, as shown in Figure 7.2a. An
\/-phase to the /.-phase in simple countercurrent attempt to step off stages shows that an infinite
flow. number of stages is required to change the compo-
sition across the point at which the operating line
and equilibrium curve intersect. The term pinch
Operating Variables
applies here to the pinched-in stages at the point of
Limiting values of the L/V ratio may be deter- intersection. If the concentration of the \/-phase
mined for several For transfer from the
cases. leaving the cascade is specified, the point (x /-, ,

/.-phase to the the minimum VlL ratio

\/-phase, is fixed, and the pinch occurs as shown in Figure
that gives the desired separation with an infinite 7.2b. The recovery
not specified; that is, x N is
is

number of stages can be determined as shown in not fixed. the operating line through
In this case,
Figure 7.2. For a specified recovery from the L- (*o# V\ that first intersects the equilibrium curve
)

phase, the point [x N , y/v + 1 ) is fixed. As the quan- gives the limiting value of VlL, which is a
tity of V-phase is reduced, the slope (L/V) of the maximum value.

(a) (b)

iV+1

yl

*0

(c) (d)

Figure 7.2. Limiting values of the ratio of phase flow rates, (a) Transfer from
/.-phase to V-phase. Recovery specified; y-\ to be determined, {b) Transfer from
/.-phase to V-phase. y-\ specified; x/y to be determined, (c) Transfer from V-phase
to /.-phase. Recovery specified; x/y to be determined, (d) Transfer from IZ-phase to
/.-phase, x/y- specified; y-\ to be determined.
100 STAGE OPERATIONS

Limiting ratios of LIV can be determined in a Ya 0.005 0.01 0.02 0.03 0.04 0.05
manner for the case where mass transfer is
similar
Xa 0.038 0.0785 0.168 0.272 0.395 0.539
from the V-phase to the /.-phase, such as in gas
absorption. For transfer from the V-phase to the These values are plotted on Figure 7.3.
/.-phase, the operating line is above the equilibrium = 600
\/ lb moles/hr
curve. the recovery of component a is speci-
When
fied, the value of y is set, since //v + 1 is known. 0.05
The entering /.-phase composition (x
y

is usually
X = 0.0, //V + 1
= 0.0526
)
0.95
known, so the point (x ,/i) is fixed; the For 90 percent recovery, the benzene leaving in the
limiting LIV is located as shown in Figure 7.2c. In
gas is (0.10) (0.05) (600) = 3.0 lb moles. Therefore,
this case, the limiting LIV is a minimum. On the
Vt = 3.0/570 = 0.00526 lb mole benzene/lb mole
other hand, when x and //v + 1 are known instead
air, since \Z/ = V% + 1 = (0.95) (600) = 570
'
lb
of the recovery, x# may be specified. Then the moles air. The conditions at the upper end of the
point (x/v,K/v + 1 ) is fixed and the limiting LIV is
tower are set and can be plotted: X = 0,
determined shown
as in Figure 1 .2d. This limiting Y) = 0.00526. The minimum liquid rate occurs
ratio is a maximum. when the operating line through (Xq,^) first
touches the equilibrium curve at a point, as shown.
Example A benzene— air mixture is to be
7.1. The slope of this line is

scrubbed in simple countercurrent absorption

a
tower using a nonvolatile hydrocarbon oil as 0.0526 - 0.00526
= 0.091
solvent. The inlet gas contains 5 percent benzene (vl-
KV 'mm 0.52-0
and the entering gas flow is 600 lb moles/hr.
Therefore, the minimum oil rate is
Solubility of benzene in oil follows Raoult's law.
The tower operates isothermally at 80° F. The (0.091) (570) (200) = 10,390 Ib/hr
average molecular weight of the oil is 200 and the At
tower pressure atm. is 1

(a) What is the minimum oil rate (Ib/hr) U.7V'). c tu.« = 1.5(/.7V/') min = (1.5)(0.091)
needed to recover 90 percent of the entering = 0.137,
benzene?
/.' = (0.137)(570)(20O) = 15,600 Ib/hr
(b) How many theoretical stages are required
if the oil rate is 1.5 times the minimum? The actual operating line has a slope of 0.137, and
it is X N = 0.345.
determined that
Solution. Since the recovery is specified and The stages may be stepped off from either
the mass transfer is from the V-phase to the end. About 5.3 stages are required for a liquid rate
/.-phase, the limiting LIV ratio will be determined of 15,600 Ib/hr, as shown in Figure 7.3. Use of
as shown in Figure 7.2c. To assure a straight mole fraction coordinates with the assumption of
operating line, mole
be used. This ratios will constant LIV in the problem would result in a
necessitates calculating the equilibrium curve for Although the gas phase is dilute,
substantial error.
mole-ratio coordinates. the phase is not. The equilibrium curve
liquid
shows appreciable curvature.

ya =
—x a for benzene

COUNTERCURRENT FLOW WITH REFLUX

at80°F, />
a
= 103 mm Hg; P = 760 mm Hg.
The introduction of reflux at one or both ends of a
103 countercurrent cascade results in two net flows and
Ka
= Xa =0 136Xa therefore two operating lines (Figures 6.1 and 6.2).
-

760
In this section general equations for the two oper-
Since ating lines and for their intersection are derived
and applied. A mass balance from the solute-rich
Xa
ya = and xa = end to include any stage n in the enriching section
1 + Ya 1 + Xa (Figure 6.1) gives

Xa =
Dx D + Vc y c
= 0.136 /rt+1 x„ + (7.4)
1 + Ya 1 +x a Vn ^
 SIMPLIFIED CALCULATION METHODS 101

0.06

(XN , y\-+i) a (x N , y.\-n) mmL/v

- 0.04

0.2 0.3 0.4 0.6

Xa , lb moles benzene/lb mole absorbent oil

Figure 7.3. Solution to Example 7.1.

When L and V are constant, the subscripts may be an extraction cascade, but no reflux at the solute-
dropped. lean end, as shown in Figure 6.1.

L Dx D + VrVc In distillation calculations Vc =0, and Equa-

(7.4a) tion 7.4a becomes
D
A mass balance between any stage in the strip- m Zn + 1
= X + — Xn
D 17.46)
ping section and the solute-poor end of the cascade
y V
(Figure 6.1) gives or

Lm-\ L N XN - Vn + -[Yn + D
u
Ym~
-
x m-1
\ ,-, ,-t
Kn + 1 = xn + *D [7.4c)
..
v„
~.
m
V,
\/.0) L+D L +D
The values of L and Inspection of the equation for the enriching-
V may change at the feed.
Therefore, section operating line (Equation 1 Ab) shows that
it is a straight line when L and V/are constant. The
Lx, Vy N +1 slope of the line is L/V and its intercept at x = is
Vm .-, *m - (7.5a]
y = Dx D /V. Furthermore, the operating line
1

V inter-
sects the diagonal (x = y) at x n =k„ + 1
=x d .

where L and V are the constant values of /.-phase Similarly, Equation 7.5a may be used in distillation
and \/-phase flow in the stripping section. Equa- where there is no reflux from a reboiler and an
tions 7.4a and 7.5a describe straight operating lines external vapor stream (V/v + i) is supplied to the
for use when there is reflux at the solute-rich end of bottom of the column. More typically, a reboiler is
102 STAGE OPERATIONS

present (Figure 6.2) and there is reflux at both (Figure 6.2), which is

ends. Then a mass balance between any stage in m Fz F = Bx B + Dx D (7.9)

the stripping section and the bottom of the column
yields and

Bx B F = B+D (7.10)
Ym ~ ry *m-\ (7.5b)
V Combination of Equation 7.8 with Equation 7.9
gives
This stripping section operating line has a slope
L/V and a /-intercept at —Bx B lV. It intersects (L- L)x -Fz F =(V-V)y
i i (7.11)
the diagonal at x m _i = ym = x B The method of .

From net flow considerations

plotting the two operating lines depends upon the
known information. The intersections with the di- V=L- A=L-B (7.12)
agonal and the slopes or /-intercepts are commonly
used.
V=L- A = L+D (7.13)

The two operating lines can be plotted on the Substitution in Equation 7.11 for V and V from
equilibrium diagram if sufficient information is Equations 7.12 and 7.13 gives
available. Equation 7.4 or 7Ab gives a straight
operating line for the enriching section of the cas-
(L-L)x, -Fz F =(L-L-B- D)y,
cade as long as the /.-phase and the \/-phase flows = {L- L — F)y,- from Equation 7. 1

are constant, of the feed may

but, the addition
and
change both of the phase flow rates. This results in
a new operating
stripping section,
line of a different slope for
as givenby Equation 7.5a or
the L-L
-r Xi - ZF= \L-L
[~~F
— 1
I

J
K '
(7.14)

7.5b. The changes in the /.-phase and \/-phase

flows depend upon the quantity and properties of The ratio / is now defined as the increase at the
the feed. The following derivation will interrelate point of feed introduction in total flow of the
/.-phase per unit of feed, or
the two operating lines with the feed character-
istics for the distillation cascade of Figure 6.2, with
=
L-L
reflux at both ends. The intersection of the two i (7.15)
operating lines properly occurs at the feed stage.
The equations for the two operating lines will be Equation 7.14 becomes
solved simultaneously to develop an equation that
form, the locus of all possible
will give, in a general
= zF (7.16)
//
values of the compositions at the intersection. This
1 /- 1

equation will be of great use in locating the oper- which the equation for the locus of all possible
is

ating lines on the equilibrium diagram. The follow- intersections of the two operating lines. The /-line
ing derivation shows that the locus of the inter- defined by Equation 7.16 can be plotted if / and
sections can be expressed in terms of properties of the feed composition are known. The line has a
the feed alone. slope of /'/(/ — 1 ) and intersects the diagonal (x = y)
At the intersection of the two operating lines, at z F .

a point on one line must be identical to a point on The concept of an /-line has been developed
the other, so that x m =x,=x n and y m =// = with reference to distillation, where it is most
K„ + 1, where the subscript refers to the intersection
/'
commonly used. An /-line may also be developed
value. Equations 7.4a and 7.5a then become, at for extraction with reflux only at the solute-rich
their intersections, end, starting with Equations 7.4 and 7.5. In distil-
lation, the enthalpy of the feed determines the
Vy,=Lxj+DxD (7.6)
value of /. By use of Equation 7.15 and an en-
Vy, = Lx, - Bx B (7.7) thalpy balance around the feed stage, the follow-
ing definition of / can be derived:
Subtracting Equation 7.6 from Equation 7.7 gives
heat required to convert one mole
(V-V)y, = (L-L)x,-Bx B -Dx D (7.8)
of feed to a saturated vapor
/
= (7.15/?)
This equation may be simplified by combining it latent heat of vaporization of one
with a mass balance around the entire cascade mole of the feed composition
 SIMPLIFIED CALCULATION METHODS 103

lines.After the operating lines are located using the

/-line, stagesare stepped off using the appropriate
operating line. The stepping-off shifts from one
operating line to the other at the stage on which
the feed is added, regardless of whether or not this
stage is the vicinity of the intersection of the
in
operating In the design of a column, how-
lines.
ever, the fewest stages will be required if the shift
is made as near as possible to the intersection of

the operating lines, and the feed is added at the

stage so indicated. A feed to another stage requires
more sometimes said to be mislocated.
stages and is

A partial condenser or reboiler may con-

Figure 7.4. Location of /-lines in distillation. tribute up to one equilibrium stage, as discussed in
Chapter 6. The equations for the operating lines
Slope of
/ /-Line
can be derived for these cases from enthalpy and
material balances.
(1) Superheated vapor <0 +
The may be
reboiler eliminated in favor of
(2) Saturated vapor
- direct addition of heat as steam in distillations
(3) Liquid and vapor Oto 1

Liquid at bubble point oo

where water is one of the components. The direct
(4) 1

Liquid below bubble point >1 + feeding of steam into the bottom of the distillation
(5)
column is called "open steam." The operating line
Equation 7.156 in the stripping section for open steam can be
an expression equivalent to the
is

/ (Equation 7.15). The varia-

original definition of derived from material balances around the lower
tion in / and in the slope of the /-line with part of the column.
the thermal condition of the feed is outlined in
Figure 7.4.
Example 7.2. A mixture of 35 mole percent a
The simplified graphical method for distil-
and 65 mole percent b is to be separated in a
lation does not usually make use of enthalpy data,
distillation tower. The concentration of a in the
since the method assumes that the latent heat of
distillate is 93 mole percent, and 96 percent of all
vaporization of all mixtures is constant, but the
product a is in the distillate. The feed is half vapor
thermal condition of the feed must be considered. and the reflux ratio (L /D) is to be 4. The relative
volatility of a to b is oc _ = 2. How many equi-
For example, if the feed is a subcooled liquid, it a b
librium stages are required in each section of the
must pick up heat as it enters the cascade, because
all liquids in the cascade must be saturated liquids.
column?
It picks up heat at the expense of the condensation
Solution. An x-y diagram (Figure 7.5) is

drawn for a-, = 2.

of some vapor, that the liquid downflow is
so
further Therefore, (L — L)>F and
increased.
*a 0.20 0.40 0.50 0.60 0.80 1.0
/> 1, as shown in Figure 7.4. The slope of the
/-line is //(/ — 1), and therefore the slope for a Va 0.333 0.571 0.667 0.75 0.889 1.0
subcooled liquid will be positive. The other cases
shown in Figure 7.4 can be treated by an analysis z,r=0.35 /=- = -
1 2
similar to that above. Consideration must also be
given to a subcooled liquid reflux from the con-
Slope of the /-line = -1
denser, and a superheated vapor reflux from the
reboiler must be allowed for in calculations on the
equilibrium diagram. A subcooled reflux yields a x D = 0.93 —= 4.0
liquid downflow in the column (L) that exceeds
the external reflux (L ). Slope of the enriching operating line
Although the /-line is located using the feed
composition, the actual feed need not be added at
=
L D 4
the stage located at the intersection of the two = = -= 0.8
L + D LQ 5
operating lines and the /-line. The /-line merely
represents the intersection of the two operating
—+ 1
104 STAGE OPERATIONS

Figure 75. Solution to Example 7.2.

To find xB : Basis: 100 moles feed The stripping-section operating line is a

straight line from the intersection of the enriching
Moles a in feed = 35 line with the /'-line to the point x B = 0.0219 =/s-
Molesa in distillate = (35M0.96) = 33.6 The stages may be stepped off from x B or x D .

Since no information on the reboiler and con-

Moles a in bottoms = 1 .4
denser is available, neither is assumed to contribute
Total moles in distillate = 33.6/0.93 = 36.1
to the separation. There are 9.5 stages in the strip-
Total moles in bottoms = 100-36.1 =63.9 ping section and 7.5 stages in the enriching section.

Therefore, Operating Variables

1.4 As the reflux ratio is reduced, the slope of the
*b = = 0.0219
enriching-section operating line decreases until one
63.9
of the operating lines first intersects the equi-
The enriching-section operating line is located by librium curve at a single point. This first inter-
the point x D = 0.93 = y y and the slope = 0.8. section determines the minimum reflux ratio,
Alternatively, the enriching-section operating line which requires an infinite number of stages to give
can be plotted by the point x =0.93 = y-\ , and the desired separation. For systems of constant
the intercept x = at relative volatility, this intersection is at the feed.

Dxd 0.93 Figure 7.6a shows such a case where the feed is
*D
y = = 0.186 optimally located at the intersection of the two
V 4 + 1
+ 1
operating lines. For systems where the relative vol-
(1) atility is not constant, the intersection point deter-
 SIMPLIFIED CALCULATION METHODS 105

Figure 7.6. Determination of minimum reflux on the x—y diagram, (a) Pinch at
feed, (b) Pinch in enriching section.

mining minimum reflux may occur anywhere be- INTERMEDIATE STREAMS

tween x D and x B , as shownFigure 7.6b. As the
in
The reasoning used in calculating intermediate
reflux ratio is decreased approaching the minimum,
streams on the equilibrium x—y diagram is the
the steps become pinched in between the operating
same as that used on the three-variable dia-
lines and the equilibrium curve; the required num-
grams considered in Chapter 6. The addition or
ber of stages increases rapidly. With a very large
removal of a stream changes the L- and V-phase
number of stages, the vapor and liquid com-
flow rates and results in a new operating line. Two
positions change only slightly; hence, this region is
methods are available for locating the new oper-
called the zone of constant composition.
ating line in the intermediate section of the cas-
At total reflux, L = V = L = V, so that the
cade shown in Figure 6.16.
slopes of both operating lines become 1.0; the
The /'-line for the intermediate stream may be
operating lines concide with the diagonal (x = y).
plotted regardless of whether the stream is a prod-
The number of stages at total reflux is then deter-
uct or a feed. If the top reflux ratio is given, the
mined by stepping-off from x D to x B
stages
upper operating line may be drawn from x D to the
between the equilibrium curve and the diagonal,
intersection with the /-line for the intermediate
as shown in Figure 7.7. The number of stages re-
stream. The operating line is continued beyond the
quired at total reflux is the minimum for the
intersection but now at a new slope corrected for
specified separation.
the addition or removal of the intermediate stream.
The intermediate operating line intersects the /'-line

for the feed, changes slope, and continues tox B In .

this case, only one intermediate stream was pres-

ent. In general, any number may be calculated as
outlined.
The second method of calculation is by de-
fining a fictitious feed (Z), which
the algebraic is

sum of the feed (F) and the intermediate stream

(/). This is shown in Figure 7.8. The cascade with a

fictitious feed is equivalent to a cascade with one

feed and no intermediate streams, so that it is
methods developed
possible to use the calculation
earlier one feed. First the three /'-lines are
for
plotted. The one for the fictitious feed is calcu-
lated from the properties of the two component
Figure 7.7. Determination of minimum stages streams. The slope of the enriching-section oper-
at total reflux. ating line is usually set by fixing the top reflux
106 STAGE OPERATIONS

y* >. + !
v„ v„*l

Stage n

J..-1 Ln

(a)

Figure 7.8. Fictitious feed for an intermediate

stream.

This operating line is drawn in and continued

ratio.
until intersects the /'-line for the fictitious feed
it

(Z). The stripping-section operating line is then

located by drawing a straight line from this inter-
section to x B = y. But the enriching operating line
has physical significance only until / is added; the

stripping operating line only from x B to the

is valid
/-line for the feed. The operating line between the
F and / /'-lines is easily located by the straight line
connecting the intersections of the /'-lines with the
stripping and enriching operating lines, as shown.
This method may be extended to multiple inter-
mediate streams, but it then is less convenient than
the first method outlined.

INTERRELATION BETWEEN CONCEPTS OF Figure 7.9. Interrelation between the delta

OPERATING LINE AND DELTA POINT point and the operating line.

Although calculation on the equilibrium x—y dia-

gram is most advantageous when straight operating vice versa, or to determine y„ + 1 from x n or vice
linesoccur, the method may be applied to cases of versa. Figure 7.96 shows the construction required
curved operating lines. Curved operating lines re- to locate the point (x n yn + ,
on the operating line
-\ )

sult when the assumptions made earlier in this by use of the delta point. The compositions x n and
chapter do not hold. For example, a variation in y n of the streams flowing from stage n are known
the molar overflow in distillation caused by non- and are plotted on the upper diagram. A straight
constant latent heats of vaporization or significant line from x A through x n locates y n + at its inter- \

sensible-heat changes would give a variable slope to section with the V-phase locus, as shown. The
the operating lines. Similarly, in liquid extraction a values of x n and y n + are now known. For calcu-
-\

variable partial solubility of the two solvents would lations on the equilibrium diagram, Equation 7.3
result in curved operating lines. shows that the point (x n yn + -\) must be on the
,

The point and the operating line are

delta operating line. The values for x n and yn + found y

both used to relate the compositions of the streams on the upper diagram may now be transposed to
flowing between two stages. For any stage n (Fig- the equilibrium diagram as shown by the construc-
ure 7.9a) either the delta point or the operating tion lines, thereby establishing a point on the oper-
line may be used to determine yn from x n _- i , or ating line. Additional points on the operating line

may be determined in the same manner, and the
curved operating line may then be drawn in, as
shown. It is not necessary to use streams actually
flowing between equilibrium stages to determine
the operating line. Any straight line from the delta
point cuts the /.-phase and \/-phase curves at com-
positions that represent a point on the operating
line.
The complete determination of curved oper-
ating lines is shown in Figure 7.10 for a case in
distillation where the latent heat varies, the con-
struction lines for each stage are shown to illustrate
how the curved operating lines are located. The
stepsbetween the equilibrium curve and the curved
operating lines are not only construction lines for
H, h
x a>y a
Figure 7.10. Determination of curved operating
lines.
SIMPLIFIED CALCULATION METHODS 107
the enthalpy-composition diagram; they are also
the equilibrium stages determined for this case,
where L and V are not constant. The variation of
the latent heat of vaporization is large for the
system in Figure 7.10, but the operating lines are
only slightly curved. Assumption of constant molar
overflow to give straight operating lines for this
system would have yielded about one less equi-
librium stage than was obtained with the curved
operating lines in Figure 7.10. An error of this
magnitude may be tolerable in some engineering
applications, so the assumption of constant molar
overflow is often made, especially when little or no
enthalpy data are available.
APPLICATIONS OF SIMPLIFIED GRAPHICAL
METHODS
In many cases in stage operations, insufficient
physical data are available to permit rigorous cal-
culations utilizingthe methods of the previous
chapters. Insuch cases, the methods outlined in
this chapter may be used with a minimum of
physical data, even though the physical system
does not fall within the simplifying assumptions
made at the beginning of the chapter. Often the
error introduced is small, and a sufficient factor of
safety may be included in the design. For example,
distillation of ammonia-water systems may be
evaluated by the methods of this chapter, even
though the latent heats — and therefore the phase
flow rates — are not constant.
An intermediate degree of accuracy may be
achieved where partial data are available for the
phases over the composition range of interest. For
example, in distillation, if only the heats of vapor-
ization of the pure components are known, they
can be plotted on an enthalpy-composition dia-
gram, and a straight line may be drawn between
the two values of the saturated -vapor enthalpy to
give an approximate vapor enthalpy curve over the
entire composition range. This method accounts
for the latent heat of vaporization but neglects
other heat effects, which are often of a much
smaller magnitude. A similar improvement for the
simplified procedure can be made by recomputing
the equilibrium diagram using for one of the
components a fictitious molecular weight, chosen
so that the molar flow rates remain constant.
UNSTEADY-STATE OPERATION
The calculation methods developed up to this
point have applied to steady-state operation in
108 STAGE OPERATIONS
which the compositions and flow rates do not vary
with time. In such cases, the feed is added,
products are withdrawn continuously, and flow
rates may change from point to point in the equip-
ment, but at any specified point there is no change
with time. An unsteady-state or transient process
involves changes of variables with time. There are
many transient operations in chemical processing.
All batch separation processes involve a change in
composition with time.
For producing relatively large quantities for
extended periods of time, continuous processes
operating at steady state are usually chosen, but
for small quantities of infrequently produced
materials, batch processes are more appropriate.
Continuous processes pass through a period of un-
steady-state operation when they are started up.
Even during continuous operation, there are some
transient fluctuations in what is steady-state oper-
ation on the average.
A rigorous analysis of unsteady-state oper-
ation extremely complex and beyond the scope
is
of this book. A more through treatment is available
in the literature on process dynamics and control
(such as Reference 2). In this section a few simple
examples involving unsteady-state operation
tillation are considered.
Start-up of Continuous Equipment
In the start-upof multistage equipment, com-
positions and flow rates may vary with time. The
length of time after start-up required to reach a
steady state is an important consideration in pro-
cess operation. The time depends upon the rate of
passage of the two phases through the equipment
and upon the rate of mass transfer. The rate of
passage depends upon the volume of material held
up in the equipment and upon the flow rates.
In the start-up of a distillation column, feed is
supplied to the column, where it runs down to the
still and is vaporized. The vapor rises to the con-
denser, where it is condensed and returned to the
column. Frequently a distillation column is started
up at total reflux after an initial charge of feed. It
is run at total reflux for a time sufficient to allow
the distillate composition to approach that desired
for steady-state operation. The time for this period
can be predicted approximately. Once the distillate
composition is achieved at total reflux, the distil-
late and bottom products may be withdrawn and
feed added to complete the transition to steady-
state operation.
There is also a transient period at the start-up
of a simple countercurrent cascade, such
extraction or absorption. The length of this period
willdepend upon the flow rates of the two phases,
the hold-up volume of the equipment, and the rate
of approach to equilibrium. The transient period
may range from minutes in small equipment,
through several hours in large-scale installations, to
several days in some special applications.
Batch Operations
When the quantity of material to be processed is
small, batch methods are often used. Although the
trend in the chemical process industry is toward
continuous processing wherever possible, batch
operations are still very widely used.
Many batch operations can be calculated
using methods developedin earlier chapters. For
example, batch operation, such as the
a single-stage
extraction of coffee from the grounds into a hot-
water phase, can be calculated assuming equi-
librium between the two phases before they are
separated. The two phases (initially pure water and
ground coffee) are mixed and undergo composition
changes with time. The time required to reach
equilibrium is an important process variable. It
depends upon the rate of mass transfer and upon
in dis-
the degree of contact between the phases. The
composition of the two product phases in any
batch single-stage process can be calculated by
methods developed in earlier chapters if the two
phases are allowed to remain in contact until all of
one phase is in equilibrium with all of the other
phase. On the other hand, in some cases one phase
may be added or withdrawn during the process, as
in differential distillation.
Differential Distillation
In batch distillation from a single stage, a liquid is
vaporized, and the vapor is removed from contact
with the liquid as it is formed. Each differential
mass of vapor is in equilibrium with the remaining
liquid. The composition of the liquid changes with
time, since the vapor formed is always richer in the
more volatile component than the liquid from
which it is formed. This results in a continuing
impoverishment of the liquid in the more volatile
component.
The liquid composition varies and so also
must the equilibrium vapor composition. The origi-
nal composition of the total liquid phase (L) is
taken as x and of the vapor phase formed, y. By an
overall mass balance for the vaporization of a dif-
ferential quantity of liquid,
as in dL =-dV (7.17)
 SIMPLIFIED CALCULATION METHODS 109

If dV moles of average composition (/) are formed the still pot, the residue will become increasingly
by vaporization, a mass balance for the more vola- depleted in the more volatile component as the
tile component gives distillation progresses. The number of equilibrium
stages in the column is constant, and the concen-
d(Lx) = -y dV (7.18] tration of more volatile component in the distillate
and decreases as the still-pot concentration decreases, if
the reflux ratio is held constant. This is shown in
xdL + Ldx = -ydV (7.19) Figure 7.1 1. F is the quantity of original charge of
Combination of Equations 7.17 and 7.19 gives composition x Fl D the quantity of distillate of
composition xq, and B the residue in the still pot
xdL + Ldx=ydL (7.20) of composition x B At any time during the distil-
.

and lation,
F=D+B (7.25)
rdL _ r dx
(7.21! Fx F = Dx n + BXf
J
L ~K-x [7.26)

Because the entire column is an enriching section,

which is called the Rayleigh equation. Integrating
there is only an enriching operating line. It extends
over a finite change from state 1 to state 2 gives
initially between xp and an initial distillate com-

Lo r*2 dx position (x Di ), which is determined by the reflux

In -^= (7.22)
ratio. The equation of the initial operating line is
L: *x] y-x
Ifthe equilibrium relationship between y and x is
L D
(7.27)
known, the right-hand side of Equation 7.22 may
be integrated. If Raoult's law holds, y = Kx, and The column represented in Figure 7.11 is equiv-
alent to three equilibrium stages. The initial distil-
1

-In — (7.23) late composition is determined by adjusting the

L, K 1 x.
operating line of a given slope until exactly three
For a constant relative volatility, stages fit between x F and x Di As the distillation .

continues, the concentration of the more volatile

—
L =
2 X 1 /, 2 1 ~X 2 \ .
component in the still-pot residue decreases to the
In -In a In- (7.24)
Li a-n X| 1 — x-\) final value x B The composition
. of the final distil-

late (x D f ) is determined by an operating line of the

Batch Distillation

Batch distillation is often used for separating small

quantities of liquids. Often the batch still is used yy x
for a large variety of separations, and therefore it

must be versatile. A batch distillation is usually

carried out in an existing column equivalent to a z.
known number of equilibrium stages, and therefore < //in.
x Df

it is necessary to determine the reflux ratio re-

J/
quired to give the desired distillate purity. Intercept =
Z D, 'b)
In the typical batch distillation, the liquid to
be processed is charged to a heated kettle, above /
which mounted the distillation column equipped
is :y
/
with a condenser. Once the initial liquid is charged, 'X
no more feed is added. The liquid in the kettle is
boiled, and the vapors pass upward through the
column. Part of the liquid from the condenser is
refluxed, and the remainder is withdrawn as distil-
late product. Nothing is withdrawn from the still

pot until the run is completed. Figure 7.11. Batch distillation at constant reflux
Because the withdrawn distillate is richer in ratio, (a) Initial operating line. (/?) Final operating
the more volatile component than the residue in line.
110 STAGE OPERATIONS

1.0
same slope as earlier and by the requirement of
exactly three equilibrium stages.
At any time during the distillation, there are

B moles of composition xB in the still pot. If dB

moles of composition x D are removed from the
column, a mass balance for the more volatile com-
Initial
ponent gives operating line

More volatile more volatile more volatile

component in _ component in component in Operating line for

the original " the dB moles the {B - dB) xD =0J

B moles removed moles remaining
or
Bx B = (dB)x D + (B- dB)(x B - dx B ) (7.28)

Neglecting second-order differentials yields

0.2 0.4 0.6 0.8 1.0
dB
—=
B xD
—
dx R
-x
(7.29)
xa . mole fraction ethanol in liquid

B (a)

Integrating from the initial to final still-pot con-

ditions (F to B) gives
1.80

dxi
= ln-= f (7.30)
'b B B JxB f Xq xb
1.60 -
This equation is of the same form as the Rayleigh
equation for differential distillation (Equation
7.22). A number of equilibrium stages are avail- 1.40

able, so the relationship between x D and x B must

be determined graphically. Then the right-hand
side of Equation 7.30 is integrated graphically,
Figure 7.12. Solution to Example 7.3. (a) Determination
after several corresponding values of x D and x B
of x B from xq. (b) Graphical integration curve. Equation
have been determined. It should be noted that
7.30.
when only one stage is available, Equation 7.30
reduces to Equation 7.22. ted as a function of x B and a graphical integration
(Figure 7.12&) is made.
Example. 7.3. One hundred pound-moles of a
mixture of 20 mole percent ethanol, 80 mole per- {b) Graphical integration of Equation 7.30.
cent water is charged to the still pot of a batch
1
distillation column equivalent to three equilibrium
stages. The distillation at a reflux ratio of 3 and *D *B *D ~ X B
normal atmospheric pressure is continued until the
0.76 0.22 1.85
residue in the still pot reaches a composition of
0.75 0.20 1.80
0.03 mole fraction ethanol. What is the quantity
0.74 0.10 1.6
and average composition of the distillate?
0.72 0.04 1.5
0.70 0.03 1.5
Solution, (a) The slope of the operating line
0.68 0.02 1.5
is 3/4 (Figure 7.12a). The three stages in the col-
umn plus the still pot give a total of four equi-
r*F dx B
librium stages. The initial x D is determined by 0.275
adjusting the operating line of slope 3/4 until ex- *Bf xD -x B
actly four stages fit between x F and x D this pro- ;
100
cedure shows that x Di = 0.75. Now various values I n = 0.275
of x D are chosen arbitrarily and x B is determined B
graphically for each x D Then M{x D - x B is plot-
. ) B = 7 6.2, D = 23.8
 SIMPLIFIED CALCULATION METHODS 111

Average distillate composition 1 o

-
Dxn Fxc — BXt
x Dav = 0.8 -
D D
-
(100M0.2)- (76.2X0.03)
/ Initial

23.8 S 0.6
- y/- s_
operating line

= 0.745
/ "
In this case, the average composition of the distil- 0.4
/
/
late is its initial value because the change
close to
during the distillation was small. In other cases, the / \y^^^~- Final operating line
- -
average distillate composition may be considerably 0.2 /

less than the initial value. / *Hf -

It is possible to maintain a constant distillate

//
composition in batch distillation if the reflux ratio 0.2 0.4 0.6 0.8 1.0

is continuously increased during the run. Calcula- xa , mole fraction benzene in liquid

tions in this case are identical with those illustrated

Figure 7.13. Solution to Example 7.4.
previously, except that x D remains constant and
the slope of the operating line (L/V) must change
to maintain a constant number of equilibrium The operating line is pivoted around x D = 0.95 = y
stages between x D and x B .
until exactly five stages fit between x Bf = 0.18 and
x D = 0.95. The slope is 76/77, and the reflux ratio
is (76/77)/(1 -
76/77) = 76. This value is close to
Example. 7.4. A mixture of 35 mole percent total reflux. Operation at such a high reflux ratio

benzene and 65 mole percent toluene is to be would require a large quantity of heat per unit of
fractionated at normal atmospheric pressure in a product and would be economically unjustifiable.
batch column to recover 60 percent of the benzene More benzene could be recovered, but the purity
in a distillate of composition 0.95 mole fraction of the distillate product would have to be reduced
benzene. The column is equivalent to four equi-
librium stages. It is frequently found most feasible to main-
(a) What is the initial reflux ratio? tain a constant distillate composition at the begin-
(b) What is the final reflux ratio? ning of batch distillation. When the reflux ratio has
increased to an economic or physical limit, it is

held constant for the remainder of the run, and the

Solution. (a) The initial reflux ratio is
distillate composition is allowed to decrease until it
determined by adjusting the slope of the operating reaches a predetermined limit, at which time distil-
line through x D = 0.95 = y until exactly five stages
lation stopped.
is
fit between x F and x D From Figure 7.13, the.

In many
batch distillation columns, the liquid
slope is 0.8, so that the initial reflux ratio is
held up on each plate of the column is appreciable.
L /D = 0.8/(1 -0.8) -4. This must be taken into account when predicting
(b) One hundred pound moles charged: 35 lb
the yield and composition of distillate.
moles benzene, 65 lb moles toluene.

Benzene recovered in distillate = (0.60) (35) STAGE EFFICIENCIES

= 21 In actual countercurrent multistage equipment, the

0.05
two phases leaving a stage are not in equilibrium,
Toluene in distillate = (21 .0) = 1 .1 because of insufficient time of contact or inade-
0.95 quate dispersion of the two phases in the stage. As
Final still-pot composition: benzene = 14.0 a result, the composition change for each phase in

toluene = 63.9 an actual stage is usually less than that possible in

an equilibrium stage.
14.0 The relationship between the composition
- = 0.18
*Bf
63.9+ 14.0 change in an actual stage and that in an equilibrium
112 STAGE OPERATIONS

*
y

(b)

Figure 7.14. Operating lines for a single stage, (a) Equilibrium stage, [b) Actual
stage.

stage is illustrated in Figure 7.14. The operating ciency. Various efficiencies have been defined to
line for a single stage is obtained by writing a mass express the failure to reach equilibrium in an actual
balance around one stage (Figure 7.9a): stage. The performance of an actual stage is de-
pendent upon many physical variables. The stage
L„ _ x n _ , y
+ Vn + yn +
1 ,
= L n xn + Vn yn must bring two phases together in intimate contact
Rearranging and assuming constant molar overflow and then separate them. More thorough contact
gives and longer time of contact give a closer approach
to equilibrium in the exit streams. Influential vari-

=
L
— /-x n _i + Vyn + y ables include the interfacial area between phases,
Yn xn + (7.31)
V V the rate of transfer (per unit area) of components
between phases, and the time of contact. These
The compositions of the two entering streams variables are in turn influenced by the construction
(x„_ 1 and y„ + 1 ) are fixed values, and L and \/are characteristics of the stages, the physical properties
constant, so that Equation 7.31 is of a straight line of the phases, and the flow rates. Because of the
of slope (— L/V). The line starts at the point complex influence of all these variables, no purely
(x n _i,K„ + i) and
runs toward the equilibrium theoretical prediction of stage efficiencies is pos-
curve. In an equilibrium stage, the leaving streams sible. Instead, efficiency data taken on existing
are in equilibrium with each other, so that the equipment are correlated to develop empirical
point (x n ,y n ) lies on the equilibrium curve, as relationships, which can then be used to predict
shown in Figure 7.14a. In an actual stage, the exit efficiencies for new equipment. Of course, such
streams approach equilibrium, but they generally correlations can be applied only to equipment and
do not reach it. The mass balance of Equation 7.31 systems similar to those on which the original data
shows that the point (x n ,y n eQ lies on a straight ) _ were taken. Although much data have been taken
line of slope (-L/V) through the point and many correlations made, the stage efficiency
(x n _! y n + ). This must be true for both an equi-
, 1 involves the greatest single inaccuracy in the design
librium stage and an actual stage. Therefore, the of multistage separation processes. Theoretical
point (x n y n for an actual stage lies on a straight
, ) developments have made it possible to relate effici-
line between {x n _ u y n + and (x n ,y n eq _, where } ) ) encies to fundamental characteristics of the sys-
the latter point represents the compositions leaving tem, but empirical data are still required. A
an equilibrium stage. thorough analysis of stage efficiencies in distil-
Although the physical model of an equi- lation columns is found in Reference 6.
librium stage is convenient for calculations, the In this section several efficiencies are defined
resulting number of equilibrium stages must be and interrelated. Chapter 16, these efficiencies
In
related to the required number of actual stages are related to the rate of mass transfer.
before the multistage equipment can be built. The The most commonly used efficiency is simply
relationship is usually expressed as a stage effi- defined as the ratio of the number of equilibrium
 SIMPLIFIED CALCULATION METHODS 113

stages calculated for a given separation to the num- positions refer to component a; that is, the more
ber of actual stages required for the separation. volatile component in distillation, or the solute in
This is the overall efficiency. absorption or extraction.
The Murphree vapor efficiency is defined by
number of equilibrium stages calculated
En = yn -yn +
E„ =
i
number of actual stages required (7.33)
[7.32)

Although it is permissible to report a fractional The term y n * is a hypothetical vapor com-

number of equilibrium stages, it is obvious that position that would be in equilibrium with the
only an integral number of actual stages can be liquid leaving the actual stage. As shown in Figure

built. The overall efficiency is simple to use in 7.146, the point (y n *,x n is on the equilibrium
)

calculations, but it gives little insight into the
physical variables that determine its value. A
theoretically more fundamental efficiency is likely
to lead to more adequate empirical correlations.
The Murphree stage efficiencies (5) describe
curve, where x n is the composition of the liquid
leaving the actual stage. Thus, E v is the ratio of
the actual change in average vapor composition to
the change that would occur if the total vapor were
in equilibrium with the liquid actually leaving the

the efficiency of a single stage. They were origin- stage.

allydefined for distillation, but they can be easily The Murphree liquid efficiency is defined by
extended to other stage operations. Figure 7.15
shows a schematic representation of a single stage.
x n-\ —x r

E, = (7.34)
The stage shown is a perforated plate where the *n- 1

lighter \/-phase bubbles up through a heavier L-

The composition**,, * is for a hypothetical liquid
phase (as in distillation with vapor bubbling
thatwould leave the stage in equilibrium with the
through liquid). The following development can be
average composition of the vapor actually leaving
applied to any stage-contacting device.
the stage.
The vapor coming to the stage is assumed to Murphree stage efficiencies may exceed 1.0 if
be thoroughly mixed and of uniform composition near-equilibrium conditions exist at all points in
y n + -\. The vapor leaving the stage has an average the stage. In distillation, for example, the com-
composition y n It will probably be well mixed
.
position y n * is in equilibrium with x n which is the ,

when it reaches the stage above, but near the liquid lowest composition the more volatile com-
of
surface on stage n the vapor composition may vary ponent (a) in the stage. Thus, y n * can be less than
because it is in contact with a liquid of varying
y n because the latter is an average of the y's all
,

composition. The vapor composition y applies to a across the stage, most of which are in contact with
single point on the stage where the vapor is in liquids richer in a than the exit liquid. Thus, the
contact with a liquid of composition x. The liquid numerator in Equation 7.33 would be larger than
enters the stage with composition x n _ and leaves \
the denominator, and E v would exceed 1.0.
with composition x n Both the x and y com-
.
Murphree point efficiencies are defined be-
cause they can be better related to the funda-
mental mechanisms of mass transfer in the stage, as
L-phase shown in Chapter 16.
t t t t
The vapor point efficiency is

y* + i

(7.35)
K*-Kn + 1

Stage n
The term y is the vapor composition leaving a
point in the stage where the liquid composition is
x, and y* is the composition of a hypothetical
t t t t
vapor in equilibrium with the liquid of com-
.

position x (Figure
7.15). Point efficiencies do not
V- phase exceed 1because y' is the maximum value y can
.0,

achieve at the point under consideration. A liquid

Figure 7.15. Phase contacting on a single stage. point efficiency can also be defined.
114 STAGE OPERATIONS

Point efficiencies can be related to stage

efficiencies by assuming a physical model of liquid
mixing. If it is assumed that all of the liquid in a

stage is thoroughly mixed so that its composition is

uniform, then x =x n all across the stage, so that
y = yn and y* = y n *. Then the point and stage effi-

ciencies are equal: E v = E v °. On the other hand, if

no liquid mixing is assumed, it can be shown (7)

that

Ev =—~ie - 1. (7.36)
mv
where

L = molar flow rate of liquid

V = molar flow rate of vapor
m = slope of equilibrium equation Figure 7.16. Stepping off actual stages for a
(y* =mx +b) Murphree vapor efficiency of 50 percent.

This equation assumes no mixing in the liquid-flow

reference to Figure 7.11/?, where it is seen to be
direction, and no composition gradients perpen-
the ratio of the two distances {y n — yn + i) and
dicular to the flow direction. It also assumes a
linear equilibrium relationship in the composition
range existing in the stage. E v ° is assumed to be
When actual stages are stepped off between
the equilibrium curve and operating line, the full
constant in the stage. E v can clearly exceed 1.0 if
vertical distance (y n * — y n + is not used. Instead,
-\ )
the liquid flow follows this model. For example,
for E v ° -0.9 and mV/L =0.5, E v = 1.12. In an
some fraction of the vertical distance as deter-
stage, the
mined by the efficiency of the stage is stepped off.
actual liquid follows neither of these
Figure 7.16 illustrates a typical problem in distil-
models exactly; it is neither totally mixed nor
lation with a Murphree vapor efficiency of 50
totally unmixed. As a result, an intermediate
percent. It is convenient to draw an "effective
model involving partial mixing may be necessary.
separation curve" between the equilibrium curve
In Chapter 16, E v ° is related to the rate of
and operating line at a distance determined by the
mass transfer. Reference 6 gives a more complete
efficiency. In this case, the curve is located at half
summary of stage efficiencies, specifically oriented
the vertical distance between the equilibrium curve
toward distillation. In this reference E v is approxi-
and operating line, because the vapor efficiency is
mately related to E Alternatively, the procedure
.

50 percent. The stages are then stepped off

given in the folowing paragraph may be used.
between the effective separation curve and the
The Murphree stage efficiency can be used
operating line.
with the methods outlined in this chapter to
calculate the number of actual stages required for a
given separation. Consider the operating lines for a ANALYTICAL CALCULATION OF STAGES
single stage shown in Figure 7.16. The operating
For simple countercurrent flow in cases where
line for the equilibrium stage runs from the enter-
both the operating line and equilibrium curve are
ing conditions (x„_ 1f /„+-,) to the leaving condi-
straight, the number of equilibrium stages may be
tions (x n ,y n ). The latter point lies on the equili-
calculated analytically. This situation occasionally
brium curve because x n and y n are in equilibrium.
arises in operations such as gas absorption.
In an actual stage (Figure 7.1 1o), the entering
A material balance around the first stage in
conditions are the same, but the compositions (x n
Figure 5.2 gives for dilute gases (where L and V are
and y n of the phases leaving do not lie on the equi-
)
nearly constant)
librium curve, because the phases are not in equili-
brium. The composition y n * is in equilibrium with Lx + Vy 2 - + Vy^
Lx-i
the liquid composition x n leaving the actual stage,
or
as shown. The definition of Murphree V-phase
efficiency can be more fully understood by L(x -Xt) = l/(yi — /2) (7.37)
 SIMPLIFIED CALCULATION METHODS 115

The equilibrium curve will be a straight line if Equation 7.46 is useful in calculating the separa-
Henry's law applies tion if A and N are known. It can be derived in a

more direct manner using the calculus of finite

V\ = K*i
differences (4). Equation 7.46 can be solved for
Substituting forx! in Equation 7.37 gives the total number of stages. Cross multiplication
gives

A N "y N + i
-A N + 'y, - //v + + Kl 1

and -* W + V* + -A»"Kx
= /\/V 1

K2 + — *o -Ay N + +AKx i

V\
= (7.38) A» + Hy,- Kx )

+ 1 = A (y N + - /Cx - y N + + K1
KV ! ) 1

The group L/KV is called the absorption factor and = (A- IMy/v-M -/Cx )+Ki-/Cx
is designated by>4. Equation 7.38 becomes

y 2' + KAxq ^+i-(4-i)(!5LtlZ^»Ui

Ki= (7.39) \ /! - Kx /
A + 11
A similar analysis for stage 2 gives

K2
+ KAx,
———
= K3 : :
=
y 3 : +Ay
— —— :

(7.40)
-(^)(
Taking the logarithm of both sides gives
K/v +

/t
1

- Kx
— K*o
K
1 A
A+^ A+1
If

in
the value of y\ from Equation 7.39
Equation 7.40
is substituted
N= ^(^K^fK] \ogA
(7.47)
2
(A + 1)y 3 + KA x
= (7.41 Equation 7.47 can be used to calculate the number
/2
A 2 +A + 1
of stages required for a given change in composi-

- tion of the gas phase in absorption. The equation

If Equation 7.41 is multiplied by (A 1)/(/4- 1),
may also be written dimensions, in mole-ratio
providing the equilibrium curve and operating line
(A
2
-\)y 2 +A (A-'i)Kx Q 2
= (7.42) are straight on mole-ratio coordinates. Use of
V2
W -D3
Equation 7.47 with mole ratios requires the
An assumption of dilute solutions to ensure a straight
expression for y 3 may be derived in a similar
manner. If this process is continued to stage N, equilibrium curve.
For stripping columns where mass is trans-
(A N -^y N + ,+A N (A-^Kx ferred from the liquid to the vapor phase,
vn =
s^~ (7 ]

/V + 1

2
A material balance over the entire cascade gives Xp —X/y g) A
(7.48)
L{x -x N )
= V{y y
-y N + i) (7.44)

The term (MA)

*-r ar-i referred to the stripping
Because x N = yrv/K, Equation 7.44 becomes is as
factor.

L(x -*!\=V(y,-yN+ ,) (7.45)

Analytical Calculation of Minimum Stages at Total Reflux

Eliminating yN between Equations 7.43 and 7.45 An first developed by Fenske (1) is
expression
gives useful determining analytically the minimum
in
number of stages required at total reflux. Use of
K/V+1-/1 A N + '-A this equation is limited to systems with constant
(7.46)
^ w -1
+ 1
//v + -/Ofo
1
relative volatility.
116 STAGE OPERATIONS

REFERENCES
By the definition of relative volatility,

_K/y_ = a
_j^v_ (7.49) 1. Fenske, M. R., Ind. Eng. Chem., 24, p. 482 (1932).

1 - Vn 1 - xn 2. Luyben, W. L, Process Modeling, Simulation, and Con-

trol for Chemical Engineers, McGraw-Hill, New York
for the last stage. At total reflux the operating line (1973).
coincides with the diagonal and yN = x N _ 1 , so that 3. McCabe, W. L, and E. W. Thiele, Ind. Eng. Chem., 17,
p.605(1925).
= a (7.50) 4. Mickley, H. S., T. K. Shewood, C. E. Reed, Applied
1 - X/v _ -, 1 - xB Mathematics in Chemical Engineering, McGraw-Hill,
New York (1957).
For stage N— 1,
5. Murphee, E. V., Ind. Eng. Chem., 17, p. 747 (1925).
6. Schoenborn, E. M., et al.. Tray Efficiencies in Distilla-
Kjv^i__
= a _J^JL_ = a
ix
2_^t_
-
(7 . 51 tion Columns, A.I.Ch.E. (1959).
1 - K/v-1 1 - xn- - xn 1 '
7. Smith, B. D., Design of Equilibrium Stage Processes,
McGraw-Hill, New York (1963).
If this procedure is continued to stage 1

/1 */V
=a» (7.52) PROBLEMS
1 -/! 1 -X/v
For a total condenser x D = Ki , and for the case of 7.1. In your own words, describe the basis of deriva-

the bottoms withdrawn from the bottom plate tion and application of the operating line.

*b = x n Therefore,
7.2. Under what conditions is it convenient to use an
operating line on an equilibrium diagram to determine
XD - N_ x B_ stages required in liquid extraction? the operating line
a (7.53) Is

- xD 1 - xB
1 useful for the system shown in Figure 3.8? In Figure 3.9?
Make a sketch of a ternary diagram for a system where the
and
operating line would be useful. Are L and V constant from
xD ( 1 - xB )
stage to stage for the sketched system? Explain.
log 7.3. Sketch an enthalpy-composition diagram for a

=
xg (1 -x p )
binary system that would be susceptible to analysis by an
/V (7.54)
log a operating line on an equilibrium diagram. Would you
expect the system shown in Figure 3.4 to give a straight
operating line?
Computer Calculations 7.4. Using a material balance around a portion of a
All the calculations of this chapter may be pro- simple countercurrent cascade from stage /7+1 to N, derive
grammed for the digital computer, providing equili- an equation for the operating line [where y n + -\
- f{x n )]

brium data are available in the form of empirical How is this equation related to Equation 7.3?
equations or some other form appropriate to 7.5. Gas a is soluble in liquid c, but gas b is com-
computer calculations. For example, stage-to-stage pletely insoluble. The solubility relationship isya = Mxa
calculations can be made by a program that where M
is a Henry's law constant. Draw operating lines
alternately applies the equations for the equili- and equilibrium curves for the following situations and
brium curve and the operating line. The program complete the calculation specified:
would follow the same steps as required in a (a) Lq, Vw+i, xq, //v+i and x/y are known. Find N
graphical solution. The choice between a computer for a gas absorption.

solution and a graphical solution depends upon the (b) Repeat (a) but for stripping.
accuracy required and upon the number of times (c) For an absorber equivalent to N equilibrium
the calculation is to be made. For example, the stages, find the outlet gas composition given Lq,xq, V/y+1,
application of the McCabe-Thiele method to //V +1 ,and/V.
determine the number of stages required to (d) Lq, Xq, x/v, and //v+1 are known. Find the
produce a very high-purity distillate may be lowest possible exit gas composition attainable in gas
impractical graphically because of inaccuracies absorption.
involved in stepping-off stages in a confined space 7.6. It is always safer to use mole ratio instead of
at high purities. Here a computer solution can give mole fraction coordinates in calculation of gas absorption
much greater accuracy, presuming the equilibrium using an operating line on an equilibrium diagram. Why?
data are accurate. What disadvantage may arise?
 SIMPLIFIED CALCULATION METHODS 117

7.7. Can a minimum LA/ be determined in stripping? The flow rate of fresh gas is 3.0 standard /7?
3 /s and the flow
Explain. rate of charcoal is 2.5 kg/s.
7.8. Solve Problem 5.13(b) and (c) using an opera- (a) Calculate the number of theoretical stages
ting line on an equilibrium diagram. required to reduce a 10 percent acetone-air mixture to 1

7.9. Solve Problem 5.17 using an operating line on an percent acetone (by volume) at 30°C and normal atmo-
equilibrium diagram. What assumptions must be made to spheric pressure.

assure a straight operating line? Check the accuracy of the (b) What effect would increasing the total pressure
assumptions by comparing your results with those of have?
Problem 5.17 solved using a ternary diagram. (c) What effect would increasing the temperature
7.10. Solve Problem 5.19 using an operating line on an have?
equilibrium diagram. What assumptions must be made to (d) What is the minimum flow rate of charcoal for
assure a straight operating line? Test the accuracy of the the conditions of part (a)?
assumptions by comparing your results with those of 7.14. Aqueous triethanolamine solutions are used to
Problem 5.19 solved using a ternary diagram. absorb carbon dioxide from a gaseous mixture as a purifica-

7.11. An absorption column equivalent to five equili- tion step in the manufacture of dry ice.The carbon dioxide
brium stages is to be used to remove ethane from a is absorbed by the amine solution, but the other compo-
hydrogen-ethane mixture at 300 psia. The entering nents of the gas are not. The triethanolamine solution is

absorbent liquid contains 99 percent hexane and 1 percent subsequently heated to recover the pure carbon dioxide.
ethane. The entering gas contains 35 percent ethane and 65 A gaseous mixture containing 25 percent CO2 is to be
percent hydrogen. The column is at a constant temperature scrubbed with a 1 molar solution of triethanolamine at
of 75° F. Because of the fluid flow characteristics of the 25 C and normal atmospheric pressure.
column, the mole ratio of entering absorbent to entering (a) What is minimum flow rate of triethanolamine

gas cannot exceed 3 to 1 at the desired flow rate of gas. solution necessary to reduce 10 moles/s of entering gas to
What is the maximum possible recovery of ethane if this an exit concentration of 1 percent?
column is used? (b) How many equilibrium stages are required at 1.5
7.12. Propane is to be stripped from a nonvolatile oil times the minimum amine flow rate? (The density of amine
by steam in a countercurrent tower. Four moles of steam solution may be assumed to be the same as water.)
will be supplied at the bottom of the tower for every 100 (c) In many cases the scrubbing solution contains a

moles of oil-propane feed at the top. The oil originally small amount of CO2 as it is recycled from the CO2
contains 2.5 mole percent propane, and this concentration stripper. What would be the minimum solution rate if it

must be reduced to 0.25 mole percent. The tower is main- enters with a concentration of 0.10 moles CO2 per mole
tained at 280° F and 35 psia. The molecular weight of the amine?
heavy 300, and the molecular weight of propane is 44.
oil is (d) How many stages would be required for 1.5 times
The equilibrium relationship is y = 33.4x where y = mole the minimum solution rate with the entering solution of
fraction propane in vapor and x = mole fraction propane in part (c)? {Note: The mole ratio of CO2 to solution may be
liquid. used, but since the ratio of water to triethanolamine is

(a) How many equilibrium stages are required? constant, the ratio X= moles CO2 per mole amine can be
(b) If the pressure is increased to 70 psia how many used directly.)
equilibrium stages would be required?
(c) What would be the effect of lowering the tem- Equilibrium Data for 1 Molar Triethanolamine at 25° C [Mason
perature of stripping? and Dodge, Trans. A.I.Ch.E., 32, p. 27 (1936)]

(d) What would be the minimum steam flow rate

Partial pressure of
necessary to give the same recovery of propane?
CO2 in gas, kN/m 2 0.19 1.44 5.79 12.9 3.45 9.63
7.13. Acetone is to be removed from an acetone-air Liquid concentra-
mixture at 30°C and normal atmospheric pressure by tion, mole
adsorption on charcoal in a simple countercurrent process. C02/mole amine 0.0587 0.161 0.294 0.424 0.612 0.825

The following equilibrium data were taken on the charcoal

to be used at 30° C: 7.15. Ammonia is to be absorbed from an air mix-
ture by contacting it with water at 20°C. The entering gas
Grams acetone composition is 10 percent ammonia (by volume) and the
adsorbed/gram charcoal 0.1 0.2 0.3 0.35 exit gas is to contain 1 percent ammonia. Flow of entering
gas is 10 moles/s at 20°C and normal atmospheric pressure.
Partial pressure acetone, (a) Determine the minimum entering water flow rate
kN/m 2 0.266 1.48 5.60 12.3 needed to accomplish this separation.
118 STAGE OPERATIONS

(b) Determine the number of equilibrium stages re- and 70 mole percent toluene is to be fractionated at normal

quired and the exit liquid composition at 125 percent of atmospheric pressure in a column with a total condenser
the minimum entering water flow rate. and a still from which the bottoms are withdrawn. The
(c) Suppose that all of the liquid leaving the absorp- distillate is to contain 95 mole percent benzene and the
tion column in (b) is fed to the top of a stripping tower bottoms 4 mole percent benzene. The feed is at its dew
operating at 20° C. Pure air at 20°C is fed into the bottom point.
at a rate of 9 moles/s. Determine the exit gas and liquid (a) What is the minimum reflux ratio (Lq/D)?
compositions for these conditions, assuming that the strip- (b) What is the minimum number of equilibrium
per is equivalent to the same number of equilibrium stages stages in the column required at total reflux?

as the absorber in (b). (c) How many equilibrium stages are required at a
7.16. For Problem 6.21 determine the number of reflux ratio of 8?
equilibrium stages required using the solvent-free equi- (d) How many equilibrium stages would be required
librium diagram. Assume that the mutual solubility of cot- at a reflux ratio of 8 if the feed were a liquid at its bubble
tonseed oil and propane does not vary with concentration point?
of oleic acid. Is this assumption reasonable? 7.28. Solve Problem 6.26 on an equilibrium diagram,

7.17. Derive Equations 7.46 and 7.56 starting with a assuming constant molar overflow.
distillation column with reflux at both ends. 7.29. Solve Problem 6.25 on an equilibrium x-y dia-

7.18. Derive the equation for the /-line starting with gram.
the typical flow pattern in a distillation column with reflux 7.30. Solve Problem 6.28 on an equilibrium x-y dia-
at both ends. gram.
7.19. Verify the list of values of / and /-line slopes 7.31. An aqueous solution containing 20 mole per-
given in Figure 7.4 using an enthalpy-composition diagram. cent ethanol is 1 atm into a distillate
to be fractionated at
7.20. (a) For a distillation system with constant of composition 80 mole percent ethanol and a waste of
molar overflow, relate Figure 7.6a to the same calculation composition 3 mole percent ethanol. The feed is at 60" F. A
carried out on an enthalpy-composition diagram. Show the distillation column equivalent to seven theoretical plates is

relationship between the two graphic procedures. available. What reflux ratio (LqID) should be used?

(b) Repeat the comparison for Figure 7.7. 7.32. Solve Problem 6.29(a) and (b) on an equilibrium
7.21. Using an enthalpy-concentration diagram, show x-y diagram.
geometrically that constant molar overflow results when 7.33. Solve Problem 6.34 on an equilibrium x-y dia-

the latent heats of vaporization are equal for all com-, gram.
positions. Neglect heat effects other than the latent heat, 7.34. A mixture of 30 mole percent ethanol, 70 mole
but do not assume that the enthalpy of either phase is percent water is to be fractionated at 1 atm into a distillate
constant. of 82 mole percent ethanol and a bottoms of 4 mole
7.22. Derive an equation for the enriching-section percent ethanol. There will be a total condenser at the top
operating line in distillation with condenser. Draw
a partial of the tower, but, instead of a reboiler, saturated steam will
a sketch to show how this line is plotted and label impor- be fed directly into the bottom of the tower to supply
tant points. necessary heat The feed is 40 mole percent vapor, 60 mole
7.23. Derive an equation for the stripping-section percent liquid. Molar steam rate is 70°o of bottoms.
operating line where the bottoms are withdrawn through (a) Draw a flow diagram for the tower, and label

the reboiler. Draw a sketch showing a plot of the equation entering and leaving streams.
and label points. (b) Derive an equation for the operating line in the
7.24. Construct an enthalpy composition diagram stripping section.
showing the information given in Figure 7.8. Show how the (c) Determine the number of equilibrium stages re-

point (Zv,/ir) can be located on the enthalpy-composition quired.

diagram. (d) Determine the reflux ratio (LqID).
7.25. Make a sketch similar to Figure 7.8 showing (e) Determine the minimum steam rate (per mole of
how the operating lines can be located for the case where bottoms) which would permit this separation.
there is one intermediate liquid product between the feed 7.35. An equimolar mixture of ethanol and water is

and the bottom of the column. to be fractionally distilled to produce a distillate of com-
7.26. Solve Problems 6.23 and 6.24 on an equi- position 0.80 mole fraction ethanol and bottoms of 0.05
librium diagram assuming constant molar overflow. For mole fraction ethanol. The feed is a saturated liquid and
which case is the assumption of constant molar overflow there is a total condenser and a total reboiler.
more valid? (a) At a reflux ratio of 2.0, how many equilibrium
7.27. A mixture containing 30 mole percent benzene stages are required?
 SIMPLIFIED CALCULATION METHODS 119

(b) How many stages are required at total reflux? and 35 mole percent toluene is distilled with the vapor
(c) What is the minimum reflux ratio? continuously withdrawn as it is formed.
(d) What percentage of the feed ethanol is recovered (a) What is the composition of the liquid after 25
in the distillate? mole percent of the original liquid has been vaporized?
7.36. A saturated liquid mixture of 40 mole percent (b) How much of the original liquid will have vapor-
ethane and 60 mole percent ethylene is to be separated by
ized when the equilibrium vapor has a composition of 65
distillation. column must operate within the
The distillation mole percent benzene?
temperature range —20 to +50° C. The lower limit is set by 7.41. A vapor mixture of 40 mole percent hexane
the temperature of the available cooling fluid for the con- and 60 mole percent heptane is slowly condensed at normal
denser. The upper limit is set because of possible reaction atmospheric pressure and the equilibrium liquid is with-
of the ethylene at higher temperatures. The distillate is to drawn as it is formed.
have a composition of 90 mole percent ethylene, and 95
(a) What is the composition of the vapor after 50
percent of the feed ethylene is to be recovered in the
mole percent of has condensed?
it
distillate.
(b) If all the liquid formed were allowed to remain in
(a) Determine a suitable pressure for column oper- contact with the vapor what would be the composition of
ation from data given in Appendix D. the vapor after 50 mole percent had condensed?
(b) Determine the minimum reflux ratio and mini- 7.42. An equimolar mixture of benzene and toluene
mum stages at this pressure. is charged to the still pot of a batch distillation tower which
(c) Choose a reflux ratio that permits operation at
is equivalent to six equilibrium stages. The distillation is

twice the minimum number of stages. There will be a total

carried out at normal atmospheric pressure at a constant
condenser and a total reboiler.
reflux ratio of 3 until the concentration of benzene in the
7.37. For Problem 7.35(a), use enthalpy-composition still pot is reduced to 5 mole percent.
data and delta points to construct curved operating lines on
(a) What is the average composition of the distillate?
the equilibrium diagram. Then determine the number of
(b) What is the overall percentage recovery of ben-
stages required using the curved operating lines. Is the
zene in the distillate?
assumption of constant molar overflow accurate in Problem
(c) What is the temperature in the still pot at the
7.35(a)?
beginning and at the end of the run?
7.38. An amount of 1000 lb moles/hr of a 25 mole
7.43. Ninety-five percent of the ethylene glycol in a
percent ethanol-water mixture is to be fractionated at 1
20 mole percent ethylene glycol 80 mole percent water
atm into two ethanol-rich streams of compositions 60 per-
mixture must be recovered in the residue from a batch
cent and 80 percent ethanol; 98 percent of the ethanol in
distillation tower equivalent to three equilibrium stages
the feed is to be recovered in these two products. That is,
(including the still pot) operating at 228 mm Hg. An overall
only 2 percent of the feed ethanol may go out in the
distillate composition of 5 mole percent ethylene glycol is
bottoms, which are withdrawn from the reboiler. Equal
desired.
60 percent and 80 percent products
quantities (in moles) of
(a) If the distillate composition is held constant,
are to be produced. The feed, the bottoms, and the 60
percent product are liquids at the bubble point. The 80
what must be the initial and final reflux ratios?

(b) If the reflux ratio is held constant, what must be

percent product is a saturated vapor withdrawn from a
its value? (Equilibrium data are given at the end of these
partial condenser. The reflux ratio (LqID) at the top of the
problems.)
column isto be 3.0. The overall plate efficiency is 25
percent. Assuming constant molar overflow and using the
7.44. A mixture of 36 mole percent chloroform and
64 mole percent benzene is to be separated using a batch
x-y diagram, calculate:
distillation tower equivalent to ten equilibrium stages
(a) The number of actual plates required for the

desired separation.
operating at 760 mm The distillate composition will be
Hg.
held constant at 96 mole percent chloroform during the
(b) The actual plate on which to put the feed.
run.
(c) The actual plate from which the 60 percent prod-
uct is withdrawn. (a) Plot the reflux ratio as a function of the percent-

(d) The reboiler duty (Btu/hr). age of original charge distilled.

(e) The condenser duty (Btu/hr). (b) What is the theoretical maximum recovery of

(f) The minimum reflux ratio at the top of the chloroform in the 96 percent distillate? (Equilibrium data

column. are given at the end of these problems.)

(g) The minimum number of stages at total reflux. 7.45. Two mixtures of a and b are to be fractionated
7.39. Derive Equations 7.23 and 7.24. at constant pressure in a distillation column equipped with
7.40. A liquid mixture of 65 mole percent benzene a total condenser and a total reboiler.
120 STAGE OPERATIONS

(c) The concentration of the more volatile com-

Composition Thermal
ponent in the feed is reduced.
Quantity xa condition
(d) The reboiler becomes fouled and the heat it can
supply is substantially decreased.
Feed 1 50 moles 0.50 Saturated liquid
(e) The feed (formerly a liquid) is now fed in as a
Feed 2 100 moles 0.35 Saturated vapor
vapor.
Distillate 0.9 Saturated liquid
7.51. What results from Equation 7.46 when = °°? N
Bottoms 0.05 Saturated liquid
show that this result is equivalent to corresponding graph-
aa _ b = 4.0 Reflux ratio (L /D) = 1.5 ical calculations.
7.52. Derive an expression for {L/V) min from Equa-
The Murphree vapor efficiency varies with the liquid tion 7.46. Can A for the minimum (LlV) be greater than
composition: 1.0?
7.53. The ammonia concentration of an air-ammonia
xa 0.05 0.2 0.4 0.6 0.8 0.9
mixture is from 3 percent to 0.05 percent by
to be reduced
Ev 0.67 0.67 0.67 0.50 0.50 0.50 volume by scrubbing the gas with water in a countercurrent
towerat 90° F and 1 atm pressure. The flow rate of entering
(a) How many actual stages are required?
(b) At what stage is each feed introduced?
gas is 5000 ft /hr (90°F, 1 atm). For ammonia in aqueous
solution in the concentration range, the equilibrium ex-
(c) What is the overall stage efficiency? is

7.46. An aqueous solution containing 35 mole per-

pressed by y = 0.1 85x.
cent ammonia at 100 psia and 150°F is to be fractionated (a) Determine graphically the minimum water rate

into a distillate of 95 mole percent ammonia and a bottoms and the number of stages required at twice the minimum
of 3 mole percent ammonia. There is a total condenser, and water rate.

the bottoms product is withdrawn from the bottom plate.

(b) Determine analytically the minimum water rate

The reflux ratio Lq/D is 1. and the number of stages required at twice the minimum
(a) Construct curved operating lines on an equi- water rate.

librium diagram for this separation. 7.54. Derive Equation 7.48.

(b) Determine the number of actual stages required 7.55. In what way should Equation 7.54 be modified
at a Murphree vapor efficiency of 25 percent. to be valid for a partial condenser and for bottoms with-

(c) Determine the overall stage efficiency. drawn through the reboiler?

7.47. In Chapter 2 the stage efficiency was defined as 7.56. Solve Problem 7.27(b) analytically.

the ratio of the composition change in an actual stage to 7.57. Develop a computer program that performs
the composition change that would result in an equilibrium calculations necessary to design a distillation column to
stage. Using a diagram similar to Figure 7.14, show how this separate binary mixtures of constant relative volatility and
definition differs from the Murphree stage efficiency. constant molar overflow. Pattern your program after the
7.48. Determine the number of actual stages required stage-to-stage calculations developed in this chapter, using

in Problem 7.27(c) for: an operating line and an equilibrium curve. Given relative

(a) An overall stage efficiency of 60 percent. volatility, feed rate, feed composition, and feed thermal
(b) A Murphree vapor efficiency of 60 percent. condition, desired product purities (xg andx ), and reflux
(c) A Murphree liquid efficiency of 60 percent. ratio, your program should be able to calculate distillate
7.49. Under what circumstances is the overall effi- and bottoms products rates, the number of equilibrium
ciency of a column numerically equal to the Murphree stages and feed location. Assume a total condenser and a
vapor efficiency? total reboiler.

7.50. An existing distillation column is being used to Use your program to determine the number of stages
fractionate a feed into the desired distillate and bottom for the following situations:
products. Predict what the effect each of the following
changes will have on the purity of the distillate and bot- a = 2.5, F = 100,Z/r = 0.5,<7= '• L /D = 2.Q

toms, other independent variables in the column remaining (a) x D =0.95x e =0.95.
unchanged. Assume that the stage efficiency never changes (b) x D =0.99 x B =0.99.
and that the feed can be relocated over several plates on (c) x D = 0.999 x B = 0.999.
either side of the original feed nozzle, if this is necessary to
maintain high purity distillate and bottoms. 7.58. Use the program developed in Problem 7.57 to
(a) The relative volatility of the feed is lowered (feed find the effect of varying reflux ratio (from total reflux to
composition and enthalpy characteristics unchanged). approaching minimum reflux) on the number of stages
(b) The reflux ratio {L /D) is increased. required for the separation of Problem 7.57(a).
 SIMPLIFIED CALCULATION METHODS 121

7.59. Modify the program of Problem 7.57 so that it be adjusted to the optimum location. Use the program to
calculates minimum reflux and then calculates the number solve the following problem. For a column equivalent to 40
of stages for an actual reflux expressed as a percentage equilibrium stages, what reflux ratio should be used for:
(exceeding 100) of minimum reflux. Then for the separ- a = 2.5, F = 1 00, ZF = 0.5, q= 1, xB = 0.99, x D = 0.99.
ation of Problem 7.57(a), calculate N for an actual reflux 7.61. Using the program developed in Problem 7.60,

ratio equal to 101, 102, 105, 110, and 200 percent of the determine what happens to the reflux ratio as product
minimum reflux. purity increases. Take several values of product purity,
7.60. Develop a computer program that will calculate assuming distillate and bottoms are equally pure in their
the reflux ratio required to produce products of a stated major component, starting with the purities and other con-
purity from a feed of known characteristics, given the ditions of Problem 7.60. What is the maximum purity
number of stages in the column. Assume the feed stage can attainable for the feed and column of Problem 7.60?

EQUILIBRIUM DATA:* CHLOROFORM-BENZENE AT 760 mm Hg

Mole percent chloroform

Liquid 8 15 22 29 36 44 54 66 79 100

Vapor 10 20 30 40 50 60 70 80 90 100

Temperature, °C 80.6 79.8 79.0 78.2 77.3 76.4 75.3 74.0 71.9 68.9 61.4

'Internationa/ Critical Tables, McGraw-Hill. New York. Copyright © 1928; by permission of Natl. Acad, of Sc.

EQUILIBRIUM DATA:* ETHYLENE GLYCOL-WATER AT 228 mm Hg

Mole percent ethylene glya )l

Liquid 10 20 30 40 50 60 70 80 90 92 95 97 99 100

Vapor 0.2 0.4 0.7 1.1 1.8 2.8 5.0 10.0 21.4 25.5 35.0 46.5 69.0 100

Temperature, ^C j 69.5 72.8 75.6 78.8 82.9 87.7 93.1 100.5 111.2 127.5 132.0 139.5 145.1 152.4 160.6

'Trimble and Potts, Ind. Eng. Chern., 27. p. 66 (1935). Courtesy Amer. Chem. Soc.
 Eight

Multicomponent Stage Operations

The typical industrial chemical process often related trial-and-error procedures. Before the
requires of a mixture containing
the separation advent of the digital computer, many simplifying
several components. For example, the product assumptions were made and approximate methods
stream from a chemical reactor may contain not were developed to estimate the number of stages
only the desired product but also remaining excess and the reflux rates required for a multicomponent
reactant and products of the various side reactions. separation. Digital computers have permitted more
These reactants and side products must be iden- rigorous stage-by-stage calculations involving
tified and separated from the product stream in successive iterations until answer is
a correct
order to recover a final product of high purity. achieved. As a result, many of the approximate
Multicomponent mixtures, such as the product shortcut methods are no longer necessary, but
stream from the chemical reactor, are frequently several of the approximate methods useful in pre-
separated by the various stage operations discussed liminary calculations are presented later.
in Chapters 2 through 7. In the earlier chapters, the The next section applies the phase rule to an
principles of phase equilibrium were applied to analysis of multicomponent stage operations. The
separation of binary mixtures. Although the same following section considers multicomponent
principles apply to the separation of multi- absorption. Then two important methods of stage-
component mixtures, the calculation methods are by-stage calculation for multicomponent distil-
much more complex, because the additional com- lation are presented.
ponents require that a greater number of intensive
variables be independently adjustable, as shown by
COUNTING VARIABLES IN STAGE OPERATIONS
application of the phase rule in the next section.
Although multicomponent mixtures may be In the previous chapters on binary distillation,
separable by any of the various mass transfer absorption, and extraction, a number of variables
operations discussed in Chapter 2, the most widely are fixed before calculation to determine the
applied are absorption and distillation. Multi- remaining characteristics of the system. For
component distillation and absorption are example, in a simple countercurrent liai'id-extrac-
especially important in the petroleum industry. tion cascade operating at a known temperature and
Crude oil is a mixture of hundreds of organic pressure (Figure 8.1), the flow rates and composi-
compounds, and most petroleum processes require tions of the feed (L , x Qa ) and of the solvent
the separation and purification of multicomponent (v'/v + K K/v + i.a) are frequently specified, along
mixtures. The industrial plants pictured in Figures with the desired percentage recovery of a given
6.3 and 8.10 are processing multicomponent mix- component in the extract. With these five indepen-
tures. Multicomponent mixtures are common in dent variables specified, the number of equilibrium
industry, whereas simple binary mixtures almost stages required for the separation may be calcu-
never occur. lated. In this case, the feed and solvent contain no
Rigorous multicomponent calculations are more than two components each, and the phase-
extremely complex, often involving several inter- equilibrium relationships are known. On the other

122

vn
y»i
Tn or Hn
Pn Stage n
Tn .Pn
n the composition of a C-component mixture, it is
n
h n -\ or Tn - 1
Pn-l
<7n
Figure 8.1. Variables in a single
hand, it would be possible to fix L x 0a V/v +
Z/v + 1 ,a> ar, d the number of equilibrium stages and
then calculate the percentage recovery. In fact,
specification of any five of these six independent
variables permits calculation of the remaining one.
Suppose now that the feed contains three
components. In this case, one independent variable
is added (either x 0b or x
0c and there are now seven
)
independent variables, six of which must be set to
permit evaluation of the seventh. Similar analyses
must be made in determining the number of stages
required or the recovery in distillation, absorption,
leaching, and other stage operations. For binary
mixtures, the number of variables that must be
specified is almost self-evident, so that it requires
little thought. But for multicomponent mixtures, a
careful analysis must be made to determine how
many and which independent variables can be
fixed to define the system completely. This section
presents a systematic method for determining the
number of variables in stage operations. The meth-
od was developed by Gilliland and Reed (2) and
later extended by Kwauk (7) and Smith (12).
Independent Variables
An independent variable is any characteristic of a
system that can in principle be varied directly at
the option of the operator. For example, in a
closed, piston-equipped cylinder of an ideal gas, the
temperature, pressure, or volume might be varied.
The temperature could be raised by heating the
cylinder. The volume could be decreased by push-
ing in the piston, and this would also tend to
increase the pressure. But clearly the three vari-
ables are not all independent. A restrictive relation-
ship exists among them. This relationship in its
simplest form is the ideal gas law:
PV=nRT
For the system the cylinder, n is constant, so
in
that fixing any two of the three variables (P, V, T)
automatically fixes the third. Therefore, only two
of the three variables are independent, because if
MULTICOMPONENT STAGE OPERATIONS 123
Vn + i two have been set at specific values, it is not
yn + \. i
possible to vary the third.
W n + i or 7'n+i
Pn + 1 Another common example is specifying the
composition of a mixture. To specify completely
*n or T„ necessary to specify C concentrations when they
are expressed in units of mass per unit volume; if
composition is expressed in mole fraction, mass
equilibrium stage. fraction, or volume fraction, only C — 1 values must
be stated, because the sum of the fractions is 1.
For example, in a system containing components a,
, , i ,
b, c, and d, the concentrations are related by
Ca + Cb + Cc + Cd - Ctota (8.1
i
where each term might be in pound-moles per
cubic foot. Four of the five quantities appearing
must be specified to define the system. On the
other hand, if Equation 8.1 is divided by Ctota i,
there results
x a +x b +x c +x d = *\.0 (8.2)
where the x's are mole fractions of each compo-
nent. Now the term on the right is fixed, so that only
three mole fractions must be specified to define
the system, and only three can be varied independ-
ently. As a result, in systems with compositions
expressed as concentrations, there are C independ-
ent composition variables, whereas when com-
positions are expressed as fractions totalling 1.0,
there are only C — 1 independent composition vari-
ables. Because any three of the four mole fractions
of Equation 8.2 serve to define the system, all four
might be listed as independent variables, but it is
understood that when three have been specified,
the fourth is no longer an independent variable
because the sum of the mole fractions must be 1.0
In these simple examples, the number of in-
dependent variables that must be fixed to define
the system is quite obvious. However, in more
complex multistream, multiphase systems (such as
a distillation column), the number of specifiable
independent variables is far from obvious. In the
following sections the phase rule, together with
mass and energy balances, will be used to analyze
complex systems in stage operations to determine
how many independent variables must be fixed to
define the system. In these analyses, the simple
considerations of this section must not be forgot-
ten.
The Phase Rule
The phase rule was discussed briefly in Chapter 3.
It is a relationship derived from thermodynamics
124 STAGE OPERATIONS

(1) used todetermine the number of intensive It should be emphasized that the phase rule
variables that may be varied independently in a applies only to intensive variables. Extensive vari-
chemical equilibrium. The phase rule may be ap- ables, such as rates of flow and heat addition, are
plied to reaction equilibria and to phase equilibria. restricted by appropriate mass and energy balances.
Stage operations are concerned with single phases
and with equilibrium between two phases, so the
phase rule may be used to determine relationships Equilibrium Stage
among the intensive variables of the fluid streams. The method for determining the number of inde-
In mass and energy balances may be
addition, pendent variables in a staged operation is based
applied to determine relationships among the ex- upon mass balances, energy balances, and on the
tensive variables of the system.
phase-equilibrium relationships, as described by the
An intensive variable does not depend upon phase rule. These three relationships may be ap-
the total quantity of the phase (or in the flow plied to a single equilibrium stage (Figure 8.1 ). The
systems, it does not depend upon the flow rate of
two streams entering and the two streams leaving
the phase). In stage operations, the temperature, have a specific composition and enthalpy associ-
the pressure, and the composition are the intensive ated with each of them. In addition, heat may be
variables usually of interest. Another intensive vari-
added or removed from the stage. q n has a
If
able frequently appearing is the enthalpy per
positive value, heat is added; for a negative q n heat ,

pound of a given phase. is removed.

By the phase rule (Equation 3.3) for a single-
The two entering streams are not in equi-
phase mixture of C components librium so they are considered separately. Appli-
cation of the phase rule to the single phase L n
F=C+2- =C+1 1 (8.3) shows that Equation 8.3 applies, and there are
C+1 possible independent variables associated
so that C+ 1 intensive variables may be specified with Ln . A similar analysis yields C+1 inde-
to define the system. With compositions expressed pendent variables for l/n + 1
. The two streams leav-

as mass or mole fractions, there are C — 1 com- ing {L n and Vn are equilibrium and are there-
) in

positions to be set, plus any two of the three
remaining variables (T,P,h). In Figure 3.4, the
pressure is fixed in a vapor-liquid equilibrium, so
that specifying a single composition and the two
intensive state variables, P and T defines the sys-
tem in the one-phase region. On the other hand, in
the two-phase region for the two-component sys-
fore considered together. For these combined
streams P = 2, so that by Equation 8.4 there are C
independent intensive variables associated with the
equilibrium mixture leaving the stage. The total
number of independent intensive variables associ-
ated with the streams flowing to and from the
stage then (C + 1 + (C + 1 + C = 3C + 2.
is ) )

tem independent extensive variables may

Several
be listed. The mass flow rates of the four entering
F=2-2+2=2 and leaving streams (L n ^, l/„ + i, L n Vn and the , )

rate of heat addition (q n add five independent )

so that only two variables may be specified. If the extensive variables, to yield a total of 3C+ 7 inde-
pressure is fixed, setting the temperature immedi- pendent variables. How many of these independent
ately defines the system and sets values for the variables can be arbitrarily specified? Certain re-
relativeenthalpy and compositions of both phases, strictions apply to the system, and these restric-
as indicated by Figure 3.4. In general, in the two- tions reduce the number of variables that can be set
phase mixtures encountered in staged operations independently. A mass balance around the stage
yields C component mass balances plus one total
F = C (8.4) mass balance. But the total mass balance is equal to
the sum of all the component balances, so that
Strictly speaking,
enthalpy is not an there are only C independent mass balances. The C
independent variable, because there is no physical independent mass balances place C restrictions on
way to set the value of enthalpy. Normally, the the arbitrary specification of the 3C + 7 indepen-
pressure, temperature, and composition may be set dent variables. That is, it is not possible to set all
physically and this fixes the enthalpy, but it is entering and leaving compositions without taking
sometimes convenient to consider enthalpy as an into account the mass balance that must be satis-
independent variable. fied for each component.
 MULTICOMPONENT STAGE OPERATIONS 125

One enthalpy balance may be written around Example 8.1. Determine whether sufficient
the aiayc.
ic stage: independent variables have been fixed to define
each of the following systems completely.
L n _- h n _- +
i i
Vn + iHn+: +q n = L n hn - Vn Hn
(8.5) (a) A stage in a distillation column
and this adds one more restriction. The number of For
independent variables that must be fixed to deter- Z.„:x a =0.6, x6 =0.1, xc = 0.30
mine completely the operation of the equilibrium V„ + 1 :
Va= 0.75, yb = 0.05, yc = 0.20
stage is equal to the total number of independent /.„_ 1 : x a = 0.50, x b =0.15, xc = 0.35
variables minus the restrictions placed upon the Vn :
y a =0.68, k 6 =0.10, yc = 0.22
stage by mass and energy balances, as summarized (b) A stage in a liquid extraction cascade
in Table 8.1.
Analysis of an equilibrium stage using the For
phase rule together with mass and energy balances = 0.6, xb = 0.4;
has shown that 2C+6 variables must be set to ,_,= 1000lb/hr
define completely the operation of the stage. The Vn + ^: yc = 1.0; V„ + 1
= 2000 lb
choice of these 2C+ 6 variables from among those
available is arbitrary, but certain variables are
(c) A distillation stage for ethanol and water
frequently specified. Reference to Figure 8.1 Vapor entering:
shows that after the pressure is fixed, the compo- 100 lb moles: 25% ethanol, 75% water
sitions, flow rates, and temperatures (or enthalpies) Vapor leaving:
of the four streams may be specified, along with 100 lb moles: 30% ethanol, 70% water
the rate of heat addition. If any component does Liquid entering:
not appear in a given stream, its fraction is fixed at 80 lb moles: 20% ethanol, 80% water
zero. Frequently the complete compositions, en-
Temperature = 90°C for all streams
thalpies, and flow rates of the two entering streams
Pressure = 1 atm for all streams
are specified, together with the pressure P and the
heat addition q N In this case, the specified vari-
. Solution, For a three-component system,
(a)
ables are 2(C+ 2) + 1 + 1 = 2C+ 6, so that the 2C + 6= 12 independent variables must be
12, so
system is exactly specified. It is not possible to fix specified to define the system. Of the twelve mass
arbitrarily another variable (such as the stage tem- fractions given, only eight are independent (two
perature or compositions of the leaving streams). If for each stream), so four more variables must be
another variable were fixed, the specified variables set. Typically, those set would be the temperature

would contain a contradiction. and pressure in the stage (which fix the equilibrium
relationships), plus the flow rates of Z-„-i and

Table 8.1 VARIABLES AND RESTRICTIONS IN AN This system contains three components,
(b)

EQUILIBRIUM STAGE but all them do not appear in both entering

of
streams. Nevertheless, the composition of each
Independent variables stream is considered to have C — 1 = 2 independent
1 Intensive variables associated with the fluid variables. For this three-component system, twelve
streams 3C + 2 variables must be specified. As stated, four compo-
2 Extensive variables associated with mass sitions and two flow rates give a total of only six
flow to and from the stage 4 variables. In addition, pressures and temperatures
And associated with energy flow to or from of the four streams may be set to give six ad-
the stage 1 ditional variables. Note that the pressures and tem-
peratures of the two streams leaving must be equal,
Restrictions
since they are in equilibrium.
1 Mass balances -C Here twelve variables are fixed and
(c)
2 Energy (enthalpy) balance -1
2(2) +6= 10 variables must be specified. There-
fore, the system as defined cannot exist at equili-
= Independent variables that must be fixed
brium. Specification of two variables (such as the
to define the system completely
flow rate and composition of Vn + must be -
[
)

3C+2+4+ 1 C- 1 2C + 6 dropped.
126 STAGE OPERATIONS

Variables in a Cascade of Equilibrium Stages Table 8.2 VARIABLES AND RESTRICTIONS IN-
SIMPLE COUNTERCURRENT CASCADE
A simple countercurrent cascade of equilibrium
stages without reflux (Figure 8.2) is used in many
Independent variables
cases in liquid extraction, leaching, and gas absorp-
tion. In distillation, the enriching section and the
1 Variables to be fixed for N stages
(2C + 6 for each stage) /V(2C + 6)
stripping section can be considered as separate
2 Total number of stages (a single variable) 1
countercurrent cascades, joined at the feed stage.
Each equilibrium stage in the cascade shown — Restrictions
in Figure 8.2 has associated with it 2C + 6 indepen- 1 Additional restriction because of flow
dent variables that must be specified, as shown in between stages — 2(/V 1MC+2)
the previous section. Therefore, the N stages in the [(C+2) restrictions for each of

cascade include A/(2C+6) independent variables. 2(/V — 1 ) streams flowing between

In addition, the number of stages N is an indepen- stages]

dent variable. The total number of independent

variables is then N(2C + 6) + 1 = Independent variables that must be fixed

In addition to the restrictions on each stage to define the system completely

given in the previous section, several new restric-
/V(2C + 6) +1 -2(/V- 1KC + 2) 2C + 2/V + 5
tions arise when stages are connected in series. The
concentrations of the two streams leaving a given
stage are the same as those of two of the streams
entering the adjacent stages. For example, L-\ has
specified, an additional 2(C + 2) variables are fixed.
Subtracting these two groups of specified variables
the same composition leaving stage 1 as it has
from the total that must be specified (Table 8.2),
entering stage 2. Each stream flowing between
stages has C + 2 independent variables associated 2C + 2/V + 5-2A/-2(C+2) = 1

with it (as shown in the previous section). There

yields one remaining variable to be specified. This
are 2(/V — 1) streams flowing between N stages, so
would typically be either the number of stages, the
that 2(/V — 1)(C+ 2) restrictions are imposed on the
recovery of desired product, or the concentration
cascade. These restrictions, when subtracted from
of the desired productin L N or Vy
the number of independent variables, effectively
correct for initially counting the independent vari-
One might be tempted to include the overall
mass and energy balances around the entire cascade
ables of a stream flowing between stages twice:
But a total of N mass and
as additional restrictions.
once when it flows from a stage and again as it
flows into the next stage. Table 8.2 summarizes the
N enthalpy balances resulting from balances
previous discussion and shows the evaluation of the
around each stage have already been included in
number of independent variables that must be the restrictions, and the overall balances are not
fixed to define the system.
independent because each is equal to the sum of
The choice of the 2C + 2/V + 5 variables to be
the N balances for the individual stages.

fixed is arbitrary, but usually certain variables are

fixed by equipment-design considerations. For
example, the pressure and heat addition for each
stage are often fixed, yielding two fixed variables
per stage for a total of 2/V for the cascade. If the
temperature (or enthalpy), pressure, compositions,
and flow rates of the two feed streams are also
8.2. An absorption column equiv-
Example
alent to 10 equilibrium stages is to be used to
separate a mixture containing 20 mole percent
propane, using a nonvolatile hydrocarbon oil as an
absorbent; the concentration of methane in the gas
leaving is to be 80 mole percent. The temperature

vi v v+ .

Hi or Ti
Pi — Ti Tn Tv *— P„ +I

U — 1

Pi tit it Pn
N-\
P\--.
N
Ps — L.v

*o0rTo Av orrv
P° Qi 92 9n <?.v-i 9.v PM
Figure 8.2. Variables in a countercurrent cascade of equilibrium stages.
 MULTICOMPONENT STAGE OPERATIONS 127

be 60° F, and L and VN +

at each stage will are at y

60° F. Are enough data given to define the system?

-$,L \Vn =
h D or TD
Solution. The figure below shows the data Pd
given, with methane designated a; ethane, b; pro-
pane, c; and the hydrocarbon oil, d. The total
Enriching section
number of independent variables specified
21 (8 is N- stages
compositions, 12 temperatures, and number of
stages). To fix the system, 2C + 2/V + 5 = 2(4)
+ 2(10) + 5 = 33 variables must be specified, so that
the system not defined with the data given. It is
is 'N "V,

not possible to calculate the remaining quantities Feed stage

hF or Tv
with the data supplied. Twelve more variables must JF tvm=1
be fixed. For example, the pressure at each stage
and at Z. ar| d VN + would be enough to define
-\

the system completely and to permit evaluation of

Stripping section
unspecified flow rates, compositions, and heat
M- stages
transferred at each stage — providing, of course,
that phase-equilibrium and enthalpy data are avail-
able for the system containing methane-ethane-
propane-hydrocarbon oil at the specified tempera-
ture and pressure.
hR oxTR
a
r
s
V: \//V + 1
Figure 8.3. Variables in a distillation column
with a total condenser and partial reboiler.
ya = 0.8 Ya = 0.2

Yb = ? Vb = 0.3
Vc= ? Vc = 0.5 independent variables to be fixed, as shown in the
previous section. Similarly, the stripping section
ya = o Yd =
* has 2M + 2C + 5 independent variables to be fixed.
10
stages * The feed stage must be analyzed to determine
Lo LN the number of independent variables to be speci-
fied. It differs from an equilibrium stage because it
xa =0 xa = ? has one additional stream flowing to it. The feed
xb = xb = ? stage may be thought of as consisting of a mixer

= (where F and L F _- are mixed) and an equilibrium

xc = ? [

stage, as shown in Figure 8.4. Analysis of the mixer

xd = 1.0 xd = ?
shows that there are three single-phase streams
flowing to and from it, for a total of 3(C+2)
Variables in a Countercurrent Cascade with Reflux variables, and there may be heat transferred, for

The addition of reflux at both ends and feed at the one additional variable. The restrictions include C
middle of a cascade adds both independent vari- mass balances and one enthalpy balance, so the
ables and restrictions to the definition of the sys- number of independent variables that must be
tem. The system is that shown in Figure 6.2. The fixed is

following analysis would typically apply to distil- 3(C+2) + 1 -(C+ 1) = 2C + 6

lation, and this analysis will be written in those The equilibrium stage also has 2C+6 specifiable
terms, based on the column shown in Figure 8.3. independent variables. There is one stream flowing
The column can be broken down for purposes of from the mixer to the stage, so that (C + 2) vari-
analysis into a countercurrent cascade of N stages ables must be subtracted (since they really have
in the enriching section, one of M stages in the
been counted twice, as discussed earlier). The net
stripping section, a feed stage, a total condenser, number of independent variables that must be
and a partial reboiler.
specified to define the feed stage is then
The enriching section is a simple counter-
current cascade and therefore has 2/V + 2C + 5 (2C + 6) + (2C + 6) - (C + 2) = 3C + 10
128 STAGE OPERATIONS

Table 8.3 VARIABLES AND RESTRICTIONS FOR A

DISTILLATION COLUMN

Independent variables:
1 Enriching section 2/V + 2C+5
2 Stripping section 2M + 2C+5
3 Feed stage 3C+10
< 1*
4 Condenser with reflux C+ 4
5 Partial reboiler C+ 4

— Restrictions (additional):
1 Eight intermediate streams -8 (C+2)
Lf Vf+i

Figure 8.4. Variables at the feed stage. Independent variables that must be specified
to define the system completely = 2(/W + /V) + C+ 12

The total condenser has one stream flowing in

and one flowing out, for a total of 2(C+2) vari-
ables. There is also a heat-addition stream q c ,
The resulting number of specifiable independent
which in this case has a negative numerical value.
variables is
The restrictions include C mass balances, one enth-
alpy balance, plus one additional restriction be- (C+2) + (C+2) + 1 -(C+l) = C+4
cause the net heat removed not only consists of a
The results of these analyses of the various
heat leak through the condenser insulation but also
units of a distillation column are given in Table
includes the latent heat of vaporization of the
8.3. It is now necessary to combine the specifiable
vapor being condensed. The number of specifiable
variables forall parts of the column and to subtract
independent variables is then
additional restrictions because of the streams flow-
2(C + 2) + 1 -(C + 2) = C+ 3 ing between the various units. There are eight such
streams (L Q V„ = L N VF L Fl Vm =
,
1 , L M and
, , y , ,

Immediately after the condenser is the reflux-

Vs ), considering the condenser-splitter as one unit.
distillate splitter, where one stream enters and two
Hence, there are 8(C+2) additional restrictions.
leave, for a total of 3(C + 2) variables. There is no
The number of specifiable independent vari-
total
heat exchange. The restrictions are C mass balances
ables then 2(M + N) + C + 1 2, as shown in Table
is
and one enthalpy balance. Additional restrictions 8.3. The choice of variables to specify is governed
arise because the compositions, pressures, and
by equipment design and process information, as
temperatures of the two product streams are ident- shown in the next example.
ical. This adds C + 1 restrictions. One final restric-

tion arises because the pressure of the system equals

Example 8.3. One hundred pound moles per
the feed pressure. The number of specifiable in- hour of a liquid mixture containing 30 mole per-
dependent variables is then cent benzene, 50 mole percent toluene, and 20
3(C + 2)-C-1-(C+1)-1=C + 3 mole percent o-xylene at 1 atm and 70° F is to be
separated in an existing insulated distillation col-
The condenser and splitter may be com-
total
umn equivalent to 20 equilibrium stages. The col-
bined toa "total condenser with reflux."
give
umn is equipped with a total condenser and a
Because there is one stream flowing between the
partial reboiler. It is desired to produce a distillate
condenser and splitter, the number of specifiable
with a composition of 95 mole percent benzene.
independent variables is
The entire column operates at 1 atm pressure.
What variables may be adjusted to achieve this
(C+3) + (C+3) -(C + 2) = C+4 distillate composition?
the partial reboiler, one single-phase stream
In
enters with C+2
variables, and the two streams Solution. Before selecting variables to ad-
leaving contribute a total of C+2 variables, be- just, it is necessary to determine how many in-
cause they are in equilibrium. An additional vari- dependent variables must be specified to define the
able arises with the heat added (q ). The restric-
s system. From the data given, + N = 20 — 1 = 19, M
tions are C mass balances and one enthalpy balance. and C = 3, so from Table 8.3, 2(19) + 3+12 = 53
 MULTICOMPONENT STAGE OPERATIONS 129

variables which must be fixed. A number of vari- uble in the liquid. Both absorption and stripping
ables have been fixed: have been widely applied in separating hydro-
carbon mixtures in the oil industry. The same
1. column fixes 21 q's at (q at 20
Insulated
theoretical concepts underlie each.
stages, plus"mixer" at feed stage — Figure 8.4). The methods
calculation for multicomponent
2. Pressure of 20 stages, plus feed mixer, re-
absorption are much the same as those covered in
boiler, and condenser is 1 atm, so 23 variables are
Chapters 5 and 7. A rigorous method would in-
fixed.
volve stage -by -stage mass and enthalpy balances
3. Flow rate, composition, temperature, and and application of phase equilibria to determine
pressure of the single-phase feed fixes five vari-
the number of stages required for a given separ-
ables.
ation.Such a calculation would require complete
4. The number of stages (/V=20) sets one
data on the enthalpy and phase equilibria of the
variable. A total of fifty variables has been fixed, multicomponent mixture. These data are often not
leaving three still to be specified.
available. Fortunately, in many systems, simplify-
Usually a desired percentage recovery of ben-
ing assumptions can be made. If heats of solution

zene in the distillate, together with the mole frac-

and other heat effects are negligible, the temper-
ature through the absorption column is constant
tion benzene in the distillate, is chosen. These two
and so are the equilibrium relationships. The pres-
variables effectively fix D, and also B and x Bb
sure is also approximately constant. As a result,
because of the overall mass balance. The feed plate
methods similar to those of Chapter 7 may be
is usually located so that the liquid on the plate has
employed.
approximately the same composition as the feed,
If the heats of solution are not negligible,
and this fixes an additional variable. Enough vari-
cooling coils are occasionally installed at each stage
ables have now been specified, and other variables
to remove the excess heat and to maintain the
(such as condenser duty, reflux ratio, etc.) may be
calculated.
column at a constant temperature; it is thus still

possible to use simplified calculation methods. In

Experiencein the design of multicomponent

columns shows that there are many
distillation
more limitations on specifying variables than have
been discussed in this section. This section has
considered only limitations imposed by phase equi-
librium, mass balances, and energy balances. In the
cases where the temperature change from stage to
stage is significant, more rigorous mass and enth-
alpy balances must be employed.
Table 8.2 shows that 2C+2A/ + 5 independ-
ent variables must be set to define an absorption
system. Assuming constant pressure and either zero

design of equipment, practical considerations such

heat removal (for the case where heat effects may
as temperature and economic availability of cool-
be assumed negligible) or a specific constant value
for heat removal at each stage (when heat effects
ing water and steam, capacity of reflux pumps, and
sizes of existing columns further limit the choice of are significant and the temperature is to be held

values for variables. constant) fixes two independent variables per stage
for a total of 2/V. In addition, the compositions,
pressures, temperatures, and flow rates of the
MULTICOMPONENT ABSORPTION entering gas and the entering absorbent fix 2C + 4
An example of a multicomponent stage operation additional variables. Only one independent variable
with reflux, gas absorption, is considered in this remains. Usually the desired recovery is specified
section. As discussed in earlier chapters, gas absorp- for the design of a new column, so that the re-
tion involves the transfer of components of a gas quired number of stages may be calculated. Alter-
phase to a liquid phase. Usually none of the liquid natively, the number of stages in an existing
absorbent vaporizes, so (unlike distillation) absorp- column is specified and the possible recovery is

tion involves interphase transfer in one direction determined.

only. In some however, the liquid absorbent
cases, It is of interest to note that if the complete

is slightly and this must be taken into

volatile, compositions of the two entering streams are speci-
account. Gas desorption or stripping is the oppo- fied, the number of remaining specifiable inde-

site of absorption. In stripping, components of the pendent variables (2N + 7) does not depend on the
liquid phase are transferred to a gas phase. They number of components in the system.
are "stripped" from the liquid. In stripping, the Both graphical and analytical methods ot
major component of the gas phase is often insol- calculation are considered below. The graphical
130 STAGE OPERATIONS

methods are presented first to demonstrate the (x«>7jr+i)'

principles; then the analytical methods, which are

6
particularly useful for digital-computer calculations,
will be illustrated.
Slope = L/V- 5 s'
Graphical Methods (12)

The procedures for multicomponent absorption are

4
s£ Equilibrium
curve
identical to those presented for binary mixtures at
the beginning of Chapter 7. Instead of having a (*o>>i>-
single equilibrium curve and operating line, there
are now an equilibrium curve and an operating line
*>
for each absorbed component of the gas. For a (a)
dilute gas and a dilute liquid, the flow rates of L
and V will be approximately constant through the
column, and calculations may be based on a mole
fraction plot utilizing Equations 3.6 and 7.3a. A
"dilute" gas is defined here as one consisting
largely of a nonabsorbed component. A dilute
liquid contains primarily the nonvolatile absorbent.
For concentrated gas or liquid phases, the
mole ratio diagram and Equation 1.2b must be
used, together with the appropriate mole-ratio (*o-yi)

equilibrium relationships. This device yields

straight operating lines, but the equilibrium curve
must now be located by trial and error.

Dilute Gas and Liquid

At moderate pressures vapor-liquid equilibria of

many hydrocarbons may be expressed by Raoult's
Figure 8.5. Multicomponent absorption with
law (Equation 3.5). At higher pressures, even
dilute gases and liquids, (a) Specified com-
hydrocarbons deviate from the ideality required by ponent b. (b) Nonspecified component a.
Raoult's law, and the empirical relationship of
Equation 3.6 must be used together with experi-
mental values of K, such as those in Appendix D-2.
For purposes of illustration, a four-com-
Because L, V, and /a/ + i ,6 areknown, specification
ponent gas with three absorbed constituents
of Rb also specifies y 16 . The operating line for
(a,b,c) and nonabsorbed component (g) will be
component b may now be plotted, along with its
contacted with a nonvolatile liquid absorbent (/) in
equilibrium curve (Figure 8.5). The equilibrium
a countercurrent cascade (Figure 8.2). It is as-
curve is a straight line if Raoult's law or K values
sumed that heat effects are negligible and that the
are used. The operating line is located with the
flow rates and compositions of the entering
point (x ,/i) and the slope L/V. The operating
streams are known. Thus, all but one of the inde-
line runs from (x ,y : to (x /v K/v + ). Because
) ,
pendent variables have been set. Two cases will be
Z/v + 1,6 is known, the terminal point of the
considered: specified recovery here, and specified
operating line can be located. Now the number of
stages later.
stages required for the specified recovery can be
Since only one remaining independent vari-
determined by stepping-off stages from either end,
able may be fixed, the recovery of only one com-
as shown; six stages are required.
ponent can be specified. The recovery of the others
Recovery of the components a and c may be
can then be calculated. In this case, the recovery of
determined by the following procedure. Exactly
b will be specified. The fractional recovery of any
the same number of stages are available for the
component in absorption is
recovery of a and c as were required for the speci-

- The operating lines for a and c

fied recovery of b.
=
VyN + : Vy\ k/v+i -y\
/?
(8.6) must have the same slope {L/V) as that for b.
Vy N + 1 K/V + 1 Furthermore, x and //v + are known for a and c.i
 MULTICOMPONENT STAGE OPERATIONS 131

This is enough information to locate the two oper- 0.18

ating For example, for a, the equilibrium

lines.
curve is plotted (Figure 8.5) and lines at x 0a and
K/v + 1 ,a are inserted. Now various lines of slope L/V
are drawn until one is found that requires exactly
the same number of stages as did the recovery of b
(Figure 8.5). This line then fixes y t and x N so that ,

recovery of a may be determined from the appro-

priate form of Equation 8.6. A similar procedure is
y a ,yb
used for component c. Figures 8.5a and 8.56 have
been separated for clarity. Normally, all equi-
librium curves and operating lines are plotted on
one figure, as shown in the following example.
As in Chapter 7, a minimum L/V ratio requir-
determined for
ing infinite stages can be a specified
recovery of the other components.

Example 8.4. Methane and ethane are to be

removed from a gas containing 80 percent H 2 15 ,

percent CH 4 and 5 percent C 2 H 6 (by volume) by

,

absorption into a heavy oil in a column operating

at 215 psia and 80° F. The molar ratio of entering
absorbent to entering gas is 10 to 1. The entering
absorbent contains no methane or ethane. Deter-
(x<,b,yib)'
yQ 0.005 0.010 0.015 0.020 0.025 0.030
Xa ,x b
mine the number of equilibrium stages required for
an 80 percent recovery of methane. What is the Figure 8.6. Solution to Example 8.4.

recovery of ethane?
The ethane operating line is parallel to that
Solution. A containing
gas 20 percent
for methane, and located so that exactly 6.5 stages
absorbable components may not
be "dilute"
fit between y N + b = 0.05 and x 0b = 0. In order to
y
enough to use a mole fraction plot. This will be
allow for 6.5 stages, the operating line approaches
checked later. Data given are Z./V = 10.0,
very close to the origin, so that y yb 0. Stepping -
K/v + i,a=0.15, y/v + 1.6 =0.05, y N + g =0.80, ^
off of stages shows barely three discernible stages
x i - 1 .0, where a = methane, b = ethane,
because a pinch is approached as y-\ b -+0. Within
g = hydrogen, 1 = hydrocarbon oil. Hydrogen is not graphical accuracy, = and
y^ b 0, recovery of
appreciably absorbed at these conditions, Equi-
ethane is 100 percent. This is expected, since
librium data are obtained from Appendix D-2. At
ethane is so much less volatile than methane.
215 psiaand80°F,/Ca = 11.5, Kb = 2.7. These two
To determine whether L is constant, Lm is
curves are plotted (Figure 8.6). The range of x a
calculated. For 100 lb mole of L Q VN+: = 10, and ,

and ya plotted on Figure 8.6 is determined by the

methane in L N is (0.8) (0.15) (10) = 1.2; ethane is
extreme values (x N y/v + ). In this case only y N +
1
0.5, so that L N = 100+ 1.2 + 0.5= 101.7. Thus,
, ^

is known, so the equilibrium curve is plotted out to

the liquid downflow increases by 1.7 percent from
this point.
the top to the bottom of the column. This is
From Equation 8.6 for methane
sufficiently small to assume a constant L. V varies
from V/yv + 1 =10 to ^=10-1.2-0.5 = 8.3.
0.15-y 1a percent variation may be too great to
80= 100 This 17
0.15 justify the assumption of constant V. This will be
checked in Example 8.5.
Therefore, y 1a =0.030.
The operating line for methane can now be Stripping can be analyzed by the same
plotted from (x y,) = (0.0, 0.030) ,
and techniques as absorption. The operating line is
L/V = 10.0. The operating line ends where it inter- defined in the same manner. But //v + i, a is now
sects //v + i,a = 0.15. About 6.5 stages are required less than y 1a and x 0a is less than x Na and the
, ,

for the specified methane recovery. operating line lies below the equilibrium curve, as
132 STAGE OPERATIONS

where / is any absorbed component, must be

>a — Ka x a
converted to these new mole ratio units. For 1 lb
mole of absorbent, there are IX, lb moles of other
components dissolved in it, for a total of (1 + IX,)
lb moles. Thus, the mole fraction at any point in

the column is
X,
*/
= (8.8)
1 + IX,

For l//v + 1
lb moles of total gas entering, there
are IV, total lb moles of gas at any point in the
column. Thus,

VI = ^r (8.9)

Combination of the last three equations gives

Figure 8.7.
gases and liquids.
Multicomponenf stripping with dilute
J5—*&-
+
(8.10)
IV, 1 IX,
Equation 8.10 can be used to evaluate the equilib-
shown in Figure 8.7. Component b is the key rium curve on mole-ratio coordinates. However,
component, and ten stages are required. Because IX, and IV, depend upon phase flow rates as well
component a is much more volatile (as indicated as upon the basic equilibrium relationship. As a

by a large K), the recovery of component a is
limited by the approach to equilibrium between
y 1a and x 0a On the other hand, y 1c =0, because
.
component c is much less volatile than component
b.
result, Yjcannot be evaluated directly from X,, but
it must be determined by trial and error.
As a first step, an approximate equilibrium
curve is located using the slope at the origin and
one point of high concentration taken at the
bottom of the column (X N Y N ). At the origin ,

Concentrated Gases or Liquids (12) (X, = 0, Yj = 0), IX, = 0, and I V, = 0, so that

Equation 8.10 becomes
If either the gas phase or phase contains
liquid
appreciable quantities of absorbable components, Y,=KX, (8.11)
L and/or V are not constant, and use of Equation
7.3a leads to curved operating lines. Therefore, This shows that the initial slope of the equilibrium
mole ratio units are used with Equation 7.3b, but curve is K (Figure 8.8). To determine the point
this leads to a curved equilibrium curve that is a (X/v,, V/v,), it is necessary to estimate V 0/ If the .

function of the composition of both phases. The recovery is given for one component, approximate
graphical solution of such problems is trial and
error. The symbols in Equation 7.3b are redefined:

V'Y, -L'X
»»+i ~ .., xn [7.3b)
V
where now
L' = pound-moles of nonvolatile liquid
absorbent entering
V - pound-moles of total gas entering = VN + :

Y = pound-moles of any component per

pound-mole of V
X= pound-moles of any component (except
the absorbent) per pound-mole of L'

The equilibrium relationship Figure 8.8. Approximate determination of

the equilibrium curve on mole-ratio coordi-
Yi = KjXj (8.7) nates.
 MULTICOMPONENT STAGE OPERATIONS 133

Y 0i may be estimated for all components. If no

recovery is given, the assumption of absorption of
half of all absorbable components is a good first = 0.01327

trial. An experienced engineer can make fairly

good estimates. Once Y Ni is estimated, the point
Xn» ^Ni s plotted, and a smooth curve is drawn
'

between the slope at the origin and the point

(Figure 8.8). The process is repeated for each
absorbed component. Next, the operating lines are
located, and the number of stages or recovery is
determined. The values of Y0i assumed to plot the
equilibrium curves are checked, and new curves
drawn if necessary. For greater accuracy, several
points on the equilibrium curve for the specified
component may be determined using the results of
the first trial, and the stage calculation may be
repeated.

Example 8.5. Repeat the solution of Example

8.4 using mole ratio coordinates.
0.005 0.010 0.015
(X ob Ylb )
,

X
The slope of the equilibrium
Solution.
Figure 8.9. Solution to Example 8.5.
curve methane is 11.5 at the origin (by
for
Equation 8.11). The recovery of methane is 80
percent, and the recovery of ethane is estimated at
For ethane, 3.4 stages give nearly complete
100 percent, since it is much less volatile than
recovery:
methane. As a result, I,XNl at the bottom of the
column is -0.0005
„
/? - mM°— O05
10
0.05
"" %
2Xyv/ =
(0.80)(0.15)l/' + 0.05V"
= 0.01 7
_.,
TT7T,
10V Analytical Methods

Also, V/v+La =0.15, and V/v + i

=1.0. So by Wherever they can be applied, analytical methods
Equation 8.10, are preferable to graphical methods, because they
are more rapid and more accurate. Analytical

1 + SX, (0.15X1.017)
methods may involve a single equation relating the
Xa = number of stages to system parameters, or they
2 V, K (1.0)(11.5)
may involve a stage-to-stage calculation on a digital
= 0.01327 computer using methods paralleling the graphical
procedures of the preceding section.
The equilibrium curve is plotted with the
For dilute gases and liquids, the graphical
initial slope 11.5 through the point (0.01327,
method for the key components described in the
0.15) (see Figure 8.9). Similarly, the ethane curve
previous section is equivalent to the equation
is located by the initial slope 2.7 and the point

=0.05, XNb = [(0.0b)

(//v + i, 6 (1.017)]/[(1.0)
\iA Vn + — Kxq
A^
1
=0.0189). Both equilibrium curves are very
- Kxo ^}
(2.7)] log
nearly straight lines, so that it will not be necessary V\
N= (7.47)
to reevaluate them later. Because L/V= 10/1, \ogA
L'/V'= 10/(0.8) (1) = 12.5. With this slope and
the point [X 0a = 0, V 1a = (0.2) (0.15) = 0.03] which was derived in Chapter 7. The absorption
only 3.4 stages are required, compared to the 6.5 factorA = L/KV must be constant for Equation
calculated in Example 8.4. The difference lies in 7.47 to hold, so that it is limited to dilute gases
the error involved in assuming constant L and V in and liquids where L, V, and K are essentially con-
the earlier example. This leads to an incorrect value stant. Equation 7.47 is useful in calculating the
of V1a and an incorrect slope for the operating line. number of stages required for a given recovery.
134 STAGE OPERATIONS

If the number of stages is specified, y-\ and liquid temperature.

hence the recovery may be determined from

lv+j:xn )

(8.14)
(7.46)
KV'ZYN + ,

Vn + — Kx i/V+1 - 1
1
L' = Z. ,andZ.'(1 + 1XN )
= LN

Equation 7.46 may also be used to calculate the L V= l/ w+1 and V/v + 1
=1.0
required for a given recovery and number of stages.
Similarly
The equivalent expressions for stripping are
+ KV'Yy
x o ~ *N '/V 1
S' = (8.15)
S n + '-K
(8.12) Z-'d +X )

xq-Yn + iIK
Equations 7.46 and 7.47 may be converted to mole
where S = MA
= KV/L, so that Equation 8.12 is
ratio units by appropriate substitution of A' for
identical to Equation 7.48. The number of strip-
A, Vt for /t etc.; Equations 8.12 and 8.13 may
,

ping stages is given by

be similarly rewritten using S' X, etc. The resulting ,

equations are only approximate.

S- 1\/ x -y/y + i/K x 1
log

N=-
Fritx N -yN + y
/K S
(8.13)
logS Example 8.7. Solve Example 8.5 analytically.

Solution. From Example 8.5, assuming

Example 8.6. Solve Example 8.4 analytically. 100 percent recovery of ethane,
Solution. To determine the number of XN =0.017, L'/V^ll.b, ATa = 11.5, Kb = 2.7
stages required for 80 percent recovery of
By Equation 8.4, with £ V/v + = 1-0,
methane, Equation 7.47 is used. For methane: 1

=0.15, /, =0.030, x =0, A = 10/11.5 =

//v + 1

0.87, so that
4' = —12—5 (1 +0.017) = 1.105
11.5
0.87- 1 \, 0.15-0 1
With Equation 7.47 modified for mole ratios:
log
0.87 'V0.030 - ' 0.87
N=- = 6.56 For a:
log 0.87
1 - 1.105 w0.15-0\ 1
log +
The recovery determined from
of ethane is
'
1.105 '' 0.03-0 '
l7l 05
Equation 7.46. For ethane: //v + =0.05, x = 0, N= = 3.2
i
log 1.105
Kb =2.7, Z./\/=10 sothat>A = 10/2.7 = 3.7
/

For 6 by Equation 7.46 modified:

0.05-/! (3.7)
756 -3.7
12.5
7.56
= 1.0 A' «-r=- (1.017) = 4.66
0.05 (3.7) -1 2.7
Solving gives 4 2
0.05 (4.66) 4.66
= 0.993
/! =0.0 0.05 (4.66)
42 -1
and Yyb = 0.00035
R b = 1 00 percent and the recovery = 99.3 percent. These results

As expected, these results agree with Example 8.4,

agree with Example 8.5.

but they are in error because they assume constant

L and V (see Example 8.5). MULTICOMPONENT DISTILLATION
Multicomponent distillation is based upon the
The analytical equations may be modified to same principles of phase equilibrium as binary
account reasonably well for variable L and V by distillation. analysis of a multicom-
The design or
defining absorption and stripping factors for the ponent tower utilizes equilibrium rela-
distillation
concentrated end, and by using a K at the average tionships and mass and enthalpy balances. The
 MULTICOMPONENT STAGE OPERATIONS 135

assumption of constant molar overflow is ponent depends not only upon its concentration in
frequently made, so that enthalpy balances may be the liquid and the pressure, but also upon the
neglected. This assumption is sufficiently correct in concentration of all other components in the
a large number of petroleum distillation applica- liquid. As a result, it is necessary to have data on
tions, but in some other situations it is unwar- the equilibrium relationship for one component as
ranted. A crude-petroleum distillation system is it is affected by the variation in composition of all

shown in Figure 8.10. of the other components. Because this would

In binary distillation, specification of the require a tremendous quantity of data, simplifying
pressure and the concentration of one component assumptions are made when possible. When the
in one of the phases was sufficient to define the components have a similar chemical structure (such
phase equilibrium, because as the hydrocarbons in petroleum), it is possible to
assume a Raoult's or Henry's law relationship. This
F=C+ 2 -P = 2 + 2 - 2 = 2
assumption relates the vapor composition of a
The situation is more complicated in multicom- given component to its liquid concentration by a
ponent distillation, where F = C, so that C variables constant that does not depend upon the variety or
must be fixed to define the equilibrium system. quantity of other components present. In many
The equilibrium vapor concentration of one com- systems the assumption of Raoult's or Henry's law

Figure 8.10. Crude oil distillation battery. The multicomponent mixture of crude
oil is separated into several product streams as a first step in refining petroleum.
The furnace shown on the left partially vaporizes the crude oil and feeds it to the
tallest column, which operates at atmospheric pressure. The short, wide column to
its left operates under a vacuum, further processing the low-volatility bottoms of

the atmospheric column. The column to the right of the atmospheric column
removes the most volatile components from the gasoline, which come off the top
of the atmospheric column. The low superstructure in the right foreground sup-
ports heat exchangers, air-cooled condensers, and other auxiliary equipment.
(Foster Wheeler Energy Co.)
136 STAGE OPERATIONS

is not valid (see Chapter 3). For example, the the number of stages is fixed. Further specification

entire object of azeotropic and extractive distil- of the reflux ratio, the feed-stage location, and the
lation is to add a component that will influence the concentration of a single component at one end
vapor-liquid equilibria among ether components. defines the system, but the calculation procedure is

still and error. The compositions cannot be

trial

Independent Variables determined until the stage temperature is known,

but the stage temperature depends upon the com-
Of the 2{M + N) + C + 1 2 variables that must be set
plete compositions of the phases in the stage. In a
to define a distillation column, usually the follow-
ing are fixed:
method first proposed by Thiele and Geddes, a
temperature profile is assumed (10). Distillate and
Feed stage (pressure and bottoms compositions are estimated by approxi-
heat addition at both mate methods, and stage-to-stage calculations are
the "mixer" and the made from each end to the feed stage. If the feed
equilibrium stage) 4 stage compositions as calculated from each end do
Stages (pressure and heat not check, the end compositions are adjusted and
addition on each the calculation is repeated. Finally, when the feed

stage) 2(M + N) compositions check, the assumed temperature

Condenser (pressure) 1
profile must be verified and further adjustments
Reboiler (pressure) 1
made. This multiple trial and error was system-
Feed (composition, atized in a method developed by Hummel (6).
pressure, temperature) C+ 2 The methods discussed above have many
Total 2(M + N) +C + 8 modifications and special procedures that prove to
be useful for specific applications. No attempt will
There remain four variables to fix, regardless of the be made to cover these methods in this elementary
number of components present. In binary distil- introduction. Instead, the principles of stage-to-
lation, specification composition and
of the stage calculations will be applied to a few simple
recovery of one component at one end of the cases of multicomponent distillation. The reader
column, the reflux ratio, and the location of the interested in a more detailed discussion is referred
feed stage defines the system completely. In multi- to References 3, 4, 5, and 13 and to the vast

component distillation, the same specifications also literature on multicomponent distillation.

define the system, but the complete compositions
at each end of the column are not known. In order
to make stage-to-stage calculations to determine
the required number of stages for the specified
recovery, it is necessary to assume the complete
compositions of the distillate and bottoms and
then calculate stage-to-stage toward the feed stage.
If the compositions at the feed stage determined
from calculations from each end do not agree, then
new compositions for the bottoms and distillate
must be assumed and the calculation repeated. This
method was first suggested by Lewis and Matheson
in 1932 (8) and has been modified by many others
since (4, 5, 6, 9, 10, 13). This long, tedious, trial-
and-error approach encouraged the development of
many shortcut and approximate methods; but with
the development of high-speed digital computers,
the time and tedium have been removed from the
more rigorous trial-and-error procedure so that
now widely applied, with modifications to
Key Components
In column with a single feed, distillate, and
a
bottoms, a multicomponent mixture is split into
two fractions. The bottoms contains the "heavier,"
less-volatile constituents of the feed; the distillate
contains the "lighter," more volatile components.
The components of intermediate volatility appear
in distillate and bottoms to a significant
both the
extent. Usually process or end-use requirements
lead to the specification of a separation between
two of the components of intermediate volatility.
For example, the gasoline fraction of crude oil
consists of many hydrocarbon constituents of
widely varying volatility. To refine raw gasoline
further for use as motor fuel, it is necessary to
remove very volatile constituents in a distillation
it is column called a "stabilizer." Usually propane
make it (C 3 H 8 and lighter constituents must be removed,
)

suitable for computer calculations. while butane (C4HK)) and heavier constituents are
If an existing distillation column is to be used retained. In this case, most of the propane should
to separate a new
feed, then the four variables go out in the distillate and most of the butane in
specified are different from the previous case. Now the bottoms.
 MULT COMPONENT STAGE OPERATIONS 137

The two components between which the split the distributions of the light key and heavy key are
isdefined are called the key components or simply specified, so that Equation 8.16 may be written for
the keys. The more volatile of the two is called the the heavy key and any other component.
light key, and the less volatile is the heavy key. In
the gasoline-stabilization example, propane is the
light key and butane the heavy key. Very volatile
log —
x Di
x b,
= /V logoff + log
X °hk
x Bhk
(8.17)

components have extremely low concentration in

Often in the initial estimate the number of moles
the bottoms, and very nonvolatile components are
rather than the mole fraction is known. Thus,
in extremely low concentration in the distillate.
These constituents are referred to as the nondistri-
Equation 8.17 may be written in the equivalent
form
buted components. On the other hand, the keys
and other constituents of intermediate volatility DxDi Dx Dhk
are "distributed" in significant proportion between log — = N log a, + log (8.18)
Bx Bi Bx Bhk
the distillate and bottoms. Some calculation
procedures assume zero concentration of each non- This is an equation of a straight line on log-log
distributed component at the appropriate end of coordinates, as shown in Figure 8.11. The line is
the column. Others, particularly computer determined by the two points defined by the light
methods, assume very small but definite concen- and heavy keys. With this plot and the component
trations (such as 0.00001 mole fraction).
mass balances around the column, the distributions
of all components may be estimated.
Minimum Stages
The distribution may be estimated directly
The minimum number of stages at total reflux may from Equation 8.17 or 8.18 by evaluating the slope
be estimated by Fenske's equation (Equation 7.54) (A/) and the intercept from data on the light and

heavy keys. The estimates are accurate only at

aN= <?DiyxBL\ (754) total reflux, but they may be applied with
\x D jJ\XBi) discretion to finite reflux ratios as a first approxi-
where / and / are any two components; for mation. More accurate methods for finite reflux
example, the / might be the light key and / the ratios are available (4, 13).

heavy key. afj is an average value of a. determined

-

at the average of top and bottom temperatures in Example 8.8. The following mixture must be
the column. separated by distillation so that 95 percent of the
propane charged is recovered in the distillate and
Distribution of Components 90 percent of the butane charged is recovered in
The exact components between the
distribution of the bottoms. Estimate the distribution of the other
distillate and bottoms may be determined only by
rigorous stage-to-stage calculations. But because
estimates of the distribution are needed as a first
step in some of the rigorous procedures, several
approximate methods for estimation are available.
One method estimating distribution of
for
components is based on the Fenske equation
(Equation 7.54). The equation is accurate only for
constant relative and total reflux. In
volatility
practice, a temperature through the
varies with
column, but the variation is often small and if
necessary an average value may be used.
For any component Equation 7.54 may be
/',

written with respect to any other component/:

Xq i
X qj
log = N log a,y + log (8.16) 100
*B i
x Bj

The relative volatility of all components may be Figure 8.11. Graphical determination of
referred to the heavy key for convenience. Usually distribution.
138 STAGE OPERATIONS

components at total reflux for a column operating It ignores the effect of all components other than

at 215 psia. propane and butane. Equation 8.18 may now be

used for calculating the distribution of the other
Feed,
components. For example, for pentane
Component Formula lb moles Kat 185°F Dx D5 3.0
log = 6.5 log 0.41 + log
Ethane C2 H 6 5 4.9 Bx B5 27.0
Propane C3 H 8 20 1.9
so that
n-Butane C4H10 30 0.85
n-Pentane C5H12 20 0.35 Dxp<
= 0.00046
n-Hexane Ce H i4 20 0.17 Bx 55
n-Heptane C7H16 5 0.08
Because
Total 100 lb moles
0x D5 + Bx B5 = Fz F5 = 20

Dx D5 = 0.009
Solution
and Bx B5 = 19.991
For propane, Dx D3 = (0.95H20) = 19.0
These values and others are tabulated below, where
and Bx B3 = 20-19= 1.0
Zf, = FzpilF, etc.
For butane, Dx DA = (0.10) (30) = 3.0 Note that ethane, hexane, heptane, and perhaps
and Bx B4 = 30 - 3.0 = 27.0 pentane can be considered nondistributed com-
ponents. The values of x Dl and x Bl could now be
Each a is determined from /C-values found in used for the first trial in the stage-to-stage calcula-
Appendix D-2, but a temperature is needed. As a tions discussed earlier.
the bubble point of the feed may be used
first trial,

asan estimate of the average column temperature. Stage-to-Stage Calculations

The bubble point is calculated by methods outlined Calculations from stage to stage are based upon the
in Chapter 3 to be about 185° F. /C-values at this bubble-point and dew-point calculations developed
temperature and 215 psia. are tabulated above. in Chapter 3, coupled with mass balances at each
Here the heavy key is butane and the light key, stage.
propane. Each tabulated a is referred to butane In calculating the equilibrium composition of
vapor and liquid at a single stage, /C-values must be
a ihk - — known, but these cannot be determined until the
^butane
stage temperature is determined. The stage
so that a for butane is 1 .0. temperature is a function of composition, however,
The propane and butane values are substituted into so a trial-and-error procedure is required. Much of
Equation 8.18 to evaluate N the trial and error can be eliminated if the relative
volatility is used in place of K, because the former
19.0 3.0
log ^-^- = /V log 2.2 + log
varies little with temperature. The relative volatili-
1.0 27.0 ties are referred to one of the key components, as
in the previous example. For any component /',

So that N = 6.5. This is a rough estimate of the

minimum number of stages required at total reflux. V, = Ki*i

Component <*ihk ZFi 0X0/ Bx Bl Xpi *Bi

Ethane 5.8 0.05 4.99945 0.00055 0.185 7.5 x 10~ 6

Propane 2.2 0.20 19.0 1.0 0.704 0.0137
n-Butane 1.0 0.30 3.0 27.0 0.111 0.370
n-Pentane 0.41 0.20 0.009 19.991 0.0003 0.274
n-Hexane 0.20 0.20 6x 10- 8 20.0 2x 10" 6 0.274
n-Heptane 0.09 0.05 2x 10~ 8 5.0 7x 10" -10 0.0685

D = 27.0 B = 73.0 1.0 1.0

 MULTICOMPONENT STAGE OPERATIONS 139

and for the reference component For the stripping section (From Equation 7.5)

Vr = K r xr B
Vm i rp Xm _ 1 , Xbj (8.27)
so that

/0 X;
D and B are determined by overall mass
= Oi, (8.19) balances on the key components. Equation 8.26 is
Vr Kr Xr Xr used alternately with Equation 8.21 to calculate
where a.ir defined as K,IKr
is .
downward stage by stage from the distillate com-
The sum of the mole fraction in the vapor is 1.0.
position to the feed stage. Equations 8.23 and 8.27
are used to calculate upward from the bottoms
S// = /»+y6+/c+.--/r=1.0 (8.20) composition.
Division of Equation 8.20 by yr gives The stage-to-stage calculations outlined above
may be applied to a wide variety of problems. If

the complete composition is known at one end of

(8.21:
Kr Vr Vr Vr the column, calculation stage by stage to the other
end is straightforward, as shown in the following
Substitution of Equation 8.19 into 8.21 gives
example. On the other hand, if the distillate and
So^x, 1.0 bottom compositions have been estimated, calcu-
lations made from top and bottom must check at
Xr Vr
the feed. This will be considered later.
so that

y r = Xrl^CXirXj (8.22) Example 8.9. A mixture of benzene, toluene,

and o-xylene is to be separated in a distillation
Similarly, for any component /
column equipped with a total condenser and a
y, = airXjH.air Xj (8.23) The feed is a saturated liquid con-
partial reboiler.
taining 30 percent benzene, 45 percent toluene,
Equation 8.23 may be used for calculating the
and 25 percent o-xylene. The bottoms are to
composition of a vapor in equilibrium with a given
contain 95 percent of the o-xylene charged with a
liquid, providing each a is known. An initial
concentration of 0.99 mole fraction o-xylene.
temperature guess permits determination of trial a,
Determine the number of stages required at a
and the temperature is later checked as in a typical
reflux ratio of 3.0 and a pressure of 1 atm.
bubble-point calculation.
A second estimate of T is obtained by calcu-
lating
Solution. Basis: 100 lb moles feed

1 o-Xylene in bottoms = (0.95M25) = 23.75

K.-±- (8.24)
Xr Zot/rXi 23.75
— —— = 23.99 and D = 76.01
Total bottoms: B=
and then determining at what temperature K has 0.99
r

the calculated value. The benzene in the bottoms is negligible, as

To determine the composition of a liquid in verified later.
equilibrium with a given vapor, an analysis slightly
simplified from the one above yields Toluene in bottoms
= (0.01)(23.99) = 0.24
(8.25)
Benzene in distillate

= 30; x Db = 30/76.01 = 0.3947

A mass balance between any stage and one
end of the column is used to calculate from stage Toluene in distillate
to stage. The an operating-line equation
result is
= 45 - 0.24 = 44.76; x Dt = 0.5889
identical to that derived in Chapter 7 for binary
distillation. For any component / in the enriching o-Xylene in distillate
section = (0.05X25) = 1.25; x Dx = 0.0165
D The complete composition of the distillate is
Yn + 1 , /
= .. Xr + -xDi (8.26)
known accurately, so the calculation will begin
140 STAGE OPERATIONS

with the distillate and proceed downward. The for toluene

vapor from the first stage has the same composi-
K„ + i.r
= 0.75x„ +0.1472 (B)
tion as the distillate product. Equation 8.25 is used
f

to find the composition of the liquid leaving the

for o-xylene
first stage. At this moderate pressure, Raoult's law

holds for these three similar compounds, so that

Vn + y.x =0.75x nx +0.0041 (C)
each cc is evaluated from vapor pressure data. An
initial temperature must be assumed. A reasonable
These equations are used to calculate the composi-
value would lie between the boiling points of
tion of the vapor flowing to a stage from the
benzene and toluene (80-1 10.6°C). The average of composition of the liquid flowing from the same
95°C will be used.
stage.
With only three components, the choice of
Calculations are summarized below.
key components is somewhat arbitrary. Toluene
For the first stage, the temperature is calcu-
and o-xylene are chosen as the light and heavy
lated to be 102.7°C, instead of the 95° C assumed.
keys, respectively. The relative volatilities are
It is therefore necessary to recalculate a at the new
referred to toluene, because it has an intermediate
temperature and to reevaluate x and a second
Hence, a for toluene is defined as 1 .0.
volatility.
calculated temperature. It very closely checks the
For the other components a equals the ratio
first, so calculations for stage 1 are complete.
of the vapor pressure of the component to that for
Calculations for the second stage begin by evalu-
toluene. Vapor pressure data may be obtained
ating y 2 from x-\ using Equation 8.26. For
from Appendix D-1 for hand calculations, but for
example
computer calculations, the following empirical
equations are useful: y 2b = 0.75(0.2051 ) + 0.0987 = 0.2525

A temperature at the second stage must be

p =e 7.599-4208.8/7"
1
assumed to evaluate a. As a first trial, the final
px =e 17. 962-4728. 6/7" first-stage temperature may be used and calcula-
tions carried out as before.
where P is in mm
Hg and T is in K. The equations The calculated temperature on the second stage is
predict the vapor pressure from temperature with sufficiently different from the assumed value to
an error of less than 0.4 percent in the temperature require a second trial.
range 80— 145°C at atmospheric pressure. Calculations are continued stage by stage until
For the enriching section L = 3D = 228.03. V = the ratio of the key components in the liquid
Z. +0 = 40 = 304.04, so that L/V= 0.75 and DlV = flowing from a stage is equal to that of the feed
0.25, so that Equation 8.26 becomes liquid. Then the mass balance changes. For a feed
at its bubble point
for benzene

K„ + 1
* = 0.75x„ 6 + 0.0987 (A) L = L + F and V=V

STAGE 1

Trial 1: assume T= 95°C Trial 2: assume T= 102.7°C

Ki/a a *1

0.3947 2.468 0.1599 0.2009 2.410 0.1638 0.2051

0.5889 1.000 0.5889 0.7399 1.000 0.5889 0.7375
0.0165 0.3503 0.0471 0.0592 0.3605 0.0458 0.0574
Ey,/a= 0.7959 1.0000 Zy^a = 0.7985 1.0000

K r
= IkiA* /V and Pr = PKr Pr = 606.9
sothat/>r = (760) (0.7959)
= 601.1 and
r= 102.7°C Therefore, T= 102.9°C
 MULTICOMPONENT STAGE OPERATIONS 141

STAGE 2
Trial 1: assume T= 102.9°C Trial 2: assume T= 108.4°C

V2 oc y 2 loc a y 2 /cc *2
0.2525 2.409 0.1048 2.361 0.1069 0.1140
0.7003 1.000 0.7003 1.000 0.7003 0.7497
0.0472 0.3604 0.1310 0.3704 0.1274 0.1363
Zy 2 / a = 0.9361 2K 2 A* = 0.9346 1.0000
Pr = 711.4 Pr = 710.3
T= 108.4°C T= 108.3°C

so from Equation 8.27, with L = 228.03+ 100 = negligible benzene in the bottoms is justified
328.03 and V= 304.04, L/V= 1.0789,5/1/= 0.0789 because x Bb = 9.54 x 10
-6 Note that the feed is
.

added where the ratio of keys is approximately the

same as the feed composition even though the
actual liquid composition at the feed stage is quite
ymt = 1 .0789x m _ 1 , f
- 0.000789
different. The overall feed composition and feed-

ymx = 1.0789xm _ 1# , -0.0781 stage liquid composition rarely agree in multi-

These mass balances are used below the feed, and
calculations are continued stage by stage until the
concentration of o-xylene equals or is greater than
(x Bx = 0.99) in the liquid flowing from any stage.
Results are summarized in Table 8.4.
The ratio of key components in the feed \sz t lzx =
0.45/0.25,= 1.8. At stage 3, x t lx x = 2.51 and at
stage 4, x t lx x = 1.21, so that the feed would be
between the third and fourth equilibrium stage.
For this calculation, the liquid downflow is
changed after stage 4. At stage 10, x x =0.9902,
which is slightly greater than the required x Bx =
0.99, so 10 stages are required (9 equilibrium
stages plus the partial reboiler). The assumption of
component distillation. In general, it is impossible
to get exact agreement. The best feed location is
usually where the ratio of key components is the
same.
The liquid compositions are plotted in Figure
8.12. Note that the toluene concentration increases
to a maximum and then rapidly decreases. The
occurrence of such maxima in concentration of
intermediate components is common.
This example was solved in a few seconds
using a simple program on a digital computer. The
program included a trial-and-error procedure to
evaluate the stage temperature to within 1°C. Test-
ing for feed-plate location and last stage was
included in the program.

Table 8.4 THREE-COMPONENT DISTILLATION

Stage No. Temperature Liquid Composition Mole Fraction

°C

Benzene Toluene O-xylene

*b *t *x

1 102.9 0.2053 0.7378 0.0569

2 108.3 0.1140 0.7500 0.1361
3 113.1 0.0734 0.6621 0.2645
4 (feed) 118.4 0.0537 0.5187 0.4276
5 126.9 0.0165 0.3583 0.6252
-3
6 134.1 4.28 x 10^ 0.2055 0.7902
-4
7 138.9 9.93 x 10" 0.1041 0.8948
-4
8 141.7 2.17x 10" 0.0491 0.9506
9 143.1 4.59x10" 5 0.0222 0.9777
10 (reboiler) 143.8 9.54 x 10" 6 9.70 x 10~ : !
0.9903
142 STAGE OPERATIONS

lr For minimum reflux, successively lower

\ /Benzene
values of reflux ratio are inserted inthe computer
program, and the required number of stages is
2 -
evaluated. As the minimum reflux is approached
/ Toluene
the number of stages increases rapidly, and some
arbitrary number (such as 100 or 1000) may be
3 •
chosen to approximate infinite stages.

Example 8.10. Determine minimum stages

4 -

and minimum reflux for the separation of Example

8.11.

Si*
\ o -Xylene Solution. Equation 8.28 may be used for an
E estimate of minimum stages. An average relative
c
volatility must be used. The temperature range in
2
CO
6 the column is approximately 100— 145°C, so
a ik-hk is evaluated at the average 123°C.

a= 2.5
7

Then

8
n _
(2.5)" =
M
0.0164 )(
—
0.5889;/ 0.99
0.01
'

so that Nm i n =8.9.
9
may be used in
Alternatively, a reflux ratio of 100
the computer solution of Example 8.11 to yield
ID
N = 8.87 stages, which is the minimum for
0.2 0.4 0.6 0.8 1.0 practical purposes.
Liquid composition, .v For minimum reflux, successively smaller
Figure 8.12. Concentration profiles in Example values of reflux ratio are used in the computer
8.11. solution, starting with the 3.0 of Example 8.11. As
shown in Figure 8.13, the minimum reflux is about
Operating Variables
0.535. From a reflux ratio of 1.0 to 100, there is a
difference of only 4.5 stages. Below 1.0, the
As discussed the minimum number of
earlier,
required number of stages increases rapidly.
stages may be
estimated from Fenske's equation
Economic considerations would probably suggest
written for the heavy key and the light key
operation at a reflux ratio somewhat greater than

«/*-/>* "= —— (8.28)

1.0.

Distillate and Bottoms Compositions Unknown

Several methods are available for estimating
minimum reflux for a desired separation of keys In the earlier example involving distillation of a
(3, 4, 5, 13).Before the advent of computers, these benzene-toluene-xylene mixture, the complete
approximate methods were widely used, together composition was known accurately at one end of
with the minimum number of stages, to estimate the column. It was possible to calculate stage-by-
the stages required at some finite reflux ratio. With stage from the known to the unknown end. This
a computer, it is possible to determine minimum situation is a special case, because more generally
reflux and minimum stages as accurately as desired the composition of neither end is specified
by stage-to-stage To determine
calculations. completely. For example, specification of the
minimum extremely large reflux ratio
stages, an number of stages; feed location, quantity, and
(such as 100 or 1000) may be used in a program composition; distillate quantity; and reflux ratio,
such as that used to solve Example 8.1 1 A value of . together with column pressure and heat addition
100 or 1000 approximates the "true" value of (if any), is sufficient to define the column com-
infinity for purposes of evaluating minimum stages. pletely. In this case, none of the end compositions
 MULTICOMPONENT STAGE OPERATIONS 143

requires specification of the number of equilibrium

stages, so that it is particularly useful for evaluating
potential performance of an existing column. The
calculation procedure begins with the choice of a
reflux ratio and the estimation of the temperature
profile through the column. The temperature
profile determined by the bubble point of the
is

liquid at each stage in the column, but the bubble

point is dependent upon the liquid composition,
which is not yet known. For this reason, it is
necessary to assume a temperature profile, then
calculate compositions, and finally calculate a new
temperature profile. The new profile is again used
to calculate compositions; the procedure is

repeated until there is no further change in

temperature or composition between trials. Such a
repetitive or iterative procedure is especially suited
for computer calculations.
Before calculation can begin, some specifi-
cation of end compositions is necessary. For the
procedure outlined below, it is most convenient
simply to specify the quantity of distillate, D
(presuming the quantity of feed, F, is known).
This, together with the reflux ratio and number of
stages, makes it possible to calculate the end com-
positions. Alternatively, sufficient data on the
desired split in two key components will permit an
estimate of the end compositions.
Calculations proceed stage by stage from each
Figure 8.13. Number of stages as a function of reflux.
end toward the feed stage. Because of the errors
are given, so the earlier method is not applicable. made in the initial estimates of temperatures and
When neither end composition known,
it might
is end conditions, the compositions at the feed stage
seem reasonable to start at the known feed com- calculated from each end will not check each
position and calculate toward both ends. Such a other. It is then necessary to adjust stage tempera-
procedure leads to erroneous results, however. tures and repeat the calculations until the com-
Generally the feed composition is never exactly positions "mesh" at the feed stage.
duplicated at any point in the column. The feed is Examination of the procedure just outlined
added where a stage composition is similar to the shows that the calculation of multicomponent
feed composition, but it is not identical. distillation is inevitably trial and error; it is neces-
Because of the large number of variables that sary to assume a value of one variable and then to
may be specified to define a distillation column, check it later. Before the advent of computers,
many combinations are possible. One was cited in many methods were proposed to alleviate and
the previous paragraph and will be considered in reduce the tedium involved in the calculation. One
detail here, but many other combinations arise in method developed by Hummel (6) is based upon
practice. For example, a separation between key the Thiele-Geddes method. It involves an initial
components may be specified, so that end com- estimation of the temperature profile by calcula-
positions are partially known. The method described tions from both ends toward the feed, and also
here can be applied to most situations commonly calculation a short distance on either side of the
encountered. feed. The resulting pieces of a temperature profile
Numerous methods have been developed for are fitted together and a check is made on the end
stage-to-stage calculations with end compositions compositions. The method can be applied where
unknown. Most of the methods are based on a the number of stages is to be determined. Success-
procedure proposed in 1933 by Thieleand Geddes ful application of the method requires a certain
(10). As originally developed, their method amount of experience and intuition.
144 STAGE OPERATIONS

Thiele-Geddes Method 6. A component balance over the enriching

based section made. With constant molar overflow, the
is
The calculation method described here is

method, but it has

balance gives for component / (see Chapter 7)
upon the early Thiele-Geddes
been extended and modified by several people. The
computer iterative procedure is the work of Lyster Kn + 1./ -
-L X ni + 9_ x Di .. (8.26)
\,
and coworkers (9). Although it is possible to con-
sider enthalpy changes and hence changes in molar A mass balance around the condenser gives
liquid and vapor rates through the column, the D = V—L, so that Equation 8.26 may be re-
simplified treatment presented here assumes arranged to
constant molar flow rates, just as was done in the
Xm
f
—L
-

McCabe-Thiele method for binary distillation in n+

1+1
/"
1 , \
= (8.29)
Chapter 7. This assumption is reasonably accurate X
Xn;
Dl VVx
VlXn;
DJ J

for mixtures of similar compounds, such as

petroleum hydrocarbons, having a modest spread For the equilibrium stage n + 1

in boiling points. More rigorous procedures

= Kn xn + 1 (8.30)
including enthalpy balances are discussed in Yn + 1 , / +1. i" ,/

References 4 and 12. The procedure applies to a where /C„ + is the equilibrium constant for com-
1 ,

column with a total condenser and a reboiler, such ponent evaluated / at the temperature of
as that shown in Figure 8.3. Each step in the stage n + 1
procedure is described below. Combining Equations 8.29 and 8.30 gives

1. The total number of equilibrium stages Xn + ^ ,i 1

and the feed stage are specified. (8.31:

^n + 1./' V x Dl
[

The reflux ratio (L /D) is chosen. XDi

2.
3. The distillate fraction (D/F) is specified. At the top of the column (where n = 1 ) with a
4. The column pressure is fixed. total condenser y-\ =x D and y-\ =/C x , so that
1 1
In the first four steps, often necessary to
it is
X\lx D - MK
Equation 8.31 can then be applied
y
.

try several values of each variable specified before

to calculate the ratio x,lx D at the second stage. ,

the best combination is achieved. This is easy to do

This latter value is then substituted into the right-
on a computer. The procedure outlined here
hand side of Equation 8.31 to evaluate the com-
requires that the variables listed above be assigned
position ratio on Stage 3. This procedure is
specific values for any given run, but many com-
repeated for each stage down to the feed stage,
puter runs may be made. where x Fl /x Di is determined. The K for each stage
5. An initial temperature is assumed.
profile
is evaluated using the assumed temperature profile.
The temperature at each stage determined by the
is
The calculation from the top to the feed stage is
bubble point of the liquid on the stage; but the repeated for each component. The resulting
bubble point depends upon the not-yet-determined
(x Fl /x Dl ratios will be used in steps 8 and 9 to
)

liquid composition. The distillate bubble point

effect a feed-stage "mesh."
must be higher than the boiling point of the most
7. A component balance in the stripping
volatile component in the feed at the column
section made. The procedure is similar to that
is
pressure, so that this boiling point is a convenient
detailed in step 6 for the enriching section. The
first estimate of the temperature at the top of the
derivation starts with a component balance
column (at stage 1 ). Similarly, the bubble point of between the bottom of the column and any stage
the bottom product (B) must be lower than the
m in the column and yields the operating line for
component in the
boiling point of the least volatile
the stripping section.
feed, so this boiling point used as a first estimate
is

of the temperature at the bottom of the column B

(stage N). For intermediate stages, the temperature Ym i ry Xr 1./' x Bl (8.27)
?
is determined by assuming a linear variation
between the top temperature and the bottom where stage (m — 1 is the stage above stage m, and )

temperature. This procedure gives a very rough L and V are the molar flow rates in the stripping
estimate of the temperature profile, but it is section. This equation is substituted into
adequate for a first trial; it will be improved in
successive trials. Y mi >^mi Xr (8.32)
 MULT COMPONENT STAGE OPERATIONS 145

and rearranged to give ing section is calculated from

{x n ;lx Di )x Di
Xm — 1 , / V ^Tri i^r
Xni (8.37)
1 + 1 (8.33) [Z(x ni /x Di )x Di
LI X-Bi
]

This equation is applied to each stage starting The ratio x nj lx DJ is known for each com-
at the bottom (stage N) where x mi = x Bl * and ponent on each stage from step 6.
continuing to the feed stage where x Fi lx Bi is For the stripping section
evaluated for each component. The ratio x Fi lx Bj is
{x mi lx Bi )x Bi
used in steps 8 and 9 to effect a feed-stage "mesh." (8.38)
The flow rates in the stripping section (L, V) may Z[(x mi /x Bi )x Bl ]

be related to those in the enriching section by

Equations 7.11 and 7.15, taking into account the The ratio (x mi /x Bi is known for each com-
)

thermal condition of the feed. ponent in each stage from step 7.

8. The ratio x Bi lx Dj is evaluated. Use of
With the liquid compositions determined
Equation 8.31 to calculate from the top to the from Equations 8.37 and 8.38, the bubble point
feed stage has yielded x Fj lx Dj for each com- can be calculated using procedures described in

ponent; application of Equation 8.33 to calculate Chapter 3.

from the bottom to the feed has given a set of The resulting temperature profile will be
x FJx Bi At the feed, x Fl in these two calculated
.
different from that initially assumed; because the
expressions should be the same, so that temperatures were evaluated from liquid com-
x Fi lx Dl positions based upon the initially assumed
(8.34) temperatures, the new profile will not be the final
\x Di i ca x Fi lx Bi correct one. Normally the new profile is closer to
9. Distillate composition and quantity are the correct profile than was the initially assumed
calculated. Separate values of x Bi and x Di may be profile. To obtain a still better profile, the entire
evaluated from an overall mass balance on the calculation can be repeated using the new profile in
column place of the assumed one. This second trial (or
iteration) yields a new profile that can be used as a
Fz Fi = Dx Dl + Bx Bi (8.35)
starting point for a third iteration; the iterative
Rearrangement gives process can be continued until the temperature
Fz Fi profiles in two trials are identical within the
*Di ~ (8.36) accuracy desired. At this point, the calculated D of
D + B(x Bi lx Di )
Equation 8.36 checks the originally assumed D
where F and the z Fj 's are known, D has been with a specified accuracy. This procedure is called
assumed, and B can be evaluated from B = F — D. the direct-iteration method of convergence. Use of
Equations 8.34 and 8.36 are used to calculate the it yields successive temperature profiles that tend

x 0/ 'sand thex^/s. to converge on the correct profile. The direct-

Because of the inaccuracy of the assumed iteration method gives a rather slow convergence,
temperature profile, the sum of the x Di 's is not so that many iterations may be required. This is
1.0, as it should be. In other words, the calculated time-consuming, even on a digital computer. There-
value of D does not equal the initially assumed fore, several methods have been developed to force
value: convergence more rapidly. One such method is
described in the next section.
D ca =DZx Dl

10. A new temperature profile is calculated

6 Method of Convergence
using the bubble point of the liquid at each stage.
The liquid composition on each stage in the enrich- The 6 method is modification of the Thiele-
a
Geddes procedure described
in the previous
section. It gives much more rapid convergence than
direct-iteration and it can be applied to a wide
*For the purposes of calculation it may be assumed that xm j = x B ,
variety of distillation problems (4). The procedure
for both partial and total reboilers. For a partial reboiler, the
number of stages in the column will equal one less than the number
begins at step 9 of the Thiele-Geddes method. Use
calculated. of the calculated value of [x Bi lx Di ca from )
146 STAGE OPERATIONS

Equation 8.34 in Equation 8.36 leads to an The procedure for finding 6 follows:
incorrect value of x Dil because (x Bi lx Di ) ca is
1. Carry the Thiele-Geddes method through
incorrect. Suppose a quantity 8 is arbitrarily
step 8 to evaluate a set of (x Bi /x Di ) ca , one value
defined to "correct" (x Bi /x Di ca )
for each component.
x Bi ' x Bi 2. Assume an initial value for 8 (such as 1.0),
= 8 (8.39) then evaluate f (8) and f (8) from Equations 8.44
x Di \
CO XDI- and 8.45. Calculate a new value of 8 by substitut-
ing the assumed value into Equation 8.46, together
The subscript co refers to the corrected value
of the ratio, whereas ca refers to the calculated
with the calculated values of f(8) and f'(6).
3. Use the new value of 8 to obtain new
value. If 8 were known accurately after the first
trial, it would be possible to reach the correct
values of f{8) and f'W), and then use Equation

answer immediately. Unfortunately, 8 is not 8.46 to find the next value of 0.

but it can be calculated by 4. Iterate on Equations 8.44, 8.45, and 8.46
known accurately,
until two successive values of 8 agree within the
successive iterations.
Substitution of (x Bi /x Di c0 ) into Equation desired accuracy (perhaps 6 t + ^ — 8 t = 0.0001 ).

8.36 gives an accurate value of x Dl After 6 is determined, it is substituted in

FZFi Equation 8.41, and a set of (x Dl co is found. The )

(x Dl )co = (8-40) bottoms composition is determined from Equation

n
D + R Bl /x Di
, B(x , , .

) C0
8.39.
Combining Equations 8.39 and 8.40 gives *Bi
Us,) (*o/) (8.47)
F, *Di
(*D/) [SAM
D + B8(x Bi /x Di )t Next the liquid composition on each stage is

determined from
The sum of the mole fractions in the distillate
must total 1 .0 when the correct answer is achieved \X n jlXQj) ca \Xqj) C0
Xni (8.48)
2 [\X n j I X o /) C a( x D /) co J
2U D/ ) C0 =1-0 (8.42)

for the enriching section, and

Combining Equations 8.41 and 8.42 gives

\X m jlX B ,) ca \X B j) C0
Fz F xm (8.49)
,

1 =0 (8.43)
i
X[{x mi/x B i) ca (XBi)co]
lD + Bd{x Bi /x Di )t

for the stripping section. The ratios (x ni /xD,) ca

This equation be solved for 8 by
can
and have been calculated earlier in
(x mi /x Bi ) ca
Newton's method, which can be applied in an
steps 6 and 7 of the Thiele-Geddes method.
iterative procedure to yield the correct solution.
The new liquid compositions calculated by
For Newton's method, a function of 8 is
Equations 8.48 and 8.49 are used in bubble-point
defined
calculations to evaluate the new temperature
Fz Fi profile. At this point, a return is made to step 5 in
f{6) = Z (8.44) the Thiele-Geddes method, and the new tempera-
D + B8[x Bi lx Di )

ture profile is used to calculate again through steps

6, 7, and 8. Then a new 6 is found using the four
and its derivative with respect to 8 is taken
steps of the 6 method.
The repetitive calculation proceeds through
B{x Bi lx Di ) ca Fz Fi
f'(d) =-Z (8.45) several iterations, alternately using the Thiele-
[D + Bd(x Bi /x Di ) ca ; Geddes method and the 6 method. The 6 method
is a better procedure than direct iteration for
Comparison of Equations 8.43 and 8.44
calculating new temperatures and compositions
shows that f(8) =0 when the correct 6 value has
been found. because the 8 method converges to the correct
The any answer with fewer iterations. The calculation is
8 at trial t can be used to calculate a
completed when the calculated 6 is equal to 1.0,
better value, according to Newton's method.
within a specified accuracy.
f+ i
=e -f(d)/f(d)
t (8.46) The 8 method was designed specifically for
 MULTICOMPONENT STAGE OPERATIONS 147

computer solution of multicomponent distillation Top

problems. Although the principles of the method

are straightforward, the repetitive calculations are
too laborious for hand calculation. To learn the
principles of the method, it is instructive, however,
to carry a hand calculation through one iteration,
even though it requires considerable time. Program-
Feed
ming the 6 and Thiele-Geddes method for the
digital computer is not difficult, once the program-
mer clearly understands the method. Prepared
programs are available on this or similar methods in
many program libraries. The following example
was solved using a digital computer. 14 H
Bottom 15
Example 8.11. A liquid feed at its bubble 240
point is to be fractionated in a column equivalent
to 15 equilibrium stages (including a total con- Figure 8.14. Temperature profiles for Example 8.11.

denser and a partial reboiler). The feed is at the

eighth stage from the top. The feed composition 6. Calculations proceed from the top of the

in mole fractions is:

column to the feed. First, it is necessary to have K
data at 45 psia. as a function of temperature. At
Isobutane (iC 4 H 10 ) 0.1 this moderate pressure, it is adequate to assume
Normal butane (nC 4 H 10 ) 0.3 that K = P V /P T , so that the vapor pressure data
may be used. The data
are obtained from Appen-
Isopentane (iC 5 H 12 ) 0.2
dix D-2d the form of empirical equations. Al-
in
Normal pentane (nC 5 Hi 2) 0.3 though the graph of Appendix D-1 might be more
Normal hexane (nC 6 H 14 ) 0.1 convenient for hand calculations, empirical equa-
tions are better for computer calculations, and
The column operates at 45 psia. For a reflux ratio they will be used here.
of 3.0 and DIF ratio of 0.4, calculate the
The detailed calculation for n-butane is shown.
percentage recovery and concentration of butanes Equation 8.31 becomes: (with
For the first stage,
in the distillate.
n= 1)

The combination of the Thiele-

Solution. *2_ = JL + 1
Geddes method and the 6 method is used. Let V\x D J
F = 100 lb moles as a basis, so that D = 40.0.
First the Thiele-Geddes method is applied: where LIV = 0.75.
K-\ may be calculated from the vapor pressure

1. From the data given, /V=15, feed data of Appendix D-2e. The metric constants
stage = 8 (from top). given there can easily be converted to the English
2.L /D = 3.0, so that L/V = 3/4 = 0.75. units required in this example:
3. DIF = 0.4.
4. /> = 45 psia.
A'=A - 1.71350
5. To determine an initial temperature
profile, the bubble point of isobutane at 45 psia is C = \C- 32.000
read from Appendix D-1 as approximately 75° F.
where A, B, and C are the metric constants given in
This value is used as a first estimate of the top the Appendix, and A', B' and are English-unit
, C
temperature (stage 1). Similarly, the bubble point constants to use with T^F and P v psia. The
of n-hexane at 45 psia (240° F) is used as the first
English-unit constants for the components in
estimate of the bottom temperature (stage /V). this example are:
These temperature estimates can be read from a
graph, because no great accuracy is required for A' B'

this initial estimate of the profile. A linear iC 4 H 10 5.03458 1589.040 400.00

temperature profile is assumed between the top nC 4 Hio 5.11679 1702.620 400.00
and bottom temperatures, as shown in Figure iC 5 H 12 5.07617 1836.0216 387.575
8.14. Thus, the temperature at stage 2 is 86.79°F, nC 5 H 12 5.13871 1916.334 385.600
and so forth. nC 6 H 14 5.16426 2108.754 371.859
148 STAGE OPERATIONS

With a total condenser x,/x D = 1//CV For i-butane 7. In the stripping section, Equation 8.33 is
at75°F used to calculate the ratio x mi lx Bi for each com-
ponent at each stage from the reboiler to the feed
1589.04
log-, o P v = 5.03458 stage. The partial reboiler may be counted as stage
400.0 + 75 15 in this problem, so that x 15 = x B for each

P v =48.891 psia component. Then for isobutane, Ty 5 = 240°F, so

AC 15 = 7.91576. V and L must be calculated. Be-
So that
cause the feed is a liquid at its bubble point, / = 1
*1 45 Now L = L+iF, and L = 3D = 120, so that L -
= 0.92041
xD 48.891
120+100 = 220^ V=V-{1-i)F, and V = LI
0.75 = 160 and V= 160. Then Equation 8.33 gives
Now K 2 -PvIPt evaluated at the second stage
where 7"2 = 86.79° F, so that P v = 58.909 and
K 2 = 58.909/45 = 1 .3091 Then .
— ™
xB
=
220
[7.91 576( 1 )
- 1 ] + 1 = 6.02962

*2 1 To calculate the ratio for stage 13, Equation 8.33

[0.75(0.92041 - 1) + 1] =0.71821 is applied again, with /C 14 =7.11070 at Ty 4 =
xD 1.3091
228.21° F so that
Then for stage 3, with 7= 98.57° F, K3 = 1.5679,
and

*3 1
xn
—
xB
=
160
220
[7.11070(6.02962) - 1] =31.45429
[0.75(0.71821 - 1) + 1] = 0.50432
xD 1.5679
The procedure is repeated at each stage for each
The procedure repeated until x Q /x D is cal-
is
component to give the ratios of x,lx B in the table
at the top of page 149.
culated for isobutane. Then it is repeated for the
other four components. The concentrations of the two most volatile com-
The ratios of (x n /x D for each of ) the five ponents are very low the bottoms, but they
in

components are given in the table below. increase rapidly as calculations move from the re-
boiler to the feed stage. These more volatile com-
The ratios for the two more volatile components
decrease gradually from the top stage to the feed.
ponents are "stripped" out of the liquid as it flows
To put it another way, the vapor phase is enriched from the feed. On the other hand, the concen-
tration of the least volatile component increases
in the more volatile components as it passes up-
gradually in the liquid flowing from the feed to the
ward from the feed to the condenser. Conversely,
reboiler. The values at stage 8 in the table are used
the least volatile component (which is present in
in the next step.
very small quantity in the distillate) increases
8. The ratio of bottoms composition to distil-
rapidly in concentration as the liquid flows from
the top downward toward the The inter- feed.
late composition is calculated for each component
mediate components may go through maxima in from Equation 8.34. For i-butane, for example:

composition (as for nC 5 H 12 and iC 5 H 12 in this

case). The ratios for stage 8 are used in further 0.10168
= 5.07341 x 10" 6
calculations in step 8 below. 20041.6713

Stage
No. 7"°F iC 4 H 10 nC 4 H 10 iC 5 H 12 nCsH 12 nC6H 14

1 75.000 0.92041 1.32095 3.51705 4.73172 16.14411

2 86.786 0.71821 1.34206 8.09337 14.15345 160.06193
3 98.571 0.50432 1.12356 14.27083 32.32769 1126.71
4 110.357 0.33911 0.81479 20.13149 58.54691 6151.83
5 122.143 0.23156 0.53988 23.18194 85.85287 26409.51
6 133.929 0.16664 0.34790 22.08449 103.18182 90163.27
7 145.714 0.12720 0.23161 17.60048 102.64746 247375.44
8 157.500 0.10168 0.16501 11.86212 85.2767 550777.6
 MULTICOMPONENT STAGE OPERATIONS 149

Stage
No. r°F iC 4 H 10 nC4Hio 1C4H10 nC 5 H 12 nC 6 H 14

15 240.00 1.0 1.0 1.0 1.0 1.0

14 228.21 6.02962 4.89579 2.55870 2.19582 1.11659
13 216.43 31.45430 20.44754 5.4143 3.96064 1 .07899

12 204.64 145.79394 75.06254 9.78843 6.05144 0.93525

11 192.86 601.08504 242.8143 15.23912 7.89992 0.75782
10 181.07 2196.5677 689.9697 20.42914 8.82299 0.60243
9 169.29 7081.9290 1714.0896 23.50775 8.42111 0.49094
8 157.50 20041.6713 3702.8673 23.11656 6.86443 0.41960

At this point, the Q method is introduced to The correct distillate and bottoms compositions
obtain a second-trial temperature profile. Equat- are
ions 8.44, 8.45, and 8.46 are used, but first an 1C4H1 nC4Hio 1C5H12
initial value of 6 is assumed. Ultimately, 6 must
(xd)co 0.249781 0.744280 0.005588
converge on 1.0, so this seems like a reasonable
(*b)co 0.0001456 0.0038126 0.329608
first assumption. Alternatively, an initial value of

0.0 might be used. For 6y = 1.0, Equation 8.44 is nC 5 H 12 nC 6 H 14

applied with F = 100 and B = 60. To evaluate f{6)
(x D ) co 0.000351 0.000000
and f(6), the calculations as tabulated at the (x B ) co 0.499766 0.166667
bottom of the page are used.
Next, Equations 8.48 and 8.49 are used to
Now this new value of 6 is used in place of 1.0 to
calculate stage compositions. For example, for i-
evaluate a further value of 6. The procedure is
butane at stage 2 with data from step 6 of the
continued until two successive values of 8 agree.
Thiele-Geddes method:
Using a computer gives 6 = 114.9599 with an ac-
curacy within f{d) =+0.0001. (f(d) should be zero (0.71249)(0.24978)
when 6 has converged on the correct value). *2 ~
[(0.71249X0.24978) + (1.34206)(0.74428)
Equation 8.41 is used to calculate (xo,) co
+ (8.09337)(0.005588) + (14.1 5345X0.000351)
with the final 6. For example, for isobutane
+ (160.06193) (0.000000)]
10 = 0.14603
\ X D ho ~ 6
40 + (60)( 114.9599X5.07341 x 10~ )

For stage 2, the other components yield

= 0.249781
1C4H10 nC4Hio iCsHi2
(xb)co is evaluated from Equation 8.47. For
i-butane x2 0.146034 0.813117 0.0368123

(x B ) co =(1 14.9599X5.07348 x 10- 6 )(0.249781)

n C5H-|2 nCeH^
= 0.0001456 x2 0.0040415 0.000000

Fz F B(x B lx D ca Fz F
)

Componem : Fz F (x B lx D ) ca D + B6(x B /x D ca ) [D + Bd(x B /x D ca )2

)

iC 4 H 10 10 5.07341 x 10~ 6 0.249997 1.9025 x 10~ 6

nC4Hio 30 4.45636 x 10- 5 0.74995 5.01273 x 10- 5
'C 5 H 12 20 0.513144 0.28253 0.12288
nC 5 H 12 30 12.42298 0.03820 0.036253
nC 6 H 14 10 1312628.4 1.2697 x 10~ 7 1.2697 x 10- 7

Sum 1.32072 0.15918

Thus/(0) = 1.32072- - 1 = 0.32072

f'(6) = -0.15918

and by Equation 8.46

2 = 1.0 -0.32072/1-0.15918) = 3.0148

150 STAGE OPERATIONS

These composition values are used to cal- important variables are discussed in this section. To
culate a new stage 2 temperature using a simple make the discussion quantitative, the system of
bubble-point calculation described in Chapter 3. Example 8.11 is used as a basis, and changes are

The resulting temperature (89.47° F) is plotted in made the various variables. The recovery and
in

Figure 8.14, together with temperatures for other purity of the butanes are used as criteria. That is, it
stages calculated from the new compositions eval- is presumed that the object of the distillation is to

uated by the same procedure as given above for recover a mixture of isobutane and n-butane in the
stage 2. This new temperature profile is used as a distillate. The percentage recovery of butanes in

starting point for a second complete iteration
(starting where the linear temperature profile was
first assumed).
The value of
(114.9599) is far removed from 1.0 because the
initial assumed temperature profile
curate. This large value of
x Di)ca t0 9' ve more reasonable compositions from
which to calculate the second temperature profile.
The 6 for the second complete iteration is found to
be 0.3623; for further iterations, d is 1.1648
(third), 0.9881 (fourth). Five complete iterations
are required to give 6 = 1.0001275, which is close
enough to 1 .0. As an alternative, it is possible to
continue iterating until temperature profiles check
within an arbitrary accuracy (such as 0.01°F). A
third alternative is to require that the stage com-
positions as calculated by Equations 8.37 and 8.38
sum to 1 .0 on each stage to within, say, ±0.001
The final composition profiles are shown in
Figure 8.15. Note that the intermediate com-
ponents may pass through a maximum concen-
tration at intermediate points in the column.
the distillate is to be high, and the concentration of
other components in the distillate is to be as low as
possible. To isolate the effect of each variable, all
6 for this first iteration other variables are held constant at the values of
Example 8.11. For example, to study the effect of
is quite inac- varying the total number of stages, the other vari-
6 corrects the [x Bi l ables (reflux ratio, DIF ratio, and feed tray loca-
tion) were held constant. In an actual design all of
these variables and others would be considered
together to determine the best economic design.
Number of Stages
As shown in Figure 8.16, the recoveries of both
butanes and their combined concentration in-
creases as the total number of stages increases. This
is as expected, since more stages give greater frac-
100

Design Variables

A number of design variables influence the re-

covery and purity of distillate and bottom prod-
ucts in multicomponent distillation. A few of the

2 lV
'
/' ' '
\
3
4 nC'4 —
5 "
/\\^ 5 ^^^ —
6 —
7
L / "^n^x^^^^
Feed 8 —
9 c6 y' \\
10 -
11 -
/

—
12
i —
13
f\ J \
14
15
-
VN^
0.1
i

0.2
i

0.3
// 1

0.4
\J
0.5
1 1
_

Figure 8.16. Variation with number of stages of total

0.6 0.7 0.8 0.9
x mole percent butanes in distillate and percentage
Figure 8.15. Composition profiles for Example 8.11. recovery of butanes.
 MLLTKTOMPONENT STAGE OPERATIONS 151

tionation of the lightest and heaviest components. D/F Ratio

The number of stages ultimately chosen depends A smaller D/F ratio (that is, withdrawal of less
upon economic factors. the increased recovery
Is
distillate per mole of feed) leads to higher-purity
worth the expense of additional stages? Are there butanes, but reduces the percentage recovery of
potential customers for higher-purity butanes at a butanes. In Figure 8.18, there appears to be an
higher cost? In the computer calculation of the optimum D/F ratio of 0.4, where both purity and
data for this figure, the feed stage was the middle recovery are higher. In this case, all variables are set
stage in each column. All variables except N are the at the values of Example 8.11, except for D/F.
same as in Example 8.1 1

Reflux Ratio
Both recovery and concentration of butanes
creases as reflux ratio is increased, as
Figure 8.17. In this figure, all conditions are the
same as Example 8.11, except that L Q /D is varied.
Higher reflux ratios for the same distillate produc-
tion increase the internal liquid and vapor flow
rates. This in turn increases steam and cooling
water costs and may raise equipment costs by
requiring a greater column diameter to handle the
greater flows. Here again, an economic balance
must be reached.
Feed-Tray Location
Figure 8.19 shows that an optimum exists in feed-
in- tray location. The optimum occurs near the middle
shown in of the column. Location near either end reduces
both purity and recovery of butanes.
This brief discussion of design variables is
intended to show the complexity involved in
choosing the best design. The design engineer
searches for an optimum in a multidimensional
system.

in

BC
10 ::

l /d

Figure 8.17. Recovery and concentration as a Figure 8.18. Recovery and concentration of C4 in

function of L /D. D.
152 STAGE OPERATIONS

1.000
relatively nonvolatile component. The "solvent"
added flows downward from stage to stage, increas-
ing the relative volatilities. It finally flows out with
the bottoms.
Azeotropic or extractive distillation is used to
separate both binary and multicomponent mix-
tures. For simplicity, only binary mixtures are dis-
cussed here, but the principles and calculations are
the same for multicomponent mixtures.
Azeotropic or extractive distillation is based
on the interactions of the components of nonideal
solutions. The azeotropic or extractive agent is

chosen because it interacts more strongly with one

of the components than it does with the other. An
azeotropic agent may be added to a nearly ideal
binary mixture of low relative volatility (that is,
the two components have nearly the same boiling
point). If the structures of the original components
are sufficiently different, the azeotropic agent may
interact with one of them (for example, by hydro-
° 99.0
gen bonding), thereby increasing their relative vol-
atility.
Theoriginal binary solution may be suf-
ficiently nonideal to contain a binary azeotrope
that blocks the production of pure components by
4 5 6 7 8 9 10 simple distillation. Addition of an azeotropic agent
Feed tray location (15 trays total count from top) may "break" the binary azeotrope and allow pro-
Figure 8.19. Effect of feed tray location on D recovery and
duction of pure components, although it may re-
concentration. quire additional separation steps to recover the
azeotropic agent. For example, ethyl alcohol and
water form a binary azeotrope at 0.89 mole frac-
tion alcohol, so that this is the richest alcohol
AZEOTROPIC AND EXTRACTIVE DISTILLATION
solution that can be recovered by simple distil-
Many mixtures are difficult to separate by distilla- lation. Addition of benzene as an azeotropic agent
tion because their relative volatilities are yields a distillate product consisting of a ternary
close to 1.0. Some of these mixtures may be sep- azeotrope. The bottoms is the desired pure alcohol.
arated by the addition of another component that Sufficient alcohol is lost in the ternary azeotrope
increases the relative volatilities of the original con- to require its recovery. On condensation, the ter-
stituents. Fractionation after the addition of a nary azeotrope in the distillate splits into two liquid
component to increase relative volatilities is refer- phases. The benzene-rich phase is returned to the
red to as azeotropic distillation or as extractive first distillation column, while the water-rich phase
depending
distillation, upon the nature of the is sent to a second column. In this second column,

added component. the distillate is again the ternary azeotrope, which

In azeotropic distillation the added com- is condensed and sent to the phase splitter (to-
ponent is relatively volatile, so that it appears in gether with the distillate from the first column).
the distillate. In many cases it forms a low-boiling- The bottoms from the second column is an aque-
point azeotrope with one of the original com- ous solution of alcohol. It is sent to a third column
ponents, the low-boiling azeotrope goes overhead that produces the water-alcohol azeotrope as the
in the distillate, and the remaining components go distillate and pure water as the bottoms. This distil-
out at the bottoms. Of course, it is then necessary late is mixed with the feed to the first column.
to separate the azeotropic agent from the original Thus, the net effect of the process is to produce
component by an appropriate separation process. pure alcohol and pure water from the alcohol-
Extractive distillation involves the addition of a water azeotrope.
 MULTICOMPONENT STAGE OPERATIONS 153

Calculations azeotropic distillation are

for PROBLEMS
nearly the same
those for simple multi-
as
component distillation. The addition of an azeo- 8.1. Attempt to list all possible combinations of in-
tropic agent to a binary mixture produces a multi- dependent variables that would exactly define a single equi-
component mixture. Stage-to-stage calculations librium stage processing a three-component mixture.
similar to those developed earlier in this chapter Determine the number of independent variables
8.2.
can be used to determine the separation possible that must be specified to define a partial condenser. Suggest
with a given column or to determine the number of a suitable combination of variables.
stages required for a given separation. Two differ- 8.3. Determine the number of independent variables
ences arise. Equilibrium data are more complex that must be specified to define a total reboiler. Suggest a
because the K for a given component is a function suitable combination of variables.
not only of temperature and pressure, but also of 8.4.some systems of multiple distillation col-
In
the nature and concentration of other components umns, the condenser of one column is combined with the
present. Thus, complete nonideal equilibrium data reboiler of another column operating at a lower pressure, so
must be available for the system of interest. A that the condensation supplies the heat for the vapor-
second difference arises from the fact that the ization. Additional heat may also be added or removed
azeotropic agent may be fed at more than one
directly. For the system shown in the figure, determine the
point in the column. Calculations must take into
number of independent variables that must be specified to
account these intermediate streams. In extractive define the system for C components.
distillation, the nonvolatile solvent is fed as an
intermediate stream near the top of the column. Lower- pressure column
A thorough treatment of azeotropic and ex-
tractive distillation is beyond the scope of this
**1 K
book. An excellent discussion is available in Refer-
ence 12.

Higher- pressure column

REFERENCES

1. Andersen, L. B., and L. A. Wenzel, Introduction to 8.5. Determine the number of independent variables
Chemical Engineering, McGraw-Hill, New York (1961). that must be specified to define the system in Problem 6.23.
2. Gilliland, E. R., and C. E. Reed, Ind. Eng. Chem., 34, 8.6. Determine the number of independent variables
p. 551 (1942). that must be specified to define the system in Problem 6.24.
3. Hengstebeck, R. J., Distillation: Principles and Design 8.7. Consider a distillation column with two three-
Procedures, Reinhold, New York (1961 ).
component feeds, a total condenser, and a partial reboiler.
4. Holland, C. D., Multicomponent Distillation, Prentice
How many independent variables must be fixed to define
Hall, Englewood Cliffs (1963).
the system? Suggest a suitable set of variables that would
5. Holland, C. D., and J. A. McDonough, Petr. Ref., 41,
define the system.
No. 3, p. 153(1962).
6. Hummel, H. H., Trans. A.I.Ch.E.,40, p. 445 (1944). 8.8. A distillation column with one feed, a total

7. Kwauk, M., A.I.Ch.E. Journal, 2, p. 240 (1956). condenser, a total reboiler, and six product streams (inclu-

8. Lewis, W. K., and G. L. Matheson, Ind. Eng. Chem., ding D and B) is to be designed to process a six-component
24, p. 494 (1932). mixture. How many independent variables must be fixed to
9. Lyster, W. N., S. L Sullivan, D. S. Billingsley, CD. define the system? List one suitable set of variables.
Holland, Petr. Ref., 38, p. 221 (1959). 8.9. A gas containing 90 percent methane, 5 percent
10. Thiele, E.W., and R. L. Geddes, Ind. Eng. Chem., 25,
ethane, 3 percent propane, and 2 percent butane (by vol-
p. 289(1933).
ume) is to be scrubbed with a nonvolatile hydrocarbon oil
11. Sherwood, T. K., R. L Pigford, and C. R. Wilke, Mass
215
in a bubble-cap column operating at psi and 60 F.
Transfer, McGraw-Hill, New York (1975).
Ninety-eight percent of the propane is to be removed. The
12. Smith, B. D., Design of Equilibrium Stage Processes,
entering oil contains none of the gas components.
McGraw-Hill, New York (1963).
13. Van Winkle, M., Distillation, McGraw-Hill, New York (a) Determine graphically the minimum L/V ratio.

(1967). (b) Determine graphically the number of stages re-

14. Wang, J. C, and G. E. Henke, Hydrocarb. Process., 45, quired and the recovery of all components for an L/V of
No. 8;p. 155(1966). 0.8.
154 STAGE OPERATIONS

(c) Solve (a) and (b) analytically. Feed: 60 mole percent benzene
What L/V must be used to maintain the same
(d) 30 mole percent toluene
recovery of propane for the same number of stages as (b) if 10 mole percent o-xylene
the pressure is reduced to 44.7 psi at 60° F?
8.10. A hydrocarbon liquid containing 4 percent n- The distillate is to contain 98 percent of the benzene
pentane, 2 percent isopentane, and 3 percent n-hexane is to charged at a concentration of 0.99 mole fraction.
be stripped with a nonreactive gas to remove 90 percent of (a) Determine the number of stages required at a

the n-pentane. The feed is 100 lb moles/hr. The operating reflux ratio of 2.0.

pressure is 44.7 psi. (b) Draw a composition-versus-stage number dia-

(a) Operating at 160°F, determine graphically the gram.

number of stages required at twice the minimum stripping (c) Estimate the minimum reflux ratio and the mini-

gas rate; determine the recovery of pentane and hexane. mum number of stages.
The column of part (a) has been built and is
(b) 8.16. One hundred pound moles/hr of a saturated

operating. The recovery of n-pentane is to be increased to liquid mixture of composition 50 mole percent benzene, 20
95 percent. At what temperature should the stripper be mole percent toluene, and 30 mole percent o-xylene is to
operated at the same V/L? be separated into a distillate with not over 0.5 lb mole
(c) Repeat (a) analytically. o-xylene and a bottoms containing less than 0.5 lb mole
8.11. A gaseous mixture containing 90 percent ethyl- toluene. Pressure is 1 atm.
ene, 6 percent ethane, 3 percent propylene, and 1 percent (a) At a reflux ratio of 3.0, how many stages are
propane is fed to an absorption tower, which is equivalent required?
to seven equilibrium stages. The gas is fed at the rate of 50 (b) Draw a temperature-stage number and a com-
lb moles/hr and a nonvolatile hydrocarbon oil is used as an position-stage number graph.
absorbent. Using Appendix D-2d, determine a temperature, a (c) Plot number of stages as a function of reflux
pressure, and an absorbent flow rate that would ensure 99.9 ratio.

percent pure ethylene leaving the top of the absorber. 8.17. A mixture of 50 mole percent benzene, 30
8.12. An absorption column equivalent to six equi- percent toluene, and 20 mole percent o-xylene is to be
librium stages is to be used to recover 95 percent of the fractionated into a distillate containing 99.8 mole percent
ethane from 80 lb moles/hr of a gas containing 78 percent benzene, 0.2 mole percent toluene, and a negligible amount

methane, 7 percent ethane, and 15 percent propane. A of o-xylene.The composition of benzene in the bottoms is
nonvolatile hydrocarbon oil is used as an absorbent. Oper- 0.0030 mole fraction. The distillation column operates at 1
ating conditions are 465 psi and —20 F. atm pressure with a total condenser and a partial reboiler.
(a) What absorbent rate should be used? (a) How many actual stages are required at a reflux
(b) What is the composition of the gas leaving? ratio of 10.0 and an overall stage efficiency of 60 percent?
8.13. One million ft 3 /hr (measured at 1 atm and (b) Draw a composition-stage number diagram. In-

32°F) of gas containing 79 percent methane, 9 percent clude the temperature profile.
ethane, 5 percent propane, 4 percent n-butane, 2 percent (c) Estimate the minimum stages and the minimum
isobutane, and 1 percent n-pentane is to be treated in an reflux.
absorption column containing 40 actual stages. The absorb- 8.18. A mixture of isobutane, n-butane, isopentane,
ent oil (density 49 lb/ft 3 ) is supplied at the rate of 20,000 and n-pentane is to be fractionated in a column equivalent
gal/hr. The molecular weight of the oil is 150. The tower to 35 equilibrium stages, plus a total condenser and a
operates at 44.7 psi and 70° F. The overall stage efficiency partial reboiler. The distillate must have a composition of
is 20 percent. Calculate the recovery of each component. 99.0 percent isobutane, with 95 percent of the isobutane in

8.14. A liquid containing 3 percent methane, 6 per- The column operates at 100 psia. The feed is
the distillate.
cent ethane, 4 percent propane, 3 percent isobutane, and 2 a saturated liquid with 24 percent isobutane, 30 percent
percent n-butane (the remainder is a nonvolatile oil) is to be n-butane, 26 percent isopentane, and 20 percent n-pentane.
stripped to remove 98 percent of the propane. The column What reflux ratio is required?
is to operate at 44.7 psi and 120°F. 8.19. The feed given below is to be separated into a
(a) Determine a suitable V/L ratio and the number of distillate containing less than 0.2 lb mole/hr propane and a

stages required for the recovery. bottoms with less than 0.5 lb mole/hr propylene.
(b) What is the recovery of the other components?
Feed lb moles/hr
8.15. The following saturated liquid is to be separ-
ated into a distillate and bottoms in a distillation column Propylene (C 3 H 6 ) 35
operating at 2 atm equipped with a total condenser and a Propane (C3H8) 25
partial reboiler. n-Butane (nC4Hio) 40
 MULTICOMPONENT STAGE OPERATION'S 155

Operating pressure is 250 psia. There is a partial condenser An existing column equivalent to 50 equilibrium stages is to
and a total reboiler be used. Determine and plot the percentage recovery of
(a) Estimate the minimum reflux ratio and the mini- propylene in the distillate as a function of reflux ratio.
mum number of stages required at total reflux. Pressure is 250 psia.
(b) Determine the temperature profile and the num- 8.24. Pure n-butane (99.7 percent) is to be recovered
ber of stages required at twice the minimum reflux ratio. from the following mixture:
(c) Plot liquid compositions as a function of stage
n-Butane 55 percent
number for twice the minimum reflux ratio.
n-Pentane 30 percent
8.20 The following saturated liquid mixture is to be
n-Hexane 15 percent
stripped of all of its constituents lighter than butane. No
more than 1 percent of the propane charged can appear in At a reflux ratio of 10, plot the percentage recovery of
the bottoms, and 2 percent of the butane may appear in butane in the distillate as a function of total number of
the distillate. stages. Pressure 20 psia. is

Feed lb moles/hr 8.25. Check the calculations in Example 8.11.

Methane (CH 4 ) 12 (a) Make stage-to-stage calculations from both ends
Ethane (C 2 H 6 ) 18 to the feed stage and determine (xg/xQ) ca for n-pentane
Propane (C 3 H 8 ) 10 only.
n-Butane (nC^irj) 60 (b) Using the value of = 3.0148, continue iterations

Total pressure is 250 psia. There is a total reboiler and a

until 6 converges on 1 14.9599.
partial condenser. (c) Calculate a second-trial temperature for stage 3.

(a) How many equilibrium stages are required at a 8.26. A mixture containing 25 mole percent n-

reflux ratio of 3? butane, 35 percent n-pentane, and 40 percent n-hexane is

(b) Plot temperature and composition as functions of to be fractionated to produce nearly pure n-butane. A
stage number for Lq/D = 3. column equivalent to seven equilibrium stages operating at

(c) Plot number of stages required as a function of atmospheric pressure is available. The feed stage is at the

Lq/D from minimum to total reflux. middle of the column and a maximum reflux ratio of 4.0

8.21. A liquid mixture of 25 percent n-hexane, 40 may be used. For a D/F of 0.23 what recovery and purity
percent n-heptane, and 35 percent n-octane is fed on the of n-butane is possible?

proper stage to a distillation column equivalent to 25 8.27. For the system of Problem 8.26, determine an

stages.The bottoms must contain 98 percent of the n-oct- optimum of D/F that gives both high purity and high
ane charged in a concentration of 99 percent. Total pres- recovery.

sure is 150 psia. 8.28. An existing distillation column equivalent to

(a) What reflux ratio is required? 12 equilibrium stages is to be used to produce nearly pure

(b) What reflux ratio is required for a 99 percent propane from the following mixture:
recovery of n-octane in a concentration of 99 percent? Ethane 20 percent
(c) Plot temperature and composition as a function Propylene 30 percent
of stage number for the reflux ratio of (a). Propane 50 percent
8.22. A mixture of n-butane, n-pentane, and n-hex-
The column operates at 250 psia. and feed nozzles are
ane is to be fractionated in a column equivalent to 30
located at the fifth and seventh stages from the top. Maxi-
equilibrium stages at 150 psia. Only 0.5 percent of the feed
hexane is to appear in the distillate, and less than 0.5
mum reflux ratio is 5.0. What percentage recovery of 99.0
percent pure propane is possible?
percent of the n-pentane is lost in the bottoms. The feed
8.29. A mixture of 30 mole percent n-pentane, 50
contains 10 percent C4Hi , 40 percent C5H12, 50 percent
mole percent n-hexane, and 20 mole percent n-heptane is
C 6 H 14 .

fed at the center of a column equivalent to five equilibrium

(a) What reflux ratio is required for the separation?
stages operating at atmospheric pressure. At a D/F of 0.25,
(b) Plot temperature and composition as a function
what reflux ratio (if any) will yield a distillate of com-
of stage number.
position 0.995 mole fraction n-pentane?
8.23. Essentially pure propylene (99.0 percent) is to
8.30. It is desired to recover 99 percent of the ethane
be produced from the following mixture:
from the following mixture in of 0.995 mole
a distillate
Propylene 45 percent
fraction ethane with the following feed:
Propane 25 percent
i-Butane 10 percent Ethane 55 percent
n-Butane 15 percent Propylene 35 percent
n-Pentane 5 percent Propane 10 percent
156 STAGE OPERATIONS

Determine the reflux ratio and the number of stages in fractions, each of which is 99.0 percent pure:
a column operating 250 psia that will accomplish the
at
Isobutane 25 percent
desired separation. Assume that the feed is at the center
n-butane 25 percent
stage. (There may be more than one correct answer to this
n-pentane 25 percent
problem.) If time permits, make a parametric study of
n-hexane 25 percent
reflux ratio and number of stages, such as might be used in

an economic evaluation of the proposed design. Assume the columns operate at atmospheric pressure, and
8.31. Design a system of distillation columns to sep- use reflux ratios of 4.0 or less. Make any other reasonable
arate all the components of the following mixture into assumptions that are required. Specify the number of stages
required in each column.
 C) a 157
_3
c
g- .~
Q. a>
3
-I Q. 4_t 4-» .Q *—
c c lm i_
1) V o .
0)
n .a
-a
n o CO IB

am
CJ c U
13 o
< < 2

3 Q.
Q.
o Ato ~CT o> ro
> u
>
</>
o a> aj re -
to 01
-Q O) 5
< 0) o c 3 ^ .y - .y
CO
*j
3 -C
.1)

>
n a It a 3=
i_ CD i— CD
V) O 3 3 qj CT
CO LL CO X -1

Z
g o- v
H Q
<; I 2 -2
2rX
3
(4
^_ O LU —
CO O c
W 00 D
<
- JO
u u
3 0) *- C
o .E
3 C *— c ."3
3 H-
.5" o o ID c a> > >
C n c ra n >
o
C-3

S2 X
c
El, 0) CO
m a a $ n CD
.52*
W ._ 4^ ro W 4-* i-
CD
l_ o
F C o
< i
3 X a> c o _c r
CT LU
*-<
3 O o cd
u >
CO
2 a m
Q.

- O CO n n X x ^ J
00 CC ^ CO LU LU S k
a
u
z a^
w ° 3*O = o
a
o r a 4««
D" >• to" >
2 _ » -o £ - S >
z ra CJ^U0;.—
O .h
a «aj *:cu
CD nj
^_
*•*. — CT

Q o a £ .2
o Ec:
£ « .2 S.S*
E E 2 C A3
z o o
CD T3

< o c c S -o *=
3
z o .!2

O 00 Q 2
-
<
- o o
c </> CO E E </)

z 0)

O
tv
o .Q XI
a>
o o
Q. C E F i_ i_
F E
O o
H £ c c
a>
-Q
^3 XJ
.Q 3 -5
o .2
< o e O 3 3 O °
PL,
:2£ 0)
Q. ^ CO
4->

CO
X !5
> 5

>
Q.
f E 5=x O u
F m
O X » n s on
a
+* 0)

"S c.
3 cu
"D oO e- > s-
a)

2 E 3 O 3 03 -*-1 fp _ O
Q. 0) aS SS <l)

c
o 2
0)
o CD u CD 9- S I~11 I S 5)
wi
CO
^-
-
*-

l"|| |"
""
C3 a n> Q. ro
X a.
(0
X ia M— CJ
£ o u
o to vi u
F
o a E a c
Cl
CD
SZ 3= -o
CI
a) a!
CD
a a 5 § |
CD "
CJ ^ CJ Li. Q > -I - U Q < <

03 .Q Q Uj U. U, i -i E
158

a
5
—
o < E
O
_
? CD —
0)
X 05
c j;
n ^ 4OJ
X n
-

Q. (U

5 E c
-
TO
X C3
CO
XQ. CD
en

< IK X,
CD
IS
~j tz
Hi. QJ a
X i_
li — C
Efl

CD
Xu
CO
t3 T3
D r
O g
E c
1.1 in o o
'->
S
™ § 0) ' »--
CD
CD
o T3
I

<- — CD o CO
1= X o c y.
O LU Q-
CO
0)
c CJ
CO
Z C E
o t/5
n •-~
?
T3 o
CO CD en
< c X _
a: o o <Z
X
— .!§ 5
o c
o
CD
j~.

w CD a
o .P.
D X
+=
CJ
-j
< CT LU
H
en
~ X ^ X
O
_
CJ q. cn
c E .*=

— CD CD <
as c fi J9

H i» O «"

<
-J
T3
C
":
in
e
to
U w o c
CD
a
t; cd
ir
z =— Q. ~J
CO 8 CJ

^ CO > lT i C tn
X X CD

c n c
o 'J! -
-

- «
z CJ
CD u
- C/3
_f-

—(J
Q u
".:

Q.
in CD
. X
z E O CD
C E CD §
< -
a. CJ
CD
a £
O O -
z E E
r
*^
Q. CD

g Q. C
X X m
IS
J3 O
- - CD
n x x o g o -5 o o o -2 o " ~ *-o CJ § CO
X -S
5
< 3 3 3 (/) *— t/» k_ tn v»- wi i-
o o o CD
i - > ^
O m X X CD

XO X
GO QQ CD
Z o T o *-'
T) ? 5 m
- -o Si
CO

c o -Q
,-
c
OJ O
<
Oh
0)
c
o
C
o .E -J
c c
O ta
E* o
C C a a. ^ x
<D
"° "° E E
a.
E
•— CD 005J
~ ~ ~ *-
5 O -x
o
= oc .
o
o a>
c
o
o CD
c _
o t. C
w
CD
CD
S
.
-Q
E ™
CD ^ -" CJ _a> co O CD
*j o " *- x
>
c
o
a I E
(D r H (C 3
CD CJ

^C
co «j
E CO u_ 81
X E <" ^1 CD
X
X3 _ "D -a
O c
"° ^ t
c <u co o " <o O <o 33l<j CD
0)
— w *- 3
•__
.2. CU OJ
E
Xn
"-
en o ~o o T3 -a -o w- = 3 o c
2 ^ T3 r
c o
CD

a u 5 - C 21^- -e
w o "> co ca aj .9
* 3 4^ > V
CO
=o S^ S- +^o .E— ^o 2
co Cl
x o —
co %-
- x°- -Ca *- rr-
ai a. o
c
<-; ** *-
* a u .E -^ cl .E E °> CO — tr co cd
a> S S S 2 £ 2 &
< X I X I -j < LL QC U. WZ a,
CO CO
CO C
£2 I O
•§• S *2 £
— o co S F
t .Q..9-

Q
i
•—
cj
Co
o
CO
~ -C5
c/>

X
- 3 3
o a & &Co >• N a <w o" Cl Co CO

Molecular and Turbulent Transport
Nearly all unit operations involve the transport of
mass, heat, or momentum. Such transport may
occur within one phase or between phases. In
many cases, it is necessary to know the rate of
transport of mass, heat, or momentum in order to
design or analyze industrial equipment for unit
operations. For example, in the stage operations
covered in the preceding chapters the rate of mass
transfer between phases was assumed to be great
enough to permit the rapid establishment of equi-
librium. In reality, however, the rate of mass trans-
fer is an important factor in determining the stage
efficiency, which must be known to determine the
actual number of stages necessary for a given sepa-
ration.
A knowledge of the principles of momentum
transport is a prerequisite to the proper design of
the pumps and piping systems so essential to the
chemical process industry. Similarly, the design of
industrial heat exchangers rests on an understand-
ing of the fundamental transport mechanism for
thermal energy.
The fundamental mechanisms of mass, heat,
and momentum transport are closely related. The
rate equations for the three transport systems are
of the same form, and in certain simple physical
situations the mechanisms of transport are iden-
tical. Molecular transport depends upon the
motion of individual molecules. Turbulent trans-
159
Part Two
port results from the motion of large groups or
clusters of molecules.
The following six chapters will consider the
fundametal mechanisms of molecular and turbu-
lent mass, heat, and momentum transport. The
three transport processes are closely related, and
they will be developed in a general way wherever
possible.
The terms transport and transfer are used
interchangeably in many cases. If a distinction is
made, transport usually is used when referring to
the fundamental mechanism within a single phase,
whereas transfer
refers to the overall process.
After the transport mechanisms have been
developed, a number of applications will be con-
sidered. Methods of evaluating and correlating
transport properties will be discussed. Application
of the transport principles to interphase transfer
will be developed. Integration of the rate equation
will be demonstrated where concentration varies
along the length of the contacting path. Finally,
after the fundamentals of transport systems have
been thoroughly developed, applications to the
design of industrial equipment will be considered
in Part Three of this book.
A table of the notation used in Part Two
together with dimensions and typical units, is given
after Chapter 14.
 Nine

Molecular Transport Mechanism

Molecular transport of mass, heat, and momentum solids, because of substantial differences in mo-
occurs in gases, liquids, and solids. It is the lecular structure of these three physical states.
basic physical mechanism underlying many impor- Gases contain relatively few molecules per unit
tant industrial unit operations. In general, mo- volume. Each molecule has few near neighbors with
lecular transport is due to the interaction of which to interact, and intermolecular forces are
individual molecules, although the specific mecha- relatively weak; the molecules of a gas therefore
nism depends upon the physical state of the trans- move relatively freely for considerable distances
portmedium and upon the entity being transported. before colliding with other molecules.
Thus, whereas the mechanisms are similar for heat Liquids, on the other hand, contain a much
and mass transfer in a gas, they are quite different higher concentration of molecules per unit volume,
in a solid. so that each molecule has several near neighbors
An understanding of the molecular transport with which to interact, and intermolecular forces
mechanism may be obtained from a study of the are stronger.As a result, molecular motion is much
kinetic theory of gases and liquids or from a con- more restricted in a liquid. Migration of molecules
sideration of solid-state physics. These fundamental from one region to another occurs in a liquid, but
subjects are extremely complex; neither is con- at a much lower rate than in gases. Liquid molecules
sidered in detail here. Instead, a simplified physical vibrate back and forth, colliding frequently with
model of a fluid is used to derive an expression for near neighbors.
the rate of transport as a function of a concentra- Molecules in solids are even more close-packed
tion gradient. The resulting general transport equa- than in liquids; and molecular motion is even more

tion is then extended to real gases, liquids,and restricted. In many solids, the intermolecular forces
solids. A brief discussion of the transport mech- are strong enough to hold the molecules in a fixed
anism in included to indicate
each physical state is array called a crystal lattice. Obviously, molecular
how real materials differ from the simple physical migration is greatly restricted, although molecules
model. Finally, consideration is given to the corre- still vibrate around their fixed positions in the
lation of experimental data for prediction of the crystal lattice.
transport properties of gases, liquids, and solids. Asa cooled (that is, as energy
given material is

is removed), it may
from the gaseous state,
pass
MOLECULAR TRANSPORT with chaotic, intense molecular motion, to the
liquid state, with restricted motion and a more
Each molecule of a system has a certain quantity of
mass, thermal energy, and momentum associated ordered structure, and then to the solid state, with
little molecular motion and a highly ordered struc-
with it. When a difference in concentration exists
ture. Obviously, these great differences in physical
for any of these properties from one region of the
system to another, interactions among molecules state yield differences in the mechanism of mo-
lecular transport.
may produce a net transport of the mass, thermal
energy, or momentum. The actual mechanism of When a system contains more than one chemi-
transport differs greatly among gases, liquids, and cal species, mass transport may occur. If the con-

161
 162 MOLECULAR AND TURBULENT TRANSPORT
centration of one species is greater in one region of
the system than in another region, molecular
motion leads to a net transport of the species from
the region of higher concentration to the region of
lower concentration. For example, if a bottle of
perfume is opened in the middle of a room of still
air, some perfume vaporizes and the perfume mol-
ecules are then transported by molecular motion
from the region of higher perfume concentration
near the bottle to a region of lower concentration
farther away. Such molecular transport of mass is
usually referred to as molecular diffusion. Because
molecular mass transport requires actual migration
of molecules, the rate of mass transport is much
greater in gases than in liquids; diffusion in solids is
extremely slow for the same difference in concen-
tration.
The molecular transport of heat occurs from
region of higher temperature to a region of lower
temperature. Because temperature is a measure of
the internal energy of a molecule, a higher tem-
perature indicates a greater internal energy. Thus,
heat is internal energy being transported from
higher-energy molecules to lower-energy molecules.
Because heat is energy in transit, some scientists
and engineers prefer to use the term energy trans-
port instead of heat transport. Not all of the
energy of a molecule can be transported across the
available temperature difference. The portion of
the internal energy that can be transferred across a
temperature difference is sometimes referred to as
thermal energy. The molecular transport of heat is
called conduction. The transport of heat occurs
not only by molecular migration but even more
effectively by collision of molecules vibrating
around relatively fixed positions. Energy is trans-
mitted by collision of vibrating molecules in liquids
and solids where little or no migration occurs. As a
result, many and solids are excellent con-
liquids
ductors of heat, whereas gases are relatively poor
conductors. A poor conductor is a good insulator,
by definition. In metals, heat is also transported by
electron migration, as discussed in a following
section.
Momentum transport occurs in flowing gases
and liquids. Under extreme forces some solids may
be considered to "flow," but not in the same
manner as liquids and gases. A crystalline solid may
fracture along crystal planes, instead of flowing
uniformly. This discussion is limited to the fluid
mechanics of gases and liquids. In a flowing fluid,
one region may possess a higher velocity than
another. For example, with flow in a pipe, the
velocity at the center can be shown to be a maxi-
mum, while that at the wall is zero. Momentum is
transported from the region of higher flow velocity
to the region of lower velocity by the mechanisms
of migration and collision. In a gas the intermol-
ecular forces are weak, so that momentum trans-
port occurs through migration and subsequent col-
lision of molecules. In a liquid, intermolecular
forces are strong, so that a substantial force must
be applied to induce flow of molecules past each
other. As a result, the rate of momentum transport
is higher in than in gases for the same
liquids
velocity pattern. Molecular momentum transport
occurs in laminar flow, the two terms are some-
times used interchangeably.
The practical applications of mass, heat, and
momentum transport developed long before the
basic mechanisms were understood. The engineers
a who applied the principles of the three types of
transport were not primarily interested in the simi-
larities among the mechanisms of transport. As a
result, three distinct systems of notation and
nomenclature developed. Thus, fluid mechanics or
fluid flow is the traditional name for the unit
operation based upon momentum transport. Heat
transfer and mass transfer are also traditional unit
operations. Although it might be desirable from a
fundamental point of view to use one set of nota-
tion in discussing the three systems, the engineer
using a unified nomenclature would not be under-
stood by the majority of engineers in industry. For
this reason, after the transport-rate equation is
derived using a completely general notation to
show the interrelation, it will then be rewritten in
the traditional terminology of mass, heat, and
momentum transport for application to industrial
problems. In this discussion it is well to remember
that there are not only fundamental similarities in
mechanism, but also fundamental differences.
The General Molecular-Transport Equation
Derivation of the general rate equation for mol-
ecular transport first requires that a physical model
be postulated. The previous discussion indicates
that no single physical model is likely to be
adequate to describe transport in gases, liquids, and
solids, because of differences in structure and
transport mechanism in the three states. In this
section a very simple physical model based upon
molecular migration alone will be used to derive
the general rate equation. Although the resultant
transport equation is strictly applicable only to the
model, it may be extended to real gases, liquids,
and solids. Because the model considers only mol-
ecular migration, ignoring turbulent transport and
 MOLECULAR TRANSPORT MECHANISM 163

other mechanisms, it resembles a gas more closely

than a liquid or solid. Thus, the physical model used
here is referred to as a simple model gas. More
complex models could be used for the derivation,
but for present purposes the simple model is ade-
quate to illustrate the general approach.
In 1860 J. C. Maxwell first derived rate equa-
tions for transport of mass, heat, and momentum.
The following derivation is a simplification of
Maxwell's approach and leads to the same final
equations. The simple model gas is based on an
ultrasimplified kinetic theory of gases:
Figure 9.1. Volume element of the model gas.
1. The gas is made up of molecules, each of
which is a perfect sphere of diameter o.
2. No attractive or repulsive forces exist +j/-direction,and so forth. Attention is focused
between gas molecules. only upon the molecules moving in the +x- or
3. The actual volume of the molecules is —x-direction. The assumption is made that the
negligible compared to the volume between mol- molecules moving the y- and z-directions do
ecules. not interact with the molecules moving in the
4. All collisions between molecules are per- x-direction.
fectly elastic. Individual molecules have certain properties
5. Each molecule is in random motion at a associated with them. These properties are trans-
mean speed the random direction.
(c) in ferred from a region of higher concentration to a
6. Each molecule movesa distance/ between region of low concentration. Examples of such a
collisions with other molecules. The distance / is transferent property would include the thermal
called the mean free path. energies of molecules, the flow momentum of mol-
7. The time required for a molecule to travel ecules, or the chemical identities of molecules in a
a mean free path traveling at the mean speed is the mixture of two components. A small volume of
mean time between collisions 6. That is, 6 = He. model gas possesses a certain concentration of one
8. The number of molecules is large enough of the transferent properties, designated by the
that statistically average values of properties can be Greek letter gamma:
used to describe all the molecules.
T = concentration of the property to be
This is a highly idealized molecular picture of transferred, in dimensions of quantity of
a gas. Molecules of real gases are not spherical, and transferent property per unit volume of
theremay be strong attractive or repulsive forces gas
among them. Furthermore, the molecules will It is initially postulated that a concentration
move at various speeds for various distances gradient exists in the volume element shown in
between collisions. Figure 9.1, where dTldx is the gradient, defined as
the model gas, all molecules move at the
In the increase in concentration with distance in the
same speed c, which represents a mean value of the +x -direction.
various speeds present in a real gas. Similarly, the Three planes are spaced a distance / apart, and
mean free path is a commonly defined quantity each has an area Ay Az. The concentration at plane
equal to the mean of the various distances traveled 2 can be related to the concentration at plane 1 by
by molecules between collisions.
The derivation considers the molecules of gas
occupying the volume shown in Figure 9.1. Since
T, =T 2 + —
dT
dx
(-/) (9.1)

the molecules are in random motion, they move in where dTldx is the gradient and —/ isthe distance
all possible directions. To simplify the situation, from plane 2 to plane 1, since +x is defined to the
the derivation considers that the molecules move in right.
directions parallel with the coordinate axes x, y, Similarly,
and z. Then, one-sixth of the total number of
molecules
one-sixth
move
in the
in the +x-direction at any instant,
—x -direction, one-sixth in the
r, = r. + —
dx
(/) (9.2)
164 MOLECULAR AND TURBULENT TRANSPORT

Plane Plane

(a) (b)

Figure 9.2. Concentration gradients in molecular transport, (a) Concentration

gradient at steady state, (b) Concentration gradient at unsteady state.

The concentration gradient is shown schematically -rv

in Figure 9.2a. The gradient is assumed to be V (2-1) (9.4)
60
constant in the volume element.
At any specific moment, molecules will be
The minus sign arises because the flux is defined as
crossing plane 2 in both the +x- and —x-directions. positive moving in the +x-direction. Thus,
if it is

because the flux from plane 2 to plane 1 is in the

The movement of molecules at plane
total 2 will be
the vector sum of molecular movement in these
— x-direction, the numerical value of this flux must
be negative.
two directions.Molecules moving in the +x-
direction will arrive at plane 2 having had their last
The net flux is positive in the +x-direction.
Since this system is defined invariant in time, and
collision at plane 1. They are accordingly assumed
with a constant and uniform gradient in the con-
to arrive at plane 2, having a concentration of the
centration of the transferent property, we can
transferent property (e.g., momentum) character-
istic of plane 1. These molecules will have traveled
evaluate the sum of the +x-directed flow and the

a distance / in time 0: hence, the molecules travel

— x-directed flow at the same point or, in any
region of uniform flow, such as between planes 1
with a velocity lid. As mentioned earlier, the mo-
lecules moving in the +x-direction are only one- and 2, to evaluate the net flux, which we will write

sixth of the total concentration of molecules in as # after Equation 9.5.

any volume. The flux of molecules moving from
#net(1-2) - #(1-2) + #(2-1) (9.5)
plane 1 to plane 2 is therefore equal to the concen-
tration of molecules mo_ving in that direction (£1^ )
From Equations 9.3, 4, and 5
times their velocity (lid), and is designated by the
Greek letter psi t ^4 + i=m (9.6)
60 \ 60 /

rv On the other hand, ifthe slope of the concen-

#1 1-2) (9.3)
tration gradient in this flat-slab geometry is not
60
constant, as shown in Figure 9.2b, the accumula-
The flux has dimensions of quantity of
#d-2) tion not zero and an unsteady state exists. The
is

transferent property per unit time per unit trans- unsteady state is considered further in Chapter 11.
port area. Because the flux is defined as constant in time and
Similarly, the flux from plane 2 to plane 1 is position, Equation 9.6 is generally applicable and
 MOLECULAR TRANSPORT MECHANISM 165

position subscripts may be dropped. We can gradient of that property. For the model gas, the
express the time between collisions as proportionality constant is g/c. For real gases, the
proportionality constant could be evaluated from
_ /
(9.7) experimental data or calculated from a more
c sophisticated kinetic theory. The result of this

Substituting into Equation 9.6 derivation completely general for any transport
is

depending upon random molecular motion. It will

dV be applied to the specific cases of mass, heat, and
= -lie (9.8)
dx momentum transport in the following sections.

If the gradient is negative, as shown in Figure 9.2a,

the flux will be positive in the +x-direction,
MASS, HEAT, AND MOMENTUM TRANSPORT
because there is a minus sign in Equation 9.8. As was mentioned earlier, the terminology for the
Equation 9.8 is the final result of the deriva- three transport systems is traditionally different.
tion. It simply states that the flux of a transferent The notation is given Table 9.1. Typical in
property is proportional to the concentration dimensional units are also included. Definitions of

Table 9.1 NOTATION FOR MASS. HEAT. AND MOMENTUM TRANSPORT

General Specific Notation (with Dimensions)

Notation
for the Thermal Energy Momentum
Model Gas Mass Transport Transport Transport

(MLy/6)
Flux of transferent property V net Na lA(MlL y L z d) qlA (HlL y L z d) Ty9c
—f
LyL

Concentration of transferent property c a (MlL x L y L z ) c P p T(H/L x L y L z ) vp —Z-

[MLyll

-x'-y'-z

Proportionality constant, the transport

diffusivity 5 = ile 2 (L x 2 6) = klpc 2 v=- (L 2 0)
OL
p (L x 6)

Typical Units

English Metric Systeme

Symbol Definition Dimensions Engineering (cgs) International

Na Rate of mass transport M/e lb mole/hr g mole/sec mol/s

Q Rate of transport of thermal energy Hie Btu/hr cal/sec J/s
Tv Shear stress on the fluid FIL y L z 2 dyne/cm 2 N/m 2
lbf/ft

9c Dimensional constant MLjie^Fj 4.18 x 10 8 1.0 1.0

2
lb ft/hr lb f g cm/sec 2 dyne kg m/Ns 2
A Transport area Ly Lz ft
2 cm 2 m2
Ca Concentration MlL x L y L z lb mole/ft 3 g mole/cm 3 mol/m 3
CP Heat capacity HIMT Btu/lb°F cal/g °C J kg C
T Temperature T °F °C Cor K
V Fluid velocity Lyld ft/hr cm /sec m/s
k Thermal conductivity H/6L y L z (T/L x ) Btu/hr ft 2 (°F/ft) cal/sec cm 2 ( C/cm) J/sm 2 UC mi
P Density MIL x L y L z lb ft 3 g/cm 3 kg/m 3
V Absolute viscosity* ML x /L y L z 6 lb/ft hr g/cm sec kg/m s
5 Generalized transport diffusivity Lxve ft
2 /hr cm 2 sec m 2 /s
'/ Mass diffusivity L x 2 /e 2
ft /hr cm 2 /sec m2 ,'s

a Thermal diffusivity L X V6 ft
2 /hr cm 2 /sec m 2 /s
V Momentum diffusivity Lx m ft
2 /hr cm 2 /sec m2 s

'Chemical engineering literature predominantly uses these dimensions of viscosity incorporating

mass, this unit differs from the standard SI dimensioning of Pa s by the conversion factor for
force to mass, which is numerically unity in SI. but includes dimensions.
166 MOLECULAR AND TURBULENT TRANSPORT

the terms for the three systems are given at the end concentration gradient is constant, and a steady
of Part II, and they will be considered in detail state exists.
below. Equation 9.8 is the general transport equa- From this point the derivation proceeds
tion for one dimension. It may be written in the exactly as in Equation 9.8. The final result may be
terms listed in Table 9.1 for any specific transport written by reference to Equation 9.8 and Table
system. In the following sections, the mechanism 9.1:
of each of the three transport systems will be
considered separately, and the three specific trans-
N
—
A
a
= — lie
6
,
,_ —
dca
dx
(9.10)
port equations will be examined independently.
The form of each equation is identical with that The term N a /A is the mass flux of a component a

used for the general equation. Special attention resulting from molecular transport along the con-
will be given to the physical significance of the centration gradient dc a ldx.
specific fluxes and gradients. Equation 9.10 applies only to the simple
model gas. It is now necessary to extend it to real
Mass Transport gases, liquids, and solids. In all three physical states
If the model gas consists of two different kinds of it is possible to measure the mass flux (N /A) and
a
molecules, mass transport may occur. It will be the concentration gradient (dca /dx), but the mean
assumed that the volume element pictured in free path / and mean speed c were postulated as
Figure 9.1 contains two different species of gas, part of the simple model gas. Thus, Equation 9.10
designated a and b. Mass transport will occur only might be extended to real materials by simply
if the concentration of the gas phase in nonuni- replacing the \lcby a new, empirically determined
form. According to the original model, all mol- quantity 9. Equation 9.10 then becomes
ecules are equal in size and move at equal speeds in
a random direction. If more molecules of gas a are
Na_ _ dc^
(9.11!
present in a given volume, they will tend to migrate A '' dx
by random molecular motion to a neighboring where J the mass diffusivity with typical units
is
region of lower concentration of gas a. If the
of ft
2
m 2 /s).* Equation 9.11 is often taken
/hr (or
difference in concentration is maintained, a steady
as a starting place for mass transport in gases,
flux of gas a will move from the region of higher
liquids, or solids. The mass diffusivity varies with
concentration to the region of lower concentra-
temperature and pressure; Equation 9.1 1 is a state-
tion.
ment of Pick's Law, which is not really a funda-
It would be possible to derive the mass- mental physical law but merely a definition of 9.
transport equation by considering individual mol-
The mass diffusivity for gases, liquids, and solids is
ecules and concentrations in terms of molecules determined experimentally or calculated from
per unit volume. But the engineer is interested in
empirical correlations, as discussed in a later
average properties of groups of molecules, and it is
section.
more convenient to express concentration in terms Molecules of b the gas mixture have the
in
of moles per unit volume, where
same random motion those of gas a. Therefore,
as
if a difference in concentration of gas b exists in a
molecules of gas a
Moles of gas a = given volume, the gas will diffuse from the region
Avogadro's number
of higher concentration to that of lower concentra-
tion. Considering gas b Equation 9.10 may be
Avogadro's number is 6.02 x 10 23
molecules per
written
mole of gas in SI. The engineering notation for
concentration in mass transport is b dc b
- = -ilc (9.12)
dx
ca = concentration of gasa —
lb
J
moles gas a
(9.9)
and Equation 9.1 1 may be written for gas b
ft of total gas

Jb__ d£b
whereas the SI units would be mol/m 3 . / (9.13)
~
The concentration of gas a in the volume A dx
element pictured in Figure 9.1 varies from plane to For dilute real gases or for liquids the diffusion
olane. It is assumed that the concentration varies
linearly from plane 1 to plane 3. That is, the 'Parenthetical units are SI.
 MOLECULAR TRANSPORT MECHANISM 167

coefficients defined by Equations 9.11 and 9.13 where

2
are identical. pa = partial pressure of gas a, atm (or N/m )

The diffusion rates of gases a and b are inter- T= absolute temperature, °R (or K)
related, as shown in the following paragraph. The R= gas constant, ft 3 atm/lb mole °R [or
total concentration c t for the gaseous mixture is (N/m 2 )m 3/mol K]
given by This equation is a common alternate form of Equa-
tion 9.11.
ct =ca +c b (9.14)

Ifthe temperature and total pressure are constant,

Example 9.1. A large tank filled with a
mixture of methane and air is connected to a
the total number of moles per unit volume (c f is )

second tank filled with a different composition of

constant. Differentiation of Equation 9.14 with
respect to the distance x gives
methane and air. Both tanks are at 100 kN/m 2 and
0°C. The connection between the tanks is a tube of
dca dcb 2 mminside diameter and 150 long. Calcu- mm
= [
latethe steady-state rate of transport of methane
dx dx
through the tube when the concentration of
or
methane is 90 mole percent in one tank and 5 mole
dca dc b
(9.15) percent in the other. Assume that transport
dx dx between the tanks is by molecular diffusion. The
mass diffusivity of methane in air at 0°C and
Therefore, a gradient exists in gas a, a gradient
if
100 kN/m 2 is 1.57 x 10~ 5 m 2 /s.
exists in gas b. The gradient of gas b is equal but
opposite in sign to that of gas a. Since a gradient Solution. Equimolar counterdiffusion of
for b exists, there must be a mass flux of b, as gases a and b will occur. The steady-state gradient
stated by Equation 9.13. Combination of Equa- is shown in Figure 9.3. Equation 9.18 will be used:
tions 9.1 1, 9.13, and 9.15 gives

Nh = -Na (9.16) A '

RT\dxl
At x = 0, p a = 0.9 P, where P is the total pressure.
This shows that the rates of diffusion are equal but
At x = 0.15/77, pg = 0.05 P. Integrating Equation
in opposite directions. Equation 9.16 is always true
9.18 between these limits gives
in binary mixtures, regardless of what other
mechanisms for mass transfer may be operating. 9 (0.05 P- 0.9 P)
The rates of transfer of the two species by molec- A RT (0.15-0)
ular transport is always equal but in opposite
directions. When molecular transport is the only Substituting P = 100 kN/m 2 , T= 273.15 K; R =
mechanism, the overall transfer process is called 8.3144 (N/m 2
) m 3
/mol K gives
equimolar counterdiffusion. It gives concentration
Na_ (1.57 x 1Q- 5 )(0.05-0.9)(105 )
profiles that are linear at steady state for the rec-
tangular coordinates used here. In Chapter 10, A (8.3144)(273.15)(0.15)
another transfer process involving two transport = 0.00392 mol/s m2
mechanisms is developed. It leads to curved con- 0" 6
centration profiles. Since A = tt(0.002) 2 /4 = 3. 1 42 x 1 m2
/Va = 1.231 x 10~ 8 mol/s
Equation 9.11 may be written in terms of
partial pressure of gas a. From the perfect-gas law
P a V=n a RT
0.9 P.
na _ pa
°a = (9.17)
V RT t

Substitution of ca from Equation 9.17 into Equa-

05/'
tion 9.1 1 gives
x—,. 0.15 m
'J dp a Concentration gradient
(9.18) Figure 9.3. for
A RT dx Example 9.1.
168 MOLECULAR AND TURBULENT TRANSPORT

It should be pointed out that the quantities yand a generally does not have the same value as > in
A were constant in this illustration. If they had real materials. Equation 9.20 is usually used in a

varied with x or ca they would be included within

,
somewhat modified form utilizing the thermal con-
the integral sign when Equation 9.18 was inte- ductivity k (since a = k/pc p )

grated.
-q = -* dT — (9.21
Heat Transport A dx
If the molecules in one region of a gas possess where k = thermal conductivity, Btu/hr ft
2
(°F/ft)
greater thermal energy than those neighboring in a
[or J/sm 2 (°C/m)].
region, part of the thermal energy is transported by
random molecular motion from the region of T= temperature, °F (or °C)
higher energy to that of lower energy. Thermal The term q/A is the heat flux and dT/dx is the
energy in transport is referred to as heat. The temperature gradient. The properties a, k, p, and
higher the temperature of a system, the greater is Cp must be determined experimentally or predicted
the concentration of thermal energy. Therefore, from empirical correlations, as discussed in a later
heat will be transported from the region of higher section. The thermal conductivity may vary with
temperature to the region of lower temperature. temperature and pressure. If k does not vary with
A constant temperature gradient may be the temperature, then Equation 9.21 is a statement
assumed to exist through the volume element of of Fourier's law. This "law" is really simply a
the model gas pictured in Figure 9.1. The concen-
definition of k. Thermal conductivity varies greatly
tration of thermal energy in each slab of gas is
3 3
among gases, and solids because of the
liquids,
equal to pc P T Btu/ft (or J/m ), where p is the
great difference in the mechanism of heat transport
density, c is the specific heat, and T is the temper-
p in the three states.
ature of the gas. This expression is a relative
measure of the thermal energy content of the gas.
It is based on an arbitrary datum of zero thermal
Example 9.2. Determine the rate of conduc-
energy such as 7"=0 K. T generally appears as a tion through a copper rod 2 mm in diameter and
differential or a difference, so the arbitrary datum 150 mm long. The temperature is 300°C at one
temperature cancels out. end and 20°C at the other. The average thermal
The concentration of thermal energy at plane conductivity is 1.36 x 10 6 J/s m 2 (°C/m). Assume
the concentration at plane 2 is heat is added or removed only at the ends.
1 is (pCpT)-i ,

(pc P T) 2l and at plane 3 is (pc/>7~) 3 The mean free .

