Professional Documents
Culture Documents
Principles of Unit Operations (1980)
Principles of Unit Operations (1980)
FOUST
LEONARD A. WENZEL
CURTIS W CLUMP
LOUIS MAUS
L. BRYCE ANDERSEN
Principles of
Unit Operations
Board of Advisors,
Engineering
R. V. Whitman Mechanics
Massachusetts Institute of
Technology
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.
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
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.
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
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.
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,
Alan S. Foust
Leonard A. Wenzel
Curtis W. Clump
Louis Maus
L. Bryce Andersen
Bethlehem, Pennsylvania
December 1959
Contents
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
One
•References are collected alphabetically at the end of each tool the engineer. Unfortunately, our under
of
chapter; parenthetical numbers refer to references. standing of mathematics is largely restricted to the
UNIT OPERATIONS IN CHEMICAL ENGINEERING
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
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
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- tions where extremely corrosive conditions force
plished with both streams flowing in the same frequent repairs, and in others for various specific
direction (i.e., cocurrent flow). If cocurrent flow is reasons. Because of the greater productivity of
used, the limit in amount of transfer that can occur continuously operating equipment and the resul-
is firmly set by the equilibrium conditions that will tant lower unit cost, it is usually advantageous to
be reached between the two streams being contacted. operate equipment continuously. This means that
If, however, the two streams being contacted are time is not a variable in the analysis of such a
made to flow in opposite directions, transfer of process, except during the rather brief start-up and
material or energy in considerably greater amounts shutdown periods. The time rate of transfer or of
is possible. Such a flow pattern is known ascounter- reaction is important in fixing the necessary size
current flow. and capacity of equipment, but the performance is
As an illustration, if a stream of hot mercury expected to be the same today, tomorrow, or next
and a stream of cold water are allowed to reach year if the operating conditions remain the same.
thermal equilibrium, the temperature attained can Conditions are not constant throughout a system
be predicted by a heat balance that recognizes the at any time, but those at a particular point are
relative quantities of the streams, their initial constant with time.
temperatures, and their heat capacities. If the When small quantities of material are to be
streams flow simultaneously from the same inlet processed, it is often more convenient to charge
point to the same outlet point, the equilibrium the entire quantity of material to the equipment,
temperature is definite, and the path is as indicated process it in place, and remove the products. This
in Figure 1.1a. If the streams are made to flow in is called a batch operation.
opposite directions, as by letting the mercury flow An operation that is variant with time is
downward through an upflowing stream of water, spoken of as a transient or unsteady state, in con-
it is possible for the entering hot-mercury stream trast with that spoken of as steady state, in which
to raise the temperature of the leaving cool-water conditions are invariant with time. Quenching a
stream to a temperature above that to which the steel part for heat treating and freezing ice cubes
mercury stream is lowered as it leaves the contact- in a domestic refrigerator are illustrations of
ing equipment, as indicated in Figure 1.1/?. The unsteady-state operations. In batch operations,
counterflow principle is used in many chemical almost the entire cycle is a start-up transient and a
engineering operations in order to permit greater shutdown transient. In a continuous operation, the
transferof a property than would be indicated time during which the start-up transient exists may
merely by the attainment of a single equilibrium be extremely small in comparison with the steady
between the leaving streams. state operation. Analysis of transient or batch
operations is usually more complex than of steady-
Continuous and Batch Operation
state operation. Because of the greater simplicity
In the majority of chemical processing operations, and the wide occurrence throughout chemical
it more economical to maintain continuous and
is processing of steady-state operations, the intro-
steady operation of equipment, with a minimum of ductory treatment is in terms of conditions that do
disturbances and shutdowns. This is not always not vary with time. Analysis of a transient opera-
practical in some small-scale operations, in opera- tion is different from the steady state only in the
i i
H2 out
-t -
.
1 L
.