Solution. Integrating Equation 9.21 with
path and mean velocity of the molecules of the constant k gives
model gas vary with temperature. In this case an
kA(T2 -T,)
appropriate constant average value of / and c will q =
be assumed. x 2 -Xy
The derivation of the heat-transport rate Now A = tt(0.002) 2 /4 = 3.142 x 10" 6 m2
equation proceeds as did that for Equation 9.8.
Ty =300°Catx =0 1
The result is
r2 =20°Catx 2 = 0.15 m
Q d(pCpT)
(9.19) So
A dx
(1.36 x 10 6 )(3.142 x 10- 6 )(20-300)
The symbols for this expression are defined in <7
=
Table 9.1. (0.15-0)
Equation 9.19 may be applied to real gases, = 7980 J/s
liquids, and solids by replacing the lie derived for
the simple model gas by an empirically determined Momentum Transport
constant a. Because the concepts of momentum flux and
momentum concentration are difficult to visualize
—q = _ a d(pCpT) (9.20) physically, the momentum-transport mechanism in
A dx the model gas will be considered in detail.
where a is the thermal diffusivity with typical units Each molecule in the model gas has a mass m
2
of ft /hr m 2 /s).
Because the mechanism of
(or and a random speed c. Therefore, each molecule
heat transport differs from that for mass transport, has a momentum mc in a random direction. The
 MOLECULAR TRANSPORT MECHANISM 169

Slab 1 Slab 2 Slab 3

(a)

Slab 1 Slab 3

Figure 9.4. Vectorial representation of momentum in the model gas. (a) Momentum
of random molecular motion, (b) Flow momentum superimposed on the random
molecular momentum Wi > v 2 >v 3 ).

momentum of every molecule is in a random direc- slab assumed to be uniform. In reality, the
are
tion, however, so the vectorial sum of all the velocitieswould vary uniformly in thex-direction.
momenta in a given volume is zero, and there is no The velocity v in the -(-/-direction is a macroscopic
excess momentum to be transferred. A gas or bulk velocity. It is equivalent to an actual flow
possesses transferable momentum only if various of the gas in the +/-direction. If the molecules in

regions of the gas possess different concentrations slab 1 are flowing at a velocity v-\ , each molecule
of momentum. Differences in momentum occur if has two components of momentum, mc and mvy
there are differences in molecular velocity. The Similarly, the momentum components in slab 2 are
molecules of the model gas pictured in Figure 9.4a mc and mv 2 , and in slab 3, mc and mv 3 In the -

are in random motion at a speed c. The length of
the arrow on each molecule indicates the magni-
tude of its velocity, and the direction of the arrow
indicates the direction of the velocity. Each of the
molecules in the three slabs shown is moving at the
same speed and consequently possesses a momen-
tum equal in magnitude to those of all other mole-
cules. Since the direction of movement
the sum of the momenta is zero.
all
On the other hand, the model gas in Figure
9Ab is shown to have in addition a flow velocity v
in the -(-/-direction. Each molecule then may be
considered to have two components of
tum, mc random direction and mv in the
in a planes. Thus, there is no random motion so far as
+/-direction. the flow velocity were constant in
If
all three slabs, the momentum concentration of
each slab would be the same, and there would be
no net transport of momentum from slab to slab.
In the gas pictured, the flow velocities in the slabs
differ, as indicated by the length of the arrows
representing the flow momentum. This is an
idealized picture; the flow velocities within any
model gas mc is uniform throughout the volume
element, and the vectorial sum of all mc is zero.
Consequently, it does not contribute to the excess
of momentum necessary for momentum transport.
If the flow velocities of the slabs are not equal, an
excess of momentum can exist in one slab com-
pared to another. The momentum caused by the
is random, bulk flow of the gas in the /-direction will be
transferred between slabs in the x-direction, as
indicated in Figure 9.46.
The model gas is considered to be flowing in a
regular manner in the +/-direction. Average groups
momen- of molecules flow parallel with the xy and yz
gas flow is concerned. Of course, individual mol-
ecules still move randomly in any direction, but
the average movement is uniformly in the flow
direction. This uniform macroscopic movement is
called laminar flow. A more random macroscopic
flow is referred to as turbulent flow; it will be
discussed in Chapter 12.
The velocity gradient between the planes of
170 MOLECULAR AND TURBULENT TRANSPORT

gas shown inFigure 9.4£> is assumed to be constant port corresponding to Equation 9.8 is

Each slab ofgas has a different momentum.

1 A(mv) _ d{pv)
Individual molecules traveling from one slab to net = — t/c
5
.

(9.25)
another by random motion in the x-di recti on carry A Ad dx
with them their y-flow momentum. This effec- The left side of Equation 9.25 has dimensions of
tively results in the transport of flow momentum force per unit area, because force can be defined as
from one slab to another. the time rate of increase of momentum. This idea
The concentration of momentum for any is further developed in the following paragraphs.

volume of gas is equal to the total momentum It is now of interest to examine further the
possessed by the gas divided by the volume of the mechanics of momentum transport in order to
gas. If there are n molecules in slab 1, the momen- define the momentum flux in conventional terms.
tum concentration is nmv^ I Ay Az I. But nm is The momentum flux cannot be directly measured,
equal to the total mass of gas in the volume Ay but it is related to the force or the shear stress
Az I. acting on the fluid phase.
The density of the gas is therefore In the model gas, the exchange of molecules
and momentum between slabs1 and 2 will result in

=
m nm accelerative forces between the slabs, as shown in
P
V Ay Azl Figure 9.5. It is assumed that v-^ is greater than v 2 .

The molecules leaving slab 1 that arrive in slab 2

where m
= the total mass = nm, and the momen-
must be decelerated so that their momentum may
tum concentration in slab 1 becomes pi^
be distributed throughout slab 2. This deceleration
Similarly, the momentum concentrations in the
exerts a force upon slab 2, and the origin of the
other slabs are pv 2 and pv3 .
force is slab 1. The force acts in the direction of
The momentum flux is equal to the rate of
flow, the y-direction. This force can be represented
momentum transport divided by the transport
by the force vector F y which is the force exerted
-\ ,
area.
on slab 2 by slab 1 at the plane between the slabs.
A general expression for momentum flux is
Meanwhile, the molecules arriving in slab 1 from
slab 2 must be accelerated from v 2 to v-\ This
Momentum flux =
1 d(mv) /n
(9.22)
_ l
creates a force —F y2 acting in the —/-direction
.

A dd from slab 2 on slab 1 at the boundary plane. (The

where numbers of molecules moving between planes in
the two directions are equal, sofK = —Fy2 .) ,

A = transport area
The negative sign appears with — F y2 because
m = total mass being accelerated the force acts in the —/-direction, and by conven-
v = velocity
tion all forces are defined as positive in the +y-
= time
direction. The two forces acting on the fluid at
plane 2 are numerically equal, but oposite in sign.
At steady state, the momentum flux is constant.
Further, in the model gas v is constant for the
molecules originating in any given slab. Therefore,

1 A(mv) v Am
Momentum flux = (9.23)
Ad A Ad

For the model gas, A = Ay Az, A8 = 6 and Am =

Ip Ay Azl, since only one-sixth of the molecules of
a volume element move in the +x-direction. Equa-
tion 9.23 becomes

J_
A(mv) (pv) Ay Az I _ (pv)l
(9.24)
A A0 60 Ay Az 60
Direction of momentum flux
From this point, the derivation follows that
of Equation 9.8. The result for momentum trans- Figure 9.5. Accelerative forces in momentum transfer.
 MOLECULAR TRANSPORT MECHANISM 171

This is true at any plane in the fluid. For example, net momentum flux across that plane. Therefore,
at plane B in Figure 9.5, there are two opposing the shear stress the +y-direction (ry g c may be
in )

forces Fy3 and — Fy4 In Chapter 10, the force

. substituted for momentum
flux in the +x -direction
acting on a fluid plane will be evaluated using a in any of the equations used in this derivation. For
force balance around a volume of gas. The forces example, Equation 9.25 becomes
acting on a plane are difficult to measure directly,
d{pv)
and the force balance permits their evaluation from Ty9c =
= -He (9.29)
other forces of the system. A force balance on a dx
volume element must consistently utilize either the For real gases and liquids the term §/c
externa/ or the internal forces acting on a volume derived for the simple model gas is replaced by an
element. For slab 2 the external forces are Fy and -|
empirical factor v (the Greek letter nu), called the
—Fy 4- The internal forces are — FY 2 and F y3 The .
momentum diffusivity. The momentum diffusivity
force balance states that the sum of the external (v) equal to \lc for the model gas, listed in
is
forces is zero Table 9.1 For real gases and liquids the momentum
.

Fy +(-/> 4 )=0 diffusivity is defined by

\

or that the sum of the internal forces is zero,

- V d(pv)
Tv9c = —J— (9.30)
-Fy2 +Fy3 =0 dx
The momentum diffusivity is more commonly
These equations assume no other forces acting
referred to as the kinematic viscosity. The absolute
upon the slab.
viscosity is defined by
The force acting between two surfaces, such
as that at planed of Figure 9.5, is called a shearing
dv
Ty gc = ~M (9.31;
force, because it tends to deform the fluid. dx
The force acting on a slab of gas accelerates
where n = vp = absolute viscosity, lb/ft hr or kg/m
the mass of gas the slab. For this case, the laws
in
sec. The absolute viscosity may vary with tempera-
of motion state that the force required to acceler-
ture, pressure, and velocity gradient. If p. is con-
ate a mass m
is simply equal to the mass times the
stant with dvldx, Equation 9.31 is a statement of
acceleration (Fgc = ma). More precisely, the shear-
Newton's law. This "law" is merely a definition of
ing force that accelerates the mass is equal to the
the absolute viscosity based upon a model assum-
time-rate of change of momentum in the slab
ing that viscosity does not vary with velocity
d(mv) gradient. Although this model is adequate for real
Fy9c - (9.26)
gases, many liquids have viscosities that vary with
dd
dvldx. They are referred to as non-Newtonian
where m the mass that is accelerated. The dimen-
is
liquids, because they do not follow Equation 9.31
sional conversion factor igc is included because
)

customarily force is expressed in dimensions of

pounds force (\bf) and mass in dimensions of Example 9.3. Two vertical parallel metal
pounds mass (lb). These two systems of dimensions plates are spaced 10 mm apart. The left-hand plate
is moving at a velocity of 5 m/sec and the right-
are related by g c as discussed in Appendix A.
,

The shearing force per unit of shear area is hand place is stationary. The space between the
plates is filled with a heavy oil with a Newtonian
referred to as the shear stress, designated by the
2
viscosity of 0.1 N s/m
Greek letter tau. In the plane between the slabs, .

the shear stress (a) Calculate the force per unit area required
is
to maintain the movement of the left-hand plate.
F F.. (b) Calculate the force per unit area exerted
(9.27)
y
A Ay Az by the oilon the right-hand plate.
(c) Determine the momentum flux at the
Combining Equations 9.26 and 9.27 gives surface of each plate.

1 d(mv) Solution. At steady state, momentum is

Ty g c = (9.28) transferred from the left plate through the oil to
A dO
the right plate; the velocity of the oil varies linearly
Comparison of Equations 9.22 and 9.28 shows that with distance between the two plates (Figure 9.6).
the shear stress acting on a plane is identical to the The velocity of the oil at the surface of the moving
1 72 MOLECULAR AND TURBULENT TRANSPORT

(b) Because the velocity gradient is linear and

the viscosity constant, the shear stress in the oil is
constant from one plate to the other. Thus, the
force exerted by the oil on a unit area of the
right-hand plate is simply the shear stress of
the same value as calculated in part (a).
Momentum (c) The momentum flux at any point is equal
flux
to the shear stress. At the left plate from part (a)

Momentum flux = r v g c = 50 —m —
kg m/s
~
s

Note that the dimensions are consistent with the

'
definition of momentum flux: momentum per unit
J0,
area and unit time. Because steady state exists, the
momentum flux is constant so that the flux at the
right plate is the same as at the left. Note also that
gc could be dropped from Equation 9.31 if the
Figure 9.6. Momentum transport between two parallel
viscosity had been expressed in Ns/m 2 In Systeme .

plates in Example 9.3.

International^ is numerically unity.

plate is equal to the velocity of that plate. The TRANSPORT IN GASES

velocity of the oil at the surface of the stationary
The general term transport properties includes
plate is zero.
the mass diffusivity, thermal conductivity, and
(a) The force per unit area required to main-
absolute viscosity of gases, liquids, and solids,
tain velocity is simply the shear stress acting at the
Much work has been done on the theoretical and
interface between the plate and the oil. The shear
experimental evaluation of transport properties.
stress is evaluated using Equation 9.31
This section will summarize a few of the more
ju = 0.1N s/m 2 =0.1 kg/ms useful theoretical and empirical methods for the
evaluation of transport properties. Although many
Integration of Equation 9.31 at constant gradient
of the theoretical derivations involve very complex
and viscosity (and hence constant flux and shear
physical and mathematical concepts, their applica-
stress) yields
tion usually depends upon the experimental evalua-
W2 - ^i tion of certain terms in the final equations. Con-
'yl
<X 2 -Xt) sequently, even the theoretical equations require
empirical verification before they may be used for
From F gure 9.6
the prediction of properties of real materials.
v :
= 5 m/s atx! =0 Nevertheless, the theoretical expressions are invalu-
able for interrelating data for transport properties,
v2 = 0atx 2 = 0.01 m
For example, it is possible to predict mass diffusivi-
Then ties and thermal conductivities of certain real gases

(0.1X0-5) from viscosity data for the gases.

Although engineers can utilize the methods
0.01
for predicting transport properties without a com-

. c n 9 m/s
k plete understanding of the theoretical background,

m2 s they will better understand the transport mecha-

nism on which they base their numerous applica-
Now tions if they have some acquaintance with the
theories of molecular structure. For this reason, a
gc = 1 kg m/s 2 N
brief discussion of the molecular theory is included
Therefore, for each physical state.

tv = 50 N/m 2
The Model Gas
This is the force per unit area required to maintain The simple model gas was used to derive the
the motion of the left plate. general form of the transport equation, but the
 MOLECULAR TRANSPORT MECHANISM 173

model was too simple to conform to reality. Never- assigns a distribution of speeds to the molecules.
theless, a consideration of the transport properties One such speed distribution is the Maxwell-
of the model gas is included here for comparison Boltzmann distribution (9)
with the real gases discussed in the following
section. =
M \
3/2
Mc /2R T
:
7
f(c) Aire (9.34)
For the simple model gas, the transport 2nRT
diffusivities were found to be equal to g/c; that is
where f(c)dc is the fraction of all molecules that
9=ot=p = $lc (9.32) are traveling at speeds between c and c + dc. Equa-
tion 9.34 is plotted in Figure 9.7 for nitrogen at
To evaluate the transport diffusivities, expres-
sions for the mean speed mean free path
c and the
two temperatures. Note that the fraction of mol-
ecules at higher speeds increases as the temperature
must be developed. Such expressions come from a
rises. This is reasonable, since a higher temperature
consideration of the kinetic theory of gases. This
theory considers physical models of increasing
indicates greater molecular energy. The mean speed
is obtained by integrating over all possible values of
complexity that describe real gases with various
c, having weighted each c according to the fraction
degrees of success. The values of / and c depend
f{c)dc that possess that speed
upon the model chosen. For the model used earlier
to derive Equation 9.32, where it was assumed that
all molecules travel at a single speed c the assump- , c = | cf[c)dc
•o
tion of the perfect-gas law and the kinetic-theory
definition of temperature lead to (9)
=
/ M \
3/2
,
3 Mc /2RT c*dc
2
„3
4tt

e =^
M\
3/?r\
1
1/2
(9.33)
%RT^
-(—
\0-rrRT!
\2irRT
12
-'

(9.35)
where
c = molecular speed, m/s where the symbols are defined as for Equation
R = gas constant, 8.314 J/mol K 9.33. The Maxwell-Boltzmann mean speed given in
7~ = absolute temperature, K
E quat ion 9.35 differs slightly in the constant
M = molecular weight, kg/mol
(x/8/7rcompared to yjl>) from Equation 9.33,
The molecules of a gas obviously do not all which assumed all molecules traveled at the same
same speed.
travel at the A more realistic approach speed. Either of these two expressions for c might

Figure 9.7. Maxwell-Boltzmann distribution for nitrogen.

174 MOLECULAR AND TURBULENT TRANSPORT

be used in Equation 9.32 to evaluate the transport C. If instead the Maxwell-Boltzmann speed distribu-
diffusivities. tion is assumed, the result is (10)
The mean free path may be determined from RT
consideration of the collision rate of a molecule in /
= (9.38)
the simple model gas. A molecule moving in the \flAPTTO
+x -direction may collide with molecules moving in The notation for Equations 9.37 and 9.38 is
the directions -x, ±y, and ±z. (No collision with
23
+x-molecules is possible because all +x-molecules A = Avogadro's number, 6.02 x 10
move at the same speed in the same direction.) The molecules/mol
relative velocity of a +x-molecule compared to a P = pressure, N/m 2
—x-molecule is 2c, because the two molecules are T = absolute temperature, K
each moving at speed c but in opposite directions. o = rigid-sphere molecular diameter, m
As a +x-molecule travels it will collide with any R = gas constant, 8.314 (N/M 2 m 3/mol K )

molecule whose center lies within a distance a of / = mean free path, m

its center. Thus, in a time interval Af, a +x-

Equations 9.36 and 9.38 differ by a small numeri-
molecule of diameter o can be considered to sweep cal factor.
2
out a cylindrical volume that is 7ra in cross-
The transport diffusivities may now be evalu-
sectional area and 2c t long. Any —x-molecule in
ated using the two sets of / and c. Combining
this volume collides with the +x-molecule. If there
Equations 9.32, 9.33, and 9.37 gives an expression
are n molecules per unit volume, only 1/6 are for the simple model gas with molecules all travel-
—x-directed, so that the +x-molecule will collide ing at speed c.
with (5/7) (7ra 2 (2cAf) = 3 nno c Af —x-molecules
2
5.74 x 10- 24 T 3/2
)

in the time interval Af. The collision rate with cy

=*=v = n 2 „ V2 (9-39)
—x-molecules is then simply 3 ttpo 2 c collisions per
1/2
Po M
unit time. Collisions with ±y- and ±z-molecules also For model gas whose rigid-sphere molecules
a
occur. Since these molecules move perpendicular travel at speeds described by the Maxwell-
to +x-molecules, their relative velocity is only>/2~c. Boltzmann distribution
For example, a -ty-molecule of speed c is moving at
4.81 x 10" 24 r 3/2
a velocity of y/2c relative to the +x-molecule mov-
ing perpendicular to it at speed c. Thus, by the Po 2 M V2
same reasoning as used for the —x-molecule, the
2 The notation for the last two equations is
collision rate for -(-/-molecules is 5-^/2^/7 o c. The
total collision rate considering all directions is T= absolute temperature, K
then 3 ( +3V2I (a7tto
2
c). P= pressure, N/m 2
The mean free path
defined as the may be a = rigid-sphere molecular diameter, m
total distance traveled in time interval Af divided M = molecular weight, kg/mol
by the total number of collisions occurring in that j
c = mass diffusivity, m 2/s
time interval a. = thermal diffusivity, m 2/s
distance traveled v = momentum diffusivity, m 2 /s
/=
number of collisions Equations 9.39 and 9.40 can be used for
evaluating transport properties, providing the mol-
cAf
ecular diameter o known. Real gas molecules are
is

Q + $y/2)(niro 2 c)At not rigid spheres, as was assumed in the simple

1
model gas. Thus, a has no real physical meaning
2 (9.36) and it cannot be measured directly. Instead, it
( 3 + §\/2)(mra )
must be determined from some measurable
From
the perfect-gas law,jhe number of mol- property, such as the viscosity. Even in the most
ecules per unitvolume \sn = (An/V) = (AP/RT) so rigorous kinetic theory, the viscosity is often used
that Equation 9.36 bee mes to determine equivalent "molecular diameters" of

=
RT real gases. Tabulated values of a should be used
/
2 (9.37) only with equations based upon the model from
{\+\y/2)APiro
which they were originally evaluated. Thus, the a's
Equation 9.36 is based upon a model that in Appendix D-6 cannot be used with the simple
assumes that all molecules travel at the same speed model gas equations.
 MOLECULAR TRANSPORT MECHANISM 175

Equation 9.40 may be rewritten to solve since examination of Equation 9.41 shows that p
7~ 1/2
directly for the transport properties. It is more varies as .

customary to use the thermal conductivity {k) 1/2

instead of the thermal diffusivity (a), and the M 673K = (2.096 x10- 5
)(^)
absolute viscosity (p) instead of the momentum
diffusivity (v). Since p = vp, and p = PM/RTby the = 3.28x 10~ 5
kg/ms
perfect-gas law, Equation 9.40 becomes
This value is considerably lower than the exper-
10~ 25 > imental value of 4.11 x 10 -5 kg/ms because of
5.79 x
= (9.41: inadequacies in the simple model upon which
M
Equation 9.41 is based.
The numerical constant includes R. The thermal (c) / is calculated from Equation 9.40 using
conductivity may be determined by noting that part (a). Atmospheric pressure is 1.013 x10 5
k = apcp. Furthermore, p = PMIRT and Cp = N/m 2
5R/2M for a monatomic perfect gas. Equation (4.81 x 10~ 24 )(273) 3/2
i/ =
9.41 then becomes
(1.013x 10 )(3.02x 10- 10
5
)
2
(39.9x 10~ 3 )
1/2

-23
1.203 x 10 y/T/M = 1.17 x 10~ 5 m 2
/s
k = (9.42)
This may be compared with the experimental value
-5
The numerical constant includes the value for R.
of 1.62 x 10 m 2 /s.
From this expression, k is obtained in units of (d) k is calculated from Equation 9.42
J/sm 2 (°C/m). 1.203 x 10~ 23 V273/39.9 x 10~ 3
Although these expressions apply strictly only k =
(3.02 x 10- 10 2
to the model gas, they may be used to obtain )

approximate estimates for real gases, as shown in = 1.09 x 10- 2 J/sm 2 (°C/m)
the following example.
This value is about two-thirds of the experimental
Example 9.4. The viscosity of argon at 0°C value of 1.63 x 10~ 2 .

and atmospheric pressure has been measured as

2.096 x 10 -5 kg/m s. Assuming that argon is a The prediction of transport properties based
simple model gas: upon the simple model gas gives values within 33
(a) Calculate the equivalent molecular percent of the experimental values. This agreement
diameter. is surprisingly good, considering the great simplifica-

(b) Predict the viscosity of argon at 400°C
and atmospheric pressure.
(c) Predict the mass diffusivity of argon at
0°C and atmospheric pressure.
(d) Predict the thermal conductivity of argon
tions in the model, but it is not sufficiently
accurate for most engineering calculations. More
rigorous kinetic theories must be used to predict
transport properties, as discussed in the following
section on real gases.

at 0°C and atmospheric pressure.

Real Gases
Solution
Real gases differ from the simple model gas in
(a) The molecular diameter is obtained by
many ways. The molecules of a real gas are rarely
rearranging Equation 9.41:
spherical; they are certainly not rigid. The surface
'5.79 x 10- 25 v/MT of a molecule is not clearly defined. A molecule
a = consists of one or more atoms bound together by
interactions among the electrons of the various
For argon, M = 39.9 x 10" 3 kg/mol, 7"= 273 K: atoms. Each atom consists of a very small, dense
nucleus surrounded by a cloud of electrons revolv-
(5.79 x 10- 25 )V39.9x 10~ 3 x 273
a = ing around the nucleus. Thus, the surface of the
-5
2.096 x 10 molecule consists of an electron cloud. Further-
= 3.02x 10- 10 more, atoms within a gas molecule do not neces-
m
sarily occupy fixed positions in the molecule; they
(b) Equation 9.41 may be solved at 7" = may oscillate in relation to each other, thereby
673 K using a calculated in part (a). Alternatively, causing oscillations in the shape of the molecule.
1 "6 MOLECULAR AND TURBULENT TRANSPORT

Gaseous molecules possess a wide range of a = distance of closest approach of two mol-
speeds. Although the Maxwell— Boltzmann distri- ecules that collide with zero initial relative
bution is a fair first approximation of the distri- kinetic energy
bution of speeds, it is not exact. A more rigorous r= distance between the molecules
approach uses the Boltzmann integro-differential
This function is plotted for a typical case in
equation, which describes the distribution of mol-
Figure 9.8. It is seen from Equation 9.43 that the
ecules as a function of position, velocity, and time.
intermolecular force is equal to the slope of the
It takes into account all forces acting upon the
curve shown, with the sign changed. Thus, at large
molecule. Application of this rigorous equation to
distances, the slope is positive and the force is
molecular transport requires the use of statistical
negative, so that there is an attraction. Inspection
mechanics and advanced mathematics. It is
of Equation 9.44 leads to the same conclusion.
assumed that the molecular size is small compared
For between molecules, r is
large separations
to the distance between molecules, so that the
much greater than a, so the second term in Equa-
theory aoes not apply to dense gases or liquids,
tion 9.44 predominates and indicates an attractive
where the molecules are closer together.
force between molecules. When the molecules are
An introduction to this complex subject is
close together, r is less than a, the first term in
given in Reference 9, while Reference 7 presents a
Equation 9.44 is dominant, and the molecules
more thorough treatment. One ultimate result of
repel each other. Many other potential functions
this rigorous kinetic theory of gases
the expres-
is
have been suggested, but the Lennard— Jones 6-12
sion for predicting transport properties given in
potential has proved most useful in predicting
this section.
transport properties for many nonpolar gases. The
Probably the inadequacy of the
greatest
constants e and o must be evaluated from experi-
simple model gas the failure to consider the
is
mental data. This potential is not useful for polar
forces acting between molecules. Molecules of real
molecules such as water or for long molecules,
gases attract each other when they are far apart
excited molecules, free radicals, or ions. The
and repel each other when they are close together.
engineer must therefore resort to more empirical
The exact mechanism of these attractive and repul-
methods of prediction of transport properties for
sive forces is not understood. Even in the most
these molecules.
rigorous kinetic theory, it is necessary to assume
A detailed treatment of the rigorous kinetic
the form of these intermolecular forces. The inter-
theory using the Lennard— Jones model is beyond
molecular force is related to the potential energy
of interaction by a simple expression
::

(9.43)
dr

where F is the force, p is the potential energy, and

r is the distance between molecules. A negative

force (F) implies an attraction between molecules, I
- Intermolcular re — -. ~ : :. ; :--•;:: :-

whereas a positive force implies a repulsion. The

form of the equation describing the potential
energy of interaction is not known. Many empirical
forms have been proposed and tested with experi-
mental data.
One such empirical expression that has gained ::
wide acceptance is the Lennard— Jones 6-12
potential:
12
p = 4e (9.44)
l\r (;)]
-1.6 -
where

p = intermolecular potential energy function

e = maximum energy of attraction of two
molecules Figure 9.8. Lennard— Jones 6-12 potential for nitrogen.
 MOLECULAR TRANSPORT MECHANISM 177

the scope of this book; however, the resulting where k = thermal conductivity, J/s m 2 (°C/m) and
equations are useful in predicting transport the other symbols are as defined earlier.
properties. The equations will be presented and Combination of Equations 9.45 and 9.46
applied to the calculation of transport properties. yields
For the viscosity of pure gas,
a

Ivff
k= —
15 R
4
—u /VT
(9.47)
25
M = 8.44x 10~ (9.45)
o'Sli This expression may be used to calculate k from ju

where for monatomic gases.

= absolute viscosity, kg/m For polyatomic molecules, Equation 9.47 must

Ijl s
be modified to take into account internal degrees
M = molecular weight, kg/mol
of freedom in the molecule. For polyatomic mol-
o = collision diameter, m (Appendix D-6)
ecules.
£2i = collision integral, a function of 7"*
(Appendix D-6) ]5 R^ 4_ c_v 3
7~= absolute temperature, K k = (9.48)
4 M L15 R 5
T* = reduced temperature = kTle
elk = potential parameter, K (Appendix D-6) where Cv is the molar heat capacity at constant

const = a dimensional number

volume.

This expression may be compared to that for the

Example 9.7. Calculate the thermal con-
simple model gas given in Equation 9.41. The
ductivity of carbon dioxide at 100°C and normal
collision integral is evaluated from a consideration
atmospheric pressure.
of the mechanics of molecular collisions. It is a
function of temperature, so that the viscosity Solution. Equation 9.46 may be modified
described by Equation 9.45 is no longer a simple for use.
function of temperature, as contrasted to the
model gas where it varied as 7~1 /2 The constants a 2.63 x 1CT 23 ^77M A_ Cv_ 3
k =
.

and elk given in Appendix D-6 are evaluated a

2
fti 15 R 5J
from experimental viscosity data, so that predic-
tion of viscosity is excellent. As shown later in this
T=373K,c v =40.5J/mol K
section, same constants may be used to
these
(Appendix D-13)
predict thermal conductivity and mass diffusivity. R = 8.314 J/mol K, M = 44.0 x 10" 3

From Appendix D-6, for C0 2 , elk = 190 K

Example 9.6. Calculate the viscosity of argon
at 0°C and atmospheric pressure. a = 3.996 x 10 -o
1
m, 7— -g
373
=1.96.
Solution. For argon from Appendix
D-6a, elk = 124 K, a = 3.418 x 10" °m.
ecular weight is
1
The mol-
39.9 x 10~ 3 kg/mol. Then T* =
^ = 1.184

273/124=2.20. From Appendix D-6b at T* = 2.63 x 10~ 23 V373/44 x 10~

3
4 40.5
2.20, ^t = 1.138. Substitution of these values in k=
(3.996 x 10" 10
2
) 1.184
Equation 9.45 gives
2
25
= 0.0243 J/s m (°C/m)
_ 8.44 x 10~ V39.9 x 10~ 3 x 273
M
(3.418 x 10- 10 ) 2 (1. 138) This value agrees well with an experimental value
of 0.0212.
= 2.10x 10~ 5 kg/ms
For calculation of mass diffusivity, the
This value is in exact agreement with the experi- following equation has been derived for binary
mental value, as expected. mixtures:

For the thermal conductivity of a pure mon- .

(!)(JL + _L)
atomic gas 24 V
/ (7 3)
\2l\Ma MJ
b =8.42x 10
Po ab 7 n 2
,

k = 2.63x 1(T 23 ^y— (9.46)

o SI, (9.49)
1 78 MOLECULAR AND TURBULENT TRANSPORT

where The experimental value of J (

is 1.65 x 10~ 5 .

%b = mass diffusivity, m /s
2 An additional higher -order correction factor
M = molecular weight of species a, may be applied to Equations 9.45, 9.46, and 9.49
a
kg/mol for greater accuracy (7). This correction is less than
M b
= molecular weight of species b, 1 percent, and it has been omitted here.
kg/mol As was mentioned earlier, the Lennard— Jones
P = total pressure, N/m 2 6-12 model does not hold for polar molecules,
oab , £2 2 , Tab * = Lennard— Jones constants free radicals, or long molecules. Two important
ammonia. Values for
polar molecules are water and
The Lennard— Jones constants are evaluated from the transport properties of these anomalous
the following relationships: molecules have in some cases been correlated by
oab =Hoa +ob ) (9.50) semiempirical methods. For example, viscosities
can be correlated with the Sutherland model (1,
where oa and o b are the collision diameters for p. 565); G Mi land has developed an empirical
i

each molecular species (Appendix D-6a). correlation for the mass diffusivities of many gases
(5).
tab _ e
_a_ £a
X (9.51 The Gilliland equation is
k V k k
where
found
the individual
Appendix D-6a. With Tab
in
potential parameters
* =kT/e ab ,
are
H2 1.38 x10- 7
vA 3
(
{m*
— —
mJ
+ )

is evaluated using Appendix D-66. 9m (9.52)

2
PiV'^ + Vt" 3 ^

Example 9.9. Calculate the mass diffusivity where V is the molar volume, as given in Appendix
for the mixture of carbon dioxide and nitrogen at D-7. The term V U3 measure of the molecular
is a
25°C and atmospheric pressure. diameter of each constituent and is analogous to o.
Thus, the form of Gil liland's empirical equation is
Solution
similar to the previous theoretical expressions for
r=298K, Ma = 44.0x KT 3 , mass diffusivity. The constant was evaluated by
Mb = 28.0x TO- 3 measuring the diffusivities of many real gases,
including water vapor.
Tables of transport properties are included in
ForC0 2 -=190, a a =3.996 x 10- 10
k Appendix D.

ForN 2 - = 79.8, ob = 3.749 x lO" 10 Example 9.10. Calculate the mass diffusivity
k for amixture of carbon dioxide and nitrogen at
Therefore,
25°C and atmospheric pressure using the Gilliland
correlation.
a a6 = ^(3.996 + 3.749) (10- 10 )

Solution. From Appendix D-7, the molar

= 3.872 x 10~ 10 m volume of Vb = 2 x 15.6 = 31.2 x KT 6
N 2 is
Cab m 3 /mol, and for C0 2 Va = 14.8 + 7.4 + 12 = 34.2
,

= V190x79.8 = 123.0 K x 10~ 6 m 3 /mol. Ma =44.0x10- 3 and Mb =

28.0 x 10- 3
.

Then
9=
T3b *= 298/123 = 2.42

Then from Appendix D-60, = 1.010. Using 1.38 x 10- 7 3

r) +( -)
V/(298)
J2 2 (

Equation 9.49 gives \44x1Q- 3 '

\28x1Q- 3 /

5 1/3 1/3 2 (10- 6 2/3

(1.013 x 10 )(34.2 + 31.2
^=8.42x 10- 24 ) )

= 1.31 x 10~ 5 m 2
/s
1
LZ)
V/
(2 98)3(1)( +
\2/\44x1Q- 3
28x10~ 3
/
The result not as close to the experimental value
is

as is that of Example 9.9. Generally, the Gilliland

(1.013 x 10 5 )(3.872x 10" 1 °)
2
( 1.010)
correlation should be used only when the
= 1.53 x 10~ 5 m 2/s Lennard— Jones 6-12 model cannot be used.
 MOLECULAR TRANSPORT MECHANISM 179

The transport properties of pure gases have

been considered in this section, The viscosity of a
gas mixture may be calculated from pure-
component viscosity data using a semiempirical obcboo
equation developed by Wilke (15). The thermal
conductivity of gas mixtures may be calculated in a oo oooo -
manner analogous to that for viscosity. Diffusion
in systems with more than two components is
oooo -
described by complex transport equations (3),
which will not be considered here.
o ooooo -
Dense Gases
oooo oo •
Gases under high pressures do not follow the
kinetic theory upon which the equations for
predicting transport properties of this section are
based. Other methods must be used. The kinetic
oo oooo •

Figure 9.9. Simple model of a liquid, o is the

theory of dense gases has not been fully developed. molecular diameter; d is the free distance
Most successful methods of correlating high- between molecules; A is a "hole" between
pressure data on transport properties are based molecules.
upon the principle of corresponding states.
References 3 and 10 summarize these methods.
holes of free space, as shown schematically in
Figure 9.9. A molecule may migrate into a neigh-
TRANSPORT IN LIQUIDS
boring hole, providing enough force is available to
The kinetic theory of gases has been well overcome the intermolecular forces holding it in
developed, so that available data may be gener- place. By migrating, the molecule creates a new
alized and predictions of transport properties may hole, which may be filled by another molecule.
be accomplished with considerable accuracy. The Thus, molecular migration is possible, but at a
kinetic theory of liquids is not as well organized, much lower rate than in gases because of the
and no single theory has been generally accepted. greater resistance to migration in liquids. As a
This lack of theoretical development is due to the result, molecular mass transport, which depends on
complex structure of the liquid, where the migration, is restricted in liquids. In a liquid, a
molecules are close together and intermolecular given molecule vibrates within a limited space, con-
forces are strong. Many proposed theories fined by the forces of neighboring molecules. The
postulate the existence of "free space" within the molecule may collide with a neighbor, thus trans-
liquid. In a gas, there is much free space, as mitting some thermal energy without leaving its
evidenced by the great compressibility of a gas fixed position. This is one mechanism for molecu-
when pressure is Liquids are much less
applied. lar transport of heat. To cause flow in a liquid, a
compressible, but a slight contraction under high relatively large force must be applied to overcome
pressure supports the postulate that a very limited the strong intermolecular forces tending to hold
amount of free space exists in liquids. Several molecules in place. As a result, the momentum
theories subdivide the free space into regions of transport coefficient (i.e., the absolute viscosity) is
approximately molecular size called holes. higher in liquids than in gases.
The kinetic-theory models for gases are made Examination of the transport diffusivities for
up of continuous free space throughout which are water illustrates the relative importance of the
distributed moving molecules. The molecules various transport mechanisms. At 0°C in liquid
occupy only a small fraction of the available water, <j = 1 .36 x 1 CT 9 a=142x1(T 9 and v =
,

-9
volume. In contrast, the models for liquids postu- 1796 x 10 all in square meters per second. The
,

late a continuous phase of molecules in close array, mass diffusivity is the coefficient for self-diffusion
held together by strong intermolecular forces. The (water in water). The values indicate that the
spacing between molecules (somewhat analogous limited molecular migration leads to a low rate of
to the mean free path in a gas) is very small, of the diffusion, but collision between vibrating mole-
order of the molecular dimension. Dispersed cules gives a moderate rate of heat conduction; the
throughout the continuous molecular phase are strong intermolecular forces lead to a high rate of
180 MOLECULAR AND TURBULENT TRANSPORT

momentum transport, for the same gradient in If the logarithm of Equation 9.53 is taken,
transferentproperty. The transport diffusivities B
may be compared to those for gases, which are all In n = + ln C (9.54)
RT
of the same order of magnitude. For example,
for C0 2 at atmospheric pressure and 100°C, This is the equation of a straight line if In is /jl

1/ = 1.6x 1fJ- 5 ,a= 1.7x 10- 5 ,and^= 1.25x 1(T 5 ,

plotted as a function of 1/7". Regular log 10 semi-
all in square meters per second.
logarithmic coordinates may be used for the
It should be emphasized that the representa-
straight line plot, noting that Equation 9.54 may

tion of a liquid shown in Figure 9.9 is only be written as

D
schematic. The molecules of a real liquid are not as
iogioM = + 'ogioC 0.55)
well ordered. Liquids may be considered to have
-
2303/?7
a degree of molecular orderliness somewhere
between the complete disorder of a gas and the Thus, the slope of the straight line resulting from
regular configuration of crystalline solids. plotting n versus 1/7" on semilogarithmic coordi-
The simple "hole" model for liquids has been nates is B/2.303R.
used by Eyring and others (6), together with the
theory of absolute reaction rates, to yield an TK - n
expression for the variation of liquid viscosity with
temperature. It takes the general form 273 0.00366 0.001792
B/RT 293 0.00342 0.001005
H = Ce (9.53)
313 0.00320 0.000656
333 0.00300 0.000469
where 7" is the absolute temperature, R is the gas 354 0.00283 0.000357
constant, and B and C are constants that depend
upon the liquid under consideration. Actually, this These data are plotted on semilogarithmic coordi-
equation form precedes Eyring's development. The
nates, as shown in Figure 9.10. It is noticed that
problem is to evaluate the constants B and C. They the data do not fall on a straight line. Nevertheless,
are not even always constant. If limited experi-
the best straight line is drawn through the data to
mental data are available, Equation 9.53 can be
evaluate the constants. The constant B could now
used to correlate data for interpolation or extra-
be evaluated from the slope of the straight line and
polation over a limited range.
C from the intercept (by extrapolating to 1/f = 0).
It is more convenient to evaluate B and C by taking

Example 9.11. The following experimental two points from the straight line through the data.
data are available for the viscosity of water:

Temperature, Viscosity,
°C kg/m s

0.001792
20 0.001005
40 0.000656
60 0.000469
80 0.000357

(a) Evaluate the constants of Equation 9.53

for water.
(b) Predict the viscosity of water at 100°C.

Solution
(a) It would be possible to use any two of the
data points given to evaluate B and C. The data are 0.001
0.0028 0.0030 0.0032 0.0034 0.0036 0.0038
subject to experimental error, however, so it is
MT
preferable to use all the data. The data will be
plotted as indicated by Equation 9.53. Figure 9.10. Solution to Example 9.1 1
 MOLECULAR TRANSPORT MECHANISM 181

Example
9.12. Estimate the viscosity of
At ju = 0.0004 -= 0.00293 liquidwater at 20°C from the following data:
Tb = 100°C,P2o°c = 0.998 x 10 3 kg/m 3
At n = 0.0015 -=0.00360 C
Solution. The molar volume at 20 C is

Then 18 x 10~ 3 kg/mol

V= = 18 x 10" 6 m 3 /mol
-+ 0.998 x 10 3 kg/m 3
In 0.0004 = 0.00293 In C (a)
R From Equation 9.56
- + In C 4x 10- 10
In 0.001 5 = 0.00360 (b) (3.8x373/293)
R 18x10~ 6
Subtracting (a) from (b) gives
= 0.0028 kg/m s

B 0.0015
0.00067 -= In-— This may be compared with an experimental value
R 0.0004 of 0.0010, illustrating the inadequacy of Equation
B 9.56 for polar liquids.
= 1970 The Eyring theory predicts that the mass
R
diffusivity is a function of temperature:
Then
0.0015 ^ = Ae -BIRT (9.57)
C = BIRT (1970)(0. 00360)
T
e e
The constant B is assumed to be the same as that in
C= 1.25 x 10~ 6
Equation 9.53, whereas the constant A is different
from the C. Combining Equations 9.53 and 9.57
Therefore
gives a relationship between 'J and [i.
M=1.25x10- 6 (e 1970/r )
ACT
(b) At T= 100°C = 373 K (9.58)
M
10- 6 1970/373 = 0.000244 kg/ms
H= 1.25 x (e )
Wilke (7) defined F a function of the molar
This is lower than the experimental value of volume as ,

0.000284 because the actual viscosity does not T 1

F = (9.59)
exactly follow Equation 9.55. This equation "
AC
should be used only over short temperature ranges.
Application of Eyring's theory permits The constants A and C in the Eyring theory are
approximate evaluation of the constants B and C also functions of molar volume. Wilke correlated
when no viscosity data are available. The derivation experimental data for numerous systems of dilute
(2,6) results in solutions of nonelectrolytes to obtain F as a
function of volume, as presented in
molar
4x1Q- 10 e 387V 7
'
Appendix D-8. The molar volume may be calcu-
(9.56) lated from density data or by Kopp's law, using the
V
atomic volumes given in Appendix D-7. The atomic
where volumes are added, according to the chemical
V= molar volume at temperature T, m 3/mol formula of the liquid under consideration, to give
T= temperature of interest, K the molar volume.
Tb = normal boiling point, K The function <p that appears in Appendix D-8
H = viscosity kg/m s takes into account various properties of the
solvent, such as hydrogen bonding. For water
This equation is approximate. It is inadequate for = 1.0, for methanol 0.82, and for benzene 0.70.
polar and long molecules. Molar volume
liquids For other solvents, <p may be determined from
data at the normal boiling point are given in known values of viscosity and diffusivity. If no
Appendix D-7. For other temperature, values may data are available, a value of = 0.90 is recom-
<t>

be calculated from the density and molecular mended. Further work by Wilke and Chang (11)
weight. led to an empirical equation correlating data on
182 MOLECULAR AND TURBULENT TRANSPORT

many aqueous and nonaqueous solutions. cent, but occasionally with errors as great as 50
percent:
-'=5.88x 10- 15 (xM) 1/2 7"/aiV°-
6
(9.60)

%p /o\
1 /3
where k = 3.59 x 10- - (9.6V
5 = mass diffusivity, m 2/s
where
M = molecular weight of solvent, kg/mol
T= temperature, K k = thermal conductivity, J/s rn°C
p = viscosity of solution, kg/m s c P = heat capacity, J/kg°C
V= molar volume of solute at the normal p = density, kg/m 3
m 3/mol
boiling point, M = molecular weight, kg/mol
X = Association parameter of solvent: water,
X= 2.6; methanol, 1.9; ethanol, 1.5; ether, Example 9.14. Calculate the thermal con-
benzene, heptane, other unassociated ductivity of liquid water at 20°C, knowing
3
solvents, 1.0. p = 998 kg/m c p = 4.18 x 10 3
, J/kg°C, M = 18.0
x 10~ 3 kg/mol.
Example 9.13. Calculate the mass diffusivity
of ethanol in a dilute solution with water at 10°C.
Solution. From Equation 9.61
1 3
Solution. The molar volume of the solute 998 v
'

£ = (3.59x10- 9 )(4.18x10 3 )(998)l

ethanol can be calculated using the atomic volume M8x10~ 3 '

data of Appendix D-7: = 0.569 J/s m°C

Ethanol = C 2 H 5 OH The experimental value at 20°C is 0.598.

2(C) = 2(14.8 x 10~ 6 = 29.6 x 10 )

6
Non-Newtonian Fluids
6(H) = 6(3.7 x 10- 6
) =22.2x 10~ 6
The correlations for prediction of viscosity given
1(0) = 7.4 x 10~ 6 = 7.4 x 10~ 6 earlier in this section apply only to Newtonian
59.2 x 10 6 m 3 /mol fluids. A Newtonian is one that follows
fluid
Newton's law (Equation 9.31 ), which assumes that
For water = 1.0, so from Appendix D-8, F = 2.4 x
the absolute viscosity p is constant as dv/dx
10 12 Then with p = 0.1308 kg/m s, Equation 9.59
.

changes and as time passes. A common way of

gives
stating this is to say the Newtonian viscosity is
£= TIFyt = 283/(2.4 x 10
1 2
) (0.1 308) independent of rate of shear (dvldx) and time of
^ 10 application of shear. If viscosity varies with rate or
= 9.0x 10 m 2/s
time of shear, the fluid is said to be non-
Alternatively, Equation 9.60 may be used with Newtonian. All gases and liquids of low molecular
X=2.6: weight are Newtonian, as are mixtures of low-
15 molecular-weight liquids. Non-Newtonian fluids
_ 5.88 x 10~ 10~ 3 1/2
(2.6 x 18 x ) (283)
include solutions of finely divided solids in liquids
6 6
(0.1308X59.2 x 10-
-

)
as well as liquid solutions of large-molecular-weight
r
y=9.bx 10" 10 m 2 /s molecules.
Non-Newtonian fluids combine some charac-
The experimental value at 10°C is 8.3 x 10~ 10 teristics of with some characteristics of
solids
m 2 /s. liquids. For example, a small force applied to a
The foregoing discussion of the prediction of suspension of clay in water does not cause flow.
mass diffusivities applies only to dilute solutions. The mixture behaves like a solid in resisting the
In concentrated solutions, the mass diffusivity is a force because it is insufficient to dislodge the clay
function of solute concentration, and no adequate particles from their positions in an interlocking
correlation is available. matrix. At this point, the suspension might be
Although many methods for predicting the considered a very porous solid soaked with liquid.
thermal conductivity of liquids have been If the force is increased, ultimately a value is
developed, they all are inaccurate (10). A simple reached at which the force is sufficient to break
relationship proposed by Weber (13) in 1880 down the interlocking structure of the clay
predicts k with an accuracy usually within 15 per- particles. Then the suspension begins to flow,
 MOLECULAR TRANSPORT MECHANISM 183

taking on the characteristics of a liquid in which a Bingham plastic

finely divided solid is suspended. Fluids exhibiting Newtonian

this property are called Bingham plastics. Another
example of combined solid and liquid character-
istics is in a solution of high-molecular-weight
polymers. At low shear rates, the polymer mole-
cules, together with some solvent molecules,
associate to create a gel structure analogous to the
crystal structure of solids. At higher shear rates,
the structure breaks down, so that there is an
apparent decrease in viscosity as the shear rate
increases.
Non-Newtonian flow is described quantita-
tively by empirical equations that are often modifi-
cations of Newton's law. In general, the form of
the equation is not based upon an underlying re =-K(£)
TyBc
"-{dx) -
dx
n<\
theory, because the theory has not developed to
Shear rate or velocity gradient — (dvldx)
this point.
Several non-Newtonian fluids are
types of Figure 9.11. Shear behavior of non-Newtonian
discussed below. Whether a given fluid is included fluids.

in a specific category depends only upon how

closely flow data for the fluid fit the empirical

Note that if any point on the flow curve
is joined
equation. Some fluids fit none of the equations
to the origin with a straight the slope of the line,
listed. Although more complex equations are line is the apparent viscosity. Figure 9.11 and sub-
available, they are still largely empirical.
sequent text describe various types of non-
Non-Newtonian fluids are characterized by Newtonian flow behavior.
viscosities that change with rate of shear or time
of application of shear. Of the two classifica-
tions, shear-rate-dependent non-Newtonian fluids Bingham Plastic Fluids

will be described first. Shear-rate-dependent non- These substances require a threshold stress (r gc )

Newtonian fluids are usually represented by an that must be exceeded before flow can occur. For
equation of the form Bingham plastics (Figure 9.1 1 ),

dv_\ dv
Ty9c -*( (9.62) -Ms if ryx >t (9.65)
dx' \dx'
where and

Ty9c = the shear stress at the point of con-

sideration
= if Tyx <T .

dx
(-) = some function of the velocity where
gradient at the point of consider-
ation T o9c = the yield value, the initial value of the
shear stress that must be exceeded
At any point the system the "apparent"
H B = defined by Equation 9.65
in
viscosity is useful in characterizing local behavior.
The "apparent" viscosity can be defined in the For Bingham plastics,
dv
same manner as the Newtonian viscosity,
Me —
dx
+ r gc
dv_ =
TyUc (9.63) Va (9.66)
y g, "Ma
dx
(-) '
\ dx
Equating Equations 9.62 and 9.63 gives
From Equation 9.66 it is evident that ^a decreases
<p(dv/dx) with rate of shear dv/dx. Examples of Bingham
Ma
= (9.64)
{dvldx) plastics are rock and clay suspensions in water.
184 MOLECULAR AND TURBULENT TRANSPORT

Dilatant Fluids viscosity for a time after a constant stress is first

The stress-shear rate diagram (Figure 9.11) shows applied. Bentonite clay suspensions and some sols

that the equation for dilatant fluids is

are rheopectic. Thixotropic fluids exhibit a de-
crease in apparent viscosity with time under con-
"~ 1
dv_ dv_ stant stress. Thixotropic behavior is found in paint
Ty9c = -K n>} (9.67)
dx dx and catsup, among many materials. After standing
without an applied force, rheopectic and thixo-
and
-1 tropic fluids revert to their original condition.
n
dv_
=K > Non-Newtonian properties are sometimes de-
Ma A7 1 (9.68)
dx sirable. For example, thixotropic behavior is desir-
able in paints. During brush working, certain paints
Thus, the apparent viscosity increases with increase
flow readily to cover the surface, but upon stand-
in rate of shear. Starch suspensions, potassium
ing the original highly viscous condition returns
silicate "solutions," and gum arabic "solutions" are
and the paint will not run.
examples of dilatant fluids.

TRANSPORT IN SOLIDS
Pseudoplastic Fluids

This is probably the largest class of non-Newtonian The structure of solids can be characterized as
fluids. The stress-shear pattern for pseudoplastic either amorphous. In a crystalline
crystalline or

fluids shows that the equation for pseudoplastic solid, atoms are arranged in a regular pattern called
fluids is
a crystal lattice. Most solids are crystalline, includ-

n -1 ing all metals and salts. Metals are distinguished

dv_ dv_
rY 9c = -K n < 1 (9.69) from other crystalline solids by an electronic con-
dx dx figuration that permits the outermost electrons of

and the atom to move more freely throughout the

n -1 solid. These free electrons account for the high
dv_
= K n<1 (9.70) electrical and thermal conductivity of metals.
dx Amorphous solids do not have a crystal lattice,
although they do have some order to their arrange-
Note that the apparent viscosity of pseudoplastic
ment of atoms.
fluids decreases with increase in rate of shear.
Examples are solutions of high polymers, paper
Many important solids are amorphous — for
example, glass, plastics, paint, and rubber. There is
pulp, and mayonnaise
no clear division between amorphous solids and
Newtonian Fluids
liquids. Some scientists consider amorphous solids
to be supercooled liquids, because they have no
On the stress-shear rate diagram (Figure 9.11),
crystal structure.
Newtonian fluids are represented by a straight line
The physics of the solid state is a vast and
through the origin. In this case, the apparent
well-developed field. No attempt is made here to
viscosity is equal to the absolute viscosity at all
cover it thoroughly, because it alone would require
parts of the curve. Newtonian be rep- fluids can
a book longer than this one. References 11 and 12
resented by Equations 9.67 or 9.69, where fi= 1,
are good introductions to solid-state physics. Only
at which time n = n a = K.
a brief summary of the mechanisms pertinent to
The equations describing dilatant, Newtonian,
molecular transport is given here.
and pseudoplastic fluids are really all the same. The
difference arises in the numerical value of the
Viscosity
exponent n. Dilatant and pseudoplastic fluids are
sometimes combined into a power-law model, Solids may deform in two ways when a force is

described by Equations 9.67 or 9.69, where n is applied. Elastic deformation has occurred when
either greater or less than 1 .0. the solid returns to its original shape after the force
is removed. Plastic deformation has occurred when

the solid does not return to its original shape after

Time-Dependent Fluids the force is removed. Plastic deformation under
Many fluids exhibit a change in apparent viscosity long-term stresses is known as "creep," and it is
with time under the application of a constant stress analogous to flow in gases and liquids. Flow in
(8). Rheopectic fluids show an increase in apparent solids occurs at high temperatures and pressures.
 MOLECULAR TRANSPORT MECHANISM 185

Table 9.2 DIFFUSIVITIES IN SOLIDS AT 1 atm* of energy; lattice vibration is an important mechan-

ism for transport of heat in crystalline solids; elec-

System Temp, °C 9. m 2 /s tronic heat transfer is the primary mechanism in
metals. Molecular migration contributes negligibly.
10
H2 in Ni 400 0.95 x 10" If an atom is vibrating with a larger amplitude
10~ 6
H2 in Pd 400 0.316 x
characteristic of a higher temperature than its near-
12
O2 in Ag 400 0.32 x 10~
8 est neighbors, it exerts periodic forces on its neigh-
CO in Mild Steel 800 0.67 x 10~
10~ 16 bors that increase their amplitude. In this way
He in Si02 400 0.81 x
-13 vibrational energy is transmitted from a region of
Au in Ag 870 0.43 x 10
Ni in Cu 400 0.23 x 10~
12 higher temperature to a region of lower temper-
ature. In the closely bound structure of many
*By permission of National Academy of Sciences, from Int. Crit. crystalline solids, atomic vibrations generate vibra-
Tab., Vol. V. p. 76. Copyright © 1929. tion of the entire crystal lattice and subsequent
dissipation of the vibrational energy as heat.
For example, the rolling, drawing, and extrusion of In metals, some of the outer electrons of the
metals involves flow in the solid state. Although atoms are free to move about the solid, carrying
there is no clear distinction, plastic deformation is with them thermal energy from a region of higher
usually considered part of the field of solid mech- temperature to a region of lower temperature. This
anics, as contrasted with the fluid mechanics of is the major mechanism of heat transfer in metals,
gases and liquids. The physical models and termin-
with a small contribution from lattice vibration.
ology for the two fields differ. Plastic deformation Electronic transport is a particularly effective
usually time-dependent non-Newtonian
involves
mechanism; metals that are good conductors of
flow, but this terminology is not normally used,
heat are also good conductors of electricity.
nor is an apparent viscosity defined. If it were Crystalline solids with strong bonds (short-
defined, the apparent viscosity would be extremely
range ionic and covalent bonds) between atoms
high because of the large stresses required to in-
conduct heat primarily by lattice vibrations. They
duce flow in solids. Momentum transport across a
are poorer conductors than metals because they do
specified velocity gradient would also be very high
not have free electrons available for heat transport.
in solids, compared to liquids and gases. Large
The crystal structure is sufficiently rigid to permit
stresses are necessary because of the great inter-
transport by lattice vibration.
molecular forces acting to hold molecules in the Molecular collision is the primary mechanism
fixed patterns of solids.
for transport of heat in amorphous solids and in
crystals with weak, diffuse molecular bonds. Here
Mass Diffusivity
the structures are either too disordered or too
Molecular transport of mass is extremely slow in complex to develop lattice vibrations. Solids in this
solids because of the immobility of the molecules. group are poor conductors when compared to
Even in crystalline solids a few holes exist, al- those discussed earlier that possess other more
though apparently not as many as in the liquid effective mechanisms for transport of heat.
model considered earlier. Molecular migration is The effective thermal conductivity of some
limited because of the small number of holes; apparent solids isdependent upon the gross state
therefore, molecular diffusivities are extremely of aggregation of the material. For example, cel-
low, as shown in Table 9.2. The values in Table 9.2 lular and fibrous materials may include entrapped
may be compared to liquid diffusivities of the gas and thus have a low effective thermal conduc-
order of 10" 9 m 2 /sec. The first two values in Table tivity. For this reason, glass wool and foamed plas-
9.2 suggests that small gas molecules diffuse more tics are used as insulating materials.
rapidly in solids than do larger solid molecules. Data on the thermal conductivity of various
Data on diffusion in solids are limited, and there solids are given in Appendix D-10c.
are no general correlations for prediction of mass
diffusivities.
OTHER TRANSPORT PHENOMENA
Thermal Conductivity The three major types of transport considered in
Transport of heat may occur by three mechanisms this chapter arenot the only transport systems.
in solids. Collisions between individual vibrating For example, mass transport may occur by any of
molecules result in a relatively small interchange a number of mechanisms. Transport across a con-
186 MOLECULAR AND TURBULENT TRANSPORT

centration gradient is the most significant mechan- Table 9.3 TYPICAL VALUES FOR DIFFUSIVITY
ism, and it has been considered in detail earlier in RATIOS
this chapter.
temperature gradient Prandtl Schmidt
Mass transport across a
No. = via No. = vl'y
is called thermal diffusion. a temperature gradi- If

ent is applied to a uniform mixture of gases, the

Simple model gas 1 1
heavier molecules will tend to migrate to the lower
Inorganic gases 0.7-1.1 0.6-1.1
temperatures, and the lighter molecules to the Liquid water 7 1200
higher temperatures. The effect can be predicted Organic liquids 10-1000 or more 300-2000
from the rigorous kinetic theory of gases (3,7), and Liquid metals 0.001-0.1 -
it has been verified experimentally on many occa-

sions. Thermal diffusion has been applied to the

separation of gaseous and liquid mixtures. It is
particularly useful in the separation of gaseous
Few phenomena described in
of the transport
isotopes. For example, two isotopes of uranium
this follow the simple flux-gradient
section
were separated from each other by the thermal relationship derived earlier in this chapter. For
diffusion of
235 F and U 238 F
U in the early
6 6 example, thermal diffusion depends upon the dif-
stages of the nuclear energy in the Second program
ference in the logarithms of the temperatures. On
World War (2). the other hand, electron and neutron diffusion
Mass transport across a pressure gradient can
follow equations of the flux-gradient form in
be predicted from kinetic theory, but the effect is
certain elementary cases. For example, the law
small and has not been investigated to any extent.
(electrical potential/resistance) = (current) is of
Mass transport under the influence of an externally
the flux-gradient form.
applied force is called forced diffusion (4). If the
force uniformly upon all molecules of the
acts
system (as does gravity), there is no net transport. DIMENSIONLESS RATIOS OF TRANSPORT
On the other hand, an electromagnetic field im- DIFFUSIVITIES
posed upon a mixture of ionized gases may cause
forced diffusion. Ratios of diffusivities are extremely useful in
A
binary mixture of gases may be separated analysis of transportphenomena, as will be shown
by allowing part of the gas to flow through a later. The two ratios that are most useful are the
barrier containing very small holes (diameter less Prandtl number (A/ Pr = via) and the Schmidt
than the mean free path of the molecules). The number (A/ Sc = v/y). For the simple model gas
separation factor is equal to the square root of the both of these ratios are equal to unity, and for
ratio of the molecular weights of the gases. The simple real gases they are close to unity. For
process is generally called gaseous diffusion and is liquids the ratios vary widely, the ratios are useful
used to separate U 235
and U 238 F 6 in the F6 for analysis when several transport phenomena
United States nuclear energy program at plants occur simultaneously. Characteristic values of /V Pr
built at a cost of over three billion dollars (2). and /V Sc are shown in Table 9.3.
The thermal-energy flux is the sum of four
fluxes (3). The flux caused by a temperature gradi-
REFERENCES
ent has already been considered. Energy flux is also
caused by the intrinsic energy associated with the
molecules in mass transport. Thermal-energy trans- 1. Barrer, R. M., Diffusion in and Through So/ids,

port also occurs across a concentration gradient. Macmillan, New York (1941).
This is known as the "Dufour effect." Finally, heat 2. Benedict, Manson, and T. H. Pigford, Nuclear Chemical
may be transferred by radiation, a topic covered in Engineering, McGraw-Hill,New York (1957).
a later chapter. 3. W. E. Stewart, E. N. Lightfoot, Transport
Bird, R. B.,
Phenomena, John Wiley & Sons, New York (1960).
The transport of subatomic particles is
4. Chapman, S. and T. G. Cowling, 77?e Mathematical
important in a number
For example,
of fields.
Theory of Non-Uniform Gases, Cambridge Univ. Press
electron transport in solids, liquids, and gases is the
(1939).
basis for electric and electronic developments.
5. Gilliland, E. R., Ind. Eng. Chem.,26, p. 681 (1934).
Neutron-transport considerations are essential to 6. Glasstone, S., K. J. Laidler, H. Eyring, The Theory of
the design of nuclear reactors. Rate Processes, McGraw-Hill, New York (1941).
 MOLECULAR TRANSPORT MECHANISM 187

7. Hirschfelder, J. O., C. F. Curtiss, R. B. Bird, Molecular 9.6. One end of an aluminum rod is heated to 400°C
Theory of Gases and Liquids, John Wiley & Sons, New while the other end is maintained at 0°C in an ice-water
York (1954). mm
bath. The rod is 20 in diameter and 1 m long. Calcu-
8. Metzner, A. B., "Non-Newtonian Technology," Ad- late the heat flux through the rod [for aluminum k =
vances in Chemical Engineering (ed. T. B. Drew and 703J/sm 2 (°C/m)].
J.W. Hoopes), Vol. 1, Academic Press, New York 9.7. A copper rod 1 in. in diameter and 2 ft long is
(1956).
heated at one end and cooled at the other end. Heat is
9. Present, R. D., Kinetic Theory of Gases, McGraw-Hill,
New York supplied at a rate of 40 Btu/hr. What is the difference in
(1958).
temperature between the two ends of the rod, assuming no
10. Reid, R. C, Prausnitz, J. M., and T. K. Sherwood, The
loss of heat through the cylindrical surface of the rod? The
Properties of Gases and Liquids, McGraw-Hill, New
thermal conductivity of copper 2
York (1977). is 215 Btu/hr ft (°F/ft).
11. Van Vlack, L. Elements of Materials Science and
H., 9.8. Two horizontal plates of 2 m2 area are spaced
Engineering, Addison Wesley, Reading, Mass. (1967). 1 mm apart. The space is filled with a lubricating oil
12. Sproull, R. 1., Modern Physics. 2nd ed., John Wiley & (fi = 0.03 N s/m 2 ). The lower plate is fixed and a force of
Sons, New York (1967). 100 Newtons is applied to the top plate.
13. Weber, H. F., Wiedemann's Ann., Ann. Phys. Chem.,
(a) At what velocity does the top plate move?
10, p. 103 (1880).
(b) What is the momentum flux at the lower plate?
14. Wilke, C. R., Chem. Eng. Progr., 45, p. 95 (1950).
15. Wilke, C. R.,J. Chem. Phys., 18, p. 517 (1950).
9.9. Two horizontal plates are spaced 3 in. apart. The
space between the plates is filled with mercury at 90° F.
16. Wilke, C. R., and P. Chang, A.I.Ch.E. J., 1, p. 264
(1955). The upper plate is moving at 3 ft/sec, and the lower plate is

moving in the same direction at 1 ft/sec.

(a) Why does the liquid velocity vary linearly be-
PROBLEMS tween the plates?
(b) Calculate the shear stress at a plane 1 in. from the
slower moving plate.
9.1. Derive Equation 9.10, starting with a constant
(c) Calculate the momentum flux at the surface of
gradient in mass concentration (c a ), and employing the
the faster moving plate.
method used to derive Equation 9.8.
9.10. Calculate the thermal conductivity, mass dif-
9.2. Calculate the mass flux of benzene through a
2 fusivity for self-diffusion, and absolute viscosity for neon at
layer of air 10 mm thick at 25°C and 200 kN/m total 2
100°C and 304 kN/m .

pressure. The partial pressure of benzene is 6 kN/m at the

(a) Assuming that neon is a simple model gas.
left side of the layer and 1 kN/m at the right side. The
(b) Assuming that neon follows the Lennard— Jones
mass diffusivity at this temperature and pressure is
6-12 model.
4.4 x 10~ 6 m
2 /s. Also express the flux in English
units.
9.11. Convert Equation 9.45 to English engineering
9.3.Calculate the mass flux of methanol in water at
units for use with the following units:
77°F (^/=4.95x 10 _5 ft 2 /hr).Theconcentrationsof meth-
anol at two points 0.1 in. apart are 0.05 and 0.10 lb M lb/ft hr

mole/ft 3 .
M g/mole
T °R
9.4. The diffusivity of helium in Pyrex glass at 20°C
2 2 ft
and 100kN/m pressure 10~ 1
is 1.16 x m /s. Calculate
the rate of loss of helium from a Pyrex flask filled with 9.12. Convert Equation 9.46 to English engineering
pure helium at a pressure of 100 kN/m 2 and 20°C into the units for use with the following units:
helium-free air surrounding the flask (at the same pressure).
k : Btu/hr ft 2 (°F/ft)
The flask wall is 2 mm thick and has a surface area of M : g/mole
2
0.1 m .
T : °R
9.5. A furnace wall consists of 2 ft of brick. The o : ft
2
brick has a thermal conductivity of 0.6 Btu/hr ft (° F/ft), a
3 9.13. Convert Equation 9.49 to English engineering
specific heat of 0.2 Btu/lb °F, and a density of 110 lb/ft .

units for use with the following units:

The temperature at the inside surface of the wall is 1 100° F,
2
and at the outside surface 200° F. 9 : ft /hr
(a) Calculate the thermal diffusivity of the brick. M : g/mole
(b) Calculate the heat loss per hour through a wall P : psia

10 ft high and 10 ft long. T : °R

(c) Calculate the temperature of the brick 4% in. 9.14. Viscosity data are often reported in the metric
from the inside surface. unit centipoise, where 1 poise = 1 g/cm s. Develop a con-
188 MOLECULAR AND TURBULENT TRANSPORT

version factor to convert viscosity in centipoise to SI units 9.22. The viscosity of liquid acetone is 2.148 centi-
(kg/m sec) and the English engineering units (Ib/fr hr). poise at -92.5°C and 0.399 centipoise at 0°C. Predict the
9.15. Methane has a viscosity of 0.0020 kg/m s at values of viscosity at — 30°C and at 30°C and compare with
380°C and 0.00226 kg/m s at 499°C at normal atmospheric experimental values of 0.575 centipoise and 0.295 centi-
pressure (101.3 kN/m 2 ). poise, respectively.
(a) Calculate the viscosity of methane at 100°C and 9.23. Contrast the flow characteristics of mayonnaise
700°C, assuming it is a simple model gas. and catsup. Why does gently shaking a bottle of catsup for
(b) Calculate the viscosity of methane at 100°C and a few seconds accomplish greater flow than a single strong
700°C assuming it follows the Lennard— Jones 6-12 model. shake?
(c) Calculate the thermal conductivity of methane at 9.24. For a pseudoplastic fluid, what SI units would
500°C assuming it follows the Lennard— Jones model. be appropriate for the constant K (Equation 9.69)?
(d) Calculate the mass diffusivity of methane in 9.25. Which of the following is likely to be a non-
methane at 500°C, assuming it follows the Lennard— Jones Newtonian fluid: a coal slurry in water, honey, apple sauce,
model. molten steel, human blood, or a brine solution.
9.16. Calculate accurately the mass diffusivity of car- 9.26. Calculate the mass, thermal, and momentum
bon dioxide in benzene vapor at 100°C and 150 kN/m .
diffusivities of air and of liquid water at 0°C and normal
9.17. Calculate the mass diffusivity of water vapor in atmospheric pressure. Compare these values. Calculate the
nitrogen at 200°C and 200 kN/m 2 .
Schmidt number of self-diffusion and the Prandtl number
9.18. Calculate the mass diffusivity of hydrogen in at 0°C and normal atmospheric pressure for air and for
water vapor at 25°C and 200 kN/m 2 .
liquid water. Discuss the meaning of these dimensionless
9.19. Calculate the thermal conductivity of argon (in ratios in terms of a unit concentration gradient. (Calculate
70°F and 100 psi.
English engineering units) at the mass diffusivity for self-diffusion.)
The thermal conductivity of air at normal
9.20. 9.27. Calculate the Prandtl number for pure ethyl
atmospheric pressure is 1.045 at — 100°F and 1.782 at alcohol, water, mercury, and glycerol at approximately
2
200°F, both in Btu/hr ft (°F/ft). Calculate the thermal 20°C. Compare the resulting values in terms of the indi-

conductivity of air at 0°F. vidual properties.

9.21. The viscosity of liquid benzene is given below. 9.28. Discuss the reasons for the differences in

Develop an equation useful in predicting the viscosity of Prandtl number and Schmidt number for the various
benzene in this temperature range materials given in Table 9.3.

perature, Viscosity x 10 3
°C kg/m s

10 0.758
20 0.652
30 0.564
40 0.503
50 0.442
60 0.392
70 0.358
80 0.329
 Ten

Differential Mass, Heat,

and Momentum Balances

The molecular-transport equations developed in ute to the inflow and outflow of transferent
Chapter 9 may be applied in the analysis of many property. The other mechanisms must be con-
problems encountered in unit operations. In this sidered in any complete accounting of transfer
chapter procedures for analysis are developed using within a system. A complete analysis requires the
one-dimensional situations, together with a few statement of all transfer mechanisms, combined to
simple two- and three-dimensional cases. In the describe the physical behavior of the system as
next chapter, these procedures will be extended to accurately as possible. Often the most effective
develop the generalized three-dimensional equa- approach is to write a mass, heat, or momentum

tions. balance including all transfer mechanisms acting in

The three basic molecular-transport equations a volume element of differential size. The resulting
developed in Chapter 9 are: differential equations may be integrated with
appropriate boundary conditions to obtain a final
1. Newton's law for momentum transport
expression describing the overall system. Here only

Ty9c = -H —
dv
(9.31;
molecular transport mechanisms will be included.
Other transport mechanisms are discussed in
dx
Chapters 12 and 13.
2. Fourier's law for heat transport:
A systematic approach is desirable in writing
-q = -k —
dT
(9.21:
mass, heat, and momentum balances where proper-
ties may vary with position and time. The basic
A dx
balance equation states that input plus generation
3. Fick's law for mass transport:
must equal output plus accumulation for any
volume element:
(9.11)
A dx
Rate of Rate of
The symbols for these three expressions are INPUT GENERATION
defined Chapter 9. As discussed in the previous
in
Rate of Rate of
chapter, the three expressions are of the same + 10.4)
OUTPUT ACCUMULATION
form, relating the flux of transferent property
(mass, heat, or momentum) to the gradient in Input or output to a volume element may be by
concentration of the property. molecular transport, or by bulk flow if the system
In a given volume element, the concentration is a fluid in motion. For example, consider water

gradient —
and hence the flux — may vary with
position and with time. Thus, it is necessary to
integrate Equations 9.1 1, 9.21, and 9.31, taking
into account variations within the volume of the
system under consideration. In addition, mechan-
isms other than molecular transport may contrib-
being heated as it passes through a steam-jacketed
pipe. The volume element might be a section of
pipe one foot long. Heat is transferred from the
hot pipe to the water by molecular transport when
the water is in laminar flow. In addition, the water
flowing into the one-foot section brings with it a

189
190 MOLECULAR AND TURBULENT TRANSPORT
quantity of thermal energy, as determined by its
temperature and flow rate. Similarly, the water
flowing out of the one-foot section carries thermal
energy with it.
The transferent property may actually be
generated within the volume element. A common
example of internal generation of heat is an electric
heating element. Heat is generated within the metal
element because of the resistance of the wire to the
flow of electricity. The heat is then transferred to
the surface of the element by molecular transport
(conduction), where it is dissipated
surrounding air. A common obvious but
example of internal generation is a fluid flowing in
a pipe. Momentum is generated at every point in
the flowing fluid and is transferred to the wall of
the pipe. An example of internal generation of
mass is a nuclear reactor, where neutrons are
generated by the process of fission of uranium.
After generation, the neutrons diffuse through the
reactor core, obeying the simple mass-diffusion
equation.
The term generation should not be confused
with creation. The law of conservation of mass and
energy states that neither can be created nor
destroyed, although it is possible to convert mass
into energy and vice versa. Therefore, the mass,
heat, or momentum generated within the volume
must have been present in a different form before
it was transformed into the form being generated.
For example, in the electrically heated element,
electrical energy is converted into thermal energy.
In the flowing fluid, a portion of the pressure
increase supplied by a pump to the fluid is trans-
formed into momentum, which
to the wall. During this process, a portion of the
mechanical energy supplied by the pump is dis-
sipated as heat in the fluid. In a nuclear reactor
the neutron was originally part of the mass of the
uranium nucleus. The fission of the nucleus liber-
ates not only neutrons and fission
also energy that appears as heat. Thus, the nuclear
reactor core is also an example of transfer with
internal generation of heat.
The rate of generation as given in
10.4 may have a negative value. For example, in
diffusion with chemical reaction, the reactants are
being consumed (to give other chemical species), so
that the rate of "generation" of reactants is nega-
tive.On the other hand, the rate of generation of
the products of the reaction would be positive.
The rate of accumulation given in Equation
10.4 takes into account the unsteady-state
behavior of the transfer within
element. If the concentration of transferent
property at a given position changes with time,
there is an accumulation, and an unsteady state
exists. For example, consider an electric heater just
after the current has been switched on. Initially,
the heater is at room temperature, but the heat
generated by the electrical energy dissipated by the
resistance causes the element to heat up. There is
an increase in temperature at any specific point in
the heating element, and thus the accumulation of
thermal energy has a positive value at that point.
by heating the Ultimately, the heating element becomes hot
less enough to transfer heat to the surroundings at the
same rate as it is being generated in the element, so
that the temperature at any point does not change
with time. Then the accumulation is zero and a
steady state exists. If the current is now turned off,
no more heat is generated, and the heating element
cools off. The temperature decreases, so that the
accumulation has a negative value. This unsteady
state persists until the heating element again
reaches the temperature of the surroundings. Equa-
tion 10.4 may be applied to the electric heating
element as a whole at any time during the heating
and cooling described above.
remainder of this chapter, quantitative
In the
expressions based upon Equation 10.4 will be
developed for various transfer systems involving
molecular transport. The general procedures also
apply to turbulent transport, discussed in
Chapter 12.
METHOD OF ANALYSIS
is transferred
A may be outlined for the analy-
general procedure
sis of transfer systems involving molecular trans-
port. The individual steps are listed below.
1 . A physical model is formulated. The speci-
fication of the physical model requires a thorough
fragments but understanding of all mechanisms that may in-
fluence the concentration of transferent property
in the system. The model should adhere closely to
physical reality, although certain physical phenom-
Equation ena that have negligible influence are often omit-
ted. For example, in a mass-transport model ther-
mal diffusion and forced diffusion are usually
omitted because they are negligible compared to
ordinary diffusion resulting from a mass concen-
tration gradient. In practice, the physical model is
frequently a compromise between physical rigor
and mathematical tractability. A physical model
including all possible mechanisms often leads to
the volume mathematical expressions that are difficult to use.

The model may be simplified to make the math-
ematics more feasible, but the simplified model
may no longer lead to an accurate description of
the actual system. Therefore, it is always desirable
to check the final expressions describing the sys-
tem by using experimental data obtained from the
system.
2. Balances are written for the pertinent trans-
ferent properties. Mass, heat, and momentum
ances are obtained for a small, finite volume ele-
ment, including all mechanisms that appear in the
physical model. Because concentration and mech-
anisms vary within the volume of the system, it is
usually necessary to obtain the balances for an
infinitesimal volume element. This leads to differ-
ential equations, which are then integrated to de-
scribe the entire system. To obtain the differential
equations, the limits of the balances on a small,
finite volume element are taken as the dimensions
of the volume element approach zero, in the usual
procedure of differential calculus. If the transfer-
ee property concentration is a function of more
than one independent variable, a partial differential
equation is necessary. For example, in the electric
heating element being heated, the temperature is a
function of both time and position in the element,
so that a partial differential equation is needed to
relate temperature to time and position.
3. The flux terms in the differential-balance
equations are replaced by concentration-gradient
expressions. In other words, \p is replaced by
-8A(dr/dx) using Equations 9.11, 9.21, or 9.31.
This step reduces the number of variables in the
differential equations by eliminating the flux vari-
able.
4. The differential equations are integrated,
using appropriate boundary conditions. The re-
quired integration may be quite complex. Some-
times the several differential equations are inter-
related, so that they must be solved simul-
taneously. In many cases, the equations cannot be
integrated analytically, so that numerical methods
must be used. The advent of digital computers has
made numerical methods much more feasible.
Bounday conditions are chosen from known data
on the system. For example, when a fluid flows
through a pipe, the velocity at the pipe wall is
assumed zero. This information could be used as a
boundary condition to evaluate one integration
constant in an expression relating fluid velocity
and position in the pipe.
5. The expressions are checked with
final
available experimental data. This step is necessary
to determine whether the model formulated in
DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 191
1 actually described physical reality with sufficient
accuracy. For example, suppose it was assumed
that the thermal conductivity and heat capacity of
a metal bar being heated were constant, when in
fact they varied considerably over the range of
temperature through which the bar was being heat-
ed. The final equation relating temperature to time
and position would not fit experimental data on
bal- temperature taken at various times and positions in
the bar. If the final expressions do not fit the
experimental data, the physical model and its
mathematical description must be examined to de-
termine how they can be improved. This procedure
often leads to a better understanding of the actual
physical mechanisms involved. It is possible that
the model may never fit the data, however, because
of a significant lack of understanding of physical
reality. If the data are accurate, they are a better
description of physical reality than is any inference
drawn from the physical model.
The five steps outlined above are applied to
molecular-transport systems of varying complexity
in the remainder of this chapter. Only a few simple
systems are described to demonstrate the pro-
cedures. Many more complex systems have been
analyzed (1,5). This method of analysis is ex-
tremely important to an engineer, who may call
upon an applied mathematician to assist in the
integration in step 4, but will have to do the other
steps alone.
Geometrical Considerations
Systems to be analyzed may take any size or shape.
Many of these shapes are not easy to describe
analytically, so that approximate numerical meth-
ods may be necessary. A few simple shapes that
may be treated analytically are presented here.
The flat-slab geometry (Figure 10.1a) is based
upon the rectangular coordinate system. In this
chapter, molecular transport is presumed to occur
only in the x-direction. The area across which the
transport occurs is constant and equal to the area
of the plane perpendicular to the transport direc-
tion. The finite volume element has an area A and a
thickness Ax.
For the cylinder (Figure 10.16), the direction
of molecular transport is assumed to be only along
a radius. It is also assumed that there is radial
symmetry — that the properties at a given r are the
same regardless of the direction of r. This elim-
inates the need to consider the angular component
step in the cylindrical coordinates.
 192 MOLECULAR AND TURBULENT TRANSPORT

geometry of the system, as shown in the following

paragraphs for a cylinder.
A2
Direction of
molecular transport
The mean area for transport through the wall
(x-direction) of a cylindrical pipe of length L, inside radius r y
and outside radius r 2 is determined by the follow-
ing procedure.
The general transport equation for radial trans-
port may be written from Equation 9.12 and
Table 9.1 as

i> = -5 — 10.5)
dr

The rate of transport (\^A for molecular trans-

port alone) is constant with radius. Equation 10.5
may be multiplied by A to give

(\}/A) = -8A —
dT
dr
= constant (10.6)

Separating variables and integrating from

(T = T, at r = /"! ) to (T = r2 at r = r2 ) gives

IW) r-f = -5(r 2 -r 1

10.7)
Direction of r, A
molecular
transport
(radial)
Forthecylinder, from Figure 10.16, A = 2irrL,
so
«jA) -r, dr
= -5(r 2 -r 1 ) 10.8)
2nL r, r

and
it A) r2
Figure 10.1. Simple geometries, (a) Flat slab: Tran.,port in- = -5(r 2 -r 1 ) (10.9)
area = A (constant);
volume of element = AAx. (b) Cylinder:
2ttL n
transport area at radius/- = 2nr(x 2 — *1 ); volume of element = Equation 10.9 is rearranged and the right-hand
2nr _V Ax. (c) Sphere: transport area at radius r = 4nr 2
volume of element = 4nr 2 Ar. side is multiplied by
r2 -ry
(r 2 -n)L Ti
In the sphere (Figure 10.1c), transport is as- (C"4) =2tt -8 10.10)
sumed to be in the radial direction, and symmetry
\r\(r 2 /r-i r2 -r y
J

around the center point is assumed. In other The mean area is defined as the area through
words, the properties at a specified radius are as- which the overall average driving force (— AT/Ac)
sumed uniform regardless of the angle of the rad- would produce the constant rate (\jA).
ius. A more general treatment would include two

angular coordinates as well as the radial coordinate rz-n

to take into account variations of properties in the
{^A)=A -6 10.11:
r 2 -/-!
angular directions.
If the rate varied as the radius changed, an
Comparison of Equations 10.10 and 10.11
accumulation would occur and an unsteady state shows that
would exist. The rate is equal to the flux times the _ 27r(r 2 - r^)L
area across which the flux is occurring.
A= . , . (10.12)
\n(r 2 /r-\

It is sometimes convenient to define a mean

Since A = 2-rrrL
transport area, so that a single value of area may be
used in the integrated transport equation. The
A 2 -A,
A = 10.13)
form of the mean area is determined from the \n(A 2 IA-
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 193

This is referred to as the logarithmic mean For the the physical model was chosen as
first step,

= ———
r-i — f-\ steady-state molecular transport in the x-direction
area (A, m ). If A = 2nrL, then r = r, m -
only with no internal generation. The second step
ln(r 2 /r 1 )

involved writing the balance on a finite volume

where r/m is the logarithmic mean radius in this
case.
element (Equation 10.16) and then taking the limit
to obtain a differential equation (Equation 10.19)
which is a balance on an element of infinitesimal
thickness dx.
Steady-State Transfer
The remaining steps of the five-step procedure
At steady state, there is no accumulation of trans- are illustrated by the examples from mass, heat,
feree property in a volume element, so Equation and momentum transfer that follow.
10.4 becomes

Rate of Rate of Rate of Momentum Transfer

INPUT GENERATION OUTPUT The momentum-balance equation for a rectangular

volume element is written from Equation 10.19.
(10.14)
d{ry gc A)
Steady-state transfer may
be further classified on = ; 10.20)
the basis of the presence or absence of internal dx
generation. Transfer without internal generation This can be integrated to give
will be referred to as simple transfer. Several ex-
amples of simple transfer and transfer with internal {r y gc A) = constant :io.2i:

generation are considered in the following sections. Equation 10.21 may from a
also be derived
consideration of the upon a fluid
forces acting
SIMPLE TRANSFER contained between two parallel moving planes. The
external stress pattern acting upon the fluid is
Where there is no internal generation, Equation
shown in Figure 10.2. Planes 1 and 2 are moving at
10.4 reduces to
velocities v-\ and v 2 as a result of forces F and F 2 :

Rate of _ Rate of nniRl applied to the planes. The planes are moving at
Iiu-'OJ
INPUT OUTPUT constant velocities; constant forces are required to
and output may be by both
In general, input overcome the resistance to flow offered by the
molecular transfer and fluid flow. If input and fluid between the planes. At steady state, an ex-
output are by molecular transport only, Equation ternal force balance upon the fluid states the alge-
10.15 applied to the rectangular volume element braic sum of forces must be zero.
shown in Figure 10.1a gives
F, +(-F2 )=0 (10.22)

(R=(^Uax dO.16)

Rearranging and dividing by Ax gives an equivalent

expression

(>M)x + ax-(<M) x
= :10.17)
Ax
Equation 10.17 is easily converted to differential
form by taking the limit as Ax approaches zero

lim
WA) x + t x -WA) x dtyA)
Ax-0 Ax dx
;10.18)

Combining Equations 10.17 and 10.18 gives

d(\pA)
= (10.19)
dx
This differential equation results from the first two Figure 10.2. The external stress pattern in simple
steps of the five-step procedure discussed earlier. momentum transfer.
194 MOLECULAR AND TURBULENT TRANSPORT

Forces are here defined as positive when act- Example 10.1. Two parallel flat plates are
ing in the -(-/-direction. In Chapter 9 it was shown spaced 2 in. apart. One plate is moving at a velocity
that a -(-/-directed force represented an +x-directed of 10 ft/min and the other is moving in the oppo-
momentum flux flowing into the plane; whereas a site direction at 35 ft/min. The viscosity of the

—/-directed force represented an +x-directed fluid between the plates is constant at 363 lb/ft hr.
momentum flux flowing out of the fluid plane. (a) Calculate the stress on each plate.
An external force balance on the slab of fluid (b) Calculate the fluid velocity at ^-in. inter-
between plane 1 and the plane atx gives vals from plate to plate.

F, - (t v A) x =0 10.23) -* v2 = — 35 ft/min

and an external force balance between plane 2 and

2 in.
the plane atx + Ax gives

-F2 + (t v A x + Ax = (10.24)

Combining Equations 10.22, 10.23, and 10.24

gives Solution (a)
1.* The physical model is for simple
(r y A) x + Ax -W y A) x
=0 (10.25)
momentum transfer in flat geometry with constant
This expression yields Equation 10.21 when viscosity and transfer area.
treated to a limiting procedure similar to that used 2,3.* Equation 9.31 applies directly
on Equation 10.16. The constant g c may be in-
dv
serted in the integrated form of Equation 10.25 to Ty9c
make it identical to Equation 10.21. Thus, for the dx
geometry, the basic expression for simple
flat-slab 4.* Boundary conditions for this system
momentum transfer may be obtained either from a are:
force balance or from the generalized transport
equation for rectangular coordinates. The differ- at x = x2 = 2 in., v = v2 = -35 ft/min = -2100 ft/hr

ential balance is easily applied to situations in-

atx=x 1 =0in..* = Ki= 10 ft/min = 600 ft/hr
volving laminar flow in a straight-line path. the
If

path is curved, as in a sphere or in the angular Integration of Equation 9.31 gives

direction in a cylinder, accelerative forces caused (v 2 - v-i

by the path curvature are present, and the simple r yUc

y g, :10.28)
differential balance is difficult to apply. In this
(x 2 -Xi)
case, it is preferable to use the general equation of Substituting the boundary conditions and the value
motion developed in Chapter 1 1 for the fluid viscosity gives the stress on either
With the five-step procedure, the flux r y g c of plate.
Equation 10.20 can be replaced with a gradient by
using Equation 9.31 (363)(-2100-600)
Tv = = 1.41 x 10- 2 lbf/ft
2
(4.17x10 8 )(2/12-0)
-ixA
dx The stress is positive because momentum is
= ! 10.26) transferred in the direction of decreasing velocity
dx
(b) Steps 1 2, and 3 are the same as for part a.
,
If /i and A are constant. Equation 10.26 may be 4.* Equation 9.31 is integrated between
written
Uf =Xi , ¥ = v-\ ) and (x = x and v = v) to give
d2 v Tv9c
10.27) V—V^ = - (x -Xt ) (10.29)

Generally, the viscosity

vary with veloc- may
ity (in non-Newtonianand area of transfer
fluids)
may vary with distance. The variations must be *The numbered sections in example solutions refer to the
expressed in terms of velocity or distance before five-step procedure described earlier, insofar as possible. Often step
the integration can be completed. 5 is not relevant.
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 195

Eliminating Tv gc l[x between Equations 10.28 and Heat Transfer

10.29 gives For heat transfer, Equation 10.19 becomes

v-v-[ = (i/ 2 - 1^1

L—
\x 2 -x v)>
i 10.30)
dx
(10.31!

or, the velocity at any point x for the stated values That the rate of conduction
is, is constant in the
of v-\,v 2 ,xy and x 2 is x-direction. For step 3 of the five-step procedure,
Equation 10.31 may be combined with Equation
„=10 + (-35-10)(^|)=10-(45)(|) 9.21 to give

d(-kA dT/dx)
Therefore, at = 10.31a)
dx
x = 1/2 in., v= 10 - (45) (~-j = -1.25 ft/min For the flat -slab geometry A is constant with
x. may be further assumed that k is constant
It

At with T. Then Equation 10.31a becomes

d2 T
x = 1 .5 in., v = 1 - (45)
(-j-J
= -23.75 ft/min -kA —
dx
7
2
= (10.31b)

Step 4 involves integration and application of

boundary conditions. Typical boundary conditions
Example 10.2. Determine the shear stress and for conduction in the flat slab shown in Figure
velocity the conditions of Example
profiles for
10.1a would be
10.1 where the fluid between the plates is a
Bingham-plastic non-Newtonian liquid described 1. AtX=X! T-Ty
by Equation 9.65 with ju fi = 363 lb/ft hr and
2
2. Atx=x 2 T=T2
r =0.01 Ib/Vft .

That the temperature is known at both surfaces.

is,

Solution Two boundary conditions are needed because

1. The physical model isfor simple mo- Equation 10.31b is a second-order equation that will
mentum transfer in a flat-slab geometry with the yield two constants when integrated. Integrating
fluid described by Equation 9.65. once gives

dv dT
T y 9c -To9c Me -kA—
dx
= C,
dx

2. Equation 10.21 was derived without re- The second integration yields
gard to fluid properties, so applies to non-
it
-kAT=C^x + C2 (10.31c)
Newtonian fluids as well as Newtonian fluids.
3. This expression for ry g c is inserted into Applying the first boundary condition to Equation
the balance equation (Equation 10.21). 10.31c gives
4. The resulting equation is integrated to ob- -kATy =C x +C 2 (10.32)
1 1
tain
and the second boundary condition gives
v2 — v, ft Ib/hr
r y gc ~ T gc = -Me " = 5,880,000
3 -kAT2 =CiX 2 +C 2 (10.32a)
x 2 -X! ft hr

and Solving Equations 10.32 and 10.32a simultaneously

leads to
Ty - r = 0.0141 Ibf/ft
2
(See Example 10.1)
2
(T2 -T y )
so that Ty = 0.0141 + 0.01 = 0.0241 Ib^/ft Thus, .
Ct -kA 10.32b)
the shear stress is constant, but at a higher value (x 2 -Xt)
than that in Example 10.1 to overcome the yield and
value t The velocity Bingham
profile for the
.

T2 -T,
plastic is identical to that for the Newtonian fluid = -kA{ x: 10.33)
in simple transfer. x2 - x.
196 MOLECULAR AND TURBULENT TRANSPORT

Substituting Ct and C2 into Equation 10.31c gives a The variables r and T are separated and the
relationship for the temperature T at any position resulting equation is integrated with constant q.
x. :T
/, dr
-=-
:

7-2-7-! q\ I kdT
- t, A T,
T=T, + (x Xi) :
10.34)
x2 - *1 Since A = 2-nrL for L feet of pipe length, and
/r= 0.5 + 5x 10- 4 7",
Equation 10.34 is an equation of a straight
line; the temperature is a linear function of dis- r2 dr ;T,

tance in a flat slab with constant thermal conduc- q\ ^—7=-! (0.5 + 5 x 10- 4 r)</7-
r, 7-nrL t,
tivity. The heat flux may be determined by differ-
entiating Equation 10.34 with respect to x and Q r2
-7-
(ln = -[(0.5)(7-2 1 )
substituting the result into Equation 9.21.
)

2-nL

dT T2 -T |5x 10" 4 _ 1

dx x2 —
:

x-\
+ (^ \,
)(t2 2
,
-tS)\

and Then for 1 ft of pipe

Q
— = —k
.
—
7"
2
7"!
(10.35)
2tt[(0.5)(350- 100)
+ (5x10- 4/2)(350 2 - 100 2 )]
A x 2 -Xt
Q =
3/12
This shows that q is constant for all values of x ,n|
1/12
between x-| and x 2 .

Equation 10.35 can be obtained more directly = 877 Btu/hr

by integrating Equation 10.26 to obtain q = con-
This problem might also be solved using a mean
stant, and then integrating Equation 9.21 with q
area. The mean area for cylindrical geometry is
constant. The longer procedure was included be-
given by Equation 10.12
cause it illustrates a useful method to be used again
later. (r 2 -r,) (2H3.14H3/12-1/12)
A=A lm = 2nL
3/12
Example A
steam pipe 2 in. in ouiside
10.3. 1/12
diameter has an outside surface temperature of 2
= 0.955 ft for 1 ft of pipe
350° F. The pipe is covered with a coating material
2 in. thick. The thermal conductivity of the Then because k varies with T
coating varies with temperature such that
-7\
k = 0.5 + 5 x 10" 4 T where T is in degrees Fahren- A kdT
I

2
heit and k Btu/hr ft (°F/ft). The outside sur-
in •7",
=
face of the coating is at 100° F. Calculate the heat q
r2 -r.
loss per foot of pipe length.
and
Solution (0.955) [(0.5X350 - 100)
1. The physical model is for simple heat
+ (5x 10- 4/2)(350 2 - 100 2 )]
transfer (conduction) radially in cylindrical co-
ordinates. (3/12- 1/12)
2. The balance equation for cylindrical co- = 877 Btu/hr
ordinates q - constant with radius.
is

3. Equation 9.21 is written for radial trans-

Mass Transfer
port and substituted into the balance equation. Let
1 designate the inside of the insulation and 2 the The basic molecular-transport mass balance is ob-
outside. tained from Equation 10.19. For species a in a
4. Boundary conditions are then binary mixture of a and b

Atr = r1 =1 in. F=F =350°F dNa

1 = 10.36)
Atr = r2 =3in. 7"=7"2 = 100°F dx
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 197

and for species b be identified as the total rate of transfer by the

two mechanisms (Na ) t , so that
d_Nt
= (10.37)
dx (Na ) t =Na +Ntf ca /c t (10.41)

These two equations can be integrated to give Na = A similar balance for component t yields
constant and Nb = constant. It should be empha-
sized that Equations 10.36 and 10.37 are based
W b ) t =Nb +N tf c b /c t (10.42)

upon Equations 10.19, so that they apply only If Equations 10.41 and 10.42 are added,
when molecular transport (diffusion) is the sole
mechanism for mass transfer.
(A/ 3 ) t + (Nb ) t =N a +Nb + N tf (c a + cb )lc t
(10.43)
A more general mass balance considers bulk
flow as well as molecular transport. As the name As shown Chapter
9, the rates of molecular
in

implies, bulk flow involves the general movement transport the two species are equal but in
of
of the fluid phase as a whole. A mass balance opposite directions. Thus, combining Equations
including both mechanisms may be written around 9.14, 9.16, and 10.43 gives
the volume element shown in Figure 10.1a, as
shown in Figure 10.3. Application of Equation
N tf = (Na ) t
+ (Nb ) t :
10.44)

10.15 to the volume element gives The total bulk flow rate is therefore equal to
the sum of the total rates of transfer of the two
* c components by both mechanisms.
N
a/
tf~ + Na = Ntf- + /Va
Equations 10.41 and 10.44 are combined to
x Ct x+Ax Ix+Ax
give the general expression for mass transfer for
(10.38) component a.
where (Na =Na + [(Na
) t ) t + (Nb ) t ] ca /c t (10.45)
Ntf = rate of total bulk flow, lb moles/hr This equation is used in the following discussion.
ca = concentration of a, lb moles/ft3
Two types of simple mass transfer are usually
c t = total molar concentration lb moles/ft 3
defined. When only molecular transport is present,
(a constant at constant T and p)
the process is called equimolar counterdiffusion.
Rearranging, dividing by Ax, and taking the As shown in Chapter 9,
limit as Ax -> 0, gives
Nh = -N a (9.16)
dN,a d{N tf ca lc t )
|
_ 10.39) and hence the name for this transfer process. Equa-
dx dx tions 10.36 and 10.37 indicate that Na and Nb are
Integration of Equation 10.39 gives constant with x. With only one mechanism opera-
tive (i.e., no bulk flow), (Na t = Na and (Nb t = ) )

Na + Ntf ca lc t
= constant ( 1 0.40) Nb so that Equation 10.45 reduces to (Na t =N
,
a ) ,

The term
Equation 10.40 is the rate of
first in as expected.

molecular transport of a, while the second term is Equimolar counterdiffusion may occur in
the rate of bulk flow of a. Thus, the constant may distillation. Mass transfer occurs between the liquid
on a distillation plate and the vapor bubble rising
through it. If the molar latent heats of vaporization
of the mixtures are constant and the system
Bulk flow of a Bulk flow of a adiabatic, the moles of more volatile component
vaporized must equal the moles of less volatile
Ca
"if r
|
x
AJ
"if c j+4i component condensed. In such a case, the more
Molecular Molecular volatile component diffuses in the direction
transport transport opposite that of the less volatile component but at
of u in
atx = atx + Ax = the same rate. Equimolar counterdiffusion was dis-
Na \
x Na|x+dx cussed in Chapter 9, where it was shown that if the
temperature and total pressure are constant, a
X X + A* gradient in component a will of necessity result in

Figure 10.3. Mass balance with molecular transport an opposite gradient in components. Example 9.1
and bulk flow. involves equimolar counterdiffusion.
198 MOLECULAR AND TURBULENT TRANSPORT

The second type of simple mass transfer is boundary. Thus

called diffusion through a stationary gas.* It in-
[Nb ) t
=0 (10.46)
volves both molecular transport and bulk flow.
Diffusion through a stationary gas occurs when one Furthermore, a mass balance on component a
boundary of the diffusion system is permeable to shows that the total rate of transfer of a must be
only one component. As a result, there is no net constant with x.
movement of the other component, which is said = constant
(N a ) t 10.47
be stationary. In gas absorption, diffusion
!

to
through a stationary gas occurs. For example, con- Combining Equations 10.45, 10.46, and 10.47 and
sider the absorption of ammonia from an air- (Na gives
solving for ) t

ammonia mixture by water. The boundary be-

tween the gas and the water is permeable only to
(/Va)r=/Va /(1 ~c a /c t ) 10.48)

ammonia, because air has a negligible solubility in It can be seen that the total transfer rate is greater

water. Therefore, ammonia diffuses from the bulk than that due to molecular transport alone. Bulk
of the gas through stationary air to the water flow increases the rate of transfer.
surface, where it is absorbed. There is no net trans- Elimination of Na between Equations 9.11
fer of air. and 10.48 leads to an expression that may be
Diffusion through a stationary gas described is integrated to give
by Equation 10.45, where component a is taken as
(/V a )r ln(c 62 /c 6 i)
the component that can pass through boundary 2 = — '/ c t 10.49)
shown in Figure 10.4. This figure shows typical A (X 2 -Xy)
concentration profiles for the two components.
where cb2 - c —
t ca .

For example, boundary 2 might be a gas-water

Equation 10.49 is the basic equation describ-
interface; component a, the ammonia; and com-
ing diffusion through a stagnant gas. Because of its
ponent b, the air, as described in the previous
importance, it will be derived again in the follow-
paragraph.
ing paragraphs from a detailed consideration of the
Component b is not transferred through the
physical mechanism.
plane at x 2 As a result, a mass balance on compo-
An alternative derivation of Equation 10.49
.

nent b shows that its total rate of flow at any x

considers the physical effect of the diffusion and
must be zero at steady state. Otherwise, compo-
bulk flow of both components. In considering the
nent b would accumulate against the impermeable
transfer of components, it is helpful to look first
at a volume element of the gas itself, without
regard to the movement of the volume element
with respect to a stationary coordinate system. If a
gradient in component a exists within the volume
element, molecular transport must occur according
ZlL to the transport equation

Diffusion of 6

Boundary
Na = _
A
^ dx
0.1 1:

permeable At aconstant temperature and total pressure, the

only to
Bulk flow of a and b total concentration (or molar density) (c t is con- )

component a
stant. Then, as shown in Chapter 9,

ct =c a + cb (9.14)

and
Diffusion of a
dc b dc a
(9.15)
dx dx
Figure 10.4. Concentration gradients in diffusion Equation 9.15 shows that a gradient must exist in
through a stationary gas. the concentration of component b if one exists in
the concentration of a. The gradient will be equal
*Most cases of interest involve gases, but the following in magnitude but opposite in sign to that of a, as
development also applies to liquids. shown in Figure 10.4.
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 199

Because a gradient exists in component b, it and

will diffuse to the left opposite to the diffusion of
Moles of gas a C_a
component a. The flux of b is 10.53)
moles of gas b
Nb =
dcb
Combination of Equations 10.51 and 10.52
(9.13) gives
A dx

Combination of Equations 9.11, 9.13, and Bulk flux of a and b = [

—
cb
— | (10.54)
9.15 shows that for molecular transport

Nh = -N Combination of Equations 10.51 and 10.53 gives

a (9.19)

for the volume element of gas. Therefore, equi-

Bulk flux of a = 10.55)
molar counterdiffusion by molecular transport
occurs at every point in the gas. Because boundary
Ji
2 is impermeable to b, however, another transfer
The bulk flow of the gas in the +x-direction contri-

mechanism occurs that just balances the transport butes to transfer of a; thus, component a is trans-
ferred toward the boundary by bulk flow accord-
of b, as shown below.
ing to Equation 10.55 and by diffusion according
The gradient in b shown in Figure 10.4 results
to Equation 9.14. The total rate of transfer of a
in the transfer of b away from boundary 2. This
toward the boundary is the sum of the two
boundary is permeable only to a and no b can be
supplied acrossTherefore, another transfer
it. Nb- Na
mechanism must supply b at the same rate it is (10.56)
A cb It A
being removed by molecular transport. This new
mechanism \sbulk flow — that is, the movement of By Equation 9.16 the diffusional flux of b is
a volume element of the gaseous mixture toward numerically equal but opposite to that of a, so
the impermeable boundary. Therefore, any mole- Equation 10.56 becomes
cule in the gas is subject to two motions: the
normal molecular transport due to concentration
N
t' v a
a\ ,
<-a \l'/V*»
v a\ N
'* a

gradient and bulk flow in the +x-direction due to

depletion at the boundary. The volume element of or
gas is made up of the mixture of a and b, which
varies with position. The rate of transfer of b
Mas,
toward boundary 2 must just balance the rate of
transfer away from the boundary, to maintain a
(5H*J)®
It is important to note that (Na /A is the flux due
steady state. That is, )

to molecular transport only and that (Na /A) t is

Rate of rate of
due to both transport and flow. Substitution for
bulk flow of b molecular transport of b
Na /A from Equation 9.11 into 10.57 gives
(10.50) N r
dc
— — 9—
ca ] r . a 1
+
(51- L
1
cb \ L dx
The minus sign indicates that the bulk flow is in a
direction opposite to the molecular transport. or
Since the rate of molecular transport of b is

Nbl from Equation 10.50, (£)

A --•(*)(£)
Cb dx *
<10.58,
t

bulk flow of b N, This the general equation for transfer through a

is
Bulk flow of b = 10.51:
A stationary gas. From a simple mass balance it can
be shown that (Na /A) t is constant. But c b varies
In Equation 10.51, the bulk flow of b was con- with c a as shown by Equation 9.14. Equation
,

sidered, however the gas in the volume element 10.58 may be integrated if the variations in 9 and
consists of both a and b. By Dalton's law,
A are known.
The total bulk flow of a and b may be shown
Total moles of gas _ c t
to be equal to the total transfer of a and b, as
10.52)
moles of gas b cb follows. Since a b N =—N
Equation 10.56 may be ,
200 MOLECULAR AND TURBULENT TRANSPORT

written as follows: p 4ft *\

/
(?).-£
Na
K"?)
Na
2

J
ft
Stationary air

__± -Liquid level

\ ic_
— (10.59)
(t)-£)(-?)-©(3 Methanol

Comparison of Equations 10.54 and 10.59 shows

that the bulk flow of a and b is equal to the total
transfer of a and that transfer through a stationary ^ *
6 ft

gas is greater by a factor c t /cb than transfer by

equimolar counterdiffusion. Figure 10.5. Cylindrical tank in Example
10.6.
Equation 10.46 states that the total transfer
of b, (Nb { is zero. There is no net transfer of b,
) ,

because the bulk flow balances the molecular trans- The tank is tapered, as shown in Figure 10.5. The

port. air within the tank is stationary, but circulation of

For the case where and A are constants, air immediately above the tank is adequate to

Equation 10.58 can be integrated analytically. assure a negligible concentration of methanol at

Since dc a /dx = —dc b /dx, Equation 10.58 becomes this point. The tank and air space are at 77° F and
upon integration 1 atm. The diffusivity of methanol in air at 77°
and 1 atm is 0.62 ft 2 /hr. Calculate the rate of loss
M.
I7I
:

*= i/c, r 2
^ of methanol from the tank at steady state.

0. (x 2 -x 1 )
= c/c f Inl — Solution
1. This is a case of methanol diffusing
through 2 ft of stationary air from the surface of
This expression is identical to Equation 10.49. A
the liquid methanol to the top of the tank. The
more conventional form of Equations 10.49 and
transfer area varies. The concentration of methanol
10.60 may be developed.
in the air at the surface of the liquid is determined
The logarithmic-mean concentration is de-
from the vapor pressure of methanol at 77° F,
fined as
which is p a = 135 Hg. mm
C b2 ~Cb1 By the law c a =
Cb, (10.61) 2,3. perfect-gas p a IRT, so
C b2 Equation 10.58 may be written as
In

A/, JP dp a
r

Combining Equations 10.60 and 10.61 gives ! 10.64)

RTp b dx
(NA =ry c t (c b2 -cb :)
(l062) 4. To integrate this equation, A must be ex-
A t cbm (x 2 - Xt ) pressed in terms of x. Geometric considerations
Since ca1 +c61 = c t = c a2 + c b2 , Equation 10.62 show that
may be written as
A=-(6-x) 2
j-
c t (ca2 -ca1 ) 4
(10.63)
Cb m(*2 ~Xi) where x = is the surface of the methanol and

This equation applies only where 9 and A are x = 2 ft is the top of the tank. Furthermore,
constant. It should be noted that at steady state P b = P — P a Then Equation 10.64 may be written
-

(A/ a f must be constant, but that N

as
)
a for molecular
transport alone varies with distance, as indicated dx P dp a
by the curved concentration gradients in Figure r (7T/4H6-X) 2
RT
-Pa2

Pa\ P-Pa
10.4.

filled
Example 10.4. An open cylindrical tank is
to within 2 ft of the top with pure methanol.
W.)f ~)
4
6 — x2 6 — X!
- —
QP
RT
.

In
j

y
P-Pa2
P-p*
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 201

Now xt = 0, x 2 = 2 ft; p a2 = 0, p a1 = 135 Hg; mm Internal

generation = <£'

P= 760 mm, R = 0.7302 ft 3 atm/lb mole °R, T =

537° R, and

(0.62)(1: / 760-0 \

'760- 135/
and

(/Va ) f
= 0.00286 lb moles/hr

Equation 10.59 may be written in terms of

pressure and rearranged to give

Na =j(Na )t
Figure 10.6. Transfer with internal generation.

At the surface of the methanol, 61 =760- 135, so

760- 135 or
Na = (/Va = 0.823(/Va
760
) f ) r
d(4/A) = VdV 10.68)

About 82 percent of the transfer of a at the meth- This is the general transferent-property balance
anol surface is due to molecular transport. At the equation for internal generation. It merely states
top of tank p b2 = 760, so that all of the transfer of that the increase in rate of transfer across a volume
a is due to molecular transport alone.
element is equal to the rate of generation within
the element. If # is constant for the entire volume
between planes 1 and 2, the equation may be
TRANSFER WITH INTERNAL GENERATION
integrated to
It is possible for some of the transferent property
to be generated within the system. It is assumed (^) 2 -(^) =<SV 1 10.69)
here that the transferent property is generated uni-
formly at all points in the system. The transferent
where the volume V=LAyAz. The transferent
property generated within the volume may leave
property is continuously generated and must be
through both planes, as determined by the prop-
continuously transferred to a boundary to main-
erty gradient. If one plane is impermeable to the
tain steady state. Obviously, the rate of transfer is
transferent property, all the internal generation
not constant with distance but increases as the
must leave the volume through the other plane.
boundary is approached.
Equation 10.68 may be integrated when the
A system with internal generation is shown in
variation of A and V are known. It applies to mass,
Figure 10.6. The transferent property is generated
heat, and momentum transfer, as shown in the
at a uniform rate of ^ units of property per unit
following sections.
time and unit volume. A property balance can be
made around the element of volume of thickness
dx. The rate of input to the element plus the rate
of generation must equal the rate of output, or Momentum Transfer

i\pA) x +VAV=(^A) X + AX (10.65) As a fluid flows through a duct because of a

pressure difference in the direction of flow, shear
where the volume A V = A Ax and A = Ay A?. forces exist between layers of the fluid, as well as
Equation 10.65 may be rearranged and div- at the surface of the duct. These shear forces
ided by Ax to obtain represent a flux of momentum. The shear forces
increase from zero at a central axis to a maximum
= <SA 10.66) at the duct surface, and the momentum flux thus
Ax increases from the axis to the surface. This in-

Taking the limit as Ax -* gives creasing flux of momentum results from the inter-
nal generation of momentum in the fluid, as the
d(\!/A) energy represented by the pressure differential
= <$A 10.67)
dx is converted into momentum. For momentum
202 MOLECULAR AND TURBULENT TRANSPORT

an expression for the internal momentum gener-

ation terms of the forces acting upon the volume
in

element. This will be accomplished by writing a

force balance.

->(i). — flfi...
The pressure of the fluid at planes / and at
y + Ay produces forces acting on the volume ele-
ment (Figure 10.7). Pressure is exerted in all
directions at any point in the fluid; however, the
pressure exerted upon the specific face of a plane

\ 4>
may be considered directional. For example, the
pressure force exerted upon the external face of
plane / over an area S is equal to the pressure times
the area upon which the pressure is applied

Figure 10.7. Internal momentum generation.

(FvV L =(PSl
>y 10.73)

and is +/-directed. The pressure force acting upon

the external face of plane / + A/ is

transfer Equation 10.68 becomes

(~Fy 'y+Ay = (-PS) y+Ay
v+ (10.74)
d(T y g c A)= %dV (10.70)
because it is -/-directed. Here P= pressure of the
where the volumetric rate of generation of
<3
r is
fluid; S= area over which the pressure is applied =
momentum. In many applications, the shear stress AzAx.
is not easily measured. It is therefore desirable to
No other forces are acting upon the volume ele-
relate it and the generation rate to a more easily
ment. Note that the transfer area (A) is perpen-
measured quantity, as shown in the following dis-
dicular to the flow area (S). Pressure acts in all
cussion. It will also be shown that the internal
directions in a fluid. Here, Equations 10.73 and
generation rate is related directly to the pressure
10.74 define the external forces that are applied to
drop in the flow direction.
the chosen volume element by the presence of a
The external forces acting upon the volume of
pressure. The external forces are balanced by
fluid are shown in Figure 10.7. The fluid at the
equivalent internal pressure forces.
plane at x + Ax is moving at a lower velocity than
A total force balance around the element
at plane x, and therefore a negative velocity gradi-
states that the sum of the external forces must
ent is present at all values of x within the element.
equal zero at steady state. Therefore, a balance of
The element is chosen so that there is no change in
the four external forces acting upon the volume
any property with z. According to Chapter 9, a
element gives
stress in the +/-direction occurs on the face of
(Fy) X + (-Fy) x + Ax + iFy)y + (-Fy)y + Ay =
a plane into which momentum is transferred.
Momentum is transferred into the external face (10.75)
of plane x, and therefore, a +/-directed force (F y x )

Substitution of Equations 10.71, 10.72, 10.73, and

is present over the area = Az Ay. A
10.74 into Equation 10.75 gives
[Fy ) x =(r y A) x (10.71)
(t v A) x - (t y A) x + Ax + (PS) y - (PS) y + Ay =
At x + Ax, momentum is leaving the external face
of plane x + Ax and therefore a —/-directed force
(r
y
A) x + Ax - (r y A) x = -[(PS) y + Ay ~(PS) y ]

is present over the area A (10.76)

' — 'y'x+Ax \~TyA x+Ax ;10.72) As Ax -* and as Ay -* 0, Equation 10.76 reduces

to
In simple transfer, it has been demonstrated that
d{ry A) = -d(PS) (10.77)
the momentum transfer in a system of this geo-
metry is constant with x. In this example, the rate for the differential distances dx and dy. This
of transfer at x + Ax is greater than the transfer atx equation states that the change in shear force (or
by an amount equal to the rate of internal gener- rate of momentum transport) in the x-direction
ation. Therefore, the generation should be included across a volume element is equal to the change in
in a momentum balance. It is necessary to obtain pressure force across the element in the /-direction.
 DIFFERENTIAL MASS. HEAT. AND MOMENTUM BALANCES 203

If S is constant and the right side of Equation pressure of the fluid across the length L. Equation
10.77 is multiplied by dy/dy, 10.81 may be used with the following substitutions

dP dP V=SL 2
AP = P7 -P,
d(T Y A)=- — Sdy = dV (10.78)
=irr L
dy dy dV=2itLrdr
whereSdy = dV A = 2irrL
Combination of Equations 10.70 and 10.78

= -9c
dP
- i
gives
&T = —gc —
AP
Ay
(incompressible fluid)
"<r (10.79)
dy
Then
This equation shows that the rate of generation of
dv
momentum is equal to the decrease of pressure
with distance in the /-direction. The pressure gradi-
dl-2TrL^r — )
=
±P9c
—IvLrdr (10.82)
dr '
Ay
ent can be easily measured in most cases. In the
Integrating once gives
common engineering application of flow in pipes,
the pressure gradient is maintained with pumps or dv *P9c 2
compressors. The loss in pressure with distance -2-nLfir TrLr + C, (10.83)
dr Ay
represents a loss in the mechanical energy of the
fluid in the absence of other energy changes. At the center of the pipe the velocity is a maximum;
If the fluid
incompressible, the density and
is
therefore, at r=0 dv/dr=0, and from Equation
velocity will be independent of pressure, and §T is 10.83 C-[ = 0. Rearrangement of Equation 10.83
a constant gives

AP
'>t
=
-9c—- (10.80) a — = APg
dv c
r 10.84)
Ay dr 2 Ay
Substitution of the transport equation into which shows that the velocity gradient varies linearly
Equation 10.70 gives with radial distance. By substitution of the transport
equation, Equation 10.84 becomes
d[-vA—)='r T dV >
(10.81)
'
dx ±Pgc
(r y g c = 10.85)
which may be integrated if the variations in A and
)

2 Ay
\/are known.
the derivation discussed previously, the ex-
In For the boundary at r-\

ternal force upon each plane of the volume ele- APgc

ment was considered. A corresponding balance of {T ' 9ch= n 10.86)
-2~A~y
internal forces upon each of the planes could have
been written with the same result because each Combining Equations 10.85 and 10.86 gives
plane has associated with it two forces on opposite
Ty9c r
sides of planes, equal in magnitude but opposite in = (10.87)
direction. (T y g c ): n
The most frequently encountered application Therefore, the shear stress also varies linearly with
of momentum transfer with internal generation is
radius. It is zero at the center of the pipe.
the flow of fluids through a circular pipe. Consider An integration of Equation 10.84 gives
a pipe of inside radius r^ and length L with an
incompressible Newtonian fluid of viscosity \x (Fig- APgc r 7
ure 10.8). Pressure gages measure the difference in
v =
4 Ay n
+ & 10.88)

The integration constant C 2 may be evaluated

from the boundary condition at r = r y
v = 0. Be-
,

cause the pipe wall is stationary, the fluid particles

--ei immediately adjacent must have zero velocity.
•Av Therefore, with this boundary condition

Figure 10.8. Momentum transfer in a fluid in laminar c2 -^^Ll (10 89)

.

flow in a circular pipe. 4 Ay n

204 MOLECULAR AND TURBULENT TRANSPORT

and Equation 10.88 becomes of the resultant equations cannot be integrated

analytically.
A/to
v = (r^-r2 ) :
io.90)
4 A/m
Example An oil is in laminar flow in a
10.5.
Examination of Equation 10.90 shows that the 6 gal/min. The oil viscosity is
Vi-in. I.D. tube at
velocity distribution is parabolic.
300 centipoises, and its density is 60 lb/ft 3 Cal- .

It is usually more convenient to measure an

culate:
average velocity of flow rather than a point veloc-
ity as defined in Equation 10.90. By definition,
(a) The pressure drop per foot of pipe length
2
(lb,/in. ft).
the average velocity of flow is equal to the total 2
volumetric flow rate divided by the total flow area (b) The wall stress (lb f/ft ).

(c) The velocity at the center of the tube.

(S-i ).Then the average velocity v is defined
(d) The radial position at which the point
•si velocity is equal to the average velocity.
vS, = vdS (10.91
Solution. The physical model includes
Because S = nr2 dS = , 2irr dr, and S-\ = irr-\
2
radial momentum transport in laminar flow of a
Newtonian fluid in a cylindrical pipe. The neces-
2 = sary equations are derived in the text preceding
v{-nr-\ ) I 2-nvr dr 110.92)
•o this example. The average velocity is calculated by
Substituting v from Equation 10.90 into Equation dividing the volumetric flow rate by the flow area.

10.92 gives 6 gal v 1 ft

3
v min
, 1
|

min ''7.48 " 60s

vtirry
2\
-M^JV-'**
=
A Ay id -o
v=
-(1/2 x 1/12) 2
gal

ft
2
= 9.8 ft/s

= 2tt
-A/frc u n
*
4 Ay lx 2 4 (a) From Equation 10.94
and AP = 32vli (32)(9.8)(300x 0.000672)

_ -APgc rv Ak~ D 2 gc (1/2 x 1/12) (32.2)

2

v = 10.93)
8A/M = 1 130 lb f /ft
2
per foot of pipe
2
Since r\ = 0/2 where D is the pipe diameter, or = 7.84 lbf/in. per foot of pipe

v = —-
-APgc D 2
(10.94)
(b) From Equation 10.86
32 Ay n ±P9c
(*V0c)i=-
This equation expresses Poiseuille's law for laminar 2 Ay
flow in circular ducts. It is useful in calculating the
; 1130) (32.2X1 748)
pressure loss in laminar flow and in determining
the viscosity of a fluid using a flow-tube viscom- 2
eter.
ft-lb/s
An
equation that is useful in calculating the = 380
2
ft
point velocity from the average velocity is obtained
by combining Equations 10.90 and 10.93. and the shear stress is

r -
-r2 2' 380
=-— 2
= 2 = 2 rYK =1.18lb,/ft
10.95)
rv '-W :

32.2

For non-Newtonian fluids where the viscosity (c) From Equation 10.95 at r=
varies with velocity gradient, an expression for the
viscosity in terms of other variables must be sub- v = 2v
stituted before the general Equation 10.81 may be
[*-ef]
integrated. The viscosities of many non-Newtonian = (2)(9.8)(1 -0)
fluids do not vary in a simple manner, and many = 19. 6 ft/s
 DIFFERENTIAL MASS. HEAT. AND MOMENTUM BALANCES 205

(d) From Equation 10.95 when v = v Equation 10.82 is

— Apq
v = 2v\-\-[- d[2nLKr(-dv/dr) n ]
= ^^ £ r
2irLrdr
Ay
Integrating once gives
= — 7
!
Apgc irLr
-(im) 2 2irLKr(-dvldr) n =^T^- + Cl
r = 0.177 in.
Ay
To evaluate C-\ , the boundary condition of
maximum velocity at the pipe center is used: at
Example Derive an expression for the
10.6.
=
r dvldr = 0. Substituting this expression into
0,
laminar-flow velocity as a function of radius and
the above equation gives C-\ = 0. The equation can
average velocity for a power-law non-Newtonian
then be rearranged to give
fluid in a circular pipe.

Solution
-dvj -hpge \"»
1 The physical model is that of an incompres- dr 1
2KAy '

sible non-Newtonian fluid described by Equation The term (— Apg c /2KAy) is a constant, so the
9.67 or 9.69, where n may have any value. The variables may be separated and the equation inte-
fluid is in laminar flow in a circular pipe, so that grated to
internally generated momentum is transferred to
-Apgc Vn r
1+1/ "
the pipe wall by molecular transport. v=- \
+ C,
2. The momentum balance is given by Equa- 2KAy '
1 + 1//7

tion 10.70 for molecular transport with internal The term the brackets is always positive, because
in
generation. This expression was obtained from the AplAy always negative. The two minus signs
is
generalized balance on a finite volume element cannot be combined because the value of n is
(Equation 10.65). unspecified.
3. Equation 10.70 applied to flow of both
To evaluate C2 a second boundary condition
Newtonian and non-Newtonian fluids. The expres- (at r = r-| , v = 0) is used:
sion for momentum obtained from Equa-
flux is

tion 9.67. In dvldr always has a

a circular pipe, -Apge \ l,n r,** Vn
l
2
negative numerical value, so that Equation 3.67 '

2KAy '
1 + 1//I
may be simplified for flow in a cylindrical duct:
This is substituted into the previous equation
dv n-1 dv to give
Ty gc = K 1n
\"" 1 r
r
1 +1//7 _
-rr 1 +1 In
dr dr :

V =
\2KAy ''
1 + 1/f
A minus sign may be inserted inside of the
absolute-value brackets without changing the sign
The average velocity is obtained by inte-
of the overall expression. Now
always —dvldr is
grating this expression for v over the flow area:
positive for flow in a cylindrical duct, so the ab-
solute value brackets are not needed. Then the
previous equation becomes dr
2KAy
v =
dv\n 7T/Y
Ty9c = K\- ,
dr
This is integrated to give
This expression is substituted for ry g c in Equation Mn
10.70 to give
2 + 6n \ 2KAy '

d [KA(-dv/dr) n ]
= GrdV
Dividing this equation into the general expression
4. The integration of this expression follows for v gives
the same procedure as for the Newtonian fluid
shown beginning with Equation 10.81. The
earlier,
expression for a power-law fluid analogous to
206 MOLECULAR AND TURBULENT TRANSPORT

This is the final result. If n = 1 this equation , tutions may be made into Equation 10.97:
reduces to that for a Newtonian fluid (Equation
V=-nr 2 L
10.95), as expected.
5. If experimental data on velocity profiles dV=2nLrdr
and pressure drop for flow in a circular pipe are A = 2-nrL
available for a specific fluid, they may be used to
determine whether the physical model of a power- dT = dT
law non-Newtonian fits the fluid with sufficient dx dr
accuracy and to evaluate the constants n and K.
where L is the length of the cylinder.
The calculations might prove lengthy, but a com-
Equation 10.97 becomes
puter program could be easily prepared.

Heat Transfer

Examples of internal generation of heat include

'(-O
which integrates to
2ttl <>
q rdr ; 10.99)

nuclear-reactor fuel elements and electrically heat-

ed wires. The generation is not necessarily uniform
within the volume. When the generation is not dr
= itL
-2irkLr — 2
+ %r
110.100) d
uniform it must be expressed as a function of The temperature at the center of the cylinder must
distance or another variable. For heat transfer, the be a maximum if heat is uniformly generated.
bajance equation, Equation 10.68 becomes Therefore, one boundary condition is at r = 0,
T= T ,ordT/dr = 0.
dq = %dV ;10.96)
Substitution of this boundary condition into
where % is the internal generation rate for heat,
Equation 10.100 permits evaluation of C^
Btu/hr ft Substitution of qIA from the transport
.
-2irkLr(0) = -nLV g (0)
2
+ Ci
equation (Equation 9.24) gives
Therefore, C^ = 0.
dT Equation 10.100 may now be rewritten as
-kA = f'
q dV 110.97)
dx J

-*(^)=^
\dr 2 1
;io.ioi;
This equation may be integrated twice for flat-slab
geometry. which shows that the temperature gradient (dT/dr)
varies linearly with distance in the radial direction.

-kA —=
dT
dx
r
|
%dV+C :
Substitution of the transport equation gives

q = % r
10.102)
A 2
which shows that the flux also varies linearly in the
(10.98) radial direction. The flux at the boundary is

To complete the integration, the variations of k, A

10.103)
V, and yg must be known as functions of 7" or x in fill
the flat-slab geometry.
Combining Equations 10.102 and 10.103 yields
Another geometry of the cylinder.
interest is

This could be an electrical resistance wire or a (qIA) _ r

10.104)
nuclear fuel element. The heat generated uniformly (q/A), n
within the element is transferred in a radial direc-
This expression shows that the flux varies linearly
tion to the surface. The geometry here is the same
with r. Equations 10.102, 10.103, and 10.104 are
as that for fluid flow in a circular pipe. The
useful in understanding the mechanism of heat
generalized model of
molecular transport with
conduction with internal generation, but they tell
internal generation applies toboth cases; the final
us nothing about how 7" varies with r.
equations for conduction with internal generation
Further integration of Equation 10.101 gives
will be of the same form as those for laminar flow.
For constant #Q and k, Equation 10.97 may be Vnr2
integrated. For the cylinder, the following substi-
-kT = + C 10.105)
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 207

The integration constant C 2 may be evaluated by For T= 7"o at r = 0, Equation 10.106 becomes
applying either the boundary condition at the cen-
ter (T=T at r = 0) or the boundary condition at
%(0-r,
g
2
)

-k(T -T,)=
the outside surface (T=T at r = r-\). Using the
:
4
latter gives 2
Since k = 10 Btu/hr ft (°F/ft), 7"i
= 1500, and
n = 1/96 ft,
-kT, =-V- + C 2

or
-(io)ir -i 5 oo) J 4 1x1o8 "°- 1/96
-
»
2

Cn=-kT,
%n and

r = 261 2° F
Substituting this expression for C2 into Equation Therefore, the wire is molten
the center. at
10.105 gives (b) The molten core extend outward to will

-tfr-r,)--3 —
%(r 2 -r,
-1
2
—)

(10.106)
the radius at which T= 2500°F. Division of Equa-
tion 10.106 by Equation 10.107 gives

T-T, r
2
-r
Use of the center boundary condition would
instead give T -T -r 2 ~\r)
2500- 1500 /1/8\ 2
-k(T -T) = (10.107)
2612-2500 •ffi-
Either Equation 10.106 or Equation 10.107 may and
be used to calculate the temperature at any point r=0.04 in.
in the cylinder, depending upon the available data.

These equations show that the radial temperature Mass Transfer

distribution is parabolic in a cylinder with uniform
internal generation.
Examples of mass transfer with internal generation
can be found in systems in which diffusion and a
This is exactly analogous to the parabolic
chemical or nuclear reaction occur simultaneously.
velocity profiles in laminar flow in the came
As mentioned earlier, the generation and diffusion
geometry in the previous section.
of neutrons in a nuclear reactor illustrate mass
transfer with internal generation. Of more interest
Example 10.7. An electrical resistance wire to the chemical engineer is diffusion with a chemi-
has a melting point at 2500°F. The electrical input cal reaction. An example of this is the absorption

to a wire 10 ft long and % in. in diameter gives a of nitrogen dioxide by water to form nitric acid.
uniform volumetric heat generation totaling Absorption and reaction occur simultaneously, and
1,400,000 Btu/hr. The surface temperature of the both phenomena progress at a given rate. The
wire is 1500°F and the thermal conductivity is 10 nitrogen dioxide diffuses a short distance in the
Btu/hr ft 2 (°F/ft). liquid water before it reacts. When it reacts, it

(a) Can the wire be safely used, or will the disappears. This is mass transfer with negative
center reach its melting point?
generation.
Calculate the radius of any molten core
(b) The general balance equation may be re-
formed assuming that the properties of the molten written for mass transfer with internal generation
metal are the same as those of the solid. d{Na = )
N dV
<S (10.108)

Solution where &N = volumetric rate of internal generation

(a) This is a case of transfer of heat with of mass and Na = rate of transport. In the typical
internal generation. The volumetric rate of gener- case, #/v varies and must be written as a variable in
ation is terms of distance x or composition c a .

Consider the case of the equimolar counter-

1,400,000 diffusion of component a through a region in
'£.= = 4.1 x10 8 Btu/hr ft
3

(10)(7r/4)(1/4x1/12) :
which the component undergoes a slow first -order
208 MOLECULAR AND TURBULENT TRANSPORT

irreversible decomposition (a -*b). The rate of the In the first boundary condition, xi is arbitrarily set

decomposition is determined by the first-order equal to zero to simplify the final result. Then x 2
model, which states that the rate of change of becomes the distance from plane 1 to plane 2.
concentration of a is directly proportional to the These two boundary conditions are substituted
concentration of a, into Equation 10.114. For the first condition.

dc a
= -kcs (10.109)
c a1 = C,e + C 2 e~°
d6 so
where k is an experimentally determined constant. Ca i = C-^ + C2
The minus sign shows that the concentration
isdecreasing with time. The left side of the equa- For the second condition
is equal to the change in moles of a per unit
~
Cye**^* 1^ + C2 e- X ^rm /
tion
ca i =
time and therefore per unit volume. This is exactly
equal to the internal generation rate of mass Solving these two equations simultaneously gives
[&N )- that is

Ca1 e* 2NA7!7 _ Ca2

f
dc a C2 = 10.115)
-
SN ~ (10.110)
dd e *:V*/v _ e
-
Ikl's

Therefore, and

% = -kCa (10.111)
Ci =
Ca2 -CalC" 'kl'J

10.116)
The minus sign indicates negative generation — that
decomposition or disappearance.
is,
Substitution of Equations 10.115 and 10.116 into
Combining the transport equation (9.1 1 ) with
10.1 14 gives on rearrangement
Equations 10.108 and 10.1 1 1 gives

Ca2\e
xJkl 'J ,-xJkl '/ '

{-'JA ^)=- kca dV (10.112)

+C l
e ^ -x)sjkl'y
z _ e ~(x, -x) s/k/'s
Ca =
Taking and A constant and V = Ax for a slab (e>
'kl'J _ e -x 2
s/k/<s

geometry gives
(10.117)
This equation may be written in a shorter form by
d(-'JA —\ = -kc a (A dx) using the definition of hyperbolic sine
dx I

z
and
sinhz = ^(e -e-') (10.118)

d_tdCa_\ k Application of Equation 10.1 17 to 10.1 18 gives

dx\dx
or c a2 sinh(x^/ - /+c a1 sinh((x 2 - x)
2
d ca ca
ca = (10.113)
dx' sinh^x 2
yp
This is a linear second-order differential equation
[10.119)
with constant coefficients. By routine analytical
techniques, this equation is integrated to where the hyperbolic sine,

= Ci£? x N/*7^ + q2( 'kl'J

10.114) :sl kl'J _ Q-x>Jk/'J
sinh x
where Cy and C 2 are integration constants that
must be ev aluated from boundary conditions.
The two integration constants are evaluated Equation 10.119 may be used to calculate the
using the boundary conditions: composition c a at any point x between planes 1
and 2.
1. At x = (plane 1), ca = ca1 . Because component a decomposes to com-
2. At x=x 2 (plane 2), c a = ca 2- ponent b, component b undergoes internal gener-
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 209

ation and will also diffuse. This could be treated by 5. Mickley, H. S., T. K. Sherwood, C. E. Reed, Applied
the methods illustrated for component a. Mathematics in Chemical Engineering, 2nd ed.,
If the rate of the chemical reaction cannot be McGraw-Hill, New York (1957).
expressed by a simple first-order equation such as 6. Wylie, C. R., Advanced Engineering Mathematics, 2nd
Equation 10.109, the integration of the resulting ed., McGraw-Hill, New York (1960).

differential equation may be difficult. Additional

discussion of diffusion and chemical reaction may
PROBLEMS
be found in Reference 1.

10.1. Calculate the mean transport area for a thick

SIMULTANEOUS MASS, HEAT, AND MOMENTUM
spherical shell of inside radius r-\ and outside radius r-^.
TRANSPORT
Assume transport is in the radial direction from r-\ to r-±.

In many cases, two or even three types of transport 10.2 Repeat the derivation of Equation 10.13 for the
occur together. For example, if a flowing liquid is specific case of heat transfer through a cylindrical shell.
being heated in a steam-jacketed circular pipe, both 10.3. Develop an expression for the velocity profile
momentum transport and heat transport occur. [v = f{x)] for a power-law fluid (of known K and n) filling a
Momentum is generated in the flowing fluid and is space between two moving planes.
transported to the pipe wall. At the same time, 10.4. Repeat Example 10.1 in SI units. Convert the
heat is transported from the hot wall to the liquid. original data to SI units and then solve for an answer in SI
The several types of transport may interact units.
with each other, thereby requiring new quanti- 10.5. Derive the equivalent of Equation 10.34 for
tative relationships describing transport, unlike heat transfer in a cylinder.

those for any transport mechanism operating 10.6. Repeat the derivation of Equation 10.34 for
alone. For example, consider the heated liquid the case of a variable thermal conductivity: k = a + bT
described above. The liquid is hotter at the wall + cT where a, b, and c are empirical constants for a
than at the center of the pipe, so that a radial particular material. Sketch the temperature profile.
temperature gradient exists. The variation in 10.7. Explain Figure 10.4 in your own words. What
temperature produces a variation in liquid viscos- is the bulk flow of a at X2?
ity. As a result, derivation of the expression for 10.8. Ammonia is being absorbed from an air-
liquid velocity profile could not assume a constant ammonia mixture by a sulfuric acid solution. The concen-
viscosity, as was done in deriving Equation 10.90, tration of ammonia in the air 1 in. from the surface of the
so that this equation does not apply to a heated acid is 40 volume percent The concentration at the acid
fluid in laminar flow. surface is percent, because the ammonia reacts with the
Interaction of the various types of transport acid. The total pressure of the system is 400 mm Hg and

may produce very complex physical situations, re- the temperature is 60° F.
quiring complex physical models and lengthy (a) Calculate the rate of absorption of ammonia
mathematical analysis. Only the simplest cases are across 0.5 ft of acid surface.

considered in this book. Many examples of multi- (b) Calculate the concentration of ammonia 1/2 in.

ple transport involve more than one transport di- from the acid surface.

rection, so they will be considered in the following (c) Calculate the rate of transfer of ammonia by
chapter. molecular transport and the rate of transfer by bulk flow 1

in. from the acid surface.

(d) Repeat part (c) at the acid surface.
REFERENCES 10.9. A well located in the desert is 10 m deep to the
water level and 1 m in diameter. The stagnant air and the
water in the well are at 30°C and normal atmospheric
1. Bird, R. B., W. E. Stewart, E.N. Lightfoot, Transport
pressure. A slight breeze of dry air is blowing across the top
Phenomena, John Wiley & Sons, New York (1960).
of the well. Calculate the rate (kg/s) of diffusion at steady
2. Carslaw, H. S., and J. C. Jaeger, Conduction of Heat in
state of water vapor from the surface of water in the well.
Solids, 2nd ed., Oxford University Press, New York
(1959). (Assume the partial pressure of water vapor in the air at the

3. Dusinberre, G. M., Numerical Analysis of Heat Flow, surface of the water is equal to the vapor pressure of water
McGraw-Hill, New York (1949). at 30°C.)

4. Kreith, F., Principles of Heat Transfer, 3d ed., Intext 10.10. Calculate the rate of diffusion of sodium
Press, Scranton (1973). chloride at 20°C through a stationary film of water 1 mm
210 MOLECULAR AND TURBULENT TRANSPORT

thick, where the concentrations are 20 and 10 weight viscosity. One element is driven at a chosen speed, and the

percent, respectively, on either side of the film. stress to hold the other element stationary is measured.
10.11. Sulfuric acid is diffusing through liquid water The stress is a function of the liquid interface area, the
~6 separation, the speed of the driven element, and the
at a rate of 8 x 10 moles/m s. What will be the concen-
viscosity.
tration of acid 3 mm from a point where the concentration is

10.16. The viscosity of a liquid is to be determined

5 mass percent acid? (Assume the density of the solution is

with the following data taken on a Couette viscometer:

that of water and that the diffusion is one-dimensional. The
viscometer: r^ = 0.003 m, r 2 = 0.005 m, length = 0.100 m.
temperature is 20 C.)
The measured moment on the bob is —0.007 Nm, when the
10.12. A chimney brick is shown in the sketch. Heat
cup is rotating at 30 rpm. What is the viscosity?
is transferred from the 3-in. end to the 5-in. end. No heat is
10.17. An electrically heated resistance wire has a
lost through the other sides, because these sides adjoin
other bricks. The 800° F because it is exposed
3-in. side is at
diameter of 2 mm and a resistance of 0.10 ohm per foot of

to hot gases inside the chimney. The 5-in. side is exposed

wire. The thermal conductivity of the wire is 20 J/m s
to the atmosphere and is at 200° F. The thermal con- (K/m). At a current of 100 amp, calculate the temperature
2 difference between the center and surface of the wire at
ductivity of the brick is 0.4 Btu/hr ft (°F/ft). Calculate

the heat loss per hour through the brick. steady state.
10.18. Thin flat plates of uranium are used as fuel
elements in nuclear reactors. Heat is generated uniformly
within the uranium metal by fission. This heat flows to the
surface of the metal and is removed by a liquid coolant.
Consider a fuel element 3 mm thick whose surface temper-
atures are both 200° F. The volumetric heat generation rate
8 3
(
Q ) is 2 x 10 Btu/hr ft . Calculate and plot the temper-
ature profile across the 3-mm thickness, starting at one
surface and ending at the other.

Uranium: p= 11 55 lb/ft 3 ; k= 17.5 Btu/hr ft

2 (°F/ft);

5 in. -T
cp = 0.032 Btu/lb°F

10.19. Heat is generated within a sphere at 2 x 10

10.13. Steam at 260°C is flowing through a thick-
Btu hr ft The sphere is 3 in. in diameter. The surface
walled nickel pipe. The temperature of the outside pipe
.

0.025 m temperature is 200°F. The thermal conductivity is 200

surface is 90°C. The inside diameter of the pipe is
2
Btu/hr ft (°F/ft).
and the outside diameter is 0.030 m.
Calculate the temperature at the center of the
(a) How much heat is lost from 5 m of this pipe in
(a)

sphere.
24 hr?
(b) What is the temperature halfway through the (b) Calculate the temperature at r = % in.

10.20. For volumetric heat generation in a flat slab

pipe wall?
whose surfaces are at different temperatures as shown,
10.14. A spherical furnace has an inside radius of 3 ft
and an outside radius of 4 ft. The thermal conductivity of (a) Derive an expression for T in terms of x and the
2 constants of the system.
the wall is 0.10 Btu/hr ft (°F/ft). The inside furnace
2000" 175" F. (b) Derive an expression for the value of x at which
temperature is F; the outside surface is at
T is a maximum.
(a) Calculate the total heat loss for 24 hr of oper-
ation.
(b) What is the heat flux and temperature at a radius
of 3% ft?

10.15. A Couette viscometer (ry = 1 in., r

2 = 1.25 in.
and length = 4 in.) is filled with a fluid of viscosity 3000
centipoises.
(a) Calculate the moment on the bob if the cup
rotates at 20 rpm.
(b) Calculate the moment on the bob, if the bob
=
i
rotates at 20 rpm in the same fluid contained in a large tank
(r 2 = °°).

Note: The Couette viscometer consists of a cylindrical bob 10.21. A heavy oil is pumped through a pipe with a
in a cup that contains the liquid for measurement of its 2-in. inside diameter. The pressure drop over 10 ft of pipe is
 DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 211

10 psi. The viscosity of the oil is 200 centipoises and the 10.93. Let the channel be 2x-| high. Ay long, and Az wide.
3
density is50 lb/ft . The channel is much wider than it is high, so edge effects
(a) Calculate the volumetric flow rate of oil through may be neglected.
3 /min).
the pipe (ft 10.26. Equation 9.67 is a general equation for di-

(b) Calculate and plot the momentum flux profile latant, Newtonian, and pseudoplastic liquids, depending
across the pipe. upon the value of the exponent n. Consider a liquid flowing
10.22. A fluid of viscosity 0.0002 kg/m s is flowing between two stationary flat plates spaced 2 in apart.
between two flat plates 10 mm apart. The plates are 1 m by (a) Plot the shear stress and velocity profiles for each
1 m in area. The average velocity if 1 m/s. Calculate the of the three liquids.
pressure drop. Plot the velocity and the stress as functions (b) List typical dimensions for K for each liquid.
of distance across the space between the plates. Data: Assume that the velocity at the center (f ) is

10.23. 20 gal/hr of benzene at 70°F is flowing 1 ft/s and that K= 1 in dimensions appropriate for each fluid.

through a pipe of 0.957 in. inside diameter.

Dilatant: n = 2.0
(a) What is the pressure drop (psi) through 100 ft of
Newtonian: n = 1.0
pipe?
Pseudoplastic: n = 0.5
(b) What is the velocity (ft/s) at the center of the

pipe? 10.27. Develop an expression for the average laminar

(c) What ismomentum flux
the at the pipe wall? flow velocity of a Bingham plastic fluid in a circular pipe,

10.24. A common channel for fluid flow is the an- analogous to Equation 10.93, which is for a Newtonian
nulus, the space between two concentric circular pipes, as fluid.

shown. The inside of the annulus has a radius ry and the 10.28. Develop an expression for the laminar flow
outside, T2- velocity as a function of radius for a power-law fluid
flowing as a thin film down the outside of a circular tube.
Flow area The motive force is gravity.

10.29. Determine the pressure drop per 100 m of

pipe for applesauce flowing at the rate of 2 x 10
-3 m 3 /s
through a pipe of 0.02 m inside diameter. Assume laminar

flow. Applesauce is a pseudoplastic fluid with K = 0.5

N(s)" Im 2 and n = 0.65. Compare this with a Newtonian fluid
-

of the same K.
(a) Derive an expression for the velocity of flow v at
10.30. Show that if there is no reaction (k = 0) Equa-
any radius r in the annulus.
tion 10.119 reduces to an expression forc a at any value of
(b) Derive an expression for the radius at which the
x for simple diffusion, which is
maximum velocity occurs, in terms of r-\ and r^-

10.25. Derive an expression for the average velocity

Cg Cg 1 X
of flow between parallel flat plates of the form of Equation Ca2 _ca1 *2

Equations of Change
The mass, heat, and momentum transport equa-
tions developed in Chapter 10 may be extended
the more general case involving three dimensions
and time. These general equations, sometimes
called equations of change, are partial
the
differentialequations describing the relationships
among the physical variables, time and the geom-
etry of the system. The equations of change are:
1. The equations of continuity, based upon
the conservation of mass.
2. The equation of motion, based
Newton's second law of motion.
3. The energy equation, based upon the con-
servation of energy.
These equations are quite complex in
general form. An analysis of the physical situation
often shows that certain terms in the equations
may be neglected, resulting in an equation that can
be solved without complex mathematics. This
chapter serves as an introduction to a very compli-
cated subject, focusing upon the concepts behind
the development of the equations. Applications to
some commonly encountered problems will
demonstrate the utility of the equations of change.
The approach to the analysis of mass, heat,
and momentum transport presented in this chapter
contrasts with that developed in Chapter 10. In the
earlier chapter, a physical system was analyzed to
develop a transport equation for that specific situa-
tion. In this chapter, equations are derived that
describe transport systems in general. These general
equations may then be applied to specific physical
systems. An analysis of the specific physical system
will show that certain terms may be neglected, and
the general equations may be simplified to describe
that specific system. If the general equations are
Eleven
applied to the simple physical systems analyzed in
to Chapter 10, they will, of course, simplify to the
same equations developed directly in that chapter.
Time Derivatives
Before developing the equations of change, it is
necessary to define three different time derivatives
that are used in these equations.
In the most general case, there are four
independent variables: time (6) and three position
variables (x,y,z). The concentration of trans-
upon
feree property (D may vary with one or more of
these four variables. Thus, to define T exactly
requires the specification of 0, x, y, and z. It is
apparent, therefore, that
their
F = f{x,y,z,d) (11.1)
The partial derivative describes the increase in
T (thedependent variable) with respect to any one
of the four independent variables. Thus, 81790 is
the increase in F with time at a fixed position and
is called the partial time derivative. Sometimes the
partial derivative is written {dF/dd x ,y,z where the
subscripts indicate which variables are fixed. In this
text the subscripts will not be used. The partial
time derivative (81790) is the time rate of increase
of the concentration that would be seen by a
stationary observer.
If is not stationary, then the time
the observer
rate ofchange of concentration varies with
position. The time rate of increase of F as viewed
by a moving observer is called the total time deriva-
tive. From Equation 11.1 the total derivative is
-SM5MSMS)- 11.2)
212
 EQUATIONS OF CHANGE 213

It is understood that the bracketed terms in point 1 m from the inlet when
Equation 11.2 have the appropriate variables held (a) The thermocouple is stationary at the
constant. For example, 0I7dx) is OI73x) /fZ>e . point.
Division of Equation 1 1 .2 by dd gives (b) The thermocouple is moving past the
point at a velocity of 1 m/s.
+ + + (c) The thermocouple is moving at the fluid
~
dd \bx 'd6 \ dy' dd \ dz? dd \dd) velocity.

(11.3)

Equation 11.3 is the mathematical statement Solution. It will be assumed that flow is in
for the total time derivative. In this equation dx/dd, thez-direction only.
dyldd, and dz/dd are the velocity components (a) A stationary thermocouple measures the

of the moving observer. Thus, the total time deriva- partial time derivative dT/dd, which is given as
tive is the sum of four terms: one resulting from 3°C/s at the point in question.
the time rate of increase of V at a fixed position, (b) The thermocouple is moving at a velocity
and three resulting from the observer's motion. If 1 m/s independent of the fluid velocity, so the
the observer stops moving so dx/dd, dy/dd, and total time derivative is needed.
dz/dd are zero, dT/dd = 31780, as expected from For the z-direction, Equation 11.3 becomes
the definition of the partial time derivative.
A third time derivative is defined by con- dT = (dT\d£ dT
+
sidering a flowing fluid and having the observer dd~ \dz/dd dd
drift along with the current, at the same time
observing the time rate of increase of T. In this dT/dd = 3°C/s from part (a) and dz/dd is given as

instance, the observer is moving at the velocity of

1 m/s; furthermore,

the stream, and the velocity components are u, v, dT_d_

w in the three coordinate directions. Equation 1 1.3 [250(1 -e-° 1z
)] =25e-° u
dz~ dz
still describes this process, but the observer's
velocity components are now those of the stream. At z = 1 m.
Thus,
(-) = 22.62°C/m
9I ={ K )u +
{
K) v + (K) w+ lK) H1.4)
Therefore,

where DT/Dd is the substantial time derivative. It dT

sometimes = (22.62)(1) + 3 = 25.62°C/s
is called the derivative following the
dd
motion, or the Stokes derivative. When the stream
is not moving, the partial time derivative again (c) The substantial time derivative is required

results. when the thermocouple travels at the fluid veloc-

ity. Equation 1 1.4 gives for w= 0.3 m/s

Example A fluid flowing through a

11.1.
DTJdT\
circular pipe being heated so that a temperature
is
Dd ~\dz)
dzl
,w +
dT
dd
—
= (22.62X0.3) + 3 = 9.79°C/s
gradient exists from one end of the pipe to the
other. This gradient is expressed by
0.1*i
250(1 -e MASS BALANCE
where Tz is the temperature °C at a given point z, A mass balance around the differential volume
measured from the inlet, in the flow direction. The element shown in Figure 11.1 leads to a differen-
fluid velocity at the center-line is 0.3 m/s. At a tial equation that is often called the equation of

particular time, the temperature of the fluid is continuity. This equation is the mathematical
increasing at a rate of 3°C/s at a point 1 m from expression for the conservation of mass. The
the inlet. A
thermocouple that can move along the balance will first be made for
pure component a
centerline is available to measure the temperature. with simple transfer (no internal generation). For
Calculate the time rate of increase of the this case, the only mechanism of mass transport is
temperature measured by the thermocouple at a bulk flow because molecular diffusion can occur
214 MOLECULAR AND TURBULENT TRANSPORT

P".vL + Ay
position. Applying Equation 1 1 .5 gives

f»z i + fe
(ptv) x Az Ay + (pv) y Az Ax + (pw) z Ax Ay
= (pu) +
x Ax AzAy + (pv) y + Ay Az Ax

* f> v* x + Ax dp
+ (pw) z+Az Ax Ay + TzAx Ay Az (11.6)
30
f>Vz\ z j Ai
Rearranging Equation 11.6 and dividing
r.v, z^ P v y\y
through by Ax Ay Az yields

_ 9p _ (pu) x +Ax - ipu)x (pv)y+Ay ~ <P^/

30 Ax Ay
(pw) z+Az - (pw) z
(11.7)
Az
Figure 11.1. Three-dimensional volume element. Taking the limit as Ax, Ay, Az approach zero
gives

only in a multicomponent mixture of variable com- 3p 3 3 3

position. 30 3x by oz
As shown in Figure 11.1, mass is considered
volume element parallel Equation 11.8 is the equation of continuity
to flow into and out of the
to each of the coordinate dimensions. At the x-face
for a single component system. All flow systems

of the cube the fluid velocity and density are u and must satisfy it. This equation can be written in
vector form:
p x At the x + Ax face the velocity and density are
.

u Ix+a* and p x + Ax The product of velocity and

|
.
3p
density gives a mass flux. = -V •
pv :n.9)
30
In words, the mass balance on the cube is

where v is the velocity vector having components

Rate of rate of mass
rate of ,

1
- ,-. u, v,and w. The symbol V (del) is a vector oper-
mass input mass output accumulation ator (2).
Equation 11.8 may be rearranged to give an
The rate of mass input into the cube is
alternateform of the continuity equation by carry-
For the x-direction (pu) x Az A/ ing out the indicated differentiation of the pv

For the /-direction (pv) y Az Ax products.

For the z-direction (ptv)^Ax Ay 3p

— ~~u —
3p
30
p —+ v—+p
+H
du dp
H — + w—+ P
3x
dv dp
p
dw
dx dy dy dz
—dz
To obtain the rate, the mass flux is multiplied
by the appropriate area to which the flux is or
normal.

+ u — +v — + w — =p + —+ —
/3p dw\
The rate of mass output from the cube is
I—
\dd
3p
-
3p
dx
3p
H
dv dz'
\ /

\d*
—
du dv
aw
3x dy
ai/ dz
For the x-direction (pu) x + Ax Az Ay (11.10)
For the /-direction AyAz Ax
{pv) y +
The left hand side of Equation 11.10 is the sub-
For the z-direction (pw) z+Az Ax Ay
stantial time derivative. Thus, Equation 11.10 may
be written as
The rate of mass accumulation within the

+—+—
Dp
cube is

=P \
l
—
du dv dw^
(11.11)
'D~6 3x dy dz >

- Ax Ay Az
dp
or, in vector form

The use of a partial derivative is necessary

Dp
p 11.12)
because p is, in general, a function of time and DO
 EQUATIONS OF CHANGE 215

The equation of continuity can also be before (Equations 1 1 .7 and 1 1 .8) gives
developed for a binary system, by considering a
fluid of two components a and b flowing through
—
oc a
b(N'+c =
a u) b(N'+c
Y
-
a v)
- + +
the volume element of Figure 11.1. Because con- 90 dx by
centration gradients of a and b may exist across the
b(Na '+c a w)
volume boundaries, both molecular diffusion and $ N =0 (1M8)
bulk flow may cause mass to enter the volume. bz
Fick's law describes the molecular diffusion. The or
molar flux* of a entering the x face is '

dca 3/Vax' bNay 3/Va / d(c a u) b[c a v)

[ t

d£a
Na '\ x J
ab 11.13)
90 9x by dz bx by
bx
b(ca w)
and the total molar flux of a entering thex-face 4N =0 (11.19)
is
bz
Wa.Tot'l* =Na '
\
x +C3 U\ X (11.14) In vector notation,

The second term in Equation 1 1 .14 is the flux

of a entering the volume by bulk flow. Similar ^+
9c.

9a
V-/Va '+ V.(c.w)-ar W| = Od 1.20)
expressions can be written for the y- and z-
directions. Spec+es a may also accumulate with
time inside the volume. An equation similar to 11.20 results for
component/?
tea
Ax Ay Az 11.15)
90 ^+
do
V-/V+ v(c6 v)- ^, 6 =0 (11.21)

The possibility also exists for a to be gener-

These equations are the continuity equations
ated within the volume by chemical reaction. This
for a particularcomponent in a binary system.
generation can be expressed as
An
important special form of the continuity
Ay Az equation for a binary system can be developed, as
f
iv.Ax (11.16)
follows. Fick's law may be written in vector form
as
where ^ Ng is the moles of a produced per unit
volume per unit time. Na' = -®abVca (11.22)
Applying the conservation principle (Equa-
tion 1 1 .5) to the volume element gives For a system in which there is no bulk flow
and no generation, Equation 11.20 becomes for
Rate at which rate at which component a
species a enters species a is produced
9ca
rate at which which
rate at + V- A/ '
= (11.23)
species a leaves species a accumulates 90

[Na + ca u) x Ay Az + (Na + ca v) y Ax Az Combination of Equations 11.22 and 11.23

yields
+ (/Va '
+ ca w) z Ay Ax + 'S Ng Ax Ay Az
9ca
= {Na + ca u) x+Ax Ay Az = ®ab*-VCa =2ab Y 2 C3 (11.24)
90
+ (Na +ca v) y+Ay Ax Az
'

+ \Na + ca w) z+Az Ay Ax +
bCa
— Ax Ay Az
bo
where V 2
is the Laplacian operator defined as

(11.17)
9
2
< 9
2
c + 9V
^ 2 c„ = 7
2
+ 5 T 11.25)
bx by 2 bz 2
Rearranging Equation 1 1.17, dividing through

by Ax Ay Az and using the limiting process as Equation 11.24 is sometimes called Fick's
second law of diffusion. It applies to unsteady-
state transport in solids and to a binary fluid
*For simplicity Na = '
Na IA as used in Chapters 9 and 10. system where N a = -N b
216 MOLECULAR AND TURBULENT TRANSPORT

Example hollow sphere with per-

11.2. A Separating variables and integrating again
meable and outer surfaces
solid walls has its inner yields the general solution
maintained at a constant concentration cai and ca0
respectively. Develop the expression for the con-
ca = + C2
centration profile for component a in the wall at r
steady-state conditions. What is the flux at each
The boundary conditions are
surface?
1. At r=R, ca = c ai
Solution. The sketch illustrates the physical 2. Atr = fl 2 c a =cao

situation. Because of the geometry, the equation of

Using both boundary conditions gives
continuity is best used in spherical coordinates.
C,
Cai + C2
"1
and

cao - + C2
no
Solving these two equations simultaneously
gives

R-iR
"2
C\ = (cao - cai )
1

(/?2-/?l)
and
From the problem statement, at R-\ ,c a = c ai and at
R2, ca -c ao The equation of continuity for this R2
.

case is given in Appendix E.

C2 =c ai + (c ao -C ai
R 2 — R-\
Substituting these values of C-\ and C 2 into

be
= 0.'ab
r

Lr
1

~b~
2
a
T\
dr
r
/

x
2
dca \
^
dr
I
'
+
r
n
2
1

sin
a

30
/

I
\
.

sin —
dca
30
the general solution produces, after some algebraic
manipulation, the desired information.
2
1 9 ca l
(Cao -Cai )R 2 r-R,
r
2
sin
2
3i//
2 J ca Ca j +
(R 2 -R^
Also, from the problem statement, because of Because there is no bulk motion of a through
steady state, dc a /dd = 0, and c a is a function of the wall, mass is transferred only by molecular
radius only. Therefore. diffusion. The molar flux is given by Fick's law.

dca
=
9ab A Na ,;=-9ab
r
2
dr (* 37) dr

The concentration gradient is found from the

Because c a is a function of r only, there is no need first integration of the continuity equation.
to use partial derivatives. Thus
dca 1

= Ci _
(c ao -c ai )R R 2 y

2 2
d_ dr r r (R 2 -R,)
dr\ ("SO- Thus.

dca \C a0 — ca j)R 2
Integrating once gives
/va 'Uff =-gab
dr r = R l
(R2 — Rl)Rl
rif=C, and
dr
dca \C ao — c a j)R-\
There is insufficient information the problem
in Na '\ r=R =-&ab = -2.
statement to solve for C-\ directly at this time. dr r = R2 \R 2 — r?i ) R2
 EQUATIONS OF CHANGE 217

If each of these fluxes

is multiplied by the

respective surface area, one finds that the rates of

transfer of a across the inner and outer surfaces are
For the Ay Az face [(pu)u] x Ay Az
the same. For the Ax Az face [(pu)v] y Ax Az 11.31)

For the Ax Ay face [(pu)w] z Ax Ay

MOMENTUM BALANCE
A momentum balance on the volume element The rate of x-directed momentum out of the
shown in Figure 11.1 leads to the equation of element by bulk flow is
motion and the Navier-Stokes equations (7,8).
The momentum balances states that For the Ay Az face [(pu)u] x+Ax Ay Az
Rate of momentum rate of momentum Ax Az
| For the face [(pu)(v)] y+Ay Ax Az
into element \ out of element
For the Ax Ay face [(pu)(w)] z+Az Ax Ay
rate of momentum
addition from external 11.32)

forces acting on the

Momentum also enters the element because of
element molecular diffusion, as shown in Chapter 10. Thus
rate of momentum t,j is the momentum flux through a face perpen-

accumulation 11.26) dicular to the /-axis for /-directed momentum.

within the element Therefore, for the volume element shown in Figure
11.2, the rate of momentum entering by molecular
Momentum is defined as the product of the processes is

mass of a body times its velocity.

momentum = mv (11.27)
For the Ay Az face txx x \
Ay Az
where v is the velocity vector, and therefore For the Ax Az face Tyx \y AX AZ 11.33)
momentum is a vector. Momentum per unit vol-
For the Ax Ay face tzx z Ax Ay
ume is of more direct interest in fluid mechanics, |

and is
The rate of momentum leaving the element
Momentum by molecular processes is
= pv = M (11.28)
volume
For the Ay Az face Txx \
x+Ax Ay Az
Because momentum is a vector, it can be described
in terms of three components having direction and
For the Ax Az face r^L
yx iy-+A .AxAz
1
(11.34)

magnitude, e.g., For the Ax Ay face tzx z+Az Ax

\
Ay
Mx = pu

MK = pv :i i.29)

Mz = pw z*U. + Ay

Itassumed that fluid can move through all

is

six faces of the volume element of Figure 11.1.

Therefore, momentum enters and leaves the ele-
ment through each face. First, it enters the element
because of bulk fluid motion. The x-directed
momentum entering by bulk flow into the element
at x, because of flow in the x-direction, is

Mxx = (pu)u 11.30)

M, is the momentum transferred in the x direc-

tion as a consequence of flow in the x-direction.

Applying Equation 11.26, the rate of x-direc- Figure 11.2. Molecular momentum transport through
ted momentum entering the element by bulk flow volume element.
218 MOLECULAR AND TURBULENT TRANSPORT

Only those external forces associated with Equation 1 1 .8, yields

gravity and fluid pressure will be considered. The
gravity force in the x-direction is pgx Ax Ay Az. Du ( drXX d T yx ,
drzx
)
bP
+ T~- P9* 11.41)
The force resulting from fluid pressure in the x- De bx by bz bx
direction is
Dv
=
(dr
bTx y bTyy
+
Z y\
+
bP ^
At plane x P\ x Ay Az P P9y 11.42)
bx *l)y
D~e~ \\~bx~ ~te' dy~
At plane x + Ax P\ x+Ax Ay Az
Dw _ /3t
— + br
br yz
+—
rv -
l< xz ~ zz
" p9z 11.43)
The resultant of these two forces in the oe \ bx dy bz y
i

x-direction is

The above three equations are components of

(P\ x - P\ x+Ax ) Ay Az + pgx Ax Ay Az (11.35)
the equation of motion showing the forces acting
on a volume element moving with the fluid. They
The x-directed momentum accumulation
may be added vectorially to give
within the element is given by

b(pu)
Ax Ay Az !
11.36)
_p _
Dv
=v .
T+v/,_ pg (11.44)
bd

Substitution of Expressions 11.31, 11.32, Equation 11.44 is a general statement of

11.33, 11.34, 11.35, and 11.36 into Equation Newton's second law of motion. This equation and
11.26, division by Ax Ay Az, and letting Ax, Ay, its component parts are useful, but not nearly as
and Az approach zero gives much as the equations in which the stresses are
written in terms of viscosity and velocity gradients.
b(pu)
— (puu) + —
b 3
ipuv) + — (puw)
3
For Newtonian fluids, these expressions are
b6 bx by bz
(5)
9^x 9t^ 3t^\ _bP
-(
bx
+
by
+
bz bx
P9x
?"xx
= -2p
„
—
bu
+ §p — + — +
Ibu ,bv
(11.45)
ox \dx by a7/
(11.37)
bu ibu bv 3w\
The y- and z-components are derived in a Tyy
2
(11.46)
similar way. They are by \ bx by ~b~z>

b(pv)
— (pvu) + — (pw) + — (pvw)\ b
3_ 1
Tzz
bw 2
/ bu bv bw\
(11.47)
bd ox by bz
)z J bz \bx by 37'
bTx y bTyy 37 Z y bP (bu bv\
bx by bz
)- + P9x 1~xy (11.48)

;11.38) (bv bw\

TVZ (11.49)
b(pw)
~ZT~
be
= ~\
r

L
V
d
dx
\pwu) + —
b
by
(pwv) + — (pww)
t
b
bz
k ,

(bw bu\
Tzx (11.50)
\- —
3rxz bTyyz d7z.
-( |
P9y
bx Ty bz i
bz
Substitution of Equations 11.45 through
11.39)
11.50 into the corresponding equations of motion
Adding Equations 11.37, 11.38, and 11.39 (11.41, 11.42, and 11.43) gives
vectorially gives the equation ofmotion in vector
bP —
form: Du
P9x
— +
b
2n
bu
|/x(V-v)]
3(pv) D6 3x 3x bx
V •
pw- V •
t- TP + pg (11.40)
ibu bv bw bu\
be b
phr. + bx
by I \ by bzl y
bx bz
Carrying out the indicated differentiation of
Equations 11.37, 11.38, and 11.39, and applying 11.51)
 EQUATIONS OF CHANGE 21

p
Dv
=
P9y
—
dP —
+
3
[*.£-§ M<Vv>]
d~b dy dy\
Flow

bx L^\3x by 'J bzl^by bz>\

(11.52) vary with r from the center of the tube to the wall.
Dw dp 9 / 3iv Application of the equation of continuity with
p =P9 §p(v •)) constant p shows that w is not a function of z. The
~Be dz dz
appropriate Navier-Stokes equations for the z
8 I" /3w 3iy\l 3 I" (dw
/3w dv component are
M +
8xL Ux 3z'J a/rUy
(11.53)
dw
dd
+ vr —
dw
dr
+
r
dw
90
+ w
dw
~dz~'~~dl
\ dP

Equations 11.51, 11.52, and 11.53 are the

2
1 d w d
2
w~
Navier-Stokes equations; they describe the be-
havior of many flow systems. Their use is limited
to Newtonian fluids flowing under laminar flow
+P
r dA
r
dr Frr22 ^ +
^. + P9z
(E-16
conditions. As derived, the only body forces con- and for the r component
sidered were those due to gravity and pressure.
For constant p and p conditions, Equation / dvr dvr v$ dvr dw\ dP
1 1.51 becomes y
dd dr r 30 dz
'
dr

Du 2 2 2 2
dP d u d u d u\ [9/1 3 1 2 dv 3

^-3r^V
i
=
^
i
t
+P +
2+
3^) «"") [d-r{-rVr^) r?
2
30
2
r
2
30 dz 2
P9r

Similar equations result for the other two direct-

(E-14)
ions, so the vector sum becomes
Because of the assumptions stated above and
Dv because there is no z-component of gravity, Equa-
= pg - V- P + fxV' :i 1.55) tion E16 becomes
Dd
A special version of this equation may be ri!( bw\
= +
written for the case where pressure and gravity ° -d~z Vr dr\
effects are small compared to inertial effects:
and Equation E-14 becomes
Dv
hp
H
Dd
M ; 11.55s)
= - — + P9r
dr
or

Dv The above equation is a statement of how

D6
= w v (11.55b) pressure varies with which is of no interest in
r, this
particular illustration. Only the equation with vel-
ocity explored further. It may be written as an
is

Example 11.3. Starting with the appropriate ordinary differential equation

Navier-Stokes equation, derive the velocity profile
for a Newtonian fluid of constant p and p flowing dP f1 d I dw \

through a horizontal cylindrical tube. Steady-state =P

7z \~r7r\
laminar flow prevails and flow occurs only in the
z-direction. (This derivation will lead to Equation
The left side of this equation is a function of
10.90).
z only and the right side is a function of r only.
This situation is possible only when each side is
Solution. To obtain the desired expression,
equal to the same constant. Thus, let
the Navier-Stokes equations in their cylindrical
coordinate form are used (see Appendix E).
= constant = —
dP AP
Flow is one-dimensional, so v r = and v^ = 0. —
The velocity in the z-direction is not zero and will dz Az
220 MOLECULAR AND TURBULENT TRANSPORT

And thus, the above equation becomes ENERGY BALANCE

1_ AP _ 1 d_. dw\ An energy balance may be applied to the volume
element shown in Figure 11.1. The element density
p Az r dr^ dr
is taken as p at x, y, z, 6, and the velocity com-

Integrating once, ponents of the fluid flowing through the element

2 are u, v, w, at x, y, z, and 0. For the volume
r
^
2p
AP
Az
= r —+C
dw
dr
: element, the general energy equation may be stated
as
Atr-0, the velocity is a maximum and dw/dr Rate of accumulation \
= 0, so that C-[ = 0. Integrating again gives
of internal and kinetic
J

— — = w + C2
2 energy
r AP /

Ap Az
net rate of internal /net rate of heat'
At r = r-[ , w= 0, so that i and kinetic energy I + j added by
2 brought in by I 1 molecular
_ r, AP bulk flow / \ transport
2 ~ Ap At
rate of thermal
net rate of work 1

Substituting for C2 into the general solution gives done on the

energy
surroundings
AP generated
w=— (r,
2
-r 2 )
by the element
AAzp
(11.56)
Thus, the velocity profile is parabolic, which
is what was shown in Equation 10.90. The net rate of accumulation of internal and
kinetic energy within the volume is

The procedure above when apply- illustrated

ing the Navier-Stokes equation is typical of the
ipU+ KE) Ax Ay Az (11.57)
analysis process. A summary of this procedure is ae
given below.
Choose a coordinate scheme describing the
1
where U
the internal energy per unit mass of
is

fluid. is associated with molecular

Internal energy
geometry of the physical situation.
2. Analyze the problem with respect to any motion and molecular interaction and is dependent
assumptions, or restrictions, that may apply — for upon temperature. KE is the fluid kinetic energy
example, steady state, one-dimensional flow, no per unit volume.
pressure gradient, symmetry, etc. The net rate of flow of internal and kinetic
3. Select from Appendix E the appropriate energy into the element is

Navier-Stokes equation and eliminate terms found

to be inappropriate. (u(pU+ KE) \
x -u{pU+ KE) \
X+ Sx ) Ay Az
Solve the resulting differential equations
4. + (v(pU+ KE) y \
- v(pU+ KE) y+Ay Ax Az
\ )

using the appropriate boundary conditions. The

following are common boundary conditions: + (w(pU+KE) |, -w(pU+KE) z+ ± z Ax Ay \ )

(a) At a solid-fluid interface, the fluid (11.58)

velocity is that of the solid. This condition is often

called "no the wall," and is an assumption

slip at The net rate of heat added by molecular trans-
that seems to be accurate in many situations. port is

(b) At a liquid-gas interface, the velocity

(Qx\x -Qx\x+ax) AyAz + (q y \
y -q y \
y+Ay ) Ax Az
gradient in the liquid phase is assumed to be zero.
(c) At a liquid-liquid interface, the mo- + (^Iz-<7zU+az) Ax Ay
mentum flux and velocity profiles are continuous
(11.59)
across the interface.
Solutions to the Navier-Stokes equations in The generation of thermal energy within the
two- and
three-dimensional flow are complex. volume may arise from several mechanisms. For
Some of the classic solutions for a variety of flow example, it can come from an exothermal chemical
systems are found in Reference 6. reaction, a heat source, or a nuclear reaction. The
 EQUATIONS OF CHANGE 221

'
rate of energy generated is bq x bq y bq z
-
9q Ax Ay Az dx by dz -
: 11.6O)

Work done by the volume element on its - \

—b

bx
(Pu) + — b
(Pv) + —
b
(Pn
surroundings arises from several sources. Work is
by bz
done whenever exerted through a dis-
a force is + p(ugx + vgy + wgz )

tance, and the rate of work

is equal to force times

velocity. Energy is added to the volume element as

fluid is pushed into it by the pressure at plane x;
- r
—
-bx
b
{Txx U + Txy V + Txz w)

energy is consumed by the volume element as fluid a

is pushed out at plane x + Ax. For the element, the + TYxU + T V+Tyz w)
yy
net rate of work by pressure on the volume ele- fy(
ment is b .

+ !rxU + Tzy[ '+ TZZ W) n.64)

dz
:

{Pu\ x -Pu\ x + Ax )AyAz+(Pv\ y -Pv\ y + Ay )AxAz

+ (Pw\ z -Pw\ z+Az )AxAy As before, the terms on the right side of
Equation 11.64 involving pU and KE can be com-
(11.61) bined with the left-hand term to yield the substan-
tive derivative pD(U + \v 2 )lDQ where KE = \pv 2
Work is also done on the volume element by ,

Thus, Equation 11.64 can be written in compact

gravity:
vector notation as
p Ax Ay Az (ugx + vg y + wgz ) (11 .62)

Finally, element does work

the volume
against viscous forces. For example, T yx AxAz is
the force necessary to produce a flow velocity u in =-V 'q+9q - V-/V + p(vg)-v(T-v)
a direction parallel to the AxAz plane. So the net
(11.65)
rate of doing work against viscous forces is

Equation 11.65 may be put into a more use-

[Wxx U + Txy V+Txz w) \
x ful form by performing the following rearrange-
ment. First, Equation 11.44 is multiplied through
-(rxx u + Txy v + Txz w) X+ Ax ] Ay Az
by v
\

+ [{r yx u + ryy v+ ryz w) y

Dm
\

- - v ^" = vVT+vrP-vp0
[ryx u + ryy v + ryz w) \
y+ Ay ] Ax Az
+ [(Tzx U + Tzy V+Tzz w) z Dv
\

= 0(V)
V
- (tzx u + tzy v + tzz w) z+Az \ ] Ax Ay DO DB
' S substituted:

(11.63) D(W 2
-p—^Q— = ^T + vTP-vpg
)

(11.66)
Substituting Expressions 11.57, 11.58, 11.59,
11.60, 11.61, 11.62, and
11.63 into Equation Next, the V-terms are expanded in Equations
1 1 through by Ax Ay Az,
.56, rearranging, dividing 11.65 and 11.66. Finally, the expanded form of
and taking the limit as Ax, Ay, and Az approach 1 1 .66 is subtracted from 1 1 .65 to give

zero yields the energy equation

DU
-=- v q+ -P(V- - t.tv
Ve
W + KE) = ~Dd

The term pDU/Dd

:o„

is
v)

the rate of accumulation

(11.67)

— u(pU + KE) + — v{pU + KE)

b b of internal energy per unit volume; — V q) is the
energy added to the volume by conduction; [9
(
•

ox ay q )

is the volumetric heat generation for an incompres-

+ — w(pU+ KE)
bz
sible fluid; (-PV-v) is the increase in energy
caused by compression or expansion of the fluid;
222 MOLECULAR AND TURBULENT TRANSPORT

and (— t : the internal energy increase from

Vv) is conducting medium is a solid, rather than a fluid,
frictional expenditure of mechanical energy in fric- so the compression and viscous dissipation terms
tion, which degrades mechanical, or work, energy are also zero. This leaves only one term in Equa-
into thermal energy — concisely referred to as tion 11.67
"viscous dissipation." This last term is often small
and can be neglected except for those situations where q = qIA
involving very high shear stresses. In cylindrical coordinates this becomes
shown Chapter 10 that q =
— kVT
It has been
and, also, for an ideal
this information, Equation 11.67
in
gas dU = C v dT. Using
may be written,
-—
r or
(rq r ) + ~
r
——
dr
+
oz
=

neglecting viscous dissipation,

Conditions do not vary along the pipe in the z-
DT = direction, so q z '= and dq z'/dz = 0. There is also
pC v kV 2
T+V a -P(V'\r) (11.68)
do no variation in the angular direction, so qe = 0, '

and bqoloz = 0. This leads to

For gases or liquids at constant pressure, the
equation of continuity states that V • v = 0. Fur-
')=0
thermore, for liquids, the density is nearly constant f
dr
(rq r

for a wide range of temperatures so that V • v =

for all liquids. In addition, for liquids, Cv = CP so ,
Integrating once gives
Equation .68 becomes
1 1
rq r = '
d
DT dT
pCp =kV , T+ *« !11.69) Now q r = —k —- , so
~D6 dr
For systems that are not flowing, the sub- dT
* + '-
Cy
may be replaced by the partial
stantial derivative
derivative to give
-4
Substituting k = 0.5 + 5 x 10 T and integrating
again gives
(11.70)
-0.57"+ 2.5 x 10~ 4 T 2 = d lnr+C2
Equation 11.70 is the general equation for
C-i and C2 are evaluated from the boundary con-
heat transfer by conduction. If <S q = 0,
ditions

= r T=T,
»C^-^T 111.71)
At r y

r = r2 T=T2
or So
dT , 0.5(72 - TO + 2.5X 10- 4 (7-2 2 - T,
2
]

(11.72) Ci =
be
In r 2 lry
Equation 11.72 is sometimes called Fourier's sec-
ond law. It is useful in the analysis of unsteady- Now qr = -kA — and A
dT
dr
= 27rrZ. (Example 10.3).
state conduction.
So, from the first integration

Example Develop the basic differential

11.3. dT C A y

equation used in Example 10.3, starting with the qr = -kA = 2nLC

dr
appropriate form of the energy equation and
making simplifications based upon the physical sit- 0.5(r2 - T y ) + 2.5 x 10- 4 (T2 2 - 7V)
uation. = 2ttL
In r 2 lr-\
Solution. Equation 11.67 may be used as a
starting point. Since the heat transfer is at steady This expression can be found in the solution of
state, there is no accumulation, so DU/Dd = 0. Example 10.3, and the solution continues as shown
There is no internal generation, so <$
q
=0. The there.
 EQUATIONS OF CHANGE 223

UNSTEADY-STATE TRANSFER T= Tal6<

The equations of change developed thus far are
perfectly
scribe many
general and, at least conceptually, de-
physical processes involving mass, heat,
f'"\
and momentum transfer. Each of the three equa-
~~ t= r, <&.

tions has been derived for the case where the trans- = " 6=
ferent property is time dependent. Of particular
interest are

dca
= 2 b V
2
ca [11.24)
W
Dv 2
i>V v (11.55/?)
D6
and

= ar- 2 T (11.72)
/
36
Figure 11.3 Infinite flat plate.
Note the similarity in form of these equa-
tions. They are identical except for symbolism.
What this means
that mathematical solutions for
is

these equations are also identical. Analytical pro- The first condition states that the sheet is

cedures for these partial differential equations are initially at a uniform temperature n7~. At time
available in textson applied mathematics (1,9), so zero, both external surfaces of the sheet are ex-
they are not covered here. In some instances, it is posed to a temperature T-\ and maintained at that
convenient to use numerical and graphical pro- value. For now, it will be assumed that x^ instan-
cedures for solution. These will be discussed brief- taneously reached 7"i at 6 = 0. The solution to the
ly, and because students often are familiar with equation involves a Fourier cosine series.

temperature, Equation 11.72 will be used to 7-7-

demonstrate these procedures. -=2
M
I
_<|)"
rre- {n+ V " a6/
*>cos(n + \) —
nx
r. T n= 7TU7+3) X-|

Infinite Flat Plate

(11.74)
A frequently encountered problem is determining
the temperature variation with time in a sheet of
Although the solution involves an infinite
series, the magnitude of each term in the series
material which heat flow is essentially one-
in
decreases rapidly as n increases, and often only the
dimensional. It is assumed that within the sheet
first few terms are needed to obtain an accurate
there are no heat sources or sinks and thus Equat-
solution.
ion 11.72 applies. According to Figure 11.3, the
Equation 11.74 is also the solution to Equa-
sheet is infinite in extent in both the y and z-
tions 11.24 and 1 1 .55/? with suitable trans-
directions. For this case Equation 11.72 becomes
formation of symbols.
2
37" d T
= a 11.73)
:
36 3x
Example 11.4. A large flat steel slab 0.300 m
The method of solution for this type of par-
thick is temperature
initially at a of 540°C. It is

tial equation is called "separation of

differential
plunged into a quench tank that maintains both
variables," and is presented in detail in References
surfaces temperatures at 100°C. Determine the
1 or 9. Two boundary conditions and one initial
temperature at the center plane of the slab and at

are required to solve the equation.

75 mm from the center plane one half hour after
cooling begins. Assume a is constant at a value of
I C 1 at < 0, 7"= 7" for all x* 9.0 x 10- 6
m 2 /s.
BC2at0 >0, T= 7i at x = ± xi
Solution. Equation 11.74 is used with
*The presubscript on T denotes time, and the postsubscript
r=540°C, 7"! = 100°C, = 1800 s, x, = 0.15 m,
denotes position. and x = at center plane. Three terms of the
224 MOLECULAR AND TURBULENT TRANSPORT

infinite series will be used — that is, {n = 0, 1, 2). more realistic solution to the equation because
recognition must be given to the possibility for a
Ty-T 7TX resistance to transfer to occur at a solid/fluid
e -U
2 :
a e/4x l
)

cos
2xi boundary. For the above example, it was assumed
that the surface temperature instantaneously
3ttx
- i e - (9 * 2a9/4x '
2)
cos
reached Ty This assumption means that there is no
.

2x7 thermal resistance between the quench fluid and

the slab surface, or that the parameter m in Table
57TX
11.1 is zero. More will be said about resistances to
: 2
(257r ae/4x, )
COS +
2x, transfer in Chapters 13 and 14. Figure 1 1.4 may be
used for heat, mass, and momentum transport
For the center plane, x = so all the cosine terms
using the definitions of Table 11.1 Graphical solu-
are1 For the data given,
.

tions to the transient equation for other geometries

2 are also given in Figure 11.4. The charts may be
TT
2
ad (3 .14) (9.0x 1CT 6 )(1800)
= 1.77 used only the following conditions are met: (1)
if
2 2
Ax 1 (4H0.15) the diffusivity is constant, (2) the body initially is
at a uniform concentration of transferent property,
Thus,
and (3) the boundary of the body is brought to a
- T new condition and maintained at that condition
7"i
e
-1.77_. e
-15.9 + l -45
[
T for all time.

Example 11.5. Rework Example 11.4 using

= - [0.169-4.1 x 10 8
Figure 1 1 .4.
IT

Solution
+ 5.72x 10~ 21 +•] =0.216
aB (9.0 x 10" 6 )(1800)
So that T = 100 - 0.216(100-540) = 195.0°C. In
X= = 0.72
xr (0.15)
2
this case,only the first term in the infinite series is

numerically significant. m = n = x/x-\ = for midplane

The temperature 0.075 m from the center From App. D-16
plane is calculated in the same way except now the
cosine terms enter. Thus, for x = 0.075 m Y = log (-1.084x0.72 +
,
0.0961)

Y = 0.21381 = -,ln°~cL T=194° 1

/4C
7TX 7r(0.075) __ 1 00 - 540

cos = cos — = cos

it
—= ,_
0.707
7x y 2(0.15) 4 At position 0.75m from midplane, n = ^^
bU
0.1
= 0.5

From what was learned from the center-plane cal-

Interpolating and solving
culation about the behavior of the infinite series, Y=log- (-1.0845x0.72 + (-0.0636)
,

only the first term in the series will be used. = 0.16564

Ty-T- 4 7"=100-(0.16564x(100-540)) = 172°C
= -[0.169(0.707)] =0.152
' 1 - ' 7T Numerical and Graphical Solution
7"= 100- 0.152(100-540) = 167.0° C Xumerical Solution: Differential equations
may be solved approximately using finite differ-
To facilitate calculation, the solution for ence methods. The methods involve replacing the
Equation 11.74 has been put into graphical form differentials by finite differences and then solving
(Figure 11.4).* These charts represent a somewhat the resulting equations algebraically. In certain
cases, finite difference equations can be solved
graphically. To demonstrate finite-difference tech-
*These figures are in the same formulation as presented
niques, the one-dimensional form of the conduc-
originally by Gurney and Lurie (3). The only differences are be-
lieved to be based on digital computation, using more eigenvalues tion equation is considered in the absence of a heat
for thesummations and more digits in each eigenvalue by Yenwo (10). source or sink:
The change in graphical scales, and replacement of graphs by linear
semilogarithmic equations for some values of the dimensionless (11.73)
variables, is believed to improve the utilitarian accuracy. dd 3x 2
 EQUATIONS OF CHANGE 225

16 13^ 14-^ 15- 16 13s 12>

(b)

Figure 1 Modified Gurney-Lurie charts (Note: An empirical equation

1.4. is given in
Appendix D-16 for values of X>
0.2 and Y <0.3).

Key to Figures 11.4

Slab (a) Cyl. (b) Sph (c)

Line m n m n m '

1 0.0 1.0 0.0 1.0 0.0 1.0

2 0.0 0.8 0.0 0.8 0.0 0.8
3 0.5 1.0 0.0 0.4 0.5 1.0
4 0.0 0.6 0.5 1.0 0.0 0.4
5 0.0 0.4 0.5 0.8 0.0 0.0
6 1.0 1.0 0.0 0.0 0.5 0.8
7 0.0 0.4 1.0 1.0 1.0 1.0
8 0.5 0.4 2.0 1.0 0.5 0.0
9 2.0 1.0 0.5 0.4 2.0 1.0
10 0.0 0.0 0.5 0.0 1.0 0.0
11 2.0 0.4 1.0 0.0 5.0 1.0
12 0.5 0.0 2.0 0.0 2.0 0.0
13 2.0 0.0 5.0 0.0 5.0 0.0
14 5.0 1.0
15 5.0 0.0
16
17

Recomputed from (3); Courtesy Amer. Chem. Soc.

226 MOLECULAR AND TURBULENT TRANSPORT

Table 11.1. NOTATION FOR FIGURE 1 .4a, 1 1 Ab and

,
1 1 Ac

Specific Notations
Parameter General
Symbol Notation Heat Mass Momentum

ri-r 7-1 -T C3 1 v\Py ~ vp

Unaccomplished change
ri- r Ty-oT ca1 - 0Ca v-[P-[ - QV p

Relative time
50 aB m vd
2 2 2 X! 2
*1 *1 *1

X X X X
Relative position
*1 *1 *1 *1

m
5 k 2
Relative resistance
SX1 hXy kc x :

X, Y, m, n are dimensionless
h, kc = transfer coefficients* The assumptions
T-
temperature & = generalized transfer coefficient of the solution are
cB = concentration a.= thermal diffusivity seldom met in

v= velocity @ = mass diffusivity momentum

x = distance from center to any point v = momentum diffusivity transport
xi = radius, or half-thickness 5 = generalized diffusivity
6 = time
k = thermal conductivity Postsubscripts:
1 = boundary
Presubscript:
a = components
= at time (initial condition)

Generally, the presubscript will refer to time, the postsubscript to position. For momentum transport v-\ = 0.

"These terms will be defined in later chapters. For this chapter m = 0. The use of the charts for
m> will be demonstrated in Chapter 14.

Figure 1 1 .5 represents the temperature history of a where (A e T/Ax) B (A e T/Ax) A are the values of
,

slab of material finite in the x-direction only, so the gradient at planes B and A, respectively. Both
heat transfer may be assumed to be one-dimension- are evaluated at 6. Examination of Figure 1 1 .5 will
al and the time, temperature, position interrelation show the meaning of these terms. The left side of
is described by Equation 11.73. Assume that the Equation 1 1 .73 in finite difference form is
temperature midway between planes A and B rep-
resents the average temperature of that "slab" and
that the fluxes into that element contribute to
:i i.76)
accumulation in the element. The potential gradi- Ad
ent at each boundary may be evaluated in the
terms of the potential at the centerline of the Equating Expression 1 1.76 to the right side of
element, and of the potential at adjacent center- 1 1 .75 and expanding the ATI Ax terms of 1 1 .75 in
lines. The accumulation may be expressed in terms finite difference form gives

of the increase of the temperature at the centerline

temperature, through a corresponding time inter- >7~x + Ax~07~x eTx — eT)x— Ax
val. + AeT,x ~~ 6'
x Ax Ax
In finite difference form, Equation .73
1 1 is
A6 Ax
AbT (11.77)

A(TCenterline'
( \ Ax K \ Ax
'
where g Tx = temperature at x and 6
= a :n.75)
Ad = temperature atx and 6 + Ad
Ax + a<?7"x
 EQUATIONS OF CHANGE 227

B x + Ax temperatures of that plane and the centerlines of

adjacent slabs (or increment) at a time A0 pre-
ceding. The left-hand term is at time 6 + Ad, all
others are at 8. Therefore, given an initial temper-
ature distribution within the slab, it is possible to
calculate the temperature as time passes based
upon increments of Ad coupled to thickness of
increments, Ax, and the thermal diffusivity of
the material.
If the modulus M
is chosen to equal 2, Equa-

tion 1 1.80 becomes

>'x + Ax + (Tx- A;
+ Tr =
A0'x (11.81)

Thus, Equation 11.81 fixes the value of Tx equal

to the arithmetic average of 7" on either side of it at
E x + Ax
one time-interval earlier. Setting = 2 fixes the M
time interval Ad if Ax is specified, or fixes Ax if

Ad is specified. Any value of M greater than 2 may

be chosen and used in the solution of Equation
11.80. Care should be exercised in choosing Ad or

Surface
Ax so as to keep errors small since a differential is
being replaced by a finite difference. As small a
value of Ax or Ad as is convenient should be used.
For graphical solution, it is necessary to choose a
Ax that will divide the slab into an integral number
of layers, and to solve for Ad.
Equation 11.81 may be written in specific
terms for
Mass transfer:
ca,x + Ax + 6 ca,x — Ai
Figure 11.5. Finite differences in transfer at time 6.
e+A0 ca,x ~ 11.82)

Momentum transfer:
Solving Equation 1 1.77 for e + AeTx 9' ves

e^x + Ax + e vx- ax ,„..„„,

ccAd
e + Ae^x
= « (11.83)
+ A0^'x ~ . o~ id^x + Ax ~ *-eTx + qTx _ ax J
+ 8'x
Ax
(11.78)
The above three equations may be used for
any initial concentration distribution.
A modulus // is defined:

Jt=
Ax 2
(11.79)
Graphical Construction: When = 2 for M
ocAd the one-dimensional case, a simple graphical solu-
tion for Equations 11.24, 11.55/?, and 11.72 is
Then, substitution of Equation 11.79 into 11.78 possible. This graphical solution is called the
gives
Schmidt or Binder (4) method. It is based totally
\M — TX + eTx - upon the successive use of Equation 11.81 and is
>7"x + Ax + 2) ax
Ae'x ~ illustrated in Figure 11.6.
+
M The solid line in Figure 11.6 represents the
:
11.8O) temperature distribution within a slab at time
d = 0. According to the method, the temperature
This equation permits prediction of the e + AeTn is determined by the intersection of plane
temperature at a centerline of a slab from the n and a line joining e Tn _ and e 7"„ + Thus, the
1 1
.
228 MOLECULAR AND TURBULENT TRANSPORT

get the surface temperature for the 0=0 value, and

thereafter to use the slab surface temperature equal
to the ambient value. A more rigorous value for the
initialsurface temperature after sudden exposure
could be evaluated from an analytical solution to
the conduction equation.

Example 11.6. A wet with

colloidal gel
alcohol is to be dried. The
packed into a
material is

pan 2 ft by 1 in. thick, and the top face is exposed

to a dry airstream. The alcohol will be transferred
through the gel by molecular transport with a mass
diffusivity of 0.00096 ft 2 /hr. The initial gel con-
centration is uniform at 3 percent alcohol by vol-
A B C D ume. Determine the concentration profile 2 hr
after the surface of the gel is exposed to the air
Figure 11.6. Schmidt graphical method.
containing no alcohol.

dotted line connecting qTc and qTa gives i7"g, Solution. The slab has one face impermeable
the temperature at B one Ad later. And the line bottom of the pan). This face
to alcohol (the
from TD and TB gives Tc etc. , ,
would be equivalent to a 2-in. thick slab with both
This convenient graphical technique is based sides exposed to air. Then the alcohol would dif-
upon proper choice of time and distance intervals fuse to both faces, but none would cross the center
such that the concentration at a point one time plane. Alternatively, it is evident that the concen-
interval later is equal to the average of the concen- tion gradient at the gel-pan interface must be
trations at the two adjacent distance intervals at zero since there is no transfer. The 2-in. slab is
the start of the time interval. The development shown in Figure 11.7. The position increment Ax
assumes that the whole volume is homogeneous in is chosen at 3 in. to give a reasonable number of
all respects except in concentration of the transfer-
increments over the total thickness for the pur-
ent property. In the derivation of Equation 11.77, poses of this illustration. A smalller increment
the property balances are based upon fluxes would give greater accuracy. Since = 2, Ad may M
through planes A and B and accumulation (express- be evaluated
ed as concentration of transferent properties)
based upon the concentration at the midpoint be- Ax'
.* = = 2
tween A and B. Obviously, the plane A' at the 9 Ad
surface (Figure 11.5) does not fit the model, and
its concentration cannot represent the average con-
A0 = = 0.40 hr
centration in the last half slab, Some kind of spec- (0.00096) (2)
ial treatment is necessary when the surface temper-

ature is to be predicted and used in a calculation.
When the surface temperature of a slab is
suddenly changed, the question of what temper-
ature applies at 8 = must be answered before the
graphical construction of the Schmidt method can
proceed. The problem is that at 9 = 0, the surface
temperature may be considered equal to the initial
temperature of the slab, to the new temperature
after the sudden exposure, or to some value be-
tween the two.
One recommendation (effective, though
without theoretical justification) is to use an arith-
metic average of the two initial conditions — that
Therefore 2/0.4 = 5.0 time increments will be re-
quired. The initial condition of the slab is shown in
Figure 11.7a. The notation ca2 refers to the con-
centration at zero time at 2 distance increments
from the surface. The presubscript will in general
give the time increment, and the postsubscript, the
distance increment from the surface of the slab.
The initial concentration gradient is horizontal at 3
percent alcohol. However, on exposure to dry air
the surface concentration drops to zero. Just be-
fore exposure it is 3 percent. Thus, c a0 ma y be
either or 3 percent. For the first time increment
only it is customary to use the arithmetic average
,

is, the ambient temperature to which the slab is of these values (i.e., c a0 = 1.5). After the first
exposed and the surface temperature at 6 = - to time interval, c a0 is taken as zero. To find the
 EQUATIONS OF CHANGE 229

0Ca2 <Ka3 0C a 2 OCal

1 0. 1 0. 1

Fictitious mirror image

of actual slab
fa)

0.1 0,1.2 o.i

' J
3
\ N -

/ /\ K
i

/ / \ \

4 />"""i \ 4

5 5

10

1,2.3.4

Figure 11.7. Solution to Examole 11.6. (a) Initial conditions, (b) After 1.A0
increment, (c) After 2. A d increment . [d) Final conditions.

concentrations after one time interval, Equation at the first time interval are determined graphically
1 1 .82 is used. For ca it becomes ! in the same manner.
1

O^aO + 0C a 2 0C a1 + oC a 3
1^31
- \Ca2 =

This can be accomplished graphically by connect- 0Ca 2 + 0C a 2

ca3 = ~
ing the points c aQ and c a2 by a straight line, as' 1

shown in Figure 11.76. Thus, the concentration \

in. from the surface decreases to 2.25 percent after This last equation shows the utility of the mirror
a time interval of 0.4 hr. The other concentrations image of the slab. ,c 3 3 is evaluated using c a2 n '
230 MOLECULAR AND TURBULENT TRANSPORT

the slab and ca2 in the mirror image. Since (a) An observer standing on the train platform sees

ca2 - 3.0, ^33 = by the previous equa-

3.0. Also, the walking passenger. At what speed does it appear that
the passenger moving? What derivative of position with
tion, since ca and 0^3 =3.0, yC a2 =3.0. The
i
is

respect to speed does this speed represent?

resulting gradient after 0.4 hr shows the slab next
(b) What derivative with respect to time is repre-
to the pan still at 3 percent, but near the surface
sented by the 3 miles/hr observed by the seated passenger?
the concentration has decreased. The concentra-
tion profile is not a smooth curve because finite
(c) A car is traveling at a speed of 60 miles/hr parallel

to the tracks in the direction of the train movement. What

differences have been taken. The smaller the dis-
is the apparent speed of the car as viewed by the walking
tance increment, the smoother will be the resulting
profile.
passenger and the observer on the platform? What deriva-
tive does each of these observations represent considering
Figure 1 .7c shows the profile after two time
the train speed as the flow velocity?
increments. It is determined as before. For ex-
ample, 11.2. A liquid is being heated in a pipe 60 m long.
The temperature varies linearly from 80°C at the inlet to
1 c a0 + 1 ca2
2C3 1 = 98°C at the outlet. The liquid flows at a velocity of 15 m/s.

Steady-state conditions prevail.

After the first interval iCa0 = percent, as shown. (a) What is the derivative of temperature with time?

Each point is labeled only with its time interval, (b) What is the substantial derivative of temperature

for brevity. After two time intervals, the concen- with time?
tration at the pan is 3 percent, but it drops rapidly (c) Can the total derivative of temperature with time
nearer the surface. be determined numerically? Explain.

The calculation is continued for five time 11.3. Using a physical situation other than those
intervals, as shown in Figure 11. Id. The points described in this book, illustrate the differences among
labeled 5 represent the final concentration at the partial, total, and substantial derivatives.
pan has dropped to slightly less than 2 percent 11.4. In your own words, describe the physical
meaning of each term in Equation 1 1 .8.

11.5. Derive Equation 10.49 starting with the appro-

REFERENCES
priate form of the equation of continuity, and justifying
the omission of terms.
1. Carslaw, H. S., and J. C. Jaeger, Conduction of Heat in 11.6. Derive Equation 10.99 starting with the appro-
Solids, Addison-Wesley, Cambridge (1955). priate form of the energy equation, and justifying the
2. Chambers, L. G., A Course in Vector Analysis, omission of terms.
Halstead Press (1974). 11.7. Derive Equation 10.108 starting with the ap-
3. Gurney, H. P., and J. Lurie, Ind. Eng. Chem., 15, propriate form of the equation of continuity, and justifying
p. 1170(1923).
the omission of terms.
4. As presented in Jakob, M., Heat Transfer, Vol. 1,
11.8. In your own words, describe the physical
pp. 383-398. John Wiley, (1949).
5. Knudsen, J. G., and D. L. Katz, Fluid Dynamics and
meaning of each term in Equation 11.40. How does

Heat Transfer, McGraw-Hill, New York (1958), Equation 11.44 differ physically from Equation 11.40?

pp. 31-33. 11.9. In your own words, describe the physical

6. Longwell, P. A., Mechanics of Fluid Flow, McGraw- meaning of each term in Equation 11.65. how does this
Hill, New York (1966) pp. 118-126. equation differ physically from Equation 1 1 .67?
7. Navier, M., Mem. Acad. Sci.,6, p. 389 (1827). 11.10. Solve Problem 10.13, starting with the appro-
8. Stokes, G. G., Trans. Camb. Phil. Soc, 8, p. 287 priate equation of change.
(1845).
11.11. Solve Problem 10.14, starting with the appro-
9. Wylie, C. R., Advanced Engineering Mathematics, 3d
priate equation of change.
ed., McGraw-Hill, New York (1966).
11.12. Solve Problem 10.15, starting with the appro-
10. Yenwo,G., Personal communication (1975).
priate equation of change.
11.13. Solve Problem 10.28, starting with the appro-
PROBLEMS priate equation of change.
11.14. Solve Problem 10.19, starting with the appro-
11.1. A passenger seated in a train observes another priate equation of change.
passenger walking down the aisle at a speed of 3 miles/hr 11.15. Solve Problem 10.24, starting with the appro-
toward the rear of the train. The train is moving forward at priate equation of change.
a speed of 20 miles/hr 11.16. For unsteady-state transfer, why is the partial
 EQUATIONS OF CHANGE 231

time derivative used for diffusion (Equation 11.24) and wall temperature is 0°F,
conduction (Equation 1 1 .72), but the substantial derivative (a) Compute the steady-state heat loss per square
used for laminar flow (Equation 1 1 .556)? foot of wall for daytime.
11.17. Derive Equation 11.73 starting with a heat (b) For nighttime compute the time after change of
balance and qIA = —k dTldx. thermostat required to accomplish 63 percent of the change
11.18. A sphere of 0.100 m radius at a temperature expected.
of 400°C is suddenly placed in a cool fluid so that its 11.26. Consider a thin film of Newtonian fluid flow-
surface is at 50°C. The thermal diffusivity of the sphere is
ing down a plane surface inclined at an angle a from the
1.15 x 10~ 6 m 2 /s. Plot the radial temperature profile after horizontal. The film has a constant thickness and temper-
1 hr and 2 hr. ature, and is in laminar flow. Liquid is supplied at the top
11.19. A platinum catalyst is made by immersing of the plane at the required rate. The surface is L feet long
spherical alumina pellets in a chlorplatinic acid solution and H/feet wide.
until an appropriate amount of acid diffuses into the pellet. (a) Develop an expression for the velocity as a func-
The acid is then reduced to release a finely divided plati- tion of film thickness for a point along the plane assuming

num on the alumina. Half-inch diameter pellets are initially the film thickness is known.
wet with pure water. They are immersed in acid solution
(b) Develop an expression for the film thickness as a

such that the surface concentration is maintained at 50 function of a for a known volumetric flow (ft /sec).

percent acid, 50 percent water. The transfer of acid is by 11.27. Consider a wetted-wall column in which a

molecular transport and = 5,x 10~ 5 ft 2 /hr. & liquid flows down the inside of a vertical cylindrical tube.

(a) Calculate the acid concentration ^ -in. from the The liquid falls in a thin film completely wetting the
center after 3 hr immersion. surface. The remainder of the column is filled with upward-
(b) Calculate the time required to reach a concen- flowing gas. the column has an inside radius R-\
tration of 40 mole percent acid at the center. (a) Neglecting the effect of gas flow, develop an

11.20. A large steel slab initially at 2000°F is quench- expression for the velocity within the film as a function of

ed in oil that maintains its surface temperature at 300°F. radius.

The slab is 6 in. thick by 6 ft square. Calculate and plot the (b) Determine the total volumetric flow.

temperature profile across the thickness after 10 min and (c) What is the qualitative effect of gas flow on the
2 liquid flow? Can this effect be neglected?
after 30 min. Let a = 0.30 ft /hr.

11.21. A steel rod 1 m long is heated until the rod 11.28. Two similar circular disks are separated by an
has a linear gradient running between 600°C at one end and oil film of thickness b. A force F compresses them together.

300°C at the other. The temperature at the 600°C end is

Oil under pressure is supplied continuously to a central

suddenly lowered to 100°C. The sides and the other end of chamber and escapes radially. For an oil of viscosity n,

the rod are insulated. Calculate the temperature profile derive an expression between F and specified values of fi, Q,
after .027 mega s. (Hint: Because the sides and one end are b, D-\ , and D2 as shown in the figure.

insulated, the rod can be considered part of a flat slab with

the 600°C end at the surface of the slab.)

11.22. Calculate the concentration profile of the gel

in Example 11.6 after 2 hr of drying, using Figure 11.4.
Compare the result with that of the illustration.
11.23. Calculate the concentration profile of the gel in

Example 1 1.4 four hours after drying begins.

11.24. A large brick wall 1 ft thick is heated on one

side and cooled on the other so that at steady state a linear
temperature gradient exists. The temperature varies be-
tween 1500°F at one face and 100°F at the other. Sud- Q, Oil flow-

denly the heating and cooling are reversed, so that the two
surface temperatures are effectively reversed. Determine the
resulting temperature gradient 3 hr and 6 hr after the 11.29. A uniform rate of heat is generated in the
reversal. Let a = 0.02 ft 2 /hr. solid section of a hollow cylinder. The inner and outer
7"
11.25. The walls of a house are built of red brick, 13 surfaces are maintained at 7"; and , respectively. Develop
in. thick. The 75°F during the day, and
walls are heated to an expression for the temperature within the wall. What is

the inside wall is allowed to fall to 60°F at night by the heat rate at each surface for steady-state conditions?
adjusting the thermostat. For a day on which the outside 11.30. Repeat 1 1.29 for a hollow sphere.
 Twelve

Turbulent-Transport Mechanism

Molecular transport, which was described in centration gradient. In practice, turbulent trans-
Chapter 9, depends upon the random motion of port is of more
importance to the chemical
individual molecules for transport of mass, heat, or engineer than molecular transport. Therefore, the
momentum. Turbulent transport, which will be concepts developed for laminar flow will now be
discussed in this chapter, is due to the random applied to transport in turbulent flow.
movement of large groups or clusters of molecules.
These groups or clusters of molecules are called
Pressure Drop in Turbulent Flow
eddies, and in some cases they are large enough to
be visible to the naked eye. Mass, heat, and Ifthe pressure drop (-AP) is measured in a pipe at
momentum may be transferred by turbulent trans- a relatively high fluid velocity, it is found to be

port in any fluid. much higher than the value predicted by the
The conditions under which turbulent trans- laminar-flow equation (Equation 10.94). For
port will be discussed in the following
occurs example, consider that water at 70° F flows
sections.Turbulent transport cannot be described through a smooth circular pipe of 1 in. I.D. at a
completely by a single mathematical expression, mean velocity of 10 ft/sec. If the pressure drop is
because the mechanism is much more complex calculated according to Equation 10.94, the loss in
than that for molecular transport. As a result it is pressure per hundred feet of pipe would be
necessary to combine theory with experimental 0.663 psi. If an apparatus is physically tested under
evidence to obtain an adequate picture of turbu- the described conditions, the measured pressure
loss per hundred feet of pipe would be 15.4 psi.
lence.
In order to understand turbulent transport, it
The experimental value would be 23 times the
is necessary to examine turbulent momentum
first
calculated value. On the other hand, if the water

transport in a flowing fluid. Once a picture of were flowing at 0.1 ft/sec, the experimental value
turbulent motion has been developed for momen- and calculated value would agree exactly. This
indicates that at low velocities the fluid behavior is
tum transport, it can be extended to heat and mass
transport.
identical to the model of momentum transport in
laminar flow chosen for derivation of Equa-
tion 10.94. At some higher mean velocity, the fluid
TURBULENT MOMENTUM TRANSPORT behavior is not the same as in the laminar regime.
Molecular momentum transport is variously named The purpose of this chapter, then, is to examine
laminar flow, streamlined flow, or viscous flow. As this new regime and to describe the regime as
was shown in Chapter 9, movement of the fluid as exactly as can be done at this time.
a whole is in the /-direction only, whereas momen-
tum is transferred by random motion of individual
molecules in the x-direction. Using this relatively The Reynolds Experiment
simple model, expressions were developed in A classic experiment reported by Osborne
Chapter 9 relating shear stress, diffusivity, and con- Reynolds (10) in 1883 is still used to demonstrate

232

the qualitative difference between laminar and tur-
bulent flow.
Consider a glass tube in which water flows at
any desired mean velocity. At the center of the
tube a fine jet of water-soluble dye is introduced
through a capillary tube so that a thin filament of
dye is injected into the stream of water; the veloc-
ity of the dye stream is equal to that of the water
at the point of introduction. Reference to Figure
12.1a shows that at low water velocity the dye
filament retains its identity in the water stream,
tending to widen very slightly during the down-
stream passage because of molecular diffusion of
dye. At a slightly higher mean velocity, as shown in
Figure 12.1b the dye filament breaks up into finite
large eddies. Farther downstream the eddies break
up further, and the dye that has been introduced
tends to become homogeneously dispersed. At
much higher mean velocity (Figure 12.1c) the eddy
activity becomes extremely violent, and the region
of homogeneous dye color approaches the point of
dye entry. From visual observation it is evident
that the eddies in normal pipe flow are of the order
of one-tenth the pipe diameter, and that the eddies
move completely random pattern. In some
in
experimental studies, the eddy formation has been
suppressed by careful control so that high mean
velocities were maintained with absence of eddies,
but these conditions demand, for example, deaer-
ation of the fluid, insulation of the apparatus from
-Dye-injection needle
Flow-*
(a)
-Dye-injection needle
Flow-*
(b)
Homogeneous-color zone-
Dye-injection needle
Flow —b
"
i i
(O
Figure 12.1.The Reynolds experiment (vertical scale
exaggerated), (a) Flow pattern at low mean velocity
with dye injection, (b) Flow pattern at higher velocity
with dye injection, (c) Flow pattern at high velocity
with dye injection.
TURBULENT-TRANSPORT MECHANISM 233
any external vibration, and supersmooth finish on
the duct work. This evidence supports the notion
that the incidence of eddies is brought about by
external imperfections of the systems of the type
that cannot be practically eliminated. In those
experiments in which high velocity was maintained
with absence of eddies, pressure-drop data were
correlated with Equation 10.94. A system oper-
ating at abnormally high velocity in viscous flow
could be made to revert to turbulent flow by
simply physically jarring the pipe. This indicates
that high-velocity viscous flow is an abnormal or
metastable state.
It should also be noted that an upper limit of
viscous flow as well as a lower limit of turbulent
flow seems to exist and that the limits are sep-
arated by a transition region. The Reynolds appara-
tus and other means of examination (8) also indi-
cate that the eddy formation begins at the center
of the tube to form a central core of eddy activity.
The diameter of the core increases with increase in
mean velocity.
The conclusions that can be drawn from the
Reynolds experiment are: (1) above a certain mean
velocity for a given system, relatively large eddies
form that flow cross-stream in some random be-
havior; (2) these eddies are larger and more abun-
dant at the center of the tube; (3) an increase in
mean velocity of the fluid widens the turbulent
core until the tube is essentially filled with the core
of eddy activity.
Momentum Transport by Eddy Activity
In Chapter 9, the transport of momentum by single
molecules moving in natural random motion was
described. The conclusion drawn there was that the
fraction of the total molecules that possessed a
cross-current component of velocity could move to
regions in which the flow velocity was different
from the origin velocity of the migrant molecule,
with consequent transport of momentum. In cer-
tain respects, turbulent flow is similar to laminar
flow, except that the scale of the migrant mass is
much larger in turbulent flow and that additional
variable dependencies exist, in the Reynolds
experiment the eddies are distinguishable by eye,
which is indicative of the magnitude of the eddies.
Since the eddy has a random cross-flow component
superimposed upon the bulk flow, it is reasonable
to suppose that momentum can be transferred by
eddy cross-flow motion as well as by molecular
cross-flow motion. The visible examinations are
sufficient to indicate that the cross-flow motion
exists on a macroscopic scale.
234 MOLECULAR AND TURBULENT TRANSPORT

The absence of agreement between

^\^ S
fluids Average velocity v in
the flow direction
at higher velocities with equations that are
fjrj
moving at radius r
derived based upon a molecular-transport model
such as Equation 10.94 is sufficient to indicate
\ Instantaneous velocities
that a heretofore unevaluated mechanism is oper- fit' and v + v at radius r
ative in addition to molecular transport. This does
not mean to imply the molecular transport is ab- ^Average velocity u in

sent, because in any fluid sample in which a veloc-

u = o the crossflow direction
(x) at radius r
ity gradient occurs, even within an eddy, random Time-
molecular motion will transport some momentum.
Figure 12.2. Instantaneous point velocities in the x-
A major difference between molecular behavior and /-directions.
and eddy behavior is apparent from the Reynolds
experiments. The eddy activity is violent at the averaged velocity and a fluctuating velocity, or
center of the pipe and decreases in intensity away
from the center. Furthermore, the eddy activity u =u+u
:i2.3)
becomes more violent with increase in velocity. V= V+ V
Subsequent developments will attempt to
where u and v' are the fluctuating velocities in the
evaluate the contribution to the momentum trans-
x- and /-directions, respectively.
port made by each of the mechanisms.
From these definitions it is evident that the
time-averaged values of the fluctuating velocities
Mechanism of Turbulent Flow must be zero.
In simplestform, fluid stress due to combined If a sensitive pressure-measuring device were

laminar and turbulent flow can be represented by used to measure local pressure as a function of
the general equation time, results similar to those of Equations 12.2 and
12.3 would result.
'"total ''molecular
"""
'"turbulent 12.1] The continuity equation and the equations of
motion are still valid for turbulent flow if the
The shear from molecular or laminar
stress
instantaneous velocities and pressures are used. For
processes is obtained from Newton's law as de-
an incompressible fluid of constant viscosity, the
veloped in Chapter 9. To evaluate r turbu ent , ad-
continuity equation (Equation 11.1) becomes for
i

ditional analysis is required.

thex-components
The Reynolds experiment would certainly
suggest that turbulence is characterized by random,

chaotic motion of fluid particles. If an extremely

—7k

ox
W + u) + —
oy
r\

[v + v) + —f\

oz
(iv + w) =
sensitive and responsive device were used to mea-
sure velocity at a point in a fluid in turbulent flow, (12.4)
and if this device were arranged so that the x-
and the equation of motion ;1 1.37)11 1.54) be-
(radial) and /-(axial) components of velocity were
comes
measured, the time response shown in Figure 12.2
would result.
—
p(u + u') = p(u + u'){u + u')
Atime-averaged velocity (that is the average ar Vox
value of the instrument measurements) may be
written + — (u + u')(v+ v)
f
oy
1
u= — r
udt
t -o
+ — (u + u')(w +w')\+ /iV 2 (i7 + u)
(12.2)
1

vdt
+ P9> 12.5)
ox
where u and v are the instantaneous velocities in
the x- and /-directions measured at a point, and Since turbulent fluctuations are random, solu-
and v are the time-smoothed values. tion to Equations 12.4 and 12.5 is impossible.
The instantaneous velocity may then be con- Taking advantage of the fact that the time-
veniently expressed as the sum of the time- averaged value of the fluctuating velocities is zero,
 TURBULENT-TRANSPORT MECHANISM 235

Equations 12.4 and 12.5 can be time-averaged to

give

+ —+—=
dw
—
du dv
(12.6)
dx by dz

dd
d _
pu = - —
dP
dx
- — puu
ox
__ —
+
3
pvu + — puw
3 3

oy bz

—
o __ —
puu+
a _,_ .

puv+--pu '-'1
w
a
-[ ox by dz
.2rrn +
+ p'Vu pg x (12.7)

The terms in the second set of brackets of

Equation 12.7 are associated with turbulent veloc- Figure 12.3. Crossflow movement of an
ity fluctuations, for convenience, let eddy.

( r turb>xx = pu u
/-direction. At region 1 the mean velocity is ^ As .

(^turb)xK
= PU V a result of a pulse, a portion of fluid at region 1
accelerates to y-, + v and adopts a cross-flow veloc-
(^turb)xz ~ P& W ity component u . With these new velocity com-
ponents, an eddy is formed at 1. It is assumed that
These terms are the components of the tur-
bulent momentum flux. They are called the
this eddy retains its identity in cross-flow over the

Reynolds stresses or eddy stresses.

distance (x 2 — Xi 2 is reached where
) until region
the mean velocity is v2 =Vt + /. At that point the

Prandtl Mixing Length

eddy is said to have decayed or been absorbed into
a host region, or lost its identity.
The equations of motion for turbulent flow can be
The velocities at 1 and 2 can be related by
solved only if the eddy stresses can be evaluated.
dv
Prandtl's mixing-length model is one attempt to do
model assumes that eddies of
"2 = vi +— ,

(x 2 -x,) = Vy + v (12.9)
this (7). This fluid dx
move distance before they lose
a definite transverse
— Xt
If (x 2 ) is designated by X, the Prandtl mixing
their identity as a consequence of being absorbed
length, Equation 12.9 can be written for any
into a new environment where the average velocity
region as
equals the instantaneous and fluctuating velocity at
the instant that the eddy started its transverse dv_
v+ v' = v+ X
travel. Prandtl's model did not recognize the ef- dx
fects of impingement of the eddy on others having
a different velocity. This travel distance is called dv
or v' = \ (12.10)
the mixing length. dx
Prandtl assumed that eddies move around in a
similar fashion to molecules. Earlier, for molecular
The rate of mass exchange between the
transfer it was assumed that molecules move from regions per unit area is pu' . Thus, the mean turbu-
lent shear stress on the fluid is
region to region within a fluid carrying with them
theirmomentum. These molecules were further
assumed to travel some mean distance before losing 7"turb -pUV 12.11)
their momentum and mingling with other mole-
cules. The Prandtl mixing length is analogous, The continuity equation requires that*/ and v' are
therefore, to that of the mean free path discussed of the same order of magnitude. Thus, the
in Chapter 9 for molecular processes. This analogy, Reynolds stress is related to the Prandtl mixing
however, is not a very good one because no fluid lengthby
has identifiable eddy particles.
Figure 12.3 represents a portion of a tur-
r turb = p(X-) 12.12)
bulent velocity profile with the mean flow in the
236 MOLECULAR AND TURBULENT TRANSPORT

2
Eddy Viscosity radius is#"i, S= irr-\ and A = 2-nr^ L; and
Boussinesq introduced the concept of an eddy
(1) -AP(TTr, 2
)=T y -
l
2irriL (12.19)
diffusivity of momentum E T whichis analogous to
,

For a part of the tubular section bounded by any

the kinematic viscosity defined by Newton's law. It
radius r,
is defined by

dv -APUr 2 )=T y (2irrL) (12.20)

r turb (12.13)
dx Equations 12.18, 12.19, and 12.13 are similar
to those derived in Chapter 10 for molecular trans-
Comparison of Equation 12.13 with 12.12 shows
port with internal generation. If Equation 12.20 is
that in terms of the Prandtl mixing length
divided by Equation 12.19,
dy_
^\2
£r=A 12.14)

From Equation 12.14 evident that the

dx
it is
-M3 112.21

eddy viscosity is not a physical property but is very so that for turbulent flow, the stress also varies
much dependent upon position and flow con- linearly with radius. If Equation 12.21 is substi-
ditions. tuted into Equation 12.15 with modification of
Equation 12.1 can now be written in more the derivative for cylindrical geometry,
by substituting Equation 12.13 and
specific detail
d(pv)
using Newton's relationship.

7-total = -V
d{pv)
-E r
d(pv)
4t>-^V
When the density is constant and if there is a
« 12.22)

dx dx
known velocity-radius relationship, E T can be
d(pv) determined at various radial positions.
= -!!> ET ) (12.15)
dx
Velocity Distribution in Turbulent Flow
And for a fluid of constant density
Nikuradse (6) measured the point velocity
(in the
dv flowing fluids and
= -<M + pET (12.16) /-direction) of turbulently
Ttotal )

dx found that the velocity profile lost the parabolic

character described by Equation 10.90 and tended
'total
-(»^m
Equation 12.15 does not imply that
» 2 "'
ET is an
to approach plug flow, almost as if the velocity of
the fluid was independent of radial position. The
tendency toward plug flow increased with increase
in mean velocity for a given system, as shown in
increase in viscosity of the fluid at high velocity. It
does imply that additional momentum is trans- Figure 12.4. Note that with increase in velocity the
ferred beyond that predicted by the laminar flow "point" of the parabola is blunted, so that as v-+°°
mechanism alone, and that the transfer may be the flat profile of plug flow appears to be the limit.
represented by the sum of molecular and turbulent Von Karman (13, 14) proposed that the
processes. Nikuradse velocity-distribution data be represented
ET varies from zero at the wall to high values by three separate empirical equations for all
in the turbulent region. It should be remembered Newtonian fluids flowing in "smooth"! tubes
that the force-balance equation developed in Chap- regardless of density, velocity, wall stress, and
ter 10 is always applicable regardless of the flow viscosity for any radial position from the wall. The
mechanism. For any section of fluid flowing velocity profile would naturally include all the
through a conduit of any shape, the net force foregoing terms. The generalized velocity distribu-
applied to the ends of the tube because of pressure tion is given in Equations 12.23, 12.24, and 12.25
drop is (—APS) and the resisting force is t y A. and is presented graphically in Figure 12.5.
These terms must be equated by the force The parameter y + is a generalized position in a
balance flowing fluid system at a particular radial position

-\PS=ry A (12.18)
f'Smooth" means glass tubing or commercial drawn-metal
where S and A are the cross-sectional area for flow tubing. Commercial pipe is somewhat rougher, which changes the
and transfer area. For flow in a cylinder whose velocity profile.
 TURBULENT-TRANSPORT MECHANISM 237

O (a)
^\

J-

(b) (c)
u

y
+ =
nonvectorial
t his is
dimensions are
dimen sionless.
= V(tv<7 c )i lp called the 'frjction_ vel-
ocity" with dimensions vT^TU/rJ^
v = the point velocity at a position r, L y /d
[(/"i
— r)u*p/n] a position param eter
,
,
in use,

Figure 12.4. Velocity distribution in laminar

with vectorial dimensions \/L y IL r In .

and turbulent flow, (a) Low-velocity laminar

nonvectorial dimensions, the param-
flow, (b) High-velocity turbulent flow
eter is dimensionless.
(c) Very high-velocity (y -*•«>) plug flow.
r-\= radius at the boundary, L r
r = any radial position, L r
20 — — ——i i i

| Mir r
2
p = density of the fluid, M/L r Ly,
= absolute viscosity, M/L Y8
I Laminar Buffer fi

sublayer (ty )i = fluid stress at the pipe wall, ? y IL L r .

y

Equation 12.23 will be examined more criti-

cally because the peculiar nature of the velocity and
position parameters requires some elaboration. The
derivation of Equation 12.23 follows. Consider
Equation 10.90 which defines the position-velocity
relationship in laminar flow through a circular
duct.
2
i i i i in (r, -r 2 )

y = ——
20 30
(r, - r)

:,
50
u'p
400

From Equation
10.90)

10.86,
Figure 12.5. Generalized velocity profile for turbu-
lent flow in tubes (14). By permission of ASME; / -APg
-A/><7cC \ (Ty9e)'
copyright 1939. 10.86)
I 2 Ay
Ai/ /

r. The parameter u* is a generalized velocity in a Substituting Equation 10.86 into Equation 10.90
flowing system including the point velocity at y + ,
gives
+
where u and y + are related in Equations 12.23,
12.24, and 12.25 below.
The first region of a turbulent-flow system is
+
bounded by y = 0, which is at the wall of a tube, Equation 12.26 can be rearranged and factored.
+
and y = 5, a short distance from the wall. This is
{T y g c )i ipiry -r)\/(r^ +r)
called the laminar sublayer. In this region, the v =
point velocity and position are related by rj*V2r) (12.27)

u^ =y n
12.23) If r very nearly equal to r u as is the case near
is
+
the wallof the tube (y <5), then the term
The region between the radial positions defined by
+ +r)/2r-\ can be equal to 1. With this modifi-
y = 5 and y + = 30 is called the buffer layer repre- {r-i

cation Equation 12.27 can be rearranged to

sented by the empirical equation
+ +
u = -3.05 + 5.00 In y ;12.24) {r^ -r)pV(T y gc )i/p
12.28)
The region between the radial positions defined by \/(ry gc )^lp
+
y = 30, and the tube center is called the turbulent Combining Equation 12.28 with the definitions of
core and is represented by the empirical equation + +
u and y gives
+ +
u =5.5 + 2.5 lny (12.25)
u- =r (12.23)
The definitions of the terms in these equations are
Equation 10.90 was derived based upon r y g c =
+
u = v/u* , pa ramet er with
a point-velocity —p(dv/dx) and consequently is a laminar-flow equa-
vectorial dimensions \/\. v IL r Where . tion. Therefore, Equation 12.23 is also a laminar-
238 MOLECULAR AND TURBULENT TRANSPORT

flow equation. In turbulent flow the laminar sub- "blunting" of the velocity profile takes place. The
layer is defined as the region extending between blunted profile is the first noticeable difference in
+ = +
y (at the boundary) and y = 5 (a short dis- velocity pattern between laminar and turbulent
tance from the boundary). Flow in the laminar flow.
sublayer is true laminar flow. For a given value of There is one point of inconsistency in Equa-
+ + +
y the actual thickness of the laminar sublayer tion 12.25. The function u = 5.5 + 2.5 In y is

(r-\ — r) depends upon u*. As the wall stress t k1 continuous for all positive values of y* and will
increases (as would be the case if the velocity were have a finite value for the first derivative for all
+
increased) the laminar-sublayer thickness decreases. positive values of y At the center of the tube
.

+
The actual existence of the laminar sublayer r = and y = r- u*p/iJL. Consequently, according
l

has been debatedt for some time. The actual veloc- to Equation 12.25, dvldr must be a real nonzero
ity in this region is very difficult to measure and number the center of the tube. All velocity
at
has always been subject to doubt. There is no measurements show that dvldr is zero at the center

doubt, however, that the velocity distribution in
this region can be approximated by Equation
12.23; therefore, in a practical sense, the assump-
tion of laminar flow in the small region would lead
to an error in theory rather than to an error in
prediction of rate of transport. The buffer
+ +
extends from y = 5 to y = 30. The Reynolds ex-
periment and others have shown that eddy activity
is most pronounced at the center of a tube and is
absent? in the laminar sublayer, so that, as the
name implies, the buffer layer is a region of tran-
of a tube; therefore, Equation 12.25 is incorrect
for the center of a tube and probably in the region
near the center. Away from the center the equa-
tion is a reasonable empirical representation of the
velocity at various points. Actually, it is unreason-
layer able to suppose that the velocity profile should
follow three distinct equations; more likely one
very complex function is correct, but this function
has not been proposed as yet. In spite of obvious
mathematical inconsistencies, the three equations
still provide a basis for calculation that yields valid

sition from the laminar sublayer with no eddy results. For example, even though a serious math-
activity to the violent eddy activity at the core of ematical inconsistency exists when Equation 12.25
the tube. As in the case of the laminar sublayer, is used near the center, it should be remembered

the thickness of the buffer region is a function of that very little momentum is transferred in that
wall stress, so that at high wall stress the thickness region, so even a large percentage error in that
is small. The buffer
region is bounded on one side region amounts to very little actual error when the
by no eddy activity and on the other by
a region of entire tube is considered
a region of considerable eddy activity, so it is In summary, the available information for
reasonable to assume that there is some degree of analysis is:

eddy activity in the region. Velocity distribution in

1. The experiments of Reynolds that show
the buffer layer is experimentally determined with
considerable precision, so that there is little doubt formation of gross eddies moving in random man-
ner, with the eddy activity starting in a central core
about the manner in which the data fit the curve.
The closeness of fit does not imply description of with the core increasing in radial dimension with
mechanism. increase in mean velocity.
The turbulent core is represented by Equation 2. The velocity-profile data of Nikuradse that

12.25. It is the region of maximum eddy activity.

were expressed as three equations by von Karman.
Comparison of the constant 2.5 in Equation 12.25 The three equations represent three separate re-

with 5.0 in Equation 12.24 indicates a decrease in gions of flow behavior: laminar flow near the walls,
rate of change of the velocity gradient with in- limited eddy activity through the buffer layer, and

crease in position parameter or approach to the the turbulent core. The point velocities are cor-
center of the circular duct. The logarithmic func- related in terms of the wall stress, geometry, den-

tion of position also inherently indicates decreasing sity, and viscosity.

gradient. It is in this region that the extreme
Example 12.1. Water is flowing through a
2
2-in. I.D. tube. The pressure drop is 10 lb^/ft per
t See penetration theory. Chapter 13.
foot of tube. Find the point velocity and eddy
$ A few experimenters visually observed some notions that
appear to be eddy activity near the walls (3). This led to another diffusivity positions between the wall
at various
model for turbulence, but the model has not been confirmed. This and the center of the tube (/u=1.0 centipoise,
3
topic will be dealt with later as "penetration theory." p = 62.4 lb/ft ).
 TURBULENT-TRANSPORT MECHANISM 239

Solution. Using the pressure-drop data and Notice the inconvenient number of significant fig-
the geometry, the wall stress t y can be calculated ^ ures required to define the limit of the laminar
from the force balance, Equation 12.18 sublayer. Is the assumption that (^ + r)l2ry =1
(-A/>)Si =(Ty A)i used in Equation 12.27 justified in this case? The
eddy diffusivity ET is zero in this region because by
Rearranging and inserting data gives +
definition the relationship u = y + is a viscous-flow
2
_ (-A/ )5 _ (10)7t(1/12)
>
1
equation based in its original definition upon
7/1 " T y 9c - -p{dv/dr).
~ A 2tt(1/12)(1)
The next logical point to test is / + = 30, the
=0.417 2
tk1 lb f /ft upper limit of the buffer region and the lower limit
From the wall stress and density data the friction of the turbulent core. The velocity at y + = 30 can
velocity u* can be calculated. be calculated from Equation 12.24 or 12.25.
+ +
Notice that between y = 5 and y = 30 Equation
U
* A^-/ (0 417)(32 2
- -
V /2
12.24 is the only equation that applies. If Equation
V p
'
(62.4) / 12.24 is used, then
u* = 0.464 ft/s +
u = -3.05 + 5.00 In 30 = 13.96
v v v = 6.48 ft/s
Then and
u 0.464 +
The position represented by y = 30 can be cal-
+
+
(r-f - r)u*p culated from the definition of y and known
and y = +
values of p, p, and u .

+
Consider first a position at the wall, r = r u y =0. (n -r)= 0.000697 ft
+
Equation 12.23 applies, so that u = and v = 0. = 0.082603
r ft
This, of course, is a basic assumption of fluid flow:
that the velocity of the fluid at a boundary is equal The same result can be calculated from Equation
+
to the boundary velocity. 12.25 since / =30 is a valid limit for Equation
+
Next consider a point at y = 5. This point is 12.24 and 12.25.
the upper limit of Equation 12.23 and the lower
+
limit of Equation 12.24. Using Equation 12.23 u = 5.5 + 2.5 In 30 = 14.0
v = 6.5 ft/s
u+ = y+ = 5
The eddy diffusivity at this point can be calculated
+
By the definition of u from Equation 12.22, provided that dvldr is
^=5(0.464) =2.32 ft/s
known at y* = 30. To evaluate dvldr, Equation
12.24 is differentiated.
If Equation 12.24 had been used,
+ du + _ 5.0
u = -3.05 + 5.00 In 5
+
dy + ~ y +
u = 5.00
+
Differentiating the definition of u gives
v = 2.32 ft/s

The by / = 5 can be
actual position represented
+
+
du+ =
.
—
dv

calculated from the definition of y given above. u

+ +
y P (5)0 x6.72x10- 4 )
and differentiating the definition of y yields
(/-! -r) =
u*p (0.464X62.4) —u*p*—
dr
dy+ =
[ry -r) =0.0001 162 ft
P
+
This is the distance of the point y = 5 from the
Then, by combination of these three equations,
wall for the conditions of this example. The same
position measured from the center is dv -5.00a* 2 p
+
r = (.jV _ 0.0001 162) = 0.0831838 ft dr py
240 MOLECULAR AND TURBULENT TRANSPORT

From the definition of u*. velocity distribution, which utilizes three distinct
equations.
{Ty9c)^
u* 2 The nextradial position that will be con-

P sidered the midpoint between r=

is and r = r 1 .

In this example the point lies well into the turbu-

so that
lent core, and therefore Equation 12.25 applies.
dv - 5.00(7-^)! At r = r, 12,

dr PV*
From Equation 12.22
'M^)= 1795 2u
=

+
5.00( TyKVc'1
u = 5.5 + 2.5ln 1795
g
v= 11.22 ft/s
py
E. can be determined using Equation 12.22 and
(T y g c )-\ cancels, so that by rearrangement the value of dvldr determined by differentiating

(rpy
+ and rearranging Equation 12.25.
)

iv+E r =
)
(/-1) (y + £ r )(2.5p)(7v<7c)i
(5PT,) , ,
\T y g c )i +
+ (2r, (py )
At / = 30, r = 0.0826 ft, and
+
4 (py
(0.0826X6.72 x 10~ )(30)
:

iv + Er = (* + Fr) =
)
(2)(2.5)(p)
[5.00)(62.4)(0.0833)

+ E 7 =6.41 x 10- 5 ft
2 (1 x6.72x 10- 4 )(1795)
(v ) /s
(v+E T )
=
(2M2.5M62.4)
H = 6.72 x 10- 4
v = — = 1.077 x 10- 5 ft
2
/s
£r = 386x 10- 5 - 1.077 x 10~ 5
P 62.4

£T = (6.42 -1.077) x 10~ 5 £T = 385x 10" 5 ft /s

2

2 It is evident that the eddy diffusivity is nearly 400

£7 = 5.33x 10~ 5 ft /s
times the molecular diffusivity of momentum at
the point r = r 1 /2.
Note that ET is five times as large as v. This
At the center of the tube, r = and y~ =
indicates that the momentum transferred by the
r y u*pl\i = 3590.
turbulent mechanism is five times as great as that
+
transferred by the molecular mechanism at a point u = 5.5 + 2.5 In 3590
r- 0.0826 ft defined by the position parameter ^ = 12.02 ft/s
+
y =30. If the velocity can be defined at y~ = 30
by both Equations 12.24 and 12.25, it is reason- Using Equation 1 2.22 at r = gives
able to expect that E 7 might be the same regardless
(^ + Fr )
=
of the equation used. But this is not the case. Using
Equation 12.25 to find dvldr gives ET = -1.077 x 10~ 5

This result shows another serious inconsistency.

The Reynolds experiments indicate visually that
dr PV^ eddy activity is a maximum at the center of the
and pipe, whereas the calculations above show a neg-
ative value. The negative value implies that a neg-
(v + E7 = 12.84 x 10- 5
)
ative transfer of momentum is taking place. Differ-
E.= 11.76 x 10- 5 ft
2
/s entiation of Equation 12.25 results in a finite value
of dvldr at the center of the tube; however, all
This value is slightly more than twice the value experimental evidence points to dv'dr = at the
found using Equation 12.24. Thus, the values of E 7 center of the tube.If dvldr = is substituted into
+
at y = 30 differ because the derivatives of the two Equation 12.22at/- = 0,
equations differ, even though the values of tr
+ (Tygc )^(0)
are the same at y = 30. This discontinuity in = -(v + E T )p(0)
function is a fault with this empirical generalized

This equation is indeterminate, so that no light can
be shed on the eddy diffusivity at the center of the
tube by this method.
All the foregoing indicates that the velocity-
distribution equations are inconsistent in some
spects. The velocity data derived from these equa-
tions are useful empirical values, but the equation
forms are apparently incorrect. If the equations
were correct, the derivatives of the equations
would be valid at all points, and dvldr would be
zero at the center of the tube.
Example 12.1 offers some indication of the
difficulties encountered when working in a flow
regime that is not rigorously described mathemati-
cally. The velocity profile equations result from the
correlation of experimental data without having
available an adequate theory as a foundation. Con-
sequently, the data are reported in a convenient
+ +
orderly manner (as u and y that is not rigorous.
)
The general aim of the experiments was to describe
the velocity profile with special attention focused
upon the condition near the tube wall. No partic-
ular effort was made to describe conditions at the
+
center of the tube, and therefore a term in y at
the center was omitted. All the above emphasizes
that the present empirical mathematical form-
ulations of turbulence are certainly not theoreti-
cally adequate. In order to complete the picture of
turbulence, a rigorous mathematical representation
is required.
From the foregoing material, a model has
been described that results in a mathematical defi-
nition of turbulent flow. There are intuitive objec-
tions to a model that considers gross particles of
fluidmaking small, sporadic motions. Although the
model is probably oversimplified, it can be visual-
ized and analyzed, and it is consistent with obser-
vation.
A more may be derived if each
rigorous model
pulsation in recognized and treated
velocity is
mathematically as an unsteady-state case. An
attempt in this direction is presented in Chapter 13
under "Penetration Theory." Other models of tur-
bulence have been described (1 1,12,14,15)
From Equation 12.15 it is evident that at any
point at which turbulence occurs, molecular trans-
port occurs simultaneously and by a parallel mech-
anism. In Example 12.1 the value of ET was usually
much greater than v. Consequently, although the
molecular mechanism existed, the magnitude was
almost negligible. In systems with high momentum
diffusivities, v may be of such magnitude that a
significant part of the momentum transport is by
the molecular mechanism.
TURBULENT-TRANSPORT MECHANISM 241
The initial production of eddies has not been
discussed. It is generally supposed and supported
experimentally that eddy activity is caused by mi-
nute imperfections in the system (5). The experi-
ments of Reynolds and others confirm this. If
re-
eddy activity begins, secondary eddies are formed
by the primary eddies so that unless some mecha-
nism exists to cause decay, the eddy activity will
multiply indefinitely. The viscous shear of the fluid
and especially the high shear rate near the walls
tend to cause eddy decay so that the system of
eddies comes to steady state as the rate of eddy
generation equals the rate of eddy decay.
Extensive investigations have been made of
the size of eddies and their energy content in
relation to that of the overall stream. It is to be
hoped that the elaborate studies of turbulence, its
origin, and other factors will give complete under-
standing sometime in the future. The over-
simplified developments of the counterpart quan-
tities in molecular and turbulent transport should
present an adequate picture of the radical differ-
ences between them. Simultaneously, the similarity
of the mean free path of the molecule and the
distance traveled by an eddy before it loses its
identity and of the molecular speed and cross-flow
velocity should be appreciated.
It is necessary to appreciate the existence of a
laminar-flowing layer and a turbulent core, both
vaguely bounded and overlapping in a buffer or
transition layer.When the flow pattern is known,
extension of the same concepts to transport of heat
and mass follows simply and logically.
HEAT AND MASS TRANSFER IN TURBULENT FLOW
The eddy activity in turbulence serves as a transfer
mechanism for momentum. In addition, since gross
particles of fluid move in cross-flow, this eddy
activity also serves as a medium for physical mix-
ing. The physical mixing of fluids is important in
mass and heat transfer. Consider boundary
a wall
that is capable of delivering heat or mass to a
flowing stream. If the concentration of transferent
property at the wall is significantly higher than in
the main stream of fluid, a gradient will be set up
at the wall, and transfer will occur into the fluid.
After steady-state operation prevails, the concen-
tration gradients in the various sections of the fluid
streams will be shown in Figure 12.6 and as de-
scribed below.
The gradient across the laminar sublayer is
1.
with most of the difference in concentration
large,
between the wall and center of the fluid occurring
242 MOLECULAR AND TURBULENT TRANSPORT

^Laminar sublayer, no mixing

/ /'Buffer region, mild mixing

Boundary -

—
Radial position (r)~*
Figure 12.7. Transfer in turbulent flow.

Figure 12.6. Heat and mass transfer in turbulent

flow. Variation of concentration
property with radial position (not to
across the very thin laminar sublayer. Molecular
transfer occurs across this layer, with the rate de-
pendent upon the gradient and molecular
fusivity.
2. The gradient across the buffer region is
considerably less than that in the laminar sublayer
with less difference in temperature or concen-
tration across the buffer region than
laminar sublayer.Eddy motion tends to carry (by
of transferent Equation 12.30 can be used in a transferent-
scale). property balance to develop a relationship for the
flux at any position in terms of the flux at the
boundary. Consider a test section extending from y
to y + Ay of radius r\ (see Figure 12.7). In Figure
12.6 the concentration at all radial positions was
dif-
described as being very nearly constant except in
the small region of the laminar sublayer and buffer
region. Assume that, after passing through the test
section, the increment of concentration Ar =
— rK
Fy + Ay constant at all radial positions.* As
is
across the
shown in Figure 12.7, the transferent property
entering the flow section at y is vST y and the ,

the mass-mixing process) some of the transferent

transferent property leaving the flow section at
property from the buffer region into the turbulent
y + Ay is vSr y + Ay The transferent property leav-
.

core and to return fluid containing less of the

ing the test section through the wall (note that the
property or solute from the turbulent core. In
transfer is taken as positive with increasing r) is
addition, since a gradient is present, some of the
A. S is the flow-tube cross-sectional area, and A is
property will transfer by the molecular-transport
the transfer area. The balance equation may be
mechanism.
written in terms of cylindrical geometry for a sec-
3. The gradient through the turbulent core is
tion extending from the center to radius r as
very small compared to the buffer region. Some
transfer will occur because of the molecular-trans- virr 'T y = virr
2
r y + Ay + ^2nr Ay
port mechanism, but most of the transport will
occur by physical mass mixing. Any heat or solute or, rearranging,

that passes through the buffer region will circulate vr

rapidly throughout the turbulent core. The rapid = r -Fy + Ay) (12.31)
T~<
Ay v
circulation tends to offset the establishment of a
gradient, and the small gradient reduces the Equation 12.31 may be written for a test section
amount of transfer that can occur by molecular extending from the center to the boundary at r-\
transport.

A general equation for transfer

the turbu- in 01 =p-(r y -r y + Ay ) (12.32)
Ay
lent regime may be written in terms of the eddy
diffusivity of transferent property. If Equation 12.31 is divided by Equation 12.32,

<// = -(S + £)^ (12.29) ^ = ^^(y) (12.33)

where E = eddy diffusivity of transferent property.

This equation can be applied to cylindrical
* For this condition to hold, the surface concentration would
geometry by modification of the concentration
have to change by AT over the range Ay. T is usually dependent
gradient.
upon conditions outside the flow tube, and the assumption would
probably not be rigorous. However, the small thickness of fluid
dT
yV = -(5 + E) (12 30) involved in that region and the low velocity of fluid in that region
dr would have little effect upon the entire behavior of the system.
 TURBULENT-TRANSPORT MECHANISM 243

Equation 12.33 the flux at any radial posi-

In momentum, equal to gc/. Using the same reason-
tion is to be linear with radius, zero at r = 0,
shown ing in turbulent transfer, it is reasonable to assume

and a maximum at r = r^ This condition is unique

. that E N = Eg = ET = A 2 | dvldx |; thus, EQ and
to "transfer with internal generation" as described £/v can be evaluated from knowledge of wall stress
in Chapter 10 and shown in Equation 10.68. The and velocity profile. Evidence substantiating this
similarity of these equations suggests that transfer assumption will be included in the next chapter
through the wall is similar in behavior to the when the analogous expressions are introduced.
internal-generation application described earlier.
This is actually the case. Even though the initial
Example 12.2. From the data given in Ex-
mechanism of transfer through the laminar sub-
ample 12.1 calculate dT/dr in the laminar sublayer
,

layer simple transfer, the intense physical mixing

is
and also at r = r 12 when the heat flux at the wall
}
in the buffer layer and turbulent core transfers the
and directed toward the wall is 10,000 Btu/hr ft 2 .
property rapidly enough so that it appears almost
Assume that the average properties of the fluid are
instantaneously in all parts of the turbulent core.
those given in Example 12.1.
In this sense, the rapid transfer is similar in be-
havior to the internal-generation applications Solution. In the laminar sublayer, all heat
described Chapter 10. The transfer is not abso-
in passing through the wall must pass through the
lutely instantaneous, as can be seen from the exis- laminar sublayer, and the mechanism is simple
+
tence of a small gradient {dT/dr) in the turbulent transfer. From Example 12.1, at y = 5, (r^ — r)
core, as shown in Figure 12.6. The small gradient = 0.0001162 ft. The laminar sublayer is thin
indicates that there is some transfer by the molec- with respect to the pipe diameter so that the
ular mechanism. Thus, two parallel transfer mech- change in transfer area through the laminar sub-
anisms are operating with transport due to turbu- layer is negligible; therefore, from Equation 9.43 in
lence or physical mixing, usually predominating. which dT/dr is taken equal to dT/dx, the thermal
An exception the case of heat transfer in liquid
is conductivity of water = 0.354 Btu/hr ft 2 (°F/ft).
metals, in which the thermal diffusivities are ex- Then,
tremely high; therefore, a significant amount of
dT i q\ 10,000
transfer can occur with a small gradient. = = = - 28 300 ° F/ft
Equation 12.33 can be substituted into Equa- * -U) -E^i -

tion 12.30
At r = r- /2, from l
Example 12.1, E T =
385 x 10~ 5 ft 2 /s, which the assumed value of
h ©• [5 +
dr
! 12.34) Eq . Other data are: ry
is

=1 in., p = 62.4 lb/ft

3
,

cP = Btu/lb°F. These values are substituted into

1
Specific expressions for mass and heat transfer can
Equation 12.36 in consistent hour units.
be written by analogy.

For mass transfer

I9I\
12
(10,000)

A Vi 12

For heat transfer = -[0.354 +(62.4)(1)(3600x 3.85 x 10" 3 )] —

/"/7"

dr
d(pcp T)

(AItH^ 1
dr
Therefore,

dT
or = -5.78°F/ft
dr

+ <12 361
(D 1 K)-* '**»* -

This chapter emphasizes the similarity be-

Equations 12.35 and 12.36 are working equa-
tions if the diffusivity and gradient can be eval-
uated. In Chapter 9, the simple molecular transport
theory shows equal diffusivities for mass, heat, and
tween heat, mass and momentum transfer for a
point between r = and r = r-\ and between y = y
and y = y + Ay. In the following chapter, the func-
tions described here will be integrated with respect
to r to form transfer coefficients that are valid at
244 MOLECULAR AND TURBULENT TRANSPORT

any axial position (/). Along with the integration, 5. Knudsen, J. G., and D. L. Katz, Fluid Dynamics and
the analogous behavior will be examined further. Heat Transfer, McGraw-Hill, New York (1958) p. 107.
Von Karman's equations are adequate 6. Nikuradse, J., VDI-Forschungsheft, 356 (1923).
7. Prandtl, L., Essentials of Fluid Dynamics, Hafner, New
empirical relationships for predicting the velocity
York (1952) p. 118.
profile turbulent flow in circular conduits. In
in
8. Prengle, R. S., and R. R. Rothfus, Ind. Eng. Chem., 47,
this chapter, the equations were differentiated in
p. 379 (1955).
order to evaluate the eddy diffusivity and Prandtl 9. Reynolds, A. J., Turbulent Flows in Engineering, John
mixing length. Although such a differentiation is Wiley & Sons, New York (1974).
it proved to
theoretically correct, lead to impos- 10. Reynolds, O., Phil. Trans. Roy. Soc. London, Ser. A.,
sible values of the eddy diffusivity at several posi- 174, p. 935 (1883).
tions in the flow channel. This is an example of the 11. Taylor, G. I., Proc. London Math. Soc, 20, p. 196
weakness of empirical equations with inadequate (1921).

theoretical bases. Such empirical equations often

12. Taylor, G.I., Proc. Roy. Soc. London, Ser. A., 151,
421 (l-IV) (1935).
p.
cannot be extended beyond their intended pur-
13. Von Karman, T., Engineering, 148, pp. 210-213
pose. It is hoped that adequate simple expressions
(1939).
for the velocity profile in turbulent flow may ulti- Von
14. Karman, T., Trans. ASME, 61. pp. 705-719
mately be developed. Such adequate expressions (1939).
would lead to correct values of eddy diffusivity 15. Von Karman, T., NACA, Tech. Mem. No. 611 (1921).
after differentiation by the methods outlined in 16. Von Karman, T., J. Aeronaut. Sci., 7, p. 1 (1934).
this chapter.
Very active research and coordination studies
concerning turbulence are going on in many PROBLEMS
places. The importance of the scale, intensity, and
distribution of any turbulence activity would
12.1 A Newtonian fluid flows through a circular duct
clearly be indicated by Figure
12.2 as being a
of radius R. The velocity distribution is given by
statistical function in time, and much of the work
is in the direction of attempts to evaluate the on
u = "max! 1
-
statistical behavior (4,9).
As has been indicated in all of the discussions,
What is the average velocity? How does this value compare
there are simultaneous transports of the various
with a fluid flowing in laminar flow having the same u max ?
properties, both within the fluid and at the bound-
Discuss the effect of decreasing the exponent from 1/7 to
aries. In addition,the behavior of jets and flows off
1/12.
of the agitatormixing operations as well as
in
12.2. Equation 12.25 may be written for conditions
floating droplets can involve the same phenomena.
at the center of a pipe where the velocity is maximum. For
For interpretation of experimental results and pre-
this condition develop an expression for the ratio of
dicting behavior, it has been found considerably
"/"max-
more effective to resort to mechanism ratio expres-
12.3. Find the radial position in the turbulent core at
sions of a phenomenon (2).
which the eddy diffusivity is a maximum.
The analogies between the transports of 12.4. Air is in turbulent flow in a smooth tube of 3
momentum, heat, and mass that constitute an in. I. D. The air is at 70°F and 1 atm, and the pressure drop
important part of the unit operations are based 2 per foot of tube.
(-A/>) is 0.025 lty/ft
upon the mechanism ratios for the different trans-
(a) Calculate and plot the velocity as a function of
ports that are occurring.
radial position from the wall to the center of the tube on
linear coordinates.

REFERENCES (b) Calculate and plot the eddy diffusivity (E T ) as a

function of radial position from the wall to the center of

the tube. Discuss the resulting plot, with particular regard
1. Boussinesq, T. V., Mem. Pres. Acad. Sci., 3d ed., + = + =
XXIII,
to the eddy diffusivity at y 5, at y 30, and at the
p. 46, Paris (1877).
center of the tube.
2. Davies, J. T., Turbulence Phenomena, Academic Press,
New York (1972). 12.5. The pressure drop for molten sodium at 300°
3. Fage, A., and H. C. H. Townend, Proc. Roy. Soc flowing through a tube 1 in. I. D. is 230 lb//ft 2 over 10 ft

London, ser. A., 135, p.656 (1932). of tube. The flow is turbulent.
4. Hinze, J. 0., Turbulence, 2nd ed., McGraw-Hill, New (a) Calculate the thickness of the laminar sublayer
York (1975). and of the buffer layer.
 TURBULENT-TRANSPORT MECHANISM 245

(b) Calculate and plot the velocity as a function of (a) Calculate the shear stress at the wall.
radial position from the wall. (b) Calculate the thickness of the laminar sublayer.
Plot the velocity profile for laminar flow which Calculate the eddy diffusivity at+ + =
(c) (c)
y = 5 and y
has the same maximum velocity as the case for turbulent 30. Discuss.

flow in part (b). 12.8. Discuss the applicability of the Prandtl mixing
(d) Would the two cases of parts (b) and (c) have the length near the center of the turbulent core.

same average velocity? 12.9. Show that with the assumption u + = y + in the
(e) What is the pressure drop in part (c)? laminar sublayer, the eddy diffusivity is not zero from
+ +
12.6. A valve on a 2 in. line carrying benzene at 70' y = to y = 5. Explain the reason and suggest a correc-
tion so that ET = +
can be used to adjust the flow rate. Pressure gages measure at all values of k between and 5.
the pressure drop across the pipe for various flow rates. For 12.10. Calculate the eddy diffusivity and Prandtl

pressure drops of 0.01 and 1.0 lb^/ft 2 per foot of tube: mixing length for the conditions of Problem 12.5:
(a) Calculate and compare the thicknesses of the (a) At/ + = 2.

laminar sublayer for turbulent flow. (b) At y + = 5.

(b) Calculate and plot the velocity as a function of (c) At k + = 20.
radial position for turbulent flow. (d) At/ + = 40.
(c) The pressure drop of 0.01 lbf/ft
2 indicates suf- 12.11. For heat transfer in turbulent flow, calculate
ficiently low velocity to give laminar flow. Calculate and the temperature gradient (dTldr). The heat flux through
plot the laminar-flow velocity profile for the pressure drop the wall is 500 Btu/hrft 2 .

on the same plot as in part (b). (a) In the laminar sublayer.

12.7. The velocity at the center of a 6 in. duct was (b) At/ + = 25.
found to be 40 ft/s for air at 70° F and 1 atm. At this (c) At r = r-i 12 for the conditions of Problem 1 2.4.
velocity turbulent flow exists.

Fundamentals of Transfer
Mechanisms
In Chapter 12 turbulence was studied in detail in
an effort to describe the individual mechanisms
that are operating at each radial position at steady
state. The mathematical manipulation of velocities,
gradients, and diffusivities at points in a system
cumbersome. would be advantageous for design
It
purposes if the same information were available in
the form of simpler functions. The simple func-
tions should respond readily to mathematical
handling. In addition, the data required to evaluate
the functions should not be difficult to determine
and, if possible, should be in terms of gross dimen-
sions, mean velocities, and the physical properties
of the system. Equations expressed in gross dimen-
sions and mean velocities are the result of integra-
tion between boundaries of equations describing
point conditions such as Equation 12.15.
integrated equations describe overall conditions at
some position along flow duct and define transfer
a mena can be analyzed adequately to yield such
cannot be
coefficients. Certain of the integrations
performed by ordinary analytical procedures, in
which case the integrated form may be indicated or
defined and then evaluated or correlated by
empirical methods based upon experimental data.
Mechanism Ratio Analysis
The form of expressions for momentum
general
transport derivesfrom differential equations
indicating the contribution of each mechanism.
For certain cases, these differential equations were
derived in Chapter 11. Even though these equa-
tions cannot be integrated in the general case, they
yield valuable information. Since they are rigor-
ously derived from fundamentals, the terms must
be dimensionally homogeneous. Each term
expresses the contribution of a particular mecha-
nism. Valuable information can be gotten by divid-
Thirteen
ing through the equation by any one term; each
quotient so generated indicates in a dimensionless
form the ratio of the contributions of the mecha-
nisms in the numerator and denominator.
is If such differential equations can be inte-
grated, there is no profit gained by investigating
these dimensionless ratios; if they cannot be inte-
grated, but each contributing mechanism can be
identified, then these ratio terms are useful in
interpreting the contribution of each mechanism.
If the phenomenon is so complex that negligi-
ble knowledge can be gained from investigation of
the differential equations, then empirical processes
are available for evolving dimensionless groupings
of the involved variables. The groupings are
extremely helpful in planning and correlating
The experimental programs of investigation, or in inter-
preting the results of experiments. If the pheno-
dimensionless groupings expressing the contribu-
tion as in Table 13.1 these can be very valuable. If
,
only the individual variables can be
identified — with some uncertainty, perhaps —
then these identified variables can be grouped into
dimensionless terms expressing certain ratios in one
of the techniques of dimensional analysis, as
described in Appendix A, and resort must be had
to these techniques.
In addition to the definition and evaluation of
transfer coefficients, the analogous behavior of
mass, heat, and momentum transfer is examined in
this chapter. Further applications of mass, heat,
and momentum transport are dependent upon con-
ditions other than those heretofore described. For
example, in certain momentum-transfer phenom-
ena, the velocity distributions vary with axial
position as well as with radial position as in the
246
 FUNDAMENTALS 01 TRANSFER MLCHAMSMS 247

Table 13.1 DIMENSIONLESS GROUPS IN TRANSPORT PHENOMENA

Group Name Mechanism Ratio

uLp inertia forces

Reynolds number
viscous forces

P pressure forces
Euler number
pu 2 inertia forces

inertia forces
Froude number
i-9 gravity forces

Tylp wall shear forces

Friction number*
u2 inertia forces

p.lp CpP molecular diffusivity of momentum

Prandtl number
klCp p k molecular diffusivity of heat

h wall heat transfer rate

Stanton number
Cppu heat transfer by convection

hL temperature gradient at a boundary

Nusselt number
k temperature gradient across the fluid to the boundary

CpPuL heat transfer by convection

Peclet number
k heat transfer by conduction

molecular diffusivity of momentum

Schmidt number
molecular diffusivity of mass (at end, and recluster)

kc L concentration gradient at a boundary

Sherwood number
Q concentration gradient across the fluid to the boundary

h heat transfer by convection

k c Ch heat transfer by mass transfer

*This ratio defines the friction factor, but it is not the one that is used for calculational purposes in

this text.

entry length to a pipe or around bodies that offer motion are used to describe a process called
large projected areasnormal to flow. These applica- mechanism ratio analysis. The mechanism ratio
tions are examined in a section devoted to skin analysis simply requires an awareness of the various
friction in the boundary layer on a smooth surface mechanisms involved with the process. There are
and form drag due to accelerative forces in fluids quite likely to be fewer mechanisms involved than
flowing over and around bulky shapes. In addition, variables. Typical mechanisms are momentum
heat and mass transfer in laminar flow, heat transfer by molecular interaction and heat transfer
transfer associated with motion due to natural due to eddy activity. Mechanisms are easier to
rather than forced convection, and heat transfer predict than variables.
associated with condensing vapors and boiling The Navier-Stokes equation (11.55) for a

liquids are considered in this chapter. fluid of constant viscosity and density may be
In Appendix A, the ordering of data by written for the x-direction as
dimensional analysis is described. According to
dimensional analysis, the variables influencing a
particular process must be known or assumed
—
du
+u —+ —+ w—=
du du du g
v
x 1
—
dp
dt dx dy dz p p dx
before the analysis can be completed. Unless some
u u u
intuitive feel about the process is known, pertinent
variables may be overlooked. Here equations of
,(
V 9x'
+ —
9
r+
by"
9

dz<
(13.1)
248 MOLECULAR AND TURBULENT TRANSPORT

Equation 13.1 is a dimensionally homoge- The equation of motion can be used also to
neous equation with each term having units of form other dimensionless groups. Some of these
2
(m/s) /m, and division by one of the terms will are listed in Table 13.1.
yield dimensionless ratios or groups (24). Further-
more, each term or groups of terms in the equation THE STRESS MEMBRANE MODEL
has physical meaning. The first term, du/dt, is the From the discussion in Chapter 12, a simple model
fluid acceleration in thex-direction. The next three of a stream can be postulated. The
turbulent
terms, udu/dx + vdu/dy + wdu/dz. are inertia stream consists of a laminar portion of the fluid
terms. The terms on the right side of the equation near the boundary and the turbulent core charac-
represent field forces, static pressure forces, and
terized by eddy activity. At a low mean fluid
viscous forces, respectively. The variables in the velocity, the turbulent core does not exist; there-
equation are velocity, length, pressure, viscosity,
fore, for simplicity, laminar flow will be examined
density, and gravity.
first. In laminar flow the fluid moves in the flow
The method of using the differential equa-
direction with no component of velocity in any
tions of motion to obtain mechanism ratios or
other direction. Stress is established in the fluid
dimensionless groups is illustrated by the following
according to the expression r v = —p[dvldx), as de-
example.
rived in Chapter 9. Thus, the linear-flow path and the
stress are interrelated, so that the stress results in
Examples Develop dimensionlessgroups
13.1. maintaining the flow linear, and the linear flow
from the equation of motion for the flow of produces the stress. For the case of laminar flow
a viscous fluid flowing in a horizontal pipe. through tubes, the tube may be considered to be
Flow is at steady state and gravitational forces can filled with an infinite number of concentric tubular
be neglected. stress membranes, each of which tends to confine
the flow within the annuli formed between the
Solution. Equation 13.1 may be written for membranes. The strength of the membranes (or the
this case as stress in the fluid) increases linearly from zero at
the center to a maximum at the wall (see Equation
u —
du
+ — + w—
du du
v
1 dp &u
+
dhj_
+
8V 10.85).
dx dy dz < dx 2
dy
2
dz
2 t
If some eddy-forming event occurs, a portion
of the fluid may be subject to a force tending to
A
give a component of velocity in a direction other

Item Mechanism Variable

than y. Opposing this force, the stress membrane
will tend to confine the flow in the orderly
A Inertia forces u 2 /L manner. If the kinetic energy of the distorted fluid
B Pressure forces PlpL
element is high enough, it can penetrate the stress
C Viscous forces vulL 2
membrane and form a true eddy. In this model, the
If each item in the equation is divided by the region of zero stress at the center of the tube is the

inertia term, two dimensionless groups

are formed. most likely portion of the tube to exhibit eddy
activity, which is in agreement with observation. In
In terms of the variables given above, the first
group is addition, if an eddy forms and penetrates the stress
membrane, the penetration continues until a region
pressure forces of high stress is encountered, high enough to over-
= Euler number come the kinetic energy of the eddy. At this point
pu inertia forces
the eddy is absorbed and the orderly flow resumes.
The second group is The region of highest stress is in the vicinity of the
wall, so that little, if any, eddy activity occurs near
v viscous forces the wall. This hypothesis is also in agreement with
Lu inertia forces observation. At the wall, the stress membrane is
supported by the wall, since the wall is strong
The latter group is most frequently seen in its enough that no eddy can penetrate it. The support
reciprocal form, or by the wall extends into the fluid to form the
laminar sublayer, in which no eddy activity occurs.
uLp _ inertia forces
This hypothesis is also in agreement with observa-
= Reynolds number
p viscous forces tion.
 FUNDAMENTALS OF TRANSFER MECHANISMS 249

REYNOLDS NUMBER-FRICTION CORRELATION (r y A), = (-AP)S, [13.5)

In Example 13.1, a ratio of inertia forces to viscous In terms of momentum, the force balance
forces was formed, and this ratio was defined as momentum to the
states that the rate of transfer of
the Reynolds number. It was originally proposed wall equal to the rate of momentum
(TyA)-i is

by Sir Osborne Reynolds about the middle of the transfer from the fluid (-AP)Si The force balance .

nineteenth century as a criterion to characterize represents total forces and therefore total momen-
the nature of flow in ducts and pipes. Many tum transfer irrespective of mechanism. For a
additional uses of the Reynolds number are found cylindrical duct, A] = -nDL and 7
S-\ = nD /4 where

in all of the areas of transport phenomena. What- L is the duct length. Therefore, Equation 13.5 can
ever its however, the physical meaning is
use, be written as
always the same.
The stress-membrane model can be used to =
( -AP)D _8pv
*V1 (13.6)
establish a relationship between the Reynolds AL D
number and the stress phenomena occurring in the
or
fluid. For ease of analysis, it is desirable to use
mean velocities and gross system dimensions. Let AP)D Z2\iv
4(tk1 ) = 13.7)
us consider the flow to be through a circular duct, D
and begin the analysis by considering laminar flow.
In laminar flow, the relation between the
In a system which turbulent flow is also
in

local velocity and the mean velocity in a circular

present, the turbulent mechanism must be included
duct is given by
in the analysis. The stress-membrane model con-
siders an eddy with cross-flow-directed kinetic
energy sufficient to pierce the stress membranes.
10.95) The kinetic energy of a particle of mass (m) mov-
\ i
r-t
2
ing in the /-direction at velocity v is mv 12. For

where v is the average velocity and r^ is the duct fluids,it is more convenient to use density than

radius. This equation can be differentiated to give mass; the kinetic energy per unit volume of a fluid
element of volume dV is d(pv 2 12). If v is known as
Avr a function of position in the flow duct and the
(13.2)
dr
7T average velocity, the kinetic energy per unit
volume of fluid may be found by integrating over
Substitution of Equation 13.2 into Equation 9.31 the duct radius (62). It becomes
for Newtonian laminar flow gives

dv _ Apvr 13.8)
r„ = (13.3) V 2<x
2
dr~ n
where a is a number between 0.5 for laminar flow

At the duct wall, t y = ty -\ and r -fy § so and unity for plug-flow conditions in which the
velocity is invariant across the duct. Because eddy
Apv Spv activity exhibits a random behavior, the mean y-
= (13.4)
*V 1
D directed kinetic energy be of all eddies may
n assumed proportional to the kinetic energy of the
2
where D is the duct diameter. Thus, the stress at fluid, as represented by pv 12a or multiples there-
2 12*
the wall is written in terms of easily measured of such as pv Thus, it may be assumed that
dimensions. Equation 13.4 states that the total pv 2 12a. is a gross mathematical description of the
transfer of momentum to the wall, rKlf is equal to turbulence mechanism.
the rate of transfer of momentum through the Consider the laminar flow mechanism
fluid by molecular transport. These two mechan- described by Equation 13.7. As the mean velocity
isms are equal in laminar flow because no other is increased, some value will be reached where
mechanisms are operative in the system described. momentum transfer by the turbulent mechanism
The determination of wall stress is usually
indirect,but is related to the pressure drop per unit
length of duct through a force balance made on an * In most instances of turbulent flow a varies from 0.9 to
element of fluid. 1.0. See Figure 20.2.
250 MOLECULAR AND TURBULENT TRANSPORT

2 constant cross section

will begin. If Equation 13.7 is divided by pv 12,

IV
8t h 2(-AP)D 64/u vp = constant
2 2
13.9)
p-v v pL Dvp
The product vp is constant for any fluid
Each term in Equation 13.9 is a dimensionless irrespective of density changes due to pressure or
ratio of two mechanisms. The two terms on the temperature effects in compressible fluids. The
left represent the ratio velocity adjusts density changes. This useful
to
product is called the mass velocity and given the
momentum transfer
Total symbol G = vp.

momentum transfer by turbulence Using this symbol, the Reynolds number is

frequently written as
The mathematical expression for this ratio
defines the friction factor (f) where Dvp DG Dv
/v Re = 13.12)

8tk1 _ 2(-AP)D
f = 13.9a)
2
pv2 v pL
FRICTION FACTOR FLOW EQUATIONS
The right-hand term of Equation 13.9 repre- Figure 13.1 is a logarithmic plot of friction factor
sents the ratio Reynolds number over a range of
as a function of
Reynolds numbers between 100 and 10,000,000
Momentum transfer by molecular transport
for flow in smooth tubes.
momentum transfer by turbulence The data for the plot were taken over a wide
range of each of the variables (velocity, density,
The above treatment is another way of tube diameter, etc., using liquids and gases) in
arriving at the Reynolds number as seen earlier in smooth tubes. The system was operating in well-
Example 13.1. developed flow. Well-developed flow means that
Equation 13.9 may be written the velocity pattern in the tube is the same at all

points along the test length. For the tests all

f=64//V R ,
(13.10) pressures were measured far enoughdownstream of
the tube entrance so that entrance effects do not
where
contribute any error. The subject of entrance
Dvp effects is covered later in this chapter in the section
/V Re = Reynolds number =
P on boundary layers. An important feature to
notice is thatthe data fall into two distinct
and the equation holds for laminar flow only. curves separated by a sharp discontinuity at
Since f and A/ Re each represent dimensionless 2100 < A/ R e < 3500. Curve A in Figure 13.1 is

ratios of two mechanisms, Equation 13.10 may be

written more generally as / =0(A/ Re where the r

functionality will be dependent upon the relative

effects of the molecular and turbulent mechanisms.
The Reynolds number can also be written in a
very useful form by recognizing that the mass flow
of fluid through a duct is given by

w= pvS (13.11

where
w = mass flow
10" 2 3 5 1Q
J
2 3 5 10
4
2 3 5 10'"
1
2 3 5 irj
h
2 3 5 10'
v = mean velocity
p = fluid density
Figure 13.1 Variation of friction factor with
S = duct cross-sectional area
Reynolds number. (A more complete chart is

At steady state for flow through a duct of presented in Appendix C-3.)

 FUNDAMENTALS OF TRANSFER MECHANISMS 251

continuous from low Reynolds numbers and ends 2

The groups (—AP/p) and (v 12) will be
at /V R e = 2100. This curve is in exact agreement encountered in the mechanical energy balance for
with Equation 13.10. It is in this range of Rey- fluids, in Chapter 20. Another definition of the
nolds number that all flow is laminar. According to friction factor that may be encountered elsewhere
the stress-membrane model, it is in this region that (1) is

no incipient eddies have sufficient energy to break

through the stress membrane. Above /V Re = 3500 (-AP)D
13.14)
in smooth tubes the fluid is normally in turbulent
2\?Lp
flow, as represented by curve B in Figure 13.1. It is
in this region that eddy activity is violent enough
Then
to break through the stress membrane and that f=Af
momentum is transferred by eddy activity as well
as by molecular transport. The
difference arises from the use of r as the
In the range between = 2100 and /V Re =
A/ Re linear term in the Reynolds number, and from
3500, unstable transition flow occurs, and laminar canceling out 2 from the expression of kinetic
flow, turbulent flow, or a combination of both energy. The form described in Equation 13.14 is
may exist. Behavior in the transition region is a not used in this book. When using friction-factor—
function of fluid properties, system geometry, Reynolds-number plots, or the equivalent equa-
system kinematics, and system history. These data tions, note carefully the definition of f associated
are consistent with the stress-membrane model and with that particular correlation.
with the observations described in the Reynolds The data reported in Figure 13.1 may be
experiment in Chapter 12. The simplicity of the expressed by several equations. Curve A, in the
plotted data for wide variation of all variables laminar-flow region is represented by Equation
confirms the usefulness of the analysis. 13.10,
As discussed above, the friction factor is pro-
64
portional to the ratio of the momentum loss of the f= (13.10)
fluid and the momentum loss by eddy activity. N Re
The friction-factor— Reynolds-number correlation
chart, Figure 13.1, based upon smooth tubes and
is
Equation 13.10 is also plotted in Appendix C-3.
Several equations all empirical, are offered for
consequently is dependent upon skin friction. Skin
friction is that portion of the fluid friction associ-
the turbulent-flow region; the first is valid over the
limited range for A/ Re between 5,000 and 200,000
ated with a tangential force on a smooth surface
oriented parallel to the direction of flow. Refer-
ence will be made to form friction or form drag,
0.184
f = 13.15)
(0.2
for nontangential forces later in this chapter. In (/V Re'
contrast to smooth tubing the inside surface of
commercial pipe is not "smooth" in the sense that A plot of f/8 as a function of /V Re is also plotted
the word is used here. The friction for new com- in Appendix C-5. A more exact equation that is
mercial pipe in the turbulent regime is some 20—30 valid over awider range, for /V Re between 3,000
percent higher than for smooth tubing. This and 3,000,000 is
behavior of pipe of varying roughness will be dis-
cussed in Chapter 20. 0.5
f = 0.00560 + 13.16)
A number of expressions of friction factor - 32
(/v Re )

appear in present-day literature. None of these

expressions differ the overall concept. The dif-
in Forms of other equations may be postulated from
ference lies in the value and placement of a con- the mechanism-ratio analysis. The curvature of
stant. The friction factor used in this book is curve B indicates that the first term of the infinite
written as series is not an adequate description of this phe-
nomenon. The choice of a graphical representation
or a more complex equation must be faced.
In Equations 13.15 and 13.16 the velocity
f= :i3.i3) appears in both the Reynolds number and the
friction factor. When velocity is the unknown vari-

2 able, these equations are inconvenient, since a

252 MOLECULAR AND TURBULENT TRANSPORT

is required. Equation 13.17

trial-and-error solution Assuming turbulent flow, Equation 13.17 can
is another equation form for the same data so be used.
arranged that velocity appears in the left term
1
only. = 2log 10 (/V Re \^F)-0.80
Vf
— = 2 log 10 (/V R eVn - 0.80 (13.17) Expanding f and A/ Re gives
y/f
z
v Lp
+
Note the similarity in form between lA/fand u
+ 2(-AP)gc D
of Chapter 12 and also (N Rey/f) and y .

The foregoing equations are the results of 2(-AP)gc D

= 2.og 10
analyses of experimental data and are therefore
empirical, but the means of presentation has at
(^ v 2 Lp
)- 0.80

least some basis in theory. The velocity may be canceled in the right-hand
term to give
Example 13J. Calculate the pressure drop 2
,72
v Lp
through 100 ft of smooth tubing for an oil flowing

at a mean velocity of 8 ft/sec. The tubing diameter 2(-AP)gc D

3
is 3 in., p = 5 centipoises, p = 60 lb/ft .

Solution. Because no knowledge of the flow

(Dp / 2(-AP)gc D \
80
regime is available, the first step in the calculation
is to determine the Reynolds number.
The right-hand term can be evaluated with the data
at hand.
Dvp (3/12)(8)(60)
/v Re = = 35,700 v
z
Lp (3/12)160)
P ~(5x6.72x 10~ 4 -=2.og" 10
,u (-
2(-AP)gc D \(5) (0.000672)
This is in the turbulent range and therefore Figure
13.1, Equation 13.15, or Equation 13.16 may be /2(548)(32.2)(3/12)>
used. From Figure 13.1, at /V Re = 35,700 0.80
! 100) (60)

2(-AP)gc D ,72
v*Lp
f= 0.023 = 2 = 6.67
v Lp V 2[-AP)gc D
Therefore,
Then
2
fv Lp (0.023) (64) (100) (60) 2
-AP) = = 548 2
lb f /ft _, (6.67) (2)(548)(32.2)(3/12)
iQcD (2)(32.2)(3/12) v2 = = 64
(100)(60)
548
<-A/J = 777 = 3 81 P si
) -
and v = 8.0 ft/sec
144

A check on the assumption of turbulent flow

Example Calculate the velocity of an
13.2.
can be made by computing the Reynolds number,
oil flowing through a 3-in. tube. The pressure drop
2
through the tube is548 lb^/ft per 100 ft of tube. Dvp (3/1 2) (8) (60)
3 = = = 3.57 x 10 4
Oil properties are p = 5 centipoises, p = 60 lb/ft /V Re _
(5 x aio
/R 6.72 x in
10 4
.

p )

Solution. In this problem, velocity is un-

known. Since velocity appears in both A/ Re and f, A/ Re is greater than 2100, so the initial

neither of these groups can be evaluated, so that assumption is valid.

for turbulent flow a direct solution is not possible

with Figure 13.1 nor with Equations 13.15 and Shapes Other Than Cylindrical
13.16. In the viscous-flow regime a direct solution For duct shapes other than cylindrical, the above
is possible because Equation 13.10 can be re- equations may be used with appropriate modifica-
arranged to Equation 13.7 for direct solution. tion for the new geometry. The flow-duct geom-

etry in the friction factor and
is defined by the diameter of the tube. For shapes
other than circular, the diameter must be replaced
by an appropriately chosen variable or group of
variables that describes the system with a single
linear dimension that is equivalent in behavior to
D. If were chosen that was not equiva-
a variable
lent, curves A and B in Figure 13.1 would be
shifted, and the curves as shown would be useless.
The geometrical factor is introduced into the
flow equations in the force balance.
(-AP)S, ={ry Ah
In cylindrical geometry after
2
irD /4 and A :
= ttDL = bL
=
Mry hL
D ;13.18)
(-AP)
where b is the wetted perimeter of the duct = irD.
For any other shape a more general form of
Equation 13.18 must be developed. If more than
one surface is present, as in the annulus, the vari-
ous surfaces probably exert different stresses on
the fluid so that the force balance is
-AP)S l
=t A +t 2 A 2 +
1 1
If the stresses are not equal, Equation 13.19 does
not reduce to any simple function. If as an approx-
imation the mean stress is used to replace the
different stresses on the various surfaces, Equation
13.19 reduces to Equation 13.18. For the shapes
normally encountered, Equation 13.18 is a reason-
able approximation except in extremes, such as
annuli made up of small wires running through the
center of large tubes.
For shapes other than those circular in cross
section, assuming that the mean stress is a satisfac-
tory replacement for the actual wall stresses, a
replacement term D eq for the geometrical factor D
may be written
4Si
D eq = 13.20)
where D eq
= the equivalent diameter.
other words, the equivalent diameter is
In
equal to four times the cross-sectional area of the
duct divided by the wetted perimeter. By substitu-
tion of appropriate terms for cylindrical geometry,
it can be shown that for cylinders D
eq =D. The
friction factor measured for the special case of
round pipe cannot be expected to fit rigorously for
noncircular conduits of various configurations.
FUNDAMENTALS OF TRANSFER MECHANISMS 253
Reynolds number Example 13.3. Calculate the equivalent diam-
eter for a rectangular duct 3 ft high and 5 ft wide.
Solution. For a rectangular duct the stress at
every point on the wall is not constant. Since the
dimensions of the duct are of about the same
order, in the absence of more exact information
Equation 13.20 may be used as an approximation
for the equivalent diameter.
D™
eq = —
4Si
b
=
(4)(5x3)
,„ r,
2(3 + 5)
= 3.75 ft
substituting S-\ =
THE BOUNDARY LAYER
The term developed turbulence mentioned
fully
4S, earlier refers tothe condition that occurs after a
fluid has progressed through a duct far enough so
b
that no further change will take place in the
velocity pattern with further progress through the
duct. Fully developed turbulence is assumed in all
work previous to this topic. This section will
examine the behavior of fluids prior to the full
development of turbulence. Examples of this area
of study are entry sections of long tubes and large-
diameter short ducts. Outside the field of chemical
. (13.19] engineering, those involved in the design of ship
hulls and airfoils are more concerned with the
study of this flow behavior.
Consider a flat plate with an edge facing the
direction of flow. The flat plate can be visualized
as suspended in an infinite duct so that the volume
of the plate has no effect upon the fluid velocity in
the duct; that is the volume of the plate is negli-
gible with respect to the volume of the duct. The
velocity of the bulk of the fluid (free stream) is
designated as v fs At the instant a differential
.
element of moving fluid contacts the leading edge
of the flat plate, the velocity of that fluid element
immediately decreases to zero, which is consistent
with the concept of no slip at a boundary. Any
acceleration or deceleration is associated with a
force; therefore, the deceleration of the fluid
element generates a stress in the flow direction on
the surface. At the leading edge of the plate, an
element of fluid of differential volume is deceler-
ated instantaneously by a differential area of the
surface. Instantaneous deceleration requires an
infinite force. Mathematically, (r / ;) = = °°. The ,
v
next element of fluid (measured in the x-direction)
would still be moving at v = v fs Successive layers
.
from the plate will be retarded as / increases and
the stress distributes itself over a layer of some
thickness.
254 MOLECULAR AND TURBULENT TRANSPORT

here, but certain significant relationships will be

cited comparison with equations already
for
9
h known. For example, the ratio of momentum
Transition-^ yS"
-; \ Intense
transfer by turbulence to momentum transfer by
_• __±s
turbulence
Buffer
6 molecular mechanism may be used to write an

/ i
/
Rat
plate "\ iS>s-- £-y
X ''reg :'

Laminar
sublayer
appropriate Reynolds number for the boundary
layer. Intuitively, this may be written (A^ Re 6 = )
v.
fs ^
Laminar Turbulent (8v fs p/p). The boundary-layer thickness (6) is dif-
boundary layer boundary layer
ficult tomeasure; consequently, since 6 is a func-
y
tion of y, for convenience the Reynolds number
Figure 13.2 Boundary- layer buildup on a flat plate
may be written [N He Y = (yv fs p)/p. Both forms )

(vertical scale magnified).

[(/V Re ) 6 and (N Re )
y ] of the Reynolds number
have specific uses.
At a distance y along the surface, a finite A boundary-layer counterpart of the friction
velocity gradient will be set up so that at some factor in ducts can also be written. First note that
small x = 5 from the surface, the fluid
distance in an infinite duct, pressure drop — or more
velocity would approach (say, within 99 percent) exactly the momentum lost by the fluid — cannot
the free-stream velocity. The region in which the be measured. In effect, this measurement would
fluid velocity is less than 99 percent of the free-
result from subtraction of a finite number, the
stream velocity is defined as the boundary layer. momentum gained by the plate, from infinity, the
The boundary layer extends from the leading edge momentum content in an infinite duct. This differ-
(y = 0) to the end of the plate (y = y t and extends )
ence is indeterminate. The momentum gained by
from the surface (x = 0) to the boundary-layer the plate, represented by the stress on the plate, is
limit (x = S). At y = 0, 6 = and 5 increases to a
a finite and measurable quantity. Therefore, any
maximum at y = y t The boundary-layer geometry
.

counterpart of the friction factor for ducts can

is shown in Figure 13.2. The point stress (or stress
only be expressed in terms of surface stress. In a
at a specific position y) on the plate surface also
finite-sized duct through which a finite amount of
varies so that the stress is a maximum at the
fluid is flowing, the pressure-drop term is no longer
leading edge and diminishes as y increases. It may indeterminate. The momentum trans-
ratio of total
be noted that a plate as described can operate at ferred to the plate to the momentum transferred
steady state, in that at any position y there is no mechanism may be written as
by the turbulence
change in any property with time, but it can never ,
2
T y) \/VfS p, which is identical in form to the fric-
(
operate in fully developed flow, because the
tion factor in ducts as shown in Equation 13.9.
boundary layer continues to change regardless of A number of equations for use in boundary
how long the plate is made. layers will be written here (8). There are three equa-
It is evident that near the leading edge 6 is
tions for laminar flow and three equations for
small; consequently, the distance between v =
turbulent flow. The equations are each limited in
and v = v fs is small. The velocity gradient must be application. The greater number of equations is
high, and therefore the stress on the fluid (r y gc =
required in this application because 5 varies with y
—pdv/dx) is large. Regions of high fluid stress and (r y varies with y. The counterpart variables
) :

inhibit the formation of eddies, and consequently in fully developed flow in cylindrical ducts, D and
laminar flow can be expected. Moving in the y- (r y )i , are constant with L. The additional variables
direction, thickness (5) increases but v fs remains call for either a more complex equation or a
constant so that there is a subsequent decrease in greater number of simple equations.
gradient and fluid stress. Under these conditions of the laminar regime, the stress at any point
In
reduced stress, the bulk of the boundary layer can y may be determined in terms of the Reynolds
exist in the turbulent regime. In the turbulent number based upon boundary-layer thickness at
regime a laminar sublayer as well as a buffer region the same point y.
exist near the wall. The high stress near the wall
and the impenetrable nature of the fluid adjacent (TyUy 1.5
13.21)"
to the wall assure this. Vh P
2
(A/ Re )6
Excellent derivations of the boundary-layer
equations are offered in most fluid-mechanics text- * The nomenclature for Equations 13.21 through 13.26 is

books (53). No complete derivations will beoffered presented immediately after Equation 1 3.26.
 FUNDAMENTALS OF TRANSFER MECHANISMS 255

In the laminar regime, the boundary-layer thick- 5 = boundary layer thickness (L x )

ness (6) may be determined at a point y in terms y = position measured from the
of the Reynolds number based upon the same leading edge of a plate (y = 0)
point y. in the direction of flow (L>,)

Vfs = free-steam velocity (L y /6)

4.64 i yvfsP \
13.22) (N Rely 1= Reynolds number based upon
(N Rely 0.5
I

^ P '
the position y, dimensionless.

In the laminar regime, the mean stress on the (A/ Re/6 I I

= Reynolds number based upon
surface between/ = and y = y is related to the ^ the boundary-layer thickness
Reynolds number based upon y, (5), dimensionless
(Ty)-[
y
= stress on the plate at a point y
(Tyh 0.65 downstream from the leading
7 0.5
13.23) edge
Vfs P (N Rely
(r,
k'1
mean value of the stress between
As stated earlier, near the leading edge the y = and y = /
boundary layer is laminar and is therefore yhy
(r,
(Ty)i
= friction factors for the boundary
described by Equations 13.21 to 13.23. On a long 2
V fs 2 p
Vfs P layer
plate as y increases, (/V Re K increases. At the posi-
)

tion y corresponding to (N ne y = 10 5 to 10 6 tran-

) , Equations 13.23 and 13.26 both presume that
sition occurs and a turbulent boundary layer the flow regime is the same between / = and
forms. Thus, any turbulent boundary layer is y = y. In practice, the laminar boundary layer
preceded by a laminar boundary layer near the exists from the leading edge to some critical point
leading edge. Equations for the turbulent boundary 5
y c where (N^ e y is between 10 and 10
)
6
There- .

layer are written below. The turbulent-boundary- fore, to determine the mean stress on a plate in
layer equations presented here are written as if the turbulent flow. Equations 13.23 and 13.26 must
entire plate were in turbulent flow. be used over the ranges in which they apply. Con-
the turbulent regime, the stress at any
In sider a plate of total length y t and width z. Transi-
point y is related to the Reynolds number based tion from laminar to turbulent flow is considered
upon the boundary layer thickness (S) at the to occur at yc The transition point y c can be
.

point/. calculated from (N^ e y = 10 5 as shown in Equa- ) ,

tion 13.27.
(t ynv
v) 0.0228
2 0.25
13.24)
Vfs P (/V Re
e'6
)
(13.27)

In the turbulent regime the boundary layer

thickness (6) at any point y is related to the The mean stress for the laminar portion of the
Reynolds number based upon the same point y. boundary layer that lies between y = and y = yc
can be calculated from Equation 13.23.
0.376 2
0.2
13.25) 0.65vfS P
(/Vr ely (Tyh = !13.28)
(10 5)0.5

In the turbulent regime, the mean stress on

The force on one side of the plate between
the surface of the plate between y = and y = y \s
y = and y - y c is equal to the stress-area product.
related to the Reynolds number based upon the
same point y.
0.65v fs 2 pzyc
^lam = [Ty )^ycZ = 13.29)
2.5
(tJ, 0.037 0-Kc 10)
! 13.26)
2
Vfs P (N Rely 0.2 Equation 13.26 is written for a plate in turbu-
lent flow between y = and y = y t Therefore, the .

The nomenclature for Equations 13.21 force between y c and y t must be computed and
through 13.26 is: added to the force due to laminar flow between
256 MOLECULAR AND TURBULENT TRANSPORT

y = and y = yc The mean stress between / =

. the actual transition value lies between 10 5 and
and y = y t is calculated from Equation 1 3.26. 10 6 The only justification for this arbitrary choice
.

lies in the fact that at the low limit of transition

2 5

m
(0.037) v fs p (10 ) the calculated stress on a plate is safely high.
{Tyh = 2
1 13.30)
'

Example 13.4. A flat plate 1.5 m long has an

edge facing the flow direction. Air is flowing over
and the force on one side of a plate between y = the plate at 12 m/s. The air properties are p = 1.2
and / = Kf if the entire plate were in turbulent flow kg/m 3 and ju= 1.8 x 1CT 5 N-s/m 2 .

(a) Calculate the boundary layer thickness

is
1.5 m from the leading edge.
2
(b) Calculate the point stress 1.5 m from the
Fturb = (Ty ) } zy t =
(0.037) v f pzy t
— -7J-5 — (13.31) leading edge.

(^r (c) Calculate the average stress over the plate

from the leading edge to
leading edge.
a point 1.5 m from the

The mean the region between y =

stress for and
y = yc for turbulent flow can be calculated as Solution. A test for flow regime is performed.

_
^ 1=
(0.037) vfs p
2

= (0.0037)^ 2 p (13.32) (A/, Re k

Y
=
yvfsP \
=
~r^
—
(1.5) (12) (1.2)
,,» R = 1.20 x 1
6

x 10~ 5
)
5 )0 .2 \ p ! 1.8
(10

The force applied The plate in this problem has a turbulent-flow

to one side of a plate in
5
turbulent flow between y = and y =
portion beyond some value y c where A/ Re = 10 .

ye is
The transition point can be calculated from
Equation 13.27.
^turb = (ryWe = 0.0037 v fs 2 pyc z (13.33)
10 5 m (10 5 )1.8x 10- 5
y*
m = = 0.125 m
77777777
The net tangential force applied to one side of the vfs p (12)(1.2)
plate is

The boundary-layer thickness at y =

(a) 1.5 is

^net
= ^turb — ^turb + ^lam (13.34) calculated from Equation 13.25 for turbulent flow.
0^y t 0-Kc O^Yc
0.376/ (0.376X1.5)
The net mean stress (r K ) net equal to the net
6 = = = 3 45xl0m
^i^
is
'
2
force per unit area of plate n ^oo,ooo)0

'turb
— ' turb + >\am (b) The point stress in the turbulent regime
7— ^net *^Vt 0-/ c 0- Kc can be calculated using Equation 13.24.
(rK )net= = ~ (13.35)
yt z Yt?
2
8v fsP \ J3.45x10- )(12)(1.2)
Equations 13.29, 13.31, and 13.33 are substituted (/V Re ) 6 = (
" \ p / (1.8x 10- 5 )
into Equation 13.35, and the equation is re-
= 2.76x 10 4
arranged.
2 2
0.037y f z 0.65/ c z\ ^ s p(0.0228) (12) (1.2X0.0228)
(TyUy
(rj net \ (A/ Rely
0.2 (A/ Re 5
25
)
-

(2.76 x10 4 025 )

2 2
Vfs P yt z
= 0.306 N/m
(13.36)
(7-
0.037 (c) The net mean stress over the plate can be
K )net
2 2
0.0017 (13.37) calculated using Equation 13.37.
Vfs P (A/Re)/ Yt

In the foregoing derivation, the transition 0.037

(7>)net
= v fs 2 pV 0.0017—')
point was arbitrarily taken as A/ Re = 10 5 although 2
(A/ Re L, -
Yt'
 FUNDAMENTALS OF TRANSFER MECHANISMS 257

0.037 entire entry length, flow in the "fully developed

= (12) 2 (1.2)( flow" section of the tube is laminar. If turbulence
(12.0 x 10
s
)
02
develops in the boundary layer within the entry
0.125 \
length, the remainder of the tube will operate in
-0.0017- turbulent flow.
1.5 /
As might be expected from consideration of
= 0.362 N/m 2 the boundary layer, the wall stress per unit length
of entry is greater than that for fully developed
For this particular example, the laminar flow, with a maximum at the point of entry, and
portion of the boundary layer corresponds to the the value diminishes with progress through the
first 8 percent of the plate. If the laminar correc-
tube to a minimum at the end of the entry length.
tion is omitted and the entire plate is considered in
The minimum value at the end of the entry length
the turbulent regime, the mean stress is 0.399
is equal to the value in fully developed flow. In a
N/m 2 which is about 10 percent higher than with-
,
bounded duct of finite dimensions, as is described
out the correction.
here, a pressure drop can also be expected. In the
infinite duct described in the previous section, the
The Entry Length pressure drop was unmeasureable. In the bounded
The entry length of a tube is an example of the duct, the usual force balance may be written
buildup of a boundary layer. Consider a pipe sus- relating the tangential force at the wall and the
pended in a region of flow of uniform velocity (see force due to pressure on the fluid.
Figure 13.3). At the immediate entrance to the Equations for the entry length can be derived
pipe, a boundary layer is set up at the inside tube from boundary-layer considerations, but the
surface. Some length of tube is required as the acceleration of the free-stream fluid within the
boundary layer grows in thickness until all finite duct must be accounted for. The derivations
boundary layers meet at the center of the pipe. are beyond the scope of this book, but several
This length is known as the entry length (L e ). equations are given.
Adjustments the free-stream velocity in the
in In the laminar regime, the entry length can be

entry length must be made because the free-stream predicted from the following equation (27):
velocity at the entry must equal the average
velocity of the fluid in the tube. In this sense, Le
= 0.0575(/V R( 13.38)
boundary-layer flow within a finite duct differs D
from the boundary layer in an infinite duct
where
described above. Thus, any flow within a tube is
essentially all boundary-layer flow. In the entry L e = entry length (L K )

length, a boundary layer and free stream exist D = tube diameter (L r )

within the tube. The entry length ends where the (/V Re = Reynolds number based upon the
)

boundary layers meet at the center after which a usual mean velocity and tube
constant-velocity profile is maintained regardless of diameter; dimensionless
position in the /-direction. After the boundary
layers meet at the center of the duct, the flow is The entry length for a turbulently flowing
"fully developed." All equations for tubing fluid can be predicted (28) from the equation
mentioned earlier apply to fully developed flow. If
the boundary layer is laminar throughout the 25
= 0.693(/V Re )
-
13.39)
D

Qualitatively, the pressure drop through the

I .
entry length is considerably greater than the
—--^^ °
1 I

6 pressure drop after fully developed flow is estab-

— " i
1
lished. The increase is the result of two effects.
D
S^~ 6
1 ,
Boundary-layer equations reveal that the local wall
4 ^ greater than the local
stress near the leading edge is

°fs wall stress downstream from the leading edge. In

Figure 13.3. Boundary-layer buildup in the addition, the fluid in the boundary layer is moving
entry length. more slowly than the fluid in the free stream. If, as
258 MOLECULAR AND TURBULENT TRANSPORT

in the entry to a tube, boundary-layer growth

occurs, then to maintain a constant mass-flow rate
in a bounded duct, the free-stream fluid portion in

the tube must accelerate. The twofold force effects

of acceleration of the free stream and high stress
near a leading edge increase the pressure drop over
the entry length. As a general rule, the pressure
drop through the entry length may be taken
approximately as twice to three times the value for
(a)
fully developed flow at the same Reynolds
number.

Example 13.5. Calculate the entry length for

water at 70°F flowing through a 5-in. tube at 0.1
ft/s.

Solution. Since no flow regime is specified,

the Reynolds number should be evaluated.
^Boundary layer
_Dvp_ (0.5/12)(0.1)(62.4) _ separates from

Nrb= = =38? <b) solid surface

m (1H0.000672) Figure 13.4. Flow around submerged bodies, (a) Stream-

This is laminar flow; therefore, Equation 13.38 lined shape-no separation, (b) Nonstreamlined shape-
may be used. separation of the boundary layer.

— = 0.0575(/VRe )
= 0.0575(387) = 22.3
close to the body. The fluid approaching the center
of the front face of a symmetrical body will
and
impinge upon the body and be split into two
portions, one half of which moves to each side of
Le = \— (22.3) =0.927 ft
the body. At the exact center of the body, the
fluid will have zero velocity, known as a stagnation
point. Boundary-layer growth begins at the stag-
nation point and continues over the entire surface.
FORM DRAG
Beyond the trailing edge, the boundary layers re-
In the examples of fluid friction that have been vert into the free stream. The tangential stress on
considered, the transfer of momentum resulted in a the body arising from transfer of momentum
tangential stress or drag on a smooth surface that originating in the slowing down of the boundary
was oriented to the flow direction. This
parallel layer is the skin friction. The boundary layer will
phenomenon is traditionally called skin friction or have the same characteristics as described earlier; it
skin drag. From the foregoing, it is evident that if will be laminar unless the Reynolds number ex-
any surface is in contact with a fluid and a relative ceeds a critical value, after which the boundary
motion exists between the fluid and the surface, layer becomes turbulent.
skin friction will exist between the surface and the The fluid outside the boundary layer is sub-
fluid. In addition to skin friction, significant ject to acceleration due in part to change in path
frictional occur because of acceleration and
losses and in part to change in linear speed. As the fluid is
deceleration of the fluid. The accelerative effects diverted in path to pass around the body, a force is
occur when the fluid changes path to pass around a exerted upon the body by the fluid. Note that this
solid body set in the flow path. This phenomenon force on the body is in addition to the skin friction
is called form drag. associated with the boundary layer. This is not a
Consider the body shown in Figure 13.4a. The tangential force, but it is directed according to the
body is suspended in an infinite duct and is subject geometries of the body and streamlines. In Figure
to a free-stream velocity v fs Streamlines are drawn
. 13.4, the space between any two streamlines rep-
to represent the path of fluid elements around the resents the duct cross section occupied by an equal
body. The boundary layer is shown as a dotted line volumetric flow rate. The position of the stream-
 FUNDAMENTALS OF TRANSFER MECHANISMS 259

lines around the widest part of the body shows
that the fluid is moving at a velocity greater than
v fSl so that the linear speed of the fluid has in-
creased in that region. Beyond the widest
the body, the fluid is subject to
acceleration opposite to that at the leading face of
the body and to a deceleration of linear speed as
the fluid returns to the normal free-stream pattern
downstream of the body. The summation of all
forces on the body due to acceleration and deceler-
ation constitutes the form drag of the body.
In Figure 13.4a, a shape is chosen such that
no sharp discontinuities exist along the length. The
boundary layer is continuous over the length of the
body but obviously must adopt a shape and char-
Form drag appears in many chemical engineering
applications. Metering devices are designed to max-
imize or minimize form drag, according to require-
part of ments. Roughness in pipe is an example of form
a directional drag. The finite constant rate at which particles
settle in fluids is another. Other engineering appli-
cations are the stress on structures in wind and
flowing water.
Correlations of flow characteristics and geom-
etry for bodies in a free stream utilize dimension-
less groups that are analogous in concept to the
friction factor and Reynolds number described in
the sections of this chapter covering flow through
pipe and momentum transfer through the bound-
ary layer. The drag coefficient (CD is defined as
)

acter subject to the changes in velocity and direc-

tion of the fluid beyond its bounds. Another shape
c ~?5» (13.40)
is shown Figure 13.4/?, with a sharp discontin-
in
Svf/p
uity in the trailing portion. In the new shape, the
boundary layer is shown, along with the stagnation and the Reynolds number is

point. The acceleration of fluid in the vicinity of LoVfsP

the leading face is the same as in the body de- A/ Re = (13.41)
scribed in Figure 13.4a. The upper and lower
boundary layers begin at the stagnation point and where
follow along the surface of the body. In the trailing
FD = total force exerted on the body (F)
portion of the body, beyond the widest point, the
S = maximum projected area normal to
behavior of the boundary layer is significantly
flow(Lx L z )

different. If the fluid streamlines were to conform

Lq = length characteristic of the geometry
exactly to the surface of the body, deceleration
must necessarily be very rapid. The boundary layer
(M
Vf S = free-stream velocity (L K /0)
must conform to the surface and the free stream.
In this case, the free stream is undergoing rapid Correlations are usually presented graphically
deceleration, and therefore the boundary layer on logarithmic plots of Co as a function of /V Re .

must decelerate also. But the boundary layer is Several such drag diagrams are presented in Figure
moving very slowly before deceleration; therefore, 13.5.
deceleration of the boundary layer brings about The diagrams are similar to the f— A/ R e plot
reversal of direction of the boundary layer and used for skin friction in pipe, but several significant
consequently separation of the boundary layer differences should be noted. Below A/ Re =0.1 all
from the surface. As can be seen in Figure 13.45, shapes are represented by a single line that may be
the boundary-layer reversal relieves the necessity of written in equation form as
the extreme deceleration of the streamline by set-
ting up a region of eddy activity after the body. 24
Cn = (13.42)
This intense eddy activity results in a sizable force A/Re
exerted on the body, much more than in the body
described in Figure 13.4a in which the separation This is the region in which the boundary layer is
eddy is absent. laminar and accelerative effects are small enough to
From the
it is evident that the
foregoing, be negligible. Equation 13.42 is analogous to Equa-
tion 13.10. In the range 0.1 </V Re < 10 several
5
geometry of determining factor in the
a system is a
amount of force exerted on the body. The "tear- different mechanisms contribute to the transfer of
drop" shape shown in Figure 13.4a exerts less drag momentum. The accelerative forces become signifi-
than the same body with the teardrop tail re- cant enough to affect the shape of the curve, and
moved. On the other hand, exaggeration of the in addition the boundary layer becomes turbulent.

trailing portion can lead to a large amount of In most instances the accelerative forces begin to
surface with consequent increase in skin friction. operate at Reynolds numbers well below the onset
260 MOLECULAR AND TURBULENT TRANSPORT

10-
- 1 III 1 1
1

\ Disk -
10'
- \
y— -

- Cylinder -

lO
1
^
- -

24 y^^^ —
- V^^^— .Disk

10°
Cylinder
=
Sphere*" \

10"
-
\_J=
— —

-
10 1 1 1 1 1 1 1 1 1 1 1

ic- ier 10
1
W 10
3
10
4
10
5 10"

Figure 13.5. Form drag for various shapes (51, 52). (By permission of John Wiley
& Sons, Copyright © 1953.)

of significant turbulence in the boundary layers. Table 13.2 VALUES OF CD THAT MAY BE USED IN
For example, the discontinuity in the drag diagram EQUATION 13.43
for spheres at /V R e s 10 5 is the result of change in
the boundary layer from partially laminar and par- Lower
tially turbulent to completely turbulent. Above Limit
Shape
A/Re = 10
5
the boundary layer is considered tur- LD of/VRe Co
bulent for all shapes and the accelerative effects
Circular disk, flat side
predominate. It is in this region that the drag
perpendicular to flow Diameter, D 10 3 1.12
diagram can be represented by the equation Cylinder, infinite height.
axis perpendicular
CD = constant (13.43)
to flow Diameter, D 5x 10 5 0.33
In this respect the drag diagram differs from the Sphere Diameter, D 3x 10 5 0.20
/-/V Re correlation. At high values of the Reynolds
number, the /— A/ Re diagram still has a slight neg-
ative slope characteristic of skin friction. This same Drag diagrams like those of Figure 13.5 can
characteristic can be seen in the turbulent-bound- be developed for use with shapes different from
ary-layer equations, namely Equation 13.26. there- the spheres or cylinders specified in Figure 13.5.
fore, it might be concluded that, where accelerative The correlation of data on the drag performance of
effects predominate, the drag diagram will ap- irregular shapes has received particular attention
proach Equation 13.43 very rapidly. This leveling- because solids involved in chemical processes are so
off will be seen in the f-/V Re plots for pipe with often irregular in size and shape. These correlations
rough interior surfaces (Appendix C-3). The limit- have required a more complicated specification of
ing constant values of C D are listed in Table 13.2 size and shape than is required by the shapes listed
for various shapes, along with the dimension used in Table 13.2. In addition to a size parameter, such
for L D and the lower limit of applicability of as particle "diameter," a shape parameter is need-
Equation 13.43. For the transition range, the dia- ed. Several such parameters have been developed,
gram must be used. the most common of which is the sphericity. The
 FUNDAMENTALS OF TRANSFER MECHANISMS 261

sphericity {^) is defined as the solutions are tedious and cumbersome. There-
fore, it is desirable to approach the evaluation of
surface area of a sphere of the transfer in terms of mean fluid properties and gross

<//=
same volume as the particle
surface area of the particle
— TT— (13.44)
dimensions rather than point conditions as used in
Chapter 12. The mean properties and gross dimen-
sions have an additional desirable feature; the ex-
Problems in the characterization of irregular par-
perimental data may be determined accurately and
ticles are treated in Appendix B.
reproducibly with the simplest of analytical tools.
Example 13.6. A chimney 100 ft high and 5 ft Useful equations that have the properties men-
in diameter is subject to a maximum wind of tioned result from the integration of the point-
100 miles/hr. Calculate the force exerted on the condition equations introduced in Chapter 12.
chimney by the wind. The friction factor defined earlier in this
3
chapter is, by its definition, an integrated form of
(M = 0.018 centipoise,p = 0.075 lb/ft )
the expressions for point conditions for turbulent
Solution. From Table flow. It would seem reasonable that similar con-
13.1, the characteristic
dimension cepts and equation forms would have been written
for a cylinder is the diameter. The
regime of flow is determined by the Reynolds for heat and mass transfer. Unfortunately, this is

number. not the case. Early in the investigations of

momentum-, heat-, and mass-transfer behavior, the
LpVfsP _ (5) [(100 x 88)7(60)] (0.075) similarities among the three transfer
that exist
/v Re = phenomena were not recognized. The various equa-
P (0.018 x 6.72 x10~ 4 ) tion forms that were developed independently are
= 4,530,000 not exactly similar, but the content of the equa-
tions can be related.
This value is above the lower limit of 5 x 10 5 given The general transport equation that must be
in Table 13.1, so that c D = constant = 0.33. Then integrated is
by Equation 13.50
dT
^=-(5 + £ 12.29)
—2~=0.33^
2Fd9c
SVfs P
r,

where
dx

2 = flux of a property at
0.33Sv fs p \p any value of x
Fn = 5 = molecular diffusivity
29c
E = eddy diffusivity
The maximum projected area normal to flow is T = volume concentration of transferent
the area of a rectangle 5 ft by 100 ft; therefore property
S=500ft 2 .

In cylindrical geometry, with appropriate modifi-

2
(0.33) (500) (100 x 88/60) (0.075) cation of the gradient, Equation 12.29 may be
Fn =
written
(2X32.2)
F D =4130 lb. dT
i// = -(5 + £: : 12.30)
dr
TRANSFER COEFFICIENTS FOR MASS AND HEAT
In Equation 12.33 the flux is shown to be linear
TRANSFER IN THE TURBULENT REGIME
with radius, and therefore Equation 12.30 may be
In Chapter 12, heat and mass transfer in the tur- written
bulent regime were introduced. Calculations were
based upon concentrations of transferent property
at various radial positions. The turbulent-flow *0-^*«? (12.34)

properties at the various positions were defined in

terms of eddy diffusivity, mixing lengths, and so
where
forth. Although the point-condition behavior is ^! = the flux at the boundary
useful understanding the mechanism of turbu-
in r-\ = boundary radius

lence and its effect upon heat and mass transfer, r = any radial position
262 MOLECULAR AND TURBULENT TRANSPORT

Equation 12.34 may also be written as Equa- position.* Then Equation 13.45 may be written
tions 12.36 and 12.35 for heat and mass transfer,
rdr n
respectively. At steady state Equation 12.34 may = -r,(r 1 -r )
= (r, -n
be integrated, •0 (5 + E)
(13.48)
rdr
M JL" = -r, '
err (13.45)
J
o (8 + E) The integration of the left-hand term of Equation
13.48 not be performed,** but the mean eddy
will

where T and r = are the conditions at the center diffusivity (E) that results from the indicated

of the cylinder and Ti and ry are the conditions at integration will be defined.
the wall. The right-hand term of Equation 13.45
may be integrated directly to —f\ (Y — To). This y
rdr 1
rv
rdr =
form is written in terms of r 0/ the concentration r (5 + £) (5 + £)-o 2(8 +
=r (13.49)
E
of transferent property at the center of the tube,
which is difficult to determine by analytical means.
Equation 13.49 can be substituted into Equation
Since the analysis is easier to perform, it is more
13.48.
desirable to use the mean value of transferent prop-
erty that results from integrating over the entire 01^17 -
duct. The appropriate mean value may be defined
2(6 + E)
-=-=-(r 1 -n (13.50)

as
The tube radius r^ = D/2. This value can be sub-
- v
T =— stituted into Equation 13.50, and, after rearrange-
>
1 r
YdV 13.46)
V, -o ment, the equation becomes

where V is the volumetric flow rate at a point 4(5 + E)

within the duct and Vy is the total volumetric flow 01
=
r—(ri -D (13.51)
rate in the duct.
The mean value described is the result of Equation 13.51 can be written as a rate equation
integrating over the fluid flowing in the duct.
all by multiplying by Ay ,

Note that dV= v dS, where v is the point velocity

and S is the applicable area of flow, and for
cylindrical geometry, dV = v2irr dr.
(^A) :
=
4(5 + E)

7O
— (Y,-Y)Ai (13.52)
Physically, the mean value can be determined
if the entire flow stream is collected and averaged. The transfer coefficient U ) may be defined as
An appropriate mean temperature will result if the
entire flow stream is passed through a mixing 4(5 + E)
13.53)
chamber so that every element of fluid is at the
70
same temperature; the resulting temperature will
be the mean value in question. The mean velocity Equation 13.52 becomes
willbe determined if the entire flow stream passes
11/AU =-^(1"! -Y)A,
through a volume measuring meter and the volu-
metric flow rate considered to pass through the or
duct atuniform velocity across the duct.
The ratio 7 may be defined as =
(r, -D 13.54)
(0/»)i
1

ri-r WA,)
:
13.47)
ri-r * Figure 20.2 is a special case of this ratio for momentum
transfer. In the figure yT = ^/^rnax- ' s tne ratio °^ m ean velocity to

The point velocity at the center of the tube. For a stationary tube v =
ratio 7 is the ratio of the difference
concen- in -\

tration of transferent property between the wall
and the mean value of the fluid to the maximum
difference between the wall and the center. The
ratio 7 is a function of the distribution of Y with
at the wall.
** For momentum transfer the integration may be per-
formed by writing E T as a function of r and dvldr as in Equation
12.22. The derivative dvldr can be written as a function of rby use
of Equations 12.23. 12.24. and 12.25.
 I I NDAMENTALS OF TRANSFER MECHANISMS 263

Equation 13.54 is an integrated form of the Heat-Transfer Coefficients in Turbulent Flow in Tubes
differential equation for transfer (Equation 12.30). Equation 13.54 is a general equation for transfer in
The variables that make up the equation can be integrated form and may be written for heat trans-
evaluated conveniently, and therefore the equation fer by substitution of the appropriate variables.
is convenient to use. The equation form warrants The heat-transfer coefficient (h) may be defined as
some attention. The integration of the gradient in
the differential equation results in the term h 4(a+F fl

(V } —
P), and the rate of transfer is proportional (13.57)
to this term, which is called the driving force (or
pc P DiQ
less frequently the transfer potential). The denomi-
nator term (1/V/4) is called the resistance to trans- The other variables for substitution are: 6 =

fer. Equation 13.54 may be written a, r=pc p T, yA=q, and E = Eq , so that the
heat-transfer equation for turbulent flow is

driving force
Rate of transfer = 13.55) -
resistance (pc P T)y ipcpT)

Qi
= pCp (13.58)
Note the exact analogy between transfer of heat, |
\

mass, and momentum as described in Equation \hA,l

13.55 and the transfer of electrical current as
or, if cp and p are not temperature-dependent,
described by Ohm's law

(7-1
Ee electrical potential (driving force)
13.59)
q^
Re resistance

(13.56)
where where
le = current, or rate of transfer of h = heat-transfer coefficient, with dimen-
electrons sions \ML y L z QT, and typical units
2
Ee = electrical potential Btu/ft hr°F
Re = resistance
The driving force for heat transfer is the difference
Because this analogous behavior exists,
in temperature between the wall and the bulk
all the
principles used for electrical transfer apply to the fluid, (7*! — 7"), and the resistance to transfer is

transfer of mass, heat, and momentum. This fact WlhA).

is the basis of analog-computing equipment. Equation 13.59 is the characteristic heat
Electrical analogs of transfer phenomena are built,
transfer equation for the convective mechanism.

and from electrical-transfer behavior the transfer Note the similarity between it and Equation 10.35
systems can be analyzed. Equation 13.55 is for simple conduction. The only difference is that
extremely useful in chemical engineering calcula- for conduction, the distance associated with the
tion, as will be shown in Chapter 14. resistance is known, whereas for the turbulent con-

The definition of the transfer coefficient

vective process, the resistance dimension is masked
given in Equation 13.53 shows it to be a complex
in the heat-transfer coefficient.
function of transfer by the molecular and turbu- By mechanism-ratio analysis, it is possible to
lent mechanism, of the distribution of concentra- show the physical meaning of the heat-transfer
tion of the transferent property as indicated by y, coefficient. For a flow system in which energy in
and of the geometry of the system. It is also a the form of heat is being transferred, four
function of the flow characteristics_of the system, mechanisms contribute to the process. Their over-
all contribution depends upon the relative impor-
because the mean eddy diffusivity (E) is dependent
tance of each mechanism. They are:
upon flow behavior. In spite of the complexity, it
is reasonable to believe that the transfer coefficient 1. Heat transfer by molecular transport.
may be related to pertinent variables by use of the 2. Heat transfer by turbulent transport.
mechanism-ratio-analysis procedure described in 3. Momentum transfer by molecular trans-
the earlier part of this chapter. This procedure will port.
be described for the specific transfer coefficients 4. Momentum transfer by turbulent trans-
for heat and mass. port.
264 MOLECULAR AND TURBULENT TRANSPORT

Consider first the ratio for the process under consideration. Why should
there be a dependence upon whether the fluid is

Total heat transfer by molecular and turbulent being heated or cooled?

transport In Chapter 12, the temperature profile of the
heat transfer by molecular transport fluid was examined, and the greater part of the
gradient appears in the laminar sublayer and buffer
The total heat transfer may be represented by region. The majority of the resistance to heat
+ Eq by transfer occurs in this small resistance layer of
(a )/yq and the molecular heat transfer
,
a.

Thus, the ratio of the two is fluid. Since the wall is at T y

and the inner edge of
the resistance layer of fluid is approximately at T,

(ct+Eq ) [a + Eq )Cp p the transferring medium in the resistance layer

(13.60) operates at some temperature other than T. Since
lq<X Iqk the properties of the fluid in Equation 13.64 are
evaluated at T, the discrepancy between the actual
If Equation 13.57 is rearranged and divided
temperature of the actual resistance layer and the
byAr, it isidentical to Equation 13.60.
temperature used for evaluation serves as an
explanation for the difference in the equations for
(a+Eq )Cp p hD heating and cooling. The presence of the tempera-
13.61!
1q k 4k ture gradient in the resistance layer sets up density
gradients due the density change with temperature,
Accordingly, if the numerical constant is viewed as so that below A/ Re = 10,000 some natural convec-
part of subsequent correlation, (hD/k) is the ratio tion currents alter the transfer mechanism. This
of the total heat transferred by both molecular and subject will be discussed in detail later in the
turbulent mechanisms to heat transfer by the mol- chapter. Furthermore, the thickness of the
ecular mechanism. The Nusselt number is defined resistance layer depends upon viscosity and density
+
by this ratio. (from the definition of y in Chapter 12), and the
viscosity of liquids is extremely sensitive to
hD temperature change. A new equation was proposed
/v Nu = (13.62)
(59) that included a term to correct for the actual
condition in the resistance layer.
The ratio of the two momentum-transfer
mechanisms has already been defined as the 0.14
Reynolds number. Furthermore, in Chapter the = 0.023(Ay Re )°- 8 1/3
9, /V Nu (/V Pr 13.65)
)
(-)
Prandtl number was defined as

Fluid properties are evaluated at T except for

/V Pr =
/i-i , which is evaluated at the wall temperature T%.
k p ex.

Equation 13.65 is valid above A/ Re = 10,000 for all

momentum transfer by molecular transport fluids with Prandtl numbers the range of 0.5 and
in

heat transport by molecular transport 100. The equation does not apply for liquid
metals.
(13.63)

Thus, there are three ratio-mechanisms, all Example 13.7. An SAE 10 oil flows through
dimensionless, that can be used to describe heat a heat exchanger tube 2 in. I. D. at 400 gal/min. At
transfer in a turbulent flow field. the point of examination the mean oil temperature
Early correlations (37) for convective heat
(71 is 120°F. Calculate the heat-transfer coefficient
transfer for flow in tubes were of the form and the heat flux if the tube wall temperature is (a)
8 n
50°Fand(b) 170°F.
/V Nu = 0.023(/V Re )
-

(/Vp r ) (13.64)
Data:
n = 0.4 when the fluid is heated
T. °F jU, centipoises Data in this column at 120 F.
n = 0.3 when the fluid is cooled

In Equation 13.64 fluid properties are evaluated at 50 100 p= 56.2 lb/ft 3

120 20 c P = 0.465 Btu/lb °F
the mean temperature of the fluid (T). It is impor- 2 (°F/ft)
170 6 k = 0.070 Btu/hr ft
tant to note that a unique equation does not result
 FUNDAMENTALS OF TRANSFER MECHANISMS 265

Solution. The Reynolds number is calculated The negative sign indicates heat transfer inward in
to determine the regime of flow, but first the flow the radial direction (heating).
rate may be converted from gal/min to ft/s.

400/(7.42 x 60) Example 13.7 shows the difference in transfer

¥ = . „... = 41 .4 ft/s coefficient if material that is at a given condition is
2
[tt(2/12) /4]
subject to heating or to cooling. The viscosity of
Dvp _ (2/12)(41.4)(56.2) oils is particularly sensitive to temperature change;
= 28,600 Water under the same conditions would respond in
H ~(20x6.72x 10- 4 )

a similar manner but to a lesser degree. The vis-

At = 28,600, the flow is turbulent.
/V R e with increase
cosities of gases increase in tempera-
(a) If the oil is cooling, Equation 13.64 may ture; therefore, gases would respond in opposite
be used, but, because of greater precision over a fashion.
wide range of viscosity variations, Equation 13.65
is preferred. It is not difficult to apply because the Mass Transfer Coefficients in Turbulent Flow in Tubes
wall temperature 7"i is known. Equation 13.54, the general equation for transfer in
tubes, may be written specifically for mass transfer
From Equation 13.65
33
by substitution of the appropriate variables. For
14
* no/W\°- /Lt\ -

mass transfer 5 = 9 ,£ = E N (\^A) = Na and T = ca

A - 0.023 -<«„..•••(-*-) (JL) , .

The mass-transfer coefficient for equimolar counter

h = 0.023 x ——
0-070
(28,600)
08
diffusion may be defined

4(y + EN )

f(0.465)(20x2.42)-|
1/ 3r 20 I - 14 *c'= AN = (13.66)
Din
L (0.070) M00 J
where k c '
is the mass-transfer coefficient
2
h= 193Btu/hr°Fft
[M/eLrUfM/UL^MUbmoles/hrft 2 (lb mole/ft 3 ).
The heat flux can be determined from Equation
13.59 After substitution of appropriate specific variables,
the mass transfer equation is written
q
i-
= -h(T,-Tl
= -k c '{ca - i c (13.67)
-= - 2 A,
-(50 1 20) ( 1 93) = 1 3,530 Btu/hr ft
A
or upon rearrangement
The positive sign indicates heat transfer is in the
~(c3 - ca
: )

outward radial direction the (cooling). In /V Re /V3 1

= (13.68)
velocity was calculated in feet per second; conse-
'
quently, u was converted to pounds per foot A,k c
second. In /V Pr the thermal conductivity k is
expressed in hour units, and consequently u is The driving force for mass transfer is the difference
converted to pounds per foot hour. in concentration between the boundary and the
(b) Equation 13.65 applies for heating_as well bulk of the fluid, and the resistance to mass
as cooling. All properties are evaluated at T except transfer is [MA k c '). y

H-\ which is evaluated at Ty For the evaluation of mass-transfer

,0.14 coefficients in gases and liquids in the turbulent
h = 0.023 -(/V R e,>°- 8 (/v Pr -
regime, a correlation* similar to that for heat
)
(-)
transfer been developed. The data for this
has
(0.070)
014 correlation were taken on a wetted-wall column,
/20\
h = (0.023) (3640) (6.88) (-) consisting of a vertical tube arranged so that liquid
-gj^
flows down the wall in a thin layer while a gas, in
2
h = 286 Btu/hr °F ft
* The mechanism ratio analysis may be written based upon
%= -(190 - 120M286) = -20,000 Btu/hr ft
2 the four mechanisms for mass and momentum transfer by molecular
and turbulent transport.
266 MOLECULAR AND TURBULENT TRANSPORT

contact with the thin layer of liquid, flows upward In the integrated expression for molecular-
through the core. If the liquid layer moves in mass transfer through a stationary film (Equation
laminar flow, the area of transfer is known. The 10.62), the ratio c t /c b/m was inserted as a correc-
wetted-wall column is useful for examination of tion to the equimolar transport equation. This cor-
mass-transfer data, but it has little commercial rection for conversion of the coefficient** to the
value when compared with the higher transfer rates case of diffusion of component a through station-
per unit of volume in packed columns; however, ary component b is

the transfer area of the packed column is not

k'D kcD
known. Liquid mass-transfer rates were determined
experimentally by permitting a liquid to flow in a ' c blm
pipe made of a soluble material. The equation for Accordingly, the equation for this case becomes
the correlation (61) of all liquid and gas data takes
,Dvp\ 83 033
the form*
kc D
i
cD / r
= 0.023 — -

{» \
13.70)
k'D
= 0.023(/V Re )°- 83 (/V Sc )
- 33
(13.69) where k c is the mass-transfer coefficient for gas a
diffusing through a stationary gas b.

where the Schmidt number (/V Sc = vl Q is the ratio Equations 13.69 and 13.70 are reliable over
)

the range between A/ R e = 2,000 and /V R e = 35,000

of momentum to mass transfer, both by the mol-
and between /V Sc = 0.6 and N Sc = 1,000. The gas
ecular mechanism. Mass-transfer data are not as
velocity is measured relative to the container and
common as heat-transfer data; consequently, the
not to the moving liquid.
exact conditions of applicability of this equation
Concentration data for gases and liquids may
cannot be stated with as much confidence as for
be reported in dimensions other than moles per
heat transfer. Suffice it to say that this equation
unit volume. These other expressions of concentra-
applies when the fluid descends the column in
tion lead to the writing of several different mass-
laminar flow and the gas ascends the column in
transfer coefficients that are commonly used in the
well-developed turbulent flow.
The mechanical features of the wetted-wall literature. For example, for gas concentrations
column are such that the column for mass transfer expressed as partial pressures, the form k c is
'

and the tube for heat transfer are very nearly exact defined by the equation
analogs. In addition, the equations resulting from N
correlation of heat- and mass- transfer data are very — a
= -k G '(Pay -Pa) (13.71;
nearly identical in form, differing only slightly in
the exponent on the Reynolds number. The greater where kc ' = the mass-transfer coefficient for equi-
part of the concentration gradient exists in the molar counterdiffusion, for use with
resistance layer— that is, the fluid in the approx- gas concentrations expressed as
imate region of the laminar sublayer and buffer partial pressure, with dimensions
region. Because a concentration gradient for two M/0LK L z (F/L 2 ) and typical units
2
gases of differing molecular weight results in a (lb moles/hr atm). ft

density gradient, some degree of natural convec- pa -\

= partial pressure of component a at
tion can occur in the resistance layer. Any natural the boundary F/L 2 (atm)
convection in this layer will result in increase in the pa = the mean value of partial pressure of
surface coefficient, and the probability exists that component a in the flowing stream
the small differences in the correlating equations F/L 2 (atm)
can be attributed in part to this convection phe- ** The mass-transfer coefficient may be written according to
nomenon. Equation 13.66 and then multiplied by c t lc b im -

* Gas data alone are correlated 4(y+f-/u) ct

(61 ) with
07/V cblm
k D
-~ = 0.023(N
Be
)°- 83 (N
Sc
)0^ (13.69a) Note that this expression implies turbulent transfer of gas a through
a stationary gas b. The assumption of existence of stationary gas b
For these gas data N$ c varied between 0.6 and 2.5. For liquids N^. in the turbulent portion of the tube is obviously incorrect, but since
approximates 1000. If the exponent 0.44 or 0.33 is used for none of the gas b passes through the boundary, traditional correla-
common gases, the numerical difference in kc is very small and may
'
on this basis, and until
tions of experimental data have been written
be considered negligible compared to the experimental error. Such more directly useful information is forthcoming, the practice will be
differences among researchers are not unusual. continued here.
 FUNDAMENTALS OF TRANSFER MECHANISMS 267

The coefficient kc' can be related to kc '

as Several other forms of mass-transfer coefficient are
follows. The rate of transfer of component a can shown in Table 13.3. These forms can be related to
also be represented by Equation 13.68. k c and k c '.

N.
= -k c '(c a} -ca ) Example A wetted-wall column 2 in.
13.8.
I.D. contains and C0 2 flowing at 3 ft/s. At one
air
For perfect gases ca = pa /RT; therefore, point in the column, the C0 2 concentration in the
air mole fraction. At the same point in the
0.1
Na _-k c '
is

~ (Pa1 - Pa) 13.72) column, the concentration of C0 2 in the water at

A y
RT the air— water interface is 0.005 mole fraction. The
By examination of Equations 13.71 and 13.72, column operates at 10 atm and 25°C. Calculate
the mass-transfer coefficient and mass flux at the
kc' point of consideration. Specify the direction of
diffusion.

Table 1 3.3 MASS-TRANSFER COEFFICIENTS FOR VARIOUS EXPRESSIONS FOR DRIVING FORCES

Na (driving force)
General equation:

" [-4-A
\ coefficient /

where Na IA-\ = mass flux, lb moles/hr ft

2

Coefficients

Expression for Units of Driving Equimolar Transfer of a through Units of

Driving Force Force Counterdiffusion Stationary b Coefficient

FOR GASES
lb moles lb moles of a transferred
ca1 — ca *c *c
ft3 hrft 2 (lb moles/ft 3 )

lb moles of a transferred
Pal ~Pa atm Kg' G ~
k to
RT RT hr ft 2 atm

Va\ ~Va mole fraction V ~

k c 'P
RT
k yv = —
kc P
RT
lb

hr
moles of a transferred
ft 2 (mole fraction)

kcP{yb y n?b ) k c P(yb ^)(Yb) moles of a transferred

ky =
'

Ya\ - Ya mole ratio kY

RT RT hr ft 2 (mole ratio)

FOR LIQUIDS
lb moles lb moles of a transferred
Ca1 -Ca
'

kL 'kc' kL =k c
hr ft 2 (lb mole/volume)

kc P lb moles of a transferred
xa1 xa mole fraction kx '
M
k* =
w hr ft
2 (mole fraction)

,
k c 'p{x b -\)(x b ) ^cP< x 6l)<Xfc) lb moles of a transferred
Xa\ — Xa mole ratio kx ky =
M M hr ft 2 (mole ratio)

where x, y = mole fraction

X, Y = mole ratio, moles of a/moles of b
p = average density of a mixture of a and b
M = average molecular weight of a mixture of a and b
268 MOLECULAR AND TURBULENT TRANSPORT

Solution. From Appendix D-11, for C0 2 in At this Reynolds number the air is in turbulent
air at 25° C and 1 = 0.164 cm 2 /s and /V Sc =
atm Q flow.
0.94. As shown in Chapter 9, the mass diffusivity Airis essentially insoluble in water, so that

of gases varies inversely with the pressure; there- this an example of diffusion through a nondiffus-
is

ing gas. In this case Equation 13.70 applies

fore,

Cblr Pbln
(P —
—
P
\

'
) = 0.164
1

10
= 0.0164 cm 2
/s
Pt
2
where subscript b refers to the stationary compo-
This answer can be converted to English units of
2
nent, air in this case.
ft /hr.

(0.0164M3600) By definition,
= 0.0635 2
2
ft /hr
(30.5) (Pbi ~Pb)
Pbh
In gases the Schmidt number vl@ is independent of
pressure because v and Q each vary in the same
"(£)
manner with pressure. The partial pressure of CO2 P6 1 =(Pr-p a i) = (10-8.2) = 1.8 atm
in the gas phase is equal to the product of mole
fraction of C0 2 and the total pressure. Pb = (Pt-Pa) = (10- 1) = 9 atm

(1.8-9) -7.2
pa =xa />=(0.1)(10) =

The value of p a1 must be

statement reports the concentration of C0 2 in the
liquid at the interface as 0.005 mole fraction. The
calculated.
1 atm

The problem
Pblm ~

o -0 = AA7 atm

partial pressure of C0 2 in the gas phase at the

Pblm
= —
4.47
— = 0.447
interface is equal to the equilibrium vapor pressure Pt 10
of C0 2 over the liquid. For the carbon dioxide-
water system, the Henry's law constant (H) can be
Equation 13.70 may be solved for k c
used to calculate p a In Appendix D-3 at 25°C,
.
t

the Henry's law constant H = 1640 atm/unit mole k c = 0.023 3w*' M3

fraction C0 2 in the liquid. Then
(f)(±K»"
Pblm> K

(0.0635) 1

Pa 1 = Hx a ,
= (0.023) (30,600)°-83 (0.94) - 33
0.447
An = 0640X0.005) = 8.2 atm

2
Then assuming that the properties of air are a = 100 lb moles/hr ft (lb mole/ft 3 )

satisfactory approximation for the air— carbon

dioxide mixture, the properties of air at 25°C and Equation 13.68 is used to calculate the mass flux.
10 atm are
Na IAy = -/r c (ca1 -c a )

= 0.018 centipose
Pair
ca =pa /RT
= 0.0808 3
Pair lb/ft at standard conditions = 0.729 atm 3
/? ft /lb mole °R
Pair at 10 atm, 25°C = 0.080 (273/298X10/1)
T= (298X1.8) °R
= 0.74 lb/ft 3

Testing for turbulence shows Na IA^ =-—MAH -p a )

100
(3)(0.74) (8.2- 1)
[Dvp) (h) (0.729X298X1.8)
/v Re = = 30,600
(0.018 x 6.72 x 10- 4 :
= -1.84 lb moles/hr ft
2
 FUNDAMENTALS OF TRANSFER MECHANISMS 269

The negative sign indicates transfer from the wall Equation 13.54 may be written for momentum
toward the center of the tube, since flux is positive transfer by substituting pv for T and [T y gc )y for
in the direction of increasing radius. This is also
evident from the high value of 8.2 atm forp a1 the ,

equilibrium vapor pressure of C0 2 over a liquid (T y g c ): =-&ApV\ ~Pv] (13.73)

with concentration 0.005 mole fraction C0 2 in
water. For the fluid adjacent to a stationary duct v-^ = 0,
so that

ANALOGIES AMONG MASS, HEAT, AND MOMENTUM ^Y gc )y = tjpv

TRANSFER and
Heretofore, mass, heat, and momentum transfer (Ty£ch
were examined almost simultaneously, with the S,= (13.74)
pv
general similarity of behavior always made evident.
The similarity among molecular-transport phen- If Equation 13.74 is divided by v, the result can be
omena was examined in Chapters 9 and 10. In substituted into Equation 13.9a.
Chapter 12, the similar dependence of mass, heat,
and momentum transfer upon eddy activity was f _(Ty g c )< l
ST
presented. Up to the present, a close-knit relation- 2
8 v p v
ship exist between the three transfer
seems to
phenomena the laminar and turbulent regime.
in or
Other evidence exists to show more quantitative
fv
similarity in the turbulent regime. In the following
13.75)
section, several of many analogies between the
transfer phenomena will be examined. Analogies
will be drawn between the transfer-rate correla- The Reynolds Analogy
tions presented previously in this chapter. In addi- The Reynolds analogy (48) is of historical impor-
tion, some further information will be presented in
tance as the first recognition of analogous behavior
the Martinelli analogy. of momentum- and heat-transfer rates. As such it
The analogies are useful tools to the student undoubtedly was the guidepost for much of the
as an aid to rapid understanding of transfer phen- later work in the same areas. Although it is limited
omena and to the professional as a sound means to
in application, it is useful in any work involving
predict behavior of systems for which limited
common gases, and in fact it has been one of the
quantitative data are available. In this section, the
more powerful tools in analyzing the complex
analogies will be used to elucidate the mechanism
boundary-layer phenomena of aerodynamics.
of transfer.
Reynolds postulated that the mechanisms for
transfer of momentum and heat are identical. The
The Transfer Coefficient for Momentum postulate may be written as
The development of transfer-rate correlations for
turbulent flow shows one inconsistency. The cor- 4(a+£a ) 4(v + ET )

relation momentum-transfer rate utilizes

of the 13.76)
a friction-factor— Reynolds-number relationship, lq D yT D
whereas transfer coefficients are used to correlate
or, if coefficients are substituted,
heat- and mass-transfer rates. These correlations are
traditional practice in engineering.* If a surface
coefficient for momentum transport is written,
h
= *, :13.77)
perhaps further similarities can be discovered. Cpp

* The momentum-transfer work was done by civil

early Equation 13.77 may be divided by v, and the
engineers, who had no professional interest
in heat or mass transfer.
friction factor substituted for £ T lv as in Equation
The early heat-transfer work was done by mechanical engineers,
13.75.
who had a small professional interest in momentum transfer. Mass
transfer lies in the province of chemical engineering. It is this diverse
h f
background that probably hindered immediate recognition of the (13.78)
similar behavior of the transfer phenomena. c P pv 8
270 MOLECULAR AND TURBULENT TRANSPORT

Equation 13.78 is a mathematical statement of the The Colburn Analogy

Reynolds analogy. The group hlc P pv is the Stanton
number (/V St ). The Stanton number is related to
the Nusselt, Reynolds, and Prandtl numbers as
follows:
N Nu
Colburn (5) examined the empirical equations for
heat and momentum transfer in the turbulent
regime and presented a more general analogy. If
Equation 13.65 for heat transfer in the turbulent
regime is divided by the product (A/ R e )(/Vp ), then
r

/Vst
= (13.79) 0.1 4
/v Re /v Pr
8
(/V Re )°- (/V P r)
033
(-)
Therefore, Equation 13.78 may be written Wnu
= 0.023
A/ Re A/ Pr (/V Re )(/V Pr )

= /V St = — 13.80)
/V Re /V Pr 8 This expression can be rearranged to

tally,
When Equation
it

for gases in turbulent flow.

13.80 was tested experimen-
was found to correlate data approximately
It did not correlate
(/V St )(/Vp
^(T = 0.023(/V R( -0.2

!
13.83)

experimental data for liquids in turbulent flow nor Equation 13.83 is also plotted in Appendix C-5.
for any fluids in laminar flow. All gases are charac- Equation 13.15 for momentum transfer may be
terized (see Chapter 9) by values of the Prandtl divided by 8 and written as
number in the general range between 0.6 and 2.5.
Upon more critical examination, it was concluded = 0.023(/V Re -0.2 ) (13.84)
(61) that the Reynolds analogy is valid only at 8
(A/ Pr = 1. By similar reasoning, an expression may
)

Equation 13.84 is also plotted in Appendix C-5.

be written based upon the postulate that mass and -0 2
The term 0.023 (A/ Re can be eliminated from
momentum are transferred by an identical mechan- )

Equations 13.83 and 13.84

ism. This postulate may be stated as

0.1 4
M?+EN ) A(v + ET )

(/V. H/V Pr )
2 '3^l) (13.85)
(13.81) ; t

yN D yT D
Ifthe appropriate transfer coefficients are substi- The term jqdefined as (/V St )(/V Pr 2/3 (mi/m) 014
is )

tuted and if Equation 13.81 is divided by v an ,

and is called the "/" factor for heat transfer. Equa-
analogy equation can be written tion 13.85 can be written

113.82)
v h (13.86)

When Equation 13.82 was tested experimentally, it
was found to correlate data approximately for
gases in turbulent flow but did not correlate data
for liquids in turbulent flow nor for any fluids in
laminar flow. Gases are characterized by values of
This equation is a statement of the Colburn
analogy between heat and momentum transfer.
Equation 13.86 is derived from two equations
that correlate experimental data over the range
e < 300,000 and 0.6 </V Pr < 100.
1 0,000 < N R

the Schmidt number near unity. More critical Equation 13.85 shows a functional relationship
examination revealed that Equation 13.82 is valid between heat and momentum transfer and the
(61) only at /V Sc = 1. Prandtl number as predicted in the Reynolds
Although the Reynolds analogy is of limited analogy. Note that in Equation 13.85, if A/ P = 1
utility, one significant conclusion may be drawn — and = 1,
(mi/m)
that, at /\/p = A/ Sc = 1, the mechanisms for mass,
r

heat, and momentum are identical. A second con-

/V S t = 13.80)
clusion may also be drawn — that, for other fluids,
!

8
transfer processes differ some manner func-
in
tionally related to the Prandtl or Schmidt number. which the Reynolds analogy.
is

Application of the Reynolds analogy appears in The

empirical equations for mass transfer and
Example 13.9. momentum transfer in the turbulent regime may

t^mmmt
 FUNDAMENTALS OF TRANSFER MECHANISMS 27

212
be compared the same manner. Equation 13.69
in Equation 13.92 may be divided by (v)

for mass transfer can be divided by the product

(/V Re )(/V Sc and rearranged to
) 13.93)
c P p(a) 2/3 @) 2/3
!

(u)
2/3

% v
(/VSc )
2/3 (/V
Re )
003 = 0.023(/V )-°- 2
Re
Equation 13.92
stant Reynolds
is

number
understood to apply to
for each
a con-
of the transfer
(13.87)
phenomena. Equation 13.93, demonstrates that
the ratio of the transfer coefficient to the mol-
03 ecular diffusivity to the 2/3 power is a constant.
If (/V Re )
"

is taken as unity,* Equation 13.87

This simple relationship is valid for the empirical
may be written
equations used for the Colburn analogy and for the
mass-transfer analog of the Colburn analogy, sub-
k' 2/3 = ject to the correction mentioned above. However,
(/V Sc ) 0.023(/V Re r - 2
(13.88)
if the mechanism definition of the coefficients is

substituted into Equation 13.92 by use of Equa-

2
can be eliminated
_0 tions 13.76. 13.66, and 13.57,
The 0.023 (/V Re
factor )

between Equations 13.88 and 13.84 to give

A (a+Eq ) 4 i<J+EN ) 4 (v + Eq )

(13.94)
kc
> 2/3 =
Dyq (a) 2'3
Dy N ((/) 213
D y T (v) 2/3
(/VV
'ScJ i
13.89)
The term AID cancels and Equation 13.94 may be
written
213
The term jN can be defined as (k c '/v)(N Sc ) and
"/'"
factor for mass transfer. Therefore
(oc+Ea )
(<s + EN (v + Er
called the ) )

0\2I2 13.95)
yq (a)
2/3
InM lAv)
2/3 !

= 13.90) This equation a restatement of Equation 13.92

In ;
is
8
and applies as long as the Reynolds number is the
same for each phenomenon. The Reynolds
transfer
Equations 13.86 and 13.90 can be combined to
number may have any value within the range of
give
validity of the equations used in the derivation.
The concentration-distribution functions y Ql y N
(13.91) and y T by definition, have a maximum value of
lq ~In ,

8 unity and a probable minimum value not below

0.7. Within the range of validity of the correlation
This equation applies over the following ranges: for equations, transfer coefficients vary over about a
heat transfer 10,000 < A/ R e < 300,000 and 0.6 < fifteenfold range. If the concentration-distribution
<100; for mass transfer*
/V Pr 2,000 <A/ Re < functions (y q y T and yu) are taken as approx-
, ,

300,000 and 0.6 Sc < 2,500. <N imately equal, whereas the molecular diffusivities
Some interesting conclusions may be drawn (a, v, and (/) vary over the range of validity of the
concerning Equation 13.91 if transfer coefficients empirical equations, then at a given Reynolds
are substituted in each of the terms. If this substi- number the mean eddy diffusivities of mass, heat,
tution is made and the velocity canceled from each and momentum will not be equal. In Chapter 12
4 = 1
term, and letting (jjl^ //i) unity, then the point value of the eddy diffusivity was assumed
equal at a given flow rate irrespective of the trans-
2/3 2/3 = f ferent property. There is no inconsistency in these
(/Vp r ) = /r c '(/V Sc ) 'r (13.92)
Cpp observations. The definition of mean eddy
diffusivity as given by Equation 13.49 shows the
* The error in the assumption may be considerable. Over the integration of the point value over the entire duct.
range of Reynolds number for which this assumption is applied, The integration includes the use of the gradients
(/V Re )
003between 1 .32 and 1 .46. The surface of the fluid in
varies d{pv)ldr, d(pcpT)/dr, and dc a ldr. If at a particular
a wetted-wall column may be rippled, which can introduce
flow condition the point eddy diffusivites are equal
geometrical disturbances and increase the rate of mass transfer in
the gas phase. The higher exponent on the Reynolds number may be
but the gradients differ with trar\sferent property,
a result of the rippling of the fluid surface. the mean values that result from integration will be
272 MOLECULAR AND TURBULENT TRANSPORT

different. Reynolds analogy, in which a =

In the Other analogies have been written that may
:
= the values of the mean eddy diffusivities
,
be applied with varying success over the entire
were equal, implying that the gradients were equal; range of Prandtl number. These analogies are based
but for the Colburn analogy, in which a = = i upon the model for turbulent flow described in
the mean eddy diffusivities are not equal, implying Chapter 12, each with its own set of assumptions.
differences in gradient for different transferent The work of Martinelli (34) is a good example of
properties. This concept will be amplified further an analogy between heat and momentum transfer
in the next section on the Martinelli analogy. that can be derived from a more detailed model.
The empirical equations in this section are Martinelli made the following assumptions.
valid for gases and liquids within the range 0.6 < 1. The temperature driving force between the
/V Pr <100 and 0.6 </V Sc < 2500.
This range
wall and the fluid is small enough so that —
nln-\ 1.
includes most fluids of practical importance except
Well-developed
2. turbulent flow exists
liquid metals. Little data are available for liquids of
within the test section.
extremely high Prandtl number because such
3. Heat tlux across the tube wall is constant
liquids must have extremely high viscosity, high
along the test section.
enough that pumping machinery to maintain the
4. Both stress and heat flux are zero at the
high velocity of turbulent flow does not exist. The
center of the tube and increase linearly with radius
liquid metals are a major class of real fluids whose
to a maximum at the wall (see Equations 10.104
heat-transfer behavior cannot be predicted with the
and 12.33).
empirical equations cited above. These fluids are
5. At any point Eq = E.
characterized by low viscosity (approximately that
6. The velocity-distribution equations, Equa-
of water) but high thermal conductivities (approx-
tions 12.23, 12.24, and 12.25, are valid. The equa-
imately 30—60 times higher than water). Other
tions
properties are such that a typical value for liquid
metals is N P = 0.01. This value is well outside the
,
d(pv)
range of validity of the empirical correlations cited Ty = -(V+ET ) (12.15)
above. If the empirical equations were faithful dx
statements of mechanism, they would undoubtedly and
apply over the entire range of Prandtl numbers.
dipcpTi
The criticism here is much the same as the criticism —q = -(a + £Q )— (12.36)
of Equations 12.23, 12.24, and 12.25, the velocity- A dx
distribution equations in the turbulent regime. The
equations are valid for predicting behavior but do are modified according to assumption 4 and
not faithfully follow the mechanism of the written for cylindrical geometry
behavior. Once again, it might be concluded that
these problems will disappear as soon as a complete d{pv)
mathematical picture of turbulence is written. «*(£)-*'*> -^ <
12.22)

Application of the Colburn analogy appears in

Example 13.9.
and

d(pCpT)
The Martinelli

The Reynolds analogy

Analogy
is a significant contribution
©,o- <a+E„)
dr
(12.30)

because demonstrates similarity of mechanism

it The velocity-distribution equations are used
among transfer phenomena, but it is limited to unit to evaluate E. in Equation 12.15, and these values
values of the Prandtl and Schmidt numbers. The may be substituted into Equation 12.36 according
Reynolds analogy does nothing to elucidate the to assumption 5. Equation 12.36 is integrated so
mechanism. The Colburn analogy demonstrates that the point value of temperature is expressed as
numerical similarity among transfer phenomena a function of position, and this equation is con-
over a wider but still limited range of Prandtl and verted into the form /V N u =p(/V Re Pr ,f). The , N
Schmidt numbers. The Colburn analogy implies equations are complex and are not presented here.
that the correlation equations are not faithful Instead, the results are presented in the form of a
statements of the mechanism, however reliable plot of /V N u as a function of /V R e at values of Pr N
they may be for predicting numerical values of varying between and «=, in Figure 13.6a.
coefficients. Because the Martinelli analogy can be applied
 I UNDAMENTALS 01 TRANSIER MECHANISMS 273

potassium alloys. The equation resulting from this

correlation is

/V Nu = 7 + 0.025(/V Pe 08 )
* (13.96)

where /V Pe = the dimensionless group called the

Peclet number. This group is equal
tothe product (A/ Re )(A/p and repre- r )

sents the mechanism ratio:

Total (molecular + turbulent) momentum transfer

heat transfer by molecular mechanism

For convenience, Equation 13.96 is plotted in

Appendix C-5.
Data for other liquid metals such as mercury,
lead,and bismuth show a somewhat lower Nusselt
number than that predicted by Equation 13.96 at
comparable values of the Peclet number. The lower
values of the Nusselt number are presently
1000 attributed to the nonwetting characteristics of
these liquid metals with respect to the heat-transfer
tubing, although the hypothesis is not clearly
established (25). For mercury, lead, and bismuth,
Equation 13.97 is recommended (31).

= 0.625 - 4
A/ Nu (A/p e ) (13.97)

Equation 13.97 is plotted in Appendix C-5 for

convenience.
The Martinelli analogy predicts the values of
/V N u for liquid metals with fair precision and also
predicts heat-transfer coefficients for those fluids
that within the valid range of the Colburn
fall

analogy. The agreement between

this analogy and
experimental data can be seen in Figures 13.66 and
13.6c. Figure 13.66 is a plot of A/ N u asa functionof
A/ Pr as predicted by Martinelli at A/ Re = 100,000.
Data for the same conditions taken from Equation
13.96 and Equation 13.65 are plotted on the same
figure. Figure 13.6c is a similar plot taken at
0.0001
/V Re = 10,000. Examination of these plots shows

(c)
that the Martinelli analogy is in fair agreement with
experimental data over the total available range of
Figure 13.6. The Martinelli analogy, (a) /Vn u as a function
Prandtl number. Thus, this work is a significant
of /V Re and /Vp r . Comparison of the Martinelli analogy and
contribution to the understanding of the mech-
empirical equations at: (b) /V Re = 100,000, /u/|Ui = 1. (c)

/V Re = 10,000, ulyn s
1 (35). (With permission of McGraw- anism of heat and momentum transfer in the
Hill Book Co., Copyright © 1954.)
turbulent regime.
Analogies have been written for mass transfer
over the entire range of /V Pr it can be tested with using techniques similar to those of Martinelli and
,

experimental data. Lyon (32) predicted the form

* This equation does not contain a term (m/mi) and attaches
of a correlation equation that is valid over the
greater importance to the Prandtl number. In contrast to gases.
Prandtl number range of by a method
liquid metals water, and organic liquids, the molten metals have viscosities that
similar to the work of
Martinelli and tested the are relatively insensitive to temperature, as well as exhibiting greater
equation with experimental data using sodium and molecular conductivity.
274 MOLECULAR AND TURBULENT TRANSPORT

others (56). The experimental data and analogies boundary layer. These correlations are shown in a
agree over a wide range provided the assumption later part of this chapter. Although the analogies
E N = 1.6£~ is used instead of the equality of eddy have definite usefulness in predicting behavior
diffusivities assumed by Martinelli for heat trans- beyond the range of known data, the tested
fer. A reasonis offered for this difference. For empirical equation is still the most valuable source
example, the surface of fluid in a wetted-wall for design data.
column may be rippled and therefore tend to
transfer mass in the gas phase because of an Effect of Prandtl Number on the Temperature Profile
increase in turbulence due to the surface irregu-
The ratio y= (Ti - H/CTt - T ) defined in
larities. Heat transfer in this same kind of appara-
Equation 13.47 can be examined qualitatively for
tus invariably is greater than under similar con-
heat transfer. Temperature— raaial-position data are
ditions in dry tubes.
plotted Figure 13.7 using a coordinate system
in
The foregoing material is impressive evidence
that is applicable to cylindrical geometry irrespec-
of the value of the analogies to define a mechanism
tive of the size of the cylinder and irrespective of
for turbulent transfer of mass, heat, and momen-
the actual values of temperatures. The temperature
tum. In summary, transfer depends upon the
function is (7i — T)I(T^ — 7" ), and the position
product of the molecular diffusivity and the
function is (r, — r)lr-\ The
. subscripts are indica-
gradient for a part of the transfer and upon the
tive of location, where
and r\ represent the 7",
product of the eddy diffusivity and the gradient
condition at the wall, 7" represents the condition
for the remainder of the transfer. The eddy diffusi-
at the center of the tube, and T and r represent any
vities for mass, heat, and momentum are probably
radial position in the duct. The temperature
identical in physically identical apparatus. These
profiles for fluids of Prandtl numbers 4, 0.7, and
observations support the assumption that
0.024 are plotted. All data are taken for systems at
E T = EN = Eg made in Chapter 12. The molecular
/V Re = 40,000. In addition, the dotted line repre-
diffusivities are certainly not identical, and the
sents the velocity profile (vlvc ), plotted against
effect of this difference is shown when heat-
(r, - r)lr^ for all fluids. The data of the three
transfer characteristics of substances such as water
temperature curves were cross-plotted to find the
and mercury are compared. The high thermal con-
temperature profile curve at /V Pr = 1. This curve is
ductivity of mercury results much higher heat-
in
congruent with the dotted velocity profile.
transfer coefficients constant
at Reynolds
The plot may be used to predict qualitatively
numbers. For water, the contribution of molecular
that y Q which is based upon the mixing-cup mean
,

transfer to the total transfer is almost negligible,

value of f, will increase with increase in Prandtl
whereas, for mercury, the contribution of molec-
ular transfer to the total transfer is of the same
order of magnitude as the contribution due to 1.0

turbulence.
The Martinelli analogy is a valuable starting
point for the prediction of transfer behavior of 0.8

substances whose properties are outside the range

previously investigated. For example, for any fluid
an equation of the form //
<0
//
= constant a 6
A/ Nu (/V Re ) (/V Pr ) (13.98) I |
//

"" "- 0.4 -

might be predicted.
The behavior of fluids of very high Prandtl
'
number may be predicted from Figure 13.6a in
02
which the curves for /V Pr = 100 and /V Pr = °° each
have a slope of approximately 0.8. The two curves
are approximately coincident, and therefore
a = 0.8 and b = in Equation 13.98. No prediction 0.2 0.4 0.6 0.8 1.0

can be made as to the effect of the viscosity (r

}
- t)It x

ratio (m/Mi)- Figure 13.7. Temperature and velocity profiles at

The analogies have also been used to predict /V Re = 40,000 (35). (With permission of McGraw-Hill
behavior for transfer to flat plates, as in the Book Co., Copyright © 1954.)
 FUNDAMENTALS OF TRANSFER MECHANISMS 275

number at constant /V Re . An examination of Fig- number in the Reynolds analogy, then

ure 20.2 shows that y T (or in this case v/v max in- )

creases with increase in Reynolds number. It may =

/v Nu (/v Re )(/v Pr
be assumed then that y q is a function of both the )(!)
Reynolds and Prandtl number.
Comparable data for mass transfer are not
available, but in the light of analogous behavior it
may be assumed that yN increases with increase in /V Nu = 164.2
Reynolds number and Schmidt number.
By the Colburn analogy, use Equation 13.85.

/u \°- 1
f
Example Compare
the value of the
13.9.
Nusselt number, given by the appropriate empirical
MWWW'«g) -3
equation, to that predicted by the Reynolds,
Colburn, and Martinelli analogies for each of the By substituting A/ Nu /(/V Re /v>,) for /V St and re-
following substances at /V Re = 100,000. (a) SAE ranging, this equation becomes
10 lube oil, (b) water, (c) air (1 atm), and
mercury. Consider all substances at
(d) liquid
100°F, subject to heating with the tube wall
/V Nu = (/V Re )(/V Pr )
1
/3^^
at 150°F.
0.14
Data: Properties at 1 atm, in Btu, hr, ft, and = (105>(0.71)1/3(«M
lb \ 8 /\0.020/

M100 M150 /V Nu = 204

(centi- (centi-
poises) /f!oo P100 c/>ioo (/Vprhoo poises) By the Martinelli analogy, the data are avail-
able in Figure 13.6a.
Air 0.018 0.0157 - 0.2401 0.71 0.02
Water 0.69 0.365 61.9 1. 4.57 0.42 A/ Nu = 170
SAE 10 31. 0.071 56.6 0.45 47.5 12.
Mercury 1.45 5.1 84.3 0.33 0.0234 1,40 Solution for mercury. Equation 13.97 is the
appropriate empirical equation that can be used to
predict the Nusselt number.
At /V Re = 100,000, f= 0.0184 (from Equation
= 0.625(/VPe - 4
13.15). /V Nu )

(/Vpe) = (/V Re )(/Vp r ); therefore

Solution for air. For air, the appropriate
= 0.625(10 2 (0.0234)° 4
empirical equation is Equation 13.65. /V Nu )

/V Nu = 13.9 (most accurate value)

0.14
/V Nu = 0.023(/V )°-8(/V )i/3^
Re Pr By Reynolds analogy,

/0 018\° -1 4
/Vs, = -
/V Nu =0.023(100,000)°- 8 (0.71) 1/3 (-^
j

/v Nu = /v Re /v Pr
-
/Vimu
= 204 (most accurate value)

By the Reynolds analogy, use Equation 13.80. = (10 5 (0.0234

/V Nu )

i°T
= 5.38
/Vst =
8
By Co/burn analogy,
0.14
f
According to Equation 13.79, A/ St = A/ N U //V R e /V Pr .
2/3
/Vst(/V Pr ) (-)
If the term f/8 is used to replace the Stanton
276 MOLECUl AR AND TURBULENT TRANSPORT

Because /VSt = W Nu /(/V Re /V Pr ), the Colbum analogy Solution for SAE 10 oil. The heat-transfer
equation can be written coefficient for the
10,000 /V Re > oil can be
M predicted by Equation 13.65 because

*.-*..«U"*(3©"' = 0.023(/V Re )°- 8 1/3 /M>°-

-
14
/V Nu (/V Pr )

14 V1.
^1.
1/ ,/0.0184./1.45v°
/V Nu = (10^)(0.0234)^3(-__)(_-) 0.1 4
= 0.023(10 5 )
08 (47.5) 1/3 (— )
= 66.1
= 935 (most accurate value)
By the Martinelli analogy, from Figure 13.6a,
By the Reynolds analogy,
A/ Nu = 20
f
A/ St =
Solution for water. Equation 13.65 can be 8

used to predict the heat-transfer coefficient for Because A/ St = /V Nu /(/V Re /V Pr ), the equation may be
water. written
0.14
«H. = «R.«P,(£)
/V Nu = 0.023(/V Re ) 08 W Pr )
033
w
(-)
Mi

= 0.023(10 5 )°- 8 (4.57)°- 33 ( —

0.69i° 14

/V Nu
= 105(47

= 10,500
. 5) (^84)

A/ Nu = 406 (most accurate value)

By the Colbum analogy,
By Reynolds analogy,

Because NSx = NM u /Nn e Np r , the equation may be

Because /V St = /V Nu //V Re /V Pr , written
,0.1 4

/V N u=A/Re/Vp r
(|)

4
/V Nu = (10 5 )(4.57)(^-
)
.03,4,5)0-33(1)^(0^)
/V Nu = 1050 = 935

By the Colbum analogy, By the Martinelli analogy, from Figure 13.6a,

/V N u = 800

The values of the Nusselt numbers are tabulated

below for easy comparison. The substances are
Because /V St = N Uu /N He N Pr , the equation may be
written in terms of increasing Prandtl number.
written
0.14
Nusselt Numbers

Empirical
Empirical Equation
0,4
= (1
0S„4.57,"3(166) (M184) Equation (Most Col- Mar-
A7 Pr Number probable) Reynolds burn tinelli
M.05 l
8 '

Mercury 0.0234 113.97) 13.9 5.38 66.1 20

= 406
Air 0.71 (13.65) 204 164 204 170
Water 4.57 (13.65) 406 1050 406 400
By the Martinelli analogy, from Figure 13.6a
SAE 10
oil 47.5 (13.65) 935 10,500 935 800
/V Nu = 400

Note that over the entire range, the Martinelli
analogy is in good agreement with the empirical
correlation of experimental data. The Colburn
analogy agrees perfectly, as it
water, and oil, because the Colburn analogy for
these substances is the same as the empirical
equation. The discrepancy for oil between the
Martinelli analogy and the empirical equation is
due to the absence of a viscosity ratio term in
the Martinelli analogy. If this factor is included,
the value becomes (800)0.142) =913, where
14 = (31/12)°- 14 = 1.142.
(M/jUi)°-
Summary
This section on analogous behavior of transfer
phenomena has described three analogies. The
Reynolds analogy is based upon the assumption
that the mechanisms for heat and
identical; subsequent extensions include mass
transport. Test of the analogy shows this assump-
tion to be true at N Pr = 1 and approximately true
at N Sc = 1. The Colburn analogy is based entirely
upon similarity of empirical equations, and there-
fore, over the range of application, it is in good
agreement with experiment Critical examination
of the analogy shows that it does not represent a
mathematical statement of the mechanism. The
Colburn analogy when applied to mass transfer
results in an error of 1.32 to 1.46, as mentioned in
an earlier footnote. The reason for the discrepancy
is attributed at least in part to surface effects
the descending film of fluid. The discrepancy then
is geometrical, and the implication may be that the
systems are not exactly analogous in a geometrical
sense. The Martinelli analogy
equality of eddy diffusivities of heat and
tum as a statement of mechanism. The Martinelli
analogy shows remarkable agreement over the
entire range of Prandtl number, indicating that the
analogy is a reasonable statement of the mechan-
ism. A similar analogy for mass transfer agrees with
mass-transfer data if EN = 1.6 £> is used. This
discrepancy is attributed to the difference in
geometry between a cylindrical tube of dry metal
and a wetted-wall column. The mechanism for
geometrically similar systems may be considered
dependent upon the general equation.
dT
rP = -{6 + E)
dx
where in the light of present knowledge
ET = Eg = E/\/
FUNDAMENTALS OF TRANSFER MECHANISMS 277
HEAT AND MASS TRANSFER IN THE BOUNDARY
LAYER
It was shown earlier that a flat plate in a stream
should, for air,
produces a boundary layer of fluid of velocity
lower than the free-stream velocity and that this
layer lies close to and is affected by the surface. If
the same shape at some temperature T is placed in
:
a free stream at temperature Tfs a temperature
,
boundary layer will also form, with a similar (but
not necessarily identical) shape as shown in Figure
13.8. Just as the boundary layer is a region of
velocity gradient in the gas phase adjacent to the
boundary through which momentum is transferred
from the fluid free stream to the boundary, the
thermal boundary layer is a region of temperature
gradient through which heat is transferred to or
from the free stream to the boundary. In the event
momentum are that the plate surface temperature is T fs between
y= and y = y^ and then at y-\ the temperature
,
changes from Tfs to Ty, the temperature boundary
layer begins to form at yy Meanwhile, the velocity
.
boundary layer would have begun to form at / = 0.
The two layers develop according to the appro-
priate fluid and system properties. An analogous
mass-transfer boundary layer can also occur. Con-
sider a pan of water subject to a gas flow parallel to
the water surface. If the surface stress of the wind
does not cause ripples on the surface, the smooth
surface is geometrically similar to the flat plate,
and if the partial pressure of the water in the air is
on different from the vapor pressure of water at the
liquid a boundary layer of water-vapor
surface,
concentration will form along the water surface.
The entrance to a heat exchanger and the entrance
is based upon to a wetted-wall tower are other examples of
momen- thermal or mass-boundary layers combined with a
momentum boundary layer. As in momentum
transfer, when the edges of the thermal or mass
boundary layer meet at the center of the tube, the
flow is fully developed, and no further growth of
boundary layer takes place with progress through
"f>
Vfs Tfs
,
Figure 13.8. Simultaneous formation of momentum
boundary layer and mass or thermal boundary layer.
278 MOLECULAR AND TURBULENT TRANSPORT

the tube. The temperature or concentration can The friction factors have been related to
continue to respond to transfer. Reynolds numbers in equations written earlier. If
As shown in Figure 13.8, the momentum Equation 13.26 is combined with Equation
boundary layer begins to form at some point y = 0, 13.100, then
and the thermal or mass boundary layer begins at a
different point /,. For example, a heat exchanger 0.037
tube may extend through a tube sheet in such a 2
13.102)
cpvfsp
-

(/V Re ),
manner that the length of tube between y and y^
is at the same temperature as the free stream, in
and, if Equation 13.24 is combined with Equation
which case no thermal boundary layer forms 13.101, then
upstream of y\ Beyond the tube sheet, beginning
.

may
at yy, the tube be heated. In this case, the 0.0228
momentum boundary layer begins at y = and the
cP v h p (/V Re ) 6
025 113.103)
thermal boundary layer begins at y = y^
Consideration of the above suggests that the where (/V Re
K
= {yv fs p/p.)
)

analogous behavior of fluids in similar geometries y - any position on the plate down-
can be analyzed using the Reynolds or the Colburn stream of the leading edge
analogy. (W Re )6 = (oVftp/M)
5 = boundary-layer thickness at some
The Reynolds Analogy Applied to the Boundary Layer position y
The Reynolds analogy was applied to turbulent
flow through smooth tubes and found to be valid if The boundary-layer thickness (5) and position
the fluid Prandtl number was unity. The same (y) are related by Equation 13.25
reasoning might be applied to a flat plate in a free
0.376
stream. The Reynolds analogy for tubes is 13.25)
2
!

(Nne)y°-
(13.99)
cP vp pv' In a similar manner, equations can be derived
for mass transfer through the boundary layer. The
If a flat plate with a turbulent boundary layer in a conditions where the Reynolds analogy holds are
free stream behaves in a manner analogous to fully that the fluid be characterized by a Schmidt
developed turbulent flow in a tube, the following number of unity, that the boundary layer for
equations may be assumed to apply for boundary
momentum and mass start at the same point, and
layers:
that the flat plate is smooth. For smooth tubes,
h Mi (13.100)
the mass-transfer
Reynolds analogy is
counterpart equation to the
2
CpVfsP Vfs p
and
hy _ ^ly (13.101) v v2 p
! 13.82)
2
CpV h p Vfs p
where k c = mass-transfer coefficient
'
for equi-
where h y = heat-transfer coefficient at a specific molar counterdiffusion
point y measured from the leading
edge. The entire plate is heated; that If a flat plate on which a boundary layer builds in a
is, the momentum and thermal bound- free stream that turbulent behaves in analogous
is

ary layers begin at the same point manner to fully developed flow in a tube, the
(t yhy = wall stress at a point y measured from following equations may be assumed:
the leading edge
h = average value of the heat-transfer co- (k c ') {r Y h
efficient for the section between the ;13.104)
2
leading edge and some position y Vfs Vfs P
r
yh average wall stress for the section and
between the leading edge and some
(Arc')
c 'y
point y
2
(13.105)
Vft = free-stream velocity Vfs Vfs p
 FUNDAMENTALS OF TRANSFER MECHANISMS 279

Equations 13.104 and 13.105 can be combined From the foregoing material, it has been
with Equations 13.26 and 13.24, respectively, to established that the analogous behavior of mass,
give heat, and momentum in smooth tubes can be
extended to other examples of transfer to a fluid
0.037
stream from a smooth surface in which the flow
13.106)
WRe ,
0.2
is parallel to the surface. The boundary-layer

and equations for heat and mass transfer find

(kc\ 0.0228 occasional application in specialized equipment
0.25
113.107) components that do not fall within the conditions
(/Vr e'6
of fully developed flow in ducts.
where
(k c ') = average mass-transfer coefficient Example 13.10. Stack gases leave a boiler,
from the leading edge to a position y pass through a steel duct 3 ft by 3 ft and 16 ft
{k c ')
y
= point value of the mass-transfer co- long, and enter a masonry chimney. Estimate the
efficient at a position y heat-transfer coefficient in the duct if the gases are
at 900° F and 1 atm. Consider the stack gases to
The mass-transfer coefficient for diffusion of gas a
through a stationary gas b may be written by have the same properties as air. The gas velocity is
noting that 50 ft/sec.
Solution. In a duct of this size and length, the
cblm
= kr (13.108) boundary-layer equations may be used, provided
that the boundary layer is small with respect to the
where duct size and the effect of the interactions of the
= logarithmic mean concentration of boundary layers at the corners of the duct can be
Cbim
gas 6 neglected. For this example the boundary-layer

c t = total concentration of gases a and b thickness can be calculated using Equation 13.25
k c = mass-transfer coefficient for diffu-
5 0.376
sion of gas a through stationary gas b
0.2

Equations 13.106 and 13.107 have been

tested experimentally and agree with experiment
The physical properties of air at 900°F and 1 atm
within the limits of the assumptions used in deriva-
can be found Appendix D.
tion. = 0.0353 centipoise
Mair9oo°F

The Colburn Analogy Applied to the Boundary Layer 28.9 lb 1 / 492

= 0.0291
If the fluid properties in the system are such that
Pair 900°F
L359ft LpV
3
ST 460 + 900
J p
v
)
lb/ft"

the Prandtl number is greater than 0.6 and less

C/»,air900°F= 0.260 Btu/lb°F
than 100, then in a similar manner the Colburn
analogy can be used to deduce equations for heat ( .
yv fs p (16)(50)(0.0291
(/V R e'y = 985000
and mass transfer in the boundary layer. For (0.0353 x 6.72 x10" 4
H )

turbulent flow in which all boundary layers begin

= 9.85x 10 5
at / = 0, and (pipy = 1, the Colburn analogy may
)

be stated
This answer is in excess of 10 5 and therefore most

_*!L_
CpVfsP
Wpr)
2/3 = V
vfs
Wsc)
2/3 = <™feV
2
Vfs P
of the boundary layer is turbulent. Equation 13.25
may be applied to determine the thickness of the
,

boundary layer.
0.0228
= 13.109) /(0.376) 16(0.376)
°- 25
(/VRe) 6 5 = = 0.38 ft

and
(/V Re L - 2
(9.85 x10 5 02 )

The boundary-layer thickness

h
CpVfsP
(/V Pr )
2/3 = —
Vfs
(/V Sc )
2/3 =
(T y g c h

2
v fs P
0.037

(/VRe),
02
is in the order of 10

percent of the duct dimension. The remainder of

the duct constitutes the free stream. Since the
(13.110) problem statement requests an estimation rather
280 MOLECl LAR AND TURBULENT TRANSPORT

than a rigorous calculation, the boundary-layer tangentially. Many of the chemical engineering
equations may be used for estimation of heat- applications of transfer occur in systems that do
transfer coefficient not meet these conditions exactly. For example,
Using the Reynolds analogy, Equation 13.102 commercial pipe of the commonest variety has a
can be used to determine the average heat-transfer sufficient degree of surface roughness to increase
coefficient over the entire length of duct the friction factor by 20—30 percent. Although for
smooth surfaces heat, mass, and momentum trans-
h 0.037 fer follow a remarkably analogous pattern, when a
CpVfsP Wr e'y 0.2 system is subject to form drag, the previous
analogous behavior is altered. In one study (7), the
__c P v fs p(0.037) internal surfaces of tubes were artificially rough-
0.2
uV Re V ened so that the friction was increased. At
maximum roughness, the pressure drop was
(0.260X50 x 3600) (0.0291) (0.037 )
increased sixfold, but the heat-transfer coefficient
(9.85 x 10
5 02
) was increased only twofold. It has been universally
observed that, for systems exhibiting form drag,
r7=3.2Btu/hrft 2 °F
heat or mass transfer is not increased to as great an
The same calculation can be made with extent as momentum transfer as a consequence of
Equation 13.110, which was derived using the the form drag.
Colburn analogy. To explain this phenomenon, an element of
roughness as shown in Figure 13.9 will be exam-
0.037 ined. In the same manner as the shape shown in

CpVfsP
W Pr )
2/3 .

W ReV 0.2

_ cpv u p (0.037
337), 1 v-
h-
/

°
Wr.) k

'/v Pr /

For air at 900° F and 1 atm, N Pr = 0.69.

Boundary-layer x^ Separated
separation eddy
__ 2.3 3.2
~ 2 3
'
" 0.78 Figure 13.9. Eddy behavior behind a roughness
(0.69)
element.
C
/T= 4.1 1 Btu/hrft 2 F

This latter values is probably more nearly correct
than that obtained using the Reynolds analogy.
More rigorous analogies have been written for
the boundary layers in a manner similar to the
work of Martinelli described above. The differences
between the analogies are dependent upon the
choice of assumptions and the velocity-distribution
function chosen for the derivation. The reader may
refer to textbooks in fluid mechanics for the
development of these equations
Figure 13.4, the boundary layer separates and an
eddy forms behind the roughness element The
eddy circulates fluid within a very small finite part
of the free volume of the tube as a result of the
position of the roughness element. The local turbu-
lence intensity in the eddy is greater than in the
region nearby. The eddy transfers momentum from
the flowing stream with an increase in measured
stress. The intense circulation within The eddy does
not contribute proportionately to the bulk mixing.
(54). It should be remembered that the local eddy is

"fed" by the surrounding streams and in turn feeds

MASS AND HEAT TRANSFER the surrounding streams. Therefore, there is a small
LN SYSTEMS
increase in circulation throughout the stream
EXHIBITING FORM DRAG
because of the interchange between the eddy and
All the material on heat and mass transfer that has the surroundings. This small interchange is a partial
been presented chapter was written specifi-
in this contribution to the mixing phenomenon as under-
cally for smooth surfaces over which fluids moved stood in turbulent flow and can explain the small
 FUNDAMENTALS OF TRANSFER MECHANISMS 281

Sources
increase in heat- and mass-transfer coefficient. That
>„-(29)
jq - (35, 39) is, the local circulating eddy transfers more
//8-(16) momentum without proportional increase in bulk
CD /8-(52)
mixing in the main stream.
Several geometric shapes have been studied
experimentally, and a partial compilation of the
work is presented graphically in Figure 13.10.
Figure 13.10a reports experimental work on a
single cylinder with flow normal to the axis of the
cylinder. The /-factors for mass transfer and heat
transfer are plotted as a function of the Reynolds
number as defined in the figure. The heat- and
0.001 mass-transfer data are in excellent agreement. In a
.0 10 10' 10"
system that offers a projected area normal to flow,
*Re
(a)
both form drag and skin friction are present.
Goldstein (16) reports a procedure by which the
Y = //8 = T,g<lvt. 2 p where Z> 2 = diameter of cylinder
measured forces normal to a body can be sub-
Y = Q>/8 = FrfJASv,*p r yg c = tangential stress
= Sum of normal and tracted from the measured total forces on a body.
F,
y=j ("p,)«
tangential force The remaining force is the tangential force or skin
\pcpv,.j
friction. For the single cylinder, these tangential
S = projected area
Y = Js forces are combined with appropriate variables to
v„ = free-stream velocity

h =
form the friction factor, and this factor is plotted
"Re = (DiV/.plft) heat-transfer
coefficient as rV8, consistent with previous considerations in
ke ' = mass-transfer the Colburn analogy. This plot is in excellent agree-
coefficient ment with the data for jN and jq Finally, to .

Sources
describe the magnitude of the total force on the
j N (gases) -(57)
jq (gases)- (40) body, CD /8 is plotted and shown to be very large
y Q (liquids) — (41) compared to jq , jNl and f/8. The foregoing
C D /8 (gases) -(53)
indicates that for systems in which form drag
occurs, if the tangential force or skin friction can
be isolated, it will correlate with mass and heat
transfer, but when total force, as in CD /8, is used,
no correlation can be written. This analogous
behavior is consistent with all previous analogies in
0.01 which skin friction was the only means of momen-
tum transfer.
Similar data for the single sphere are plotted
0.001
in Figure 13.10/?, but the experimental data for f/8
are not available. In the case of the sphere for the
data presented, the agreement between jq and//v is

as good as it was for single cylinders.

Y = CD j% = FTg l4Sv,*p c where D = diameter of sphere
_. _I h \ /3
v„= free-stream velocity
Y r
\pcpVfJ FT = total force on the
Y=jJN -\
k
A
\v„f
(N )"
5
SphCre
= projected area to flow Example 13.11. The "Lister" bag is a
h = heat transfer coefficient common means of storage of potable water by
NRe = = mass-transfer army field forces. The bag is made of porous
(Dvf,p/fi) k/
coefficient canvas. A small amount of water diffuses through
the canvas and evaporates from the surface. The
Figure 13.10. Mass and momentum transfer with form
drag, (a) Heat, mass, and momentum transfer to single
evaporation of the water cools the surface of the
cylinders, (b) Heat, mass, and momentum transfer to single
bag and a temperature driving force is established.
spheres. From the source indicated in the numbered ci- Estimate the heat- and mass-transfer coefficients if
tation, with the permission of the copyright holder in each the bag is a 2.5-ft diameter sphere hanging in a
case. 0.5 miles/hrwindat90°F.
282 MOLECULAR AND TURBULENT TRANSPORT

Data (Appendix D): Air at 90° give empirical equations useful in predicting trans-
fer coefficients, as discussed in earlier sections.
p = 0.0183 centipoise
Parallel with the increased use of such coeffi-
2
k = 0.0140 Btu/hr ft (°F/ft) cients, attempts to explain them physically have
been made. Understanding of the physical mechan-
cP = 0.25Btu/lb°F
ism of heat and mass transfer in turbulent flow
_ (28.9)r(460 + 32)- 3 rests on an understanding of the mechanism of
P ~ = 0.0721 lb/ft
(359) 1(460 + 90)- turbulent flow itself. In the earlier part of this
2
chapter, transfer coefficients were defined using
^H,o-air = 1 ft /hr
the eddy -diffusivity model. Two other physical

/V P
Pr = —
cP p

k
=
(0.25X0.0183x2.42)
(0.0140)
= 0.79
models for transfer will be considered here. These
are the widely applied film theory and the more
recently developed penetration theory.

NSc = —
pQ
p
=
(0.0183x2.42)
(0.0721)(1)
=0.613
Film Theories

Solution. The Reynolds number can be cal-

The most rudimentary film theory states that all

culated the resistance to transfer in a turbulently flowing

fluid is concentrated in a stagnant film next to the
Dv fs p (2.5)(0.5 x 5280)(0.0721) wall or stationary boundary of the fluid. According
/v Re =
p (0.0183x2.42) to this model, all the driving force (or concentra-
tion gradient) acts across the stagnant film, and
/V Re = 10,750
also the concentration in the bulk of the fluid is

Figure 13.106 is entered at /V Re = 10,750 and jN constant, because it is highly turbulent motion.
in

and jq can be read This profile is shown in Figure 13.1 1a. The transfer
in the stagnant film would be by molecular trans-
jN = 0.007
port, and the general equation for transfer would
jq = 0.010 be

From the definition of jN

= — 2/3 = 0.007
*= — (r,-n (13.111)

//V (/V Sc )

Vfs where x f is the thickness of the stagnant film, is i//

(0.007X0.5x5280) the flux and 5 is the diffusivity. Since the film

k,' = thickness cannot be measured directly, it is com-
2/3
(0.613)
bined with the diffusivity to give a transfer coeffi-
2
k c = 25.6
'
lb moles/hr ft (lb mole/ft3 ) cient ), where
(

From the definition of jq t

. =-6
<
(13.112)
Xf
2'3 = 0.01
(N Pr )

PCpVfs and Equation 13.111 becomes

(0.01)(0.0721)(0.25)(0.5 x 5280) = -<5(r 1 -f) (13.113)

h =
2/3
(A/ Pr ) A more satisfactory concept of the film states that
h = 0.557 Btu/hr ft
2
°F it not stationary but is in laminar flow. Transfer
is

in laminar flow would still be by molecular trans-

port, and the gradient would still be only across
OTHER MobELS FOR TURBULENT TRANSFER
the film, as shown in Figure 13.116. The resulting
Transfer coefficients have been in use for many equations would be identical to Equations 13.111
years in the design of heat-transfer and mass trans- and 13.113.
fer equipment. In most cases, they have been con- Neither the stagnant film nor the laminar-
sidered to be empirical coefficients that could be flowing film is consistent with what is known
determined from experimental data. In certain about turbulent flow, as outlined in Chapter 12.
cases, the experimental data could be correlated to The most accurate observations show no stagnant
 FUNDAMENTALS OF TRANSFER MECHANISMS 283

1*T
r„sr rft
=r

Turbulent Turbulent
core core
Buffer layer

Stagnant film
_ Laminar
flowing film Laminar
c
f h- '/ h- sublayer

(a) —^x (b) (O

Figure 13.11. Concentration gradients in film theory.

film near the wall. Although evidence supports the and for mass transfer
existence of a laminar sublayer next to the wall,
this not equivalent to a laminar-flowing film,
is kc' =
2_ (13.116)
since only part of concentration gradient is in the Xf
laminar sublayer and all of it is postulated to occur
across the laminar film. These inconsistencies with
N± _
A
.
— kc (c a i — Cg) (13.117)
observed data have led to a redefinition of the film
as a fictitious thickness of laminar-flowing fluid The fictitious-film thicknesses for heat, mass, and
next to the boundary that offers the same resis- momentum at equal flow rates not equal
are
tance to transfer as actually exists in the entire except under the limiting conditions of the
turbulently flowing fluid. That is, all resistance to Reynolds analogy.
transfer is concentrated in a fictitious film in which Because of its apparent lack of adequate
transfer is by molecular transport only. The thick- physical basis, the film theory is losing favor, even
ness of this fictitious film is shown with a typical though it has proved very useful in interpreting and
concentration gradient for turbulent flow in Figure correlating turbulent transfer data for equipment
13.11c. The fictitious film extends beyond the design. The term film coefficient for heat or mass
laminar sublayer to allow for the change in concen- transfer is still widely used, but it is less desirable
tration in the buffer layer and turbulent core. The than the terms transfer coefficient or surface
centerline composition T is somewhat greater coefficient.
than the average composition r, as expected.
Equations 13.111, 13.112, and 13.113 may be Penetration Theory
used to define transfer across the fictitious film,
The penetration theory was originally applied by
where x f is now the thickness of the fictitious film.
Higbie (18a) to analyze the liquid phase in gas
Obviously, x f can never be measured, since it does
absorption where the liquid could be assumed to
not exist. For this and other reasons, this most
be in laminar flow or stationary. It was more
sophisticated version of film theory is of little use
recently applied to turbulent flow by Danckwerts
in understanding the mechanism of turbulent trans-
(9) and Hanratty (18).
fer. Equations 13.112 and 13.113 may be written
Higbie considered mass to be transferred in
specifically for heat transfer
the liquid by unsteady-state molecular transport.
This concept resulted in an equation for the mass
h a flux at a point on the surface of a liquid exposed
13.114)
Cpp Xf to an absorbent gas

T) 113.115) \c a \ 13.118)
r-* (ri
A nd
284 MOLECULAR AND TURBULENT TRANSPORT

where More recently, Pinczewski and Sideman (45)

have presented a model that visualizes the fluid
6 =time of contact of the liquid with the
near the wall as a mosaic of patches of periodically
gas
replaced boundary layers. This leads to an analysis
c a1 = concentration in the liquid at the gas
that is partially steady state and partially unsteady
interface
c a = average bulk concentration in the liquid state. The model is able to predict heat and mass
transfer rates from mechanical data alone.
fluid A
This relationship has proved useful in analyzing summary of various penetration models is found in

absorption data from wetted-wall columns. Reference 58.

Danckwerts applied this unsteady-state
concept to absorption in a turbulent liquid. He Comparison of Models
postulated random the liquid at
surface renewal in The general equations for the three models con-
the gas-liquid boundary. In this model, there is no sidered in this chapter are
laminar sublayer at the boundary. The liquid
the turbulent core migrate Film theory:
eddies that originate in

to the gas-liquid boundary where they are exposed

to the gas for a short period before they are *=-(r,-n (13.120)
displaced by other eddies arriving at the surface.
During the short exposure period (8) of an eddy, it Eddy-diffusivity model:
absorbs gas at a rate according to Equation 13.1 18.
It then is displaced from the surface and returns to 4(5 + £)
the bulk of the liquid, where its absorbed gas is *=
Dy
(n-n (13.121)

distributed by turbulence. Fresh surface is contin-

ually created by the random motion of the eddies. Penetration theory (random surface renewal):
The mean rate of production of fresh surface is
constant for a given degree of turbulence and equal
^ = -v/ 67(r -f)1
(13.122)

to s. The surface-renewal factor (s) is defined In each case it is possible to define a transfer
statistically and cannot be determined directly at coefficient.
present The rate of absorption for the turbulent
liquid with random surface renewal is Film theory:

Na (13.112)
(13.119) Xf

Eddy-diffusivity model:
1
where s the surface renewal factor, (1/6), sec"
is .

Hanratty extended this model and showed that it 4(6 + E)

(13.123)
was consistent with experimental data for mass and Dy
momentum transfer. To do this it is necessary to
Penetration theory:
assume the form of the statistical probability func-
eddy residence time at
tion for the distribution of £ =V5s 13.124)
the boundary. Microscopic examination of the
motion of dust particles in a turbulent-flowing In the film theory, the fictitious-film thickness (x f )

liquid very the boundary (14) indicates

near cannot be measured or predicted directly. Al-
random motion that may be interpreted as large though the film theory has proved useful, it con-
eddies. This observation tends to cast doubt on the tributes to the understanding of turbulent
little

existence of a laminar sublayer and to support a transport. As

turbulent exchange at the boundary.
Toor and Marchello (33,60) developed a
"film-penetration" model, combining features of
both earlier models. At low Schmidt numbers, the
steady-state film model seems appropriate, whereas
at high Schmidt numbers, the unsteady-state
surface-renewal model more adequately describes
the situation.
indicated by Equation 13.112, the
film theory predicts that the transfer coefficient
willvary as the first power of the diffusivity, 5.
Experimental evidence does not support this con-
clusion.
The eddy-diffusivity model considers the
random motion of eddies and attempts to define
the effect of this gross eddy motion upon the
transport in the system. At the present time, the
 FUNDAMENTALS OF TRANSFER MECHANISMS 285

prediction of either point or average eddy diffusivi- heat transfer, produce results of the form
ties is uncertain.
In penetration theory with random surface
renewal, rudimentary statistical methods have been
applied to develop expressions for predicting trans-
where
N Nu $
K I)
13.125)

fer caused by random eddy motion at a boundary.

In this case, the behavior of a single eddy is exam- A/ Pe = Peclet number, (Dvpc P lk)
ined and generalized by statistical means. The LID = length-diameter ratio
mathematical treatment is based upon unsteady- 3> = unspecified function
state transfer, in contrast to the steady state
assumed in the eddy-diffusivity model. As indi- With simplifying assumption, several of the
cated by Equation 13.124, the penetration theory analyses reduce to

indicates that the transfer coefficient is propor-

tional to the square root of the diffusivity. The NN u = constant (A/ Pe )
a
I -j ( 1 3. 1 26)
Toor— Marchello model yields an exponent on dif-
fusivity of from 0.5 to 1.0. Experimental data
In gases the viscosity increases with increase in
show that the exponent varies from 0.5 to 0.75 for
temperature, whereas in liquids the viscosity de-
various physical situations.
creases with increase in temperature. The shear
stress in a tube is linear with radius with the
maximum at the walls; this stress pattern is in-
HEAT AND MASS TRANSFER IN LAMINAR FLOW dependent of the transfer mechanism. At any
radial position, the stress is constant. If a change in
In isothermal laminar flow of fluids in tubes, it is
viscosity occurs at that the velocity
position,
possible develop expressions for the velocity
to
gradient must change to maintain constant stress.
profile analytically, as shown in Chapters 9 and 10.
In a heated or cooled tube with laminar flow, a
In the absence of turbulence, heat and mass are
temperature gradient can be expected between the
transferred only by molecular transport. As a re-
wall and the center of the tube. The thermal
sult, heat and mass transfer in laminar flow has
gradient tends to distort the velocity profile in a
been subject to extensive rigorous mathematical
manner similar to that shown in Figure 13.12, for a
analysis for various physical situations, but these
hot wall in contact with a cold fluid.
analyses do not often correspond to practical situa-
The rigorous analysis of heat transfer in
tions actually encountered. Variations in tempera-
laminar flow is complicated further by the estab-
ture or concentration create differences in density lishment of density gradients in the fluid as a result
and viscosity, which alter the flow patterns and of thermal expansion. The density gradients estab-
hence the rate of transfer. lish natural convection currents in the fluid, with
Transfer coefficients for laminar flow are
typically less than for turbulent flow, so it is pref-
erable to operate in the turbulent regime where
possible. But there are situations where the laminar
regime is unavoidable, such as with very viscous
liquids or with gases in small diameter tubes.
For heat transfer in laminar flow, numerous
rigorous solutions have been developed for various
boundary conditions. Analyses have been made for
the parabolic velocity distribution and constant
wall temperature (17), constant wall heat flux, and
with the wall temperature varying linearly orsinu-
soidally with position in the flow direction (55).
Other analyses were written for heat transfer dur-
ing the development of the boundary layer within
a tube at constant wall temperature, constant heat
input, and constant temperature difference (21).
All the analyses, which involve the solution of Figure 13.12. Velocity profiles with a
partial differential equations for two-dimensional thermal gradient.
286 MOLECULAR AND TURBULENT TRANSPORT

consequent crosscurrent flow in horizontal ducts, derivations that are written without accounting for
or locally accelerated flow in vertical ducts. These natural convection such as Equation 13.126. The
crosscurrents contribute to heat and momentum distortion of the velocity profile is recognized by
transfer in the same manner as the crosscurrents in use of the viscosity ratio (p.ln-\ ).
turbulence. At low values of the Reynolds number For convenience, Equation 13.127 may be
(that is, below A/ Re =2100), the effect can be written in terms of the /-factor for heat transfer by
considerable, and, in fact, the natural convection first substituting /V Re /V Pr for /V Pe so that Equation
effect detected until A/ Re = 8000, at which time
is 13.127 becomes
natural convection currents are overshadowed by
1 "
the crosscurrents of turbulent flow. 1/3 1/3
The distortion of the velocity profile and the
(A/ Nu )a.=1.86(/V Re ) (/Vp r ) (^)
(^)
effect of convection currents can be
natural
(13.128)
expected to affect the rate of heat transfer. Two
empirical equations are presented here with a state- substitution introduces the viscosity into two
The
ment of range of usefulness. The first equation dimensionless groups. This does not necessarily
(59) is limited to horizontal tubes one inch or less imply that the Nusselt number is dependent upon
in diameter and temperature driving forces less
viscosity. Note that viscosity cancels from A/ R e and
than 100°F. when both have the same exponent. Equation
A/ Pr

1/3 °- 14 13.128 may be divided by /V Re /V Pr to give

ID \ / At \

WnuU =1.86(/V Pe )
1/3
(-^j (-) (Nt4u) am

(13.127) W Re /V Pr )

1/3 14
ID \ / u \
-

where 2/3 2/3

1.86(/V Re )- (/Vp r )-
(-^) (^)
(N Nu am = Nusselt number (h am D,/k)
) (13.129)
h = heat-transfer coefficient for
a m a
tube of length This heat- L.
Since /V St = /V Nu /(/V Re /V Pr ), Equation 13.129 may
transfer coefficient is based be rearranged to
upon the equation qIA = 4 1/3
ham (AT) am 2/3
(^)°' =1.86(/V Re )- 2/3
(A/st)a m (/Vp r )

(AT) am = arithmetic-mean-temperature (^)

driving force. The general sub- (13.130)
ject of mean driving force will
be discussed in Chapter 15 2/3 14
= viscosity of the fluid at the
and since (/V St ) am (/V Pr ) Wm) "

=/„.
Hi
wall temperature (7"i
= 1.86(A/ Re )- 2/3 ^-j (13.131)
k, c P p, h = fluid properties evaluated at
,
/Q

the mean bulk temperature (7")

A/p e = Peclet number (D-\ vpc P lk)
The fluid variables in
evaluated at the arithmetic-mean bulk temperature
(7~) with the exception of which is evaluated at
p.
the temperature of the tube wall. However, Equa-
tion 13.127 does not acknowledge natural convec-
tion, and the equation is limited to small tubes
(£>1 < 1Jn.) and small temperature driving force
({T, -D<100°F). At a small driving force in
small tubes, the natural convection phenomenon is
limited in magnitude. Although the natural convec-
tion phenomenon is not recognized with appro-
priate variables, the constant 1.86 is somewhat
higher than the constants found in the theoretical
For convenience Equation 13.131 is plotted
in use,
in Appendix C-5, which jq is plotted as a func-
in
Equation 13.127 are tion of A7 Re and the parameter L/D. Although the
actual heat transfer is not dependent upon a
: , Reynolds number, the equation form of Equation
13.131 is desirable because the Reynolds number is
the criterion for designation of laminar and turbu-
lent flow. The chart in Appendix C-5 is recom-
mended for all heat-transfer calculations for cylin-
drical tubes because the chart applies to both
laminar and turbulent flow.
In the range between A/ R e = 2,100 and A/ R e =
10,000, the mechanism of heat and mass transfer is
uncertain. This transition region may produce un-
stable flow patterns, and this transition region may
 FUNDAMENTALS OF TRANSFER MECHANISMS 287

depend upon the frequency of disturbances. Fluc- Solution. First, the Reynolds number should
tuations in pressure drop have been observed, sug- be calculated. The data for this problem can be
gesting that the flow is fluctuating between laminar found in Appendix D.
and turbulent. The smooth curves representing the
transition region may not describe actual perform- ju = 0.58 centipoise
ance. If possible, operation of equipment
the in
p = 61.5 lb/ft
3

transition region should be avoided because of un-

certainties, not the least of which is the frequency
of the disturbance, and possible instability. In
N Re = Dvp/n = —(3/8 x 1/12X0.3X61.5)
-

(0.58x0.000672)
using Appendix C-5 for any flow regime, it should
= 1490
be remembered that the curves are based upon
empirical correlations of data that showed some At A/ Re = 1490, the flow is laminar and Equation
scatter around the curves. Use of the correlations 13.131 or Appendix C-5 applies. D,/L= (3/8 x
should take into account uncertainties in the 1/12)/3.21 = 0.01, L/Dy = 100. If Equation 13.131
data. is used
A correlation that takes into account natural
/3
convection in laminar flow in horizontal tubes in-
iD \ 1

2
/,= 1.86(/V Re )- /3(_!)
cludes the Grashof number, a group characterizing
natural convection (11): 2/3
\
(a011 " 3
f 1

(/VnuU=1.75 (-)
/M°- 14 r TT/Vpe^! '•"'"W
* Ml 4L
1. .86(^(0.216)

0.12(
7r/Vpefli/V G r'
M Wp
1/3a/
—
0.36\0.88 1/3

jq = 0.00309

;13.132) This value is in agreement with the chart of Appen-

where
dix C-5 for A/ Re = 1490 and UD\ = 100. Then
2 0.1 4
= Grashof number (D^ 3 p 2 g(3 AT/p ), /JU x
/V Gr
dimensionless
Jq = W S t) am Wp = 0.00309
2
g = gravitational acceleration = 32.2 ft/sec
/3 = volumetric coefficient of thermal ex- Numerical values can be substituted into A/ St , /V Pr ,

_1
pansion of the fluid (°F) Hi//u. Mi =0.23 centipoise, cp = 1 .0 Btu/lb °F,
2
A 7"= the temperature difference between A- = 0.370 Btu/hrft (°F/ft)
the wall and the bulk of the fluid
(7-,-r),°F h __. h
(/Vst),
cpvp (1)(0.3x 3600X61.5)
The general significance of the Grashof number
will be discussed in a following section.
(1)(0.58x2.42)
N 9r = J—
cP fx
Mass transfer in laminar flow can also be = = 3.8
analyzed analytically (2). If the components of the k 0.370
mixture have different densities, then variations in
bulk fluid density because of concentration gradi- /AH\ 0-23
= 0.397
ents will induce natural convection, thereby in- U / 0.58
creasing the rate of mass transfer over that ex-
Then
pected by diffusion alone. Too few data are
available to develop an empirical correlation.
2/3 0.14 _
(3.8) (0.397) 0.00309
1X0.3x3600X61.5)
Example 13.12. Calculate heat-transfer co-
efficient for a |-in. I.D. horizontal tube in which _ 0.00309(1X0.3 x 3600X61.5)
water flows at 0.3 ft/sec. The mean bulk tempera-
ham ~
(2.44) (0.88)
ture of the water is 125°F. The tube wall is main-
tained at 225° F. The tube is 3.21 ft long. h am =95.7 Btu/hrft 2 °F
288 MOLECULAR AND TURBULENT TRANSPORT

HEAT TRANSFER IN NATURAL CONVECTION to the element of fluid under the influence of the
large driving force, so that the fluid expands, with
If temperature gradient exists in a fluid, the
a consequent decrease in density. A natural convec-
density of the fluid will also vary within the fluid. tion current is created by the air of reduced den-

This variation in density induces a flow of fluid
called natural or free convection. A review of free
convection is given in Reference 36.
A common example of natural convection is
the steam-heated radiator. The air near the radia-
tor's surface is heated, thereby reducing its density
so that it is displaced upward by the cooler, more
dense, surrounding air. A steady-state flow pattern
is established, with cooler air flowing into the
bottom of the radiator and warmer air flowing out
of the top. The warmed air is eventually cooled in
the room, so it settles and flows back to the base
of the radiator, thereby completing the flow cycle.
The relatively low flow velocities and heat-
transfer rates of free convection contrast with the
higher velocities and transfer rates of forced con-
vection where a pump or fan maintains fluid flow,
as discussed earlier in this chapter. In cases of
relatively low flow velocity, as in laminar flow of
the previous section, both free and forced convec-
tion may play a role. Free convection occurs
wherever a temperature gradient exists in a fluid,
sity,with the gas motion upward along the surface.
An increase in velocity for a given element of fluid
means that the element sweeps out a greater
volume per unit time, which tends to bring new air
into the element in diagonal path from the outside
cold-air supply. The process is shown qualitatively
in Figure 13.13. All phenomena described here
take place within a momentum and thermal bound-
ary layer established near the wall, with conse-
quent velocity and thermal gradients developing in
the process. At the start, the velocity pattern of
the boundary layer is characteristic of laminar
flow, but with progress up the surface, turbulent
flow may occur. After some progress up the sur-
face, the original element that is diluted somewhat
with additional "outside" air reaches a temperature
closer to the surface of the wall, and the thermal
driving force is diminished. Further qualitative
analysis of the boundary-layer behavior can be
made by examining the effect of the various trans-
port mechanisms on the rate of acceleration of the
fluid. The always positive, and the
acceleration is

so it is not easily controlled. For example, in the velocity of the fluid always increasing, but the
is

air space between the two panes of a double-glazed rate of increase of acceleration may be positive or
window, a natural convection pattern is set up, negative. The effects of the mechanisms, each
thereby increasing heat transfer and reducing the taken alone, is given below.
insulating effectiveness of the double glazing, par-
1 . The decrease in thermal driving force tends
ticularly the separation is significant.
if
to decrease the rate of heat transfer, rate of expan-
The analysis of natural convection involves a
consideration of two- or three-dimensional fluid
av< away from th e
flow and heat transfer (47). A simple case involves
heat transfer from a heated vertical surface in con-
Hot
tact with a large mass of air at a lower temperature.
wall"
The air near the surface is heated, so that its
density decreases relative to the cooler air nearby.
Higher velocity
The buoyant heated air rises along the surface, to
be replaced by heavier cooler air from a region Additional air moves
away from the surface. toward the wall

At steady state, the heat added to the air at

the hot wall must be removed at some other re-
mote part of the cycle, or the entire air mass would
eventually reach the wall temperature and heat
transfer would cease. The history of an element of
fluid may be traced during passage along the hot
wall. The particular element starts at the bottom of
the hot wall, where the motion is slow. The tem-
perature of the air element is low compared to the
wall, since no heat has yet been transferred to the
element during this cycle. Consequently, a large Figure 13.13. Motion of an element of gas along a

temperature driving force exists. Heat is transferred vertical heated surface.

 FUNDAM1 NTALS 01 TRANSI1 R MtCHANISMS 289

sion, and consequently the rate of increase of

acceleration.
2. The dilution of the boundary layer with
new air from the outer region tends to dilute the
element of air with colder air and thus maintain
the thermal driving force, with consequent increase
in acceleration of the fluid.

3. Acceleration of the fluid tends to increase

the flow of air from the outer regions.
4. The increase in velocity tends to increase
the rate of heat transfer, rate of increase of vol-
ume, and consequently the rate of increase of
acceleration. This is especially true if the boundary
layer is turbulent.
5. The increase in velocity tends to increase
the wall stress, which tends to decrease the rate of 2 4 6 8 10 12
Distance from wall, mm
increase of acceleration of the fluid. Again, this is
Figure 13.14. Local temperature and local velocity in
especially true if the boundary layer is turbulent.
the natural-convection boundary layer at a heated
6. The increase in temperature of a gas results
verticalplate. Lower curve of each pair, 1 cm from
in increased viscosity of gases, or decreased vis-
bottom; upper, 24 cm (36). (With permission of
cosity of liquids, with the usual effects on wall McGraw-Hill Book Co., Copyright© 1954.)
stress and acceleration of the fluid, especially if the
boundary layer is laminar.
7. At all times, the acceleration of the gas
where
results from the gravitational effect upon the ex- NHu =hL/k
panded gas. The acceleration may be due to change
in speed or change in direction. N Pr = Cplji/k
L 3p 2 gi5(-AT)
H G r= ~
The foregoing shows that common natural
list

convection is an extremely complex process for

A/ G r is a new dimensionless group, called the
mathematical analyst.
Grashof number, and represents the ratio of buoy-
The net result of the actions of these various
ant forces to viscous forces. In /V Gr ,/3 is the co-
mechanisms is shown in Figure 13.14, where local
efficient of thermal expansion and is the fractional
velocitiesand temperatures are plotted at two posi-
volume increase with increasing temperature, so
tions on the wall. The temperature gradient is
the combination pA7 is the fractional decrease in
steepest near the bottom of the wall, and it de-
density with increase in temperature.
creases farther up the wall. The average velocity is
The Grashof number characterizes free con-
low near the bottom, and it increases as the fluid
vection just as the Reynolds number characterizes
moves up the wall.
forced convection. While the analysis leading to
The simplest analysis of natural convection at
Equation 13.133 includes some faulty assump-
a heated vertical wall assumes that the wall is at a
tions, it does introduce the Grashof number and
uniform temperature and that the velocity and
results in useful empirical correlations (13).
temperature of the gas are functions only of the
For heat transfer from a vertical surface
distance from the wall. (At any given distance from
the wall, the temperature and velocity do not vary /V Nu = 0.555(/V Gr /V Pr 1/4 when A/ Gr A/ Pr <10 9
)

as the fluid moves up the wall.) This ignores the

acceleration of the fluid as it moves up the wall, (13.134)
but it leads to relatively simple analysis. The result- /V Nu = 0.0210(/V Gr /V Pr 2/5 when /V Gr /V Pr >10 9
)

ing equation, while not an accurate description of

actual heat transfer by natural convection, does (13.135)
suggest that (24)
Figure 13.15 shows the fit of these equations
with experimental data. Expressions are available
N Nu = f(N Gr ,N Pr ) (13.133) for other geometries, such as horizontal flat sur-
290 MOLECULAR AND TURBULENT TRANSPORT

10-

N Nu = 0.021 0( NG ^Pr) 2'5

o Vertical cv inder L

o Vertical plane

a 9.01 in. plate

v 2.99 in. pi ate

oj?
103

Transition st±
Or a
region

BJ

< D
La o /Cr
10 :
80
60
v^5
40

^S Nui =,
0.55(-vGri>r) V.
20
Q^ 1

10
10 5 10" 10 9 10 1
10' 10' 5
V
'

"Gr Pr

Figure 13.15. Correlation of natural-convection heat-transfer data for vertical

surfaces (13). (Courtesy of NACA.)

faces and horizontal cylinders (36). The fluid in Solution. The air properties are evaluated at
free convection is often air, and simplified equa- the average of the plate temperature and the room
tions are available for this case (36). For example, air temperature. At 58°C,
for vertical surfaces with air near room tempera-
ture 90 + 25
'
— _
— 3o L>
ave

= 10 4 9
Laminar flow: (/V Gr )(/V Pr ) to 10
14 M = 0.020 centipoise = 2 x 10~ 5 kg/m s
h = 0.29(-AT/LV (13.136)
k = 0.016 Btu/hr 2
ft (°F/ft)
Turbulent flow: (/V Gr )(/V Pr )
= 10 9 to 10 1 2

= 0.19(-A7") 1/3
= 0.027 J/s m 2
(°C/m)
/7 (13.137)
cp = 0.25Btu/lb°F
where h is in Btu/hr ft
2
°F, and L in ft.
°F, AT in
For horizontal cylinders, equivalent expres-
= 1045J/kg°C
sions are
The density of air can be calculated from the
perfect gas law:
Laminar flow: N Gr N Pr = 10 3
to 10 9

/?=0.27(-A77Z.) 1/4 (13.138) 29.9 g/mol 273 K

Turbulent flow: /V Gr /V Pr = 10 9
to 10 1 2 22.4m 3 /kmol 273 x 58 K
1/3 = 1.10 kg/m 3
/7=0.18(-A7") (13.139)

The volumetric coefficient of thermal expansion

The appropriate units are as given above, and L is
can be computed from the perfect -gas law. (3 is the
the diameter of the cylinder.
fractional volume increase per degree of tempera-
ture rise, or fractional density decrease per degree
Example 13.13. Calculate the rate of heat of temperature rise, so
transfer from a vertical flat plate 1 m high and 1 m

wide. The plate at a temperature of 90°C and the ]_dp_

is
= -
room air is at 25°C. pdT
 FUNDAMENTALS OF TRANSFER MECHANISMS 291

For perfect gases

=
P
RT
and
1 dp 1

P-- pdT T
where T is absolute temperature.

When T = 33 1 K, = 1 /33 1 = 0.00304

(a) (b)
2
Then, with # = 9.8 m/s
Figure 13.16. Condensate growth over ver-
(1)
3
(1.10)
2
(9.8)(0.00304)(90 - 25) tical surfaces and horizontal tubes, (a) Ver-
WG r
=
2 tical wall or tube, (b) Horizontal tube.
(2x10~ 5 )
Tv = saturation temperature of condensing
= 5.85x 10 9 vapor. T2 = temperature of the solid surface.

(1045)(2 x 10" 5
) presence of water vapor), then it is necessary to
/v Pr = = 0.774
0.027 consider both heat transfer and mass transfer of
the condensable vapor through the noncondensable
9
Since /V Gr /V Pr = 4.53 x 10 , Equation 13.136 gas. Only the case of pure vapors is considered
should be used: here.

/V Nu = 0.021(4.53 x10 9 )
2/5 = 152 Condensers are widely used in the chemical
industry. For example, the condenser on a distilla-
and tioncolumn converts the vapor from the top of the
152x0.027 column into liquid distillate and reflux. Power
h =
nii
= 4.13 J/s m 20o°C plants use water-cooled condensers to convert ex-
1
haust steam from the turbine into liquid. The
Alternatively, Equation 13.136 could be used typical home air-conditioning unit uses an air-
cooled condenser to liquefy the refrigerant.
h = 0.19[(90 - 25X1.8)]
1 /3 = 0.89 Btu/hr ft
2
°F
The usual design for condensers employs
= 5.05 J/s m2 °C tubes, for simplicity of fabrication, with the con-
densate formed inside or outside of the tubes. A
Using the earlier value,
cold fluid on the opposite side of the tube wall
q= hAAT= (5.05) ( 1 ) (90 - 25) = 328 J/s serves to remove the latent heat of condensation.
In film-type condensation, the condensate that
forms at a point flows downward over the remain-
HEAT TRANSFER FROM CONDENSING VAPORS
ing length of film-covered surface until a discon-
In order to condense a pure vapor, it is necessary tinuity is reached so that a pendant drop can form
to remove its latent heat of vaporization; in cer- and break off. Because condensate forms over the
tain cases, it is also necessary to subcool the result- entire surface, the thickness of the film will in-
ant liquid condensate. This heat removal is usually crease and will be a maximum at the point of exit
accomplished by a cooled solid surface on which (see Figure 13.16). As might be expected, the film
the vapor condenses. If the liquid condensate may flow in the laminar regime or in the turbulent
"wets" the surface, it forms a continuous liquid regime, depending upon the rate of condensation,
film through which heat must be transferred. If the length of path of the growing condensate film,
condensate does not "wet" the surface, droplets fluid properties and geometry. The film velocity
are formed instead of a film. In either case, the will be zero at the solid surface and a maximum at
condensate flows down the surface under the in- the condensate-vapor interface. For heat transfer
fluence of gravity. to occur, a temperature gradient must exist
Condensation of pure vapors can be treated through the flowing film of condensate, so the
by considering the heat transfer needed to con- vapor-liquid surface of the condensate is usually
dense and subcool the condensate. If a noncon- considered to be at the vapor temperature with the
densable gas is also present (for example, air in the liquid surface in thermodynamic equilibrium
292 MOLECULAR AND TURBULENT TRANSPORT

with the vapor. The condensate in contact with the length increases and the temperature difference
solid surface is considered to be at the temperature between the vapor and surface increases. The rea-
of that surface. The driving force for the down- son for this effect is that the film thickness in-
ward flow of condensate is the gravitational field creases, with length and temperature driving force,
of the earth. offering more resistance to heat transfer. Although
the analysis was carried out for a vertical plane
Film-Type Condensation on Vertical Surfaces surface, the result is applicable to condensation
inside or outside vertical tubes because the conden-
Nusselt (43) derived the basic equation for film-
sate film is thin compared to typical tube diam-
type condensation on a vertical plate. This deriva-
eters.
tion will not be given here, but is readily available
Nusselt's equation has been checked many
(22, 256-266, 19, pp. 661-680). Nusselt
pp.
times and has been found to give fair agreement
made a number of assumptions in his analysis that
with experimental values. The assumption in the
are worthy of mention:
analysis that is most suspect is the one implying
1. Pure vapor is at its saturation temperature. laminar flow behavior of the condensate film. Al-
2. The condensate film flows in laminar flow, though the film may start in laminar flow, vapor
and heat is transferred through the film by conduc- shear will cause ripples; as the film thickens near
tion. the bottom of the surface, something like turbu-
3. The temperature gradient through the film lent flow can appear. Because of these turbulent
is linear. effects and the effect of vapor shear, most experi-
4. The temperature of the condensing surface mental values for h are higher than those predicted
is constant. from Equation 13.140; McAdams (38) recom-
5. The physical properties of the condensate mends the values of h from Equation 13.140 be
are constant and evaluated at a mean film tempera- increased 20 percent. Thus, the recommended
ture. equation is
6. Negligible vapor shear exists at the inter- 4
3 1 '

face.
*, P/(P/-P,> A/tol
6=1.13
PlUTy
The resulting equation for the length of the
plate is :13.142)

1/4 mean condensing heat

k,
3
p,(p,- p v ) AH v g Calculation of the
h = 0.943 transfer coefficient from Equation 13.142 is direct
p,L(Tv -T 2 )
if L is known, but is not straightforward if L is to
113.140) be determined for a specified heat load. For this
where case, let A = mass flow of condensate for unit
length of drainage perimeter. For a vertical tube
h = mean heat transfer coefficient over
the tube length w w
pi, p v = densities of liquid and vapor
A= (13.143)
PT ttD
g = gravitational acceleration
k/ = liquid thermal conductivity The heat load per tube is

Pi = liquid viscosity
AH V = latent heat of vaporization Q = AH v w = nhDL(Tv - T2 ) (13.144)

Tv = vapor saturation temperature or

Ty = surface temperature AH V FnrD h
L = tube length 13.145)
UTy-T2 ) w A
In Equation 13.140, liquid properties are
Substitution of Equation 13.145 in Equation
evaluated at a mean film temperature 7> (12),
13.142 gives
where
3 1 /3
Tf=Tv -3/MTy-T2 (13.141
k, p,(p,- p v )g
)
h= 1.18 13.146)
p,A
From Equation 13.140 it is seen that the mean
heat-transfer coefficient decreases as the tube For horizontal tubes Pj= L and A = wIL.
 FUNDAMENTALS OF TRANSFER MECHANISMS 293

Film- Type Condensation on Horizontal Surfaces and there is a reduction of the overall average
Nusselt (44) also analyzed condensation on a hori- performance. This reduction is due to the thick-
zontal tube and for the same set of assumptions ar- ening of the film on the lower tubes as shown in
rived at Figure 13.17.
It has been found that for a bundle of hori-
2
ki Pi(pi pv ) AH v g 1 I zontal tubes splashing of condensate as it flows
/? = 0.725 from tube to tube causes the Nusselt assumptions
P,D 2 (TV -T2 )
to be invalid, and it is suggested (22) that A is

!13.147) more nearly inversely proportional to A/ 2/3 than to

N. Therefore, Equation 13.148 becomes
Banks of Horizontal Tubes
The results for a single horizontal tube can be
[k, 3 p,(p,- p v AH v g 1 I

extended to the case of a bank of N vertical tubes

)

/? = 0.725
where the total condensate flows smoothly from H,D 2 N
2/3
(T V -T2 )
J

one to another without splashing. In this case 13.150)

Equation 13.147 is modified by replacing D 2 with
ND 2 where
, N is the number of tubes in a vertical

column.
Example 13.14. A condenser
is fitted with

1 -in. O.D. tubes 3 the condenser is

ft long. If
2
ki Pi(Pi - Pv) &H v g 1 I

mounted horizontally, the condensate will contact

h = 0.725 13.148)
p,ND 2 (Tv -T2 ) five tubes in its descent. If the tube surface tem-
perature is 150°F and steam is condensing at
1 atm, recommend whether to mount the conden-
Condensate properties are again evaluated at 7>.
ser (a) vertically or (b) horizontally.
This equation gives a mean heat transfer coefficient
over N tubes. Thus, the mean condensing coef-
ficient for the entire row of tubes is related toN Solution. The liquid film properties may be
the condensing coefficient for the top tube by
evaluated at the average film temperature, 7>
(Equat. 13.141) = 165°F. At this temperature, for
hN = h :
N -1 I
;13.149) liquid water:

H = 0.36 cp = 0.80 lb/ft hr

k = 0.383 Btu/hr ft 2 (°F/ft)
3
P/= 60.6 lb/ft (steam tables)
pv = 0.014 lb/ft 3 vapor at 21 2° F, 1 atm
AH V = 970 Btu/lb (steam tables)
cP = 1 Btu/lb °F

(a) For vertical mounting, Equation 13.140 is

appropriate:
1 /4
r (60.6X32.2 x3600 2 )(0.383) 3 (60.06 -0.014)970
h = 0.943
(3)(0.87)(212- 150)

= 796 Btu/hr °F 2
ft

(b) For horizontal mounting, Equation 13.150

applies with the indicated corrections

IA
2
(60.6X32.2 x 3600 )(0.383) 3 (60.6 - 0.014)970 , 1

h = 0.725
(0.87X5) (tM(212- 150)
Figure 13.17. Condensate flowing
= 1147 Btu/hr °F 2
down N = 3 rows of tubes. ft
294 MOLECULAR AND TURBULENT TRANSPORT

Hence, horizontal mounting is preferred in this pressure and the outside surface temperature is
case. Shorter tubes would swing the advantage maintained at 90° F. Determine the condensation
toward the vertical tubes. rate.

Solution. This solution requires the calcula-

Turbulent Condensation
tion of a condensing heat transfer coefficient and
Equation 13.140 was derived for laminar flow of the subsequent determination of the heat transfer
the condensate film. If a condensate Reynolds rate. From this rate, the steam condensation can be
number is defined as determined.
Liquid properties are evaluated at the film
D eq vp, ASvp, temperature determined from Equation 13.141.
N Re* = (13.151)
Pi p tPi
Tf =Tu UT V r,) =212-1(21 90)
but iv = Svp, so, therefore,
= 120.5°F
Aw 4A
N R i- 13.152) At 120.5°F, the condensate properties are:
PtPi Pi
3 2
p, = 62 lb/ft k, = 0.368 Btu/hr ft (°F/ft)
Turbulent condensation on vertical surfaces = 0.56 centipoise
Pi
can begin when /V Re exceeds a critical value of
about 1800. Because of the short flow path, turbu- At 21 2° F and 1 atmosphere,
lent condensation on horizontal surfaces is infre-
p v = 0.0373 lb/ft 3 ±H V = 970.3 Btu/lb
quent.
Condensation under turbulent flow condi- Since the condensate flow is not known,
tionswas studied experimentally by Kirkbride (23) Figure 13.18 cannot be directly used, but laminar
and analytically by Colburn (6). Figure 13.18 flow will be assumed. Then, from Equation 13.142
shows this work, which has been reasonably well
although extrapolation to
verified experimentally, k/
3
P/{pi- p v ) AH v g 1 /4

higher Prandtl numbers is not recommended. For h = 1.13 = 1.13

the turbulent region the lines are represented
P,L(TV -T,)
(0.368) (62)(62-0.0373)(970.3)(4.17 x 10 8 )1 1/4
3
by(1)
L (0.56x2.42)(3)(212-90)
hclr^ r
.1/3

=0.011(/V Rec )
1
'^^ /VPr,
/? = 934 Btu/hr ft
2
°F

13.153) Q = M(-AD = -934x(txtS x 1){90-212)

= 29,848 Btu/hr
and
Example 13.15. A vertical tube 1 in. O.D.
Q 29,848
and 36 in. long has steam condensing on the out- w = —— = = 30.7 Ib/hr
side surface. The steam is saturated at atmospheric AH„ 970.3

::-
-

^ ^^^ -
i
7 -
6
5

4
-""-
^^ Recommended
3 ^^-^Eq. 13.142

^-<>
Eq. 13.14CT"-^^^\.

::"

102 !910 J 3 4 5 6 7 8910* 4 5 6 78910 5

Figure 13.18. Film-type condensation on vertical surfaces. [(1)— by courtesy of the

author; (6)-by permission of Amer. Chem. Soc. Copyright © 1934; (23)-by
permission of A.I.Ch.E. Copyright © 1934.]