(a) (b>
Figure 1.1. Flow and temperature in a contactor, (a) Concurrent, (b) Counter-
current
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
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 omy in learning and is believed to contribute a
chemical properties under extremes of conditions greater breadth of understanding of all the oper-
such as temperature and pressure. The unit oper- ations when the interrelations are understood.
ations used to separate mixtures into more or less
pure substances are unique to chemical engineer- Analysis of the Unit Operations
ing. The materials being processed may be natu- The unit operations may be analyzed and grouped
rally occurring mixtures or they may be the prod- using any one of three possible methods. A unit
ucts of chemical reactions, which virtually never operation may be analyzed using a simple physical
yield a pure substance. model that reproduces the action of the operation;
it may be analyzed by considering the equipment
INTEGRATION OF THE UNIT OPERATIONS used for the operation; or it may be analyzed
starting with a mathematical expression that de-
Early studies of the unit operations as independent scribes the action and is tested using experimental
steps laid an important part of the foundation for process data.
the phenomenal growth of the chemical industry. The physical model is established by careful
In a complex manufacture, the interactions of the study of the basic physical mechanism. The model
steps force engineers to consider the total process, is then applied to a real situation either through
With increasing information has come broader cules in their disregard of mathematics. Because of
recognition of the basic similarities. Conversely, the nonlinearities involved and because boundary
recognition and exploitation of the similarities conditions of one phase usually respond to changes
have contributed to a broader understanding of occurring in an adjacent phase, it is frequently
each operation. It now appears that the compart- impossible to formulate the boundary conditions
mentalization of information by unit operation for solution of a mathematical expression in man-
leads to unnecessary repetition and waste of time ageable terms.
and that study of basic principles common to a Each of the three modes of grouping could be
group of the operations will lead to a better under- used as a basis. The physical model of the funda-
standing of all of them. mental operation is the most satisfactory approach
This book presents under single headings those and is used in this presentation. Wherever possible,
operations having similar fundamentals, using the physical model is described mathematically,
generalized nomenclature and concepts. This pre- and the performance is expressed in mathematical
sentation has been found to result in time econ- relations derived from the fundamental principles.
UNIT OPERATIONS IN CHEMICAL ENGINEERING
This formulation gives the best basis for under- out sometimes in continuous-contact equipment
standing and refining those operations in which the and sometimes in staged equipment. In some oper-
art is ahead of theory. This is true in spite of the ations, the advantage of one or the other mode of
fact that the models are oversimplified and that the analysis may be obvious. In many others, the
mathematical formulation of the behavior of the choice is dictated by availability of the necessary
model cannot be transposed perfectly into an ex- data and constants. Equilibrium data are a part of
pression of the behavior of the prototype. the stock in trade of the physical chemist and are
It should be obvious that there is no universal available for a large number of substances under
criterion dictating a particular choice of method of various conditions. To some extent, the con-
analysis and that allcontributing factors should be venience in analysis is related to the work of earlier
recognized in deciding upon a particular mode. investigators, in that their results may have been
Any grouping requires some arbitary choice and interpreted in a fashion thatmakes one or the
always leaves one with some of the operations that other analysis more convenient. Choice of one
fit poorly into the general scheme. Such operations method of analysis does not necessarily restrict the
must be studied individually. actual operation to the same model.
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^ (1.1
cations usually approach Equation 1.2 much more
30 3x 2 closely than the rigorous diffusion equation.
For the rate operations, analysis must be based
where Y = concentration of the property to be
upon the driving force causing a change, the time
transferred
during which a driving force is allowed to act, and
8 = time
the quantity of material upon which it acts. The
x = distance measured in direction of
diffusion equation above expresses the transient
transport
behavior of a large number of properties under the
6 = proportionality constant for a system
influence of a driving force for transport of the
This equation is frequently called the diffusion property. In chemical engineering, mass, momen-
equation. It is a general expression that reduces to tum, and thermal energy are the three properties
Ohm's law for electrical flow for specified con- whose transport is the most frequently involved.