It is necessary to check on the assumption of
laminar flow.
4A 4w 4 x 30.7
" ec
= 347
ft irDft (3.14)(tM(0.56x2.42)
Thus, the laminar flow assumption is valid.
Effect of Noncondensable Gases
If noncondensable gases are present, the condensa-
tion coefficient is materially decreased. The non-
condensable gas collects in the vicinity of the sur-
face, and the condensing component must diffuse
through the film. Introduction of the diffusion
resistance into the path the rate of
decreases
condensation far below that for a pure material.
For example, as little as 1 or 2 percent of noncon-
densable gases in a steam system may reduce the
rate of transfer of heat by 75 or 80 percent. In
practice, a steam condenser is operated with appro-
priate auxiliaries (steam traps or air vents) to
eliminate air in the system. The calculation of
coefficients for systems in which noncondensables
are present requires consideration of simultaneous
mass and heat transfer (36).
Dropwise Condensation
If the condensate does not wet the surface but
collects in droplets, the entire surface does not
become covered with fluid. Instead, as condensate
forms, it quickly coalesces to form discrete larger
drops. The adhesion of the larger drops to the
surface is weak, and they soon fall, leaving behind
a dry path. Since the total surface is only partially
covered with drops, the uncovered surface at T^
provides film-free heat transfer with a large
driving force maintained.
The obvious advantages of dropwise con-
densation would lead to the opinion that his condi-
tion should be used wherever possible. However,
the physical preparation of surfaces that exhibit
permanent dropwise condensation is difficult.
Special temporary coatings on surfaces that
produce dropwise condensation were tested experi-
mentally with resultant increases of five to ten
times the value for film-type condensation coef-
ficients. A summary of work on dropwise con-
densation is given in Reference 42.
Boiling-Liquid Heat-Transfer Coefficients
Boiling, the opposite of condensation, is also of
primary importance. Types of apparatus in which
boiling may occur are many and varied in applica-
tion; for that matter, it is almost safe to say that
FUNDAMENTALS OF TRANSFER MECHANISMS 295
every condensing apparatus must have associated
with it some boiling apparatus. In spite of the
many necessary applications, boiling of liquids
remains phenomenon about which little is under-
a
stood. The apparatus under discussion consists of a
submerged solid, usually in the form of tube or a
nearly flat plate, as in a kettle jacket, whose tem-
perature exceeds the boiling point of the liquid
which contacts. Thus,
it the driving force for boil-
ing the temperature
is difference 7^ —Tb, the
difference between the surface temperature and
the boiling point of the liquid under the existing
pressure. As boiling progresses, heat transferred to
the liquid appears as latent energy of vaporization,
and vapor bubbles are formed. In a microscopic
sense, an element of fluid, however small, must be
elevated in temperature above its boiling point to
the superheated state so that it will contain energy
in excess of that necessary to bring about boiling.
When this element of liquid is in the described
state, the change of state is initiated in the form of
a bubble nucleus. The nucleus formation requires
this abnormal superheated state for some short
period of time to overcome what seems to be a
"reluctance" on the part of the fluid to change
phase, and a metastable high-energy state must be
achieved before this change can occur. In some
systems, the amount of liquid superheat necessary
is greater than in others. The conditions favorable
to bubble-nucleus formation are at best only
partially understood. For example, the "bumping"
of a laboratory flask occurs when a high degree of
liquid superheat accumulated before formation
is
of a bubble nucleus. With the formation of the
nucleus, the superheated liquid "flashes" — that is,
forms a volume of vapor with nearly explosive
force. The addition of boiling chips of rough por-
ous media increases the ease of nucleus formation
and therefore reduces the accumulation of liquid
superheat.
It is known that a pit in the hot surface of a
diameter and cone angle favorable to overcoming
the surface tension of the liquid is a most favorable
spot for nucleus formation, and growth to a size
greater than the critical dimension below which the
nucleus is unstable, and above which it will grow.
On an actively boiling surface, the pit seems to
retain some vapor from previously grown and dis-
engaged bubbles, which facilitates the growth of
the new nucleus.
If a system were operated at constant boiling
temperature (TB but increasing surface tempera-
)
ture (T-\), the response of the system could be
traced. As T^ increases, formation of bubble nuclei
296 MOLECULAR AND TURBULENT TRANSPORT

increases, and the increase in bubble activity pro- 7~i

— TB the hydrodynamic domain of the depart-
,

duces turbulence near the surface with consequent
increase in heat-transfer coefficient. The
heat-transfer coefficient favored by the decrease
is
in viscosity in the high-temperature region near the
surface. In this range the bubbles originate at
favored sites on the surface. These sites are surface
imperfections at which the shape of the imperfec-
tion, and perhaps a residual fragment of vapor or
dissolved or adsorbed gas, results in conditions
favorable for bubble formation and growth. As T-\
is increased, additional less-favorable sites
active until the entire surface is actively forming
bubbles. This condition is called nucleate boiling.
If the increase in driving force is continued by
increasing 7^ at constant TBl the heat-transfer
coefficient decreases after passing through a
ing bubbles makes it impossible for liquid to reach
increase in the surface, hence starving the interface of liquid
and generating a relatively stagnant layer of vapor
on the hot surface.
Any alteration of the conditions of surface
and fluid can affect the heat-transfer coefficient.
For example, if the pressure of the system is
increased, the bubble volume per mole of vapor is
decreased. This decrease will tend to suppress the
incidence of film boiling by decreasing vapor
become volume and permit increased heat-transfer
coefficients and increased values of the maximum
temperature difference. The vapor blanket in film
boiling is present because the density difference
between liquid and vapor in the gravitational field
is not sufficient to remove the vapor as rapidly as it

maximum because the heating surface becomes is formed. One exploration (15) in which the rate

"blanketed" by a film of vapor. This second regime of migration of bubbles away from the surface
of boiling is called film boiling. This decrease con- was increased in a centrifugal-force field showed
tinues until 7"i reaches such a level that additional marked increase in heat flux and heat-transfer co-
heat is transferred by radiation. The thermal con- efficient.
ductivity of vapor is considerably less than that of Surface tension (26,49) to have an
is said
the liquid (for water at 212°F, k yapo = 0.01 15 , effect upon film the liquid "wets" the
boiling. If

and /C| iquid =0.350). Therefore, the resistance to
heat transfer in the vicinity of the wall increases
sharply because of the vapor blanket. Figure 13.19
is a plot of h as a function of 7", — TB and shows
the maximum value.
One well-supported theory proposes that as
the generation of vapor increases with increasing
surface, the liquid will spread under vapor bubbles
tending to "snip" them off and away from the
solid surface, whereas, if the liquid does not wet
the surface, there will be less tendency for the
vapor bubbles to disengage. (Contrast this with
dropwise condensation.) Thus, a low surface
tension or tendency to wet a solid is favorable to

Region 1 [-Region 2 jRegion 31 Region 4 — Region 5

Natural convection Nucleate (Nucleate) Transition from Stable film boiling
boiling boiling nucleate to (with radiation
individual bubble film boiling coming into play)
bubbles columns

01 1.0 10 100 1000 10.000

m - to
Figure 13.19. Pool boiling flux as a function of temperature difference.
 FUNDAMENTALS OF TRANSFER MECHANISMS 297

heat transfer. The wettability of liquid-solid substances and binary mixtures. Figure 13.20a
systems is related to adsorption phenomena and is plots the ratio of maximum heat flux (g/4) max to
not well understood at this time. the critical pressure Pc as a function of the reduced
The design of apparatus for boiling liquids pressure P/Pc Figure 13.206
. is an expansion of the

introduces an unusual problem. If for some reason, same plot the low-pressure range, and the choice
in

the demand for vapor from a boiling apparatus of the plot depends upon the reduced pressure.
should increase, the usual practice of increasing the The temperature difference (7*! — TB qrTBX which )

temperature of the boiling surface will satisfy the occurs at (q/A) max can be estimated from Figure
,

demand if the system normally operates at a 13.20c in which [T^ — TB qmax is plotted as a )

temperature difference below that which gives function of the reduced pressure. Note that the
maximum flux. If the system is operating at or surface temperature is significant. Estimation of
above the temperature difference for maximum surface temperatures in practical equipment is
flux, an increase in the heating-surface temperature described in Chapter 15. The heat-transfer co-
will result in a decrease in vapor output. efficient for boiling at the maximum condition can

The maximum heat flux (q/A) max and the be calculated from the equation
temperature difference for maximum flux
(Tt — TB q m ax can be estimated in terms of the
(f)'max =/w(7"i
-TB qmax (13.154)
)

operating pressure and critical pressure of the

^ )

system with two separate criteria existing for pure where ftmax = the heat-transfer coefficient at

400

380 240

220
360
200
340
180
320
160
300
140
280
120
260
100

240 '
80

220 60

200 40

180 20

n
160 0.01 0.02 0.03 0.04 0.05
Reduced pressure (p/p c )
140
(b)
120

100

80

60

40

20

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10 0.01 0.02 0.04 0.1 0.2 0.4 0.6 1.0

Reduced pressure (plp c ) Reduced pressure (plp c )

(a) (c)

Figure 13.20. (a) Plot of maximum heat flux for boiling as a function of reduced
pressure— high-pressure range (3). (b) Plot of maximum heat flux for boiling as a
function of reduced pressure— low-pressure range (3). (c) Plot of maximum tempera-
ture difference for boiling as a function of the reduced pressure (3). (With permis-
sion of the Amer. Inst, of Chem. Eng.; Copyright © 1945.)
298 MOLECULAR AND TURBULENT TRANSPORT

[q/A\ This heat-transfer coefficient can be The heat flux (q/A) max and temperature
used to compute the minimum area to satisfy the difference (T :
- TB )
q max can be used to deter-
duty. Practical designs should always be made con- mine h max according to Equation 13.154.
siderably larger, frequently corresponding to a
coefficient 1 — 10 percent of this maximum.
Further discussion on this will be found in max (
' 1 ' B lq max
*"-'max
Chapter 15.
In recent years there has been a great deal of 1,103,000
most of
research on boiling heat transfer, although max
37
this work focuses on the detailed mechanism of the
2
phenomenon and on transfer rate prediction and h max = 29,900 Btu/hr °F ft

correlations that relate closely to high-pressure

boilers. Much of this work is summarized by This discussion has been based upon the
Tong and is reported in the original liter-
(59a) inertialbehavior of a static pool of liquid in which
ature, for example, Ref. 17a. The correlations that the only momentum contributions trace to bubble
have been presented (59a) are based on limited growth. For the case of a liquid flowing in a
data and yield conflicting results. Thus the en- conduit with heat addition, such as in a reboiler
gineer attempting to predict heat transfer rates tube, the increase in velocity incident to the
during boiling has a great deal of source material acceleration of liquid accompanying the increase in
available but should not use any part of this mat- volume results in a momentum increase and a pres-
erial without thorough cross-checking and exam- sure gradient that may generate unique problems
ination of alternate, possibly conflicting, results. and phenomena (20).
A flowing liquid and its equilibrium vapor
represent a highly compressible "fluid," and one of
Example 13.16. For purposes of design, cal-
fairly high density. The elastic consequence of this
culate the maximum heat flux and temperature combination is a limiting rate of flow of sur-
difference for maximum flux for water boiling at
prisingly low value. It is generally accepted that a
500° F. fluid cannot flow in a uniform conduit faster than
Data: The critical pressure of water is
the velocity of a compression wave in that fluid
3206 psia and the operating pressure at 500° F is
(this will be discussed further in Chapter 20). The
681 psia.
term acoustic velocity for a completely homo-
geneous continuum such as air is generally
Solution. The reduced pressure for this accepted, but in this case of two dispersed phases,
system can be computed there is serious question as to whether one can
characterize an acoustic velocity of the mixture.
681 Certainly, there is some maximum velocity at
= 0.222 which a compression wave can be transmitted
3206
through this mixture with resultant evaporation
and condensation, but such a velocity is surpris-
Figure 13.20a is entered at P/Pc = 0.222, and ingly low at pressures where the specific volume of
2
(g/4) ma x//'c 'S read as 355 Btu/hr ft psia. the vapor is large.
Animportant consequence of this phenom-
enon of simultaneous compression and condensa-
=(3206X355) tion for the liquid and its vapor is that there are
(j)
\A 'max likely to be three different mass flow rates, each
for a different fraction of vapor, which generate
the same pressure gradient along the conduit;
=1,103,000 Btu/hr ft
2 hence, an invitation to instabilities exists for low-
(j)
\A pressure operations, such as in vacuum evaporators,
max
where this limiting velocity must be recognized. In
The temperature difference designated (7", - high-pressure operations, such as power-generation
7e)max associated with (q/A) max can be esti- boilers, the difference in specific volumes of the
mated from Figure 13.20c. The chart is entered at liquid and vapor phases is small enough that it can
P/Pc =0.222 and (T - TB qmax is read as 37°F.
: ) be accommodated by proper engineering design.
 FUNDAMENTALS OF TRANSFER MECHANISMS 299

REFERENCES 30. Lorenz, L., Wiedmann Ann. d. Phys., 13, p. 582

(1881).
31. Lubarsky, B., and S.J. Kaufman, NACA Tech. Note,
1. Bell, K., Process Heat Transfer Notes (1975).
No. 3336 (1955).
2. Bird, R. B., W. E. Stewart, E. N. Lightfoot, Transport
32. Lyon, R. N., Chem. Eng. Prog. ,47, p. 75 (1951).
Phenomena, John Wiley & Sons, New York (1960).
33. Marchello, J. M., and H. L. Toor, Ind. Eng. Chem.
3. Bonilla, C. F., and M. T. Cichelli, Trans. A.I.Ch.E.,51
Fund., 2. p. 8(1963).
p.761 (1945).
34. C, Trans. A.J.M.E.,69. p. 947 (1947).
Martinelli, R.
4. Bromley, L. A., N. R. LeRoy J. A. Robbers, Ind. Erg.
35. McAdams, W. H., Heat Transmission, McGraw-Hill,
Chem.,45, p. 2639 (1953).
New York, 2nd Ed., p. 221, Fig. 111 (1942).
5. Colburn, A. P., Trans. A.I.Ch.E.,29, p. 174 (1933).
36. McAdams, W. H., Heat Transmission, McGraw-Hill,
6. Colburn, A. P., Ind. Eng. Chem.,26, p. 432 (1934).
New York, 3rd Ed., Chap. 7 (1954).
7. Cope, W. F., Proc. Inst. Mech. Engr., London, 145,
37. Ibid., Chap 9.
p. 99 (1941).
38. Ibid., p. 335.
8. Coulson, J. M., and J. F. Richardson, Chemical
39. Ibid., p. 267, Figs. 10-12.
Engineering, Vol. 1, McGraw-Hill, New York (1955).
40. Ibid., p. 267, Figs. 10-11.
9. Danckwerts, P. V., Ind. Eng. Chem. 43, p. 1460 ,

41. Ibid., p. 267, Figs. 10-13.

(1950).
42. Merte, H., Advances in Heat Transfer, Vol. 9,
10. Davis, E. J., and M. M. David, Ind. Eng. Chem. Fund.,
Academic Press, New York, p. 181 (1973).
3, p. 111 (1964).
August, Trans. A.S.M.E., Ser.
43. Nusselt, W., Z d. ver. Deutsch Avg., 60, p. 514 (1916).
11. Depew, C. A., and S. E.
44. Ibid., p. 560.
C, 93 (1971).
45. Pinczewski, W. V., and S. Sideman, Chem. Eng. Sci.,
12. Drew, T. B., in McAdams' Heat Transmission,
29, p. 1969 (1974).
McGraw-Hill, New York, 3rd Ed., p. 330 (1954).
46. Powell, R.W., Trans. Inst. Chem. Engrs. London, 18,
13. Eckert, E.R.G., and T.W. Jackson, NACA, Tech. Rept.
p. 36 (1940); and Powell, R. W., and E. Griffiths,
1015, p. 255 of 37th Ann. Rept. (1951 ). Chem.
Trans. Inst. Engrs. London, 13, p. 175 (1935).
14. Fage, A., and H. C. H. Townend, Proc. Roy. Soc.
47. Rathby, G. D., and K. G. T. Holland, Free Convection,
London, Ser. A., 135, p. 656 (1932).
Adv. in Heat Transfer, Vol. II, Academic Press, New
15. Gambill, W R., and N. D. Greene, Chem. Eng. Progr.,
York, 1975.
54, No. 10, p. 68 (1958).
48. Reynolds, O., Proc. Manchester Lit. Phil. Soc, 8,
16. Goldstein, S. (ed.). Modern Developments in Fluid
(1874).
Mechanics, Oxford University Press, London, p. 651
49. Rohsenow, W. ASME, 74, p. 969 (1952).
M., Trans.
(1938).
50. Rouse, H., and W. Howe, Basic Mechanics of Fluids,
J.
17. Graetz, L., Ann. Phys. u. Chem., 25, p. 337 (1885).
John Wiley & Sons, New York (1953).
17a. Grigull, U., and E. Hahne, Eds., "Heat Trans-
51. Ibid., p. 181, Fig. 107.
fer, 1970." Preprints of papers presented at the Fourth
52. Ibid., p. 188, Fig. 112.
International Heat Transfer Conference, Paris-Versailles,
53. Schlicting, H., Boundary Layer Theory, 5th Ed.,
1970.
McGraw-Hill, New York, Chap. 7 (1968).
18. Hanratty.T. 1., A.I.Ch.E. Journal. 2, p. 359 (1956).
54. Ibid., Chap. 14.
18a.Higbie, R., Trans. A.I.Ch.E., 31 , p. 365 (1935). 55. Sellars, J. R., M. Tribus, Klein, Trans. ASME,
J. S. 78,
19. Jakob, M., Heat Transfer, Vol. 1, John Wiley & Sons, 441 (1956).
p.
New York, p. 661 (1949). 56. Sherwood. T. K., Trans. A.I.Ch.E., 36, p. 817 (1940).
20. Kalinin, E. K., et al., "Film Boiling Heat Transfer,"
57. Sherwood, T. K., and R. L. Pigford, Absorption and
Adv. in Heat Transfer, Vol. II, Academic Press, New
Extraction, McGraw-Hill, New York, p. 74. Fig. 22
York (1975).
(1952).
21. Kays, W. M., Trans. A.S.M.E., 77, p. 1265 (1955). 58. Sherwood, T. K., R. L. Pigford, C. R. Wilke, Mass
22. Kern, D. Q., Process Heat Transfer, McGraw-Hill, New Transfer, McGraw-Hill, New York, Chap. 5 (1975).
York, p. 266 (1950). 59. Sieder, E. N., and G. E. Tate, Ind. Eng. Chem., 28,
23. Kirkbride, C. G., Trans. A.I.Ch.E., 30, p. 170 (1934).
p. 1429 (1936).
24. Klinkenberg, A., and H. H. Mooy, Chem. Eng. Prog.,
59a. Tong, L. S., Boiling Heat Transfer and Two- Phase
44, p. 17 (1948). Flow, John Wiley & Sons, New York (1965).
25. Knudsen, J. F., and D. L. Katz, Fluid Dynamics and
60. Toor, H. L, and J. M. Marchello, A.I.Ch.E. Journal, 4,
Heat Transfer, McGraw-Hill, New York, p. 467 (1958).
p. 97 (1958).
26. Kreith, F., Principles of Heat Transfer, International
61. Treybal, R. E., Mass Transfer Operations, McGraw-Hill,
Textbook Co., Scranton, p. 406 (1960). New York, 63 (1968).
p.
27. Langhaar, H. L, Trans. A.S.M.E.,64, A-55 (1942). 62. Walker, W. W. K. Lewis, W. H. McAdams, E. R.
H.,
28. Latzko, H., Z Angew. Math. u. Mech., 1, p. 268 Gilliland, Principles of Chemical Engineering, McGraw-
(1921). Hill, New York, p. 43 (1937).
29. Lohrisch, W., Forsch a.d. Geb. d. Ingen. (VDI), 332,
p. 1 (1929).
300 MOLECULAR AND TURBULENT TRANSPORT

PROBLEMS (b) What is the thickness of the boundary layer at 3

in. from the leading edge?
(c) What is the thickness of the boundary layer 3 ft

13.1. Calculate the pressure drop in 100 m of from the leading edge?
smooth tubing of 0.02 m I.D. for: (d) Calculate the point stress 3 ft from the leading
(a) Carbon dioxide at 40 C, atmospheric pressure, edge.
and flowing at a rate of 5 x 10~ 2 m 3 /s. (e) Calculate the average stress over the plate from
(b) Benzene at 40 C, atmospheric pressure, and the leading edge to a point 3 ft from the leading edge.
flowing at a rate of 5 x 10~ 2 m 3 /s. 13.11. Air at 200° F and 1 atm flows over a long flat
(c) SAE 10 oil at 40 C, atmospheric pressure, and surface at 200 ft/s free-stream velocity.
flowing at a rate of 5 x 10
-2 m 3 /s. (a) Calculate the point at which the boundary layer
(d) Mercury at 40 C, atmospheric pressure, and becomes turbulent.
flowing at a rate of 5 x 10~ 2 m 3 /s. (b) Calculate the maximum thickness of the laminar

What flow regime exists in each of the cases? boundary layer.

13.2. Estimate the flow rate at which the transition (c) Calculate the thickness of the boundary layer 10

from laminar to turbulent flow will begin in a 2-in. I.D. ft from the leading edge.
smooth tube for: 13.12. Calculate the entry length for SAE 10 lubrica-

(a) Propane at 70° F, 1 atm. ting oil at 100 F flowing into a 1 -in. I.D. tube for an
(b) Acetone at 70° F. average velocity in the tube of:

(c) SAE 10 lubricating oil at 70°F. (a) 0.05 ft/s.

13.3. The pressure drop is 10 psi over 100 ft of (b) 1 ft/s.

smooth 1^ in. I.D. pipe with SAE 10 lubricating oil flowing 13.13. Determine the entry length for air at 100°F
at 100° F. and 1 atm flowing into a 1 -in. I.D. tube at 0.5 ft/s average

(a) What is the average velocity of flow? velocity in the tube.

(b) What is the flow regime? 13.14. A support structure for a spherical gas storage

13.4. Calculate the pressure drop per meter of tank 15 m in diameter is being designed. Meteorological
6x10~ 3 m I.D. smooth tubing for methane at atmo- records indicate maximum winds of 50 m/s have been
spheric pressure, 500 K, and flowing at 5 x 10~ 4 m 3 /s. experienced in the region where the tank will be. Calculate
13.5. Air at 150" F and 1 atm flowing through a the force exerted on the tank. Air temperature is 293 K.
smooth 6-in. I.D. duct 100 ft long gives a pressure drop of 13.15. A hood ornament for a new car is essentially a

10 in. of water. Determine the flow. circle about 0.06 m in diameter. Estimate the power needed
13.6. For the two cases indicated below, calculate to overcome the drag of this ornament when the car is

the pressure drop per foot when the flow is in the annular traveling at a speed of 25 m/s. Air temperature is 293 K.
space made from a 2-in. O.D. pipe inside a 3-in. I.D. pipe. 13.16. A service station sign is 30 m above the
(a) Water at 1 00° F and flowing at lOOgal/min. ground and is exposed to hurricane force winds of 54 m/s.
(b) SAE 10 oil at 100°F and flowing at 100gal/min. The sign is a 2.5-m circle. Estimate the force acting upon
13.7. An oil (/i = 20 centipoise, p = 55 lb/ft 3 flows ) the sign during the hurricane. Air temperature is 300 K.
through a 1 -in. I.D. pipe 1000 ft long. 13.17. Water at 60°F is flowing past a cylindrical

(a) What is the maximum flow in gallons per minute bridge pier 1 ft in diameter submerged to a depth of 10 ft.

that will ensure laminar flow? Calculate the force on the pier when the water velocity is

(b) What would be the pressure drop for this flow? (a) 0.1 ft/s.
13.8. Air at 1 N/m 2 and 50°C flows through a (b) 10 ft/s.

0.01 m x 0.005 m rectangular channel 10 m long. If the 13.18. Integrate Equation 13.48 for momentum
flow -3 m 3 /s, what the pressure drop?
is 1 x 10 is transfer according to the directions given in the footnote on
13.9. An oil (specific gravity = 0.88) flows through a p. 262.
4-in. diameter horizontal pipe 1000 ft long. The pres-
I.D. 13.19. Water is heated in a copper tube of 1 in I.D.
sure drop over this length is 25 lbf/in 2 The oil flows at a . At a point along the tube the average water temperature is

rate of 0.4 ft 3 /s. Estimate the viscosity, in centipoise, of 100°F and the tube wall temperature is 210°F. Calculate
the oil. the heat-transfer coefficient and the heat flux for a water
13.10. Methane at 70°F and 1 atm is flowing parallel velocity of 5 ft/s.

to a flat plate 4 ft long. The methane velocity is uniform at 13.20. Air at atmospheric pressure and an average
10 ft/s as it passes the leading edge of the plate. temperature of 100°F is flowing at 10,000 ft 3 /min through
At what point (if any) from the leading edge does
(a) a 30-in. circular duct. The wall temperature is 300 F. Calcu-
the boundary layer become turbulent? late the heat flux.
 FUNDAMENTALS OF TRANSFER MECHANISMS 301

13.21. SAE 10 lubricating oil flows at 40 gal/min (a) The Reynolds analogy.
through a smooth tube of 1 in. I.D. At a point in the tube (b) The Colburn analogy.
the average oil temperature is 300° F and the tube wall (c) The Martinelli analogy.
temperature is 60°F. Determine the heat flux at this point. (d) The recommended empirical equation.

13.22. Calculate the heat-transfer coefficient for the 13.31. Under what conditions are the various

cooling of hydrogen from the following data at 1 atm: analogy expressions not valid?
13.32. A cylindrical steel chimney 3 ft in diameter
Average hydrogen temperature = 300 F and 50 ft high is exposed to a wind of 50 miles/hr. The
Tube-wall temperature = 60 F
surface of the chimney is at 500° F and the average air
Tube diameter = 0.5 in.
temperature is 70°F.
Hydrogen velocity = 200 ft/s
(a) Calculate the force on the chimney.

13.23. Calculate k c , kg, k y , ky at a point in a (b) Calculate the heat loss from the chimney.
wetted-wall column where air containing negligible 13.33. The Appendix C-5 shows that the
plot of

moisture is contacting water. /Vn u decreases with increasing values of L/D for /Vr 6 less
than 2100. Why should this occur?
Data: Column diameter = 1.5 in. I.D. 13.34. SAE 10 lubricating oil flows through a 1-in.

Air temperature = 70 F I.D. tube at an average velocity of 0.1 ft/s. At one point,
the oil is at 70°F and the tube wall temperature is at
Air pressure = atmospheric
Air velocity = 12 ft/s 120 F. The tube is 10 ft long, and well-developed flow may
Water velocity = 4 ft/s
be assumed. Calculate the heat flux at this point.
13.35. Water flows through a 0.05-m I.D. pipe at a

13.24. Ethyl alcohol is flowing downward counter- mass rate of 2 kg/s. At a point along the tube the water

current to a stream of air in a wetted-wall column 2 in. I.D.

temperature is 15°C, and the tube wall is at 70° C. What is

Both the alcohol and air are at 77°F. The total pressure is
the heat flux at this point, assuming well-developed flow?

700 mm Hg. At a point in the column the average partial 13.36. A panel is placed in a hallway for heating. The

pressure of ethanol vapor in the air is 25 mm Hg. The gas panel is a flat, vertical surface 0.3 m x 0.5 m. Estimate the

velocity is 2 ft/s. Calculate the mass flux at this point. heat transfer rate by free convection if the panel surface is

13.25. Show that (a) k y = kc P/RT and (b) kG = 320 K and ambient air is 290 K.

kc IRT 13.37. A horizontal steam pipe 2.5 in. O.D. is

13.26. Show that Ar x = k c p- IM exposed to ambient air at 40 F. The outer surface of the
13.27. Show that k Y = k c P(y b steam pipe is at 200° F. Calculate the heat loss over 100 ft
)(k&)
13.28. Air at 212°F flows over a streamlined naph- of pipe.

thalene body. Naphthalene sublimes into air, and its vapor 13.38. Condensing equipment may be mounted
pressure at 212 F is 20 mm Hg. The heat-transfer coeffi-
vertically or horizontally. In which position should a con-

cient for the same shape and air velocity was previously denser be placed to condense the maximum amount of a

found to be 4 Btu/hr ft 2 °F. vapor for a given surface temperature? All things being the
The concentration of naph-
"same," what are the relative amounts of vapor condensed
thalene in the bulk air stream is negligibly small. The mass

2 for each installation?

diffusivity of naphthalene in air at 212°F is 0.32 ft /hr.
13.39. Condensing equipment may be mounted
Calculate the mass-transfer coefficient and mass flux for the
horizontally or vertically. In Example 13.14 horizontal
system.
mounting was indicated based upon the size of the coeffi-
13.29. 310 K and atmospheric pressure flows
Air at
For laminar flow of condensate, derive an expression
cients.
through a smooth tube 0.025 m I.D. The tube is 6 m long.
in D, L, and A/ that defines the limiting tube length beyond
Pressure taps at the ends of the 6-m section indicate a
which condensers should be mounted vertically for maxi-
pressure drop of 7 x 10 3 N/m 2 . Estimate the heat transfer
coefficient using:
mum value of the heat-transfer coefficient.
13.40. Acetone is to be condensed on a single 1-in.
(a) The Reynolds analogy.
O.D. tube at 1 atm. The tube surface is at 60° F. What
(b) The Colburn analogy.
length of tube will give the same average heat-transfer
(c) The Martinelli analogy.
coefficient whether mounted horizontally or vertically?
(d) The recommended empirical equation.
13.41. Make a qualitative plot of heat-transfer

13.30. Liquid sodium at 400 K flows through a coefficient and surface temperature for the case of pool
0.025-m I.D. tube at 1.5 m/s. Calculate the heat transfer boiling of water on a horizontal metal surface. How would
coefficient by the nature of the metal to materially alter the plot?
 Fourteen

Interphase Transfer

Many industrial operations involve the transfer of The transfer process is more complex if the
heat, mass, and momentum from one phase to transfer path is made up of resistances in parallel.

another. In a heat exchanger, for example, heat is
being transferredfrom a hot fluid through a tube
wall to a cold fluid. Gas absorption involves the
transfer of mass from a solute-rich gas phase to a
solute-poor liquid phase. In the case of a fluid
flowing through a pipe, momentum is transferred
through the fluid to the pipe wall.
In Chapter 13, the rate equation was
integrated in the transfer direction at a given posi-
tion in an apparatus to define a transfer co-
efficient. It is the purpose of this chapter to show
the interrelation of resistances and driving forces in
an interphase transfer process at a particular point
in a piece of equipment and to integrate the rela-
tionships over the entire transfer area. In addition,
a design equation relating the rate of transfer to
the total required transfer area will be developed.
As an example, situations arise in heat transfer
where radiation, conduction, and convection occur
simultaneously. The conduction part of the transfer
process has one resistance while parallel to the
conduction resistance; there is a different resist-
ance along the radiation path.
A further complication occurs if the heat-
transfer process is considered at another point in
the exchanger where the temperatures of both
phases have changed. Although the same qualita-
tive arrangement of resistances is present, the
changes in temperature have resulted in different
temperature gradients and different resistances at
this point.
Series Resistances at a Point: Steady State
It has already been established that the rate of
transfer, be it heat, mass, or momentum, is equal

MULTIPLE PHASE RESISTANCES to the driving force divided by a resistance. This

may be expressed mathematically as
As before, the rate of transfer may be expressed as
the driving by the resistance. In
force divided
operational equipment where the rate of transfer
Rate =\pA = - —
Ar =
driving force

resistance
(14.1!

dictates the transfer-area requirements, a new

situation arises that has not been encountered The above equation in its earlier application
before. Multiple resistances are encountered, and was limited to a single phase with the driving force
the rate of transfer varies from position to position and resistance indigenous to that phase. The same
within the apparatus. For example, at a particular equation, however, may be applied to more than
point in a heat exchanger, an analysis of the trans- one phase. Imagine simple transfer through two
fer path reveals that several resistances are encoun- resistances in series as illustrated in Figure 14.1.
tered in series. First, there is the resistance to Assume steady-state operation exists, in which case
transfer between the hot bulk phase and the tube the rate of transfer (\pA) is constant with respect
wall. Then there is the resistance of the tube wall to time.
itself, and finally the resistance between the tube In Figure 14.1, R^ and R fl schematically
wall and the cold bulk phase. Within each of these depict resistances associated with phases A and B.
resistances is found a temperature gradient. It should be recalled that these resistances are

302
 INTERPHASE TRANSFER 303

iM
A
M \pA
where
r- *a\ r2 ' &b™B
Ar^ = the driving force across R^
AT B= the driving force across R B
Transfer direction
*- Ar„ = the driving force across R„
Figure 14.1. Series resistances.
Parallel Resistances at a Point: Steady State
comprised of the transfer distance and the diffusi- Consider Figure 14.2 where the transfer process
vities. Solving Equation 14.1 for the driving force occurs through two parallel resistances R„ and R
across each phase gives for phase A B
under conditions of steady-state operation. Let

(AO, = -(r 2 - T, ) = (^A)R A (14.2) r 2 - r, = afa

and for phase B and

{£T)b = -(ra - T2 ) = (iPA)R B (14.3) r 4 - r 3 = Ar fi

Addition of Equations 14.2 and 14.3 gives For resistance (R^,) in accordance with
Equation 14.7
-(r 3 - T! ) = (^)R„ + bl>A)R B -

(14.4)

For simple transfer at steady state, Equation 14.4 AI\

becomes WA)a=- :i4.9)

-(r 3 - r, )
= m) [{R A + rb )] (14.5) and for R a
or

-(r 3 -r 1 )
= (^)[— *— + —*— } (14.5a) =
Ar,
6 A & A L A A B B J (\I*A) B (14.10)

Solving Equation 14.5 for the rate of transfer gives but

(\I>A) = - (ra-ro (rs-r,)

(14.6)
(\pA) A +(^A) B = {^A) tota[ (14.11)
Ra +R B 1 1
Therefore combining Equations 1 4.9 and 1 4.1 with
SA A A >B*B Equation 14.11

Note that Equation 14.6 is written in terms of ai^ Ar g

the driving force (r 3 — 1^ across both resistance ) («M) total [
[14.12)
A and resistance B with the intermediate concen-
R/4 Rb
tration T 2 not appearing. This driving force is The nature of the driving force is such that if
calledthe total, or overall, driving force, and the driving force in each branch is not equal, it will
^a + Rfi is the total, or overall resistance. The adjust itself until AI^ and Ar s do become equal,
same reasoning as used in the development of or
Equation 14.6 can be applied to any number of
resistances in series. In general, then, for steady-
-AT A =-Ar B (14.13)
state, series resistances add As a result, Equation 14.12 may be written

2 (ADy
\pA=-
7=1
(14.7)
(^)totai=(-Ar> , orfl )(—
\ Ha
+—
Ho
)
'
(14.14)
/=n
2 R,
/-i

For steady state, a useful relationship showing

the driving-force-resistance relationship fo. the
total system or any part may be written by restat-
ing Equations 14.2, 14.3, and 14.6.

SAr AI\ Ar,

(<M) =
ZR R,
Transfer direction
»-

(14.8) Figure 14.2. Parallel resistances.

304 MOLECULAR AND TURBULENT TRANSPORT

Generalizing Equation 14.14 for any number of

branches gives (showing that parallel conductances
add)

-AT, - Ar ? UjAj)
(iM)totai = 's 7T - /
(14.15)

Equations 14.7 and 14.15 may now be

applied to steady-state heat and mass transfer.
Series or parallel resistances for momentum
transfer have no significance because momentum
transfer usually stops at a boundary.

HEAT TRANSFER
Distance from interface
Heat transfer as carried out industrially involves
the contacting of a hot phase and a cold phase, Figure 14.3. Temperature conditions in the

separated by a well-defined boundary. Both series heat exchange.

and parallel resistances are frequently encountered.

For example, a typical heat exchange involves at temperature profile proceeding from the hot phase
least three resistances in series, such as the fluid to the cold. Since both phases have the same

resistance of the hot phase, the resistance of the temperature at the interface, the interface is In
tube wall that keeps the two phases separate, and equilibrium.
the fluid resistance of the cold phase. A similar For of resistances, Equation 14.7 is
a series

physical situation exists in a furnace wall con- applicable. Because the nature of the resistance
structed of several different types of building layers is such that their thickness cannot be pre-
material where each material constitutes a separate dicted, it will be necessary to use the surface-
resistance in series in the direction of heat flow. coefficient concept mentioned earlier. For heat
This same furnace wall, on the other hand, with a transfer, the rate \jA becomes q, and the resistance
steel door or a peephole, will now present the M6A is designated as MhA.
more-complicated case of parallel resistance paths, The rate of heat transfer from the bulk of the
each containing several resistances in series. hot phase to the interface is

Series Resistances in Heat Exchangers

Qa =- \T,-TA )
(14.16)
Consider a point in a heat exchanger where the hot
phase has a temperature TA and the cold phase has hA A t

a temperature TB with both phases being in

,

motion. For the time being, it is assumed that the and the rate of transfer from the interface to the
two phases exchanging heat are kept separated by a bulk cold fluid is
boundary of negligible thickness and, hence,
negligible resistance. The actual case of a tube wall (TB - Tj)

serving as a boundary will be considered later in Qb = (14.17)

1
the chapter. The apparatus at the point in question
will have two resistances in series in the transfer
hB A B
direction. As previously established in Chapter 13,
Each of the above equations is made up of a
because of boundary-layer phenomena, only a por-
single resistanceand a single driving force.
tion of the phases on either side of the boundary
Rearranging Equations 14.16 and 14.17 yields
willoffer significant resistance to transfer. Figure
14.3 schematically depicts this physical situation.
Line /— / represents the imaginary interface -^i-TA )^qA (-~-\
\h&AA '
(14.16a)
common to both phases, and the dotted lines on
either side of /— / represent the extent of the region
and
that comprises most of the resistance to transfer.
The steep or large temperature gradients will occur
(TB -T )=q B [-^—\
i
(14.17a)
in these regions, and the solid line illustrates the n dm d
 INTERPHASE TRANSFER 305

Addition and rearrangement of Equations through point P; it is an equation of a straight line

14.16a and 14.17a, with the condition that for of slope — (h B A B /h A A A ). The intersection of this
steady state q A = q B results in
,
line with the equilibrium curve determines the
interfacial temperature 7",. Figure 14.4 serves to

~{TB - TA demonstrate a concept that will be useful when

= (14.18) considering series resistances in mass transfer.
Q
1 1
Equation 14.18 or a more complete form including
.h A AA hB A B all resistances is appropriate for studying heat
transfer.
Equation 14.18 is identical in form and mean- A practical case of heat exchange might be a
ing with Equation 14.7. The use of Equations furnace wall consisting of several layers of brick of
14.16 and 14.17 is perfectly valid, except that various resistances. On one side of the wall will be
difficulties arise when measurement of 7", is hot gases, while on the other will be ambient air. In
attempted. A relationship between bulk and inter- this both conduction and convection
situation,
facial temperatures is easily obtained by dividing occur. There the turbulent transfer by convec-
is

Equation 14.16 by Equation 14.17, however. For tion to and from both sides of the wall, and there
steady state is a molecular transfer, or conduction, through the

wall. Both mechanisms occur simultaneously.

TA - Tj _ ( hB A B \ Figure 14.5 illustrates this physical situation.
(14.19)
TB -T, \h A Aj brick
In this case there are three different types of
making up the furnace wall. The brick layers

Both h B and h A can be calculated by suitable are Ax 2 Ax 3 and Ax 4

, , feet thick, and each layer

correlations found in Chapter 13, and T, can be will have a thermal conductivity dependent upon
determined by solving Equation 14.19 with AA = the material of construction and its temperature.
AB for well-defined boundaries, the interfacial
The solid line represents the temperature profile
;

through the various resistances, assuming constant

areas are common for both phases. The interfacial
thermal conductivity within each kind of brick.
temperature can also be determined graphically
T,
Applying the development used in Equation 14.7
by plotting the usual equilibrium diagram. In this
for heat transfer, (\pA) is equal to q, and
case the hot-phase temperature will be plotted
against the cold-phase temperature with equili-
brium being represented by a 45-degree line. Figure
-AT, = qA 14.20)
14.4 shows this plot. \h A A A )
In Figure 14.4, point Ogives the temperatures
corresponding to the bulk conditions of both The heat-transfer coefficient is used in Equation
phases at the point of interest. This point would 14.20 as an expression for the resistance to heat
correspond to a point on the operating line relating transfer offered by the fluid. Continuing across
the temperature history of the two phases through-
out the exchanger. Equation 14.19 is plotted

Cold-phase temperature *-

Figure 14.4. Equilibrium diagram for heat ex-

change. Figure 14.5. Series resistances in a furnace wall.
306 MOLECULAR AND TURBULENT TRANSPORT

the wall, assuming intimate contacting at the brick of Equation 14.28 shows that
interfaces,
-2A7" -AFt -AT2
Ax 2 Q =
/ \
(14.21)
2R

! 14.29)
(14.22)

/ Ax 4 \
Example 14.1. A flat furnace wall consists of
(14.23) a 6-in. layer of fire brick [fc=0.95 Btu/hr ft 2
(°F/ft)], 4 in. of insulating brick {k = 0.14), and
the transfer process
In these three equations,
4 in. of common brick (k = 0.8).
The fire-wall temperature is 1800°F, and the
is only molecular and the transfer path length is
outer surface temperature is 120°F. Determine the
known. Finally across the outer resistance layer
heat loss through the wall and the temperature at
the junctions between the different types of brick.
-ATs = qB (—-) (14.24)
Solution. Equation 14.28 clearly applies
In these equations the symbols are: here. Since surface temperatures are given, there is

no need to calculate the thermal resistance of the

AT = temperature driving force, °F across each layer of gas on either side of the wall. Basis: 1 ft 2
resistance of transfer area
q = heat-transfer rate, Btu/hr
k = thermal conductivity, Btu/hr ft 2 (°F/ft) Ax: (6/12)
Resistance of fire brick =
A = heat-transfer area, ft 2 k2 A (0.95M1.0)
h = heat-transfer coefficient, Btu/hr ft 2 °F
= 0.527 hr°F/Btu
Adding Equation 14.20 through 14.24 gives
Ax 3 (4/12)
- 1 Resistance of insulating brick = =
-(A7 ! + A7"2 + A7"3 + A7"4 + AT5 )
= qA k3 A (0.14X1.0)
hA A A
= 2.38hr°F/Btu
Ax 2 Ax 3 Ax 4 1

kzA k4 A 4 hB A B
Ax 4 (4/12)
koA
2^2 3^3 " Resistance of common brick =
k4 A (0.8M1.0)
(14.25)
= 0.417 hr°F/Btu
Now,
steady state, all the heat passing
at
through R, must pass through R 5 so q A = q 2 - Therefore, 2R = 0.527 + 2.38 + 0.417 = 3.32 hr
.
= q. Therefore, Equation 14.25 may be written
. .
°F/Btu, and -XAT = -(120 - 1800) = 1680°F.
as
-2 AT- 1680
Q = = 506 Btu/hr 2
Thus ft
Ax 2 Ax 3 Ax 4 1
IR 3.32
-XAT Q + + -+ +
lh A AA k2 A 2 k3 A 3 kA A 4 hB A B i Equation 14.29 used to evaluate junction
(14.26) temperatures.
or
-HAT -A7"2 -A7", -AT*
-XAT=q[R 1
+ R2 + R3 + R4 + R5 ] =qZR £R R2 R3 R4
(14.27) The temperature at the junction of the fire
and the heat-transfer rate is brick and insulating brick is determined.

-2Ar A 7*2
q= (14.28) q= -
SR R2

Equation 14.28 is, of course, identical with _A_r2

506 = - _
Equation 14.7. Reexamination of the development 0.527
 INTERPHASE TRANSFER 307

or system, each area would have a unique numerical

value of the overall coefficient as
-A 72 = 0.527(506) = 267°F

1 1 1 1
which gives a temperature of 1800 - 267 = 1533°F. Z/? =
For the interface between the insulating brick UA UA A A UB A B hB A B
and common brick
+ z
7=1
—Ax.
kjAj
- +
1

hA A A
14.30)
A7, AT-*

where
or U= overall heat-transfer coefficient,
2
Btu/hr °F ft
267 A 7,
Aa or AB = transfer area of hot or cold phase,
0.527 2.38 2
ft
h B = heat-transfer coefficient, Btu/hr
and
F ft 2 transfer area of cold phase
A73 =2.38(506) = 1205°F h A = heat-transfer coefficient for hot
phase, Btu/hr °F ft 2 transfer
which gives a junction temperature of 1533 — area of hot phase
1205 = 328° F. This temperature could also have /=* AXy
been calculated using the A7 across both the fire 2 —— = sum of all resistances associated
/-i kjAj
brick and insulating brick and the sum of the w t f, trie moiecuia,- transfer of
j

2
resistances associated with both these materials. heat, hrft °F/Btu
This illustration was for a case of constant
heat flow through a series of resistances of con- The heat-transfer equation may now be writ-
stant area. the transfer area varies, the appro-
If ten as
priate area must be used (see Chapter 10).
A7 A7 A7 total driving force

Overall Heat-Transfer Coefficient 1 1 1 total resistance

A glance at Figure 14.5 and Equation 14.29 shows UA A A Ub A b UA

that the heat-transfer rate for a steady-state process
(14.31)
can be calculated if the temperatures at the two
boundaries of any one resistance are known. Norm- where A7 is the total temperature drop between
ally the bulk stream temperatures are the easiest to the two phases.
measure. The temperature difference between the By examination of Equation 14.30, some
two bulk phases will represent the temperature qualitative conclusions may be drawn regarding the
driving force across all resistances encountered relative effect of the various resistances
between the two phases. This AT is referred to as encountered along the transfer path. Consider a
the total driving force. Accordingly, an overall co- heat-transfer apparatus where two fluids, separated
efficient may be defined as a function of the total by a tube wall, are exchanging heat. In this case
resistance and transfer area as MUA = I/?, where U Equation 14.30 reduces to
is the overall heat-transfer coefficient, Btu/hr
2
ft °F, and A is the transfer area consistent with
1 1 1 1 Ax w 1
— i t

the definition of U. UA UA A A UgAg hA AA kw Am h B Ag

For example, the area term associated with
(14.32)
the overall heat-transfer coefficient may be the
transfer area of either the hot or the cold phase. In where A m the appropriately averaged mean of
is

a heat exchanger in which the two phases are AA and A B .consider the pipe wall to be very
First,
separated by a tube wall, the outside surface area thin; then Ax w approaches zero and the tube
of the tube would be the transfer area of, say, the resistance is negligible, and with negligible wall
hot phase. On the other hand, the inside surface resistance
area of the tube would be the transfer area of the
1 1 1 1
cold phase.
14.33)
Since the total resistance is fixed in a given UA A A UB A B hA A RA
h B^g
308 MOLECULAR AND TURBULENT TRANSPORT

but for a thin wall A A =A B =A, thus,

1~JL + J_ 14.34)
U~hA
!

hB
Depending upon the magnitudes of h A and
h B the resistances at both surfaces may be signific-
,
TB =95'F
ant The value of h may vary between about 2.0
2o
Btu/hr ft F and 40,000 or more, depending upon Resistance layer
the nature of the heat-transfer process. For cold fluid

example, in a heat exchanger using condensing Resistance layer

hot fluid
steam to heat water, the magnitude of the steam
coefficient is about 2000, whereas that of the
Figure 14.6. Sketch for Example 14.2.
flowing water might be about 400. From this one
can deduce, using Equation 14.34, that the water-
side resistance is a major portion of the total resist- The transfer area may be calculated as follows:
ance. It is usual base the overall
practice to
For the cold phase:
coefficient upon the area associated with the
greatest resistance. In general, however, before any AB = ttD b L = (3.14)(0.902/12)(1) = 0.236 ft
2
simplification of Equation 14.30 can be made,
each resistance should be analyzed individually and For the hot phase:
in comparison to the overall resistance before it
may be said to be negligible. Correlations found in A A =nD A L = (3.14X1/12X1) = 0.262 ft
2

Chapter 13 will permit the evaluation of the

For the tube wall:
individual heat-transfer coefficients.

0.236 + 0.262
Example A
double pipe heat exchanger
14.2. A = 0.249 ft
2

consists of a 1 in., 18 BWG copper tube inside a

2 in. sch. 40 steel pipe.* Water is flowing through The arithmetic mean A am area is justified as
the inner tube, while in the annular space lube oil opposed to the use of A lm in this case because the
flows countercurrent to the water. At a particular tube is thin-walled and the inside and outside areas
point in the exchanger the bulk temperature of the are nearly equal.
oil is 350°F, while the bulk water temperature at

the same point is 95° F. The heat-transfer co- 0.049

Tube wall thickness = = 0.00408 ft
efficients at this point are 100 Btu/hr ft 2 °F for the 12
oil and 400 for the water. Calculate the heat-

transfer rate at this point. What is the temperature Thermal conductivity of copper
= 215 Btu/hr 2
at the outer surface of the copper tube? ft (°F/ft)

Solution. The heat-transfer rate is found by Therefore

Equation 14.28; because bulk temperatures fluid
are given, the total driving forces and the total 1 1 0.00408 1
+ +
resistances must be employed. UA (400M0.236) (215K0.249) (100X0.262)
Total driving force = 95 - 350 = -255° F
It should be evident from the above equation

Total resistance = —
UA
that the resistance of the copper wall
for two reasons: first, and, second,
it is thin,
is negligible

copper has a high thermal conductivity. Also, by

Equation 14.32
examination, both oil phase and water phase resist-
1 1 1 Ax 1 ances are of the same order of magnitude.
+ +
UA A A Solving for the total resistance
UB A B hA A A kA m h B A B
Basis: 1 ft of tube length
= 0.0106 + 0.0000762 + 0.0382
* See Appendix C for dimensions of pipes and tubes, and
UA
see
= 0.0488 hr°F 2
Appendix D for physical properties of fluids and metals. ft /Btu
 INTERPHASE TRANSFER 309

Thus able solutes contacted with a solute-free liquid or

is

a liquid containing little solute.Mass is transferred

=
-AT 255 from the gas to the liquid at every point along the
<7 = 5230 Btu/hr
1 0.0488 gas— liquid interface at a rate depending upon the
driving force and resistance at each point. For this
UA
operation, assume steady state, simple transfer, and
that true thermodynamic equilibrium exists at the
The temperature of the outer surface of the
interface and the interface itself offers no resist-
copper tube may be calculated by using the single
ance to mass transfer. The equilibrium condition is
resistance associated with the oil film as in Equa-
in some manner common to both phases.
tion 14.29.
It should be pointed out that noticeable in-

-(7-2-7-,) consistencies in some of the assumptions exist. For

5230 = example, in a contacting device packed with
0.0382 ceramic rings, the liquid phase usually flows down
each piece of packing in laminar flow, so that the
-AT = (5230)(0.0382) = 200° F
entire liquid phase should offer resistance to
transfer. The liquid phase is thoroughly mixed as it
or
passes to the next piece of packing. In a stage
72 = 350 -200=1 50° F contactor, the gas phase is usually in laminar flow,
and, because of a relatively short flow path, well-
This problem is solved in a manner identical
developed turbulent velocity profiles do not exist.
with the previous example. Furthermore, there is some evidence (1) that
equilibrium does not exist at the interface. Despite
MASS TRANSFER these limitations, the model is useful, and data
Industrial mass-transfer operations involve the con- analyzed according to the above assumptions are
tacting of two phases in various apparatus. Bubble- extendable to situations that may not necessarily
and sieve-tray columns and cascade, packed, and fit the working model.

spray towers are frequently used. Wetted-wall The problem is to relate the series of
columns have the serious disadvantage of small resistances encountered between the two bulk
transfer area; the equipment employing fine phases with the driving force across these resist-
bubbles or sprays is much more practical. Whatever ances for the stated absorption process. At a partic-
the mode of contacting, two dissimilar phases are ular point in the absorption tower the solid lines of
brought together for transferring mass across the Figure 14.7 illustrate the concentration variation
interface between the two phases. from the bulk gas phase to the bulk liquid phase.
One phase usually flows countercurrent to the Line /— / represents the interface between the two
other phase, with the phases in contact such that phases; the dotted lines indicate the extent of the
an interface exists between them. The solutes being effective resistances. As shown in Figure 14.7, the
transferred from phase to phase must accordingly liquid- and gas-phase solute concentrations have
pass through this interface. Because of the nature unrelated units. Obviously, the concentration must,
of the contacting apparatus, the length of the have continuous profile,
a and if the solute concen-
transfer path or time of contact cannot be tration were expressed in terms of fugacity for the
accurately established, indeed, although a well- liquid and gas phases, the potential curve would be
defined interface exists, it is far from explicit in its a continuous function in Figure 14.7 because the
geometry. However, it can be postulated that this fugacity has the same significance and units for
interface is analogous to the tube wall in heat both phases. However, useful relationships can be
transfer; both phases are kept separated and on developed without the concept of fugacity.
either side of their boundary the transfer-resistance It is evident from Figure 14.7 that the
layers are established. It is in these layers that the transfer path is through two resistances in series so
largest portion of the resistance to mass transfer is that Equation 14.7 applies. Because the transfer
found. mechanism is a combination of molecular and
To establish useful concepts in regard to turbulent processes whose relative effects cannot
resistances and driving forces in a mass-transfer be predicted, it is necessary to employ the
operation, consider the process of gas absorption. transfer-coefficient concept.
/ Here a gas stream containing one or more transfer- The rate of solute transfer from the bulk gas
310 MOLECULAR AND TURBULENT TRANSPORT

driving forces and the liquid-phase coefficient is in

terms of liquid-phase driving forces.
For steady-state transfer the two rates given
by Equations 14.35 and 14.36 must be equal, or

E - (Pai - Pac) (CaL ~ Cai )

(14.37)
1 1

Ak, Ak L
Equation 14.37 is of the form: rate is equal to
the driving force divided by the resistance. Each
side of the equation is made up of a single resist-
ance and a single driving force. To use Equation
14.37, therefore, requires the interfacial composi-
tions which are exceedingly difficult to obtain
experimentally. However, bulk-phase solute con-
centrations are easily and reliably measured. Here
Distance from interface
then is an excellent case for the use of overall
Figure 14.7. Concentration conditions in mass driving forces and resistances.
transfer.
Equations 14.35 and 14.36 may be rearranged
as
phase to the interface is

(Pai ~ PaC -iPai-Pac) = Na Tr\ (14.35a)

Na = (14.35) Ak i G
1
and
Ak G
and the rate of transfer from the interface to the ~(caL - cai )
= Nt
1 14.36a)
lAk L
bulk liquid is

Equation 14.35a determines the rate of

(c aL — Cai )
transfer of solute from the bulk gas phase to the
Na = (14.36)
1 interface, and consequently the driving force is
expressed in terms of gas-phase concentration
Ak L
units. Partial pressure is a useful gas-phase concen-
where tration unit and is solely a gas-phase phenomenon
relating the concentration of the solute in the gas
Na = rate of solute transfer, lb moles/hr
phase to the total gas-phase pressure. Equation
p a = solute partial pressure in gas phase,
atm 14.36a also determines the rate of transfer, but
here the transfer is from the interface to the bulk
ca = solute concentration in liquid phase,
liquid. Consequently, driving forces are in terms of
lb moles/ft 3 of liquid
= transfer area, ft 2 liquid-phase concentration units. Concentration
A
units in the liquid phase employ the two com-
k c = mass-transfer coefficient for gas phase,
2 ponents that make up the liquid — that is, moles
lb moles/hr ft atm
k L = mass-transfer coefficient for liquid solute per unit volume of liquid.

phase, lb moles/hr ft 2 (lb mole/ft 3 Since pa is the concentration of the solute in

)

the gas phase and ca is the concentration of the

subscripts solute in the liquid, Equations 14.35a and 14.36a
G = gas phase cannot be directly combined because they are
/= interface expressed in dissimilar units. A relationship for
L = liquid phase solute concentration at the interface, where the
concentrations of both phases are common, is
It is important to reemphasize the units of the needed before combination of these equations is

mass-transfer coefficients. Note once again that the possible. This relationship is the general phase-
gas-phase coefficient is in terms of gas-phase equilibrium equation, which relates the concentra-
 INTERPHASE TRANSFER 311

tion of the solute in the liquid phase to the or, for simplicity, Equation 14.39 may be
concentration of the solute in the gas phase. expressed in terms of the distribution factor as in
The composition of solute in the gas phase in Equation 14.38a.
equilibrium with liquid of solute composition ca is
Pai ~ mc ai (14.39a)
f(ca (14.38)
Multiplying Equation 14.36 by m yields upon
where rearrangement

ca = liquid-phase concentration, moles m

-(mcaL - mcai = ) Na ; 14.40)
solute/unit volume of liquid Ak,
* = partial pressure of the solute in the
pa
gas phase that is in equilibrium Combining Equations 14.38a and 14.39a with
with ca Equation 14.40 results in

For dilute systems this function is frequently m

linear and becomes -(Pa* ~ Pai) = Na (14.41;
Ak,
p a *=mc a (14.38a)
Equation 14.41 is now written in terms of
where m is a distribution factor relating p a * and equilibrium-equivalent driving forces for the liquid
ca ; it might be pure component vapor pressure as phase. Equation 14.41 may now be added
to Equa-
in Raoult's law, or a Henry's law constant, or the tion 14.35a since the driving force the gas phase in

slope of the equilibrium curve at ca . is expressed in pressure units and the driving force
the stage operations, two phases were
In in the liquid phaseis expressed as the pressure

brought together in a stage, mixed, and separated. equivalent of the liquid-phase composition. Driving
The two phases that resulted from the separation force units are now on the same basis. The addition
were said to be in equilibrium. If one phase of Equations 14.41 and 14.35a for a constant (Na )

composition is known, the composition of the gives

other phase in equilibrium could be calculated
from an equation such as Equation 14.38a. For the
case under study, the two phases are not in equilib-
rium except at the interface. Nevertheless, an or
(Pa* -Pac) = Na [—
.Ak G
— Ak L J
! 14.42)

equilibrium equivalent for each phase may be con-

(Pa PaG
ceived. For instance, through the use of Equation Na =- 14.43)
m
:

14.38 or 14.38a, a vapor composition that would 1

+
be in equilibrium with a given liquid composition Ak c Ak L
could be calculated, and this vapor sample could
conceivably physically exist if conditions were Equation 14.43 is now expressed in terms of
appropriate. This vapor composition is referred to total resistance and total driving force, in gas-phase
as the gas-phase equilibrium equivalent, for a stated units.
liquid composition. In continuous-contacting A similar expression may be written using
equipment the equilibrium equivalents do not exist liquid-phase driving forces,
in the physical sense, but the compositions may be

calculated and plotted on Figure 14.7, as indicated ~\CaL ~~ ca

by the dotted lines. The equilibrium-equivalent
Na = ! 14.44)

composition profile now serves to make contin-

mAkn Ak, .

uous the composition profile from bulk phase to

bulk phase. Although the equilibrium equivalent
Either Equation 14.43 or 14.44 is appropriate
does not exist, it can be used to relate the solute for calculating mass-transfer rates. The choice
concentrations anywhere in the gas and liquid depends upon the information at hand.
phases for particular calculations.
At the interface, by assumption of equilib- Overall Mass-Transfer Coefficient
rium
In the same manner as the total resistance for heat
Pai = fic ai ) 14.39) transfer, a total mass-transfer resistance may be
312 MOLECULAR AND TURBULENT TRANSPORT

defined with common areas A, as manner identical to that for the heat transfer case.
Rearranging the interfacial portions of Equation
1 1
+
m 14.47 yields
(14.45)
AK G Ak G Ak L Pac -Pa, = Ak L kL
and = 14.48)
1 1 1
caL ~ Cai Ak c
-+ (14.46)
AK L mAk G Ak L The reader is referred to Equation 14.19 for
comparison. In the usual mass-transfer equipment
where both phases are in intimate contact, so that the
Kq = overall gas-phase mass-transfer coeffi- transfer areas for each phase may be assumed
cient, lb moles/hr ft
2
atm. equal. Equation 14.48 plotted through point M
KL = overall liquid-phase mass-transfer locates the interfacial compositions p ai and ca at ,

coefficient, lb moles/hr ft
2
(lb mole/ point D. It must be emphasized that point D cor-
3 responds only to bulk conditions (M), since con-
ft )

centrations will vary throughout the apparatus.

Thus, for steady-state transfer,
Using Figure 14.8 and Equation 14.47, the follow-
-(Pa*-Pac) -(caL -ca *) ing can be observed:
N
Al
a =
1 1 In terms of gas phase units:
A Kc AKL 1. The driving force from bulk gas to inter-
face = (p ai -p a G)-
-(Pai-Pac) ~ (caL — Cal )
2. The driving force from interface to bulk
(14.47)
1 1 liquid = [p a * - ai ).

Ak G Ak L 3. The total driving force from bulk gas to

bulk liquid = (p a * -p a c)-
The driving forces and resistances of Equation In terms of liquid phase units:
14.47 may be visualized by referring to Figure 1. The driving force from bulk gas to inter-
14.8. This diagram is analogous to that for heat face = (c a * -c ai ).

transfer given in Figure 14.4. 2. The driving force from bulk liquid to inter-
In this diagram, point M represents the con- face = (caL — cai ).
centration of solute in the bulk phases at the 3. The total driving force from bulk gas to
particular point under study. The gas-phase com- bulk liquid = (c a * — c aL ).
position isp a G and the liquid-phase composition is
c aL The relationship between bulk compositions
.
Any driving force along with the appropriate
and interracial compositions is obtained in a resistance may be used to determine the transfer
rate.
From Equation 14.47
1

Pai - PaG k_c_

(14.49)
Pa* -PaG J_
KG
or

Partial-pressure driving force in gas phase

overall driving force in partial-pressure units

resistance in gas phase

total resistance
and

caL ~ ca kL_
: 14.50)
Solute concentration in liquid (c a ), lb moles/unit volume c aL — ca J_
Figure 14.8. Phase relationships for mass transfer. Kl
 INTERPHASE TRANSFER 313

or and smaller or, in the limit, as m -* 0,

Concentration driving force in liquid phase
(14.51)
overall driving force in concentration units
or from Equation 14.49
resistance in liquid phase

total resistance 14.52)

Equations 14.49 and 14.50 are analogous to

From Equation 14.52 it may be concluded
the equations for heat transfer through series resist-
that the ratio of the gas phase resistance to the
ances. The terms of Equations 14.49
ratio of the
is approximately unity and the gas
total resistance
and 14.50 be constant only if the equilibrium
will
phase may
be said to be the major resistance. The
curve is a straight line. Consequently, the ratio of
same conclusion may be reached by examination
individual phase resistance to total resistance may
of the liquid-phase driving forces. From Figure
vary from point to point throughout the unit. It is
14.9
possible for the transfer resistance to change from
the gas phase being the major resistance to the (c aL ~ cai < ) (c aL - ca *) (14.53)
liquid phase being dominant between the terminals
or from Equation 14.50
of the apparatus.
Some qualitative conclusions regarding the rel-
ative magnitude of the phase resistances and their
— 11
<— ,

14.54
kL K L
influence on the total resistance may be made by
using Equations 14.49 and 14.50 and a constant
and (Mk L )l(MKL ) is very small.
value of k L /kc
Now consider the case where the solute in the
Consider a case where the solute in the gas
gas phase is relatively insoluble in the liquid phase.
phase is very soluble in the liquid phase such that
Figure 14.10 shows the phase behavior of such a
small changes of solute in the gas phase will pro-
system. Once again, the same point conditions of
duce large changes of solute concentration in the
bulk-, interfacial-, and equilibrium-equivalent con-
liquid phase. For this example, the equilibrium
centration are plotted.
curve and bulk composition are illustrated in
Using the liquid-phase driving forces,
Figure 14.9.
Again, let M
represent the concentration of (CaL - Cai )
""*"
(CaL - Ca i
14.55)
the bulk phases at the particular point under study,
or by Equation 14.50
with the interfacial and equilibrium-equivalent con-
centrations also shown. For this case, if Equation
J 1_
14.39a is valid, m
is relatively small, and from
kL ~KL (14.56)

observation of Figure 14.9 as m becomes smaller

CaL

Solute concentration m liquid (c a ) Solute concentration in liquid (c a )

Figure 14.9. Phase relationships for soluble Figure 14.10. Phase relationships for in-

solute. soluble solute.

314 MOLECULAR AND TURBULENT TRANSPORT

The conclusion reached from Equation 14.56 P* = f(Ca)

is that the liquid-phase resistance approximates the

total resistance. In this case, the liquid phase is said
to be the dominant resistance, or, traditionally, the
"controlling" resistance.
To illustrate an earlier statement that the
"controlling" resistance may change from liquid to
gas phase, consider Figure 14.11. This curve is a
combination of the phase diagrams of Figure 14.9
and Figure 14.10.
At the bottom of an absorption tower, for
example, the gas and liquid streams passing will be
rich in solute. This condition is shown at point M.
From the previous discussion, at this point in the
Solute concentration in liquid (c a )
unit the liquid-phase resistance is controlling. On
the other hand, at the top of the column both Figure 14.11. Typical phase diagram.

streams are lean in solute concentration, as

illustrated at point E. This is the exact situation moles per unit volume of liquid; again, the same
depicted in Figure 14.9 and the same conclusions principles apply.
derived from can be made.
it
Example14.3. Air is sometimes dried by a
To summarize the conclusions regarding the
sulfuric absorption process. The humid air
acid
overall mass-transfer coefficient, it may be stated
flows countercurrent to the H 2 S0 4 through a
that just as in heat transfer, the overall coefficient
column packed with suitable packing material. A
is based on the sum of the resistances, and if the
test on a unit as described indicates that air enters
gas-phase transfer resistance dominant, KG may
is
the bottom of the tower with a relative humidity
be used. If the liquid-phase resistance is dominant,
of 50 percent and leaves with a relative humidity
KL is used. The emphasis for deciding which phase of 10 percent. The entering acid at the top of the
is controlling appears to have been placed upon the
column is 67 percent by weight H 2 S0 4 and the
value of the characteristic of the equilibrium rela-
acid leaving is 53 percent by weight acid. Experi-
tions. This not the whole story, however. In
is
mental evidence indicates that k G = 2.09 lb moles/
Figures 14.9, 14.10, and 14.11, k L lkG was held 2
hr ft atm driving force and kL = 0.068 lb moles
constant for illustration. The relative values of the
H 2 0/hrft 2 (lb moles/ft 3 of liquid). For unity
individual transfer coefficients kL and kG are like-
transfer area, a temperature of 25°C, and assuming
wise important. The ratio k L lkG represents the
kG and kL to be essentially constant, determine
ratio of the gas-phase resistance to the liquid-phase
for the terminals of the tower:
resistance. In Chapter 13 it was shown that mass-
transfer coefficients are functions of flow rates of (a) The instantaneous rates of transfer.
the liquid and gas phases and physical character- (b) The percent of total diffusional resistance
istics of the system, including diffusivity. Con- encountered in each phase. Use both gas-
sequently, operating conditions can also influence and liquid-phase driving forces.
the relative magnitudes of the two phase resist- (c) The interfacial compositions.
ances. More be said about the use of overall
will (d) The numerical values of /C G and K L .

mass-transfer coefficients in Chapter 16.

The development of the previous concepts Solution. The equilibrium data at 25°C for
pertinent to mass transfer has been accomplished the H 2 S0 4 — water system are reported by
using the operation of gas absorption as the Wilson (2) and tabulated below.
instruction vehicle. The conclusions reached may Percent relative humidity
be extended to other mass-transfer operations 0.8 2.3 5.2 9.8 17.2 26.8 36.8 46.8 56.8
using suitable transfer coefficients. Distillation and
Weight percent H 2 S0 4
extraction carried out in continuous-contacting
80.0 74.9 70.0 64.9 60.0 55.1 50.0 45.0 40.0
devices are both analyzed using individual and
overall transfer coefficients. It is at times more The equilibrium data as tabulated are not in
convenient to express solute concentration in as convenient a form as could be desired. This is
terms other than partial pressure in the gas and frequently the case. For illustrative purposes, these
 INTERPHASE TRANSFER 315

data will be converted to partial pressure of water Thus, for this particular acid concentration,
in the gas phase and moles of water per unit
/17.2\ /23.75\
volume of To
convert the liquid phase (the
p* =
liquid.
calculation of only one point will be demon- (^)b^) =a00538atm
strated):
The above calculations are repeated using the tabu-
For 60 percent H 2 S0 4 : 40 lb H 2 0, 60 lb H 2 S0 4 lated equilibrium data. The recalculated data are as
Pound moles of water: 40/18 = 2.22 follows and are plotted on Figure 14.12.

Specific gravity of a 60 percent acid = 1.498 p a *, atm 0.000248 0.000721 0.001626

lb moles H 2
Density of a 60 percent acid = (1.498) (62.3) 1.1780 1.451 1.670
3
ft solution
3
5\
=93.5 lb/ft
/93
c a = (2.22) \— '—) = 2.08 lb moles H 2 0/cu ft
0.00305 0.00853 0.0115 0.01464 0.01775

1.88 2.25 2.41 2.57 2.71

solution

The equilibrium relative humidity for a 60

percent acid is 17.2 percent. Entering acid, ca2 =
(^)(^) (1.576)(62.3)
Vapor pressure of water at 25°C = 23.75 mm
= 1 .802 lb moles H 2 0/ft 3 solution
Percent relative humidity = 100

Exit acid, ca1

where =(^)(^j) (1.302X62.3)
pa = partial pressure of water in air
Pa = vapor pressure of water at bulk air
= 2.12 lb moles H 2 0/ft 3 solution
temperature
/ 50 \/23.75\
0.018 Entering
u air, pa1 = = 0.01565 atm
-
i i 1 1
i 1
1 1 1 1

-
^ I 100 A 760 /
0.017

0.016 - M - / 10\/23.75\
Pol "I Exit air, pa2 =
Ha2 1 =0.00313 atm
0.015 - l\ - \ 100/ \ 760 /

| 0.014 - -
Conditions are indicated in Figure 14.13.
1 0.013 _ _
~ For the bottom of the absorption tower
o 0012 - "

o -
p a} = 0.01565 atm
0.011 - Cal*

* 0.010 - — pa1 * = 0.0062 atm (from Figure 14.12)

'
<Pai > 1

° 0.009 - Thus, the total driving force (gas units) =

o /l
as
en
/
/ 1

— 0.01565 -
0.0062 = 0.00945 atm. From the state-
0.008 1

'o 1
ment of the problem and the use of Equation
2 0007 (CaJ l 14.48, -k L lk G = 0.068/2.09 = 0.0325. This is the
VI

2!
o.
0.006
Pol*— y slope of the line relating bulk phase conditions to
those at the interface.
artial o do
(a) The instantaneous rate of mass transfer is
0.004 Cal
P«2
0.003
Pj
<n
1
p, =0.00313 atm
—
r1 *n ca2 = 1.802 lb rnol/ft
3

0.002
— L-
0.001 c <z2
-

—p
1 1 1
1 i i 1 1
1 1

1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 J
pai = 0.01565 atm ^ 1
ca i
= 2 12 lb mol/ft
3
ca , lb mol H2O per ft solution

14.12. Graphical solution for Example 14.3. Figure 14.13. Sketch for Example 14.3.
316 MOLECULAR AND TURBULENT TRANSPORT

calculated using Equation 14.35 and obtaining p ai and

from Figure 14.12.
Liquid-phase resistance
-iPai-Pao) _ -(0-00965 - 0.01565) = 0.370
M MAk c 1/(2.09)(1)
total resistance

2
= 0.0126 lb moles/hr ft One would expect these fractional resistances
to be the same regardless of which set of driving
Using Equation 14.36 should give the identical
forces is employed. The slight difference in this
rate, or
instance is due to errors in reading values from
(CaL— Cgj) (2.12-2.315)
Figure 14.12 and to the curvature of the equilib-
Na =
MAk L 1/(0.068)(1) rium line.

= 0.0126 2 The interfacial compositions have already

(c)
lb moles/hr ft
been used and are (by Figure 14.12)
(b) The fractional resistance across each p ai = 0.00965 atm
phase is calculated from Equations 14.49 and
14.50 with the aid of Figure 14.12. cai = 2.305 lb moles H 2 0/ft 3 solution

In gas-phase units (d) The overall coefficients KG and KL can

be evaluated by several methods.
Gas-phase resistance = p aG — p ai (1) From Equation 14.47
= 0.01565-0.00965
= 0.006 atm -
NM 0.0126
Kn =
A[p a *-paC 1x0.00945
p ai — pa
* )
Liquid-phase resistance =
= 1.335 2
= 0.00965 - 0.0062 lb moles H 2 0/hr ft (Ap)

= 0.00345 atm and

p aG — p a
*
Total resistance, both phases = rV. 0.0126
K, =
= 0.00945 atm A{c a -c a *) 1x0.5
Thus, = 0.0252 lb moles H 2 0/hr ft
2
(Ac)
Gas-phase resistance 0.006
= 0.635 (2) An attempt to use Equation 14.45 and
total resistance 0.00945
14.46 for obtaining Kq and KL will be made.
and Since a point slope at (c ai p ai must be used, good , )

Liquid-phase resistance agreement is not expected.

= 0.365
total resistance m = 0.01955 (slopeatc a andpa,) ,

In liquid-phase units From Equation 14.45

Liquid-phase resistance = cai - ca
= 2.305-2.12 +
m _ 1
+
0.01955
„^o =0-766
Kg kG k L ~ 2.09 0.068
= 0.185 3
lb moles/ft
2
— ca
and KG = moles H 2 0/hr
1 .308 lb ft (Ap).
Gas-phase resistance = ca *
,
From Equation 14.46
= 2.62 - 2.305
= 0.315 lb moles/ft 3
Jl_
-+
1
—1 1.0 1

= 39.1
KL mkc k L [0.01955M2.09) 0.068
Total resistance, both phases = {ca * - ca )

and KL = 0.0256 lb moles/hr ft

2
(Ac).
= 0.500 lb moles/ft 3
The agreement between d(1) and d(2) is

Thus, better than one can usually expect.

Exactly the same procedures as above are
Gas-phase resistance 0.315
= 0.630 used to calculate the conditions at the top of the
total resistance 0.500 tower. In this case, only the results will be written.
 INTERPHASE TRANSFER 317

For the top of the tower dimensionless parameters (see Table 11.1). In
Chapter 11 transfer was limited to a single phase
pa2 = 0.00314 atm ca2 = 1.802 lb moles H 2 0/ft 3
where the condition at the phase interface was
pa/ = 0.00255 atm cai = 1.818 lb moles H 2 0/ft 3 known. In this case the relative-resistance term (m)
pa * = 0.00239 atm ca * = 1.885 lb moles H 2 0/ft 3 was equal to zero. For m to be zero requires a
negligible surface resistance to transfer, correspond-
(a) Instantaneous rate ing to an infinite value of the surface coefficient.
In the practical case of either heat or mass transfer
(Pai -Pac) _ (0.00255-0.00314)
Na ~~ the values of the surface coefficients are finite, and
MAkG 1/(1)(2.09) the surface temperature or concentration is differ-
= 0.001235 lb moles/hr ent from that of the surroundings.
Knowledge of surface resistances now makes
(b) Using gas units it possible to solve Equation 11.16 in terms of
more practical boundary conditions. For geo-
Gas-phase resistance 0.00314 - 0.00255 metries discussed in Chapter 11 having a finite
= 0.788
total resistance 0.00314 - 0.00239 surface resistance, corresponding to a definite value
of 6 the boundary conditions become: T =
, r at
Using liquid units
= 0; T = T 2 at 6 = °°; at steady state
Gas-phase resistance 1.885—1.818
= 0.832
total resistance 1.885 - 1.802 = ^(r, -r 2 )
V3x'
(c) p ai = 0.00255 atm
3
cai:= 1.818 moles/ft solution
T2 is the average concentration of transferent
property in the surroundings.
(d) KG = 1.54 lb moles/hr ft
2
(Ip) This interface condition simply states that the
rate of transfer from within the solid to the surface
KL = 0A8 lb moles/hr ft
2
(Ac)
must be equal to the rate of transfer from the
surface to the surroundings. For heat transfer the
Note how different the rates of absorption are
at the tower ends. Examination of Figure 14.12
interface condition is kCdTlbx) = h(T^ 2 ), or —T
the rate of heat transfer by conduction to the
and comparison of (p a * — p aG at the terminals )

surface equals the rate of heat transfer by convec-

show that the driving force at the bottom of the
tion away from the surface. Similarly for mass
tower is considerably less than that at the top of
the tower. The gas-phase resistance is also higher
transfer, the interface condition is 9 (del
3x) = kc (c ai -c c2 ).
at the top of the tower than at the bottom.

If conditions with regard to driving forces, Example 14.4. A cylindrical steel shaft 4 in.

resistances, and rates vary from position to posi- in diameter and 8 ft long is heat treated by quench-
tion within the tower, some means for accounting ing it in an oil bath. The shaft is heated to a
for these variations must be found, so that they uniform temperature of 1100°F and plunged into
=
may be related to the size of the equipment needed an oil bath maintained at 300 F. The surface co-
2
to do a specified transfer. This will be considered efficient of heat transfer is 150 Btu/hr ft °F. Cal-
later in this chapter. culate the temperature of a point 1 in. from the
center after 2 min of quenching time. Use a value
2
of a = 0.35 ft 2 /hr and k = 26 Btu/hr ft (°F/ft).
UNSTEADY-STATE TRANSFER BETWEEN PHASES

The general unsteady-state transport equation has Solution. This example is similar to Example
been developed in Chapter 1 1 11.1, with the exception that the oil bath is main-
tained at a constant temperature of 300° F rather
2
^= 5
a r
(11.16)
than the shaft surface being held at a constant
temperature. Because of the resistance to heat
be Lax 2 J

transfer between the oil and steel, the temperature

Equation 11.16 has been solved for several simple at the surface of the steel will be somewhat higher
geometries and boundary conditions in terms of than the bath temperature.
318 MOLECULAR AND TURBULENT TRANSPORT

Figure 1 1.4 will be used for this solution. For

= 2 min u rw
c

=
T2 -T _ 300- T o
~-2

Y
T2 -qT ~300- 1100 c

ad (0.35X2/60)
K
c
D

>. In -
p\TaseJ^->r-^' — r^
= 0.42 9Vv3^-^'
2 2 a
(2/12) c
a
c-
n = n
x7~2 c
n
t—
k 26
m = = 1.04
1 Length of contactor &y 2
/»x, (150)(2/12)
Figure 14.14. Concentration-length profile in
From Figure 1 1.4,
process equipment.

= 0.58 =
300 -T
Y 14.14 the terminals of the contactor
In Figure
300- 1100
are designated as 1 and 2. The overall driving force

and r=764°F has been established as being the difference in the

concentration of transferent property between the
This temperature may be compared with the
two bulk phases. Note that in Figure 14.14, the
value of 364° F at the same position for the solu-
overall driving force is continuously varying from
tion to Example 11.1.
terminal to terminal. It is quite likely that the rate

of transfer will also vary along the length of the

THE DESIGN EQUATION equipment. Consequently, to calculate a rate of
transfer for the entire apparatus will require an
Thus far the transfer of heat or mass has been
integration of the general rate equation. Although
focused upon point conditions within a transfer
a knowledge of the rate of transfer is important, an
apparatus. An understanding of driving forces and
equally important relationship is needed between
resistances at a point permits the evaluation of the
the rate and length of contactor, or total transfer
rate of transfer only at that point. A need now
area.
exists for relating the rate of transfer at all points
Figure 14.15 represents an apparatus that is
within the apparatus to the total transfer area
suitable for the general contacting of phases for
required for process equipment. Specifically, it is
exchange of heat or mass. For this apparatus,
desired to know how much heating surface is

required for a specified heating assignment or how

tall a contactor should be for a given separation in
mass transfer. The answer to these questions is best
formulated as a design equation that will relate the
rate of transfer across the total transfer area to the
total change in transferent property. Further
details of design will be considered in a later
chapter.
Consider two phases in countercurrent flow
undergoing an interchange of a particular trans-
ferent concentration (T). The phases are kept
separated either by solubility characteristics as in
mass transfer or by solid boundaries as in a heat
exchanger. As these phases pass each other
between the terminals of the contacting equip-
ment, one phase will become depleted in the
quantity of transferent property while the other
V=volumetric flow rate of phase V, ft 3 /hr
L = volumetric flow rate of phase L, ft 3 /hr
Y v = transferent-property concentration in phase
3
V, quantity of transferent property per ft
of V
VL = transferent-property concentration in phase
3
L, quantity of transferent property per ft
of L
subscripts: (1) = tower bottom
(2) = tower top
Consider a differential section of this unit
through which the two phases pass countercurrent
to each other. This section will include a transfer
area dA. Across this section, phase L will be
changed in amount by dL and in composition by
oT L and phase Vwill be changed by quantity d\J
,

will become enriched. This is illustrated in Figure and by composition dT v .

14.14. For steady-state operation, a material balance

 INTERPHASE TRANSFER 319

Equation 14.60 defines the design equation

] £ for equipment where the rate of
operational
transfer, rather than the amount of enrichment
V+dV L+dL
I> + dI\T rL +drL attained at equilibrium, dictates the size of equip-
ment. Since the area must be a positive number,
Equation 14.60 will adjust itself to account for the
W///////A
transfer proceeding toward a low potential. Inte-
gration of Equation 14.60 between the terminals
of the apparatus will give the total transfer area
required. Thus

2 cr(IVV) 2 d(T L L)

iVi rL1 f>-{ 1 2(Y V -T L )

-{ 1 S(T V -T L )

Figure 14.15. Continuous contactor.

over the differential section yields

= -{-${Td(^A)
-T
2
(14.61)
V L )

d\/ = dL (14.57) The corresponding design equation for heat

and a component balance over the same section transfer is

d(r v V)=d(r L L) (14.58) d(Vc P pi)

pT) A 2 a\Mc
d{Vc pP
rt _ rz
•2 a\\ic r2 P pT)
i) B l
=
Equation14.58 is referred to as the K J
1 U(TA -TB )
J
1 U(TA -TB
operating-line equation, which relates the composi-
dq
tions and flow rates of the two phases passing each
other at a particular point in the apparatus. This is
-[ 1 U(TA - TB
! 14.62)

the same operating line that appeared in Part One

of this book.
where
Examination of Equation 14.58 shows that ^a Vg = i
hot and cold phase volumetric
each term is a rate of addition or depletion of flow rates, ft 3 /hr
transferent property and is really a statement of cpa . Cpb = hot and cold phase specific heats,
the rate of transfer of transferent property across Btu/lb°F
Pb = hot and cold phase densities, lb/ft
the total transfer area. Accordingly, Equation 3
Pa
14.58 must therefore be equal to the total driving TA,Tg = hot and cold phase temperatures,
force divided by the total resistance. Or, by °R
analogy with Equation 14.1,
For mass transfer, Equation 14.61 may be
written as
=
-OV-IY)
d(r v V)=d(r L L)=d{\!;A)
MIdA d(Mca 2 d(Lca

! 14.59)
J
1
J
1
2

K(ca
)

-1 1 K(c a
)

-c a *)
where
(14.63)
— (IV — F L )
= total driving forcebetween
bulk phases V and L at the where
differential section
V= volumetric flow rate of phase V, ft 3 /hr
3f= overall transfer coefficient
L = volumetric flow rate of phase L, ft 3 /hr
The area term in Equation 14.59 is of partic- ca = solute concentration (must be consistent
ular interest since it is a direct measure of the size with V and L), lb moles/ft 3
of the equipment that is needed. Solving Equation K= overall mass-transfer coefficient (must
14.59 foraW gives be consistent with V and L), lb moles/hr
2
ft (lb moles/ft 3 )

d(T v V) d(T L L)
dA = -
*(r„-r*) s{T v -rL The design equation for momentum transfer
14.60) takes a different form from Equation 14.61 since
d[jjA)
momentum transfer isnot an interphase phenom-
s(T v - r. enon.Momentum is simply transferred through a
320 MOLECULAR AND TURBULENT TRANSPORT

fluid to the boundary where a stress is established. of glass wool. Calculate the steady-state heat flux. Thermal

The design equation simply becomes a force 45 W/nvK. for

conductivities for the materials are: for steel

balance, or the plywood 0.21 W/nvK, and for glass wool 0.056
:
W/nvK. The cold face is at 40 F and the hot is at 100"F.
d(APgc S) =d(T y gc A) = -d( S 7 vpA) (14.64) 14.4. A furnace wall is exposed to hot gases at

where 1100"K. The wall consists of 0.12 m of fire brick and

S = the cross-sectional 3rea, ft

2 0.25 m of common brick. Heat transfer coefficients on the

•-=the momentum-transfer coefficient, hot side are 3000 W/m 2 «K and 22 W/m 2 -K on the outside.
Ambient air is at 300 K.What is the heat transfer rate per
ft/hr
= the stress area, ft 2 square meter of wall, and what is the temperature at the
A
= the pressure applied to S, Ibf/ft 2 interface of the two bricks?
AP
2 14.5. Molten bismuth flowing through a 2-in I.D.
Ty = shear stress at the boundary, lb//ft
is

type 304 stainless-steel pipe that is insulated with a 2-in

It is apparent that the solution of the heat- layer of molded diatomaceous earth = 0.15). An
Or ad-
and mass-transfer design equations is quite com- ditional layer of laminated asbestos felt 1 in. thick is to be
plex and depends upon evaluation of resistances placed around the pipe and diatomaceous earth insulation
and driving forces at various points throughout the to reduce the surface temperature of the complete assembly
apparatus. In the general case, integration of Equa- to 200" F. If asbestos felt is not recommended for use above
tion 14.62 or 14.63 is likely to be done graphic- 500" F, is this addition feasible? The inside surface temper-
ally, since the probability that an analytical rela- ature of the tube is 650° F, and the tube wall is 0.15 in.
tionship exists among all the variables involved is thick. Use a thermal conductivity for the asbestos felt equal
small. to 0.044 Btu hr 2 f F/ft).
The next few chapters devote consider- will
14.6. The addition of insulation to a pipe increases
able detail to the solution of these design equations the resistance of transfer path and hence decreases the heat
along with descriptive material pertinent to an un- loss from the hot fluid within the pipe. Under certain
derstanding of the process equipment being used. conditions the addition of too much insulation may cause
an increase in the heat loss since the effect of added
REFERENCES heat-transfer surface offsets the advantages of insulation.
Show that the critical thickness of insulation occurs when
rc = k/h, where rc is outside radius of insulation where heat
1. Schrage, R. W., A Theoretical study of Interphase Mass
loss least, h heat-transfer surface coefficient, and k is
Transfer, Columbia University Press, New York (1953). is is

2. Wilson, R. E. Ind. Eng. Chem., 13. p. 326 (1921 1.

thermal conductivity of the insulation. List the simplifying
assumptions made.
14.7. A furnace wall exposed to hot gases at 1400" F
PROBLEMS
is constructed of 4 in. of fire brick and 10 in. of common
brick. For safety purposes it is desirable to keep the outer
NOTE: In ail problems involving thermal resistances in SL.face wall temperature at 160" F in the surroundings at
series, assume perfect contact between the elements. 80" F. Can an additional layer of magnesia insulation be
14.1. Heat is flowing from steam on one side of a used if it cannot withstand temperatures greater than
vertical steel sheet 0.375 in. thick to air on the other side. 650" F? How thick a magnesia layer would you recommend?
The steam 1700 Btu/hr 2 °F Assume
heat-transfer coefficient is ft flame-side heat-transfer coefficient is 12.0 Btu/hr
and that of the air is 2.0. The total temperature difference ft
2 "F and the room-side coefficient is 4.0.
is 120"F. How would the rate of heat transfer be affected 14.8. Water flows through a 1 in. sch-40 steel pipe at
by making the wall of copper rather than steel? By in- a velocity of 3 ft/s. At a point along the flow path the
creasing the steam coefficient to 2500? By increasing the water is at a bulk average temperature of 100" F. Outside
air coefficient to 12.0? Express results in percentages of the the pipe saturated steam at 5 psig is condensing. If the
rate for present operation. condensing heat transfer coefficient for the steam is 1500
2C
14.2. If a scale layer 0.003 in. thick is deposited on Btu ir ft F, calculate:
the steam side of the steel sheet of Problem 14.1, what (a) The overall heat transfer coefficient based upon
would be the rate of heat transfer. Thermal conductivity of the inside surface of the pipe.
the scale 0.5 Btu/hr 2 (°F -
ft (b) The overall heat transfer coefficient based upon
14.3. The wall of a refrigated truck body consists of the outside surface.
a 1.6 x 10~ 3 outer steel sheet and a 9.5 x 10 -3 m plywood (c) The percent of the total temperature drop that
inner wall. Between the two materials is a 1 x 10
_1 m layer takes place across each resistance.
 INTERPHASE TRANSFER 321

(d) Repeat part (c) for the same size pipe, but made 14.15. A vapor mixture of ethyl alcohol and air is

from 18— 8 stainless steel. separated from nitrogen bymeans of a membrane

14.9. Repeat Problem 14.8 except that the fluid permeable only to the alcohol. The alcohol vapor is
flowing through the system is a fluid whose density is 55 diffusing at a constant rate through the membrane into the
lb/ft 3 and viscosity is 95 centipoise. nitrogen area. The air— alcohol mixture analyzes 1 percent
The inside surface temperature of a hollow
14.10. by volume ethyl alcohol, and the nitrogen— alcohol phase
silver sphere is 500 K and the outside surface temperature is analyzes 0.01 percent alcohol. It is estimated that on both
200 K. The wall thickness is 0.013 m. What is the heat rate? sides of the membrane a 1 mm thickness of gas exists and
For silver use Ar= 430W/m-K. Use r-, =0.012m. that the membrane offers no resistance to transfer. A con-
14.11. The term Thermopane is given to a window stant temperature of 25 C and a pressure of 1 atm prevails.
that in certain cases consists of two layers of glass, each 1/4 Calculate (a) the rate of diffusion of ethyl alcohol and (b)

in. thick separated by a layer of stagnant air, also 1/4 in. the concentration of ethyl alcohol at the membrane.
thick. If the temperature drop (surface-to-surface) across 14.16. Whitney and Vivian [Chem. Eng. Prog., 45.
the composite system is 30°F, find the heat loss through p.323 (1949)] studied the absorption of SO2 in water in a

the system per ft

2 . column packed with 1-in. ceramic rings. Their results show
If this window replaces a single glass pane, 1/4 in. that
thick at a difference in cost of $120, find the number of 82
c a = 0.044G^
-
/r
days of operation under the stated temperature-drop con-
ditions to pay for the window. Coal with a heating value of k G a = 0.028G v 0J G L - 25

13,000 Btu/lb costing $48 per ton is burned in a furnace

where
having an efficiency of 40 percent.
2
14.12. Derive an expression for the overall heat- Gi_ = liquid rate, Ib/hr ft of tower cross section
transfer coefficient for the following case of a flat furnace G\/ = gas rate, Ib/hr ft 2 of tower cross section
a = ft
2 of interfacial area/ft 3 of packing
wall in the x direction.
k G a = liquid-phase surface coefficient, lb mole/hr ft 3

(lb moles/ft 3 )

k G a = gas-phase surface coefficient, lb mole 'hr ft 3 atm

1 1

Note in this case,

KGa kGa k ca

An air— SO2 mixture is to be contacted counter-

currently with water at 20°C and. 760 mm total pressure

under such conditions that the findings of Whitney and

Vivian are valid. The gas flow rate 200 2 and
is to be Ib/hr ft

the water rate is 6000 Ib/hr ft 2 . At a particular point in the

absorber the gas phase analyzes 0.12 mole fraction SO2
while the liquid phase is 0.002 mole fraction SO2.
14.13. A cylindrical rod of diameter D-, and thermal (a) Determine the local mass-transfer rate.

conductivity of k-\ is covered with two different insulating (b) What percent of the resistance to mass transfer is

materials of diameters D2 and D3, respectively, and con- in the gas phase?
ductivities of Ar2 and ^3. If one end of the system is (c) Repeat parts (a) and (b) for the case where the
maintained at a temperature T-\ and the other at T2, where gas mass velocity is doubled.
7"i > T2. derive an expression for the heat transfer rate (d) Repeat parts (a) and {b) for the case where the
assuming heat flows axially only. liquid mass velocity is doubled.
14.14. A small electric furnace is 6 by 6 by 12 in. 14.17. Benzene and toluene are being separated by
inside dimensions and has fire-brick walls (/r=0.65) 2 in. distillation in a packed column. At a point in the distilling
thick. The front of the furnace is a movable wall that column, the gas phase is 63 mole percent benzene, while
permits entry into the furnace. In this section is a 2 by 2 by the liquid is 50 mole percent benzene. The local mass-
The inner 2 85 percent of the
3 in. quartz observation window (Ar= 0.040). transfer rate is 0.05 lb moles/hr ft . If

surface temperature for all sides is 1200 F, and the outer resistance to mass transfer is in the gas phase, determine the

surface temperature is 250° F. Assuming all joints perfectly interfacial compositions and the local mass-transfer co-

made and neglecting the influence of the corners on the efficients.

temperature distribution, what is the heat loss from this 14.18. Solve Problem 11.18 such that the sphere is

furnace? placed in a surroundings of 100 F. A thermal conductivity

322 MOLECULAR AND TURBULENT TRANSPORT

of k = 3.3 Btu/hr ft 2 (°F/ft) is valid for this material. Plot (b) If the cost of the heat lost is 5 cents/kWh, what is

the radial temperature profiles if: the worth of this heat loss per season?
(a) The sphere is placed in air at 100°F. Use h = 2.0 (c) If a 50 mm layer of styrofoam, or urethane foam,
2 °F.
Btu/hr ft thermal conductivity 0.05 W/m K is installed in perfect
(b) The sphere is placed in water at 100 F. Use thermal contact with the wood of the roof, what is the
2
h = 20 Btu/hr ft °F. saving in $/season, 100 m2 ?
14.19. A reflecting mirror for a telescope is to be 14.22. A refrigerant pipeline of 20 mm O.D. copper
relieved of stresses by annealing in an oven in preparation for is insulated with a 20 mm thickness (radial) of slag-wool in

grinding. The mirror is a disk of plate glass 12 in. in perfect thermal contact. If the pipe is at — 20° C outside
diameter by 1 in. thick and is at a uniform temperature of surface, and the surroundings at 20°C, what is the heat loss,

75°F before being placed in the annealing oven. How long W/meter of tubing? Take h = 5 W/m2/C Because of a poor
will the disk have to remain in the oven so that the surface installation job, and abuse over several years, the slag-wool
of the disk reaches a temperature of 750°F? The air temper- has become ice-loaded, so that is has the conductivity of
ature in the oven is 790°F, and the heat-transfer coefficient ice. What is the heat loss in ratio to the original?
2
is 5.0 Btu/hr ft F. For the particular glass being used, 14.23. One of the copper finned-tubes for air-
2 (°F/ft) and 2 /hr.
k = 0.40 Btu/hr ft a = 0.015 ft conditioning tested by Beatty [Trans. ASME (1945)] had
14.20. A steel sphere 4 in. in diameter is at a uniform an inside area of 0.045 m 2 /m of tubing, and a total outside
temperature of 900°F. This sphere is to be quenched in a area of 0.448 m 2 /m of tubing. Cold water at 5°C flowed at
water bath at 100 F. A thermocouple is embedded in the 3 m/s, resulting in an inside h of 3407.0 W/m 2 /C. The
center of the sphere and after a period of 10 min the overall coefficient was 48.0 W/m 2 of outside area K. If the
temperature at the center of the sphere is 325°F. Estimate thermal resistance of the tube-metal was negligible, what
the value of the heat-transfer coefficient between the was the outside air coefficient?
sphere and the water bath. 14.24. Water at 1.22 kg/s is being heated by steam
14.21. The roof of a house consists of slate shingles condensing inside a finned-tube. The inside area is 0.045
fastened to "one-inch" edge-grain pine boards. The average m 2 /m of tube, and the outside is 0.45 m
2 /m. The
steam
thickness of the slate layer is 5 mm, and of the board, coefficient is 17034.0 W/m 2 inside area K, and the overall

15 mm. Assume perfect thermal contact between them. coefficient is 1749 W/m 2 inside area K. What is the water-

(a) If the average outside temperature 0°C and the is film coefficient, W/m 2 outside area K, if the metal
attic temperature is 15 C, what is the heat loss in kWh/100 resistance is negligible?

m2 for a heating season of 4000 hr? Assume an outside h of

11 W/m 2 /C, and an inside one of 3 W/m 2 /C
 INTERPHASE TRANSFER 323

PART II. NOTATION AND NOMENCLATURE

Wherever a variable in the table below has physical ky,k y gas-phase mass-transfer coefficient [M/8Ly,Lz (M/
dimension, the dimensions as well as typical units will be M)] (Table 13.3)
stated. ky.ky gas-phase mass-transfer coefficient [M/8L K LZ (M/
M)]

Dimensions K constant
Kq over-all gas-phase mass-transfer coefficient [M/
Fy Force, with subscript indicating direction of 8LK LZ (F/L 2 )] , lb moles/hr sq ft atm
application K[_ over-all liquid-phase mass-transfer coefficient
H Thermal energy [M/eL^M/LjfLyL;,)] , lb moles/hr sq ft (lb moles/
Lj Length, with subscript indicating direction cu ft)

M Mass I mean free path (Ly), cm

T Temperature L length (Lj), ft
8 Time L volumetric flow rate of phase L, (L x L K L z /8),
cu ft/hr
i-e entry length (Ly,), ft
Symbols
m total mass (M), lb

A transfer area (L^L^), sq ft m Slope of equilibrium curve (Equation 14.39a)

b wetted perimeter (L), ft m parameter (Table 11.1)
c concentration (M/LX L K LZ ), lb moles/cu ft m molecular mass (M), gm
c mean speed (L//8), cm/sec M molecular weight (M/M), lb/lb mole
Cp heat capacity (H/MT), Btu/lb°F M moment (FL,-), ft-lbf (Chapter 10)
CD drag coefficient M momentum
D diameter (L r ), ft Jt modulus (Chapter 1 1)

9 mass diffusivity (Lx 2 /8), sq ft/hr n number of moles

E eddy diffusivity (Lx 2 /8), sq ft/hr n parameter (Table 11.1)
4 transfer coefficient N rate of mass transfer (M/8), lb moles/hr
f friction factor, dimensionless N number of rows of tubes
f friction factor (f = 4f') NGr Grashof number (L 3p 2g§ AT/fi 2 ), dimensionless
F force (Fy), \bf /V Nu Nusselt number (hD/k), dimensionless
2 2 number (Dvp/cpk), dimensionless
9 acceleration of gravity (L/8 ), 32.2 ft/sec Wpe Peclet
2 32.174 2 number (cpp/k), dimensionless
9c dimensional const (MLy/8 Fy), lb-ft/sec \b f A/ Pr Prandtl
G mass velocity (M/Ly- 2 8), lb/sq ft, hr Nsc Schmidt number (pIpQ), dimensionless
<$ volumetric generation rate, quantity of transferent /Vst Stanton number (h/cppv), dimensionless
property/Lx Lj,Lz 8 /V Re Reynolds number {Dvp/p.), dimensionless

heat-transfer coefficient (H/8L^LZ T), Btu/hr sq P partial pressure (F/L 2 ), psi

*
ft°F P equilibrium partial pressure (F/L 2 ), atm
H3 Henry's law constant (Appendix D) P total pressure (F/L 2 ), psi

latent heat of vaporization (H/M), Btu/lb mole Q rate of heat transfer (H/8), Btu/hr

IN Colburn mass-transfer factor (Equation 13.90) r radius (L,-), ft

Colburn heat-transfer factor (Equation 13.86) R resistance to transfer

k thermal conductivity [H/8L K L Z (T/L X )] , Btu/hr R gas constant (Appendix A)

sqft (°F/ft) s surface renewal factor (Equation 13.124)
mass-transfer coefficient [M/8Ly,Lz (M/L x L K L z )] S area in flow direction (L X LZ ), sq ft

lb moles/hr sq ft (lb mole/cu ft) T temperature (T), °F or °R

l<G> kG gas-phase mass-transfer coefficient [M/8L / LZ (F/ T* reduced temperature (T* = kT/e)
L 2 )] (Table 13.3) TB normal boiling point (T), °F
k Ll k L liquid-phase mass-transfer coefficient [M/8Ly, L Z (M/ u velocity in the x-direction (L x /8), ft/sec
x L y Lz )] (Table 13.3)
\- u + generalized velocity (u + = vlu*), {\/L y /L r )

*X' k x liquid-phase mass-transfer coefficient [M/8Ly,L z (M/ u friction velocity (u* = \f(Ty gc )ylp), VL K L r /82 )

M)] (Table 13.3) ft/sec

kx.kx liquid-phase mass-transfer coefficient [M/8L / L 7 (M/ fluctuating velocity component in the x-direction
M)] (Table 13.3) (L x /8), ft/sec
324 MOLECULAR AND TURBULENT TRANSPORT

PART II. NOTATION AND NOMENCLATURE Contd.

U over-all heat-transfer coefficient (H/6L L ? T)

/ p density (M/LX L/ LZ ), Ib/cu ft

v velocity in the /-direction (Ly/8), ft/sec a molecular diameter (L r ), cm

v fluctuating velocity component in the /-direction a Stefan— Boltzman constant
(L„/8), ft/sec t stress (F/L 2 ), lb/-/sq ft
V volume (L x Ly,Lz ), cu tt T
y
shear stress on a fluid (F^/Ly Lz ), lbf/sq ft

V volumetric flow rate of phase V (L x LyLz /8), \j/ generalized flux of transferent property, quantity
cu ft/hr per unit time and unit area
w velocity in the z-direction (Lz /6), ft/sec cj angular velocity (1/8), 1/sec

w mass rate of flow (M/8), Ib/hr ft Lennard-Jones collision integral

w fluctuating velocity component in the z-direction
(Lz /8),ft/sec Subscripts
x direction of molecular transport a component
x mole fraction, liquid phase (M/M) am arithmetic mean
Xf film thickness (Lx ), ft A phase A
X mole ratio (M/M), dimensionless b component
X parameter B phase B
y mole fraction, gas phase (M/M), dimensionless B boiling
y direction of flow c component
y
+
generalized position [/
+ =
(ri — r)pu*lp], {\jL t /L r ) c critical, transition
Y mole ratio (M/M) D drag
Y parameter e eddy, turbulent
2 direction of flow f film
3 generalized over-all transfer coefficient (L x /8), fs free stream
ft/sec i interface
a absorptivity component, phase, position,
/ general etc.
a thermal diffusivity (a = klcpp) (L x 2 /8), sq ft/hr / liquid
|3 volumetric coefficient of thermal expansion 1/T, Im logarithmic mean
1/°F m molecular, laminar
7 ratio (Equation 13.47) N mass transport
7 surface tension (F/L) q heat transport
T generalized concentration of transferent property, r any radial position
quantity per unit volume t total
5 generalized transport diffusivity (5 = lie), (L x 2 /8), v vapor
sq ft/hr x position in x-direct ion
5 boundary-layer thickness (Lx ), ft
/ position in the /-direction
A increment e electrical
e effectiveness of heat transfer 6 time
e emissivity r momentum transport
elk potential parameter, °K center position
6 mean travel time (6 = lie), (6), sec 1 position
X Prandtl mixing length (Lx ), ft 2 position
A flow rate based upon circumference (M/L8), Ib/hr 3 position
ft (Equation 13.143)
p. absolute viscosity (MLX /L K LZ 8), Ib/ft-hr Specialized subscripts are included only in the general
M a apparent viscosity of a non-Newtonian fluid notation.
(MLX /LK LZ 8), Ib/ft-hr An overbar on a variable designates an average.
v momentum diffusivity (v = pip), (L x 2 /8), sq ft/hr A prime on a flux designates the value per unit of area.

Applications to Equipment Design
In the next several chapters, the fundamentals of
heat, mass, and momentum transfer will be applied
to the design of industrial equipment. As it is used
here, design in most cases will be limited to the
evaluation of the transfer area required. This
evaluation is only the first step in a complete
design. The detailed process and plant design that
must follow the determination of transfer area will
not be considered here. Such a design must include
complete specification of mechanical details, ma-
terials of construction, electrical components, pro-
cess control and operation, and many other factors.
In addition, a detailed economic analysis must be
made on the many process alternatives to deter-
mine which is the optimum — that is, which will
produce the product at the lowest unit cost.
The engineer may be asked to design
equipment or to evaluate and suggest improve-
ments in the performance of existing equipment.
The calculation methods considered in the succeed-
ing chapters may be applied to either the synthesis
of new processes or the analysis of existing pro-
cesses. Although the fundamentals covered in part
two are completely appropriate in describing the
transfer phenomena under study, it is in
practical application that many compromises must
be made. Often the theoretical development under-
lying a particular application is incomplete. This
lack of complete understanding leads to the use of
empirical equations most commonly based upon
Part Three
experimental data taken in the laboratory or pilot
plant. An engineer asked to design a plant must do
it with the data available or recommend what
additional data are needed. The engineer cannot
expect a complete and precise theoretical basis for
the proposed process therefore and must be willing
and able to use empirical methods and approxi-
mations based upon experience whenever the theor-
etical background is inadequate to do the job.
There is no such thing as the permanently
perfect design, because of the possibility of new
and cheaper materials of construction or because of
a shift in either the technical or economic speci-
fications under which the equipment was originally
designed. It is necessary that the design be tech-
nically sound, economically close to the optimum,
new and cognizant of safety and working conditions for
operating personnel. In many plants, the value of
continuing production justifies reasonable expendi-
tures to avoid interruptions.
The design and operation of any chemical
plant require the close cooperation of specialists in
many technical and nontechnical fields. Engineers
must understand and appreciate the problems of
their their coworkers whenever they influence her or his
own area of specialization.
This section will consider the unit operations
based upon heat transfer, mass transfer, simul-
taneous heat-and-mass transfer, and momentum
transfer.
325

Heat Transfer
The chemical process industries all utilize to a very
large extent the transfer of energy as heat. The
purpose of this chapter is to apply the design equa-
tion as it has been developed to the practical solu-
tion of industrial heat-transfer problems. Two
mechanisms of heat transfer have already been
established:
Molecular — the transfer of heat by molecular
action; this is referred to as conduction.
Turbulent — the transfer of heat by a mixing
process; this is usually referred to as convection.
It has been established that these two mech-
anisms can occur simultaneously or individually.
A mechanism of heat transfer that is
third
common but has not yet been discussed is radiation,
which is the transfer of heat by emission and
absorption of energy without physical contact.
Unlike conduction or convection, which depend
upon physical contact for thermal-energy transfer,
radiation depends upon electromagnetic waves as a
means for transferring thermal energy from a hot
source to a low- temperature sink. Radiation may
occur simultaneously with, or independent of, the
other two mechanisms of transfer.
INDUSTRIAL HEAT-EXCHANGE EQUIPMENT
The heat exchangers used by chemical engineers
cannot be characterized by any one design; indeed,
the varieties of such equipment are endless. How-
ever, the one characteristic common to most heat
exchangers is the transfer of heat from a hot phase
to a cold phase with the two phases being separ-
ated by a solid boundary.
327
Fifteen
Double-Pipe Heat Exchangers
The simplest type of heat exchanger is the double-
pipe heat exchanger shown in Figure 15.1. The
double-pipe heat exchanger is essentially two con-
centric pipes with one fluid flowing through the
center pipe while the other fluid moves cocurrently
or countercurrently in the annular space. The
length of each section is usually limited to standard
pipe length, so that if an appreciable heat-transfer
surface is required, banks of sections are frequently
used. If the required area is too large, a double-pipe
exchanger is not recommended. The use of a double-
pipe heat exchanger is not limited to liquid— liquid
heat exchange but may also be used for gas— liquid
exchange and for gas-to-gas exchange. Materials of
construction may vary, depending upon the fluids
being handled. Either fluid may be moved through
the tube or annulus at relatively high velocities,
thereby aiding in the heat-transfer process.
Shell-and-Tube Exchangers
When the required heat-transfer surface is large, the
recommended type of exchanger the shell-and-
is
tube variety. In this type of heater or cooler, large
heat-transfer surface can be achieved economically
and practically by placing tubes in a bundle; the
ends of the tubes are mounted in a tube sheet. This
is very commonly accomplished by expanding the
end of the tube into a close-fitting hole in the tube
sheet by a process called "rolling." The resultant
tube bundle is then enclosed by a cylindrical casing
(the shell), through which the second fluid flows
around and through the tube bundle.
The simplest form of the shell-and-tube ex-
changer is shown in Figure 15.2— a single-pass
tubular heat exchanger. The fluid flowing through
328 APPLICATIONS TO EQUIPMENT DESIGN

Annulus
fluid

—c

Tube
fluid
1
out

Figure 15.1. Double-pipe exchanger.

the tubes enters a header or channel where it is
distributed through the tubes in parallel flow and
leaves the unit through another header. Either the
hot or the cold fluid may flow in the shell of the
exchanger surrounding the tubes.
Parallel flow through all tubes at a low veloc-
ity gives a low heat-transfer coefficient and low
pressure drop. For higher rates of heat transfer,
multipass operation may be used. Such a heat
exchanger is illustrated in Figure 15.3. In this type
of construction, the fluid in the tubes is diverted
by baffles within the distribution header. The
liquid passes back and forth through some fraction
of the tubes at high velocity, which gives good
heat-transfer coefficients. The number of tube
passes employed depends upon the economics of
the design and operation and upon the space avail-
able. Complexity in design sometimes results in
expense in fabrication that must be balanced
against improved performance. Another disadvan-
tage of the multipass exchanger is the added
frictional loss due to higher linear velocities and
the entrance and exit losses in the headers. Only an
economic balance can indicate the most judicious
design.
Note that in Figure 15.3 baffles are placed

Tube Shell Shell

fluid fluid fluid

in out in

^ y Tubes ^- Baffle

Outlet
header

Tube
fluid
zz: :zZ
out

^\\S\\\\\\\\\\\\\\\^ !
j
y////////////////^///////////////z^

Drain
Figure 15.2. Single-pass shell-and-tube heat exchanger.

Shell
fluid
out
Figure 15.3. Heat exchanger with four tube passes and one
within the shell order to divert the flow of the
in
shell fluid predominantly across the
into a path
tubes of the exchanger. The constantly changing
velocity of the shell fluid tends to impart turbu-
lence to the stream, thereby improving heat trans-
fer. Baffles in their simplest form consist of semi-
circular disks of sheet metal, pierced to accommo-
date the tubes. The baffles direct as much of the
flow as possible normal to the tubes in the shell
side as well as serve to keep the tubes from sagging.
It may be desirable to have several shell-side passes,
but this may cause complexities in construction
and higher friction losses. Multiple passes on the
shell side usually appear only in larger installa-
tions. Normally, close baffle spacing is employed.
If a condensing vapor serves as the heating med-
ium and is in the shell, baffles are usually not
needed.
The two heat exchangers illustrated indicate
that cleaning of both shell and tube bundle is
difficult, at best. In addition, since large temper-
ature differences between the two fluids ex-
changing heat are likely to exist, thermal expansion
can be expected, for which no provision is indi-
cated. Therefore, to allow provisions for easy re-
moval of the tube bundle for cleaning and to allow
for expansion, a floating-head exchanger is used, as
shown in Figure 15.4. Here, one of the tube sheets
is independent of the shell so that the entire tube
bundle can be removed for cleaning of the shell
and of the outside of the tubes. This also allows for
the differential expansion between the tubes and
the shell.
HEAT TRANSFER 329
shell pass.
A satisfactory choice and design of shell-and-
tube exchanger will depend upon a compromise of
such factors as cost, cleanability, temperatures,
corrosion, operating pressure, pressure drop, and
hazards. Large heat exchangers are designed speci-
fically for one particular application, but small
exchangers may be designed for multipurpose use.
A fixed tube sheet is the cheapest because of its
ease of fabrication. Unfortunately, high stresses
between tubes and shell may cause loosening of the
tube joints. In addition, shell-side cleaning is very
difficult without removing the tube bundle. Float-
ing heads eliminate stresses but add to cost of
fabrication.
The nature of the shell-side fluid is also
important and will influence the selection of the
type of exchanger. Since the shell side of the
exchanger is difficult to clean, the least corrosive
and cleanest fluid should be placed in the shell. To
retard or eliminate corrosion requires the use of
expensive alloys, and so the corrosive fluid should
be placed in the tubes to save the cost of an
expensive alloy shell. The viscosity of the fluids is
important in choosing which fluid should be on the
shell side. In the shell, induced turbulence and
added form friction can be attained by using baf-
fles. A fluid that would normally be flowing in
laminar flow in the tubes should be placed in the
shell to improve the heat-transfer characteristics.
High-pressure fluids should flow through the tubes
to avoid expensive high-pressure shells. Only a
complete analysis of local design specifications can
answer all the questions of selection.
330 APPLICATIONS TO EQUIPMENT DESIGN

• Floating head

(a)

Floating head

(b)

Cover end shell flange

-Shell nozzle flange
Shell cover flange

Shell cover

Shell cover bolt-

•
Floating tube sheet
Shell cover gasket

Floating cover

Floating cover gasket

Splice bar (split 90* to backing ring)
Boating cover bolt-
Floating cover flange

Backing ring (split)-

(c)

Figure 15.4. Two-tube-pass, one-shell-pass, floating-head heat exchangers, (a) Internal floating-head split-ring
type. (6) External packed floating-head type, (c) Detail of floating head.

Extended-Surface Heat Exchangers air, compared with 2000 for the steam or salt. This
If heat exchange is occurring between two fluids poor heat-transfer situation will require much
where one fluid has a very high resistance to heat transfer surface for a reasonable flow rate of air or
transfer in comparison to the other, the higher- oil.
resistance fluid "controls" the rate of heat transfer order to compensate for the high resistance
In
(see Chapter 14). Such cases occur, for example, in of the or air, the heat-transfer surface exposed
oil
the heating of air by steam or in the heating of a to these fluids may be increased by extension of
very viscous oil, flowing in laminar flow, by a the surface, as in the addition of fins to the outside
molten salt mix. The relative magnitude of the of the tube, as illustrated in Figure 15.5. The fins
heat-transfer coefficient is about 10 for the oil or are referred to as an extended surface; they

*\«.
 HEAT TRANSFER 331

Figure 15.5. Extended surface tubing. (Courtesy The Griscom-Russell Co.)

332 APPLICATIONS TO EQUIPMENT DESIGN

increase the transfer area substantially in a given equal to the sum of the individual resistances.
amount of space. Some automobile radiators are
good illustrations of extended-surface heat ex- 2/? = — =/? c + /? vv + /? d + /? L (15.1)
changers.
where

HEAT-TRANSFER COEFFICIENTS
Rc = condensate resistance
Rw = wall resistance
Typical overall heat-transfer coefficients for indus- Rd = scale or dirt resistance
trialheat exchangers are listed in Table 15.1. These RL = liquid-side resistance
are only order-or-magnitude values. Individual film
The vapor-side resistance depends upon the
coefficients may be calculated from the empirical
temperature driving force and upon the temper-
correlations of Chapter 13. Individual coefficients
ature of the condensate. The dirt resistance and
may be combined into over-all heat-transfer coef-
tube-wall resistance also depend upon their respect-
ficients by the methods of Chapter 14.
ive temperatures and upon the influence the tem-
perature has upon the thermal conductivity. Wilson
Evaluation of Heat- Transfer Coefficients
found study of steam-condenser operations
in a
The heat-transfer coefficient most easily measured that the sum
of the condensate, scale, and tube-
experimentally is the overall coefficient. The overall wall resistances remained substantially constant for
temperature difference and the total heat transfer varying cooling-water rates. For the turbulent flow
can usually be measured directly for a heat ex- of water, the water-side resistance is an inverse
changer of known area. By the relationship function of the fluid velocity through the tubes, as
q = -UA{AT), m the overall coefficient (U) may
, indicated by Equation 13.64.
be calculated. Heat-exchangers are often designed
Thus
using overall coefficients rather than individual
coefficients. The determination of individual co- 1 1 1

= Rr + /?w + RM + = Ci +
efficients is rather difficult because of the un- UA C2 (v) 08 CM*) 08
certainty regarding the measurement of surface
(15.2)
temperatures.
where
A method of calculating film coefficients was C-\ ,C2 = empirical constants depending upon
proposed by Wilson (39) for condensing vapors. It geometry and physical properties
is based on the overall resistance (MUA) being v = average water velocity, ft/s

Table 15.1 TYPICAL OVERALL COEFFICIENTS OF HEAT TRANSFER (7), (15)

Inside or Tube Outside or Shell

Type of Exchanger Fluid Velocity, ft/s Fluid Velocity, ft/s Btu/hrft 2 °F Ref.

Shell and tube Brine 1-3 Water 1-5 50-400 7

Shell and tube Water 2 Gas oil 3.0 50-70 7
Shell and tube Water 2 Lube oil 0.2 15 7
Shell and tube Water 5 Gasoline Condensing 90 7
Shell and tube Crude oil 2 Gasoline Condensing 20-30 7
Shell and tube Crude oil 10 Gas oil 6.0 80-90 7
Shell and tube Water 4-6 Steam Condensing 400-800 7
Double pipe Water 3-8 Brine 3-8 150-300 7
Coil in box Gas oil Condensing Water Natural convection 8-20 7
Tube bank Steam Condensing Air 10 9 7
Basket evaporator Brine Boiling Steam Condensing 150-225 7
Vertical tube evaporator Water Boiling Steam Condensing 400-1000 7
Vaporizer Steam Condensing Organic Boiling 300 15
Vaporizer Steam Condensing Acetic acid Boiling 450 15

(With the permission of John Wiley; copyright© 1950, and the Amer. Inst of Chem.
Engrs., copyright© 1959)
 HEAT TRANSFER 333

HEAT-EXCHANGER CALCULATIONS
The design calculations for heat exchangers are
based upon the application of Equation 14.62. The
equation as written is perfectly general but may
present difficulty of integration in some applica-
tions. Consider again the exchange of heat between
two fluids within a heat exchanger. At a particular
point in the exchanger, the rate of heat transfer
Slope = 1/C 2
Intercept, C
may be expressed by a modification of Equation
x

Intercept, Ci
14.62, or

dq = -U ATdA 115.3)

most heat exchangers, the temperature of

In

l/v os at least one of the fluids will vary, and hence the
driving force between the hot and cold fluids will
Figure 15.6. Wilson plot
vary. Not only will AT vary throughout the ex-
changer but so will q and U. The overall coefficient
of heat transfer will vary as a result of variations in
A plot of MUA as a function of 1/(i7)
- 8
on physical properties and in the flow regime brought
rectangular coordinates will determine the con- about by the transfer of heat. Consequently, the
stants in Equation 15.2. the average temperature
If
integration of Equation 15.3 is complex. Solving
of the water varies widely, C2 will vary enough Equation 15.3 for the area gives
that the analysis is not dependable. This relation is
indicated in Figure 15.6. dq
If the straight line is extrapolated to an infinite
(15.4)
J
o
J
o U AT
velocity (zero water-side resistance) the intercept
Ci of Equation 15.2 will be determined. For the The temperature
variations of the fluids with-
moment, consider that the condenser tube is clean in heat exchanger are shown in Figure 15.7,
a
so that R d = 0. Therefore, the value of C-\ equals where the temperature of each stream is plotted as
the sum of the condensate and tube-wall resist- a function of the length of the exchanger.
ances. The tube-wall resistance, which is a function In a true parallel-flow heat exchanger, Figure
of thermal conductivity and wall thickness, can be 15.7a, both the hot and cold fluids enter the ex-
calculated and subtracted from Cy to give the changer at the same end and flow through the
condensate resistance; its reciprocal is the conden- exchanger in the same direction. At the entrance of
sate heat-transfer coefficient. the exchanger, a large driving force prevails, giving
The slope of this line equals 1/C2 , and, at a a relatively large heat-transfer rate. As the fluids
water velocity of 1 ft/s, the water-side resistance progress through the exchanger, the temperature
also equals 1/C2 The data for a Wilson plot must
. driving force becomes less and less so that the rate
be obtained over a varying range of water velocities drops off asymptotically as the streams approach
with both condensate and water temperatures, on some limiting temperature. The net result of this
the average, remaining as nearly constant as pos- type of variation in AT is that the exchanger is
sible, so that the Prandtl number remains essen- much more effective for a unit area of heat-transfer
tially constant. Also, as high a water velocity as surface at the entrance than it is near the exit.
possible should be used to give a short extrapola- Simply extending the length of the exchanger will
tion. result in little improvement in the amount of heat
Now, suppose that the same condenser tube is transferred. On the other hand, Figure 15.76 illus-

tested after several months service, during which trates true countercurrent operation of a heat ex-
time the tube has become fouled. Equation 15.2 is changer. In this type of operation, the two fluids
still valid but the straight line resulting from the exchanging heat pass each other in opposite direc-
test will now lie above the clean-tube case and will tions. Here, the driving force is much more nearly
give a larger intercept. The new intercept will in- constant throughout the length of the exchanger.
clude in its value a dirt or scale resistance that The net result isthat a unit surface gives about the
would be equal in magnitude to the difference same rate of exchange through the entire ex-
between intercepts for clean and dirty tubes. changer. In general, the variations in driving force
334 APPLICATIONS TO EQUIPMENT DESIGN

Length Length
(a) (b)

Figure 15.7. Temperature in parallel and countercurrent heat exchangers, (a) Parallel.

(b) Countercurrent.

and resistances necessitate the rigorous integration There are 10 pipes per pass, each 0.622 in.

of Equation 15.4 to size heat-exchange equipment. I.D. by 6 ft long.

Example 15.1. (8). It is desired to heat 1000 Inside area / 0.622 \

bbl/dayof cottonseed oil from 70° F to 180°F by

= 7r^— -J(6)(10) = 9.78 ft
2
/pass
pass
means of steam condensing at atmospheric pressure
in a shell-and-tube heat exchanger. Ten copper Thus, the total heat-transfer area becomes 9.78/V
pipes (0.622 in. I.D. and 0.840 in. O.D.) 6 ft long where N is the number of passes.
will be used in parallel in each pass. Assuming a Therefore,
steam heat-transfer coefficient of 2000 Btu/hr
2
ft °F, how many passes in series will be required?
CpdT
9.78/V = w, j (b)
The properties of cottonseed oil are as fol- r, -UiAT
lows:
Density = 0.8836 - 0.00028797° F, g/cm 3 In many simple cases
heat transfer, the in

Specific heat = 0.45 + 0.000625(7° F - 60), Btu/lb right-hand side of Equation could be integrated (/?)

°F analytically, assuming constant values of Cp and U,.

Thermal conductivity = 0.0808 Btu/hr ft 2 (°F/ft) However, in this problem both cp and U, vary with
Viscosity = 11.2 centipoises at 100°F; 2.5 centi- temperature to an appreciable extent, and it is
poiseat210°F necessary to evaluate them at several temperatures
(Assume log viscosity versus temperature is
between Ty and T2 Equation (b) is then inte- .

linear.) grated graphically using the point values of

Cp/UjAT.
Solution. The heat-transfer equation is
To aid in the solution of this problem, it will
(A r q dq be advantageous to plot the physical properties of
J
dA=- the cottonseed oil as a function of its temperature.
J
o U AT
This is done in Figures 15.8, 15.9, and 15.10.
but dq may be evaluated by the heat gain of the oil,
The overall resistance based upon the inside
dq = [wjC p dT] surface area is

Thus,
1 1 Ax 1
t2 WjCp dT T CpdT
i ', -U AT
= Wil
r, -U AT UiAj hjAj
+
kA,
+
ho A
(c)

The transfer area per pass may be evaluated * It is common practice in heat-transfer calculations to base
from the pipe data. Because of the physical charac- the overall heat-transfer coefficient upon either the inside or the
teristics of the cottonseed oil, it will offer great outside surface area of the tube when cylindrical geometry is con-
sidered. Throughout this chapter the subscripts / or o will denote
resistance to heat transfer in comparison to the
inside or outside surface. Previously, subscripts 1 , 2, etc., were used
steam or pipe-wall resistances. Therefore, the inside to denote phase boundaries. Thus sub 1 now becomes sub /. Sub-
pipe area is associated with the controlling resist- scripts 1 and 2 will be used to indicate the ends of flow paths in

ance and will be used in these calculations. heat exchangers.

 HEAT TRANSFER 335

of the inside resistance. In this case, both the steam

resistance and copper-pipe resistance are constant.
The mass flow rate of cottonseed oil is com-
puted as follows:

E 6
" 5 W,= (1000^)(42^)(1^)
=* 4
\ day/v bbl/'24hr/
3
x (8.34) [0.8836 - 0.0002879(70)] lb/gal

70 80 90 100 110 120 130 140 150 160 170 180 190 200 210
Wj= 12,600 Ib/hr feed to exchanger (constant for
Temperature, "F steady state and independent of temperature)
Figure 15.8. Viscosity of cottonseed oil.
will be convenient to express the flow rate
It

as a mass velocity so as to avoid calculations of

linear velocities. Linear velocities are dependent
upon the specific gravity of the oil which is tem-
perature sensitive.
Thus,
60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
2 2
Temperature, °F (12,600 Ib/hr) (1 44 in. /ft )

Figure 15.9. Specific heat of cottonseed oil. 2

(0.304 in. /tube)( 10 tubes)

= 596,000 Ib/hr 2
ft
0.86

0.85
At any point within the exchanger, — (A7~) =
(212 — 7")°F, where T is the bulk average temper-
0.84
ature of the oil. Thus, all terms in Equation {b)
0.83
have been determined in a general manner so that
0.82
70 80 90 100 110 120 130 140 150 160 170 180 190 200 210
it is now possible to evaluate them explicitly for
Temperature, "F specific within the exchanger. For this
points
Figure 15.10. Specific gravity of cottonseed oil.
example, points in the exchanger corresponding to
cottonseed-oil bulk average temperatures of 70,
100, 130, 160, 180°F will be selected. Only the

or — = — + AxA, + A
1 1
70°F point will be completely calculated.
Ui hj kA m h A t
For 70° F. First, the flow character must be
determined through the use of a Reynolds number.
For circular cross sections, A is proportional to D;
Since frequent computations of A/ Re are necessary,
therefore,
it will be expressed as a function of oil viscosity, or
AxDi D DG
J_ _ _L _ (0.622/12X596,000) _ 12,800
<
_ = "

Df = 0.622
U,

in.
hi kD m
Ax = 0.109 in.
h D A/Re=
T (Z427) ^"T
7

where // is in centipoises.
D = 0.840 in. k Cu = 220 Btu/hr ft
2
(°F/ft) The same technique will be used for the
Dm = 0.731 in.
Prandtl number.
(for thin-walled tubes the arithmetic mean is satis-
factory) so that =
cpix JcP ){2A2n') =
/v Pr 29.9M( M ')

k 0.0808
1 1 (0.109/12X0.622) (0.622)
J[ +
~ The plot (Appendix C-5) for heat
/-factor
Ui hi (220X0.731) "
(2000X0.840)
transfer convenient to use in this case because it
is
or (d) eliminates the need for concern regarding the
appropriate coefficient correlation to use. Since
— = — + 0.000405
1 1

use of this plot may involve an L/D,

Ui hj
L 6ft
Equation (d) is a general equation for this = 116
problem expressing the overall resistance in terms D (0.622/12) ft
336 APPLICATIONS TO EQUIPMENT DESIGN

Cp
Now the sample calculations for 70°F will be
7/ -A7=(212-71 h, Ui Cp
completed. From Figures 15.8 and 15.9, Ui(-AT)

(/V Re ) 70
= — 800 = 762
1 2
(definitely laminar flow)
70
100
142
112
43.7
42.5
44.8
42.0
0.456
0.475
0.0000732
0.000101
16.8
130 82 36.4 35.8 0.493 0.000168
= 29.9(0.456)(16.8) = 229; 2/3
= 37.2 160 52 53.5 52.4 0.512 0.000188
(N Pr ) 7o
(/V Pr )

180 32 72.3 70.3 0.525 0.000234

Before the oil coefficient can be evaluated, it is
necessary to know the oil-side pipe-surface temper-
It is apparent from the above table that
ature; it must be assumed for a preliminary calcula-
neither the total resistances Mil nor cp are con-
tion and checked by the series-resistance concept
stant. A graphical integration of cpdTIU A7" from
after completion of the preliminary calculation.
Equation {b) will therefore be required since there
Assume Tis = 208°; then y. is = 2.6 centipoises '

014 = is no analytical relationship between all the vari-

and M/s/M = 2.6/16.8 = 0.155 and (m/s/m)
ables.
0.77. From Appendix C-5, at a Reynolds number
= 0.0046 and so from Cp/U/AT is 7", and the
plotted as a function of
of 762 and an LID = 1 1 6, /,
area under the curve determined between the
is
the definition of jq (Chapter 13)
limits of 77 = 70 and T, = 180. A value of 0.0157 is
0.1 4 found. The number of passes may be determined
2/3
y<7
= 0.0046 = (^)(/V Pr ) from Equation (b),
(?)
12,600x0.0157
and N= = 20.3 « 21 passes
9.78

/>/=/«CpG/(/VPr ) 2 '3 Logarithmic-Mean Temperature Difference

p) The solution to the previous illustration recognizes
or all variations in properties and therefore is as rig-
orous as accuracy in graphical integration will per-
(0.0046) (0.456) (596,000) mit. The procedure is necessarily tedious, but,
fortunately, under some conditions simplifying
assumptions leads to negligible error. Equation
and from Equation (d) 15.3 may be solved using the following assump-
tions, which are very often valid.

— = — + 0.000405 = 43.7 + 0.000405 = 0.0223 1. The overall coefficient (U) is constant.

U, hi
This not rigorously valid, but, for fluids whose
is

physical properties are not too sensitive to temper-

Now check the assumed Tis by proportioning re-
ature variations, the overall coefficient based upon
sistances and driving forces as in Chapter 14.
average fluid properties leads to little error.
2. The specific heats of the two fluids are
212 Tis
212-70
independent of temperature.
0.000405 0.0223 3. The mass rates of flow of both fluids are
constant.
The left side of the equation
the ratio of the is

temperature drop across the steam and pipe to the the second and third assumptions are valid,
If

sum of their resistances, and the right side is the a simpler integration of the design equation may be
ratio of the total AT to the total resistance. Solving used. The development that follows is for true
for T!s gives parallel-flow operation but applies equally well to
countercurrent operation. Figure 15.11 shows the
Tis = 209° F temperature history as a function of the heat-
which is a satisfactory check. transfer rate throughout such an exchanger. Be-
The same sequence of
calculations must be cause of constant specific heats the temperature
repeated for each of the other temperatures. The lines are linear with heat transferred, the hot fluid
results are tabulated below along with the quanti- flows at a constant rate of (wA Ib/hr and changes )

ties necessary for the solution of Equation(b). in temperature from TA to TA2 The cold fluid
-\ -
 HEAT TRANSFER 337

grated between the limits of A7"i and A7"2 and

TA _—-T=" TAi
•^^AT, between and q to give
1 1 ^
-(AT2 -AT,) = qA
2> TR lwB c PB WA CPA
AT2 Tm
TB
(15.10)

It would be advantageous in many calcula-

tions to evaluate the integral of Equation 15.3
directly when the simplifying assumptions hold.
This would be possible if a rigorous average AT
LOW q High q
were known. Derivation of a rigorous mean value
Heat transfer rate (q), Btu/hr
follows.
Figure 15.11. Heat-transfer rate as a func- Combining Equation 15.9 with Equation 15.3
tion of driving forces. for steady state yields

will flow at a steady rate of w B Ib/hr and be heated dq A =-U ATdA =-

from TB y to TB2 The temperature driving force,
. 1 1

in the transfer direction, at any point in the ex- w B c PB wA c PA

changer is
(15.11)
-AT=(TA -TB ) (15.5)
Rearrangement of Equation 15.11 to separate
Differentiating Equation 15.5 gives variables for integration, with U being assumed
-d(AT) =cfTA -dTB (15.6) constant or appropriately averaged, gives

Now, the heat

through the exchanger
given up by the hot fluid as it passes 1
-+
-\
1
\U\J
rA
dA = -
r
Ar d(AT)
2
—
is
wB c PB wA\Cp
c PA
Ai J
A7", AT
-dq A = -wA c PA dTA ( 1 5.7)
(15.12)
and the heat gained by the cold fluid is
Upon integration, Equation 15.12 becomes
dq B = wB c PB dTB (15.8)
1 1 AT2
From Equation 15.7
UA = In I15.13)
iwB c PB WA Cp A AT,
dq A But from Equation 15.10
dT^ = (15.7a)
wA c PA 1 1 AT2 -AT,
and from Equation 15.8 wB c PB wA c PA \- Qa
dq B :15.10a)
dTR = (15.8a)
WB C PB Therefore, substituting Equation 15.10a into Equa-
tion 15.13 gives
Combining Equations 15.7a and 15.8a with Equa-
tion 15.6 yields rAT_2_ A7", A7"2
UA = In 15.14)
Qa AT,
dq A dq B
-d(AT) = Rearranging,
wA c PA wB c PB
ATn - AT,
= dq A
1 1
Qa = -UA = -UA{AT) lm
w B c PB wA c PA AT2
In
AT,
15.9)
(15.15)
since dq A + dq B = at steady state.
By the assumptions previously stated, the Equation 15.15 is the simplified design equa-
terms within the brackets of Equation 15.9 are tion for cases where the aformentioned assump-
constant. Therefore, Equation 15.9 may be inte- tions are valid. The bracketed term of Equation
338 APPLICATIONS TO EQUIPMENT DESIGN

15.15 defines the logarithmic-mean temperature temperatures of the streams versus heat transferred
driving force (AT/m
Thus, the variation in driving
). will reveal the inaccuracies.
force is handled very simply by using the log mean In the widely used tube-and-shell exchanger,
of the driving forces at the terminals of the ex- the complex flow paths can bring a stream into
changer. Equation 15.15 applies also to counter- contact with the other stream at points where the
current flow as well as to a situation where one of allegedly hot stream is actually at a temperature
the fluids is at constant temperature such as a lower than the other, generating a temperature
condenser. It is rigorous for any case where AT is cross. This should be avoided.
linear with q. In the tube-and-shell exchanger, there is virtu-
Colburn (10) has suggested that a more accu- ally certain to be leakage and bypassing of part or
rate calculation can be made using a logarithmic- parts of a stream, so that not all elements of that
mean value of the combination U AT. It should be stream are exposed to the same thermal environ-
emphasized that this is the product of coefficient ment, invalidating the log-mean AT. This should
at one terminal and A7"at the other as indicated by become clear when this type of equipment is stud-
the subscript. ied.
Accordingly, It must be reemphasized that
in the general
case where the simplifying assumptions are not
Uy AT2 - U 2 AT, valid, heat-exchanger area must be evaluated by
q = -A ;15.16)
Uy AT2 graphical integration of Equation 15.4.
In
U 2 AT :

Example 15.2. Solve Example 15.1 using a

where logarithmic-mean temperature difference and
Uy,U2 = overall coefficients at exchanger Equation 15.16. Base properties on an average oil
terminals temperature.
AT\ , AT2 = temperature driving forces at
Solution. The total heat load on the exchanger
terminals
will be the same as Example 15.1.
In Equation 15.16 it is assumed that the over- Therefore,
all coefficient varies linearly with T through the
exchanger, as well as that specific heats are con- q = wc P (T2 - T, ) = (12,600X0.48) x (180 - 70)
stant and that steady state exists. = 665,000 Btu/hr
In several common classes of heat exchange,
the assumptions made in deriving the logarithmic
(212 -70) -(212- 180)
(AT) lm = _,_ = 74.3°F
mean of AT are not satisfied; hence, a design based 212-70
In
upon it is fallacious, or even unworkable. In some 212- 180
cases where the assumptions are but the satisfied,
stoichiometry calls for a disappearance of A7" at The average oil temperature is (180 + 70)/2 =
one end, the unit is unworkable. 125°F. Therefore, ^ 12 5°=37.0 Btu/hr ft 2 °F
Most of the erroneous designs arise because from the table in Example 15.1,
wcp of one or both streams is not constant.
Change in the quantity of a stream would be recog- 665,000
and A = = 244 ft
2
nized as invalidating the logarithmic mean. Where U(AT) lm (37)(74.3)
wcp is not constant, because of a varying Cp or
244
the involvement of either latent heat or heat of
Thus N = —— = 24.9 passes
reaction, the temperature-enthalpy history of that 9.78
stream is nonlinear. If these plots are convex toward
each other, the AT is less throughout the exchang- Here the total heat exchanger is about 20
er than at the ends; if concave toward each other, percent larger than that of the detailed solution.
the AT is greater. For the convex case, with a If Equation 15.16 is used, the following area
relatively wide spread of temperatures of each results:

stream, and relatively small A7" (temperature

a
approach) at the ends, there may develop a pinch H
\iu\tJ
U 2 ATi
point within the exchanger, where A7" approaches A =
zero. This is obviously unworkable. A plot of the U AT2 - U2 AT
y y
 HEAT TRANSFER 339

or A/ Re = 2,100, according to Appendix C-5 the rigor-

ous graphical solution is required. If both Reynolds
numbers exceed 10,000, the resistance may be
\70.3x 142' nearly enough constant so that the logarithmic-
A = 665,000 = 153 ft
2

44.8X32) - (70.3X142) mean temperature difference may be used. For

Thus A/= 153/9.78 = 15.6 passes, or about 30
cent lower than that of Example 15.1
Figure 15.12 shows one reason for the differ-
ence between Examples 15.1 and 15.2.
Curve A represents the actual overall resist-
ance as a function of position in the exchanger.
The reason for the maximum can be seen by refer-
ence to the Reynolds number—/-factor relation-
ships. The oil enters the exchanger at a /V R e = 762,
which is laminar flow and leaves at a /V Re = 3420,
so that the flow character goes from laminar into
the transition zone, and the heat-transfer coeffic-
ient will go through a minimum. Examination of
the tabulated data of Example 15.1 indicates also
that U will have a minimum, and hence the total
resistance becomes a maximum at some point in
the apparatus. Curve B is the constant resistance
both Reynolds numbers between 2,100 and 10,000
per- or for both less than 2,100 about a 10 percent
error is likely if either Equation 15.15 or 15.16 is
used. The recommended procedure is to examine
both terminal Reynolds numbers, the specific heat,
and the overall heat-transfer coefficient. If the
Reynolds number exceeds 10,000 and the specific
heat is constant an overall resistance based upon
the average fluid temperature may be used in con-
junction with the log mean A 7". If these conditions
are not met, a rigorous point-to-point solution of
the design equation is required for an accurate
result.
Example15.3. A double-pipe heat exchanger
is be designed to cool benzene from 180°F to
to'
100°F, under sufficient pressure to maintain the
benzene in the liquid phase. Water enters the annu-
lar space at 70° F and flows countercurrent to the

selected as that for 7~= 125°. In this case, the benzene at a velocity of 5 ft/s. The inside pipe is
resistance is relatively high and constant; therefore, sch.-40,* l^-in. steel pipe; the outside pipe is

a larger area is required. Curve C is the resistance sch.-40, 2-in. steel pipe. The pipe comes in 20-ft
assumed in the Colburn relationship. This curve sections. How many sections in series must be used
shows the resistance to vary linearly with temper- to cool 7500 lb of benzene per hour? Although
ature of the oil but to be considerable lower than pipe is available in 20-ft sections, the jacket must
either of the other two cases, thereby indicating a be shorter than the inner tube, as can be seen by
smaller heat-exchange area. reference to Figure 15.1.
The general conclusion from this illustration
Solution. The solution of this problem is
is that the simplified procedures are not appro-
clearly one of evaluating the various resistances
priate when the flow character of either fluid is
that are encountered along the transfer path. In
such that the fluid passes from laminar flow into
this particular case the resistances are: 1) the ben-
moderate or extreme turbulence, which causes a
zene resistance, 2) the pipe wall, and 3) the water
variable overall heat-transfer coefficient. Thus, if
resistance. Fluid properties will be evaluated at the
one examines the Reynolds number of the fluid at average bulk temperatures.
the terminals of the exchanger and if it brackets

The Benzene Resistance (inside pipe)

\iu The inner diameter of the sch.-40, U-in. pipe

1.38
12
= 0.115ft

The cross-sectional flow area of the inner pipe

= 0.0104 2
ft

Leneth
* The schedule number for a pipe is a convenient method for
Figure 15.12. Comparison of calculations characterizing the strength of the pipe. More will be said of nominal
for Examples 15.1. and 15.2. pipe size and schedule number in Chapter 21.
340 APPLICATIONS TO EQUIPMENT DESIGN

The average bulk temperature of the benzene 1.66

The outer diameter of the inner pipe =
180+ 100 12
= 0.1385 ft
2
= 140° The equivalent diameter for this geometry is

Density of benzene at 140° = 52.3 lb/ft

3
defined as
4S
D eq = — (see p. 253)
Viscosity of benzene at 140° = 0.39 centipoise
/ 1 kr \ Theannulararea,S = -[(0.172) 2 - (0.1385) 2 ]
Mean velocity of benzene = (7500 Ib/hr) | 4
'3600 s' = 0.00816 ft 2

The wetted perimeter, P= 7r(0.172 + 0.1385)

x )
(
r) = 3.83 ft/sec
(
\52.3lbAo.0104ft 2 / = 0.977 ft. Thus,

Dvp (0.115)(3.83)(52.3) AS (4X0.00816)

(/V Re = = 8.78x 10 4 0e q = = 0.0356 ft
)/ 4
H (0.39 x 6.72 x 10" )
f 0.977

It quite apparent from the magnitude of

is
The exit temperature of the water may be
A/ R e that the benzene is flowing in complete turbu-
obtained by an overall heat balance.
lence. r Ta2 [
t B2
Specific heat of benzene at 140° =0.45 j
wA c PA dl A = J
j
w B c PB dl B
ta ^ t b:
Btu/lb°F
Thermal conductivity of benzene at 140° = wA c PA {TA2 -TA ,) = wB c PB (TB2 -TB ,)
0.085 Btu/hr ft 2 °F/ft.
Water at 5 ft/s = (5X3600X62.3X0.00816)

/V Pr = —
cpp
k
=
(0.45X0.39x2.42)
r-—:
0.085
= 5.0
= 9180 Ib/hr. Thus,

(7500X0.45X100 - 180) = (9180)0X70 - Tfi1 )

Thus, the benzene coefficient can be estimated

-(7500X0.45X80)
(70 51 = -29.4
using Equation 13.64. 9180
C
hjD -TB ^
= -29.4 - 70 = -99.4° or 7"
fi1
= 99.4
= 0.023(/V Re - 8 - 3
) (/V Pr )

k 99.4 + 70
Average water temperature = = 84.7
hjD
8
= 0.023(87,800)° (5)° 3 = 337 ^85°F
k
Viscosity of water at 85° F = 0.8 centipoise
337)0.085)
= = 249 Btu/hr ft
2
°F Thermal conductivity of water at 85° F = 0.342
hi
0.115
D eQ vp (0.0356X5X62.3)
The area associated with h, is (3. 1 4) (0. 1 1 5) ( 1 (N Re'o = 20,600
1-4
= 0.361 ft /ft. 2 H (0.8x6.72x10-
Thus, the benzene resistance becomes The water is also flowing in well-developed
1 1
turbulence. Again, Equation 13.64 is used, but in
R,= this case the exponent on /V Pr is 0.4 because the
h,A, (249)(0.361) water is being heated.
/?/ = 0.0111 hrft°F/Btu
cP n (1X0.8x2.42)
A/ Pr = -^— = = 5.67
The Water Resistance (in annulus) k (0.342)
An equivalent diameter of the annulus will be
required to obtain A/ R and h ho D eq 4
e = 0.023(/V Re )°- 8 (/V Pr )°-

2.067
The inner diameter of the outer pipe =
12 ho D eq 8 4
= 0.023(20,600)° (5.67)° = 130.5
= 0.172 ft
 HEAT TRANSFER 341

(130.5X0.342) Therefore, four lengths will have to be used,

h -' =12 50Btu/hrft^F which allows about one foot per length for jacket
(0.0356)
couplings as ineffective for heat transfer.
The area associated with the water resistance is the This illustration was solved using the A7~/m
outside surface area of the inner pipe or concept, which necessitates the validity of several
assumptions. The assumption of constant values of
AQ = (3.14)(0.1385)(1) =0.435 ft
2
/ft
cp for both benzene and water is rather good in
Thus, the water-side resistance becomes this case since only a slight variation may be ex-
pected over the temperatures involved. But the use
1
of ATlm also requires the constancy of the total
1

fl.= = 0.001 805 hr ft °F/Btu

h A (1250) (0.435) resistance. To check the validity of this assump-
tion, U may be calculated at the terminals of the
Resistance of pipe wall exchanger, and from these values the appropriate-
ness of the use of AT/m may be determined.
0.140
Pipe-wall thickness = = 0.01 1 7 ft Calculation of U at the terminals (hot end of
12 exchanger)
Thermal conductivity of steel = 26 Btu/hrft 2 (°F/ft) For benzene:
Arithmetic mean diameter of wall Temperature of benzene = 180°F
0.1385 + 0.115 Specific heat of benzene at 180°
= 0.127 ft °
= 0.47 Btu/lb
Viscosity of benzene at 180°
The arithmetic average is justified since the
= 0.3 centipoise
wall thin. The mean wall area is
is
Density of benzene at 180°
2
irDL = (3.14)(0. 127)(1) = 0.398 ft /ft = 50.8 lb/ft 3
Thermal conductivity at 180°
Thus, the pipe-wall resistance is
= 0.082 Btu/hr ft 2 (°F/ft)
Ax 0.0117 Dvp _ (0.115X3.83X50.8)
nw - = 0.00113hrft°F/Btu =
(/v Re = 111 ,000
kA m (26) (0.398) ),
4
(0.3 x 6.72 x10~ )

The total resistance to heat transfer based upon a

CpP (0.47X0.3x2.42)
preliminary estimation of the surface coefficient is (/v Pr ),= = 4.16
k (0.082)
!/? = /?/ + Ftp + R
h,D
= 0.01 1 1 + 0.001 13 + 0.001805 = 0.01404 = 0.023(/V Re 08 ) (/V Pr )°-
3

Log-Mean Driving Force = 0.023(111,000)° 8

(4.16)° 3 = 381
AT,-AT2 (180-99) -(100-70)
(381X0.082)
(AT) /m = hi
= = 272 Btuhrft 20 F
A7-i 180-99) (0.115)
In In-
A7"2 (100-70) For water:
c
= 51
Temperature of water = 99°
c
The total heat requirement is 99
Specific heat of water at
= 1.0 Btu/lb °F
q = (7500(0.45X180 - 100) = 270,000 Btu/hr
Viscosity of water at 99° = 0.7 centipoise
The total resistance ZR = 0.01404 hr ft °F/Btu. Density of water at 99° = 62.3 lb/ft
3

Thermal conductivity at 99°

Therefore, the total length of heat exchanger is
2
= 0.350 Btu/hr ft (°F/ft)
found from
(AT) ,~ 51 D eQ vp_ (0.0356X5X62.3)
= = 23,500
q/ft = = 3640 Btu/hrft (/V R e)
~ (0.7 x 6.72 x 10- 4
1 0.01404 p )

or the total length

UA
is 270,000 3640 or 76.4 ft
(/V Pr )o = —
cP p
=
(1X0.7x2.42)
(0.350)
= 4.84
342 APPLICATIONS TO EQUIPMENT DESIGN

^^k
= 0.023(/V Re )° 8
(/V Pr )°-
4
Thus, it

made between U
seems view of the comparison
justifiable in
the terminals and U based
at
upon average properties that the average temper-
= 0.023(23,500)° 8
(4.84)° 4 = 135 ature is satisfactory and the (AT)/ m assumptions
(135) (0.350) are valid. Actually, the use of (AT) /m is usually
h °= = "30 Btu/hr ft* °F justified whenever the flow character is fully turbu-
(0.0356)
lent and physical properties do not vary widely, so
The total resistance at the hot end of the heat checking terminal values is rarely necessary.
exchanger is

1 1

S/? = + 0.00113 + SHELL-AND-TUBE HEAT EXCHANGER

(0.361 )(272) (0.435X1330) CALCULATIONS
= 0.01306 hr 2
ft °F/Btu Under certain restricted conditions, the heat-trans-

This value of 0.01306 is about 7 percent lower fer design equation may be written in the form of
than that based upon the average properties; Equation 15.15 as q = —UA{AT)/ m It is a useful .

equation despite its many restrictions. In most

Cold End of Exchanger commercial heat exchangers of the shell-and-tube
variety, however, the flow is neither cocurrent nor
For benzene:
countercurrent. On the contrary, Figure 15.3
temperature = 100°F shows that the flow is really a complex pattern of
cP = 0.43Btu/lb°F mixed flows. Note, however, that if either stream is
p = 0.52 centipoise at a constant temperature as in a condenser or
p = 53.8 lb/ft 3 vaporizer, the heat transfer behaves as in a simple
k = 0.09 Btu/hr ft 2 (°F/ft) exchanger.
Dvp (0.115)(3.83)(53.8) If the same restrictions that are applicable for

{N ^ = =
/nn
= 67 700
simple-flow exchangers are assumed to hold, the

W'-t- —m)—
]
>
p «7o
(0.52 x 6.72 m-4i
x 10 *)
'

(AT), m may be used only in conjunction with a

Cpn (0.43X0.52x2.42) suitable correction factor to account for the geo-
602 metry of the exchanger and for the fact that flow
is neither countercurrent nor cocurrent. Thus, for

—
h.D
k
= 0.023(67,700)° 8
(6.02)° 3 = 290 complex exchangers,

q = -UAY(AT) lm (15.17)

hi = —tttz—
(290)(0.09)

0.115
= 227 Btu/hr ft
2 °
F where Y is the geometry factor and (ATl, m is the

temperature driving force evaluated from terminal

For water: A7~'s based upon countercurrent operation.
Bowman, Mueller, and Nagle (5) have corre-
Temperature = 70° F lated the /-factor in terms of two dimensionless
3
p = 62.3 lb/ft ratios.
c P = 1 Btu/lb°F
Figure 15.13 schematically indicates a one-
p = 0.94 centipoise shell-pass, two-tube-pass exchanger with phase A
k = 0.334 Btu/hr ft 2 (°F/ft)
entering the shell at temperature TA :
and leaving
h„D at TA2 . Phase B enters the tubes at temperature
°— ^ = 0.023(1 7,500)° 8 (6.81 )° 4 = 123.0 TB and leaves at 7g Two dimensionless ratios
-\
2 -

k
123.0x0.334 2
ho = 1160 Btu/hr ft (°F/ft)

1 1
2/? = + 0.00113 + +T R
(0.361X227) (0.435X1160)
= 0.01521 hrft 2o
F/Btu

The cold-end total resistance is about 9 percent Figure 15.13. Schematic representation of a
higher than that based upon the average properties. one-shell-pass, two-tube-pass exchanger.
 HEAT TRANSFER 343

are defined. One is a temperature ratio be represented by

400° -» 200°
TA y-TA2 t

Z= 15.18) 200° «- 100°

T,B2 T,si

Ar =200°
1 AT2 = 100
C

Equation 15.18 is in reality a ratio of the hourly

heat capacities -
the heat required to raise
that is,
Therefore
the hourly flow 1°F for each of the two phases. AT -AT2y
200-100
The second dimensionless ratio, which is also a ATlm = = 144 c
AT, 200
temperature ratio expressing the effectiveness of In In
the exchanger, is defined as AT? Too

The Y factor can be obtained from Figure 15.14b.

T,B7 T,B
X= TA y~TA2
1

15.19) 400-200
T*, - T,51
Tb2 200- 100
Figures 15.14 and 15.15 show the graphical Tb2 TB y
200- 100
X= = 0.33
correlation proposed by Bowman for several basic TA y-TB ^ 400-100
configurations. In the derivations of these dia-
grams, itwas assumed that the shell-side fluid is so
and from Figure 15.14b, Y= 0.96
well mixed that an average temperature at any, Thus
cross section is applicable. Also, the assumptions of
constant overall resistance, constant specific heat,
(AD corrected = Y(AT) lm = 0.96 x 144 = 138°F

no phase change, and equal transfer area in each

pass are implied. In the event that these assump- Tube-Side Coefficients
tions are not valid, reference should be made to Heat-transfer coefficients for the fluid flowing
Gardner (12, 13), who treats more complex situa- through the tubes are calculated in accordance
tions. with correlations presented in Chapter 13. The
tube-side coefficient should be evaluated first and
Example 15.4. A shell-and-tube heat ex- analyzed to ascertain whether the tube-side resist-
changer having two shell passes and four tube ance is dominant. This should be done considering
passes is being used for cooling. The shell-side fluid the nature of the shell-side-fluid. Calculation of
enters at 400° F and leaves at 200° F and the tube- tube-side coefficients first may eliminate the com-
side fluid enters at 100°F and
200° F. leaves at plex procedures necessary in evaluation of shell-
What is the mean temperature difference between side coefficients. In any event, the usual heat-
the hot fluid and the cold fluid? transfer-coefficient correlationsapply for the
tube-side fluid.
Solution. A
simple sketch will aid in this
solution. The 2—4 exchanger can be represented as
Shell-Side Coefficients
follows
The heat-transfer coefficients for the shell-side
fluid cannot be accurately calculated by any of the
TA2 = 200"F
correlations presented thus far. A continually

H — * TBl = 100°F
changing path of the shell-side fluid prevails as a
result of the baffles influencing its flow direction
and because the cross-sectional flow area will vary
as the fluid flows among the tubes, the heat-
transfer coefficient must be determined entirely
-»- T„, = 2(XrF from experimental data. In addition, leakage past
baffles and short circuits of flow paths also limit
TA =
, 400'F
the effectiveness of the exchanger. It is not the
purpose to present here the myriad of techniques
First, the (A7~) /m based upon true counter- available to account for leakage, variable areas, and
current operation is evaluated. This evaluation may other corrections in the evaluation of the shell-side
344 APPLICATION TO EQUIPMENT DESIGN

1.0 ,, - |
£- 1
;
i^

t \^
09 V A \ \ \ \\\ N. >.

n \ \
\\ \\\\
\ \ \
\ \ \

i \ ^ \o
0.8

\ \
\
|
o
b
m
o II
00-
b b |5 '

\5
N>
iO\C0 te-
3
*

VL4
\
\ \
0.7
o
b \ \ V L
\ \

0.6

\ \\ V
ft \ \]
|f
0.5
0.1 0.2 0.3
TK V VA 0.4 0.5 0.6 0.7
1 1
0.8
] 0.9 1.0

J I

^-^2
-*"^.

""if
(a)

1.0

^ S"\V\
Y
0.9 \\ N^\
J>

) V3
o V
v

\ \ \
y 0.8
II
p \
o O P° c»
V*
b b o ° b 1
\'c > \'oo

f f
\
\ \ 1 1
°.0.7 I \l \ l\ 1

\\
0.6

05
0.1 0.2 0.3 0.7
1

111 0.8 0.9

II

1.0

Figure 15.14. (A7") //77 correction factor for mixed-flow heat exchangers (5). (With
permission of ASME, Copyright © 1940.)
TB2 - -
Z = ^1
7"g1 T"^ 2
X=
TA l
- 7fi 1 7~52 - 7*81
 HEAT TRANSFER 345

\ so
10
>> N N-X. 1

-«^
>&
\
09
\\ \<3

\o
l»"
V
t3 0.8 L'°
\- V

N i- V
li
lr" \*
tv>
o m o oo o^ .& oj' M o 00 \
o o o o o o o fjl
I"

07

06 —
-
05
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

T
A ;
shells

"W
<c)

10 ^

\"S^^
N ~"==
N
1

> \
"

09 1

Vrf> \ \

\Y*

T3 0.8
1'4 »

N \
IV)
ro
o
oo y
o 071 o oo OJ
o O o o o o o en

07 1 1

D.6

0.5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

id)

(a) One shell pass; two or more tube passes. (6) Two shell passes; four or more tube
passes, (c) Four shell passes; eight or more tube passes, {d) Six shell passes; twelve
or more tube passes.
346 APPLICATIONS TO EQUIPMENT DESIGN

<o)
A
V
Tax

A
if

Coils

Tb, - TBi

Figure 15.15. (A71 /m correction factor for cross-flow heat exchangers, (a) Cross-
flow, both fluids unmixed, one tube pass, {b) Cross-flow, shell fluid mixed, two
tube passes, (c) Cross- flow drip, helical coils with two turns (5). (With permission of
ASME, copyright© 1940.)

coefficient. The reader is referred to Tinker (37) Each tube arrangement presents certain ad-
for a more detailed analysis. vantages. For instance, the square pitch makes the
For most exchangers, a reasonable approxima- tubes more available for external cleaning, and the
tion can be made by simple procedures. The sim- triangular pitch increases turbulence of the shel-
plest procedure is to use 60 percent of the co- side fluid. Accordingly, the triangular pitch will
efficient calculated for flow normal to rectangular give a higher coefficient but also a higher pressure
tube banks, assuming no leakage around baffles drop.
and no ineffective transfer area of tubes. The tube pitch is defined as the shortest
Heat-transfer coefficients have been investi- center-to-center distance between adjacent tubes.
gated (3, 4, 16) for fluids flowing normal to banks The tube clearance is the shortest distance between
of tubes in several different types of heat ex- adjacent tubes.
changers. The resulting correlation is shown in
Figure 15.16 along with the friction-factor corre- Shell-Side Coefficients for Liquid Metals
lation enable
to one to predict pressure drop The significance of the Prandtl number in heat-
through the shell.
transfer coefficients for liquid metals has already
been discussed Chapter 13. Lyon (24) has shown
in
Tube-Sheet Layout
that, for low Prandtl number flowing
fluids of
Several different tube-bundle arrangements are inside tubes, the heat transfer is a unique function
shown in Figure 15.17. of the Peclet number. On the same bases, Rickard,
 HEAT TRANSFER 347

Model Rows Do Pitch /D, >

FRICTION 1 10 \ 1.25

2 2 10 h 125
3 14 h 125
4 10 h 1.50

5 10 h 1.50
1
,. _ {-APg r
)p

Model No. N = ma ss minimum

0*v Gmax velocity at area

04 QilG \ D = tube diameter

^
L^t \/
flow C p Model No. 2\
N. N. N= number of transverse rows

02 Oil

flow
Q-Q
O-O
•^s^^^
w >v

1
\\MC >v
^sSosS^ No.3^ A
.——' —s !

^^O^ ^ 5
Model

—
0.04
^^^ Oil

O
r£*r\ Model No. 4

Qi^<^0 L
Model No.

OO
5~

^S flow
flow O-O
02
HEAT TRANSFER ^^xT^"^^;^ /

001 "**

1
-*\—\ — 1
-Transition zo re -

0.004
10 20 40 10 2 200 10 I0 4
^mai ^n

Figure 15.16. Heat-transfer coefficients and friction factors for flow normal to
tube banks (3). (With permission of ASME, copyright © 1952.)

Clearance

ODOO
Flow

direction
c8£o°
OQQO Pitch

(a) (b) (c)

Figure 15.17. Common tube-sheet layouts, (a) Square pitch, (o) Triangular pitch,
(c) Rotated square pitch.
348 APPLICATIONS TO EQUIPMENT DESIGN

Dwyer, and Dropkin (34) postulated and experi- Table 15.2 TYPICAL FOULING FACTORS (36)
mentally verified that a similar relationship is valid
for flow of fluids of low Prandtl numbers across Fouling Factor, Rj = Mh d
banks of tubes. Their correlation is hr°Fft 2 /Btu

0.67 Water Ve ocity

h Do DoVmzxPCp
= 4.03 + 0.228 (15.20) 3 ft/s or less 3 ft/s or more

Seawater (up to 125°F) 0.005 0.0005

where
Well water 0.001 0.001
h= average heat-transfer coefficient, Delaware and Lehigh
Btu/hrft 2 °F river waters 0.003 0.002
Brine 0.001 0.001
D = outside diameter of tube, ft
= shell-side velocity of fluid based Fuel oil 0.005 0.005
I'max
upon minimum flow area, ft/hr With permission of TEM A, copyright © 1 978.

This correlation is valid for Reynolds numbers

between 20,000 and 200,000.
Fouling
The buildup of dirt or scale on heat-transfer sur-
faces problem that must be considered in
is a
heat-exchange equipment. As a result of the for-
mation of scale, another resistance is added to the
heat-transfer path causing a reduction in the rate of
exchange. If the thickness and thermal conductiv-
ity of the scale can be established, the magnitude
of the scale resistance can be calculated. Unfortu-
nately, the scale continually builds up so that its
resistance is a varying quantity.
Some fouling factors recommended by the
Tubular Exchange Manufacturer's Association (36)
are found in Table 15.2. These factors are treated
as resistances and added to the "clean" tube resist-
ance to give the operating resistance that might be
expected.
Example 15.5. A steel mill desires to cool
3000 gal/hr of an inorganic liquid from 250°F to
180°F by Lehigh River water flowing through a
shell-and-tube exchanger at a rate of 5000 gal/hr.
The liquid is somewhat acidic in nature and so an
alloy exchanger is probably required.

The deposit that increases resistance may be
simple dirt, mechanically deposited; it may be a
true crystalline scale deposited because of the in-
verted solubility curve of some dissolved constit-
uent present; or it may be a deposit originating
from chemical reactions within the fluid or be-
Within the plant is an unused shell-and-tube,
floating-head exchanger have one shell pass and
two tube passes. The shell is mild steel, and the
tubes are type 304 stainless. Specifications of the
surplus exchanger reveal a 20-in. I.D. shell with 158
1 -in. 14 BWG tubes 16 ft long laid out on a 1 ^ -in

tween the fluid and the tube. The rate of deposit is
a complex function of heat duty, flow rate, geo-
metry, and the particular characteristics of the fluid.
Because scale buildup varies with time and its
increase accompanied by a drop in the transfer
is
square pitch. Baffles are placed 6 in. apart.
Will the surplus exchanger be suitable for the
cooling requirements? Assume the inorganic liquid
has the physical properties of water.

rate, some optimum time exists for operation Solution. Since the liquor to be cooled is
before cleaning. True scale will be shown in corrosive, should be placed in the tubes. The
it

Chapter 19 to be a parabolic function of time. An
exponential decrement in coefficient has been used
to express buildup in other cases. Scales may ap-
proach an asymptote, or they may "mushroom"
rapidly once they start (29).
The absence of a definitive theory regarding
scale formation usually requires actual perform-
ance data to help in determining length of service
before cleaning. The limit in permissible fouling
may be expressed in resistance units or as a per-
centage of the clean coefficient below which opera-
tion becomes impossible or uneconomical.
tubes of 304 stainless will probably be acceptable
for this purpose. The river water will in all prob-
ability be dirty, and the exchanger should have a
removable tube bundle. The square pitch of the
tube layout is good for cleaning the outside of the
tubes. Consequently, from a point of view of
materials of construction and ease of maintenance,
the surplus exchanger seems adequate. But, this is
not enough, because the heat load on the ex-
changer will demand a certain transfer area. It will
therefore be necessary to estimate the cooling sur-
face required and check this against the available
 HEAT TRANSFER 349

area. load can be evaluated from the

The heat The viscosity of the liquid at
7"
av =
temperature reduction of the inorganic material. (250 + 180)/2 = 215°F is 0.25 centipoise. The flow
character is determined by evaluating A/ R e
w= (3000 gal/hr)(8.34 lb/gal) = 25000 Ib/hr

- TB: DG 83,000
q = wc P {TB2 ) /V Re = (0.834/12) = 9550
(0.25x2.42)
= (25000 lb/hr)(1 Btu/lb °F)(250 - 180°F)
L 16)
= 1,750,000 Btu/hr = 230
D (0.834/12)
From this heat duty, the temperature rise of the
cooling water can be calculated Thus, j q (Appendix C-5) = 0.0038.
To calculate the Prandtl number,
1,750,000 Btu/hr 2
(TA 2- Tai)- cP = 1.0 Btu/lb °F, k = 0.402 Btu/hr ft (°F/ft)
(5000 ga /hr)(8 34 b/ga |) | ,

Therefore,
=42°F
To be safe, the cooling-water
(1 Btu/lb °F)

entrance tem-
A/ Pr = —
cP n

k
=
(1)0.25x2.42)
0.402
= 1.5; /V Pr
2/3 _
1.31

perature should be based upon the maximum river-

and
water temperature, say 80° F. Thus, the water will
enter at 80° F and leave at 122°F. Therefore, the hi_
=a003e
true countercurrent (AT) lm for this exchanger is
Id
c P G' Mr) (7)
(250- 122)- (180-80)
tAT) >» =
(250-122)
° m5 ° As a first trial,
the above equation for/7, gives
assume (pt /s //Lt)
01 4
1, then solving

In
(180- 80)
(0.0038) 1)(83,000) (

hi = = 240 Btu/hr ft
2
°F
For a one-shell-pass, two-tube-pass exchanger,
(1.31X1)
Figure 15.14a is used

X= —7-T
TB
^-T
TA
— B

B
^

,
=
180-250
80-250
= 0.41
Shell-Side Fluid (cooling water)
Flow rate = 5000 x 8.34 = 41 ,600 Ib/hr
Flow on a 1| in. square
Z= —
7^1-7^2
TB2-T
—
B \
=
80-122
180-250
= 0.60
The number
area: 1-in. O. D. tubes
pitch gives a clearance for fluid flow of 0.25
tubes at the center of the ex-
of
in.

Therefore, from Figure 15.14a, at X= 0.41 and changer may be evaluated by dividing the shell
Z=0.60, V = 0.97 diameter by the tube pitch, and subtracting one
from the resulting number, or
(A7) C orr= Y(AT),m =(0.97)113.5= 110
20
Tubes at center = 1 = 16-1 = 15
Tube-Side Coefficient (inorganic fluid) 1.25

1-in. 14 BWG tubes: Taking the length of the shell-side flow area as
equal to the baffle spacing, the minimum flow area
Dj = 0.834 in. (Appendix C-7)
for the shell-side fluid thus becomes
2
tt „ /(0.834) \
2
Sj = -Dj 7 = (0.785) ( = 0.00381 ft /tube
S sh e„ = = 0.1566 2
4 \ 144 / <15)(^)(^) ft

Therefore, Therefore,
(no. of tubes) (cross-sectional

Total flow area =

area per tube)
G m ax = ^^- = 266,000 Ib/hr ft
2

(no. of passes) 0.1566

av of 101° F = 0.68 centipoise

7"
158x0.00381 2
ju at
S/ = = 0.301 ft
2 and

25,000 D o G max 266,000

= 2 = = .
m 13 46Q
Mass velocity G = 83,000 Ib/hr ft Re ( f }

0.301 u (0.68x2.42
350 APPLICATIONS TO EQUIPMENT DESIGN

From Figure 15.16 Gm ax '

= 0.0085 A7* = 31 °F. Therefore, the average outside surface
tamperature = 132°F, and Hos = 0.5 centipoise.
Therefore
h — 0.14
=
/
o 5 x 0.14
=0.958
&)(?ffir-' u ""
i

\ u I ^0.68'

0.1 4 _ Adjusting,
at 101°, c P = 1, k = 0.350, assume in os /n) = 1

483
CpH (1X0.68x2.42) h = r^Z^ = 505 Btu/hr ft
2
°F
2/3 _
/v Pr = 4.7; (/Vp r ) 2.81 0.958
k 0.350
and Tube-side surface temperature is

(0.0085) 1X266,000)(
0.0042 AT/
h = 805 Btu/h rft 2 °F
(2.81)(1) ~
0.00663 VI4
Take 60 percent: Ar, = 72.0°F
2t
h„- = 0.60 x 805 = 483 Btu/h rft >F and the surface temperature is 215 — 72 = 143°F,
The the larger of the
tube-side resistance is
and n is = 0.45 centipoise
two fluid resistances; consequently, the overall 0.1 4 4
/0.45\,0.1
resistance will be based upon the inside tube area. = 1 .085
Therefore, summarizing the resistances on this
(?) ~\0.25/

basis and neglecting the fouling factor for the Adjusting,

moment give
240
= = 22,Btu/hrft20F

Tube-side fluid,
i
i
1
I
Resistance
= 0.0042
"'
T^
hi' 240 The new total resistances becomes
0.083(0.218)
Tube,
kA (12X9X0.240)
= 0.0007
Tube side, -1
= —-
1
= 0.00453
hi 221

Shell-side fluid,
A 1(0.218)
= 0.00173
AxAj
Tube, = 0.00070
h Ao 483(0.262) kA

Total resistance
'
1

U;
= 0.00663

hr ft 2 °F/Btu
Shell side, —
h
A,

A
=
0.834
-nr
505
=0.00165

2
and Total,— = 0.00688 hr ft °F/Btu
1

Ui = = 150 Btu/hr°F ft
2
inside area
0.00663 Thus the corrected overall coefficient (dirtfree) is

Using this overall coefficient, the surface co- 1

= 145.0 Btu/hr 2
efficients may be readjusted to account for ((y' )c,ean ft °F
14
0.00688
(/i /V)
s not being unity. The surface temper-
-

ature may be calculated by the usual manner of The required area is.

proportioning resistances and driving forces based 1,750,000

upon coefficients calculated in the first trial. The A =
UiY(AT) lm (145X0.97X113.5)
= 1 1 1 ft
2

outside surface temperature is determined from

Shell-side resistance shell-side driving force The total surface actually available in the exchanger
is
total resistance total driving force
,0.834;
0.00173 = (158)(16)(3.14)(—-) = 550 2
A,- ft

0.00663
Thus, the required area is considerably less
A7"
than the available area so that, even with scale
(215- 101: buildup as operation continues, the heat exchanger
 HEAT TRANSFER 351

should perform the necessary cooling. To verify Either part of Equation 15.21 is usable, de-
the allowable dirt factor, calculate 1/1/ using the pending upon the information at hand. For film
actual surface area, or condensation on horizontal tubes (25)

AiY(AT) lm
i
1 -
(550) (0.97) 113.5) (

h = 0.725
k f 3 p f 2 g(AH) 1/4
= 0.95
3 2
[kf p f gL
1/3

\Uj>) dirty q 1,750,000 NDp f {Tsv -Ts ) llfW

= 0.0346 hr 2
ft °F/Btu (15.22)
The permissible scale resistance is then
where N = the number of tubes in a vertical row of

R d =- — = 0.0346 - 0.00688 tubes.

Uid Uic
= 0.0277 hr
This allowable scale resistance is significantly
greater than the value for Lehigh River water
reported in Table 15.2. In the event that the re-
quired area and available area were closer together,
the dirt resistance could be much more important
because frequent cleaning would be necessary to
keep the exchanger performing as specified. This
solution is not complete in that pressure-drop re-
quirements should also be examined, as well as the
economics of the operation.
The condensate properties are evaluated at
2
the average film temperature (TV), recognizing that
ft °F/Btu
the variation in condensate thickness around the
tube, or down it in the case of vertical tubes, will
cause a considerable variation in the tube-wall
temperature, McAdams (25) suggests for 7>
Tf =[Tsv -l(Tsv -T s )] ;15.23)
Equation 15.22 is written on the assumption
that the condensate flows on or over the tube
surface in laminar flow. Equation 15.21 includes a
28 percent increment above the theoretical equa-
tion to account for the presence of turbulence. No

Condensers conclusive correlation for turbulent-flowing films is

yet at hand, but Kirkbride (23) and Badger (1,2),

Heat exchangers whose prime purpose is to bring
investigating the condensation of diphenyl and
about a change of phase from a vapor to a liquid
"Dowtherm" on vertical nickel tubes, suggest
by means of a coolant are commonly called con-
densers. Chapter 13 introduced the condensation 1/3
4w l 0.4r*.3 ,2
coefficient and the mechanism of dropwise and p y
h = 0.0077 15.24)
film condensation. Most commonly encountered PfnDl p;
condensation operations in the process industry are
of the film type. if the flow at the bottom of the tube is turbulent.
The same basic design equation used for heat
exchangers is valid, but the nature of the co- Vaporizers

efficient correlations is somewhat different, de- A vaporizer is a heat exchanger specifically de-
pending upon whether the condensation process signed to supply latent heat of vaporization to a
takes place on vertical or horizontal tubes. fluid. the vapor formed is steam, the exchanger
If

For vertical surfaces and film-type conden- is commonly called an evaporator. If the exchanger
sation, McAdams (26, 27) recommends is used to supply the heat requirements at the
1/4 bottom of a distillation tower, it is called a re-
3 2
k f p f g(AH) kftpSgnDV' 3
h = 1.13 = 1.18 boiler. Vaporizers satisfy the latent heat require-
LLnf(Tsv - Ts ) HfW ments to a boiling fluid, and the heat duty for this
:i5.2i: type of exchanger is easily computed. Many dif-
where ferent design arrangements are available depending
upon the service.

AH = latent heat of condensation, Btu/lb Heat-transfer coefficients for vaporizers are a

2 complex function of such items as flow rates, frac-
g = gravitational acceleration, ft/hr
L = length of tube or vertical surface, ft tion of liquid vaporized, transfer area, physical
Tsv = saturated-vapor temperature, °F design, etc. Practical design heat fluxes are of the
2
Ts = surface temperature, °F order of magnitude of 20,000 Btu/hr ft . Design
D = outside diameter of tube, ft procedures for vaporizers and reboilers may be
w = condensate, Ib/hr found in Kern (21).
35: APPLICATIONS TO EQUIPMENT DESIGN

EXTENDED SURFACES FOR HEAT TRANSFER Referring to Figure 15.18, let

The desirability of extending the surface of a tube L = the height of the fin
Ty = temperature at the base of the fin
by addition of fins or spines has already been
discussed. Design calculations for finned heat ex- T= temperature of the fin at a distance x
changers are complex because of uncertainties in from the base
knowing the temperature driving force distribution T = temperature of fluid surrounding fin, <Ty
over the resistances between the fluids exchanging A = transfer area of fin
heat.
P = fin perimeter
h f = heat-transfer coefficient for fin
Longitudinal Fins Consider the element of the fin {dx) as shown in
The simplest type of fin is the longitudinal metal Figure 15.18
strip extending lengthwise down a tube. To under- The net heat flow into this elemental slab by
stand the mechanism of this type of exchanger, conduction from the base of the fin is
one may consider the steady flow of heat along a
thin fin extending from a tube wall. Figure 15.18 q tn = -kA\-^)dx (15.25)
illustrates a portion of such an exchanger.
The mathematical solution to the heat-transfer The heat flow from the surface of (dx) by
equation for this configuration has been offered by convection must be
Gardner (14), based upon the following assump-
tions: <7out =h f (T- T )Pdx 15.26)

Steady state exists so that the temperature

1
But to maintain a constant temperature in dx re-
quires that Q in = -Qout- Adding Equation 15.25
distribution is constant and heat flow through the
fin is uniform. and Equation 15.26 gives
2. No heat source exists within the fin.
3. The fin has a constant thermal conduc-
-kAl-r)dx+h
z
(T-To )Pdx f
= (15.27)
y
dx '

tivity.
4. The heat-transfer coefficient is uniform Subst ituting the defined factors 6 = T — T
over the entire fin surface. and X -\I hfPIAk into Equation 15.27 and re-

5. Ambient temperature is constant. arranging gives

6. The temperature at the base of the fin is

d 2d
constant. —z-\ ,
2
6 = (15.28)
7. The fin is thin with respect to its height
dx z
and so no temperature gradient exists across the The solution to Equation 15.28 is

width of the fin.

8. The heat transferred from the fin tip is
6X = C y
cosh Xx + C2 sinh Xx (15.29)

negligible compared with the heat leaving the fin

The constants C-\ and C2 can be evaluated by
sides.
substituting the boundary conditions.
9. Good contact exists between fin and tube.
Atx = 0, 6 =0 1f and d =6,

At x = L, = —
dd
by assumption 8 of Gardner;
dx
therefore, from Equation 15.29

—
•
dd
= Xd sinh XL + XC 2 cosh \L = (15.30)
dx
and

sinh XL
C 2 =-0i
cosh XL

The temperature distribution is given by

Figure 15.18. Longitudinal fin. 6 X = Qy cosh Xx — d y tanh XL sinh Xx ( 1 5.31

 HEAT TRANSFER 353

or the entire length of the fin. Equation 15.32 applies

cosh X{L — x) only to the fin and not to the unfinned surface of
e* = 115.31a) the tube. The fin efficiency will always be less than
cosh XL
100 percent, since a unit area of fin surface isnot
Rewriting Equation 15.31a gives as effective as a unit area of tube surface. A tem-
perature gradient within the fin must exist to cause
(AT) X coshX(Z--x)
:i5.3i/7> conduction through the fin.
AT, cosh XL An empirical correlation for evaluating fin-
where side coefficients for longitudinal fins in double-
(A7") x = temperature difference between the pipe heat exchangers has been developed by Kern
fin and the surrounding fluid at a (22) for numerous heating and cooling experiments
distance x from the base of the fin = on different varieties of equipment arrangements.
TX -TQ The correlation is shown in Figure 15.19.
AT, = temperature difference at the base
of thefin = T, — T Transverse Fins

Figure 15.5 illustrates several types of transverse

A fin efficiency may be defined as
fins.The fin-efficiency equation for transverse fins
(AT) m tanh XL is somewhat more difficult to develop than that for
Vf = 15.32) the longitudinal The mathematical develop-
fin.
A7-i XL
ment of transverse fins has also been accomplished
where (AT) m is the temperature difference be- by Gardner (14). Consider the most general type of
tween the fin and surrounding fluid averaged over transverse fin, that of varying cross section. The

1000 1
;
1

100

f/
Jr
=* a
i

/
/
a.
a. -as

10

G = mass velocity

10
100 1000 10.000 100.000

--^
'Re

Figure 15.19. Heat transfer in longitudinally finned heat exchangers (22). (With
permission of McGraw-Hill, copyright © 1950.)
354 APPLICATIONS TO EQUIPMENT DESIGN

TRANSFER UNIT AND EFFECTIVENESS

As indicated in Chapter 13, the complexity of

turbulent transport has defied theoretical analysis.
For the estimation of required equipment in de-
signs, for the analysis of performance and the
evaluation of production, we must resort to ana-
logs and similarity theory. Fortunately, these do
give us very usable transfer proportionality con-
stants and the potentials of trans-
(coefficients),
port can frequently be evaluated on the basis of an
appropriate mean.
Two concepts of chemical equipment per-
formance have recent years been joined and
in
Figure 15.20. Varying-cross-sectional transverse fin.
incorporated into heat-transfer calculations with
significant advantage in certain types of calcu-
simplest fin of this type is sketched in Figure
lations. The transfer unit as an expression of the
15.20.
performance of equipment, or the difficulty of the
In this type of fin, the surface available for
transfer in a system, has been used for many years
heat transfer is a function of L.
in mass-transfer calculations. Fuller development
The net heat entering the differential section
of the concepts will be given in the later treatment
[dx) by conduction from the base is
of mass transfer. The transfer unit is an evaluation
of the amount of transport between phases ex-
d (
~\kA
dx\
—
dT\
dx'
\ pressed in terms of the increase in the intensive
property accomplished by the transfer, in ratio to
the driving force for transfer. Both of these are
Heat will leave this elemental section by con- expressed in terms of a difference in an intensive
vection in amount dq = h f ATdA. For thermal factor. Thus, the transfer unit for heat is the ratio
equilibrium to exist in dx, the decrease in conduc- of the temperature increase resulting from transfer
tion through the fin at that point must equal the {jdT) to the driving force (AT). If this ratio is
convection from the fin, or unity, the equipmentperforming as one transfer
is

unit; if the number is greater than unity, more

— (kA — )-h f ATdA = ; 15.33) transfer is being accomplished per unit of driving
dx\ dx! force (9).
In recent years, the great advantage of the
If the thermal conductivity is assumed constant, transfer unit for certain cases of heat transfer has
Equation 15.33 becomes been recognized (19), and it is being used more and
more frequently in heat-transfer calculations in
d*T
kA—^-
2
dx
+ k[ — — \-h ATdA=0
ldA\[dT\
\dxJ\dx'
f (15.34) connection with another old, though infrequently
used concept — the effectiveness e of a piece of
or heat-transfer equipment. Effectiveness is the heat
quantity actually transferred divided by the trans-
d2T 1 [dA\ (dT\ hf fer of heat that is thermodynamically possible.
—
dx
~ \— )\—
T + A\dx'\dx/
2 I
kA
~AT=0 These heat quantities are most conveniently ex-
(15.34a) pressed in terms of a time-heat-capacity c min =
(wCp )m in for that stream having the lower value of
This second-order differential equation has mass flow per unit time times heat capacity (wcp )

been solved by Gardner using Bessel functions. in units of Btu per hour or J/s. The actual transfer
Efficiencies calculated from Equation 15.34a may expressed in terms of the smaller stream then
be found elsewhere (20). It is generally advisable to would become (wcp min x (T2 — T-\) min for the
)

base design calculations on experimental values of smaller stream and is divided by (wcp min x )

h f for the particular fin design being used. This (7"m,n m. - 7"
max i
Obviously, the
n .) to get €.

information may usually be obtained from the smaller stream is the one having the wider range of
manufacturer. temperature span in the equipment. Choosing the
 HEAT TRANSFER 355

hot stream to have c min makes the direction of x(liquid-phase concentration) for the analysis of
fluxes and increments more comfortable intuit- mass-transfer operations as well as distillation. In
ively. these a wide variety of concentration units compli-
These concepts combine to facilitate calcu- cate the problem. Figure 15.216 plots as the oper-
lations of heat transfer in evaluating the perform- ating line the temperatures of the two streams at
ance of existing equipment, and in expressing the each point through the exchanger. Obviously,
performance of equipment handling multiple when time-heat-capacities (C) are constant, this
streams. They become particularly useful if mul- will be between the terminal temper-
a straight line
tiple streams are flowing in complex patterns atures. The arbitrary choice of the hot stream as
through a multiplicity of channels in the same the minimumcan be shown to be inconsequential,
piece of equipment, and with different degrees of but does keep the signs on fluxes and driving
it

mixing of the two streams. In the design specifi-
cation of new equipment, with conventional
geometry and known coefficients, these tools offer
little if any advantage over conventional techniques
using UAAT calculations. For a true counterflow
exchanger, with no change in phase, the advantage
appears only in evaluating q and the outlet temper-
ature, thus avoiding a trial-and-error solution for
forces more comfortable.
With the hot stream assumed the minimum,
the operating line is always above and steeper than
the equilibrium (45°) line. Its span on the ordinate
represents the temperature increment of the hot
stream, the span on the abcissa represents the tem-
perature increment of the cooler stream, and the
slope (m) represents the ratio of C max /Cmjn The .

temperature attained in an exchanger of known AT at any point is represented by the separation of

surface. the operating line from the equilibrium line,

A graphical representation such as in Figure measured parallel to either axis.

15.21 facilitates the presentation and development If we consider an increment of contact area
of the design equation. Figure 15.21a is very dA, we can express the transfer in a differential

similar to Figure 15.76 and assists in visualizing expression including each stream:
coexisting temperatures through a counterflow ex-
-Ch dTh = dq = -Cc dTc = UdA{Th - Tc )
changer; it can be plotted against area or heat
transferred as desired. Figure 15.216 presents the (15.353,6, c)
counterpart of the operating and equilibrium lines
It will be necessary at times to use the relation
in an x—y field for the McCabe-Thiele analysis of
between the expression of heat transfer in classic
distillation. In heat exchangers, the concentrations
(intensive factors) are the temperatures in the two UAAT terms, and that in terms of the number of
streams. In the analysis of heat exchangers, the
transfer units Nq We can
. write

temperatures coexisting in the two streams are UdAAT = wepdT = G S p dT

plotted against each other to generate an operating
line; the equilibrium curve is always the 45° line.
Arbitrarily, 7"hot is plotted on the ordinate and
7"coid on the abcissa. The same form will be widely
used in a plot of /(gas-phase concentration) versus
where, as usual, AT" refers to a temperature differ-
ence between the hot and cold streams, and w(the
mass flow rate) can be expressed in terms of mass
velocity and cross section GS. Cp is defined as wc p

Maximun stream temp. (Tc )

(b)

Figure 15.21. Development for true counterflow case.

356 APPLICATIONS TO EQUIPMENT DESIGN

or GScp. Rearranging, priate terms for fouling resistances and extended

surface as needed.
U dA Development of the relation of e and Nq will
(15.36)
Gcp S AT be illustrated first by the simplest case of pure
counterflow, with constant Cc Chl and U. It is
which leads to ,

developed for the case of Ch min but the =C ,

UdA dT counterpart case leads to the same result. Figures

= d(N :15.37) 15.21a and 15.21/? are used.
Cm in A7"

A counterpart equation can be written in terms of True Counterflow

the larger stream, but it is incompatible with
Writing an energy balance on dA
thermodynamic effectiveness. Integrating,
dQ = -Ch dTh =-C c dTc
Nq =
1

j
UdA = — 15.38) 1 15.35a/?)
^m in
and a rate equation
For the case where essentially all resistance is

in one film, and most of the temperature drop is

dq = U(Th -T c )dA (15.35c)

across that film, the parallel is obvious: The operating line has the slope

1
"a vg^ dTh Cc
N< = hdA = 115.39) m= = constant (15.40)
o C, dT,

In the analogies of heat, mass, and momen- By definition,

tum transfer, it is shown (Figure 13.10) that Th i n — Ti/tout

e = (15.41)
U
<7 ,//V ^/8)=0'(/V R e,/Vp r ,/V Sc ) ' h in ' cin

For a dominant-film resistance, Equation 15.39 Where 6, 5', and A are indicated in Figure 15.21/?
becomes and are used only as convenient shorthand nota-
dT h dA -2/3
dA tion.

at" Gcp S
= h A/Pr
~S~ Subtracting Equation 15.35a from 15.35/?,

which simplifies to 1 1
-T
:

d(Th )
= dq (15.42)
c
A LCC Ch
Nq - h N Pr -2/3 '

S Equations 15.42 and 15.35c yield

For some of the complex shapes of heat ex- d(Th -T c) U

changer apparatus, it is well to realize that AIS is ['- dA (15.43)
Th — Tc L Cc iCh
related to the hydraulic radius. The force balance
used in Chapter 13 to develop the equivalent diam-
eter rearranges to Integrating from I to II, substituting Nq =
D S • L flow volume
^m in
_
^wetted heat transfer area
In[^^]=-/Vjl-^) (15.44.
13.20)
or
ch
n, s Th 1
1
— Tc \

= = e } (15.45)
L A Th\ — Tc \

That is, in a complex or irregular exchange surface, Now, from the geometry of the figure
S/A may be used to express the D/L ratio. r*„-7-c „=(A-5)
an overall heat transfer coefficient is to be
If

used to evaluate Nq through Equation 15.38, it is

evaluated as from Equation 14.32, with appro-
Thi -T ci =(A-5') = (a--)
 HEAT TRANSFER 357

or

Th 1 1
— T,ell -5/A 1 Hot fluid (WC) h = c h
^—~* Heat transfer
Thi -T ci ^-8/(A^m)
Cold fluid (WC) C =C C
and from Equation 15.41
1 -e
15.46) 100
^-e(Ch /Cc ) Cmin /Cmax — 4>--* *^^^** ^ ^L^--*-
,
- - '

Combining Equations 15.46 and 15.45

80
i i s* (.

tiSfi

,.0-'
^-^^_-
1 -e -N q (y-ch ic c voo-
= )

1~(CJCC ////
60
Rearranging
/////
1 - e -N q a-ch ic c )

€ = 15.47) 40
W\
1
- N q (y-c h ic c )

In the general case, 20 1

becomes Cmin
Cf,

Cc becomes Cmax n r 12 3 4 5

This equation is easily graphed as in Figure

No. of transfer units, NTUmax = At//C min

15.22. For Ch /Cc =0 (condenser or vaporizer), Figure 15.22. Counterflow exchanger performance (19).
e = 1 -e-'V ForC„/Cc = 1,* (With permission of McGraw-Hill, copyright © 1964.)
i\>.

€ = 15.48) tube-and-shell exchangers with straight tubes as

1 +N r presented in Figure 15.3 accomplish the "1—2
parallel-counterflow" contact by proper arrange-
True Cocurrent Flow ments of the flow-channels at the ends. As illus-
Similar development yields trated in Figures 15.14a, b, c, and d, large ex-

changers use serias arrangements of that element to

provide large areas in a minimum volume. Counter-
_ p \ ^max/
parts of Equations 15.47 and 15.49 have been
e = 15.49)
!

evaluated and are available as numerical or graph-

^m in
1 + ical solutions similar to Figure 15.22 in Reference
19.
The correction factors presented in Figures
This equation may be graphed most conveniently
15.14 and 15.15 were developed by a slight alter-
ase versus Nq with C min /Cmax as a parameter.
of the method used here; the equations
,
ation
These developments for pure coflow and
yielding Figure 15.14 can be altered to yield
counterflow are important for appreciation of the
concepts, but most industrial heat exchange equip-
€ =
ment uses complex flow patterns, either as a conse- + e -r
quence of choice of a hairpin-shaped tube or to d +Cmin /Cmax )+Vl +(Cmin /Cmax )2 (-
1
—
1-e"
allow flexibility in the relative cross section for
flow of each fluid. These were discussed and illus- 15.52)
trated in Figures 15.13, 15.14, and 15.15; the
drawings imply hairpin tubes in the "1—2 parallel-
where r = Nq Vl + (Cmin /Cmax )
2

counterflow" element of Figure 15.13. Many Conceptually identical with multipass exchangers
* Equation 15.48 requires use of consisting of several one- and two-pass units in
Simplification to
L'Hopital's rule. series is the case of a very large exchanger broken
358 APPLICATIONS TO EQUIPMENT DESIGN

up into several identical smaller units arranged in about 0.75—0.8 (note the sharp drop in Y in
overall counterflow. This is usually done to facili- Figures 15.14 and 15.15), or an N greater than
q
tate field-handling (cleaning, retubing, warehousing approximately 4.0 (note the negligible increase in e
of spares). For these operations, it is easier to beyond Nq = 4 in Figure 15.22), serious considera-
handle modest-sized units than to handle a tion should be given to a different concept.
behemoth.
In each of these configurations, the overall
effectiveness e of 77 identical units of individual
THERMAL ENERGY TRANSFER BY RADIATION
effectiveness ep can be evaluated: So far, the transfer of thermal energy has been
considered when it occurs by the mechanisms of
-1 conduction and convection or turbulent diffusion.
1 -e„ In these cases, the transfer rate can be determined
€ = 15.53)
by integration of the usual rate equation, and
equipment sizes can be determined by combining
1 -e„ the rate equation with appropriate material and
Kays and London (19) have plotted relations energy balances and the proper design equation.
of e as ordinate against N q as abcissae for most Simultaneously and independently, energy is trans-
of the common configurations. These bear an inter- ferred by radiation. Thus, the relation
esting relation to the Bowman, Mueller, Nagle
AT-correction plots (Figures 15.14 and 15.15), in Q = Qc + Qk +Qr 115.54)

which X is V is q and
related to effectiveness, MN ,
where
Z is a CIC not restricting the assignment of
ratio,
C min to the numerator. q = total rate of energy transfer as heat
One additional variation must be included, qc = rate of thermal-energy transfer by the
relative to the uniformity across a path of the mechanism of convection
temperature of the fluid. If a fluid is constrained qk = rate of thermal-energy transfer by the
by tubes, or between flat fins, there is little chance mechanism of conduction
that the temperature at any point of progress
qr - rate of thermal-energy transfer by the
through the exchanger will be uniform across mechanism of radiation
the path of the stream; conversely, for the shell-
side fluid in a tube-and-shell exchanger, tempera- is applicable in every case. Of course, in many
ture approaches uniformity across the path at any cases, one of the mechanisms may transfer a
point of progress. The effectiveness is obviously a negligible amount of thermal energy compared to
function of the extent of mixing of each stream. one or more of the others. In this section, the

Figure 15.15 shows some other configurations in phenomenon of energy transmission by radiation
which the streams may have different extents of will be considered in some detail. The fundamental
mixing within themselves. mechanism involved and the quantitative relations
For cross-flow, one fluid "mixed," other describing this mechanism will be developed, for
"unmixed" for: = Cu n m ixed ;C n the transfer of radiant energy does not seem to be
adequately described in terms of rate equations of
^m ixed
— F Cmax/*-min the form considered in the preceding chapter.
e = 1 -e (15.50)
where Equipment Transferring Energy Mainly by Radiation

F= 1 — Nq C m j n /C max Radiation is the major energy-transfer mechanism

in a large variety of chemical-process equipment.
for:
Electric heaters, direct -fired kettles, steam boilers,
''max ^rriixed' ^rnin ''unmixed rotary kilns, blast furnaces, and petroleum pipe
are examples of such units. Figure 15.23
e- rc m[n/c ma x) (15.51)
stills is a
e = Cma x/Cmin (l -
schematic drawing of a petroleum pipe still. Fur-
where naces like this are built for heat-release rates up to

r' = 1-e
-
^ 160,000,000 Btu/hr and are a common unit in
most petroleum refineries.
A rule-of-thumb generalization should be The major heat exchange occurs from the
recognized: when a design produces a Y less than flame to the single row of vertical tubes around the

wall of the furnace. The refractory-brick
walls and the metal reradiating cone
radiant energy back to the heating tubes. In this
furnace, convective heat transfer
between the flue gases confined in an annular ring
around the reradiating cone and the heated tubes,
which are finned in this section. A final
convective-heating section is installed at the base of
the stack. These vertical furnaces require an un-
usually short stack to supply the necessary draft
because the flue-gas travel is uniformly vertical,
eliminating the flow-direction changes that cause
much of the pressure drop in horizontal furnaces.
The monorail attached to the top of the stack
permits the removal of the heating-element tubes
vertically through the header-box doors.
The Nature of Radiation
Although the mechanism of radiant-energy transfer
is not completely understood, the associated
phenomena are explainable in terms of a dualistic
theory. This theory deals separately with the
emission and reception of radiation, and with its
transmission. Radiation is emitted and received in
discrete particles or pulses called photons. The
energy transmitted in each photon is a function of
the frequency of emission but is otherwise con-
stant. The emission of energy must lower the
energy of the emitting body. In terms of the Bohr
model of the atom, which pictures electrons orbit-
ing around the atom nucleus at various energy
levels, emission would occur as electrons jump
from a high-energy orbit to a lower-energy orbit.
Emission also occurs by lowering the vibrational
energy of the nucleus with respect to other nuclei
in the molecule or by decreasing the rotational or
molecule. The
translational motion of the
frequency of emission is a function of the energy
levels of the emitting atoms, and, since any finite
body contains atoms in a wide range of energy
states, the emitted radiation will have a complete
In transmission, radiation
spectrum of frequencies.
has all the properties of an electromagnetic
— that
wave is, a vibrating electric
similarly vibrating magnetic field. The planes of
vibration of these fields are perpendicular to each
other, and the directions of vibration are perpendi-
cular to the direction of radiant transmission. The
speed of the beam is fixed by the medium through
which it travels. In a vacuum, the speed is about
186,000 miles/s, whereas in denser media it is
slightly less. In any case,
C=A/7
HEAT TRANSFER 359
chamber where
reflect the
C= velocity of the electromagnetic beam
X = wavelength, or distance from energy
is obtained
peak to energy peak
n = frequency of radiation, or the number
of energy peaks passing any point per
unit time
The problem of the wave and
reconciling
particle properties of radiant energy is identical to
that for electrons and atoms. It has gradually
become understood that these properties are dif-
ferent aspects of the same phenomenon and that
the aspects are necessary and complementary. The
uncertainty principle of Heisenberg states that it is
impossible to determine simultaneously the
velocity (or associated energy or momentum) and
position of a particle. Thus, experiments designed
to determine particle velocity find that radiant
energy has the characteristics of waves, whereas
experiments designed to determine particle posi-
tion find that the energy has particle properties.
The wavelike appearance of energy probably arises
from the identical form of equations describing
wave motion and those describing the probability
of finding a photon at a given location.
Upon photons travel
striking a receiver, the
into it they strike an electron or nucleus that
until
is susceptible to the energy level of the photon.
The collision results in an increase in energy of the
receiving atom, as an increased amplitude of vibra-
tion of the nucleus, as an increasing rotational or
translational energy of the molecule, or as an
increase in energy level of one of the orbiting
electrons. Since solids have closely spaced atoms,
radiation is usually absorbed very close to the
surface. For liquids, the penetration would be
greater, whereas for gases, much greater penetra-
tions are obtained. The depth of penetration will
depend upon the characteristics of the incident
photons as well as upon the characteristics of the
receiving body.
Thermal Radiation
field and a
All solids, liquids,and most gases receiva and give
off energy in the form of electromagnetic waves,
which are similar whether they are perceived in the
form of thermal energy, light, X-rays, radio waves,
or radar signals. These various forms differ only in
wavelength. Light waves, those perceived by the
human eye, have wave lengths ranging between
0.35 and 0.75 micron, whereas electromagnetic
waves received as thermal energy have wavelengths
between 0.3 and more than 10 microns (10 3
360 APPLICATIONS TO EQUIPMENT DESIGN

Monorail-

A-A

B-B

Inlet Outlet

C-C

Figure 15.23. Petroleum pipe still with auxiliary convective heating sections.

(Courtesy Petro-Chem Development Co.)

 HEAT TRANSFER 361

microns = 1 mm). The spectrum of wavelengths and partly transmitted, the proportions of the
radiated from any body will depend upon its incident energy that are absorbed, reflected, and
temperature as well as its surface characteristics, transmitted depend primarily upon the character-
with lower temperatures producing longer wave- istics of the receiver, and to a lesser extent upon
length radiation. the wavelength of the radiation and the tempera-
Thermal and light radiation are fundamentally ture of the receiver. For example, most solids
identical, and the laws of optics apply to both absorb or reflect all the radiant energy that strikes
these radiation forms. Radiation travels in straight- them, but glass is transparent to radiation of short
line paths from any emitter. It can be received wavelength, and for long-wavelength radiation it is
directly only by bodies that can be "seen" by the almost opaque. This is shown in Figure 15.24.
emitter. Receivers shade the area away from the These considerations can be stated more concisely
emitter, so that objects in this space receive no as
radiant energy directly from the emitter under
p + oc + t= 1 (15.55)
consideration. Of course, objects so shaded receive
radiation from all the other objects they can see. where
Such objects may also receive reflected radi-
p = reflectivity, the fraction of the total
ant beams, so that an object is rarely shaded
incident radiation that is reflected
completely by another body. As a radiant beam a= absorptivity, the fraction of the total
travels through space from object to object, it may
incident radiation that is absorbed
be partly absorbed or deflected by the gas through r = transmissivity, the fraction of the total
which it passes. Many common gases such as water, incident radiation that is transmitted
carbon dioxide, and hydrocarbons absorb and through the body
generate appreciable amounts of radiant energy.
On the other hand, monatomic and symmetrical This equation obviously only a form of energy
is

diatomic gases such as hydrogen and nitrogen are balance. It means of dispersal of
traces the possible
almost completely transparent to thermal radiation the radiant energy arriving at a point. For most
because their nuclei and electrons cannot be ener- solids and liquids, it simplifies to

gized by photons of the available energy levels.

p + oc= 1 (15.56)
Deflection of the radiant beam occurs when the
gas space carries dust or other finely divided solid For glass, the situation is quite different. For
particles. energy of some wavelengths striking the glass (for
example, light in the near infrared range), r would
The Reception of Thermal Radiation, Kirchoff s Law approach 1, and therefore p and would be very
ex

As with optical waves, thermal radiation falling on small. The beam of infrared radiation would pass
an object will be partly absorbed, partly reflected, through virtually unchanged. On the other hand,

02 0.4 0.6 08 1 20 2.4 26 28 30 3.2 34

Wave length (X), microns

Figure 15.24. Transmission of radiant energy by window glass (33). (With per-
mission of Chemical Engineering, copyright ©
1958.)
 362 APPLICATIONS TO EQUIPMENT DESIGN

for radiation in the far ultraviolet frequency range, For the black body,
t would be nearly 0, and almost all the incident
^furnace-to-black body
radiation would be either absorbed or reflected. = eb
Here, Equation 15.56 applies. In this case, most of A
the incident radiation is absorbed, so that a is
*7black body-to-furnace
almost 1 and p is small. Another example of un- = eb
expected behavior is snow which, despite the blind-
ing reflection of visible light from its surface, and
absorbs almost all the thermal radiation striking it.

Thus, for snow, a approaches 1, whereas p is quite eb = eb

small.
where e b is the rate at which thermal energy is
As with other forms of radiation, thermal
transferred to and from the black body, Btu/hr 2
ft .

radiation penetrates the surface of any receiver to

For the nonblack body,
some extent before the energy is completly con-
verted to heat. In metals, the penetration about is Qf urn ace-to-body
=e
b
1 micron, whereas, for electric nonconductors, it
may be as much as 1000 microns (11). Radiation
^absorbed by body
may be reflected diffusely from a surface (that is, = ae b
in all directions), or it may be specularly reflected

as are light beams from a mirror.

Q reflected by body
= d -cc)eb
The Ideal Radiation Receiver

A body can be imagined that has r = and p = 0; and

allthe incident radiant energy would be absorbed.
eb = aeb + ( 1 - a) eb
Such a body is called a "black body" but its
surfaces need not be black in color. Black bodies (15.57)
do not exist in nature, although many materials
approach black behavior. It can be shown as If e is defined as the actual rate at which radiant

follows that a black body emits radiation at the
maximum rate possible of any body of equal size
and shape at the same temperature. Consider two
bodies identical in every characteristic except that
one is black, whereas the other has an absorptivity
(a) less than 1. If each body is put into a furnace
held at a constant temperature (7" ), both bodies
7"
will eventually reach This will occur because
.
thermal energy will be transferred to each of them
by radiation from the furnace until such time as
there is no longer a temperature difference
between the furnace and the enclosed bodies. Once
they reach T the bodies must maintain this
, Kirchhoff's law holds only at constant tempera-
temperature, growing neither hotter nor cooler
than the furnace. This is necessary by the second law
of thermodynamics, which explicitly prohibits the
spontaneous generation of a temperature differ-
ence between two bodies that can transfer thermal
energy only to each other. At this condition of
thermal equilibrium, then, each body must be
absorbing and emitting thermal radiation at equal
rates; otherwise, the temperature would not stay
constant. The rate of absorption and emission for
the black body will be different from that for the
nonblack body. The situation may be clarified by
writing thermal-energy balances around each body.
energy is absorbed and consequently emitted by
the nonblack body per unit of surface (Btu/hr ft 2 ).
e = ae b = ee b ! 15.58)
which defines e, the emissivity, as the ratio of the
rate at which energy is radiated from a unit surface
and that at which it would be radiated from a
black surface of equal area. From Equation 15.58,
a = e (15.59)
which is known as Kirchhoff's law. Since a is
temperature-dependent, e must be also. Thus,
ture, or, in other words, thermal equilibrium
between emitter and receiver is required.
Although black-body characteristics cannot
be found in naturally occurring surfaces, they are
very closely approximated by a cavity in a solid
substance that is open to the atmosphere only
through a pinhole in the solid surface. Any radiant
beam coming to this opening will pass into the
cavity. There it will bounce from wall to wall,
being partly absorbed at each contact. Because its
chances of finding the opening again are almost nil,
the beam's energy will be completely absorbed,
even though only a modest fraction of this energy
 HEAT TRANSFER 363

is absorbed at each contact. The surface pinhole A pressure effect of radiation must result
thus absorbs all the radiant energy striking it. from the momentum change occurring at collision
Under the condition that the pinhole and its between photons and the enclosing walls of the
surroundings are at thermal equilibrium, the system. The derivation is almost identical to that
emission from the pinhole must be at a rate equal for the pressure exerted by a model gas. By defini-
to that at which radiant energy strikesit from the tion, pressure is the change in momentum resulting
surroundings. Then, since e b both absorbed and
is from the collisions against unit area of wall in
all
radiated, e as well as a must be equal to 1. This unit time. For a single photon, the momentum
argument can be followed regardless of the change per impact is 2 mu where u is the velocity ,

temperature chosen for thermal equilibrium. component normal to the wall and m is the mass of
the photon. The factor 2 occurs because of the
The Stefan-Boltzmann Relation elastic nature of the impact. The number of
The between emitting black-body tempera-
relation photons striking unit area of the surface per unit
ture and the rate of radiant-energy emission was time will be half of all the photons within the
first deduced empirically by Stefan and later distance u from the surface. Within the unit time,
proved by Boltzmann (6) through considering the the photons will move toward the surface, strike it,
analogy between black-body radiation and perfect- and rebound. The factor /2 occurs because half the
1

gas behavior. photons will be moving away from the surface,

The photons of energy can be treated in the Thus, the number of impacts per unit time per unit
same way as the molecules of a model gas. The surface area is y2 uN/V, where N is the number of
difference is that the photons all move at a uni- photons in the system. Multiplying this impact rate
form rate and obey much more closely the criteria by the momentum change per impact, and defining
of no interaction and of elastic collisions. Since the pressure effect of radiation (/>,) as momentum
radiant beams surely travel at finite velocity, a change per unit time per unit surface
volume of space through which radiant energy
flows will have a certain concentration of this n = N
—
Pr mu 2
(15.61)
energy. Consider a system consisting of an enclosed
isothermal volume of space in which the concentra- where Pr is the pressure effect of radiation.
tion of radiant energy is Ur This concentration
.
Since
must be a function of temperature alone, or
c 2 = u 2 + v2 + w2 ! 15.62)

= Ur = (15.60)
where
<t>(T)
V
c = velocity of light
where u, v, w= components of the velocity of light
in the x-, y-, z-directions,
Er = internal energy due to radiation, Btu
V = volume of the system and since
= some unspecified function
! 15.63)

The insignificance of such variables as the proper- because of the random paths of the photons within
ties ofthe boundary walls or the size and shape of the system,
the system can be shown through the second law 2
c =3u 2 (15.64)
of thermodynamics by inserting, reversibly, a new
wall. If such an insertion creates a change in the Combining Equations 15.61 and 15.64,
system, the change must be spontaneous and must
1 N 2
—
be accompanied by an increase in entropy. On Pnr = - mc 2

reversibly removing the new wall, the system, con- 3 V

sisting of the enclosed space, must return to its From Einstein's relation and Equation 15.60,
original state. This change also must be spontane-
ous, occurring with increasing entropy. A contra- Nmc 2 = Er
diction has appeared, for a reversible cyclical and
process has produced a net increase in entropy.
Therefore, the insertion of new walls cannot
change the thermodynamic state of the system.
P=±3r 15.65)
364 APPLICATIONS TO EQUIPMENT DESIGN

Equations 15.60 and 15.65 can be inserted where b is an arbitrary constant of integration.
into the first two laws of thermodynamics to give From this equation,
the relation between absolute temperature and the
rate of radiant-energy emission. The first and
U =bT*
r (15.73)

second laws of thermo-dynamics may be written This equation the desired relation between rate
is

of energy emission and temperature, although

dE=TdS-PdV (15.66) perhaps not the most useful form of the relation.
where Before converting this equation to a more useful
form, it is well to note that b arose as a constant
S = entropy independent of temperature and radiant-energy
E= internal energy
concentration. Surface properties cannot affect b,
Because only radiant energy affects the system, all of course, as long as the original black-body restric-
the thermodynamic properties can be written with tion maintained.
is

the subscript r. Radiant-energy transfer is being considered,

and Equation 15.73 should be written in terms of
Thus,
q rb rather than Ur Since q rb is the amount of
.

dEr = TdSr -Pr dV (15.66a) radiant energy emitted by a black surface per unit
time
Dividing by dV at constant 7",

idEr bSr
—
A
= (a constant) cU, (15.74)

b) ^^
i
\ i

w
I™) = r BV' (15 67
- >

T where c is the velocity of light. The constant here

but can be obtained by integration of the radiant

energy emitted by a small area to the whole visible
(bSr lbV) T =[bPr lbT) v hemisphere. Such integrations show this constant
to be 14. Then
(a Maxwell relation based on Helmholtz free
energy), and then
—
Qrb
=
°b
—T 4.
= aT 4
„
(15.75)

mT P' 15.68)
[jvl \*rh~ Equation 15.75 is the Stefan-Boltzmann
equation basic to all radiant-energy- transfer
From Equation 15.60 Er = VU r and
calculations. The constant o, the Stefan-Boltzmann
constant, is a true constant with a value of
~9 2 4
1.73 x 10 Btu/ft hr °R The total radiant
-« (15.69)
.

($)r energy emitted from a unit area of black-body

surface per unit time \sq rb /A. No account is taken
U = Er
From Equation 1 5.65 Pr = — r
and here of the fact that radiant energy is absorbed as
3 3V well as emitted or that all the emitted energy may
not strike a given receiver. The Stefan-Boltzmann
d equation also gives no information on the wave-
- 1 J±L
(-) 3 dT
!
15.70) length distribution of the emitted radiation.
'v

Placing Equations 15.69 and 15.70 into Equation Wavelength Distribution

15.68 The distribution of wavelengths of radiation

emitted by a black body can only be found by
U =-T
r
1
—
dU —U
-- rr
considering the mechanism of radiant transfer.
3 dT 3 Attempts to arrive at a satisfactory relation for this

or distribution finally led to the quantum theory

(31, 32). Using this theory, a satisfactory relation
dUr between emitting-body temperature, wavelength,
4Ur = T :i5.7i;
dT and the intensity of radiation is obtained. Calcula-
tions based on this theory led to Figure 15.25.
Separating variables and integrating, this equation
Here, the increase in radiant-energy emission for a
becomes
differential wavelength range, or the mono-
\nUr = A\u T+\nb (15.72) chromatic radiation intensity, is given as a function
 HEAT TRANSFER 365

16.000 xlO 10

14.000

12.000

10.000

8.000 -
. Locus of maxima of monochromatic
radiation strength curves

6.000

4.000

2.000 -

0.000
0.000 2.000 4.000 6.000 8.000 10.000
Wave length (micrometers)
Figure 15.25. Monochromatic radiation strength as a function of wavelength and
temperature for black bodies.

of wavelength for various constant temperatures. Analytical integration at constant temperature of

For any given temperature of the emitting black the equation from which Figure 15.25 comes must
body, the monochromatic radiation intensity also give q rb As would be expected, this integra-
.

increases with increasing wavelength until a maxi- tion gives the Stefan-Boltzmann equation.
mum is reached and then decreases. The maxima
shift toward shorter wavelengths as. the tempera- Qrb =oAT A
(15.78)
ture increases. Wien first showed this shift to be
necessary by developing his displacement law (38),
Net Radiant-Energy Transfer
from considerations similar to those of Boltzmann
Though the Stefan-Boltzmann equation is the
B = 7m a x Tm
'n a x ;15.76)
fundamental relation for radiant-energy transfer, it
Here B is a constant found by experiment to be only gives the amount of radiation emitted by a
5193 microns °R. As indicated on Figure 15.25, black body without regard to what it absorbs. In
the area under any one of the isothermal curves practice, interest always centers on the net energy
gives the total rate of radiant-energy emission of a flow. This must, of course, be the difference
black body at that temperature. Thus, between the radiant energy emitted and that
absorbed per unit time. Considering only black-
body radiation,
Qrb
15.77) = —
A \Qrb )net \Q rb> emitted \Qrb 'absorbed (15.79)
366 APPLICATIONS TO EQUIPMENT DESIGN

But since

(Qrb) emitted " oAT y

and
- oAT2
(<7nb Absorbed

the net rate of energy flow is

(q rb )nex = oATy 4 - oAT2 * = oA(TS - r2 4 )

(15.80)

Completely
where surrounding
walls at T2
7"i
= temperature of emitting body
T2 = temperature of surrounding surfaces
A = surface area of body which is emitting
and absorbing Figure 15.26. Radiant-energy transfer from
and to an imperfect radiator.
In differential form,
(1) Radiation from small body:
d{Qrb)r\et ._ (Remitted = oATy*e
= a(7i A T2 4
~Ta
dA
) (15.81) (2) Radiation from wall:
= oA T2 £w
£3
( /-)wall emitted
r
w
For Equations 15.80 and 15.81 to be valid, (3) Radiation from wall striking body:
= oAT 4
very special circumstances must exist. A black to/-) wa il to body 2
(4) Radiation reflected from body:
body must be emitting energy only to its surround-
(^reflected = oAT2 4 p
ings and receiving black-body radiation from its
(5) Radiation absorbed by body:
surroundings. Thus, the radiating body cannot
(<7r)absorbed= oAT A a 2
"see" any of itself; that is, all its faces must be flat
or convex. Moreover, the surroundings must be
until it is completely absorbed. Since the originat-
black and isothermal. One case would be that of
ing body is so small in relation to the walls, there is
two parallel infinite planes, both black and at con- a negligible chance that any significant quantity of
stant temperature. Another possible case would be
the reflected radiation strikes it. At the same time,
that of a small black sphere, cube, or cylinder
the walls emit radiation at a rate characteristic of
enclosed in a large isothermal space. In this case, all
their temperature, area, and surface emissivity
radiant energy emitted by the black body would be (beam 2). Most of this radiant energy merely
ultimately absorbed by the walls of the space even
ricochets from wall to wall, but part of it strikes
though the walls are not black, since the original
the small body. Since the wall area is so large
radiant beam would have negligible chance of being
compared to the area of the body, the energy
reflected back to the body. Moreover, the walls of
striking the body will equal the maximum radia-
the space are so large that even though they are
tion that could be emitted by a black body of area
imperfect radiators, the small black body receives
A and temperature T2 (beam 3). This can be
over its entire surface radiation characteristic of
shown by a thermal equilibrium mechanism such as
the wall temperatures.
that used in defining a and e. Part of the energy
Actually, even with the case of a small
striking the small body is reflected, part absorbed
convex-surfaced body radiating to a large en-
(beams 4 and 5). Under these conditions, the net
closure, Equation 15.80 will not apply if the body
rate of radiant-energy transfer must be
is not black. The situation for a real body is dia-

Wr'net
= (Remitted — 97 'absorbed
gramed in Figure 15.26. (

The small body energy at a rate

radiates = (beam 1) — (beam 5)
characteristic of temperature, area, and surface
its
or
emissivity (symbolized by beam 1). This energy
strikes the walls, where it is partly absorbed and (qv)net = oATy 4 e - oAT2 4 oc
partly reflected by the wall. The reflected energy
which can be written as
continues to pass from surface to surface of the
enclosure, being partly absorbed at each contact, {q r )nex=oA(TSe-T2 A ct) (15.82)
 HEAT TRANSFER 367

Inthe relations given for the individual radiation p + a = 100%

and the net rate of transfer given by Equation
rates 100

15.82, a, p, e, and A are properties of the small % j [

body, and A w and e w are properties of the sur- 80 "a

rounding walls.
60

Constants for Imperfect Radiators b\ Aluminum

40 /
a — polished
The values of a, p, and e used in these equations b = anodyzed
require further consideration. The radiation 20 I

1
|

emitted by the small body is characteristic of its

i
!

temperature (T-\). Therefore, e will be that

emissivity applicable to this particular surface at
7^ The radiation striking the body will, however,
.
100
% ~/-C
——ps J<" Fire clay, white
l\
be at a wavelength spectrum characteristic of an 80
\a \ \
--Tile, white
emitting black body at 7~2 the temperature of the,

60
surroundings. Therefore, p and a must apply to the
Wall plaster —
surface of the small body receiving energy \ \ \
40
characteristic of 7~2 By Kirchhoff's Law, a = e at
. \\ I

thermal equilibrium, and hence the absorptivity (a) 20

will equal the emissivity (e) of the small body itself
7~
at 2 This is true despite the fact that the
.

temperature of the small body is not T2 .

Kirchhoff's law implies that a, e, and p are % \/\Y \a Water

1

w 1 A \ thickness
20
functions of temperature and therefore of the a = 0.01 mm
6
wavelength of radiant energy. This dependence is \| b = 0.05 mm
40
illustrated for a few materials in Figure 15.27. The
behavior shown for polished aluminum is typical of 60
most metals, which have higher absorptivity for
short-wavelength radiation. The curves shown for
other materials are typical of nonmetal surfaces.
80

100
V
For them, the absorptivity is a very irregular func- 1 3 4 5 6 7 9u
tion of wavelength but generally increases with Wave length (X)

increasing wavelength. For some materials, the vari- Figure 15.27. Reflectivity and absorptivity as a
ation of absorptivity with wavelength is small, so function of wavelengths for different materials
that a and e may be considered to be independent (35). (Reproduced with permission of Johann

of wavelength and temperature. These materials are Ambrosius Barth Verlag, copyright © 1943.)
called gray bodies. Many materials are considered
to be gray bodies for engineering calculations even applying the wavelength distribution of Figure
when the assumption is not justified. The reason 15.25, as is done in Example 15.7. More usually,
for this practice is that, since e is a surface the total emissivity is directly measured experi-
property, it will change markedly with small mentally. Such total emissivities are tabulated in
changes in surface condition. Therefore, values of e Appendix D-15 and shown graphically in Figure
are seldom known to high accuracy. 15.28 (35). normal use, the e versus 7" data of
In

In engineering applications, the radiant Appendix D-15 may be interpolated linearly.

energy being transferred is not monochromatic but
rather of a range of wavelengths characteristic Example 15.6. An uninsulated steam pipe
of the temperature of emission. Then, in order runs through a dark warehouse room. The pipe is
to use Equation 15.82 or any similar equation, 2-in. sch.-40 steel pipe; the steam is saturated at
emissivities and absorptivities integrated over the 100 psia, and the warehouse is kept at 40° F. How
entire range of wavelengths must be used. Because much steam condenses per hour per foot of pipe?
the wavelength distribution is given in Figure The coefficient of heat transfer by natural convec-
15.25, the proper emissivity or absorptivity may be tion from the outside pipe surface may be taken as
2
obtained by integrating the monochromatic 2.0 Btu/hr ft °F, and the pipe surface may be
emissivity curve, like those of Figure 15.27, while treated as a gray body.
368 APPLICATIONS TO EQUIPMENT DESIGN

100 relating temperature driving force to resistance

SO
^^^^ i

20
along each step of the total path— steam to inside
pipe wall, inside pipe wall to outside pipe wall, and
outside pipe wall to warehouse— by the parallel
mechanisms of convection and radiation. The
4
/"^ condensing-film resistance and pipe-wall resistance
/
are so low compared to the outside-surface-to-
W 60 /si 40 r;
8
warehouse resistance, however, that the total AT
may safely be approximated as equal to the AT
/ / Is* [h from pipe surface to warehouse. Then, the pipe
/ y/
/// surface temperature [Tp will equal)the
1 40 11 60 condensing-steam temperature of 328° F. More-
over, the temperature of the warehouse walls and
air will be identical at 40°F. Then.

20 SO T 4 T- 4n

7 q = 0.173 xAx 0.95

moo' moo' J
pjlj-T + 2.0x4 x {Tp - Tw ) (c)
100
600 1000 2000 4000 6000 10.000
= 0.173 x 0.622 x 0.95[(7.88) 4 - (5.00)
4
]
Temperature. "R

Figure 15.28. and absorptivity of various

Reflectivity
+ 2.0 x 0.622(788 - 500)
materials (35). p + a=
100%. (1) White fire clay; (2) = 329 + 358 = 687 Btu/hr ft of pipe
asbestos; (3) cork; (4) wood; (5) porcelain; (6) con-
crete; (7) roof shingles; (8) aluminum; (9) graphite.
where the Stefan-Boltzmann constant has been
(Reproduced with permission of Johann Ambrosius
used in the form 0.173 x (1/100) 4 Rate of con- .

Barth Verlag., copyright © 1943.) densation = q/AH = 687/888.8 = 0.78 lb of steam

per hour per foot of pipe. Note that even at these
Solution. be due to
The total heat loss will
relatively low temperatures, radiation has
radiation These mechanisms
and convection.
accounted for almost 50 percent of the total heat
operate in parallel and independently to transfer
transfer.
energy from the pipe surface to the warehouse.
Thus,
Example 15.7. Determine the total emissivity
q = qr +q c (a)
of anodized aluminum at 1800°R.
where
Solution. This problem requires integration
q= thermal energy, Btu/hr
total loss of of the wavelength distribution function of Figure
qc = loss of thermal energy by convection 15.25 with monochromatic emissivities taken from
from the pipe surface, Btu/hr Figure 15.27.From Equation 15.77,
or
Qrb
q = eAoiT* - Tw *) + h c A(TD - Ta (b)
qf b = A d\
where o L d\ J

'
P l ' \A , Ta = temperature of pipe surface,
q rb
warehouse wall, and warehouse
air, respectively, °R d\
e = surface emissivity of pipe at "0 d\
€ =
average of emitting and re-
qr b qrb
ceiving temperatures from
Appendix D-15 r dX
A = area of outside surface of pipe d\
The pipe surface temperature (Tp ) should be calcu- The integration can be most conveniently done
lated by setting up the series-resistance equation graphically using finite wavelength intervals.
 HEAT TRANSFER 369

X AX
d[q rb /A)
d\
e\
~d[q rb /A)1

- d\
AX
-m* or

Wr/net emitted by wall

= oA T
=
w 2
(1
— boA w T2
-y)oA w T2 4
1 15.7 0.12 15.7 1.88 where 7 is the fraction of the total radiation
1 emitted that is received by the walls. In another
1 47.1 0.05 47.1 2.35 form
2 —
1 182.0 0.30 182.0 54.6
Wz-'net emitted by wall ' w— OA w T 2
1

3 (15.83)
188.5 0.35 188.5 66.0
1
where Fw ^ ^ is called the geometric or view factor
4
for radiation w to 1. Since 7 is a fraction of
from
135.0 0.20 135.0 27.0
1
the originally emitted energy, F w ^^ must be a
5 number from to 1 representing the fraction of
1 104.0 0.20 104.0 20.8
the total radiation from surface w, the wall, that
6 strikes surface 1, the small body.
2 67.5 0.42 135.0 56.5
The net rate of energy emission from the wall
8
to the body equals the rate of energy emission from
2 37.6 0.90 75.2 67.7
the wall less the emission from the wall to itself.
10
Then,
10 15.7 0.95 157.0 149.0

2= w /-'emitted from wall — Wr'emitted from

20 ]= 1039.5 445.83 wall to itself

"*"
Wr'emitted from wall to body
e =
445.83 „
= 0.43 M or
1039.5
oA w T2 A - Fw ^ w oA w T2 * + Fw ^ioA w T1 4 '

Geometric Factors 1 =F U + Fw^\ 15.84)

Thus far, the geometric relationships between
emitting and receiving surface have been kept very This equation is a restricted example of the general
simple by arranging that the emitting surface sees result that all the energy emitted by a surface must
only the receiving surface. All the radiation strike some receiver.

emitted struck the receiver. Moreover, it was 2F=Fw1 +F F^ n

1 = 1- +2 + /ri-3 + /r i
+ --- +
always necessary that none of the emitted radia-
tion returned to the emitter. In most cases, these (15.85)
restrictions cannot be maintained, and a more
In the interchange of radiant energy between
general approach is necessary. This approach uses a
two black bodies
geometric factor that relates the radiant energy
4
striking a surface to the total radiant energy w7a)i-2 = a>4 1
F ^2 7
1 "i

emitted.
The situation illustrated in Figure 15.26 can (<7rt»>2—1 = ^2^2-1 7"2 4
be reexamined with attention focused on the sur-
r
rounding walls. In this consideration, the walls will (<7r6)net1~2 = oAy (/ 1 ^ 2 7"l' F 2 ~:T2
Ay
be presumed to be black. Thus, the total rate of
radiation emission from the walls will be (15.86)

At thermal equilibrium (q rbhex 1-2

= 0/ ar| d
= °A
\Qr 'emitted from wall W I 2
7"/= r2 4 Then .

All of this radiant energy will not strike the small

body; most of it will strike other parts of the wall Fv .r/ = —
Ay
f i7"a
surfaces and will be absorbed there. Therefore,

—
_A 2
wr'net emitted by wall '<7 /-'emitted from wall Fy 2 ~
A ^2-1 or Fy .
2 Ay = F 2 .
}
A2
*1
\Qr 'radiated from wall to itself (15.87)
370 APPLICATIONS TO EQUIPMENT DESIGN

dA 2 must be proportional to the area of the

projected surface on the hemisphere (dA 2 ) and
inversely proportional to the square of the distance
2
(r between the point and the projected area.
)

Similarly, the area of dA, as seen from dA 2 must

be the projected area dA, on the plane perpendic- '

ular to the connecting beam. As before,

dA y
'
=dA, cos0i

The fraction of the total energy flux emitted

by dA, that strikes dA 2 is the view factor. This
fraction must be proportional to the projected
areas of dA, and dA 2 and inversely proportional
to the square of the distance between them. The
numerator of this fraction must be
idA,' dA 2
Figure 15.29. Radiation between two differential
fori dA\-dA1
surfaces in space. ?
/ dA , cos 01 , dA 2 cos 2
Since the view factors are functions of geometry
2
r
only, Equation 15.87 is general and not restricted
where (15.89)
to thermal equilibrium. Equation 15.86 can thus
be rewritten as ^Qr)dA\-dA2 = radiation from dA, inter-

4 cepted by dA 2 Btu/hr
(£7^)net1^2=^l/
r
2 <
7~1 — Tl*)
,

/ = a
i -
proportionality constant,
= oA 2 F 2 •i(7-,
4
-r2 4 )
Btu/hr ft 2
(15.88) dA,' dA 2 = projected areas of dA, and
,

dA 2 on the planes normal to

The exact relation between the view factor the radiant beam, ft 2
and the system geometry will now be determined. r = distance from dA, to dA 2
This will be done by determining the view factor
01 02 = angles between the perpen-
,

for a perfectly general geometry. The general case diculars to the centers of dA,
of a differential surface area radiating to another and dA 2 and the ray connect-
surface in space is illustrated in Figure 15.29. Here ing them, radians
the differential area (dA,) is radiating to surface
dA 2 A reference hemisphere has been drawn at a
.
Equation 15.89 is known as the cosine law. It can
fixed radius (r) through which all radiant energy be written in terms of the solid angle, dw, over
must pass. The geometric factor will then be the which a point on dA, sees dA 2 Since a solid angle .

is defined as the area subtended on the surface of a

ratio of the energy flux through dA 2 the '
,

projected area of dA 2 located on the hemisphere sphere divided by the square of the sphere radius,
surface, to that through the entire hemisphere. If (dA 2 )

the hemisphere were drawn through the center of (Qr)dA:-dA2 = i dA,' = i cos 01 dA 1 do:
2
(r
dA 2 then )

(15.90)
dA 2 =dA 2 cos /3 2

The denominator of the fraction that is

where (j 2 is the angle between the perpendicular to the geometric factor must be the total rate of
dA 2 and the radial beam between dA, and dA 2 . radiation emission from dA,. This rate can be
Under this same condition, r is the distance found by integrating Equation 15.89 or 1590 over
between dA, and dA 2 * The amount of radiation the entire hemisphere.
emitted from any single point on dA, that strikes
for) dA 1-*hemisphere
= 1 1 COS 01 dA 1 d0J
* As drawn in Figure 15.29, r is the distance from dA, to
dA 2 but is shorter than the distance from dA to dA 2 This is done = W2
, .

, 27rrsin far dfa

for convenience in visualization and does not alter the basic = idA cos 01 5 :i5.9i:
development. =o r
 HEAT TRANSFER 371

where the integration is done in rings of r tan d&y and therefore

in width sweeping outward from the normal axis of
the hemisphere to the horizon. Each of the rings is lQr)A^A2= joFdA ^ A iTS dA, (15.98)
r sin 0! in radius and 2irr sin /3i in circumference.
Since o /3 1 is very small the tan d(ly has been
f

Combining Equations 15.95, 15.96, 15.97, and

taken as equal to dfiy in determining the width. 15.98,
Integrating,

(Qr)dA |-» hemisphere = I dAy2ir[\ Sin 2 0] g

2
= I dA y 7T
F i-2 = —
A f
-a,
FdA ^ A2 dA, (15.99)
y

(15.92)
or
Equation 15.92 gives the total radiation from a ~

black-body surface. By the Stefan-Boltzmann equa- =

"1 cos fa cosj3 2 dA 2
F^2 r
' dAy
tion, Ay 'A, L7T 'A 2 r2 J

= O dAyTy 4
(<7r)<M1-»hemisphere (15.93) 15.100)
Combining Equations 15.92 and 15.93 and solving
for/' Equation 15.100 is perfectly general and may be
integrated to any desired geometry.
0Ty Such integrations have been carried out for
/
=
certain fixed geometries and presented in graphical
form. Some of these solutions are given in Figures
Because both numerator and denominator of the 15.30 and 15.31. Other solutions are available
fraction that is the view factor have been obtained,
(18). For cases not previously solved, Equation
the view factor itself is 15.100 must be integrated. Usually the exact
integration cannot be done. Then an approximate
(dr) dA"\^dA2
FdA\->dA2 answer may be obtained by evaluating the radiant
\Qr'dA 1 — hemisphere interchange between portions of each body as
2
/ dAy cos j3-| dA 2 cos fi 2 lr follows:

i dA y
IT 1. Divide each surface into roughly equal
parts.
COS @y cos |3 2 dA 2
(15.94) 2. Using Equations 15.97 and 15.95 directly,
FdA\ dA2
7ir evaluate the interchange between representatively
selected parts of surface 1 and all visible portions
The view factor applicable for transfer from dAy
of surface 2.
to a finite receiver (A 2 ) can then be found by
3. Average the interchange energies so
integration.
obtained and multiply by the number of portions
1 r cos j3i cos/32 dA 2 into which surface 1 was originally divided. The
FdA^A7=- 2 ^
15 95)
-

averaging procedure must be consistent with the

original division of the total surface. The appro-
engineering applications involve transfer
In general,
priate weighting procedure should be used for un-
between two finite bodies. A view factor is desired equal portions of area.
that applies to the entire geometry present. Such
an overall view factor could be simply used to
calculate the entire energy-transfer rate.
Example 15.7. Determine the geometry
factor to be used for radiation between a wall and
l<tr)A^A2 = 6AiF^ 2 T, A (15.96) adjoining floor. Both surfaces are black; the wall is

This energy-transfer rate must be the integral of 10 ft by 20 ft and the floor is 20 ft by 40 ft. The
the transfer rates from all the differential sections other walls and ceiling may be considered to be
oM,. transparent to radiation.

(Qr)A1^A2 = ^(dr)dA\ -A2 Solution The physical situation is shown in

Figure 15.32. This geometric arrangement must be
but
used in defining the terms of Equation 15.100
(qr)dAi-A2 = odA FdAy , A2 Ty A
: (15.97) before integration.
372 APPLICATIONS TO EQUIPMENT DESIGN

0.50

0.40

0.30

0.20

0.10

0.3 0.4 0.6 0.8 1.0 2.0 3.0 4.0 6.0 8.0 10
C, dimension ratio (y/z)

Figure 15.30. Geometric shape factor for radiation between adjacent rectangles in

perpendicular planes (17).

[From C. O. Mackey, L. T. Wright, Jr., R. E. Clark, N. R. Gay, Cornel/ University

Engineering Experiment Station Bulletin 32, Ithaca, New York (August 1943) by
permission.]
B= ratio (length of unique side of that rectangle on whose area the heat-transfer
equation is based/length of common side) = x/z in sketch
C = ratio (length of unique side of other rectangle/length of common side) = y/z in
sketch

From Figure 15.32, the following substitu- This equation has been integrated with respect to
tions can be made: each of the variables successively by Hottel (17) to
give
dA !
= dy dz-i

dA 2 = dx dz 2
x F^
2-1
J_
- An
cos ft
Vx 2 + y2 + Az 2 (1 + B2 + clf-iUBWeiB^Blf^C
where Az =z 2 — ^1 x In
(1 + B 2 B ~ iVB) {] + c2) (c2/eMl/5) (ff 2 + c2 B+{c7,B)
) )

COS |3 2
= , 1
- + — tan
C 1 1
- //
C
+ -92 tan
, 1

Vx 2 + y2 + Az 2 + tan
B B C J 1

B
,

s/B* +
- ,,

C2
r=Vx 2 + y 2 + Az 2
1 - - xy dx dz 2 dy dz where B = x/z, C = y/z, and where z is the common
y

'1-2 edge and x and / are the other edges of the surface.
2
itAi -a, -a :
(x + y 2 + Az 2 )
2
For this case, x/z = 2.0 and y/z = 0.5. Putting these
1 ,-z, = 20 ,/ = 10 -z, = 20 -x=40 values in Equation (a) gives /r ^ 2 = 0.079. This 1

Fv I I I" (
result can also be read from Figure 15.30.
2007T z, = -/=0 z :
=0 • x=

xy dx dz 2 dy dzy Example 15.8. Determine the net heat flux

U +K 2 2 + (Z 2 -Z 1
)2]2 between a furnace door 2 ft by 2 ft and a wall
 HEAT TRANSFER 373

The determination is readily made once the view

factor determined. This factor can be found by

is

using the general procedure given on p. 371

coupled with the geometric factors read from
Figure 15.31. The geometric arrangement of
furnace door and wall is probably too unsym-
metrical to expect that already-calculated view
factors like those of Figure 15.30 would be avail-
able. In cases where already-calculated view factors
are available, they are, of course, convenient. But
the procedure to be shown here is powerful enough
to allow calculation of many cases for which view
factors are not available.
A front view of the system geometry is
shown. The general procedure calls for dividing the

Figure 15.31. Geometric factor for a system of differential

surface and a finite rectangular surface parallel to it (17).
[From C. O. Mackey, L. T. Wright, Jr., R. E. Clark, N. R. -r—
Gay, Cornell University Engineering Experiment Station
Furnace
Wall
Bulletin 32, Ithaca, New York (August, 1943), by per- door
mission.]

10'

furnace door into a number of representative sec-

tions and finding the view factor from several of
these sections. Let the door be divided as shown in
the illustration. As shown in Figure 15.31, only

x = 40
"
(a) (b)

Figure 15.32. Radiant transfer from a wall to the floor, (a)

i
Physical arrangement. (6) Geometric relation of terms.

located as shown in the adjoining illustration. The

2
wall is 8 ft high, and the bottom of the furnace
door is 3 ft from the floor. Assume all surfaces are
black. The furnace door is at 600°F, and the wall is -3

100°F.
Wall 4

1 differential areas are used, so the five points shown

2'
will be taken as representative. The use of so few
points can only be justified because of the sym-
Furnace
metry of the system. The geometric factor will
door now be determined for each of these points.

Solution The net heat flow can be deter- Point 7. The spatial relation is as shown.
mined from Equation 15.88. The desired view factor will be

F-\~w ~ F-\ .| i+ii + F, -dn+iv) F^x-F 1 -iv

374 APPLICATIONS TO EQUIPMENT DESIGN

*~
?
10' — ^3^w ~ ^3^1 + F 3 -^n + F 3 ^||| + F 3 -.| V
= 0.06 + 0.24 + 0.22 + 0.057 = 0.575
I W^9M^
lllll ^m^^^^M,
/

^^Wall^^^^^
By similar constructions
0.615. Averaging the five geometric factors gives
FA ^ W = 0.22 and F5 ^ w =

Fa v£//n-*w =0.443. When the average geometric

factor thus obtained is used, Equation 15.99 gives
IV
^§§§§p yy ^
111 4i

^^^Hwm !
1
f
A,
1

Ay
A,

ForfHU+ii) B= -=
b
a
—10i
2
=5.25 so that Fdoor-vwll = 0.443

Then
C = - = -f =1.75 (<7r)neti-2 = 0.173 x 4 x 0.443[( 10.60) 4 - (5.60)
4
]

where the factors
Figure 15.31, and
a 2
= 3580 Btu/hr transferred from the fur-
b, c, B, and C are as defined on
nace door to the wall
Fi-»(i+u) = 0.22.

For ^-(in+ivj: Allowance for Real Surfaces

5=5.25 C=2.25 F When the bodies are not black but are real liquids
1 -»(|| H .|v)-0.24
and solids, further difficulties arise in evaluating
ForF^i: the proper emissivity. If the surface is opaque to
5= 0.25 C=1.75 F,-| =0.057 thermal radiation, any radiant beam striking it is
partly absorbed and partly reflected. The reflected
For F - iv
t beam may return to the original source, may strike
5 = 0.25 C=2.25 F,-, v =0.06 another portion of the receiver, or may strike a
third object. In thisway, every beam bounces from
Then F,_> w = 0.22 + 0.24 - 0.057 - 0.06 = 0.343.
surface to surface, at each contact losing part of its
intensity, until it is completely absorbed. More-
Point 2. Here the spatial arrangement is
over, the characteristics of the surface influence
slightly simpler, as shown. Again using Figure
the manner of travel of the beam, because from
some surfaces reflection is specular, whereas from
others it is diffuse.
The "effective" emissivity under such condi-
tions will be found for the case of diffuse radiation
with parallel plates that are essentially infinite in
extent. Here, the geometric factor (F) is unity. The
effective emissivity would also be unity if the sur-
faces were black. With nonblack surfaces, the
15.31 to determine partial view factors gives
geometric factor will remain as with black surfaces,
F2- w = F 2 ^i + /Vn = 0.23 + 0.23 = 0.46 but an interchange emissivity must be found. The
situation for parallel plates is shown schematically

Point 3. The spatial arrangement is shown in in Figure 15.33. There, the amount of heat
the illustration. absorbed by surface 2 of that emitted by surface 1
per unit time per unit of area 1 is

<7l-2
= (1 - /o 2 ai - 2
P2 Pi«i - P2 Pi
3 2
<*i
A,
4
-P2 Pi
3
«i •^oT, 4
which shortens to

<7l-2 «1P2
= 1 e,oT, A (15.101)
Ay 1 -P1P2
 HEAT TRANSFER 375

P> 2 P, 2 a 2 for cases in which the geometric factor is less than

unity.

qjA = aT *t
x l x
= t
Effect of Zero-Flux Surfaces
e l
P 2 "i 2
:,

In industrial furnace calculations, the engineer

T must usually account for the fact that the radiation
|f2 u l
le i
p.i 2 p x
a i
emitter and ultimate receiver are not the only
Figure 15.33. Radiant interchange between parallel non- surfaces present but are connected by refractory
black surfaces. surfaces. These surfaces are insulated; con-
sequently, any incident radiant energy must be
Considering the surfaces to be gray and reradiated. They usually have high absorptivities
opaque, we can apply Kirchhoff's law and replace and act as preventing radiant energy from
a shield
the a and p terms by e's being Thus, the geometry factors and inter-
lost.

= - ei(1 -e 2 change emissivities must be altered to account for

<7l-2 1 e^oA,T{ the additional surfaces. Geometry factors for black
€i +e 2 ~ e i e2
bodies which include the effect of the refractory
e 2 e. surfaces are designated F _ 2 Such factors for
1
.

-oA^TS (15.102) parallel planes enclosed by refractory walls are

6! +e 2 -eie 2
given in Figure 15.34. It can be seen that the
Applying the same reasoning to a beam originally presence of refractory walls increases the rate of
leaving surface 2, we get heat transfer markedly by preventing radiation
leakage.
02 e 1 e2
= oTS For
a single radiation source and a single
A, e, +e 2 -e,e2 receiver,both of which are black and are con-
The net interchange then is
nected by refractory walls, the F can be deter-
mined from the values of F1-+1 and F 2 _i that
4
q^-q 2 =
e 1 e2 r 1
-e 2 e r2 4 1
—
qv apply.
=
A, e! +ee 2 -e^
+ -e e 2 Ay From the meaning of aci- 2
by tracing the ,
*i y

possible paths of radiant energy leaving the source

or
a
r/-r 4 2 )
F\^7 = ^1-2 + F R ^ 2 + Fi^ R F R ^ R F R ^ 2
F-\^ R
*1 —+—-1 1
1 + F\^r F R ^ R F R ^ 2 + F-\-^ R F R ^ R F R ^ 2 + •

€t e2
1-2 - Fy^ 2 + Fy-. R F R ^ 2 {\ + F R -. R +F R ~ R
Thus, the interchange emissivity (ei_ 2 ) is given by
R-R
1

ei-2 15.103) F\-^ R F R ^ 2

—+—-1 1
, 1-2 = FV (15.107)
1 (1-/7?-**]
ei e2
where the subscript 1 refers to the emitting sur-
for the case of parallel plates. For concentric
face, the subscript 2 to the receiving surface, and
cylinders or concentric spheres with diffusely
the subscript R to the refractory surfaces. This
reflecting surfaces, a similar development leads to
relation may be put in more convenient form by

^1-2=-^ ——. (15.104)

applying the general restrictions on F factors given
in Equations 15.85 and 15.87. As indicated above,
the thermal-equilibrium restriction also requires
ei A2 \e 2
that the net heat flux from source to refractory
where 1 the smaller body. The inter-
refers to walls must equal that from refractories to receiver.
change emissivity used in one of the may be Thus,
following two equations:
9i -r ~ Qr -2

c7 ,= a^ 1
e1 _ 2 (r 1
4
-T 2
4
) (15.105) or

for cases where the geometric factor is unity, or F,-. R A :

o(T,
A - 7V) = FR 2 A R o(T R * - T7
qr =oA 2 e^ 2 F2 ^,(TS-T2 *) (15.106) (15.108)
376 APPLICATIONS TO EQUIPMENT DESIGN

0.9
—
cs
^" ^--Z^^^iZ^^^
0.7
y^^^^^Z^^Z^^^
it*.
6
//CzZ^^^Z^^^
1 / // ^/ x sZ^
0.4
III// / //^
Curves (F) Shape of Surfaces Curves (F\
Disks 5
5 1 1 If
/ / ° /r 1

2 Squares 6
3 Rectangles. 2:1 7
4 Rectangles, long and narrow 8
2 Iff/ / /A

01

lA 1 3
Side or diameter
Ratio (-jDistance between planes /

Figure 15.34. Geometric factors F and F for radiation between identical parallel
opposed black planes (28).
(Curves 1, 2, 3, 4— no refractory surfaces. Curves 5, 6, 7, 8—opposed planes of an
enclosure completed by refractory surfaces.)
(With permission of McGraw-Hill, copyright © 1954.)

These considerations and the restriction that both Geometric Factors for Imperfect Radiators
receiver and emitter are flat or convex (F^^ = y Ifthe surfaces are_ imperfect radiators, the geo-
F ^2 = 0) ma y De used with Equation 15.107 to
2 metric factors [F ^ 2 and F y ^ 2
y
must be )

give combined with the proper interchange emissivities.

These factors are frequently combined into the
*2-f?-2*1 single factor &1-+2 so that the simple equation
'1-2 115.109)
Ay + A 2 -2F^ 2 A,
(q r )^ 2 =A, ?^ o(T/-T A
2 2 ) (15.111)
This relation no knowledge about the
requires
refractory surfaces except that they are at a con- may be used to determine the net heat flux from
stant temperature and that they receive and reemit source to sink including the effects of reradiating
all the radiant energy that does not go directly walls and emissivities less than one. If the enclosure
from Ay to A 2 .
has only a single source and single sink, both of
The same relations and restrictions can be which are gray bodies, the factor &\^ 2 becomes
used to determine the uniform refractory tempera- (28).
ture, giving
1

1-2
4 _ Ui-
F^ 2 A )TS y
+ {A 2 - F^ 2 A, )72 4 Au 1
T
(A y -F^ 2 A,) + (A 2 -F^ 2 A } )
"1-2 ^ ' A 2 \e 2
(15.110) (15.112)
 HEAT TRANSFER 377

Example 15.9. In an electrically heated be used; although there are a multitude of sinks,
dryer, the stock is moved on a belt conveyor these sinks are all at the same temperature, are
through the drying chamber. There it is heated by geometrically similar, and hence can be treated as a
convection and radiation from glow bars along the single surface.
top of the drying chambers. The glow bars are 1 in.
1
in diameter and are spaced in a row 4 in. from the ?y >2
1
roof of the dryer and 4 in. from center to center,
the stock surface is 8 in. from the bar surfaces in a Fy V«i ' A 2 \e 2 '

bed wide enough to be considered infinite com- (15.112)

pared to the rest of the system dimensions. If the 1

glow bars are at 2800°F with an emissivity of 0.9

4/12
while the stock surface is at 200°F with e = 0.5, -L + (-L_
determine the rate of heat transfer to the stock 0.6 '0.5 ttx 1/12^0.9 <iH
surface by radiation alone. The refractory walls of
1 1
the dryer may be considered black. = 0.356
1.667 + 1.0+1.29(0.111) 2.808
Solution From Figure 15.35, F^^, 2 ma Y De
read. Here the ratio of center-to-center distance of Here the unit length of bed has been chosen as 4
tubes in a row to the outside tube diameter is
in.,which corresponds to one complete glow bar.
BID = 4/1 = 4, and a single row of tubes is present. Then
From the proper curve on Figure 15.35, F\^ 2 =
4
0.6 where 1 refers to the stock surface. Since the (<7r)i-2 = 1 x 0-356 x 0.173(6\60 - 32.60 4 )

surfaces are not all black, the J* factor must be = 0.0616 x (1900 -1,132,000)
used, or interchange emissivities combined with the = - 69,600 Btu/hr ft
2
view factor. For this purpose, Equation 15.112 will
The minus sign indicates that thermal energy is
being transferred to the stock surface, and the unit
of area is 1 ft 2 of stock

Radiation through Absorbing Media and from Luminous

Flames
As mentioned earlier, certain unsymmetrical gas
molecules absorb significant quantities of radiant
thermal energy. The most common of them are
water vapor and C0 2 although most organic vapors
and oxides of nitrogen and of sulfur also absorb
radiant energy. In all cases, the absorptivities of
these gases are significant through narrow-
wavelength bands but are nearly zero through the
rest of the wavelength spectrum. The pattern of
the absorption bands corresponds to the natural
vibration frequencies of the atoms within the mol-
ecule and hence is characteristic of the individual
molecules. In other wavelength ranges (infrared,
X-ray, etc.), the strength .and position of the
absorption bands serve as analytical devices. Beer's
law states that for a given wavelength the strength
of energy absorption (or emission) is logarithmic-
ally proportional to the number of molecules
struck by the energy beam. For perfect gases, the
3 4 5 6 7 strength will be proportional to the partial pressure
Center-to-center distance of tubes in row\
Ratio of absorbing gas and the length of beam travel
( Outside diameter of tubes /
through the gas. Thus,
Figure 15.35. bank of staggered tubes (28).
Radiation to a
(With permission of McGraw-Hill, copyright © 1954.) ln(1-aj=7p/ (15.113)
378 APPLICATIONS TO EQUIPMENT DESIGN

where Radiant-Heat-Transfer Coefficients

p= partial pressure of absorbing gas In some cases where there are many repetitive
/ = length of beam travel through the gas calculations of radiant-heat transfer to be made, or

7 = a constant where heat transfer by convection and radiation

ot x = absorptivity for radiation of wavelength X
must be considered, it is convenient to reduce the
calculations to the form of the heat-transfer
The constant 7 depends upon the wavelength of coefficient used in convection calculations. Thus,
the incident radiation and upon the mono- for interchange between black surfaces,
chromatic absorptivity of the gas at fixed values of
p and /. Thus, in considering the total absorptivity q rb = oA(TS - V) = h^Air, - T2 )

of a gas receiving radiant energy, the temperature

of emitting source, the atomic and molecular (15.116)
properties of the gas, the partial pressure of receiv- from which
ing gas, and the length of travel of an average beam
will all influence the result. ITS-T2 A )

h rb - a
Design charts showing these relations have Ty-T2
been obtained experimentally for C0 2 H 2 0, and ,

S0 2 and are given in standard texts (28, 30). Also where h rb = radiation "coefficient" for black-body
given are factors by which a characteristic dimen- interchange. For nonblack surfaces the proper
sion of various shapes may be multiplied in order emissivities and F factors can be inserted into Equa-
to obtain the length of a mean radiant beam. With tion 15.116. Such coefficients can obviously be
water vapor ax pi = atm, the emissivity varies
1 .0 ft used only for the conditions applied in their
from 0.1 to 0.3 as the temperature varies from development, but they do significantly reduce the
3500 to 500° F. Over the same temperature range, labor of some calculations of radiant-heat transfer.
the emissivity varies from 0.3 to 0.6 when pi has
been increased to 20 (28). REFERENCES
From the and absorptivities so
emissivities
obtained, the radiant-energy interchange between a
1. Badger, W. L, Ind. Eng. Chem.,22 p. 700 (1930).
black surface and the gas will be
2. Badger, W. L., Trans. A.I.Ch.E.,33, p. 441 (1937).
3. Bergelin, O. P., G. A. Brown, S. C. Doberstein, Trans.
Qrb = oA s (e G 7"6 4 - 06 T/ (15.114)
ASME, 74, p. 960(1952).
where 4. Bergelin, O. P., A. P. Colburn, H. L. Hull, Eng. Expt.
eG = gas emissivity evaluated at the gas Sta., Bull. 2, University of Delaware (1950).
temperature 5. Bowman, R. A., A. C. Mueller, W. M. Nagle, Trans.

ol c = gas absorptivity evaluated at the surface ASME ,62, p. 283 (1940).

temperature 6. Boltzmann, L, Wiedemann's Anna/en, 22, p. 291
(1884).
For a gray surface, the interchange must be evalu- 7. Brown, G. G., and Associates, Unit Operations, John
ated by tracing typical beams as discussed previ- Wiley & Sons, New York (1950), P. 425.
ously. For surfaces of high emissivities (el>0.7), 8. Chemical Engineering Problems, A. I. Ch. E. 1 946) , (

the resulting surface emissivity factor is approx- 9. Chilton, T. C, and A. P. Colburn, Ind. Eng. Chem.,27,

imately (e s + 1 )/2. Thus, Equation 15.1 14 becomes p. 255(1935).

10. Colburn, A. P., Ind. Eng. Chem., 25, p. 873 (1933).

1+e< 11. Eckert, E., Forsch. Gebiete Ingenieurw., 9, p. 251

Qr = oA t ie G TG 4 -occTs (1938).
12. Gardner, K. A., Ind. Eng. p. 1215 (1941).
Chem., 33,
(15.115) 13. Gardner, K. A., Trans. ASME,
33 (1945).
67, p.

The soot particles in most luminous flames 14. Gardner, K. A., Trans. ASME,
67, p. 621 (1945).
15. Gilmour, C. H., Preprint of paper presented at 2nd
also emit and absorb radiant energy. The strength
National Heat Transfer Conference, A.I.Ch.E.-ASME,
of such emission can be obtained from an empirical
Chicago (1958).
design chart based upon optical pyrometer
16. Grimison, E. D., Trans. ASME, 59, p. 583 (1937); 60,
measurements of the apparent flame temperature.
p. 381 (1938).
Such a chart is given by Perry (30). The resulting 17. Hottel, H., Trans. ASME, FSP53-19b, p. 265 (1931).
emissivity be used to obtain radiant-energy
can 18. Jakob, M., Heat Transfer, Vol 2, John Wiley & Sons,
interchange through Equation 15.115. New York (1957).
 HEAT TRANSFER 379

19. Kays, W. M., and A. L. London, Compact Heat 15.2. A gear oil is flowing through a 2-in. sch. 40
Exchangers, 2nd ed., McGraw-Hill, New York (1964). steel pipe at a velocity of 3 ft/s and is heated from 80° F to
20. Kern, D. Q., Process heat Transfer, McGraw-Hill, New 200 F by steam condensing at 5 psig. The steam-side co-
York (1950), p. 542. may be assumed to be 1500 Btu/hr ft °F. A scale
efficient
21. Kern, D. Q., Process Heat Transfer, McGraw-Hill, New resistance of 0.003 hr ft F/Btu is predicted. How many
York (1950), Chapter 15.
20-ft lengths of pipe are necessary for the exchanger?
22. Kern, D. Q., Process Heat Transfer, McGraw-Hill, New
Oil Properties
York (1950), p. 525.
= 40 centipoises
23. Kirkbride, C. G., Ind. Eng. Chem., 26 , p. 425 (1934). Ml oo
24. Lyon, R. N., Report ORNL-361, Oak Ridge National M200 = 5.4 centipoises (assume In p. vs T°P is linear)

Laboratory (1949).
25. McAdams, W. H., Heat Transmission, 3rd ed.,
k, Btu/hr ft 2
np., F 3 c P Btu/lb
McGraw-Hill, New York (1954), p. 330. (°F/ft) p, lb/ft ,

26. McAdams, W. H., Heat Transmission, 3rd ed., 80 0.081 53.3 0.473
McGraw-Hill, New York (1954), p. 335. 100 0.0803 52.0 0.485
27. McAdams, W. H., Heat Transmission, 3rd ed., 150 0.0792 50.8 0.512
McGraw-Hill, New York (1954), p. 337. 180 0.0785 50.1 0.530
28. McAdams, W. H., Heat Transmission, 3rd ed., 200 0.0780 49.7 0.541
McGraw-Hill, New York (1954), p. 82-124.
29. McCabe, W. L., and C. Robinson, Ind. Eng. Chem., 16, 15.3. A straw oil flows through a horizontal uninsu-
p. 478(1924). lated 1 40 steel pipe which is exposed to quiet room
-in sch.

30. Perry, J. H., Chemical Engineer's Handbook, 5th ed., air at 50°F. The oil flows steadily through the pipe entering
McGraw-Hill, New York (1973) 10-56-10-60. at an inlet Reynolds number of 3000. The average oil inlet
31. Planck, M., Verhandl. deut. physik Ges., 2, pp. 202 and temperature is 200°F. How far from the entrance will the
237 (1900).
oil reach a temperature of 130°F?
32. Planck, M., Ann Physik, 4, p. 553 (1901).
Oil Properties
33. Polents, L M., Chem. Eng., 65, No. 7, p. 137 (1958).
34. Rickard, C. L, 0. E. Dwyer, 0. Dropkin, Trans ASME, p = 50 lb/ft 3
SO, P- 646(1958).
c A>=0.50Btu/lb/°F
35. Sieber, W., Z. Tech. Physik, 22, p. 139 (1941).
36. Standards of Tubular Exchanger Mfg. Assoc, 6th ed.. k = 0.08 Btu/hr ft 2 (°F/ft)

New York (1978). 7", F 70 80 95 110 130 160 200 240

37. Tinker, T., Proceedings of the General Discussions on 22
p, centipoises 18 13.8 10.8 8.1 5.7 3.8 2.7
Heat Transfer, IME-ASME (1951), pp. 89-116; Trans
ASME, 80, p. 36(1958). 15.4. 100,000 Ib/hr of water are to be cooled from
38. Wien, W., Ben Preuss, Akad. Wiss., 55 (1893). 200 F to 100 F by a coolant entering the exchanger at
39. Wilson, R. E., Trans. ASME, 37, p. 47 (1915). 60 F and leaving at 95° F. Assuming an overall heat-transfer
2
coefficient of 400 Btu/hr ft °F, determine the heat-
PROBLEMS transfer surface required for:
(a) A double pipe heat exchanger with parallel flow.

(b) A double pipe heat exchanger with counter-

15.1. It proposed to heat 140,000 Ib/hr of a non-
is
current flow.
volatile oil from 60° F to 400° F by means of a vapor
15.5. For Problem 15.2, calculate the heating area
condensing at 500 F. The heat exchanger will consist of a
required by the method indicated:
bundle of 1 in. I.D. copper tubes inside a steel shell with
(a) Find p at 80° F and 200° F, and determine an
the oil flowing through the tubes. Calculate the heat-
arithmetic average. For the temperature corresponding to
transfer area required for the exchanger.
Values of the overall heat-transfer coefficient and the
this average viscosity, calculate U based upon properties
evaluated at this temperature. Use (AT)/ m .

specific heat of the oil are tabulated below. (Normally, the

(b) Average 80° F and 200° F, and use this tempera-
overall coefficients would have to be calculated using
ture to determine physical properties and coefficients. Use
methods of Chapter 13.)
(An /m .

mp., °F c P Btu/lb/°F
, U, Btu/hr (c) Use arithmetic average of h at 80°F and h at

60 0.48 470 200° F for the exchanger. Use (AT) /m .

100 0.50 540 (d) Compare answers with the rigorous solution of

200 0.56 680 Problem 15.2.

300 0.62 880 15.6. Process steam at atmospheric pressure is con-
400 0.68 1184 densed in a shell-and-tube condenser and the condensate
380 APPLICATIONS TO EQUIPMENT DESIGN

cooled to 80°F in a simple double pipe exchanger, the the benzene shows that some acid impurity is present. Tests
exchanger is a 1g-in.. 16 BWG, copper tube inside a 3 in. on the steam indicate no reason for excessive corrosion
sch.-40 steel pipe. Cooling water enters the annulus at 60° inside the inner pipe. From the test data, calculate the

and leaves at 190°F. If the steam rate is 2500 Ib/hr to the magnitude of the scale resistance.

condenser, how much coolant is required (gal/min), and

how long must the exchanger be? Flow rate = 4400 Ib/hr benzene
15.7. Nitrogen at 30 psig is flowing in the annulus of Benzene temperature = inlet 70°F, outlet 130°F

a double-pipe heat exchanger made from a sch.-40, 2-in Steam temperature = 240°
Steam-heat-transfer coefficient = 2000 Btu/hr 2 °F
steel pipe and a 1-in., 16 BWG copper tube. The gas flow is ft

6500 ft 3 /hr, measured at 60° F, 760 mm Hg. Water flows in Effective heat-transfer length = 15 ft

the tube at a velocity of 2 ft/s, entering at 50° F. The

nitrogen is to be cooled from 250° F to 60° F. How long 15.13. A shell-and-tube heat exchanger is to be used
to cool 200,000 Ib/hr of water from 100°F to 90° F. The
should the exchanger be?
being cooled from 180°F to exchanger is a one-shell-pass, two-tube-pass unit with the
15.8. Butyric acid is

140°F by heat exchange with benzene entering double water flowing through the tubes. River water flows through
a
the shell and enters at a temperature of 75°F and leaves at
pipe heat exchanger at 100°F and leaving at 150°F. A
90° F. The heat-transfer coefficient of the shell-side fluid is
100-ft long exchanger is fabricated from a 3-in. sch.-40
1000 Btu/hr ft
2 °F. Design specifications state that the
outer pipe and a 2-in., sch.-40 inner pipe. The flow rate of
2 tube-side fluid shall have a pressure drop as close to 10 psia
the butyric acid is 20,000 Ib/hr. /r
acid = 0.09 Btu/hr ft
(°F/ft).
as possible and that the tubes be 16 ft long made from 18
(a) What is the dirt factor if the hot stream is placed
BWG copper tubing. For these specifications, what
in the annulus?
diameter tube and how many tubes are needed?
15.14. In the typical petroleum pipe still, crude feed
(b) What is the dirt factor if the cold stream is in the
annulus?
may be 50,000 bbl/day, and residuum one third of this
(87500 and 29000 gal hr). Consider the interchanger for
15.9. 3000 Ib/hr of a lube oil flow through a smooth
recovering sensible heat of the residuum by interchange
f-in. I.D. copper tube whose inner surface is maintained at
with the feed, available at 700° and 100°F, respectively.
a constant temperature of 250°F by steam condensing on the
This interchanger will be 77 units of one-shell-pass, two-
outside. The tube is 20 ft long, and the oil enters at 100° F.
tube-pass exchangers arranged in overall counterflow. Each
What will be the exit temperature of the oil?
Oil Properties
is expected to function with a coefficient (U) of 40 Btu/hr
ft °F. Compute the incremental cost of heat recovered for
Specif ic gravity = 0.82 - 0.00047 where T= °F 17 = 1,2,3, and 4 if the exchangers cost $25/ft 2 installed
1 25 (use as an annual cost of $6100 year unit). The area is
Viscosity = 5.5
/60v
(
— )
-

centipoises 1250 ft 2 each. Resid is in tubes.

c P = 0.5Btu/lb°F Properties:
Feed Residuum
= 0.1 Btu/hr 2
A: ft (°F/ft) 3
p lb/ft 50 60
15.10. It is desired to preheat 30,000 Ib/hr of air from CP Btu/lb°F 0.5 0.6
70° F to 200° F at constant pressure of 1 atm by steam
condensing at 220°F inside vertical 1-in., sch.-40 steel 15.15. 20,000 Ib/hr of air is to be cooled from
pipes. The heater consists of a staggered bank of pipes, each 150°F to 100°F by water flowing in a shell-and-tube heat
4 ft long, spaced on 2.5-in. equilateral centers. The heater is exchanger. Cooling water is available at 70°F and should
40 pipes wide. What is the number of pipe rows required? leave the exchanger at 95°F. Available to do this cooling

15.11. A horizontal 3-in., 16 BWG copper con- assignment is a one-shell-pass, two-tube-pass exchanged con-
denser tube is 20 ft long and has water entering at a sisting of fifty 1-in.O. D. 18 BWG copper tubes, each 10 ft

velocity of 5 ft/s and 70° F. Dry saturated steam at 175°F is long. The tubes are arranged inside a 10-in. shell on a

fed to a well-insulated jacket surrounding the pipe. Assum- staggered l|-in. triangular pitch with baffles spaced every
ing film-type condensation, how many pounds per hour of 6 in. Is this exchanger satisfactory?
steam can be condensed? 15.16. A typical space heater (steam-heated core, air
15.12. Benzene flows through the annulus of a blown through by a fan) has been heating 100 m 3 of air per
double pipe heat exchanger made up of a 2-in., sch.-40 steel minute from 12°C to 62°C with steam condensing at
pipe inside a 3-in., sch.-40 steel pipe. This exchanger has 112 C. For this case of cross-flow, air unmixed in fully
been operating for several months, and the exit temperature turbulent flow, e=1— e _/v q. Using and stating such
of the benzene has been decreasing with time. Analysis of simplifying assumptions as you need, predict the discharge
 HEAT TRANSFER 381

doubled. What the relative duty 500°C? Use 2 °F to

temperature if the flow is is a convection coefficient of 2 Btu/hr ft

in J/s? all surfaces and ignore temperature gradients within this

15.17. In the acceptance test of a process unit a well- tile.

designed baffled single-shell-pass, two tube pass ex- 15.24. Repeat Problem 15.23 for the case in which
changer performed as designed. Oil is in tubes in the tiles are placed on edge 2 in. apart on a table to cool.
turbulent flow, 2268.0 kg hr, having cp = 2.3 J/g°C and en- 15.25. It is sometimes desirable from an architec-
tering at 71°C and leaving at 21°C. The area is 19.1 m
2
. tural and safety standpoint to shield steam radiators. With
(a) What is the heat transfer coefficient? steam at 212°F and room temperature fixed at 70°F,
(b) For a study of the dynamics of the unit under determine the heat transfer to the room for the case of a

upsets, it is necessary to know the performance for over- shielded and unshielded radiator. Use a convection co-
load and underload of 10 percent of the flow rate of oil at efficient of 1.5 Btu/hr ft 2 °F for both cases at all surfaces.

the same inlet temperature. Predict the outlet temperatures State and justify any assumptions.

of the oil. 15.26. Determine the heat loss by radiation from an

(c) Similarly, what would be the estimated outlet oven wall to the wall of the room within which it is located.

temperatures for the same flow rate, but at inlet tempera- The oven wall is square, 4 ft on a side, and is 2 ft from the
tures 10°C above, and below design conditions? room wall and parallel to it. The oven wall is of steel sheet
and is at 400° F, and the wall is at 80°F and is of asbestos
15.18. A household electric iron is rated at 1000
2 heated surface of stainless steel. board.
watts and has a 3ft
15.27. Determine the total absorptivity of white tile
(a) What is the rate of radiant-energy transfer from
for black-body radiation at 2200 F.
the heated surface when it is at 450 F?
15.28. A thermocouple probe used to measure flue-
(b) What is the net radiant-energy-transfer rate (in
gas temperature from a cement kiln reads 800° F. The probe
watts) from the iron at 450° F to the room in which it sits if
extends 2 ft into the flue from a fitting in the wall. The
the room is at 60°F?
probe is made of ^ -in., sch. 40 pipe with the thermocouple
15.19. If the thermostat on the iron of Problem tip and pipe end welded together. The |-in. pipe is insul-
15.18 were what equilibrium temperature
inoperative, ated with 1 -in thick magnesia pipe insulation out to 1 in.

would be attained by the heated surface? Assume all the from the tip. Duct wall temperature is measured to be
energy loss is by radiation. 300° F. The flue gas is flowing at 25 ft/s at which velocity
15.20. On a clear night, objects on the surface of the its convective coefficient to the probe tip is calculated to be
earth exchange thermal energy with the sky as if the sky 18 Btu/hr ft F. Determine the flue-gas temperature, ignor-

were a black body at — 40°F. What will be the steady-state ing heat transfer through the insulation to the probe.
temperature of a galvanized iron roof on such a night when 15.29 Liquid N2 boiling at — 320°F is held in a glass

the air temperature is 80° F? A convection-heat-transfer Dewar flask. This flask is made of a double layer of silvered
coefficient of 5 Btu/hr ft 2 °F from roof to the air may be glass 4 in. I.D. and \ in. between the concentric glass
used. Assume the roof is perfectly insulated from the room walls. The interwall space is completely evacuated. The flask
below. bottom is hemispherical. The top, covered with a 1$ in.

15.21. An asbestos block 4 in. by 4 in. by 12 in. is thick black rubber stopper with one small hole for vapor
put in a furnace at 3000° F. What is the rate of temperature removal, is 6 in. above the boiling N2 surface. The N2
rise of the block because of radiation when it is at 600 F? liquid is 6 in. deep at the center. Determine the rate of
Assume uniform temperature in the block. The specific evaporation in liters per hour.
heat of asbestos is 0.25 cal/g°C, and its density is 120 15.30. Performance data on a recently constructed
lb/ft 3 . heat exchanger yielded the following data:
15.22. A horizontal steam line of 3-in sch.-80 pipe
Toluene side:
300 ft long carries steam at 250 psig through an unheated
warehouse. The line is insulated with \ in. of 85 percent Flow, gpm* 2050
magnesia and painted black with shiny lacquer of negligible Entering temperature, °F 281.9
thickness. If the steam is dry and saturated as it enters, how Exit temperature, °F 75.6

much liquid water should be trapped out hourly at the

Water side:
other end of the warehouse on a winter day when the
warehouse temperature is 10 F? Flow, gpm* 159,000
by 6 Entering temperature, F 74.2
15.23. White fire-clay tiles \ in. by 6 in. in.

room Exit temperature, °F 75.3

are removed from the kiln at 3000°C and cooled in a

at 100°F. If these shapes are well separated from each other

during cooling, how long will it take them to cool to "Corrected to 60° F.
382 .APPLICATIONS TO EQUIPMENT DESIGN

Assuming the data are accurate and the heat transfer (measured at standard conditions) is to be heated from
coefficient calculated fiom the data is approximately what 40" F to 160 F by passing it through a heater consisting of

one would expect, what do the test data mean? 500 standard 1 -in. steel pipes, in parallel, with steam at
15.31. A heat exchanger consisting of a bundle of 20 6 lb/in gage outside the pipes.
parallel tubes surrounded by a steam jacket is available for (a) Calculate the length of the pipes.
use in a semiworks scale plant. Each of the tubes is thin (b) If it is desired to increase the exit temperature of

walled, of 0.902 in. inside diameter, and 7.5 ft long. It is the air by an additional 10 F, what new steam pressure
planned to heat 600 Ib/hr of air at 1 atm with steam con- would be required?
densing at 220" F. The air is to enter the exchanger at 70" F (c) If a new heater was to be built, using steam at the
and make one pass through the tubes, the steam being in original pressure, what length of tubes would be needed to
=
the jacket outside the tubes. If the available exchanger is to heat the air to 170°F instead of 160 F?
be used under the specified conditions, what exit air tem- 15.36. Water is being heated from 80° F to 160° F in
perature can be expected, assuming the resistance to heat a 2-in. standard steel pipe by saturated steam at 21 8°
transfer on the steam side to be negligible? condensing outside the pipe. If the water velocity is

1532. It is desired to design a simple, adiabatic, 3.0ft's, what length of pipe will be required? If a scale

double-pipe heat exchanger to cool 6000 Ib/hr of water deposit with a thickness of 0.003 in. and a thermal con-
from 200° F to 150°F by means of an equal weight of cold ductivity of 0.5 forms on the inside of the pipe, what
water entering at 100 F. The coolant is to flow through the length of pipe will be required if the capacity and terminal
inner tube, and the water to be cooled is to flow counter- conditions are to remain the same?
currently in the annular space between the outer and inner 15.37. Steam at 100 lb/in. pressure (gage) (satu-

tubes. It is proposed that the annulus be formed by stand- rated) is flowing through a standard 2-in. steel pipe located
ard 2- and 3-in. steel pipes mounted concentrically. If in a large room whose temperature is 70 F. Estimate:
this plan is followed, what is the necessary length of the (a) The percentage of heat saved by lagging the
exchanger? pipe with a 1-in. layer of 85 percent magnesia lagging
15.33. An air preheater, with air flowing inside 1 -in. (k = 0.036).

O.D. iron tubes 10 ft long, is used to heat air from 60" F to (b) The percentage of heat saved by wrapping one
250" F. Stack gases, at an average temperature of 600" F, layer of aluminum foil (e - 0.04) around the pipe.
flow at right angles to the tubes. The tubes are staggered 15.38. A perscn standing 25 ft from a point source
2
and are placed at the vertices of equilateral triangles 2 in. of radiation receives energy at 250 Btu/hr ft . How far

on a side. The preheater is 50 rows deep. The furnace away should the person stand to receive 100 Btu/hr ft ? Or
producing the stack gases burns coke (assumed to be 100 400 Btu/hr ft 2 ?
percent C) at the rate of 2 tons/hr. It produces a flue gas 15.39. A room is heated by a uniformly distributed
containing 10 percent CO2. Compute number of tubes
the heat source in the plaster ceiling, maintaining it at such
needed. Consider the flow of air and gases as follows: Dry temperature that 3000 Btu/hr maintains the concrete floor
air enters the preheater and is discharged into the furnace, at 75" F. The room is 1 5 x 1 5 ft and 8 ft high. Its walls are
where it is used to burn the coke. The excess air and flue nonconductive and reradiating. What is the ceiling temper-
gases pass through the preheater (where they give up part of ature?
their heat content to the incoming air) and are then dis- 15.40. A surburban house has a solar heating panel
charged from the system. of 200 ft . on which 24-hr average insolation is

15.34. A heat exchanger is to be designed to cool oil 40 Btu'ft 2 hr. It is coated with a highly absorptive asphalt
from 450" F to 225" F by passing it through 1 -in. I.D. tubes paint. The hot water supply is heated by this heater to a
at a velocity of 5 ft/s. Another oil will flow outside the temperature of 140 F. How many gallons per day of hot
=
tubes and will be heated from 100 F to 400°F. If the water are available? If this energy is collected during 10 hr
overall coefficient of heat transfer is assumed constant at on most days and a safety factor of two totally cloudy days
50, calculate the length of each tube (i.e., sum of all passes) must be accommodated, how large a storage tank would
required in the following cases: you recommend? Inlet water is at 40 F or above. :

(a) For a double-pipe countercurrent exchanger. 15.41. Air at an arithmetic mean temperature of
(b) For a baffled heat exchanger with 2 shell passes 95" F is being heated while flowing through a duct 4 x 6 in.

and 4 tube passes. The rate is 1200 scfm. If the wall temperature is 250"F,
(c) For a baffled heat exchanger with 6 shell passes compute the temperature rise per foot of duct traversed.
and 12 tube passes. Use the Sieder and Tate equation. Predict the pressure drop
The density and specific heat of both oils may be taken as per foot by the Colburn analogy.
3
55 lb/ft and 0.5 Btu/lb°F, respectively. 15.42. An air cooler is being specified as a 1-1 tube-
15.35. Ten thousand cubic feet of air per minute and-shell exchanger consisting of 200 tubes, 6 ft long and
 HEAT TRANSFER 383

1 in. O.D., in 15 rows. Air is flowing inside the tubes, 300 F. Air enters at 2000 scfm. The overall coefficient is

maintaining the outside surface at 200° F; water at 100°F 3.5 Btu/hr ft F. Determine the corrected logarithmic
flows transversely outside at a rate of 2000 Ib/s. mean AT and the required area for a flow pattern:
Compare the rates of heat transfer and the pressure (a) 1—2 shell and tube counterf low.
drops of water for square-and-triangular pitch, 2 in. center- (b) Single-pass cross-flow, both streams mixed.
to-center spacing in each case. (c) 2—4 shell and tube exchanger.
15.43. A bid is being requested for an economizer (d) Single-pass cross-flow air mixed and gas unmixed.
for a small power plant, large enough to heat 60,000 Ib/hr 15.46. Glycerine is to be cooled by air from 150 F
of water from 150°F. There are 100,000 Ib/hr of flue gases to 90 F in the tube of 1—2 shell and tube exchanger at a

available (Cp = 0.24 Btu/lb°F) at 800° F. Estimate the out- rate of 5 gpm. Find the corrected ATlm, the number of
let gas temperature and area required for a chosen counter- transfer units and the length of tubing (1 tube) required
flow arrangement, capable of performing with a coefficient when 1 in. O.D. 18 ga copper tubing is used, and the overall
2 2
of 10Btu/hrft °F. U is 20 Btu/ft hr°F.
15.44. Air in the amount of 45,000 Ib/hr is being 15.47. The temperature of air flowing through a 10-
heated from 70° F to 170°F, using saturated steam at in. diameter duct whose walls are at 600 F is to be
20 psia. It is found necessary to double the air flow for the measured by a thermo-couple that is inside a thin

same temperature rise. In studying the appropriateness of 1/2 in. O.D. tube, and bonded to the inner surface. The
the present equipment, you must first determine the new tube ends at the centerline of the duct and is closed. Air

steam pressure required. What is your predicted new steam flows in the duct normal to the tube. The thermocouple
pressure? indicates a temperature of 400° F; estimate the true temper-
15.45. Air for a furnace is preheated from 70° F to ature of the air. Comment on the factors that can introduce
120 F in a preheater; flue gases are cooled from 450 F to errors. The air is flowing at 50ft s.
 Sixteen

Mass Transfer

Mass transfer is the basis for a number of important is transferred from the fluid to the solid, there is a

unit operations that involve the separation of the buildup of solute on the solid. Ultimately, saturation
components of gas or liquid mixtures, as discussed is approached, so that the bed must be regenerated.

in Chapter 2. The basic mechanisms of turbulent Fixed-bed adsorption is therefore an unsteady-state

mass transport was examined in Chapters 12 and or batch process, subject to a different physical
13. The principles of interphase mass transport were model, to be considered toward the end of this
developedin Chapter 14; the design equation devel- chapter. At this point, continuous-contact equip-
oped there serves as the basis of the design pro- ment for steady-state operation will be examined.
cedures presented in this chapter.
The equipment considered in this chapter STEADY-STATE TRANSFER
provides more or less continuous contact of the two
phases between which mass is being transferred. Packed towers are widely used to provide con-
The physicalmodel describing continuous contact tinuous contact between two fluid phases, as re-
differs from that introduced in Chapter 2 for dis- quired in absorption, distillation, and extraction.
continuous contact. In that case, the equilibrium Columns employing stages or plates may also be
stage was the physical model for the repeated mix- used for these separation processes, as discussed in
ing, equilibration, and separation of phases en- Chapter 2. These two contacting devices offer con-
countered in staged contacting equipment. trasting advantages and disadvantages:

The starting point for the design of con- Plate columns provide more positive con-
1.
tinuous-contact equipment is Equation 14.63. tact between the two fluid phases by repeated
mixing and separation; whereas packed columns
d(Lca
may be subject to bypassing or back-mixing.
J
1 -1 K(ca -c ,*) Ji K(c a -c3 *) 2. Plate columns can handle greater liquid

(14.63) loads without flooding.

3. columns are more easily cleaned.
Plate
This equation may
be applied to various separation
Packed columns give lower pressure drop
4.
processes, including absorption, distillation, extrac-
for gas flow, which is of particular importance in
tion, and adsorption.
vacuum operation.
In absorption, distillation, and extraction, the
5. Liquid holdup is less in packed columns.
two fluid phases are fed at constant rate to the Packed columns may be more economical
6.
contacting equipment, and normal operation is
in processing corrosive liquids, because of cor-
essentially steady state. On the other hand, where
rosion-resistant ceramic packing.
one of the phases is a solid, as in adsorption or ion
7. Small-diameter packed columns are usually
exchange, it is difficult to design equipment to
cheaper than plate columns of the same size.
maintain a flow of the solid. As a result, these
operations often use a fixed bed of the solid phase, Plate columns are much more widely used than
with the fluid phase fed continuously. As the solute packed columns, especially for distillation. Packed

384
 MASS TRANSFER 385

-4 — Phase L inlet 2. It should have a large void volume. This

Phase V outlet
-=; Iw Distributor will allow reasonable throughput of phases without
excessive pressure drop.
3. It should have good wetting character-

•Packing istics.

4. should be corrosion resistant.

It

5. should have a low bulk density. In large

It

packed towers, the weight of the packing can be

quite large, thereby resulting in serious support
problems.
— Column
Phase
—-d
V inlet E
-Packing support
6. It should be relatively inexpensive.

common pack-
Figure 16.2 illustrates several
Phase L outlet ing Raschig rings are widely used in the
shapes.
Figure 16.1. Packed tower components. process industries because of their low cost,
although they may not be as efficient as some of
columns are typically employed in operations where the newer packing materials. The wall thickness of
smaller-diameter columns are sufficient to handle the Raschig ring is an important factor because, as
the fluid flow rates. the wall thickness is decreased, mechanical strength
Packed towers used in continuous counter- decreases. A greater wall thickness will result in an
current contacting of two phases are vertical shells increased pressure lower free space, and
drop,
filledwith a suitable material having a large surface reduced surface area.results are obtained
Best
area. Figure 16.1 illustrates the main components when walls are relatively thin. The diameter and
of a packed tower. the height of the Raschig ring are equal. Raschig
Phase L enters the top of the column and is rings may be fabricated from porcelain, clays,
distributed over the surface of the packing either by carbon, or metals. The "Intalox" saddle gives a
nozzles or distribution plates. In most of the mass- greater degree of randomness than the Raschig
transfer operations, phase L is a liquid. Thus, the ring. It is designed to give a more effective nesting
liquid, upon good initial distribution, flows down of pieces so that good wetting characteristics are
through the packing by tortuous routes, thereby obtained. This packing is more expensive in initial
exposing a large surface area for contacting with the cost than the Raschig ring. Berl saddles are costly
rising V-phase. The V-phase enters the bottom of to produce but do have some advantage over other
the tower and rises upward through a similarly packings. They can be packed with more random-
tortuous path. The V-phase for absorption and ness than rings, and they give a relatively large
distillation is a gas or vapor, and in extraction it is amount of surface area per unit volume. The Pall
a liquid. ring has sections of the ring wall stamped and bent
The column itself is nothing more than a inward to give better circulation of the contacting
simple cylinder, such as a length of pipe. Care must phases.
be exercised in choosing the proper materials of For many packed-tower operations using rings
construction depending upon the chemicals being in large-diameter towers, the packing is usually
handled. The simplicity in shell and in the entire arranged stacked configuration. The cyclohelix
in a

assembly is another advantage of packed columns. is good for use in stacked-packing installations. The

internal helix affords good contact between gas

Tower Packings and liquid. The cross-partition and Lessing rings are
simple modifications of the Raschig ring to
Many types of packing materials have been used,
improve operating characteristics.
ranging from simple, readily available solids such as
Actually, no one packing possesses all the above
stones or broken bottles to expensive complex
desirable qualities and so compromises must be
geometric shapes. In general, the packing material
made. Although the packing material itself may be
should have the following characteristics:
designed to give excellent phase contacting, the
1. It should have a large wetted surface area method of packing into the tower must also be
per unit volume of packed space so as to present a considered because, if the liquid and gas do not
potentially large interfacial area for phase con- contact everywhere within the tower, the packing
tacting. is not completely effective.
386 APPLICATIONS TO EQUIPMENT DESIGN

Table 16.1 PHYSICAL CHARACTERISTICS OF DRY

COMMERCIAL PACKINGS*

Specific
Surface Packing Dumped
Percent (a v ) Factor Weight
Packing Voids (e) ft
2 /ft 3 F lb/ft 3

Ceramic Raschig rings

\ in. 73 240 1600 46
i

2 63 111 580 54
1 73 58 155 40
2 74 28 65 38
Metal Rasch g rings (1/16 in. wall)
i

2 73 118 410 132

1 85 57 137 71
2 92 31.4 57 37

Berl saddles
k in. 60 274 900 56
i

7 63 142 240 54
1 69 76 110 45
2 72 32 45 40

Pall rings

1 in. 93.4 66.3 48 33

2 94.0 36.6 20 27.5

Cyclohelix and single spiral

3k in. 58 40 60
4 60 32 61
6 66 21 59

* Courtesy of Norton Co.

(e)

diameter columns, it is customary to employ liquid

redistributors at spacings of several diameters to
redirect the flow toward the center of the packing.
With good initial liquid distribution and a ratio of
tower diameter to packing size greater than 8 to 1
the tendency of the liquid channeling toward the
wall is diminished but not eliminated, except with
liquid redistributors. Channeling is a principal
contribution toward poor performance in packed
(f) <«»
columns.
Figure 16.2. Common packing shapes. (Courtesy Norton
Table 16.1 lists some physical characteristics
Co.) (a) Rashig rings, (b) Intalox saddle, (c) Pall rings. (cO
of various types of industrial packing materials. In
Cyclohelix spiral ring, (e) Berl saddle, (f) Lessing ring, (g)
this table, the percent voids is the percentage of
Cross-partition ring.
total packed volume not occupied by the solid
packing itself. The voids are the empty space
The downflowing tends to select pre-
liquid
between the pieces of packing. The specific surface
is the surface area of the packing per unit volume.
ferred its flow down the
paths, or channels, for
packing. This tendency is referred to as channeling. The packing factor is an empirical quantity used in
As may be expected, the fluid tends to move a pressure-drop correlation later in this chapter.

toward the region of greatest void space, which is

the region near the wall, because the packing Design of Packed Towers

material cannot nest so tightly with the plane wall The basic question that must be answered in the
as it can with itself. Consequently, in large- design of packed towers concerns the interfacial
 MASS TRANSFER 387

area between phases that must be provided to

ensure that the specified mass transfer can take
place. As shown in Equation 14.63, that area is
directly proportional to the composition change to
be accomplished, and inversely proportional to the
concentration driving force between phases. The
proportionality constant is the mass-transfer co-
efficient, which is influenced by many system
factors, including type of packing, flow rates and
properties of the fluid phases, and the diffusion
characteristics of the particular chemical species
being transferred. The area of transfer can be re-
lated to the volume of packing required and hence
to the size of the absorption tower.
In packed columns, the two fluid phases are
in contact with each other throughout the height
of the tower. The concentration of the component
being transferred varies in each phase as it flows
through the column; hence, the concentration
driving force varies along the height of the column.
Figure 16.3. Packed tower.
The phase flow rates and mass transfer coefficient
may also vary. To cope with these variations, it is As shown in Chapter 14, compositions of the
necessary to describe the mass transfer rate for a fluid phases can be represented in a variety of
differential height of column and then integrate ways, such as partial pressure for gases, concentra-
over the total height. tions in moles per unit volume for liquids, or mole
In the development that follows, certain con- ratios for either phase. For simplicity, only mole
ventions will be employed to avoid confusion: fractions will be used to express compositions in
this development. The various composition expres-
Transfer of mass from the /.-phase to the
1.
sions and resulting mass transfer coefficients are
V-phase will be regarded as positive.
summarized in Table 13.2.
2. Integration of the design equation will be
At steady-state operation, a mass balance over
carried out from the bottom of the tower (desig-
the differential tower section, shaded in Figure
nated position 1 to the top of the tower (designated
)
16.3, yields
position 2).
3. The upward-flowing phase will be designated dV = dL 16.1

the V-phase, whereas the downward-flowing phase and a component balance over the same section
will be the /.-phase. gives

A packed tower for mass transfer is shown d(Vy)=d(Lx) (16.2)

schematically in Figure 16.3. The following nota- Equation 16.3 relates the bulk compositions
tion is used: of the two phases in contact at any point through-
out the length of the tower. Upon integration
V= molar flow rate of phase V, lb moles/hr
between the tower bottom and any point within
(or moles/s)*
the tower, Equation 16.2 becomes
L = molar flow rate of phase L, lb moles/hr
(or moles/s) Vy - V-^y^ Lx - L^x,
y = mole fraction transferent component in the
or
V phase
x = mole fraction transferent component in the V> + /.!*! = Lx + V : y, 16.3)
L phase
z = tower height, ft (or meters) Equation 16.3 is the operating-line equation and is

Subscript 1 = tower bottom valid for all x between x, and x 2 and all
values of
Subscript 2 = tower top values of y between y^ and y 2 In the general case
.

in which V and L vary with change in concentra-

* SI units in parentheses. tion or position, this equation describes a curved
388 APPLICATIONS TO EQUIPMENT DESIGN

operating line. For certain cases of extreme dilu- In a similar manner, the tower height can be
tion which the change in concentration has
in evaluated using /.-phase changes in composition
negligible effect upon V and /., the operating line along with the corresponding rate, with all
as defined in Equation 16.3 is substantially linear. terminology based upon the liquid phase.
In accordance with the development of Equa-
d{Lx) .x2 d(Lx)
tion 14.60, the rate of change of a component
is 'aSlv — yA
~
within a phase must be equal to the rate of transfer
J„
'x, k x 'aS(x-Xi)
J„
J*, Kx 'aS(x-x*)
to the phase. Thus, for the V-phase in equimolar
(16.8)
counterdiffusion (as in distillation)
To use Equation 16.7 or 16.8 requires know-
d( Vy) = ky '(yi - y) dA = Ky '(y* - y) dA ledge of the following:
(16.4)
1. The value of kx 'a, k 'a, Kx 'a, or K 'a as a
y y
where dA is the interfacial transfer area associated function of the flow rates of V and L and the
with the differential tower height. This transfer transport properties of these two phases.
area in a packed column is extremely difficult to 2. The relationship between point composi-

measure and is more conveniently expressed as tions in either phase and point driving forces,
(x-x,), (x-x*), tyj-y), or iy*-y). Bulk
dA=aSdz (16.5)
compositions in either phase may be calculated
where from Equation 16.3 and driving forces may be
determined by methods described in Chapter 14.
a = interfacial area per unit volume of packing
S = empty tower cross-sectional area The design equation for the case of diffusion
through component may be obtained
a stationary
The term Equation 16.5 is not to be
a in
in same manner as described for Equations
the
confused with a v in Table 16.1, which is written
16.7 and 16.8 by employing the appropriate mass-
for dry packing. This new term is a function of a
transfer coefficient to account for the difference
combination of packing characteristics and re-
in mechanisms, as explained in Chapter 9. Thus,
tained liquid. It is therefore dependent upon the
for diffusion through a stationary component such
flow rates of the two phases. Because a is usually
as is encountered in absorption.
unknown in packed columns, it is combined with
the surface coefficient to give a composite d(Vy) d(Vy)
coefficient k with units of amount transferred
y
'a
y t
k y aS{y, - y) -V, Ky aS{y* - y)
per unit of (time) (driving force) (volume of pack-
ing). Thus, Equation 16.4 may be combined with (16.9)
Equation 16.5 to give
and in terms of the /.-phase
d( Vy) = ky'aiy,- - y)S dz = Ky 'a(y* - y)S dz d(Lx) d(Lx)
dz =
(16.6) •'o -x, k x aS{x — Xj) -x, Kx aS{x - x*)
Equation 16.6 may be used to solve for the tower (16.10)
height required by integrating over the total change
in concentration between the tower terminals. Thus, Equations 16.7, 16.8, 16.9, and 16.10
are design equations for mass-transfer
rigorous
d(Vy) y: d(Vy) operations involving equimolar counterdiffusion or
dz =
•V, ky'aS(yi-y) J„,
Ky 'aS(y* - y) diffusion through a nondiffusing phase. These
equations are the exact counterpart to Equation
(16.7)
14.63, the design equation for mass transfer. The
Equation 16.7 is actually two separate equations, equations will now be examined in more detail in
either ofwhich may be used to determine the an effort to attach some physical significance to
tower height. Use of the gas-phase mass-transfer them.
coefficient k y 'a requires knowledge of the inter-
facial gas-phase composition Equimolar Counter Diffusion
yh whereas use of the
overall gas-phase coefficient K y 'a requires the gas Equimolar counter diffusion occurs in distillation
composition (/*) in equilibrium with the bulk since there is an equal rate of interchange of the
liquid composition. more and less volatile constituent between phases.
 MASS TRANSFER 389

If the assumption of constant molar overflow is where the various symbols are defined in Table
made (Chapter 7), Equation 16.9 may be written 16.2. The choice of which combination of H and N
as to use depends upon the form in which the mass-
transfer coefficient is available. Before discussing
dZ=V " V the uses of these expressions, similar relationships
\o I, ky'aSW, - Y) V, Ky'aSW* - y)
will be derived for diffusion through a stationary
(16.11) component.
where the molar flow rate (V) is now considered
constant through the enriching section of the Diffusion Through a Stationary Component
tower. be constant in the stripping
It will also Mass transfer operations involving diffusion
section but of a different magnitude, depending through a stationary component include gas
upon the feed quantity and conditions. For the absorption, liquid extraction, adsorption, and ion
same assumptions, Equation 16.8 becomes exchange. The starting point for the development
dx x1 dx of working equations will again be Equation 16.9
dz = L = L or 16.10. However, unlike distillation, the assump-
1 \
'x, k x 'aS(x — Xj) J
x, Kx 'aS(x — x*) tion of constant V and L is not valid. For example,
(16.12) in gas absorption both phases will have variable
flow rates as a result of the transfer of solute to or
Some simplifications may be made to put from the phases. This complication can be circum-
Equation 16.11 or 16.12 into a more usable form. vented by recognizing that the quantity of solute-
The phase flow rates (L and V) have already been free gas, which is essentially insoluble, is a con-
assumed constant. Since the individual and overall stant. Therefore, let V
be the molar flow rate of
mass-transfer coefficients are functions of these the solute-free gas and V' = V(1 — /); thus
flow rates, the coefficients must also be constant
providing physical properties do not vary appreci-
ably. Therefore, d(Vy) =
i y
V'd(-^-)= V'—-L 2
\ dy
— = V- dy :

\1 -// (1 -y) (1 -/)

r Z V
V .v, dy dy
(16.16)
J
o ky'aS J
K , (/,- - y) K'aS
,
v, W* -
y a^ I y)
Consequently, in accordance with Equation 16.6,
(16.13)
the design equation may be written as
and
* L dx L *2 dx dy_
r
dz =
r P
d( Vy) = V ;
= k a(y,
y
- y)S dz
J
o kx 'aS Jx, ix-Xj) Kx 'aS Jx, (x-x*) (1-W
(16.14)
= Ky a(y* - y)S dz

Tower heights required for a given separation (16.17)

can be obtained by use of either Equation 16.13 or Solving Equation 16.17 for the tower height,
Equation 16.14, provided that the assumptions
stated are valid. Note thatboth equations the
in
r
£ v,
r 'j V dy
dz=\
integral term is equal to the total change of compo- k y aS (1 - y)(y/ - y)
sition for the particular phase divided by the avail- V dy
k2
able driving force. This is a measure of the ease or
difficulty of separation and upon integration gives , Ky aS (1 -y)(y*-y)
a quantity defined by Chilton and Colburn (3) as (16.18)
the number of transfer units (A/). The physical
In terms of the /.-phase, the design equation is
significance of this term will be considered later in
the chapter. The quantity outside the integral is
called the height of a transfer unit (H). The tower r**-f*-* *
j
o k aS Jx, x (1 -x)(x-x,)
height is determined by multiplying the number of
transfer units by the height of a transfer unit, *5 L dx
z = H G N G =H OG N OG =H L N L =H 0L N 0L Jx, Kx aS (1 -x)(x-x*)
(16.15) (16.19)
390 APPLICATIONS TO EQUIPMENT DESIGN

Table 16.2 THE NUMBER OF TRANSFER UNITS (A') AND THE HEIGHT OF A TRANSFER UNIT (//)

Driving
Mechanism Number of Transfer Units Height of a Transfer Unit
Force

r v, dy V
NG hg n-y
k Y 'aS

rV* dy V
Nog Hog y*-y
Equimolar V, y* -y Ky 'aS
counter
diffusion r*> dX L
NL »L X — Xj
Jx, X — Xj k x 'aS

r*j dx L
Nol HOL x-x*
Jx, x-x* Kx 'aS

V
NG Hg Yi y
-V, d -y)(Vi-y) kyasn -y), m

?v* H-y)imdy V #
Nog Hog y -y
Diffusion through •V, d -y)(y*-y) KyaStf -y),m
a stationary
component
NL
* (1 -x), m dx
hl
L
X — Xj
Jx, (1 -x)(x-x,) k xaS(} -x), m

r* a (1 -x), m dx L *
Nol hol X X
J
x ,
(1 -x)(x-x*) /C x aS(1 -x), m

Equations 16.18 and 16.19 are rigorous design developed, this is not possible. As shown in

equations that are applicable to diffusion through Chapter 13,

stationary nondiffusing phases. These equations are
always appropriate, although it is often possible to k c 'D
= 0.023(/V Re 083 (A/ S c)
- 44 (13.83a)
make simplifications in their use. As before, the 9 )

equations may be expressed in terms of transfer above equation is for the mass-
Although the
units and the height of a transfer unit, although the
transfer coefficient k c it will be shown later that a
'
,

latter quantity remains within the integral. Or

similar correlation for the composite coefficient
V: k c a exists. Thus, for the time being it is postulated
'

j
dz= j
'
H G dN c = ' H 0G dN 0C
y \

k'aD ,.
6
^-=C(/V Re )
a
(/V Sc )
16.21!
.X
= I
"
H L dNL = I
'
H 0L dN 0L (16.20)
J
x, -x, For a given system of specified flow condi-
tions, geometry, and physical properties, k y 'a is a
where G H H
0G are defined as before. The
, constant. The relationship between equimolar
transfer-unit definition includes the term (1—/) or counterdiffusion and diffusion of a solute through
(1 — x) in the denominator, as shown in a stationary component is, from Chapter 13,
Table 16.2.
It would be convenient for ease of solution if Ky 3 ~ Ky3 (16.22)
Yt
the terms equivalent to H G or H OG could be
removed from the integral sign and treated as a Because k y 'a is constant for a specified system, k y a
constant. However, as the equation has been YbimlYt must be independent of concentration.
 MASS TRANSFER 391

This fact will be used to simplify the integral term area for transfer large, then the height of a transfer
of Equation 16.18, the design equation by manipu- unit will be small. Some individuals contend that
lation of equations. the height of a transfer unit is easier to use because
Thenumerator and denominator of the it has a simple dimension (length), in contrast to

second term of Equation 16.18 are divided by the complex dimensions of the mass-transfer co-
Ybimlyt- This ratio will be expressed in terms of efficient. This is only a superficial advantage,
component a by expressing yb as (1 — y), and by because evaluation of the height of a transfer unit
taking the logarithmic mean of (1 — y) in the bulk requires knowledge of the mass-transfer coefficient
stream and at the interface. in units consistent with those in which the phase
This results in composition is expressed, as shown in Table 16.2.
Before examining methods for calculating
V (1 -y)im dy
dz = numbers of and heights of transfer units, it would
Vi k Y aSC\ -y),m d -/)(/,-/) be well to consider the physical meaning of a single
transfer unit. As an example, the expression for the
V
1 y , KY aSC\ -y) lm
(1

(1
~y)imdy
-y)(y<
number of overall gas-phase transfer units (N OG
found in Table 1 6.2 can be set equal to 1 .0.
)

(16.23) . v, dy
L-= y2 — /i
/V 06 = 1.0= ~r I (16.24)
By Equations 16.21 and 16.22, the term y*~y .
,
•V, (K*-K)mean
k r^3(1 —y)/ m in the denominator is equivalent to
ky'a and is independent of concentration. It is
A single transfer unit therefore gives a change
of composition of one of the phases equal to the
dependent upon the geometry of the apparatus and
average driving force producing the change. The
on the flow rate of the \/-phase.
action of a transfer unit can be compared to that
Correlations for mass transfer in general
of an equilibrium stage, as shown in Figure 16.4.
involve flow rates, frequently expressing a propor-
The equilibrium is shown by the solid
V08
stage
tionality between k and As a reasonable,
.

line to V-phase composition of

give a change in
though obviously not rigorous, approximation, k v a
y2
'

Ki —
which is equal to the driving force (y-\ * —
will be assumed proportional to V over the range
y-\ based upon the liquid composition leaving the
)
involved in one application. On the basis of these
stage (x ). The transfer unit shown by the dotted
assumptions, the group of terms Vlk y aS (1 — y) /m 1

line gives a change in composition y 2 — y-\ equal to

and VlK y aS{\ —y)\ m are assumed constant for a
the average driving force in the transfer unit. The
particular contactor, usually at an average of the
driving force changes as the composition changes.
terminal and may be removed from the
values,
The driving force is Y\* — y\ at (x, , y^ ) and
integral inEquation 16.23, to yield the definitions
of H and N shown in Table 16.2, for diffusion
through a stationary component.

What Is a Transfer Unit? Equilibrium line

>2
The manipulations of the basic design equations a
A
carried out in the previous paragraphs have led to sL
c
an apparent simplification of the equations, so that >1 / v

//
a i

the height of a packed tower can be easily obtained xi

c
by multiplying the number of transfer units o
a.
r / 2
E x2
required by the height of the transfer unit. In o
u
reality, each of these terms is a complex expression c
g
that may prove difficult to evaluate. The number o
/B
of transfer units required is a measure of the o
o
difficulty of the separation. Hence, if very high- £
>\
*>
purity products are desired, a larger number of S x\
transfer be required. The height of a
units will Operating line

transfer unit is a measure of the separation

effectiveness of the particular packing for the r. mole fraction component a in /,-phase

chemical species being processed. If the rate of Figure 16.4. Comparison of a transfer unit and an equilib-
interphase mass transfer is high and the surface rium stage.
39: APPLICATIONS TO EQUIPMENT DESIGN

y2 * —
Yi at (x x » Yi)- In the most general terms,
the average driving force for the gas phase is

=-
<y*-v) m 16.251
ra dy
L
v. y -/
Liquid phase concentrations and driving forces
may show the difference between
also be used to
an equilibrium stage and a transfer unit. Solution
of Equation 16.25 depends upon the relationship
z note fractal aflate - L-z-n-e
between y* and y. If both the equilibrium curve
Figure 16.5. xy diagram for absorption.
and operating lines are straight or nearly so, solu-
tion of Equation 16.25 gives the arithmetic
a.e-age. the tower Ui Y\ ) to the compositions at the other
,

W'-y
^'-*-;^'-^ <16.26.
end (x 2 y 2 ). The graphical method can be used for
,

nonlinear operating lines and equilibrium curves, as

shown. For any value (x, y) o n the operating line, a
This equation requires a trial-and-error solu- line of slope (— k x alk Y a) may be drawn to the

tion to evaluate y 2 but the solution is not very

.
complex. In Figure 16.6, the dotted line indicates
one transfer unit with the separation in this unit
equal to (y 2 —Y\) which is determined by trial
such that (y 2 — Ki ) = AB, the average driving force
of y 2 and yj The use of Equation 16.26 will offer
.
a convenient tool toward a graphical determination
of the number of transfer units required for a given
separation.
equilibrium curve. As shown in Chapter 14,* the
point of intersection of this line with the equilib-
rium curve has the coordinates (x,-, /,-), which are
the compositions at the interface between the two
phases. Values of y, determined in this manner can
be used with corresponding values of y to perform
a graphical integration of Equation 16.28. Values
of y (and corresponding values of y,) are needed
over the range of integration, from y-, to y 2 The .

graphical integration is shown in Figure 16.6.

Calculation of the Number of Transfer Units. Equation 16.27 can be integrated even if the
Evaluation of the number of transfer units for a mass transfer coefficients for the individual phases
are not known. As shown in Figure 16.5, the
given separation made by integrating the appro-
is

priate expression from Table 16.2. Several methods

equilibrium composition y* can be determined for
of integration may be used. As an example, the a given y by dropping a vertical line to the equilib-

expressions for V-phase transfer in diffusion rium curve. Equation 1627 can then be integrated
through a stationary component (as in gas absorp- by methods similar to those used for Equation
tion) 16.28, as shown in Figure 16.6.
should be noted that the slope (—k Y a.k x a\
It
,k, (1 -y),m dy can be related to H G and HL if the rates of mass
M6.27)
(1 -y)(y*-y) transfer are given in this alternate form, using the
expressions found in Table 16.2.
U-Vhmdy Equation 16.27 or
Nn = M6.28) :erical Integration.
d-y)(y,-y) 16.28 can be integrated numerically, using any of a
variety of numerical integration techniques. This
will be integrated using several methods.
procedure is especially convenient if the equilib-
rium data are available in equation form. The equa-
Graphical Integration. If data are available
tion may be analytical (such as one based upon
on the individual mass-transfer coefficient for each
Raoult's law, if applicable), or empirical (as fitted
phase, then a graphical method can be used to
to a set of equilibrium data).
integrate Equation 16.27 or 1628. First, the
equilibrium curve and operating line are plotted on
the same figure, as in Figure 16.5. The operating * Since Na = k xa{x - x,) = kyaiyj — y). then [y — y/)f
- J
-:~ :-e :;~c:;: :~$
~e e->:5 :s a: z~e e~z z~ Or — xj) = —kga/kya.
 MASS TRANSFER 393

Equation 16.29 is specifically for four equal seg-

ments, but it can be easily generalized to any
number. As Figure 16.7 shows, the trapezoidal rule
tends to overestimate the area under the curve for
functions of the shape shown, a shape often
encountered in the integration of expressions for
transfer units. Dividing the integral into a larger
number of segments increases the accuracy of the
numerical integration.
Simplification of the Integral. For many
instances of diffusion through a stationary compo-
nent, such as gas absorption or extraction, Wiegand
(15) showed that (1 — y), m may be approximated
Figure 16.6. Graphical integration for Nq or Nqq.
by an arithmetic average of (1 — y) and (1 — /*).
If the small error introduced by this approximation

is acceptable Equation 16.27 integrates to

A
simple method of numerical integration is dy
based upon the trapezoidal rule, illustrated for N G
= + *ln-
-y
— (16.30)
Equation 16.27 in Figure 16.7. This method k, W*-y) 1 y

simply replaces the continuous function f(y) with a Normally the term
integral must be evaluated
series of straight-line segments. The integration
graphically, but it is much simpler than the integra-
interval from y to y 2 is divided into several seg-
y
tion of Equation 16.27. In many instances, the last
ments, preferably equal, if convenient, by selecting term of Equation 16.30 is negligible, but the term
several intermediate values of y (/,, //, /* in Figure
must be examined for each case. Often for dilute
16.7). Then f(y) is evaluated at each value of y
solutions, (1 — y) and (1 —y)/ m are approximately
from /t to and the areas of the resulting
y2 equal, so that they may be assumed to cancel in
trapezoids are evaluated and summed. For Equation 16.30, again resulting in the simplified
example, the area of the trapezoid bounded by y 2 integral. In any event, it should be emphasized that
and Yi is (y 2 -//) [ffo) + fWM I 2 lf equal seg- -
the simplifications cannot be made without regard
ments are chosen (such as the four shown in Figure for the accuracy of the assumptions and the error
16.7), the integral is given by a simple numerical
to be encountered.
expression number
The other expressions for of transfer
units given in Table 16.2 can be simplified in the
'
f(y)dy = [f(y 2 + ) 2f{y,) + 2f(yj) + 2f(y k
J )
above manner. For equimolar counterdiffusion, the
final term of Equation 16.30 does not appear.
+ ^i)](/2-Ki)/8 16.29) Equation 16.30 is given in terms of mole-
fraction units, but other units may be used. For
instance, in terms of mole ratios,

N,OG
y 2 dY
Yi-Y
In —
1 + Y2
(16.31)

The choice of equation to use is somewhat

arbitrary, but, just as in heat transferwhere the
overall coefficient is usually based upon the area
associated with the largest resistance, it seems
logical for purposes of accuracy to use the equa-
tion that involves driving forces in terms of the
principal resistance.That is, if the gas phase can be
said tobe controlling, then the driving force across
the gas phase is large and can be determined with
good precision. In this case, Equation 16.28 would
Figure 16.7. Numerical integration of Equation be recommended.
16.27. Graphical Method of Baker. From Equation
394 APPLICATIONS TO EQUIPMENT DESIGN

16.24, one transfer unit may be defined as For the equilibrium curve, let

=
y* = mx + C (16.33)
Noc = 1 ;
16.24)
iy*-y), Then
A simple graphical procedure for evaluating y2 * = mx 2 + C
the number of transfer units needed to accomplish and
a certain change has been proposed by Baker (2)
yy * = mxf + C
based upon the use of an arithmetic average of the
driving forces and the assumption that, over the or, eliminating C
length of one transfer unit, both operating line and
Vi -Ki
equilibrium line are essentially linear. The pro- m ; 16.34)
cedure is construction
illustrated in Figure 16.8. A *2-*1
line AB is located midway (vertically) between the The equation for the operating line is
operating and equilibrium lines, and the evaluation
is carried out as follows. Starting at point F, a line
L(x-x,) = Viy-y,)
is drawn horizontally to the halfway line (AB) at or
point G, and is extended a distance equal to FG V
From H, M on the
x= x^ + jiy-yi)
,

!16.35)
beyond to H. a vertical line to
operating linedrawn. FHM is one transfer unit,
is Therefore,
since it gives a change in gas-phase composition of
V
Vm —Vf, which is equal to the average driving x^+-[y-y^) + C 16.36)
force KP. The procedure is now continued so that
MN = NO, etc. Substitution for y* in Equation 16.31 from
Log-Mean Driving Force. Another simplifi-
Equation 16.36 gives
cation that can be made of Equation 16.27 is of
importance the case of dilute concentrations. It
in rV* dy
is of interest because of the similarity with the
N OG
simplified heat-transfer design equation using log-
mean driving forces. In a dilute mixture, Equation y dy
16.37)
16.30 reduces to Y V
m *i+-ty-y-\) + C-y
Nog=
r
y7
—dy (16.32)
•V. V -V After the denominator is rearranged to put Equa-
For such mixture, over the range of
a dilute tion 16.37 into the form dx/(a + bx), the equation
concentrations involved, both equilibrium and may be integrated to give
operating lines are assumed to be linear. This
assumption is analogous to the assumption of con- m *1 +7-1/2 -/l) + C-/2
stant Cp in heat-exchange design which yielded a N OG
linear T- versus- q curve. mV mxy + C — Ki

(16.38)

Equation 16.38 may be simplified by substituting

appropriate forms of Equations 16.33, 16.34, and
16.35.

=
1 /2 - /2
Nqg
V7 Ki x2 -x 1 /i /i

*2 - *1 /2 - /1
(16.39)
Further rearrangement gives

OG
A/,
K2-/1 _ K2* ~/2
mole fraction component a in L-phase (/2*~/2)-(/l*-/l) V\*-V\
Figure 16.8. Graphical evaluation of N 0G .
(16.40)
 MASS TRANSFER 395

But by definition To calculate V2

the total molar gas rate out of the
column, the recognition that all the
requires
(/2* -/2> ~ (/1* ~/l) solute-free gas entering the column must also leave
, K2*~/2 the top of the absorber in the gas stream if no
solubility in the oil is assumed. Therefore,

Therefore, Equation 16.40 reduces to Solute-free gas in = solute-free gas out

= 0.8 lb moles/mole gas in
K2-/1
A/ OG (16.41!
(y* - y) /m Hydrocarbon out = (0.2M0.05) = 0.01 lb moles

Equation 16.41 is restricted to dilute solu- Therefore

tions and straight equilibrium and operating lines.
0.01
Unfortunately, unlike the use of (AT) /m , the y2 = = 0.01235
— y*) /m 0.01 +0.8
applicability of (y is not so general. Many

times however, for dilute-solution situations, plot- and

ting of the equilibrium curve and operating line in
(156.2X0.8)
mole-ratio units will make these lines linear and V2 = : = 1 26.6 lb moles/hr ft
2

validate the use of Equation 16.41 in mole ratio (1-0.01235)

units.
Thus, the flow rates and compositions of the
gas stream are determined. For the oil,
Example A gas stream containing a
16.1.
10,000
valuable hydrocarbon (MW = 44) is to be scrubbed L2 = = 33.3 lb moles/hr ft
2

with a nonvolatile oil (MW = 300) in a tower 300

packed with 1 -in. Raschig rings. The entering gas
and
analyzes 20 mole percent hydrocarbon and 95 per-
cent of this hydrocarbon is to be recovered. The x2 =0
inert gas molecular weight is 29. The gas stream
2
Pound moles of hydrocarbon in exit oil
enters the column at 5000 Ib/hr ft and hydro- (0.95)(0.2) = 0.19 lb moles/mole gas in. Therefore,
carbon-free oil enters the top of the column at
10,000 Ib/hr ft
2
. Determine the N 0G for this Hydrocarbon
operation. in exit oil = 0.19 x 156.2 = 29.7 lb moles
Solute-free oil = = 33.3 lb moles
Solution. Because gas absorption involves dif- =
Total moles
fusion through a stationary component, the appro-
priate expression for the number of transfer units and
is (Table 16.2)
*1 =
—
29.7

63.0
- = 0.472
* (1 -y)im dy
Nog
I
=
K, (1 "/)(/*-/)
(16.27)
A summary of the material balance
reference to the following sketch.
is made by

To evaluate the integral, it is necessary to know the

V2 = moles /hrfP ~|
126.6 lb | L9 = 33.3 lb moles/hr ft
2

limits of and y 2 that can be

integration [y^ )
y 2 = 0.0124 x2 =
"I
obtained by a mass balance on the column.
Basis: 1 lb mole of entering gas

Solute-free gas in = (0.8) (29) = 23.2 lb

Hydrocarbon = (0.2)(44) = 8.8 lb

Average molecular weight of gas = 32.0 lb

The total molar gas rate entering the bottom of the 2

V, = 156.2 lb moles/hr ft Lj = 63.0 lb moles/hr ft'
absorber is = 0.20 i, = 0.472
y,

5000 quite evident that the phase-flow rates

V^ = = 2 It is
1 56.2 lb moles/hr ft
32.0 are not constant throughout the column so that
396 APPLICATIONS TO EQUIPMENT DESIGN

the treatment used in the development of Equation 0.20

16.16 is needed. To plot the operating line, a
material balance is written from the tower bottom
to any point within the tower, or

V y yy + Lx = Vy+ L^x, (a)

This equation may be rewritten in terms of

the nondiffusing component where

I/' =1/(1 -y) (b)

and
/.' = /.(! -x) (c)

substituting Equations b and c into Equation a

gives

(d)

0.1 0.2 0.3 0.4 0.5

Equation (d) is valid throughout the column, x, mole fraction hydrocarbon
and simply by knowing \f L' x-\ and yy any , , ,
Figure 16.9. Operating and equilibrium lines.
point (x, y) intermediate between the terminals
Example 16.1.
may be obtained. Substituting the known values of
V L' X! and y-\ into Equation (d) gives
, , ,

evaluating Nqg are available. The graphical evalua-

(0.2) (x) (y)

tion of the integral term Equation 16.27 is
in
1 25.2 7^17 + 33.3 :
= 1 25.2 always valid since it is substantially free of any
(0.8) (1 -x) (1 -y) assumptions. This solution, of course, is somewhat
(0.472) tedious so that the solution to the problem should
+ 33.3 (e) be judged from the point of view of the simplifica-
(0.528)
tions that can be made and the accuracy of the
result required.
where V' = (156.2)(0.8) = 125.2 and Z.' = 33.3.
First, examination of the equilibrium line and
Thus the equation for the operating line becomes
operating line shows them both to be curved over
the range of concentrations involved. This curva-
x/(1 - x) = 3.76(//1 - y) - 0.045 (f)
ture would introduce serious error if the log-
mean driving force were employed. Furthermore,
Data for the operating line are calculated because the concentration of the hydrocarbon in
from Equation (f) and are tabulated below and the gas stream is relatively high throughout most of
plotted in Figure 16.9. the column, the assumption that (1 — y) is equal to
(1 — y)im is n °t completely sound. The graphical

0.0 0.0307 0.1325 0.271 0.383 0.472 procedure of Baker might be adequate, but it
works best when the curvature of both operating
0.0124 0.02 0.05 0.10 0.15 0.20 line and equilibrium curve is slight. It would appear
that the best solution is the detailed evaluation of
On the same diagram
plotted the equilib- is the integral of Equation 16.27.
rium curve for Having now satisfied
this system. The following tabulation indicates the data
the material balance and taken into account the necessary for the graphical evaluation of N g-
equilibrium relationships, several procedures for Values of y and y* are read from Figure 16.9.
 MASS TRANSFER 397

(1 -y) lm
1-/* d-K)/m y -y
(1 -y)(y* -y)

0.0124 09876 1.0 0.9938 -0.0124 -81.0

0.03 0.003 0.97 0.997 0.9885 -0.027 -37.7
0.05 0.0117 0.95 0.9883 0.9692 -0.0383 -26.6
0.08 0.03 0.92 0.97 0.9450 -0.050 -20.6
0.10 0.0475 0.90 0.9525 0.9265 -0.0475 -21.7
0.14 0.0945 0.86 0.9055 0.8825 -0.0455 -22.6
0.15 0.106 0.85 0.893 0.8715 -0.044 -23.3
018 0.1445 0.82 0.8555 0.8352 -0.0355 -28.7
0.20 0.1730 0.80 0.8270 0.8135 -0.0270 -37.6

The column in the above tabulation is plotted

last

against y in Figure 16.10, and the area under the

curve between the limits of y^ and y 2 gives the
Nog- Since y 2 is less than yy the integral is ,

positive. In this case, A/ 0G =4.91. Note that the

quantities (1 — y) /m and (1 — y) are nearly equal.
For dilute concentrations this assumption of equal-
ity is usually valid.
The integration can be performed numerically
using the trapezoidal rule, which may be applied to
the tabulated data above even though the y inter-
vals are not equal. For example, the area of the
002 004 0.06 008 10 012 0.14 0.16 018 020 22
interval from 0.0124 to 0.03 is
y

(0.0124 - 0.03)[(-81.0) + (-37.7)] 1/2 = 1.045 Figure 16.10. Graphical integration for
Example 16.1.
The sum of the intervals gives 5.6 for Nqg This
all •

value is higher than that obtained by the graphical unit (H) can be defined in various ways to cor-
method because of the tendency of the trapezoidal respond to the specific expression for the number
rule to overestimate the integral. of transfer units (A/), as shown in Table 16.2. All
It is instructive to use Equation 16.31 to these definitions require knowledge of the rate of
evaluate N Q g With values from the table above mass transfer as expressed by the mass-transfer
coefficient,which is a function of many variables.
0.0124-0.20 Mass-transfer coefficients for packed towers cannot
A/ OG = 10.0
(0.173-0.20)- (0-0.0124) be predicted theoretically, because of the complex
flow patterns of the two phases. Experimental data
(0.173-0.20)
In
must be obtained for each type of packing, because
(0-0.0124) the packing itself plays an important role in deter-
mining phase flow patterns.
This answer differs sharply from that found in Mass-transfer data are reported in the litera-
the more rigorous graphical integration because ture either in terms of the mass-transfer coefficient
Equation 16.31 was based upon the assumption of or in terms of the height of a transfer unit. A great
both phases being dilute. In this case, the liquid quantity of information has been published, and
phase is concentrated as it leaves. This results in a severalattempts have been made to correlate the
nonlinear operating line and nonlinear equilibrium published data, with only limited success.
data in this concentrated region. Several factors help explain the inconsistency
often encountered in mass-transfer data in packed
Evaluation of the Height of a Transfer Unit columns. Most of the data were taken in laboratory
To complete the determination of the height of a columns of small diameter. In such cases, the
packed tower, it is necessary to evaluate the height inability of the packing to fit as snugly against the
of a transfer unit, once the number of transfer wall as it does against itself in the middle of the
units has been calculated. The height of a transfer column can lead to preferential flow of the phases
398 .APPLICATIONS TO EQUIPMENT DESIGN

along the wall, thereby giving incomplete contact phase are typically available for specific systems
between phases. This channeling can be a problem where the resistance to mass transfer is predomi-
whenever the packing size is a significant fraction nantly in one phase. The available data are examined
of the column diameter. In large-diameter indus- in the following paragraphs.
trial columns, it is extremely important to have the For gas absorption and desorption, consider-
entering liquid uniformly distributed across the able data have been collected for gas- and liquid-
packing at the top of the column. Maldistribution phase mass transfer coefficients and heights of
of liquid can severely reduce the mass-transfer rate, transfer units. Data on distillation are more
because the two phases are not adequately con- limited. Even though the transfer coefficients are
tacted. If the packing is not carefully installed, defined differently for absorption and distillation
there may be variations in void volume, which will (Table 16.2, k Y 'a versus k y a), in the absence of
encourage channeling or bypassing by the phases. data for a specific distillation, the generalized cor-
Mass-transfer data are available for both the relations for gas absorption may be used for distil-
L-phase and the V-phase. As shown in Chapter 14, lation. Reference 12 presents an extensive sum-
the individual mass-transfer coefficients for the mary of data and correlations available for gas
two phases can be related to an overall mass- absorption and distillation. References 5 and 11
transfer coefficient. develop- generalized correlations using extensive
1 1 m data from various sources.
(16.42) Mass-transfer coefficients and heights of
Ku kx
transfer units depend upon phase flow rates to a
As shown earlier in this chapter, convenient to it is greater or lesser extent. If the phase flow rates vary
combine the interfacial area per volume with unit significantly within a column, then the coefficients
the mass transfer coefficient, because it is difficult would also vary. In many applications of gas
to measure them separately. Therefore, Equation absorption, the phase flow rates are sufficiently
16.42 may be modified. constant to use an average value for calculating
mass-transfer coefficients or heights of transfer
1
=
1
+
m units. In distillation, on the other hand, phase flow
(16.43)
l\yd Kyd KyO rates mayvary. In particular, the flow rates often
change the feed point, so different values must
at
This expression can be further modified to relate it
be used to calculate coefficients in the enriching
to the expressions for heights of transfer units
section and in the stripping section. Furthermore,
found in Table 16.2.
the model of constant molar overflow implies a
V V varying mass flow rate if there is a significant
K Y aC\ -y),m k Y aC\ -y) /m difference in the molecular weights of the two
components. In situations where varying flow rates
mV L(1-x) /m 1 1
cause variation in mass-transfer coefficients, it may
! 16.50)
(1 -x),m (1 -y) lm kx a be necessary to leave the coefficient within the
integral of Equation 16.23, and perform a point-
Using the definitions found in that table yields to-point integration graphically or numerically.
mV An increase in liquid rate typically causes L H
Hog ~ Hc + ~T~ HL (16.44) to increase, whereas a change in gas rate has no
effect (up to the loading point). The height of
because the ratio (1 — x) /m /(1 — y), m in the right packing apparently has a small effect on H L ,

term approximates unity for the most commonly because of the increasing maldistribution of phase
encountered mass-transfer operations. Equation flows as height increases. The column diameter has
16.44 relates H 0G to mVIL as a straight line of no effect on H L As the liquid Schmidt number
.

slope equal to H L and intercept of H G If H 0G . increases, H L increases, as would be expected from

were expressed as a function of mVIL, Equation diffusion theory.
16.44 could be used to determine H G and H Ll the The height of a gas phase transfer unit (H G )

relative magnitudes of resistances encountered in
the transfer process. Unfortunately, this is not so
simple because a constant value of is not usually
encountered outside the range of dilute solutions.
Data for heights of transfer units in either
decreases as liquid rate increases, but varies little
with gas rate (except at very low gas rates, where
m H G approaches zero as the gas rate approaches
zero). H G increases as tower height increases, but is
insensitive to column diameter.
 MASS TRANSFER 399

The correlations of Onda, et al. (11), are a v = packing surface area per unit volume
presented below. These correlations are based upon of packing, ft 2/ft 3 (m 2/m 3 ) (Table 16.1)
2 3
a large quantity of data for a variety of packings a w = surface area of wetted packing, ft /ft

with gas absorption and desorption using water and (m 2 /m 3 )

organic solvents. The gas phase correlation also Ci = dimensionless constant. Cy = 5.23 for
includes data on the vaporization of pure liquids. packing larger than 1/2 in.; C-\ = 2.0
Most of the experimental data fall within ±20 for packing less than 1/2 in.
percent of the value calculated from the correla- QL = liquid-phase diffusion coefficient,
2
tion, but some data deviate as much as ±50 per- ft /hr (m 2 /s)
cent. Such uncertainties are not unusual in mass- Dp - nominal packing size, ft (m)
2
transfer correlations. They may be due to Qv = gas-phase diffusion coefficient, ft /hr
2
inaccurate data or to an inadequate correlation. (m /s)
8
These correlations include data on the follow- g = acceleration of gravity, 4.17 x 10 ft/
2
ing packings: hr 2 (9.81 m/s )

2
G L = liquid-phase mass velocity, Ib/hr ft
Raschig rings: { to 2 in. (kg/s m 2
)

Berl saddles: \ to \\ in. 2

Gv = gas-phase mass velocity, Ib/hr ft
2
Pall rings. 1 in.
(kg/s m )

Spheres: \ and 1 in.

k G = individual gas-phase mass transfer co-
Rods: \ and 1 in.
efficient, lb moles/hr ft
2
atm [mol/s
The correlations are in terms of mass-transfer co- m2 (N/m 2 )]
efficients, which can be easily converted to heights k L = individual liquid-phase mass transfer
2
coefficient, lb moles/hr ft (lb moles/
of transfer units, HG or HL , as necessary. These
workers developed a method for determining the ft
3
) [mol/s m2 (mol/m 3 )]
actual interfacial area between phases in the wetted HL = liquid viscosity, lb/ft hr (kg/m s)

packing, so that k can be calculated directly. Hv = gas viscosity, lb/ft hr (kg/m s)

3 3
Liquid phase: p L = liquid density, lb/ft (kg/m )

3 3
p v = gas density, lb/ft (kg/m
— o = liquid surface tension, dynes/cm
oc = critical surface tension of packing
)

material, dynes/cm (Table 16.3)

2/3 -1/2
= 0.0051
(
—GlM
\a w n L
\

i
PM
/

W»i/
Ml
(a v D p )° The range and scatter of the data upon which
Equations 16.45 and 16.46 are based are shown
(16.45) graphically in Reference 12 as well as Reference
Gas phase: 11. The range of data for k L extended from a
liquid Reynolds number (G L la w \x L of 4 to 400 for

pM
)

kG RT n 0.7 ., 1/3
water and 1 to 40 for organic absorbents. For kG
mC,lJ*.\ (.,0,)-" ,

a v-* I
avn v Pv2 the gas Reynolds number {G v la v yL V ranged from )

5 to 1000.
(16.46)

The expression for a w , the wetted surface of the

packing, is
Table 16.3 CRITICAL SURFACE TENSION OF PACK-
ING MATERIALS
0.75 oc dynes/cm
,

= 1
«pf-1.46p|-|^|
\o! La p,' y
v
Carbon 56
2 -0.05 n 2
GL 2 Ceramic 61
av
(—
-
v

16.47) Glass 73
Pl 9
2 )
Paraffin 20
Polyethylene 33
Each term Equations 16.45, 16.46, and 16.47 is
in
Polyvinylchloride 40
dimensionless, so any consistent set of units may Steel 75
be used. Typical English units are given below with
the definition of terms. (SI units are given in paren- (With permission of J. Chem. Eng of Japan; Copyright
theses) 1967)
400 APPLICATIONS TO EQUIPMENT DESIGN

Example 16.2. Calculate HG and HL for the 210/29

c
absorption of S0 2 from air into water at 87 F and
2 0.389 x 37.3 x 1
1 atm, for the conditions: G v = 210 Ib/hr ft ,

2 2 = 0.50
G L =8205 Ib/hr ft 5=4.76 ft 1-in. ceramic ; ;
ft

Raschig rings
where M for air is approximately 29
Solution. For S0 2 in air at 87° F and 1 atm,
H G = 0.044 lb/ft hr; p G = 0.0726 lb/ft
3
9G = ; HL = with (1 — x), m = 1 for dilute liquids
2 k x aS
0.50 ft /hr (Equation 9.52); and R = 0.7302 atm
ft /lb mole °R; 7~= 87 + 459 = 546° R. From Table
3

2
16.1, a v = 58 ft /ft
3
Substituting numerical values
.
since G L IML = LIS and k x = k L plML
k L ap L
in Equation 16.46 yields

58x0.50 210 07 8205

,
, ,

*G = 5.23( 2.00 x 37.3 x 62.2

0.7302 x 546 '58x0.044'
1/3 = 1.77 ft
0.044
(58 x 1/12)" 20
t

"0.072x0.50 Pressure Drop in Packed Columns

= 0.379 2
lb moles/hr ft atm In a packed column, the liquid entering at the top
C normally flows downward through the column,
For S0 2 in water
1 atm, p L = 2.42
at 87 F and driven by the gravitational force. In gas absorption,
lb/ft hr;p L =62.2^/^; L
= 6.58 x 10- 5 ft 2 /hr; -<
the gas entering at the bottom must be moved by a
and o = 71 dynes/cm = 2,027,000 Ib/hr 2 For this .

blower or fan. In distillation, the reboiler creates a

packing, o c = 61 dynes/cm (Table 16.3). The
pressure that causes an upflow of vapor. In order
wetted surface must be evaluated before k L can be
to maintain the upward flow of gas or vapor, the
determined.
pressure at the top of the column must be less than
0.75 oont; 0.1 at the bottom. This pressure drop is an important
61 8205
= 1 — exp -1.45(^1 I factor in the design of packed columns. Because
71 58 x 2.42
the liquid downflow occupies the same channels as
2 0.05
8205 x 58 the gas upflow, the pressure drop is really a func-
«! tion of both flows, as shown in Figure 16.1 1.
62.2 2 x4.17 x 10 8
'

Flow of the gas through the tower packing is

8205 2 usually turbulent; consequently, the slope of curve
*'
62.2x2027000x58 i"l C is about 2, as would be predicted by a turbulent-

flow pressure-drop equation. For a constant gas

= 0.643
velocity, the pressure drop increases with increas-
So
ing liquid rate as can be observed from line G in
2 3
a w = 0.643 x 58 = 37.3 ft /ft

/L 21
Then Equation 16.45 may be solved for k L .

2/3
2.42 x4.17 x10 s 8205
--
-
k L = 0.0051 (-
-*
_
62.4 37.3x2.42 Q. ^2 n /
O
1/2 t
2.42
a, G / / ^/£°
*(;62.4 x 6.58 x 10" 5 '
< 1 1 "/$
.-.

s
/l
/
h v/^
/a
= 2.00 lb moles/hr ft
2
(lb mole/ft 3 — /cv
)
•

To find the heights of the transfer units, reference .-

.-
i.

is made to the definitions in Tables 13.3 and 16.2:

- "<• / 5*7
-
_ >/
V ±1
Hr ,
= with (1 - y), m = 1 for dilute gases o/
k y aS /
Log gas velocity

——
GylMy
—since G V IM V
knar
= VIS and k y = k G P Figure 16.11. Pressure-drop
packed towers.
characteristics in
 MASS TRANSFER 401

Figure 16.11. Each type of packing material has a gas must bubble through it. This condition is some-
fixed void volume for liquid passage so that as the times called "visual flooding." At a gas rate
liquid rate increases, the voids fill with liquid, thus somewhat than that corresponding to
greater
reducing the cross-sectional area available for gas "visual flooding," a second change in the slope of
flow. the pressure-drop line occurs (point 2). This point
The nature of random packing is such that a is more reproducible than the visual observation

myriad of expansion and contraction losses and and is generally taken as the flooding point. At this
considerable turbulence are created by the flow of point, the drag force of the gas bubbling through
the two fluids around the individual solid packing the liquid is quite important. Each liquid rate has
elements. The pressure drop is a combination of its own loading and flooding points.

skin friction and form drag with form drag pre- Operation of packed towers is not practical
dominant at the higher velocities. By the analogy above the loading point. Colburn (4) presents
equations of Chapter 13, there is a direct relation- methods for evaluation of the optimum gas veloc-
ship between heat- and mass-transfer surface ity through a packed column. For simplicity and
coefficients and skin friction, so that it is advan- safety packed towers are designed using gas veloc-
tageous to have a high percentage of the total ities of about 50—75 percent of the flooding
pressure drop attributable to skin friction. It has velocity at the expected liquid rate. This design
been estimated (2) thaf not over 10 percent of the will normally ensure a stable operating condition
pressure drop is the result of skin friction. somewhere below loading and should proyide for
In Figure 16.1 1, a line of constant liquid rate thorough wetting of the packing surface.
(lineA) can be examined as if the flow patterns Numerous investigators have determined
were being viewed through the transparent wall of flooding velocities for a variety of fluids and pack-
a packed column. Up to point 1 on this curve, the ing materials. A generalized correlation for flood-
pressure-drop characteristic is quite similar to that ing and pressure drop for flow rates less than
of the dry packing (C). The slope of this portion of flooding is presented in Figure 16.12. The ordinate
the line is about the same as that of the dry curve, of that figure includes a packing factor F that is

but the pressure drop is greater. The larger pressure determined by fitting drop data for a
pressure
2
drop is due to blocking of part of the voids by the specific packing. At one time the ratio a v le was
liquid and roughening of the surfaces by waves. used in place of F, but this quantity was found to
Visual observation would indicate orderly trickling vary, depending upon how the packing was put
of the liquid downward through the packing with into the tower. Such variations in packing methods
no observable liquid buildup. At point 1, a change still present a problem, so the correlation of Figure

in slope occurs, indicating the pressure drop 16.12 should be regarded as giving estimates of the
increases more rapidly with an increase in gas veloc- pressure drop that might actually be encountered.
ity. This point may not be distinct enough to Values of F are found in Table 16.1.
allow observation of any change in the flow pat-
tern or characteristics. Perhaps it might be possible Example 16.3. A gas mixture containing 6.0
to observe an increase in the quantity of liquid percent S0 2 and 94.0 percent dry air is to be
retained in the packed section. The retained liquid scrubbed with fresh water in a tower packed with
is referred toholdup. Point 1 is called the
as 1 -in. ceramic Raschig rings to remove the S0 2 so
loading point, and the increased dependence of the that the exit gas will contain no more than 0.1
pressure drop on the gas velocity is a consequence mole percent S0 2 The tower must treat 1000
.

of drag between the phases. As point 1 is passed, Ib/hr of gas and is to be designed using 50 percent
visual observation shows a greater amount of liquid of flooding velocity. The water flow is to be twice
holdup. Perhaps observation will show a liquid the minimum required to achieve this separation in

layer at the top of the packing and a gradual filling a tower of infinite height. Operating conditions
of the packing voids with liquid. The liquid now will be isothermal at 87° F and 1 atm pressure.
has filled a large portion of the packing, and the Determine the required tower diameter and height.

EQUILIBRIUM DATA AT 87°F

PSO mm ^9 0.6 1.7 4.7 8.1 11.8 19.7 36.0 52.0 79.0
c, gS0 2
per 100 g H 2 0.02 0.05 0.10 015 020 0.30 0.50 0.70 1.00
Density of solution
3 6216 62.17 62.19 62.21 62.22 62.25 62.32 62.38 62.47
lb/ft
402 APPLICATIONS TO EQUIPMENT DESIGN

0.60 1

1 1

Generalized Pressure Drop

0.40
Correlatior i

i i

Paramet€ r of curves is pressure

drop in ir ches of water foot
0.20
of packec height; parenthetical
values ar i Pascals meter
J^50 ^-:
0.10
5;"^
0.060
v^soVaosT
0.040 ~0.25
&04)
s
^ 0.020
k, >1 >K 1 \
_0^£_ J82)
sj rvO*
0.010

0.006 0.05" ~ -'

)

0.004

0.002

0.001
0.01 0.02 0.04 0.06 0.1 0.2 0.4 0.6 1.0 2.0 4.0 6.0 10.0
G
Gl pG y
T
i

G\
'V W~ PL Pal
PC'

Figure16.12. Flooding velocities and pressure drop in packed towers (14).

(Courtesy of The Norton Co.) .

G = mass velocity of gas, lb/sec, ft j t * -4.n*»o \\lhr"*

L = mass velocity of liquid, lb/sec, ft

3
PC = U qQib density, lb/ft
Pl_
= liquid density, lb/ft

mX= liquid viscosity, centistokes (.CP^

C= conversion factor: 1 in English and 10.764 in SI units
F = packing factor (Table 16.1)
AP in lb/ft 2 = 5.2 x in. of water

Solution the gas leaving the tower will be y 2 = 0.001, and

Equilibrium Curve. First the equilibrium the concentration of the liquid entering will be
data are converted to mole fraction units and x 2 =0. The operating line equation may be
plotted. written, in terms of solute-free streams, as

V7 *2 *1
y 0.00079 0.00224 0.0062 0.0107 0.01566 V Li -y 2 1-yJ 1 -X 2 1 -*1
X 0.0000565 0.000141 0.000281 0.000422 0.000564

y 0.0259 0.0474 0.0684 0.104 Now =0.001, and x 2 =0, and

/! =0.06, y 2
the minimum flow will correspond to a line
liquid
X 0.000844 0.00141 0.00197 0.0028
through (x 2 ,y 2 and {x^*,y^ since the equilib-
)

where rium line is concave toward the operating line.

0.18 From Figure 16.13 X!* =0.001740. Also
~64~
Pso,
y
= and x = 1000
760 0.18 Vy =
1 + c (0.06x64) + (0.94x29)
64
= 32.1 lb moles/hr total gas flow
Mass Balance. Before the mass balance can
be completed, it will be necessary to determine the W = IM1 -/1 )=32.1(0.94)
minimum liquid flow rate. The concentration of = 30.1 lb moles/hr S0 2 -free gas
 MASS TRANSFER 403

phase transfer units, using Equations 16.45 and

16.46. It is not clear from the data to this point
whether one or another phase is controlling, so no
assumption can be made. To obtain /r-values, it is
necessary to know mass velocities, but these
require knowledge of cross-sectional area of the
column, as well as mass flow rates. The cross-
sectional area of the column is determined from
the correlation of Figure 16.12, using the specifica-
Figure 16.13. Operating line and equilibrium curve tion of 50 percent of flooding velocity for the gas.
for Example 16.3.
First, total flow rates must be determined.

Therefore, the minimum water rate is V\ = 1000 Ib/hr = 32.1 lb moles/hr

S0 2 entering = 32.1 x 0.06 x 64 = 123.2 Ib/hr

30.1
0.001 0.06
W(o- 0.00174
Is
L' = 1085
0.94/

lb
0.99826
'
V

moles/hr S0 2 -free
'

water
S0 12 a =
leaving
——
0.001

0.999
x 30.1 x 64 = 1 .93 Ib/hr

'

For a design water rate of 2Z. min =123.2

Therefore, S0 2 absorbed by the water
L' = 2 x 1085 = 2170 lb moles/hr 1.93= 120.3 Ib/hr.

The design flow rate of L' = 2170 lb moles/hr Fresh water entering = 2170 x 18 = 39,000 Ib/hr
can now be used in the equation for the operating
Total liquid leaving = 39,120 Ib/hr
line to determine the exit-liquid composition.
Assume the density of the gas stream to be essenti-
,0.001 0.06
30.1 =2170
(=2170(0-^-)
1
ally that of air at 1 atm and 30°C.
\0.999 0.94
29 273
x, = 0.000880 = x = 0.0726 lb/ft
3
pv
359 303
All terminal compositions and flow rates are
3
now established. p L = 62.2 lb/ft

Xt = 0.000880 y, = 0.06 GL _ 39,000 / 0.0726

x 2 = 0.0
fpv = 1.33
y 2 = 0.001 G„v ~ 1000 v 62.2
L' = 2170 lb moles/hr V' = 30.1 lb moles/hr pL

The entire operating line may now be From Figure 16.12 at a value of
evaluated.

30.1
1
0.06
0.94
)
=2170(
0.00088
0.99912*
>

^ — / =1.33

Concentration of the liquid phase is dilute

enough so 1 — x) = therefore = 0.023
( 1 ;

9cPvPl
x = 0.01 388 0.00000655 From Table 16.1, F - 1 55. Since p w = p L , there-
fore
Neglecting the last term, the operating line becomes 2 01
G l/ (155)(0.77)
= 0.023
y 0.06 0.04 0.03 0.02 0.01 0.001
4.17 x 10 8 x (0.0726)(62.2)
x 0.00088 0.000591 0.000430 0.000283 0.000141 -0
2
G v =420 Ib/hr ft
These data are plotted in Figure 16.13, and
the operating line is found to be substantially Specifications are to use a velocity 50 percent
linear. of flooding; therefore,
Heights of Transfer Units. It is now neces-
3
sary to calculate the heights of the gas and liquid Gv = 482 Ib/hr ft
404 APPLICATIONS TO EQUIPMENT DESIGN

Since the gas flow at the bottom of the tower used to calculate N 0G
is on this flow rate, or
greatest, base the area Vi ~ V^
N oc
v 1000
2
(/*-K) 1m
S= = 2.07 ft
From the
Gv 482 least-squares equation for y,y^* = 0.022,
and y 2 * = 0, so that
The tower diameter is therefore
0.001 - 0.06
0G ~ - -
4(2.07! (0 -0.001 ) (0.022 0.06)
D = = 1.62 ft
(0-0.001)
In
Consequently (0.022 - 0.06)
= 5.80
= 482lb/hrft 2
Gv
The height of the packed section of the column is
(2170H18) then
G, = = 18840 Ib/hrft 2
2.07 z = Ho G No G
Now that the mass velocities of the two = 1.85x5.8
phases have been calculated, it is possible to calcu- = 10.73 ft
late the heights of the transfer units. A method is
The total pressure drop for the column is calcu-
shown in Example 16.2, and similarly we determine
lated using Gv
= 482 and the pressure drop curves
of Figure 16.12. For the abscissa
Hc = 0.555 ft

HL = .97 ft
Gj^ /p^_18840 /0.0726
1
= 1.33
These values are functions of the flow rates
482 V 62.2

through the column, and in general the flow rate The value of the ordinate is

varies from the top to the bottom. In this case, 2

G v 2 Fn° - 1
(482) (155)(0.77)
- 1

however, both phases are dilute, and the change in

flow rates is small. In fact, an examination of QcPvPl (4.17x 108 )(0.0726)(62.2)
Figure 16.13 shows that both the operating line = 0.0186
and the equilibrium curve are essentially linear, and
These values for the ordinate and abscissa give a
therefore simplified calculation techniques can be
pressure drop of approximately 0.6 in H 2 ft
used (both phases are dilute).
packing, on Figure 16.12. The pressure drop for
The equilibrium curve is nearly linear, so it
the tower is therefore about 16 in. of water.
could be fitted by a least-squares equation

y = 25. 7x - 0.00063 Liquid Extraction

Liquid extraction is carried out in packed towers

This equation is fitted to the first four points in using the same packing
as for gas absorption. But
the table above, plus the point (0,0), since most of the flow patterns are substantially different
the transfer units will occur at the dilute end. The because the physical properties of the two liquids
above equation states that m
= 25.7 in the expres- are less different than those of a gas and a liquid.
sion y = mx for the range of data in the tower Care must be taken to disperse one liquid phase as
under consideration. Therefore, Equation 16.44 small droplets in the other (continuous) liquid phase.
can be used to calculate H c The continuous phase should be the one that wets
the packing, because otherwise the dispersed phase
mV mGw
Hog = Hg +~H L
l =H G +-—H
b
L
droplets will coalesce on the packing and reduce
the interfacial area for mass transfer. Of course, the
L
two phases must be relatively insoluble in each
25.7 x 482
0.555 +- -(1.97) other. Typically, one is an aqueous solution and
18840 the other contains an organic solvent. An aqueous
= 1.850 ft phase will preferentially wet ceramic or metal
packing, whereas an organic phase will wet plastic
With dilute solutions, Equation 16.41 can be or carbon.

Although the design methods covered earlier
in this chapter can be used for liquid extraction in
packed columns, the practical application is quite
different from that for gas absorption or distilla-
tion. Correlations predicting the flooding point are
available (10); but no generalized correlation exists
for mass-transfer coefficients, because of the many
factors that influence the rate of mass transfer in
liquid extraction in packed towers. These factors
include packing size, shape, and material, tower
diameter and packing depth, flow rates and
physical properties of both liquids, which phase is
dispersed, and the direction of mass transfer. The
height of transfer unit for the dispersed phase does
not vary with phase flow rates, but the H for the
continuous phase has been found to vary with the
ratio of the flow rates of the two phases.
In the absence of a generalized correlation or
specific data for a new system, it is necessary to
gather experimental data on a pilot-plant scale
before designing an industrial packed column for
liquid extraction. Available data and correlations
for specific systems are summarized in References
12 and 14.
Often the data available are for relatively
small laboratory columns, so that they are difficult
to scale up to larger industrial columns. An
example of the data available is that of Leibson
and Beckman (8), who studied the extraction of
diethylamine from a continuous water phase into a
dispersed toluene phase using a 6-in. diameter
column packed with unglazed porcelain Raschig
rings. The overall continuous-phase height of a
transfer unit (H c) was found to be a linear func-
tion of the ratio of the velocities of the two phases.
Vc
H oc =a +b (16.49)
mVD
where
H c
= height of the overall continuous-phase
transfer unit, ft
Vc
= superficial velocity of the continuous
water phase, based upon the empty
tower cross-sectional area, ft/hr
VD = superficial velocity of the dispersed
toluene phase, based upon the empty
tower cross-sectional area, ft/hr
m = slope of the equilibrium curve, = m
dy* Idx (for this system, m = 0.865)
a, b = empirical constants for a particular
packing size
For 1/2-in. packing, a = 3.8 and b = 2; whereas for
1 -in. packing, a = 8.2 and b = 2; so larger packing is
less effective.
MASS TRANSFER 405
Spray Towers
A spray tower can be thought of as a packed tower
without the packing. It is simply a cylindrical shell
with the /.-phase at the top of the tower dispersed
by a nozzle, or other dispersing device, into the
upward-flowing V-phase. It is sometimes used in
the absorption of a very soluble gas or in situations
where pressure drop must be low. It has that
advantage, but axial backmixing of both phases
reduces the effective mass transfer, even when
operation is nominally countercurrent. The liquid
must be widely dispersed to cover all the cross-
sectional area at the top of the column, but with-
out significantly impinging on the wall, where
droplets coalesce and run down the wall, effec-
tively reducing the interfacial area for mass
transfer. Adequate dispersion of the liquid through
the spray nozzle requires a substantial liquid pres-
sure.
Spray towers have been used for liquid extrac-
tion, but the problems of axial backmixing are
severe. Typically, only one or two transfer units
can be achieved, regardless of tower height. As a
result, spray towers are not often used for extrac-
tion in industry.
Spray towers are primarily used in the drying
of liquid slurries and solutions. This application,
which involves both mass transfer and heat
transfer, will be covered in Chapter 18.
Related to the spray tower are the cyclone
scrubber and the Venturi scrubber. The former
disperses a liquid absorbent into a gas stream that
is moving in a spiral path. The liquid droplets are
thrown to the outside wall by centrifugal force,
where they coalesce and run down to a collector.
The scrubbed gas exits by a central duct, essen-
tially free of liquid droplets. The Venturi scrubber
disperses a liquid absorbent into a high-speed gas
moving through a narrow constriction in a flow
duct, thereby obtaining a fine dispersion and large
interfacial transfer area. The resulting mixture is
then separated in a settling tower or cyclone. Both
the cyclone scrubber and the Venturi scrubber are
used to collect aerosols and dusts, as well as for gas
absorption. They require considerable energy for
gas and liquid flow compared to other contacting
devices.
CYCLIC TRANSFER
The mass
unit operations involving the transfer of
between and a solid phase include adsorp-
a fluid
tion, ion exchange, drying, and leaching. Adsorp-
tion and ion exchange typically involve the transfer
406 APPLICATIONS TO EQUIPMENT DESIGN

of a solute from the fluid phase to the surface of Although the equipmentused for continuous
the solid. In drying of a solid, the adhering liquid is solid-fluid operations not as simple as a packed
is

transferred to a gas phase. This process involves tower, the same basic fundamentals apply to these
both heat and mass and it will be
transfer, operations as to the operations of absorption and
examined in Chapter Leaching involves the
18. distillation. An example of continuous counter-

dissolving of a component of a solid phase by a current ion exchange is shown in Figure 16.14.
liquid solvent. Continuous countercurrent adsorption is used
Continuous contacting of a solid phase with a industrially to separate mixtures of light hydro-
liquid or gas phase presents serious problems in carbon gases (14). Generally, however, the
equipment design. The continuous countercurrent problems associated with the flow of a solid phase
movement of solids and fluids calls for specially discourage the use of continuous-contact equip-
designed equipment to allow for the steady intro- ment for solid-fluid contact. Instead, the solid is
duction of solids in one end of the apparatus and maintained in fixed beds, with the fluid flowing
their removal at the other end. This must be through the bed.
accomplished with little loss of the fluid phase. Fixed beds are widely used for adsorption and
ion exchange. A fixed bed may superficially
resemble a packed tower, but there is a funda-
mental difference. The packing in a packed tower
is present simply to facilitate the development of a

large interfacial area between the two fluid phases

involved in the mass transfer. The solid particles in
a fixed bed are in fact one of the phases actually
involved in the mass transfer. As the solute is
transferred from the fluid phase to the surface of
the adsorbent or ion-exchange resin, the concentra-
tion of solute on the solid increases as time passes.
Ultimately, the solid reaches a saturation point
where it can no longer accommodate additional
solute. At that point the fixed bed must be
regenerated. Because the concentration of solute
changes with time at various points in the fixed
bed, the process is unsteady state, and the design
methods developed for packed towers are not
applicable. Before examining design methods for
fixed beds, the basic mechanisms of adsorption and
ion exchange will be considered.

Adsorption
Adsorption, as the term is used here, applies to the
physical transfer of a solute in a gas or liquid to a
solid surfacewhere the solute is held as a result of
intermolecular attraction with the solid molecules.
The adsorbed solute does not dissolve in the solid
Figure 16.14. Continuous countercurrent ion exchange.
but remains on the solid surface or in the pores of
The units pictured process brackish water (1500 ppm sod-
the solid. The adsorption process is often reversible
ium) to produce boiler feed water at a chemical plant. The
so that by changing pressure or temperature the
ion-exchange resin flows clockwise around the loop in the
solute may be readily removed from the solid. At
right-hand unit pictured. The flow is intermittent, with
adsorption, regeneration, backwash, and rinse taking place equilibrium the adsorbed solute has a partial pres-
when the resin is stationary. The adsorption takes place in sure equal to that of the contacting fluid phase,
the unit immediately in front of the worker at the middle and by simply changing pressure or temperature of
levelon the right. The wash section is the tAbend at the the operation the solute may be removed from the
top, and the regeneration, using nitric acid and ammonia, solid.
takes place the L/-bend at the bottom. {Courtesy Chemi-
in The selection of adsorbents is quite impor-
cal Separations Corporation. tant. The solids should offer low pressure-drop
 MASS TRANSFER 407

characteristics, and good strength to withstand the
rigors of In addition, adsorbents are
handling.
selective in their ability to adsorb specific solutes.
Consequently, the chemical nature of the solid
must be considered to assure satisfactory perform-
ance. Commercial adsorbents include bentonite,
bauxite, alumina, bone charcoal,
carbon, and silica gel.
Equilibrium data for adsorption are usually
presented in the form of adsorption isotherms (see
ions. Anion-exchange resins may have fixed ionic
groups of primary, secondary, or tertiary amines,
phosphonium, and arsonium. Many types of
exchange materials in various physical states have
been synthesized. For example, spherical beads,
membranes, fibers, and liquids have been prepared.
fuller's earth, The "zeolites" are naturally occurring clays that
are capable of exchanging cations.
Typical of the operation that occurs during
ion exchange is that illustrated by water softening.

Chapter 3). These data are necessary before the The reaction is very simply indicated by the
design equation can be applied. In addition, know- following equation, in which [RS0 3 ]~ represents
ledge of adsorption transfer coefficients is the resin that has as a component a fixed ionic
necessary. These coefficients are usually available group. In this case, it is the sulfonic group, and the
only for specific conditions rather than as general resin is said to be a cation type.
correlations. Design of equipment, therefore, must
2[RS0 3 ]-Na + +Ca ++ %(RS0 3 ) 2 -Ca + + 2Na +
be made using data that closely parallel design
conditions. (solid) (solution) (solid) (solution)

Adsorption may be used where the fluid

In this process, the hardness ions, such as
phase is either a liquid or a gas. Liquid applications ++ ++
Ca and Mg are removed from water by pass-
,

include decolorizing of food materials such as

ing the water through a bed of resin particles. Since
vegetable oils and sugar syrups, drying of gasoline,
the reaction is reversible, the resin may be returned
and purifying liquid wastes for water pollution +
to the Na form simply by passing a strong salt
control. Gas treatments include recovering of sol-
solution over the resin. In this manner, it is ready
vent vapors from paint drying and dry cleaning
for reuse in the next softening cycle.
operations, drying gases, and removing toxic
Water can also be deionized by ion exchange
components of waste gases for air pollution con-
so that nearly all dissolved solids are removed.
trol.
Deionization requires that the water first be passed
through a cation exchanger and then an anion
Ion Exchange
exchanger. The reactions are
Ion exchange can be thought of as a special case
of adsorption, where a specially prepared ion- [RS0 3 ]-H + + Na + Cr - [RS0 3 ] ~Na + + H + Cr
exchange resin serves as the adsorbent. Ion
(solid) (solution) (solid) (solution)
exchange was originally applied primarily to water
treatment, but many other applications have and
developed. It is now also used for brackish-water
desalting, separation of rare earths, recovery of [RN(CH 3 ) 3 ]
+
OH- + H + CI- ^
metals from leaching solutions or waste streams, (solid) (solution)
and recovery of antibiotics from fermentation
+
broths. [RN(CH 3 ) 3 ] Cr + HOH
Itconvenient to think of an ion-exchange
is (solid) (solution)
resin as a homogeneous gel throughout which is
distributed a network of hydrocarbon chains. The cation exchanger is regenerated usually with
Attached to these chains are ionic groups that are sulfuric acid, and the anion exchanger with sodium
immobile. Their ionic charges are balanced by hydroxide.
diffusible ions of opposite charge. For example, a The expression for phase equilibrium rela-
cation-exchange resin has incorporated into it tionships in ion exchange is similar in type to that
immobile anionic groups bonded to each other of adsorption isotherms. Another widely used
through cross linkages. The charges of these fixed method for ion-exchange equilibrium is based upon
anionic groups are balanced by diffusible cations. the mass-action relationships. For example, con-
It is possible to manufacture ion-exchange resins
sider the reaction

with many groups of fixed ions. Cation-exchange

[RS0 3 ]
" H + + Na + Cr ^ [RS0 3 ]~Na + + H
+
Cr
resins may contain such fixed ionic groups as
phenolic, carboxylic, sulfonic, and phosphonic (solid) (solution) (solid) (solution)
408 APPLICATIONS TO EQUIPMENT DESIGN
The equilibrium constant for this reaction
+ +
(cone. [RS0 3 -]Na )(conc. H Cr)
K equil
(cone. [RS0 3 ]-H + )(conc. Na + Cr)
For most engineering applications, this equilibrium
constant may be expressed as
1
•^ equil.
1-/a
where
x a = equivalent fraction of solute
liquid phase
ya = equivalent fraction of solute
resin phase
The rate of transfer in ion exchange is compli-
cated by the complex mechanism of the transfer
process. A postulated mechanism of ion exchange
consists of the following steps:
1. Diffusion of ions from the bulk
phase to the surface of the resin.
2. Diffusion of the ions from the surface of
the resin into the resin to the site of exchange.
3. Exchange of the ions at the active site.
4. Diffusion of the replaced ions from within
the resin to its surface.
5. Diffusion of the replaced ions from the
resin surface to the bulk liquid.
Any one of the five steps may be
limiting although too few data exist to
factor,
permit any generalizations. The rate for step 3 is
best described by a kinetic reaction rate, whereas
the rate for the diffusion steps may be described
by the usual rate equation for counter diffusion as
indicated by Equation 16.14. Once again, a suit-
able mass-transfer coefficient is needed. If the
exchange of ions (step 3) is very rapid compared to
the diffusion steps, then ion-exchange equipment
may be designed by employing Equation 16.11. If
the rate-controlling step is the exchange step, then
kinetic procedures are needed.
Analysis of Fixed-Bed Adsorption
Because of the several interacting physical mechan-
isms involved in adsorption or ion exchange, and
because such calculations are usually applied to a
finite quantity of adsorbent in a cyclic operation,
the design of fixed beds for adsorption can be an
involved procedure. Extensive summaries of the
principles and procedures are available in Refer-
ences 1 2 and 1 3. A simple procedure for the analy-
sis of fixed beds is presented in the following
paragraphs, as applied to certain limited situations.
is
Solute-rich
Absorbent
gas i
bed
•Effluent
V =
s cross- sectional
area of bed
V
Y2
s = Z = Zi z = z
Figure 16.15. Adsorption in a fixed bed.
A schematic representation of a fixed-bed
adsorber is given in Figure 16.15. For simplicity,
the system will be discussed in terms of adsorption
a in
of a gas, although the development also applies to
liquids and to ion exchange.
a in
As adsorbent
a solute-rich gas enters a fresh
bed, the solute the gas begins adsorbing on the
in
solid adsorbent. The rate at which adsorption
occurs is influenced by one or more of several
factors:
1. Diffusion of solute from the flowing gas
liquid to the external surface of the solid particle.
2. Diffusion of the solute into the gas-filled
internal pores of the particle.
3. Reaction at the gas-solid interface.
4. Diffusion of the adsorbed solute within
solvent phase.
If none of these steps are slow, it is possible to
imagine instantaneous and complete adsorption
upon the first increment of the bed as the first
a rate
increment of gas enters. The solid adsorbent will
immediately adsorb solute until it is in equilibrium
with the entering gas stream. Thereafter, the enter-
ing gas flows unchanged through this first incre-
ment, and onward to have its solute adsorbed in
the next increment. This action continues until
enough feed has been introduced to saturate the
entire bed to equilibrium with the gas, at which
time the effluent gas instantaneously increases
from zero concentration of solute to the concen-
tration of the entering stream.
During the time the adsorption front is pro-
gressing through the bed, the unadsorbed compo-
nent has been flowing through the remainder of
the bed and out as the effluent. If the concentra-
tion of solute in the effluent is measured, it will be
found to be zero until the time at which the bed
becomes saturated; then it suddenly increases to
the entering composition.
The time required to saturate the bed is a
function of the adsorptive capacity of the bed, the
flow rate, and the concentration of adsorbable
constituent in the feed. Accordingly, a direct pro-
portionality exists between the time required to
saturate the bed and the cumulative effluent mass,
 MASS TRANSFER 409

v = ves where
v,
3
p B = bulk density of adsorbent, lb/ft (or
kg/m 3 )

X-\ = mass solute/mass adsorbent in equi-

(a) librium with the entering gas (of com-
position Y-i)

In addition to the adsorbed solute, the unit volume

i'l
of bed retains solute in the gas phase, filling the
void space between the adsorbent particles. This
gas has a concentration Yy and a volume e for the
unit volume, where e is the fraction of void space,
a characteristic of the adsorbent. The solute in this
gas filling the void space is eY-\ . When
the front has
M. cumulative mass of effluent, with time moved a distance z, into the bed, the inflow of
Figure 16.16. Adsorption breakthrough curves, (a) For solute will equal the accumulation in the solid and

instantaneous adsorption, (b) For adsorption at a finite in the gas filling the void volume, since there is no
mass transfer rate. outflow.

V'Y^d=X^p B Sz +eY i y
Sz i
(16.50)
as shown in Figure 16.16a. With instantaneous
Only 6 and z, are variables, so the rate at which the
adsorption, the front between the fully saturated
front moves through the bed is
bed and the fresh adsorbent moves through the bed
until it reaches the efflux end, at which time a
dz V'Yy
concentration change from zero to y^ occurs as the 16.51)
last increment of adsorbent becomes saturated. dd S(X,p B +eYi)
This sudden increase of solute concentration in the
The time at which the bed is fully saturated (0J
effluent has been called breaking through the bed
occurs when the front has moved completely
leading to the designation breakthrough curves as
through the bed. Rearranging Equation 16.50 gives
the appropriate name of a plot of exit concentra-
tion as a function of time.
(16.52)
This graphical representation of exit con- V'Y,
centration with time is very similar to the graphical
representation of concentration at points through
The cumulative mass of effluent M is simply V"0at
the bed at a point in time during the adsorption any time, so at saturation of the entire bed
cycle. The similarity of these curves of concentra-
B + eY,)Sz
(X,p *—
tion as a function of different independent vari- M S
= V'6 S = (16.53)
ables invites carelessness in associating perfor- M
mance factors with one or the other of the variables
The foregoing development applies only to the
indicated. Confusion is the inevitable consequence hypothetical model of instantaneous adsorption.
of such carelessness. In any actual adsorption, the
rate of mass
The which the adsorption front moves
rate at transfer is the solute is not adsorbed
finite, so that
through the bed can be related to system variables. instantaneously. As a result, the gas carries some
The rate of flow of solute into the bed is Y V y ,
solute forward into the bed while other solute is
where being adsorbed; therefore, some solute can pass
beyond an adsorbent region even before the region
V' = flow rate of gas on a solute-free basis,
is saturated. As saturation is approached, the rate
Ib/hr (or kg/s)
= mass solute/mass solute-free gas at bed of adsorption slows. These two effects produce a
V-,
blurring of the planar adsorption front of
entrance
instantaneous thereby creating an
adsorption,
Each unit volume of bed contains a mass p B of adsorption zone of finite width. At the leading
adsorbent that is able to adsorb a mass p B Xy of boundary of the adsorption zone, the advance
solute when it reaches equilibrium with the enter- solute is being absorbed, while at the back of the
ing gas phase zone the bed is just reaching saturation. The result-
410 APPLICATIONS TO EQUIPMENT DESIGN

ing effluent concentration reflects this variation in of gas through the bed (both on a solute-free basis)
solute concentration in the bed at a particular
time, as shown in the breakthrough curve of Figure _M S

The slope of the breakthrough curve is a 16.54)

16.166. !

function of the adsorption rate; the higher the rate,

the steeper the curve.
Adsorption Zone
The adsorption zone can be related to the break-
through curve, as the following development will
show. Some arbitrarily chosen low value of solute
concentration (Yb can be defined as the "break-
)
point" — the point in time at which solute is first
noticed in the effluent. This value could be the
maximum permitted if, for example, the solute
were an air pollutant. This solute concentration
corresponds to a cumulative effluent of b (Figure
Some of this time was required to form the adsorp-
tion zone at the beginning of the bed when the gas
was first introduced. If 6 f is the time required to
form then d s - d f is the time available
the, zone,
for the zone to move through the bed once it is
formed. It is useful to define 6 a as the time re-
quired for the adsorption zone to move a distance
equal to its own width (za ). The effluent accumu-
lated in the time between the breakpoint and the
saturation point is simply (Figure 16.166)
M a = M - Mb 16.55)
M s

16.16). Similarly, a value of solute concentration So that

(Ys approaching that of the entering gas (V^ can
) )

be chosen arbitrarily as the value representing a (16.56)

practical saturation of the bed — the point at
V
which the bed can adsorb essentially no more In the interval 6 S — d ft the adsorption zone moves
solute. This saturation value has a corresponding a distance z a every d a of elapsed time. So the bed
cumulative effluent s M
Typical arbitrary choices
.
length (z) is traversed when (6 S — d f )/d a move-
might be Yb = 0.05 Y, and Ys = 0.95 V, ments of length z a have occurred if the front's
The adsorption zone can be defined as that shape does not change after it is established
length of bed in which the concentration change
Yb to Ys is occurring at any time. As gas is first fed
z = za 16.57)
to a fresh adsorbent bed, the adsorption zone must
be established; it then moves through the bed until
it emerges at the discharge end as the breakthrough
Equation 16.57 may be rearranged to give an
curve. Under certain circumstances, the adsorption explicit expression for the width of the adsorption
zone maintains a constant width as it passes zone
through the bed. These circumstances are:

Pi — Vf
1. Isothermal adsorption (the heat of adsorp-
tion must be small). It is now necessary to find an expression for the
Feed dilute in adsorbable solute.
2.
time required to form the adsorption zone df for
3. Adsorption equilibrium curve concave up-
use in the above equation. The breakthrough curve
ward. (Those shown in Figure 3.12 meet this con- was defined as a function of time, but its represen-
dition.)
tation in terms of cumulative effluent for a
constant-flow case allows a simplification of the
At constant inflow of feed, an adsorption zone of equations.
constant width moves at constant rate through the The last part of the adsorption occurs as the
bed. adsorption zone moves out of the bed,
The following development (14) will make it corresponding to a change in effluent concentra-
possible to plot the breakthrough curve for a
tion from Yb to Ys (Figure 16.166). If this zone
specific adsorption system. With the appropriate
had been saturated, it would contain a V 1 pounds M
rate data, it will then be possible to determine how
of solute, as represented by the rectangle of sides
long a bed of given adsorbent and length will last,
M a and Y^ on Figure 16.166. In fact, it contains
before the breakpoint is reached.
only
The total time any portion of the adsorption
.M,
zone is in the bed is simply the cumulative mass of
effluent at saturation divided by the mass flow rate
YdM
•Mb
 MASS TRANSFER 411

pounds of solute, as represented by the area under yields a new expression for the width of the
the breakthrough curve within the rectangle a V, M adsorption zone
The final section of the bed could still adsorb
za = z (16.61)
0, -d -f)d a

(V, - Y)dM
J
Mb Since, by definition, M= V'd,

pounds of solute, as represented by the area above

Ma
za = z 16.62)
the breakthrough curve within the rectangle M a Yy /tf.-{1 -f)Ma
Thus, the unsaturated adsorption capacity of this This expression will be used below.
region as a fraction of its total capacity is
The slope of the adsorption breakthrough
curve is function of the rate of adsorption. A
a
[Yy-Y)dM solute that is adsorbed rapidly gives a nearly ver-
Mb tical breakthrough curve, approaching the step
f= (16.59)
M a Y, change shown in Figure 16.16a. A lower adsorp-
tion rate yields a breakthrough curve of finite
The formation of the adsorption zone at the slope, as in Figure 16.166. An even slower adsorp-
beginning of the bed may be assumed to follow the tion would yield a curve of still lesser slope. Lower
same sort of pattern as has just been described for rates of adsorption produce longer adsorption
the departure of the zone; so Equation 16.59 can zones and hence a lower fraction of saturation of
be used to describe the formation of the adsorp- the entire bed, as indicated by Equation 16.63. At
tion zone. the breakpoint, a length of bed (z—z a is satu- )

The fraction f obviously some function of

is rated with (z — z a )Sp B X- pounds of solute, while
l

the shape and slope of the breakthrough curve, the adsorption zone, not fully saturated, contains
normally approaching 0.5. Hypothetical systems za Sp B
C\ —f)X-\ pounds of solute. If the whole
can be visualized in which it approaches 1 or 0. Its column were saturated, it would contain zSp B Xy
real importance appears only when its contribution pounds solute. So the fraction of saturation of the
to total solute content is reckoned in terms of the entire bed (F) is

portion of the total bed that is fractionally satu- - za )Sp B X^ - f)X

(z + za Sp B (1 y
rated. For a bed of given length, a flat breakthrough F=
curve results in a large fraction of adsorbent being zSp B X :

partially saturated, whereas a steep slope indicated z- fza

that most of the bed is fully saturated. This partial (16.63)
saturation of a large or small fraction of the total
bed is important in total adsorption capacity at This expression can be used to calculate the total
breakthrough. solute actually adsorbed when the breakpoint is
The fractional saturation is also useful in reached. If the breakpoint is the maximum permis-
establishing the relationbetween the time required sible effluent concentration, then F determines the
to establish the absorption zone and that required solute capacity of the bed, reflecting the portion of
to advance through the bed a distance equal to its the total length of saturated adsorbent, and the
thickness. The amount of solute contained in the remainder of the length in which a lower concen-
adsorption zone is (1 — f) times the amount con- tration exists. The rate of adsorption, as measured
tained in the bed that reaches saturation while the by a mass-transfer coefficient, will be related to the
front is moving its thickness. Since feed is assumed width of the adsorption zone in the next section.
steady, the same fraction applies to time and to
effluent. The value of 6 f at any fractional saturation Transfer Units
may be fitted by the equation The rate and extent of adsorption in the adsorp-
= f)d t 16.60) tion zone can be analyzed using the concept of the
(1
transfer unit developed earlier in this chapter. To
For any value of f between zero and 1 cor- make the analysis, it is necessary to change the
responding to a real adsorption rate, this expres- coordinate system so that it moves through the
sion provides a linear relationship between f and adsorbent bed at the same rate as the adsorption
6g. Combination of Equations 16.58 and 16.60 zone. The adsorption zone is then stationary in this
412 APPLICATIONS TO EQUIPMENT DESIGN

coordinate system, and the gas flows counter-

current to an apparently moving solid adsorbent.
This can be visualized as a continuous counter-
current tower, with fluid phases flowing at such
rates that a pair of concentrations and Y remain X
fixed at the origin. The adsorbent zone can then be
thought ofas though it were a section of a counter-
current column offering continuous contact
between two moving phases, as shown in Figure
16.17. The column is infinitely tall, so that adsor-
bent leaving it (of composition X-\ is in equilib- )

rium with the gas entering (of composition Y, ). Figure 16.18. Compositions in

Equilibrium also exists between the vapor leaving the adsorption zone.

and the adsorbent entering. Because the latter con-

tains no solute (X 2 = 0), the vapor leaving would
The rate of adsorption can be related to a
contain no solute (Y 2 =0). mass balance around a differential volume element
The operating line for the tower can be of height dz by rewriting the design equation in

obtained by writing a mass balance between the mass ratio units (Equation 16.18)
top of the tower and a position within the adsorp-
d(VY)=K Y a(Y*-Y)Sdz (16.65)
tion zone. For a unit area of column cross section
Since V and S are constant, and K Y a can be
L'X 2 + V'Y = L'X+ VY 2 assumed constant in many applications, Equation
Since X2 and Y2 equal zero, 16.65 can be integrated over the height of the
adsorption zone
V'Y=L'X (16.64)
za = dz =
V :
ys dY
\

Equation 16.64 represents a straight line through Jn K Y aS -Y b Y*

the origin, of slope L'/V. This operating line is
= HqgNog 16.66)
plotted with the equilibrium curve on Figure 1

16.18. should be noted that equilibrium can

It

exist at both ends of the column only if the

Equation 16.64 could also be integrated between
z = and some intermediate position within the
equilibrium curve is concave upward, as shown.
adsorption zone, where the concentration is Y
v v V' -Y dY
y2 = q_f_ zY = (16.67)
K Y aS -Y b Y* - Y

Dividing Equation 16.66 by Equation 16.65 gives

y dY
Yb
£ i
x. Zy
za
Y*

dY
(16.68)
~1 .Y

Y* - Y
t
vb

T" The z Y lza can also be related to the cumula-

ratio
xb tive masses of effluent using Equation 16.62. For
z Y the equation becomes
,

M-Mb
Zy = :16.69)
~ ~
Ms _(1 -f)Ma
V T
T where M the cumulative mass of effluent cor-
is

Figure 16.17. Mass transfer responding zY The

to 3 is
. retained in theM
in a stationary absorption denominator because it is part of the expression
zone. for a constant 6f. Division of Equation 16.69 by
 MASS TRANSFER 413

Equation 16.62 gives where

Zy _M Mb - K = equilibrium constant (=1.2 for this
16.70)
za M a
resin)
caS = concentration of sodium ion in the
The right-hand side of Equation 16.70 is a dimen- solid resin, milliequivalents/cm 3 of
sionless expression for the horizontal coordinate of bulk resin volume
the breakthrough curve of Figure 16.16&. This caS,t = concentration of sodium ion in the
expression can be evaluated using the integrals of solid resin when the resin is totally
Equation 16.68 for a given value of Y. It is then saturated (= 1.94 for this resin)
possible to plot the breakthrough curve on dimen- caL = concentration of sodium ion in the
sionless coordinates, V/V-i versus [M — Mb I a ) . liquid, milliequivalents/cm 3 of liquid
The breakthrough curve can be determined with- CaL, =
1 concentration of sodium ion in the
out knowing the mass transfer rate, but determina- liquid entering the bed
tion of za requires knowledge of K Y a. A fixed bed of resin one meter long and 100 cm 2
The following procedure may be used to com- in cross-sectional area is to be used to process 40
plete an analysis of a fixed-bed adsorber, providing cm 3 /s of brackish water. How long can the bed be
the conditions stated earlier are met.
used before it must be regenerated?
1. Plot the operating line and equilibrium Solution. The steps outlined above will be
curve, as Figure 16.18.
in
followed; because the data are given in concentra-
2. For several values of Y between Yb and
tion units, the necessary equations will be con-
Ys , calculate (M - M
b )lMa using Equations 16.68 verted to these units.
and 16.70. 1. Equilibrium Curve and Operating Line.
3. Plot the breakthrough curve Y/Y^ versus
For the entering 2 percent NaCI solution, assum-
(M-Mb )IMa .

ing the density is 1 g/cm

3
(MW of NaCI = 58.4)
4. Calculate f using Equation 16.59 and the
breakthrough curve. caZ.j =0.02/58.4
5. With data on KY a or H 0Gl calculate z a
using Equation 16.66.
= 0.342 milliequivalents NaVcm 3
6. Knowing the total bed length z, as well as The equilibrium curve can be plotted from:
za and f, calculate F using Equation 16.63.
7. Knowing the bed characteristics and F, caS 0.342 -caL
1.2 =
calculate the total mass of solute adsorbed and the 1.94 -caS CaL
time at which the breakpoint occurs.
The operating line runs from (0,0) to intersect
This procedure is demonstrated in the following
the equilibrium curve at c aL = 0.342, as shown in
example.
Figure 16.19.
2. Graphical Integration. The effluent is to
Example 16.4. As a first step in desalination, contain no more than 0.2 percent NaCI, so this
brackish water is passed through a cation-exchange value must represent the breakpoint. Hence,
resin, where the sodium ion concentration is re-
caL ,b =0.002/58.4
duced from the equivalent of 2 percent NaCI to the
equivalent of 0.2 percent NaCI by weight. The = 0.0342 milliequivalents NaVcm 3
mass-transfer coefficient for the resin is given by
(8).
The integration of Equation 16.68, rewritten in
concentration units, is accomplished graphically.
KL a = 0.86(v L )
05
where CaL dc aL

KL a = mass transfer coefficient, s 1 •CaL, b CaL ~ caL

vL = superficial liquid velocity, cm/s c aL,s dC,aL

The equilibrium data are given J\ c

a L,b CaL -CaL ~
CaS caL,1 ~ caL The effluent concentration when the bed con-
K= is

CaS.t -Ca s *aL sidered to be saturated is arbitrarily chosen at 95

414 APPLICATIONS TO EQUIPMENT DESIGN

0.35
from the operating line relationship

0.342
caL = CaS
1.94

rather than to read values from Figure

trying
16.19. Column C
obtained from the equilibrium
is

relationship given earlier. Column D uses the data

of columns B and C to calculate the abscissa for
the graphical integration of Figure 16.20. Column
E gives the value of the integral up to the particular
value of caL which is the numerator of Equation
,

16.68. The last value in column E is the denomi-

nator of Equation 16.68. It is divided into each
entry in column E to obtain column F, using
Equations 16.68 and 16.70 as a basis. Finally,
column G is obtained by dividing each entry in
column B by c aL §1 (which has the value 0.342).
3. Breakthrough Curve: The data of columns
F and G are now used to plot a dimensionless

200

Figure 16.19. Operating line and equilibrium 150 -

curve for Example 16.4.

c *, ~ c *i

percent of the concentration of the liquid inflow. 100 -

CaL.s =(0.95X0.342)
50 -
= 0.325

The accompanying table provides the data for the

integration. In column A of the table, values of c 0.05 0.10 0.15 0.20 0.25 0.30 0.35
aS
are selected to cover the range of data encountered
in the example. In column B, c aL is calculated Figure 16.20. Graphical integration for Example 16.4.

E F G

CaL dc aL M -Mb CaL

CaS CaL CaL ~ caL
c aL -CaL c aL.b c aL
'
CaL.y

0.0 0.0 0.0

0.194 0.0342 0.0290 192 0.0 0.0 0.100
0.2 0.0352 0.0296 179 0.19 0.006 0.103
0.4 0.0704 0.0608 104 5.17 0.168 0.206
0.6 0.1056 0.0929 78.7 8.38 0.270 0.309
0.8 0.1408 0.1250 63.3 10.66 0.346 0.412
1.0 0.1763 0.1607 64.1 12.93 0.419 0.516
1.2 0.2116 0.1966 66.7 15.28 0.495 0.619
1.4 0.246 0.234 83.3 18.09 0.586 0.719
1.6 0.282 0.273 111 21.88 0.709 0.825
1.8 0.317 0.312 200 27.94 0.906 0.927
1.85 0.3250 0.3231 526 30.84 1.000 0.950
1.94 0.342 0.342
 MASS TRANSFER 415

The absorption zone is somewhat more than one-

fifth the length of the bed.
6. Calculation of F At the breakpoint, the
.

absorption zone is still in the bed, so that only the

fraction F of the entire bed is saturated (Equation
16.63).

F= —
100- (0.426) (22.7)

100
= 0.903

7. Calculation of Breaktime. If the bed

were completely saturated, it would retain 1.94
+ 3
milliequivalents of Na in each of its 10,000 cm .

The actual retention is (0.903) (19400). The rate

of inflow of sodium ion is (40)(0.342) milliequiva-
lents/s. The bed will therefore reach its break point
Figure 16.21. Breakthrough curve for Example
in f6 seconds.
16.4.

_ (0.903) (19400)
breakthrough curve (Figure 16.21). It runs from tb
(40) (0.342)
caL,b t0 c aL,s> as expected.
4. Calculation of f: Equation 16.59 can be = 1279 s

converted to dimensionless form to correspond to

The bed must be regenerated after barely 20 min
the coordinates of Figure 16.21.
of operation. To
achieve a longer period between
r
1.0 m -Mb \
regenerations, a larger bed could be used. To com-
f= plete the desalination of the brackish water, the
effluent should next be treated with an anion

This expression shows that f is equal to the area
above the breakthrough curve in Figure 16.21. The
area below the curve is easily obtained from the
data of columns F and G by numerical integration.
The area is found to be 0.574; hence, f = 0.426.
5. Calculation of H OL Since the overall
:
liquidphase mass-transfer coefficient is given, it is
convenient to calculate H 0L using a modified ver-
sion of Equation 16.66
za = HolNol
exchange resin to remove the Cl~. In practice,
mixed resins are used to remove both ions simul-
taneously.
Although several figures were used to aid in
the visualization of the solution of this example,
none of them were necessary for the calculation.
The operating line and the equilibrium curve were
expressed by analytical functions, and both
integrations were carried out using a numerical
method. Hence, problems of this type can be
readily solved using the computer.

where H 0L =
VL /K L a and N 0L =30.84 from
column E. In the expression for H0L the V'/S of ,
STAGE EFFICIENCIES
Equation 16.66 becomes simply VL the superficial , In stage operations, the efficiency of an individual
liquid flow rate, to be consistent with the volu- stage depends upon many factors, including the
metric units of K L a. Combining this expression for rate of mass transfer. Within a single stage, the
H l with the earlier equation for K L a for this contact between the two phases is continuous from
resin gives the time a phase enters the stage until it leaves. The
flow patterns within the stage are usually very
5
'OL =t4t(^)°'
Hn, complex, and often they cannot be accurately pre-
0.86
dicted or described. Because of this, it is difficult
In this case, vL =40/100 = 0.4 cm/s, so H 0L = to predict the contact time and the interfacial area
0.735 cm. Then the width of the adsorption zone is between phases. Therefore, any attempt to predict
stage efficiencies from the rate of mass transfer
za = (0.735) (30.48) involves many approximations. Because knowledge
= 22.7 cm of stage efficiencies is essential in the design of
416 APPLICATIONS TO EQUIPMENT DESIGN

stage-contacting equipment, much research has Ky = overall mass-transfer coefficient, lb

been directed toward the theoretical and empirical moles/hr ft 2 (mole fraction)
prediction of stage efficiencies. These correlations A B = bubble interfacial area, ft 2
are usually based upon a particular type of stage
If the total quantity of gas does not change, as
equipment, such as the bubble-cap tray. In addi-
would be the case in equimolar counterdiffusion or
tion, the correlations are often limited to specific
in dilute gases, Equation 16.71 becomes
chemical species, such as hydrocarbons.
This section will consider the relationship of VB dy = K y 'A B (y*-y)dd
the rate of mass transfer to the stage efficiency or
using a simple model for the phase flow pattern.
The complications of more complex flow patterns
dy Ky 'A B dd
116.72)
will be discussed, but no attempt will be made to V -V V,
cover in detail the numerous empirical correlations.
This equation may be integrated over the path of
Since distillation efficiencies have been most
travel of the bubble through the liquid.
thoroughly investigated, the following discussion
will be based upon a gas-liquid model for bubble- Yn dy KJA>
cap or sieve trays. dd
j
k»+i y* - y VB -o

Point Efficiency (y*-y) Ky 'A B d

-In (16.73)
The point efficiency refers to a single point on the W*~Vn+y) VB
horizontal surface of the contacting tray. At this
The use of a constant value of y* implies that the
point the model visualizes a single stream of gas
liquid composition is constant in the vertical
bubbles rising through the liquid, as shown in
bubble path. Rearrangement of Equation 16.73
Figure 16.22. The bubbles form at the opening,
gives
such as in the sieve tray as shown or at the slot in a
bubble cap (Figures 2.1 and 2.2). The bubbles rise —v* — —
v
= e -<.Ky A B e/vB )

through the liquid and finally break through the y* - y n +\

liquid surface into the gas above the plate. Mass is
If both sides of the equation are subtracted from 1,
transferred between the gas and liquid as the
bubble rises. The interfacial area of the bubble is
A B and its time of contact with the liquid is 6. - y -y
,
1 = 1 _ e -iKy A B 3/VB )

The change in quantity of the component in the y* - y*+i

gas bubble will be equal to the transfer across the
or
interface; that is, for the time interval dd

d(VB y) = K'A B {y* - y) dd (16.71) Kn + 1

= _ e -(Ky A B e/vB
1 )
(16 74)
where /* -K/7 + 1

VB = the total quantity of gas

in the bubble, The left-hand side of Equation 16.74 is the
moles
lb Murphree point efficiency for the V-phase (E v °)
y = vapor composition, mole fraction a as originally expressed by Equation 7.35. There-
y* = composition of the vapor which would fore.
be in equilibrium with the liquid at the
e -(KyA B e/vB )

16.75)
point under observation mole fraction
,

The point (E v °) is not equal to the

efficiency
Gas

Liquid
M -<>----- ^=^U plate {E v
efficiency for reasons discussed in
Chapter 7. The molar flow rate of the bubble
)

(VB /6) may be replaced by the total molar flow

rate (V), assuming that the gas flow is uniform
across the plate. The bubble interfacial area (A B is )

Gas I

„ difficult to evaluate accurately. As in packed

Single point columns, it may be replaced by the average inter-
on tray n
facial surface area per unit volume.
Figure 16.22. Point efficiency in gas-liquid con-
tacting. A o = aSz (16.76)
 MASS TRANSFER 417

where PROBLEMS
a = bubble interfacial surface area per unit
2 3
volume of liquid ft /ft (or
2
/m 3 ) m 16.1. Convert Equation 16.9 to a form where partial
=
S cross-sectional area of column, ft
2
(or
2
) m pressure units p express composition, with the appropriate
z = effective height of liquid, ft (or m) mass transfer coefficient. Convert Equation 16.10 to a form
where concentration units c express composition, with the
Equation 16.75 may then be written as appropriate mass transfer coefficient. Define H and N for
each case.
_
o = - e~-KJazAV/S)
KY
Eu? 1 16.77)
16.2. Does a transfer unit represent a physical model
of mass transfer, as does the equilibrium stage? Under what
where VIS molar mass velocity of gas
is the circumstances does a transfer unit yield the same composi-
through the column, moles/hr ft 2 (or mol/s m 2 ).
lb
tion change as an equilibrium stage?
By definition, H 0G = V/Ky 'aS, so Equation 16.3. Derive Equation 16.30 from Equation 16.27.
16.77 becomes Derive a similar approximate expression for Nqi_.

N OG 16.4. Derive an expression equivalent to Equation

Eu° = 1 -e~ 16.78)
16.41 for N 0L .

16.5. Hydrogen sulfide is to be scrubbed from air by

N g can be related to the individual-phase number initially pure water at 10 atmospheres and 20°C in a packed
of transfer units, which are calculated from avail-
column. The initial H2S composition in the air is 5 percent
able correlations (12). The point efficiency can
(by volume). Determine Nqq if the exit gas composition of
then be related to the stage efficiency, as discussed
H2S is (a) 1 percent; (b) 0.1 percent. Use L = 100 mol/s,
in Chapter 7.
V/= 1 mol/s.
16.6. Using the Onda equations (Equations 16.45,
REFERENCES 16.46, 16.47) as a basis, sketch the following:

(a) a w as a function of liquid mass velocity.

1. Bubble-tray Design Manual, A.I.Ch.E., New York
(b) k{_ as a function of liquid mass velocity.
(1958).
(c) kfj as a function of gas mass velocity.
2. Chilton, T. H., and A. P. Colburn, Ind. Eng. Chem., 23,
(d) kf_ as a function of liquid diffusivity.
p. 913 (1931).
3. Chilton, T. H., and A. P. Colburn, Ind. Eng. Chem., 27, (e) kQ as a function of gas diffusivity.

p. 255(1935).
4. Colburn, A. P., Collected Papers, ASEE Summer school 16.7. Ammonia is to be absorbed from a 30 volume
(1936). percent mixture in air, using pure water as the absorbent.
5. Cornell, D., W. G. Knapp, J. Fair, Chem. Eng.
R. The entering flow rate of gas is 1 mol/s, and the entering
Progr., 56, No. 7, p. 68; No. 8, p. 48 (1960). flow rate of liquid is 3 mol/s. The desired final gas composi-
6. Eckert, J. S., Chem. Eng. Progr., 66, No. 3, p. 39 tion is 0.1 volume percent NH 3 A . packed tower of 0.3-m
(1970). diameter, filled with 1 -in. ceramic Berl saddles, operates at
6a. Fair, J. R., Chem. Eng. 76, p. 90, July 14 (1969). 20°C and normal atmospheric pressure.
7. Gerster, J. Chem. Eng. Progr., 59, p. 35 (1963).
A.,
(a) Plot the equilibrium curve and operating line.
8. Leibson, I., and R. Beckmann, Chem. Eng. Progr., 49,
(b) Calculate the individual mass transfer coefficients
p. 405(1953).
for each phase.
9. Michaels, A. S., Ind. Eng. Chem., 44, p. 1922 (1952).
10. Nemunaitis, R. R., et al., Chem. Eng. Progr., 67, No.
(c) Calculate Nqq, using each of the five methods
11 p. 60 (1971); also, Chemical Engineers' Handbook, outlined in this chapter. Which is most accurate?
5th ed., p. 21-20. (d) Calculate the height of the tower required using

11. Onda, K., et al, J. Chem. Eng. Japan, 1, p. 56 (1968); Nqq and an Hqq calculated with an average value of the
also, Chemical Engineers' Handbook, 5th ed., slope of the equilibrium curve.
pp. 18-35 - 18-39. (e) Calculate the height of the tower required using
12. Perry, and C. H. Chilton, Chemical Engineers'
R. H., Nq and Hq.
Handbook, 5th ed., Sect. 18, McGraw-Hill, New York (f) Calculate the height of the tower required using
(1973). N[_ and HL
13. Sherwood, T. K., R. L. Pigford, C R. Wilke, Mass
.

Transfer, McGraw-Hill, New York (1975).

16.8. A packed tower is to be designed to distill

14. Treybal, R.E. Mass Transfer Operations, III Ed., continuously an equimolar mixture of benzene and toluene

McGraw-Hill, New York (1980). to obtain products of 95 mole percent purity. The feed and
15. Weigand, J. H., Trans. A.I.Ch.E., 36, p. 679 1940).
( reflux are at their boiling points, and a constant value of
418 APPLICATIONS TO EQUIPMENT DESIGN

a = 2.58 can be assumed. If, for the conditions stated fraction acetone. The acetone is to be recovered by absorp-
Hq = 0.5 ft and HL = 1.0 ft, calculate: tion with water in a packed tower. The gas composition is

(a) The relative importance of the iiquid-phase resis- to be reduced to 0.0005 mole fraction. The absorption will

tance, expressed as percent of the total resistance at the top be isothermal because of the cooling coils within the tower.
and the bottom of the tower for conditions of total reflux. For the conditions of the absorption, equilibrium relation-
(b) Repeat part (a) for each end of the stripping and ships may be expressed as / = 1.8x, with compositions in

enriching sections, for a reflux ratio 25 percent greater than mole fractions. The rich gas enters the tower at a flow rate
of 1000 Ib/hr 2
the minimum. ft , and the water enters the top of the tower
(c) For the reflux ratio of part (b) determine NL , at a rate of 1350 Ib/hr ft 2 . The tower is packed with 1-in.

NG , and/VoG- Berl saddles. How tall should the tower be?

(d) For the conditions of part (b) determine the 16.13. A mixture of benzene and a nonvolatile
tower height. absorption oil containing 3.8 percent by weight benzene is

16.9. It is desired to design a packed tower to scrub preheated under vacuum to 300° F and fed to the top of a
ammonia gas from air by means of ammonia-free water fed packed column where the pressure is 120 mm Hg. Just
to the top of the column. Under anticipated conditions, the before entering the column, the flashed vapor is removed
equilibrium conditions are given by the expression Y = and separately condensed so the column is fed only with
0.8X. Two gas streams are to be treated: (a) 16 moles/s of a liquid. The molecular weight of the 220. The waste is oil is

concentrated gas containing 4.76 mole percent ammonia to to contain 0.15 weight percent benzene. The column is fed
be fed to the bottom of the tower, and (b) 17 moles/s of a with live steam at 300°F, and 0.02 mole of steam per mole
dilute gas containing 2.44 mole percent ammonia to be of benzene-free oil is used. The column is kept at 300°F by
introduced at the proper point. The tower is to be tall a suitable heater. The pressure drop through the column is

enough to have an exit-gas concentration of 0.005 mole of 60 mm Hg. Assume Raoult's law is valid and that benzene
ammonia per mole of ammonia free air. Calculate the total has a vapor pressure of 4220 mm Hg at 300° F. Determine:
packed height required. Use a water flow rate of 25 (a) The composition of the feed to the column.
mol s. The maximum velocity of the air stream at any (b) The percent of the benzene vaporized in the
2
point is to be 40 mol/s m empty tower. For the packing preheater.
57
used K Ya = 7.3 (VIS) with WS = mol/s m2 and (c) The composition of the overhead from the tower.
Kya = mol/s m . (d) The number of transfer units required.
16.10. Ammonia is to be removed from a 13 percent 16.14. The purge gas from an ammonia-synthesis plant
mixture of air and ammonia by scrubbing with water in a contains 4.0 percent ammonia; the remainder is assumed

tower packed with 1-in. Raschig rings. If 99.9 percent of substantially air. The gas is to be scrubbed at 5 atm pressure
the entering ammonia is to be removed, how tall must the with water in a tower packed with 2-in. Raschig rings. The
1000 Ib/hr 2 while the 70° F. The
column be? Entering-gas rate is ft tower temperature will be essentially constant at

water rate is 700 Ib/hr ft 2 , both at 20°C. water rate is to be 600 Ib/hr ft 2 and the inert gas rate will

16.11. In the paper industry, large amounts of be 250 Ib/hr ft 2 . Estimate the tower height needed if 99.9
H2SO3 solution are used as a bleach in some processes. percent of the NH3 entering the tower is absorbed.
Bleach is prepared as it is used by absorbing the flue gas 16.15. One kg/s of a benzene-ethylene dichloride
from a sulfur burner in water to produce the bleach liquor. solution containing 50 weight percent benzene is to be
In a particular plant, 11 percent SO2 gas at 1 atm is continuously rectified in a tower packed with 2-in. Raschig
made in the burner and is scrubbed to recover 97 percent of rings to give an overhead and bottom product of at least 95
the SO2. Absorbing liquid is pure water, and the bleach is percent purity. Feed and reflux will enter the column at
to be 0.75 percent SO2 by weight. Operation is assumed their bubble points, and a reflux ratio 1.5 times the mini-
isothermal at 30° C. mum is specified. Design the packed tower with regard to
15,000 SCFH (dry basis) of gas is fed to the absorber, tower diameter and tower height. See I.E.C. 34,1088 (1942).
which is 6 ft in diameter, and packed with 3-in. spiral tile 16.16 Acetic acid is to be extracted from an aqueous
rings to a depth of 10 ft. solution by continuous contacting with benzene in a

(a) How much bleach solution is made per hour? packed tower. The aqueous solution is fed to the top of the
(b) What is Kga for this operation? column and contacted with a rising benzene phase. The
(c) What error (expressed as percentage of the inte- inletaqueous phase enters at a rate of 1000 Ib/hr ft and
grated value) is introduced by using a log mean driving analyzes 0.04 lb mole acetic acid/ft and leaves the unit
force? analyzing 0.001 lb mole acetic acid 'ft Pure benzene enters .

16.12. Air at 80° F is used to dry a plastic sheet. The the tower bottom at a rate of \.2(G v ) rri \ ri
. Experiment
solvent in which the plastic is dissolved is acetone. At the indicates that a mass-transfer coefficient of K c a= 2. 5 lb
3
end of the drier, the air leaves containing 0.020 mole mole/hr ft (lb mole/ft based upon the benzene phase.
)
 MASS TRANSFER 419

The equilibrium relationship for this system is given by (b) The time at which the bed is completely satu-
rated.
c v = 0.025 c L
(c) The total mass of air treated when the bed needs
regeneration.
where
16.20. The definition of the adsorption zone
requires an arbitrary choice of the breakthrough composi-
C\/ = concentration of acid in benzene, lb mole/ft 3
tion Yfr and the saturation composition Ys in the fluid
Cf_ = concentration of acetic acid in water, lb mole/ft 3
phase. Often the values Yb = 0.05 Yy and V = 0.95 Yy are
s

chosen. Suppose instead that Yb = 0.01 Yy and Ys = 0.99

Determine the packed height needed.
Yy were used.
16.17. Diethylamine (DEA) is to be extracted from (a) How would the width of the adsorption zone be
aqueous solution in a 6-in. diameter tower packed with affected by this change?
1-in. unglazed porcelain Raschig rings, using toluene as a (b) How would the total capacity of the bed be
solvent. The aqueous solution analyzes 0.02 lb mole affected, assuming it is regenerated when the effluent
DEA/ft 3 and enters the column at a flow rate of 0.6 ft 3 /hr. reaches the breakpoint composition?
Pure toluene enters the bottom of the tower at the same (c) The bed is actually saturated when Ys = Yy. Why
flow rate. What must be the height of the tower to assure not use this criterion instead of those listed above?
recovery of 90 percent of the entering DEA? For a system 16.21. How does the rate of adsorption affect the
like this (8), and in the contemplated equipment, operating shape of the breakthrough curve, the width of the adsorp-
at 31 °C, it was found that: tion zone, and the time to the breakpoint? (The rate is

HL =
mV proportional to k L a.) How does the capacity of the resin to
3.9 + (18)^
adsorb solute affect these three characteristics? (The
Cone, of DEA in water = 1.156 x cone, in toluene. capacity is proportional to the equilibrium constant AC.)

16.18. In adsorption in a fixed bed, Equation 16.51 16.22. If the adsorption were instantaneous in

gives the rate at which the adsorption front moves through Example 16.4, how long would the bed last before it

the bed if the adsorption is instantaneous. The expression needed regeneration?

contains Xy, the solute composition on the adsorbent that 16.23. Repeat Example 16.4, but with a reduction of
is in equilibrium with the entering gas of composition Yy NaCI equivalent concentration to 0.1 percent by weight in

The equilibrium could be expressed Xy = K Yy for a given the effluent (maximum).

solute and adsorbent. 16.24. In Example 16.4, the desired effluent
(a) As AC decreases, how is the rate of movement of composition, 0.2 percent, was taken as the breakpoint
the adsorption front affected? composition. All the effluent up to that point would in fact

(b) What is the physical meaning of K= 0? Explain have a lower concentration of NaCI. Suppose, instead, that
the physical meaning of the expression that results from the average mixed composition of all the effluent processed
Equation 16.51 in this case. was to be 0.2 percent. How would this change affect the
(c) What is the physical meaning of large AC? How can time the bed could be used before regeneration?
Equation 16.51 be simplified in that case? 16.25. The resin of Example 16.4 is to be used to
(d) For a finite value of AC,how does the entering gas process 100 cm 3 /s of brackish water containing 4 percent
composition influence the rate of movement of the adsorp- NaCI (by weight), to reduce the sodium ion concentration
tion front? in the effluent to no more than the equivalent of 0.1
16.19 Humid air (V = 0.02 kg water vapor/kg dry percent NaCI. How long can the bed be used before it must
air) is to be dried completely using an ideal adsorbent that be regenerated?
instantaneously removes the water vapor from the air. The 16.26. Saline water (8 percent NaCI by weight) is to
equilibrium relationship is 10 Y. The adsorbent has a X= be reduced to a sodium ion concentration equivalent to no
density of 400 kg/m 3 and a porosity (fraction void space) more than 0.5 percent NaCI, using the resin of Example
of 0.3. For an inlet airflow of 10 kg/s, with a bed 3 m long 16.4. The bed, which has a cross-sectional area of 1 m 2 ,

and 1 m2 cross-sectional area, calculate: must be able to process 100 cm 3 /s of water. How long must
(a) The rate at which the adsorption front moves the bed be to permit regeneration once every 24 hr of
through the bed. operation?
 Seventeen

Simultaneous Heat and Mass

Transfer Humidification —

In every case involving mass transfer, heat must component, and the gas phase consists of a non-
also be transferred. When a component is trans- condensable gas, usually air, in which some vapor
ferred from a gas phase to solution in a liquid from the liquid phase is present. Because of the
phase, the latent heat associated with the conden- inherent simplicity, the basic relations for opera-
sation is evolved. When a component is transferred tions involving simultaneous heat and mass transfer
from solution in one solvent to solution in a will be discussed in this chapter in relation to
second, as in liquid-liquid extraction, the differ- humidification and dehumidification.
ence between the heats of solution of the solute in The principles of simultaneous heat and
mass
the two solvents is evolved. Similar heat effects are transfer developed here for evaporation of a
as
present in distillation, adsorption, leaching, drying, liquid without an external heat source are also
etc. In every case, the interfacial temperature will fundamental to reaction kinetics. The problems of
adjust itself so that at steady state the rate of heat mass transfer to a point of reaction and of heat
transfer will balance the equivalent rate of heat transfer away from it are exactly parallel. The
transfer associated with the mass transfer. In opera- chemical reaction does however, allow
not,
tions where mass transfer proceeds by equimolar thermodynamic equilibrium to exist at the inter-
counterdiffusion, as in distillation, or in operations face, whereas equilibrium does exist at the inter-
where the latent-heat effects are small, as in liquid- face in vaporization cases. The rate of the chemical
liquid extraction, gas absorption in dilute solu- reaction must be included in an evaluation of a
tions, and leaching, heat transfer is of minor reaction kinetics situation.
importance as a rate-limiting mechanism. In others,
particularly where there is a net transfer of mass
HUMIDIFICATION: INDUSTRIAL APPLICATIONS AND
from the gas phase to a condensed phase, or vice
EQUIPMENT
versa, the heat-transfer rate is important. In these
cases, it significantly limits the rate at which mass Humidification processes may be carried out to
can be transferred. In still other operations, such as control the humidity of a space or, more usually,
boiling, condensation, evaporation, and crystalliza- to cool and recover water by contacting it with
tion, mass and heat transfer occur simultaneously low-humidity air. The water that has lost heat to
in large amounts, but the rates at which the simul- the atmosphere can then be reused in heat
taneous transfer of mass and heat occur can be exchangers throughout a plant. Alternatively, the
determined by merely considering the rate of heat water could be cooled in surface heat exchangers.
transfer from an external source. The choice is one of economics, with the designer
Of those operations where both heat transfer balancing the loss of cooling water inherent in the
and mass transfer affect the rate, humidification air-water contact cooler against the cost of supply-
and dehumidification are the simplest and are also ing and handling the cooling source to the surface
the most direct applications of the theory. Here, cooler and the higher cost of the surface units.
only two components and two phases are involved. Dehumidification is practiced most commonly
The liquid phase, most often water, is a single as a step in an air-conditioning system. Itmay also

420

be used as part of a solvent-recovery system, but in
these applications the condensable vapor is not
water but instead a solvent such as trichlorethy-
lene, benzene, or methanol. Any one of these
applications can, in principle, be carried out in
similar pieces of equipment. The directions of mass
transfer and heat transfer are determined by the
relation between the humidity and temperature of
the inlet gas phase and the temperature of the
contacting liquid. The unit size, the convenience of
recovering gas or liquid phases, and the materials of
construction used all limit the applicability of a
single piece of apparatus.
The most obvious form of humidification
equipment the spray chamber. Here, the contact-
is
ing liquid sprayed as a mist into the gas stream.
is
Gas velocity is kept low so that the contact time is
high and so that there will be only a small amount
of liquid physically entrained in the gas stream.
These units are usually restricted to small-scale
operations and are frequently used in humidity
control of a room or plant where either humidi-
fication or dehumidification of the inlet
required.
The fume scrubber shown Figure 17.1 is a
in
variation of a spray chamber. Here, throat pieces
are used to assure close contact of gas and liquid
streams, and the alternate open areas allow separa-
tion of the two phases. Nozzle capacities are norm-
ally 5—10 gal/min of water with throat pieces
handling 300ft 3 /min of gas per gallon per minute
Water inlet
Spray nozzle
Figure 17.1. Spray chamber type of fume scrubber. (Courtesy Schutte
SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 421
of liquid. The scrubbers are often applied to
removing dusts from a gas stream or for reacting
the liquid with a component of the gas stream as
well as for removal of a condensable component
from the gas phase.
As water for process cooling has become
more precious, the recooling and reuse of this
water have become common. Most often mechan-
ical or natural draft cooling towers are used,
although this cooling can also be done in spray
ponds or in large holding ponds. Spray ponds have
the disadvantage of small capacity for a given pond
area and large water losses. Typical spray nozzles
for these applications deliver 50 gal/min each at
about 7 psig. They are located about 6 ft above the
pond surface and spaced about every 4 ft. In
exposed areas the pond might be surrounded by a
louvered fence about 12 ft high to reduce wind
losses. Empirical correlations are available to allow
estimation of the size of spray pond necessary for a
given duty at known air conditions and spray
nozzle spacing and design (9). Typical water load-
2
air is ing rates for spray ponds are about 500 Ib/hr ft of
pond area.
More typically, and especially for larger cool-
ing duties, cooling towers are used. These towers
are most often of wood construction with multiple
wood-slat decks. Aluminum, steel, brick, concrete,
and asbestos-board casings have also been used.
The water is sprayed above the top deck and
trickles down through the various decks to a
Stack
& Koerting Co.)
422 APPLICATIONS TO EQUIPMENT DESIGN

bottom collection basin. Corrosion is prevented by the tower chimney. Water to be cooled comes from
construction entirely of inert materials such as the plant and is sprayed over the top of the pack-
redwood, stainless steel, and porcelain. ing. It flows down over the material, which is
In the largest applications, such as recooling usually wood 30 ft high,
or plastic slatting, perhaps
of the cooling water used in power plants, natural and pond below the tower. Water from
rains into a
draft cooling towers are often used. Figure 17.2 this pond is returned to the plant. The air leaving
shows such an installation with the natural-draft the top of the tower is warm and moist. On con-
cooling tower looming over the rest of the plant. tacting the colder ambient air, much of the mois-
Each of these towers is roughly the size of a ture condenses forming a steam plume that drifts
football field stood on end, some 300 ft tall by with the wind. This plume contains salts from the
250 ft in diameter. The tower is built of reinforced water treatment and corrosion products from the
concrete varying from 5 to 12 in. thick. The hyper- plant which have been known to cause problems as
bolic shape results from the fact that the moist they deposit downwind. Also, the plume itself can
airflow up such a chimney takes such a form natur- cause fogging, road icing, and additional rainfall
ally, and considerable saving in concrete can be downstream from the plant. Control of these
obtained by conforming to this shape. A cross- ecological problems is much
a part of the design
as
sectional view of such a tower is shown in Figure engineer's responsibility as the prediction of the
is

17.3. Air enters the tower through the open ring thermal and mechanical performance of the tower.
around the bottom, itself some 25 ft high, flows up General thermal design of these towers follows the
through a packing material over which the warm methods described in this chapter, but mechanical
water is distributed, and goes finally up through design, plume flow behavior, and moist in-flow

Figure 17.2. A hyperbolic cooling tower in operation at a power plant. (Courtesy

The Marley Co.)
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 423

Reinforced concrete shell

support columns
Cold water basin

Column support pedestal

Shell foundation ring footing

274.00 ft diameter <£ columns at top of support pedestal

356.83 ft inside diameter of cold water basin

Stairway, walkway.
and flume cover
Precast concrete at inlet
canopy deck

Figure 17.3. Sectional drawing of the hyperbolic cooling tower shown in Figure
17.2. (Courtesy The Marley Co.)
424 APPLICATIONS TO EQUIPMENT DESIGN

prediction are beyond the scope of
Further information is
5, 12, and 24.
intermediate-sized water-cooling applica-
In
tions, forced or induced draft towers are usually
used. Induced-draft towers are preferred because
they prevent the recirculation of the humidified
air. In these towers, fans are located on the top of
the tower. Air is pulled into louvers around the
bottom of the tower and up through the decks
countercurrent to the water flow. In the largest
towers (capacities up to 100,000 gal/min of water),
air may be pulled in through two open sides of the
tower and up to a central fan through inclined
baffles. Empirical correlations are available to
facilitate cooling-tower selection and design (1).
this
available in References 3, 4,
book. Figure 17.4 is a cutaway drawing of a
double-flow, induced-draft cooling tower. Here air
enters both sides of the tower and flows across the
path of the falling water to a central duct. A fan
pulls the air up the duct and blows it out the top
of the tower. Water is distributed initially by flood-
ing a distributor plate containing porcelain nozzles
through which the water flows onto the packing.
Packing is of wood-slat construction assembled
without nails. The drift eliminators, which are
wood baffles through which the air passes before
entering the fan, reduce entrainment of water and
thus reduce mist content in the air discharged from
the fan. Such a tower would be about 48 ft wide
by 30 ft high with a fan 14 ft in diameter. Several
units may be constructed end to end to give a

Vertical Veloctty-recovery
stave fan cylinder
cylinder

Multiblade fan
Access door m cylinder
Motor mounted on
directly opposite motof
transverse l£ of
for gear reducer removal
each cell
s~ Handrail around fan deck
Flow control
valve
Open d.str,bution
system -\ g
^ ,

Flow control
valve
Inlet pipe-

Horizontal corrugated
Removable porcelain
• asbestos-cement board
distribution nozzles
endwall casing

Nailless fill

Removable louvers-

Concrete basin

Concrete interior piers required Access through \ For towers with extended columns.
for towers with standard columns each cell
partition
^- interior piers are optional
or when interior- column anchorage for all unanchored columns
is necessary

Figure 17.4. Transverse cross-sectional view of double-flow induced-draft cooling

tower. (Courtesy The Marley Co.)
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 425

Figure 17.5. A double-flow induced draft cooling tower bank. (Courtesy The
Marley Company.)

single long tower. In somewhat smaller cooling pressure. On this diagram, the concentration of
towers, directly countercurrent flow may be used. saturated gas is plotted against temperature from
Figure 17.5 pictures this type of tower. Internal data that must in general be obtained
construction is similar to that shown in Figure experimentally.
17.4. For the special case of ideal solutions in both
Various other contacting devices are used to phases, the saturation locus can be plotted from
dehumidify air or to remove a single component the results of Raoult's and Dalton's law calcula-
from a noncondensable gas. Often packed towers tions.
are used, or wash columns in which contacting is
essentially stagewise.A contactor of this sort is Pa = Pys = Pa* (17.1)
shown Figure 17.6. These devices are used to
in

contact the gas stream with several liquids, keeping where

the liquid streams separate. Water jet ejectors are
frequently used to condense or "scrub" a vapor pa = partial pressure of condensable com-
from a gas stream. The ejectors handle large ponent the gas phase
in
volumes of water and produce intimate cocurrent P = total gas-phase pressure
contacting of the two phases Pa = vapor pressure of condensable com-
ponent
PHASE RELATIONS AND DEFINITIONS ys ,x = saturation mole fraction of con-
densable component in the equilib-
As in previous consideration of mass-transfer and rium vapor and liquid phases,
heat-transfer processes, in the simultaneous respectively.
transfer of heat and mass, the direction of transfer
and the extent of this transfer are controlled by In this case, x = 1 since the liquid phase is a single
the equilibrium condition toward which the component. Thus,
transfer tends. Therefore, a statement of phase
equilibrium is necessary as a basis for application
ys = (17.2)
of the rate equation.
The phase conditions of gas-vapor mixtures
are conveniently shown on a concentration- Lines of constant percent saturation can also
temperature diagram plotted for a single constant be plotted on this diagram, where percent
426 APPLICATIONS TO EQUIPMENT DESIGN

ratios. Onthe concentration-temperature diagram,

lines of constant percent saturation will be shaped
like the saturation curve but will be at proportion-
ally lower concentrations.
An alternative concentration expression is
the relative saturation, which is defined as the
partial pressure of condensable component present
divided by the partial pressure of condensable com-
ponent present at saturation.

100p a 100/
Percent relative saturation =
(Pa)s Ys

(17.4)

where again the subscript s refers to saturation.

Relative saturation is the measure normally used in
reporting weather data. As shown by Equations
17.3 and 17.4, relative saturation is a ratio of mole
fractions, in distinction from percent satura-
tion, which is a ratio of mole ratios. Since Y = yl-
(1-K),
100/
1-K 100/
Percent saturation =
Ys Ys d-/)
1-/,
or the percent saturation is equal
Figure 17.6. A contactor allowing a multi- to the percent relative saturation (17.5)
ple wash to be used in cleaning dust fumes multiplied by (1 - /5 )/(1 - /)
or vapor from a gas stream. The design takes
advantage of turbulent and wet-cyclone Since y s > y, the percent saturation will always be
action to give good contacting and effective less than, or in the limit equal to, the percent

separation at several stages. (Courtesy relative saturation.

Schneible Co.) Traditionally, mole-ratio units are preferred
because they are based on the unchanging unit
saturation is defined as quantity of noncondensable gas, hence, they will
Ma be used here. However, most practical applications
100/ involve the air-water system at 1 atm total pressure
Mb —Y and temperatures between 50 and 130°F. For
percent saturation = = 100
M these cases, concentrations are low enough so that
Ys —
a

M
b
/ % V, and the percent saturation and percent
relative saturation are practically equal. In systems
(17.3) other than air— water or at high temperatures with
where the air— water system, the concentrations / and y s
may be significantly higher, so that Y is not equal
Y = mole ratio of condensable to non-
to /. Therefore, the generality that / * Y should
condensable component present
be used cautiously.
Ys = mole ratio present at saturation
For most of the systems of industrial import-
Ma M , b = molecular weights of condensable ance, the gas phase is dilute, the pressure is low,
and noncondensable components,
and Raoult's law is followed. Equation 17.2 can be
respectively.
used to calculate the saturation locus, and partial
Note that the percent saturation can be defined in saturation can be expressed in terms of vapor
terms of mole ratios as well as in terms of mass pressures and partial pressures.
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 427

100/ 10CW(/>-p a )
computed relative to the enthalpy of some
Percent saturation = arbitrary base condition at which H=0. The
base
Ys PJ[P-Pa )
conditions cannot be identical for b and a because
Pa(P-Pa ) different components are being considered. They
= 100
may be at the same temperatures, pressures, and
Pa(P~P S
phase conditions, although this is in no way neces-
17.6) sary. For the air— water system, it is common prac-
tice to take the enthalpy to be zero for liquid
100/ 100pa
Percent relative saturation = water at the triplepoint. The base for the enthalpy
Vs Pa of air is often taken as the dry gas at 32° F, 1 atm
(17.4a) pressure. These bases will be used here when
English units are used. For S.I. units the bases are
Whenthe air— water system is discussed, it has 0°C and 10 5 N/m 2 On these
. bases, the enthalpy of
become common practice to call the weight ratio moist air may be related to the appropriate latent
of water vapor to dry air the humidity. The mole and specific heats.
ratio of these components is called the molal
humidity, whereas relative humidity and percent H = cb (T-T ) + Y[\ +ca (T-Tt )}
humidity are synonymous with relative and
percent saturation. This nomenclature has (17.10)
frequently been used for systems that do not con- where
tain water, so that it is not uncommon to hear of
7"
= the base temperature, here 32° F for
the humidity of ethanol in air or to see a humidity
both components
diagram for the benzene— 2 system (15).
In humidification calculations, it is fre- Xo = the latent heat of evaporation of water
at the base temperature, Btu/lb mole
quently necessary to know the volume and specific
heat of the gas-vapor phase. For this use, the and the effect of pressure on the enthalpy of the
humid volume is defined as the volume of 1 lb of liquid is ignored. In this equation, both c b and c a
dry gas plus its contained vapor; similarly the molal have been taken as constant. If this assumption is
humid volume is the volume of 1 lb mole of dry not tenable, these specific heats must be the
gas plus its contained vapor. Thus, integrated mean values applicable between T and
To-
V„ = (1 + /)x359x^ (17.7) other terms used to describe vapor-gas mix-
tures are the dew point, wet-bulb temperature, and
applies to those conditions where the perfect -gas adiabatic-saturation temperature. The dew point is
law is valid. Here, Vh
the molal humid volume.
is that point at which condensation begins when the
Similarly, the molal humid heat is defined as the pressure or temperature is changed over a mixture
heat capacity of 1 lb mole of dry gas plus that of of fixed composition. Usually the temperature is
its associated vapor. That is, lowered at constant pressure, so that the dew-point
temperature is obtained. The wet-bulb temperature
Ch =c b + Yc a :17.8)
is the steady-state temperature attained by a wet-

where ch c b and c a are the molal humid heat and

, ,
bulb thermometer exposed to a rapidly moving
the molal heat capacities of components b (non- stream of the vapor-gas mixture. The bulb of the
condensable gas) and a (condensable vapor) wet-bulb thermometer is coated with the same
respectively. liquid that forms the vapor in the vapor-gas mix-
The enthalpy of the vapor-gas mixture is not ture. The wet-bulb temperature is related to the

generally given any special name. However it fre- humidity of the gas phase, as will be shown below.
quently appears in calculations. By analogy to This relation allows the use of the wet-bulb
Equation 17.8, temperature coupled with the temperature itself as
a measure of humidity.
H = Hb + YHa [17.9)
The adiabatic-saturation temperature is the
where H is the enthalpy of 1 lb mole of dry gas temperature that the vapor-gas mixture would
plus that of its contained vapor, and Hb and Ha are reach if it were saturated through an adiabatic
the molal enthalpies of components b and a, process. Many of the processes discussed in this
respectively. These enthalpies must, of course, be chapter and in Chapter 18 occur approximately
428 APPLICATIONS TO EQUIPMENT DESIGN

adiabatically, so that the adiabatic-saturation concentration diagram are shown schematically in

temperature is and important

a particularly useful Figure 17.10 and quantitatively in Appendix D-14,
quantity. The relation between this temperature the humidity chart for the air-water system.
and the gas-phase humidity will also be shown
below. Adiabatic -Saturation Temperature

For convenience these definitions are col- The relation between the adiabatic-saturation
lected in Table 17.1. The saturation locus and temperature and the gas-phase humidity can be
partial-saturation curves on the temperature- developed by considering the process by which the

Table l".l DEFINITIONS OF HUMIDITY TERMS

Term Meaning Units Symbol

1 Humidity Vapor content of a gas Mass vapor/mass noncon-

densable gas
Y = Y —
Ma
Mb

2 Molal humidity Vapor content of a gas Moles vapor/mole noncon-

densable gas

3 Relative saturation Ratio of partial pressure atm/atm, or mole fraction/

or relative humidity of vapor to partial mole fraction, expressed
100
pressure of vapor at as percent
Ys
saturation

4 Percent saturation or Ratio of concentration Mole ratio/mole ratio,

percent humidity
of vapor to the con- expressed as percent
centration of vapor at
100
saturation with con-
centrations expressed
as mole ratios

5 Molal humid volume Volume of 1 lb mole or ft 3 /lb mole of dry gas or Vh = (1 + /)x359
492/>
of 1 mol of dry gas m 3 /mol of dry gas
plus its associated in English units, or
vapor
—T 1.013 x10 5
Vh = (1 + Y) x 0.0224 - x
273 P
in SI units

6 Molal humid heat Heat required to raise Btu/lb mole of dry gas "
F
the temperature of joule/mol of dry
1 lb mole or of 1 mol gas°C
ch =cb + Yca
of dry gas plus its

associated vapor 1°F

or1°C
: r c
7 Adiabatic-saturation Temperature that would F, R, C,or K
temperature be attained if the gas
were saturated in an
adiabatic process
5
8 Wet-bulb tempera- Steady-state temperature F,°R,°C,orK
attained by a wet-bulb
Tw
thermometer under
standardized conditions
: C =
9 Dew-point tempera- Temperature at which F, R, C, or K
ture vapor begins to
condense when the
gas phase is cooled
at constant pressure
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 429

Hy 2 , V, T V2 L 2 TL2 • . HL2 reach equilibrium at the top of the tower, the gas
Y2 ,V
"
phase will be saturated, and TL2 L
= =T <
l

T V2 = T2 The temperature top of the tower at the

mm
.

(7" will then be the adiabatic-saturation tempera-

Q 2 )

ture, and Y 2 will be the molal humidity of gas

saturated at 7~2 The gas-phase temperature at the
.

bottom of the tower (T v -\) may be simply desig-

TV1 . HV1 , V,, L •
TL\ • Hu nated as T x Applying these restrictions, Equation
.

V 17.15 becomes
Figure. 17.7. General humidification
process. c L (T2 -T )(Y2 -Y,)=[ch2 (T2 -T )

+ a V2 ~] [c »i (7"i -"M + AoVil (17.16)

gas can be saturated. A general humidification
process shown in Figure 17.7. Here the sub-
is Either of the humid heats may be replaced in terms
scripts and 2 refer to the bottom and top of the
1 of the other because
column, respectively; subscripts L and V refer to = Cb +
Chi y^a and ch2 = cb + Y2 ca
liquid and vapor phases respectively; L and V are
the molal rates of flow of liquid and vapor or
respectively; and V
is the molal flow rate of non-
ch2 =ch: -Y,ca +Y2 ca (17.17)
condensable gas, lb moles/hr. A material balance
around the tower gives so that

(L 2 -L,) = V'(Y 2 - Y,) (17.11) c L (T2 -T )(Y2 -Y,) = Cfn (T2

[
-T )

and an enthalpy balance gives -Y,ca (T2 -T )

-L,H Ly + Y2 ca (T2 -T )+\ Y2

q + L 2 H L2 = V(H V2 -//„,) (17.12) ]

-[ch ,{T, -T )+\ Y,]

For this general process to be specialized and
to be made adiabatic, several restrictions must be (17.18)
imposed. First, no heat can be transferred to the
Rearranging and collecting terms will give
tower, or q = 0. Second, the liquid stream will be
recycled. In this way, at steady state TL = TL2 -\ . C/,1 (T2 -T,) = (Y2 - Y,)[c L (T2 -T )

As the process continues, the liquid temperature -X -ca (r2 -r (17.19)

)]
will become constant, and thus no net sensible heat
will be brought to the tower or taken from it by The bracketed terms in this equation are equal to
the liquid. The only effect passage through the — 2 as can be shown by the following enthalpy
,

tower will have on the liquid stream is that some of balance:

it will be vaporized into the gas stream. Under
HL2 — Ha2 = —\ 2
these conditions, Equation 17.12 becomes
= WlI ~ HlOi + '^L0 — HaOl
HL (L 2 -L,) = V'{H V2 -H V ,) (17.13)
+ (Ha Q— Ha2 )

Combining Equations 17.1 1 and 17.13 to eliminate = c L (T2 - T + (-An) + i-ca (T2 -T
) )]
(l 2 - Z-1 ) gives
(17.20)
H L (Y2 -Y,) = H V2 -H V , (17.14)
Equation 17.20 expresses the result obtained when
The enthalpies can now be expressed in terms of determining H a2 — H L 2 = > 2 by using a path in
molal latent heat and humid heat as given by which the system, here one mole of component a,
Equation 17.10. 7"
is first cooled as a liquid to is then vaporized at ,

7"o, and is finally heated as a vapor back to 7"2 The

c L (TL2 -T )(Y2 - V,) - -T[ch2 {T v2 )
.

summation of enthalpy changes along the entire

+ ~k Y2 ]
- [c„(Tvi -T )+\ Y,] path will equal the difference between final and
initial enthalpies because enthalpy is a state func-
(17.15)
tion. Making the indicated substitution gives
Now, if the further restriction is made that
the tower be so tall that the gas and liquid phases C/,i(r2 -r 1 )
= X2 (V 1
-Y 2 ) (17.21)
430 APPLICATIONS TO EQUIPMENT DESIGN

Since in this development 2 and Y 2 were

7" where the subscripts w and 1 refer to the wick
conditions at the adiabatic-saturation temperature, surface at the wet-bulb temperature and to the
Equation 17.21 can be written bulk-gas temperature. The rate of sensible heat
transfer to the wickq follows the usual convection
Ch\ \Tsa -T,)=\ a {Y,-Ysa ) (17.21a) and radiation mechanisms. The rate of mass
transfer of component a toward the wick Na also
where the subscript sa refers to the adiabatic-
follows the usual gas-phase mass-transfer mechan-
saturation condition corresponding to the initial
ism. The coefficients h c and h r are those for heat
gas-phase conditions designated by the subscript 1.
transfer by convection and radiation to the bulb.
Equation 17.21a gives the relation between
the temperature and humidity of a gas at any
By using h r it is assumed that the radiant-heat
,

transfer may be approximated by

entering condition and the corresponding condi-
tions for the same gas at its adiabatic-saturation
qr =h r A\J - y
7"
M (17.24)
temperature. Thus, if both the adiabatic-saturation
temperature and the actual temperature of a gas At steady state, all the heat transferred to the wick
are known, the gas humidity is obtainable. The is used to vaporize Na moles of liquid, or
humidity at the adiabatic-saturation condition can q = -Na \ (17.25)
be found from the saturation locus, and the gas
humidity can then be found through Equation which expresses the condition that the rate of
sensible-heat transfer to the wick is exactly equal
17.21a. Note also that Equation 17.21a was de-
to that of the latent heat carried from the wick by
veloped from overall material and enthalpy bal-
the mass (Na ).
ances between the initial gas condition and the
adiabatic-saturation condition. Thus, it is applic-
Combining Equations 17.22, 17.23, and
17.25 yields
able only at these two points but may not describe
the path followed by the gas as it becomes satu- (h c +h r )A(T, -rw )=-k Y A{Y, - Y w )\,
rated.
or, if the areas available for heat and mass transfer
Wet-Bulb Temperature are equal,

One of the oldest, and still the most common, K Y \w

methods of measuring the humidity of a gas stream T:-Tw = (Y w -Y } ) (17.26)
{h c +h r )

is to measure "wet-bulb temperature" as well as

its
Equation 17.26 relates the wet-bulb tempera-
its temperature. This is done by passing the gas

rapidly past a thermometer bulb kept wet with the ture to the gas humidity, just as Equation 17.21a
did with the adiabatic-saturation temperature,
liquid that forms the vapor in the gas stream.
Usually the thermometer bulb is encased in a satur- provided that predictable values of the term
ated wick. During this process, if the gas is not k Y \ w /(h c + h r ) can be obtained. This ratio of
saturated, some liquid is evaporated from the satu-
depends upon the
transfer coefficients obviously

rated wick into the moving gas stream, carrying

boundary, and temperature condi-
particular flow,

with it the associated latent heat. The latent-heat tions encountered. In measuring wet-bulb tem-

removal results in a lowering of the temperature of perature, several precautions are made to ensure

the thermometer bulb and wick, and therefore that reproducible values of k Y \ w l(h c +h r ) are

sensible heat will be transferred to the wick surface obtained. The minimized by
radiation coefficient is

by convection from the gas stream and by radia- shielding the wick from the surrounding surfaces if

tion from the surroundings. At steady state, the the surfaces are hotter than the wet-bulb tempera-

net heat flow to the wick will be zero, and the ture. Gas movement past the bulb is made rapid,

temperature will be constant. The wet-bulb often by swinging the thermometer through the gas
as is done with the sling psychrometer or by insert-
temperature is the temperature attained at steady
state by the thermometer exposed to rapidly mov- ing the wet-bulb thermometer in a constriction in

ing gas. Thus, the rate of heat transfer to the wick the gas-flow path. Under these conditions Equation
is
17.26 reduces to
kyX
Y AW
q = {h c +h r )A(Ty-Tw ) (17.22) T,-T„ = Yw -Y,) (17.26a)

and the rate of mass transfer from the wick is

where the k Y lh c ratio is that obtained under
Na =k Y A(Y, -Yw ) :
17.23) turbulent-flow conditions.

M
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 431

For turbulent flow past a wet cylinder such as For systems other than air-water, the behavior
the wet-bulb thermometer, the accumulated in a saturating contactor is quite complex. If the

experimental data give (2) column overall conditions satisfy the adiabatic

0.56
model, the liquid and the gas entering and leaving
— =8.50 Btu/lbmole°F (17.27)
must be related by the adiabatic-saturation equa-
ky P Q ab
\P?J tion. This equation does not reveal anything of the
humidity-enthalpy path of either the liquid phase
when air is the noncondensable gas, and
or the gas phase at various points within the con-
056 tacting device. Each point within the system must
hc_ //Vsc\
;17.28)
h conform to the wet-bulb relation, which requires
ky'° \Nj that the heat-transferred be exactly consumed as
for other gases. The data upon which Equation latent heat of vaporization of the mass transferred.
17.27 is based are summarized in Figure 17.8. Consideration of the adiabatic-saturation tower for
Equation 17.28 is based upon heat- and mass- such a system indicates that the liquid entering and
transfer experiments with various gases flowing leaving must be at the temperature of adiabatic
normal to cylinders. saturation, as must also the gas leaving. Speculation
Point A on Figure 17.8 is for pure air for on the path taken indicates that the gas could
which the /V Sc = /V Pr = 0.70. For this case, both follow the straight line joining initial conditions and
Equations 17.27 and 17.28 gw/eh c /k Y =c h =6.94 adiabatic-saturation conditions, with consequent
Btu/lb mole°F (or 29.08 joules/mol C.) Experi- swinging of the temperature of the liquid to follow
mental data for the air-water system give h c lky the required wet-bulb temperature, or that the
values ranging between 7.82 and 6.83 Btu/lb liquid could remain constant in temperature at the
mole°F. The latter figure is recommended. Thus, adiabatic-saturation point, with the gas following a
for the air-water system, the h c /ky value can be path during its humidification that would satisfy
reasonably replaced by c h within moderate ranges the wet-bulb relations at every point within the
of temperature and humidity, provided that the column. More probably, there is some change in
flow is turbulent. As a result. Equation 17.21a, the temperature of the liquid and some predictable
which relates the humidity to the adiabatic- path of gas temperature and composition so that at
saturation conditions, and Equation 17.26a, which all points within the system the wet-bulb relation is

relates the humidity to the wet-bulb temperature, satisfied and so that at the ends of the system the
are identical. Therefore, the adiabatic-saturation adiabatic-saturation conditions are satisfied.
temperature equals the wet-bulb temperature for
the air-water system. For systems other than air- The Lewis Relation
water, this coincidence does not occur, as can be
The identity h c lk Y with c h was first found
of
seen from the psychrometric charts given by Perry
empirically by W. K. Lewis and hence is called the
(15).
Lewis relation. The basis for its general validity
may be determined by examining more closely the
1.0
heat- and mass-transfer processes occurring at a
0.8 Vapor Components
07 saturated surface. The gradients of interest are
o *ater
0.6 shown in Figure 17.9.
0.5 ^ • :e":e-e
~ carbon tetrachloride

..1
•~ chlorooenzene
04
f
CD
a ethyl
7
acetate

tetrachloroethane
- k .i--.
>* 56
= 0.294 AT* > propanol
tTM. methanol
<
0.2 • naphthalene
p-dichlorobenzene
015 a camphor
p-dibromobenzene

0.1
04 0.5 06 08 10 1.5 2.0 30
Schmidt number (N& = u/p'/aA )
Distance *-
Figure 17.8. The ratio h c /k Y for wet-bulb thermometry in
air-vapor systems (2). (By permission of Amer. Chem. Soc; Figure 17.9. Transfer from a saturated
surface.
copyright 1950.)
432 APPLICATIONS TO EQUIPMENT DESIGN

For the transport of heat and of mass occur- Thus, the Lewis relation holds for transfer within a
ring point in the main
between the interface and a completely turbulent regime of A/ Pr =/\/ Sc Near .

bulk of a turbulently flowing gas stream such as the fluid boundary, as for instance with transport
point 2 of Figure 17.9, the general transport equa- between points / and 1 of Figure 17.9, laminar
tion (Equation 13.61) can be written for the two flow_prevails,_and a and 9 are large in comparison
mechanisms. For heat transfer this becomes to Eq and EN In this region Equation 17.31
.

reduces to

-q = -h c AT =
-4(a + £J
-^-pCpAT (17.29) a 7/v
A yq L D -g—pcpRT (17.34)
"9 7q
and for mass transfer Equation 13.61 is written As before, yN = y q when /V Sc =/V Pr but this ,

equality of Prandtl and Schmidt numbers also de-

Na -4(9 + N -A{9 + EN )Apa
—
A
= k Ap =
^ a
g
)
Aca =
RT
mands that a = Q. Thus, again,
yN L D yN L D
hc
(17.30) = cP (17.34a)

where the nomenclature is as given in Chapter 13.