ditions. The broad study of electric and magnetic As mentioned above, it is universal that these three
transport is the important "field theory" of the properties, along with a number of others with
electrical engineer. These two phenomena follow which chemical engineers are less frequently con-
well-established laws and involve relatively con- cerned, tend to flow from regions of high concen-
stant proportionality factors (such as 5 in the tration to regions of low concentration. Accurate
above equation). Since boundary conditions can prediction of the amount of the property that
usually be evaluated, analytical solutions are fre- flows from a donor region (source) to a receiver
quently possible for engineering calculations. region (sink) can be made if the driving force, the
Chemical substances are less well behaved mathe- area of the path, and the unit resistivity of the path
matically, and the proportionality "constants" are (the proportionality constant used in Equation
seldom really constant. The boundary conditions 1.1) are accurately known. Throughout the study
are more elusive; hence, chemical engineers are of the rate operations, the importance of a clear
seldom in position to apply mathematically elegant understanding of the meaning of concentration
and rigorous solutions of the diffusion equation. In cannot be overstressed. In every case, the con-
order to arrive at a solution of the diffusion equa- centration expresses the amount of property per
tion, it may be simplified into a finite-increment unit volume of the phase being processed. The
UNIT OPERATIONS IN CHEMICAL ENGINEERING
amount being transferred can usually be expressed ment in quality or quantity. It is therefore neces-
in some absolute unit measuring that quantity, sary that mathematical and/or graphical techniques
such as British thermal units or pound moles. It be available to permit the prediction of any un-
can also be expressed in terms of the decrease in known answer for a particular system regardless of
concentration of the property in a known amount whether the unknown is a composition, quantity,
of phase having a known capacity for this prop- temperature, or number of stages required to ac-
erty. For example, a quantity of energy leaving a complish a specified amount of enrichment of any
system as heat can be expressed in terms of the chosen property.
number of Btu's or calories of energy. It can also Although this book is devoted exclusively to
be expressed in terms of the decrease of temper- the principles of the unit operations of chemical
ature of a known amount of the phase. These engineering, it should be emphasized to the prospec-
generalizations will become more meaningful as tive chemical engineer that unit operations are only
different operations are analyzed and the trans- one sector of chemical engineering. The real objec-
ported quantities are expressed in terms of the tive is the engineering of the most economical pro-
various possible units. Because the basic principles cess. The unit operations are techniques in arriving
of transport are identical for the three properties, at this process, but they must not be allowed to
an analysis will be offered in completely general crowd out of consideration the other important
terms before specification of the particular prop- scientific principles that must be recognized.
erty in specific operations. The best process can be designed only with
proper regard for the basic chemistry, kinetics, and
Unsteady-State Operation
thermodynamics, with adequate recognition of the
The diffusion equation, Equation 1.1, is applicable limitations imposed by materials of construction
for transfer one direction only and that is a
in and auxiliaries to the plant. The equipment design
function of time. However, transfer may occur in will involve work by engineers trained in disciplines
more than one direction. Rigorous solution requires normally not covered by the chemical engineer.
knowledge of boundary conditions and interaction The ultimate object of the engineering is the ac-
of variables. Computer techniques make solutions cumulation of a profit from the operation. The
quite accessible, and they usually employ finite- largest return of profit, after all costs are account-
difference techniques. ed for, demands full exploitation of all the tech-
nical factors involved, favorable human relations
within the producing team, and accurate know-
GENERAL CONSIDERATIONS
ledge of the amount of product that can be sold
Understanding the basic physical principles of an for maximum eventual profit.
operation and formulating these principles into a
mathematical expression are the first requirements
for applying the principles of the unit operations.
REFERENCES
In engineering practice, however, numerical values
must always be incorporated and a practical answer 1. Constitution of the American Institute of Chemical
obtained. Engineers.
The same problem may be met by the design 2. Little, A. D., Report to the Corporation of M.I.T., as
engineer in specifying equipment, by the operating quoted in Silver Anniversary Volume, A. I.Ch.E. (1933),
engineer checking the performance of installed
in p. 7.
Stage Operations
11
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.