Equation 17.29 may be divided by Equation 17.30 The Lewis relation then has been found to
to obtain, after simplification, the value of the hold for each part of the transfer path from /'to 2,
ratio of transfer coefficients. or from the interface to any point in the main gas
stream, and thus for the entire transfer path, if
(« + £„ N Pf = N Sc If /V Pr i= N S
7/v the Lewis relation must
pCpRT 17.31)
.
c_

kg (Q+EN ) yq be modified to include Eq IE N for turbulent trans-

port and <xl & for laminar transport. This equality
For completely turbulent transport, such as of /V Pr with /V Sc is found to hold true for simple
that between points 1 and 2 of Figure 17_J9, ot and monatomic and diatomic gases, but it is not valid for
9 are negligibly small compared to Eq or EN Then
.
more complicated fluids. Of course, the most
important system to which it applies is air with
he Eg 7/V
pcpRT 17.32) small concentrations of water vapor.
kg En yq Where it applies, the Lewis relation implies
that the mechanisms of heat and of mass transfer
The terms yq and yN represent the ratio of the are identically dependent upon the flow condi-
difference in property between the
transferent tions. Both the eddy and molecular diffusivities for
interface and the bulk fluid to the difference in heat transfer equal the eddy and molecular dif-
this property between the interface and the maxi- fusivities for mass transfer. Hence, the relative
mum value in the fluid. Thus, their values usually importance of eddy to molecular transport of heat
are slightly less than 1 .0 and yN « yq .
is the same as the relative importance of eddy to
From this molecular transport of mass (see Chapter 13). The
simplifications of concept and of calculational pro-
5.
cP j- 17.33) cedures that result for those systems to which the
kg P l=N
Lewis relation applies are major ones. The resulting
identity of the wet-bulb-temperature equation
for turbulent transport in an ideal gas system under
any conditions of (Equation 17.26a) and the adiabatic-saturation
/V Sc with /V Pr .

equation (Equation 17.21a) has already been

For the Lewis relation to hold, Eq must equal shown. It will be shown below that this is one of
EN This requires equality of /VSc with /V Pr even in
.
the simplifications permitting the combination of
the regime of complete turbulence.
temperature difference and mole-ratio difference
For equal Prandtl and Schmidt numbers y q = driving forces for heat and mass transfer into a
y N and also Eq = E N as shown in Chapter 13 (see single driving force expressed as a specific enthalpy
Equation 13.109). Also at low concentration difference. With this simplification, calculations
k g P » kg p bm = ky From this .
involving equipment in which heat and mass are
transferred simultaneously are no more cumber-
he some than those for equipment where only heat or
= Cp :17.33a)
kgP ky mass is transferred.
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION
The Humidity Chart
The previously described concepts and definitions
are normally presented graphically on a "humidity
chart." Such a chart applies to a single noncon-
densable and a single condensable component at a
fixed total pressure. From the chart, information
on the humidity, relative or percent saturation,
humid volume, and humid heat or total enthalpy
should be obtainable from conveniently measured
quantities such as the dry- and wet-bulb tempera-
tures.
The saturation locus and lines of constant
percent or relative humidity are of primary import-
ance on the humidity diagram. As discussed
previously, the saturation locus is directly related
to the vapor-pressure curve, and it will appear
roughly as shown in Figure 17.10. Of similar curva-
ture will be lines of constant percent saturation or
of constant relative saturation. If lines of constant
percent saturation are plotted, they will fall at
ordinates directly proportional to the ordinate of
the saturation locus at the same temperature. For
example if Ys = 0.10 at T u Y at T^ for the 30
percent saturation curve will be 0.03. If lines of
constant relative saturation are plotted, no such
direct proportion exists. This is evident from the
definition of relative saturation, Equation 17.4.
The relative-saturation curves will always be lower
than the equivalent percent saturation curves. Only
one of the measures of partial saturation is norm-
ally presented on a single chart.
In order to allow immediate determination of
the absolute humidity from wet-bulb temperature
Temperature (T)
Figure 17.10. Partial and total saturation lines on
humidity chart.
433
measurements, lines of constant wet-bulb tempera-
ture are usually plotted using Equations 17.26 and
17.27. If the /V Sc is constant through the range of
temperature and humidity covered by a single line
of constant wet-bulb temperature, Equation 17.27
shows that the line will be straight. In Chapter 9 it
was shown that for gases of fixed composition
3/2
both Qi and v vary approximately as 7" The .
mass diffusivity (£?) varies only slightly over a wide
concentration range. Thus, the variation in A/ Sc is
slight and may be considered negligible when satura-
tion occurs at low molal humidity. For the lines to
be straight and parallel, the ratio k y \ w lh c must be
constant throughout the entire range of T and Y
covered by the chart. Thus, /V Sc and \ w must
either be constant or vary in a compensating way.
For the air— water system, the lines are approx-
imately straight and parallel and are often so rep-
resented.
Curves connecting all points having the same
adiabatic-saturation temperature are also plotted
on the humidity chart and are useful in calcula-
tions concerning humidification and drying pro-
cesses. From Equation 17.21a, the slope of the line
connecting any point and its adiabatic-saturation
temperature will be —c h -[/\sa If c h is constant
.
over the range of temperature and humidity
covered by the locus of all points having the same
adiabatic-saturation temperature, c h l\ sa will also
be the slope of this locus. Generally, c h is not
constant but increases as the humidity increases.
Then, the locus of points of fixed adiabatic-
saturation temperature, usually called an
"adiabatic-saturation line," will be concave upward
as shown in Figure 17.11. In Figure 17.11, the
adiabatic-saturation temperature and the wet-bulb
temperature are shown for a single point [T^ Y^ ). ,
For the air-water system or any other system for
which the Lewis relation holds, the constant
adiabatic-saturation-temperature line and the con-
stant wet-bulb temperature line would be identical
(i.e., rw =
7"
sa ).
Information on humid volume cannot be
presented on coordinates of humidity versus
temperature, and so a separate volume-versus-
temperature plot is usually made. On this plot are
shown the specific volume of noncondensable com-
ponent and also the specific volume of saturated
vapor-gas mixture. From these curves, the humid
volume of any vapor-gas mixture can be deter-
mined from the percent saturation.
Vf, = Vdry + ( Vs - l/drv ) (percent saturation)
a
(17.35)
434 APPLICATIONS TO EQUIPMENT DESIGN

Temperature [T)
Figure 17.11. Constant wet-bulb- temperature and adiabatic-saturation-temperature
curves on a humidity chart.

where solution laws, and the reader should be especially

careful when the vapor concentration becomes
Vh = humid volume of moist gas at 7"i
large.
Vdry = specific volume of dry gas at 7"i
The enthalpy of a vapor-gas mixture may also
Vs = humid volume of saturated gas at 7^
be determined more directly from the humidity
For convenience, these curves are usually super- chart without use of Equation 17.36. This method
imposed on the humidity chart using a common uses the fact that the lines of constant adiabatic-
temperature scale but an independent volume saturation temperature are also lines of nearly con-
ordinate. stant enthalpy.
Enthalpy information is plotted separately in This fact can be shown by rearranging Equa-
the same way as is the volume information. As tion 17.21a
with humid volume,
Ch^+X^Y^c^Tv + XvYsa (17.216)
W-Wdry + (Hs — H drw ) (percent saturation)
and adding —ch yTQ to both sides of the equation
(17.36)
C/,1 (7", - r ) + \ M Y = c h ,(Tsa
:
- T ) + XsaYn
where the subscripts "dry" and s refer to dry and
(17.21c)
saturated conditions, respectively, and enthalpy all

values must be read at a single temperature. As The right side of Equation 17.21c equals the
with the volume plots, the enthalpy curves for dry enthalpy at the adiabatic-saturation point {Hsa ) if
7"
and for saturated gas are superimposed on the C/n equals the mean value of ch between and
humidity chart using the same temperature scale =
Tsa and if \sa An- Similarly, the left side of
but an independent enthalpy ordinate. Equation 17.21c equals the enthalpy at any point
Both Equations 17.35 and 17.36 apply rigor- on the constant adiabatic-saturation temperature
ously only to ideal solutions — that is, solutions for curve (H if c h ^ equals the mean c h between
: )
7"

which there no change in volume and no heat

is and 7"! =
and if again X sa X Thus, .

effect when the two components are mixed regard-

H^ *//M (17.37)
less of the mixture concentration. In most cases

where Equations 17.35 and 17.36 are used, the gas and the adiabatic-saturation curve would appear as
phase is a very dilute solution of components that a straight, horizontal line when plotted on
are chemically inert to each other, and the assump- enthalpy-temperature coordinates as shown in
tion of ideal-solution behavior is closely followed. Figure 17.12. The fact that an adiabatic process is

Gaseous solutions do not always follow ideal- being considered itself indicates that AH = 0.
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 435

l/l
p ture are plotted on coordinates of molal humidity
ra /•
oa
^. versus temperature in this chart. Since the chart is
& <?/
for the air-water system, the lines of constant
o
E s adiabatic-saturation temperature are also lines of
<$y "V constant wet-bulb temperature. The lines are
S \ Ao
Adiabatic- nearly straight and parallel and are spaced at even

>-
/"
saturation
curves temperature intervals along the saturation curve.
Q.
n
y Overlaying the basic chart are two others: one
e with coordinates of humid volume versus tempera-
Temperature (T) °F ture on which are plotted the volumes of dry and
Figure 17.12. Adiabatic-saturation curve on enthalpy saturated air, and one with coordinates of enthalpy
temperature coordinates. versus temperature on which are plotted the
enthalpies of dry and saturated air. The tempera-
The approximation in this equation arises from ture scale is identical for all three plots.
the same causes and is of the same order of magni- As with any presentation of thermodynamic
tude as the inaccuracy in assuming the adiabatic- data, the details of the chart depend upon the
saturation curves are straight and parallel. Using system represented, the units chosen, the bases
this relation, the enthalpy ofany vapor-gas mixture used, and the ranges of P, T, and Y shown. The
can be determined directly from the saturated- chart given in Appendix D-14 applies at a total
vapor enthalpy curve plotted on the humidity dia- pressure of 1 atm for the binary air-water system.
gram. Errors are insignificant at low vapor concen- Significant changes in total pressure or quantities
trations. of a third component such as C0 2 will prohibit the
A humidity chart for the air-water system at use of the chart. Units are given based upon 1 lb
atmospheric pressure is given in SI units in Figure mole of dry air, and enthalpies are calculated using
17.13 and in English units in Appendix D-14. The bases of H = for dry air at 32° F and 1 atm total
saturation locus, lines of fixed percent humidity, pressure and for liquid water at 32° F and at its
and lines of constant adiabatic-saturation tempera- triple point pressure. Humidity charts have been

100,000
2800 0.28
/]
0.09
/ /
1 1 i 1

/ 1/ 80,000

60,000

2400 0.24 0.08 50,000

40,000
.§/ §/
30,000
2000 o 0.20 - 0.07
E £7 / /

. /
/ / //^ 1 £7
HI IX}'Jl
1
20,000
<$ /
Zo 1600 0.16 - 0.06
<V / It}
E 1/ / ^
/

9/
1
/
/

/ 3>
A /

1 "&
± 10,000

1200 C-0.12 _
*y^ T—J/£
/
/ /£>
/ /&
~^
—/S\°
£
1 O ~
/-c - 0.05
8000

6000
7 5000

800 0.08 — ^/ / / 0.04 4000

As V-^4£|a6a(, c satl /
/ 3000
' ,0/
400 0.04 ~y a<r___ .
0.03
^^Sp. v olume_of_dry^ 2000
0.02

—- 1
1 - 0.02
10 20 30 40 50 60 70 30 90
1000
Temperature (T) °C

Figure 17.13. Psychrometric chart for air-water system, atmospheric pressure,

SI units.
436 APPLICATIONS TO EQUIPMENT DESIGN

published for systems other than air-water (15), adiabatic saturation temperature curve passes
with bases different from those chosen here (16), through the 80 percent saturation curve at 99°
with significant quantities of a third component and Y = 0.0540. These conditions will be those of
present (17), and with different variables used as the air leaving the drying chamber.
major coordinates. The chart given in Figure 17.13
is for the same system at the same total pressure
Example 17.2. Determine the enthalpy,
and using the same bases for H = for both dry air relative to dry air at 0°C and saturated water at
and water. Here quantities are given based upon 1 0°C, of air at 80°C, 10 percent humidity.
mol of dry air and its associated water, and all
values are given in SI units. Solution The desired answer can be read
Various representations of the information in directly from the humidity chart Figure 17.13. At
the humidity chart have been developed for special the conditions given, the adiabatic saturation
convenience in doing particular calculations that temperature is 47°C. At 47°C the enthalpy of
must be performed repeatedly. The broad applica- saturated air is read to be 6400 joules/mol of dry
bility of the arrangement used here to a variety of air.

chemical engineering calculations has made this This answer can be checked by finding the
chart a favorite since Grosvenor proposed it (6). In enthalpy using the proper latent and specific heats.
many humidification and drying problems, the For any vapor-gas mixture,
convenience of the unit "moles of vapor per mole H = c b (T-T + Y[c.(T-To) + \
) ]
of dry gas" and the ease with which material
balances may be evaluated for changing quantities (17.10)
of vapor in a constant quantity of dry gas over-
Here Y
= 0.088 mol H 2 0/mol dry air. Then c b the ,

shadow other considerations. The use of the specific heat of air, may be taken as 29.08 joule/
humidity chart is illustrated below. mole°C; ca the specific heat of water vapor, as
,

33.89; and \ as 45,070 joule/mol. Substituting

Example 17.1. Air enters the drying chamber into Equation 17.10 gives
of a tray dryer at 210°F after having been heated
from an ambient condition of 70° F and 50 percent H= 29.08(80 - 0) + 0.088[33.89(80 - 0) + 45,070]
relative humidity. If the air leaves the drying = 6531 joules/mol of dry air
chamber at 80 percent humidity as the result of an
adiabatic-saturation process within the dryer, what Note that the difference between these two
is the temperature and humidity of this exhaust answers is only a little greater than 1 percent, even
air? though the conditions were chosen in the high-
humidity region where Equation 17.37 is no longer
Solution The humidity of ambient air must
rigorously correct.
firstbe determined, since this humidity will not
Still a third method of determining this
change in the heating process. From the humidity
enthalpy is through use of the curves for the
chart (Appendix D-14) Ys = 0.0262 at 70°F. Then
enthalpy of dry and of saturated air as indicated by
from Dalton's law,
Equation 17.36. At 80°C, the enthalpy of satur-
ated air is 44,000 joule/mol of dry air, and that of
0.0262 = Pa = 0.0256 atm the dry air is 2300 joule/mol. The enthalpy of air
1 -Pa
at 10 percent saturation is then
and from Equation 17.4a, pa = 0.5 x 0.0256
= 0.0128 atm. Then from Equation 17.6, H= 2300 + (44,000 - 2300) x 0.10
= 6470 joule/mol dry air
0.0128/0.9872
Percent saturation = x 100
0.0256/0.9744 CALCULATIONS FOR HUMIDIFICATION AND
= 49.3 percent DEHUMIDIFICATION OPERATIONS
From the humidity chart, this saturation gives Y = The design of equipment to carry out humidifica-
0.0129 which will be constant through the heating tion and dehumidification operations depends
process. At the entrance to the drying chamber, upon the concepts and equations developed in
Y = 0.0129 and T = 210°F. The adiabatic satura- Chapters 15 and 16. As discussed there, the direc-
tion temperature will be 93° F, and the constant tion of the transfer processes and their rates can be
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION
Molal humidity (Y)-
Figure 17.14. Conditions in dehumidifier.
determined by writing enthalpy- and material-
balance equations and rate equations and by com-
bining these equations into the applicable design
equations. In this section, these concepts will be
applied to the special conditions surrounding the
operation of humidification and dehumidification
equipment. The development is exactly parallel to
that used earlier for heat -transfer and mass-transfer
equipment.
In a dehumidification process, a warm vapor-
gas mixture is contacted with a cool liquid. Vapor
condenses from the gas phase, the gas phase cools,
and the liquid is warmed. Both sensible and latent
heat are transferred toward the liquid phase. These
conditions are shown schematically in Figure
17.14. In the normal convention, the gas phase is
chosen as the system, and heat- and mass-transfer
rates and quantities are positive when transfer is
toward the gas phase. This sign convention is equiva-
lent to measuring distance in the direction toward
the gas phase. Then, heat- and mass-transfer quan-
tities for this physical situation will be negative.
Gas phase
transfer
Figure 17.15. Conditions in a cooling tower
(humidifier).
437
In a water-coolingprocess (Figure 17.15),
warm water contacted by a cooler gas-vapor
is
mixture. Mass and heat are transferred toward the
gas phase. As a result, the signs of the driving-force
terms and transfer rates are positive based upon the
convention described above.
Cases may be met where the sensible heat and
latent heat transfer in opposite directions. This
often occurs at the lower end of water-cooling
towers where the water has cooled below the bulk-
gas temperature. This overcooling is possible
because the latent-heat transfer far overshadows
the sensible-heat transfer. The liquid cannot cool
below the wet -bulb temperature of the gas, as was
shown above. This case is shown in Figure 17.16.
In these operations, rates and quantities of both
heat and mass transfer are significant; enthalpy
balances and heat-transfer rate equations must
therefore be written parallel to the material bal-
ances and mass-transfer rate equations. Therefore,
the analysis of a general transfer process previously
done in Chapters 15 and 16 will be repeated here,
but transfer of both heat and mass will now be
considered simultaneously.
The nomenclature and physical arrangement,
are shown in Figure 17.17. The conditions here
differ from those of Figure 16.5 in that the liquid
phase is presumed to be a single component,
identical with the solute in the gas phase. As
before, the subscripts 1 and 2 refer to the bottom
and top of the column.
=
Z.2 liquid flow rate into the top of the
column, lb mole/hr (or mol/s)
V-\ = flow rate of gas phase entering the column,
lb mole/hr (or mol/s)
V'=i\ow rate of solvent, or "dry" gas,
lb mole/hr (or mol/s)
Gas phase
Molal humidity (Y)
Temperature (Tv )
Figure 17.16. Conditions for latent- and
sensible-heat transfer in opposing directions.
438 APPLICATIONS TO EQUIPMENT DESIGN

'-
I •
H i I
V Y H Vi .V 2
Most commonly, the column will operate
nearly adiabatically with q * 0. The approach to
adiabatic operation will be closer the larger the
column diameter. For this situation, similar bal-
ances for the differential height {dz) will be
written. The condensable component balance
becomes

VdY = dL (17.41)

The corresponding enthalpy balance is

VdH v =d(LHL ) (17.42)

If the rate of solute transfer between phases is

small compared to the total flow stream, an
average value of L may be used, and the change in
enthalpy of the liquid phase may be expressed as if
it resulted solely from the change in temperature at

constant specific heat. Thus,

Figure 17.17. Nomenclature for the general

d(LH L = ) L av c L dTL 117.43)
humidification process.
L^+L 2
where
Y 2 = mole ratio of solute to solvent gas at the
top of the column
For the change in gas-phase enthalpy, the expres-
HV y
= enthalpy of gas phase entering the
sion in terms of temperature is rigorous if c h is
column, Btu/lb mole of "dry" gas
(or joule/mol)
constant.

H L2 = enthalpy of liquid phase entering the top V'dH v = V'd[c h (T v -T )+ YX ]

of the column, Btu/lb mole of liquid
(or joule/mol)
= V'ch dT v +V'\odY (17.44)

q = heat transferred to the column from its

Rate equations for heat and mass transfer can also
surroundings, Btu/hr (or joule/s)
be written. Here, however, complications arise
TL , T v = temperature of liquid and gas phases,
because of the fact that heat is transferred from
respectively
the bulk of the liquid phase to the liquid-gas inter-
dz = a differential height of column packing,
face entirely as a result of the temperature poten-
ft (or m)
= interfacial surface, ft 2 (m 2 ) tial, but from the interface to the bulk of the gas
A
phase, heat is transferred by two mechanisms. On
a = interface area, ft 2 /ft 3 of column volume
(or m 2 /m 3 of column volume) the gas-phase side of the interface, heat is trans-
ferred as the result of a temperature potential, and
S = tower cross section, ft 2 (or m2 )

the latent heat associated with the mass transfer is

On this basis, for a tower of constant cross section, transferred as the result of a concentration driving
an overall material balance gives force. The quantities of heat transferred by these
two mechanisms are separated as the two terms on
L, -L 2 =V, V- 17.38)
the right side of Equation 17.44.
With these ideas, equations for the heat-
A material balance for the condensable component
transfer processes indicated in Equations 17.43 and
gives
17.44 can be written separately. For the liquid-
V(Y 2 - Y,)=L 2 -L, (17.39) phase transfer,

and an enthalpy balance gives

-^c L dTL =h L a(TL - 7",) dz (17.45)
L 2 H L2 + VH V +q = L)HL: , + V'H V2
iJ

(17.40) where 7", = the interracial temperature, °F

 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 439

For the gas-phase sensible-heat transfer, Equation 17.51 is a design equation similar to
Equations 14.62 and 14.63 except that the perti-
V
-ch dT v =h c a{Ti-T v )dz (17.46) nent driving force is expressed as an enthalpy dif-
o ference. Enthalpy is an extensive thermodynamic
property. As such, it cannot be a driving force for
and, for the gas-phase latent-heat transfer,
any transfer operation. Thus, the mathematical
treatment that leads to Equation 17.51 should be
V
\ dY=\ k Y a(Y i -Y)dz {MAI) examined. First, in the energy balances, the total
flow per unit time is fixed. The enthalpy terms are
where then "specific enthalpy," based upon a fixed mass
of material. Second, in the basic design equation,
Y, = the gas-phase mole ratio of solute to
Equation 17.49, the driving force is a function of T
solvent at the interface and Y, the quantities that would be expected to
control rates of heat and of mass transfer. Only in
Development of the Design Equation the fortuitous case that r= 1 can H be substituted
The enthalpy-balance equations and the rate equa- for these T and Y functions. In all other cases,
tions given above can now be combined to give a Equation 17.51 would be written as
design equation in the form of Equation 14.61.
V'd(ch Tv + XoY)
These equations will relate the change in gas-phase
Sk Y a[ich rT + - (c h rT v + X Q Y)]
-/*
temperature and molal humidity to the rates of i 'KoYi)
heat and of mass transfer to, or from, the gas
(17.51a)
phase. Thus, combining Equation 17.44 with Equa-
tions 17.46 and 17.47,
Integration of the Design Equation
V
— Integration indicated by Equation 17.51 is
dH v = h c a(Tj -T v )dz + \ Q k Y a(Yj - Y) dz
usually performed by using values of and k Y a V
o
averaged over the column height. This introduces
(17.48) small error in light of the low concentration of
for the gas phase. Separating k Y a from the right water vapor in the gas stream. Beyond this, know-
side of the equation and designating h c lk Y c h asr, ledge of the relation between the enthalpy in the
the psychrometric ratio, give main gas phase and that at the gas-liquid interface
is necessary. Such a relation can be obtained by
V - now considering the transfer process on the liquid
YdH v = k Y a[(ch rT i
+ \ Yi ) (c h rT v + \ Q Y)] dz
side of the interface. Combining the enthalpy
balance (Equation 17.42) with the liquid-transfer
(17.49)
rate (Equation 17.45) gives

—V dH v =h
In putting r into this equation for /? c a/Ayac h the ,

assumption is made that a, the area per unit L a(TL - Tj) dz (17.52)
volume of tower, is the same for heat transfer as it
is for mass transfer. This will be true only at such
and combining this equation with Equation 17.50,
high liquid rates that the tower packing is com-
pletely wet. If r is equal to 1, as it is for the hL a H v -H,
(17.53)
air— water system under normal conditions, the kYa TL — Tj
terms within parentheses in Equation 17.49 are
enthalpies as defined by Equation 17.10. Equation 17.53 applies at any point in an
air— water contacting device. From it, the tempera-
V
— ture and the enthalpy of the interface can be
dH v = k Y a(Hi-H v )dz (17.50)
determined at any point for which the liquid
o
temperature (77.), the gas enthalpy [H v ), and the
or ratio of the liquid-phase heat-transfer coefficient to
V dH v the gas-phase mass-transfer coefficient based upon
"1/2
/h
V2 Sk Y a{H, - Hv )
-/>
J
o
mole-ratio driving forces are known.
The conditions can be obtained
interface
17.51) through Equation 17.53 using a graphical method.
440 APPLICATIONS TO EQUIPMENT DESIGN

A plot is drawn with coordinates of liquid-phase 2400

temperature versus enthalpy of the gas phase. On
2200
it, the locus of interface H,- and 7/ values can be
plotted by realizing that at the interface the vapor
phase will be saturated at the interface temperature
if the assumption that equilibrium exists at a phase

boundary is tenable. From the saturation curve on

the air— water humidity chart, the saturation molal
humidity can be obtained for any desired tempera-
ture. The saturation, or interface, enthalpy can
then be calculated or read from the humidity
chart.
On the same plot, an "operating line" of H v
versus TL can be plotted by combining Equations
17.42 and 17.43 and integrating. This curve repre-
sents the path of bulk-phase conditions as the C(TlA .HvO
600
fluids pass through the unit. Thus,

400
70 80 90 100 110
/ V'dH v = I L av c L dTL (17.54)
Liquid temperature <TL or T )
t

Figure 17.18. Graphical representation of adiabatic gas-

where the limits again refer to the bottom and top liquid contacting operation.
of the column. Integrating,

V'(H V 2-H V :) = L aw c L (TL2 -TL ,) have the values of TLn and H Vn The interface .

conditions at this point can be found by using

(17.55) Equation 17.53. Thus, a tie line starting at point B
and rearranging, and having a slope equal to —h L a/k Y a will inter-
H v2 — Hy\ sect the equilibrium curve at the interface condi-
Lav c L
(17.56) tions corresponding to point B. Point / represents
Tl2 — TL i V the interphase conditions designated by (Tin Hin ). ,

This equation gives the slope of the H v versus TL In this way, interface conditions can be found

"operating" line as L av c L IV'. For the air— water corresponding to any point between A and Con
system and for most other dilute gas-solution the operating line. Of course, values of k Y a and
systems, this ratio is constant over a moderate hL a applicable to the column conditions are
humidity range, and the straight H v -versus- TL line necessary. Usually, these values must be obtained
can be determined from knowledge of the liquid- experimentally.
and gas-phase flow rates and the conditions of both From this construction,, Equation 17.51 can
streams at one end of the column, or alternately now be integrated graphically. The necessary
the line can be located from the conditions at both driving force in terms of enthalpy determined for is

ends of the column. representative points throughout the column. For

Figure 17.18 shows such a diagram for a point B on Figure 17.18, this would be (/-/,-„ —
humidification operation. The equilibrium curve H Vn ). Values of V/Sk Y a(H; - H v are then )

represented on it was obtained from data on the plotted against H v and the area under this curve
air— water psychrometric chart, Appendix D-14. determined between the limits of H v and H V2 in -
[

On this curve are located all possible conditions of order to find the column height (z). Solutions of
T,, Hj throughout the column. Line ABC is the
this sort in terms of temperature and concentration

operating line containing all values of H driving forces have been carried out in Chapters 14,
v cor-
responding to liquid temperature (TL) throughout 15, and 16.

the column. This line could have been obtained

from knowledge of the two end conditions, Example 17.3. A water-cooling tower of the
(T^ Hvi and [Ti_ 2 H V2 ), or from either one of
1 , ) , wood-slat, forced-draft, countercurrent-flow
these two plus the slope {L av c L /V). On this line, variety cools water from 45°C to 25°C when the
point B represents an arbitrary point in the column ambient air is at 32°C with a wet-bulb temperature
at which the liquid temperature and gas enthalpy of 15°C. Previous experience with towers of this
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 441

design leads the engineer to predict that h L al

A>a = 600 joule/mol°C and that kya (mol/hr) = 14,000

0.2 VIS. A liquid rate of 900 mol/hr m 2

of tower
cross section will be used.
(a) Determine the minimum gas rate. At this
rate, there will be a zero driving force somewhere
in the column, and an infinitely high tower will be
necessary. Usually, at minimum gas rate, the exit
gas will be in equilibrium with the inlet water.
(b) Determine the necessary tower height if a
gas flow rate twice the minimum found in part (a)
is used.

Solution This problem can be solved using

the graphical method outlined above. For this, an
enthalpy-temperature plot must be made, and the
locus of interface conditions must be plotted on
this diagram. The plot is made from Figure 17.13
by replotting the curve of saturated air enthalpy as
a function of temperature.
The inlet and outlet known conditions are
shown on Figure 17.19. The enthalpy of the moist
30 40
air can be determined from the humidity diagram
TL or r„ "C
by taking advantage of the fact that the adiabatic-
Figure 17.20. Graphical representation of driving forces.
saturation curves are lines of nearly constant
Example 17.3.
enthalpy. Thus, the inlet air enthalpy is the same as
that of saturated air at 15°C. From Figure 17.13,
sect the equilibrium curve at one of these end
H V y = 1200 joules/mol dry air. The conditions at conditions. Thus, at the top of the column, the gas
the bottom of the tower {H V ,TL -\) can now be y
stream is still not saturated at the liquid tempera-
plotted and are shown at point Con Figure 17.20.
ture. The resulting operating line is Cfi'on Figure
The H v TL operating line will be straight and
,
17.20
have a slope of L aw c L /V. For the minimum gas
From the slope of this line
rate called for in part (a), this line will be tangent
to the equilibrium curve at some point between L^c L 5900-1200
inlet and outlet column conditions, or will inter- = 235 joule/mol °C
I/'
•'mm 20
and

V' 900 x 75.31

= 288.4 mol/hr m2
235
| = 900 mol/hr m2
which is the answer to part (a).
~ = 600 joule/mol °C For a gas flow rate = 2V' m in V , VIS = 576.8
kYa =0.2 V'/S (mol/hr ft
2
)
mol/hr m 2 The slope of the operating
. line is

235
= 118
V
and the line CB results. The
position of point B
givesH V2 =3550, but nothing about T V2
tells .

From H V2 the value of [TW V2 the exit wet-bulb

) .

TVI = 32 °C TLl = 25 "C

temperature, equals 34°C.
Twl = 15 °C Since the operating-line position is fixed, tie
(H vl = 1200 (joule/mol) lines running from any point on the operating line

Figure 17.19. Column conditions for to the equivalent interface condition on the equi-
Example 17.3. librium curve can be plotted. From Equation
 442 .APPLICATIONS TO EQUIPMENT DESIGN

0.007

: 906
Equilibrium curve
IA= 11.02

5. 0.005

^ 0.004

0.003
Liquid -phase temperature TL
Figure 17.22. Minimum gas rate with zero driving

0.002 force between the column ends.

1000 1400 1800 2200 2600 3000 3400
ffa joule/mol corresponding to point E, the outlet gas stream
Figure 17.21. Graphical integration of the design equation. from an infinitely tall column operating at mini-
Example 17.3. mum vapor rate would be saturated at TL2 .

In Example 17.3, integration of Equation

17.53, these tie lines will have the slope = 17.51 to get the required tower height was done
—h L alk Y a = —600. Such tie lines are shown as CC,, graphically. In contrast to other rate operations,
PPj.QQj, RRj, and BB, on Figure 17.20. the analytical integration of the design equation
Corresponding values of H v and H, can now for humidification is not very difficult. The reason
be taken from the diagram, and the design equa- is that all information necessary for the solution is

tion integrated graphically, as shown in Figure given in equation form except the equilibrium
17.21. curve. Because this curve is always the same for the
air— water system, it can be represented by one
TL Hv H, (H, - Hv )
empirical equation to any desired degree of
Skya(Hj - Hv )
accuracy for all humidification problems. For
instance, the relation
25 1200 2000 800 0.00625
30 1800 2650 850 0.00588 = 2011 - 2
/V, 107.1 7", + 4.51 7", (17.57)
35 2400 3400 1000 0.00500
40 3000 4250 1250 0.00400 represents the equilibrium curve between 15 and
45 3550 5250 1700 0.00294
50°C within about ±3 percent, or the equation
From which z - area under curve 1A = 1 1 .02 m.
- 63.857", + 0.598 7", 2 (17.57a)
Hi = 2295

The situation occurring as the vapor rate used gives a similar representation in English units
in Example 17.3 is decreased toward the minimum between 60 and 120°F. Using Equation 17.57 or
rate should be examined again. As the vapor rate
17.57a along with the tie-line equation
decreases, the operating line, originally CB, moves
toward the position of CB' As it does so, the .
hL a Hy/ — Hj
(17.53)
column height would have to be increased to allow kYa 7V. -7}
the required transfer to occur despite the smaller
driving forces. Eventually, at the minimum vapor and the equation for the operating line,

rate, an infinitely tall column would be required. Hu-H v\ ^-av c L

Even for such a tower, the outlet air would not be (17.56)
saturated at the liquid temperature. The situation
TL -TL y
V
is emphasized in Figure 17.22, where the outlet the design equation,
gas-phase enthalpy is shown at point D for a finite
column and at D' for an infinite one. Of course, if
•h v2 V'dH v
Az = [17.51)
the outlet water temperature were reduced to that hV i
Sk Y a{Hj- Hv )
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 443

can be written in terms of a single variable and Substituting the last two equations into the design
formally integrated since four equations are avail- equation results in
able in the variables H,-, H v T,-, and TL The most , .

1I
r
Wi/ 2 = 3550
r"V2-' (81.07 + 1.48477) </77
convenient solution is obtained by writing the
design equation in terms of 7", and then integrating. 0.1 ''W l/1 =1 200 3143 -188.277 + 3.7687/ :

In contrast, in gas absorption and heat transfer,

(g)
rarely does the same equilibrium condition apply
to more than one problem. Thus, the graphical The indicated integration can be performed after
solution is usually easier than the derivation of an the limits are expressed in terms of 7}. Solving
analytical expression for the equilibrium curve. Equation (e), Tn = 23.65°C and 7, 2 = 41.78°C.
The integration of Equation (g) can be done in
two parts by breaking the integral into two separate
Example 17.4. Solve Example 17.3(b)
integrals of the forms
analytically using the results obtained in Example
17.3(a). dx 2cx+b
tan"
Solution The conditions imposed by a + bx + ex 2
vAac — b 2
VAac - b 2
Example 17.3 result in the following numerical
and
substitutions into Equations 17.53, 17.56, and
17.51: x dx
Hv -Hi a + bx + ex 2
-™=T L -T,
(a)
dx
^-1200= 118(77. -25)
= —
1
In (a + bx + ex 2 ) - —b c
;

(b) 2c 2c J
a + bx +
•"1/2 dH v Inserting the proper numbers and solving gives z =
-/ .94 meters.
HV y 0.2(Hi-H v )
1 1

The answers obtained column height

for the
Combining Equations (a) and (b) and solving for
in Examples 17.3 and 17.4 by about 8 per-
differ
Hj yields
cent. This is the sort of precision one could expect
H,-H v = 6O0(TL -7» (a) from a graphical integration based upon Figure
17.13 and the use of a crude representation of the
#1,-1200 = 11877. -2950
equilibrium curve. Probably the answer obtained in

=
Hy+ 1750 Example 17.4 is more precise, but both results are
77 (b) well within the precision of the knowledge con-
118
cerning the transfer coefficients. Other large un-
Combining (a) and (b) and simplifying: certainties include nonuniform flow distribution,
ability to predict operating conditionsand require-
///„+ 1750 \

T>) ments, unsteady-state effects, and inaccuracy in

*-"*-»(—fii humidity measurements. These errors and uncer-
tainties do not show up here, but may be much
Hi = 6.08H V - 6007, + 8898 (d)
more damaging to design performance capability
Now this equation can be combined with the equa- than are the errors involved in these calculations.
tion of the equilibrium curve (Equation 17.57) to
obtain Hv as a function of 7", Overall Coefficients

201 1 - 107.17" +4.51 T/

r
2 = 6.0SH v -600Ti + 8898 As shown earlier (see Equations 14.49 and 14.50),
the individual mass-transfer coefficients may be
Collecting terms
converted to overall coefficients if the equilibrium
Hv = -1132 + 81.077", + 0.74277 2 (e) curve is straight. In this case, an overall driving
force must also be used. In Figure 17.18 the ver-
From Equations (e) and 17.57
tical distance BD represents such an overall driving

Hj - H v = 3143 - 188.27",- + 3.7687",- 2 force. Point D(TL ,H*) represents saturated vapor
at the temperature of the bulk liquid and is
and differentiating Equation (e) gives
identical in concept to the "equilibrium equiva-
dH v = (81 .07 + 1 .48477) dT-, (f lent" used in mass-transfer work. The appropriate
444 APPLICATIONS TO EQUIPMENT DESIGN

rate equation is terms of gas-phase resistance and driving force.

Thus,
—V dH v = KY a(H* - H v ) dz (17.58) V'dH v
= H c NG =z (17.51/?)
Sk Y a(Hj - H v
for constant V' and KY a )

h V2 dH v SK Y a Determination of Bulk-Gas Phase Temperature

:17.59)
/.
HV y H* - H v V It was shown earlier (Example 17.3 and Figure
17.22) that for simultaneous heat and mass-
As with the equilibrium equivalent for mass transfer operations, as well as with mass-transfer
transfer, point D(TLn H*) is not, in general, the
,
and with operations, tangency or
heat-transfer
interface condition at a point in the column where intersection operating and equilibrium lines
of
the bulk-phase conditions give point B(TLn H Vn ); ,
represents a point of zero driving force and hence
nor does it represent the liquid-phase conditions can be attained only with an infinitely large con-
equivalent to some known gas-phase condition. It tactor. This zero driving force condition then forms
represents an imaginary interface condition where one limit on the range of feasible operating condi-
the interface temperature is that of the bulk-liquid tions.
phase at the same point in the column TLn Of .
Another, and often more limiting, restriction
course, the enthalpy is that of saturated air at this on the range of practical operating conditions
temperature. occurs when fog forms in the vapor phase. Fog will
If the liquid-phase resistance to heat transfer
form when the bulk-gas phase becomes super-
is very small compared to the gas-phase resistance
saturated. The carrying of water droplets by the
to mass transfer, the actual interface temperature
gas phase disrupts the basic material and energy
will approach the bulk-liquid temperature. The
balances with respect to V and L' and hence
slope —h L alk Y a approaches — °°, and point / on
makes the mathematical procedures outlined here
Figure 17.18 approaches the equilibrium equiva-
useless. Physically, the fog represents a serious in-
lent of B located at point D. For this special
convenience. If fog forms, separation of the drop-
situation, K Y a = k Y a, and the use of the overall
lets from the main gas stream is costly and incon-
mass-transfer coefficient is rigorously correct.
venient. Water losses are high in a water-cooling
Where a significant liquid-phase resistance to operation; in a dehumidification operation, the
heat and mass transfer exists, the interface tempera- major process aim is thwarted. With systems other
ture will differ significantly from the bulk-liquid than the air— water system, the presence of fog
phase temperature. In other words, points / and D may be a serious health hazard or economic loss.
will be separated as shown on Figure 17.18. For Prevention of fog formation is the main reason that
this case, the use of an overall mass-transfer co-
air fed to a contact sulfuric acid plant must be
efficient with an appropriate enthalpy-difference
dried.
driving force(H* — H v ) is correct only if the inter- On the Hv — T diagram used here, the bulk-
face locus between / and D straight.
is
gas phase temperature has not been shown. If a
Usually, the use of K Y a brings in negligible
curve of Tv Hv
could be plotted on a
versus
error, since hL a is usually large compared to k Y a diagram such as Figure 17.18, the intersection of
and the curvature of the interface locus is not
this curve with the interface locus would represent
great. The overall coefficient (K Y a) has the
a fogging condition; that is, the bulk gas phase
practical advantage of being easily measured.
would reach saturation. If no intersection occurs,
The left-hand side of Equation 17.59 is the
the column would be operable. Any point on the
number of transfer units defined on p. 389 and is
T v — H v curve would indicate a gas-phase condi-
exactly analogous to similar parts of Equations
tion occurring in the column corresponding to a
16.14 and 16.15. Rearranging Equation 17.59 gives
liquid-phase temperature read from the TL — H v
V curve at the same value of H v Without plotting
z
-^7J7F^rv
dHy
- H °° N°° mm such a T v — H v curve, a design may be inadvert-
ently developed in which the column may fail as a
.

which is identical to Equation 16.14 except that an result of fogging conditions even though it would
enthalpy, instead of a mole-fraction, driving force otherwise be operable.
is used. To determine the gas-phase T v — H v plot,
Equation 17.51 is the equivalent equation in Mickley (13) developed a convenient graphical
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMID I FIC ATI ON 445

method. Dividing Equation 17.46 by Equation Hv curve will equal the slope of the straight line
17.50 gives from point F to the interface condition cor-
- Tv dz responding to points C and F The interface condi-
V'ch dTv h c a{T, )
.

(17.61) tion can be determined from Equation 17.53, using

V'dH v k Y a(Hi-H v )dz the ratio of coefficients to get the tie line from a
point on the TL — H v curve to the corresponding
or
interface point. The stepwise procedure is as
dT v hc a — Tv
( Tj follows, with the letters referring to points on
(17.61a)
dH v k Y ac h \Hj-H v ! Figure 17.23:

1. From Equation 17.53, get point D. Draw

but, by the Lewis relation, h c a/k Y ach = 1, and
DF. As long the interface conditions are con-
as
therefore
stant FGD will represent the path of the gas-phase

dTv T, - 7\ A7\ conditions as indicated by Equation 17.62. Let this

(17.62) suffice for an arbitrary short distance FG.
dhi v Hi - Hv A/A 2. The operating-line conditions correspond-

where A but finite difference. If

refers to a small
ing to G will be at H. Thus, TH will be the liquid
temperature at the point in the column where TG
the gas-phase conditions at either end of the
column are known, a stepwise method can be used
is the gas-phase temperature. From H, determine /

to draw the locus of gas-phase conditions through-

by Equation 17.53. Draw line IG, and arbitrarily
presume this to be the gas-condition curve to point
out the tower. The procedure is illustrated in
J.
Figure 17.23 where the situation is the same as was
Repeat. The construction determines the
3.
shown in Figure 17.18. Warm water enters the
points alphabetical order. Points C, H, K, N, Q,
in
tower at TL2 and leaves cooled to TL -\. This is
T, and A fall on the operating line, whereas points
done using a countercurrent airflow that enters the
F, G, J, M, P, S, V, and Z are on the gas-phase
bottom of the column at H v ^, T v The air
-\.
condition locus. Point Z concludes the gas-phase
temperature may be greater or less than TL but^ ,
condition locus at the gas-outlet enthalpy.
here Tv ^ >TL ^. It must be greater than T, at
HV i . By Equation 17.62, the slope of the Tv — Obviously, this construction accumulates
errors due to the approximation inherent in using
finite differences rather than differentials in Equa-
Interface
condition locus
N>v/ tion 17.62. Each line segment of the curve FZ
represents a line tangent to the true gas-phase con-
dition locus.
at
u/ As the line segments FG, GJ, JM, etc., are
"O
Operating shortened, becoming closer to true differentials,
"o R /l\ ne
<u / \ the errors become less, and the correct gas-
o
E \Z\ \,A\TLZ MV1
condition locus is approached. In any case, the
\

- °x
curvature of the gas-condition locus found by using
3 l/ \
X lv\ \ / finite differences is less than the true curvature of
;£"
l/\ris~\ this line when the calculations are started from one
DA X "
Vtd
end of the column and is also less than the true
a, [\ /
curvature when the calculations are started from
c
n \ \\a vrV*'
the opposite column end. Thus, one way to
m
TO
\ \ /*"^v approach the true gas-condition locus closely is to
a.
\ \l ^^g carry out the stepwise solution from both ends of
\/H
o
o.
>(0
C[TL1 ,H VI )
/—
£-
y^s.

Gas phase
condition locus
Fl7Vl "n
the column and then draw a final locus with a
shape curving between that of the two stepwise
solutions.
Equations 17.61a and 17.62 state that the
slope of a tie line extending from a point on a
Temperature (TL ,TV or T, , ),°F phase-condition locus (here the vapor phase) to the
Figure 17.23. Graphical construction of gas-phase- equivalent interface condition is equal to a con-
condition locus. stant times the slope of the phase-condition line
446 APPLICATIONS TO EQUIPMENT DESIGN

itself. This statement is markedly different from Water 7;,= 60 F

o

that given by previous equations involving the 1 L'/S = 1000 Ib/hr ft

2

—
i

slope of a tie line as, for example,

Equations
17.53 and 14.48. In each of the previous cases, the
slope was related to the ratio of appropriate co- ^- = 200
efficients. For instance, in Equation 17.53, the
enthalpy driving force for the gas phase divided by
Air
the temperature driving force in the liquid phase t
equaled the inverse ratio of gas-phase mass-transfer T, =100°F 90% humidity
3 2
coefficient to liquid-phase heat-transfer coefficient. 27,000 std ft /hr ft

Equation 17.61 shows no such ratio. If it did, the Figure 17.24. Conditions given for
gas-phase operating line would have to be straight, Example 17.5.

since both h c a and k Y aare usually nearly constant

throughout the column. The basic difference lies in and
the quantities related by these different equations.
Equation 17.53 relates the total heat-transfer rate Lc, 1000x18
= 14.1
through the vapor phase by the mechanisms of V 70.6 x 18
convective heat transfer and mass transfer to the
total heat-transfer rate through the liquid phase. These conditions are plotted in Figure 17.25. The
Equation 14.48 for a mass-transfer situation location of point (7~
L1 , H vy ) and the slope (Lc L /
equates the total mass-transfer rate through the V) determine the location of the operating line,
liquid phase to that through the gas phase. In and with the inlet water temperature given they
contrast, Equation 17.61 or 17.62 relates the rate also fix H V2
= 1325. Thus, the air-outlet wet-bulb
U
of heat transfer through the vapor phase by the temperature is 88 F. The air-outlet temperature is
mechanism of convection to the total rate of heat determined with a Mickley solution starting from
transfer through the vapor phase by convection (7Vi, H V ). Note that, in this case, the gas-phase
y

and by latent-heat transfer associated with the condition locus actually crosses the equilibrium
mass transfer. The vapor-phase coefficients are curve. Fogging would occur in this column, and an
involved but are themselves related through the exact answer is impossible. The outlet air would
psychrometric ratio (h c a/k Y ac h = r). Thus, the probably be saturated at 88° F, giving a humidity
gas-phase condition locus differs fundamentally of Y 2 = 0.0467 lb mole/lb mole dry air.
from the operating lines used in heat-transfer,
mass-transfer, and simultaneous heat- and mass- The method outlined above breaks down for
transfer calculations. the case of an adiabatic saturator. In this case, the
operating line reduces to a point located on the
equilibrium curve. The gas phase T v — H v line
Example 17.5. A packed tower is to be used becomes a straight line at constant enthalpy mov-
to dehumidify air as a part of an air-conditioning ing from the inlet air condition toward the equilib-
system. Air enters the tower at 100° F, 90 percent rium curve. Under these conditions. Equation
humidity, and is contacted with water entering at 17.46 can be integrated directly since T, is con-
60" F and leaving at 90° F. Flow is countercurrent stant and equal to TL The . integration gives
with a liquid rate of 1000 Ib/hr ft 2 and a gas rate
of 27,000 std ft 3 /hr ft 2 of column cross section. V2 h c az
The ratio of coefficients (h L a/l<Ya) is expected to In (17.46a)
be 200. Determine the outlet air conditions.
TL -T V , ch V'IS

Determination of Coefficients in Operating Equipment

Solution The given conditions are shown in
Figure 17.24. From Appendix D-14, the humidity The stepwise construction of Mickley may be done
of inlet air is Y = 0.0629 mole/mole dry air and in reverse to determine the rate constants,
:

H vy = 1738 Btu/lb mole dry air. Also and h L a) from a single set of test data.
(k Y a,h c a,
From the inlet and outlet bulk gas- and liquid-

V 27,000 1
= 70.6 moles/hrft 2
phase temperatures and gas-phase humidities, the
end points of the operating line and the gas-phase
359 1.0629 condition line are fixed. The operating line can be
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 447

200C

lTu .H n )

Slope = —
LC L

(TL2 Mvi>

Equilibrium curve

200

50 60 70 80 90 100
Temperature (T^TiTv). °F

Figure 17.25. Graphical solution to Example 17.5.

immediately drawn. The gas-phase condition curve bulb. leaves with a 96°F dry-bulb temperature
It

can be obtained by assuming a value of —h L alk Y a and 95° F wet-bulb temperature. The liquid flow
a
2
and plotting the curve stepwise. If this curve does rate is 900 Ib/hr ft and the tower is 8 ft high.
,

not meet the experimental end condition, a new (a) Determine the values of h L a, h c a, and
value of —h L alkya must be chosen. Once a proper kya obtained in the column.
—h L alkya ratio has been found, the "driving" (b) If the gas inlet temperature rose to 140°F

force" {H, — H v ) is read from the construction, with a 111°F wet-bulb temperature, what would
and the design equation in the integrated form be the air outlet conditions? Water inlet tempera-
ture and air and water flow rates would not
dH v Sk y az change. Use the coefficients from Part (a).
/Hi - H V'
:17.51:
v
Solution (a). The column conditions are
is solved directly to give k Y a. This value of kya shown in Figure 17.26. From the gas wet-bulb
then gives h L a through the previously determined temperatures, the enthalpies of inlet and outlet gas
ratio —h L alkya. Finally, the Lewis relation can be streams can be found. From Appendix D-14, the
used to get h c a. As before, these solutions presume enthalpy of inlet air is H v = 2500 Btu/lb mole of
^

that the areas for mass and heat transfer are equal. dry and the enthalpy of outlet dry air is
air,

H V2 = 1600 Btu/lb mole of dry air.

Example 17.6. A dehumidifying tower Points A and B on Figure 17.27 can now be
packed with wood slats operates with water enter- plotted, and the TL — H v operating line can be
ing at 82° F and leaving at 100° F. The air comes in drawn between them.
at a temperature of 125°F dry bulb, 1 1 1°F wet The slope of this line (AB) is 900/18 = 50
448 APPLICATIONS TO EQUIPMENT DESIGN

TV; = 96*F Wa ter 82° F = TL2 Btu/lb mole. Since

Tu2 = 95'F Gi=900lb/hrft"
900
x 18

Slope = 50 = —
Lc v V 18
=
-;
z 8ft S 50
= 18lbmoles/hrft 2

Moist air TLi = 100'F or

>
TVI = 125°F
T„, = lll'F
2
Gv '
= 18 x 29 = 521 lb of dry air/hr ft
Figure 17.26. Column conditions
for Example 17.6. of column

3500

3250

3000

2750 -

250C -

2250 -
h a
.

-r-=
n, a
10
2000

1750 -

r
D D
1500 T V2 = %'F
(Tu ) v , = 95'F

1250

1000

750

50C

250

90 90 100 110 120 130

Temperature (TV, TL . 7]J. *F

Figure 17.27. Mickley method solution of Example 17.6(a).

 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDI F1C ATION 449

Now pointsC and D, representing and inlet From this, h L a and h c a are
outlet air temperature and enthalpy,
can be
h L a = 275x4.10
plotted, and the bulk-gas condition locus connect-
= 3
ing points C and D can be found. This requires a 1 1 70 Btu/hr (ft of column
trial-and-error construction in which various slopes volume) °F
of the tie lines are tried. A convenient starting
h c a = ch x k Y a
point might be to try h L alkya = <*> giving vertical
= 7.60x4.10
tie lines such as AF'. Using these vertical tie lines
= 31.2 Btu/hr 3
with the Mickley stepwise construction, the curve (ft of column
CD' is obtained. Since D' is not coincident with D, volume) °F
the ratio h L alkya is not <». Trying h L alkya = 1.0,
where h c a has been found using the Lewis relation.
the curve CD" is obtained from tie lines parallel to
(b). Part (b)second, almost
requires a
AF". The correct solution is obtained with tie lines independent solution. The known conditions are
parallel to AF. These tie lines give curve CD with shown in Figure 17.29. The data available do not
the construction shown. The slope of AF is —275. fix the operating line since liquid temperature and
Thus h L a/kya = +275. gas enthalpy are not both available at either end of
Interphase conditions corresponding to the the column. But the number of transfer units is
bulk conditions throughout the tower are now fixed as well as the slope of the operating line and
known. That is, where the liquid temperature plots the slope of the tie lines. Thus, the operating line
at M the interphase temperature plots at G. This can be located by a trial-and-error procedure. Lay-
permits the solution of Equation 17.51. ing down parallel lines of slope Lc L IV = 50
1
between 7"
Z 2 =82°F .
and Btu/lb #^=2500
TL Hy Hi Hv - H,- mole, the N G is found as was done in part (a). The
correct operating line will be found when NG =
100 2500 1940 0.00178 1 .821 This line will then determine TL and H V2
. .
t
97 2350 1770 0.00173
A Mickley stepwise solution will then fix T V2 .

94 2200 1630 0.00176

Figure 17.30 shows the temperature-enthalpy
91 2050 1500 0.00182
plot. The equilibrium line is plotted on it, as are
88 1900 1370 0.00189
85 1750 1250 0.00200 limiting lines at H vy =2500 and TL2 = 82°F. The
82 1600 1150 0.00223 operating line shown is a first trial. This line runs
from TL2 to Hv Lc L IV' = 50. Tie
-\ at a slope of
The Equation 17.51 [fdH v /(H v - /-/,)]
integral of lines of slope — h L alkya = —275 are drawn in.
is the area under the curve of M{H V — H,) versus Resulting determination of N G requires graphical
H v as shown in Figure 17.28. This is found, and
, integration.

2500 dH v _= Sk Y a Az tl Hv H Hv ~ H
1

Nn = 1.821 = i i T, 77
-/, 600
1 H v — Hj V
from which 82 1550 1130 420 0.00238
86 1750 1290 460 0.00217
1.821 x 1

k Ya = = 4.10 lb moles/hr (mole ratio) 89 1900 1420 480 0.00208

8 3 92 2030 1560 490 0.00204
ft of column volume
95 2200 1700 500 0.00200
0.0023 98 2350 1850 500 0.00200
101 2500 2000 500 0.00200

The actual integration out in Figure

is carried
17.31. The = 1.980, which is
resulting area is NG
too large. Subsequent locations of the operating
line will be parallel to that shown but farther from
the equilibrium curve.
The final operating line is also shown in
Figure 17.30. From \\ H V2 = 1600, TL
0.0017
= 100°F.
1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600
,

H
v Btu/lb mole dry air
, The Mickley stepwise construction results in an air
Figure 17.28. Determination of transfer units in dehumidi- outlet temperature of 98°F. From H V2 ,the outlet
fication. Example 17.6(a). wet-bulb temperature is found to be 95° F.
450 APPLICATIONS TO EQUIPMENT DESIGN

Water are few. The lack of data is mostly the result of

7] .2 = 82°F some rather formidable experimental difficulties.
2
3,. = 600 Ib/hr ft In long columns with large contacting areas, the
outlet vapor will be essentially saturated. There-
h, a
T^
Ky Q = 275 fore, the driving forces are small, and small errors
Az = 8 ft
inconcentration are magnified several fold. The
ATG = 1.821
remedy is to use short columns. Here, though, the
i I end effects get large and can only be corrected by
Air
taking several sets of data with varying column
TV. 140T height and extrapolating the result back to zero.
ffy, = 2500 Btu/ b r nole dry air
The method is rife with areas for error.
V'/S = 12 lb moles/hr fr
If the end-effect correction is not made, com-

Figure 17.29. Conditions of
Example 17.6(b).
Thus, the Mickley construction forms a tool
by which rate coefficients may be easily deter-
mined for full-size operating cooling towers and
dehumidifiers. Unfortunately, almost no data of
this sort have been reported in the technical litera-
ture.
If coefficients obtained
operating-plant equipment
designer must use coefficients obtained on experi-
mental or pilot-plant equipment or coefficients
obtained on the basis of analogies between heat,
mass, and momentum transfer. Despite the
simplicity of humidification systems where all the
mass-transfer gradient occurs in the gas phase, data
mercial towers based upon the experimental results
will be oversized if they are shorter than the
experimental column but undersized if they are
longer. This is probably one of the reasons for the
generally held belief that increasing the height of a
packed tower decreases its efficiency. Another
cause of this effect is the fact that bed porosity at
the walls is always higher than that at the center of
the tower. Therefore, the liquid tends to flow
by measurements on down the column wall instead of being distributed
are unavailable, the evenly throughout the packing. This channeling
effect is greater in tall columns, and elaborate
redistribution systems have sometimes been used.
The mass-transfer coefficients for humidifica-
tion are not fundamentally different from those
obtained on absorption equipment. Only gas-phase
mass-transfer coefficients are pertinent. As with

3000

2750 -
First operating
S Slope = ^ = 50

2500
line trial ~~
^^y^/ / y Hn = 2500

•*"
T = 140'F
Final /Y
operating-^ // \

E
line //\
€ 2( -

4; -
1750 - // \ \ V
-

TV2 = 98'F
•e. 1500
- Tie-line slope = -h L a/kya = -275

1250

1000 -
"-.
-
TL2 = 82°F

750
90 100 no 120 130 140
Temperature, °F

Figure 17.30. Graphical solution to Example 17.6(b).

 SIMULTANEOUS HEAT AND MASS TRANSFER- HUMIDIFICATION
0.0025
0.0020
0.0019
0.0018
1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500
Hv
Figure 17.31. Determination of operating-line location
Example 17.6(b).
other gas-liquid mass-transfer coefficients, they will
depend upon the gas flow rate, viscosity, and dif-
fusivity and upon the geometry of the packing.
They will also be dependent upon liquid flow rate,
primarily because at low liquid rates the packing is
not completely wet. Thus, gas-absorption mass-
transfer rates, such as those calculated from Equa-
tion 16.74, should be usable for humidification or
dehumidification operations if the geometries are
identical.
Part of the mass-transfer data available from
experimental work are summarized in Table 17.2
(21). Some understandable inconsistencies among
data taken from a variety of sources are present.
Also, since the relations shown are empirical results
of experimental data, they should not be extra-
polated beyond the range of the data without full
consciousness of the uncertainty that results.
A similar situation exists in relation to appro-
priate heat-transfer coefficients. For the air— water
system at high liquid rates, h c a may be calculated
through the Lewis relation. At low liquid rates,
incomplete wetting of the packings makes the area
for heat transfer (a h considerably larger than that
)
for mass transfer (a M). The result is that measured
lkyaM may be two or three times as
values of h c a h
large asc h depending upon the liquid flow rate (7).
The liquid-phase heat-transfer coefficient may not
be so readily calculated. Here, experimental data
on the packing to be employed must be used, or
analogies between heat and mass transfer must be
utilized.
451
, Btu/lb mole of dry air
in dehumidification.
The methods discussed here apply to the air-
water system or to any system where h c al
k Y ac h = 1. If this restriction cannot be met but
h c alkYac h = r, the development must be based
upon Equation 17.51a. Instead of enthalpy, the
graph would have (c h rT+\,Y) plotted against T
and used as the "driving force" in the rate-equation
solution. Simplified procedures for this general
case are available (10). Such procedures retain the
graphical method used here but adjust the
interface-locus curve to account for the fact that r
is not unity.
Natural-Draft Cooling Tower Design
In the design of natural-draft cooling towers, such
as those shown in Figures 17.1 and 17.2, the design
is complicated by the fact that the gas velocity is
not directly controlled. The warm, humid air leav-
ing the tower packing is less dense than the cool,
dryer air that enters the packing. Thus, it is dis-
placed up through the hyperbolic chimney by the
force of gravity acting upon the air surrounding the
tower. Airflow velocity thus depends upon the air
density decrease caused by the transfer processes
occurring in the tower and by the tower height. Of
course, the extent of the transfer processes, in
turn, depends upon the airflow velocity.
The development of the simultaneous equa-
tion set describing 1) the effect of humidity and
temperature on air density, 2) the heat and mass
transfer in the column packing (Equations 17.51,
17.56, and 17.57), and 3) the momentum transfer
452 .APPLICATIONS TO EQUIPMENT DESIGN

Table 17.2 EXPERIMENTAL MASS-TRANSFER RATES FOR CONTACT OF A PURE LIQUID WITH A GAS
(All data are given in English units)

Item Equipment Process and Notes Range of Flow Rates Equation Reference

8-in. diam. tower, 1-in. Liquid cooling with Gl 1600 k a =

g
0A86G L °'i5 G v OJ2 20
Raschig rings airand water, Gv 150-500
methanol, ben-
zene, ethyl
butyrate k a
g
corrected for end
effectsand
independent of
G for G[_ =
i_

1600-5000

GL 02
10-in. diam. tower; Humidification = 200-4160 /r
/ a = 0.0155G l/ G L 25
12.5 in.depth of cooling, liquid
15-, 25-, and 35-mm cooling, and
Raschig rings dehumidification
with air-water

21.5-in. square tower, Humidification G, = 120-6800 k ya = 0.0431 G,/' 039

li-in. Berl saddles cooling with air-

water; corrected
for end effects

4-in. diam. tower, i-in. Humidification IpGy-ln 304-927 ^G^-

13.4 /0„G,/'i 01 22
spheres cooling with air- Gl 85-1895
water; corrected
for end effects

6-ft square tower, 1 1 -ft, Liquid cooling with GL = 350-3000 K Ya = 0.0068G L ° A G V '° 5 1

3-in. packed height; air-water

wood slats, | x 2 in.,

spaced parallel, 1 5 in.

between tiers

41|- x 23g-in. tower, Liquid cooling with GL = 880-1500 K ya = 0.00001G L G^- 19

packed height 4lf in. air-water
-o.oosgeGy-
wood slats, k x 2 x
23j bottom edge
in.,
-0.00458G/. + 10.7
serrated, on f-in.
horiz. centers, 3|-
to 2|-in. vert,
centers; 1 8 spray
nozzles

Spray tower, 31.5 in. Liquid cooling and GL = 300-800 0.0526G i 18

diam. by 52 in. high; dehumidification Noc
Gv,'°-58
6 solid cone spray with air-water
nozzles

Perforated plate (sieve Humidification cooling Gu- = 670-1920 Liquid depth, in Nq 23

tray); 83§-in. holes with air-water 0.5 ca. 1.5
on |-in. triangular 1 2
centers 2 2.5

After Treybal, R. E., Mass Transfer Operations, McGraw-Hill, New York (1955), by permission. Individual items
with permission of publisher and copyright holder.
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDI FICATION 453

equations determining airflow velocity (see 21. Treybal, R. E., Mass Transfer Operations, McGraw-Hill,
Chapter 20) and their solution have occupied many New York (1955), p. 190.
engineers. Computer capabilities now allow these 22. Weisman, J., and E. F. Bonilla, Ind. Eng. Chem., 42,
simultaneous, iterative solutions to be done p. 1099 (1950).
23. West F. B., W. D. Gilbert, T. Shimizu, Ind. Eng. Chem.,
numerically and semiquantitatively. The general
44, p. 2470(1952).
approach is well described in the literature
24. Wood, B., and P. Belts, Proc. IME, 163, p. 54 (1950).
(3,4,5,12,24).
25. Yoshida, F., and T. Tanaka, Ind. Eng. Chem., 43,
These design methods require the use of
p. 1467 (1951).
transfer equations such as those given in Table
17.2. Few data are available on packing perform-
ance in full-sized cooling towers. Data have been
presented on splash-grid cooling tower packing (9). PROBLEMS
However, data on more modern, often proprietary,
plastic packing material are closely held by tower 17.1. An air-conditioning system is to be built to
manufacturers. change completely the air in a laboratory every 10 min with
no recirculation. The laboratory building has two floors, each
50 by 200 by 10 ft and is to be supplied with air at 75°F,
REFERENCES 40 percent saturation. The air-conditioner takes outside air,

refrigerates it, separates out the condensed water, and then

1. The Fluor Corp., Ltd., Bulletin T337 (1939); also, reheats it in an exchanger against condensing 50 psig steam.
Perry, Chemical Engineers' Handbook, 3d ed.,
J. H., (a) When the outside air is at 100°F, 90 percent
McGraw-Hill, New York (1950), p. 791 relative humidity, determine the volume of air brought in
2. Bedingfield, C. H., and T. B. Drew, Ind. Eng. Chem.,
to the air-conditioner and the temperature to which it must
42, p. 1164 (1950).
be cooled.
3. Chilton, H., Proc. Inst. Elect. Eng., 99, Part II (71),
(b) For the same conditions, determine the tons of
p. 440 (1952).
refrigeration required (1 ton of refrigeration = 12,000
4. Furzer, I. A., British Chem. Eng., 13, No. 9, p. 1287
Btu/hr removed) and the pounds per hour of saturated
(1968).
5. Furzer, I. A., Ind. Eng. Chem. Proc. Des. & Devel., 7, steam used.
p. 555(1968). 17.2. A tray drier takes in ambient air at 90°F with a

6. Grosvenor, W. M., Trans. A.I.Ch.E., 1, p. 184 (1908). 75° F wet-bulb temperature, heats it to 220° F along with
7. Hensel, S. L, and Ft. E. Treybal, Chem. Eng. Progr., some recycled air in a steam-heated exchanger, and passes it

48, p. 362 (1952). over the wet stock. The stock dries from 50 percent mois-
8. Kelley, R. C, paper presented before the California ture by weight to 10 percent moisture and may be assumed
Natural Gasoline Association, December 3, 1942. to enter and leave the drier at the wet-bulb temperature of
9. Kelly, N. W., and L. K. Swenson, Chem. Eng. Prog., 52,
the drying air. After leaving the drying chamber, part of the
p. 263(1956).
air is recycled back to the steam-heater inlet; the rest
10. Lewis, J. G., and R. R. White, Ind. Eng. Chem., 45,
discharged. If 20 lb of dry air is used per pound of dry
p. 486 (1953).
11. Lichtenstein, Trans. ASME.65, 779 (1943). stock handled and the exit air is at 135°F, determine
J., p.
12. Merkel, F., V.D.I. Forschungsarbeiten, Berlin, No. the fraction of moist air that is recycled and the percent

275(1925). humidity of the outlet air.

13. Mickley, H. S., Chem. Eng. Progr., 45, p. 739 (1949). 17.3. A mixture of air and methanol at 1.5 x 10 5
14. Norman, W. S., Trans. Inst. Chem. Engrs., London, 29, Newtons/m 2 and 50°C is 60 percent saturated. Calculate
p.226(1951). (a) the humidity, (b) the molal humidity, (c) the partial
15. Perry, J. H., Chemical Engineers' Handbook, 5th ed., pressure of methanol, (d) the relative humidity, (e) the
McGraw-Hill, New York (1973), pp. 20-7-20-8. volume percent methanol, (f) the humid heat in SI units,
16. Perry, J. H., Chemical Engineers' Handbook 5th ed.,
(g) the humid volume in SI units, and (h) the dew point.
McGraw-Hill, New York (1973), pp. 12-3-12-5.
Assume the mixture is an ideal solution obeying the perfect
17. Perry, Chemical Engineers' Handbook, 5th ed.,
J. H.,
gas law.
McGraw-Hill, New York (1973), pp. 12 - 6.
17.4. For the mixture of Problem 17.3 calculate the
18. Pigford, R. L., and C. Pyle, Ind. Eng. Chem., 43,
wet-bulb temperature.
p. 1649(1951).
19. Simpson, W. M., and T. K. Sherwood, Refrig. Eng., 52, 17.5. Prepare a psychrometric chart for the air—
p. 535 (1946). ethylene-dichloride system at 1000 mm Hg total pressure.

20. Surosky, A. E., and B. F. Dodge, Ind. Eng. Chem., 42, Cover temperatures ranging between and 200 F; mass
p. 1112(1950). ratios between and 1 lb vapor per pound dry air. Use air
454 APPLICATIONS TO EQUIPMENT DESIGN

The follow- The tower handles 1000 3 /min of airand

and liquid C2H4CI2 at 0°F as reference states. 6-ft -diameter std ft

ing curves should be plotted against temperature. is operated at 1.5 times minimum water flow. In a similar

(a) Absolute humidity at 100, 75, 50, and 10 per- tower, hi_lky has been found to be 60. Determine, for a

cent humidity. height of packing of 1 .5 ft:

(b) Absolute humidity at constant adiabatic- (a) Outlet water temperature.

saturation temperatures of 70, 100, and 130 F. (b) Tower height.

(c) Absolute humidity at a constant wet-bulb (c) Water flow rate.

temperature of 100 F. (d) Outlet air temperature and humidity.

(d) Dry and saturated humid volumes. (e) Whether fogging will occur in the tower. If fog-

(e) Dry and saturated humid heat. ging occurs how should the conditions be changed
(f) Enthalpy of dry and saturated mixtures to eliminate it?

expressed as Btu/lb dry gas. 17.11. A natural-draft cooling tower uses tower
17.6. Prepare a psychrometric chart for the air- packing (as described in Table 17.2, item 6) that is 4 m
methanol system at a pressure of 4 x 10 5 Newtons m2 . high. On one summer day the water to be cooled is sprayed
Include lines of constant relative humidity, adiabatic satura- onto the packing at 49°C whereas the air flowing up the
tion temperature, and wet-bulb temperature. Cover a tower enters at 29 C. Originally the air is at 30 percent
temperature range from C to 100 C. saturation, and its flow rate is measured at 0.35 m/s. The
17.7. A pan of water sits on the ground exposed to water flows through the packing (60 m tower diameter) at a

the sun. rate of 5000 /

/s. The ratio of coefficients (hj_alkya) is

(a) In the early morning the air is at 70 F with a expected to be 630 watt s/mol°C. Calculate the tempera-
65°F wet-bulb temperature. The water is at 50 F. ture of the cooled water, and the fraction of the water fed

(b) At 10 A.M. the air is at 80°F with a 70°F to the tower that is lost by evaporation.
wet-bulb temperature. The water is at 70°F. 17.12. An engineer is called upon to design a coke-
packed dehumidifier that 3
(c) At noon the air is at 90 F with a 75 F wet-bulb will handle 2000 ft of 70 per-

temperature. The water is at 120 F. cent saturated air per minute. The airstream will enter the
For each case draw a diagram showing temperature tower at a temperature of 130°F and should be cooled until
and humidity on both sides of the air-water interface (like the humidity of the exit airstream is 0.046 lb water per
Figure 17.14). Show the direction of water-vapor transfer pound dry air. The operation is to be conducted at atmos-

and of sensible and latent-heat transfer. pheric pressure. Cooling water is available at 73° F and can
17.8. (a) In Table 17.2, item 7, data are given on the rise to 90°F. It is agreed to use a gas velocity of 1200 lb of

performance of a spray humidification tower. Assuming dry air per hour per square foot of tower cross section and

that mass- and heat-transfer rates are uniform throughout a water feed rate of 249 Ib/min. Under these flow condi-
the length of the tower, convert the equation given to tions, h c a = 300 and h L a = 1 500.
Kya = f(Gi_, Gy). (Probably this assumption of uniform (a) Determine the height and diameter of the tower

rate is a poor one. Indications are that mass transfer is most required to do the dehumidification.

rapid in the drop-formation area.) (b) After the tower of part (a) was constructed, it is

(b) A spray tower is to be used in a water-cooling discovered that the engineer used the incorrect entering air

operation. Air at 140°F with 80° F wet-bulb temperature humidity. Instead of the 70 percent saturation as specified,

will be contacted countercurrently with a water spray the engineer read the humidity chart at a 60 percent satura-

initially at 140°F to cool the water to 100°F. Gy will be tion.

400 Ib/hr ft
2 and G\_ will be twice this. If the Kya relation 1 How will this error affect the tower design?
obtained above is valid, what column height will be 2. Can the constructed tower be made to operate to
required? do the specified dehumidification? How?
17.9. Table 17.2, item 5, gives an equation for the 17.13. A nuclear power plant produces 1100 mega-
performance of a square cooling tower with wood-slat pack- watts of electric power at an overall efficiency of 33.3
ing. Convert this equation to one in the form Nog = percent; that is, for every watt of electricity produced, 3
a watts of heat energy produced by the nuclear reaction,
f{Gi_ , G\/' ) where G^ and Gy' are in SI units. Make the is

assumption that the tower performance is not variable with and 2 watts are dissipated as heat into cooling water. Under
its height. design conditions, the cooling water enters the condenser at
17.10. A dehumidifier consisting of a tower packed 20°C and is heated to 40°C. It then flows to natural-draft
with 1-in. Raschig rings contacts air entering at 120°F, 80 cooling towers where it is recooled to 20 C, and makeup
percent saturated, countercurrently with water entering at water is added as necessary. Ambient conditions are taken

70 F. The air is to leave at a wet-bulb temperature of 80°F. as air at 25°C with a wet-bulb temperature of 12 C. Tower
 SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIHCATION 455

packing m high of a proprietary design drip-point style,

is 10 current forced draft of air entering at 1 10°F and 20 percent
for K G a = (2.2 x 10 _5 G L + 5.3 x 10" 2 )G
which 8
l/
-

. saturation. Measurements indicate that the air leaves at

Tentative designs call for Gi_ = 3.4 kg/s m and Gv = 96 Fwith a 94° F wet-bulb temperature.
2
2.75 kg/s m . The plant manager wishes to cool the water to as cold
(a) What is the required number and diameter of a temperature as possible. One possible means would be to
cooling towers? increase the airflow rate, and toward this end it is found
(b)Can the cooling requirement be met at any value that fan speed can be increased without overloading the fan
of hi_a/kya? If not, what minimum tower height would be motors so that airflow is 1.5 times that previously used.
required if -h^alkya = °°? If so, what value of h\_alkya Tower flooding will not occur at this higher gas rate.
would be required? What will be the outlet water temperature attained
(c) How much makeup water is needed? with the higher air rate, and what will be the air outlet
17.14. A cooling tower with wood-slat packing cools condition?
1000 gal/min of water from 105°F to 90°F using a counter-
 Eighteen

Simultaneous Heat and Mass Transfer — Drying

As used in this chapter, the term drying applies to
the transfer of liquid from a wet solid into an
unsaturated gas phase. The removal of moisture
from gases, also frequently referred to as drying, has
been discussed under the names of dehumidification
and adsorption.
The factors discussed in relation to humidifi-
cation and dehumidification and the mass- and
heat -transfer rate equations written in Chapter 17
apply also to drying operations. The drying
process, then, is identical with a humidification
process except for the influence exerted by the
solid itselfon the drying process. This influence is
considerable. The study of drying and the calcula-
tion of required dryer size must take into account
a host of problems in the areas of fluid mechanics,
surface chemistry, and solid structure as well as the
transfer rate problems dealt with in the discussion
of humidification. In many cases, these physico-
chemical phenomena are so complicated and so
incompletely understood that the quantitative
design of the dryer is impossible. For example, in
the drying of timber some of the liquid is held
within the wood fibers. This moisture can migrate
into the drying air only by diffusion through the
fiber walls. Since this moisture diffusion through
the wood is slow, the wood may dry nearly com-
pletely at the surface before the liquid escapes. The
uneven drying can cause cracking and warping of
the wood. The whole problem has been attacked
empirically so that conditions can be set to dry
timber successfully, but the basic mechanism of
liquid movement remains doubtful. Again, in spray
drying of detergents and of many other materials,
the required drying time depends to a large extent
upon the size of droplets sprayed into the hot
chamber. Any atomizing device produces a range
of droplet sizes that, of course, dry to different
extents under the same conditions. In the deter-
gent drying, exposure of the droplet to the drying
medium produces a tough, partly dry shell around
the droplet that inhibits moisture escape. Heat
flows readily through the shell, however, and
evaporates the liquid inside the droplet. The vapor
produced swells the droplet, often rupturing it and
sometimes even blowing new "balloons" on the
sides of the original droplet. Thus, the dried
particles are mixtures of hollow spheres, sphere
fragments, and clusters of hollow spheres. The size
distribution cannot be readily predicted from the
size distribution of the spray itself. The rate of fall
of the particles through the hot air, and therefore
the drying time, is very difficult to predict, and the
initial droplet-size distribution is itself only
occasionally calculable. Again, in drum drying
of
such products as milk and baby cereal, the slurry
coats a hot drum and adheres to it during drying.
The thickness of the drying layer is a function of
the surface tension and cohesive characteristics of
the slurry, but the exact relationship is still un-
known. Of course, the thickness controls the
extent of drying produced and hence the final
moisture content. These factors take on added
importance because, in general, the dried solid is
the valuable product. Its shape, color, stability,
stickiness, and hence its salability all depend upon
the drying process to which it has been subjected.
GENERAL DRYING BEHAVIOR
In drying a wet solid with a gas of fixed tempera-
ture and humidity, one general pattern of behavior
always appears. Immediately after contact between
the sample and the drying medium, the solid

456
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING
temperature adjusts until it reaches a steady state.
The solid temperature and the rate of drying may
increase or decrease to reach the steady-state con-
dition. At steady state, a temperature probe would
find the temperature of the wet-solid surface to be
the wet-bulb temperature of the drying medium.
Temperatures within the drying solid would also
tend to equal the wet-bulb temperature of the gas,
but here agreement would be imperfect because of
lag in movement of mass and heat. Once these
stock temperatures reach the wet-bulb temperature
of the gas, they are found to be quite stable, and
the drying rate also remains constant. This is the
so-called constant-rate drying period. The period
ends when the solid reaches the critical moisture
content. Beyond this point, the surface tempera-
ture rises, and the drying rate falls off rapidly. The
falling-rate period may take a far longer time
than the constant-rate period, even though the
moisture removal may be much less. The drying
rate approaches zero at some equilibrium moisture
content which is the lowest moisture content
obtainable with this solid under the drying condi-
tions used. Figures 18.1 and 18.2 show typical
drying curves, one on the basis of moisture content
versus time and the other on the basis of rate of
drying versus moisture content. The moisture-
content— time plot (Figure 18.1) is the form in
which drying test data might be obtained. Figure
18.2, the rate-of-drying— moisture-content plot, is
much more descriptive of the drying process. It is
obtained by differentiating data in the form of
Figure 18.1 and thus is subject to considerable
scattering of data and resulting uncertainty.
These typical drying curves are related to the
mechanism by which drying occurs. The drying
Time (0), hr
Figure 18.1. Typical drying curve for constant drying
conditions, moisture content as a function of time.
457
X'E
Moisture content (X'), mass liquid/mass of dry solid
Figure 18.2. Typical drying rate curve for constant dry-
ing conditions, drying rate as a function of moisture
content.
period represented by segment AB
of the curves of
Figure 18.1 and Figure 18.2 the unsteady-state
is
period during which the solid temperature reaches
its steady-state value. Although the shape shown is
typical, almost any shape is possible, and AB may
occur at decreasing rate as well as the increasing
rate shown. During the constant-rate period (seg-
ment BC of the drying curves of Figures 18.1 and
18.2), the entire exposed surface is saturated with
water. Drying proceeds as from a pool of liquid
with the solid not directly influencing the drying
rate. It is possible that the roughness of the solid
surface over which the liquid film extends may
increase mass- and heat-transfer coefficients, but
this effect has not been firmly established. The
surface temperature reaches the wet-bulb tempera-
ture, as would be expected. The constant-rate dry-
ing regime continues with the mass that is trans-
ferred from the surface continuously replaced by
movement of liquid from the interior of the stock.
The mechanism of liquid movement and conse-
quently the rate of this movement vary markedly
with the structure of the solid itself. With solids
having relatively large, open void spaces, the move-
ment is likely to be controlled by surface tension
and gravity forces within the solid. With solids of
fibrous oramorphous structures, liquid movement
is by diffusion through the solid. Since the dif-
fusion rates are much slower than the flow by
gravity and capillarity, solids in which diffusion
controls the liquid movement are likely to have
short constant-rate periods, or even to dry without
a measurable constant-rate period. At point C, the
moisture content of the solid is barely adequate to
supply the entire surface.
458 .APPLICATIONS TO EQUIPMENT DESIGN
During the drying period between points C
and D of Figure 18.2, called the "first falling-rate
period," the surface becomes more and more
depleted in liquid because the rate of liqiud move-
ment to the surface is slower than the rate of mass
transfer from the surface, until at point D there is
no significant area of liquid-saturated surface. The
part of the surface that is saturated dries by con-
vective transfer of heat from and mass to the
drying gas stream. Vapor from lower levels in the
sample diffuses to the part of the surface that is
not saturated and then continues its diffusion into
the gas stream. This mechanism is very slow com-
pared to the convective transfer from the saturated
surface.
At moisture contents lower than that at point
D of Figure 18.2,all evaporation occurs from the
interior of the solid. As the moisture content
continues tofall, the path for diffusion of heat and
mass grows longer, and the concentration potential
eventually decreases until at XE '
the equilibrium
t
moisture content, there is no further drying. The
equilibrium moisture content is reached when the
vapor pressure over the solid is equal to the partial
pressure of vapor in the incoming drying gas. This
period is called the "second falling-rate period."
Gasses of Materials According to Drying Behavior
Materials may be divided into two major classes on
the basis of their drying behavior. Granular or
crystalline solids that hold moisture in the inter-
stices between particles or in shallow, open surface
pores constitute the first of these classes. In these
materials, moisture movement is relatively un-
hindered and occurs as a result of the interplay of
gravitational and surface tension or capillary
forces. The constant-rate period continues to
relatively low moisture contents. Although the
falling-rate period divides into the two regions
mentioned above, it usually approximates a
straight line on a rate-versus-moistu re-content
basis. The which is usually inorganic, is
solid itself,
relatively unaffected by the presence of the liquid
and therefore is unaffected by the drying process.
As a result, drying conditions can be chosen on a
basis of convenience and economic advantage with
little concern over the effect of the conditions on
the properties of the dried products. In the case of
hydrates, the drying conditions will affect the pro-
duct by changing the hydrate obtained, but other-
wise the materials are not affected by the drying
conditions over wide temperature and humidity
ranges. Examples of this class of materials would
be crushed rock, titanium dioxide, chrome yellow,
catalysts,zinc sulfate monohydrate, and sodium
phosphates. For these substances, equilibrium
moisture contents are usually very close to zero.
Most organic solids are either amorphous,
fibrous, or gellike and constitute the second of the
major classes. These materials hold moisture as an
integral part of the solid structure or trapped
within fibers or fine interior pores. In these sub-
stances, moisture movement is slow and probably
occurs by diffusion of the liquid through the solid
structure. As a result, the drying curves of the
substances show only very short constant-rate
periods, ending at high values of critical moisture
content. For the same reasons, the first falling-rate
period is much reduced, and most of the drying
process is liquid-diffusion controlled; that is, the
drying rate is controlled by the rate of diffusion of
liquid through the solid. The bulk of the drying
occurs in the second falling-rate period. The
equilibrium moisture contents are generally high,
indicating that a significant quantity of water is
held so intimately in the solid structure, or in such
fine pores, that its vapor pressure is significantly
reduced. Because the water present is such an
intimate part of the solid structure, the solids are
affected by moisture removal. The surface layers
tend to dry more rapidly than the interior. If the
drying rate is high, it may cause such differences in
moisture content through the sample that warping
or cracking occurs. In other cases, it may cause
formation of a relatively impervious partly dry
shell that further inhibits interior drying and may
accentuate the unevenness of moisture content
through the sample and consequent tendency for
solid deterioration. Because of these reactions, the
conditions under which drying occurs are critical.
The conditions must be chosen with primary con-
sideration for the effect of the conditions on
product quality, and process economy or operating
convenience must be subordinated. Examples of
such materials include eggs, detergents, glues,
soluble coffee extract, cereals, starch, animal
blood, and soybean extract.
Moisture Movement - Diffusion Mechanism
In relatively homogeneous solids such as fibrous
organics, gel like substances, or porous cakes, mois-
ture probably moves toward the surface mainly by
molecular diffusion. The rate of moisture move-
ment is then expressed by Ficks's law (Equation
11.13) altered to apply to this particular case.
2
d X'
= 9, (18.1:
dd dx 2
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 459

where ing drying, so that yL * is seldom constant. More-

over, the moisture distribution at the critical moist-
Q>
L
* = liquid-phase diffusion coefficient appli-
ure content is seldom uniform. For some sub-
cable for movement through the solid
phase, ft
2
/hr
stances such as wood and clay, the distribution has
been found to be nearly parabolic, and solutions of
The integration equation requires that
of this Equation 18.1 for this boundary condition have
boundary conditions be chosen and that the been given (30).
characteristics of ®t * be specified. For the simplest
case, JL
(
would be considered constant, and dry-
* Moisture Movement — Capillarity Mechanism
ingwould occur from one face of a slab, the sides For beds of particulate solids or for substances
and bottom of which are insulated. Applying these with a large open-pore structure, the molecular
restrictions and assuming the initial moisture diffusion mechanism is obviously incorrect. For
content to be evenly distributed through the slab, these materials, movement of liquid within the
Sherwood (30) and Newman (23) obtained solid results from a net force arising from differ-
ences in hydrostatic head and in surface-tension
X'-XE '
8 , -QlMtY 1 -«*t »(?
#
'
effects Surface tension causes the pressure
(3).
+ -e i

Xr '-XF '
7T
2 1 under a curved liquid surface to be different from
that of a flat surface. For a sphere of radius r, it
can be shown (15) that
+ — e " + . . 18.2)
25
-AP = —7
2
18.3)
where r

= distance from face to center of a slab

/
where
drying from both faces, or total thick- —AP= decrease in pressure caused by surface-
ness of a slab drying from one face, ft tension effect, lbf/in. 2
XE = equilibrium moisture content, lb of
'

7 = surface tension for contact between

liquid/lb of dry solid liquid and gas phases, Ib^/in.
X'= moisture content at time 6, lb of r = radius of curvature of sphere, in.
liquid/lb of dry solid
Xq = moisture content at the start of the The radius here is positive for a bubble surrounded
period during which drying rate is by liquid and negative for a liquid droplet in a gas.
diffusion controlled, lb of liquid/lb of If a small tube is inserted in a liquid, as shown in

dry solid. Figure 18.3, the the tube can be

rise of liquid in

determined from a force balance at point A. The

Since the liquid movement by diffusion is
liquid surface in the tube has a radius of curvature
relatively slow, the drying-rate curve may show equal to the radius of the tube only if the liquid
almost no constant-rate period. In any case, c X '

wets the tube so thoroughly that the contact angle

will be the moisture content at the end of the
at the tube wall is zero. If this occurs, the force
constant-rate period and is identical to the critical balance gives
moisture content. Equation 18.2 then gives the
moisture-content— time curve during the falling-
rate period. -AP = Az — (p L -
9
pv = )
—
2y
9c r
Since Equation 18.1 differs from Equation
11.13 only in the definition of Q L * and the con- 9c ^1
Az = 18.4)
sistent units used for concentration, the methods rgiPt Pv
of solution of Equation 11.13 outlined in Chapter
1 1 are also usable with Equation 18.1 Specifically, . and r may be taken as the inside-tube radius. In
the Gurney-Lurie charts are useful graphical solu- any drying solid, pore sizes are not uniform, and
tions to Equation 18.1, and Figure 11.4 may be wetting may not be complete, but the mechanism
used to solve Equation 18.2. of movement is as described.
Even if diffusion does control the moisture At lower moisture contents (those between C
movement through the solid, Equation 18.2 does and D of Figure 18.2), the liquid interface begins
not usually fit the measured drying-rate curve. to retreat from the surface. The retreat is not
Many solids change their pore characteristics dur- uniform, as the radii of curvature of the liquid
460 APPLICATIONS TO EQUIPMENT DESIGN

Drying-gas flow

Figure 18.3. Capillary effect.

Figure 18.5. Moisture distribution in a particulate-

menisci at the surface are not uniform. Liquid in
solid bed during the second falling-rate period.
larger pores is pulled down into the sample to
supply the menisci in smaller pores by flow and frequently the drying-rate curve may be
through the surface-tension driving force. As dry- smooth from C to E. The curves vary greatly in
ing proceeds, liquid in the larger pores continues to shape and slope depending upon the structure of
retreat either until it reaches a narrow "waist" in the solid and the ease of movement of moisture
the pore and thus takes on a curvature matching within it. Toward the end of the drying process,
that in the filled pore or until it retreats far enough the moisture present exists in small pockets in pore
so that the unbalance in surface-tension force is corners scattered throughout the solid as shown in
matched by an unbalance ingravitational head. As Figure 18.6. The actual drying surface is scattered
the moisture is depleted, more and more surface and discontinuous, and the mechanism controlling
pores lose their moisture in this way so that, the drying rate is that of diffusion of heat and mass
between C and D, the proportion of the total through the porous solid.
surface that is saturated becomes less and less.
Figure 18.4 shows the solid phase during this Drying-gas flow

period. Drying proceeds from the saturated surface

at the same rate as that observed in the constant-
rate period. The overall drying rate is reduced
because heat and mass must both diffuse through
the top layers of the solid. By the time point D,
GO
the "second critical point," is reached, moisture
has retreated from all the surface pores. Further
drying involves a longer and longer diffusion path
for both heat and mass.
a granular-solid
The physical situation for
bed is shown in Figure 18.5. Dur-
ing the later drying stages, the solid surface
C3«^3oc^B
Figure 18.6. Moisture distribution in a solid bed to-

approaches the temperature of the drying medium,
but the surface from which evaporation actually
occurs remains at the wet-bulb temperature. The
second critical point is hard to find experimentally,
Drying-gas flow
ward the end of the drying process.
CALCULATION OF DRYING-TIME
In calculations involving drying, the drying-rate
curve must be considered in its major sections, for
the controlling factors differ along different parts
of the curve, the drying rate is defined as

=
-W s dX' Na Ma
R (18.5)
Add
where
R = drying rate, lb of liquid evaporated per
2
hr ft of solid surface
Ws = weight of dry solid, lb
Figure 18.4.Moisture distribution in a particulate- X' = bulk moisture content of the solid, lb
solid bed during the first falling-rate period. of liquid/lb of dry solid
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 46

which can be rearranged and integrated to obtain If the drying-chamber walls are at the bulk-gas
the drying time. temperature,

,0

•o
dd=—
A
1
-W s :

%' R
x 'dX'
_
2

— (18.6)
hv =h c +h r + Uk
In most cases, the heat transferred by radiation and
18.10)

where by conduction through the bed is minor. Then the

total heat-transfer coefficient is essentially a con-
X\ = bulk moisture content at time vection coefficient and may be correlated by a
X2 '
= bulk moisture content at time 8 relation in the usual /-factor form. Thus,

The Constant Rate Period. For the con-

stant-rate period R will be constant at Rc , and
Equation 18.6 may be readily integrated to
Few data are available to allow the constants in

6c= A~R~ {Xc - X:) (18.7) Equation 18.1 1 to be fixed with certainty. Almost
c all reported experimental results use air as a drying
where medium and hence cannot be used to substantiate
the exponent on the Prandtl number. The 2/3
Xq - moisture content at the end of the
exponent proved satisfactory for drying in super-
constant rate period, lb of water/lb of
heated steam (34) and for the evaporation of butyl
dry solid
alcohol, water, and benzene into their superheated
X\ - moisture content at start of drying
vapors (5). The data result in the equation
process
6 C = time of constant-rate drying, hr -0.7

(18.11a)
Rc will depend upon the heat-and mass-transfer CpG\/
coefficients from the drying medium to the solid
surface, for For air drying, the extensive experimental results
have been correlated (28) by the relation
v h
RC = f< Y (Y l
- Y v )Ma =^-{T v -Ti) (18.8)
(expressed in English units)
A
h v =0.0128G 08 ;18.12)
l/
where the transfer coefficients apply from the
evaporating interface to the bulk-gas phase Equation 18.12 is recommended for determining
Since the total amount of heat will be trans- the coefficient during the constant-rate period
ferred by the mechanisms of conduction through when air is the drying medium. In this case, the
the solid, convection from the gas, and radiation surface temperature (7~) may be taken as the wet-
from the surroundings, all to the evaporating sur- bulb temperature of the air.
face, the total rate of heat transfer (q T is )

Critical Moisture Content

q T = h v A(T v - Ti) = h c A{T v - 7»
The moisture content existing at the end of the
+ h r A{Tw -T i)
+ Uk A(T v -T i) (18.9) constant-rate period is called the critical moisture
where content. At this point, the movement of liquid to
h c = coefficient for heat transfer by con- the solid surface becomes insufficient to replace the
vection from gas to solid surface liquid being evaporated. Therefore, the critical
h r = coefficient for radiant-heat transfer moisture content depends upon the ease of moist-
between material surface and walls of ure movement through the solid, and hence upon
drying chamber the pore structure of the solid in relation to the
Uk = overall coefficient for heat transfer to drying rate. The complicated and uneven structure
drying surface by convection and con- of porous solids makes predicting the critical
duction through the bed to the evapor- moisture content very difficult. At present, the
ating surface engineer must depend upon experimental measure-
Tw = temperature of the walls of the drying ments made under simulated production condi-
space X
tions in order to determine c The critical moist-
'
.

Tv , Tj = temperature of the drying gas and the ure content may be expected to depend upon the
liquid-gas interface, respectively pore structure of the solid, sample thickness, and
462 APPLICATIONS TO EQUIPMENT DESIGN

0.12 content upon constant drying rate when asbestos

i>
1-inch bed depth pulp is dried into moist air (21 ).
J 2-inch bed depth
t»
3-inch bed depth
Falling-Rate Period
9. 1 0.08 o
The shape of the drying curve during the falling-
&tr^ a> o
rate period is as difficult to predict as the critical
«•" moisture content. The shape will depend primarily
o^>j^*tro upon the structure of the solid being dried and also
0.04
upon the drying rate during the constant-rate
period and on the critical moisture content. The
gas flow rate, which so strongly influences the
constant drying rate, becomes less and less import-
0.4 0.8 1.2 1.6
ant as the drying rate falls. This decreasing import-
Drying rate during constant-rate period (R, ) ance of the gas flow rate results from the fact that
2
lb H 20/hr ft the drying rate is controlled more and more
Figure 18.7. Influences of constant drying rate and bed completely by diffusion of heat and mass through
thickness on critical moisture content when drying sand in the porous solid. In many cases, the drying-rate—
superheated steam (34). (By permission of Amer. Chem. free-moisture-content curve during the falling-rate
Soc, copyright© 1951.) period approximates a straight line from the
critical moisture content to the origin (14). Then,
drying rate. Thus, the test conditions should use
production widths of the actual solid to be dried. R
(18.13)
The dependence of critical moisture content upon
drying rate is weak and may often be ignored,
although this dependence has been illustrated where the subscript C refers to the critical moisture

repeatedly (34). Drying conditions should be fixed content. Substituting in Equation 18.6,
so that the drying rate during the constant-rate
-W s
rX' dX'
period will match that expected in the plant. | dd =
Figure 18.7 illustrates the dependence of the ec A xc Re -.
critical moisture content upon bed depth and dry-
ing rate during the constant-rate period for the and integrating,
drying of sand with superheated steam (34). Figure
18.8 illustrates the dependence of critical moisture (6 -6 C )=
-W Xc
A
s
— —
RcX
In
X'
c
'
18.14)

The parallel to logarithmic mean (AT) can be seen;

accordingly, a logarithmic-mean drying rate for a
straight-line falling-rate period can be used in Equa-
tion 18.6. This logarithmic-mean drying rate
appears to be a good approximation when the
material being dried is a rigid particulate solid such
as sand. It is not suitable for use in drying fibrous
material such as leather or gellike material such as
soap.

Equilibrium Moisture Content

For many materials, such as the leather and soap

mentioned above, the solid will contain a signifi-
cant moisture content when the drying rate has
dropped to zero. The moisture remains no matter
0.05 0.10 0.15 0.20 how long drying continues if the drying-gas condi-
Average free moisture content, lb /lb tions do not change. It is in equilibrium with the
Figure 18.8. Drying of asbestos pulp in air (211 (By vapor contained in the drying medium and is called
permission of A.I.Ch.E., copyright 1933.) © the equilibrium moisture content.
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 463

The amount of moisture held by a solid in

equilibrium with humid gas depends upon the
structure of the solid, the temperature of the gas,
10

5
^^~~
and the moisture content of the gas. For many Adsorption y s^
materials, it also depends upon whether the solid
originally held more or less liquid than the equilib- 6 / /
rium value. //
//' Desorption
1

A few typical equilibrium-moisture-

content— humidity curves (17) are shown in Figure
18.9. The data are for the equilibrium water con-
04
w
tent of solids in contact with moist air. The graph //
0.2
shows that, for example, a sample of wool contain-
ing 0.2 g H 2 0/g dry wool placed in air at 25° C and
60 percent relative humidity will gradually dry
0.1 0.2 0.3
until a moisture content of 0.145 g H 2 0/g dry Equilibrium moisture content, mass water/mass dry solid
wool is reached. Further exposure will produce no
Figure 18.10. Equilibrium moisture content of sul-
further decrease moisture content. When the air
in
fite pulp showing hysteresis (28). (By permission of
humidity increases to 90 percent saturation at Amer. Chem. Soc, copyright© 1937.)
25°C, the wool will absorb moisture until it
reaches a water content of about 0.23 g H 2 0/g dry pletely dry air indicates that none of the water
wool. In contact with air at zero relative humidity, content is held chemically.
the equilibrium moisture content is zero. The same Figure 18.10 shows the equilibrium moisture
pattern holds for all the materials shown. The zero content curve for sulfite pulp (28). This material is
equilibrium moisture content in contact with com- an example of those in which the equilibrium
moisture content curve shows a hysteresis effect. In
drying operations the desorption curve is always
the one of interest. In Figure 18.11, the effect of

1
iii
- paper, newsprint
1

1/
i

temperature on the equilibrium moisture content

0.28
2 — wool, worsted of raw cotton is shown.
3 — nitrocellulose If the curves of Figure 18.9 were extended to
—
4 -silk
100 percent relative humidity, they would show
5 - leather, sole, oak tai ined
0.24-
r 6 -kaolin
i

7-N.C. tobacco leaf 0.20

8- soap
-
9-glue / / Al
t 0.20
7/ 5/ / h 0.16
// JC
s' /1/ f/ / //
1
E 0.16

// 2y / Ay
" 0.12
/f / l0
/ft' yf j/ 0.12

/
/
/
/ ^s/Z^
J^//
'/ q
>
/
/ 2 0.08
3 0"3
i / s&iy^ i/
0.04 0.04
f

'
6

20 40 60 B0 100 40 60 80 100
Relative, humidity % Relative humidity. %
Figure 18.9. Equilibrium moisture content of Figure 18.11. Equilibrium moisture content of raw cotton,
some solids at 25°C (17). (By permission of Nat. desorption data (32). (By permission of Text. Res. J.,

Acad, of Sc., copyright © 1926.) copyright© 1947.)

464 APPLICATIONS TO EQUIPMENT DESIGN

the maximum moisture content that could exist in Vapor pressure of pure water
23.8
equilibrium with moist air. At this point, the water
Unsaturated
would exert its full vapor pressure. Any additional Solution
solutions
water would only continue to exert the same vapor
pressure and would be in equilibrium with satur-
I 23.1 —i Saturated solution
Saturated
solution

1=
ated air. The added moisture acts like free water +
and is called "unbound moisture." The moisture CuS04 5H 2 -
CuS0„-5H 9
contained in the solid in equilibrium with partly
saturated air is called "bound moisture." It exerts a
vapor pressure lower than that of pure water for ~5
I-—4
CuS0 4 -3H 2
several possible reasons. It may be held in fine
4
capillary pores with highly curved surfaces, it may
5.6
CuSO.
>04
!^S0 4 -H 2 0|
contain a large concentration of dissolved solids, or 0£ » ' I
!

it may be in physical combination in a natural ' 1 3 5 39.6

organic structure. The extent of vapor pressure Equilibrium moisture content (XE'). moles H 2 0/mole C11SO4
lowering due to these mechanisms can vary widely Figure 18.13. Equilibrium moisture content for CuS0 4
with variations in moisture content as shown. at 25°C.
Moisture held in excess of the equilibrium moisture
content is called "free moisture." Figure 18.12 medium, CuS0 4 5H 2 would be obtained at 25°C.
shows the relation among the various terms. On Ifthe partial pressure of water dropped to 7.8 mm
this figure at any gas humidity, such as points, Hg, an undetermined mixture of CuS0 4 5H 2 and #

the corresponding equilibrium moisture content CuS0 4 *3H 2 would result. If the water partial
(XE ') can be read. pressure were only 2 mm
Hg, CuS0 4 *H 2 would
To this point, the discussion of equilibrium be the product. Here, too, a zero equilibrium mois-
moisture content has mainly been concerned with ture content is found in contact with completely
the moisture held in the structure of insoluble dry air.
solids. If the solids are soluble in water, all the
various phase relations are possible that are found
in the normal two-component condensed-phases APPLICATION TO DESIGN OF DRYING EQUIPMENT
system. These phase relations are discussed at The application of basic principles to the design of
length in texts devoted to the phase rule (9). drying equipment is not straightforward. In
Figure 18.13 shows schematically the equilib- addition to the difficulty in predicting the drying-
rium moisture content relation for CuS0 4 Here, .
rate curve, questions arise concerning variation of
several different hydrates form as well as a drying conditions through the dryer, difference of
CuS0 4 — 2 solution. The product obtained in area for heat transfer and for mass transfer, gas
drying this material would depend markedly upon flow pattern, and effect of operating variables and
the humidity of the air used for drying. For choice of equipment upon the condition of the
example, with air containing between 10 and dried product. Also, the usual economic factor of
20 mm Hg partial pressure of water as the drying processing cost must be considered in relation to
the most desirable product condition from a sales
point of view. For these reasons, the choice of
^100 dryer is usually based upon preliminary tests, dry-
yr
/ E;j iibnum-moisture
curve
ing the material under simulated production condi-
tions. Most dryer manufacturers maintain service
laboratories in which pilot-plant-sized dryers are
Bound Unbound
available. The customer may dry samples of
moisture moisture
Equilibrium Free moisture material on these dryers at various operating con-
mois".
ditions to attempt to find the optimum combina-
relative humidity humidity of
A% tion of equipment type and operating conditions.
Here, some major types of dryers will be
described and their operation discussed emphasiz-
Moisture content. (X), mass water/mass dry solids ing the mechanisms controlling the process. The
Figure 18.12. Types of moisture involved in the drying of dryers will be compared so that their relative
solids. advantages may be seen.
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 465

Tray Dryers
The simplest of dryers is the tray dryer, which is
shown schematically in Figure 18.14. The tray
dryer, essentially a cabinet into which the material
to be dried is placed on trays, is a batch unit used for
small-capacity operation. The dryer may have space
for ten, twenty, or more trays. They may be solid-
bottomed pans with the air circulating across the
top and bottom of the pan, or they may have a
screen base with the air circulation controlled so
that it passes through the tray and the solids con-
tained on (Figure 18.15). The material to be
it
Figure 18.15. Through-circulation tray dryer. In
dried may
even be hung on racks or hooks if it
the standard size, the dryer is about 9 ft high, 9 ft
exists in the form of a sheet. The drying conditions
long, and 6 ft wide. It has room for six trays with
are simply controlled and readily changed so that
a face area of 44 ft . (Courtesy Proctor &
this dryer is particularly well suited for laboratory
Schwartz, Inc.)
operations, or for drying materials that require
altering the drying conditions as the drying As mentioned above, for some loosely packed
progresses. If the external controllable conditions material or material that can be formed into small
are constant, the drying conditions will be constant discrete shapes before drying, the tray dryer may
for any given tray of moist solids. Thus, the drying be equipped with screen-bottomed trays. The air-
history at any given point will be just as discussed flow is then directed so that it passes through the
above. But the trays nearest to the air inlet will be material placed on the tray. This arrangement
subjected to conditions markedly different from results in much shorter drying times, but usually a
those located near the end of the airflow path. As a smaller quantity of material can be placed in the
result, material on some pans dries more rapidly, dryer than would be the case with a solid-tray
and that on others dries less rapidly than the aver- dryer. Figure 18.15 is a schematic diagram of this
age. This is of little concern with material that is kind of dryer.
not heat sensitive, but, if overdrying can cause In tray dryers, steam, gas, or electrically
scorching, the pans must be removed at varying heated air is usually used as the drying medium.
times, or the air temperature must be reduced at Energy cost major part of the total processing
is a
the approach of the end of the drying process. In cost. To conserve energy and also to control the air
some dryers, the manufacturers have at least par- humidity at the level that would result in the best
tially overcome this difficulty by arranging for product, part of the drying air may be recycled. In
reversal of the airflow path. this case of air recirculation, the humidity-

p
Fresh-air
inlet N
/?B
/ =

Fan
Heater
screen"

Figure 18.14. A 30 by 40 in. trays on

typical 20-tray dryer. This unit has
4 in. centers. normally by heat exchange with steam, although
Heating is

gas-fired or electrically heated dryers are often used. (Courtesy Proctor &
Schwartz, Inc.)
466 APPLICATIONS TO EQUIPMENT DESIGN

V//
4-

r>
//
-5 *//b
^\
E
3
/ /

x:
/ / >
<
//
/c/ X*
//
/ B'
/ V
/ t ^\
s t ^\
s t ^\
yS ^
^/ E^
/
-^D

Air temperature (TG )

Figure 18.16. Air reheat for a dryer. Figure 18.17. Vacuum shelf dryer. These dryers are
used for biologicals and other products that may be
heat sensitive; they are of high unit value. Note the
temperature path will be as shown in Figure 18.16.
multistage evacuation system terminating in a high
The air mixture containing both fresh and reheated vacuum ejector. (Courtesy of the Pennwalt Corpo-
air enters the dryer at point A and, after passing ration.)
over the trays, leaves at point B. For A—B to be
part of an adiabatic-saturation curve, the solids from the solid state at very low temperatures. This
must be at the wet-bulb temperature of the air. is fundamentally an expensive drying method
Otherwise, curve A—B will have a slope different practicable only for products of high unit price.
from that of an adiabatic-saturation curve but not Thus, major uses have been with pharmaceuticals
markedly different from that shown. The moist air and premium coffee.
is then mixed with fresh air at condition £ to give

air at point C. Air of condition C is reheated to

Example 18.1. Raw cotton of 0.7 g/cm 3
condition A where again it enters the dryer. Had
density when dry is to be dried in a batch tray
none of the air been recycled, the air path would
drier from a moisture content of 1 g H 2 0/g dry
have been E—D—B' with E—D being the air-heating
solids to 0.1 g H 2 0/g dry solids. Trays are 60 cm
step and D—B' the solids-drying step. The net
square and 1 cm thick and are arranged so that
result of air recycling, asidefrom control of inlet-
drying occurs from the top surface only with the
air humidity, is that discharge-air humidity would
bottom surface insulated. Air at 75°C with a 50°C
increase to point B rather than to B' as would have ,
wet-bulb temperature circulates across the pan sur-
been true for drying without recycle. Air dis-
face at a mass flow rate of 2500 kg/hr m 2 Previous.

charged from the dryer receives more energy as

experience under similar drying conditions indi-
latent heat but no more energy as sensible heat.
cates that the critical moisture content will be
The result is a direct saving of fuel energy, though
0.4 g H 2 0/g dry solids, and that the drying rate
a proportionately longer drying time is required.
during the falling-rate period will be proportional
Some materials cannot be satisfactorily dried
to the free moisture content. Determine the drying
in air atnormal atmospheric pressure because they
time required.
deteriorate at temperatures necessary to obtain
reasonable drying rate, or because they react with Solution. For this batch dryer, the drying-
the oxygen in air. Tray dryers can be built to rate-versus-moisture-content curve will be of the
exclude atmospheric air and use other media such typical The problem gives no information
shape.
as superheated organic vapor or evacuated air for on stock temperature, so that the initial
initial
drying. The freeze dryer shown in Figure 18.17 period during which the stock temperature
dries material under vacuum so that in many cases approaches the wet-bulb temperature of the air
the materials being dried lose liquid by sublimation cannot be specified. Thus, the constant-rate period
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 467

will be presumed to start at the initial moisture

content. This is equivalent to assuming the initial
^R c = 0.4018 kg/hr m2
stock temperature to be 50° C. £
The drying-rate curve can be completely fixed 1
1

once the equilibrium moisture content is deter- 1

1
1

mined and the rate during the constant-rate period 1

1
1

is calculated. The equilibrium moisture content I

depends upon the relative humidity of the air, as X'E = 1

1

1
1

0.038
shown in Figure 18.11. From the humidity chart, 1

V = 0.122 mol H 2 0/mol dry air and the humidity / A, = 0.1

1

=
*o = 1.0^
0.4
is 20 percent for the conditions given. Thus, for
•\

saturated air at 75°C, V, = 0.122/0.20 = 0.61. X.. Xf X'c

Since / = y/(1 -/), y = 0.107 and ys =0.375 for
Figure 18.18. Drying rate diagram for Example
the actual and saturated air, respectively. Then,
18.1.

0.107
Relative humidity = x 100=28.5% The drying time during the falling rate period
0.375
can be obtained by modifying Equation 18.14.
and, from Figure 18.11, the equilibrium moisture
X
content E is 0.038 g H 2 0/g dry solids.
'
„ „ -WAXc'-Xe') X -XE
f
'

For the constant drying rate the heat transfer AR, Xr — Xc.
coefficient may be calculated from Equation
(18.14a)
18.12. This equation is given in English units, and
the constant in that equation is not dimensionless, _ -2.52(0.4 - 0.038) 0.1-0.038
nor are the terms h v and Gv The equation could 0.36 x 0.3876 " 0.4 - 0.038
be converted to SI units, but it is probably easier
= 11.54hr
to convert G v to English units, calculate h v in
English units, and reconvert to SI units.
Total drying time,
2.2 lb/kg
G^fEnglish units) = Gv (SI units) x ef -di = 10.84+ 11.54 = 22.38 hr
10.8ft 2/m 2

2.2 Conveyor and Tunnel Dryers

= 2500 x = 509.3 Ib/hr 2
ft
10.8 Tray dryers may be made continuous by moving
2 the moist solids through the drying chamber
/7„ = 0.0128(509.3)°- 8 = 1.874 Btu/hr ft °F
continuously. This may be done by mounting the
= 10.64/m Ks 2
trays on dollies, by conveying the feed through the
and dryer on a belt as in the dryer of Figure 18.19, or,
10.64 for material in sheet form, by moving the moist
= 10.64x3600 ,_ _^
c
/?c v -T,)=
(TV 75-50 sheet through the dryer over rollers. Airflow may
X ' 1054.4x2.2x1065
be directed perpendicular to the flow of material,
= .3876 kg/hr m2 countercurrent to it, or cocurrent with it. Usually,
the flow path is not simple but successively takes
Using these values of and R c the drying XE '

,
several of these directions. In any case, the stock is
rate curve is as shown
Figure 18.18. From the
in
subjected to a drying medium of varying condi-
pan dimensions and the density of dry raw cotton,
tions along the drying path. Consequently, the
the total weight of dry solids is
characteristic drying curve is greatly altered. For
example, the "constant-rate" period no longer
W s
= 60 x 60 x 1 x 0.7 x = 2.52 kg of dry solids shows a constant drying rate. The rate decreases as
1000 the drying-air temperature decreases, even though
The drying time during the constant rate period is the surface temperature of the stock stays con-
then stant. The time required for the constant-rate
period may be calculated through use of the heat-
(1 -0.4) x2.52
9C -6 = = 10.84 hr transfer-rate equation, but an appropriate mean-
° 0.3876x0.36 temperature driving force must be used. The
468 .APPLICATIONS TO EQUIPMENT DESIGN

Air circulating fan

Figure 18.19. Single conveyor dryer. The fans pull the air through the heater
section located beneath them. The air then passes through the wet stock carried
on the conveyor and back into the bottom of the heater. The conveyor belt is
made of perforated-metal plate or woven wire, and the heater is usually of
steam-heated finned tubes. Conveyor widths vary between 15 in. and 9 ft and
lengths vary up to 160 ft. Preforming of the feed to increase drying rate and
obtain a more desirable product is often practiced. It may be accomplished by
extrusion or granulating steps or merely by scoring the cake removed from a
continuous filter so that it breaks into reasonably uniform pieces. Thickness of
material on the belt will depend upon the material being processed, but it can
vary between % and 6 in. (Courtesy of Proctor & Schwartz, Inc.)

falling-rate period may be calculated by success- equal to that for heat transfer. This approximation
ively treating incremental moisture
decreases in is much less satisfactory than the first one, because

content during which the drying-medium tempera-
ture and humidity are considered to be constant.
The calculation of drying rate and time in
continuous dryers requires enthalpy and material
balance calculations as well as the use of the heat-
and mass-transfer rate equations. Complete and
rigorous calculations involve successive approx-
imation solutions of the proper equations written
for each small segment of the dryer. To reduce the
labor involved, a number of approximations are
usually made, although they are not essential to
dryer calculations.
The first of the approximations is
tion that the heat transferred is transferred from
all
air to stockby convection mechanisms. This
assumption requires that the apparatus operate
nearly adiabatically and that heat transfer by con-
duction and radiation be negligible.
Second, the area for mass transfer is assumed
some heat is always transferred to the sides and
bottom of any pan or conveyor system used. Only
in through dryers, where the drying air flows
through the solid being dried, is this assumption
relatively good.
Third, no evaporation occurs during the initial
warm-up or cool-down period. This assumption
brings in a varying error depending upon the condi-
tion of the feed relative to its steady-state tempera-
ture maintained during the constant-rate period.
By using it, the condition of stock and drying
medium at the start of the constant-rate period
the assump- may be more readily calculated.
Fourth, moisture removal occurs entirely at
the adiabatic-saturation temperature. The heating
of the sample that occurs during the falling-rate
period is thus assumed to occur after the last of the
moisture has been removed. This final condition is
closely followed in the drying of a granular sand-
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 469

like material. For porous or fibrous materials, it

may be seriously in error. The net result of the

(2)^
assumptions is that the drying process occurs
entirely at a constant temperature. This tempera-
ture is the adiabatic-saturation temperature of the
drying air, which follows a constant adiabatic-

saturation temperature curve during moisture

«>"
i5cm-
removal.
E 120°F — 4- —
(2)
80°F
Example 18.2. The raw cotton dried in a tray
dryer in Example 18.1 is to be dried in a tunnel x; A',. =0.1 Xr A,, = 1.0

dryer through which it passes on a conveyor belt Solids moisture content (X), lb H L,0/lb dry cotton

2ft wide in a layer \ in. deep. Sixty pounds of air Figure 18.20. Temperature-moisture-content history
per pound of dry solids flows through the dryer through the continuous dryer of Example 18.2.
countercurrent to the stock at a mass rate of
2
500lb/hr and enter at 200° F with a 120°F
ft For the dried-stock heating step,
wet-bulb temperature. The stock enters at 80°F,
and as before it contains 1 lb of moisture per Wt 1

pound of dry solids. It leaves at 150°F with a A ''solids + moisture AHan X X — (b)

moisture content of 0.1 lb of water per pound of

dry solids. The specific heat of dry raw cotton may where the subscript b refers to dry air.
be taken as 0.35 Btu/lb°F. Determine the drying Assuming a 120°F wet-bulb temperature,
time required.
1 x 0.35 x 30 + 0.1 x 1 x 30 = —
60
29
x (-AH air ) (b')

Solution. Although the raw cotton specified

here is not one of the granular materials for which AHak = -6.52 Btu/lb mole dry air
the assumptions stated above apply most accur-
ately, the assumptions will be used in this solution. At the inlet condition (120°F wet-bulb temper-
ature,200° F dry-bulb temperature) the air enthalpy
Using the assumptions and applying the data
developed in Example 18.1, the problem can be is 3240 Btu/lb mole, and therefore H 2 =
attacked directly.
3234 Btu/lb mole. For air at a wet-bulb temper-
ature of 119°F, # = 3154 Btu/lb mole. Thus, the
Writing an overall material balance based on
1 lb of dry stock gives the humidity of the outlet
120°F wet-bulb temperature is still substantially
7"
correct for the air at point 2; and the 30° F A
air.
inserted in Equation (b') above is substantially
correct, since the stock reaches the wet-bulb
_ Ma VMb
— temperature of the drying air.
Xo -X '
= (Y - Y x (a)
Mb W /6f-
)
f f
s
The wet-bulb temperature of the air will not
change between points 2 and 3 of Figure 18.20, so
that H 3 % 3234 Btu/lb mole dry air.
1.0-0.1 =( V - 0.098) —x 60; Y = 0.122 Now, writing a heat balance over the initial
solids warm-up part of the cycle, assuming no
evaporation,
where the subscripts and f refer to the con-
ditions at the ends of the dryer where stock enters
and exits, respectively. These subscripts apply 60x(-A/Yair )
1 x 0.35 x 40 + 1 x 1 x 40 =
whether air or stock properties are being consid- 29
ered; for exam pie, Vq is the mole ratio of water to dry 1.35 x 40 x 29
air at the end of the dryer from which the air leaves, -Atfair = = 26.1 Btu/lb mole dry air
60
but in which the stock enters.
The path of stock and air temperature "air leaving the drier ~ 3207 Btu/lb mole
through the dryer is shown in Figure 18.20. Heat
balances can be made from the stock-outlet end of The wet-bulb temperature is still substantially
the dryer (air-inlet end). 120°F, and, since the humidity is 0.122 lb mole
470 APPLICATIONS TO EQUIPMENT DESIGN

H2 per pound mole dry air, the temperature is The falling-rate period can be attacked in the
144°F. Neither the humidity nor the wet-bulb same way. Here, the evaporating liquid is assumed
temperature of the drying air changes during the to stay at 120°F, but the effective heat-transfer
2 =200°F
7~
initial and final warm-up periods, so coefficient (h Veff ), decreases as the drying pro-
and r3 = 144° F. gresses. Again,
For the entire operation, the heat-transfer
-Wt—
\'dX'
coefficient is fixed and can be calculated from dq = h Vefi A AT=
Equation 18.12. dd
8 2
h v = 0.0128(500)° = 1.86 Btu/hr ft °F
jh Vefi A ATdd = - \Ws \'dX' (d)
2
On the basis of 1 ft of conveyor area, If it is assumed that the falling-rate period exhibits
a linear-rate-versus-moisture-content curve, h VeU
Solids held = 1 x -x
2
1
—
12
1
x 62.4 x 0.7 will also be a linear function of X' Also, /7^eff = .

h v = 1.86 at *'=,? c = '

0.40, and /Veff=0 at
= 1.82 lb dry cotton X'=XE '.

For the period during which the dried solids

final The inlet air, which enters at the end of the
are q = 1.82(1 x 0.35 x 30 + 0.1 x 1.0 x
heated, drying path, has a relative_humidity of 3 percent.
2
30) = 1.82 x 13.50 = 23.6 Btu/ft of conveyor Then, from Figure 18.11,Xf = 0.008, and '

23.6 A/7 l/eff 1.86

= h v A AT, m = 1.86 x 1.0 = constant = = 4.75
Of — do AX' 0.4

(200 -120) -(200 -150) 1.86-/V e ff

X 8^
In
— 0.40 -X'
50 or
23.6
Of — d? - = 0.199 hr /?v,
= 4.75X' - 0.04
1 .86 x 63.9
Rearranging Equation d and solving,
For the constant-rate period, the heat-transfer-rate
equation can be used directly. x f
'

AATlm j dd=-W
d0 = -W,\'\
\'j.s
x c 4.75X'-0.04
,

-W
'

\'dX' '•c
s
dq = h v AAT = (c)
-W \' 1 4.45X/ - 0.04
dd = s
02 - 0c In =-;
A ATlm 4.75 4.75Xc '-0.04
c ec rXc'
h vA ATd6=- W s \'dX' -1.82x1025

h vA ATlm (6 C - 63) = W S \'(X3 '

- Xc ')
(c'J 1
——-x4.75
80-60
x—
M80/60)
As before, the moisture content (Xc ') is
critical
0.475 - 0.04
0.4lbH 2 O/lb dry To get AT, m the air
stock. ,
x In = 8.22hr
temperature at the point where the solids reach 1.90-0.04
X c must be known. At this point, the wet-bulb
'

Finally, the time required for the initial

temperature is still 120°F, but the humidity is
warm-up period can be determined as was that for
the final warming period.
(1.0-0.4) = -(V c - Vo )^60
q = 1.82(1 x 0.35 x 40 + 1 x 1.0 x 40)
Vc = 0.106 moles of water/mole of dry air = 98.4 Btu/ft 2 of conveyor area
From this value of Y c Tc, is fixed as 180°F. Now, 98.4
= h v A AT,m = 1.86 x 1.0
substituting in Equation (c'),
03 — 00

,a
(0c
ax = 1-82x0.6x1025.8
-03) ———= icou
15.3 hr
144 _ 80) -(144- 120)
60-24 64
1.86 x 1.0 x i

In (60/24)
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 471

from which, v = gas flow velocity based upon total bed

cross section
98.4 x In
—
64
= void fraction
e
24 /Vr 6 = Reynolds number, dp vpln. (/V Re
09- = 1.3 hr
1.86x40 should be <4000)
dp = particle diameter
Total time = 1.3 + 15.30 + 8.22 + 0.20 =
drying k G = mass transfer coefficient, mol/hr m 2
25.02 This is an extremely long processing
hr. atm
period for continuous operation. If the conveyor is
Note that in Equation 18.15 the constant is not
200 ft long, the rate of movement is 200/25.02 =
dimensionless.
7.98 ft/hr, and the production rate from the 2 ft
wide conveyor belt would be Particulate Dryers

1 .82 x 2 x 7.98 = 29.0 Ib/hr of dry raw cotton Free-flowing particulate material may be difficult
to hold on a woven wire or perforated metal con-
or veyor belt. In such cases, the material can be
29.0 x 1 .1 = 31 .9 Ib/hr of dried product [X' = 0.1 cascaded through the flowing gas (as is done in the
rotary dryer), moved countercurrent to the gas
This value requires an airflow rate of 29.0 x 60 = flow in a column arrangement (as is done in the
1740 Ib/hr (dry basis) and permits a free dryer plate dryer), or blown with the gas stream (as is
cross section around the conveyor belt of done the flash dryer).
in
2
1740/500 = 3.48 ft .
In the rotary dryer, the solids are tumbled in
a continuous spray through the center of the rotat-

The results of Example 18.2 indicate that, for ing drum, and the air is blown through the spray.
continuous production at high throughput, drying Internal flights lift the solid and control its cascade
by passing air over the surface of a bed of wet through the airstream. The dryer is tilted so that

solids is likely to be impractical. Circulation of the solids gradually work their way from feed end
drying gas through the bed as is done in the dryer to discharge end. Flue gases as well as superheated
of Figure 18.19 is one possible solution. As indic- steam or even electrically heated air may be used
ated there, the feed must be made into such a form directly as the drying medium. In some dryers,
that the gases can pass through the bed, and the steam-heated tubes run through the drying cylinder
particles must be large enough to be retained on to maintain the air temperature and to act as drying
the conveyor belt. It is also obvious that the surfaces. These dryers are built in sizes up to 9 ft
temperature of inlet air should be as high as in diameter in standard models with lengths up to
possible in order to increase AT. Since the stock 80 ft.

remains the wet-bulb temperature during

at A much higher solid surface is exposed in

constant-rate drying, heat-sensitive materials may rotary dryers than is exposed in tray or tunnel
be dried with air entering at temperatures some- dryers, and the drying rate will be much higher.
what above the safe stock temperature. The high an advantage only if the air can be
rate is

In these dryers, the gas flow rate through the kept unsaturated. Thus, the engineer must use a
bed is be low because of pressure drop
likely to high air rate or must heat the air as it passes
limitations or because of the approach to fluidiza- through the dryer. It is for this reason that steam-
tion, as discussed in Chapter 22. On the other heated tubes may be inserted into the rotary dryer,
hand, the drying rate will be large because of the as has been done in the dryer shown in Figures

very large transfer area represented by the surface 18.21 and 18.22. The tubes usually are inadequate
of all of the drying particles. Also, most of the to keep the air temperature constant, although

drying process will occur in the constant-rate they make the air-temperature— humidity curve
period. Theoretical and experimental studies of the considerably steeper than an adiabatic-saturation
flow through curve. In some cases, the internal heating may even
mass transfer occurring in gas a
be strong enough to increase the air temperature as
packed bed give (2, 4)
it passes through the dryer. Thus, the air path may

)-°- 5 (/V \-0.67

1.2i/(A/ Re -0.5/a/ take any of the directions indicated by Figure
Ar G = Sc ) ;18.15)
18.23.
where In designing rotary dryers, the retention time

v = actual gas flow rate through the bed = of solids passing through the dryer must be esti-
v le, m/hr mated. The retention time depends upon the
472 .APPLICATIONS TO EQUIPMENT DESIGN

Figure 18.21. An internal-steam-tube rotary dryer. The foreground end is the

feed-inlet end. The steam enters the tubes through a central inlet at the far end of
the dryer. It is distributed to a header and then into the tubes. Condensate is

discharged at the near end of the dryer into the collection ring shown in the
foreground. From this ring, it flows back to the far end of the dryer through an
internal tube for ultimate removal.

-^ o
*-*
^" o
-2-E

Temperature. °F

Figure 18.23. Possible air-condition paths in an in-

ternally heated rotary dryer.
Path AB— No internal heating. Point B may be off
of the adiabatic-saturation curve through A, because
of heating or cooling of solids.
Path AC— Internal heating inadequate to maintain
air-temperature constant.
Figure 18.22. Internal view of rotary dryer of Figure 18.21, Path AD— Enough internal heating to maintain iso-
from the feed end. The internal arrangement of steam tubes thermal operation.
is shown. Each tube has a transverse spiral fin that roughly Path AE— Excess internal heating so that air
doubles the external surface area. temperature increases as it passes through the dryer.
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 473

density and angle of repose of the solid, the particle shape) shows no significant dusting at
2
arrangement of flights in the dryer, the slope of the 1000 Ib/hr ft airflow.
dryer, and the mass of material present in the The loading of moist solids in rotary dryers
dryer. markedly affects the operation. Too low a quantity
The movement of solid through the dryer is of solids will reduce the production rate. Too great
influenced by three distinct mechanisms. First, as a quantity of solids will result in uneven and
the dryer turns, each particle is lifted up by the incomplete flight action with some of the solid
flights dropped again. At each drop, the
and merely rolling on the bottom of the dryer, and it
particle advances a distance Ds where D = dryer may result in a moist product. As a result, drying
diameter and s = dryer slope, ft/ft. Thus, with a will be uneven, and the power required to turn the
dryer of length / rotating at N rpm, the time of dryer will increase. There is more danger of over-
passage of a particle would be proportional to loading a dryer than underloading it, with experi-
HsDN. This is called "flight action." In addition, ence indicating a loldup of 3—10 percent of the
the particles striking the bottom of the drum dryer volume as giving satisfactory operation.
bounce, and those striking other particles roll over Heat-transfer coefficients in such dryers are
them, while solids that are not lifted by the flights obviously indeterminable since the interfacial area
still move forward in rolling over others on the is unknown. Friedman and Marshall (11) deter-
bottom of the dryer. These effects are called "kiln mined Ua from tests in a 1 ft by 6 ft rotary dryer
action" and significantly alter the holding time of a and compared their results with those previously
particle in the dryer. Finally, the drying gas blow- published. The equation used to correlate their
ing through the dryer either advances or hinders results was
the travel of solid, depending upon whether gas 16
0.63G,, -

flow is cocurrent or countercurrent. The carrying Ua = -£- [G F +BG vPb - 5

]

of particles by the gas stream can only be esti-

mated empirically because the settling velocity is (18.17)
hindered by the large number of particles present. where
On these considerations and on data obtained in
pilot-sized dryers, Friedman and Marshall (11)
Ua = product of overall coefficient and sur-
3
face area, Btu/hr °F ft of dryer
offer the following relation:
volume
0.35/ BIG V D = dryer diameter, ft
18.16) 2
Gp = feed rate to dryer, Ib/hr ft of dryer
cross section
where 3
p b = bulk density of dry solids, lb/ft
6 = average time of passage, min Gv = airflow rate, Ib/hr ft 2
=
/ dryer length, ft B = 5.2Dp -°- 5 where Dp is the weight
,

=
s dryer slope, ft/ft average particle diameter, microns
N- rate of rotation, rpm
D= dryer diameter, ft This equation is far from exact, with devia-
B= constant dependent on particle size of tions reported in the order of ± 100 percent, but it

material handled, B = 5.2 (D p )-°-

5 will give an indication of required dryer size. The

Dp = weight average particle size, microns materials dried were granular, and hence most of
Gv = gas mass velocity, lb/ft 2 hr the drying corresponded to constant-rate condi-
Gp = feed rate to dryer, lb dry material/hr tions. The temperature remained constant
solid
2 near the air wet-bulb temperature until it suddenly
ft of cross section
rose to approach the air temperature. One un-
and where the plus sign is for countercurrent flow
explained observation was that, particularly for
and the minus sign for cocurrent flow.
cocurrent operation, the constant-solids tempera-
Permissible air velocities are limited by the
ture was not always the wet-bulb temperature but
tendency of the material being dried to "dust."
was sometimes as much as 5°F above it. This could
This tendency obviously varies greatly with the
be a result of radiation transfer.
properties of the material handled. For instance,
Friedman and Marshall report that sawdust begins to
dust excessively at flow rates of about 250 Ib/hr ft
2 Example 18.3. A rotary dryer 5ft in
but Ottawa sand (a white sand of very round diameter by 60 ft long is to be used to dry
474 APPLICATIONS TO EQUIPMENT DESIGN

titanium dioxide from a moisture content of 0.3 lb Substituting in Equation 22.15,

water per pound dry solids to 0.02 lb water per
^= — (240-0.05)32.2
1

pound dry solids. The Ti0 2 has a mean particle

3 18
size of 50 microns and a density of 240 lb/ft The .

-4 2
dryer operates at 4 rpm with a slope of 0.5 in. in 50 x 10 v
1

10 in of length. Determine inlet-air rate and 2.54 x 12 '

0.02x0.000672
required temperature, and rate of production of
240x32.2 2.69 x10~ 8
dried Ti0 2 if drying air is obtained by heating
room air originally at 90° F with a 60° F wet-bulb
=
7^
18
x 777^ —
7^~~Z
0.135 x 10~ 4
"
m
°- 860 ft / s

temperature.
at which velocity
4
Solution. This problem not stated so com-
is 1.64 x 10~ x 0.860x0.05
/v Re =
pletely only one
that answer is possible. The 0.02 x 0.000672
answer obtained will depend upon gas-flow rate = 0.524
and dryer-loading values used. They can be chosen
arbitrarily provided reasonable limits are not Thus, if G v = (0.860 x 3600) x (14.7 x 29)/-
exceeded. The airflow rate must be low enough to (10.73 x 1000) = 123 Ib/hr ft 2 there would be a ,

prevent excessive dusting. The value of 1000 lb/-
hr ft 2 given above probably applies to much larger
particle sizes than are involved here, so that
another method of estimating the limit of possible
airflow must be used. Dusting must occur when the
particles are picked up by the airstream. Actually,
the airstream does not flow directly opposite to
chance of entraining significant amounts of solid
into the airstream. Here an air temperature of
1000°R is assumed. Of course, some entrainment
may occur at lower gas-flow rates because of the
complex flow pattern and the interference of
particle against particle. Therefore, a gas-flow rate
[G v of 50 Ib/hr ft 2 should be appropriate.
)

the particles as they fall, but a safe limit may be The time of passage may now be calculated
obtained by keeping the airflow below the velocity from Equation 18.16 and from the dryer loading.
at which it could entrain the solid particles if it Substituting in Equation 18.16,
were flowing directly opposite them.
0.35 x 60 5.2 x 60 x 50
The relation giving the maximum settling
0.05 x (4)
09 x 5 + 0.6 (50)
05 x G
velocity of particles in a fluid when these particles F
do not with each other
interfere is derived in 1320
Chapter 22. The equation 6 = 24.2 + (a)

(ps - p)gD 2 The loading the dryer may be chosen

of
vt = (22.15) within the 3-10
percent range noted above.
18m
Choosing 5 percent of the dryer volume,

is derived by combining a force balance on a 1 6

particle with the relation for a drag force (Equa-
G/rX-x— =0.05/
p 60
tion 13.40). The force balance specifies zero
acceleration when the buoyant and drag forces 0.05 x 60 x 60 x 240 43200
(b)
balance any external force on a particle. Simplifica- Gf
tion is obtained by assuming the particle is
Combining Equations a and b,
spherical and the relative particle-to-fluid velocity
is low enough to give laminar flow. The student is
referred to the development beginning with Equa-
—G—
43200 _
=24.2 + ——
1320
GF
F
tion 22.1 for the complete derivation.
and
In Equation 22.15, v t is the maximum rate at 2
GF = 1770 Ib/hr ft
which a spherical would fall in still gas if
particle
the fall were slow enough that the Reynold's The inlet-air temperature must be high
number (A/ Re would be less than about 0.5. Con-
) enough to give an adequate AT so that the required
versely, of course, vt is the vertical gas velocity drying occurs in the available dryer volume and
necessary to suspend a spherical particle in the gas still remains unsaturated when it leaves the dryer.

in laminar flow. From the previous conditions, 1 lb of air must dry

 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 475

35.4 lb of solids, or it must absorb 35.4(0.3 — Thus,

0.02) = 9.9 lb of water. The air might possibly
0.63(50)
016 5.2x50x2401 0.5
absorb this much water if it were heated as it r
Ua = 1 1
ftrt +
9°
17/.yu
-
,
passed through the dryer. With a simple rotary 5 i
L (50)1/2
(50) J

dryer with no internal heating, as this dryer must 5 3

= 0.236(17.90 + 8820)° = 22.8 Btu/hr °F ft
be presumed to be, such a moisture pickup is
absurd. Since the gas-flow rate cannot be con- The required AT/m then is

trolled by the designer, the dryer loading must be

altered. To obtain an outlet humidity in the air-
5095
Ar/m = = 3.72°F
stream of about 1 lb of water per pound of dry air, 22.8 x 60
the required feed rate would be 1770/9.9 =
179 Ib/hr of dry Using this feed rate, the
solids.
Since AT y
is about 414°F, A7"2 will be very close
to zero. Thus, for a solid-feed rate of 17.9 Ib/hr of
required inlet-air temperature can be found by a
trial-and-error procedure. At V" = 1.0, the satura-
dry solid, the inlet-air temperature would have to be
2
tion temperature is 189°F. Since the adiabatic- 544°F, and 50 Ib/hr ft of air would have to be

saturation curve, along which the drying air will used.

move, is approximately represented by In this example, a dryer was used that was not
suitable An internally heated
for the operation.
C^(Ta -T,)=\ a (Y, -Y a ) (17.21a) dryer would have allowed a much greater feed rate.
The low permissible gas-flow rate resulted from the
the path of the drying air is known. Also Yy is very fine particles handled. Here, pelletizing the
fixed by the problem statement as particles in a preliminary step might be econom-
0.0173 mole H 2 0/mole dry air. Then, by Equation ically attractive. Finally, better throughput and
17.21a, utilization of equipment would have been obtained
had the dryer been larger in diameter but shorter.
/ 29 \ 985x18 The shown diagram form
(r1 -ra ) = (ix- + 0.0173 - 0.0173]
713
plate dryer, in in
Figure 18.24, can also be used to dry particulate
= 4000°F solids and is especially useful where dusting is a
problem or where variable temperature regions are
This temperature is beyond practicality. Using the desired. In this dryer, each plate is jacketed so that

moisture pickup as 0.1 lb of H 2 per pound of dry the material may be heated as it passes over some
solids, the outlet humidity is 0.1x29/18 + plates, and cooled as it goes by others.

0.0173 = 0.1783 mole H 2 0/mole dry air. The In flash drying, particulate solids are fed into

saturation temperature for this humidity is 130°F. a moving stream of hot gas. Some distance down-
stream of the feed point they are separated from
0.1 x —
29
x1019x18 the gas, usually in a cyclone collector. The gas-flow
requirement and the pressure loss in such a gas-solid
T,-Ta = : =414°F flow system are discussed in Chapter 22 in con-
7.13
sidering the conveying of a solid by a gas stream.
T, =544°F Here the additional complication of the heat- and
mass-transfer processes must be handled in the
This inlet temperature can be attained only if the
dryer design. Experimental measurements of heat
exit air is saturated; thus, an infinitely long dryer
transfer rates in flash dryers (7) have produced
would be necessary. The heat load on each pound
correlating equations such as
of inlet air is

/V Nu =0.16(/V Re *) 1 - 3 - 67 (18.18)
q = 0.1 x 1019 = 101.9 Btu (/V Pr )

which gives a total heat transfer rate of where

101 .9 x 50 = 5095 Btu/hr ft

2
of dryer cross section /V Re * = Reynolds number based upon particle
diameter and slip velocity v S \, as de-
Since the material being dried is granular and is fined in Chapter 22.
being dried as individual particles, it is safe to
assume that all drying occurs in the constant-rate Variations caused by the forces acting to
period. The volumetric overall heat-transfer co- accelerate the particles as they enter the gas
efficient can be calculated from Equation 18.17. stream, and by bends in the flow duct, are large.
476 APPLICATIONS TO EQUIPMENT DESIGN

Blower

Cyclone
separator

Solids,
feed
Heater

IlhQ- Air
feed
Blower

Figure 1&25. Flash-dryer installation. The

wet material (possibly mixed with dried
material) is introduced in a hot airstream
carrying the solid material upward. After
drying in the vertical duct, the solid material
is separated from the exhaust air in one or
more cyclones, and the exhaust air can nor-
mally be conveyed directly to atmosphere
by the exhaust fan. (Courtesy of Niro Atom-
izer Inc.)

required residence time, and a solids-gas separator.

Design of this separator is often crucial, for the loss
of even a small fraction of the product can be
costly where tonnages are processed — and
large
can be a pollution source. The design of cyclone
separators is discussed in Chapter 22. Figure 18.26
shows how this simple system appears as it is
developed for use in a large-scale processing situa-
tion.
Where longer contact times are desired, or
where the material partially dried in a flash dryer

Figure 1&24. The plate dryer. Units such as this are avail-
requires additional drying, a fluid bed dryer may
able in sizes ranging from a 5-plate unit 1.2 m in diameter be used. The principles of fluidization are discussed
with 3.8 m" of plate area with flights driven by a 1.5 hp in Chapter 22. In a fluid bed dryer, the moist feed
motor to a 35-plate unit 2.7 m in diameter with 176 m :
of is added to a fluid bed of material being dried at
plate area with flights driven by a 16 hp motor. The drying the center of the bed, or at one end of a rec-
medium can be air, inert gas, or flue gas at atmospheric or tangular bed, and the dried product is withdrawn
vacuum conditions. (Courtesy of AMETEK, Inc., Process from the bed at the edge, or at the opposite end.
Equipment Division.)
Residence time would typically be 30 minutes or
more. Drying rates in a fluidized bed dryer would
Hence, Equation 18.18 should be used with be determinable from the heat- and mass-transfer
caution. rates to particles in a fluidized bed. Mass transfer
Flash drying is particularly useful when a rates may be obtained from (1):
paste or moist cake is to be partially dried using a
very short contact time. A flow diagram for such a k GPBMMa DP G .- 051
dryer is shown in Figure 18.25. Note that the basic = 67
In (/V Pr )
-

=2.25|
requirements are only a solids-feed device, perhaps Md-e)
a screw conveyor, a pipe section allowing the :18.18a)
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 477

Figure 18.26. Two-stage flash drying plant for 250 tons/day cellulose pulp. The
two stages are in countercurrent arrangement, thus improving the heat economy of
the drying system. (Courtesy of Niro Atomizer Inc.)

for the air-water system Rotary vacuum dryers, porcupine dryers, and
pan dryers are often operated in batches. The varia-
h 67 =
(/V Pr
-

1.12/*, (18.186) tion in agitator power requirement versus solids

lq )

CP G moisture content (13) of such a batch dryer is

rather interesting. A typical plot is shown
Batch drying of particulates may be done in a schematically in Figure 18.31. At high moisture
conical rotating dryer, where, again, the particu- contents, the slurry is relatively free moving, but,
lates cascade through the drying medium. These as drying proceeds, it becomes more viscous and
dryers often operate under vacuum, as would the pasty. It finally becomes a heavy, adhesive mass
one shown in Figure 18.27, and can be supplied that very difficult to move. If drying continues,
is
with internal heated tubes. These dryers are advan- the mass breaks into lumps that move much more
tageous where complete vapor containment is readily, and the lumps continually break up until a
desired, as with polymer drying, and where heat free-flowing granular product is obtained. The
sensitivity requires low-temperature operation with enormous changes in physical properties of the
gentle solids movement, as with pharmaceuticals. material being dried seem to affect the heat-
transfer coefficient only to a minor extent. The
Drying of Heavy Slurries and Pastes moisture content at which the power peak occurs
Materials that are not free-flowing must usually be and the height of the power peak must depend
mechanically stirred, spread on the drying surface, upon the slurry being dried and the dryer design,
and scraped from it as drying proceeds. Such units but no studies have been reported along this line.
include the rotary vacuum dryer (Figure 18.28), Very few quantitative performance data are
the porcupine dryer (Figure 18.29), the pan dryer available for use the design of these dryers.
in
(Figure 18.30), and drum dryers. Overall heat-transfer coefficients varying between
478 APPLICATIONS TO EQUIPMENT DESIGN

Figure 18.27. Conical rotating vacuum dryer. These dryers are com-
3
monly available in sizes from 3.5 to 400 ft working volume, and rotate
from 6 to 2 rpm, respectively at these limits. (Courtesy of the Pennwalt
Corporation.)

Figure 18.28. Interior view of a rotary vacuum dryer. Heat is supplied by the vessel

jacket, with the rotor blades keeping the drying mass in motion and scraping the
surfaces clean. (Courtesy of The Bethlehem Corporation.)
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 479

Figure 18.29. A view into the drying chamber of a porcu-

pine dryer. Here the agitator blades are heated as well as the
jacket. These dryers are used for drying heavy pastes and Figure 18.30. Vacuum pandryer. This dryer, of 4 ft.

can be operated batchwise or continuously. (Courtesy of diameter by 24 working depth, has a capacity of
in.
2
The Bethlehem Corporation.) 140 gal and a heating surface of 35 ft Internal
.

scraper blades keep the charge moving as drying pro-

ceeds and scrape caked material off the heated sur-
5 and 175 Btu/hr ft2 °F have been reported (19),
faces. (Courtesy of The Bethlehem Corporation.)
but for design purposes, where no pilot-plant in-
formation is available, coefficients between thickness of slurry layer held on the drum will
2o
10-15 Btu/hr ft F should be used. depend upon the characteristics of the slurry and
of the drum surface as well as the spacing between
Drum Dryers drums. Figure 18.32 shows such a dryer, and
Slurries and pastes may be dried continuously on Figure 18.33 shows the construction arrangement
drum dryers, provided that the dried product is not of the dryer. Here, the feed may also be admitted
hard and gritty enough to score the drum surface.
Thus, drum dryers are used most often to dry
organics. These dryers are rotating horizontal
cylinders heated with steam on the inside. The
slurry being dried is spread over the outside surface
of the dryer, clinging to it and drying as the hot
drum rotates. When the slurry has made about
three-quarters of a complete rotation on the drum
surface, it is scraped off with a doctor knife. Rota-
tion rates of 1 — 10 rpm are normal. The rotation
rate, drum-surface temperature, and slurry-layer
thickness are adjusted to give the desired moisture
content in the dried cake scraped from the roll.
The product comes off in flaked form.
Various drum arrangements and feed methods Moisture content, mass H20/mass dry solids

are available. In double-drum dryers, the slurry is Figure 18.31. Performance of a pan dryer drying a
fed into the nip between two rotating drums. The material that passes through a paste or gel stage.
480 APPLICATIONS TO EQUIPMENT DESIGN

^r ^=44
Figure 18.34. Twin-drum dryer with top feed.
The feed is held in place by end boards and by the
closeness of the two rolls. (Courtesy of Blaw-Knox
Food & Chemical Equipment, Inc., Buflovak Di-
vision, One of the WCI Industries.)

Figure 18.32. A 42 in. diameter x 120 in. long atmospheric

double-drum dryer of new design complete with vapor slurry. Other feed devices are shown in Figures
hood and pneumatic adjustable endboards and knives. Vari- 18.34 and 18.35. The top feeding of Figure 18.34
speed drive is located in the right rear background. Thus, permits a thick coat to be formed on the roll
drum separation and rate of rotation can be adjusted to
surfaces, which is advantageous when the material
offset changes in feed properties. The slotted end frames
is granular and easily dried. The splash feed of
allow easy drum removal if surface refinishing is needed.
Figure 18.35 forces the slurry against the dryer
The discharge screw conveyors are removed to show the
rolls, thus helping to prevent it from falling off as
drum. (Courtesy of Blaw-Knox Food & Chemical Equip-
ment, Inc., Buflovak Division, One of the WCI Industries.) drying proceeds.
Single-drum dryers and vacuum single- and
double-drum dryers are also available and operate
through a pipe which swings like a pendulum from
on the same principle as those described above.
end to end of the dryer.
These drum dryers are shown schematically in
Twin-drum dryers operate on the same
Figures 18.36 and 18.37.
principle as double-drum dryers except that they
Operation of drum dryers and design of drum
rotate in the opposite direction — that is, away dryers for a particular service are still more art than
from each other the top. Drum spacing no
at
science. Usually, the design must be based upon
longer influences the thickness of the drying layer
pilot-plant tests under the proposed operating
but can be adjusted to break up bubbles and
conditions. Even then, scale-up methods are un-
smooth out the coating of slurry on the drums
certain. Experiments with drum dryers have
where the drums are dip fed. The simplest feed
yielded a considerable amount of information on
device is merely to have the rolls dip into the feed
drying particular materials under a few conditions.
The results cannot be directly extrapolated to
Feed pipe plant-sized units, nor can they be applied to the
drying of other materials.
Problems arising in operation of drum dryers
are concerned with obtaining an even, unbroken
sheet of drying material on the roll, with adhesion
to the roll, and with obtaining the proper con-
ditions to give a product of desired properties. The
need for a high production rate can be met only
with a thick, dense, and continuous layer of drying
solid. For this purpose, the feed slurry should be as
-Conveyor Conveyor- concentrated as possible. Of course, as the concen-
Figure 18.33. Schematic drawing of double-drum tration increases, so does the viscosity, so that an
dryer showing feed arrangement, drum-rotation even distribution along the roll length is hard to
direction, and product-discharge system. maintain, and the slurry may not wet and adhere
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYLNG 481

Figure 8,35. Twin-drum dryer with splash feed. Splashing the feed onto the drum
1

making it adhere to the drums and produces a dense coating. (Courtesy of

aids in
Blaw-Knox Food & Chemical Equipment, Inc., Buftovak Division, One of the WCI

.=::' :.'="

Figure 18 3-6

be dip fed, in which case the drum rolls through

the pan holding the feed slurry, or splash fed as Figure 18.37. Schematic diagram of a vacuum double-
shown in Figure 18.35. (Courtesy of Blaw- drum dryer. (Courtesy of Blaw-Knox Food & Chemical
Knox Food 8t Chemical Equipment, Inc., Equipment, Inc., Buftovak Division, One of the WCI
Buftovak Division, One of the WCI Industries.) Industries.)
48 2 APPLICATIONS TO EQUIPMENT DESIGN

to the roll surface. Adjustment of roll spacing and kaolin, alumina, and iron oxide furnace fume wash-
revolutions per minute be required by minor
may ings.

changes in feed conditions. Production rates in the Figure 18.38 shows a typical spray dryer
2
order of 5 lb of dry solids/hrft of drum surface installation. Air, fed through a filter and heater,
are normal when steam at 50 psig is fed to the enters at the top of the drying chamber flowing
drums. down cocurrent with the droplets being dried,
these are formed in a spray nozzle or rotary disk
Spray Dryers atomizer. As the atomized droplets fall, the
Spray dryers require a pumpable feed (solution, moisture evaporates into the hot gas, leaving the
slurry, gel, emulsion, etc.) and give a powder or solid material as particles. The larger particles fall
small bead as product at rates varying from pilot to the bottom of the dryer. Smaller particles are
size to capacities such as 25 tons/hr of solids. In swept with the gas into cyclone separators. The
contrast, drum dryers produce flakes and are very finest particles continue through the blower
limited to low or moderate throughputs. Another to a wet scrubbing system such as those described
advantage of the spray dryer is that the product is in Chapter 17. The slurry or solution produced in

exposed to the hot gas for short times, and the
evaporation of liquid from the spray keeps the
product temperature low even in the presence of
very hot gases. Spray dryers are relatively large
units and may be inefficient in their energy use.
Because of their wide range of utility and
convenient product form, spray dryers are used
with an enormous range of products. A partial list
would include coffee, milk, detergents, dyestuffs,
pesticides, polymers, tile slips, blood
enzymes, penicillin, starch, metal
the wet scrubber may be returned to the feed.
Wide variations on this system are possible, usually
dictated by the properties of the feed and the
desired product characteristics. The drying
chamber may operate with airflow countercurrent
to the product flow — or in a complex flow path.
Secondary air may be fed into the drying chamber
as shown here, or may be omitted, or may enter at
several points. The product-collection system often
plasma, uses only one cyclone collector, but it may also
concentrates, include a bag filter, and even an electrostatic pre-

EXHAUST

WET SCRUBBER
AND
MIST SEPARATOR

AIR FILTER
PRODUCT

PRODUCT
Figure 18.38. Flow diagram of a typical spray drying system. (Courtesy of
Stork-Bowen Engineering, Inc.)
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 483

production and delivery system, a drying chamber,

a solids-gas separation system, and a product-
discharge system. The design of each of these
systems depends upon the material being dried and
is influenced by the design of the rest of the unit.

Thus, the form of the finished spray dryer may vary

enormously from product to product and even
from installation to installation for the same
product.
Perhaps the most important part of the
spray-dryer system is the feed atomizer. Three
classes of atomizers are normally used in spray
dryers. Two-fluid nozzles, such as that shown in
Figure 18.40, are used in low-production-rate dry-
ing, particularly if a fine particle size is desired.
The mechanism of atomization in these nozzles is
Figure 18.39. Spray-dryer installation used to produce 25 by air shattering the liquid streams. At low air
tons/hr of nickel concentrate in a twin installation. The two pressure, the gas blows a bubble of liquid, which
spray chambers are operated in parallel. The feed gas heater collapses into At higher pressures, the
droplets.
is at the left. The cyclones and wet scrubbers are at the liquid issues from the nozzle as ligaments, which
right. The small tank at the bottom right is for liquid are then torn into droplets by the gas stream, the
recycle to the scrubbers. Feed storage is behind this tank. average droplet size decreases as the pressure of
(Courtesy of Niro Atomizer Inc.) both streams to the nozzle increases (20). Two-
fluid nozzles are used in pilot and pharmaceutical
cipitator. Figure 18.39 shows a spray dryer installa- dryers where the chamber volume is small, thus
tion. requiring small droplets, and in drying ceramics,
Any spray drying system has as its major parts etc., where the external gas-slurry mixing prevents

a feed delivery and atomization system, a hot-gas erosion of the nozzle orifice.

Liquid feed control

valvestem

Spray fluid feed

Figure 18.40. Two-fluid nozzle with air-actuated liquid-feed control

system Here air atomizes the liquid stream external to the valve body.

Air pressure, supplied at the central valve chamber, forces the valvestem
away from the seat, thus allowing spraying to start. Dropping the

control air pressure immediately stops the liquid feed, but the atom-
izing airflow continues. Atomizing air should be at 10 to 60 psig,
whereas control air should be about 30 psig. (Courtesy of Spraying
Systems Co.)
484 APPLICATIONS TO EQUIPMENT DESIGN

Single-fluid pressure nozzles operate at higher
throughput and give larger and more uniform
droplet size than does the two-fluid nozzle shown
in Figure 18.40. Thus, they are often used in
production-sized spray dryers. A pressure nozzle
built for spray-dryer service is shown in Figure
18.41. The nozzle produces a high-velocity tangen-
tial motion the liquid being sprayed. The result-
in
ing centrifugal force makes the fluid swirl around
the circumference of the nozzle hole forming an air
core along the axis of the hole. The fluid then swirls
nozzles must be completely homogeneous. If the
feed is formed by mixing several ingredients, agi-
tation must be thorough enough to eliminate any
of these small particles. Even so, fine screens are
usually placed in the feed line ahead of the nozzle
and may be actually built into the nozzle. Homo-
genizers are often incorporated in the high-pressure
positive-displacement pumps used to feed pressure
nozzles. Homogenizers force the feed through
short capillary tubes at high velocity, thus breaking
up the feed stream and violently remixing it.

out into a hollow, cone-shaped sheet that finally
breaks into droplets. Figure 18.42 shows this form
of atomization from a swirl type of pressure nozzle
operating at 30 psig (20) spraying water. To get the
same degree of atomization with a viscous liquid, a
pressure of several thousand pounds per square
inch might be necessary.
Both pneumatic and pressure nozzles require
the fluid being sprayed to flow through narrow
passages. Thus, any lumps, crystals, or other solids
suspended in the fluid will clog the spray nozzle.
Moreover, even the finest grit will wear the nozzle
tip, causing enlargement and uneven atomization.
For these reasons, the material delivered to the
Core
Centrifugal disk atomizers may be used to
spray fluids that cannot be made homogeneous
enough to pass through a nozzle. Moreover, they
produce an extremely uniform droplet size and do
not require a high-pressure feed or impart an axial
velocity to the sprayed droplets. Centrifugal
atomizers are also less affected by variations in the
feed properties such as percent solids or viscosity,
or even by variations in feed flow rate. Such an
atomizer is shown in Figure 18.43. In the atomizer,
the feed liquid is delivered to a wheel rotating at
6,000-20,000 rpm. The fluid is accelerated to a
high centrifugal speed on the disk and forced off
the disk. Usually the fluid leaves the disk as a
liquid film that breaks into droplets — or into liga-
ments and then into droplets. Some disk atomizers
have edge serrations or interruptions that may
prevent the film formation, producing ligaments
directly.
In application, usually only one disk atomizer
is used in a spray dryer, whereas several nozzles
may be used simultaneously. In this way, eroded or
plugged nozzles may be replaced with only a minor
reduction in throughput for the entire dryer. This
ease of maintenance compensates for the fact that
the nozzles are more likely to require replacement
than the disk atomizer.
The choice of atomizer influences the design
of the entire spray chamber. Nozzles spray axially
in the spray chamber, and hence require a tall,

relatively thin tower. Disk atomizers spray radially.

Thus, the spray tower must be larger in diameter
but can be shorter. The cone angle at the bottom
Figure 18.41. Single-fluid pressure nozzle. of the tower is usually 60—70 degrees, and thus
The nozzle is of the grooved-core design.
most of the tower height where a disk atomizer is
Nozzles of this will spray 60 to
design
used will be occupied by this cone. In both cases,
1600gal/hr of Throughput for a
water.
most drying occurs within a few feet of the
single nozzle varies by a factor of 2.5 to 3 as
atomizer. By the time a drying particle can strike
the pressure changes from 1000 to
7000 psig. The nozzle body is normally the chamber wall, if this can occur at all, it must be
whereas the replaceable core
stainless steel, dry enough to avoid sticking.
and nozzle are of hardened steel or
insert After leaving the atomizer, the liquid droplets
tungsten carbide. (Courtesy of Spraying Sys- fall through the hot gas in the drying chamber. If

tems Co.) the feed were pure water, the droplet would
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 485

Figure 18.42.
i
Breakup of liquid issuing from a lucite model of a swirl pressure
nozzle operating at 30 psi. The hollow liquid cone in which waves break into
droplets is clearly shown as is the hollow core within the nozzle (20). (By
permission of A.I.Ch.E., copyright© 1954.)

evaporate at the wet-bulb temperature of the dry-
ing air until it completely disappeared. In drying
solutions or emulsions, the drying particle reaches
a temperature somewhat higher than the wet-bulb
temperature as drying proceeds. Initially, the liquid
evaporates from the droplet surface. The relatively
dry surface may form a tough shell through which
liquid from the interior of the droplet must diffuse
in order to escape. This diffusion is a much slower
and whole spheres. Figure 18.44a shows a photo-
micrograph of a spray-dried organic acid. Here, the
rupturing of droplets was relatively nonviolent,
producing secondary hollow spheres attached to
the original ones. In Figure 18.446, a similar
photomicrograph is shown of a spray-dried mold-
ing powder. Here, rupture of the droplets resulted
in the production of fragments.
It is obvious that the time required for drying

process than the transfer of heat through the drop-
let shell to the interior, so the liquid tends to
evaporate in place. As a result, the droplet swells,
making the shell thinner and diffusion through it
faster. the shell is relatively inelastic and
If
impermeable, the internal evaporation is usually
great enough to cause rupture of the shell, either
producing fragments or making a new hollow bulb
beside the original one. Thus, the typical spray-
dried product consists of ruptured hollow spheres
cannot be calculated by the simple mechanism for
drying beds of porous material. The time will
depend upon the temperature, humidity, and flow
conditions of the drying gas, the size of droplets
produced by the atomizer, and the properties of
the material being dried. Moreover, the properties
of the finished product will depend upon the same
factors. The importance of these factors has long
been recognized, and a considerable amount of
theoretical and experimental work has been done
486 APPLICATIONS TO EQUIPMENT DESIGN

Figure 18.436. Atomizer wheels for various

applications. The upper wheel is a straight
Figure 18.43a. Centrifugal atomizing as-
vane wheel used for many applications for
sembly with an inverted saucer atomizing
low to medium capacities. The lower one is a
by the high shearing
disk. This disk atomizes
higher-capacity wheel with abrasion-resistant
action occurring at the sharp peripheral
inserts for handling of materials such as
edge. It is relatively inexpensive and is used mineral concentrates. Wheels of this design
where frequent replacement is expected. are able to handle feed rates reaching
(Courtesy of Stork-Bowen Engineering. Inc.)
200 tons/hr. (Courtesy of Niro Atomizer
Inc.)
to try to relate them to each other and to the
design of the dryer. Most of this work has pointed
where
toward relating the drop-size distribution from the
atomizer to its design, the properties of the feed, D vs = Sauter mean diameter, microns
and the throughput. The mechanism of drop form- va = relative velocity between liquid and
air, ft/sec
ation differs in the three major classes of
atomizers, and the correlations obtained have had a = surface tension, dynes/cm
to differ for the different atomizers.* H= liquid viscosity, centiposies
3
Two-fluid atomizers form a spray by imping- p L = liquid density, lb/ft
ing a high-velocity gas jet cocurrently against the
Q, , Q A = volumetric flow rates of liquid and air

liquid issuing from a nozzle as discussed above. For

In pressure nozzles, the spray is usually
the breakup of alcohol-glycerine mixes in an air jet,
formed by forcing a centrifugally swirling liquid
Nukiyama and Tanasawa (24) obtained through an orifice. Normally the swirl produces an

D« =
1410 N

'a\ Tl
^
+ 191 ^rc-^r
Vplo Qa
(18.19)
air core through the center of the nozzle to the end
of the swirl chamber. In such a nozzle, the mean
drop size, usually expressed as a Sauter mean, is a
function of nozzle diameter, pressure applied,
•The description of a statistically large sample of small
liquid and, to a lesser extent, surface
viscosity,
particles so that the particle-size distribution may be known is

discussed along with other properties of particulate-scJid systems in

tension. Figures 18.45 and 18.46 give some indica-
Appendix B. In the following paragraphs the reader is expected to tion of the mean drop sizes expected from pressure
be familiar with the material in this appendix. nozzles. Empirical correlations have been
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 487

Figure 18.44a. Photomicrograph of an organic spray dried from a true solution.

(Courtesy Stork-Bowen Engr. Co.)

Figure 18.446. Photomicrograph of a ceramic molding powder spray dried from

a true emulsion. (Courtesy of Stork-Bowen Engr. Co.)

developed by assuming that the Sauter mean these lines. A

modified standard deviation was
diameter is independently a function of the tangen- found to be function of tangential velocity and
a
tial and the axial velocity components at the orifice diameter. Other empirical correlations for
nozzle tip (31). The measured distributions were drop-size distribution have been offered, based
found to give a on probability
linear distribution upon more measured properties of the
easily
paper. The drop-size uniformity was correlated in nozzle, but they are usually applicable only to the
terms of the standard deviation as measured from nozzle design studied. In most of the studies, water
488 APPLICATIONS TO EQUIPMENT DESIGN

at constant feed pressure as long as atomization

occurs. This is because the air-core diameter
decreases as the viscosity increases. Nozzle capaci-
« 500 ties are frequently given in terms of a "flow
number" that is pressure insensitive. Defining the
flow number by the equation

Q
K' = 18.19a)

where
Q = capacity, vol/unit time
K' = flow number, which is related to the
efficiency of conversion of pressure to
kinetic energy in the nozzle
100
0.06 0.10 0.20 0.30 0.40 it is found that K' is relatively independent of
Orifice diameter, inches
pressure but varies somewhat with liquid viscosity
Figure 18.45. Effect of orifice diameter on mean and markedly with nozzle design. Thus K' can be
drop size from pressure nozzles. />=50psig, fluid used to obtain an estimate of the capacity of a
H2 (26). (By permission of Amer. Chem. Soc, nozzle at P 2 if its capacity at Py is known. Figure
copyright© 1951.) 18.47 shows the relation between orifice diameter,
pressure, and capacity of grooved-core nozzles with
140 . .

fixed grooved area spraying a fluid of 50-centipoise

viscosity.
In spinning-disk atomization, the fluid leaves
El20 the disk in stream; that break into droplets from
N^ ^"*\^ friction as they pass through the relatively still gas.

^v. >^,18.5 centistokes

100
v^ ^^^.^^
^L
>. 80
». 2_

60
100 150 200 250 300
1
P. lb/in/

Figure 18.46. Influence of pressure and viscosity on

the Sauter mean diameter of droplets from a swirl
pressure (By permission of
nozzle (18). Inst, for
Fuels, London, copyright © 1949.)

was the primary atomized. The effect of fluid

fluid
upon the drop size
properties, particularly viscosity
produced has not been adequately determined.
The capacity of the nozzles depends upon
nozzle diameter, pressure, internal-nozzle design,
and fluid viscosity. The importance of internal 0.02 0.03 0.04 0.05 0.06
design is in the variation in tangential velocity Orifice diameter, in.

component imparted to the swirling liquid and in Figure 18.47. Effect of nozzle diameter on the cap-
pressure drop through the nozzle for a given size. acity of grooved-core pressure nozzles (22). Groove
The influence of viscosity is minor, but, in general, area 7.7 x 10 in . Viscosity 50 centipoises. (Cour-
tesy A.I.Ch.E.)
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 489

The feed rate can be controlled independent of the kg = gas-phase mass-transfer coefficient, lb
disk. Although the hydrodynamics of this breakup moles/hr ft 2 atm
are complicated, the correlation of mean drop sizes M = mean molecular weight of gas phase
from spinning disks and the generalizing of drop- 9> = gas-phase diffusivity ft 2 /hr
size distribution have been more successful than
which is of practical and theoretical importance as
they were for two-fluid atomizers or for pressure
it fixes the conditions during evaporation to dry-
nozzles. Friedman, Gluckert, and Marshall (10)
ness of a liquid drop. Experimental studies (27)
correlated Sauter mean diameter through the
give

«
dimensionless equation
k MD p P
- = a
(g) tV) (ia20)
°
</
p
=2 + ^ 1
(/V s<r(/V Re )'
7
(18.22)

where where K^ , m, and n are constants. This equation

should give by analogy (12, 20)
=r disk radius
=G mass velocity, lb/(ft of Lw ) (unit time)
o= surface tension = 2+ K2 (N Pr )P(N Be )« (18.23)
Lw = wetted periphery, ft
®k
N= disk speed, rpm where
The spray distribution was found to form a straight Ki = K2 = 0.6
line on log-normal probability paper. Thus, the
entire spray pattern could be determined through q=n=\
the Sauter mean diameter and one other point. The
At high A/ Re the effect of the constant 2 becomes
point chosen was the maximum drop size. ,

Friedman found the maximum particle size D pm = negligible so that Equations 18.22 and 18.23 con-
vert into the usual /-factor relation.
3D VS through a dimensionless correlation, but
other workers found more uniform spray distribu- The time of exposure of a drop to the spray-
tions (20).
drying atmosphere depends upon the drop size,
particle shape and density, and gas-flow path and
The rates of heat and mass transfer to a falling
velocity. The gas-flow pattern in a dryer is norm-
drop and hence the drying rate have been studied
ally quite complex, and hence the particles travel
using boundary-layer theory. The
show that
results
these rates should be a maximum
the side of the longer paths than would be calculated from the
at
drop facing the oncoming air, should decrease to a minimal trajectories. Models of flow patterns in a
minimum near the boundary-layer separation point dryer with countercurrent air-spray flow (25) have
behind the drop equator, and should increase again shown that the flow is divided into separate spiral-
to a higher value at the rear center of the drop. ing flow in the chamber core and along the walls.

From these considerations, combined with heat- In the region near the spray point, these spirals

and mass-transfer equations, a set of simultaneous are broken and flow proceeds more directly. The

partial-differential equations results, from which usual equations for free fall of a particle through a

the rates of heat, mass, and momentum transfer fluid (see Chapter 22) will apply after the initial jet

can be determined. The general solution of these action from the nozzle has been spent. Of course,

equations involves so many assumptions as to be of the numerical values employed must change
no practical value. In the limit for a finely
throughout the particle-size range that is present,
dispersed system, the A/ Re approaches zero, and,
and will also change as the particle changes shape
for this case, the solution gives (20)
and density during fall.
The time required to dry a droplet can be
h c Dp kg MDp P estimated from the usual drying-rate equations if
^ ;18.21: they can be solved. The constant-rate period will
k ®p end when the drop surface becomes solid. This
where
may occur while the interior of the drop is still a
h c = convective heat-transfer coefficient, relatively dilute solution. If swelling and rupture
2o do not then occur, the rate during the falling-rate
Btu/hrft F
k = gas-phase thermal conductivity, Btu/hr period will be controlled by diffusion through the
ft°F shell.
490 APPLICATIONS TO EQUIPMENT DESIGN

The informationgiven above should give a The dryers discussed above are only a few of
clear picture the fundamental
of principles the many commercial types. They have been dis-
involved in spray drying and an awareness of the cussed here because they illustrate a wide variety
extent to which these principles can be used in of dryers and because they are among the most
design. Seldom will this be sufficient to design a common. Under special conditions, some dryers
spray dryer. The usual design approach is to esti- may differ fundamentally — for instance, infrared
mate roughly the cost of the required dryer using dryers, where most of the heat is transferred by
accumulated experience as a basis (6), and then radiation. But most dryers differ only in minor
explore the drying process itself by submitting detail from the ones discussed here. For example,
samples of the material to be dried to dryer manu- cylinder dryers such as those used in drying paper
facturers for pilot runs. are essentially drum dryers to which the material is
The production engineer must know how vari- fed in sheet form.
ations in feed or operating conditions affect the
spray-dried product. The effects differ markedly
from product to product, and only a few generaliz- REFERENCES
ations can be made concerning them. Duffie and
Marshall (8) studied the effect of feed temperature, Bradshaw, R. and Myers, A.I.Ch.E. Jour., 9,
1. D., J. E.
airtemperature, and feed concentration upon the p. 590 (1963).
bulk density of product for typical spray-dried 2. Carberry.J. J., A.I.Ch.E. Jour., 6, p. 460 (1960).
materials and compared their results with other 3. Ceaglske, N. H., and 0. A. Hougen, Trans. AJ.Ch.E.,
published results. Only in the case of air tempera- 33, p. 263 (1937); Ind. Eng. Chem.,29. p. 805 (1937).
ture was the behavior consistent for all the 4. Chirife, J., and R. G. Gardner, Ind. Eng. Chem.,
materials studied, ranging between inorganic salts Process Des. Dev., 11, p. 312 (1972).
and corn syrup. In all cases, increased inlet-air 5. Chu, C, A. M. Lowe, D. Conklin, Ind. Eng. Chem.,
J.

temperature decreased the bulk density. In the case 45, 1586 (1953).
p.
6. Cook, E. M., Chem. Eng. Prog.. 62, No. 6, p. 93
of feed concentration and temperature, effects
(1966).
upon atomization and drying behavior influenced
7. Debrand, S., Ind. Eng. Chem. Proc. Des. & Devel., 13,
the product bulk density but in conflicting ways.
p. 396 (1974).
Increasing feed concentration put more solid in 8. Duffie, J. A., and W. R. Marshall, Chem. Eng. Progr.,
each droplet of fixed size but also increased the 49, p. 480(1953).
average particle size. The net effect upon most 9. Findlay, A., A. N. Campbell, N. O. Smith, 77?e Phase
materials studied was a slight decrease in product Rule and Its Applications, 9th ed., Dover Publications,
bulk density with increased feed concentration. New York (1951).
Increasing the feed temperature decreased its vis- 10. Friedman, S. J., R. A. Gluckert, W. R. Marshall, Chem.

cosity, which decreased the pressure drop through Eng. Progr., 48, p. 181 (1952).
11. Friedman, S. J., and W. R. Marshall, Chem. Eng. Progr.,
a nozzle and resulted in a slightly smaller sprayed
45, pp. 482-573 (1949).
droplet. On the other hand, the initial warm-up
12. Frossling, N., Gerlands Beitr Geophys., 52, p. 170
period may be faster, the particle swelling greater,
(1938).
and final moisture content lower. On most materials Fulford, G. C, Can. Chem. Eng., 47, p. 378 (1969).
13. J.
sprayed, the final bulk density decreased 14. Gilliland, E. R., Ind. Eng. Chem., 30, p. 506 (1938).
slightly with increased feed temperature, although 15. Glasstone, S., Textbook of Physical Chemistry, 2nd
gelatin appeared to act in a reverse manner. ed., D. Van Nostrand, Princeton (1946), p. 487.
Qualitatively, the normal result of other oper- 16. Hinze, and H. Mulborn,
J. O., J. Appl. Mechanics, 17,
ating changes may be inferred from experience. p. 145(1950).
Increasing the pressure to a spray nozzle should 17. International Critical Tables, Vol 2, pp. 322-325
increase the throughput and the droplet size. The (1926).
18. Joyce, J. R., J. Inst. Fuel, 22, p. 150 (1949).
increase should result in lower air temperatures and
19. Laughlin, H. G., Trans. A.I.Ch.E., 36, p. 345 (1940).
a product of higher moisture content, larger
20. Marshall, W. R., Jr., "Atomization and Spray Drying,"
particle size, and higher bulk density. Increasing
Chem. Eng. Progr. Monograph Series No. 50, 2 (1954).
the throughput to a disk atomizer should have 21. McCready, D. W., and W. L. McCabe, Trans. A.I.Ch.E.,
similar effects. If, at the same time, the air temper- 29, p. 131 (1933).
ature is increased, the net result may be a product 22. Mclrvine, J. D., M.S. Thesis, University of Wisconsin
of larger particle size but relatively unchanged (1953).
moisture content or bulk density. 23. Newman, A. B., Trans. A.I.Ch.E., 27, p. 203 (1931).
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 491

24. Nukiyama, and Y. Tanasawa, Trans. Soc. Mech.

S., (a) The critical moisture content.
Engrs. {Japan), 4, No.
14 p. 86 and No. 15, p. 138 (b) The drying rate during the constant-rate period.
(1938); 5, No. 18, pp. 63 and 68 (1939); 6, No. 22 (c) The heat-transfer coefficient observed during the
ll-7and No. 23 II-8 (1940). constant-rate period.
25. Paris, J. R., P. N. Ross, Jr., S P. Dastur, R. L. Morris,
Compare the answer for part (c) with that obtained
Ind. Eng. Chem. Proc. Des. & Devel. 10, p. 157
by using Equation 18.12.
(1971).
26. Pigford, R. L., and C. Pyle, Ind. Eng. Chem., 43,
18.2. A fibrous material is dried in a countercurrent,

1649 (1951).
p.
adiabatic, tunnel dryer in sheets 1 m by 3 m by 2 cm in

27. W. E., and W. R. Marshall, Jr., Chem. Eng.

Ranz, size. The sheets are carried through the dryer on racks
Progr.,48, pp. 141, 173 (1952). holding 20 sheets separated so that each sheet dries from
28. Seborg, C. 0., Ind. Eng. Chem., 29 p. 169 (1937). both sides. Fifty kg of air at 130°C with a 35°C wet-bulb
29. Shepherd, C. B., C. Hadlock, R. C. Brewer, Ind. Eng. temperature enters the dryer per pound of dry stock and
Chem., 30, p. 388 (1938). flows around the stock at a rate of 4000 kg/hr m2 of free
30. Sherwood, T. K., Ind. Eng. Chem., 21, 12, p. 976 area. The stock enters at 30°C with a moisture content of
(1929); Trans. A.I.Ch.E., 32, p. 150 (1936). 0.80 kg H20/kg of dry stock and leaves with a moisture
31 Tate, R. W., and W. R. Marshall, Jr., Chem. Eng. Progr.,
content of 0.04 kg H 2 0/kg of dry stock. The stock has a
49, pp. 169,226 (1953).
bulk dry density of 200 kg/m 3 and a heat capacity of 1.3
32. Toner, R. K., C. F. Bowen, J. C. Whitwell, Textile
J/g°C. Pilot tests indicate that a critical moisture content of
Research J., 17, p. 7 (1947).
0.25 kg H20/kg of dry stock can be expected, but there is
33. Wenzel, L. A., Ph.D. Thesis, University of Michigan
(1949). negligible equilibrium moisture content. Determine the dry-
34. Wenzel, L. A., and R. R. White, Ind. Eng. Chem.. 43, ing time required and the condition of the exhaust air.
3
p. 1829 (1951). 18.3. Brick clay of bulk density p B = 110 lb/ft is

being dried in a conveyor dryer with a 3 ft wide belt. The

PROBLEMS brick-clay layer is 1 in. thick, and air flows countercurrent
to the belt. The space for airflow above the belt is 1 ft high
and is equal to the belt width. Air enters the dryer at 80° F
18.1. The following data were obtained when drying with a 50° F wet-bulb temperature. It is mixed with an
sand in superheated steam (33): equalamount (on a weight-of-dry-air basis) of recycle air,
and the combined air stream is heated to 250 F in a
Drying Sample Drying Sample
Time, hr Weight, lb steam-tube heater. After passing over the sample, half the
Time, hr Weight, lb
air is discharged from the dryer; the other half is reheated.

43.72 5.00 37.93 Exit air is at 80 percent relative humidity, product is at

0.25 43.32 5.25 37.70 Xf = 0.2 lb H 2 0/lb dry solids, and feed enters at X = '

0.50 42.95 5.50 37.48 1.5 1b H20/lb dry solids. For this material the critical
0.75 42.54 5.75 37.28 moisture content, Xc is 0.15 lb H 2 0/lb dry solids. Assum-
'

1.00 42.21 6.00 37.12

ing that heat and mass are only transferred to the top
1.25 41.85 6.25 36.90
surface of the clay and that the clay enters the dryer at the
1.50 41.52 6.50 36.73
ad iabatic-satu ration temperature of the exit air, determine:
1.75 41.20 6.75 36.58
The mass flow rate of air, Ib/hr
(a) ft
2 passing over
2.00 40.89 7.00 36.42 ,

2.25 40.57 7.50 36.22 the belt if this belt moves at a velocity of 1 ft/min.

2.50 40.30 aoo 36.05 (b) The length of belt required.

2.75 40.03 8.50 35.83 18.4. A conveyor dryer is used to dry wood chips
3.00 39.81 9.00 35.69 from an initial moisture content of 0.3 kg/kg of wood to a
3.25 39.59 9.50 35.61 final moisture content of 0.02 kg H 2 0/kg dry solids. The
3.50 39.36 10.00 35.50 chips move on a screen conveyor at a rate of 0.1 m/sec and
3.75 39.08 10.50 35.39
at a depth of 10 cm. Drying air, initially at 120°C with a
4.00 38.84 11.00 35.35
humidity of 0.005 kg H 2 0/kg dry air, flows through the
4.25 3&60 11.50 35.33
bed at a rate of 5 cm/sec. The wood chips are about
4.50 3a40 1200 35.31
0.5 cm-cubes. Laboratory tests on a single chip have yielded
4.75 38.16
a critical moisture content of 0.1 kg/kg of wood, and an
The sample was 2 in. thick and weighed 35.28 lb when dry. equilibrium moisture content of 0.01 kg H 2 0/kg of dry
It was dried in steam at 50 psia, superheated 53.5 F, and solids when exposed to air at the initial air conditions.
2 Estimate the length of dryer required to produce an
flowing at 1000 Ib/hr ft Drying was from the top face
.

only, and this face measured 8 in. by 29* in. Determine: average product that meets the desired specifications.
492 APPLICATIONS TO EQUIPMENT DESIGN

18.5. The following data on the drum drying of of 0.003 lb H 2 0/lb dry air, and at a flow rate of
2
sodium acetate on a trough-fed double-drum dryer have 1 500 Ib/hr ft of dryer cross section. Solids, flowing
been reported (Harcourt, paper presented at the Niagara countercurrent to the air, enter at 0.4 lb H 2 0/lb dry solids

Falls Meeting, ASME, September 17-23, 1936): and leave at 0.02 H 2 0/lb dry solids. Product is obtained
lb

at a rate of 1500 Ib/hr. Throughout the dryer the solids

Mc>isture exist at the wet-bulb temperature of the drying air.
Content, Feed Capacity,
If the dryer is sloped at 1 in./ft of length and rotates
weight percent Steam Drum Tempera- lb
at 5 rpm, determine the length of dryer required.
Pressure, Speed, ture product/
°F hrft 2 18.8. A Cr-Ni catalyst (p = 2.65 g/cm 3 in the form
Feed Product psia rpm )

of a washed filter cake is dried in a flash dryer. It enters the

39.5 0.44 70 3 205 1.57 dryer at a moisture content of 0.5 g H 2 0/g solids at a
40.5 10.03 67 8 200 5.16 temperature of 30° C. The drying airstream enters at 300°C
63.5 9.53 67 8 170 3.26 and at a flow rate of 15 m/sec. Its mass rate is 5 kg/kg of
solids flowing. The drying particles are typically 0.5 mm in

proposed to drum-dry sodium acetate from an initial

diameter and reside in the dryer a median of 10 sec.
It is

60 percent solids by weight to 95 percent solids in a Determine the median moisture content of the product
catalyst particles.
double-drum dryer 10 ft long by 3 ft in diameter. Steam at
70 psia will be used as heating medium, and the feed will
18.9. Coal of particle size averaging 2 cm in diameter

be preheated to 200° F. Recommend a drum speed to give is to be dried in a fluid bed dryer. The fluidizing medium is

the desired product concentration, and estimate the pro-

combustion gas, 80 mole percent N 2 , 15 percent C0 2 , 5

duction rate in pounds per hour of dried product. Discuss percent CO at 500 C. The bed
expanded to an average 10
is

the areas of uncertainty in your solution, list any percent solids, with the gas flowing at 5 m/s. The coal

assumptions you have made, and suggest any additional enters at 0.2 g H 2 0/g dry solids, and the equilibrium
physical or chemical data that would decrease the uncer- moisture content is 0.02 g H 2 0/g dry solids. 15,000 kg/hr

tainty in your answer.

of coal is to be dried to a final moisture content of

18.6. In a pilot process, mashed potatoes are dried in 0.04 g H 2 0/g dry solids. The coal being dried has a density
3
pans \ in. deep and insulated on the bottom. Drying air is
of 1.6 g/cm and a heat capacity of 0.8 J/g°C. It enters
,

at 180°F with a 10 percent relative humidity. Drying from the bed at 30°C.

an initial moisture content of 0.6 lb H 2 0/lb of dry solids to Determine the residence time required by the coal
a final moisture content of 0.15 lb/lb of dry solids requires being dried, assuming the bed is perfectly mixed, and the

6 hr. All the drying is in the falling-rate period and is

volume of bed required.
diffusion controlled. The equilibrium moisture content is
18.10. A rotary countercurrent dryer is fed with wet

0.1 lb/lb of dry solids when in contact with air at this

sand containing 50 percent moisture and is discharging sand

temperature and humidity. Bulk density of the product is

containing 3 percent moisture. The entering air is at 120°C
0.75 g/cm 3 .
and has an absolute humidity of 0.007 g H 2 0/g air. The
In the plant process, the potatoes are to be dried wet sand enters at 25°C and leaves at 40°C. The air leaves

from the same initial moisture content as in the pilot at 45°C. The wet sand input is 10 kg/s. Radiation amounts

process to a final moisture content of 0.25 lb/lb. The to 5 cal/g dry air.

process will take place in a pan dryer with pans 2 in. deep Calculate the pounds of dry air passing through the

but with perforated metal bottoms so that drying occurs dryer and the humidity of the air leaving the dryer.

from both faces. If the drying-air conditions are controlled

to duplicate those in the pilot dryer, what drying time will Latent heat of H2 at 40°C = 575 cal/g

be required? Cp for dry sand = 0.21 cal/g°C

A steam-tube rotary dryer Cp for dry air = 0.238 cal/g°C

18.7. is to be used to dry
crushed dolomite = 180 lb/ft 3 of -14 + 20 mesh Cp for H2 vapor = 0.48 cal/g°C
(p e )

particle size (Dp = 0.039 in.). The dryer is 4 ft in

range
diameter and can be purchased in 5-ft increments of length. 18.11. A single-shell, direct, countercurrent rotary

The 40 steam tubes are 3 in. in diameter located in two dryer needed to dry 10 tons/hr of manganese ore at 80° F
is

concentric rings around the inside of the shell. They are of with an average particle diameter of 0.04 in. from an initial
the low-finned variety with a total surface area of 40 ft 2 /ft moisture content of 0.1 lb H 2 0/lb dry solids to a final
of dryer length. Previous experience indicates that the over- moisture content of 0.005 lb H 2 0/lb dry solids. At these
all coefficient of heat transfer from the condensing 325° F conditions, the critical moisture content will be less than
steam to the air in the dryer will be about 10 Btu/hr ft
2 °F. 0.005 lb/lb. The drying medium will be air initially at

Air enter's the dryer at 300°F with an initial humidity 80° F with a 70° F wet-bulb temperature heated to 260°
 SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 493

before entrance to the dryer. Specify dryer length, production rate of 1000 Ib/hr of dried product. Use Equa-
diameter, pitch, rpm, and airflow rate. tion 22.15 for settling velocity.
18.12. Soluble coffee is spray dried from a 25 percent (b) Why does the answer obtained above differ so
3
solids solution having a density of 0.95 g/cm and a radically from the height of commercial spray dryers?
viscosity of 5 centipoises. It is sprayed from a swirl-type 18.13. In the spray described in Problem 18.12 the
pressure nozzle at 250 psia. For the purposes of this drop-size distribution is found to follow the normal prob-
problem, consider the drops to be uniform in size. Drying ability function; that is

occurs concurrently. Assume that the sprayed particles

neither swell nor shrink and that the drying history follows 1 -(Dp -Dp) 2
f(DD )o
a constant-rate— linear-falling-rate pattern. Critical moisture 2o2
\/27T
content is 0.5 lb H^O/lb dry solids, and equilibrium
moisture content is 0.1 lb H20/lb dry solids. Air enters the wberef(Dp) is the population density of drops of diameter
dryer at 450°F and a humidity of 0.01 lb H 2 0/lb dry air, Dp Dp
, is mean drop diameter, a is standard deviation,
and leaves at 200°F. o =\fZ(Dp — Dp) 2 In, and n is the total number of drops.
(a) Determine the height of dryer required to dry to Plot this drop-size distribution when 2a = CL S , the Sauter
an average moisture content of 0.03 lb H 2 0/lb dry solid, mean diameter. Outline the steps needed to solve Problem
and the required airflow rate and tower cross section for a 18.12 using this drop-size distribution.
 Nineteen

Simultaneous Heat and Mass

Transfer —
Evaporation and Crystallization

In an evaporation operation, a solution is concen- composition diagram for the sodium— hydroxide-
trated by boiling off the solvent. Usually, the water system at 1 atm total pressure. Equilibrium
desired product is the concentrated solution, but concentrations of vapor and liquid phases occur at
occasionally the evaporated solvent is the primary the intersection of saturated liquid and vapor
product — for example, in the evaporation of sea- boundary lines with a line of constant temperature.
water to make potable water. The concentrating Note that a discernible concentration of sodium
may be continued until the solution becomes hydroxide in the vapor phase is not obtained until
saturated and further until the solute precipitates the liquid phase reaches 95 percent NaOH and
as a crystalline solid. In this case, the operation is boils at some 700° F.
o fte n c a ed c/ys talliza tio n
1 1
Despite the multitude of rate processes occur-
In either case, several rate processes occur. ring, the engineer concerned with these operations
First, heat is transferred from the heating medium can usually consider the entire process in terms of
to the solution. The occurs
transfer usually the heat transfer from heater to solution. This step
through but may be a direct transfer
a solid surface controls the overall rate and is most easily
from combustion gases to the evaporating solution. analyzed. The contributions of the other transfer
Second, mass and heat are simultaneously trans-
ferred from the liquid to the vapor phase. For this
transfer process, all the rate, balance, and equilib- 2530T
rium equations that were discussed for a humidifica-
tion or a distillation process apply. Finally, for a
crystallization operation, there is also the simul-
taneous transfer of heat and mass between solu-
tions and solid phases. Here again, all the considera-
tions mentioned above apply. This process
sequence differs from the operations discussed in
Chapters 17 and 18 mainly in that it is not nearly
adiabatic.
The evaporation process is really equivalent to
a single-stage separation. But the components norm-
ally separate so sharply that the composition
equilibrium is frequently forgotten. In applications
where the vapor phase is the valuable product, this
equilibrium may be important. For example, if the
vapor is to be condensed and recycled to a boiler
0.2 0.4 0.6 0.8
for reevaporation, the must usually be
stream Mass fraction NaOH
specially treated to adjust pH and remove trace Figure 19.1. Phase equilibrium for H 2 0-
impurities. Figure 19.1 is a temperature- NaOH system at 1 atm pressure.

494
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 495

processes are minor in their effect on the overall ness of metal. Some
experimentally determined
operation. Moreover, so little is known, outside of but they are
overall coefficients are available (16),
the few companies that build evaporators, of most seldom adequate as a basis for commercial evapor-
of the refinements in design that there is no reason ator design.
to treat here more than the heat and material
balances and the simple formulation of heat- EVAPORATION
transfer rate through the heating surface.
As discussed above, evaporation is the operation of
The initial heat-transfer process usually
concentrating a solution by boiling away the
requires the transfer of heat from a heating fluid
through a dividing wall to the evaporating solution.
solvent. The concentration is normally stopped
before the solute begins to precipitate from solu-
Thus, the series-resistance concepts developed in
tion. Basically, then, an evaporator must consist of
Chapter 14 apply. Equation 14.31 may be
a heat exchanger capable of boiling the solution
modified to
7"! -T 2
and a device to separate the vapor phase from the
q= - —r- (19.1? boiling liquid. In its most simple form, it might be
1 Ax n 1
a pan of liquid sitting on a hot plate. The surface
+ - - + Rd +
hyAy kA, h 2 A; of the hot plate is a simple heat exchanger, and
vapor disengaging is obtained by the large area for
where R d is resistance of wall fouling, 1 refers to
vapor flow and its consequent slow rate of flow. In
the heating fluid, and 2 refers to the evaporating
industrial operation the equipment is usually
solutions. The heating fluid is usually condensing
arranged for continuous operation, the heat-
steam so that hy is the steam-condensing co-
exchange surface is vastly increased, boiling is
efficient, although some small fraction of the heat
much more violent, and vapor evolution is rapid.
may be transferred to the solution from super- Such problems as foaming, scaling, heat sensitivity,
heated steam before the steam reaches the con-
corrosion, and space limitations are met. These
densing temperature. The wall fouling depends
problems have resulted in variations and refine-
upon the solution being evaporated, the time in ments in evaporator design to meet different com-
service since cleaning, and the temperature dif-
binations of solution properties and economic
ference across the surface. The coefficient h 2 is
conditions.
that for boiling solution, although some heat may
be transferred to the solution before it reaches the Evaporator Construction
boiling point. Obviously, all three of the terms can
Evaporator-body construction falls into a few
be estimated only very roughly from theoretical
general categories, illustrated in Figures 19.2
equations or empirical correlations. The overall
through 19.9.
coefficient that results is then extremely uncertain.
As a result, the heat-transfer rate is usually
expressed by the simple equation Vapor

q = U A {-AT) (14.31)

where U the overall heat-transfer coefficient

,

based upon outside surface area, is determined

experimentally in pilot equipment or from previ-
ous experience. Experience has shown U to
depend upon the properties of the solution, the Feed Steam
heating medium, and the surface geometry and
type. The solution properties are fixed only by Boiling solution
outside tubes
specifying the components and their concentra-
tions, thelocal pressure, and the details of fluid
motion. The usual phase-rule considerations would
allow the substitution of local temperature for Condensate I

either the pressure or one of the component con- or drips I

centrations. Similar specification is necessary to

Thick
determine the heating-medium conditions. The sur- i liquor

face conditions would include cleanliness and Figure 19.2. Cross-sectional diagram of horizontal-tube
smoothness of surface and composition and thick- evaporator.
496 APPLICATIONS TO EQUIPMENT DESIGN

The horizontal-tube evaporator of Figure 19.2

is one of the classic construction types and has
been widely used for many years, The solution to
be evaporated boils outside of horizontal tubes
within which steam condenses. The horizontal
tubes interfere with the natural circulation of the
boiling liquid and thus minimize liquid agitation. Steam (condensing
outside tubes)
As a result,the overall heat-transfer coefficient is
lower than in other forms of evaporators, especi-
tH-
J
ally if the solution is viscous. No provision is made
Impingement baffle for
for breaking foam that occurs because of the boil- separating liquid
ing action. Overall coefficients of 200— from vapor

400Btu/hrft 2 °F (1100-2300 W/m 2 K) result,

depending upon the overall temperature difference,
Liquid return
boiling temperature, and solution properties. More-
over, fouling from the evaporating solution builds
Vent
up on the outside of the tubes where it cannot be
as easily removed as it could be from the inside of
the tubes. Traditionally, the horizontal tubes have
been inserted into the steam chests with packed
glands rather than by rolled or welded joints. For
these reasons, horizontal tubes are used mainly in
where the solution to be treated
small installations
is foams nor deposits solids on
dilute and neither
the evaporator tubes, or where materials of con-
struction preclude welding or rolling of tubes.
The vertical-tube evaporators of the basket Figure 19.3. Cross-sectional diagram of basket evaporator.

and standard vertical varieties, which are shown in

Figures 19.3 and 19.4, respectively, are distinct
improvements over the horizontal-tube evaporator.
In both of them the solution boils inside vertical
tubes with the heating medium, usually condensing
steam, held in a chest through which the tubes
pass. In the basket evaporator, the steam chest
forms a basket hung in the center of the evapor-
ator. Boiling or heating of the liquid in the tubes
causes flow upward through the tubes, and un-
evaporated liquid flows downward through the
annulus around the basket. In the standard
vertical-tube evaporator, the steam chest is dough-
nut shaped. Liquid flows upward through the tubes
and downward through the central hole. In large
installations, there may be several return ducts
instead of the single central one shown here. In Steam
both types, the tubes are inserted into tube sheets
^team condensing
by rolling or welding, which considerably reduces outside tubes
the cost compared to the packed glands tradition-
ally used in the horizontal-tube evaporator.
These evaporators overcome most of the
operational disadvantages of the horizontal-tube
evaporator. Natural circulation is promoted, having
been measured at between 1 and 3 ft/sec in the Figure 19.4. Cross-sectional diagram of stand-
tubes (4). As a result, coefficients are somewhat ard vertical-tube evaporator with natural circu-
higher than in horizontal-tube evaporators and lation.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 497

20
range between 200-500 Btu/hr ft F (1100-
2800W/m 2 K), depending again upon solution
properties, overall AT, and boiling temperature.
Any solid deposit will build up inside the tubes
where it is readily removed by mechanical cleaning.
Foam breaking is ineffective, although entrainment
separators and impingement baffles, which are
usually supplied, will reduce the foam buildup,
Viscous liquids can be handled, but circulation is
slow, and the coefficients obtained are poor. Thus,
vertical-tube evaporators are completely satis-
factory for most evaporation demands and are
impractical only where the liquid being evaporated
is very viscous, foams markedly, or may be subject

to evaporator temperatures for very short periods

only.
Forced-circulation evaporators are shown in
Figures 19.5 and 19.6. In these evaporators, the

Vapor

Figure 19.6. Cross-sectional diagram of forced-circulation

evaporator with an external horizontal heater.

evaporating liquid is pumped through a heat

exchanger where the heating medium surrounds
the tubes carrying the solution. Pressure drop and
hydrostatic head in combination are frequently
Impingement baffle
for separating liquid great enough to prevent the solution from boiling
from vapor in the exchanger tubes, so that the vapor generated

is flashed off as the liquid enters the disengaging

space. Since the velocity of the flashing mixture is

high, impingement baffles are important to mini-

mize entrainment. A properly designed baffle will
Steam (condensing promote coalescence of small bubbles as well as
outside tubes)
merely changing the direction of flow. Modern
forced-circulation evaporators are usually equipped
with external heaters, as shown in Figure 19.6,
rather than with heat-transfer surface built into the
evaporator body. This makes cleaning tubes and
replacing any corroded or eroded tube much
simpler that with the internal heating element. It
also permits the construction of a more compact
unit, so that it can be installed in spaces with low
headroom. In evaporating some solutions, it is
important to prevent boiling in the tubes in order
Feed to reduce solids deposition. In the evaporator with
Circulating pump
external heating, boiling can be easily prevented by

w
Figure 19.5. Cross-sectional diagram of vertical-
tube evaporator with forced circulation.
lowering the heater relative to the disengaging
space. It cannot be prevented so easily when the
heat-exchange surface is inside the evaporator
body.
 498 .APPLICATIONS TO EQUIPMENT DESIGN

In forced-circulation evaporators, the heat-

transfer coefficient will depend upon circulation
rate as well as overall A T, boiling temperature, and
system properties. At low circulation rates, boiling
occurs throughout much of the tube length. Boil-
ing increases the turbulence and may make the
boiling-side coefficient twice as large as it would be
without boiling. The fraction of the liquor vapor-
ized while passing through the tube is very small;
hence, total circulation through the tubes is many
times larger than the feed rate. At circulation rates
over about 4 ft/sec, boiling in the tube is almost
completely suppressed, and the liquor-side co-
efficient can be reasonably predicated from
Equation 13.65 for forced convection inside tubes.
Overall coefficients varying between 200—
1 200 Btu/hr ft
2 ° F ( 1 1 00-7000 W/m 2 K) have been
reported, with normal experience giving values of
500-1 000 Btu/hr ft
2 ° F (2800-5700 W/m 2 K).
Such higher coefficients permit reduction in the
size of unit required for a given application, but
the saving in first balanced by the power
cost is

cost of pumping the solution in the recirculation

loop. Thus the decision to use a forced-circulation
evaporator depends upon a favorable economic
balance that includes the evaporator first cost, the
power costs of operating the recirculating pump,
the cost of maintaining the pump, and the compara-
tive cost of cleaning the evaporator tubes. High
circulation velocities may be attractive when
moderate- or high-pressure steam is to be used,
with the steam used first in a turbine to drive the
Figure 19.7. Cross-sectional diagram of long-tube
pumps and then used as the heating medium. In
vertical evaporator.
the case of viscous fluids or of solutions that must
be kept from boiling in the tubes to minimize
solids deposition, the choice of a forced-circulation pump is required. The fraction evaporated per pass
evaporator may be almost mandatory. is usually larger than for forced -circulation opera-
The long-tube vertical evaporator, shown in tion. If the exchanger surface can be
desired,
Figure 19.7, obtains reasonably high liquor flow located external to the evaporator body to permit
through the tubes by natural convection. The tubes easy cleaning. Although viscous fluids cannot be
are usually 12 to 20 ft long. The vapor-liquid mix- handled, foaming materials can be, and, for certain
ture issues from the top of the tubes and impinges heat-sensitive materials, this evaporator has the
on a baffle. The liquor velocity is high enough that advantage of being readily operated without re-
the baffle acts as an effective foam breaker. Little circulation (i.e., as a once-through unit). Evapora-
published information is available on coefficients tion per pass through the tubes is normally far
obtained in these evaporators, but overall higher than in other types of natural-circulation
coefficients ranging between 200—800 Btu/ evaporators and can be increased still more by
2o 2
hrft F (1100-4500 W/m K) can be expected. further lengthening the exchanger tubes or by coil-
Thus, among natural-circulation evaporators, the ing them.
long-tube vertical evaporator competes most favor- Heat-sensitive materials are often evaporated
ably with the forced-circulation evaporator for in a single pass in a falling film evaporator (21). In
application in large modern installations. The this unit the fluid is fed to the top of long, vertical,
coefficients are lower than they may be for the rather large-diameter tubes (2—4 in.) so that the
forced-circulation evaporator, but no circulating fluid flows down the tube walls in parallel with the

^^B^n
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 499

flow of the evaporated gas. The result is short

contact time with the possibility of large fraction
evaporation. The largest problem with this type of
evaporator is even distribution of the fluid on the
tube walls. This may be done by using carefully
leveled weirs or by spraying the fluid directly on
the tube walls. Heat-transfer coefficients in these
evaporators may be calculated theoretically using
Nusselt-type equations to express the film
Dry vapor
coefficients for both the evaporating liquid film outlet

and the condensate film on the steam side of Separator fins arrest
entrained droplets
the tube wall (5). Evaporating film coefficients
vapor and
rising with
2 2
of 800-1500 Btu/hr ft °F (4500-8500 W/m K) return them to the
heating zone below
have been measured when evaporating water (6).
Small clearance
Modern evaporator developments have causes material
to form turbulent,
centered upon the problem of successfully hand- thin film

ling viscous and corrosive materials. The Steam

inlet
turbulent-film evaporator shown in Figure 19.8
will handle viscous liquids and can even be adapted
to the evaporation of a solution or slurry to dry-
ness. The unit consists of a vertical tube heated
over the bottom two-thirds with a steam jacket,
and containing a central rotor. Mounted on the
rotor are blades extending almost to the heated
walls. The top third of the cylinder is of larger
Steam
diameter and is unheated. In this region, the inlet

vertical rotor blades carry horizontal baffles that

collect entrained droplets and return them to the
evaporator walls. Feed enters at the top of the
Heating wall
heated section. In the evaporator, the feed is
hurled against the heated walls by the action of the
rotor blades. The blades also keep the heating
surface free of solid deposits. The concentrated
solution gradually works its way to the bottom of
the evaporator, being continuously forced against
the heated walls and thoroughly stirred by the
Concentrated
rotor.These units are built in sizes up to 42 in. in product outlet

diameter by 33 ft long with heat-transfer areas up Figure 19.8. Cross-sectional view of turbulent-film evap-
to 198ft 2 Overall coefficients of 40-400 Btu/
.
orator.
2o
hrft F (350-3500 W/m 2 K) are claimed when
evaporating materials of viscosities up to 20,000 materials are virtually thermal insulators, since
centipoises. An additional advantage lies in the low they are not used as heat-transfer surface. Transfer
holdup and consequent short time of residence of of energy from the combustion gases to the process
the fluid in the evaporator. fluid complete enough that substantial savings in
is

Another unit successfully used with both fuel are claimed compared with the use of steam
viscous materials and corrosive fluids if the sub- generated on the site. Units are built for energy
merged combustion burner illustrated in Figure release rates up to 22 x 10 6 Btu/hr (6.5 x 10 6 W) in
19.9. With this burner, no metal heat-transfer sur- applications such as heating dilute H 2 S0 4 , oxidiz-
face is required, since the products of combustion ing asphalt, and evaporating arsenic acid, muriatic
bubble up through the process fluid. The small acid, or clay slurry.
amount of submerged equipment and its simple The selection of the appropriate evaporator is
design reduce replacement costs to a minimum. ultimately a matter of economics, but, the proper-
Moreover, the parts can be made of ceramic or ties of the material being evaporated may sharply

other resistant material, even though these limit the choice. For instance, a highly viscous
500 APPLICATIONS TO EQUIPMENT DESIGN

Main air feed particular problems from a physicochemical view-

Main gas feed point. The temperature of the boiling liquid is
_ Pilot
and air
gas
feed —
i
Sight port
fixed
may
by the pressure under which it exists, and
be readily once the relation
calculated
it

between vapor pressure and temperature is known.

If the liquid exists in depth, the pressure that must

be used in calculating the boiling point at any

depth must include the hydrostatic and accelera-
tion head of liquid above that point. This is very
important in long-tube vertical evaporators, where
liquid depths of 20 ft are common. In such a case,
the liquid enters the heated tube subcooled by flow
through the downcomer and addition of the feed.
As it flows up the tube, it is heated to its boiling
point at some particular depth. Superheating is
probable as the solution flows upward and the
pressure drops. When boiling starts, the superheat
is rapidly lost, and thereafter the liquid tempera-

ture drops with the decreasing pressure. The

acceleration of the mixture imposes additional
pressure beyond the static ones; this pressure may
Figure 19.9. Submerged combustion evap-
be the major one if the flow approaches the
orator. Here the combustion casing acts to
velocity of sound in the two-phase mixture. A
channel and control the combustion gases and
protects the burner parts from corrosion during resulting temperature-height curve is shown in
operation. Figure 19.10.
When a solution is evaporated, the effects of
solution might not move readily past the heating liquid depth and acceleration are present, and the
surface by the action of natural convection alone. effect of concentration upon the boiling point
In this case, use of the forced-circulation or must also be considered. For ideal solutions, the
turbulent-film evaporator might be necessary. In concentration effect may be estimated through
the evaporation of detergent solutions, foaming Raoult's and Dalton's laws for
occurs in certain concentration ranges, whereas
P = Pa+Pb=PaXa+PbXb (19.2)
gelling occurs in others. Thus, the evaporator body
should contain as nearly a uniform concentration where a and b refer to solute and solvent,
as possible.This is achieved by using forced- respectively. the solute is nonvolatile, as is norm-
If

circulation evaporators where the circulation rate is ally the case for evaporation operations, then
large compared to the evaporation rate. In the
P = Pb xb (19.3)
concentration of fruit juices, the food value and
taste depreciate rapidly upon exposure to heat. With P° defined as the total pressure of the vapor
Therefore, no recirculation can be tolerated over a single-component liquid phase, then for the
because some of the solution would remain in the
evaporator beyond the mean holding time. Here,
the falling film evaporator is the normal choice.
In evaporating spent pickle liquor, the cor-
rosive nature of the solution prohibits the use of a
solid heating surface. In this case, submerged com-
bustion is used. Because of the better effectiveness
of heating surface and the resulting reduction in
evaporator size, most evaporators installed now are
of the long-tube vertical or forced-circulation
types.
Depth below tube outlet *•

Prediction of Solution Boiling Temperature

Figure 19.10. Temperature traverse in long,
The vaporization of a pure liquid presents no vertical evaporator tubes.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 501

pure solvent P° = Pb since x b = 1, and the vapor known as a of concentration at one

function
pressure lowering due to the presence of solute pressure. For even
nonideal solution like the
a
becomes NaOH— 2 system, it would be adequate to know
P° -P 1 - xb the boiling-point-concentration data at two widely
= xA (19.4) separated pressures, because the Duhring lines are
P° 1
straight except for the most extremely nonideal
Thus, for solutions obeying Raoult's law, the frac- systems.
tional decrease in vapor pressure is equal to the Note that the equilibrium vapor rising from
concentration of nonvolatile solute. For these solu-
any solution exhibiting a BPR will exist at a
tions, then, the vapor-pressure— temperature curves
temperature and pressure such that it is super-
for fixed concentrations will be parallel. If the heated with respect to the pure condensate. The
curves are assumed to be straight parallel lines in vapor rises at the solution boiling temperature, but
the region of the boiling point, the vapor-pressure the vapor is free from solute, not concentrated as is
lowering (P° — P) will be proportional to the boil- the solution, and therefore it condenses only after
ing-point rise (BPR), or the BPR has been removed. It is superheated by
the extent of the BPR.
TB - TB ° = kx* (19.5)

Calculations on a Single Evaporator Stage

where k constant for a given solvent in ideal
is a
solutions, and (TB — TB °) is the difference From the considerations presented above, the cal-

between the boiling point of the solution and that
of the pure solvent at the same total pressure. This
simple relation holds only under stringent restric-
tions. The range of boiling point must be narrow,
and the solution must obey Raoult's law. This
restriction implies that the solution be dilute and
culations necessary to fix the heating-area require-
ment of a single evaporator from a knowledge of
its duty are straightforward. The heat-transfer-rate
equation need only be solved simultaneously with
the overall heat balance and the total and solute
mass balances. As mentioned above, the overall

that the solute be such that ionization, complex heat-transfer coefficient is usually obtained from
formation, etc., do not occur. Obviously, when previous experience.
solutions of 20—50 percent solute are to be
obtained by evaporation, only qualitative agree- Example 19.1. Determine the heating area
ment can be expected. However, one useful genera- required for the production of 10,000 Ib/hr of 50
lization results. For most solutions, the BPR (TB — percent NaOH solution from a 10 percent NaOH
TB °) is a function of the solution constituents and feed entering at 100° F. Evaporation is to be
the concentration. In other words, the BPR is carried out in a single-body standard evaporator for
relatively insensitive to pressure. Hence, for 2 °
a solu- which an overall coefficient of 500 Btu/hr
F is ft
tion of fixedconcentration, the BPR will not expected. Steam is available saturated at 50 psig,
change significantly over wide pressure ranges. This and the evaporator can be operated at 10-psi
generalization was first noted by Duhring in 1878. vacuum relative to a barometric pressure of
It permits the boiling-point-concentration network 14.7 psia.
for a system to be simply represented and based
upon a minimum of experimental data. The Solution. Figure 19.12 is a diagram of this
Duhring chart is a plot of TB
against TB ° at the evaporator with the given information included.
same total pressure for solutions of constant com- The solution of this problem is based upon several
position. On such a plot, the zero-concentration assumptions typical of evaporator calculations:
line obviously a 45-degree line passing through
is (a) The solution in the evaporator body is

the origin. Lines for other concentrations will be
found to be roughly straight and parallel to this
line but displaced above it. Such a plot for the
NaOH— 2 system is shown in Figure 19.11. This
system is highly nonideal, as indicated by its large
heat of solution. Nevertheless, the Duhring lines
are reasonably straight and parallel up to 60 mass
percent NaOH. Thus, for most systems, the
Duhring lines can be plotted if the boiling point is
violently mixed and therefore homogeneous in
composition. This assumption is realistic except in
the case of long-tube vertical evaporators and
special types such as the turbulent-film evaporator.
Since 50 percent NaOH solution is being removed
from the evaporator, the solution within the
evaporator must be uniformly 50 percent NaOH.
(b) The boiling point of the solution is that
of 50 percent NaOH at 4.7 psia. This statement
502 APPLICATIONS TO EQUIPMENT DESIGN

1000

20 40 60 53 100 120 140 160 180 200 220 240 260 280 300
Boiling point of water, °F

Figure 19.11. Diihring lines for the NaOH— 2 system.

ignores the effects of hydraulic head. Here, this vapor at the temperature of the boiling solution
simplification maybe allowable because the liquid and the pressure given.
depth will not be great, because acceleration is (d) The condensate, or drips, leaving at D is a
small, and because liquid enters the tubes near its saturated liquid at the steam pressure. Subcooling
boiling point in the body. In the case of long-tube will occur, but its contribution to the heat flux is
vertical evaporators, hydraulic and acceleration minute.
heads should be considered, at least by using an (e) There is no heat loss from the evaporator
average depth. body to the surroundings. This approximation is
(c) The vapor leaving at lA, is pure water closer to fact the larger the unit, since the
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 503

Vapor, V, Ib/hr
The enthalpy quantities must be determined rela-
tive to a constant-base condition. If steam-table
data are to be used, all enthalpies must be fixed
10 psi vacuum relative to H = for water at its vapor pressure at
or 4.7 psia
32° F. Any convenient base may be used for the
enthalpy of the solute such as crystalline solid at
Steam. V Ib/hr, 32° F or infinitely dilute solution at 60° F. Here,
A = ?
saturated. 50 psig there are four equations in the four unknowns, F,
U = 500 Btu/ Product, L! Ib/hr
Feed,
hr ft
2
"F
V-\, VQ
and A, so that the solution of the problem
,

10,000 Ib/hr
F, Ib/hr
0.50
is direct once the necessary data have been
7> = 100* F supplied. The enthalpy of vapor (H^ and latent )
x F = 0.10
< r Condensate or heat [\ V q) can be obtained from steam tables (6).
drips, Z) lb/ hr
Liquid-phase enthalpies (h F and hy are obtained )

Figure 19.12. Conditions for Example 19.1.

from an enthalpy-concentration plot such as that
of Figure 19.13. This plot is for 1 atm pressure,
shell-surface-to-heat-transfer-surface ratio decreases but because the effect of pressure upon liquid
markedly the unit size increases.
as enthalpy is small, the pressure restriction need not
On assumptions, the area can be
these be of concern. The boiling point of the solution is
obtained through the solution of the rate equation, most conveniently obtained from a Duhring chart
coupled with heat and material balances. for the NaOH-H 2 system (Figure 19.11). At
4.7 psia, water boils at 160°F (8). From Figure
Overall material balance: F+V =V<\+L-\+D 19.11, the boiling point of a 50 percent NaOH
(Note:D=V ) (a) solution at a pressure such that water boils at
160°F is 233°F, corresponding to a BPR of 73°F.
NaOH balance: Fx F = '
L^' (b) Solving the material balances, Equation (b)
gives
Overall heat balance:
F(0.1) = 10,000(0.5); F= 50,000 Ib/hr
V \ v0 + Fh F = ViH^+L,h^ (c)
From Equation (a), V^ = 40,000 Ib/hr.
Equation (c) may be now solved for V .
Heat transfer-rate equation:
V x 920.0 + 50,000 x 60
V X V0 = UA(-AT) (d)
= 40,000 x 1,160 + 10,000 x 245
where
45,850,000
= quantities of streams V = = 49,800 Ib/hr
F, Vo, V-[,L-[,D enter-
920
ing or leaving the evapor-
ator as noted on Figure The enthalpy of steam leaving the evaporator
19.12. All these quantities {Hi has here been read from the steam tables at a
)

on the same basis;

are given pressure of 4.7 psia and a temperature of 233° F.
one hour of operation
here, The superheat results from the BPR. Finally, from
H = enthalpy per unit mass of a Equation (d),

vapor stream designated by

subscript, Btu/lb 45,850,000 = 500,4 x (297 - 233) = 64 x 500/4
h = enthalpy per unit mass of a A = 1400 ft
2
liquid stream designated by
subscript, Btu/lb In this any single-effect evapor-
example, as in

X= latent heat of evaporation ation, each pound of vapor

is produced at the cost

per unit mass of stream of condensing approximately 1 lb of steam. Here,

designated by subscript, more than 1 lb of steam was required primarily
Btu/lb because the feed entered much colder than the
x' = mass fraction of solute in solution boiling point. If the feed entered super-
stream designated by sub- heated, it is possible that 1 lb of steam would
script, lb solute/lb of total produce more than 1 lb of vapor. The fact that the
stream latent heat of evaporation of water decreases as the
504 APPLICATIONS TO EQUIPMENT DESIGN

40 50
NaOH, percent by weight

Figure 19.13. Enthalpy-concentration diagram for aqueous solutions of NaOH

under a total pressure ofone atmosphere. The reference state for water is taken as
liquid water at 32° F under its own vapor pressure. This reference state is identical
with the one used in most steam tables (8). For sodium hydroxide, the reference state
is that of an infinitely dilute solution at 68° F. [From McCabe, W. L., Trans.
A.I.Ch.E.,31,p. 129 (1935), by permission of A. I.Ch.E., copyright ©1935.]
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 505

pressure increases tends to make the ratio of vapor
produced per pound of steam condensed less than
unity.
Multiple-Effect Evaporation
In any evaporation operation, the major process
cost is in the steam consumed, therefore, methods
of reducing steam consumption (or of increasing
economy, defined as mass of vapor produced per
unit mass of steam consumed) are very attractive.
The most common of the available methods is to
use the vapor generated in the first evaporator as
the heating fluid for a second evaporator. Ideally,
this method should produce almost 2 lb of vapor
for every pound of steam consumed. The method
is feasible if the second evaporator is operated at a
lower pressure than the first, so that a positive
value of —AT is obtained across the steam-chest
surface of the second evaporator. Obviously,
sequences that combine these two (i.e., mixed
feed), or they may be fed in parallel, with fresh
feed evaporating to final concentrate in each
effect. multibody operation, the number of
In
stages may
be different for steam and liquor. Three
common feed arrangements for multiple-effect
evaporation are illustrated by schematic flow
sheets in Figure 19.14.
In addition to the economy increase in
multiple-effect evaporation, a capacity variation
would be expected. Note, however, that the
temperature difference from initial steam to the
final condenser, that was available for a single-
effect evaporator will be unchanged by inserting
any additional effects between the steam supply
and the condenser. For the simplest case, where
each effect has area and coefficient equal to that of
every other effect and where there are no boiling-
point rises,

several evaporators can be connected in series in

this way with the intent being to obtain a number q t = dy +q 2 +q 3 +" '
'

of pounds of vapor for each pound of steam con-

sumed equal to the number of evaporator bodies.
The increase in latent heat with decreasing pressure
and additional radiation losses results, however, in
the economy being increasingly lower than the
number of evaporators used as this number is
increased. This method of evaporator operation in Feed c -:~ jet

series is called multiple-effect evaporation, and Drain

each stage is called an effect. (a)

Multiple-effect evaporators may be con-
nected in a variety of ways. In a forward- feed
system, the flows of process fluid and of steam are
parallel. Forward feed has the advantage that no
* To condenser
and vacuum
system
pumps are needed to move the solution from effect

ftw
1st 2nd 3rd
effect effect effect
to effect. It has the disadvantage that all the heat- Steam
Feed
ing of cool feed is done in the first effect, so that
less vapor is generated here for each pound of
steam, resulting in lower economy. It has the
further result that the most concentrated solution (b)

is subjected to the coolest temperatures. Cool

temperatures may be helpful in preventing
decomposition of organics, but the high viscosity To condenser
*" and vacuum
that may be found sharply reduces the coefficient system

in this last effect. In a backward-feed system, the

process solution flows counter to the steam flow.
Pumps are required between effects. The feed solu-
tion is heated as it enters each effect, which usually
results in better economy than that obtained with
forward feed. The viscosity spread is reduced
because the concentrated solution evaporates at
the highest temperature, but organic materials may Figure 19.14. Schematic flow sheets of feed arrangements
tend to char and decompose. For best overall per- for multiple-effect evaporation, (a) Forward feed, (b)

formance, evaporators may be operated with flow Backward feed, (c) Mixed feed.
506 APPLICATIONS TO EQUIPMENT DESIGN

where q t is the total heat-transfer rate in all effects, Table 19.1 TEMPERATURE IN SINGLE-EFFECT AND
and <7i <72» Q2 are tne heat-transfer rates in each of
,
DOUBLE-EFFECT FORWARD-FEED
the individual effects. EVAPORATORS OPERATING UNDER
IDENTICAL OVERALL CONDITIONS
qt =U y Ai (-AT, ) + U2 A 2 (-A7"2
)

+ U3 A 3 (-AT3 ) + •
Single Double
Effect Effect
Since the areas and transfer coefficients are equal,
Steam temperature, 1st effect 297° F-^ -^297°F
qt = U A
A,[~AT - AT2 - AT3
y
)
267° F
Solution temperature, 1st effect 233° F-^
= c/ 1 ^ (-Artota
1 ,) (19.6) Effective — A7", 1st effect 64° F 30° F
Steam temperature, 2nd effect 250° F
This rate of heat transfer is the same as that Solution temperature, 2nd effect L^233°F
obtained with a single effect operating between the Effective —AT, 2nd effect 17°F
same ultimate temperature levels. Thus, multiple- Total effective —AT 64° F 47° F
effect evaporation using n effects increases the
steam economy but decreases the heat flux per
effect by about \ln relative to single-
a factor of
economy is also reduced. Another method of
effect operation under the same terminal condi-
presenting the same information is by a bar graph,
Therefore, no increase in capacity is as is done in Figure 19.15. Again, the effective
tions.
obtained, and, in fact, the additional complexity of
— AT" applicable to the single-effect evaporator is
equipment usually results in increased heat losses apportioned to the two effects in the double-effect
to the surroundings and a reduction in capacity.
evaporator. Moreover, in the double-effect evapor-
ator, the BPR in the first effect subtracts from the
The increased steam economy must then be
balanced against the increased equipment cost. The
available— A7\ This is a general result. The BPR in
result is that evaporation using more than five or each effect subtracts directly from the available
six effects rarely economical.
is
-AT; that is,

When the solution being evaporated has a — A7"total effective = — A7"tota — 2 BPR
|
(19.7)
significant BPR, the capacity obtained is very
much reduced, for the BPR reduces the —AT in Each effect of a multiple-effect evaporator,
each effect. Consider again the situation of except possibly the last one, is in essence a vapor-
Example 19.1. Here the total —AT was izer and a surface condenser, both operating on the
(297-160) or 137°F. Of this, the BPR used same fluid. the first effect of a multiple-effect
If

73° F, leaving an effective — AT" of 64°F. If two evaporator considered, then the content of the
is

effects are used for this evaporation, about equal first effect is the fluid; the steam chest of the first
evaporation would occur in each effect. Then the effect vaporizes a portion of the fluid, and the
product from the first effect, if forward feed is steam chest of the second effect condenses the
used, would be 5,000/30,000 or 16.7 percent vapor produced in the first effect. A general
NaOH. The pressure in the first effect will depend analysis of this sort of single element may be
upon the relative areas and coefficients present. examined as follows.
Assuming that they are such that water would boil
in this effect at 250°F, the boiling point of the
297° F 297° F
16.7 percent NaOH solution would then be about
267°F, giving a 70°F BPR. The resulting tempera-
tures throughout these two evaporator systems are
then as listed in Table 19.1.
The total effective —AT is seen to decrease by 233° F ,

the amount of the BPR in the first effect. The

vapor from the first of the two effects leaves at
267° F but at a pressure such that it condenses at
250° F. This superheat results from the first-effect 160° 160° F

BPR and before the major heat transfer

is lost
occurs in the second effect. The solution tempera- Figure 19.15. Distribution of total tempera-
ture in the final effect is identical in the two cases. ture difference in (a) single- and (b) double-
As a result, the capacity is reduced, and the effect forward-feed evaporation.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 507

the pressure in the effect is a function of Tf .

According to Equation 19.8 the fluid temperature

{Tf depends upon the transfer rates and terminal
)

temperatures only. Therefore, the unit could be

used with any fluid, and the temperature Tf would
be the same as long as the heat-transfer coefficients
did not change. The pressure in the vapor chest
would, however, depend upon the fluid used, since
it is related to T f through the vapor-pressure curve
Fluid. T,
of the fluid.
As stated above, this action controls the
pressure in the body of each effect except the last
one in a multiple-effect evaporator. The last effect
Condensate, Ts is controlled to a predetermined pressure and
Figure 19.16. A generalized unit of a multiple- related boiling point. The conditions in the other
effect evaporator. effects adjust themselves so that the resulting
temperature— drops with existing transfer— coeffi-
Consider a fluid contained in the unit shown cients accomplish heat-transfer rates called for

in Figure 19.16, in which the unit is equipped with by the material and energy balances.
a vaporizer unit and a condenser unit, arranged so
With these concepts, the calculation of size of
that the vapor that is condensed returns to the pool multiple-effect evaporators is a simple extension of
of boiling liquid. The surface areas and overall the methods used for single-effect evaporators.
heat-transfer of the system may be
coefficients Material and heat balances and rate equations are

stated as A v and U v for the vaporizer and A c and written for each effect. They are then solved simul-
Uc for the condenser. The heating medium to the taneously to get the required information. Thus,
vaporizer enters at Ts and the coolant to the for an n-effect evaporation, we have 3n indepen-
,

condenser is maintained as an average at Tc The .

temperature of the operating fluid may be taken as
Tf If the operating fluid is a single pure com-
.
dent equations and can solve for 3n unknowns.
Unknowns might include the feed and steam quan-
tities, the composition and temperature in the first

ponent, Tf will be the same in the condenser as it is effect, and the heat-transfer area in each effect.

in the boiler. If this fluid is a solution with signific- The equations can be written and a solution
ant BPR, the fluid temperature in the boiler will obtained directly if the evaporating liquor has no
exceed the fluid temperature in the condenser by BPR and if the latent heat of steam can be con-
the value of the BPR. Here, for simplicity, the sidered constant. If a significant BPR is found or if

operating fluid will be taken as a pure liquid. more than two effects are used, a successive-
The rate of transfer of heat from the approximation solution is involved.

vaporizer can be written as

qv = -U v A v {Tf - Tv )
Example 19.2 An aqueous solution contain-
ing 2 percent dissolved organic solids is to be
and the rate at which heat is transferred to the
concentrated to 25 percent solids by double-
condenser is
effect, backward-feed evaporation in forced circul-
2
ation evaporators of 200 m heating area each. The
Qc=-UC A C (TC -T f ).
coefficient in the first effect is 2800 W/m 2 K and
For the simplified system in which there are no that in the second is 400 W/m 2 K. What production
heat losses through the walls of the vessel, q v = q c ;
rate will be obtained? The solution exhibits no
therefore, boiling-point rise; feed enters at 30° C; steam is
TC UC A C + TVV UV A V available at 0.7 x 10
6
newtons/m 2 and the con-
7> =
-JLJL (19.8)
,

UC A C + UV A V denser operates at 7 x 10 3 newtons/m 2 .

In the derivation of Equation 19.8, no refer- Thermodynamic tables for steam

Solution.
ence was made to the nature of the fluid used in (9) give steam condensing temperature as
the
the evaporator element, nor is any specification 164 .9°C and the condenser temperature as 38.9°C.
written for the pressure. It should be evident that Taking 1000 kg/hr of product flow as the basis, an
508 APPLICATIONS TO EQUIPMENT DESIGN

overall material balance gives Heat-transfer-rate equation, effect 1

1000 x 0.25 = Fx 0.02, so F = 12,500 kg/hr of feed Wi/o=Qi = ^1 >M 164.9 -T y

(f)

and Heat-transfer-rate equation, effect 2:

V, + V2 = 12,500- 1000= 11,500 kg/hr vaporized. V,[H,-(T,-0)C ytq ]

= V,\ v:

With these results included, the schematic flow

V:\vi =q2 = U2 A 2 {T, -38.9) (g)

diagram is given by Figure 19.17. The procedure Here, there are seven independent equations involv-
will be to determine T, so that Ay = A 2 Then the .
ing the variables F, L 2 Lu V V u V2 x 2 \ vo , , , , ,

basis chosen [1000 kg/hr of product (Ly)] will be h u h 2l h F H u H 2 T U A U A 2 U u and U 2 Of

, , , .

multiplied by the ratio 200/Ay to get the actual the eighteen, the variables F, L 2 \ VOr h 2 ,h F H 2 , , ,

production rate. Because no exact information is Uy, and U 2 are already fixed by the problem
,

available on the thermal properties of the solution, statement. Three other equations are needed. These
and because the solution exhibits no BPR, the are A-\ 2 =A
which has already been written.
,

specific heat of the solution will be taken as that of Hi = 0(7^ and h^ =

water, and all enthalpies will be calculated from
the basis of liquid water at 0°C and the triple-point
pressure. As a result, all thermodynamic properties
can be read directly from the thermodynamic tables
for steam (9).
The following equations can be
Material balance around effect 1 equal. Assuming q^ = q 2 gives
<p'{T, ). The last two of the
),
equations express the fact that the steam tables
give H-\ and h-\ immediately, once 7^ is fixed.
These steam-table relations are not explicit equa-
tions, so the final solution will involve a trial-and-
error method.
A
good estimate of the final condition is
written: obtained by realizing that q^ and q 2 will be almost

L 2 = Li + V, (a) U,A,(-ATi) = U 2 A 2 [-AT2 )

or
Solids balance around effect 1 \

-A7"! U2 A 2 U-2 = -—-=

4000
L 2 x 2 = 1000x0.25
'
(b) 1-43
-AT2 U^A^ c/t 2800
Material balance around effect 2:
-
(-A7 !) + (-AT2 = 2(-A7) = 164.9-38.9 = 126.0
)

F=L 2 + V2 (c)
Here there is no BPR, so the total —AT is also the
Heat balance, effect 1
— A7" that is effective for heat transfer. Then
V \ vo + L 2 h 2 = L,h, + V,H, (d) (-AT2 + 1.43 (-A7"2 = 126.0, -A7"2 = 51.85°C,
) )

-A7", = 74.15°C, which gives 7", = 90.75°C. From

Heat balance, effect 2:
the steam tables H } = 2661.2 kJ/kg and A^ =
V^H^-h D2 )+Fh F = V2 H 2 + L 2 h2 (e) 2281.3 kJ/kg.

Vi kg/hr V2 kg/hr to condenser

38.9 °C

Effect 1 Effect 2
7\ 38.9 °C

V Ui = 2800 U2 = 4000
Ib/hr
A = A2
t A 2 = Ai Feed, 30 °C
164.9 °C
h <
100° F

Product F= 12,500 kg/hr

L 1 = 1000 kg/ ir
2 percent solids
25 percent solid \
D x
kg 'hr
y
\ •
D2 kg /hr

Figure 19.17. Schematic flow sheet for the backward-feed, double-effect evap-
orator specified in Example 19.2.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 509

From Equations (d) and (e), coupled with the The initial trial is guided by the physical restriction
overall material balance, that each kilogram of condensing steam evaporates
about 1 kg of vapor. Adjustment of the first trial
ViXin + Fh F = (11,500- Vi)H 2 + L^h^
comes from information gained in that trial. For
+ ViH,-V \vo (h) example, in the first trial where q^ = q 2 the Vy as ,

has been assumed that gr-, = q 2 and calculated from the heat and material balances was
It ,

1/1.11 times the V-\ found through q 2 from Equa-

Vo^vo = V1X1/1 tion Therefore, q 2 was decreased by a factor of

(g).

becomes 1.1 for the second trial. The factor 1.1 was used
and Equation (h)
rather than 1.11 because a change in q 2 will result
Fh F = (11,500- V,)H 2 +Z.1/71 + 1/1^1 - 2\/ l X l/1 in an even greater change in the relative values of

This equation can be solved for V^ to give required transfer areas.

The trial-and-error solution was necessary
V A (2\ V i +H2 -H y
)= 11,500H 2 + L :
h, - Fh F here because tabular thermodynamic data had to
(i)
be used for steam. If the system had a significant
BPR, a graphical equation, the Duhring line chart
V y
(4562.6 + 2571.9 -2661.2) = 11,500x2571.9 or its would also have been involved. If
equivalent,
+ 1000 x 90.75 x 4.87 - 12,500 x 30 x 4.187 the thermodynamic data had been expressed ana-
lytically in terms of specific and latent heats, the
from which V, = 6345.8 kg, V2 = 5154.2 kg, and
example could have been solved analytically.
L 2 = 7345.8. From Equation (b), x 2 = 0.0340,
The existence of a BPR would not have com-
and from Equation (d),
plicated the solution greatly. At the intermediate

V \ vo =Q^ = 16.071 GJ compositions, the BPR is small and almost

independent of temperature. The composition of
The validity of the first assumption may now be
the intermediate stream could be estimated from
checked by noting that, from Equation (g), q 2 =
the equal evaporation principle, and, from this, the
= V}\ V y. From this, V^ = 7045 kg, which is
q-i BPR obtained. During the successive approxima-
1.11 times the V-\ calculated from Equations (d)
tions, changes in intermediate temperature or com-
and (e) through the use of the assumed tempera-
position usually do not require changes in BPR.
ture distribution.
The addition of a third or fourth effect results
next assumption, q^ might be chosen
As a
in additional heat and material balances and an
equal to 1.1 q 2 Then, by the method used above,
.

additional heat-transfer-rate equation for each

-A7"2 =49.0°C and -A7", = 77.0°C. From this, effect. The calculations are more tedious but
7"!= 87.9° C, H = 2565.6 kJ/kg, and \y t =
:
involve the same trial-and-error solution of simul-
2288.6 kJ/kg. Solving Equation (i) with this new taneous equations. The steps of such a solution are
value of T, gives Vy = 6315.9 kg. From Equations
as follows:
(a) and (c), V 2 = 5184.1 kg and L 2 = 7315.9 kg.
Use of Equation (d) to check these results gives 1. From the given product concentration and
V \ v0 = <7i = 15,712,000 kJ/kg; then, from Equa- condenser pressure, determine the temperature,
tion (g), V] =6241.2. This check is satisfactory. BPR, and enthalpies in the last effect. They may be
The area may then be found from both Equations directly determined if forward feed is used.
(f) and (g) and averaged. The result gives From an overall material balance, deter-
2.

15,712,000 mine the total amount of evaporation and appor-

1
Avg = A,=A 2 = tion it among the effects by assuming equal evapor-
3600
2800 x——
1000
(164.9-87.9) ation in each effect. If the feed is very cold or is
very superheated, the portions may be modified
appropriately. The assumed vapor quantities will
= 20.24 m2
give an estimate of the concentration in each effect
for producing 1000 kg/hr of product. and consequently of the BPR. Only the crudest
The production rate is then 200/20.24 x 1000 guess as to the pressure in each effect need be

or 9880 kg/hr of product having 25 percent solids. made, for BPR is virtually independent of pressure.
This example illustrates typical evaporator 3. Find the — A7" available for heat transfer
calculations. The method is essentially a trial-and- by subtracting the sum of all BPR's from the total
error solution for the intermediate temperature. — A7". The available — A7" can be apportioned
510 APPLICATIONS TO EQUIPMENT DESIGN

among the various effects by assuming that q-i = ator bodies will come to steady-state temperatures
so that (-A 7", 1x^4, = (-AT2 x U 2 A 2
= and pressures that are determined as much by the
q 2 =03 )

(-AT3 ) x U3 A 3 = : physical properties of the evaporating solution and

the physical arrangement of equipment as they are
4. Calculate the amount
of evaporation in
by any control exercised by the operator.
each effect through energy and material balances. The procedure for determining evaporator
If the amounts differ significantly from the values
operating conditions consists of writing heat and
assumed in step 2, steps 2 and 4 must be repeated material balances around each effect and solving
with the amounts of evaporation just calculated.
them in conjunction with the rate equation for
Usually, this recalculation will represent a very
heat transfer in each effect. In the procedure out-
slight revision in the BPR and enthalpy values
lined above, steps 1, 2, and 3 merely fix a reason-
previously used.
able starting point for the solution. Actually, any
5. By means of the rate equations for each
assumed condition would do as a starting point for
effect, calculate the surface required for each
calculation. The advantage of the method outlined
effect.
is only that it shortens the trials necessary by
6. If the surfaces calculated do not fit the
allowing the calculator to start at a point reason-
required area distributions (usually A-\ = A 2 = A 3
ably close to the correct answer.
= •••), revise the temperature-difference distribu-
tions of step 3. Unless BPR's are very large, the
revision will not affect the values assumed in step
Example 19.3. A triple-effect evaporator
system be used to concentrate 5 percent
is to
2.
7. Continue adjusting the temperature differ-
NaOH to 50 percent NaOH. Forward feed is to be
used with the feed entering at 60° F. Overall co-
ences and recalculating surface areas until the areas
efficients of 800, 500, and 300 Btu/hr ft 2 °F are
are distributed satisfactorily.
expected for the three effects in the given order.
Steam is available at 125 psia, and ejectors capable
The mechanism involved in solving evaporator
of maintaining a pressure of 1 psia will be used.
problems should not camouflage the fact that the
The heating areas of each effect are to be equal and
evaporator operates under a balance of natural
large enough to produce 10 tons/hr of concentrate.
physical forces. All the designer does is try to
What heating area is needed for each effect? What
determine how the interplay of these forces affects
will be the steam consumption and the economy?
the evaporator conditions. The operator of an
evaporator can control the steam pressure, the con- Solution. A schematic diagram of the evapor-
denser pressure, and the feed-flow rate. The evapor- ator system is shown in Figure 19.18. By reference

To condenser
101.7' F
Vt Ib/hr V2 Ib/hr V, Ib/hr

Steam
344.3° F
125 psia

L\ = 800 U2 = 500 U3 = 300 Product

T,,
50 percent NaOH
Feed L x
Ib/hr L2 Ib/hr L3 =
60° F
20.000 Ib/hr
5 percent
NaOH
F Ib/hr - D- b In "0, Ib/hr ID, Ib/hr

Figure 19.18. Flow sheet of triple-effect evaporator with forward feed with con-
ditions for Example 19.3.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 511

to Figure 19.1 1, the solution and the vapor in the Heat and material balances can now be solved
third effect are found to be at 171.7°F, as is the around each effect to find a Vy, V 2 and V3 ,

product (L 3 ). Thus, the BPR in the third effect is consistent with these assumed temperatures.
70°F. Enthalpy data for water or steam can be taken
An overall material balance gives from the steam tables (8) and that for solutions
from Figure 19.13. These balances are
0.05F = 0.5 x 20,000
F= 200,000 b/hr I
VoX^o + FhF = Vy H y + [F- Vy )hy (b)

Vy Hy+iF- Vy )hy = V2 H 2 + Vy h D2
and Vy + V2 + V3 = 180,000 b/hr I (a)

Assuming equal evaporation, Vy = V2 = V3 = + {F-Vy-V2 )h2 (c)

60,000 Ib/hr. From this, the solution in the first (F-Vy- V2 )h 2 + V2 H2 =

effect is 10,000/140,000= 7.1 percent NaOH, and V2 h D3 + 180,000
( - Vy - V2 )H 3 + L 3 h3 (d)
the solutionthe second effect is 10,000/
in
80,000= 12.5 percent NaOH. If 28Q°F and 220°F In these three equations, the only unknowns are

are used as rough guesses for the boiling point of

Vy, V 2 and VQ Inserting numerical values from
, .

water in the first and second effect, the solutions the steam tables and from Figure 19.13 gives
boil at 285° F and 230° F, respectively, which gives 875.4 V + 200,000 x 26 = 1 185.1 Vy
BPR 's, 5°Fand 10°F for these effects.
of
-
+ (200,000 Vy )260 (b)
The total effective —AT is then
875.4 V - 925.1 Vy = 200,000 x 234
(-A7%ff =.(344.3- 101.7)
-(70+10 + 5) = 157.6°F 1185.1 Vy + (200,000 - Vy )260 = 11 69.9 V2

The —AT is apportioned among the vari-

available + 279.3 V2
ous assuming that qy = q 2 = q 3 Although
effects, .
+ (200,000- Vy - V2 208
) (c)
this assumption is in conflict with the previous
843.8 Vy - 961.9\Z2 = -200,000 x 52
Vy = V2 - V3 assumption, so that a revision in one
or the other of them will be necessary eventually, (200,000 -Vy - l/2 )208 + 69.9 \Z2 = 220.0 V21 1

it is a reasonable starting point. Calculating —AT + (180,000- Vy - V2 )1 137.4

gives
+ 20,000 x 200 (d)
UyAy [-ATy ) = U2 A 2 (-AT2 )
= U3 A 3 (-AT3 )

1879^2+929.4^ = 167,200,000
800 _ -AT2 500 _ -AT3
" ~ Solving simultaneously gives Vy = 57,000 lb, V 2 =
500 -ATy '
300 -AT2
60,800 lb, V3 = 62,200 lb, and VQ = 110,800 lb.
I- AT, ) + (-AT2 + (-Ar3 ) )
= 1 57.6 The check of these results with the Vy = V2 = V3
-
from which -A7 !
= 30.0°F, -A7*2 = 47.9°F, assumption, although not good, is adequate for a
-Ar3 = 79.7°F. first approximation. The large value obtained for

Thus, V indicates that the equal-area requirement will

= 171.7 + 79.7 = 251.4; 7" not be met and a temperature adjustment will be
T3s 2 = 261.4
needed. Accepting Vy, V2 and V3 as calculated, ,

309.3°F; 7^ * 31 4.3° F; r 1s = 344.3°F

r2s = and solving the rate equations for the areas results
where the nomenclature is as shown in Figure in

19.18. Q\ l/oX^c-
The guesses as to boiling point of water in the Ay =
Uy (-ATy ) Uy (-ATy )
first two effects are seen to be greatly in error, but a
recheck of Figure 19.11 shows that the BPR's _ 110,800x875.4
= 4040 2
ft
calculated need not be changed. Temperature in 800 x 30
the three evaporator bodies will then be as follows
for the first trial solution of the heat and material Vy (Hy -hy)
balances:
A2 = ,,.._
U (-AT
=2160 ft
2

2 2
Effect Effect 2 Effect 3 Condenser
V2 (H 2 -h 2
1

TU = 344.3° F |-r2s 309.3° F [-7-35 = 251 .4° F prc = 101.7° F A* = = 2410 ft

2

r-| =314.3°F-I T2 261.4°F-J r3 = 171.7°F-l 6M-A7-,)

512 APPLICATIONS TO EQUIPMENT DESIGN

These unequal areas indicate that the original BPR values change. Determining Am in this way,
apportionment of the total available —AT among
30.0x4040 + 47.9x2160 + 79.7x2410
the various effects was improper. In adjusting the *»m
— A7~'s, it is common practice to choose them so 157.6
that 2645 ft
2

and
-AT", '
= -A7"! ill (e)
4040
- Ar ,=30
A2
' -

*^r 45 8F -

-AT?' = -AT (f)

7

-W - 47.9 x
|| = 39.2° F

-AT? '
= -AT, (g)
- Ar3 =79 7x °

i^ =72 6
'

- - F

where —AT/, — A7"2 ', and -A7"3 are the newly '

assumed effective — A7~'s for each of the three 2(-Ar') = 157.6°F

effects, and — — A7"2 — A7"3 are those that
A^ , , Am
had been calculated by averaging A-\
If ,

were originally assumed. A m is the mean area of A2 A3

a value of 2870 ft
, and 2
would have been
,

one effect obtained by proper averaging of A u A 2 ,

obtained, from which — A7" = 42°F, —AT2 = 1
' '

and A 3 Equations (e), (f), and (g) are corrections

. 35.9°F, -A7V = 66.7°F, and S(-AD = 144.6°F,
on the original assumptions that q^ = q 2 = q 3 Considerable adjustment would have been required
which led to the values of Ay, A 2 and A 3 , to make Z (-AT) = 157.6
obtained above. The result
is that each of the areas Using the — AT
values calculated by Equation
will approach Am The equal-evaporation .
(j) gives the significant temperatures as

assumption has been discarded; it has served its Effect 1 Effect 2 Effect 3 Condenser
purpose.
Tu = 344.3° F iTk =293.5° F ^T^ = 244.3° F --rc = 101.7°F
The simplest method of calculating A m is 7
_

1
=298.5 / 72 = 254.3 / r3 = 171.7 '
merely to average A^ A 2 and A 3 as was done in , , ,

Indication of these temperatures to fractions of a

Example 19.2. This method, however, will give
degree should not mislead one as to the precision
—ATs that will not add up to the total effective
with which actual operating conditions are pre-
— A7". One common practice is to calculate Am by
dicted
this averaging technique and then arbitrarily adjust
Equations (b), (c), and (d) can again be
the -AFs so that (-AT"/) + (-A7"2 ') +
written using enthalpy values obtained at the new
(-A7y)=-A7-eff .

temperatures.
An alternate, and more satisfactory, method
is to apply the restriction of a constant sum 875.4 V + 200,000 x 26 = 1 180.3 ^
(-A7-, ') + (-A7V) + (-A7V) = 2(-A7-eff )
+ (200,000-^)245 (b)

= (-Ar + (-Ar2 + (-Ar3 1 ) ) ) (h) 1180.3^ +(200,000- IM245= 1167V/2

+ 263 V + (200,000 - l/, - V2 ) 1 98 (c)
in the calculation of A m Combing . Equations (h),
y

(e), (f), and (g) for this purpose, 198(200,000- V - y l/2 ) + 1167V/2 = 212V/2
+ (180,000- Vy - l/2 )1 137.4
(-AT, )
—
A
^m
i
+ (-A7"2 )
—^1 + (-A7-
^m
-
3 )
A3
-f-
r\m
+ 20,000 x 200 (d)

Solving them gives 1^=57,300, V2 = 61,100,

= (-Ar,) + (-Ar2 + (-Ar3 ) ) (i)
V3 =61,600, 111,000. and V =
from which Because these values do not coincide with the
initial assumption of equal evaporation, the result-
(-AT,)A, + (-AT2 )A 2 + (-AT3 )A 3 ing concentrations should be checked to see how
A =
(-Ar ) + (-Ar2 + (-Ar3 ) )
w
(i)
the BPR's have been affected. Such a check gives
1

x 1 = 0.07, x 2 = 0.123, and a recheck through

'
'

The Am value obtained from Equation (j) will give Figure 19.1 1 shows no change in the boiling point
2(— AT') = 2(— A7"eff ) without adjustment unless rises.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 513

Calculating the required areas as before gives is used to entrain and compress a major part of the

A y
= 2640 = 2640 ft 2 and A 3 = 2720 ft 2
ft
2
, A2 . vapors from the evaporator. Some of the vapor is
The values are equal within the accuracy range separately condensed to compensate for the motive
practically required. Moreover, they nearly equal steam added.
the value of A m obtained from the first tempera- A schematic diagram of a double-effect
ture approximation. Final answers to this problem evaporator system using vapor recompression is
would be 2700 ft 2 surface per evaporator body; shown in Figure 19.20. Here the feed is preheated
11 1,000 Ib/hr of steam required, and a steam before entering the first of two rising-falling film
economy of 1.62. evaporators. Some of the vapor from the first
Large-scale evaporation operations are effect is recompressed in a steam jet compressor A
frequently done in as many as six effects and and returned both to the first effect B and to the
require units large enough to represent a small preheater C. The rest of this vapor goes on to the
plant all by themselves. Typical applications would second effect D. Vapors generated in the second
include concentration of black liquor from a effect, plus noncondensables from both evaporator
paper-pulping operation, sugar concentration, units, go to a barometric condenser E for final
caustic soda evaporation, and ammonium sulfate discharge. Details of the evaporator itself are
concentration. Figure 19.19 shows such a large- shown in the insert. Here steam generated from
scale multiple-effect evaporator system. vaporization travels with the liquid, both in a rising
and a falling mode, producing a turbulent liquid
Vapor Recompression film against the heat transfer surfaces, as described
earlier.
Another method of increasing steam economy is
by recompressing the vapors from the evaporator A schematic flow diagram of a vapor-
and feeding them back into the steam chest of an
appropriate evaporator effect. Mechanical compres-
sion or steam-jet injection may be used. In the case
of the steam-jet injection, a high-pressure steam jet

Figure 19.19. A forced-circulation, triple-effect evaporator Figure 19.20. Flow diagram for a Buflovak double-

system concentrating caustic solution in a chlor-alkali plant. effect, thermal-recompression, rising-falling turbulent-
The vertical heat exchangers are shown in front of each film evaporator system. Details of the evaporator
evaporator body with steam lines entering near the top. For itself are also given in the insert. (Courtesy of
scale, note the operators working in front of the closest the Buflovak Division, Blaw-Knox Food &
Chemical
(third) effect. (Courtesy Swenson Division of Whiting Cor- Equipment, Inc., One of the White Consolidated
poration.) Industries.)
514 APPLICATIONS TO EQUIPMENT DESIGN

-Motor the feed is heated. The heat released by the con-

densing vapor {H c - HD must be enough to )

-Compressor evaporate an equal quantity of liquid. {H B — HA )

Vapor heat exchanger and to supply all the heat losses. The available
(evaporator)
—AT in the exchanger is that between the boiling
Flow control valve
solution at low pressure and the condensing vapor
/ • Pressure regulator at higher pressure. With this method, evaporation
/ / ^Strainer
equivalent to 15 or more evaporator stages can be
•p^xtq^j^ Feed in
obtained.
rFlow meter In this country, these systems are used only in
r=s: Distillate out special applications because of several mechanical
Distillate pump and economical disadvantages.

1. The compressor is expensive and is subject

to higher maintenance costs than is the remainder

of the evaporator system.

exchanger 2. In reasonable operating ranges, the — AT
obtained is small — about 10°F. As a result, the
evaporator equipment must be large for any
Figure 19.21. Schematic diagram of vapor-recompression
reasonable production rate. Higher —A Ts could be
evaporator as used for pure water generation from brine.
obtained with a higher compression ratio, but the
efficiency would decrease proportionately.
recompression evaporator in which mechanical 3. A recompression evaporator cannot be
compression of the vapor is used is shown in Figure started up without an auxiliary steam or heat
19.21. Here the compressor is located on top of supply.
the evaporator body to prevent heat losses in any 4. If the solution has a large BPR, the cost of
external piping. The principle of operation is recompression increases rapidly because the vapor
shown in Figure 19.22, where the cycle is shown must be compressed to such a pressure that its
on a schematic pressure-enthalpy diagram for saturation temperature is above the temperature at
water. Water at A is evaporated in the evaporator which the solution boils.
to become saturated vapor at B. This vapor is
In a situation where
fuel costs are high but
compressed along path B—C, the work done by the
electrical power cheap, recompression evapora-
is
compressor being H c — HB if the compression is
tion is economical and is used. In the United
adiabatic. The compressed vapor at C enters the
States, fuel costs have been low relative to power
steam chest where it condenses by giving enthalpy
costs and will probably continue to be coupled to
to the evaporating liquid. Condensate at condition
them; thus, recompression evaporation is not
D leaves the unit through an exchanger in which
generally attractive. was regularly used to supply
It

potable water at advanced military bases and on

ships at sea. In Europe, fuel is relatively scarce and
expensive, whereas power from hydroelectric
4>/ stations is quite inexpensive in some areas. As a

o
result, there are several large recompression-
a.
3|n
>
evaporation units in operation in European
1o

Di
</)
_ chemical plants.
T^C

Integration of Evaporators into the Total Plant Economy

B
The need for economy in the generation of steam

/
Figure 19.22.
Enthalpy

Schematic pressure-enthalpy dia-

1
dictates careful integration of the evaporator unit
into the overall
economic considerations
pressure
at high pressure
is available in a
plant economy. Such general
will dictate what steam
plant where steam is raised
and is used in an extraction tur-
gram for water showing the operation cycle of a bine for generating electrical energy. Economics
vapor-recompression evaporator. may also fix conditions of feed preheating and
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 515

product withdrawal temperature and may point to service.However, even at high circulation velocities,
advantages in a particular feeding arrangement or scale does deposit and must be removed. In one
to the withdrawal of some steam at one or more evaporator design, scale is removed by periodically
points between effects for ultimate use elsewhere switching the flow of solution and of steam
in the plant. This latter procedure allows the through the two sides of the evaporative
evaporator to act as an efficient method of generat- exchanger. Thus, surfaces on which scale deposits
ing low-pressure steam when higher-pressure steam from the boiling solution are later washed clean by
is available. The advantage of steam withdrawal the steam condensate. The exchanger is made with
from an evaporator rather than mere throttling of square tubes on a checkerboard pattern so that
the high-pressure steam is in the "free" evaporation liquor and steam passages are essentially identical.
obtained by the evaporator. The disadvantage is Many solutions containing organic matter cause
that the steam generated may contain minute deposits that are not true scales. If the operating
amounts of salt or other impurities that promote conditions are held constant, the amount of true
the corrosion of equipment in which it is used. scale buildup should be proportional to the total
amount of heat transferred through the surface
Scaling since the start of the operation. Also, the overall
In addition to the usual problems of process heat-transfer coefficient must decrease as the scale
control and maintenance, the operator of evapor- builds up because an additional heat-transfer resist-
ators is sometimes faced with serious loss of ance is being formed in series with those already
capacity resulting from the deposit of solids from present. On this basis (12),
the evaporating solution onto the evaporator heat-
transfer surfaces. This problem is particularly acute 1 1

when evaporating materials exhibit "inverse

U= (19.9)
R + Rs J_ + aQ
solubility" — that is, where the solubility decreases
as the temperature increases. When such a solution
Uo
passes through the tubes of an evaporator, the where
material nearest the tube walls is heated to a higher
temperature than the material in the center of the U= overall coefficient of heat transfer at
tube. If the solution has a bulk temperature close time 8 after the start of evaporation
to its saturation temperature, the fluid close to the Uo = overall coefficient of heat transfer at the
wall may become hot enough so that the satura- evaporation
start of

tion concentration becomes lower than the existing /?o

= to heat transfer existing at
resistance
concentration. The precipitated solid will cling to the start of evaporation
the tube wall, increase in temperature still further, Rs = resistance to heat transfer resulting
and promote further precipitation. Solutions of scale formation

sodium sulfate (see Figures 19.27 and 19.28) and Q = total amount of heat transferred from
calcium sulfate are among those that exhibit time to time 6
inverse solubility and behave in this way. The scale a = a constant
that forms adheres tightly to the tube surface and
The rate of heat transfer at any time (9) is given by
must eventually be removed. Methods for removal
the basic rate equation as
include opening the evaporator and drilling out the
tubes with a special cleaning tool and boiling out
the evaporator with a dilute acid solution. Scaling
dQ
:i4.3i:
can sometimes be greatly reduced by maintaining a dd
suspension of the solid phase in the liquor, thereby
The relation between the overall coefficient (U)
providing extensive surface for crystallization
and the time (6) may be obtained by eliminating Q
within the slurry itself.
between Equations 19.8 and 14.31. Solving Equa-
Scaling solutions can be best handled where
tion 19.9 for Q and differentiating gives
the velocity through the tubes is high. As well as
physically obstructing the deposit of scale in the 1 1
aQ =
tube, the resulting turbulence reduces the tempera- U~ U
ture variation across the tube diameter. Because
they maintain a high circulation rate, forced- -dU
adQ = [19.9a)
circulation evaporators are most suitable for this u-
516 APPLICATIONS TO EQUIPMENT DESIGN

This relation may be substituted into Equation polyhedron shape. This is called an invariant
14.31 to give the desired result in differential form crystal.For such a crystal a single dimension can
be used to characterize the volume, total surface
-dU area, or such factors as the surface area per unit
= A(-AV>add (19.10)
U3 volume. More often, even in an unhindered
environment, uneven surface potentials will force a
Integrating between the limits of 8 = and 8 = 8
crystal to grow more rapidly in one dimension than
gives
others, producing elongation and distortions.
rUdU_ re In commercial crystallization processes, un-
A(-AT)add hindered crystal growth seldom occurs. Crystals
agglomerate, impurities are occluded on the growth
1
—z-
2
—~=2A(-AT)ad
2
1
19.11)
surfaces, nucleation occurs both in solution and at
and crystals are broken by pumps
U U crystal surfaces,
and agitation. The net result of all these factors is
Equation 19.1 shows that a graph of
1
2
plotted MU spoken of as the habit of crystal growth. This habit
is of great concern to crystallizer operators because
against the evaporation of a true scaling
8 for
it affects the product purity, appearance, tendency
solution should give a straight line. If this is true,
the entire path of U as a function of time can be to cake or to powder, and hence customer accept-
found from values of the overall coefficient at two ance.
points in time after the start of operation. This The crystallization habit is strongly affected
information, coupled with knowledge of the shut- by the degree of supersaturation, the agitation
down time required to clean the tubes, permits the intensity, the population density and size of
engineer to determine the optimum time of opera- crystals in the environment, and the purity of the
tion between shutdowns. solution. Thus, the choice and detailed design of
the crystallizer itself are important not only for its

economy and operability, but also for its influence

CRYSTALLIZATION
upon the crystal habit, crystal size distribution,
In many cases, the salable product from a plant and ultimately produce salability.
must be in the form of solid particles. Where the
process yields a solution, the solid may be most Crystallization Equipment
conveniently formed by concentrating the solution Crystallizers can be conveniently classified in terms
to saturation and forming crystals from the solu- of the method used to obtain deposition of
tion. Throughout the history of the modern particles. The groups are:
chemical industry, crystals have been produced by
methods ranging between those as simple as setting 1. Crystallizers that obtain precipitation by
pans of hot concentrated solution out to cool and cooling a concentrated, hot solution.
those as complex as continuous, carefully con- 2. Crystallizers that obtain precipitation by
trolled, many-step processes tailored to give a
evaporating a solution.
product of uniform particle size, shape, moisture 3. Crystallizers that obtain precipitation by
content, and purity. Customer demands as to adiabatic evaporation and cooling.

product quality have gradually forced the dis- In the first group are found the pan coolers
continuance of the simpler crude crystallizers, mentioned above, agitated batch crystallizers, and
because a modern crystal product must usually the Swenson-Walker continuous crystallizer. In the
meet very rigid specifications governing the proper- second group are those evaporators in which
ties listed above plus such others as color, odor, crystallization as well as evaporation takes place,
and caking characteristics. called salting evaporators, draft-tube crystallizers,
A crystal is a highly organized array of atoms, and Oslo group are the
crystallizers. In the third
molecules, or ions arranged in three-dimensional vacuum Cooling crystallizers operate
crystallizers.
space lattices. The lattice is regular, with fixed best on systems with a steep solubility-temperature
distances and angles between particles, and gives a curve so that modest cooling produces a large drop
characteristic x-ray diffraction pattern from which in solubility and hence a large yield of crystals. For
the dimensions of the array may be determined. intermediate solubility slopes, the adiabatic crystal-
When a crystal grows without hindrance from lizer is indicated; typically, about 5—10 percent of
other crystals or solids, it may maintain a fixed the solvent is vaporized, producing greater yield
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION
than would be obtained by cooling alone. For
systems with very little change in solubility with
temperature, crystallization must be obtained
largely by evaporating the solvent, as in an
evaporator-cry stal izer.
I
Of the crystallizers that operate by cooling,
the simplest are various forms of batch crystal-
lizers. Pan crystallizers are pans in which a hot
solution isallowed to cool and crystallize. They are
seldom used in modern practice, except in small-
scale operations, because they are wasteful of
floor space and of labor and usually give a low-
quality product. Agitated batch crystallizers are a
simple improvement. They consist of an agitated
tank, usually cone bottomed, containing cooling
coils. In small-scale or batch processing, such
crystallizers are quite convenient because of their
low first cost, simplicity of operation, and
flexibility. They are too wasteful of labor and give
too uneven a product to be attractive for large-
scale continuous processing.
The Swenson-Wa/ker crystallizer is a cooling
crystallizer designed to operate continuously. It
consists of an open, round-bottomed trough 24 in.
wide by 10 ft long, jacketed for cooling water, and
containing a long ribbon mixer that turns at about
7 rpm. As many as four of these units may be
connected together with the agitators driven from
a single shaft. If longer lengths are required, a
second set of crystallizers may be set up at a
slightly lower level and fed by overflow from the
end of the first crystallizer set. Figure 19.23 shows
a crystallizer consisting of two of these connected
units. The hot, concentrated solution is fed con-
tinuously at one end of the crystallizer and flows
slowly toward the other end while it is being
cooled. The function of the agitator is to scrape
the crystals from the cold walls of the unit and to
pick up the crystals and cascade them down
through the solution so that precipitation occurs
mainly by buildup on previously formed crystals
instead of by the formation of new crystals.
Of the evaporating crystallizers, the salting
evaporator is the most common. In older forms,
this crystallizer consisted of an evaporator below
which were chambers into which the salt
settling
settled. The chambers were connected in tandem
so that salt could be removed from one of them
while the other was connected to the system.
Modern practice no longer has these settling
chambers directly connecting to the body. They
are most commonly replaced by a classifying
settler, through which a stream of the liquor from
the evaporator body is pumped continuously. The
517
Figure 19.23. Swenson-Walker crystallizer. (Courtesy
Swenson Division of Whiting Corporation.)
dimension of the settler and the circulation rate are
so related that only the coarser crystals settle out.
The fine ones remain in suspension and are returned
to the evaporator body for further growth. The
thick slurry from the bottom of the settler is
pumped continuously to filters or centrifuges or
other appropriate processing equipment. Mother
liquor may be returned to the evaporator or dis-
carded in whole or in part to eliminate impurities
from the system.
The Oslo crystallizer, one form of which is
the "Krystal" crystallizer shown in Figure 19.24, is
a modern form of evaporation crystallizer. This
unit is particularly well adapted to the production
of large-sized, uniform crystals that are usually
somewhat rounded. It consists essentially of a
forced-circulation evaporator with an external
heater containing a combination salt filter and
particle-size classifier on the bottom of the evapor-
ator body. If desired, the external heater may be
used as a cooler, in which case crystallization
occurs by solution cooling. Flow rises in the ex-
ternal heater shown in Figure 19.24. The flow tube
that pierces the crystallizer body continues to the
bottom of the crystal collector and classifier.
Hence, flow is upward in this classifier, allowing
simultaneous contact between the crystals and the
slightly supersaturated solution and classification
of the crystals. This is the unique feature of this
crystallizer. The largest particles are thus the ones
that progress to the bottom of the classifier and are
withdrawn in the product magma. Fine crystals
51 APPLICATIONS TO EQUIPMENT DESIGN

r - "* at the expense of sensible heat in the feed. As a

^r ^ 7/ *>^
result, the temperature of the vapor-liquor mixture
after flashing is much lower than that of the liquor
before flashing. In Figure 19.25, two designs of a
vacuum crystallizer are shown. These units may be
operated continuously as one or more stages or
batchwise. In batch operation, the hot feed is
pumped into the vessel and agitation started. Then
the ejectors are started, and the system pressure
and temperature are gradually lowered. The run is
finished when the vacuum pump develops the
minimum obtainable pressure. During the run,
liquid been vaporized at all pressures from
has
atmospheric to the final minimum pressure. Much
of this vapor has to be compressed only slightly so
that the ejectors can operate at high capacity. Only
toward the end of the run is the compression ratio
high; during this period, a booster ejector may be
The overall result is that the vapors are
necessary.
removed much more economically than would be
possible if the unit operated continuously at the

lowest pressure.
For high-capacity units, usually above 10,000
gal/hr of feed, batch operation is impractical and
the crystallizer must be operated continuously. To
Figure 19.24. "Krystal" crystallizer for production avoid the poor energy economy mentioned above,
ammonium sulfate. (Courtesy Struthers-Wells Corp.) the continuous units may be operated in stages. As
the number of stages increases, the steam consump-
and saturated solution leave the top of the bed and tion approaches that of batch operation, but ulti-
are recycled. The recycle stream is treated to attain mately the capital costs become excessive. For
supersaturation in one of three ways: 1) it may be continuous operation, the feed is put into the first
subjected to evaporation, 2) it may be cooled, or effect, which operates at only a slightly reduced
3) it may be enriched in solute by addition of a pressure. The point of feed injection is selected so
strong feed solution. The first will be described. that minimum superheating of any significant
The two body and the external
sections of the quantity of solution is obtained. Usually, the
heater are clearly shown in Figure 19.24. The superheating is limited to 5°F to prevent caking
lower section of the body contains the bed of of the unit and formation of large numbers of new,
crystals through which the liquor flows upward to small crystals. The product withdrawn from the
exert its classifying action. Solute corresponding to crystallizer contains both crystals and mother
the supersaturation is largely on the
deposited liquor and is pumped to the second effect. The
existing crystals. The saturated solution and fine second effect operates at a pressure somewhat
crystals then heated in the external heater,
are below that of the first effect. In each effect, some
which is maintained under enough hydrostatic evaporation takes place; the solution cools and
head to suppress boiling on the heating surface. deposits solids, and the pressure is reduced com-
The heated liquid passes upward to the top section pared to the preceding stage. In the last stages, a
of the body. At the higher elevation, some of the booster ejector may be necessary.
solution flashes to vapor, leaving a supersaturated Of the crystallizers discussed, only the Oslo
solution that is led downward through a central crystallizer effects any classification in the crystals
duct to the bottom of the crystal bed. The large formed. Control of crystal size distribution (CSD)
crystals are discharged through an opening in the is vital in order to produce quality products, and

bottom of the lower section of the body. this control has therefore been attacked directly by
In vacuum crystallizers, evaporation is crystallizer designers. Figure 19.26 (18) shows one
obtained by flashing a hot solution into a low- such design, the draft-tube-baffle crystallizer
pressure space. Energy for vaporization is obtained equipped with an elutriation section and an

** <
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 519

4^ ^ Steam
Water
Steam

Steam

(a) (b)
Figure 19.25. Two types of vacuum crystallizers. Both units use a booster ejector
feeding a barometric condenser, (a) Batch draft tube crystallizer. (Courtesy of the
Buflovak Division, Blaw-Knox Food & Chemical Equipment, Inc., One of the White
Consolidated Industries.) {b) Continuous vacuum crystallizer with recirculation of
crystal magma. (Courtesy of Swenson Division of Whiting Corporation.)

internal fines-separation system. Fines flow with particles present there. The elutriation leg assures
the liquid in the external flow trap and are mixed that only the largest particles fall to the bottom of

with the recycle feed and heated. They are redis- the crystallizer and are removed.
solved when the stream reenters the crystallizer. In each of these units, attempts have been

Supersaturation occurs with the pressure drop in made to solve the problems of preventing caking
the crystallizer, and precipitation occurs on the on the walls, of separating vapor from liquid and
520 APPLICATIONS TO EQUIPMENT DESIGN

To barometric

Circulation
pump
Figure 19.26. A draft-tube-baffle vacuum crystal-
lizer. Note internal and crystal
fines separation
5 10 15 20 25 30
elutriation (18). (By permission of Chem. Eng.,
Weight percent Na 2 S04
copyright© 1970.)
Figure 19.27. Solubility diagram for Na2S04
liquid from crystals, of obtaining low operating in H2 at 1 atm total pressure. (Data from
cost and first cost, of conserving floor space, of International Critical Tables, Vol. 4 page 236.

promoting the desired crystal growth of pure By permission of Nat. Acad, of Sc, copyright ©
1928.)
crystals, and of keeping maintenance costs low.
The choice, from among these and other crystal-
pressure of one atmosphere. Over the temperature
lizers, of a unit to do a particular job depends upon
range shown, only three solid species precipitate.
the economics of each individual situation and
At temperatures between 30° F and 90.5° F and at
upon the product limitations imposed by the sales
concentrations above 4.5 percent, the saturated
situation.
solution is in equilibrium with the decahydrate.
Above 90.5° F, the solid phase is anhydrous
Solubility Relations
Na 2 S0 4 Note that the anhydrous salt exhibits an
.

As with other two-phase systems in which transfer inverse solubility between 90.5° F and about
occurs, an equilibrium statement is necessary in 220°F. At 220°F, a minimum solubility of 29.5
order to set the of the driving force for
limit percent is found. Above this temperature, increas-
transfer. In crystallization, mass is transferred from ing the temperature increases the solubility. At
the solution to the crystal surface. The concentra- concentrations below 4.5 percent Na 2 S0 4 the ,

tion necessary for crystals to form and the
chemical species that separates can be determined
from a temperature-composition phase diagram,
which is often called a solubility diagram. Data on
the solubility of solid compounds in solvents as a
function of temperature can be found in most
handbooks, and a very complete collection is given
by Seidell (10). In Figure 19.27, data for the
Na 2 S04-H 2 system are plotted to show solu-
bility as a function of temperature for a total
solid phase in equilibrium with saturated solution
will be ice. This system exhibits an eutectic at
about 4.5 percent Na 2 S0 4/ at which concentration
the minimum freezing point temperature of
29.84° F occurs. Although the data for this diagram
are presented at a total pressure of 1 atm, the
diagram can be used over a reasonable pressure
range because only liquid and solid phases are
involved. For crystallizer calculations in which the
final temperature is known and in which there has
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION
been no evaporation, the saturated-solution con-
centration, the weight of crystal per given weight
of feed of known concentration, and the species
precipitated can be determined from a solubility
diagram. These problems require only a material
balance for their solution.
For crystallization problems involving energy
balances, such as those in which evaporation occurs
or in which the final temperature of an adiabatic
crystallizer is unknown enthalpy data as well as
solubility data are needed. In such cases, the
enthalpy-composition diagram is convenient.
Figure 19.28 shows the enthalpy-composition
diagram for the Na 2 S0 4 — 2 system. This is a
relatively simple diagram, and the principles
involved in its construction and interpretation
should be familiar to the reader from a study of
Chapter 3. Within any field, the relative quantities
are calculated from the familiar lever-arm ratio. The
CaCl2— 2 diagram shown in Figure 19.29 is
much more complicated, and some explanation of
-150
40
Concentration, weight percent
Figure 19.28. Enthalpy-concentration diagram for the
1atm total pressure. Bases: H = for water at 32°F and the triple point pressure,
H= Ofor Na 2 S04 solid at 32° F and at 1 atm pressure.
521
its features are in order. This system has four
hydrates and a eutectic at point b. Points c, d, e,
and f represent transition points from one hydrate
in equilibrium with solution to another. As with
the Na 2 S0 4 — 2 diagram, most of the data used
here apply at 1 atm pressure. In addition to the
1-atm data, the vapor-liquid saturation lines for
0.5 atm and 0.2 atm are given. All other lines on
the chart apply at 1 atm.
It may be instructive to follow on Figure
19.29 the course of a process in which a mixture of
CaCI 2 *6H 2 and CaCI 2 *4H 2 having an overall
mixture composition of 60 percent CaCI 2 40 per-
,
cent H 2 is heated under 1-atm total pressure. At
low temperatures, the system consists of a mixture
of CaCI 2 -6H 2 and CaCI 2 -4H 2 0. Use of the
inverse lever-arm principle shows the mix to con-
tain about 90 weight percent CaCI 2 -4H 2 0. As the
mixture is warmed, no phase change occurs until
the temperature reaches 86° F. At this point, solu-
tion of composition and enthalpy represented by
-100
-150
50 60
Na 2 S0 4
Na 2 S04-H 2 system at
522 APPLICATIONS TO EQUIPMENT DESIGN

400
proportion of vapor phase present, and increases
the concentration of the boiling liquid phase. The
vapor phase may be assumed to be pure water, and
its enthalpy may be obtained from the steam tables

at the proper temperature and 1 atm total pressure.

Example 19.4. One pound each of Na 2 S0 4

and H 2 50° F are mixed and allowed to reach
at
equilibrium at atmospheric pressure. If the system
is perfectly insulated so that equilibrium is reached

with no gain or loss of enthalpy, what will be the

temperature and phase condition of the product?

Solution The adiabatic mixing process would

be represented on the enthalpy-composition dia-
gram (Figure 19.28) by a straight line joining the
two components of the mixture. At 50° F,
^h 2 o = 18 Btu/lb, ^Na,so 4 =4Btu/lb. Then, the
enthalpy of a 1 1 mix will be 11 Btu/lb, and the
:

concentration will be 50 weight percent Na 2 S04.

These values of H and x put the mix point on the
C.CU-6H O
H— x diagram in the 90° F three-phase region.
-300 Therefore, the temperature is 90° F, and the mix-
20 40 60 80 100
Percentage of calcium chloride ture consists of Na 2 S0 4 Na 2 S0 4 10H 2 O, and
,
-

Figure 19.29. Enthalpy-concentration

saturated solution at 32.7 percent Na 2 S0 4 .
diagram for the
CaCl2-H 2 system. [Hougen, Watson, Ragatz, Chemical The quantities of the three phases present can
Process Principles, Part I, 2nd ed., John Wiley & Sons, New be found using the inverse-lever-arm principle. A
York, (1954).] tie line is drawn from any corner of the three-phase
region to the opposite side through the mix point
point c begins to form. This formation occurs (J). By the inverse-lever-arm principle, the relative

mainly by dissolving of the CaCI 2 -6H 2 phase to quantity of the phase used in drawing the tie line
form saturated solution and some CaCl2 -4H 2 0. As and of the mixture of the other two phases can be
heating continues, the temperature remains at 86° F, found. The inverse-lever-arm principle will also give
but the CaCI 2 '6H 2 continues to dissolve until the relative proportion of the other two phases
only CaCI 2 -4H 2 0and solution remain. During this present. The method is illustrated schematically in

period, the solution composition and enthalpy are Figure 19.30. On this diagram, the relative

continuously represented by point c. From this distances are as follows:

point, further heating increases the system tempera-
AJ 1.79 £7'^ 4.06
ture until it reaches 113°F. At the same time, some
of the CaCI 2 -4H 2 dissolves to form saturated JJ' 3.33 rC~1.88
solution, and the concentration and
solution From these relative distances,
enthalpy move along line cd. Further heating at
113°F causes the CaCI 2 -4H 2 to dissolve and
A
- = —
3.33
- =
.
1
^
.860;
B
- = —— = 0.464
1.88

recrystallize forming CaCI 2 -2H 2 and solution of J' 1.79 C 4.06

condition d. This continues until no further CaCI 2 •

also A + B + C = 2.0 and A + J' = 2.0

4H 2 remains. Still further heating raises the
system temperature, dissolves CaCI 2 -2H 2 0, and Then 4 = 1.860(2.0-4); 4 = 1.300
forms more concentrated solution until, at about
5 + C=0.700
190°F, the fast of the CaCI 2 -2H 2 dissolves,
leaving only the solution phase. From this point, 5 = 0.464(0.700-5); 5 = 0.228
the temperature rises with no phase change until,
and by difference
at 290° F, boiling begins. Beyond this point, addi-
tional heating raises the temperature, increases the C= 2.00-0.228- 1.300 C= 0.472
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 523

which rearranges to

Solution: x = 0.327
Na 2 S0 4 :x = 1.00 C= — — r—^ (19.12a)
Ma
Mh
where
C= mass of crystals in the product magma
per unit time
Ma = molecular weight of anhydrous solute

^x
Na 2 SO4 -10H 2 O
= 0.441
Mh = molecular weight of hydrate crystal
H = 80 Btu/lb X/r' = mass fraction of anhydrous solute in
feed
Composition, weight percent Na 2 S04
X' = solubility of the material at product
Figure 19.30. Application of the inverse-lever-arm principle
temperature expressed as a weight ratio
to a three-phase region, Example 19.4.
of anhydrous salt to solvent

From this, the final answer is

F= total mass of feed per unit time
V= evaporation in pounds mass of solvent
T= 90° F Solution (32.7%) = 1 .300 lb per unit time
Na 2 SO 4 -10H 2 O =0.228 lb
This relatively simple material balance is

Na 2 S0 4 = 0.472 lb applicable to all single-stage crystallization units or

to multistage units from which only one liquid-
Crystallizer Material and Energy Balances solid product withdrawn. In the calculation of
is

the solute in the final liquor, account must be

As was done for evaporation operations, material
taken of the solvent lost by evaporation and of the
and energy balances and a heat-transfer-rate equa-
solvent lost as water of crystallization.
tion can be written for the crystallization process.
The material balance will give the process yield — The determination of V or of the final magma
temperature depends upon a heat balance and a
that the mass of crystals formed from a given
is,
rate equation. For adiabatic crystallization, as in a
mass of solution — if the extent of evaporation or
of cooling can be determined. The general crystal-
vacuum crystallizer, the rate equation is unneces-
lization process shown schematically in Figure
sary, for the extent of evaporation or cooling may
be fixed by an enthalpy balance. In any case, the
19.31 gives the nomenclature involved.
enthalpy balance can be written with the aid of an
Assuming that this process operates at steady
enthalpy-composition diagram and the steam
state with liquid feed and a product magma con-
tables. As in evaporation calculations, the rate
taining crystals and solution characteristic of the
equation is written in terms of an overall
crystallizer content, a solute material balance can
coefficient, and this coefficient must usually be
be written in the form
determined on the basis of experience.
Solute in feed = solute in product crystals + solute
in product liquor
Example 19.5. A single-stage, continuous,
M — a M -M Krystal crystallizer to be used to obtain CaCI 2 -

Fx F '
= C
Mh
+ F(1-x F ')
V-C h a
X' 4H 2 product from
is

a feed containing 40 weight

M h
percent CaCI 2 in water at 180° F. The vacuum
19.12)
system on the crystallizer will give an equilibrium
Evaporated solvent, magma at 90° F.
PTb/hr (a) What range
of heat input per pound of
feed solutioncan be used to obtain a product
Feed, containing only CaCI 2 -4H 2 crystals?
Crystallizer
Flb/hr (b) What would be the maximum yield of
crystal in the product (pounds of CaCI 2 as crystal
Pre •duct magma
per pound of CaCI 2 in feed)?
Llb/hr
Figure 19.31. Schematic diagram of general- Solution This problem can be solved directly
ized crystallization process. using the graphical heat- and material-balance
524 APPLICATIONS TO EQUIPMENT DESIGN

methods developed in Part 1. From Figure 19.29,

the enthalpy of the feed is read to be -8 Btu/lb of
solution.
The products of the crystallizer will be vapor
at 90° F and magma consisting of CaCI 2 -4H 2
crystals in saturated CaCI 2 solution also at 90°F.
The enthalpy of 90° F vapor cannot be read from
Figure 19.29 but is nonetheless fixed and can be
determined from the steam tables. Note also that
the system pressure, although it is much less than
1 atm, cannot be determined from Figure 19.29.
This fact does not prevent the determination of the
vapor enthalpy because, at these pressures,
enthalpy is practically independent of pressure.
Thus, from the steam tables, H v = 1100.9 Btu/lb. Locus of possible
feed + heat
The product magma must have a composition < z F+h- h r*k>
and enthalpy lying on the 90°F tie line between points

saturated solution and CaCI 2 -4H 2 crystals. There-

fore, its overall concentration is limited to 52-61
percent and its enthalpy to a range between —90 Locus of possible
and -120 Btu/lb. magma (x sl h M ) .

points
Using this information, a graphical or an
<x r . hr )
analytical solution of the energy and material
balances can be carried out. The graphical solution
20 30 40 50 60 70 80 90 100
gives a less precise answer but illustrates the Concentration, weight percent CaCl 2
principles better than does the analytical solution.
Figure 19.32. Graphical solution of heat- and material-
Such a graphical solution is shown in Figure 19.32. balance equations, Example 19.5.
The enthalpy and concentration of the feed put it
at point (x F h F ). The vapor conditions are
',
Products from the crystallizer now consist of vapor
plotted at point [y v H v ), and the magma condi-
'

'
,
at (y v
'

, Hv ) and magma at (x L ', hL ). Then,

tions are located along the line from (x s H s ), the '
,

saturated solution at 90° F, to {x c h c ), the crystal '

—
Feed heating required = 270
,
(—8)
conditions. The heated feed must then lie in the
= 278 Btu/lb of feed
range of (x' F + h h F+h indicated. From this infor-
, )

mation the feed heating is fixed. 58.5 - 40

(a) Minimum feed heating = 185 - (-8) = Mass of vapor generated =
58.5
193 Btu/lb of feed. Maximum feed heating =
305 -
(-8) = 313 Btu/lb of feed. = 0.316 lb/lb of feed
Using the maximum heating, the products
(b) 58.5-52
would consist only of crystals and vapor. The yield Mass of crystals obtained = (1 — 0.316)
would be 1 lb of CaCI 2 as tetrahydrate crystal per
61 -52
pound of CaCI 2 the feed. This yield is imprac-
in = 0.494 lb/lb of feed
tical because, the product could not be re-
first,
111
moved from the crystallizer, and, second, the tie 0.494 x
line connecting crystal and vapor passes through 183
Yield of CaCI 2 =
the saturation curve to the right of point d. This 0.40
means that CaCI 2 -2H 2
initially will precipitate = 0.75 lb CaCI 2 in crystal per
and, although at equilibrium it will be converted to
lb of CaCI 2 fed
CaCI 2 -4H 2 0, the conversion will be slow and may
not be complete by the time the product leaves the This yield of CaCI 2 however, gives a magma ,

crystallizer. If the maximum

taken as cor- yield is of (0.494/0.684) x 100= 72 percent crystals, 28
responding to just enough heating so that the percent solution, which is more concentrated than
vapor-magma tie line passes through point d, the most withdrawal systems will tolerate. A magma
construction line (//, H v )-d-(x L ', h L results. ) containing about 55 percent by weight of crystals
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION
is about the most concentrated that is recom-
mended by most manufacturers.
The material and energy balances may not
always be made in as direct a way as is indicated
above. There is some evidence that the assumption
of equilibrium in the interface inherent in the use
of the energy balance is not accurate in the case of
vacuum crystallizers. With these units, rates of
mass transfer from the liquid into the vapor are so
great that the "absolute vaporization rate"
approached (20). This phenomenon is somewhat
analogous to the sonic velocity phenomenon in
that increasing the driving force no longer propor-
tionally increases the rate of transfer. As a result of
this phenomenon, and of the accumulation of
crystals at the interface, the pressure in the vapor
space may be significantly less than would be pre-
dicted on the basis of the liquid phase temperature.
For example, the products from a vacuum crystal-
lizer operating at 40° F may have to be pumped out
from a pressure that would correspond to a 35°F
liquid temperature. This behavior cannot now be
predicted without experimental data.
Crystallization Mechanism
The mechanism by which crystallization occurs is
of more than passing interest because it influences
the conditions within the crystallizer and the
properties of the product obtained. As would be
expected, the deposition of a solid crystal can
occur only as the result of a concentration driving
force from the bulk solution to the interface of the
solid surface. Because the interface concentration
must be the equilibrium concentration for solid
and solution coexistence — that is, because the
interface concentration is that of saturated
solution — the concentration of the bulk fluid
must be greater than saturation. The extent of the
super-saturation depends upon the number and
shape of crystals upon which precipitation occurs,
the temperature level, the solution concentration,
and the violence of the agitation present.
Crystallization begins with the nucleation
mechanism by which a small crystal forms. In a
homogeneous liquid, the process starts with a
random association of solute molecules occurring
by normal molecular motion. In most cases, this
cluster dissassociates, again through normal
molecular activity. In some cases, additional molec-
ules join the cluster and it begins to assume a
regular spacing of molecules and form a new phase.
At this point, it is called an embryo. The embryo is
also usually short-lived, redissolving easily because
the concentration gradient favors transfer from the
525
embryo. As the supersaturation increases, the addi-
tion more molecules to the embryo be-
of still
comes more likely, the embryo grows and stabil-
izes, and a crystal nucleus grows. Study of the
thermodynamics of surfaces shows that the solu-
bility of crystals decreases as the size increases.
Thus, once a crystal nucleus is well formed, it
tends to grow.
Primary nucleation, which includes homo-
is geneous nucleation as described above, and nuclea-
tion on minute insoluble particles, called hetero-
geneous nucleation, occur only to a minor extent
in a crystallizer magma. In these situations,
secondary the precipitation onto
nucleation,
crystal surfaces, accounts for most solids forma-
tion. Additional crystals also form by attrition
when the crystals break in the violent agitation of
the crystallizer magma. Secondary crystallization
occurs through the formation of new nuclei at
surfaces and corners of "seed" crystals already
present in the magma and the breaking away of
these nuclei, and through the formation of nuclei
in areas of low-energy content close to crystal sur-
faces or to surfaces of the crystallizer. Crystals may
also form via heterogeneous crystallization where
foreign matter forms a site for nucleation and
crystal growth. The situation is complex, but most
of these mechanisms result in increased nucleation
in situations of high supersaturation and in concen-
trated magmas (19).theories concluded
Earlier
that nucleation could occur only at supersatura-
tions greater than some minimum value and that in
the region of lower supersaturation no nucleation
could occur (13,14). This is an oversimplified view,
even for homogeneous nucleation, and clearly is
not realistic for practical crystallizer conditions.
Crystal growth is at least as complex a process
as nucleation. The most probable of growth-
mechanism theories pictures crystal growth occur-
ring at dislocations on the surface. The dislocation is
self-perpetuating, and the crystal grows in a spiral
as a result of surface forces. Screwlike surface
irregularities have been found on photomicro-
graphs of crystal surfaces, thus substantiating this
theory. Despite the complex growth mechanism, it
is instructive to approach crystal growth using the
classical rate-equation forms. Rate equations for
crystal growth have been written using the two-
film theory. Separate equations were written for
the diffusion of solute torn the edge of the
laminar layer to the crystal face (15) and for the
turbulent transfer of solute from the body of the
solution to the laminar interfaced). These equa-
tions were then combined to give an equation for
526 APPLICATIONS TO EQUIPMENT DESIGN

transfer across the entire path from bulk fluid to For example, the factor for a cube is g because
solid surface in terms of a liquid-phase transfer the area 6l 2 the volume is Z. 3 and the mass is
is , ,

coefficient.As shown in Chapter 13, equations of L 3 p = lA Lp, Then Equation 19.15 can be written
this can be expressed in a single relation
sort as
applicable from solid boundary to bulk fluid.
dL K
Application of Equation 13.81 gives — —
dd
— =
(j>p
L
(Cc
s
— ca )
' (19.17)
-4{EN + 9)A(cs -ca )
/-iojioi
NA MA = (19.13) This equation states that the rate of linear crystal
7/v* growth is independent of crystal size, an expression
where of the McCabe AL law that was initially proposed
NA =
lb moles of solute deposited per unit from empirical evidence (11,12). Of course, the
time rate of volume growth will not be constant, for
M^ = molecular weight of solute
EN = integrated mean eddy diffusivity V=0'l 3 ; dV=3<p'L 2 dL
applicable over the entire transfer where 0' is a shape factor different from 0. Expres-
path, ft 2 /hr sing Equation 19.17 in terms of volume increase
® = diffusivity, ft 2 /hr gives
cs , ca = concentrations of solute at the solid
surface and in the main bulk of the
dV _ 3<f>'L
2
KL
3
(cs - ca (19.18)
fluid phase, lb/ft dd <$>p

A = area of crystal surfaces, ft 2

This equation fails in two ways: 1) for many
7/v = ratio of concentration gradients, see
crystallization systems the growth rate is a func-
Equation 13.47
x = length of entire transfer path, ft
tion of crystal size (the McCabe law
not valid), is

and 2) at high values of the

supersaturation,
The group 4(E N +@)/y N x is the liquid-phase growth rate is more usually constant than directly
mass-transfer coefficient (k L ) in Chapter
defined varying with supersaturation level. The failure of
13. Moreover, here there is no transfer through the the McCabe law has been attacked by written
although there is a resistance to incor-
solid phase, empirical relations between growth rate and size,
poration new molecules into the surface ks
of .
but none of these have been generally accepted.
Defining an overall transfer coefficient K L = 1/ Still, the deviation occurring at high supersatura-
(Mk L + Mk s ) tion seems to support the notion that under these
conditions the growth rate is limited by surface
-NA M A =KL A(c -ca s ) (19.14)
nucleation kinetics, not by diffusion.
Equation 19.14 can be written for a single crystal Generally, crystallization is carried out at very
ifA is taken as the area of a single crystal. In this low supersaturation so that nucleation is slow. For
case these conditions, the crystal size distribution

-NA MA =-
dm
—
dd
= KL A(cs -ca ) (19.15)
(CSD) can be at least roughly estimated on the
presumption that nucleation does not occur — that
is, the number of crystals is constant — and that

where the initial CSD is known. In some cases the crystal-

lizer is "seeded" with fine crystals upon which
m= mass of a crystal, lb
growth can occur. Usually this is not necessary,
6 = time, hr
because a large number of crystals are continuously
because steady state exists for any differential available in the magma. In both cases, the crystal-
section of crystal surface. lizer is operated to maximize crystal growth and
It has been found experimentally that often restrict nucleation, thus producing large crystals
the shape of a crystal does not change during that are easily filtered, relatively pure, and gener-
growth, so that ally more desirable in the marketplace. Where the

<pALp=m (19.16)
initial CSD is known, it is common practice to
calculate the final CSD using the McCabe AL law.
where
The calculation of CSD for a seeded crystals
= shape factor, dimensionless (see App. B) lizer by the method indicated above gives at best a
L = characteristic crystal dimension very crude estimate of the actual distribution.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 527

Nucleation cannot be entirely prevented, and often product particle-size range will be determined.
a classification process also takes place in the Since the fractional growth in mass of small
crystallizer, resulting in longer holding times for particles is larger than for large ones, the shape of
small crystals. Since agitation is frequently violent, the screen analysis curve will change.
large crystals may break and erode. A reverse effect
results from the fact that small crystals are more
Example 19.5. A that forms cubic
solute
soluble than large ones as the result of a surface-
crystals is to be precipitated from solution at a rate
energy effect. Thus, large crystals may grow at the
of 10,000 lb of solid (dry basis) per hour using
expense of small ones. crystals. If no nucleation
1,000 Ib/hr of seed
Despite the inexactness of the results, such an
occurs and the seed crystals have the following size
estimate does give a useful first approximation of
distribution, determine the product size distri-
the particle size of the product to be expected
bution:
from a given seed. The seed crystals used are un-
Weight Fraction
avoidably in a range of particle sizes. Even so, the
Tyler Sieve Mesh Retained
relation between seed and product particle sizes
may be written as -48 + 65 0.10
LP =L +AL S
(19.19) -65+ 100 0.30
-100+150 0.50
where L is again a characteristic particle dimension -1 50 + 200 0.05
and the subscripts s and p refer to seed and -200 + 270 0.05
product respectively. In Equation 19.19, AL is
constant throughout the range of sizes present, as
Solution. Here, the seed CSD and the
shown by Equation 19.17. From this equation, the
product-to-seed weight ratio are given, and so the
seed and product masses may be related (11) for
product CSD can be determined by trial-and-error
mp = <t>'pL ,
3 = <t>'p(L s + AL) 3 (19.20) selection of AL values followed by a summation-
of-finite-parts approximate solution of Equation
and 19.23. The characteristic particle dimension may
m s
= 0'pZ. s
3
119.21) be taken as equal to the average of the smallest
screen opening through which a particle passed and
which combine to give that on which it was retained. Thus, the seed
crystal dimensions are
('£)' 119.22)
Weight Fraction Average Dimension
lAms ) (Ls ), in.
Here, the crystals in the seed have been assumed
all

to be of the same shape, and the shape has been 0.10 0.0099
assumed to be unchanged by the growth process. 0.30 0.0070
This assumption is reasonably close to the actual 0.50 0.0050
conditions in most cases. Equation 19.22 was 0.05 0.0035
written for the entire crystal mass. It may also be 0.05 0.0025
written for differential parts of the crystal masses,
each consisting of crystals of identical dimensions. As a first attempt, a AL = 0.005 in. will be used
The resulting differential equation can then be Then,
integrated over the entire range of particle sizes.
/ AZ.V3
Ams
mp ms ALV
r

Jo ^ =
r

J
I

l
1+
r)
L s l
dms=m
0.10 3.37 0.337
119.23) 0.30 4.94 1.480
0.50 8.00 4.00
The indicated integration can be carried out step- 0.05 14.45 0.723
wise for each small but finite particle-size range in 0.05 27.0 1.35
the seed crystals, using successively assumed AL's 1.00 7.89
until the product-to-seed ratio reaches a desired
value; conversely, it can be integrated with known Since mp lm s is specified as 10.0, the AL chosen
AL to get the mass of product. In either case, the was too small. Succeeding trials lead to a final
528 APPLICATIONS TO EQUIPMENT DESIGN

AL = 0.0058 for which

Ams
Kf Amp h
0.10 3.95 0.39 0.0157
0.30 6.00 1.80 0.0128
0.50 10.10 5.05 0.0108
0.05 18.70 0.94 0.0093
0.05 36.60 1.83 0.0083
1.00 10.01

The particle sizes of product crystals can be

obtained by adding AL to the dimension L of each
fraction of the seed crystals. The weight fraction of
product crystals (Amp ), represented by each incre-
ment of seeds, is computed for all increments.
Mass Fraction Product Cumulative
Product Fraction Smaller
(mp rZAm p ) Lp = Ls + AL than L p

0.039 0.0157 0.961

0.180 0.0128 0.781
0.504 0.0108 0.277
0.094 0.0093 0.183
0.183 0.0083 0.000
1.000 0.1 0.2 03 04 0.5 0.6 0.7 0.8 0.9 1.0
Cumulative weight fraction of total mass smaller
The product and feed size distributions are than specified size
plotted Figure 19.33 as cumulative weight per-
in Figure 19.33. Crystallizer-seed and product particle-size
cent of the total sample smaller than the size distribution, Example 19.5.
indicated. Note the flattening of the product distri-
bution curve at the small-particle size range. It is Tyler Sieve Screen Analysis
caused by the uniform increase in dimension Mesh Weight Fraction
experienced by the vastly larger number of small
particles compared to the same size increase -28 + 35 0.01
experienced by the smaller number of large -35 + 48 0.19
particles. In other words, the small crystals have a -48 + 65 0.70
very large surface-to-volume ratio compared to that -65+ 100 0.10

of the large crystals. Therefore, the small crystals

receive a much higher quantity of mass per unit The flattening of the size distribution mentioned
volume of original seed than do the large crystals. above is again evident.
The product size-range increments are not The results obtained in Example 19.5 are
expressed in the screen-size units that were used to typical of those obtained with the McCabe AL law.
describe the seed particles. A conversion to screen Two major assumptions have been made: first,
sizes can be made through use of Figure 19.33 as there is no nucleation and, second, every particle

follows: grows through the same increase in linear dimen-

Weight Fraction of sion. Neither of these assumptions fits the facts of
Tyler Sieve Screen Opening, Product Smaller than typical crystallization behavior, but the second is
Mesh in. Opening much closer than is the first, particularly when AL
is small compared to L s For relatively large values
.

100 0.0058 0.0 of AL, the crystals may grow unsymmetrically.

65 0.0082 0.10
When large crystals are grown, the larger crystals
48 0.0116 0.80
experience a greater AL than the small ones.
35 0.0164 0.99
28 In practice, nucleation cannot be prevented.
0.0232 1.00
In typical recycling crystal lizers, a given particle in
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 529

the crystallizing mass passes cyclically through and the number leaving by growing beyond L 2 will
regions of high supersaturation and regions of low be
supersaturation. In high-supersaturation regions,
such as that immediately after addition of hot feed Vn 2 GAT 19.256)
in a vacuum crystallizer or after flashing into the If particles also enter because of input and output
vapor separation space of an Oslo crystallizer, flows, these will be
nucleation rates will be high. In regions of low
supersaturation, there will be virtually no nuclea- Q,A7,AZ.Af (19.25c)

tion. If high concentrations of crystals are main- and

tained in the recycling mass, supersaturation can
often be kept low enough throughout the crystal-
QnALAt ;19.25g0

lizer to minimize nucleation. The presence of respectively, where n-, and n are the average popula-
relatively few crystals will be unfavorable for tion densities in the range L^ to L 2 in the feed and
total deposition, and thus the level of super- exit streams. Combining these relations in a number
saturation will be raised. Growth of each crystal balance gives
may thereby be faster, but then nucleation
generates many new small crystals.
QirijALAt + Vn :
GAt = QnALAt + Vn 2 GAT
In the analysis of the performance of pilot or (19.25)
Rearranging, one gets
plant-sized crystal izers, the model most often used
I

is that of the mixed-suspension— mixed-product- VG[n 2 - /?i ) = {(2,*, - Qn) AL (19.26)

removal (MSMPR) crystallizer. The model require-
ments are that the unit operate at steady state, that As AL approaches zero, the values of n become
there is no product classification within the crystal-
point values, and hence
lizer magma, that the product withdrawn be
typical of the magma in the crystallizer, and that VGdnldL = Q,n-, - Qn (19.27)
the feed contain no crystals. In addition, it is
common to assume that the McCabe AL rule If there are no crystals originally large enough to
applies, and that there is no crystal breakage. The be in the size range being considered, n, = and
basic conditions of the MSMPR crystallizer are
dn Q
easily approximated in the laboratory and are close
to those found in many commercial units. Under
—
n
=
VG
dL (19.28)
these assumptions, one can write expressions for
the crystal growth in the magma arriving at CSD If we now define V/Q as the residence time, r, and
and nucleation rates. define n° as the population density of embryo-
Define the population density distribution of sized crystals, very close to zero size, Equation
particles in the magma as 19.28 can be integrated

A/V = f
•L 2
'
n dL (19.24)
r

V
n
dn/n = - rL

J
o
—
dL
Gt
(19.29)

to give
where A/V the number of particles in the size
is

range L^ to L 2 with n being the population density n = n° exp(-L/Gr) 19.30)

expressed as numbers of particles per unit volume
per measure of particle size. This definition can Equation 19.30 is the fundamental equation
then be used to develop an expression for/7 under for the MSMPR crystallizer, but is restricted by all
the MSMPR crystallizer assumptions. the simplifications initially imposed. These include a
Consider an arbitrary size range, L-\ to L 2 in , constant crystal growth rate and a fixed retention
which the growth rate of crystals will be G and in time. From this equation, other relations can be
which the population densities will be n^ and n 2 ,
derived for all the properties, such as crystal sur-
respectively. For an increment of time Af, the face area and mass, that depend upon CSD. These
number of particles entering the size range through equations can also be used to estimate crystal
growth will be growth rate G and nucleation rate from data on an
MSMPR crystallizer.
Vn^GAT 119.25a) Some of the more useful functions derivable
530 APPLICATIONS TO EQUIPMENT DESIGN

from the basic population density are: These relations depend upon the MSMPR
limitations. A more fundamental population
CnlSdL balance equation can be written along a one-
•o
Mass-weighted mean size Lm = = 4Gt dimensional size coordinate for a mixed suspension
as

•0 bn 3 (log V) b(Gn) Q/n,- Qo"o

_

+n + — + B-D
(19.31) bt bt bL V V
(19.36)
\~nLdL
where B and D are functions expressing the birth
Population-weighted mean size L p =
—•o
and death rate of crystals, the second term allows
\~ ndL for changes in magma volume, and the third term
can include a growth rate that varies with particle
size. In general, relaxation of the MSMPR bounds is
(19.32)
not possible at this time, although work has been
2
done on crystallizers in which fine crystals were
Specific surface area >4 =p" nL 2 dL = 2<t>"n°(GT) destroyed, or in which the exit stream resulted
•o
after crystal classification. Selective removal of
(19.33)
large crystals reduces the general crystal size range,
and as would be expected. Accelerated removal of fines
not only increases the mean particle size but also
Solids concentration X= p0' 2
nL dL
•o broadens the size range of crystals obtained. It can
even produce a bimodal CSD that may be very
= 6p0'n°(Gr) 4
difficult to filter. These and other cases of interest
(19.34) to crystallizer operators are discussed by Randolph

In these equations p' is the volume shape factor

(18).

introduced above and p" the area shape factor.

Both specific surface area and solids concentration Crystallization from Mixed Solutes

are expressed per volume of magma. Crystallization from solutions of several solutes
The predominant crystal size is generally generally can be made to result in deposition of
defined as that size in which the largest fraction of crystals of one of the solutes instead of crystals
the total mass exists. It can be obtained by getting that are a mixture of the two. Thus, separation of
the graphical or algebraic expression for the mass one of the components from siich a solution is
as a function of crystal dimension and noting the possible by crystallization. The prediction of the
node where the curve is a maximum. The dominant products to be obtained from such a system cannot
size Lq is then be made based upon the two single-solute— water
phase diagrams, because the presence of the second
Ld =3Gt (19.35)
solute affects greatly the phase behavior of the
Because both laboratory and full-scale first. For example, where the two solutes contain a
crystallizers can be operated under MSMPR condi- common ion, such as Na 2 S0 4 and Na 2 C0 3 the ,

tions, these relations allow for empirical evaluation solubility of each solute is greatly reduced.
of crystallization kinetics and for a study of size Figure 19.34 is the solubility diagram for the
effects upon the ability of a crystallizer to improve system Na 2 C0 3 — Na 2 S0 4 — H 2
at temperatures
operations via fines distinction, classification ranging between 5— 25° C.
Within this temperature
mechanisms, staging etc. and composition range, only the decahydrates of
Nucleation and growth rates could be the two salts can crystallize out of solution. At
obtained in an MSMPR crystallizer directly from higher temperatures, crystals of Na 2 S0 4 Na 2 S0 4 — ,

Equation 19.30 using experimentally determined Na 2 C0 3 Na 2 C0 3 -7H 2 0, and Na 2 C0 3 -H 2

CSD. Generally, the operation should be adjusted
to reduce the nucleation rate and increase growth
rate on the assumption that large crystals are
, can
form depending upon the temperature and compo-
sition range. In Figure 19.34, point D represents
Na 2 C0 3 10H 2 O, whereas point C represents
-

desired. Thus, high circulation rates arranged to Na 2 SO 4 -10H 2 O. At 25°C, the area bounded by
reduce supersaturation should be beneficial. GHBI represents unsaturated solutions of the two
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 53

G 100% H 2 begin to precipitate, but this would not occur until

the solution reached about 10°C.
The product from any crystallization separation
will consist of the crystals of precipitated solute
and the entrapped mother liquor. This trapped
solution will carry with it all the impurity to be
found in the product. Occlusion of impurities is
particularly troublesome with coarse precipitates,
but fine crystals hinder washing and physical
removal of solution.

AUXILIARY EQUIPMENT
Evaporation and crystallization processes require a
variety of auxiliaries for successful operation.
Slurry pumps, vacuum pumps, mixers, and surface
condensers are described elsewhere and hence will
Weight % Na 2 S04 *
not be discussed here. Steam traps, barometric
Figure 19.34. Solubility of Na 2 C03-Na2S04-H 2 condensers, and entrainment separators are not
system at 5 to 25°C. [Hougen, Watson, Ragatz, Chemical otherwise discussed; because they are vital to
Process Principles, Part I, 2nd ed., John Wiley & Sons, New
evaporation and crystallization operations, they
York, (copyright ©1954).]
will be dealt with here.

salts. The rest of the diagram represents hetero-

geneous mixtures of crystals and saturated solutions.
In the area BIC, the crystals are pure Na 2 S0 4
10H 2 O; in the area HBD, they are pure Na 2 C0 3
10H 2 O; below line DBC, they are mixtures of
Na 2 CO 3 -10H 2 O and Na 2 S0 4 10H 2 O. Similar
regions exist at other temperatures as indicated
on the diagram. For example, 5°C area ACJ
corresponds to area BIC at 25°C. This diagram
is not useful in the area below UneDBC, as can be
inferred by comparing it with the binary
Na 2 S0 4 — 2 data given in Figures 19.27 and
19.28. From this diagram, it is
one or the other of these salts may be fractionally
crystallized from any given solution. For example,
a solution of composition represented by point 1
at 25°C is in the unsaturated region.
ated at 25° C, it will move as indicated by line 1—2.
At point 2, crystals of pure Na 2 SO 4 -10H 2 O will
begin to precipitate and will continue to crystallize
as evaporation continues. During this step, the
solution concentration will move along line 2—B
until point B is reached. At this point, Na 2 C0 3
10H 2 O crystals will begin to form. As another
approach, the same original solution could have
cooled to lower temperatures. When the tempera-
ture reached about 21°C, precipitation of
Na 2 SO 4 *10H 2 O would begin. This precipitation
Steam Traps
-
Steam traps must be placed in the discharge line
-
from any steam-condensing unit to prevent the
steam chest from filling with condensate and to
-
prevent live steam from escaping from the dis-
charge line. Some traps are arranged to allow non-
condensables to be removed from the process unit.
This is particularly important in condensers, for
small amounts of noncondensables blanket the sur-
faces, greatly reducing heat transfer.
Three types are common: mechanical traps,
quite evident how orifice traps, and thermostatic traps.
Mechanical traps may be of the bucket or the
float type. In either case, the mechanical trap
operates on a buoyancy principle. In the bucket
If it is evapor- trap, condensate is admitted under a bucket, flows
around it, and discharges from a valve above it. If
steam enters the trap, the bucket fills with steam
and rises to float. This action closes the discharge
valve and prevents further flow. Noncondensables
are prevented from floating the bucket by a small
*
vent hole placed in the top of the bucket. Of
course, a small amount of steam is lost through this
vent. The sequence of operations of an inverted
bucket trap is shown in Figure 19.35.
In the float trap, a mechanical member is
lifted by the collecting condensate until it opens a

would continue as cooling continued, with solution discharge valve, thus allowing flow through the
concentrations moving along line 1—3. When the trap. As the condensate layer flows out of the trap,
solution reached point 3, Na 2 CO 3 '10H 2 O would its level drops, thus lowering the float and closing
532 APPLICATIONS TO EQUIPMENT DESIGN

(a) (b) (c) (d) (e)

Figure 19.35. Sequence of operation of an inverted-bucket trap. (Courtesy of

Armstrong Machine Works.) (a) Trap newly installed-steam off, bucket down,
valve open, (b) Steam turned on— condensate reaches bucket and flows through it
without lifting the bucket, (c) Steam reaches trap-condensate is displaced from
under the bucket. This condensate displacement floats the bucket and closes the
valve. Noncondensables escape through the pinhole in the bucket and collect at the
top of the trap, (d) Condensate flows to trap— the steam under the bucket
condenses, and, as more condensate reaches the trap, the bucket begins to fill with
water and lose buoyancy, (e) Condensate flows through trap— as more condensate
flows to the trap, the bucket sinks, opening the valve and allowing condensate to
pass through. When steam again reaches the trap, the bucket will again rise and
close the valve as in (c).

it may also be used to control liquid level in a tank,

or to control the interface level between two
liquids. In this the light liquid would be
case,
admitted through the top of the trap and the heavy
liquid through the side opening.
Thermostatic traps operate on a thermal-
expansion principle. The two major ones are the
balanced-pressure type and the metal-and-liquid
expansion type. In the expansion type, condensate
flows around an operating cylinder and out a valve
port. The operating cylinder contains liquid held
between the cylinder and a metal bellows.
Attached to the bellows is a plunger that can close
the valve port when the bellows contracts. If steam
enters the trap, the liquid around the bellows
expands as its temperature rises, thus contracting
Figure 19.36. Cross section of a compound- the bellows and closing the discharge valve. Figure
lever, ball-float trap. (Courtesy of Armstrong 19.37 shows the principle of operation of the
Machine Works.) liquid expansion trap. This trap releases subcooled
condensate and hence conserves some of the
the discharge valve. Figure 19.36 shows a sensible heat in the condensate. This means that a
compound-lever float trap. significant quantity of condensate may build up in
Ball-float traps are operable over wider ranges
of pressure and load than are inverted bucket traps Drips

and will handle larger quantities of noncondens-

ables than an unmodified inverted-bucket trap.
k-^w-v-
However, the inverted-bucket trap can be supplied
with a thermostatic air vent to handle large
Steam and
condensate a W^y :

3
quantities of noncondensables; it is smaller, and it cD Bellows ' ^ Uquid
-Hs
is less likely to be damaged by air hammer. The

ball-float trap has an additional connection so that Figure 19.37. Liquid expansion thermostatic trap.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 533

Valve

Outlet

Figure 19.38. Balanced-pressure thermo-

static trap.

the system and reduce the heat-transfer area avail-

able. The extent of subcooling may be adjusted by
turning the knob N so that the condensate leaves at
any temperature below 212°F.
The balanced-pressure thermostatic trap is FLOW
shown diagrammatically in Figure 19.38. In this
trap, the steam passes around a bellows containing
a small quantity of volatile liquid. As this liquid is
heated, its vapor pressure increases until it exceeds
the pressure of the steam around the bellows, at
which point the bellows expands and closes the
discharge valve. When the bellows and contained
liquid are cooled again by condensate, the vapor
pressure of the contained liquid decreases and the
valve opens again. The contained liquid is selected
so that its vapor pressure is greater than the steam
pressure, regardless of how the steam pressure
varies.
Thermostatic traps are small and light, and Figure 19.39. Cross section of Yarway Orifice Trap:
they may be installed in any position. Since they (1) body, (3) seat adapter, (4) valve seat, (5) bonnet,
are wide open when cold gas surrounds the (6) name plate, (7) split washer, (8) cap nut, (9) lock pin,

bellows, they cannot air bind. They are unaffected (10) lock nut, (12) control cylinder, (13) valve, (17) screen,

by vibration or motion and are not easily damaged (19) lock nut (strainer), (20) blow-down body, (21) valve
(strainer), (2,11,14,15,16,18) gaskets. Courtesy Yarway
by water hammer.
Corporation.
Orifice traps operate on the difference in
throttling action of steam and of water passing
through a succesion of orifices. With steam flow,
sonic choking limits the rate of discharge and a cross-sectional view of the Yarway orifice trap, a
builds up the intermediate pressure between two patented form of steam trap. The two orifices that
orifices in series. With water flow, no such choking control flow are at the opposite ends of the central
occurs, and the intermediate pressure between two passage through the valve body (13). With water
orifices in series is the median between the up- flowing through this passage the pressure drop is
stream and downstream pressures. This difference small alowing the valve to rise in the control cylinder
between intermediate pressure when steam flows through the action of a spring. The main flow of wa-
and when water flows can be used to operate a ter will then proceed through the valve seat (4). With
valve closing and opening the trap. Figure 19.39 is steam flow the pressure drop is great enough to
534 APPLICATIONS TO EQUIPMENT DESIGN

'Live steam inlet

Condensate inlet

Figure 19.40. Pressure-return trap.

Outlet

High-vacuum
ejector
Steam inlet

Water
inlet

Water-inlet for
conical
sprays

Condenser

Vapor inlet
Manhole

Handhole

Barometric leg

Hotwell

Figure 19.41. Barometric condensers. (Courtesy of Acme Process Equip.,

Company.) A two-stage countercurrent barometric condenser, {b) multijet
(a)

barometric condenser. (b)

 SIMULTANEOUS HEAT AND MASS TRANSFER
force the valvebody down against the seat. As long
as steam is present the valve remains closed, but a
small leak blows through the valve passage.
The orifice trap is small and light It will operate
under a wide range of pressures and handle non-
condensables as well as steam. It is, however, af-
fected by corrosion and by clogging of the orifices
with dirt or scale.
Other types of steam traps that operate on
energy balance principles are available, but they are
not as common as the types discussed above.
Condensate Discharging Methods
In multiple-effect evaporators and in vacuum
crystallizers,condensate must be discharged from a
vacuum space to the atmosphere. The discharging
is often done with surface condensers followed by
a pressure return trap in conjunction with check
valves. Figure 19.40 is a schematic diagram of this
kind of trap. Condensate enters the trap but is
prevented from leaving because the trap is at sub-
atmospheric pressure. As the liquid level rises in
the trap, a ball float actuates two valves, closing
the condensate-feed valve and opening a live-steam
Cleaned
air
Separated
Figure 19.42. An entrainment separator using both centrifugal and impinge-
ment mechanisms to effect separation. (Courtesy of Schneible Co.)
- EVAPORATION AND CRYSTALLIZATION 535
valve. Steam enters the trap, forcing the discharge
check valve to open and thus emptying the trap.
When the level drops to a preset point, the float
returns the condensate valve to the open position
and closes the live-steam valve.
Another, and even more common, solution to
the problem is through a barometric condenser.
Here, a surface, or more commonly
a jet, con-
denser is on
top of a long tail pipe.
located
Condensate stands in the tail pipe high enough so
that the hydrostatic head balances the pressure
difference between the vacuum space and the
atmosphere. Figure 19.41 shows two types of baro-
metric condensers. In the two-stage barometric
condenser, vapors are contacted countercurrently
with a falling-water shower. Noncondensables are
handled in a jet ejector after each stage of the
condenser. In the jet barometric condenser, vapors
are entrained into a high-speed water jet. This jet
acts as both steam condenser and entrainer of
noncondensables. The jet condenser eliminates the
steam use of a steam-jet ejector with a barometric
condenser, but operates best at pressure greater
than 2 in. Hg. The water required by a barometric
Air plus
entrained
liquid
liquid
536 APPLICATIONS TO EQUIPMENT DESIGN

condenser may be estimated from a simple heat 6. Dukler, A. E., and L. C. Elliot, "Experimental and
balance. Water discharged to the hot well is usually Analytic Studies of Heat Transfer in a Falling Film

at 100-1 20° F. System," Final Report, Part II, Houston Research Inst.

These condensers are always much smaller (March 1968).

7. Frank, F. C, Discussions Faraday Soc, No. 5, p. 48
than surface condensers are for the same duty.
(1949).
They require about a 40-ft headroom and may use
8. Keenan, J. H., and F. G. Keyes, Thermodynamic
more water than a surface condenser. Because of
Properties of Steam, John Wiley & Sons,New York
operating convenience and reduced first cost, they
(1936).
have become almost standard in those applications 9. Keenan, J. H., F. G. Keyes, P. G. Hill and J. G. Moore,
where the vapors are neither valuable, corrosive, Steam Thermodynamic Properties of Water
Tables,
nor hazardous to health. Low-level jet condensers Including Vapor, Liquid, and Solid Phases (Inter-
are available in which the water is injected at national Edition — Metric Units), John Wiley & Sons,
sufficiently high velocity to recompress the non- New York (1969).
condensables and discharge both water and gases to 10. Lenke, W. F., Seidell's Solubility of Inorganic and
atmospheric pressure without a barometric leg. Metal-Organic Compounds, 4th ed., 2 vols., Van
Nostrand Co., Princeton (1958).
11. McCabe, W. L., Ind. Eng. Chem., 21, pp. 30, 112
Entrainment Separators
(1929).
The entrainment of liquid droplets in the vapor 12. McCabe, W. L, and C. S. Robinson, Ind. Eng. Chem.,
phase can become a serious problem in evaporators 16, p. 478 (1924).
and crystallizers as well as in mass-transfer equip- 13. Miers, H. A.,./ Inst. Metals ,37, 1,p. 331 (1927).
ment. In standard vertical-tube and horizontal-tube 14. Miers, H. A., and F. Isaacs, J. Chem. Soc, 89, p. 413
evaporators, the large vapor space in the evaporator (1952).

body is usually sufficient to let all but the very 15. Noyes, A. A., and W. R. Whitney, J. Am. Chem. Soc,
19, p. 930(1897).
smallest droplets fall out of the vapor phase. In
16. Perry, J. H., C. H. Chilton, S. D. Kirkpatrick, Chemical
forced-circulation and long-tube vertical evapor-
Engineers' Handbook, 4th ed., McGraw-Hill, New York
ators and in Oslo crystallizers, the vapor space is
(1963).
seldom sufficient to separate effectively these 17. Preckshot, G. W., and G. G. Brown, Ind. Eng. Chem.,
droplets from the high-velocity vapor-liquid 44, p. 1314 (1952).
streams. In these cases, auxiliary entrainment 18. Randolph, A. D., Chemical Engineering, pp. 89—96
separators are often necessary. These separators (May 4, 1970).
usually operate on one or both of two principles. 19. Randolph, A. D., and M. A. Larson, Theory of Partic-
Either they operate as cyclone separators on a ulate Processes, Academic Press, New York (1971).
centrifugal-force mechanism (see Chapter 22), or 20. Schrage, R. W., A Theoretical Study of Interphase Mass
they coalesce the droplets by impingement with Transfer, Columbia University Press, New York (1953).
baffles or with metal mesh. Figure 19.42 shows an 21. Sinek, J. R., and E. H. Young, Chem. Eng. Prog., 58,
No. 12, p. 74(1962).
entrainment separator that operates on a combina-
tion of these two mechanisms. The feed enters
tangentially so that some droplets are thrown to
the walls by centrifugal force. It then passes
PROBLEMS
through a set of baffles that cause further separa-
tion by impingement. 19.1. A standard, two-effect evaporator with forward
feed, barometric condenser, two-stage jet ejector, and
manual controls has been operating at stable conditions for
some time. If, with no change in operating controls, the
REFERENCES operator notices each of the following effects, what might
be probable causes of each of these effects, and how might
they be eliminated?
1. Berthoud, A., J. Chem. Phys., 10, p. 624 (1912).
(a) Product rate increases, but concentration falls off
2. Bunn, C. W., Discussions Faraday Soc., No. 5, p. 132
proportionately.
(1949).
Butler, R. M. Thesis, University of
(b) Steam consumption increases, but process-stream
3. London (1950).
4. Caldwell, H. B., and W. D. Kohlins, Trans. A.I.Ch.E., flow rates remain unchanged.
42, p. 495(1946). (c) Production rate and steam consumption decrease.
5. Chun, K. R., and R. A. Seban, J. Ht. Trans., 94, p. 432 Steam pressure, vacuum, and feed and product concentra-
(1972). tions remain unchanged.
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 537

(d) Steam consumption increases, and pressure in the

first effect increases.
(e) Production rate decreases while temperatures
increase in both effects.
19.2. Characteristics of liquids that offer problems in
evaporation are:
Steam
1. Salting.

2. Foaming.
3. Scaling.
4. Corrosive as requiring replacement of tubes.
5. Corrosive as involving expensive materials of con-
struction.
(a) With the vacuum line sealed off and water in the
6. Heat sensitivity.
system, what are the conditions at points 1 and 2 {P, T, and
7. High viscosity.
flow rate)?
8. High boiling-point rise.
(b) What would these conditions be for a 30 percent
List which of these troublesome characteristics are taken NaOH solution?
care of in evaporation of the following types: (c) What would they be for a 50 percent NaOH
solution if water T= 10°C?
(a) Horizontal tube.
19.7. An evaporator is being installed to concentrate
(b) Vertical short tube.
caustic soda solution from 48 percent NaOH to 80 percent
(c) Long-tube vertical.

(d) Forced circulation.

(e) Turbulent film.
VAPOR PRESSURES OF AQUEOUS SOLUTIONS OF
NaN0 3 (I.C.T. Ill, p. 372. By permission of Nat. Acad. Sc.
19.3. Recommend a type of evaporator for each of Copyright ©1928).
the following applications, and briefly give the reason for
your choice: Concentration, Vapor Pressure
g INdlMU3/
(a) Concentration of grapefruit juice, the solution is

nonfoaming. Stainless
100gH 2 O 0°C 25° C 50°C 75°C 100°C 125°C
steel is to be the material of con-
struction.
4.58 23.76 92.54 289.32 760.0 1740.5
(b) Concentration of glycerol solution from soap
5 4.50 23.34 90.9 284.1 746.3 1709
making. Water vapor goes to the condenser. Glycerol pro- 10 4.42 22.93 89.2 278.9 732.5 1677
duct is somewhat viscous at the operating condition. 20 4.28 22.14 86.1 268.8 705.6 1615
(c) Corrosive "black liquor" used in sulfate pulp 30 4.15 21.39 83.1 259.1 679.6 1554
mills is to be concentrated through a small range. Nonsalt- 40 4.04 20.69 79.6 249.9 654.8 1496
ing. No foam problems. 50 3.93 20.04 77.5 241.1 631.3 1442
60 3.83 19.42 74.9 232.9 609.0 1390
19.4. Glycerol— water solutions may be assumed to
70 3.73 18.83 72.56 225.1 588.1 1341
obey Raoult's law with the vapor pressure of glycerol being
80 3.64* 18.29 70.25 217.8 568.2 1294
negligible. Using water as a reference liquid, construct
90 17.77 68.1 210.8 549.4 1250
Diihring lines for solutions having 0,20,40,60,80,100, 100 17.29* 66.1 204.3 531.6 1209
and 120 g of glycerol per 100 g of water. Plot these lines 110 64.2 199.4 514.8 1170
for reference water temperatures from 20 to 120°C; 1 cm of 120 192.1 498.9 1135
the plot should not represent more than 5 C. 130 186.5 483.7 1097
19.5. The data given below concern NaN03 140 181.2* 469.4 1064
solutions
150 455.7 1032
and were taken from the International Critical Tables.
160 442.6 1002
From it construct a Diihring line plot for the NaNC>3— H2O
170 430.1 973
system.
180 418.3* 945
19.6. Shown in the accompanying figure is a boiler
190 919
connected directly to a condenser. Steam supply is at
200 893
700,000 pascal, and cooling water is at 30°C average 210 869
temperature. The boiler has a surface area of 10 m and the 220 846
condenser area is 5 m2 . Overall coefficients of 230 824
3500 W/m 2 K for the boiler and 4500W/m 2 K for the con-
denser are realized. Supersaturated solution.
538 APPLICATIONS TO EQUIPMENT DESIGN

NaOH by weight with feed at 100°F. A forced-circulation concentrate 5000 kg/hr of a 10 percent sugar solution to 30
evaporator will be used with Dowtherm "A" (a eutectic of percent, the feed enters the second effect at 20 C. Satur-
diphenyl and diphenyl oxide) condensing in the external ated steam at 1 10°C is fed to the first effect, and the vapors
steam chest. The pressure in the evaporator will be from this effect are used to supply heat to the second
10.4 psi a with the Dowtherm "A" condensing at 20psia effect. The temperature in the final condenser will be 45°C.
(B.P. = 525°F). The circulation rate will be sufficient to The overall coefficients are estimated to be 2500W/m 2 K
2 K for the W/m
give an inside film coefficient of 800 Btu/hr ft 2 °F in the for the first effect and 2000 second effect.
external heater. A barometric jet condenser will be used The heating areas for the two effects are the same. The
with water supply at 90°F, 50 psig. Heater tubes will be specific heats of the solutions may be taken as constant and
nickel, 1 in, O.D. x 16 BWG, 12 ft long, with 273 of them in equal to 4000 J/kg K. Calculate:
What capacity
parallel. in Btu/hr will be required for the The temperature in each effect.
(a)

Dowtherm boiler? (b) The heating surface per effect.

Data for Dowtherm (c) Steam consumption in pounds per hour.

(d) Pounds of water evaporated per pound of steam.

,2 \ 1 /3

Temp.,°F yu, lb/ft hr Uw X, Btu/lb

Solve this problem, using X = 2300 x 10 3 J/kg for steam at
all

point
pressures and assuming the solutions have no boiling-
rises, by writing appropriate enthalpy balances,
450 0.970 0.001110 129
material balances, and rate equations and by solving them
475 0.930 0.001082 126
simultaneously by direct algebraic methods.
500 0.895 0.001062 123
525 0.870 0.001060 120 19.11. A basket type of double-effect evaporator
connected for backward feed is to be used to concentrate a
19.8. A batch evaporator 6 ft in diameter with a NaOH solution. Each evaporator body has a 2000 ft
2

steam coil 6 in. above the bottom surface is to be used to heating area. The caustic solution enters at 80° F and 5
concentrate NaOH solution from 2 weight percent to 15 weight percent NaOH. It is to be concentrated to 50 weight
weight percent by condensing 100-psig steam in the coil. percent. Previous operation indicates that overall coeffic-
The coil surface is 30 ft in area, and an overall heat-transfer ients of 400 Btu/hr ft
2 °F and 550 Btu/hr ft
2 °F may be
coefficient of 200 Btu/hr ft F is realized from coil to obtained in the first and second effects, respectively.
solution throughout the boiling period. With the pressure Saturated steam at 150 psia is available, and a vacuum of
over the solution constantly 1 psia, how long will the 2 in. Hg absolute may be obtained with existing ejectors.
evaporation take if the solution is initially 6 ft deep? What is the maximum production rate obtainable?
19.9. A NaNC>3 solution is to be con-
5 percent 19.12. A double-effect evaporator that has been idle
centrated to 25 percent NaNC>3 at a rate of 5000 kg of for some time is to be replaced in service evaporating
concentrate per hour in a standard vertical-tube evaporator. glycerol solution. The evaporator is piped for backward
The evaporator will be run at 25,000 pascal with saturated feed. In previous operation an absolute pressure of
steam at 350,000 pascal used as heating medium. The follow- 1 15 mm Hg was maintained on the second effect and super-
ing additional information is available concerning this opera- heated steam at 5 psig and 250° F was supplied to the first

tion: effect; these conditions will be maintained in the proposed

operation. The evaporator bodies are of the standard verti-
1. Uq, the overall heat-transfer coefficient based
cal-tube type with heating areas of 750 ft
2 each. Previous
upon outside surface area = 400 2
Btu/hr ft °F.
operation indicates overall coefficients of 300 Btu/hr ft
2 °F
2. Feed temperature = 100°F.
for each of the effects. If a 10 percent glycerol feed
3. Specific heat of feed = 0.90 cal/g°C.
entering at 60° F is to be evaporated to a 40 percent
4. Condensate from steam chest leaves at steam
glycerol product, what throughput can be maintained with-
saturation temperature.
out altering the operating conditions?
5. Effect of hydrostatic head of solution in the
19.13. It has been decided to withdraw 30,000 Ib/hr
evaporator on AT is negligible.
of low-pressure steam from the vapor line leaving the first
6. The latent heat of vaporization of the solution
effect of the evaporators of Problem 19.11. What will be
may be taken from steam tables at the operating pressure.
the pressure and temperature of this steam, and what
7. Heat losses amount to 5 percent of the total heat
transferred.
production rate will now be obtained from the evaporators?
19.14. A triple-effect evaporator system is to be used
How much heat-transfer surface is required? If 1 in. to concentrate glycerol-water solution from 10 to 40 per-
O.D. x 20 BWG tubes are used, what would be a reasonable cent glycerol by weight at a rate of 1000 kg/hr of concen-
length of tube and number of tubes in the steam chest? trate. Identical long-tube natural-circulation evaporators are
19.10. A double-effect evaporator is to be used to to be used, with overall heat transfer coefficients of 3500,
 SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 539

2500, and 2000 W/kg K expected for the first, second, and 4. AH of crystallization of NaCI = -1 250 cal/g mole.
when using forward feed. Feed
third effects, respectively,
Assume this to be independent of temperature and
enters at 25° C. Saturated steam at 150°C is used to heat
concentration.
the first effect. The condenser temperature is kept at 40°C.
Specify the evaporator sizes required in terms of steam- 19.17. A vapor recompression still is to be used to
chest surface area. generate fresh water for shipboard use. The feed to this unit
19.15. A triple-effect evaporator is to be used to is seawater at 60° F containing 3.5 percent solids, which
evaporate 20,000 b/hr of 10 percent I NaN03 solution to may be assumed to be NaCI. This is concentrated to 10
obtain a 50 percent NaNC>3 product. The system is arranged percent NaCI in the evaporator. The evaporator loses

for forward feed, which enters at 70°F. The steam supplied 100 Btu to the surroundings per pound of vapor generated,
to the first effect is saturated at 50 psig, and a barometric and the vapor is compressed to a pressure two times the
condenser— ejector system maintains the pressure in the last evaporator pressure before being returned to the steam
effect at 4 in. Hg absolute. The heat-exchange areas of all chest.
these effects are to be equal. For this system determine: (a) If the unit runs continuously with no added
(a) Steam economy, pounds of steam per pound of steam required and no excess steam available and with a
vapor formed. compressor operating adiabatically at 60 percent efficiency,
(b) Temperatures in each effect. determine the evaporator pressure.
(c) Heating surface required in each effect. (b) If the steam-chest area is 1000 ft
2 and an overall

The following additional information may be help- coefficient of 400 Btu/hr ft 2 °F is realized, what is the
ful: potable water production rate?
19.18. What phases are present and what is the
1. Overall coefficients for the three effects are 550,
relative quantity of each when 55 weight percent CaCl2
500, and 400 Btu/hr ft 2 °F, respectively.
solution is cooled from 180°F to:
2. Specific heats of NaNC>3 solutions can be con-
(a) 100°F.
sidered to be independent of temperature and to be given
(b) An enthalpy of -120 Btu/lb relative to pure H2
by the equation:
at32° F and CaCI 2 at0°F.
c P = 4.175 - 3.742x + 2.166x 2 (c) 60° F.

x= wt. fraction NaNC>3 (1—39 percent cone.) (d) product from (c) is heated to 300°F.
19.19. 10 1b of ice at 0°F is added to 10 1b of
Cp = specific heat, joules/g °C at 20°C and 1 atm pressure
CaCI 2 , 4H 2 also at 0°F. What would be the temperature
19.16. A double-effect, forward-feed evaporator
and phase conditions (species and amount) that would
system is to be used to crystallize NaCI from a 5 weight
result if this mix ultimately came to equilibrium adia-
percent solution originally at 80° F. Each evaporator body
batically?
has a 1000 ft heating area, and overall heat-transfer
2 °F are
19.20. A 20 weight percent solution of Na 2 SC>4 at
coefficients of 600 and 400 Btu/hr ft obtained in
200° F pumped continuously to a vacuum crystallizer
is
the first and second effects, respectively. Steam saturated at
from which the product magma is pumped at 60° F. What is
320° F is available, and a total pressure of 1 psia can be
the composition of this magma, and what percentage of
maintained on the second effect. It is desired to obtain
40 lb of NaCI crystals per 1000 lb of feed. Calculate the
Na 2 S04 m tne f eeQl ' s recovered as Na 2 SG*4'10H 2 O
crystals after this magma is centrifuged?
feed rate under these conditions.
19.21. A solution of 32.5 percent MgSC>4 originally
Data on NaCI Solutions at 150°F is to be crystallized in a vacuum adiabatic crystal-

1. Boiling-point rise for solutions at 1 atm

lizer to give a product containing 4000 Ib/hr of MgSGv
7H 2 crystals from 10,000 Ib/hr of feed. The solution
Cone, g NaCI/1 00 g water 5 10 20 30 40 boiling-point rise is estimated at 10°F.
BPR,°F 1 3 7 11 16.5 Determine the product temperature, pressure, and
2. Solubility of NaCI in H 2 0, g NaCI/100 g H2 weight ratio of mother liquor to crystalline product.
For data see Perry, J. H., Chemical Engineer's Hand-
Temp.,°F. 32 50 90 120 160 212
book, 5th ed., McGraw-Hill, New York (1973), pp. 3-94
Solubility 35.7 35.8 36.3 37.0 37.8 39.8
and 3-95.
3. Specific heat of NaCI solutions in Btu/lb °F 19.22. Using the data given in Perry [Chemical
[cp^f(T)] Engineer's Handbook, 5th ed., McGraw-Hill, New York
NaCI g/100g (1973), pp. 3-94 to 3-95], plot a solubility diagram for

water 10 20 30 40 solid crystal the Na 2 HPC>4— 2 system between the temperature limits
Specific heat 0.91 0.86 0.80 0.72 0.37 of 0°C to 150°C for a system pressure of 0.101 megapascal.
 540 APPLICATIONS TO EQUIPMENT DESIGN

Sketch the general form you would expect the enthalpy- What would be the screen analysis expected for the product
concentration diagram to take. if 10 lb of seed crystals are used per 100 lb of feed
19.23. Trisodium phosphate is to be recovered as solution?

Na3P04'12H 2 from a 35 weight percent solution 19.25. Crystals of CaCI 2 -6H 2 are to be obtained

originally 90°C atby cooling and seeding in a Swenson- from a solution of 35 weight percent CaCI 2 10 weight ,

Walker crystallizer. From 10,000 kg/hr of feed 3500 kg/hr percent inert soluble impurity, and 55 weight percent water
of product crystals in addition to the seed crystals are to be in an Oslo crystallizer. The solution is fed to the crystallizer
obtained. Seed crystals fed at a rate of 250 kg/hr have the at 100°F and receives 250 Btu/lb of feed from the external,
following size range; heater. Products are withdrawn from the crystallizer at
40°F.
Weight Fraction,
percent Size Range, mm (a) What are the products from the crystallizer?
(b) The magma is centrifuged to a moisture content
10 -0.500 + 0.250 of 0.1 lb of liquid per pound of CaCI 2 -6H 2 crystals and
20 -0.250 + 0.130 then dried in a conveyor drier. What is the purity of the
40 -0.130 + 0.065 final dried crystalline product?
30 -0.065 + 0.025 19.26. A solution containing 7 weight percent
Latent heat of crystallization of trisodium phosphate is Na 2 SC>4, 20 weight percent Na 2 CC>3, and 73 weight percent
6 H2 is to be crystallized to obtain pure Na C03* 10H O.
64.0 x 10 J/kg mole. Specific heat for the trisodium phos- 2 2
phate solution may be taken as 3350 J/kg K, and solubility This operation can be carried out in a Swenson-Walker
data may be found in Perry (see Problem 19.22). crystallizer for which cooling can be done by brine at 0°C,
(a) Estimate the product particle-size distribution. or it can be carried out in a batch vacuum crystallizer on
(b) To what temperature must the solution be which the ejector can pull enough vacuum to give a solution
cooled, and what will be the cooling duty in watts? boiling at 20°C. What is the maximum yield of pure
19.24. A solution of 28 weight percent Na2S04 is to Na 2 CO3 10H 2 O obtainable by each of these crystallizers,
#

be crystallized to yield Glauber's salt (Na 2 SGv 10H 2 O) in and which unit would you recommend?
a four-section Swenson-Walker crystallizer. It is fed to the 19.27. A 30 weight percent solution of MgSC>4 is fed
unit at 100°F and is cooled with 50°F water flowing at 100°F to a Swenson-Walker crystallizer where it is cooled
countercurrently. The crystallizer has a cooling surface of to 60°F. As it enters the crystallizer it is seeded with
2 /ft of length and operates with an overall coefficient of
3 ft MgS04*7H 2 crystals of the following size range:
2 °F across the cooling surface.
heat transfer of 20 Btu/hr ft

The water leaves at 75°F. Tyler Screen Size Weight Fraction

(a) What feed rate to the crystallizer should be used
to obtain 75 percent of the Na 2 SC>4 in the feed as -65+100 20
Na 2 SO 4 -10H 2 O crystals? -100+150 55
(b) The crystallizer is seeded with Na 2 SC>4-10H 2 O
-150 + 200 25
crystals analysing as follows:
The products from the crystallizer are wet-screened with
Tyler Sieve Size Mass Fraction
the—28 + 35 mesh material retained but with the undersize
-48 + 65 0.04 and oversize material recycled back into the feed solution
-65+ 100 0.18 before it enters the crystallizers. Determine the optimum
-100+150 0.46 rate of addition of seed to the crystallizer in terms of
-150 + 200 0.32 pounds of seed per pound of feed solution.
 Twenty

The Energy Balance in Flow Systems

In any system where a fluid flows past a solid The accumulation term may be positive or nega-
boundary, momentum is transferred from the fluid tive, depending upon the physical situation. For

to the boundary. This transfer of momentum steady-state operation, the accumulation becomes
results in the loss of system pressure that is of zero. In a like manner, the conservation of mass of
interestto the practical designer. The physical a particular component leads to a component
mechanism of turbulent transfer was examined in balance. A component may be produced or con-
Chapter 13. Here this mechanism will be applied to sumed chemical reaction that is part of the
in a
practical design situations. process. production may be
In such a case, the
The previous chapters in Part 3 have empha- considered as part of the mass input term of
sized the relationships among transfer rate, driving Equation 20.1, whereas consumption may be in-
force, and resistance; these relationships were cluded in the mass output term.
applied to the design of equipment for heat or Figure 20.1 portrays a generalized flow
mass transfer. In all practical situations, it is also system, showing the physical qualities that must be
necessary to knowenergy and material
the considered in writing material and energy balances.
balances for the process under consideration. For The generalized system can be applied to specific
example, the pressure loss of a fluid flowing practical situations such as pumping requirements
through a pipe is a function of the flow velocity; in flow through pipes, or cooling requirements in a
this velocity may be calculated by a material chemical reactor.
balance on the process. Similarly, the power re- Equation 20.1 can be applied to the flow
quired to pump a fluid through a complex piping system of Figure 20.1. At steady state, the
system can be calculated using both material and accumulation is zero, so that the mass flow rate
energy balances. into the system must equal the mass flow rate out.

v-[S\ v2 S 2
THE MATERIAL BALANCE w-\ = w2 = = G^S y
=G 2 S 2 (20.2)
Vy v2
In its simplest form, a material balance is nothing
more than an accounting of mass. The law of where
conservation of mass states that the total mass of v = average velocity, ft/hr (m/s)*
2
all substances taking part in a process remains S= cross-sectional flow area, ft (m 2 )

constant. Although there are exceptions to this law V= specific volume of material, ft /lb
3
(m 3 /kg)
for nuclear reactions and processes where the G = mass velocity, Ib/hr ft 2 (kg/s m 2 )
reacting materials are moving at or near the w = mass rate of flow, Ib/hr (kg/s)
velocity of light, for general engineering purposes
the law is inviolate. Equation 20.2 is known as the continuity
Semantically, this law may be written as equation

Mass input = mass output + mass accumulation

(20.1) * SI units given in parentheses.

541
542 APPLICATIONS TO EQUIPMENT DESIGN

© point within the system are constant with respect

Flow in | to time.
«>i, Ib/hr (_ Work done by mass 4. Mass rates of flow into and from the
dy, ft/sec system are constant.
Ei, Btu/lb (Wf ')ft-\b force /lb mass
P x , psia 5. The rates of addition of heat and pro-
^.cuft/lb ">2' l D / nr
duction of work are constant.
Si, sqft U2. ft/sec
E2 , Btu/lb In accordance with the principle of con-
Heat added
(Q) Btu/lb ,^\ ^2. Psia
^) V2 .cu ft/lb servation of energy, the total energy entering the
S2 . sqft system must equal the total energy leaving the
Flow
'2
system. The energies under consideration for this
out
Datum plane system are those carried by the fluid and that
'7777777777777777777777777777777777777777777777777777777777777777777777:
transferred between the fluid and surroundings.
Figure 20.1. Process flow system.
The energy carried by the fluid includes:
DEFINITION OF SYMBOLS
An internal energy (£), which is an in-
1.

trinsic property of the fluid. The system under

Typical Units
study is comprised of molecules in random motion.
English Systeme This motion produces the internal energy. The
Engineer- Inter- orientation and movement of molecules may be
Symbol Definition ing national likened to internal potential and kinetic energy.
2. A potential energy (zg/gc ) due to the
w Mass flow rate Ib/hr kg/s position of the fluid with respect to an arbitrary
V Velocity ft/hr m/s
datum plane.
E Internal energy Btu/lb J/kg
P Pressure lbf/in.
2 N/m 2
3. A kinetic energy (v 2 I2g c a) due to the
V Specific volume ft 3/lb m 3/kg motion of the fluid. The term a must be included
in the kinetic-energy term to account for the effect
S Cross-sectional flow area ft
2 m2
Q' Heat added to system Btu/lb J/kg of velocity distribution in the flow channel upon
Wf Work done by system ft \bfl\b J/kg the average kinetic energy. If little velocity
z Height above datum ft m gradient exists, such as in fully developed turbulent
flow, a approaches unity. For laminar flow, the
Subscript 1 refers to reference point in entering flow value of a is not unity and must be included in the
channel.
kinetic-energy term. Figure 20.2 shows the
Subscript 2 refers to reference point in leaving flow
relationship between a and the Reynolds number
channel.
for flow of fluids in pipes.

THE ENERGY BALANCE 4. An

To obtain a relationship between the various
types of energy involved in the flow system
depicted in Figure 20.1 requires several assump-
tions. Consider for simplicity that the material
flowing through the system is a fluid. The equation
that will be developed is not limited to fluids,
however, but is pertinent to any material involved
in a flow process.
The definition of a flow system
state conditions requires that:
energy of pressure (PV) carried by the
fluid as result of being introduced into the
a
system. In reality, the PV product is the work done
by the surroundings to push the fluid into the
system and a corresponding term recognizing work
on the surroundings at the discharge. This energy is
the force exerted by the fluid immediately behind
the entrance point times the distance through
which it acts. The distance through which the force
acts is equal to the specific volume of the material
in steady- divided by the cross-sectional area at the entrance
point. Thus, the work done is force times distance,

©
The or
1. fluid flowing into the system is

uniform in properties and velocity and these items

(PS) PV (20.3)
are nonvariant with time.
2. The fluid leaving the system is uniform in
properties and velocity and these items are non-
The energy transferred between the system
variant with time. The exit conditions will usually
and surroundings includes:
be different from entering conditions. 1. The heat (Q') absorbed by the flowing
3. The physical properties of the fluid at any material from the surroundings. Addition of heat
 THE ENERGY BALANCE IN FLOW SYSTEMS 543

-.0.9
3

/
^ :s
/

/
5 07
/
/

- :f /
/

/
15

0.4
::-" 4 ::~
10 1C"
Reynolds Number

Figure 20.2. Kinetic-energy correction factor as a function of Reynolds number.

to the fluid may or may not demand that its The summation of the terms on the left side of
temperature must rise. It is possible for the fluid to Equation 20.4 represents the energy transferred to
remain in isothermal flow while heat is added, for the system, and the terms on the right signify the
the added energy can find outlet in other forms. transfer of energy from the system to the sur-
The quantity of heat from the surroundings must roundings. Equation 20.4 is referred to as the total-
include the total net heat passing through the energy balance. Because it is based only upon the
boundary of the system. This excludes the heat concept of conservation of energy, its validity for
generated by friction because frictional heat must steady state is rigorous.
come from a dissipation of other forms of energy. The total-energy balance may also be written
By convention, the sign of Q' is positive if heat is in the differential form. This is frequently con-

transferred from the surroundings to the system. venient when manipulating mathematically the
The work (W /) done on the surroundings
2. terms of Equation 20.4, or when applying
by the flowing fluid as it moves through the Equation 20.4 to compressible fluids. Thus, for
apparatus. This term is frequently referred to as unit mass. Equation 20.4 becomes
shaft work because it is often associated with a
work quantity,
rotating or reciprocating shaft. This 9 \ vdv
i

dE + d(PV) +d\-z) + = 5Q' - 51V/ (20.5)

like must pass through the boundary of the
Q',
\9c ' We
system. For the fluid to do work on the surround-
ings would require a turbine or some other work-
The 8Q' and the 8W/ terms in Equation 20.5 are
producing apparatus connected from the system to
not exact differentials. Their evaluation depends
the surroundings. By convention, the sign of W/ is
upon the manner in which the heat and work have
positive if work is done by the fluid and trans-
been added to the system.
ferred to the surroundings.
The internal energy is an intrinsic thermo-
The balance of all the energies involved in the dynamic property of the system and is obtainable
flow system of Figure 20.1 may be written as only by difference or by establishing an arbitrary
basis from which E can be evaluated. Similarly, the
2
wv<\ g
— wQ PV term is also an intrinsic property of the system.
wEy +
,

+ WZy + wP-\ V-\ +

Furthermore, since £ and PV appear as sums on
?-9c& 9c
— o both sides of Equation 20.4, it is sometimes con-
= wE 2 + —~
WVo
+ wz 2 —Q + wP2 V2 + wW/ (20.4)
venient to combine this sum into a single quantity
that will also be an intrinsic property of the
29 c* 9c
544 APPLICATIONS TO EQUIPMENT DESIGN

system. Thus, the enthalpy (H) may be defined. atus because of fluid friction. The average linear
velocity of the inlet air 5 m/s. Assuming the air
is
H = E + PV (20.6)
to behave as a perfect gas and to have a value of
Then, Equation 20.4 may be written, for unit mass, c P = 1000 J/kg°C, calculate the net heat input to
as the air in J/s.
2 2

H y
+—- + Zy -Q + Q' = H +-^ + z -9 + W
v-\

2
v-i

2 f
.
'
Solution. The net heat input to the air can be
2agc gc 2gc ot gc obtained directly from Equation 20.4, although it
will be more convenient to use Equation 20.9 in
(20.7)
this case. There is no exchange of work between the
or air and the surroundings and so W/ is zero. The

iv 7 2 — 2 energy balance is written with each term expressed

(H2 -Hy)+l^-
7i
— \
)
+ (z 2 -Zy)- = Q'-W f
g '

as J/kg. Reference point 1 will be taken at the

\ 2gc a I
gc bottom of the exchanger and point 2 will be taken
(20.8) at the top.

Writing Equation 20.8 as a difference equation 2

v2 i-V
gives
(H2 -H,) + + (z2 -z )- = Q'
1
9c
AH + a(
\2g c a!
—) + a(z -) \ gc t
= Q'- W f
'
(20.9)
2<7c<*

From the statement of the problem,

where A is the increase in going from state 1 to

state 2. Ty =20°C = 293 K
Enthalpy, like internal energy, can be evalu- T2 = 90°C = 363 K
ated only as a difference. In general, the enthalpy
(z 2 — Zy) =4 m (datum plane at bottom inlet of
change of a system is a function of the specific
exchanger)
heat and the temperature change of the system
along a constant-pressure path or where the vy = 5 m/s
specific volume is very small, as is the case for P = 120 kN/m 2
y

solids and liquids. Therefore, the change in

P2 = 120- 10= 110 kN/m 2
enthalpy with temperature is

t In order to solve the energy balance for the net

2
c
AH = CpdT (20.10) heat input, the quantities (H2 — Hy ) and v2 must
be evaluated. The outlet velocity can be calculated
where directly from the perfect-gas law because the
velocity is directly proportional to the specific
AH = enthalpy increase, Btu/lb (or J/kg)
volume for a constant cross-sectional area {Sy =
c P = heat capacity at constant pressure,
S2 )
Btu/lb °F (or J/kg °C)
T2 ,T y
= final and initial temperatures, °F (or
°C) = '
72 1

In using Equations 20.4 through 20.9, each of (r7)(S)

the terms including heat and work must be /363\/120\
dimensional ly consistent. The reader is referred to 5.4 m/s
Appendix A for material concerning dimensions.

The quantity {H 2 - Hy) can be calculated from

Example
20.1. Air flowing at a constant mass
themean specific heat and temperature change of
rate of 0.1 kg/s is heated from 20° C to 90° C. The
the air. cP is independent of pressure for a perfect
heating is accomplished in a vertical heat exchanger
gas.
of constant cross-sectional area.
The exchanger is
4 m The pressure of the airstream entering at
long.
the bottom of the exchanger is 120 kN/m 2 and a (H 2 -Hy)=c P (T2 -Ty) = (1000) (363 - 293)
,

2
10 kN/m pressure drop occurs through the appar- = 70,000 J/kg
 THE ENERGY BALANCE IN FLOW SYSTEMS 545

Substituting into the energy equation (Equa- zontal pipe, if the velocity of a fluid increases, the
tion 20.1) gives pressure must decrease.
For any real fluid, work done against fluid

7 )-
6.4 4 friction will result in the dissipation of mechanical

Q' = 70,000
j
—+ —
7 )
energy, so that the right-hand side of Equation
20.11 will be less than the left-hand side. The
mechanical energy thus dissipated appears de-
graded to heat, which may be absorbed by the
fluid itself. Hence, not all of the work done by the
fluid is transferred to the surroundings. The energy
(4-0)m(9.8™)
lost because of friction, £F, appears in the fluid as
heat, so that the total heat absorbed by the fluid is
/ Nm\/1 kg m\
Q = Q' + 2F (20.12)

10.8 work done by the fluid is the

Similarly, the total
= 70,000 + + 39.2
a. sum work done to overcome fluid friction
of the
(2F) and the work transferred to the surroundings
a will be between 1.0 and 0.5. The entire kinetic-
energy term is small compared to the other terms
and can be called zero. Thus,
W f
'
= W- IF (20.13)

Q' = 70,039 J/kg

where
Q = heat absorbed by the fluid
and for 0.1 kg/s Q' = heat transferred from the surroundings
Q' = (0.1) (70,039) = 7004 J/s W = total work done by the fluid
W f = work transferred to the surroundings
'

In the calculations above, gc = 1 kg m^N is in- (shaft work)

cluded so that the units work out correctly, even IF = total fluid friction
though it has a numerical value of unity. Similarly,

the mechanical energy equivalent of heat,

Substituting Equation 20.12 into Equation
J=1Nm/J, is included. In SI, it is not really 20.4 and expressing the result as a difference
necessary to insert these constants, providing one equation gives
recalls that units of m 2 /s2 are equivalent to J/kg.
A£ + a(-^—) + Az — + A(PV) = Q - IF - W/
\2gc a' gc
Fluid Friction

In any real fuild, the viscosity tends to resist flow (20.14)

of the fluid, thereby resulting in a transfer of The first law of thermodynamics states that
momentum to the wall of the flow channel. An the increase in internal energy is equal to the heat
ideal fluid would be one where the viscosity is added to the system less the work done by the
zero, and there are no shearing stresses or momen- system
tum transport to the wall. If it is also assumed that
the fluid is in isothermal flow and Q' = and AE=Q-W (20.15)
W = 0, Equation 20.4 reduces to
The work done by the system is defined as
-2
v\
2

P^V, +Z-, —9 + - = P 2 V2 +z 2 -+-

9 ^2

9c 29 c <* 9c ?-9c* W= |
PdV (20.16)

(20.11)
Combining Equations 20.14, 20.15, and 20.16
Equation 20.1 1, which is commonly referred to as yields
the Bernoulli equation, relates the pressure energy,
potential energy, and kinetic energy of a perfect . v.

fluid. While it cannot be applied quantitatively to

+ A(PV) - I
'
PdV+ZF = -W/
\2g c (x/ gc
any real fluid, the equation is useful in drawing
qualitative conclusions. For example, in a hori- (20.17)
546 APPLICATIONS TO EQUIPMENT DESIGN

By calculus, where
rV, fV* fPi w= mass flow rate, lb/sec (kg/s)
A(PV)= d{PV) = PdV + VdP v = average fluid velocity, ft/sec (m/s)
•V, V, />,

p = fluid density, lb/ft 3 (kg/m 3 )

and therefore Equation 20.17 becomes 5= cross-sectional area of flow, ft 2 (m 2 )

Therefore, the theoretical power required becomes

a(-^—)+Az — + f
'
VdP+XF = -W f
'

Power = (W/)(w) (20.20)

\2gc oc/ gc "V,

(20.18) where power may be expressed as foot-pounds

force per second or J/s. Horsepower may be calcu-
Either Equation 20.4, 20.7, 20.14, 20.17, or lated by applying the proper conversion factor to
20.18 may be used to solve a flow problem. The Equation 20.20.
choice depends solely upon the desired result and
the information at hand.
Each term of Equation 20.18 has dimensions Example A pump
takes water at 60°
20.2.
of energy per unit mass of fluid flowing. In the from and delivers it to the bottom
a large reservoir

English engineering system, these dimensions yield
units of ft lbf/lb. In this system of dimensions with
its definition of the pound force (lb^) and the
pound mass (lb), there is a temptation to cancel
the two, thereby producing units of simply "feet"
for each term of Equation 20.18. Each term is then
referred to as a "head," expressed in feet of fluid
of an open elevated tank 25 ft above the reservoir
surface through a 3-in. I.D. pipe. The inlet to the
pump is located 10 ft below the water surface, and
the water level in the tank is constant at 160 ft
above the reservoir surface. The pump delivers
150gal/min. If the total loss of energy due to
friction in the piping system is 35 ft-lb^/lb, calcu-

flowing. For other systems of units, this tempta- late the horsepower required to do the pumping.

tion is less attractive, since it is obviously improper The pump-motor set has an overall efficiency of
to cancel dynes with grams, pounds force with 55 percent.
slugs, or Newtons with kilograms.
The term head is used in some disciplines and Solution. The best technique for solving
among some groups to identify each of the energy problems of this type is to draw a simple sketch
terms of Equation 20.18, even though the practice with appropriate data and label the reference
is based upon a misunderstanding of the English points about which the energy balance is to be
engineering system. Whether this practice will con- made. To demonstrate that the reference points
tinue as SI units become more widely used remains
to be seen.
The only term of Equation 20.18 where there
might be some justification to use head is the
160 ft
potential energy, where head does represent a Pump
difference in height. However, examination of the
potential energy term in Equation 20.18 shows 10 ft
25
«-GJ Tank
that use of head expressed as feet is tantamount to
canceling g with gc even though they have differ-
,

ent units. As discussed in Appendix A, the accel-
eration of gravity g and the force/mass proportion-
ality constant gc are not identical, even though
similar symbols are traditionally used to represent
them. They cannot be canceled with each other.
The power supplied to the fluid in a flow
Reservoir
may be placed anywhere, several different sets will
be used. The only requirement is that all the
energy between the points must be accounted for.
(a) Reference points 1 and 4 — the water sur-

system is equal to the energy supplied to a pound face of the reservoir and the water surface in the
of fluid multiplied by the mass flow rate of fluid, tank, respectively.
since power is the time rate of doing work. The From the statement of the problem,
mass flow rate of fluid is given by Q' = 0, £4 = £, zy = z 4 = 160 ft
, 0,
w= vpS (20.19) P = PA =
y
1 atm
 THE ENERGY BALANCE IN FLOW SYSTEMS 547

1 where P2 is the pressure at the pipe inlet, and v2 is

\A = Va = = 0.0161 3
ft /lb (incompres-
62.3 the fluid velocity at this level. Admittedly, the
above equation does not include the frictional loss
sible, isothermal fluid)
associated with the fluid entering the pipe.
2F = 35ft-lb,7lb Although the total frictional loss is known, the
portions of the I,F term that comprise the en-
V\ = Va = 0, since liquid levels remain approxi-
trance, and the exit frictional losses cannot be
mately constant
ascertained at this time. Instead, the 2F term will
For this problem and the information sought, be included as a total entity upon application of
the most convenient and appropriate energy the energy balance between points 2 and 3.
equation to use is Equation 20.14 or 20.18. Rearranging the above equation gives for P2
Using Equation 20.14,
z2 9
,"
2
2
n2 - p -
^
A£ + A (-^— ) + Az — + A(PV) = Q'- W/ - ZF /"1
V2 gc
,
2g c aV2
\2Qr-a' a,.

and
+ 0+ 160 + = 0- W/ - 35
2
Therefore, (-10) (6.8)
P2 = (14.7)(144)-
-W/ = 195ft-lb,/lb (1/62.3) (2)(32.2)(a)(1/62.3)

The minus sign connotes work being done on

P2 = 2120 + 623- —
45
the fluid by the pump. In traditional jargon, it a
could be said that the pump provides a head of
kinetic-energy correction factor (a) may
The
195 ft.
be evaluated by calculating the Reynolds number
The cross-sectional area of the pipe (S) is
and by use of Figure 20.2.
(tt0
2
)/4=(3.14/4)(3/12) 2 =0.049 ft
2
. The bulk
average fluid velocity in the pipe may be evaluated
from the volumetric flow rate, as follows: Dvp (3/1 2) (6.8) (62.3)
/V Re = = = 1 58,000
'8.34lbH 2 0\/ ft
3
\
n (1x0.000672)

V min/\60s/\ gal M 62.3 lb H 2 0/ At this Reynolds number a = 0.94, and

2
P 2 = 2695 \b f /ft absolute
" 6.8 ft/s
2
\0.049 ft /
The pressure at 3 is determined in the same
Thus, the actual found by
power required is manner by taking an energy balance between the
application of Equations 20.19 and 20.20 and the liquid surface of the storage tank and the exit of
use of the pump-motor efficiency. the 3-in. pipe. P3 is found to be equal to
2
10,521 lb,/ ft absolute.
ft-lb^\/_ ft _\/62.3lb^
Power
4
=195
lb
_
.049 ft
2
)(t)
Now the total energy balance
between 2 and 3
may be applied

x
(^U(iH -
3 - Mhp E2 + P2 V2 +z 2 - + -p- + Q' = £
9c 29c &
3 +P3 V3 +z 3 -
9c
(b) Reference points 2 and 3 the inlet and
V3'
exit of the 3-in. pipe, respectively.
+ W/ + XF
Application of the energy balance between 2g c oc
these two points necessitates the evaluation of the
For these reference points
pressures at these points. This evaluation in itself
requires an energy balance applied between a point Q' = 0, z 2 = 0, z = 35 ft, E2 = E3
3
at the reservoir surface and the entrance to the P2 = 2695,P3 = 10,521
3-in. pipe. Therefore,
v 2 = v 3 = 6.8 ft/s (because the pipe diameter is
the same at both reference points)
P:V,=P2 V2 +z 2 -+-
9 v2

9c 29 c a 2F= 35 ft-lb/r/lb
548 APPLICATIONS TO EQUIPMENT DESIGN

or volve the fluid state. The fluid is usually trans-

ferred from one part of the process to another
£2 + 2695( —
^62.3/
-) +
(6.8)

(2)(32.2)(a)
:

+ through pipe or tubing of circular cross section.

Most chemical engineering flow problems dictate
2
the use of closed ducts rather than open channels.
(6.8)
Pipe or tubing may be manufactured from any
= £, + 10,
),52l( ) + 35 + W/ + 35
V62.3/ (2)(32.2)(a available material of construction depending upon
the corrosive properties of the fluid being handled
and
and the flow pressure. Such materials as glass,
W/= 195.4 ft-lb/r/lb concrete, asbestos, steel, plastics, wood, and many
others are frequently used in construction, but
This is same theoretical work
essentially the
iron, steel, copper, and brass are the most common
Note that in this case P2 and P 2
as needed before.
pipe materials encountered in the process indus-
do not equal atmospheric pressure nor are they
tries. As mentioned above, the choice depends
equal. Also, the two velocities, although equal, are
upon the application.
6.8 ft/s instead of zero.
may be joined by several tech-
Pipe sections
(c) Reference points 2 and 4 (the inlet to the
niques. Threaded connections are most common
3-in. pipe and the liquid surface of the storage
for small pipe, and emphasis will be placed upon
tank, respectively).
this type. Pipe may be manufactured with different
Again, Q' = 0, z = 0, z 4 = 170
2 ft diameters and various wall thicknesses, so some

P2 = 27 A3 \b f /ft
2 standardization is necessary. A method of identify-
ing pipe sizes hasbeen established by the American
2
P4 = 14.7 x 144 = 2120 lb,/ft absolute National Standards Institute (ANSI). By con-
i/ 2 = 6.8 ft/s vention, pipe size and fittings are characterized in
terms of a nominal diameter and wall thickness.
i/
4 =
For steel pipe, nominal diameters may vary
Z£=35 between 1/8 in. and 30 in. The nominal diameter is
Since the velocities at the two reference neither an inside nor an outside diameter but for
points are not equal, the kinetic-energy correction 12 in. size and smaller is an approximation of the
factor will again have to be used. inside diameter. But, regardless of wall thickness,

As before, the energy balance becomes pipes of the same nominal diameter have the same
2
outside diameter. This permits interchange of
/ 1 \ (6.8)
fittings.
2743 + =2120
V62.3/ (2)(32.2)(0.94) (^)
\62.3 The wall thickness of pipe is indicated by a
schedule number, which is a function of internal
+ 170 + W/ + 35
pressure and allowable stress. Approximately,
or

W/ = -194.6
Again, tne same theoretical work requirement
results. Other reference points could be used, and
exactly the same result would be found, providing
all energies at each reference point are considered.
It may be concluded that although any set of
reference points may be used, wise selection of
reference points can simplify calculations.
PIPE, TUBING, AND FITTINGS
Before considering the 2£ term in Equation 20.14
in more detail, it is important to give some thought
to the means of containing the fluid in a given flow
system.
Chemical-process applications frequently
ft-lb,/lb
schedule number « 1000 —
s
where
P= internal working pressure, force/area
s = allowable stress, force/area
Ten schedule numbers are in use: 10, 20, 30, 40,
60, 80, 100, 120, 140, and 160. The pipe-wall
thickness increases with the schedule number. For
steel pipe, schedule 40 is the number corre-
sponding to "standard" pipe. In Appendix C-6 are
tabulated the dimensions and other useful data for
schedule 40 and 80 threaded steel pipe. For other
schedule numbers and other types of pipe, Refer-
ence 8 is recommended.
Some demarcation between pipe and tubing
in- can be set forth. Unlike pipe, tubing is sold on the
 THE ENERGY BALANCE IN FLOW SYSTEMS 549

basis of actual outside diameter. The wall thickness of fluid.The basic design of the valve dictates its
may be expressed as Birmingham wire gage (BWG). use either as a stopping device or as a flow-rate
Appendix C-7 tabulates various dimensions of control device. Two common valves are the gate
several sizes of copper tubing. valve and the globe valve.
The International Organization for Standard- The gate (Figure 20.5a) is a simple
valve
ization (ISO) has established standard metric pipe design possessing a disk that slides at right angles to
and tube sizes. For pipe, which the ISO calls "tube the flow. It is principally a valve used for stopping
suitable for screwing," the actual outside diameter flow because small lateral adjustments of the disk
corresponds to American standards, but the wall give extreme changes in the flow area.
thickness varies somewhat from standard American The globe valve (Fig. 20.5/)) because of its
schedule numbers, as an inspection of Appendix design is more suitable for flow control. In this
C-6 reveals. Because the outside diameters are valve, the fluid passes through an opening whose
equivalent, the pipe and fittings are inter- area is controlled by a disk placed somewhat
changeable. For tubing, there is no official parallel to the flow direction. Good control of flow
standard established by ANSI, although tubing can be achieved with a globe valve; however, the
manufacturers have a standard, as tabulated in pressure loss is higher than with a gate valve.
Appendix C-7a. A few values of the ISO metric The check valve permits flow in one direction
steel tube standards are tabulated in Appendix only. Plug valves are used for on-off service. Butter-
C-7/?, for comparison with the American standards, fly valves operate on the same principle as a
which are given in inches. It can be seen that in damper in a stove pipe. Diaphragm valves are useful
some cases outside diameter approximately agrees. for fluids that tend to clog the moving parts of
Wall thickness closely agrees with the BWG scales. other valves, such as very viscous liquids, slurries,
or corrosive liquids. In ball valves, the moving part
Fittings is a sphere through which a hole is drilled to pass

flow when it is aligned with the pipe. Reference 12

The term fitting refers to a piece that can do one
covers valve selection.
of the following:
In modern chemical processes, many flow
1. Join two pieces of pipe — for example, rates are controlled automatically using electrically
couplings, unions. or pneumatically driven valves. A pneumatically
2. Change pipeline direction — for example, driven globe-type valve is shown in Figure 20.6,
elbows, tees. and a butterfly control valve is shown in Figure
3. Change pipeline diameter — for example, 20.7.
reducers, bushings.
4. Terminate a pipeline — for example, plugs,
EVALUATION OF FLUID FRICTION
valves.
5. Join two streams to form a third — for As discussed in Chapter 13, fluid friction falls into
example, tees, wyes. two broad classifications — form friction and skin
6. Control flow — for example, valves. friction. In the generalflow system, both types of
friction are encountered, and
Figure 20.3 illustrates typical pipe fittings.
Fittings for steel pipe are usually made of cast iron
SF = F c + Ft (20.21!
or malleable iron and can be obtained in several
wall thicknesses. For pipe sizes greater than 2 in., where
screw fittings are frequently encountered.
less
Fs = skin friction
Large-size pipes can be joined by the same types of
Ff= form friction
fittings, but it is usual to install flanged fittings or
to employ welded connections. Figure 20.4 shows A physical flow system in which only one
flanged joints and flanged pipe fittings, respec-
type of fluid friction is encountered is impossible.
tively.
Perhaps the closest approach to this is in a long,
smooth horizontal pipe without change in flow
Valves direction. In this case, skin friction predominates.
A valve is also a fitting, but it has more important However, as soon as one considers commercial steel
uses than simply to connect pipe. Valves are used pipe in a flow system, including many pipe fittings
either to control flow rate or to shut off the flow and changes of direction, boundary-layer separa-
550 APPLICATIONS TO EQUIPMENT DESIGN

90* Elbow Plain Tee Reducer

90* Reducing Street Elbow Cap

Coupling Plug Union

Nipples Bushing

Figure 20.3. Typical screwed pipe fittings.

tion is likely to occur, and form drag will enter the friction-factor equation, however, was limited to
SFterm to a relatively large extent. smooth tubes. In the actual case, fluid friction is a
In the usual piping system in the process function of pipe roughness as well as pipe size,
industries, the fluid friction occurring within the fluid properties, and fluid velocity. A commercial
system may be calculated by the methods dis- pipe inherently rough as a result of the manu-
is

cussed in Chapter 13. The development of the facturing procedure. This roughness will cause

-a
 THE ENERGY BALANCE IN FLOW SYSTEMS 551

Screwed flange Slip-on flange Blind flange

Elbow Tee Lateral Taper reducer

Figure 20.4. Flanged fittings.

(a)

Figure 20.5. Sectional view of (a) gate and (b) globe valves.
A -Wheel E- Packing nut H- Bonnet K- Disk nut
B- Wheel nut F- Gland /- Disk holder /.-Body
D- Spindle G- Packing J- Disk
552 APPLICATIONS TO EQUIPMENT DESIGN

Figure 20.7. Air-pressure-actuated butterfly valve. Air

entering the cylindrical unit at the upper left pushes down
a piston, which motivates a mechanical linkage to produce
the rotating movement needed to open or close the butter-
fly valve shown on the right. (Courtesy Fisher Controls Co.)

straight pipe. This equivalent length is added to the

length of straight pipe to give the total equivalent
Figure 20.6. Pneumatically operated control system length. The advantage of this approach is

valve. The valve is shown in the open pos- that both pipe and fittings are expressed in terms
ition and the direction of flow is indicated. of a total equivalent length of pipe of the same
When the air pressure supplied to the volume relative roughness.
above the diaphragm at the top of the unit is Appendix C-1 shows the variation in rough-
low, the spring shown forces the diaphragm ness factors for various types of pipe with nominal
and the valve stem up, thereby opening the
pipe size. In Appendix C-2a are tabulated the
valve. On the other hand, an increase in air
equivalent lengths of fittings and of other fre-
pressure will force the diaphragm downward
quently encountered physical shapes. Equivalent
against the spring action, thereby closing the
lengths of contractions, expansions, or fittings may
valve. It is therefore possible to adjust the
flow rate by adjusting the air pressure. also be correlated by a resistance coefficient (AT).
(Courtesy Leslie Co.) This coefficient is defined as the number of
velocity heads lost because of the fitting. One
2
velocity head equals v /2gc Appendices C-2b and
.

boundary-layer separation and eddy formation

within the roughness pockets, resulting in energy C-2c give K for various expansions and contractions,
dissipation and additional frictional losses. Experi-
whereas Appendix C-2d correlates with equiva- K
lent length in terms of pipe diameters.
ments have confirmed this fact by showing
artficially roughened pipes to have a greater The friction factor as developed in Chapter 13
frictional pressure drop than the corresponding
should more appropriately be written for applica-
smooth pipe tion to piping systems as
(4).
In a like manner, the presence of fittings and
changes in flow direction will cause boundary-layer = (2o 22)
separation that results in form friction. Indeed, the
f
[£]-4Ms)]- -

frictionencountered in a fitting may be a signifi- 2g c

cant portion of the total pressure drop for short
flow systems. The friction losses through a fitting where
and through expansions and contractions are most f= the friction factor
easily obtained in terms of equivalent length of D = pipe diameter, ft
 THE ENERGY BALANCE IN FLOW SYSTEMS 553

v = average fluid velocity, ft/s No velocity term appears in the von Karman
2F = total fluid friction loss, ft lb^/lb number. If this number is plotted as a function of
T.L = total equivalent length of fittings, 1A/77 as in Appendix C-4, the velocity can be
expansions and contractions, and obtained directly for a known pressure drop.
straight pipe, ft
e/D = relative roughness factor
Example 20.3. Water is pumped from a reser-

Although the friction factor of Equation voir to a storage tank atop a building by means of a

20.22 is a function of pipe roughness in addition to centrifugal pump. There is a 200-ft difference in
Reynolds number, it is still a ratio of the total elevation between the two water surfaces. The inlet
pipe at the reservoir is 8.0 ft below the surface, and
momentum transferred to the momentum trans-
local conditions are such that level is substantially
ferred by the turbulent mechanism. Pipe rough-
ness, in this case, contributes to momentum trans-
constant. The storage tank is vented to the
fer through form drag.
atmosphere and the liquid level is maintained
constant. The inlet pipe to the storage tank is 6.0 ft
Equation 20.22 is presented graphically in
Appendix C-3. In the event that noncircular cross below the surface. The piping system consists of
sections used, the simple diameter term in
are
200 ft of 6-in. sch. -40 steel pipe containing two
90° elbows and one open gate valve from the
Equation 20.22 cannot be used. Instead, the
equivalent diameter as previously defined can be
reservoir to the pump; 6-in. pipe follows the pump
substituted, where
for a length of 75 ft after which the pipe size is
reduced to 4-in. sch.-40 steel pipe for 250 ft to the
cross-sectional area tank. The 4-in. pipe contains one gate valve and
13.20) three 90-degree elbows. It is desired to maintain a
0e q = 4
. wetted perimeter flow of water into the tank of 625 gal/min. Water
temperature is 68° F. If the pump-motor set has an
This equivalent diameter is subject to conditions
overall efficiency of 60 percent, what would the
stated in Chapter 13.
pumping costs be in dollars per day if electricity
Note that Equation 20.22 or its
to use
costs S0.08/kWhr?
graphical representation requires knowledge of the
average velocity to determine the frictional pres- Solution. For the flow system sketched in the
sure drop. There are times when it is necessary to accompanying illustration, point 1, the reservoir
solve for the quantity of fluid flowing to give a
J- Vent
specified pressure drop. Since in this case velocity
appears in both ordinate and abscissa, the solution
of Equation 20.22 would require successive
approximations until the stipulated conditions
were met. For instance, an assumption of a
velocity would have to be made so that a Reynolds
200 ft

number could be calculated. With this Reynolds 250 ft 4-in. sch 40

one gate valve
number and the appropriate roughness factor, a 3 - 90° elbows
friction factor could be obtained from Appendix
C-3. Equation 20.22 could then be solved for EF
using the f and v based upon assumption. If the
2F calculated meets the requirements, the
solution is completed. If not, the process must be 6-in sch 40
repeated. 200 ft 6-in. sch 40 x 4 reducer
2 - 90° elbows
A direct approach is possible. If the Reynolds
number is multipled by the y/T, the von Karman surface, and point 2, the liquid level in the storage
number results. Thus tank, will be used as reference points. The datum
plane will be at the reservoir surface.
Dvp Writing the energy balance in the form of
/V K a =
Equation 20.18, gives

(I/ 2 \ Q P 2

+Az-+ j
VdP+ZF = -W/
(20.23) 2g c (xl gc p,
554 APPLICATIONS TO EQUIPMENT DESIGN

The flow is isothermal and water is incom- The friction factor (f) may be determined from
pressible so that V\ = V2 For the reference points
. Appendix C-3 using /V Re and elD.
/•/»,

chosen, Py=P2 and / VdP=Q. Since both f=0.017

= v2 = Thus, the
So
remain constant,
liquid levels v-\ 0.
energy balance becomes 2F for the 6-in. pipe = 2F 6in .

-Wf ' = Az — + XF (0.017)(6.92) (339)

2

9c
(2)(32.2)(0.506)
To solve for W/, only the potential energy
= 8.5 ft-lbf/lb
and frictional losses need be evaluated. In this case,
2F be equal to the friction occurring in both
will The A-in. pipe. Assuming steady state, by
sizes of pipe includingall fittings, expansions, and material balance the linear velocity of the water in
contractions. For both pipe sizes, Equation 20.22 the 4-in pipe will be inversely proportional to the
must be employed. cross-sectional areas of the 4-in. and 6-in. pipes.
The 6-in. pipe. For this size pipe, the Thus,
average velocity is
/S6in \ 28.89
3 2 2
/ galwminw ft \/144 in. /ft \
= 625
^'"•
( ^n)k)(ai334 ^)(^^7) So
= 6.92 ft/s
2
Dvp (15.7) (62.3)
The 28.89 in. is the cross-sectional area of a /V Re = = (4.026/12) = 489,000
M
fi nnnng
x 0.000672)
., ol
6-in. sch.-40 pipe obtained from Appendix C-6. (1

The /V Re for flow through this pipe can be

The total equivalent length of 4-in. pipe is (from
calculated:
Appendix C-2)
D = 6.065 in. = 0.506 ft (Appendix C-6)
Sudden reduction (6-in. to 4-in. pipe)
M = 1.0 centipoise (Appendix D-9)
(K = 0.24) 5.6 ft
3
p = 62.3 lb/ft (Appendix D-12) Straight pipe 250.0

/V Re = —
Dvp
- = (6.065/12)
(6.92M62.3)
x 0.000672)
90° elbows, L/D = 30, L = 1 ft each
Gate valve (fully opened) L/D = 13
30.0
4.4
(1
Sudden enlargement (pipe into tank)
= 325,000 (AC=1.0) 20.0

To use Equation 20.22 requires the total Total 310.0 ft

equivalent length of straight pipe. From Appendix

Relative roughness factor for 4-in. pipe, e/D =
C-2,
0.00044. Therefore, f= 0.0175 and
Entrance from reservoir to pipe;
2
K = 0.78; L = 27 ft fi/ (SZ.) (0.0175)(15.7) 2 (310)
90° elbows, L/D = 30, 4in ~
2gc D 2(32.2X4.026/12)
L = 15.2 ft each 30.4 ft
Gate valve (fully opened) L/D =13 = 62 ft-lb/r/lb
6.5 ft
Total 63.9 ft Thus the total friction loss in both pipes, including
Thus the total equivalent length of fittings, expansions, and contractions, is 2F =
6-in. pipe is
8.5 + 62 = 70.5 ft-lb^/lb.
2Z. 6in = 63.9
.
+ 200 + 75 = 339 ft Returning to the energy balance as written for
From Appendix C-1, the relative roughness factor the problem,
'
2
(e/D) for 6-in. commercial steel pipe is 0.0003.
Rearranging Equation 20.22 to solve for XF, -Wf = (z 2 - z\ )
— + 2F = 200 / 32.2 ft/s

9c I 00 fHb
1 32.2 -„
2 2
/v (ZL) \ lb/ s ,

ZF =
2gc D + 70.5 = 270.3 ft-lb^/lb
 THE ENERGY BALANCE IN FLOW SYSTEMS 555

This work term may be converted to power as For smooth tubing e =

previously described
D= 0.05 m
ft-lbA/ (hp)(min) \/ 625gal
\ /625 gal \ S= 0.00196 m2
-w; = 270.5
( lb 33,000 ft-lb ft\ min /
The desired flow rate may be calculated from
/ 746 kWhr
8.34 II \/0.746kWhr\
8.34lb the average velocity which is most easily obtained
31.7kWhr/hr from Appendix C-4.
^ gal )\ hphr /
From Equation 20.23,
With 60 percent efficiency, the actual work
31.7
Dp /2g c D(XF)
— — = 53.9
required = kWhr/hr. 7V XL
Pumping costs per day = (53.9)(0.08)(24) = (0.05) (820) /(2)(1)(0.05)(170.7)
$103.5 per day. 10- 3
(0.9 x ) 40
Example 20.4. A pump discharges ethanol at = 29,800
38°C into a smooth tube of 50 mm I.D. at a gage Then, from Appendix C-4,
pressure of 280kN/m 2 This pipe is 40 m long .

(total equivalent length) and supplies ethanol at v

2
= 7.9
140kN/m to a reactor. What is the volumetric ^2g c D(XF)/XL
flow rate of ethanol to the reactor? There is
Solving for v gives
negligible change in elevation between the pump
and reactor. v = 5.2 m/s

Solution. A simple sketch is, as usual, war- and the volumetric flow rate is

(5.2)(0.00196) = 0.010 m 3 /s
50 mm I.D.

£7Pump COMPRESSIBLE FLUIDS

For operations involving the flow of compressible
fluids, the solution of the energy balance is more
complicated because of the variation of specific
volume with pressure changes. The integrated
Reactor equations previously presented do not apply.
ranted. According to the statement of the problem Instead, the differential form of the energy balance
and for the reference points chosen must be used and integrated over the terminals of
Zy = Z2 = the system.
2 Thus, Equation 20.18 may be written as
P, = 280,000 N/m
P2 = 140,000 N/m 2 — dz+f'
L - + /: VdP+(ZF) =-Wf
3 9c a 9c J
Pi = P2 = 1000 x 0.82 = 820 kg/m
p,

= v2 = (20.24)
v-\ ?

M = 0.9 cp = 0.9 x 10~ Ns/m

3 2 The friction term in this case will be equal to

Wf = ' l 2 fv
2

(SF) = f d(!L] (20.25)

Q' = '*-, zg c D

The energy balance therefore becomes Therefore, substituting Equation 20.25 into Equa-
tion 20.24 gives
P*V, = P2 V2 + ZF
e
v7 v dv r
z2
q ./
P-.
Since the specific volumes are equal for an incom-
pressible fluid, ?c«

-= 280,000-140,000 = 170.7 —
Nm 7
(P,-P
— 2 ) fv
ZF = :
'
d(2L) =-W/ (20.26)
p 820 kg 2g c D
556 APPLICATIONS TO EQUIPMENT DESIGN

Actually, Equation 20.26 is appropriate for Example 20.5. It is desired to have

incompressible fluids as well as for gases but is 150 x 10 6 ft
3
of methane, measured at 60° F and
most frequently encountered in gas flow. The 760 mm Hg, per 24-hr day delivered to a synthesis
solution to Equation 20.26, and in particular plant located 2 miles from the compressor station.
JVdP, requires a knowledge of the PVT relation- Delivery conditions are to be 60° F and 10 psig at
ship for the gas under study. Often this relation- the plant. If the gas leaves the compressor at a

ship may be obtained by use of the perfect-gas law. temperature of 75° F, what must its pressure be at
In any case, the PVT properties may be evaluated this point? Assume negligible change in elevation
using the compressibility factor in conjunction between line terminals. The pipe is steel and is

with the gas law. 24 in. I.D.

2
Since no analytical expression exists for f\/ as
a function of L, the friction term is more difficult
Solution. If the reference points for the
to evaluate. A series of stepwise calculations
energy balance are chosen so as to have point 1 at
involving short increments of length using average
the outlet of the compressor and point 2 in the
values of v and V over that length may be used
pipe at the synthesis plant, then — W f = 0. Also by '

with little error.

statement of the problem dz = 0. For this case.
If steady state is assumed, the mass velocity
Equation 20.30 will be used.
will be a constant for a pipe of uniform cross
section. Or 2 G 2 VdV .2 .2 fG 2 V2
j
+ I VdP+ j
^——dL=0
D
(a)
i
gc oc -S i 2g c
G = vp = 10.27)
V To determine the character of flow, the
Reynolds number at the metered conditions is
where
calculated.
G = mass velocity, lb/sec ft 2 (kg/s m 2 )
_ 150,000,000 b 1 4i 492
v = average linear velocity, ft/sec (m/s) IV
\iJL\i \

3 3 D '359'' 520 '

p = fluid density, lb/ft (kg/m )

V= specific volume, ft 3 /lb (m 3 /kg) = 73.0 Ib/s (constant)

Solving Equation 20.27 for v gives Viscosity of methane at 60° F = 0.011 cp

(Appendix D-9)
v=GV (20.28)

and Cross-sectional area of 24-in. pipe = — (2)

2
4
dv=GdV 2
= 3.14 ft

Replacing v in Equation
expressions yields an expression in terms of the
20.26 by the above
G
73.0
= -
3.14
—= 23.3 Ib/s ft
2

specific volume.
DG (2X23.3)
2G 2 VdV -2 g 2 (A/ Re ) 2 =
4
j
+ -dz+ VdP H (0.011 x 6.72 x 10- )

•1 9c& "i 9c -i
= 6.30x 10 6
,2 fG 2 V2
+ dL = -W f
'

c / 0.011
2g c D = (6.3 x 10 6 5.78x 10 6
1
•1 =
(20.30)
(/V Re )i
)(^^|
Analysis of a particular gas-flow problem will (viscosity of CH 4 at 75° F = 0.012 cp)
Relative roughness factor (e/D) is 0.00007.
best indicate the appropriate method needed in
solving Equation 20.26 or Equation 20.30. For Therefore,
instance, it might be that for nearly isothermal
f2 = 0.0128 and fy =0.0130.
conditions the gas viscosity will vary little, and, if
flow is fully turbulent, f will be substantially These friction factors are so close that an
constant, so that the friction term is easily cal- average value of f= 0.0129 may be used with
culated. negligible error. To eliminate V2 in the friction
 THE ENERGY BALANCE IN FLOW SYSTEMS 557

term, Equation (a) is divided by V2 . in Equation (d)

2
r 2 G 2 dV r
2 dP r 2fG 2 dL G P, (23.3)
:

58.5
2
+ -+ _ _ =0 (b) In lnr-^= 14.5 lb/ft
V
I

2g c D ugc P2 (0.98X32.2) 24.7

1
g c Va
By the magnitude of /V Re a = 0.98 from ,
Therefore, the assumption made above that the
Figure 20.2. Assume that the perfect-gas law may velocity term was negligible contributed an error of
be used to describe the PVT relationships for 0.16 percent in the final answer.
methane. As shown byExample 20.5, the friction
1
factor approaches constancy as velocity increases,
Thus PV= - RT and the only variable in the fourth term of
M Equation 20.26 is the length itself. At higher
where M= molecular weight velocities, the size of the first term, the momen-
tum-increase term, in Equation 20.26 grows
RT rapidly so that it and the pressure-drop term, the
V=
PM third term in Equation 20.26, dominate. At very
high velocities the equation for flow in a constant-
Substitution of l/from above into the second term
cross-section duct becomes.
of Equation (b) gives
2GVdU r
2
r 2G 2 dV ,2MPdP r 2fG 2 dL + VdP^O (20.26a)
+ /
+ = gc u 1

•'i agc V j iRT J

1 2gc D
Thermodynamic arguments can be used to
Integrating Equation (c) gives show that a maximum exists in the velocity of a
fluid in a constant-cross-section duct. From the
G2 V
—
\A 2 M 2
fG 2 {ZL)
In + IP* PY physics of fluids it develops that flow velocity
ocg c 2RT
V, a 2g c D cannot exceed the velocity of an infinitesimal
The use of 7"av is not in great error in this case. For pressure wave in the fluid itself. It should be
an ideal gas, intuitively acceptable to expect this maximum to
be the velocity of transmission of a sound wave in
M fG 2
{I.L)
the medium. Proof of this beyond the scope of
,n^ + (P 2
2
~P^) + ~2g~ = is

ug c 2RTa c D~ this text.

For the high velocity attained under large
(d)
pressure gradients in a constant-cross-section duct,
Assume as an initial trial the first term of Equation this sonic, or acoustic, velocity is attained only at
(d) to be negligible, then the discharge end. It will also be attained at the
discharge of a convergent nozzle, or a frictionless
P2 2_ p ,2 = _*^l<^\ restriction (orifice). Velocities above the acoustic,
QrD \ M /
or critical, velocity may only be attained if a

or stream at or above sonic velocity is introduced into

a divergent conduit. The pressure drop to accom-
fG 2 CLL )(RTa ^
V =
gc D I M )
+ Po plish sonic velocity is a function of the physical
properties of the gas — typically, about half of the
Substituting the known values from the statement
upstream pressure for common permanent gases.

of the problem and physical data,

The velocity attained the velocity of sound in the
is

gas at the conditions of the gas at that point in the

2 system. The momentum term in Equation 20.26
2 _ (0.0129)(23.3) (2x5280) / 1544x_528\
P, = becomes significant at about one-fourth of the
32.2x2.0 \ 16
velocity of sound; above this velocity, simple
+ [(10+ 14.7)144] 2 friction equations are inadequate.
For a perfect gas, the velocity of sound is a
P y
2 = 71.0 x 10 6
function of temperature alone
= 8440 2
/»! lb/ft = 58.5 psia
=
v* g c RT (20.266)
Check the validity of neglecting the velocity term Cv
558 APPLICATIONS TO EQUIPMENT DESIGN

where or
vs = sonic velocity in an ideal gas
Cp , Cv = heat capacity for a constant pressure AP = - —
Az g
V
- = - pAz—=
9
p{z y
- Z2 )
—9
process and for a constant volume 9c 9c 9c
process, respectively
(20.32)
This velocity limit is also the velocity that can be
obtained in the throat of a converging-diverging Thus, the pressure difference can be expressed
nozzle or that can be realized in an opening such as in terms of a height of a vertical column of fluid of
would be made by a ruptured pressure disk. density p.
The field of compressible flow is enormous, Equation 20.32 may be applied to the U-tube
with vital applications in aerodynamic design, manometer. This instrument is shown schemati-
design of rotating machinery, and in combustion cally in Figure 20.8. Fluid H cannot be identical to,
and propulsion systems. Chemical engineers meet it or miscible with, the flowing fluid. It is sometimes
in dealing with safety relief systems, nozzle per- desirable to use a fluid that is heavier than the
formance studies, throttling processes, and rapid flowing fluid as is shown and sometimes one that is
reaction studies, to name a few. A reasonable lighter, in which case the U-tube is inverted. In

mastery of the field requires a strong background Figure 20.8, it is desired to measure the difference
in thermodynamics,
classical hydrodynamics, between P, and P2 The pressure difference will be
.

advanced mathematics, and physical chemistry. related to the difference in height between points a
References 13 and 14 are typical of a large number and c, which is called the manometer reading, R.
of excellent treatments on the basics of com- The pressure at point b can be determined by
pressible flow. Equation 20.32; that is,

Pb =P2 + (z d - c )p L + {z c - zb )p H
FLUID METERS
(20.33)
The metering of fluids is an important application
Since point b is stationary, the pressure as
of the energy balance. Basically, most flow meters
indicated by Equation 20.33 must be balanced by
are designed to cause a pressure drop that can be
measured and related to the rate of flow. This a pressure acting on it in an opposite direction. Or
pressure drop can be brought about by changes in Pb =P: +(z e -za )p L (20.34)
kinetic energy, by skin friction, or by form
friction. Some types of meters emphasize one or a
Equating Pb as determined from Equations 20.33
combination of these mechanisms. The pitot tube, and 20.34 gives
for example, is a metering device based upon a ^2 + Ud ~ *c)PL + (*c - Zb )PH = Pi + (^e ~ Za )p L
pressure difference resulting almost solely from
(20.35)
kinetic-energy changes. On the other hand, the
pressure change across an orifice plate is due But ze = zd and za =zb ; therefore,
mostly to kinetic-energy change and form friction.
In any event, a general equation resulting from an
P2~P\ = iUe -za -izd -zc )]p L -p H (zc -zb
) )

energy balance may be derived to relate flow and P2 -Py= {Zc - Za )p L - (Z c - Za )p H

pressure drop.

Manometers Pi P-2

Since most fluid meters tend to cause a pressure \ I.

difference across the metering section, some Fluid L
simple, easy-to-use, pressure-measuring device must
be employed to indicate this difference. One of the Fluid L- T
simplest pressure- measuring instruments is the
Fluid//
U-tube manometer.
If a duct is filled with an incompressible fluid

and no flow occurs, the energy equation, Equation Fluid H-

20.18, may be written as

Az — + V(P -/M=0 2 (20.31)

Figure 20.8. Simple manometer.
 THE ENERGY BALANCE IN FLOW SYSTEMS 559

or the length of the manometer would become

-AP=P, -P2 = (zc -za )(p H -p L
In Equation 20.36, the term (z c — z a would
correspond to the manometer reading as indicated
by an appropriate scale. This balancing of forces to
determine the pressure difference read by a man-
ometer is general. It may be used regardless of the
unwieldy. Various mechanical and electrical
(20.36)
devices are also used to measure pressure, including
)
higher pressures. These include the Bourdon tube
or bellows element where the pressure produces a
displacement in a metallic element that can be
measured and related to the pressure. Strain gages
are also used, where the stretching of a conducting

manometer orientation, presence of intermediate
fluids, or any other variation in manometer design.
Note that the manometer reading is multiplied by
the difference in density of the two fluids to get
the pressure drop.
As indicated above, variations of the man-
ometer shown in Figure 20.8 may be made. It may
be inverted so as to use a metering fluid of lighter
density than the flowing fluid. For small readings,
the tube may be inclined at some angle to the
horizontal so as to magnify the linear displacement
of fluid. Another technique used to magnify small
wire changes its electrical resistance, which can be
measured and related to the pressure.
The General Meter Equation
Consider a flow system containing a fluid meter
about which an energy balance is to be made. This
meter produces a change in the kinetic energy of
the flowing fluid that can be sensed as a pressure
difference between a point 1 just upstream of the
meter and a point 2 at the meter. Points 1 and 2
represent locations about which an energy balance
is to be made. The two points are separated by a

readings is to have one leg of the manometer of negligible distance compared to the total length of
significantly larger cross section than the other and the system. For this case, Az, W/, and Q' are zero.
to incline the smaller-diameter tube. This pro- Thus, Equation 20.18 becomes, for a fluid of
cedure will allow the pressure to be read in only substantially constant density,
the smaller-diameter tube since the meniscus level - 2 - 2
in the larger tube will remain substantially con-
stant. A two-fluid manometer as illustrated in
V(P2 -Pi)+^r-a
2pc
— + ZF=0 (20.37)

Figure 20.9 is often used to indicate low pressure

Equation 20.37 may be rearranged to give
differences.
In the two-fluid manometer, the liquid levels
v2 2 -vf = -2gc a{V(P2 -Py + 2F]
in the top reservoirs remain essentially constant,
but the reading can be multiplied a great deal if (20.38)
fluid B and fluid C have nearly the same density.
From the continuity equation,
Whatever the type of manometer used, the prin-
ciples behind Equation 20.32 can be applied to
v^P^S^ - v 2 p 2 S 2 (20.39)
give an equation that can be used to meter the
pressure drop across a test section of a duct. Since p^ = p 2l
Manometers are used to measure relatively v } S,
low pressure differences, since at higher pressures i/
2 = (20.39a)

Therefore, substituting Equation 20.39a into Equa-

Fluid A Fluid A tion 20.38 results in

vS (|V-l) =-2gc [\AP2 -Pi) + 2r\a

Fluid B Fluid B

Fluid C fT =

Fluid C

Figure 20.9. Two-fluid manometer. (20.40)

560 APPLICATIONS TO EQUIPMENT DESIGN

The bracketed term within the square root
represents a decrease attributable to
pressure
changes in kinetic energy between points 1 and 2.
The friction term will include losses due to form
friction and skin friction, as has been shown in
Equation 20.21. The manometer connecting points
1 and 2 will give a reading that is indicative of the
pressure difference between these two points. This
pressure difference is produced by two effects: the
change in kinetic energy caused by a change in
velocity because of a change in flow area, and
permanent pressure loss due to skin and form
friction. The manometer indicates only the total
pressure difference between these points and does
not differentiate between the two effects.
Equation 20.40 is referred to as the general meter
equation. It could be written in terms of v2 or in
terms of the mass rate of flow by the appropriate
use of the continuity equation. The general
equation will now be applied to specific types of
flow meters.
The Orifice Meter
An orifice meter is an extremely simple apparatus;
it normally consists of a flat plate with a centrally
relative to the orifice plate is a function of the
fluid velocity as well as the relative diameters of
the orifice and the duct. The position of the
downstream pressure tap should approximate the
vena contracta to ensure a maximum reading on
the pressure-difference indicator.
The application of Equation 20.40 to the
orifice permits calculation of the fluid
plate
velocity and hence of the mass-flow rate through
the duct. Some modification of this equation is
necessary to make it more useful. Consider again
the bracketed term of Equation 20.40. The term
(—AP/p) represents the total pressure difference
between points 1 and 2, and the SFterm includes
all between the same two points. The
fluid friction
difference between these terms represents the
kinetic-energy change between points 1 and 2 as
indicated by Equation 20.38. This difference can
therefore be expressed as a fraction of the total
pressure difference (— AP), or
AP\
(20.41
P
where C^ 2
is a proportionality factor always ess ...

drilled hole beveled to a sharp edge. The drilled

than one. Equations 20.40 and 20.41 may be
plate is inserted perpendicularly to the flow
combined to give
direction and the fluid passes through the hole. A
schematic representation of an orifice meter along
with the fluid stream lines is shown in Figure
20.10. v, = C,a
Mff) (20.42)
By the nature of the streamlines, it is appar-
ent that boundary-layer separation occurs on the e-<)
downstream side of the orifice plate. Conse-
quently, the pressure loss from form friction is Although the pressure tap at point 2 can be
likely to be considerable, and, in fact, the orifice approximately located at the vena contracta, it is
plate is a meter that maximizes form drag. The quite difficult to know with any reliability the
flow lines actually reach a minimum cross-sectional cross-sectional area of the vena contracta. How-
area one half to two duct diameters downstream ever, by geometry,
from the plate. This point is known as the vena
contracta. The location of the vena contracta Sj — C901
2°0 (20.43)
where
C2 = a constant of geometry
S = cross-sectional area of the orifice hole,
2
ft

Therefore, combining Equation 20.43 with Equa-

tion 20.42 gives
2 At vena contracta
-Di—ZZ
D
0.5 x

i/! =da
M=r) (20.44)
%=0
- 1

Figure 20.10. Sharp-edged orifice meter. C2 Sq

 THE ENERGY BALANCE IN FLOW SYSTEMS 561

Now, define a coefficient C such that

H=f) = Cia
*«ra
-1
C2 S
(20.45)

This new coefficient C is referred to as the orifice

coefficient, and the final orifice equation may be
written as
2 3
10 20 40 60 10 10
Reynolds number

Figure 20.11. Meter coefficients for sharp-edged orifices

vy =C (20.46) and for rotameter (10). (Orifice Curves by permission of
instruments, Copyright © 1973. Rotameter curves, cour-
tesy of Fischer and Porter Co.)

D v p
Bear in mind that the orifice coefficient is a For orifice: A/r 6 =
M
complex function of pressure drop and geometry.
It is sometimes convenient to use mass flow
D G qV 2 P
For rotameter: /Vr =
6
instead of Equation 20.46 is directly
velocity. P
converted into the mass form by applying the
D = orifice diameter
continuity equation. Since v-y = w/S-\ p
D eq = equivalent diameter of annular opening between tube
and float.

orifices, it should not be used. If the orifice is

w = pC S
Either Equation 20.46 or Equation 20.47 may be
used to calculate the flow rate using an orifice
meter if the pipe dimensions, orifice dimensions,
the pressure drop across the orifice, and the orifice
coefficient are known. The orifice meter measures
average velocity.
The orifice coefficient is most appropriately
correlated as a function of the Reynolds
through the orifice and the ratio of the diameters
of the orifice to the pipe. This practice seems
justified in that the orifice equation includes the
diameter ratio and the total pressure drop across
the orifice, part of which is due to fluid friction.
This correlation is shown in Figure 20.11 for
sharp-edged orifices. When the Reynolds number
through the orifice exceeds 30,000, the orifice
coefficient is approximately constant at a value of
0.61 independent of the diameter ratio. Figure
20.11 is correct only if the pressure taps are
located as shown in Figure 20.10; for flange taps
is useable. For other locations or for round-edged
(20.47) located in the piping system in such a way as to
alter the flowthrough the orifice, the orifice
lines
coefficient will be different from that predicted
from Figure 20.11. It is recommended that the
orifice should be located at least 50 pipe diameters
downstream and 10 pipe diameters upstream from
any disturbances (3). If these undisturbed lengths
of piping system are not available, straightening
vanes may be used upstream from the orifice.
The orifice meter, although simple, has one
serious disadvantage in that a large percentage of
number the pressure drop across the orifice is unrecover-
able. The fluid velocity is increased at the orifice
opening without much loss of energy, but, as the
fluid issues from the opening and starts to slow
down, much of the excess kinetic energy is lost.
The permanent pressure loss is a function of the
ratio of the orifice and pipe diameter, as shown in
Figure 20.12.
When compressible fluids are metered by an
orifice, the pressure drop across the orifice may be
a significant fraction of the total pressure of the
system. In this case, the specific volume of the
it fluid and also the fluid velocity will change
appreciably between the two pressure taps. For this
562 APPLICATIONS TO EQUIPMENT DESIGN

2
i.O
2
/3 5\
2
Orifice area: irD /A = 0.785 --] = 0.0667 ft
^ | 0.8
1

2
06 Pipe area = 0.139 ft

Linear velocity through the orifice:

c o =

0.2 0.4 0.6

(300
V
— )
min/\
(0.1338 —\ (^zM-r^zz)
gal/\60/\0.0667/
= 10.0 ft/s

0.8 1.0
:• ;e Jiarrerer
Ratio.
t

pipe diameter
Orifice Reynolds number:

Figure 20.12. Permanent energy loss in D v p (3.5/12)(10.0)(62.3x0.87)

sharp-edged orifice (3). (By permission of Wr.)o =
(15 x 6.72 x 10 -4i
ASME, Copyright© 1971.)
= 15,650

situation, the orifice equations as

are inappropriate. Instead, the energy balance
developed earlier
must —= £>!
3.5

5.047
= 0.695
be written in the differential form and integrated
between the two pressure taps, accounting for
From Figure 20.1 1, Co = 0.635. The linear velocity
variation in specific volume and velocity. This will
of the oil in the pipe can be evaluated from the
entail the use of some thermodynamic equation of
continuity equation.
state relating the PVT characteristics of the system.
Erroneous results will also be obtained when
the orifice equations are used for situations in v\ = = 10.0(0.695) 2 = 4.83 ft/s
Si
which pulsating flow is encountered if the fre-
quency of pulsations is considerably greater than Solving Equation 20.46 for —AP/p gives
the natural frequency of the manometer. The 2
errors are caused by the manometer or other
5
-,/ S1i \ T/ 0.139
ol7 \
2 I

pressure-measuring device indicating an arithmetic -AP \Sn

--ll
,V(^-1) "
(4.83) 2
[(—
I
\
) -1]
average of the differential pressure. The net result
C 2 2gc (0.635)
2
2 x 32.2
is that the indicated differential pressure will be
greater than that of the actual flow since the flow = 2.96 ft of oil

varies with the square root of the differential

or
pressure. If the period of pulsations approaches
that of the manometer, oscillations will preclude -AP= (2.96X62.3 x 0.87) = 161 lb,/ft
2

accurate reading. From Equation 20.36,

=
-AP 161
Example 20.6. A lube oil is flowing through a /? = 0.203 ft

5-in. sch.-40 steel pipe at a rate of 300 gal/min. Ph-Pl 62.3(13.6-0.87)

Inserted in this pipe is a 3. 5-in. standard sharp-
edged orifice to which is attached a mercury
This manometer reading would result if the

manometer. At the flow temperature, the oil has a

manometer were vertical. Since it is inclined 30
degrees, the reading measured along the sloping leg
specific gravity of 0.87 and a viscosity of 15
centipoises. If the manometer tube is inclined at an
is2x 0.203 x 12 = 4.86 in.

angle of 30 degrees to the horizontal, what would

The Elbow Meter
be the manometer reading measured along the
sloping leg? The elbow meter (7) functions somewhat similarly
to the orifice but uses the differential centrifugal
Solution.Equation 20.46 is appropriate for head between the inner and outer radii of the
use in and its use will require the
this solution, bend. These meters should preferably be calibrated,
determination of C To determine the orifice
. but a discharge coefficient for use in E quation
coefficient entails the evaluation of the Reynolds 20.46 could be approximated as C = 0.98 \JR C I2D,
number through the orifice and the use of Figure based upon the radius of curvature of the center-
20.11. line and the inside diameter of the pipe.

.
 THE ENERGY BALANCE IN FLOW SYSTEMS 563

The Venturi Meter
A properly designed venturi meter (Figure 20.13)
is one in which form friction is minimized. The
streamlined shape of the meter virtually eliminates
boundary-layer separation so that form drag is
negligible. The converging cone is about 25—30
degrees, and the diverging cone should not exceed
7 degrees. Venturi meters are difficult and expen-
sive to manufacture and in large sizes are very
tubes arranged parallel to flow. The outer tube is
perforated with small holes perpendicular to the
flow direction that lead into the annular space. The
annular space is otherwise sealed except for a
manometer lead. The inner tube has a small
opening pointed into the flow. This tube is
connected to the other side of the manometer. At
equilibrium there is no fluid motion within the
pitot tube. The annular space serves to transmit the

bulky. The permanent pressure loss in a venturi is static pressure. The flowing fluid is brought to rest

about 10 percent of the total pressure drop across at the entrance to the inner tube, and the tube

the meter, substantially less than in an orifice transmits an impact pressure equivalent to the
meter. Equations 20.46 and 20.47 are directly kinetic energy of the flowing fluid.

applicable to a venturi meter if the coefficient Co
is replaced by C v which has a value of about 0.98
,
for commonly encountered conditions. A venturi
meter also measures an average velocity.
Figure 20.14 illustrates a pitot tube. An
energy balance may be written between points 1
and 2. At point 2, the fluid velocity is zero so the
energy balance becomes

Flow
("i)' = 2a
ra- SF (20.48)

Equation 20.48 may be combined with Equation

20.41 and rearranged to give

Figure 20.13. Venturi meter.

v, =C,
M¥) (20.49)

The coefficient Cp of Equation 20.49 is usually

The Pitot Tube unity for a well-designed pitot tube. This means
A pitot tube is a device that measures a point that fluid friction between 1 and 2 is very small
velocity. It normally consists of two concentric and that the pressure drop measured by a pitot

Flow duct

i' 9 =

(D "i

Flow
^\P"
Static pressure
^
openings

Sectional view
of tip
of pitot tube

Figure 20.14. Pitot tube.

564 APPLICATIONS TO EQUIPMENT DESIGN

tube is attributable only to a kinetic-energy and

change. Q r, 2vrdr
The pitot tube measures only a point velocity, v= „= r
|
—=- (20.52)
and to get accurate values requires a well-designed
7T/-1 -o n
instrument and perfect alignment of the instru- If point velocities are measured with the pitot tube
ment with the flow. An percent
error of less than 1
at various radii from the center of the duct to the
is possible with a pitot tube for a wide range of duct wall, the integration of Equation 20.52 will
Reynolds numbers. This error may be contrasted give the average velocity of the fluid flowing
with the orifice and venturi meters, which in through the duct.
normal practice give results that have an error of If the velocity distribution within the duct
about 2 percent. is normal — that is, the velocity distribution is
To get the average velocity of a flowing fluid in agreement with the momentum-transfer equa-
with a pitot tube requires a point-to-point traverse tions — Figure 20.15 may be used. This graphical
across a duct diameter. Consider a circular duct of representation shows the relationship between
radius r\ across which a pitot tube traverse is to be average velocity and the maximum velocity for
made. The average velocity through this duct may fluids flowing in circular ducts. The maximum
be defined as velocity would be measured at the center of the
duct. To assure normal velocity distribution within

volumetric flow rate Q duct requires a straight run of at least 50

a circular

v =- (20.50) duct diameters without fittings or other obstruc-

cross-sectional area 7rr 1
tions.

where Q = volumetric flow rate of the fluid, ft 3 /s. Area Meters

The quantity of fluid flowing through any ring of Fluid meters of the previous group indicate a flow
radius r is 2irrv dr, where v is the velocity at radius by a change in pressure. An area meter, on the
r. The total flow is then other hand, is one in which the pressure drop is
constant and the reading is dependent upon a
variable flow area. The fluid stream passes through
Q= J
2-nrv dr (20.51) a constriction that accommodates itself to the flow

Dv0

:c 10 4 10
_ 0.9

>-0.8 t
( lufuuienl Mow)

A^
1

E0.7
/
/
/

f 0.6
/ L_
/ 1

/
1
/
1

* / J
0.5
(Laminar
flow)

0.4
10 3 10 4 1(V

DvmzxP

Figure 20.15. Relationship between average velocity and maximum velocity for
flowthrough circular ducts (5). (By permission of McGraw-Hill, Copyright © 1933).
 THE ENERGY BALANCE IN FLOW SYSTEMS 565

Outlet

Stuffing box packing

Stuffing box follower

Tube wall
(Taper exaggerated)
Capacity graduations
etched on
Tapered glass
metering tube
Metering float

Figure 20.16. Sectional drawing of a rota-

meter.

so that a constant pressure difference is main-

tained. The most important meter in this classific-
ation is the rotameter.
The rotameter (Figure 20.16) consists of a
plummet or float that is free to move inside a
Flow
vertical tapered glass tube. The fluid enters the
bottom of the tube, and as it flows upward it Figure 20.17. Schematic representation of
exerts a force on the bottom of the float. When the rotameter.
upward force on the plummet is equal to the
gravitational force acting downward on the plum- around the float. Thus, at steady state,
met, the plummet becomes stationary at some
point in the tube. The area available for flow is the Fd - Fq — FB - Vfpf Vfp = Vf(p f - p) -
annulus between the tube walls and the plummet. 9c 9c
The constant pressure drop arises from two fac- (20.53)
tors: the change in kinetic energy and fluid where
friction. Form friction is of major significance in
this instance. Floats may be designed so that form Vf = volume of the float
friction may or may not be of significance. The Pf = density of the float
"Stabl-Vis"* float, for example, is a float whose p = density of the fluid
behavior is generally independent of viscosity. The The factors that contribute to the drag force
behavior of a spherical float, however, is quite can be analyzed by writing an energy balance
dependent upon viscosity. between point 1 just before the float, and point 2
Figure 20.17 schematically shows a rotameter at the largest float cross section. Between these
with the float at its steady-state position. At this points, Equation 20.16a reduces to
condition, the forces acting on the float must be — 2 -
balanced so that no net force acts to move the "2 ~"1 2
float. The forces present are a gravity force (Fq )
+ 2^ = (20.54)
2ge a
acting downward on the float, a buoyant force
Using the continuity equation, Equation 20.54 can
(FB ) acting so as to raise the float, and a drag force
be rearranged to give
(FD ) resulting from form and skin friction for flow
P,-P
1^1 - 1 = ~2gc a + SF (20.54a)
* Registered trade mark, Fischer-Porter Co., Hatboro, Pa.
566 APPLICATIONS TO EQUIPMENT DESIGN

As with the orifice, the bracketed term in Equation Proper design of the rotameter float will make
20.54a can be represented by Equation 20.41, so the coefficient CR constant over wide ranges of
that Reynolds numbers. This constancy of CR is par-
ticularly useful if the fluid being metered is subject
to wide viscosity variations while its density is
nearly constant.
v, =C (20.55)
For a fluid of fixed density in a single
rotameter, the terms within the square root symbol
(S-) of Equation 20.58 are nearly constant and inde-
pendent of the flow rate. Combining these terms
The on the maximum
pressure difference acting
not identical to (—AP)-l2
with CR and Sf to give a new constant {CR ') yields
float cross section is

because some fraction of this is recovered as the 7t ^Cr 'S 2 (20.58a)

stream returns to the full tube diameter, and the
velocity decreases proportionally. The recovery
fraction is likely to be small, and, like the
permanent pressure loss for the orifice, it is not
dependent upon the velocity. Then
F D = (-APf)S f = (-AP,
where (— AP f is the pressure drop acting on the
) relation no longer applies. Usually the rotameter is
2
top of the float and Cf is the fraction of the
maximum pressure drop (— APt 2 that is not
recovered to act on the top of the float. Therefore
Equation 20.55 when combined with Equations
20.53 and 20.56 gives
2g c Vf(pf-p)
7i =CR
which shows that the flow velocity is a constant
times the minimum cross section for flow. Equa-
tion 20.58a emphasizes the great advantage of a
rotameter relative to an orifice, venturi, or pitot
tube. The rotameter calibration curve can be varied
2 in shape at the desire of the designer by varying the
2 )Cf Sf (20.56)
tube geometry and float density. The square-root
designed to give a nearly linear calibration, so that
)
instrument sensitivity is constant throughout the
entire operating range.
The major disadvantage of the rotameter is
that is becomes prohibitively expensive in the
larger sizes, so that rotameters are most often used
— in installations with pipe sizes less than 2 in.
9c Another disadvantage is that inherently the flow
(20.57)
rate must be read at the instrument rather than at a

Mi?- 1
)
central control panel. This disadvantage can be
overcome by any of several remote-reading meth-
where CR is the rotameter coefficient, which is ods, but these methods also increase the cost.
equal to C^ICf. Since the tube of a practical
rotameter tapers very gradually, S f = S^ — S2 and ,
Positive- Displacement Meters

Equation 20.57 simplifies to The meters described above depend upon a pres-
sure drop to measure the flow. Some meters
'2gVf [ Pf -p) indicate flow by means of mechanical devices.
71=C "| (20.58) Metal parts are in contact with a flowing fluid, and
p(S, + S2 )

the parts integrate in various manners so that a

The rotameter coefficient (CR ), like the total discharge is indicated. This discharge can be
orifice coefficient, is a function of the Reynolds timed over some convenient interval to determine
number through the minimum cross section. It is the flow rate. Prominent among this group are the
also a function of the shape of the rotameter float. wet and dry gas meter and the rotary disk meter.
Coefficients are given in Figure 20.11 for rota- The wet gas meter, (Figure 20.18) consists of a
meters with floats of three different shapes. These segmented drum that is caused to rotate by the
coefficients apply to Equation 20.59, which is a flow of gas. Gas enters under the liquid level, filling
simplified equation obtained by assuming that S 2 a segment, and causes the segment to rotate,
is very small in comparison to S and by converting
f expelling an equal volume of gas from another
Equation 20.58 into the mass-flow form through segment. The dry gas meter uses bellows with the
the use of the continuity equation. number of oscillations of the bellows being re-

corded on a series of dials.

2gVf(p f - p)p
w = C R S2 (20.59) The rotary-disk meter that is commonly used
Sf to measure the water flow in households is a meter
 THE ENERGY BALANCE IN FLOW SYSTEMS 567

Water
manometer
through the meter. There is a distinct advantage in
being able to meter directly the mass of material
that might be flowing through a system. Direct
Thermometer
measurement of mass avoids errors that are likely
Gas outlet on
back of meter
to be encountered in metering a fluid either
Gas inlet on volumetrically or by a velocity measurement with
back of meter
the subsequent conversion to volumetric flow rate.
Such variables as temperature, pressure, and vis-
Gas-inlet slot
Calibrating to bucket cosity can often lead to considerable error in
point
Gas-measuring volumetric-flow measurements.
rotor
For a mass meter to be independent of all
other variables, it must be based upon principles
Figure 20.18. Sectional view of wet-test gas
meter.
governed only by the mass of the flowing material.
One device that does this has gyroscopic precession
having a cylindrical measuring chamber with a flat as its operating principle. In this type of meter, the
diaphragm extending radially across the chamber. fluid is caused to flow in a circular path. In so
This diaphragm is pivoted in such a manner that it doing, the fluid establishes a torque that makes the
wobbles as a result of fluid action upon it. As seen instrument precess (move about its axis) in pro-
in Figure 20.19, the liquid enters the measuring portion to the mass velocity of the flow. The
chamber at one side and forces the disk ahead of it torque that is developed is directly proportional to
to move in an oscillating manner. The disk then the mass rate of flow.
forces the fluid out the discharge end. A counter Another type of gyro meter depends upon
indicates the number of wobbles, which can in turn Coriolis acceleration. In this meter, the fluid
be converted to volumetric flow. accelerates radially outward between vanes of an
impeller. There is a tendency for the impeller to lag
Mass meters behind a casing that rotates with it, and this lag
A group of recently developed meters depend for furnishes a measurable torque indicative of the
operation upon the mass of the fluid flowing mass-flow rate.

Figure 20.19. Sectional view of a disk meter. (Courtesy Neptune Water Meter Co.
568 APPLICATIONS TO EQUIPMENT DESIGN

Axial-Flow Turbine Meter which the fluid flows; this pipe is enclosed by an

The turbine flow meter utilizes an axial flow electromagnet. The fluid flows through the mag-
in the flowing stream. Its rate of rotation is
netic field. In so doing, the liquid generates a
turbine
voltage directly proportional to the flow through
almost linearly proportional to velocity of the fluid
over quite a range. An electric pulse is generated by the meter. Its use is restricted to fluids having some

a magnetic element on the turbine. The readout of

electrical conductivity, such as liquid metals.
these pulses can be translated to rate or cumulative
flow. Its extremely low pressure drop permits its Liquid Vortex Flow Meter

use in liquid streams very near the saturation The vortex flow meter depends upon the
liquid
temperature. eddies or vortices formed in the wake of a
nonstreamlined object, such as that shown in
Ultrasonic Flow Meter Figure 13.4/?. The rate at which the vortices form
The ultrasonic flow meter passes a pulsed pressure and separate from the object is a function of the
from one side of a pipe to the other, and the flow rate of the fluid stream. The rate of vortex
retardation of the pulse by the flowing material is a separation can be sensed electronically and con-
measure of the flow rate. Figure 20.20 illustrates verted into an electric signal, which is convenient
the ultrasonic flow meter. This meter is unique in for automatic control purposes. A modern vortex
flow meter is pictured in Figure 20.22.

Flow

Figure 20.20. Ultrasonic flow meter.

that nothing need be placed within the flow duct

for metering purposes. Two sonic beams are used:
one transmitted diagonally upstream and the other
diagonally downstream. Both beams are trans-
mitted in short bursts. The pulses of energy from
the two beams are integrated electronically and can
be converted to flow rate. This meter can be used
to meter practically all fluids.

Magnetic Flow Meter

As is shown in Figure 20.21, the magnetic flow

meter consists of a nonmagnetic pipe through

Electrode Figure 20.22. Vortex flow meters. Two sizes are shown.
Electrode Vortices are shed from the nonstreamlined vertical

element within the circular flow channel. The housing

mounted above the flow channel contains the electronics
necessary to sense the vortex shedding and convert it
into a signal suitable for a control system. The ring in
the foreground is adapting the meter to various
for
Electromagnet
classifications of flanges. (Courtesy Fischer and Porter
Figure 20.21. Magnetic flow meter. Company.)
 THE ENERGY BALANCE IN FLOW SYSTEMS 569

AGITATION OF LIQUIDS agitated material — as the pilot unit required. In

other cases, such as the dispersion mentioned
Agitation, or mixing, is one of the oldest and most
above, the shear stress at the impeller tip must be
common chemical engineering, yet
operations in
constant for effective scale-up. Scale-up of mixing
understanding of the subject is limited. Although
for many operations characterized as "flow vel-
the effectiveness and energy consumption in agi-
ocity sensitive" is most successful if torque/unit
tation depend upon basic principles of fluid mech-
volume is held constant (2). For these applications,
anics, the flow patternsthe typical agitated
in
less magnification of power input for a scale-up
vessel are so complex that rigorous application of
results than for a more classical prediction based
the basic principles is impossible. Instead, empirical
upon equal power per unit volume.
approaches must be used.
Because of the wide variety of applications,
Agitation is used in various applications,
agitation equipment takes many forms. A thorough
including:
review of the theory and practice of agitation is
1. Dispersion of a miscible solute throughout found in Reference 11. A sampling of agitator
a solvent. impellers is shown in Figures 20.23a—e. In Figure
2. Blending of two miscible liquids. 20.23a an flow impeller with flow-directing
axial

3. Production of a slurry of a finely divided stabilizers at the tip is shown. This impeller allows

solid in a liquid. minimum cost agitation over a wide range of

4. Mixing of reactants in a chemical reactor. modest viscosity applications requiring rapid and
5. Stirring of a homogeneous liquid to im- thorough movement. A flat-blade turbine is shown
prove heat transfer to the liquid. in Figure 20.236. This impeller produces intense
turbulence, largely in the radial direction, but at
It is obvious from this partial list that for agitation
the expense of higher power requirements. Figure
to be effective it must be carefully tailored to the
20.23c shows a turbine for use as agitator-
particular result desired. A dispersion application
circulator in a draft tube. Here all movement
may require the presence of high-intensity shear
would be axial, with the draft tube controlling the
forces acting in a restricted flow area if agglom-
circulation pattern. A very uniform agitation in-
erates of fine particlesadequately
are to be tensity results. The anchor impeller shown in
distributed. No such requirement may exist in the Figure 20.23tf is used where a low level of
mixing of reactants in a chemical reactor, although turbulence is required in a viscous fluid. Such an
the viscosity of the mix may increase markedly as impeller is particularly effective for use in a vessel
the reaction proceeds, thus requiring a high power heated or cooled by a wall jacket. Figure 20.23e
input and sturdy agitator design. Thus, almost shows a helical impeller. This type of unit is most
every application of agitation requires that the often used to blend solid— liquid ingredients or to
design be based upon past industry practice or agitate doughs, pastes, or sludges. In many appli-
upon tests carried out on the particular system cations a much heavier construction is required.
involved. For moving even heavier materials — elastomeric
For specification of required equipment in a mixes, for example — the impeller, although of a
new application, and the prediction of its perform- concept similar to the helical impeller, reduces to a
ance, similarity theory* is a useful empirical tool. series of knives or wedges, and the material is
The performance of a large-scale agitator system kneaded or cut instead of agitated. Obviously,
can be predicted from test measurements in a agitation of heavy and/or viscous material, non-
small, but similar, system. Analysis of test results Newtonian fluids, and multiphase materials is an
on the basis of fluid-dynamic principles leads to a extremely common situation, but one where the
scale factor for the important variables. These scale practice is more art than science.
factors allow empirical magnification from test
apparatus to large-scale installations. Power Requirements of Agitators
In many situations, the scale-up can be based Agitation is typically, although not necessarily,
upon the principle that the full-scale equipment carried out in a cylindrical tank, with the agitation
must supply the same energy intensity — that is, provided by blades rotating on a shaft coincident
the same power consumption per volume of with the vertical axis of the tank, as shown
schematically in Figure 20.24. An agitator may
* See Appendix A. operate on either a batch or continuous basis. In
570 APPLICATIONS TO EQUIPMENT DESIGN

Figure 20.23c. An axial-flow circulator for use in a draft

Figure 20.23a. An axial-flow impeller. (Courtesy of Mixing

tube. (Courtesy of Mixing Equipment Co., Inc., a unit of
General Signal Corp.)
Equipment Co., Inc., a unit of General Signal Corp.)

Figure 20.23o'. An anchor impeller. (Courtesy of

Figure 20.236. A flat-blade turbine impeller. (Courtesy of Mixing Equipment Co., Inc., a unit of General
Mixing Equipment Co., Inc., a unit of General Signal Corp.) Signal Corp.)

-
 THE ENERGY BALANCE LN FLOW SYSTEMS S71

shown in Figure 20.246. The vortex forms because

of centrifugal force acting in the liquid and the free
liquid surface that can respond to the force.
Agitation power requirements can be ana-
lyzed using the energy equation developed earlier
in this chapter. For example, Equation 20.16 can
be applied to Figure 20.24. As shown, there is no
inflow or outflow from the tank, so the flow terms
in Equation 20.16 are zero, and the equation
becomes

-ZF=W f (20.60)

This equation simply states that work must be

done to overcome the form and skin friction
produced by agitating the fluid in the tank. Even if
there were inflow and outflow, as would be the
case in a continuous agitator, the magnitude of the
flow terms would be small compared to the terms
remaining in Equation 20.60, because a substantial
amount of energy is needed for the agitation.
The analysis of the baffled tank is the simpler
of the two cases and is considered first. The
motion of the agitator blade in the fluid and the
resultant movement of fluid past the baffles and
wall have associated with them skin friction and
Figure 20.23e. A helical impeller. (Courtesy of Mixing form drag, depending upon the speed of the
Equipment Co., Inc., a unit of General Signal Corp.)
rotation and the design of the blades and tank. For
a geometrically similar* system of agitator, tank,
and baffles, the drag coefficient will be the same
continuous operation, the materials to be mixed
function of Reynolds number
are added continuously and the mixture is removed
continuously. An example is a stirred-tank chem-
CD =M/V Re )
ical reactor, where the reactants flow in and the

products flow out continuously. as discussed in Chapter 13. Applying the defi-
The agitator blades take many different nitions found there yields
forms, depending upon the job to be performed.
Usually the tank contains baffles, typically metal
plates mounted vertically along the wall (Figure
2Fgc
£a 120.61)
20.24a). The baffles promote greater mixing and
produce a more-or-less horizontal liquid surface.
Without baffles, a vortex forms at high speeds, as * See Appendix A.

(pn ) Liquid level

Baffle t- -

'

f Agitator

(b)

Figure 20.24. Agitated tanks, (a) Baffled, (o) Unbaffled.

572 APPLICATIONS TO EQUIPMENT DESIGN

where This equation is written expressly for geometrically

similar systems in which the tanks are baffled.
F = force applied to the solid surface by
Several typical plots of /V Po as a function of
fluid motion
/V Re are shown in Figure 20.25. Note that, for
S = projected area of the solid normal to
baffled tanks, the two limits of the curve have
fluid motion —
characteristics universal to all CD —N Re plots
L' = characteristic dimension
namely, at low Reynolds numbers,
v = characteristic velocity of the fluid
p = density of the fluid constant
= absolute viscosity of the fluid N Po = (20.67)
p N Re
With the exception of the fluid properties, the and, at high Reynolds numbers,
terms in Equation 20.61 are difficult to measure in
the typical agitation system, so the equation must /V Po = constant (20.68)
be modified before it is used. In a geometrically The region defined by Equations 20.67 and 20.68
similar system, all length dimensions bear a con- represents transition of the boundary layer and
stant ratio; therefore, any measurable dimension development of significance of the accelerative
may be used to replace L' The agitator diameter .

effects associated with form drag.

D' will be used in any instance in which a
unbaffled tanks, the formation of vortex
In
characteristic length is required.
introduces an additional mechanism: the fluid
The term F/S, the force applied to the fluid forces associated with gravity. In the vortex a part
more conveniently if the
per unit area, can be used
of the tank contents is supported against the
term is converted to power input. Recall that gravitational acceleration of the earth, and there-
power is the product of force and velocity of fore fluid forces in the system must supply the
application; then,
force to maintain the head of fluid that constitutes
vF P
— =- (20.62)
the vortex.
analyzed
The nature of the forces can be
an element of fluid is examined in the
5 S if

vortex, as shown in Figure 20.26. The fluid

where P = power, ft lb f /s (or N m/s). The area term
element is taken at z , the lowest point of the
may be written in terms of the characteristic
vortex; therefore, at any radial position, a "head"
2
length; that is, the area is proportional to D' .

of fluid will exist above the element. At steady

F P state, the pressure on any element in the fluid must
(20.63) be such that the forces on every face of unit area
of the element are the same. As elements are
The velocity (v) is a linear velocity. In this system, examined at Zq, from r = to r — f\ the pressure
the tangential tip velocity of the agitator is
must be increasing because an increasing head of
proportional to linear velocity so that fluid is present. The force associated with the head
is represented by F bl the body force, and the radial
v a tt/VD' oc ND' (20.64)
force is represented by Fc and, surely
where N - speed, revolutions/s. Equations 20.63
and 20.64 may be substituted into Equation 20.61. Fb =Fc (20.69)

The force Fcobviously a centrifugal force

then is
(20.65) resulting fluid through the
from the motion of the
N 3
D' p5 \
n )
volume element. Therefore, each of the forces can
or
be related to system variables. The body force on
(20.66) the element Ax Ay Az is

where
Fb = Ax Ay(z - z )p — (20.70)
Wpo = Pgc /N 3 D' 5 p, the drag coefficient for 9c
agitation systems referred to as the
and the centrifugal force on the element is
"power number"
2
A/ Re = D' Nplp, the Reynolds number ex- Ax Ay Az prco'
pressed in variables convenient for /> = (20.71)
agitation 9c
 THE ENERGY BALANCE IN FLOW SYSTEMS 573

z1 Baffles
Type of impeller No.
D' C D No. wID

See*1 3 2.7- -3.9 0.75- -1.3 4 0.17 1

See#1 3 2.7- -3.9 0.75- -1.3 4 0.10 2

See#1 3 2.7- -3.9 0.75- -1.3 4 0.04 4

Same as # 1, 3 2.7- -3.9 0.75- -1.3 4 0.10 3

curved 2 blades

Marine propeller
3 2.7- -3.9 0.75- -1.3 4 0.10 5
3 blades, pitch = D'

See #2 3 2.7--3.9 0.75- -1.3 4 0.10 6

Same as 5 but
0.75- -1.3
3 2.7- -3.9 4 0.10 7
pitch = 2D'
*
See *± 1 3 2.7- -3.9 0.75- -1.3 0.00 8

In non-Newtonian fluids (6). In unbaffled tanks (9)

Turbine 1

Ch1 D

6 blades

Flat paddle
2 blades

10.000

D' = diameter of impeller Pitch = the axial distance

D, = diameter of tank a free propeller would
hi = width of baffle move through a non-
z, = elevation of impeller yielding liquid during one
Zi = height of liquid revolution

Figure 20.25. Plots of Np as a function of Np e for various agitator-tank systems,

(6) By permission of AlChE © Copyright 1961, (9) By permission of AlChE ©

Copyright 1950.
574 APPLICATIONS TO EQUIPMENT DESIGN

represents the ratio

/Forces associated with kinetic energyx

\ forces associated with gravity /

and the group is dimensionless.

The additional
gravitational effects in an
unbaffled agitator system can be included in the
correlation equation previously derived in terms of
the Froude number so that

/V Po =f(/V Re )(/V Fr ) (20.78)

Figure 20.26. Forces associated with an Data are plotted using a relationship suggested by
element of fluid in a vortex. Equation 20.78 in Reference 9 for a variety of
agitator blades.The curve for a single flat-bladed
where turbine an unbaffled tank is included in Figure
in

Ax, Ay, Az = the element dimensions, ft 20.25. It can be seen that the absence of baffles
z — z = the height of fluid above the does not affect the correlation until the Reynolds
element, ft number exceeds 300.
r = the radial position, ft
to = the angular velocity, s~ 1 Non-Newtonian Liquids
The agitation of non-Newtonian liquids has re-
These forces can be equated. ceived considerable attention (1,6). Metzner and
associates examined a variety of dilatant and
—g Ax Ay Az prco 2
Ax Ay{z - z )p (20.72) pseudoplastic liquids and found that the shear rate
9c is a linear function of agitator speed (6)

p, Ax, Ay, and gc cancel from the equation so that dv

= 11 N (20.79)
dr
(z-z )g= Azrto 2 (20.73)
where the constant 11 was determined experi-
The right-hand term of the equation is related to
mentally.
the kinetic energy of the fluid, and the left-hand This expression can be used with data on
side is related to forces associated with gravity.
rheological properties of the liquid to find the
This equation may be written apparent viscosity at the average shear rate. By the
Az rco 2 definition introduced in Chapter 9, the apparent
— =1 (20.74) viscosity is
(z-z )g T y gc
Ma
= (20.80)
The exact numerical value of each term is not -dv/dr
known, but in a geometrically similar system, all
length dimensions can be taken proportional, so
For dilatant and pseudoplastic liquids
that dv
rY g c = ~K (20.81
Azoc (z -z ) ~r~0' (20.75) dr

and in a dynamically similar system For agitated vessels, dv/dr is normally negative, so a
combination of Equations 20.80 and 20.81 yields
(20.76)
dv n-1
Therefore, Equation 20.74 may be written as Ha=K (20.82)
dr
2
7V Combining Equations 20.79 and 20.82 gives
= constant (20.77)
pa =K( 11 A/)"-
1
(20.83)

for a dynamically similar system. The group Given the rheological characteristics (K and n) and
D'N 2 lg is called the Froude number (/V Fr ) and the agitator speed N, the apparent viscosity pa can
 THE ENERGY BALANCE IN FLOW SYSTEMS 575

be calculated and used to calculate the Reynolds From Figure 20.25 at /V Re = 270, A/ Po = 3.7. Solv-
number, for use with Figure 20.25 to find the ing for power gives
power number. The power required for the agi- 3 5
(3.7)(3.05) (62.3)(1)
tation can then be calculated from the power P=
number, as shown in the example below. 32.2
Forthedilatant fluids (n > 1), Metzner recom- = 203 ft \b f /s = 0.369 hp
mends use of the curve for Newtonian fluids
(curve 4, Figure 20.25). For pseudoplastic fluids (b) With p = 1.0 lb/ft s and A/ Re = 270
(n < 1), the dashed line shows the deviation in the (270)(1;
A/ = = 4.33 rev/s
transition region that has been found. 2
!1) (62.3)
It was also found that a minimum Reynolds

number must be maintained to ensure adequate The power number does not change, because the
circulation through the mixing zone. For the single curves for Newtonian and pseudoplastic liquids
turbine, this value ranges from /V R e =270 for a coincide at this Reynolds number. Therefore,
tank/turbine diameter ratio (D t /D') of 3 to A/ Re =
A/ Po =3.7
50 for D t /D' = 1.33. This latter value occurs in the
region where the pseudoplastic curve deviates from and
the Newtonian curve, but for an agitation rate 3
(3.7)(4.33) (62.3)
significantly above the minimum, the two curves P=
again coincide. Data for other agitator blades show 32.2
deviations from the Newtonian case (11). = 581 ft lb,/s = 1.06 hp

(c) By the method of (a) with K= 1.0, n - 1.1

Example 20.7. Calculate the power required = 6.65,
/V />=3.83hp
for agitation with a single-bladed turbine in a
cylindrical tank for the conditions of curve 4,
Figure 20.25, for the cases given below. The liquid
density is 62.3 lb/ft 3 The minimum Reynolds
. REFERENCES
number for adequate mixing is to be used. Turbine
diameter is 1 ft. 1. Calderbank, P. H., and M. B. Moo- Young, Trans. Inst.

(a) Pseudoplastic liquid (K = 1.0, n = 0.9) Chem. Eng. (London) ,39, p. 22 (1961).
(b) Newtonian fluid (p = 1 .0 lb/ft s) 2. Connolly, J. R., Chem. Eng. Prog., 72, No. 5, p. 52
(c) Dilatant fluid (K = 1.0, n = 1.1) (1976).
3. Fluid Meters: Their Theory and Application , 6th ed.,
Solution ASME, New York (1971).
(a) For D t /D' = 3, use/V Re = 270. 4. Kays, W. M., Trans. ASME, 72, p. 1067 (1950).
5. McAdams, W. H., Heat Transmission , McGraw-Hill,
D' 2 Np New York (1933).
= 270
6. Metzner, A. B., R. H. Feehs, H. L. Ramos, R. E. Otto,
Ma
J. D. Tuthill, A.I.Ch.E. Journal, 7, p. 3 (1961).
so 2 Murdock, J.W., C.J. Foltz, C. Gregory, J. Basic Eng.,
l (/V)(62.3) 7.

Ma
= = 0.231 N 86, pp. 498-506 (1964).
270 8. Perry, R. H., and Cecil H. Chilton, Chemical Engineers'
From Equation 20.83 Handbook, 5th ed., McGraw-Hill, New York (1973).
9. Rushton, J. H., E. Costich, and H.J. Everett, Chem.
9 " 1
ju a = 1.0(11 A/)
-

Eng. Prog., 46, pp. 395, 467 (1950).

10. Tuve, G. L., and R. E. Sprenkel, Instruments, 6, p. 201
Equating these two expressions for the apparent
(1933).
viscosity yields
11. Uhl, V. W., and J. B. Gray, Mixing-Theory and Prac-
1"
0.231 N= 1.0(1
- 1
) AT - 1
tice, Academic Press, New York (1966).
12. Weir, Chemical Engineering, November 25, 1975.
J. T.,
so that
13. Owczarek, J. A., Fundamentals of Gas Dynamics,
N= 3.05 rev/s International Textbook Company, Scranton (1964).
14. Shapiro, A. H., The Dynamics and Thermodynamics of
and Vol., Ronald Press, New
Compressible Fluid Flow, II

= (0.231X3.05) =0.704 lb/ft s York (1953).

pa
576 APPLICATIONS TO EQUIPMENT DESIGN

PROBLEMS must be maintained to deliver the required 150 gal/min.

The line from the tank is 2-in. sch.-40 steel pipe. In
addition to its vertical run, the line has a 10-ft horizontal
20.1. A 500-watt heating element is used to heat
section and contains four 90-degree elbows and one gate
3 m 3 /min of air flowing into a 30-mm tube (2.6 mm wall
thickness) at 20°C and 200 kN/m pressure. When the air
valve. What is the minimum height above the spray nozzle

leaves the heater its pressure is 110 kN/m 2 What is the .

at which the liquid level in the tank must be maintained?
temperature of the air leaving the heater? 20.8. Rework Problem 20.7 on the basis that glycerol

20.2. Air is flowing through a 12-in. I.D. iron pipe

at 70 F is stored and fed to a reactor through a similar
piping system and delivered to the reactor at the same
that discharges into a 3-in. I.D. iron pipe. At a particular
point in the larger pipe, the pressure is 1 5 psig, the
conditions as in Problem 20.7.
80° 20.9. Liquid benzene at 25°C is to be pumped from
temperature is F, and the air velocity is 5.0 ft/s. At a
point a storage tank to a chemical reactor through 100 meters of
in the smaller pipe, the pressure is 5 psig, the
temperature is 80° F. The point in question in the 3-in. pipe
50 mm nominal size steel pipe (medium weight). The line
contains two 90-degree elbows and an open globe valve.
is 100 ft above the point in the 12-in pipe. Assuming
perfect-gas behavior, calculate the energy lost due to
The reactor is located 50 m above the liquid level in the

friction between the two points.

storage tank. The tank is vented to the atmosphere and the

20.3. A closed tank kept partially filled with oil

reactor operates at a pressure of 500 kN/m 2 Calculate the .

(specific gravity = 0.9) has a pressure in the gas space above

power (watts) that must be delivered to the liquid by the
the oil of 10 psig. If the oil is discharging through a hose at
pump to maintain a flow of 0.1 m 3 /s.
20.10. Pressure in a storage tank is maintained at
50 gal/min, estimate the pressure at the nozzle located 10 ft

lower than the oil surface and with a 1 -in. I.D. discharge.
1 MN/m 2 to supply water at 20 C and atmospheric

Assume a total frictional loss in the line amounts to

pressure through 200 m (equivalent length, including fit-

1.0ft-lb/7lb.
tings) of drawn steel tubing (30 mm O.D., 2.6 mm wall

20.4. A research team is designing a flow system for

thickness). What flow rate of water can be maintained?

a nuclear reactor to study corrosion problems. The equip- 20.11. A copper smelter is located in a small city

ment as constructed consists of 230 ft (total equivalent

whose water pressure is insufficient to meet the company
length) of 1 -in. I.D. stainless-steel pipe. Molten bismuth demands. It is decided that a large elevated tank will be
is

pumped from a melt tank maintained at 350°C and 5

erected to supply water for company needs. Requirements
are: water at 65° F and a delivery pressure of 40 psia at
microns absolute pressure, through a test section included
the 230 and back to the melt tank. 100 gal/min. The line leading from the tank to the point of
in ft, If the bismuth
use is standard 2-in. steel pipe. In addition to its vertical
velocity is 1.0 ft/s, how much theoretical power must be
run, the line has a 10-ft horizontal run and contains two 90-
supplied to a pump placed in the line? Assume a constant
degree elbows and a gate valve. The tank is closed but
temperature of 350°C is maintained by suitable insulation.
Liquid bismuth properties:
vented to the atmosphere. At what height above the point
of use must the bottom of the tank be?
Viscosity = 1.28 centipoise
20.12. A paint factory keeps its solvents stored in
Density =613 lb/ft 3
vented tanks on the second floor of its mixing building.
20.5. A hydroelectric plant is supplied with water by One particular tank contains linseed oil that is fed through
a 10 km duct leading from a dam to the turbines. The duct a pipe to a mix tank on the first floor through 75 ft of pipe
is made of concrete and is 2 m in diameter. The duct inlet at (equivalent length of straight pipe). The bottom of the oil

the dam is 10 m below the water surface and 60 m above tank is 15 ft above the point of discharge into the open mix
the turbine entry. The turbine discharges to the atmo- tank. Calculate the minimum pipe diameter that will ensure
sphere. What flow of water at 20°C can be expected to the a flow of 10 gal/min to the mixer. Express pipe diameter to
turbine? How much power would the turbine develop for nearest commercial size of sch.-40 pipe. Viscosity of
this flow? Pressure at the turbine inlet is 120 kN/m 2 .
linseed oil = 15 centipoises. Specific gravity of oil = 0.92.
20.6. A summer home is to be supplied with water 20.13. Water at 55° F is to be pumped from a pond
by a pipeline leading from a spring 600 ft above the home. to the top of a tower 60 ft above the level of the water in

The line will be comprised of 650 ft of straight, sch.-40, the storage basin. It is desired to deliver 12 ft
3 of water per
1 -in. brass pipe, three 90-degree elbows, and one gate valve. minute at 15 psig. The transmission line consists of 400 ft

The pipeline discharges into a vented tank. What water flow of standard 3-in. steel pipe with eight 90-degree ells and
at 50 F could be expected? four gate valves. What horsepower is required for pumping?
20.7. A large elevated tank is used to supply water at What will be the electrical input to the motor if the
50 F to a spray chamber. To ensure proper atomization of motor-pump set has a 40 percent efficiency?
the fluid, a pressure of at least 40 psig at the nozzle inlet 20.14. A fountain is supplied by a large reservoir
 THE ENERGY BALANCE IN FLOW SYSTEMS 577

whose surface is located 30 m above the fountain nozzle, centipoises). The solution is subsequently drained into a

which points straight up. The reservoir and the fountain are reaction tank B located as shown in the following flow
connected by 100 m (equivalent length including fittings diagram. If the outlet gate valve on tank A and the inlet

and nozzle) of 25 mm nominal size steel pipe (medium gate valve on tank B are simultaneously fully opened, how
weight). The water is at 20° C. What is the maximum height long will it take to drain tank A down to a 1 -ft level?

a jet of water from the fountain can achieve? What would 20.18. Two summer cottages receive their water
be the maximum height if there were no frictional losses? supply from a dammed mountain stream. The piping
20.15. A light lube oil (specific gravity = 0.87, fi
= system is as shown in the accompanying illustration. The
3.0 centipoises) is pumped by a rotary gear pump to a height difference from water surface to the faucet in house
header at a rate of 1000gal/hr. At the header, the flow A is 50 ft, and that to the faucet in house B is 30 ft. What
system is branched into two lines. One is a 4-in. sch.-40 rate of flow will be delivered to house A if both faucets are
steel pipe and the other an 8-in. sch. 40 steel pipe. The wide open?
smaller pipe is 250 ft long, whereas the larger pipe extends 20.19. What is the mass velocity of air that can be
for 100 ft. Neglecting changes in elevation and pressure handled in a \ -in. sch. -40 horizontal steel pipe, 300 ft long,

drop through fittings and the header, what is the volumetric if the air flows isothermally at 100°F through the pipe?
flow rate of oil in each pipe? The pressure drops from 50 to 5 psig through the pipe.
20.16. A supply of gasoline at 20°C having a 20.20. In cases where open liquid manometer
viscosity of 0.667 centipoises and a specific gravity of 0.76 columns would be unusually high or when the liquid under
is pumped through a 150-mm standard horizontal pipe at a
rate of 2 m /min. At the end of 200 m, this pipe branches
into three lines consisting of 80-, 50-, and 25 mm standard Fluid B
pipes. If the pipes have respective lengths of 200, 100, and
50 m and discharge at atmospheric pressure, what is the
volume percent of total flow through each branch?
20.17. Tank A is filled with a 10 percent NaOH, 10
percent NaCI solution (specific gravity = 1.10, viscosity = 3

K6'H

r
10' Tank A
L

20'

30'
All 1 in. sch. 40 pipe 12'
Vent (open)

15' TankB

li
2'-J h6'H

100 ft

Open
2" pipe --3V pipe, 10 ft
10 ft

1200 40 ft 25
ft
— 1" pipe. 20 ft 30 ft
ft

Closed
40 ft

40 ft

200 ft

20 ft

v Closed

Elbows and tees as shown. All valves are globe valves.

578 APPLICATIONS TO EQUIPMENT DESIGN

pressure cannot be exposed to the atmosphere, the inverted atmospheric pressure and a plot prepared from which the
U-tube illustrated is sometimes used. Develop a general volume of air per minute referred to 32°F and normal
expression for the pressure difference between points 1 and barometer may be read directly. If this meter were used to
2 in terms of /?, a, b, and the fluid densities. measure the flow of dry CO2 gas at 70° F and normal
20.21. Show that the pressure-drop reading for the pressure, would it give high or low results? What correction
two-fluid manometer pictured in Figure 20.8 is given by factor would be necessary to make the plot usable?
20.27. It is desired to meter, by the installation of a
sharp-edged orifice, a stream of approximately 500 Ib/hr of
AP=R Pc -pb + ~<pa Pb) flowing at 70° F and 1.0 psig through a standard 4-in.
SB air

iron pipe line. It is agreed that the orifice will have flange
where
taps and that, for ease in reading the flow, a minimum
R= reading pressure difference of 2.0 H 2 must exist between the
in.

Sj = cross-sectional area of the tubes two taps. What diameter do you recommend for the orifice
Sq = cross-sectional area of the bulbs to be installed? What is your estimate of the static pressure
= density of fluids A, B, and C, respec-
PA- PB' PC of the air flowing at a point 3 ft downstream from the
tively orifice?

20.28. Brine (specific gravity 1.20) is flowing

20.22. A simple open-end U-tube, using mercury, is
through a standard 80-mm pipe at a maximum rate of
used to measure the pressure in a 2-in. pipe carrying CO2 0.9 m 3 /min. In order to measure the of flow, a
rate
gas at 70° F. The mercury level at zero reading is 2 ft below
sharp-edged orifice, connected to a simple U-manometer is

the pressure tap.

to be installed. The maximum reading of the manometer is
the reading on the manometer 1-in. Hg, what
(a) If is
to be 400 mm Hg. What size orifice should be installed?
is the pressure in the line?
Repeat, assuming a venturi meter is used instead of an
(b) If water were flowing in the line under the same
orifice.
pressure as the CO2, what would the manometer reading be?
20.29. A standard 0.500-in. orifice is installed in a
20.23. Petroleum oil of specific gravity 0.9 and
2-in. standard steel pipe. Dry air at upstream conditions of
viscosity 13 centipoises flows isothermally through a
70° F and 15 psig flows through the orifice at such a rate
horizontal sch.-40, 3-in. pipe. A pitot tube is inserted at the
that a U-tube manometer connected across the taps
center of the pipe, and its leads are filled with the same oil
indicates a reading of 35 cm of red oil. The red oil has a
and attached to a U-tube containing water. The reading on
specific gravity of 0.831 referred to water at 60° F.
the manometer is 3.0 ft. Calculate the volumetric flow of
(a) Calculate the weight rate of airflow in the pipe.
oil in cubic feet per minute.
(b) Estimate permanent head loss across the orifice.
20.24. A pitot tube is inserted into the center of an
(c) What percent of the power requirement could be
air duct 1 m in diameter. A pressure gage attached to the
saved by using a venturi meter in place of the orifice
pitot tube reads 7.9 N/m 2 . Calculate the mass flow rate of
(assuming a pressure loss from venturi meter equal to 20
air, at a temperature of 40°C and a pressure of 100 kN/m 2 .
percent of head across venturi).
20.25. A duct traverse is made with a pitot tube of a
20.30. A 25-mm sharp-edged orifice is installed in a
20.0-in. I.D. galvanized-iron duct through which air is
50-mm standard pipe (medium weight). What will be the
flowing at 100°F. A water manometer is used in connection
reading on a differential pressure gage attached across the
with the pitot tube. The following readings were obtained:
orifice if water at 20°C is flowing through the pipe at a rate

Position (r), in. APin. water of 0.002 m 3 /s?

20.31. An oil of 0.87 specific gravity and 6 centi-

0.0 3.67 poises viscosity flows through a pipeline. An orifice with

3.0 3.27 opening diameter of one-half of the inside pipe diameter is

5.0 2.67 used to measure the flow. It is proposed to replace this

7.0 1.90 orifice with a venturi with throat diameter equal to the
8.0 1.40
orifice diameter. If the coefficient of the orifice is 0.61 and
9.0 0.80
that of the venturi 0.98 and the flow rate is unchanged,
9.75 0.198
calculate:

(a) The ratio of the venturi reading to the orifice

Estimate the average flow rate of the air in cubic feet per reading.
minute. (b) The ratio of the net pressure loss due to the
20.26. A sharp-edged orifice in a thin plate has been venturi installations to that found with the orifice.
calibrated with dry air at 70° F and substantially standard 20.32. A standard 0.500 in. -flange-tap orifice is
 THE ENERGY BALANCE IN FLOW SYSTEMS 579

installed in a straight horizontal length of standard 2-in. power required for a 0.06-m tank at the same Reynolds
steel pipe.The pipe carries a steady flow of dry air at 85 F. number. Both operate at 100° F. Usezi ID' = 3.5.
A vertical U-tube manometer indicates a differential head 20.37. Using the conditions of curve 7, Figure 20.25,
of 70 mm Hg when connected across the orifice. When the together with a Reynolds number of 300 and a tank
upstream leg of the manometer is disconnected from the diameter of 1 meter, calculate the power required for fluids
orifice and left open to the atmosphere, the manometer kg/m
of viscosity 0.001 , 0.01 , and 0.1 s relative to that re-
reading indicates a static head of 10 mm Hg gage at the quired for a viscosity of 1.0, all of the fluids being of the
downstream tap. The barometric pressure is 745 mm Hg.
same density.
Calculate the volumetric flow of dry air in standard cubic
20.38. Repeat Example 20.7 for a dilatant fluid with
feet per minute. K = 0.1 and n = 1.4 (to yield an apparent viscosity in units
20.33. A venturi meter with a 1 2-in. throat is
of lb/ft s). Compare the results for the three types of fluids.
inserted in a 24-in. I.D. line carrying chlorine at 70°F. The 20.39. For the conditions of curve 4 baffled. Figure
barometer is 29.5 in. Hg, the upstream pressure 2 in. Hg 20.25, calculate the power required for agitation using the
above atmospheric pressure, and the head measured over minimum Reynolds number that provides adequate mixing.
the venturi (upstream to throat) is 0.52 in Hg. Calculate the The liquid has a density of 800 kg/m 3 and the turbine
rate of flow in pounds per hour. diameter is 0.3 m. The rheological properties are /C = 0.1
20.34. It is proposed to measure a given flow of and n = 0.7, 1.0, or 1.3. In each case, the constants given
water for municipal purposes by the use of either an orifice yield an apparent viscosity in units of kg/m s (or N s/m 2 ).
having a D^/Dp = 0.3 or a venturi. If it is desired to have the
20.40. Components for a liquid detergent (/z = 10
same reading for this particular flow on both instruments, centipoises) are blended in the pilot plant in a 10-gal,
how would the power loss compare in the two cases? baffled, flat-bottomed tank 10 in. in diameter. A double-
20.35. Using the agitator-tank system described in turbine agitator with blades 6 in. in diameter is used. A
curve 1 of Figure 20.25, calculate the speed for an agitator 1/2-hp motor turns the agitator at 500 rpm for 30 min to
that imparts 0.01 hp/ft 3 of tank contents in which the fluid attain complete dispersion.
under test is water at 70 F, and the tank holds 50 ft 3 .
In the plant, a geometrically similar unit is planned to
Calculate the power required for a 10-ft 3 tank at the same blend 200-gal batches of this solution. Determine consistent
Reynolds number. Use z\ /D' = 3.0 values of agitator and tank diameter, revolutions per
20.36. Using the agitator-tank system described in minute, power requirement, and batch time for the plant
curve 6 of Figure 20.25, calculate the speed required for a unit. Base the design on (a) constant N^ e , (b) constant
0.6 m3 tank in which 100 watts/m 3 is imparted to the agitator peripheral speed, (c) constant rpm.
tank. The fluid under test is SAE 10 oil. Calculate the
 Twenty- One

Fluid Motive Devices

In the typical flow system, such as those con- pumps. Positive-displacement pumps deliver a
sidered in Chapter 20, it is usually necessary to add definite quantity of fluid for each stroke or
energy to the fluid to maintain flow. The energy is revolution of the device; whereas centrifugal
added by a fluid motive device, such as a pump or pumps deliver a volume that is dependent upon the
a compressor. The device, part of the "surround- discharge pressure, or energy added. A detailed
ings," does work on the fluid, which is the discussion of pump types and applications is found
"system," and hence the energy added is equal to in Reference 3.
— W/ as defined in Chapter 20. According to
Positive-Displacement Pumps
Equation 20.18, the energy added may compensate
for frictional losses or provide for increases in The positive-displacement category may be sub-
velocity, pressure, or height of the fluid. This chap- divided further into reciprocating pumps and
ter is concerned with the various devices that are rotary pumps. As the names imply, the former
used to add energy to the fluid. involve a back-and-forth motion of a piston in a
Fluids involved in chemical processes may be cylinder, while the latter depend upon a rotating
gases, liquids, or combinations of both. It is con- motion.
ventional to speak of pumping when liquids are Reciprocating Pumps. In the reciprocating

involved, and the devices are usually called pumps. pump, the rate of liquid delivery is a function of
It is sometimes necessary to transport slurries (mix- the volume swept out by the piston in the cylinder
tures of a liquid and a finely divided solid) using and the number of strokes the piston makes per
specially designed pumps. Gases have lower density unit time. For each stroke of the piston, a fixed
and viscosity and greater compressibility, so volume of liquid discharged from the pump. The
is

different devices are used to move them. actual delivery may

Depending upon the pressure increase desired, fans,
blowers, and compressors are used to add energy to
gases.
The energy supplied to the fluid-motive
devices chemical processes often is a major
in
fraction of the energy consumption of the process.
To conserve energy, it is important to choose
properly sized pumps and compressors in designing
the process. Many fluid-motive devices are avail-
able; only a few representative examples of the
more commonly used types are discussed in this
chapter.
PUMPING LIQUIDS
Most pumps fall into one of two major
tions, positive-displacement pumps and
be less than the swept-out
volume because of leakage past the piston and
failure tofill the cylinder. A volumetric efficiency
may be defined as the ratio of actual discharge to
that based upon piston displacement. For well-
maintained pumps, the volumetric efficiency is at
least 95 percent.
A mechanical efficiency may be defined as
the energy supplied to the fluid divided by the
energy supplied to the pump. This efficiency is less
than 100 percent because of losses caused by
mechanical friction and fluid friction.
Two types of reciprocating pump are shown
in Figures 21.1 and 21.2. Power-driven pumps,
such as that shown in Figure 21.1, are driven by an
electric motor, a belt, or a chain, through a driving
classifica- gear and a crankshaft, to convert the rotary motion
centrifugal to linear motion. Steam-driven pumps use a steam

580
 FLUID MOTIVE DEVICES 581

Air
»^~ chamber

Driving
gear

Cylinder

Figure 21.1. A power-driven double-acting simplex pump. Power is supplied by a

belt on the drive wheel shown behind the driving gear. The driving gear and
crankshaft convert the rotary motion of the drive wheel to the back-and-forth
linear motion of the piston rod. The inlet and discharge valves open and close as the
piston moves back and forth. These valves are check valves that allow flow in only
one direction. As the piston moves toward the right, the inlet valve is opened, and
the discharge valve is closed at the left end of the cylinder; the reverse is true at the
right end of the cylinder. Thus, the left end of the cylinder is filling with liquid and
the right end is discharging liquid. As the piston then reverses direction and moves
toward the left, the valves reverse their positions; the left end discharges as the right
end fills. (Courtesy Novo Pump and Engine Co.)

cylinder such as shown in Figure 21.2, with the cating pump varies with time because of the
piston on the same shaft as the pump pistons, periodic nature of the piston movement, shown
as
which aremost commonly double-acting. The in Figure 21.3. For a single-acting simplex pump,
pump Figure 21.1 employs a single cylinder, but
in liquid issuesfrom the discharge valve until it closes
uses the cylinder volume on both sides of the near the end of the stroke, when the piston stops
piston to deliver liquid on both the forward and and reverses. At this point the intake valve opens
backward stroke. It is therefore called a double- and the cylinder fills, but there is no discharge
acting simplex pump. The pump of Figure 21.2 during intake, and an intermittent flow results, as
employs two cylinders with separate plungers in shown in Figure 21.3a. The simplex double-acting
each. (External tie-rods drive the right-hand pump almost eliminates the periods of zero dis-
plunger from the double-acting steam cylinder.) charge, as shown in Figure 21.3/?, but the flow still

Fluid is pumped only on the forward stroke of drops to zero just as the piston reverses itself. The
each plunger, so this unit is a single-acting duplex air chamber shown in Figure 21.1 provides a surge
pump. The plunger is a variation on the piston used volume so that the discharge does not drop all the
in higher-pressure applications, in which the way to zero as the piston reverses itself. A still
packing rings are held in the cylinder wall rather more uniform flow results from a duplex double-
than being carried by the piston. acting pump, as indicated in Figure 21.3c. In this
The rate of discharge of liquid from a recipro- case, the discharge of one cylinder is displaced half
582 APPLICATIONS TO EQUIPMENT DESIGN

Valves

Figure 21.2. Steam-driven tandem duplex plunger pump. The tie-rods that drive
the right-hand plunger are shown only in part. Both plungers are driven by the
piston rod powered by a steam-driven piston. The high-pressure steam alternately
enters and leaves either side of the cylinder, thus driving the piston back and forth.
The steam flow is controlled by the sliding valve, so that inlet and discharge occur
alternately at either end of the cylinder. (Courtesy Worth ington Pump Inc.)

a stroke from the other, so the total discharge (the small compared to centrifugal pumps. Liquids con-
solid curve in the figure) undergoes less fluctua- taining abrasive solids can damage the machined
tion. surfaces of the cylinder and piston. These pumps
Reciprocating pumps have long been used in may be used for metering of moderate flows,
many applications, including pumping of boiler because of their positive-displacement character-
feed water, and mud. Reciprocating pumps can
oil, istic.

deliver the highest pressure of any type of pump. Rotary Pumps. There are several varieties of
On the other hand, their capacities are relatively rotary positive-displacement pumps, all of which
operate on basically the same principle. The
rotating elements of the pump create a reduced
pressure on the inlet side, thereby allowing ex-
ternal pressure to force liquid into the pump. As
Simplex single the elements rotate, the liquid is trapped between
acting
Discharge Intake
faj the elements and the pump casing. After further
*—<
rotation, the liquid is forced out of the discharge
side of the pump. The flow rate of liquid from a
Simplex double
acting rotary pump is a function of its size and speed of
Full stroke
*—
Full stroke rotation, but it is only slightly dependent upon the
discharge pressure, within the limits of pump
Total flow
design. Rotary pumps deliver at nearly constant
flow rates, in contrast to the pulsating flow of
--vr-^ Duplex double
N/
.Cylinder *1>.
V N/
-Cylinder
\/
*2
V \ actin e
(el
reciprocating pumps.
Rotary pumps are used on liquids of any
^

viscosity, so long as they do not contain abrasive

Time
solids. They are particularly effective with high-
Figure 21.3. Discharge curves for reciprocating pumps. viscosity liquids, including greases, molasses, and
 FLUID MOTIVE DEVICES 583

H.

Figure 21.4. Flow pattern for an externa! gear pump. The two gears are rotating in
opposite directions, as shown in the figure. A lighter-colored liquid, drawn in from
the left, is trapped between the rotating teeth and the pump casing, so that it is

drawn toward the pump outlet on the right. (Courtesy Roper Pump Company.)

paints. Rotary pumps operate in moderate pressure

ranges and have small to medium capacities. They
are often used for metering liquids. A few
representative rotary pumps are described below.
The simplest rotary pump is the gear pump.
An example of an external-gear pump is given in
Figure 21.4, and typical performance curves for
this type of pump are given in Figure 21.5. An
internal-gear pump is shown in Figure 21.6.
Screw pumps may have one, two, or three
screws turning along their axis, with liquid flowing
between the screw threads and the casing. A
double-screw pump is shown in Figure 21.7.
An interesting variation of the screw pump is
the "traveling cavity" pump shown in Figure 21.8. 60 120 180 240 300
This pump consists of a rotor that revolves within a Discharge pressure, psi

stator, executing a compound movement; the rotor Figure 21.5. Performance characteristics of an external
is revolved about its axis while the axis itself travels gear pump.

Idler
gear

(at (b) <c) (d)

Figure 21.6. Flow pattern for an internal gear pump. The internal gear pump has a
single powered rotor, a stationary crescent, and an idler gear that rotates as required
by the main rotor. To illustrate the flow pattern, a dark liquid is shown entering
from the left. It fills the space available between the counterclockwise-rotating
main rotor, the idler, and the casing, and it moves toward the discharge. At the
discharge, the dark liquid is completely forced out by a close mesh of the rotor and
the idler gear. Discharge is essentially continuous. (Courtesy Viking pump Co.)
584 APPLICATIONS TO EQUIPMENT DESIGN

Figure 21.7. Double-screw pump. The closely meshed screw teeth rotate in oppo-
site directions. The liquid enters at the bottom and is moved from the ends of the
screws to the middle, where it is discharged upward. (Courtesy Sier Bath Pump
Division, Gilbarco. Inc.)

in a circular path. The rotor is a true helical screw, continuously moved toward the discharge. This
and the stator has double internal helical thread
a pumping action gives continuous flow with low,
pitched opposite to the spiral on the rotor. In each smooth, and uniform velocities. The pump is
complete revolution of the rotor, the eccentric capable of handling highly viscous materials. Such
movement enables the rotor to contact the entire materials as chocolate, greases, plaster, cake icings,
surface of the stator. The voids between the rotor potato salad, and putty are easily pumped.
and stator will have entrapped material that is

Inlet

Drive
shaft

Rotor

Discharge

Figure 21.8. Traveling cavity pump. (Courtesy Robbins & Myers Inc.
 FLUID MOTIVE DEVICES 585

Example 21.1. It is necessary to pump a bination of the actions of these basic types are
constant flow of a liquid with density and viscosity available, characterized as "mixed flow" impellers.
similar to water into a reactor at a rate of Some typical configurations are shown in Figures
90gal/min. The pump must operate against a 21.9 and 21.19.
pressure of 200 psi, as determined by an energy For handling corrosive, abrasive, or shear-
balance on the flow system. A pump with the sensitive fluids, detailed alterations of materials,
characteristics shown in Figure 21.5 is available, clearances, and flow paths are possible but are
with a variable-speed drive. At what speed should usually designed by specialists. For example, it
the pump be operated? What horsepower would be would be an unusual challenge to a chemical
required to maintain flow? engineer to specify a pump to handle blood with-
out damaging red corpuscles.
Solution. When plotted on Figure 21.5, the
The impeller is the heart of the centrifugal
point representing the required flow and pressure
pump. It consists of a number of curved vanes or
falls between the 400 and 600 rpm curves. The
blades shaped in such a way as to give smooth fluid
variation of capacity with speed seems to be non-
flow between the blades. Common impellers are
linear, as evidenced by comparing the change from
pictured in Figure 21.9. In the straight-vane,
200 to 400 rpm with the change from 400 to single-suction, closed impeller (Figure 21.9a), the
600 rpm. With this in mind, interpolation between surfaces of the vanes are generated by straight lines
400 and 600 rpm yields a speed of about 550 rpm parallel to the axis of rotation. The double-suction
for the case at hand. Interpolating on the horse- impeller (Figure 21.96) is, in effect, two single-
power curves yields a power of 21 hp at a discharge suction impellers arranged back to back in a single
pressure of 200 psi. casing. For handling liquids containing stringy
materials and soft solids, the impellers of Figures
Centrifugal Pumps 21.9a and 21.96 are likely to become clogged be-
The centrifugal pump is widely used in the process cause of restricted flow passages. A nonclogging
industries because of its simplicity of design, low impeller (Figure 21.9c) is designed to have large
initial low maintenance, and flexibility of
cost, flow passages to lessen the possibility of clogging.
application. Centrifugal pumps have been built to Open impellers (Figure 21. 9d) have vanes attached
pump as little as a few gallons per minute against a to a central hub and are well adapted for pumping
small head and also as much as several hundred abrasive solids.The semiopen impeller (Figure
thousand gal/min against a pressure of several 21. 9e) has a single shroud, and the closed impeller
hundred psi. In its simplest form, the centrifugal has shrouds on both sides of the vanes. The semi-
pump consists of an impeller rotating within a closed impeller shown has pump-out vanes located
casing. Fluid enters the pump near the center of on the back of the shroud whose purpose is to
the rotating impeller and is thrown outward by reduce the pressure at the back hub of the impel-
centrifugal action. The kinetic energy of the fluid ler. The mixed-flow impeller (Figure 21 .9/0 is a
increases from the center of the impeller to the tips design in which there are both a radial component
of the impeller vanes. This velocity energy is con- and an axial component of flow.
verted to a pressure as the fluid leaves the impeller Centrifugal-pump casings may be of several
and enters the volute or diffuser. designs, but the main function is to convert the
Not all centrifugal pumps have the purely velocity energy imparted to the fluid by the im-
radial-flow pattern above. Detailed
described peller into useful pressure energy. In addition, the

description of the modifications and elaboration of casing serves to contain the fluid and to provide an
the specific advantages and applications are outside inlet and outlet for the pump. Casings may be of

the usual activity of chemical engineers, but aware- either the volute type or the diffuser type. Figure
ness of the variations is needed. The turbine pump 21.10 shows a Here the impeller
volute casing.
usually has rather small straight vanes and is driven discharges continuously expanding flow
into a

at high speed to produce relatively high pressure at area. This increase in flow area causes the fluid
moderate flow rates. Axial-flow pumps and com- velocity to decrease gradually, thereby reducing
pressors consist of multibladed propellers, gener- eddy formation. By this means, most of the velo-
ating a purely axial flow of large-volume, relatively city energy is converted to pressure energy with
low-pressure fluids. low turbulence losses.
For particular balances of flow rate and The diffuser pump casing (Figure 21.11) has
pressure, impellers shaped to use some com- stationary guides that offer the liquid a widening
586 APPLICATIONS TO EQUIPMENT DESIGN

(a) (b)

Pump-out vane

(O fej

(d)
If)

Figure 21.9. Centrifugal-pump impellers, (a) Straight vane single-suction closed

impeller, (b) Double-suction impeller, (c) Nonclogging impeller, {d) Open impeller,
(e) Semi open impeller, (f) Mixed-flow impeller. (Courtesy Worthington Pump, Inc.)
 FLUID MOTIVE DEVICES 587

-Volute
Drive shaft

Figure 21.10. Volute centrifugal-pump casing. Figure 21.11. Diffuser centrifugal-pump

casing.

Volute

Liquid

Liquid out

Figure 21.12. Cutaway view of a centrifugal pump. The liquid flows in past an
inducer and into the impeller, where it is thrown outward into the volute, from
where it flows out of the pump. The inducer is actually a small axial-flow impeller
that effectively reduces the required suction pressure to the pump, which is seldom
incorporated except in pumps working with a low inlet pressure.The drive shaft, to
which a motor would be attached, must be sealed to avoid leakage of the pump
fluid.Such seals must be adjusted or replaced from time to time. (Courtesy
Worthington Pump, Inc.)
588 APPLICATIONS TO EQUIPMENT DESIGN

path from impeller to casing. The diffusers serve

the same purpose as the volute, and both types of
pumps have about the same efficiency. The major
applications of the diffuser centrifugal-pump casing
Ai
are in multistage high-pressure pumps and with
mixed-flow impellers.
Motor
Because of their high speed, centrifugal
pumps must be carefully designed for smooth
mechanical operation. Figure 21.12 is a cutaway
drawing of a typical centrifugal pump, showing
mechanical components as well as the flow path.
Historically, the seal around the shaft has consisted
of compacted, moderately soft, nonabrasive
material that can be formed into rings in an annu-
lar chamber where the shaft leaves the pump body.

A follower ring of one of several designs has been

provided to exert a controllable pressure on the
packing rings. Asbestos, impregnated with graphite,
was probably the favorite packing material. An
experienced mechanic could quickly and effectively
repack a pump when the packing was worn. In
recent years, mechanical seals have been developed
in which two nonabrasive and nongalling surfaces,

ground extremely smooth and mating precisely,

provide a seal that has some advantages over the
classical packing gland. Replacement of mechanical
seals cannot always be done as a field repair, which
is the major disadvantage of their use.

There are unbalanced forces on the impeller,

hence, adequate bearings must be used so that the
shaft can rotate evenly and with minimal strain and
vibration.
A centrifugal pump with one impeller is
Impellers

referred toa as pump. If the total

single-stage
head-capacity combination to be developed is
greater than can be developed from a single
Liquid
impeller, multistage operationmay be used. Multi- inlet
stage pumps may be thought of as several single-
stage pumps on one shaft, with flow in series. In Figure 21.13. Two-stage turbine pump for
effect, discharge from a single-stage pump is fed submerged service. This type of pump is
into the suction side of a second stage where the used in water wells, among various other
discharge pressure of the first stage is preserved. applications.The length of the drive shaft

The fluid after entering the second stage will have would depend upon the depth of the well.
(Courtesy Worthington Pump, Inc.)
added to it the pressure energy developed in the
stage, and so on.
A cutaway view of a two-stage turbine pump design, and the suction conditions. The best way to
is given in Figure 21.13. This unit employs verti- describe the operating characteristics of a centri-
cally mounted impellers and is installed with the fugal pump is through the use of characteristic
impeller casings submerged in the liquid to be curves (Figure 21.15). This figure shows the inter-
pumped. Figure 21.14 illustrates a six-stage centri- relation of discharge pressure or head (H), capacity
fugal pump with a relatively low capacity and a (Q), efficiency (77), and power input (P) for a given
high discharge pressure. A centrifugal pump usually pump at a particular speed. The H—Q curve shows
operates at constant speed, and the capacity of the the relation between total head and capacity. The
pump depends only upon the total head, the pressure increase created by a centrifugal pump is
 FLUID MOTIVE DEVICES 589

Bearing

Bearings

Impellers

Figure 21.14. Six-stage centrifugal pump. This pump delivers a relatively low flow
rate at a high pressure, as is needed in the case of boiler feed water. It can be seen
that the outflow from one impeller becomes the inflow for the next impeller.
(Courtesy Worthington Pump, Inc.)

commonly expressed in terms of feet of the fluid that stable operating characteristics are available.
flowing. The discharge head, when reported as feet The P—Q curve of Figure 21.15 shows the relation
of fluid flowing, is independent of the density of between power input and pump capacity. The r\—
the fluid. In Figure 21.15, the head increases con- curve relates pump efficiency to capacity. For a
tinuously as the capacity is decreased; this type of pump having the characteristics of Figure 21.15,
curve is referred to as a rising characteristic curve. maximum efficiency would occur at a capacity of
A is one in
stable head-capacity characteristic curve 2500 gal/min and a total head of 80 ft.
which only one capacity can be obtained at any When the pump is capable of being operated
one head. Pump selection should be made such at variable speeds, characteristic curves such as

150 - - 90

-
«^^~ -
80

70£
5"

0)

flowing

60^
o
'2 110 - 50
of -
8 B- 9^
feet
8 100

| 80 s^ 90
|
2o 70 ^v - 80 |
g
- 70
af 60 |
o>
50 60
P-Q- |
-O
40 50^-
. .
i i 1

40
10 15 20 25 30 10 15 20 25 30
100 gal/min Q, capacity in 100 gal/min
Q, capacity in

Figure 21.15. Characteristic curves for a centrifugal pump. Figure 21.16. Effects of speed change on pump
(Courtesy Worthington Pump, Inc.) characteristics. (Courtesy Worthington Pump, Inc.)
590 APPLICATIONS TO EQUIPMENT DESIGN

Figure 21.16 are obtained. A higher speed of rota- suction inlet (point 2) is

tion gives a higher capacity at the same pressure

and requires more horsepower to supply this
P2
+ ^_ *
— + z, —9 IF (21.2)
increased flow of liquid. P 4cc
2g P 9c
Usually a given pump casing can accommo- From Equation 21.2, the pressure at the suction
date impellers of several different diameters. Figure inlet is
21.17 shows the effect of changing impeller dia- 2
9 pvi
meter on the pump characteristics. P2 =Pi+pzi pZF--^- (21.3)
9c 2g c

Example 21.2. A pump with the character- At the eye of the impeller, it is reasonable to
istics given in Figure 21.17 is to deliver assume that the pressure than at the
will be less
350 gal/min at a head of 80 ft. What size impeller suction inlet. This pressure
can be difference
should be used? What power will be required? assumed to be related to the velocity at the eye by
the expression AP = $v 3 2 /2g c where is a ,

Solution. The 1750-rpm pump is appropriate. pressure-drop coefficient characteristic of pump

When the required head and capacity are plotted, geometry and v 3 is the fluid velocity at the eye.
the point falls between the curves for the 9-in. and The net positive suction head (NPSH) is de-
the 10-in. impeller. The 9-in. impeller can deliver fined as the difference between the static head at
only 175 gal/min against a head of 80 ft, so it is the suction inlet and the head corresponding to
necessary to use the 10-in. impeller. A control the vapor pressure of the liquid at the pump inlet.
valve in the system would dissipate the excess Thus combining this definition with Equation
pressure energy. The required power for the 10-in.
21.2.
impeller would be 1 1 hp, by extrapolation.
P\
Cavitation. When a centrifugal pump is
NPSH=| — + t^ —= [- + Z1—-ZF
1-7
operating at a high capacity, low pressures may \ D 2Qr- I o \ P 9c / P
develop at the impeller eye or vane tips. When this (21.4)
pressure falls below the liquid vapor pressure,
vaporization may occur at these points. The Cavitation is probable if the total head at the
bubbles of vapor formed move to a region of high suction minus the acceleration head from there to
pressure and collapse. This formation and collapse the eye of the impeller is equal to or less than the
of vapor bubbles is called cavitation. The bubble vapor pressure; that is
— 2
collapse is likely to be so quick that the liquid will Py Q
— + zi-
9 V2
Vo V3
" 2/:
2
_P
"o2-9c "*f
I

hit the vane with extreme force and is likely to

P 9c 2g c
-
p
(21 5)
-

gouge out small pieces of the impeller. In addition

to this pitting of the impeller, noise and vibration Equation 21.5 may be rearranged to express
will be created. Cavitation may be reduced or incipient cavitation,
eliminated by reducing the pumping rate. If cavita- 2 2
(P\ P v
tion is not reduced or eliminated, serious mechani-
9
(- + Z1- -IF \
)- — = ^- + <t>^-
v 2 v2
(21.6)
cal damage to the pump is likely to result. 2gc 2 9c
Net Positive Suction Head. Cavitation may and from the definition of the net positive suction
be minimized by paying attention to the design of head.
the pump installation on the suction side. In Figure
2 2
21.18, Pi is the pressure acting on the surface of *2 V3
NPSH = + (21.7)
liquid inside the tank, and Zi is the height of the 2g c 2g c
liquid surface above the pump center line at the
suction inlet. If Y.F is the total friction loss in the Therefore, to avoid cavitation, the NPSH
piping system, application of the energy balance must be greater than the maximum possible value
2
for a unit mass gives of W 2 2 l2g c + (<pv~3 /2g c ). The values of v 2 and v 3
)

will depend upon the flow rate so that some con-

— + z - + — +ZF
2 2
trol over cavitation may be achieved by variation in
—+ zi
V\
= —
Pi 9
2
v 2
(21.1)
flow rates, as previously mentioned. The value of
2gc P 9c 29c
the required NPSH for the particular pump being
If the datum plane is taken at z 2 and Vi is negli- used may be obtained from the pump manu-
gible compared with v 2 the total head at the, facturer.
 FLUID MOTIVE DEVICES 591

900 gal/min

m 3 /hr

Figure 21.17. Centrifugal pump characteristics. The pump has an inlet diameter of
4 in.and an outlet diameter of 3 in. Its 10-in. casing can contain impellers of
6, 7, 8, 9, or 10 in. as shown. The pump is normally operated at one of two speeds,
thereby producing two different characteristic curves: (a) 1750 rpm or
(b) 3550 rpm. (Courtesy Goulds Pumps, Inc.)
592 APPLICATIONS TO EQUIPMENT DESIGN

Figure 21.19 shows that the normal range in

(1)
121
specific
pumps
speeds encountered in single-suction
for various impeller designs is between 500

(2)/(3
— *- Discharge
and 15,000. On the same figure is shown the type
At impeller eye
Suction of pump characteristic that is associated with
inlet
pumps of various specific speeds.
Figure 21.18. Suction conditions for a cen-
trifugal pump. Example 21.3. It is necessary to pump
2500 gal/min against a head of 50 ft. What type of
Specific Speed. Specific speed is the speed
pump might be used?
in revolutions per minute at which a theoretical
pump geometrically similar to the actual pump Solution. This relatively high rate and low
would run at its proportioned to
best efficiency if head suggests some type of centrifugal pump. A
deliver 1 gal/min against a total head of 1 ft. The reasonable operating speed for such a pump might
specific speed serves as a convenient index of the be assumed to be around 1800 rpm. Equation 21.8
actual pump type, using the capacity and head then gives
obtained at the maximum efficiency point.
Specific speed may be determined from ny/Q 1800V2500
N* = 4800
Equation 21.8 for a single-stage pump or one stage #0.75 50 075
of a multistage pump. This equation results from
This value of specific speed falls within the range
model theory and dimensional analysis.
of mixed-flow pumps in Figure 21.10. Therefore, a
nVQ mixed-flow centrifugal pump might be used.
(21.8)
H 0.7 5
Other Pump Types
where
Ns
= specific speed, rpm Only a few of the more important pump types
n = actual pump speed, rpm have been covered in the previous paragraphs.
H = total head per stage, ft Some other types are discussed briefly below.
Q = pump capacity in gal/min at speed n The diaphragm pump is a positive-displace-
and total head z ment pump that depends upon the movement of a

Efficiency,

Power, bhp

Radial --flow field Mixed-flow field Axial-flow field

Figure 21.19. Characteristic curves and specific speeds for various impellers. Impel-
lers that tend to promote radial flow, as in a standard centrifugal pump, operate at
the lowest specific speed (up to 4200). Mixed-flow impellers, which produce both
axial and radial flow, operate at higher specific speeds (4200 to 9000). Axial flow
impellers operate at the highest specific speeds (above 9000). (Courtesy
Worhtington Pumps, Inc.)
 FLUID MOTIVE DEVICES 593

STAINLESS STEEL AIR DISTRIBUTION SUCTION

VALVE SEATS VALVE MANIFOLD CHECK VALVES

D<"i_5* Si

;.APHRAGM
CONNECTING
SHAFT DISCHARGE
MANIFOLD

Figure 21.20. A diaphragm pump. This model of diaphragm pump

employs two diaphragms connected by a shaft. There are air chambers
behind each diaphragm. Air is alternately charged to and then exhaus-
ted from each chamber, so that the shaft has a back-and-forth move-
ment, and one liquid chamber is filling while the other is being dis-
charged. Appropriately placed check valves are used to prevent
discharge during the filling cycle, and vice versa. The flow from the
pump is pulsating, but it can be smoothed by using a surge tank.
(Courtesy Warren Rupp Company.)

diaphragm to achieve fluid flow. The diaphragm The acid egg, an inexpensive pump involving
may be actuated mechanically or pneumatically. no moving parts, consists of a tank from which
The back-and-forth movement of the diaphragm liquid can be forced by gas pressure. Typically,
produces a pulsating flow typical of a reciprocating compressed air is used to force the liquid out of
pump. Figure 21.20 shows one type of pneumati- the tank into a pipeline. The operation of a single
cally operated diaphragm pump. The diaphragm tank would be intermittent, but several could be
pump eliminates all packing and seals exposed to combined to give a more nearly continuous flow.
the liquid being pumped. The diaphragm itself can be The electromagnetic pump works on the same
of any flexible material (metal, rubber, or plastic)
that is resistant to the liquid. Diaphragm pumps are
used to pump abrasive slurries and very viscous
liquids, although they will handle liquids of all Jet nozzle

viscosities.
The jet pump uses the momentum of a high- Driving
fluid
velocity fluid stream to impart momentum
to a
second stream, by actually mixing the two. An Suction^
box
-TT
Entrained \Mixing tube
'Diffuser
example of a jet pump is shown in Figure 21.21. flow or throat

Jets, or ejectors, have low efficiencies and can Entrained

produce only small pressures. Mixing the propelling fluid
—
fluid and the fluid being pumped may be disadvan-
Figure 21.21. Jet pump principle. The driving fluid enters
tageous, but one advantage is the complete lack of from the left through a nozzle that imparts high velocity.
moving parts in jet pumps. An example of applica- This jet penetrates into a stagnant or slow-moving fluid, so
tion is the use of steam as the motive fluid to that it transfers its momentum by entraining and acceler-
pump boiler feed water. It is also used to pump ating the fluid. The diffuser at outlet converts kinetic
mixtures of solids and liquids. energy into pressure energy, after mixing is complete (2).
594 APPLICATIONS TO EQUIPMENT DESIGN

Oversized pumps cost more initially and consume

more energy than necessary during operation.
Head. The head (or pressure) a pump must
develop is determined by a careful analysis of the
piping system using the methods of Chapter 20.
Friction losses and changes in the velocity, pres-
sure, and static heads must be taken into account.
The energy balance of Chapter 20 makes it possible
to calculate the pump work (— W f with reasonable )

accuracy. Nevertheless, some safety factor for un-

certainties should be included. Possible fluctu-
ations in suction or discharge pressures required
should be considered. If a temperature variation is

Figure 21.22. Electromagnetic pump. This anticipated, its effect on liquid properties should
schematic representation shows the principle be taken into account. Here again, care should be
of the electromagnetic pump. The conduct-
taken to avoid an overly cautious design that speci-
ing liquid carries a current that is at right
fies too great a work delivery, because of the
angle to an imposed magnetic field. This
resulting unnecessarily higher equipment and
results in a force that causes the liquid to
operating costs.
flow.
Liquid Properties. All types of pumps result
in more or less frictional dissipation of mechanical
principle as the induction motor, as shown in
energy; this means that for a viscous fluid more
Figure 21.22. It is useful only with liquids of high energy must be supplied. For a perfect positive-
electrical conductivity, such as liquid metals. It has displacement pump, there would be no other
no moving parts and hence no seals.
effect. For a real pump, flows are more sluggish,
resulting in partially filled cylinders or velocities
Pump Selection from the tip of an impeller element lower than
With the large variety of pump types available, the would be attained by a nonviscous fluid. The pres-
selection of a pump for a specific task might sure developed in a centrifugal pump is a function
appear to be a complex endeavor. In fact, many of the kinetic energy of the fluid leaving the im-
pump selections for routine, previously encoun- peller. Increasing viscosities result in decreased
tered situations are quite straightforward. In such velocities and therefore lowered pressure energy;
cases, the design engineer can select the appro- increasing density results in higher pressure energy,
priate pump for a specified task from manu- leading to the convenient fact that discharge pres-
facturers' catalogs. In unusual cases, such as highly sure is independent of density of the fluid if it is

viscous or abrasive liquids or extreme pressures, the
engineer should consult with the manufacturer,
perhaps asking for a recommendation (and guaran-
tee) of apump for the application.
Whether the design engineer selects the pump
or asks for a manufacturer's recommendation,
detailed information on the capacity and head
required and on the properties of the liquid to be
pumped must be immediately at hand. A general
consideration of these factors is given below.
Capacity. The flow rate required by the
process itself determines the capacity of the pump
needed. Not only steady-state flow, but also
expected surges must be considered. Some safety
factor for uncertainties in
cluded. Possible increase in flow rate because of
process expansion might be taken into account.
Care should be taken not to overestimate needed
capacity by putting in too large a safety factor.
design should be
expressed terms of the height of a column of
in
fluid it support - the head. For example, an
will
oil of 200 cp viscosity reduces pressure and ca-
pacity about 10 percent and increases power by
about 25 percent relative to water. If high-viscosity
liquids are to be pumped, appropriate performance
data should be obtained from the manufacturer.
Corrections for high viscosity are available in
Reference 3.
If the liquid contains solids in suspension, as
in sewage, paper stock, slurries, and some foods,
special pump designs may be necessary. For
example, one of the major problems encountered
in the development of the artificial heart (which is
in- a small, specialized pump) is the destruction of red
blood corpuscles, The corpuscles are suspended in
the liquid plasma, and many conventional pump
designs lead to corpuscle attrition; hence, special,
very gentle pumps had to be developed. If very
 FLUID MOTIVE DEVICES 595

abrasive solids are suspended, centrifugal pumps sures without a flow controller. Instead, a positive-
lined with rubber, plastic, or hard metal may be displacement pump of the rotary or reciprocating
required. type would be used. The seal problems of centri-
Selection of materials of construction must fugal pumps sometimes dictate use of seal-less
take into account the corrosive properties of the pump types.
pumped liquid. If the liquid is unusually volatile at Pump Sizing. The pump size is determined
the temperature of operation, care must be taken by pressure developed and capacity requirements.
to avoid cavitation or vaporization on the suction Once the type of pump is chosen, the final selec-
side of the pump. Pumps in chemical processes tion of the size may be accomplished by referring
sometimes must operate at extremes of tempera- to manufacturers' catalogs. Usually, each model of
ture and pressure, so special designs may be re- pump is available in a range of sizes for which the

quired. performance characteristics overlap somewhat. For

Type of Pump. Once the pumping job is example, Figure 21.23 shows the operating ranges
fully specified, the type of pump
can be selected. of a family of centrifugal pumps. A diagram of this
For example, a high-pressure, low-flow application type would be used to choose the pump size appro-
may suggest a reciprocating pump. A low-pressure, priate for the capacity and head required by the
high-capacity job may suggest an axial-flow pump. job to be done. Then the detailed head-capacity
General pump characteristics discussed earlier are efficiency curves would be examined for that
summarized in Table 21.1. The numerical values pump size to choose the best impeller diameter.
are intended to be only representative; exceptions Figure 21.176 gives the detailed curves for one of
can be found. the pump sizes of Figure 21 .23. Where the required
Therenot always a clear-cut basis for choice
is head and capacity on the border between two
fall

among pump types. It is necessary to balance capa- pump sizes, it may be desirable to examine the
bilities and limitations, including durability and detailed curves for both before making a final
previous experience with a given manufacturer's choice. Reference 4 deals with pump sizing in
pumps. Centrifugal pumps, in all their varieties, are piping systems.
so versatile that they are more widely used than
other types, but they have limitations. Because Example 21.4. It is necessary to pump a
flow varies with discharge pressure, they are not liquid with properties similar to water at a rate of
used for direct metering or constant flow rates 300 gal/min against a head of 70 ft. Recommend a
under conditions of variable input or output pres- pump type and size.

Table 21.1 PUMP CHARACTERISTICS

Centrifugal

Standard Turbine Propeller Rotary Reciprocating

(Radial Flow) (Mixed Flow) (Axial Flow) (Gear or Screw) (Piston or Plunger)

Head High, single stage — Intermediate, up to Low, up to 60 ft Intermediate up Highest available

(or discharge up to 600 ft; 200 ft to 600 psi up to 100,000
pressure) multistage — psi

up to 6000 psi

Capacity Low (100 gal/min) Intermediate, up to High, up to 100,000 Low (1 gal/min) Intermediate up
(or delivered to very high 16,000 gal/ gal/min to intermediate to 500 gal/min
flow rate) (200,000 gal/ min (500 gal/min)
min)

Liquids handled Clean or dirty High solids Abrasive Up to high Clean; no solids
contents viscosity;
nonabrasive

Metering or flow- No No No Yes Yes

control
capability
596 APPLICATIONS TO EQUIPMENT DESIGN

1750 rpm

Figure 21.23. Operating ranges of pumps. Each labeled area

a series of centrifugal
represents the head/capacity region which a specific size pump can operate. The
in

model number is given for each size, where the first number is the diameter of the
discharge line, the second is the diameter of the suction line, and the third is the
maximum diameter of impeller that will fit in the pump casing. All diameters are in
inches. Figure 21.17a gives detailed operating characteristics of one size pump in

this series. This pump's range overlaps the ranges of several smaller-size pumps in
this series, as can be seen by comparing the figures. (Courtesy Goulds Pumps, Inc.)

Solution. The type of pump can be suggested
by checking the specific speed. No pump speed is
given, so a reasonable value of 1800 rpm
used.
1800-
/V5 = = 1288
0.75
(70)
Examination of Figure 21.19 shows this specific
speed to be at the upper end of the range for the
standard centrifugal pumps, so the catalog of a
manufacturer of this type of pump is consulted
(Figure 21.23). The actual speed of 1750 for
these pumps would lower the calculated specific
speed somewhat. Examination of the figure shows
that size 3x4—10 can do the job. This is a pump
with a 3-in. discharge line, a 4-in. suction line, and
necessary to use a throttling valve in the line to use
up this extra head. This is, of course, wasteful of
will be energy.
COMPRESSING GASES
Gases are frequently involved in chemical pro-
cesses; they must be moved through pipelines and
process equipment using appropriate fluid motive
devices, just as process liquids must be pumped.
Fans, blowers, and compressors are used to impel
gases. Fans produce a small pressure increase (up to
0.4 psi); blowers produce a larger increase (up to
4.0 psi), and compressors produce the largest in-
crease (from 4.0 psi up to 60,000 psi in routine
applications). The distinction among these three
terms not always clear, nor should the pressures
is

a maximum impeller size of 10 in. Figure 21.17a is
now consulted to choose the proper impeller size.
The required head and capacity give a point that
falls slightly above the curve for the 9-in. impeller;
hence, that size could not quite handle the job.
The 10-in. impeller is selected, even though it can
deliver a head of 90 ft at 300 gal/min. It may be
given above be viewed as rigid boundaries among
the three types of equipment. Fans are sometimes
called blowers, and vice versa. Higher-pressure
blowers are frequently called compressors. In this
book, the term compressing gases refer to the
action of any gas motive device, regardless of the
pressure increase involved.
 FLUID MOTIVE DEVICES 597

The equipment used in compressing gases

often bears a resemblance to liquid
superficial
pumps, because the designs are based upon the
same physical principles. Hence, there are recipro-
cating, rotary, and centrifugal compressors, paral-
leling pump types. One type of gas motive device is
actually called a pump; the vacuum pump com-
presses a gas that below atmospheric pressure so
is

that can be discharged to the atmosphere.

it

The compression of gases differs from the

pumping of liquids in a number of ways. Gases
have lower viscosities and densities than liquids,
but gases are much more compressible. According
to the perfect-gas law, the density of a gas is
inversely proportional to the pressure at constant
temperature. Hence, as a gas is compressed, its
volume decreases substantially, in contrast to that
of a pumped liquid. When a gas is compressed in a
typical compressor, its temperature rises signifi- Figure 21.24. Inlet end of axial-flow fan, showing
cantly, as predicted by thermodynamic laws. propeller. (Courtesy Buffalo Forge Co.)
Hence, the gas must be cooled to maintain a
reasonable temperature. Although frictional losses large ventilating and industrial fans. Ventilating
in pumps cause a small increase in liquid tempera- fans are designed for large volume and low pressure
ture, the effect is usually negligible compared to increase (up to 130,000 ft 3 /mi n at 6 in. of
the temperature rise in gases. water (0.2 psi). Ventilating fans may be either
Fans and Blowers. axial or centrifugal, with the latter giving the
higher pressure increase. Pressures up to 12 in. of
Fans operate at pressures low enough that the
water (0.4 psi) are attainable with centrifugal fans.
compressibility of the gas may be neglected. In
Blowers provide higher pressures, up to 10 ft of
other words, inlet and outlet volumes of gases are
essentially equal, and fans are simply movers of
gas. Fans are classified as either radial or axial flow.
Radial-flow fans depend upon centrifugal force for
propelling the gas; whereas axial-flow fans impart
energy to the gas as it flows parallel with the
central axis of the fan. Figure 21.24 shows an
axial-flow fan, and Figure 21.25 illustrates a centri-
fugal fan.
Exact fan performance can be determined
only through testing, but a number of general
statements can be made. For a specific fan and
operating conditions, the volume of gas moved
varies directly with fan speed; the static pressure
varies as the square of the fan speed; the power
consumption varies as the cube of the fan speed.
Fan operating characteristics obviously depend
upon design. For axial-flow fans, the propeller
design is a prime variable. For centrifugal fans, the
blading design has primary influence on perfor-
mance.
Typical performance characteristic curves for
centrifugal and radial fans are given in Figure
21.26. Fan sizes range from small portable models Figure 21.25. Centrifugal fan. (Courtesy
used for circulating air in an automobile to very Buffalo Forge Co.)
598 APPLICATIONS TO EQUIPMENT DESIGN

oxygen and nitrogen is accomplished by distillation

et level
SoWM"!
—
|

^- of liquefied air produced by a refrigeration process

Sa??^-^. ^-<<5 involving the compression of air.
^
'^"v^ uS
To*
aipress level P>
As with pumps, compressors may be classed
""""
S< attc pre Sio
'*r^v as positivedisplacement or centrifugal. The
L*<< : positive-displacement category can be further
divided into reciprocating or rotary. The centri-
a01 x y^-r j<^ fugal category involves radial-flow and axial-flow
Z* 1
J"^'
E
— \ ^
<*.< t/
5 s
xT J designs.
s2
c^>
Specific rf, Jm
inn

NX Reciprocating Compressors.
6 8 10 12 The reciprocating compressor can furnish gas at
3
Capacity. '000 ft /min
(a)
pressures of a few pounds or at extremely high
pressures, such as 35,000 psig. The characteristic
features of reciprocating compressors are the same
| 120 =^°mo_powe • level as those of reciprocating pumps — a piston, a
" 110 — 1
~>.
1

cylinder with suitable intake and exhaust valves,

Specific
1 1

sound power
V and with drive. Single-stage or multi-
a crankshaft
level plus 40 dB
rJ"
1
^*~ stage operation is common, with double-acting
'C/ &$ chan, Ca',

rT>& '-
cylinder usage being general.
\ i.

Gas being compressed enters and leaves the

303
Si 2:

£ c
i S" 2
hp
^
^J*
^>
-
cylinder through valves set to be actuated
pressure difference between cylinder contents and
when the

outside conditions that desired. multistage

\ v^ is If

compression is used, it is general practice to cool

the gas between stages. Multistage compression
8 10 14
12
3
with cooling between stages saves power, compared
Capacity. 1000ft /min

(b)
to the same compression accomplished in a single
stage. Figure 21.27 shows a double-acting single-
Figure 21.26. Typical performance characteristics for fans
stage piston compressor; Figure 21.28 shows a
at constant speed, (a) Centrifugal ventilating fan. (b) Vane
double-acting two-stage compressor.
axial fan. (Courtesy Buffalo Forge Company.)
Discharge characteristics of reciprocating
3 compressors are similar to those for reciprocating
water (4 psi), and 200,000 ft /mi n for various
pumps. Compressor operation is ideally thought of
industrial applications.
as being reversible and adiabatic, and efficiencies
are reported relative to this basis. Thermodynamic
Compressors.
losses and fluid friction are grouped together as a
Compression of gases to high pressures is fre- compression inefficiency. Mechanical friction
quently encountered in the chemical process losses are termed mechanical inefficiency An over- .

industry. Many chemical reactions require high all compressor efficiency will be the product of the
pressures for favorable yields. For example, nitro- compression and mechanical efficiencies. The over-
gen and hydrogen have been compressed to all efficiency of most reciprocating compressors is
15,000 psi in the synthesis of ammonia. Recent 65—80 percent.
practice in the ammonia industry tends strongly
toward centrifugal compressors, even though the
top pressure in their optimum range is lower than Rotary Compressors.
for reciprocating machines. The disadvantages of Rotary blowers and compressors come in a variety
the reciprocating masses are avoided if possible. of types. The lobe blower is shown in Figure
3
Higher pressures promote the liquefaction of gases 21.29. It can deliver up to 15,000 ft /min at
for a variety of purposes. For example, the basic 12 psi, so it is a high-volume, intermediate pressure
refrigeration process involves the compression, device for gases. The helical screw compressor is
cooling, and subsequent expansion and heating of available over a wide range of pressures and capaci-
the refrigerant. The separation of air into pure ties, ranging from low-pressure units of 15 psi and
 FLUID MOTIVE DEVICES 599

-Drive wheel

Air out

z — Crank shaft
Valve

Connecting

Air in

Figure 21.27. A single-stage, doubling-acting piston air compressor. This particular

model can be obtained in a number of sizes, up to a 14-in. bore and an 11-in.
3
stroke, which delivers 521 ft /min at 100 psi, the maximum pressure attainable.
The cylinder is water-jacketed to remove heat of compression. The unit would be
belt driven from an electric motor in most applications. (Courtesy Ingersoll-Rand
Co.)

12,000 ft 3 /min to high-pressure units of 170 psi pumps do not. An impeller for a high-pressure com-
and 25,000 ft 3 /min delivered. Helical screw com- pressor is shown in Figure 21.31, and a cutaway
pressors in some cases use injected oil as a sealant drawing of a multistage centrifugal compressor is

and coolant. An example of the screws used in this shown in 21.32. In the compressor of
Figure
type compressor is pictured in Figure 21.30. The Figure 21.32, gas flows into the eye of the im-
sliding-vane compressor is used for a variety of peller, where it is accelerated radially, leaving at
applications, delivering 3000 ft 3 /min. at 125 psi in high velocity at the outer edge, and flowing into a
a two-stage model. diffuser that converts the velocity energy to pres-
sure energy. The gas is then directed into the eye
Centrifugal Compressors.
of the next impeller. Figure 21.33 shows a multi-
Centrifugal compressors are widely used in petro- stage centrifugal compressor using impellers of an
leum and chemical plants. For example,
refineries open type in an unusual design.
radial and compressors are used to deliver air
axial The axial-flow compressor bears little resem-
to the regenerators of catalytic crackers at flow blance to the axial-flow pump, although it is
3
rates of 170,000 ft /min at a pressure of 40 psi closely related to the gas turbine used in power
above atmospheric. Nitric acid plants use radial generation. Most aircraft turbine engines include an
3
compressors with airflows up to 80,000 ft /min at axial-flow compressor for the entering air. Each
pressures of 130 psi. stage of an axial-flow compressor consists of a row
Radial-flow centrifugal compressors bear a of blades attached to the rotating shaft and a row
superficial resemblance to centrifugal pumps, and of blades attached to the stator. In addition to the
the physical principles of operation are similar, but axial-flow stages, a compressor may include centri-
there are distinct differences in construction. Com- fugal compressor stages, as illustrated in Figure
pressor impellers are narrower and rotate at higher 21.34.
speeds. Multistage compressors usually include Axial-flow compressors are higher in effi-
some provision for cooling, vhereas multistage ciency than comparable centrifugal compressors,
600 APPLICATIONS TO EQUIPMENT DESIGN

Intercooler

Figure 21.28. Double-acting, two-stage compressor. In its largest size, this model
can deliver 650 3
ft /min at 100 psig, with cylinders of 14^ and 8% in. diameter,

driven by a 125 hp motor. To remove the heat of compression, both cylinders are
water-jacketed and an intercooler is used between stages. (Courtesy Ingersoll-Rand

Co.)

Discharge port

Figure 21.29. Lobe blower with two-lobe rotors. The

gas enters at thebottom and is forced out the top by
the rotating lobes. Precision design and manufacture
ensure close tolerances and minimize leakage between Figure 21.30. The heart of a helical-screw com-
the discharge and inlet. pressor. (Courtesy of Ingersoll-Rand Co.)
 FLUID MOTIVE DEVICES 601

Figure 21.31. Impellers for a multistage compressor. The impellers shown are
for a high-pressure compressor. They are fully shrouded and are produced by
welding. (Courtesy Dresser Industries.)

Suction
Discharge
Inlet guide vanes

Thrust bearing
ntegrally

Integrally supported supported

Main s haft
shaft bearings / Return bend \ \
/ ' Interstage . .
Balancing shaft seals bearings
Impeller eye Interstage-
Case
drain
labyrinth seals drum

Figure 21.32. A multistage centrifugal compressor. (Courtesy Dresser Industries.)

602 APPLICATIONS TO EQUIPMENT DESIGN

Cooler Moisture separator

Diffusor

Inlet

Intermediate- Discharge
pressure
impeller

(a)

Figure 21.33. Multistage integrated centrifugal compressor. This unique design

combines up to five-stage compression with interstage cooling in a compact
3 2
configuration. The smallest unit delivers 1170ft /min at 125 lb/in. whereas the ,

3 2
largest unit delivers 12,500 ft /min at 325 lb/in. The smallest unit has a power
.

rating of 300 hp, the largest 4000. The smallest unit weighs 10,350 lb, the largest
100,000.
Figure 21.33a shows the flow pattern for a four-impeller model. The gas flows
first through the lowpressure impeller, then through a cooler, then back through an
intermediate-pressure impeller, then through a cooler, a second intermediate impel-
ler, a cooler, and finally through the high-pressure impeller, cooler, and discharge.
The two intermediate-stage impellers and coolers are hidden from view; only the
first and last stage impellers and coolers are exposed in the cutaway.
Figure 21.336 shows the four impellers, driven by a common bull gear. The
lowpressure impeller is the largest, because it has the greatest volume of gas to
process, while the high-pressure impeller is the smallest, because it has the least
volume to process. The mass flow rate is, of course, constant through each impeller.
Advantages of simplicity of construction — and hence ruggedness — and compact-
ness make this a very popular compressor type. (Courtesy Ingersoll-Rand Co.)
 FLUID MOTIVE DEVICES 603

Figure 21.34. Multistage axial-flow compressor. This axial-flow compressor has

nine axial stages and one centrifugal in the low-pressure (right-hand) side and two
centrifugal stages in the high-pressure (left-hand) side. This particular machine
develops 1 00 psia air from atmospheric suction and has one intercooler between
low-pressure and high-pressure sections. (Courtesy Dresser Industries, Inc.)

and they are available in sizes that deliver higher Vacuum Service
flow rates (up to 860,000 ft 3 /min). They are In addition to the two devices described immedi-
smaller and lighter than centrifugal compressors,
ately above, other positive-displacement com-
but they may cost more for a comparable job. pressors can be used to achieve vacuums. A vacuum
Moreover, axial compressors have a more limited isany system pressure below atmospheric pressure
operating range and are more subject to corrosion.
(29.92 in. Hg). Pressures of 0.5 in. Hg can be
Typical operating characteristics for centri- obtained using the rotary or reciprocating com-
fugal and axial compressors are given in Figure pressors already discussed. Of particular use are the
21.35. sliding-vane compressor and the rotary-piston com-
pressor,which are capable of producing pressures of
Other Compressors.
0.001 mm Hg (with a lubricant having a vapor
Many other types of compressors are available for pressure equal or less than that value). For applica-
various applications. A few are mentioned here. tions requiring very low absolute pressures, on the
A rotary compressor employing a liquid ring order of 10
-12 mm Hg, a multistage diffusion
is shown in Figure 21.36. It is used primarily in pump is used. Table 21.2 lists the operating ranges
vacuum service, but it can be used for other low- for various vacuum pumps.
pressure applications.
Another device used primarily in vacuum Compressor Selection
service the jet ejector, a variation on the jet
is Selection of the proper gas motivating device for a
pump discussed earlier. The basic principle is illus- particular application is a complex process, re-
trated in Figure 21.21, and a two-stage version is quiring a thorough knowledge of the advantages
shown in Figure 21 .37. In this application, steam is and disadvantages of each type of compressor. Key
used to evacuate air. As many as six ejector stages factors include the gas flow rate, inlet and dis-
may be used to obtain absolute pressures as low as charge pressures, gas properties, and compressor
0.1 mm Hg. power consumption and cost. The complex
604
100%
110 ^Peea
/
i

£ 100
s
o /

S 90 90<»^ /
D
Q.
/
i

S 80 ^~7
Limit of /
— stable >>
8 :>
°o-
§ 70
to
operation 7
£
Q.

8
in
a>
60

50
/
j ^ t
X maximum
^60%
q. 40 f/
E /
8 30 50 Be
(5
/
/
20 1

10
/
^
y •

/„. ^

§ 120
tu

°:100 ee^.

^/
O.

Z 80 M$?£
fo 60
/L*-*'*'
80^ >

VI
$ 40 70
a.
I 20
^ fiffl
;>

-^ ;
-j—
"" 50SB

10 20 30 40 50 60 70 80 90 100 110 10 20 30 40 50 60 70 80 90 100 110

Compressor inlet capacity, percent Compressor inlet capacity, percent
(a) (b)

Figure 21.35. Operating characteristics for centrifugal and axial compressors, (a)

Centrifugal, single and multistage, (b) Axial.

IN THIS SECTOR, LIQUID MOVES ©IN — MOVES

THIS SECTOR, LIQUID INLET
OUTWARD —
DRAWS GAS FROM
INLET PORTS INTO ROTOR
INWARD
IN
COMPRESSES GAS
ROTOR CHAMBERS
CONNECTIONS
CHAMBERS
LIQUID

INLET
PORT
ROTATING
LIQUID
COMPRESSANT

w
C\ THIS SECTOR,
IN
COMPRESSED GAS DISCHARGE
ESCAPES AT DISCHARGE PORTS CONNECTIONS

KEY
ROTOR-one moving part

X CAST IRON BODY

H LIQUID COMPRESSANT (a)

Figure 21.36. Liquid ring vacuum pump and compressor.

(a) Principle of operation, (b) Disassembled view of rotor
and casing. (Courtesy of Nash Engineering Co.)
 FLUID MOTIVE DEVICES 605

High-pressure
steam inlet
2nd stage
Discharge

High-pressure
steam inlet

1st stage

Figure 21.37. Two-stage jet ejector. (Courtesy Croll-Reynolds Co., Inc.)

Table 21.2 OPERATING RANGES OF SOME Table 21.2 (continued)

COMMERCIALLY AVAILABLE VACUUM
PUMPS (5)

Operating Range, Operating Range,

Type of Pump mm Hg Type of Pump mm Hg

Reciprocating piston
4-stage 760-3 x 10~ 1
-stage 760-10
1
5-stage 760-5 x 10- 2
2-stage 760-1 2-10-2
Oil ejector (1 -stage)
Rotary piston oil-sealed
Diffusion-ejector 2-1 0- 4
1 -stage 760-10- 2
Mercury diffusion with trap
2-stage 7C0-10- 3 1 -stage 10~ -<10- 6
1

Centrifugal multistage (dry) liquid jet 760-200 1-<10- 6

760-1
0- 3 2-stage
Mercury Sprengel 10-<10- 6
3-stage
Water aspirator (18°C) 760-15
0- 4 Oil diffusion
Two-lobe rotary blower Roots type) 20-1
(
1 -stage 10- 1
-5x 10~ 6
Turbomolecular io- -io- 10
1

5x 10~ -10- 9
1
4-stage fractionating (untrapped)
Zeolite sorption (liquid nitrogen cooled) 760-10- 3
4-stage fractionating (trapped) 5x 10- -10" 12
1

Vapor jet pumps Getter-ion (sputter-ion) 10- 3 -10- 11

Steam ejector Sublimation (titanium) io-4 -^- 11
1 -stage 760-100 10- 2 -10- 10
Cryopumps(20 K)
2-stage 760-10 10-2-10- 12
Cryosorption (1 5 K)
3-stage 760-1
606 APPLICATIONS TO EQUIPMENT DESIGN

thermodynamic relationships of gases make com- Centrifugal compressors are the most widely
pressor selection than pump selec-
more difficult used type in the chemical process industry because
tion. Consultation with compressor manufacturers of the great variety of simple, low-maintenance
is usually required. No detailed consideration of designs that provide long periods of continuous
compressor selection will be attempted here. An operation. They are used at low-pressure rises
introduction to the subject is given in Reference 6, (8 psi) as blowers and at very high-pressure rises

and References 7 and 8 give information on com- (5000 Reciprocating compressors are exten-
psi).

pressor types. sively used at low to intermediate flow rates, and

A preliminary selection of compressor type at intermediate to high pressures. Axial com-

can be made using the concepts of specific speed pressors are limited to use at high flow rates at

and specific diameter, as shown in Figure 21.38. pressures up to 125 psi. Rotary compressors are
The specific speed defined in the figure uses some- used for vacuum service and for low-flow, low-
what different units for flow rate than did the pressure rise applications.

earlier definition for pumps. Use of the figure

requires some knowledge of the speed and dia- Example 21.5. Suggest a compressor type for
meter of the compressor motivating element. For moving 100,000 ft3 /min of air originally at
the reciprocating compressors, piston diameter is 14.7 psia and 70° F, with a discharge pressure of
typically in the range of 3-15 in. and speeds range 40 psig
300-1000 rpm. Rotary compressors operate at
Solution. Some assumption must be about
speeds from a few hundred revolutions per minute
the speed. Since centrifugal compressors are widely
for blowers to 12,000 rpm for screw-type compres-
used, use of 3500 rpm seems appropriate. Air at
sors. The motivating elements vary in diameter from
the stated conditions has a density of
a few inches (screws) to a few feet (lobe blowers).
Centrifugal compressors operate from 3500— 1 2,000
rpm, although some applications as high as
few
— x—
359
29 492
530
= 0.075 lb/ft
3

30,000 rpm are known. Impellers vary from a

inches to several feet. The turbine shown in Figure So the head is

21.32 is a high-speed centrifugal compressor with

40 x 144
short blades at the outer extremity of the impeller = 76,800 ft \b f /\b
imparting kinetic energy to the gas. 0.075

30

10

0.6

3 DS = DH H /-y/Q
N= speed, rpm
3
Q= flowft /s
H= head, ft

D = impeller diameter, ft
0.1
0.1 0.3 0.6 1 10 30 60 100 300 600 1000 3000 10.000
Specific speed. N,
Figure 21.38. Operating ranges of various types of single-stage pumps and compres-
sors (1). (Courtesy of ASM E, copyright © 1962.)
 FLUID MOTIVE DEVICES 607

The specific speed is then 21.3. Contrast the results of closing a valve just
downstream of a positive-displacement pump with the case
^ 3500 V
100,000/60
31
of a centrifugal pump, assuming power to the pump con-
s
~ (76,800)
075 ~ tinues.
21.4. Check a point on the efficiency curve of Figure
This value the radial-flow region of Figure
falls in
21.15 by calculating the efficiency using head and horse-
21.38. A reasonable radial compressor diameter power curves.
might be 2 ft, so the specific diameter would be 21.5. A pump with the characteristics given in Figure
20.17 and a 10-in. impeller are available in the storage
2(76,800) 0.25
D = = 0.82 warehouse. Can it be used to pump 400 gal/min against a
s
V 100,000/60 head of 300 ft? What will be the speed and power consump-
tion? Is there any disadvantage in using this pump?
These values fall out of the ranges of all com- 21.6. A centrifugal pump must be selected to deliver
pressors shown. Perhaps the assumed values were 50 m 3 /hr of water against a head of 20 m. Can any of the
inappropriate. Centrifugal compressors can be pumps of Figure 20.17 be used? What power (watts) must
operated at higher speeds to increase head and be supplied?
capacity. If the speed is 12,000 rpm and the dia- 21.7. A centrifugal pump with characteristics shown
meter 3 ft, then Ns = 106 and Ds = 1.23. This falls in Figure 20.16 is operating at 1600 rpm, delivering
well within the region of radial-flow centrifugal 2500 gal/min of water at a head of 61 ft. Over a period of
compressors. months, scale buildup in the line increases the head to
70 What happens to the delivery rate of water if no
ft.

REFERENCES other changes are made? Suggest a modification to restore

the original delivery rate.
21.8. A plastic intermediate is to be pumped from a
1. Balje, O. E., Trans. ASME, J. Engr. Power, 84, p. 103
storage tank into a batch reactor. Pumping time is to be
(1962).
held to a minimum. Preliminary plans call for placing the
2. Bonnington, S. T., and A. L. King, Jet Pumps and
storage tank on the second floor, the pump being on the
Ejectors, British Hydraulic Research Association, Fluid
Engineering, Cranfield, England (1972). ground floor. System specifications are:

3. Hicks, T. G., and T. W. Edwards, Pump Application Fluid properties:

Engineering, McGraw-Hill, New York (1971). Cp = 0.65Btu/lb°F
4. Kern, R., Chemical Engineering, 82, No. 11, H = 500 cp
pp. 113-120 (1975). p = 45 lb/ft 3
5. Kirk, R. E., and D. Othmer, Encyclopedia of Chemical k = 0.36 Btu/hr ft
2 (°F/ft)
Technology , 2nd ed., John Wiley & Sons, New York M. wt. = 3000
(1963).
Vapor pressure at 80° F = 100 mm Hg
6. Neerken, R. F., Chemical Engineering, 82, No. 2,
System geometry: See the accompanying illustration.
pp. 78-94 (1975).
7. Perry, R. H., and C. H. Chilton, Chemical Engineers'
Handbook, 5th McGraw-Hill, New York (1973).
ed.,
8. Rollins, J. P., ed., Compressed Air and Gas Handbook, 100 ft of pipe & 3 elbows & 1 globe valve
90'
4th ed., Compressed Air and Gas Institute, New York
(1973).

PROBLEMS

21.1. From the data in Figure 21.5, plot the pump

efficiency as a function of capacity at 400 rpm. Pump: Any one of a line of centrifugal pumps of from 10 to
21.2. A rotary pump with the characteristics shown 1000 gal/min capacity can be purchased. All of these
in Figure 21.5 is to be used to pump and meter 25 gal/min pumps specify a minimum NPSH of 10 ft of water, and
of a liquid of 2 cp viscosity and 0.8 specific gravity through develop ample head. What pump capacity should be speci-
a piping system equivalent to 100 ft of 1 -in. sen. -40 steel fied (use a friction factor of 0.033)?

pipe. The suction pressure is 25 psig. Velocity head and 21.9. Figure 21.17 gives the characteristics of a series
static energies are negligible. Can the pump be used? At of centrifugal pumps. Calculate the specific speed for the 6,

what speed? What power is required? 7, and 8 in. impeller sizes.

608 APPLICATIONS TO EQUIPMENT DESIGN

21.10. The pump of 10-in. impeller diameter, for (a) 20,000 gal/min against a 20 ft head.

which characteristics are given in Figure 21.176 is used in a (b) 3000 gal/min of water against a pressure rise of
piping system taking cooling water from a river and deliver- 500 psi.
ing it to a standpipe. The required lift is 30 ft, and flow is (c) 100 m 3 /hr at a head of 100 m.
through 100 ft of 3-in. sch.-40 steel pipe (equivalent length, (d) 100 m 3 /hr at a head of 10 m.
including fittings), (a) What is the maximum flow the pump 21.13. Water is to be pumped from a river to a large

can deliver? (b) What flow rate is attained when a gate valve storage tank for plant service. The piping system shown
in the line is half-closed, to add to the equivalent length of below consists of 180 ft of 3-in. sch.-40 pipe on the suction

fittings given above? side of the pump and 700 ft of the same pipe in the

21.11. Recommend a pump, and defend this recom- discharge line. When the level in the storage tank falls

mendation, for below the control point, water is pumped into the tank
until the control level is reestablished. A centrifugal pump
(a) Pumping cutting oil to a machine tool.
having the characteristics indicated below is used for this
(b) Delivering water from the Hudson River above
intermittent steady-state pumping assignment.
Albany to New York City.
Determine the flow rate through the piping system in
(c) Delivering H2 gas at 1 5,000 psi to an ammonia
gallons per minute and the power required.
converter.
(d) Pumping concentrated tomato soup to the Centrifugal Pump Characteristics
canning machines (100 gal/min).
Capacity, Total Head, Efficiency,
(e) Metering caustic solution into a reaction stream
gal/min ft of water percent
in response to a pH controller.
(f) Feeding liquid oxygen to the first stage engine of
280
an Atlas rocket. 20 260 45
(g) Pumping detergent slurry to a spray dryer nozzle 40 220 60
(1500 psig, 5 gal/min.) 60 160 60
(h) Recirculating water for a laboratory constant- 80 110 56
temperature bath. 100 63 50
120 28 43
(i) Transferring sugar solution from stage to stage of
140 10 37
an evaporator system. Feed temp=120°F;
160 5 30
flow = 30 gal/min.

Vent (open)

Water pumping system. Pump

(j) Recirculating molten salt between a storage tank 21.14. A hydraulic cylinder, consisting of a 4-in.
and a heat exchanger. T= 1000°F, ju = 10 centi- diameter piston that travels at a rate of 9 ft/min, is actuated
poise; flow = 50 gal/min. by a rotary-gear pump which supplies hydraulic fluid to the

(k) Recirculating molten NaK alloy from a nuclear cylinder. The length of the stroke is 20 in., and the load on
reactor to a steam generator in a nuclear power the cylinder is 4 tons. The piping system consists of 30 ft of

plant. j-in. sch.-80 pipe (total equivalent length), and there is a

negligible change in elevation from the fluid reservoir to the

21.12. Suggest a centrifugal pump of the type of cylinder. The hydraulic fluid used has a specific gravity of
Figure 21 .23 to supply at 1 750 rpm: 0.9 and a kinematic viscosity of 100 centistokes.
 FLUID MOTIVE DEVICES 609

A manufacturer's catalog lists the following gear Note: This problem may be solved by (1) considering
pumps, which are judged satisfactory to meet hydraulic- the effect of time dependent heads or (2) the use of an
fluid pumping assignments: average head over the pumping cycle.

Minimum Capacity, gal/min*

21.16. A centrifugal pump with the characteristics
given below is to be used to supply water to a house from a

Pump Pipe 150 300 400 800 1000 nearby lake. The water surface in the vented storage tank in

Size Size, in. psi psi psi psi psi the house is 2 m higher than the lake surface. There is 50 m
(equivalent length) of 50 mm nominal-size steel pipe
1 1/2 1.57 1.5 _ — _ (medium weight) connecting the lake, pump, and house.
3 1/2 3.35 3.15 - - - What flow rate of water at 20°C can be delivered to the
5a 1/2 5.47 5.12 - - - house? What size motor (watts) is needed?
5 1/2 - - 6.73 6.15 5.85
10 3/4 - - 10.07 9.2 a75 Capacity, Efficiency,
10a 3/4 10.05 10.4 — - - m 3/hr Head, m percent

*Courtesy Roper Pump Company. 10 15.0 38

20 14.7 52
From this information, select the pump that will 30 14.3 60
accomplish the required specifications. 40 13.5 64
21.15. A tank car is to be emptied of 10,000 gal of 50 12.0 63
benzene at 80" F in 3 hr (see figure below). The plant piping 60 9.7 56
70 6.0 45
system is as indicated below. Available to do the pumping is

a centrifugal pump having the following characteristics:

21.17. By examination of Figure 21.23, indicate the

Capacity, Total Efficiency,

general trends in pump capacity and head that result from:

gal/min Head, ft percent (a) An increase in maximum impeller size.

(b) An increase in suction and discharge line size.

110 21.18. Given the choice between pumps in two

20 106 29.2 adjacent and overlapping regions of Figure 21.33, explain
40 90 40.0 what criteria could be used to make the choice.
60 63 45.0 21.19. Select a pump size from Figure 21 .23 for each
80 41 47.0 of the following cases, if appropriate. If the pump is in the
100 22 48.3
proper region, use Figure 21.17 to select an impeller size.
120 12 46.5
(a) 1000 gal/min, 120-ft head.
140 7 40.0
(b) 750 gal/min, 140-ft head.
(a) Is this pump satisfactory for the job? (c) 300 gal/min, 80-ft head.

(b) How long will it take to "empty" the tank car? (d) 250 gal/min, 80-ft head.

(c) How much work is necessary? (e) 100 gal/min, 160-ft head.

35 sen 40 - 3 Vent
ft in pipe
Vent plus 3-90° ells

and 1 gate valve

15 ft

20 ft

6ift

110 ft. 3 in. pipe

plus 1 gate valve
and 4-90* ells
4ft

Tank car discharge of benzene.

610 APPLICATIONS TO EQUIPMENT DESIGN

21.20. The centrifugal fan of Figure 21.26a is to be istics for the compressors of Figure 21.38. Can any general
used for ventilating a room. The expected total pressure conclusions be drawn about head, capacity, and position on
drop through the ducts is 1.8 in. of water. What ventilation that figure?
rate is possible and what power would be required? Could a 21.22. It is necessary to compress 600 ft
3 /min of
air
ventilation rate 10 percent higher be obtained with this at 70° F and 14.7 psi to a final pressure of 2000 psig.

fan? (Take into account the concomitant increase in head Suggest a compressor type.
because of greater pressure drop.) 21.23. Suggest a compressor type, rpm, and impeller
21.21. Compare the specific speeds and specific dia- size to compress 1000 m3 /hr under a 10-m head.
meters of the fans of Figure 21.26 with those character-
 Twenty -Two

Flow and Separation

Particulate Solids
Through Fluid Mechanics

Many chemical engineering unit operations involve exerted on the solid by the fluid. This force is a
particulate solids as well as fluids. Often the solids combination of boundary-layer drag and form drag,
are an integral part of the material being and it can be expressed in terms of a drag coefficient.
processed — example, conveying spray-dried
for By Equation 1 3.40, the drag coefficient is

product in a gas stream or feeding pulverized coal in

2FD
air to a burner. Sometimes the fluid motion is used CD =—^-z: (13.40)
to separate a particulate solid from the liquid phase,
as occurs in filtration or sedimentation. Sometimes
where F is the force acting on the solid, v fs is the
the fluid motion is used to separate the solid
free-stream velocity relative to the particle, and 5 is
particles from each other, as in the classification of
the projected area of the solid normal to the flow.
particles of various sizes or densities. In all cases,
This equation is important wherever momentum
however, the equations of fluid mechanics describe
transfer at a fluid-solid boundary must be
the systems, and momentum transfer is the major
examined. Thus, may
be applied to the design of
it
transport process occurring.
particle-separation equipment as well as to that of
The separation methods discussed may be
piping systems. Over the range of flow conditions
classified asmechanical separations, as opposed to
likely to be encountered, the drag coefficient may
those separations requiring vaporization or conden-
be determined from Figure 22.1. Note that the
sation. For example, salt crystals can be separated
sphericity term is introduced to account for non-
from their mother liquor by filtration or centrifuga-
spherical particles. (See Appendix B for sphericity
tion. Several different sizes of crushed ore can be
determinations.)
separated by screening, elutriation, jigging, or
Consider a particle moving through a fluid in
be separated from a liquid
classification. Sludges can
one dimension only, under the influence of an
by sedimentation. The mechanical methods of separ-
external force. This external force may be from
ation may be grouped into two general classes: those
gravity or from a centrifugal-force field. The basic
whose mechanism is controlled by fluid mechanics,
theory of the flow of solids through fluids is based
discussed here, and those whose mechanism is not
upon the concept of freely moving bodies,
described by fluid mechanics — for example, screen-
ing.
*.-«£ (22.D

APPLICATIONS OF THE MECHANICS OF PARTICLE
MOVEMENT THROUGH A FLUID
General Principles
In Chapter 13 the concept of form drag was intro-
duced. For steady flow of a fluid past a solid,
where F is the resultant force acting on any body,
dvldd is the acceleration of the body, and is the m
mass of the body.
In Figure 22.2, the forces acting on the falling
body are the external force (FE ), a buoyant force
{F B ), and the drag force (F D due to fluid friction
)

boundary layers are established, and a force is in the direction of the velocity of fluid relative to

611
612 APPLICATIONS TO EQUIPMENT DESIGN

10,000
5§i^
x^*
— i — -4+

\->,
\^N —— —^_— — —
1000
600
400
1

V^
Nmv
— i
, . . . i

^_s
M 200

100 X%
vn^n
i

II

G 60
\ XSX\
I
1
20

io
V ^v
5 _____ ILLU* = 0.125
—
:_ =3- \P = 0.220
!

"^Lr w = 0.6C
—J—
= 0.8C 6

-p — i
,
. . .

* =
06 ^ ^~^. l.OO
1
\
4

|\
2

1
"1
0.001 0.01 0.1 1 2 4 6 10 100 1000 10.000 10 10

Reynolds number |iVRe = — —

Figure 22.1a. Drag coefficient as a function of Reynolds number (60). (By
permission of J. Franklin Inst., copyright © 1934.)

c "^T
611 \. \
il i/)

8 o
V E
QO __ y- Slope = -2
Q .
nj
Q0
= °
<J -

14

1.0

NRe , Reynolds number

(logarithmic scale]

Figure 22.1b. Evaluation of Nr c at v t ,

Figure 22.2. Forces on a body,
Example 22.1.

obtained from Equation 13.40

the particle. Then, by Equation 22.1,

CD v fs 2 pS
Fb)9c =m ~^ (22.2)
Fd9c = « ! 13.40)
dd

The external force (F E ) may be expressed by

Newton's law as where p is the fluid density and v fs is the velocity
of the particle relative to the fluid.
F E 9c = m 3£ (22.3)
the buoyant
Archimedes's principle yields
where a E is the acceleration of the particle force. The mass of the fluid displaced by the solid
resulting from the external force. The drag force is is (m/p s )p, where p s and p are the solid and fluid
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 613

densities, respectively. Therefore, Equation 22.10 may be used to calculate the

terminal velocity in laminar, transition, or turbu-
F B 9c =
|
— \pa E [22.4) lent if CD is evaluated from Figure 22.1.
flow
An
expression for the terminal velocity inde-
Substituting Equations 22.3, 13.40, and 22.4
pendent of Cq may be developed for particles in
into Equation 22.2 gives
laminar flow. The resisting force due to fluid fric-
dv pa E CD v fs 2 pS tion acting on a sphere when the relative motion
= aE (22.5) produces laminar flow has been shown by Stokes
dd 2m
(54) to be
Equation 22.5 is a general equation for the
FD g c = 2-nD p.v
p (22.11
total force acting on a body in any force field. Its
solution requires a knowledge of the nature of the
external force and the drag coefficient. If the This resistance term may be substituted in
external force is gravity, a E is equal to the accelera- Equation 22.2 along with the other force terms to
tion of gravity g and Equation 22.5 becomes give

m — = ™/l
2 dv
dv CD v pS — i p<
ml --\g-3iTD
dd -K) 2m
(22.6)

Since m=
dd
p pv
\

{nDp 2 IQ)ps for spheres,

1
ps
(22.12)

If the external force is from a centrifugal

2
field, a E = rcj where r is the radius of path and oj
,
(itDp 3 \ dv irDn
3

the angular velocity in radians/sec. Equation 22.5 - - 3TrD p pv

becomes for this case l~6~>^r (ps p)g

dv
— = z
i p^ Cop^S 122.13)
ro: (1 ) (22.7) or
dd \ pj 2m
dv (ps - p)g 18pv
Equations 22.6 and 22.7 are both important 2
(22.14)
in solving mechanical-separation problems.
dd~ ps
Ps D
"p HS
p s

Terminal Velocity. Consider the particle of At the terminal velocity, dv/dd = and
Figure 22.2 to be falling in a gravitational field in
(ps - p)gD,
such a manner that other particles which might be = (22.15)
Vt
present do not hinder its fall. As the particle falls, 18p
its velocity increases and will continue to increase
Equation 22. 1 5 is a statement of Stokes's law, which
until the accelerating and resisting forces are equal.
is applicable to the fall of spherical particles in
When this point is reached, the particle velocity
laminar flow. It is used, for example, for calculat-
remains constant during the remainder of its fall
unless the balance of forces is upset. The ultimate
ing viscosity using a falling-ball viscosimeter. A ball

constant velocity is called the terminal velocity. of known diameter falls through a fluid of un-
For spherical particles the projected area
known viscosity in a tube. The time of fall between
normal to flow is irDp 2 /4, and the mass is two index points is measured, and by Equation
22.15 the viscosity can be determined.
(nD p 2 /6)p s Using Equation 22.6 for the gravita-
.

If Equation 22.15 is substituted into Equa-

tional-field case, and substituting for S and m gives
tion 22.10, an expression for the drag coefficient

—
dv
dd
3Cq
=gfa--)- AD
y
/ p \ v^-p
(22.8)
for laminar flow results. First, Equation 22.10 is
solved for CD
\ ps J pPs
.

At the terminal velocity, dv/dd = 0; therefore, 4(p, - p)gD p

Cn = (22.10a)
2 3v t 2 p
3CD vt p

4D pHS
pp
-'(-3 [22.9)
If one v t in Equation 22.10a is replaced by Equa-
tion 22.15,
Solving Equation 22.9 for the terminal velocity (v t )

yields 4(p, - p)gDp 18m 24p 24

Cn =
'
MPs - p)gDt 3vt p (p s -p)gDp 2 Dp v t p /V Re
Vr = (22.10)
3CD p (22.16)
614 APPLICATIONS TO EQUIPMENT DESIGN

where A/ Re
is the Reynolds number for the 22.20 on Figure 22.1. Its intersection with the
particle. Equation 22.16 is for laminar flow. A proper sphericity curve gives the terminal Reynolds
particle may be considered to be in laminar flow number from which Dp can be calculated.
up to a particle Reynolds number of 0.1, where
transition to turbulent flow begins. Above a Rey- Example 22.1. Calculate the terminal velocity
nolds number of 1.0, Equation 22.10 must be for rain drops of 1 mm diameter falling through
used. air at 20° C.
Equation 22.10 is referred to as Newton's law
and used to evaluate terminal velocities of falling
is
Solution. In this problem, the terminal
spherical particles. It may also be used for non-
some characteristic dimension velocity can be calculated using Equation 22.10;
spherical particles if

used for Dp evaluated at the proper

and CD is
but, since the velocity is unknown, CD cannot be
is ,

sphericity (see Appendix B). Note that in the use

directly evaluated. This method would require a
trial-and-error solution, but the problem is easily
of Equation 22.10, CD is also a function of velo-
solved using Equation 22.19. This equation will be
city, resulting in one equation with two unknowns.
Figure 22.1 A technique plotted on Figure 22.1 using the specified data. It
The second "equation" is .

for simultaneous solution follows.

willpass through the point A/ R e = 1 .0 and CD =

Equation 22.10 may be solved for CD and the

[4gD p 3 p(p s - p)/3p 2 ] with a slope of -2.
results expressed in the logarithmic form. Dp = 0.001 m
4gDp (p s -pY - 29 273 1
log CD = log 2 log v t
Pair
= x ttt x —rz = 1 .206 kg/m 3
3p 0.0224 293 100
(22.17) s/m 2
M air = centipoise = 0.00002 N •

Expressing the Reynolds number at the terminal 3

ps = 1000 kg/m
velocity in logarithmic form yields
4gDp 2 p(ps -p)
DP P CD = 2
log/V Re = log + logi/ f (22.18) 3m
3
4 x 9.80665 x (0.001 ) x 1.206(1000- 1.206)
Eliminating vt between Equations 22.17 and 22.18
3x(20x 10- 6 2
gives )

= 39,375
4gDp 3 p(ps - p)
log CD = -2 log/V Re + log
3|u' On Figure 22.1 at (CD = 39,375, /V Re = 1.0) draw
a line of slope—2 (see Figure 22.1a). At its inter-
(22.19) = 1 .0, /V R e = 1 90 = D v t plp. There-
section with <//
p
Equation 22.19 the equation for a straight line
is fore,
of slope (—2) passing through the point A/ Re = 1
190/i_ 190x0.00002
and Co = 4g D 3 p(ps — p)/3p 2 In this equation, v t
p .
vf = = 3.15 m/s
does not appear, but
it may be determined by Dp p ~ 0.001 x 1.206
plotting Equation 22.19 on Figure 22.1 The inter- .

section of this line with the proper sphericity curve In Example 22.1 and in the development of
will give the terminal Reynolds number from Equation 22.10, it was presumed that the moving
which v t can be calculated. particles were of rigid shape. This is not true of
In a manner similar to the development of liquid drops, which take shapes that vary with the
Equation 22.19, an expression may be derived in droplet motion. Small drops are nearly spherical,
which the size of the particle does not appear. The but larger ones tend to streamline, to oscillate, and
expression is to break up. Gas bubbles moving in a liquid are of
even more varied shape. Small bubbles
4g{p s - p)p. (Dp < 0.2 mm) are nearly spherical, but tend to
log CD = log /V Re + log (22.20)
flow in swirling patterns. Larger bubbles have such
2„
z 3
2p v t
variable and unstable shapes as to require the use
The size of a particle having a specified terminal of specialized correlations to determine their rate
velocity can be determined by plotting Equation of movement.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 61

Hindered Fall of Spherical Particles. In the

usual operational case, many particles are present,
and the surrounding particles interfere with the
motion of other individual particles. The influence
of the neighboring particles affects the velocity
gradients surrounding each particle. For this
hindered flow, the settling velocity is considerably
less than what would be calculated from Equation
22.15. The mechanism of fall differs in that the
particle is settling through a suspension of particles
in a fluid rather than through the fluid itself.
The density of the fluid phase becomes
effectively the bulk density of the slurry, which is
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
the quotient of total mass of liquid plus solid Volume fraction of liquid in slurry (.2")

divided by the total volume. The viscosity of the

Figure 22.3. Settling factor for hindered set-
slurry is considerably higher than that of the fluid tling (By permission of Amer. Chem.
(52).
because of the interference of boundary layers Soc., copyright © 1944.)
around interacting solid particles, and because of
the increase of form drag caused by the solid reported in the form of Figure 22.3, which repre-
particles. The viscosity of such a slurry is sents the slurry viscosity fitted to Equation 22.21.
frequently a function of the rate of shear, of the For non-Newtonian slurries, such a curve must be
previous history as it affects clustering of particles, used only in situations paralleling the measure-
and of the shape and roughness of the particles ments. Caution must be exercised that all Theo-
insofar as these factors contribute to a thicker logical factors be recognized and controlled in such

boundary layer. comparisons.

As would be surmised from the above, genera- Two-Dimensional Motion. The previous dis-
lized predictions of the viscosity of slurries are cussion was limited to simple vertical flow of a
inappropriate. Experimental measurements are particle in one-dimensional flow, but the particle
necessary for accurate values, and extrapolation of may have a horizontal velocity component as well.
any variable should be made with caution. For one Such would be the case of a liquid being dispersed
particular slurry of nonclustering the
spheres, from a spinning-disk atomizer where the liquid is
measurements can be correlated into the
a ratio of thrown from the disk in a horizontal path. In this
effective bulk viscosity (p B to the viscosity of the
)
case, a significant initial horizontal-velocity compo-
liquid expressed as a function of the volume frac- nent exists. As the fall continues, the horizontal-
tion of liquid {£) in the slurry for a reasonable velocity component decreases. For a development
range of compositions (6): and solution of the equations of motion for parti-
cles having a second velocity component, the
HB IQl-8211-g]
reader is referred to Lapple and Shepherd (33).
" (22.21
M St Applications of Particle-Flow Theory. The
mechanics of particle motion discussed thus far can
For greatest convenience in use, an equation using
be directly applied to industrial operations of
only the physical properties of the solid and liquid
particle separation from fluids. The mineral-
phases would be most desirable. A correction
dressing industrieswidely use some of the
factor (R) incorporating both the viscosity and
techniques, but the chemical industries also use
density effects can be derived for a given slurry,
many of them. For example, the elimination of
allowing use of a more convenient equation based
dust and dirt from flue or stack gases and the
upon Equation 22.15
removal of solids from liquids prior to discharge as
(ps-p)gDp 2 waste are cases common to the chemical-process
vH = R (22.22) industries. The emphasis upon clean streams and
18m clean atmosphere makes the chemical engineer
where vH is the terminal velocity in hindered more and more concerned with the "cleaning" of
settling. fluids. The chemical engineer may also be involved
Experimental measurements of effective with solid separations as a consequence of one or
viscosity as a function of composition may be more steps in a chemical process.
616 APPLICATIONS TO EQUIPMENT DESIGN

Classification 24m
The separation of solid particles into several frac- Cob ~ (22.27)
Db v t P
tions based upon their terminal velocities is called
classification. Suppose, for example, that two Substituting Equations 22.26 and 22.27 into
particles having different settling velocities are Equation 22.24 gives, for different-sized particles
placed in a rising current of water. If the water of different density settling at the same velocity
velocity adjusted to a value between the terminal
is
(Pb ~ p)
velocities of the two particles, a separation will -r (22.28)
result. The slower-settling particle will move up- Db (Pa ~ P) D
ward withthe water, while the faster-settling or
particle simultaneously settle out to the
will
2
bottom. Another way to separate the particles (Da) ^Pb~
(22.28a)
would be to feed the suspension through a tank of \ DJ pa -
large cross-sectional area. When the fluid stream
enters the tank, the horizontal velocity component Separation is possible only if a separation

decreases, and the particles start to settle. The ratio, defined asthe ratio of the size of the smallest
faster-settlingparticles will tend to accumulate particle of a to that of the largest particle of b, is

near the but the slower-settling particles will

inlet, greater than

be carried farther and will concentrate nearer the -p

exit. (22.29)
Db Ypa
- p
Suppose two different materials a and b are
present in a solid particulate mixture with a more where
dense than b. If the size range of the two materials
is large, no complete separation is likely because
n = % for laminar flow
the terminal velocities of the largest particles of b n = 1 for turbulent flow
may be greater than that of the smallest particles \ <n < 1 for transition flow

of a. The range of sizes that will result from a

separation can be calculated from the ratio of sizes
Examination of Equation 22.29 indicates
that by choosing a fluid with a density very nearly
of the particles of a and b with the same terminal
equal to that of one of the solid materials, the size
velocities.The equality of terminal velocities
ratio of the separation can be made larger or
occurs when
smaller. For example, if p approaches p b the ,

'MPa - p)gD a 4(p b - p)gD b settling ratioapproaches zero and particles of any
vt = size range can be separated. This is the basis for
3pCDi 3pCDi
"heavy-medium separations," which can be much
(22.23) more effective than hydraulic classification. The
or liquid phase may be made heavier by dissolving a
(pb - p) CDi soluble material in it which increases the density.
(22.24) More frequently, the effective density of the liquid
Db (pa ~ P) CDb
is increased by dispersing in it a heavy solid so
From the shape of the drag coefficient- finely ground that its settling velocity is negligible.
Reynolds number diagram, it is evident that the The "liquid" phase in heavy-medium separations is
value of the drag coefficient is essentially constant actually a relatively stable suspension, in which the
at high values of the Reynolds number. Therefore, differences in terminal velocities of particles of
for particles of the same sphericity settling at high different densities are magnified.
Reynolds numbers, Equation 22.24 becomes

Da Example 22.2. A mixture of silica and galena

Pb~ P is to be separated by hydraulic classification. The
(22.25)
Db Pa~ P mixture has a range betwen 0.08
size and mm
3
Under laminar-flow conditions, CD = 24//V Re 0.7 mm. The density of galena is 7500 kg/m and ,
,

and the density of the silica is 2650 kg/m 3 Assume the .

sphericity, = 0.806.
i//
2Ap
Coa - (22.26)
(a) What water
velocity is necessary for a
Da v t p pure galena product? Assume unhindered settling
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 617

of the particles and a water temperature of 20°C. smallest galena particle in the galena product. By
(b) What is the maximum size range of the Equation 22.20,
galena product?
WPgal ~ p)p
Solution
log CD = log/V Re + log
2„
3pV f
3

(a) For equal-sized silica and galena particles,

the heavier galena will settle faster. Therefore, the
log CD = log A/ Re

settling velocity of the largest silica particle will 4 x 9.80665(7500 - 1000X0.001

determine the water velocity. A water velocity + log
2
3(1000) (0.04) 3
equal to this settling velocity should give a pure
galena product. By Equation 22.19, using the SI log CD = log A/ Re + log 1.33
units This is a straight line on Figure 22.1 passing
through (A/ R e = 1 .0, CD = 1 .33) with a slope of +1
4gDp 3 p(psii- p) This line intersects the \p = 0.806 curve at A/ Re =
log Co = -2 log/V Re + log
2 9.0. This Reynolds number corresponds to a
3P
diameter of
The viscosity of water at 20°C = 0.001 N • s/m 2 ,

/V- Re /i (9X0.001
DP = = 0.000225 m
logCD = -2logA/ Re + vp (4.0X1000)

Thus, the galena product size ranges between

r4x(9.80665)(0.0007) 3 (1000X2650 -1000)1
109
0.225 mm and 0.70 mm. Galena particles smaller
L 3(0.001) 2 than 0.225 mm are carried overhead along with all
log CD = -2 log A/ Re + log 7400. silica.

Classification Equipment. The simplest type

This is a straight line on Figure 22.1 passing of classifier consists of a large tank with provisions
through (A/ R e = 1, CD = 7400) with a slope of -2. for a suitable inlet and outlet. Figure 22.4
This line intersects the = 0.806 curve at /V R e =
\jj
illustrates such a gravity settling tank.

28. A slurry feed enters the tank through a pipe.

A Reynolds number of 28 corresponds to a Immediately upon entry, the linear velocity of the

settling velocity of feed decreases as a result of the enlargement of

cross-sectional area. The influence
of gravity causes
(28H0.001 the particles to with
the faster-settling
settle,
vt = = 0.04 m/s
10~ 3 )( 1000) particles falling to the bottom of the tank near the
D vp (0.7 x
entrance. The slower-settling particles will be
This velocity must also be the water velocity carried farther into the tank before settling to the
to ensure a clean galena product, since it will carry bottom. Placing vertical baffles within the tank
overhead.
all silica allows for the collection of several fractions. Very
Calculation of the size of a galena particle
(b) fine particles will be carried out of the tank with
that settles at a velocity of 0.04 m/s fixes the the liquid overflow. The position in the tank at

Fluid in, Fluid out,

wide range some fine

of particle particles

sizes

Coarse Intermediate Fine

particles particles particles

Figure 22.4. Gravity-settling tank.

618 .APPLICATIONS TO EQUIPMENT DESIGN

which a certain size particle may be expected can desired, is fed into the top of the column. The
be calculated on the assumption that the particles large particles, which settle at a velocity higher
quickly reach terminal velocities. The relationships than that of the rising fluid, are collected at the
developed earlier apply to the deposition location bottom of the column, and the smaller particles are
of any given particle size. carried out of the top of the column with the fluid.

The gravity -sett ling tank is referred to as a Several columns of different diameters in series
surface-velocity classifier. The resulting separation may be used to bring about a further separation.
is not a sharp one, since considerable overlapping A double-cone classifier (Figure 22.7) is a
of size occurs. conical vessel inside of which is a second cone. The

The Spitzkasten (Figure 22.5) is another type inner cone is slightly larger in angle and is movable
of gravity-settling chamber. It consists of a series of in a vertical direction so that a variable annular
conical vessels of increasing diameter in the direc- area is available for flow.

tion of flow. The slurry feed enters the top of the The feed material flows downward through
first vessel where the larger, faster-settling particles the inner cone and out at a baffle at the bottom of
are separated. The overflow, including unseparated
solids, feeds into the top of the second vessel,
where another separation occurs since the velocity
is lower than that in the first vessel. Each succeed-

ing vessel will separate a different range of particles

depending upon settling velocities of the particles
and upon the fluid velocity in the vessel. In each
vessel, the velocity of the upflowing fluid is inde-
pendently controlled to give the desired size range
of product from that vessel.
An elutriator (Figure 22.6) is a vertical tube
through which a fluid passes upwardly at a spec fic ;

Hydraulic water
velocity while a solid mixture, whose separation is

Heavy, coarsest material

Less coarse solids

Still less coarse solids

' Fine solids

Figure 22.5. Schematic representation of a Spitzkasten.

 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 619

Feed

Water and
O I
fine-solid
overflow

Movable cone

Fixed cone

Water
Fine
Water
solids

Coarse solids
Figure 22.6. Elutriators.
Figure 22.7. Double-cone classifier.

the inner cone. Rising upward through the annular sion. The heavy material sinks to thebottom where
space, fluid is fed at a controlled velocity into the the rakes scrape it upward toward the top of the
unit in the vicinity of the exit of the inner cone. tank.The stroking action of the rakes is such that

The solids from the inner cone and fluid are
and then flow through the annulus whose
sectional area varies. Classification action occurs in
this annular space with the small particles leaving
with the liquid and the larger particles settling to
the bottom for removal.
Another piece of equipment for solid separa-
tion whose mechanism is based upon settling
velocity is the rake classifier (Figure 22.8). The
rake classifier is a tank with an inclined bottom in
which are provided movable rakes. Feed is intro-
duced near the middle of the tank. The lower end
of the tank has a weir overflow over which the
fines that have not settled leave in dilute suspen-
mixed when the stroke is completed, the rakes are lifted
cross- and returned to the starting position. The slurry is
thus kept in continuous agitation. The time of the
raking stroke is adjusted so the heavy particles have
time to settle while the fines remain near the
surface of the slurry in the rake compartment.
Thus, the heavy material is moved upward along
the floor of the tank and removed at the top of the
apparatus as a dense slurry.
For classifying fines material, a bowl classifier
(Figure 22.9) is used. It consists of a shallow
cylinder with a gentle conical bottom. The feed
slurry is fed at the center near the surface, and the
liquid flows outward in a radial direction. The

Figure 22.8. Heavy-duty rake classifier. Cutaway shows rake assemblies. (Courtesy
Dorr-Oliver, Inc.)
620 APPLICATIONS TO EQUIPMENT DESIGN

Figure 22.9. Bowl classifier 8 by 12 ft in diameter. (Courtesy Dorr-Oliver, Inc.)

solids settle out according to their particle size of solids from fluids in a cyclone usually involves
with the fines overflowing at the bowl wall. The particle sizes small enough so that Stokes's law
heavy sludge is scraped toward the center for dis- may be assumed valid, irrespective of gas velocity.
charge. Equations 22.7 and 22.16 give, for a spherical
particle obeying Stokes's law,

Centrifugal Classification

The most widely used of the centrifugal-type separ- —£-ru 2 {p - P s )

= 2nDp v Rix (22.30)
ation equipment is the cyclone separator for separ-
ating dust or mist from gases (Figure 22.10). where v R is the radial velocity of the particle. But
The feed enters the cyclone tangentially near 2
roj = v\ an lr where vxan is the tangential velocity
the top and is given a spinning motion as it enters of the particle at radius r. Thus,
the chamber proper. The tangential velocity of the
particles tends to carry them toward the periphery
*Dp 2 vl n
{p s -p)=3TrDp v R ii (22.30a)
of the chamber. The spiral motion of the fluid
results in some inward radial acceleration of the
Solving Equation 22.30a for v R gives
particle, and simultaneously gravitational force
imparts downward acceleration. The result is a (p, - p) l^n
downward and spiraling path of increasing radius vR (22.31
L 18 M J r
until the particle reaches the boundary. Thereafter
the particles continue a spiraling path down the but, by Equation 22.15, the bracket term in Equa-
wall, and the gas, freed of solids, moves upward in tion 22.31 the gravitational terminal velocity of
is

the central core. At high tangential velocities, the the particle with the constant, g, omitted. There-
outward force on the particle is many times the fore,

force of gravity; hence, cyclones accomplish more vLn

rapid and more effective separation than vR = (22.32)
gravitational-settling chambers for particles of sizes
down to a few microns. For very small sizes, the Thus, the higher the terminal velocity, the
energy represented by the tangential velocity is greater the radial velocity, and the easier it should

insufficient to overcome the centripetal force of be to separate the particle. The evaluation of the
the rotating fluid, and separation is ineffective. radial velocity is complex because it is a function
It is assumed that the particles in a cyclone of terminal velocity, tangential velocity, and posi-
quickly reach their terminal velocities. Separation tion from the center of the cyclone. For a given-
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 621

sized particle, the radial velocity is a minimum near

CLEAN AIR the center of the cyclone and increases toward the
OUTLET
wall.
SCROLL OUTLET -

Theoretically, the size of the smallest particle

retained by the cyclone can be calculated, but
experience indicates that agglomeration of particles
and entrainment tend to introduce deviations from
the calculations. If a cyclone is designed to separ-

ate a definite particle size, a cyclone efficiency may

be defined as the mass fraction of the particles of
DUST LADEN
AIR INLET
that size that is separated in the cyclone. Cyclone
efficiency increasesas the size of the particles
increases. If cut diameter is defined as that
a
diameter for which half the particles by weight will
be separated, the efficiency of a typical cyclone
separator is shown in Figure 22.1 1
Cyclones are usually designed to operate with
a specified pressure drop. The pressure drop
through cyclones may be one to twenty times the
inlet velocity head (51). Lapple (32) suggests an
inlet gas velocity of 50 ft/sec for ordinary installa-
tions.
Through the use of plastic cyclones, the
effects of size, shape, and air velocity have been
VORTEX SHIELD
studied by several investigators (19, 56). Signifi-
cant cyclone dimensions are height, diameter, exit-
duct diameter, and the ratio of the diameter of the
DUST TRAP cylinder to that of the exit duct.

Wet Scrubbers
For gases containing very fine particles, liquid
scrubbing is sometimes used for effective separa-

Figure 22.10. Cyclone separator. (Courtesy the tion. The dirty gas passes upward through water
Ducon Co.) sprays, which tend to wash out the dirt particles

0.9 1

0.8
0.7

1 06
* 0.5
c
o
o 0.4
o
~ 0.3

S 0.2
o
o
nj

0.1
0.1 0.2 0.3 0.4 0.5 0.6 0.8 1.0 4 5 6 7 8 9 10
Particle-size ratio, DP /DCU[
Figure 22.11. Cyclone efficiency
622 .APPLICATIONS TO EQUIPMENT DESIGN

CLEAN AIR OUTLET obtained, and the solids concentration that can be
handled (20). The first of these is the tubular-bow/
centrifuge. This centrifuge rotates at high speeds
developing centrifugal forces of the order of
13,000 times the force of gravity, but it is built for
low capacities in the range between 50 and
SUPPLEMENTARY
500 gal/hr. Since it has no automatic solids
WATER N.Ei removal system, it can handle only small concen-
trations of solids. The second is the disk-bowl
centrifuge, which is larger than the tubular-bowl
centrifuge and rotates at slower speed, developing a
centrifugal force up to 7000 times gravity. This
INSPECTION
— DOOR centrifuge may be designed to handle as much as
. -_ r s.r
5000 gal/hr of a feed containing moderate quanti-
ties of solids that are discharged continuously in a
concentrated stream. Both of these types of centri-
fuge are primarily designed to separate liquid-liquid
systems. However, the disk-bowl centrifuge can be
adapted to separate liquid-liquid-solid systems, or
liquid-solid systems where the major product is a
clarified liquid. The solid-bowl centrifugal, which
forms the third class, is primarily a solid-liquid
separator and operates like a thickener (to be
described later). These centrifugals are built to
handle solids at rates up to 50 tons/hr.
The tubular-bowl centrifuge is shown in
Figure 22.13. It consists of a tubular bowl that
rotates within a housing. The bowl stands verti-
cally, hung from a thin flexible, solid shaft sup-
ported by a thrust bearing and is belt driven from
SLUDGE OUTLET
an electric motor or direct driven with an electric
Figure 22.12. Centrifugal washer. (Courtesy The motor or from an air or steam turbine. At the
Ducon Co.) bottom, the bowl is loosely guided by a spring-
supported bushing. The feed is delivered to a nozzle
and entrain them for removal at the bottom of the at the bottom of the bowl and jets into the bowl
scrubber. Figure 22.12 shows a cyclone scrubber in where it is quickly accelerated to bowl speed by
which both centrifugal force and scrubbing action the action of light metal vanes loosely fitted into
work toward elimination of dirt. Many variations the bowl. The heavy phase collects along the walls
of this type of scrubbing equipment are available. of the bowl, the light phase forming a concentric
layer on the inside of the heavy phase. Droplets of
Centrifugation light liquid in the heavy phase move toward the
Centrifuges of the solid-bowl type are settling center of the bowl, and droplets of heavy liquid
devices that utilize a centrifugal field rather than a move toward the wall of the bowl. The 2- to 5-ft
gravity field to cause separation of the components length of the bowl affords sufficient residence time
of liquid-solid or liquid-liquid systems. Gas-solid for the droplets to reach their proper phase layer.
systems are separated by the same effects in the The layers are maintained in the bowl, and their
cyclone separator discussed earlier. The centrifugal separate discharge is controlled by ring dams at the
field causes particles of the heavier phase to "fall" top end of the bowl. There is no arrangement for
through the lighter phase away from the center of removing solids; if they are present, they usually
rotation. This action is exactly what occurs under build up on the walls of the bowl until the unit is
the influence of the gravity field in classification. stopped and cleaned.
Centrifuges. There are three main types of A cutaway view of a disk-bowl centrifuge is
centrifuges distinguishable by the centrifugal force shown in Figure 22.14. This type of centrifuge was
developed, the range of throughputs normally invented by DeLaval in 1878 and has been widely
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 623

Driving mechanism used since then for such diverse applications as

%
1 cream separation, catalyst separation, dehydration
of marine lubricating oils, and refining fish oils. In
the unit shown, feed, which may consist of two
liquid and one solid phases, enters from the
bottom and is accelerated to the centrifuge speed
as it is thrown out toward the perimeter of the
spinning disks. It feeds up into the disks through a
series of holes in these disks. Here the long flow
path but short distance from disk to disk promotes
easy separation, with the heavy liquid flowing out
on the underside of the disks and the light liquid
flowing toward the center on the top surface of the
disk. Solids collect in the heavy liquid in the area
outside of the disk stack. They are discharged by
extrusion through a series of nozzles around the
perimeter of the spinning bowl along with some of
the heavy liquid. Light liquid and the bulk of the
heavy liquid overflow at the top of the bowl and
are discharged from the flanged pipes shown. In
cases where the feed contains only one liquid
phase, the feed is conducted to the outside of the
disks and flows in toward the hub between the
disks. The solids are then thrown into the outer
area and extruded out through the nozzles.
Centrifuges of the disk-bowl type are oper-
ated at speeds up to 10,000 rpm producing centri-
fugal forces of 14,000 times gravity force g. They
Figure 22.13. Cutaway view of tubular-bowl centrifuge. offer capacities up to 30,000 gal/hr.

Figure 22.14. Cutaway view of a disk-bowl centrifuge with top drive, bottom feed,
and peripheral nozzles for continuous solids discharge. (Courtesy Sharpies-Stokes
Div., Pennwalt Corp.)
624 APPLICATIONS TO EQUIPMENT DESIGN

EFFLUENT

Figure 22.15. Typical cross-sectional view of solid bowl centrifugal, right-hand

design. (Courtesy Bird Machine Co., Inc.)

A cross-sectional view of a solid-bowl centri- bowl diameters ranging between 4 and 54 in. The
fugal is shown in Figure 22.15, and the action of 54-in. machine handles up to 50tons/hr of solids,
this centrifugal is diagrammed in Figure 22.16. The although this rate must be reduced if the particles
major parts of this machine are a cylindrical bowl are particularly fine or if the liquid phase is vis-
with a truncated cone-shaped end and an internal cous. These machines may also be operated as
screw conveyor for solids that closely fits the classifiers, in which case the feed rate and bowl
interior of the bowl. These parts rotate together, speed are adjusted so that satisfactorily small
but the screw conveyor rotates at a rate 1 or 2 rpm particles willnot settle out and leave with the
below the rate of rotation of the bowl. In opera- filtrate. This type of operation might occur with
tion, the feed is admitted through the central the centrifugal accepting the product from a wet-
screw and enters the bowl about halfway along the grinding step and feeding the large-particle solids
side. The centrifugal action forces both liquid and back to the grinder while the filtrate with its fine
solid phases to the walls of the cylinder. The solids, particles passes on for further processing. These
being denser, concentrate against the walls and units develop a centrifugal force up to 3000 times
along the bottom of the pool of liquid held in the that of gravity, with speeds up to 6000 rpm, so
bowl by the position of the filtrate discharge parts. that sharp separations can be made in the 1 -micron
The solids conveyor, however, has a net rotation particle-size range.
toward the small end of the bowl and scrapes the Similar centrifugals are made with a perfor-
solids from the walls of the cone toward the small ated wall on the bowl. These centrifugals act
end. As the solids move in this direction, they may exactly like filters (to be described later) with the
be given a freshwater wash with water entering in filtrate draining through the cake and bowl wall
the same manner as the feed entered. They are into a surrounding collector. As with other centri-
ultimately discharged at the small end of the coni- fugal filters, they operate best with coarse-granular
cal bowl section. or coarse-crystalline, free-flowing solids.
These centrifugals are made with maximum Centrifuge Theory and Calculations— Rate of
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 625

rEED PORTS

PILLOW BLOCK AND

MAIN BEARING

"EED PIPE
BRACKET
SOLIDS
DISCHARGE
-WASH
COMFARTMENT
Figure 22.16. Cutaway view of a solid-bowl centrifugal showing internal flow of
liquid and solids phases. (Courtesy Bird Machine Co.)

Separation. The basic force balance around a the unique terminal velocity characteristic of its

particle falling in a centrifugal force field was given position. Thus, for any position, dvldd = 0; but,
earlier as for any instant in the movement of a single
particle, dv/dr is positive. Then considering a
r<X> (22.7) particular position, dv/dd = 0, and Equation 22.33
dd \ ps ) 2m becomes
In the separation of phases by settling in either a
gravitational or a centrifugal field, the perfection ru 2 (p s -p)Dp 2
vr = (22.34)
of separation is limited by the rate of fall of the ^8p
smallest particles In most cases, these
present.
low enough that laminar flow
particles fall at rates where
exists and Co = 24//V Re/ as shown by Figure 13.5.
Making this substitution in Equation 22.7 and con- vR = terminal falling velocity of spherical
sidering spherical particles so that particles of diameter Dp at radius r in a

7T
centrifugal field rotating at rate cj
Dp 3 p s
m 4p s a The radial distance traveled by the particle may be
~S

Then Equation 22.7 becomes

s
W obtained by multiplying Equation 22.34 by the
differential time (dO).

ru 2 (ps -p)Dp 2
vR dd = dr = dd (22.35)
dv /ps -p\
~/ps-p 18^ 18m
2
(22.33)
PsDp
which, upon integration, gives
As particles move radially in a centrifugal field, the
field strength changes with their position. Con-
—r
2
2
oj (p s -p)Dp 2 n
u 2 (p -p)D p 2
s V
sequently, the terminal velocity of the particles is a in
18m 18m Q
function of radial position. In these developments
a particle at any position is considered to move at (22.36)
626 APPLICATIONS TO EQUIPMENT DESIGN

where
V= volume
Q
centrifuge
of material

= volumetric feed rate to the centrifuge

V/Q = residence time of a particle in the
held in the
where

m eff
= effective value of {r 2 — r\ )lr to
be used in Equation 22.37
when r 2 — r-\, the thickness of
the liquid layer in the centri-
centrifuge fuge, is not negligible compared
The diameter in Equation 22.36 is that of a to either r-\ or r 2
particle falling from
to r 2 during the residence
r^ A very useful characteristic of a centrifuge
time available in the centrifuge. The significance of can be derived by manipulating Equations 22.37
this equation is perhaps most readily seen in rela- and 22.15. Equation 22.37 is solved for Q and the
tion to the parallel equation, which holds when the gravitational constant inserted.
liquid layer within the centrifuge is very narrow
compared to the radius. In this case, the centrifugal -p)gD
'
2
Vu 2

field is considered constant, and Equation 22.35 q = -

(p s
—-— p - •
—— r
— : = 2vt '
Z
may be written directly in terms of the residence 9m g(r2 -o
time(WQ). (22.39)
in which
ru 2 (ps -p)D 2
V (ps -p)9Dp
'
2
vR 6 = x= (22.35a) vt = (22.15)
18m Q
18m
where
where vt is the terminal settling velocity of a
x= radial distance traveled by a particle of particle in a gravitational field and
diameter Dp in the residence time avail-
able Vu 2 r
2= (22.40)
9(r 2 - n
If x is taken as half the thickness of the liquid layer
(^2
—r \ )/2, half some diameter Dp
the particles of
'

where Z is a characteristic of the centrifuge itself

will settle to the wall, whereas half of them will and not of the system being separated. The 2
still be in suspension when the fluid leaves the factor can then be used as a means of comparing
centrifuge. Dp '
, is the "critical diameter." Particles centrifuges (1). the cross-sectional area of a
It is

of diameter larger than predominantly be Dp will settler that will particles down to the same
remove
settled from the fluid phase, and particles of diameter as those separated in the centrifuge when
diameter smaller than D p will predominantly its volumetric feed rate equals that of the centri-

remain in solution. Solving Equation 22.35a for fuge. If two centrifuges are to perform the same
D p and substituting (r 2 - r )/2 for x, the critical
'
,
y
function.
particle diameter is
9i 9i (22.41)
9MQ 22
r2 -n
<v= (ps -p)u 2 V
(22.37)
The quantity I can be determined for com-
mercial centrifuges, although in some cases the
where
determination requires approximation methods.
For the tubular-bowl centrifuge, applying Equation
i"2 — ry = thickness of the liquid layer
22.38 gives
Dp '
= critical particle diameter
2
iru> / (r 2
2
-r, 2 )

Comparison of this equation with Equation 2 = (22.42)

22.36 shows that for the case where the liquid-
In
layer thickness is great enough that the variation of rv
centrifugal field with radius must be considered, where / = bowl length
the effective value of (r 2 — r\ )/r is
For the disk-bowl centrifuge Ambler (1) gives

(^—-) =2 In— (22.38) 2nir(r 2

3 - r^J1
V r / ef f /-! 2= (22.43)
3g tan £2
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 627

where Example 22.3. A liquid-detergent solution of

100-centipoise viscosity and 0.8 g/cm 3 density is
n = number of spaces between disks in
to be clarified of fine Na 2 S0 4 crystals (p s =
the stack
3
r-\ = outer and inner radii of the disk stack
1.46 g/cm ) by centrifugation. Pilot runs in a
r2 ,

12 = conical half angle laboratory supercentrifuge operating at

23,000 rpm indicate that satisfactory clarification is

Table 22.1 gives values obtained for several types obtained at a throughput of 5 Ib/hr of solution.
of centrifuges based both upon calculations from This centrifuge has a bowl 7| in. long internally
geometry and upon laboratory and plant data. The with r 2 = % in., and (r 2 — ry ) = 19/32 in.
table shows both the comparative performance of (a) Determine the critical particle diameter
the centrifuges and the effectiveness of the for this separation.
machines in comparison to what is calculated on (b) If the separation is to be done in the
the basis of their geometry. Note how much more plant using a No. 2 disk centrifuge with 50 disks at
effective the disk centrifuge is than any of the 45° half angle, what production rate could be
other centrifuges listed. This is the result of the expected
high residence time and short disk-to-disk separat-
(a) The critical particle diameter
Solution,
ing distances built into this machine.
(Dp can be determined directly from Equation
')

In the development of Equations 22.33 to

22.37. Since the liquid layer is thick, Equation
22.43 given above, the diameter (Dp was defined )
22.38 is used to get the effective value of (r 2 — r )/r. y
as a particle diameter, with the inference being that
the particle would be solid. If a liquid-liquid separ-
ation is to be made, the mechanism is no different '9 x 100 x 2.42 x (
\62.4x0.8/
— x 1728
0.875
from that of solid-liquid separations. Droplets of D' = x2 In
liquid rather than particles of solid migrate, and 62. 4( 1 .46 -
0.8 0) x (2 tt x 23,00 0.281
they migrate across one liquid phase and coalesce x 60) 2 7.75 x tt(0.875 2 - 0.281 2 )
into the other phase instead of migrating through
Dp' = 0.0000033 ft or 1 .08 microns
the fluid phase to the wall. The rate of migration is
still calculable through use of Equation 22.37 (b) Table 22.1 gives 2 = 1290 for the labora-
properly modified for the centrifuge being used. tory supercentrifuge and 2 = 72,600 for the No. 2

Table 22.1 COMPARATIVE PERFORMANCE OF CENTRIFUGES (1)

2 Values, ft
2

From Experimental Extrapolation on

Calculated from Data Clarifying Ideal Commercial Systems
Geometry Systems (Supercentrifuge Tests)

Laboratory supercentrifuge (tubular bowl I3 in. I.D. x l\ in. long) operating at:
10,000 rpm 582 582 582*
1 6,000 rpm 1,485 1,485
23,000 rpm 3,070 3,070 1,290
50,000 rpm 14,520 14,520 not used
No. 16 supercentrifuge (tubular bowl 4| in. I.D. x 29 in. 3) operating at:
long)
1 5,000 rpm 27,150 27,150 27,150
No. 2 disk centrifuge, 1| in. r\ x 5| in. r 2 on disks
52 disks, 35° half angle, 6000 rpm 78,800 98,000 89,400 to 178,800
50 45° half angle
disks, 34,000 72,600 67,900 to 134,000
Super-D-Cantor (solid-bowl centrifugal)
PN-14 (conical bowl), 3250 rpm
(£>= 14 in.-8 in., L = 23 in.) 4,750 2,950 2,950*
PY-14 (cylindrical bowl), 3250 rpm
[D - 14 in., L - 23 in.) 8,940 5,980 5,980*

*For relatively low throughput rates.

628 APPLICATIONS TO EQUIPMENT DESIGN

disk centrifuge. Since these centrifuges are to per- final radius (/>). Upon simplification
form identical functions, Equation 22.41 gives
2
ft pu) r dr
Q^ (22.44)
Q? = 22 = x 72,600 = 281 Ib/hr n gc
•

1290
which, upon integration, gives
2
Centrifuge Theory Calculations— Outlet Dam
Settings. In liquid-liquid separations, the position
P= 7r-(r f 2
t
-n 2 ) (22.45)
2gc
of the outlet dam becomes more important than it

is for solid-liquid separations, for, instead of Applying Equation 22.45 to the physical situation
merely controlling the volumetric holdup in the of Figure 22.17/? where the pressure must be the
centrifuge and the critical particle diameter, the
same on either side of the liquid-liquid interface at
position now also determines whether a separation r 2 results in
can be made at all. Figure 22.17 shows the physical
situation in a centrifuge arranged for clarifying a Ph^> Pi^>
2
liquid phase of entrained solids and that in a centri- lr2 -r4 2 )
= (r 2
2
-n 2 ) (22.46)
2gc 29c
fuge arranged to separate two liquid phases. In
Figure 22.17/? the distances have the following or
significance: 2
r2 ~rS BL (22.47)
r-\ = radius to top of light-liquid layer ri
2
~r? Ph
r2 - radius to liquid-liquid interface
where
r3 = radius to outside edge of dam
r4 = radius to surface of heavy-liquid down- Pi = density of the light phase
stream from the dam. ph = density of the heavy phase

The location of the interface is fixed by a balance In order for the centrifuge to separate the two
of forces arising from the hydraulic heads of the liquid phases, the liquid-liquid interface must be

two liquid layers. Expressing these forces as located at a radius smaller than r 3 but greater than
pressures gives that of the top of the overflow dam (r 4 ). More-
over, it is usually found that one of the phases is
more difficult to clarify than the other. To com-
• rf
dP=
-r f
—
dF
=
.rf adm •r f
2
p(oo r)(2irr/dr)
pensate for this, the volume of this phase must be
J
n
J
n A J
ri gc A 'r, (2irr/)gc made larger than the volume of the easily clarified
phase. This can be done by adjusting the height of
where the general case is used in which the liquid
the two overflow dams. Note that in the liquid-
layer extends from any initial radius (/-,) to any
clarifying centrifuge of Figure 22.17a, only one
dam is used, and the sole function of this dam is to
control the volume of liquid maintained in the
centrifuge.

Example 22.4 In the primary refining of

vegetable the crude oil is partially saponified
oils,
with caustic and the refined oil separated immedi-
ately from the resulting soap stock in a centrifuge.
In such a process, the oil has a density of
0.92 g/cm 3 and a viscosity of 20 centipoises, and
the soap phase has a density of 0.98 g/cm 3 and a
viscosity of 300 centipoises. It is proposed to
(a) (b) separate these phases in a tubular-bowl centrifuge
Figure 22.17. Overflow-dam arrangements in a tubular- with a bowl 30 in. long and 2 in. I.D. rotating at
bowl centrifuge, Tubular-bowl centrifuge arranged for
(a) 18,000 rpm. The radius of the dam over which the
clarifying a liquid of entrained solids, (b) Tubular-bowl light phase flows is 0.50 in., whereas that over
centrifuge arranged for separating two liquid phases. which the heavy phase flows is 0.510 in.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 629

(a) Determine the location of the liquid- Sedimentation

liquid interface within the centrifuge. The separation of a dilute slurry by gravity settling
(b) If this centrifuge is fed at a rate of into a clear fluid and a slurry of higher solids
50 gal/hr with feed containing 10 volume percent content is called sedimentation. The mechanism
soap phase, what is the critical droplet diameter of of sedimentation may be best described by obser-
oil held in the soap? vation of what occurs during a batch settling test as
solids settle from a slurry in a glass cylinder. Figure
Solution, (a) The interface location can be
22.18a shows a newly prepared slurry of a uniform
found directly from Equation 22.47.
concentration of uniform solid particles through-
r4
4
0.92 r2
2 - 0.510' out the cylinder. As soon as the process starts, all
r-i'
EL particles begin to settle and are assumed to approach
r
r
2
- r 2'
r. 0.98 r2
2
- O500 2
Ph 2 rapidly the terminal velocities under hindered-
settling conditions. Several zones of concentration
0.939(r 2
2
- 0.250) = r 2 2 - 0.260
will be established (Figure 22.18b). Zone D of

r2
2 = 0.426, r2 = 0.654 in.
settled solids will predominantly include the heavier
faster-settling particles. In a poorly defined
(b) The volume of soap phase held in the transition zone above the settled material, there are
centrifuge is channels through which fluid must rise. This fluid is
forced from zone D as it compresses. Zone C is a
30 xtt(1 2 - 0.426) region of variable size distribution and nonuniform
= 0.0313 3
ft
1728 concentration. Zone B is a uniform-concentration
zone, of approximately the same concentration and
This represents a residence time of distribution as initially. At the top of region B is a
boundary above which is clear liquid, region A. If
0.0313x3600 V the original slurry is closely sized with respect to
= 1 6s s =
50x0.10/7.48 Q the smallest particles, the line between A and B is
sharp.
The droplet size, of which 50 percent will
critical As sedimentation continues, the heights of
be separated from the soap phase in the 169-sec each zone vary as indicated in Figure 22.18b, c, d.
residence time, can be determined from Equation Note that both A and D grow larger at the expense
22.37. of B. Eventually, a point is reached where B and C

9 x (300 x 2.42) x 3600 1.0

D' =
- 2
2 In
(0.98 0.92)62.4 x (2tt x 1 8000 x 60) x 1 69 0.654
Dp '
= 26.6 x 10~ 6 ft or 8.1 microns

A
Clear-
liquid

.4-
B
Uniform —
concentration
B-
Transition
C i) HU i' lU
Variable
size and *4W
cone, zone c-
D \K'P' 'fl

C D- -*.
sZ1~M
ib) [c] ri;

Figure 22.18. Batch sedimentation.

630 APPLICATIONS TO EQUIPMENT DESIGN

Overflow
disappear and all the solids appear in D; this is V. ft
3
/hr
referred to as the critical settling point (Figure c.-,= 0(no solids)

22.18e)-that is, the point at which a single

distinct interface forms between clear liquid and
sediment. The sedimentation process from this point

on consists of a slow compression of the solids, with

liquid from the boundary layer of each particle
being forced upward through the solids into the
clear zone. Settling rates are veryslow in this dense
slurry. The phase is an extreme case of
final Lu , ft7hr
3
hindered settling. Equation 22.22 may be used to c u lb/ft
.

estimate settling velocities. It accounts for the Figure 22.20. Schematic diagram of thickener.
effective density and viscosity of the fluid but does of the bottom for discharge. The motion of the
not account for agglomeration of particles, so that rake also "stirs" only the sludge layer. This gentle
the calculated settling rate may be in considerable stirring aids in water removal from the sludge.
error. Continuous-Thickener Calculations. The
In batch-sedimentation
a operation as purpose of a continuous thickener is to take a slurry
illustrated, the heights of the various zones vary of some initial concentration of solids and through
with time. The same zones will be present in the process of sedimentation produce a slurry of
continuously operating equipment. However, once some higher concentration. The calculations
steady state has been reached (where the slurry fed necessary for the design of a continuous thickener
per unit time to the thickener is equal to the rate are governed by the settling characteristics of the
of sludge and clear liquor removal), the heights of solids in the slurry. The design of a thickener
each zone will be constant. The zones are pictured requires a specification of a cross-sectional area and
in Figure 22.19 for a continuous sedimentation. a depth. It is possible through the use of batch-
Industrial sedimentation operations may be settling information to design a unit to produce a
carried out batchwise or continuously in equip- specified product in a continuous manner. The next
ment called thickeners. The batch thickener oper- few paragraphs will indicate the calculation pro-
ates exactly like the example cited above. The cedures.
equipment is nothing more than a cylindrical tank Laboratory measurements of the type dis-
with openings for a slurry feed and product draw- in connection with Figure 22.18 are
cussed earlier
off. The tank is filled with a dilute slurry, and the the most satisfactory tests for determining the
slurry is permitted to settle. After a desired period settling characteristics of a given slurry or sludge.
of time, clear liquid is decanted until sludge Although these are batch tests, their analysis is

appears in the draw-off. The sludge is removed useful for the design of continuous thickeners.
from the tank through a bottom opening as indi- Figure 22.22 shows the results of this test where
cated in Figure 22.20. the height of the liquid-solids interface is obtained
Continuous thickeners (Figure 22.21) are as a function of time. Slopes of this curve at any
large-diameter, shallow-depth tanks with slowly point of time represent settling velocities of the
revolving rakes for removing the sludge. The slurry suspension at that instant and are characteristic of
is fed at the center of the tank. Around the top a specific solids concentration.
edge of the tank is a clear liquid overflow. The The early portion of the curve tends to be
rakes serve to scrape the sludge toward the center linear corresponding to a constant-rate settling of
Feed the sludge at the initial concentration. In thicken-
Ufr 7^>iClear liquor ing, this initial region represents a negligible part of
Clear liquor
overflow
Uniform feed
the total thickening time. As time increases, the
cone, zone
In-/Transition settling velocity decreases. Coe and Clevenger (14)
zone
in 1916 and in 1952 Kynch (30) proposed a model
variable -cone, zone
to explain this occurrence by assuming that the
settling velocity is proportional to the suspended
solids concentration. Once the constant settling
'

Thickened sludge outlet rate region of Figure 22.22 is passed, every point
Figure 22.19. Settling zones in continuous on the curve corresponds to a different solids con-
thickeners. centration.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 631

Figure 22.21. Continuous thickener. (Courtesy Dorr-Oliver, Inc.)

It must be emphasized that the settling curve interface was

function of local concentration.
a
determined in the laboratory applies only to the Batch-test clearly show that the settling
results
slurry tested and that the results may have some velocity decreases with increasing concentration.
uncertainties. For example, the largest-diameter But the decrease is less rapid than the increase in
test cylinder possible should be used to minimize concentration. This balance of decreasing linear
wall effects. Tests should also be carried out at velocity and increasing mass flow as a consequence
depths comparable to the depth of the proposed of greater density of solids in the compacting layer
unit. is unpredictable but is important in the design
process.

Design Calculations A necessary condition for functioning of a

continuous thickener is that the rate at which
Nearly all of the current methods for design calcu-
solids settle through every zone must be at least
stem from the work of Coe and Clevenger.
lations
fast enough to accommodate the solid being
As mentioned earlier, their primary assumption
delivered to that level. In the upper part of the
was that the rate of descent of the solids-liquid
thickener, the slurry is quite dilute, and settling is
quite rapid. In the bottom, the density and solids
concentration are extremely high and settling
velocity is low.
In athickener operating continuously, solids
pass through layers of all concentrations between

that of the feed and that of the final underflow, as

indicated in Figure 22.19. If solids entering any
layer do not pass through at a rate equal to that of
the arrival of solids into the layer, the layer will
increase in thickness, and it will grow upward until
solid ultimately appears in the overflow. Thickener
Time, 8
designis based upon identifying the concentration

Figure 22.22. Batch-settling results. of the layer having the lowest capacity for the
632 APPLICATIONS TO EQUIPMENT DESIGN

cross-sectional area of the cylinder in which the

(V + dv + vL ) (c - dc)
1

test is being performed. If a limiting layer exists, it

must first form at the bottom and move upward to

'//////////
C> V +

1
VL
W///////,
1k the clear-liquid interface.
the limiting layer
the interface is
is
If

c L and the time for

the concentration of

d Ll the quantity of solids passing

it to reach

through this layer is c L S6 L (v L + v L ). This quantity

Figure 22.23. Selected concentration layer
must equal the total solids present, since the layer
in thickener.
having this limiting concentration started forming
at the bottom and moved upward to the interface.
passage of solids through it under operating condi- Therefore
tions. This particular layer is called the rate-
limiting layer. Sufficient area must be provided to
cL Sd L (v L + vL ) =c z S (22.52)
ensure that the specified solids flux does not If z L is the height of the interface at d
L with
exceed the capacity of the rate-limiting area. vL being constant in accord with Equation 22.51,
Figure 22.23 depicts such a layer of concen- then
tration c in a batch test. This layer is assumed to be
zl
the rate-limiting one, so it may be viewed as rising vL = (22.53)
at a velocity v L . Solids settle into this layer from
just above, having concentration (c — dc) and Substituting the value of v L from Equation
velocity + dv) with respect to the column and
(v 22.53 into Equation 22.52 and simplifying gives
(v + dv + v L with respect to the layer. Solids settle
)

cqZq
out of this layer at a velocity v with respect to the cL I22.54)
walls and v + vL with respect to the layer. If the layer Z L + VL
is assumed to have a constant concentration of The may
laboratory-test data be treated by
solids, then by material balance
plotting the height of interface as a function of
(c - dc)S6 {v + dv + v L )
= cSBW + vL )
time, as Figure 22.24. From this plot, the value
in
of v L the slope of the curve at 6 = 6 L as shown
is
(22.48)
by Equation 22.55. The tangent with the curve at
where S is the area normal to the solids flow. L intersects the ordinate at z,. The slope of this
Solving Equation 22.48 for v L gives line is

- zL
vL = c —
dc
—
dv
-v-dv
— dv v (22.49)
Z;
= vL (22.55)

and by neglecting dv, or

dv Zi = zL +6 L vL [22.56)
vL = c—
dc
-v (22.50)

It has been assumed (14,30) that settling

velocity is function of concentration (that is,
a
v = f(c)), and it follows that dvldc = f (c). Equation
22.50 therefore becomes

vL =cf(c)-f(c) (22.51)

Since cconstant for this layer, f(c) and f{c)

is

are also constant, and v L is therefore also constant.

The constancy of vL in the rate-limiting zone may

be used to determine the concentration of solids at
the upper boundary of the layer from a single
batch-settling test. Let c and z represent the
initial concentration and height of the suspended

solids in a batch-settling test. The total weight of Time (0)

the solids in the slurry is c z S, where S is the Figure 22.24. Batch-settling results.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 633

Combining Equations 22.56 and 22.54 yields

c L Z; =c z (22.57)
40

It follows, therefore, that the height the slurry

z,- is

would occupy if all the solids present were at

30
concentration c L In terms of the model postulated
.

above, c L is the minimum concentration at which

boundary layers interfere. Z 20
The settling velocity as a function of concen-
tration determined from a single settling test as
10
follows. For several arbitrary d's, the slope of the
tangent and its intercept at d = are determined
from a plot of z as a function of 6. The value of
6 8 10 12 14 16
the intercept is used in Equation 22.57 to obtain
Time, hr
the corresponding concentration. From this, vL as
Figure 22.25. Height of interface as a function of time for
a function of c is obtained.
Example 22.5.

Example 22.5. A single batch -sett ling test velocity at that time is the slope of the curve,
was made on a limestone slurry. The interface dz/dd = v = 2.78 cm/hr, and c = 425gm/l. Other
between clear liquid and suspended solids was points are obtained in the same way, tabulated in
observed as a function of time, and the results are Table 22.2, and plotted in Figure 22.26.
tabulated below. The test was made using 236 g of
limestone per liter of slurry. Prepare a curve show- Table 22.2 SOLUTION TO EXAMPLE 22.5
ing the relationship between settling rate and solids
concentration. 6, hr Zj, cm v, cm/hr c. g/l

TEST DATA 0.5 36 15.65 236

1.0 36 15.65 236
1.5 23.8 5.00 358
Time, hr Height of Interface, cm
2.0 20 2.78 425
3.0 16.2 1.27 525
36.0
4.0 14.2 0.646 600
0.25 32.4
8.0 11.9 0.158 714
0.50 28.6
1.00 21.0
Thickener Area
1.75 14.7
3.0 12.3 The required thickener area is fixed by the layer
4.75 11.55 requiring the maximum area to pass a unit quantity
12.0 9.8 of solids. Therefore, it is necessary to determine
20.0 8.8 the concentration for such a layer.
Consider the batch -sett ling test in which the
total flux of solids is the sum of the flux of the
Solution. Using the test data, the height of settling particles and the flux to provide bulk flow
the interface (z) is plotted as a function of time id) to the underflow. This total flux is

(Figure 22.25). From the solid concentration of the Luc

initial slurry, F=F f + Fu = cv + (22.58)

CqZq = 236 x 36 = 8500 g cm/I where

From Equation 22.57, c = layer composition

v = settling velocity of the solids at c
8500 ,
Lu = underflow volumetric rate
c = g/l
A= cross-sectional area normal to flux

The tangent to the curve at 6 = 2 hr is found The first term on the right of Equation 22.58
to have an intercept of z, = 20 cm. The settling is the flux of the settling particles cailed the batch
634 APPLICATIONS TO EQUIPMENT DESIGN

16.-

:-

::

::

Co Ci C,
"
Solids concentration

Figure 22.27. Solids flux for continuous

thickener.

The area calculated from Equation 22.59 is

the minimum area that must be provided for a
continuous thickener operating at steady state.

Example 22.6. For the batch data of

Example 22.5, determine the area and required
underflow rate for a feed rate of 3.785 x 10 6
liters/day (1 x 10 6 gal/day). Feed concentration is
::: -:•: 500 600 700 s::
236 g/l (grams/liter) and the thickened sludge is
cL Solids concentrator!, g/
.

specified at 700 g/l. The underflow velocity is to

Figure 22.26. Settling-rate-concentration relationship, for
be 500 cm/day.
Example 22.5.
Solution. The solution is shown in Table 22.3.
flux and is obtained from the batch test data. The cm day
—g
I

second term is the flux associated with the with- Fu = cv = 500 x x c x =

u
day 24 hr 1000 cm 3
drawal of solids due to the underflow. A plot of
I

Equation 22.58 showing the relative contributions 0.0208c,

of the two fluxes is shown in Figure 22.27. cm 2 hr
In Figure22.27 the solids flux FL at concen-
tration c L is less than solids flux at other concen-
The total flux, F=FB +FUl is plotted against

trations between the feed and underflow concen-

composition in Figure 22.28. From Figure 22.28,
trations. Therefore, F L is the limiting solid flux
F L = 11.5 g/cm 2 -hr. Thus,
capacity, and an area must be provided so that the
6
day U 236 g
flux does not exceed F L Thus,
. the required area
^-oco
(3.785 x 10
day 24hr M l(^) I

for the thickener is

A =
*-oCo
FL 11.5g/cnrvMir
A = (22.59) 6 2 2
A = 3.24x 10 cm (=3483 ft )

Table 22.3 SOLUTION TO EXAMPLE 22.6

c, g/l* v, cm/hr* Fg, g/cm 2 -hr Fu g/cm 2 -hr

.
F, g/cm 2 -hr

236 15.7 37.05 4.91 41.96

358 5.0 17.9 7.45 25.4
425 2.78 11.8 8.84 20.6
525 1.27 0.667 10.90 11.6
600 0.646 0.388 12.5 12.4
714 0.158 0.113 14.9 15.1

Table 22.2.
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 635

26
—
section of this angle bisector line with the settling
- curve gives an estimate of the time d c at which the
24
solids enter the compression zone, and the con-
22 - centration at 6 C \sc c .

During the test, the mass of solids in the

20 -
cylinder is constant. A material balance on the
IF solids yields
18
c Az = cc Azc = c u Az u (22.60)
£ 16 -
1
or
Cqz q = Cqz = cu z u (22.61)
.§14
OS
where the subscript u denotes underflow values. If
I 12 /* the thickening process is thought of as going from
i/)

concentration c c to c Ul an amount of water must

1

TJ
1

o 10
(ft 'F u be removed to accommodate this change in con-
I

8 1
centration. The volume of water that is squeezed
I

out is

V = A(zc -z u
1

6 1
(22.62)
)

4
1
and the time required to discharge this volume of
water is (6 U — 6 C ). So the volumetric flow is
1

Co 1

2
1

V A[zc -z u )
c"
-d c -e c
1 i i
1 1

() 200 400 600 800 (6 u ) (e u )

Concentration, g/l
and solving for {d u — 6C )
Figure 22.28. Solution to Example 22.6.

=
A{zc -zu \

Use of the above procedure is straightforward,

(d u -d c ) (22.63)

but it may be of interest to know at what time, From Figure 22.29, the settling velocity at 6 C
and at what concentration, this critical flux is is obtained from the slope of the curve at 6 C or ,

realized. Talmadge and Fitch (55) presented a

method for this calculation, as well as a means to *1 ~Zc
vr = (22.64)
determine the minimum area.
The batch cylinder test is again the basis for
Under continuous flow conditions, the clear
the procedure. At the start of the test, the solids
liquid upflow cannot exceed vc if thickening is to
are uniformly dispersed throughout the cylinder at
occur. Therefore, the flow at 6 C when thickening
a concentration c The total mass of solids in the
.
starts is
cylinder is c Az where A is the cross-sectional
,

area of the cylinder and z is the initial height of

(z } -zc )\
(22.65)
the interface, which is, in this case, the liquid
depth. Let 6 C be the time needed to reach the
critical concentration c 2 Physically, this concen-
-

tration is reached at a time when the settling

velocity decreases rapidly.
Once again, the test data are plotted as in
Figure 22.21. This curve is used to determine d 2 Angle bisector line

and c 2 in the following way. The early portion of

Corresponds to c c
the curve represents "free settling" at nearly con-
stant velocity. Draw a tangent to this portion of
the curve. At the very end of the test, where
concentrations are high and velocities low, the
curve also shows nearly constant velocity. Draw a
tangent to this portion of the curve. The two Time.0

tangents are extended until they intersect. At the Figure 22.29. Determination of critical concen-
intersection construct the angle bisector. The inter- tration.
636 APPLICATIONS TO EQUIPMENT DESIGN

Substituting L from Equation 22.65 into Equation

22.63 gives
Zc ~ zu z1 ~~ zc
(22.66)
6,i — 9t

Equation 22.66 is used to calculate 6 U from Figure

22.29 as follows: z u can be calculated from Equa-
tion 22.60.
z oCo
zu = (22.67)
cu 8 10 12 14 16 18 20 22
Time, min.
On Figure 22.29 locate zu and 6 U can be calcu-
,

lated from Equation 22.66 or obtained from the

plot where the horizontal line through z u intersects
the tangent to the curve through 6 C
A horizontal line through zu = 11.7 intersecting
the tangent to the settling curve at 6 C = 6.5 min
gives d u = 1 1 .2 min. So,
.

The Talmadge and Fitch procedure applied to LqQu

A =
the batch-test analysis described above is used to zo
determine the minimum area for a continuous
day
thickener. The average mass rate at which the layer 4.5 x 10 6 11.2 min) 1000
of concentration c u forms in the thickener is day/ \1440 min t)
c u z u A/d u.But, since c u z u = c z according to
51 cm
Equation 22.60, the average mass rate is c z A/6 u .

= 6.92x 10 5 2 2
For continuous operations, the rate at which the cm (743 ft )

layer of concentration c u is formed must equal Thickener Depth. Comings (15), in a series
that at which solids enter the unit. of tests on a continuous thickener, determined the
LqCq =c z A/d u effect of underflow rate on thickener capacity.
or Essentially clear overflow was obtained from all
LqQu runs. The depth of the thickening zone increased as
A = (22.68) the underflow rate was decreased. He concluded
zo
that the depth of the thickening zone for incom-
pressible slurries is less important than the reten-
Example 22.7. A biological sludge from a
tion time of the particles within the thickener.
secondary waste treatment facility is to be concen-
trated from 2500 mg/l to 10,900 mg/l in a continu-
That is, the area of a thickener may be dependably
predicted from the settling limit described above
ous thickener. Flow to the unit is 4.5 x 10 6 l/day.
Determine the area required from the batch-test and from the volume of the compression zone
data given in Table 22.4.
determined to give necessary retention time. The
retention time may be determined from laboratory
Solution. The test data are plotted as shown data in a batch test. Therefore, since depths repre-
in accompanying illustration. Tangents are
the sentative of those expected in the full-scale
drawn and angle bisector constructed as described thickener were used in the batch tests, no separate
above, from which z c = 20 and 6 C = 6.5 min. The calculation for the thickened-sludge volume is re-
tangent to the curve at d c gives a value of Zy = quired.
32.5 cm. In addition to the depth of the settling zone,
From Equation 22.67, some provision must be made for inventory and
mechanical design features. The total depth of the
z c (51)(2500)
= = 11.7 thickener may be estimated (7) by adding to the
cm
cu (10900) depth of the settling and thickening zones.

Table 22.4 TEST DATA

0,min 1 8 12 16 20 25

Height of
interface, cm 51 43.5 37.0 30.6 23.0 17.9 14.3 12.2 11.2 10.7
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 637

Bottom pitch 1 to 2 ft in the two channels. The flow behavior in

Storage capacity 1 to 2 ft expanded, or fluidized, beds will be very similar to
Submergence of feed 1 to 3 ft that for packed beds, except that the flow passages
important to note that if the required
It is will be more open and almost continuously inter-
underflow concentration is less than the critical connected. In cases where the bed is falling down
value determined from the batch test, area calcula- through the rising fluid or where the fluid is con-
tions are based upon the unhindered settling rates. veying the solid phase, the mechanism is more like
that of the hindered settling of particles through
APPLICATIONS OF THE MECHANICS OF FLOW OF the fluid which has been treated above.
It is apparent from the physical mechanism
FLUIDS THROUGH PARTICULATE SOLIDS
discussed above that the momentum transfer from
In many industrial operations, a fluid phase fluid to particles arises from a form-drag and a
flows through a particulate-solid phase. Examples kinetic-energy loss, that is,
include filtration, heat transfer in regenerators and
T 9c = '^c'form drag + T 9c 'kinetic energy
pebble heaters, mass transfer in packed columns, \ 'total \

and skin friction

chemical reactions using solid catalysts, adsorption,
and flow of oil through the reservoir toward an oil (22.69)
well. In many cases, the solid phase is stationary, as At low rates of flow through very small passages,
it is in a packed distillation column; in some cases, the kinetic-energy losses are small compared to the
the bed moves countercurrent to the gas stream, as form-drag losses, but for high rates of flow through
it does in a pebble heater or in some catalytic large passages or fluidized beds, the kinetic-energy
reactors. In some cases, the fluid velocity is great losses may completely overshadow the form-drag
enough that the momentum transferred from the However, the transition will not be sharp,
losses.
fluid to the solid particles balances the opposing because the wide variety of parallel passages
gravitational force on the particles and the bed permits a wide variety of flow conditions.
expands into a fluidlike phase; in still other appli-
cations, the fluid phase carries the solid phase with Flow Through Packed Beds
it, as it does in pneumatic conveying.
Fluids are forced to flow through stationary beds
The rate of momentum transfer from the fluid
of particulate or porous solids in a wide range of
to the solid particles, and therefore the pressure
practical situations including moisture assimilation
drop for flow through the bed, is related to the
by soils, adsorption, ion exchange, and many of
physical mechanisms by which flow occurs. In a
the examples noted in the introductory section. In
packed bed, the flow path is made up of many
these applications, the form-drag losses can be
parallel and interconnecting channels. The channels
related to the flow conditions through modifica-
are not of fixed diameter but widen and narrow
tion of the relations previously given for friction
repeatedly, and even twist and turn in varying
losses in ducts. For this case,
directions as the particles obstruct the passageway.
The channels do not even have the same average 2(-AP) f gc D
cross section or total length. In flowing through f= = 0(/V Re )
= 0(—
r,2.
v*pL n
\

these passages, the fluid phase is repeatedly acceler-

ated and decelerated and experiences repeated (22.70)
kinetic-energy losses. In addition, the rough sur-
where -APf = pressure drop due to friction and
faces of the particles produce the usual form-drag
and skin-friction losses. The flow through large 8(r„fe )i
f= 13.9a)
open channels will be at higher velocity than the 2
v P
flow through parallel narrow constricted channels,
because the pressure drop per foot of bed length have been previously presented. For noncircular
must be constant regardless of the channel under ducts, an equivalent diameter must be used.
consideration. For this reason, the transition from 4S
laminar to turbulent flow will occur at a much = 13.20)
£>eq
lower bulk flow rate in open passages than it will in
restricted channels. Moreover, at the convergence This defines the equivalent diameter of a duct as
of two channels, eddy currents and turbulence will four times the cross-sectional area divided by the
be promoted because of the inequality of velocity wetted perimeter. In terms of a packed bed, Equa-
638 APPLICATIONS TO EQUIPMENT DESIGN

tion 13.20 can be modified to Inthe region of low flow rate and small
particle size or, in other words, at low A/ Re
4S L total volume of voids
(laminar flow). Equation 22.74 can be used to
b L total surface area of particles express the entire pressure drop, because the
kinetic-energy losses are small. Under these condi-
AeNVn
;i3.20a) tions, the drag coefficient and hence the friction
(1 -e) • A/A factor are inversely proportional to the Reynolds

where number.

e = porosity, fraction of total volume that (13.10)

is void N Re
N = number of particles
NVP = total volume of solid particles where k] is a proportionality constant.
NA P = total surface area of solid particles Combining this equation with Equation 22.74
yields
Inserting Equation 13.20 into Equation 22.70, 3
e (-AP) f9c D p .
(1-e)/i
(22.75)
8(-AP) fgc eVp ( 4eVp vp \
(1 - e)vs 2
pL '

D p vs p
= =
f
v
2
pL^-e)A p Mm
<

*\{1-e)ATT~)
\(1 - e)A p pf
(2Z71)
which rearranges into

It is usually convenient to express this equation in (-AP) fgc (1-e)V*

= k (22.76)
terms of the superficial velocity, that which would 3 2
e Dp
occur with the actual mass-flow rate flowing
through the container holding the bed if this con-
where and k 2 are constants. Equation 22.76 is
k-\
tainer were empty. By definition, then, vs = ev
known Carman-Kozeny equation and has
as the
where vs is the superficial velocity. Also, it is
been successfully used to calculate pressure drop
common practice to replace the Vp and A p terms
for laminar flow through packed beds. It was
with a "particle diameter." The particle diameter is
originally derived by Kozeny (28), who used the
defined in terms of the diameter of an equivalent
simplified model of a number of parallel capillary
sphere. (See Appendix B.)
tubes of equal length and diameter to describe the
For a sphere,
packed bed. If this model were exact, k, would be
tt(D sp ;
64, as given by Equation 13.10. Carman (13)
3
applied this equation to experimental results on
Va spi) /6
tt(D sd
flow through packed beds and found that k 2 =
from which 180.
If the A/ R e is high, kinetic-energy losses

Dsp= (22.72) become significant. These losses may be found by

AJVD modifying the kinetic-energy term in the general
Equation 20.18.
relation,
where D sp = diameter of a spherical particle

For an irregular particle of known A p to Vp ratio,

there is only one size of sphere having this same P 2# c
ratio. The diameter of this sphere is taken as
But, if the energy loss is to occur repeatedly in a
characteristic of the particle.
unit channel length,

Dp = — =D sp (22.73) (-AP) k gc
=
n
— Hpv* (22.77)
_£ L 2

where
Writing Equation 22.71 in terms of vs and Dp gives
n = number of repetitive kinetic-energy
losses in a unit length
8{-AP) fgc e 2 D,
f= -APk = pressure drop due to kinetic-energy
y I22.74)
6vs 2 pL(1 -e) U(1 -e)p/ losses
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 639

In the channels under consideration, the expan- Equation 22.81 and the superficial velocity, (vs )

sions in channel width probably occur at distances may be inserted into Equation 22.78.
roughly equivalent to the channel diameter. This 2
follows because the particulate bed has been
replaced by a model consisting of many parallel
circular ducts. The diameter of these ducts will be
—i-AP) k gc
;

L
=k
2
3
e

4eD,
2
= kt
(1 -e)

proportional to the particle diameter, and the fact

of one expansion occurring for each particle is thus 6(1 -e)
approximated. Using n proportional to 1/D C gives (22.82)
r,2
(-AP) k g c k 2 pv This relation was first derived by Burke and
(22.78) Plummer (11) to express the pressure drop result-
D r
ing from turbulent flow through packed beds.
were £3 is constant and D c is the channel
a Equation 22.76 for the pressure drop caused by
diameter. As with the pressure drop due to form form drag and Equation 22.82 for the pressure
drag, it is convenient to convert this equation to an drop caused by kinetic-energy losses may now be
expression in terms of D p and vs As before, .
added to obtain the total pressure drop resulting
from flow through the bed (18, 36).
v =
i-AP)gc €) JWs_ 2
(1 (1 -e) pv
~ ^2
6Vn 1

Dn = (22.73)
(22.83)
and
This equation has been developed under the
Ne LitD, assumption that the velocity is constant through-
(22.79) out the length of the bed. With gas flow at high
VD LnD 2 pressure drop, this would not be the case, and a
1 -e)
differential form of Equation 22.83 would have to
be written and integrated for the full bed depth. If
where
such an integration is carried out by assuming
Nc = number of channels in the area of the isothermal expansion of an ideal gas, Equation
bed 22.83 becomes
D = bed diameter

In Equation 22.79 the bed has again been likened

l-AP)gc
= k:
!1-e) Mm- + k , d -e) Gvsm
r—
2 4
to a largenumber of capillary tubes of fixed length
DP
and diameter. The particle surface area (A p ) has (22.84)
been taken as equal to the wall surface of these
where
capillaries. Also,
vsm = superficial velocity at the average of
Nc itD c 2 ttD
2
inlet and outlet pressure
(22.80) = mass rate of flow (pws
G ) based upon
total bed cross section
so that
Equation 22.84 can be rearranged into an equation
NC D C <

D2 = (22.80a) relating a modified friction factor and a A/ Re as

follows (17):

Combining Equations 22.79 and 22.80a gives AP)gc D, N Re

= k? + k i
NC D C 4€ VVsm (1-e)' 1 -e)
V* ND
— C C
2
M 0cd-e) P
(22.85)
(1 -e)
x

4e Here, the term on the left side of the equation is an

(22.81) averaged friction factor modified by the inclusion
640 APPLICATIONS TO EQUIPMENT DESIGN

of the porosity terms. Using this relation, the con- of these equations. Typical units might be as
stantsk 2 and £ 4 can be readily determined from follows:
experimental data by plotting
—AP= pressure drop through the packed bed,
2
(-AP) DP 2 lb,/ft
9c L = bed length, ft
Mm -e)'
Dp = particle diameter, ft
3
as a function of Since Equation
/V Re /(1-e). p = fluid density, lb/ft
22.85 is groups of variables, straight
linear in these vsm = superficial velocity at a density aver-

lines result where the slope is k* and the intercept

aged between inlet and outlet con-
at /V R e /( 1 - e) = is k 2 A large amount of experi-
.
ditions, ft/s
e = bed porosity, dimensionless
mental data on flow through beds of granular
/V Re = average Reynolds number based upon
solids show that k 2 = 150 and k A = 1.75 (17). The
superficial velocity (D p vsm plp), dimen-
modified friction-factor term of Equation 22.85 is
sionless
the ratio of the total pressure drop to the pressure
drop due to form drag and skin friction. Usually,
For liquids, vsm may be taken at any point in the
the friction factor is given as the ratio of the total
bed, but gas-density variations require the use of
pressure drop to the kinetic-energy losses. Re-
the average velocity as specified above.
arranging Equation 22.85 in this way by dividing
In Figure 22.30 is shown the mean curve
by [D p vsm pm /n)/^ -e) and inserting the values
obtained when experimental data from several
of k 2 and k 4 found experimentally gives
sources (11, 17, 43, 45) are plotted as

[-AP)gc Df (1 -e) [(-AP)g c Dp /Lpvsm 2

][e
3
n-e)}
= 150— + 1.75
L pVsrr 1 -e) N Re as a function of [/V Re /(1 — e)] . The solid line is a

(22.86) plot of Equation

and the dotted lines
22.86,
marked Carman-Kozeny and Burke-Plummer are
Both 22.85 and Equation 22.86 are
Equation plots of Equations 22.76 and 22.82, respectively.
dimensionless,and hence any set of consistent The data points themselves have been removed for
units may be used throughout any complete term clarity. They scatter smoothly around the line

80

\^
g
40

S
%
^
Q.
20 \ \
K

•= 10
iv
\\ \
iquat ion 22 S 5

\
\ '

V>
i

5 4
^>^^
^""-o^.
>^_
Burke-Plummer

1.0 III
1.0 2 3 4 567 8 10 20 40 60 100 200 400 1000
Afc d-e)

Figure 22.30. Pressure drop for flow through packed beds.

 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 641

plotting 22.86 showing no systematic

Equation bed porosity (e), and the mean superficial velocity
deviation, and a maximum
deviation of about 20 must first be determined.
percent if a few erratic points are eliminated. This For the £-in. cubes,
is excellent agreement in view of the wide range of

variables covered by Figure 22.30. Note the simi- 6Vn 6x0.25 3

larity in shape between the curve shown here and
D = 0.25 in.
>=A, 6 x 0.25 :
the drag-coefficient— Reynolds-number plot
(Figure 13.5). The gradual transition between The porosity can be determined using Figure
laminar and turbulent regions, which covers a range B-1 2, Appendix B, which gives porosity as a function

of /V Re /(1 — e) between 5 and 2000, results from of sphericity (<//). By Equation B-26,
the widely differing flow passages through the bed,
with the resulting wide variations in actual velo- (a25) 2/3
city.
* =
*(-?) *U
Other relations for the pressure drop through
6 x 0.25 2
packed beds have been found by empirically corre-
3/2 -6 \ 2/3
lating experimental data. In this way, Leva and / tt
= 0.806
Grummer (36) obtained
U 3 '2
-

(-A/3 )^ _ 0.0243G 1
V- 1
\ 1 - 1
(1 -e) From Figure B-1 2, this gives

~ 11 3
L p p e e = 0.44
(22.87)
Fundamentally, vsm must depend upon the
pressure drop, since the gas specific volume is
to which they applied different constants for
particles of different surface roughness. Here X is a
pressure dependent. Here, however, as in many
pressure-drop computations, the pressure drop will
shape factor obtained by the relation
probably be small relative to the total pressure, so
that the effect of this —AP on gas density will be
X = 0.205 (22.88) small. With this simplification, fs1 and vs2 can be
2/3
V, fixed directly. Otherwise, a trial-and-error solution
would be necessary to find v2 . On this basis,

Leva's experimental data were all in the range of

= TT-n— —
high /V Re and relatively large particles. PM 100x29
relatively
Data on both liquid and gaseous fluids were in- Pi = =" = 0.500 lb/ft
3
RT) 10.73x540
cluded in the correlation so that G can be either
Gv or GL PM _ 100x29 3
.
= = 0.314 lb/ft
P2 ~
Brownell and Katz (10) correlated results RT2 10.73x860
obtained mainly for very fine particles by using the
and
standard f versus /V Re plot by multiplying f and
/V Re by factors dependent upon the particle spheri- 1000 1000
city and the bed porosity.
+
0.500 0.314
vsm = = rtBgvex /u
2595 ft/hr
Example 22.8. A bed of 4 -in. cubes is to be
used as packing for a regenerative heater. The Using these values, the group /V Re /(1 — e) can be
cubes are poured into the cylindrical shell of the calculated, and the modified friction factor can be
regenerator to a depth of 10 ft. If air flows through read from Figure 22.30.
this bed, entering at 80° F and 100 psia, leaving at
400° F, and flowing at a mass rate of 1000 Ib/hr ft 2 DP G
of free cross section, determine the pressure drop 0.25 x 1000
/V Re Mm
across the bed. = 669
1-e 1-e 12(1 -0.44) x (0.023x2.42)
Solution. This problem can be solved by 3
l-AP)gc D p e
applying Equations 22.85 and 22.86 or Figure = 2.00 (from Figure 22.30)
2
22.30. In any case, the particle diameter (D p ), the Lp m Vsm d-e)
642 APPLICATIONS TO EQUIPMENT DESIGN

This equation may be solved for (-AP). Here, For a given bed, Equation 22.91 may be written
Pm vsm is replaced with G. for both theunexpanded and the expanded state at
the pressure drop required for bed expansion.
2.00 x 1 x 1 000 x 2595 x 0.56 2 From the difference in these two states,
-AP =
!32.2x3600 2 ) x —25
12
x 0.44 3
= 40.4 ltv/ft

150a/
-Vsm °) =
-0

2 (Vsm
g(ps - p)D p -e°
-AP = 0.28 psi
1

(22.92)
The assumption of a relatively small —AP has then
been justified for this case. Larger relative pressure where the superscript refers to the condition
drops can be expected if the particle diameter is where the fluid velocity is just insufficient to
small. expand the bed. Equation 22.92 relates the bed
porosity and therefore the extent of bed expansion
Bed Fluidization to the mean superficial velocity of fluid passing
The packed bed expands when the pressure drop through the bed. It applies only where the pressure
due to the upward flow of fluid through a granular drop is constant and balances the force of gravity
unrestricted bed equals the weight of the packing. acting on the solid particles and where Dp is small.
As the bed expands, it retains its top horizontal The pressure-drop behavior of a packed bed as
surface with the fluid passing through the bed the velocity of flow up through it increases is
much as it did when the bed was stationary. Now, illustrated in Figure 22.31. Between points A and
however, the porosity is much greater, and the B, the bed is stable, and the pressure drop and
individual particles move under the influence of Reynolds number are related by Equation 22.86.
the passing fluid. The bed has many of the appear- At point B, the pressure drop essentially balances
ances of a boiling liquid and is referred to as being the bed solids weight. Between points B and Cthe
"fluidized." Writing the force balance on a section bed is unstable, and particles adjust their position
of bed of length L when the pressure drop equals to present as little resistance to flow as possible. At
the gravitational force gives point C, the loosest possible arrangement is
obtained in which the particles are in contact.
(-AP)gc Beyond this point, the particles begin to move
(1 -e)(ps -p) -g (22.89) freely but collide frequently so that the motion is

similar to that of particleshindered settling.

in
where ps = solid-particle density. The extent of bed Point C is referred to as the "point of fluidiza-
expansion may be obtained by eliminating tion." By the time point D is reached, the particles
(-AP)gc /L between Equations 22.84 and 22.89. are all in motion, beyond this point, increases in

150 —^
(1 -e) pv,
+ 1.75^— (p,-pfc
(22.90)

Since pressure affects the density and hence the

velocity of a gas in the fluid phase, this equation
should be solved rigorously by successively insert-
ing vs varying between inlet and outlet. The bed
porosity would show equivalent gradation. This
effect is usually small, and the use of mean values
of p and vs is usually satisfactory. In most indus-
trial applications involving expanded or fluidized

beds, the particle diameter is small, and vsm is also

small. In these cases, the second term of Equation
22.90 is negligible compared to the first, so that log .v

Figure 22.31. Fluidization of a bed of par-

150^
(22.91 ticulate solids (61). (By permission of Chem.
(Ps ~ p)gD p 2
1 - e Eng. Prog., copyright © 1948.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 643

/V Re result in very small increases in —AP as the and after the particles have had a chance to adjust
bed continues to expand and the particles move in to a position offering minimum pressure drop.
more rapid and more independent motion. Ulti- Thus, at point C,
mately, the particles will stream with the fluid, and
-A/>
the bed will cease to exist. This occurs at point E. = ( 1 -e c )(ps - p G ) (22.94)
Two main types of fluidization have been
noted experimentally (61 ). In cases where the fluid The porosity minimum fluidization e c will be
at
and solid densities are not too different, where the slightly greater than that for a packed bed, and
particles are small, and therefore where the velo- hence L c will also be greater than the height of the
city of flow is low, the bed fluidizes evenly with bed without flow. e c will also depend upon the
each particle moving individually through a rela- shape, surface characteristics, and size of the solid
tively uniform mean free path. The dense phase has particles. For example, for mixed round sand
many of the characteristics of a liquid. This is called (t// = 0.86) e c — 0.42, for absorption carbon
particulate fluidization. Where the fluid and solid e c — 0.72, and for anthracite coal (i//=0.63)
e c ^ 0.60 (35).
densities are greatly different or the particles are
large,the velocity of flow must be relatively high. As the fluid velocity is increased from C
In this case, fluidizationis uneven, and the fluid
toward E (Figure 22.31) the bed expands, some-
passes through the bed mainly in large bubbles. times up to 2.5 times L c The exact amount of bed .

These bubbles burst at the surface spraying solid
particles above the bed. Here, the bed has many of
the characteristics of a liquid with the fluid phase
acting as a gas bubbling through it. This is called
aggregative fluidization. In addition to these two
main classifications, fluidized beds show consider-
able bypassing and slugging — that is,
and unstable flow of large gas bubbles through the
bed. Both these behavior patterns are strongly
affected by the design of the distributor that holds
up the packed bed and distributes gas flow into it.
The presence of particulate or
fluidization must be the result of the influence of
gravitational forces on the phases present in a
fluidized bed as well as the bulk fluid mechanics of
the system. Thus, the Froude number (v2 IDp g),
which is the ratio of kinetic to gravitational energy,
would be expected to be one of the criteria. Orig-
inally, the Froude number was taken as the entire
expansion depends upon the distributor design and
bed diameter, and hence is difficult to predict on
the basis of laboratory or pilot results (29).
Equation 22.92 gives an estimate of this expansion
for particulate fluidization in a bed where wall
effects and recirculation are absent. In designing a
intermittent fluidized bed, the engineer must allow not only for
this expansion but also for the height above the
top of the bed required to get adequate solid-fluid
separation, the transport disengaging height (TDH).
This also depends upon the characteristics of
aggregative the fluid-solid system and has not yet been
adequately correlated. Figure 22.32 gives an
empirical correlation for estimating TDH for fluid
catalytic cracking catalyst (64) and shows typical
TDH values. At heights above TDH, there will still
be catalyst where there is particle
losses, especially
40 ^ 1 !N^ 1

criterion (61), but more extensive information has \ s.

indicated that if

10

W Fr )(/V Re )(^—^)(-)<100 (22.93a)

=

_
^>$% \V^ ^t)^^
^Ss,
s.
l uq
OvW.22
= 1.53

92
cm/sec -
:

particulate fluidization occurs, whereas

- 0.61 </>0\*
-y^ ^0>
\\\\
if
0.46
0.31 S \
(/V F r)(/V Re )
(^—^)(p) > 10 ° (22.936)
E
-
aggregative fluidization exists (48). In Equations 04 .
i i.i i ?>
0.03 0.10 1.00 10.00
22.93 the /V Fr , /V Re , and L should all be taken at Vessel diameter d,(cm)
the point of minimum fluidization. Figure 22.32. Empirical correlation for estimating
As indicated above, the minimum fluidization the transport disengaging height for fluid catalytic
point, point C in Figure 22.31 , will occur when the cracking catalyst (64). (By permission of A.I.Ch.E.,
pressure drop due to flow balances the bed weight, copyright© 1958.)
644 APPLICATIONS TO EQUIPMENT DESIGN

breakage or widely varying particle sizes within the that the catalyst is eroded and broken by its con-
bed, but these losses will not be greatly reduced by stant motion, so that there continual attrition of
is

further free space. the particles requiring continual makeup of fresh

Industrially, it has been found advantageous catalyst.
to carry out many solid-catalyzed reactions in The industrial application of
largest-scale
fluidized beds. The circulation bed and
of the fluidization the fluid catalytic cracking of
is in

uniform agitation within it prevents the occurrence heavy crude-oil fractions to give gasoline compo-
of hot spots and dead regions. It also makes nents. This process was developed during the early
possible the continuous circulation of the catalyst part of World War when
there was a critical need
II

between the reaction vessel and a regeneration from crude oil and has been
for high gasoline yields
vessel. The major disadvantage of this system is very widely applied since then. Figure 22.33 shows

Figure 22.33. A 23,000 bbl/day fluid catalytic cracking unit. The large vessel
just to the right of the elevator shaft is the regenerator. The reactor is the next
vessel to the right of the regenerator. This is an Esso Model IV unit built at Port
Jerome, France. Units of this type first went on stream at the end of 1952 and
have been built in sizes from 5,000-55,000 bbl/day. (Courtesy Exxon Research
and Engineering Company.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 645

Example 22.9. A catalyst having spherical particles

of Dp = 50 microns and p s = 1.65 g/cm 3 is to be
used to contact a hydrocarbon vapor in a fluidized
reactor at 900° F, 1 atm pressure. At rest the bed
has a porosity of 0.35 and a height of 3 ft. At
operating conditions, the fluid viscosity is 0.02
centipoise and its density is 0.21 lb/ft 3 Determine .

the superficial gas velocity necessary to fluidize the

bed, the velocity at which the bed would begin to
flow with the gas, and the extent of bed expansion
when the gas velocity is the average of the velo-
cities previously determined. The porosity at mini-
mum fluidization velocity is 0.42. Does aggregative
of particulate fluidization occur?

Solution. The mean

superficial velocity at the
start of fluidization
can be obtained directly from
Figure 22.34. Schematic diagram of fluid catalytic-cracking
process.
Equation 22.91. Thus,

a typical fluid-cracking unit, and the catalyst and 3 2

=
e ( Ps-p)gDp
fluid flow through this unit is diagrammed in (vsm °)
Figure 22.34. 1 -e 150/x
Regenerated catalyst at about 1050°F is
blown by the vaporized feed into the reactor
—- (1.65 x 62.4- 0.21 )32.2f --^—)
vessel. The hot catalyst preheats the feed and is 0-42 3 \25.4x12/
conveyed at a rate of about 7 lb of catalyst per X
0.58 1 50 x (0.02 x 0.000672)
pound of feed as required by the system energy
balance. Cracking takes place in the reactor at
about 900° F with coke depositing on the surface 0.128 x 102.8 x 32.2 x (0.0269 x TO" 6 )

of the activated natural clay or synthetic silica- 0.002014

alumina catalyst. Spent catalyst collects in a stand-
pipe and flows through a valve into an airstream
that conveys it back into the regenerator. In the
regenerator, the coke is burned off the catalyst,
whereupon the regenerated catalyst collects in a
second standpipe for return to the reactor. A
25,000 bbl/day unit such as the one shown in
Figure 22.33 contains about 500 tons of catalyst
of 20- to 400-micron particle diameter circulating
= 0.0056 ft/s
isthe velocity at which fluidization will begin.
The bed will disintegrate and stream with the
flowing gas when the gas velocity equals the velo-
city of free fall of the particles. For these small
particles, the flow is laminar, and the settling

at a rate of 20 tons/min and transferring velocity can be determined directly from Equation
100x 10 6 Btu/hr of heat from the regenerator to 22.15.
the reactor. Only about 1 percent of the circulating
catalyst is entrained by the fluids, and it is separ-
(Ps p)g
2
D1£_- 1 02.8 x 32.2 x (0.0269 x 1
0" 6 *

ated in cyclones before the fluids leave the con- D

Vr =
tacting chambers (25). ^8p 18 x (0.02x0.000672)
Although the fluid-cracking process is the = 0.37 ft/s
largest-scale application of fluid-bed reacting,
fluidized reactors have been developed for many
processes including ore roasting, cement manu- The extent of bed expansion at any inter-
facture, extraction of oil from shale and bitumi- mediate flow rate can be determined from
nous sand, production of phthalic anhydride by Equation 22.92 by solving for e. This was not
oxidation of naphthalene, and oxidation of ethy- possible when finding the maximum value of vsm
lene to form ethylene oxide (50). because Equation 22.92 becomes indeterminate
646 APPLICATIONS TO EQUIPMENT DESIGN

when e -> 1 . Solving Equation 22.92, sizes at flow rates high enough to fluidize all the
bed particles.At these flow rates, a large propor-
-03
150/i tion of the fines would be expected to be entrained
v< m - v. ') +
in the gas phase. For example, cracking catalyst
1 - e g(ps - p)D p 2
normally ranges between 20 and 500 microns in
0.002014
diameter (40) and is fluidized at superficial
32.2 x 102.8 x 0.0269 x 10~
6
velocities of 1 to 2 ft/sec. As seen in Example 22.9,
0.42 3 these velocities are great enough to entrain
/0.37 + 0.0056 „ nn \
x 0.0056 + particles somewhat greater than 50 microns. How-
\ 2
'
0.58
ever, in operation, relatively little entrainment of
solids occurs. This apparent violation of Stokes's
= 23.5(0.1 87 + 0.0027) + 0.1 28 = 4.18 + 0.1 28
law probably results from the flocculation of the
= 4.308 small particles into agglomerates perhaps held

= 0.855
e at
previously determined velocities.
=0.190 the average of the together by static electricity. Such agglomerates
vsm ft/s,
have been observed and seem to be particularly
large and stable in particles of about 10 microns

The type of fluidization that will occur can be and smaller diameter. Similar effects were noted by
inferred from the use of Equations 22.93. Morse (44) in examining the available experimental
data on pressure drop and porosity of fluidized
2
( Ps
~
P\ L = v DpVp p s - P L beds. Below A/ R e of about 10, he found that the
x/V Re x x x x
A/ Fr
\
)n
p 1 D\ D p g 7r
ii P
n
D modified friction-factor term of Equation 22.86
was lower for fluidized beds than for fixed beds,
v3 (ps -p)L but, above /V Re = 10, the modified friction factor
gDp for fluidized beds was greater than that for fixed
1-0.35 beds. In the low A/ Re range, this deviation is
3
(0.0056) (1.65x 62.4 -0.21) x3
1-0.42 explained on the basis of flocculation, since the

a05 ° experimental data in this range were obtained by

32.2 x 0.02 x 0.000672) Leva and coworkers (37) with small particles. The
25.4 x 12
data of A/ Re > 10 were obtained by Wilhelm and
= 856 Kwauk (61) with larger particles. The deviation
here was explained as the result of the kinetic
Thus, we would expect aggregative fluidization, in
energy imparted to the solid particles by the fluid
accordance with Equations 22.93.
stream. The comparison between fixed- and fluid-
Even though we do have particulate fluidiza-
ized-bed pressure-drop data given by Morse is
tion, it is probably necessary to redo the calcula-
shown in Figure 22.35 (44).
tion on the bed expansion at the stated conditions.
An examination of that result leads to that con-
10.000
clusion. At the porosity calculated, the bed height
would be
1000

L_ 1-e (
0.58
= 4.00
1-e 0.145 5 loo

Thus, the bed is four times its original height. Such

an extreme expansion is sure to produce bypassing
and slugging so that the bed would be much less
expanded than calculated.

Equations 22.91 and 22.92 were obtained by

1.0 10 100 1000 10.000 100.000
applying to the Carman-Kozeny equation the re-
Mod fied Reynolds number .\'
Rt
. 1-e
striction of constant pressure drop equivalent to
the bed weight (Equation 22.89) and allowing a Figure 22.35. Comparison between fixed- and fluidized-
resulting variation in porosity. Fluidized beds are bed momentum transfer (44). (By permission of Amer.
normally operated with a wide range of particle Chem. Soc, copyright © 1949.)
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 647

Since fluidized beds are used for such diverse Equations 22.95a and 22.95b can be used to
purposes as catalytic reactions, coal combustion, obtain the unknown velocities and void fraction at
and particulate drying, the rates of heat and mass choking with two different sets of known condi-
transfer between the particles and the fluid are of tions. In one situation, the gas velocity and void
special interest. Determination of these transfer fraction at choking can be obtained by simul-
rates has been the subject of much work (29, 64). taneous use of these equations when the superficial
A correlation for the particle-fluid heat transfer solids velocity is known, as would be the case if the
coefficient is (4) solids were fed into the gas flow at a measured
,2/3
dM
-B
-e)
-
1/4 rate. In the other situation, the void fraction at

" 2 = 0.043 choking can be determined directly from Equation

-
gp(p s p)
Gk 2l2 c P 22.956 if the gas velocity itself is known.
(22.95) Where particles of a distribution of sizes are
where all the properties are those of the fluid being conveyed, as would almost always be the case.
unless indicated. Equations 22.95a and b can be modified by in-
Fluid-Solid Conveying, Pneumatic Con- cluding a volume fraction distribution of the
veying. When the fluid-phase flow rate exceeds particles (38)
the free-settling velocity of the particles, the fluid-
ized bed loses its identity because the solid *ftt = *ti(1 - ec ){vec - vtl ) (22.96a)
particles are conveyed in the fluid stream. This -°- 47
[Xx ti (ve c - vXl )]
2 = 200 <70(e c - 1
method of conveying is frequently used through-
out industry usually with air as the fluid phase; for (22.966)
instance, to unload grain ships, to convey the 2xt = SXfi = 1 (22.96c)
i

product from spray dryers, to fill and empty

cement silos, etc. It has the advantage of cleanli- where x fi = volume fraction of feed particles with
terminal velocity v t[
ness, low loss, and the ability to move large quanti-
where x tj = volume fraction of particles in the riser
ties of solids rapidly. On the other hand, significant
with terminal velocity v X[
breakage of solid particles may occur, and pipe
erosion may be excessive.
The transition from moving bed flow to slug-
As the vertical gas-phase flow rate increases ging dense phase flow occurs when the voidage in
beyond the fluidization region, the solid particles
the conveying tube equals that at incipient fluid-
are initially moved with the fluid almost as a
ization (39). If a typical void fraction at initial
moving fluidized bed. This flow region is usually
fluidization is taken as 0.45, then
called moving bed flow. At higher gas rates, the
solids are carried in gas slugs, much as occurs in 0.55 ve c - 0.45 vs = 0.55 v, mf (22.97)
aggregative fluidization. This is called slugging
The use of Equations 22.95 and 22.97 allows
dense phase flow. Finally, at still higher gas rates,
the solid bed loses all cohesion and the particles
the phenomenon of gas-solid vertical conveying to
be broken into three regimes, as shown in Figure
stream with the gas as dispersed individuals. This is
22.36. Since each of these exhibits a different
called dilute phase flow (38).
physical situation, would be expected that each
it
The transition from slugging dense phase flow
must be treated separately in predicting the
to dilute phase flow is referred to as the choking
pressure drop. This is the case, and hence separate
point and usually, but not always, is an abrupt
pressure drop correlations for these regions are
transition. Conditions at which this transition
presented below.
occurs have been correlated by Yang (62) to about
a ±30 percent accuracy
1. Dilute Phase Flow Regime. Since this
vs = (ec v-vt )C\-ee ) (22.95a) region shows steady, independent flow of gas and
solid phases, the energy balance can be written for
[e c v- v t ) 2 /2gD = 100(e c -°- 47 - 1)
the two phases and simplified to give the needed
(22.956) relationships. From such a general expression (53)
where vs = superficial solids velocity
vt = terminal velocity of a solid particle in 2-.L

the gravity field AP = pev' + + g\ [(1 -e)ps + ep)dZ

v = gas velocity at the choking point
1 -eJ
e c = fraction voidage at choking + ArL/D (22.98)
648 APPLICATIONS TO EQUIPMENT DESIGN

2. Dense
Phase Slugging Flow. Here the
pressure dropalmost entirely a result of the
is

weight of the solids, and Equation 22.99 can be

/Equation simplified to
6 / 22.97
Moving bed / A/> = ps (1 -e)gc L (22.99a)
flow /
Use of even this simple equation requires deter-
M
O / Dense phase mination of the void fraction. The following
/Equation
/ slugging flow / 22.95 equation has been proposed (41)

/ / Dilute phase
/ / flow 1 -e mf
Log (dimensionless gas velocity). log(i7i' m /) iv+Vb-vm f+ws /ps ) (22.103)
Figure 22.36. Quantitative flow regime dia-
gram in vertical pneumatic conveying for a where vb = bubble rise velocity in a f luidized bed
choking system (38). (By permission of
vb = 0.35 y/gD for slugging conveying
Amer. Chem. Soc., copyright © 1976.)

3. Moving Bed Flow. This flow regime has

where the three terms on the right-hand side of the
practical application in conveying systems in
equation represent pressure drop due to accelera-
limited situations, such as for short-distance solids
tion, gravity, and wall friction, respectively, and
moving. In this regime, the pressure drop can be
where r is the shear stress at the wall. Assuming
calculated by modifying Ergun's equation
that AP is small relative to the pressure, that the
(Equation 22.86) by the use of an appropriate slip
acceleration is short, and that the gas
section
velocity (63)
density is very small, Equation 22.98 reduces to
APD p e 3
=150//V Rej/ +1.75
AP = ^-+n-e)p
-e
gL+4rL/D s (22.99)
Lph 1-e) (22.104)
1
1 -e
The wall friction term is typically broken up into a
term due to fluid friction, and one due to solids where /V Re is the Reynolds number of gas relative
.

flow (53). to the solid particles, and is defined as

2fs Lp s vs 2 D pe
ArL/D = APfg + AP fs = fL v2 /2g c D + -e
1 -e 1

(22.100) A*d —el

In this equation f is the friction factor as defined in The equations and the approach presented
Chapter 20, and fs is a particle-wall friction factor here offer a logical attack on the design of vertical
that must be determined empirically. Since the solid-gas conveying. However, the wide variation in
particle-wall friction loss is usually a small factor, solid and the tendencies of solids to
properties
Leung and Wiles (39) recommend a simple expres- agglomerate, and to adhere to wall surfaces, makes
sion that roughly correlates most of the low- the use of these equations hazardous except when
pressure conveying data dealing with free-flowing solids of nearly uniform
size and high sphericity. Conveying fibers and
fs = 0.005(1 -e)lvs (22.101) solvent-refined coal are but two examples of appli-
Since the constant in Equation 22.101 is dimen- cations for which these equations would be inade-
sional, vs must be expressed in m/s. Data on quate. Thus, a test of the actual conveying situa-
particle-wall friction loss in high pressure con- tion should be carried out if at all possible.
veying have been correlated by the equation (27) Moreover, horizontal conveying has not been
considered. This might be attacked by writing
Equation 22.99 without the term for solid particle
f< = 0.0251 5(w,/pi/) 00415
1(1 -e)v' J weight. In fact, for short horizontal runs such a
procedure is feasible and results in pressure drops
0.03 (22.102)
that are small compared to those in vertical
where w s is the mass flow rate of solids, kg/s m2 conveying. Problems occur where low gas velocities
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 649

are used, for here the tendency of solids to settle in slightly pressurized system, such as would be the
horizontal sections limits the velocity that can be case with an upstream blower, can probably be
used. done with a star feeder. Problems can occur as the
Bend losses must also be recognized. Bend pressure drop caused by solids flow builds up,
loss estimates ranging from 0.5 to ^.5(v 2 p m /2g) however. Another possibility is the use of a flow-
have been recommended with the constant de- through feeder. This operates like the star feeder
creasing as the bend radius increases (38). Here p m except that the air flows through the feeder
is the density of the flowing mixture, cylinder (see Figure 22.37). Another possibility is
Pm = Ps d -e) +Pe. the mechanical screw feeder shown in Figure
The design of a pneumatic conveying system 22.38. Here the material enters the screw from a
requires careful consideration of the properties of small hopper and is carried through the barrel by

the material to be conveyed

such as caking — the screw action. The screw is of decreasing pitch,
tendencies, ease of particle breakage, and any
tendencies of the powder to explode — and of the
operating conditions such as pressures, tempera-
ture, solids/gas ratios, etc.
The physical arrangement requires, at a mini-
mum, an air blower, a method of feeding the solids
into the airstream, and a method of later removing
the solids from the air. The household vacuum
cleaner is a simple example of such a system. Use
of such a vacuum system is often convenient on a
large scale, too, especially where the solids are
friable and where both feed solids and ultimate
solids location are atatmospheric pressure. Unload-
ing barges or trucks an example. Often the solids
is
are fed into a moving gas stream by a mechanical
device such as a star feeder. This is a set of paddle-
which compacts the materials and forms a
pressure-sealing zone. The material forces the
check valve open as it leaves the barrel, and is
broken up and picked up by the air nozzle that
forces it into the conveying line.
Feeding solids into a high-pressure chamber
on a continuous basis is one of the more difficult
requirements of chemical processing. In fact,
feeding coal into high-pressure reactors has been,
and continues to be, a major barrier to the develop-
ment of coal conversion processes. Most successful
systems have been semicontinuous. For instance,
the solid is fed into a vessel, sealed in and pres-
surized, and then a valve is opened that permits the
solids to flow into a high-pressure conveying line.
If continuous feed is desired, two such tanks are

vanes that rotate in a horizontal cylinder allowing needed, both continuously weighed, with one
solids to flow into the cylinder at the top, to be being filled and pressurized while the other is
carried between the vanes to the bottom of the feeding the conveyor. Other feeding mechanisms
cylinder, and to fall out at that point into the fluidize the solid in the pressure chamber and
conveying air. Separation of the solids from the remove the solids from the bottom and/or top of
airstream is usually done in cyclone separators. In the fluidized bed.
vacuum systems, it is usually necessary to protect For such flow systems, there is always the
the blower with an air filter placed between the danger of static buildup and spark ignition of a
cyclone and the blower. combustible dust— air mixture. Safety precautions
If system is used, either because the
a pressure appropriate to the system are mandatory.
blower located upstream from the solids addi-
is

tion point (thus ensuring that solids do not enter Example 22.10. In manufacturing synthetic
the pump), or oecause the solids are taken from or detergents, the spray-dried bead is pneumatically

delivered to a pressurized chamber, solids addition conveyed from the bottom of the spray drier to
becomes much more difficult. Feeding into a cyclone separators at the top of the manufacturing
building. In a plant producing 3000 Ib/hr of

Material
inlet
let /
Lined
Drive
motor
J
/ Screw/ barrel
Check valve
Mixing chamber
. ^-Conveying line

Figure 22.37. Schematic diagram of a blow-through solids

Pe nozzle

Figure 22.38. Schematic diagram of mechanical screw feed-

feeder. feeder.
650 APPLICATIONS TO EQUIPMENT DESIGN

product, the conveyor is a 10-in. I.D. tube running From a consideration of the form of Equations
vertically for and horizontally for 30 ft. The
60 ft 22.95a and 6, both as shown in Figure 22.36 and
detergent has a mean particle size of 600 microns as seen by noting the equations themselves, it
3
and a packaged bulk density of 20 lb/ft The . seems clear that choking does not occur, that is,
blower pulling air through this conveyor has a flow is in the dilute phase region. Thus Equations
rating of 500 standard ft 3 /min. What is the 22.99, 22.100, and 22.101 will be used to obtain
pressure at the blower inlet if atmospheric pressure the pressure drop. To use Equation 22.101, the
exists at the bottom of the spray drier? Ignore void fraction e is needed. This can be obtained
pressure drop through the cyclone and assume the from the superficial gas and solids velocities, and
conveyor operates at a constant 80° F. the terminal velocity

Solution. First the flow regime will be deter-

mined by use of Equations 22.95 and 22.97. Here,
v — vr * = v t
as is typically the case, the superficial solids
velocity is obtainable. The solid density is obtained where vs * = actual solids velocity
from the bulk density in the package. vs * = -5.05 + 1 6.77 = 1 .72 ft/s
2 1 - e = 0.0473/1 1 .72 = 0.00404

Ps =
1 - ^
0.38
= 32.3 lb/ft
3
which confirms the presence of dilute flow. Then,
from Equation 22.101,
This gives the superficial solids velocity
3000 f, = 0.05 x 0.00404/0.0473 = 0.00427
v<
=——— x 1 1
= 0.0473 ft/s
32.3 "tt/4x(10/12) 2 "3600
The friction factor at the wall can be obtained
The gas velocity is given by the blower capacity from Appendix C-3.
540
v =
500
——x
60 "tt/4x (10/12) 2 "492
1

x— = 16.77 ft/s
'a/c
VRe
= Ovpe =
(10/12)x16.77x0.074
=77,950
n 0.02 x 0.000672
and the terminal velocity is obtainable from Figure
f= 0.019 if e = 0.0002 as for commercial steel pipe
22.1 via Equation 22.19

4gDp 3 p(ps -p) _ Equation 22.100 can now be used to get the
3ju
2 wall friction term of Equation 22.99.
492
4 x 32.2(600 x 10- 6/0.305) 3 x— x—29
-(32.3-p)

2g c D 1 - e
3 x (0.000672 x 0.02) 2
2
- 0.019 x T677"
128.8(0.00197) 3 x 0.074(32.3 0.074)
= 4334 64.4 x 10/12
3x (0.00001 344) 2
Thus,

and
^ = 55
2 x 0.00427 x 32.3 x 0.0473' x 10/12

= 0.0996 + 0.1273 = 0.2269

0.00404

— T^zz— —=
55 0.000672
x x 0.02
Vt = 5.05 ft/s Now, Equation 22.99 must be broken into
0.074x0.00197 two one for the vertical section of 60 ft and
parts,

Now, using Equation 22.95a, one for the 30 ft horizontal section. For the verti-
cal section
0.0473 = (16.77e c - 5.05) (1 - ec )

2
e c = 0.306
A/V = ^-+(1 -e)ps gL + 4T/DL
Checking with Equation 22.95o, 1 — e

(0.306 x 16.77 - 2
5.05) /64.4 x (10/12) =
32.3 x O0473 2
+ 0.00404x32.3x32.2x60
100(0306-° 47 -1) 0.00404
(5.13 - 2
5.05) /53.67 = 100 x 0.745 = 74.5 + 0.2269 x 60
2
0.0001 19 ¥=74.5 = 1 7.83 + 252. 1 1 + 1 3.61 = 282.55 lb/ft
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 651

For the horizontal section, the second term on the were flowing alone. This add-on is based upon the

right-hand side of Equation 22.99 should be conventional drag coefficient for particles of the
omitted, since that represents the energy needed to solid, which avoids the necessity of using a vis-
lift the solids. Thus, cosity or pseudoviscosity for a slurry. One form of
the equation is
APH" = 17.83x— +
30
60
13.61 x —
30
60
= 15.72 lb/ft
2
1.5
= K VCc
Finally, at least one bend loss should be included. VgD

A/>fl= [p,d - e)
2
+ pe] v l2g where f and fw are friction factors for slurry and
= (32.3 x 0.00404 + 0.074 x 0.996)16.77764.4 for liquid alone, and c is the weight fraction of
= 0.89 2 solids.
lb/ft

Then the total pressure drop is Co = drag coefficient for a free-falling sphere

AP = (282.55 + 1 5.72 + 0.89)/1 44 = 2.08 psi

CD = 4/3 gd
-^r
Vm
Thus, the pressure at the blower inlet is

P= 14.70-2.08= 12.62 psia s = density ratio —

p s

P
Fluid-Solid Conveying, Slurry Transport. In

addition to the pneumatic transport described It has been well known that the constants in

above, which widely used for short-distance

is the equation must be appropriate to the flow
movement of solids, the economy of long-distance regime. As is typical in two-phase flow situations,
moving of pulverized solids in slurry form indicates the mode of flow depends upon the flow rates of
that such moving will be used with increasing fre- each phase and the total of both. At very low
quency. In view of the expected increase in use of velocities of the fluid, the solids comprise an essen-
coal, extensive studies have been made in the trans- tially stationary bed with very little movement
portation characteristics of a slurry of coal in water (regime 0). As the rate of flow of the fluid is

and some slurry pipe lines are in operation. For increased, increasing amounts of the solid are
example, the Mohave Power Station receives thrown up into the liquid phase in bursts, or slugs.
slurried coal through the Black Mesa Pipeline. This This has been characterized as "saltation" flow
line runs for 273 miles between Arizona and (regime 1). At progressively higher fluid velocities,
Nevada. It is 18 in. in diameter and is designed to the entire portion of the solid is lifted into the
move 5.5 million tons/year of coal in a 50 percent stream and is moving, but the larger particles
by weight slurry (57). Other solids are also moved congregate toward the bottom of the stream;
in other places. For instance, in Tasmania a hence, this is called "heterogeneous" flow (regime
valuable deposit of iron ore that had been con- 2). At a sufficiently high velocity, the solids can
sidered inaccessible for conventional transport is become essentially uniformly distributed —
now being moved to dockside by hydraulic trans- homogeneous flow (regime 3). Turian (57, 58) and
port through a pipeline 6 in. in diameter and 53 hiscoworkers examined and classified thousands of
miles long. published experimental measurements into these
Generalized correlations upon which designs four regimes of flow and fitted for each a set of
can be based are virtually nonexistent, because of correlation constants to generate empirical
the difficulty in fitting most slurries into conven- equations for pressure drop in the general formula-
tional fluid mechanics correlations. Many data have tion and variables of Durand's equation (Table
been collected, but unifying principles have been 22.5 and Figure 22.39).
hard to develop. The heterogeneity of the slurry
seems to preclude the application of continuum
f-fu
expressions for viscosity and density. *9c Dgis DJ
The classic generalization is that of Durand (22.106)
(16) and his coworkers who correlated slurry flow
behavior in terms of an increment on the friction A calculation of the pressure drop for a
factor expressing the contribution of the solid flowing slurry must recognize the different flow
phase to the friction factor for the liquid phase if it mechanisms discussed above; it must identify the
652 APPLICATIONS TO EQUIPMENT DESIGN

Table 22.5. CONSTANTS FOR THE SLURRY FLOW FRICTION FACTOR

EQUATION (EQUATION 22.106)

Equation Regime
Designation (m) a b c d e

a 0.4036 0.7389 0.7717 -0.4054 -1.096

b 1 0.9857 1.018 1.046 -0.4213 -1.354
c 2 0.5513 0.8687 1.200 -0.1677 -0.6938
d 3 0.8444 0.5024 1.428 0.1516 -0.3531

uuu
- guide, and the ratios have been used as the identifi-
/
I
I
I

cation number.
—
800 / The boundaries between adjacent pairs of
Flow with a regimes are generated by an empirical combination
stationary bed Saltation flow /
700 (regime 0) (regime 1) <© / of the pressure-drop functions for each adjacent

fy
£/
/ Heterogeneous _
regime (actually, a "cut-and-fit" process). A plot of
600
c- / flow principal boundaries is shown in Figure 22.39 for
Of £/ (regime 2) 3 the locus of values of regimes number equaling
500 unity for the contiguous regimes, applicable to one
particular system. The constants tabulated in Table
1/ 22.6 represent an interpretation of many measure-
400 - 3
iS7
/
/ /& ments. These are fitted into Equation 22.107 for a
given flow system; several sets must be evaluated
and studied to find the pair of regimes separated
300 Homogeneous flow — by a ratio of unity for the pressure drops expected
(regime 3)
in these two regimes.
t

V T =
22.5 °C
3
v
2
= 1

Ps = 2.977 g/cm d'

200
c = 5% by volume •c*' CDt Dg(s -1)
/fa* Pipe ID. = 0.957 in.
<V (22.107)
1 I 1 i I i

Inspection of Equation 22.106 in relation to

7 8 9 10
Mean slurry velocity, v (ft/sec) Figure 22.39 indicates that the ratio computed in

Figure 22.39. Flow regime diagram for solids-water flow in

all cases increases with velocity, becoming unity at
1-in. PVC pipe (57). (By permission of A.I.Ch.E., copyright a velocity corresponding to the maximum velocity
©1977.) for that regime and for the system being studied.
Identification of the existing regime is based upon
existing regime and be applicable to it. This is true applying to the existing operating conditions sets
for both design of a new installation and analysis of constants appropriate to regimes in increasing
of performance data from an existing system. In order of velocity. The existing regime is the first
general, determination of the existing regime one in which R< 1.

cannot be made by observation. The transitions occur in zones, rather than the
Identification of the existing regime is
sharp increments indicated by Figure 22.39. Par-
effectedby comparing the pressure drops predicted ticularly for fine particles, the regime boundaries

by the appropriate functions for each and analyz- cross so that transition is expected from stationary

ing the order of changes indicated by regimes in bed flow directly to heterogeneous, or homo-
order of increasing flow velocity. At a boundary geneous flow. The reader is referred to the original
between regimes, the pressure drops should be publication if these transitions are encountered.
equal, or the ratio ofthem should be unity. Since
the differences of friction factors are small num- Example 22.11. Determine the pressure drop
bers, they are not satisfactory as an identification per mile that would be expected in the Black Mesa
 PARTICULATE SOLIDS FLOW AND SEPARATION THROUGH FLUID MECHANICS 653

Table 22.6. CONSTANTS FOR THE SLURRY FLOW REGIME

EQUATION f EQUATION 22.107)

Boundary
Equation Between
Designation Regions a b' c d'

a 0-1 31.93 1.083 1.064 -0.06160

b 1-2 2.411 0.2263 -0.2334 -0.3840
c 1-3 1.167 0.5153 -0.3820 -0.5724
d 0-2 0.4608 -0.3225 -1.065 -0.05906
e 0-3 0.3703 0.3183 -0.8837 -0.7496
f 2-3 0.2859 1.075 -0.6700 -0.9375

pipeline described above. As stated, this pipeline is and flow is in the laminar region. For the flow of
18 in. I.D. 273 miles long. It supplies 5.5 million water alone.
tons/year of coal to the Mohave Power Station as a 1.5x5.36x62.4
50 weight percent slurry. Consider the median /V Re = = 747,000
mm
in diameter with a spher-
1.0x0.000672
particle to be 0.07
icity of 0.8. fw =0.0125, using Appendix C-3

Solution. The first step

type of flow regime using Equation 22.107 with
is to determine the —
fw
=
0.125
= 0.003125

the constants given in Table 22.6. Each of the

Now the regime boundaries can be determined
groups in Equation 22.107 must first be obtained,
starting with the bulk mean velocity. 2
v /Dg(s - 1
= 77.4
083 (0.0031 1 064 (600)-° 0616
5.5x10 6
1
x 2000x1/0.5 31 .93(0.5) 25)
= 2.51 x10 6 lb/hr of slurry
365 x 24 (22.107a)
1.70
p= (0.5 x 1.0 + 0.5 x 1.35) = 1.18 g/cm 3 for the slurry = 2.50
2.411 (0.5)° 2263 (0.003125)-°- 2334 (600)-° 3840
1
(22.1076
2.51 x 10 6 x
1.18x62.4
v= = 5.36 ft/s 1.70
= 8.929
5724
3600 x-x 1.5" 1 .1 67(0.5)° 51 53
(0.0031 25)-°- 382 (600)-°
4
(22.107c
:
[5.36)
1.70 1.70
Dg(s--\) 1.5 x 32.2 x (1.35- 1) = 0.0091
0.4608(0.5)-°- 3275 (0.003125)- 1065 (600)-° 05906
To calculate the free-fall drag coefficient CDoo ,
(22.107cO
Figure 22.1 can be used along with Equation 22.19. 1.70
= 4.25
At/V Re =1, 0.3703(0.5)
- 3183 (0.003125)-° 8830 (600)-° 7496

49D P 3 p(ps-p) (22.107e)

Cn =
: 1.70
3m = 105.5
3 1075 (0.003125)- 067 (600)-° 9375
0.07 x 0.1 \
0.2859(0.5)
2
4x32.2x X6Z4 (1.35-1) (22.107A
v2.54x 12
3x (1.0x0.000672) 2 Comparing these results with the configuration
9 shown in Figure 22.39, it is apparent that the flow
128.8 x 0.012 x 10- x 1362
= —z
10- 6
= 1 .554 is in the homogeneous region. In the various forms

1.355 x of Equation 22.107, a large calculated number

Then, using Figure 22.1 for = 0.8 C Doo = 600, means a greater velocity than the transition velocity,
654 APPLICATIONS TO EQUIPMENT DESIGN

or perhaps a smaller particle diameter than that reason orifices, Venturis, and area meters are hardly
equivalent to the transition, all other things being suitable, and obstructionless meters such as those

equal. that compare the time required for a small pressure

The next step is to calculate the frictiona! pulse to travel a given distance diagonally across a
loss, this time using Equation 22.106c'. tube with the slurry flow and against it are often
used.
428
f-f
4g c
0.8444C 5024 -
if
!
— \
)
1 -

(CD )
01516
Pressure-sensing tubes and sample tubes need
to be continuously flushed to prevent sediment
buildup, and measurements of density or of flow
0.3531
must account for the fact that solids can and do
onto the bottom of the pipe or other con-
settle
\Dg(s- 1)/
tainer.
.428
n ___. 0.0125 1
/ \
0.8444(0.5 0.5024 . Relation Between Regions of Fluid-Solid Flow
Ux32.2/
4x32.2/ As evident from the treatments above, flow
(6 00)0.
1516 (17 Q)-0. 3531 through packed beds, sedimentation of solids,
fluidization, and fluid-solid conveying are all opera-
f-fu