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
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 Ion-exchange resins, molecular sieves, and
solutions. Because one of the phases is a solid that clathrate compounds are relatively expensive com-
does not flow like a fluid, special types of equip- pared to conventional adsorbents such as charcoal
ment are required for solid-liquid extraction. and silica gel. Hence, these special adsorbents are
used in separations that cannot be accomplished by
Adsorption more conventional means. These special separation
Adsorption involves the transfer of a constituent of processes are carried out in batch processes in
a fluid to the surface of a solid phase. To complete fixed-bed adsorbers, rather than in multistage and
the separation, the adsorbed constituent must then continuous-flow equipment.
be removed from the solid. The fluid phase may be The mass-transfer operations involving trans-
either a gas or a liquid. If several constituents are fer to a solid phase are sometimes referred to as
adsorbed to varying extents, it is often possible to sorption processes, thereby including all types of
separate them into relatively pure components. solid attractive forces.
Many solid adsorbents are used. Strictly A mixture of several adsorbable solutes can be
speaking, the term adsorbent is usually applied to a separated using a fixed bed of adsorbent. First, a
solid that holds the solute to its surface by physical small volume of the mixture is fed to the bed
forces. An example would be the adsorption of where completely adsorbed near the entrance.
it is
organic vapors by charcoal. The lighter fractions of Then continuous flow of solvent is passed
a
natural gas are separated industrially using a through the bed. Part of the solute mixture is
moving bed of adsorbent. Most other industrial desorbed and passes out with the solvent. The rate
processes use fixed beds and batch or cyclical pro- at which each solute passes out of the bed depends
cesses rather than multistage equipment because of upon its equilibrium adsorption on the solid. If a
the difficulty of moving the solid. component is strongly held by the adsorbent, it
Molecular sieves (4) are special synthetic passes slowly through the bed, whereas a weakly
adsorbents that can be used to separate mixtures adsorbed component moves quickly through and
by differences in molecular size, polarity, or car- out of the bed. As a result, the components flow
bon-bond saturation. For example, water mole- out of the bed at different times and can thereby
cules are relatively small and highly polar, so they be separated. This separation process is called
are preferentially adsorbed by molecular-sieve chromatography (8). Chromatographic separation
adsorbents. Thus, these adsorbents are useful in is a widely used method of chemical analysis of
drying gases and liquids. Molecular sieves are being small samples. Large-scale industrial applications
used to separate normal paraffins from hydro- are limited. The ion-exchange separation of rare
carbon streams in an oil refinery for use in the earths mentioned earlier is an example of chroma-
manufacture of biodegradable detergents. tographic separation.
Closely related to adsorption are separation
processes that retain solutes in solids by various Membrane Separations
means. One such process is ion exchange, where A number of separation processes involve the trans-
the solute is retained by a chemical reaction with fer of mass through a thin plastic membrane (1).
the solid ion-exchange resin (1,6). As the name Although these operations have relatively limited
implies, ions in solution may be removed by this application, they show promise for special sepa-
process. It is widely used to produce highly pure ration problems. The theory of membrane separa-
water. Other applications include recovery of anti- tions is not fully developed, but as a first approxi-
biotics from fermentation broths and separation of mation, the separation can be considered a matter
rare earths. of molecular size. Smaller molecules more easily
Clathration is a physical phenomenon that has pass through the small pores of the membrane. As
been applied to separate mixtures on the basis of a result, if some driving force is used to "push" the
molecular shape. The solid clathrate compound at- molecules through the membrane, the smaller
tracts and retains molecules of a particular shape. molecules are selectively separated. Some larger
For example, an industrial process separates para- molecules also pass through, so that the separation
xylene from meta-xylene by trapping the former in is not perfect.
a clathrate compound. These two compounds have Dialysis is a separation process in which mass
nearly identical physical and chemical properties so is transferred across a membrane by a con-
that conventional separation processes cannot be centration gradient driving force (13). In the larg-
used. The clathrate compound might be considered est industrial application, sodium hydroxide is
as a special type of adsorbent. separated from an aqueous solution with hemi-
MASS-TRANSFER OPERATIONS 17
cellulose in the manufacture of viscose rayon. The porous metal barrier (2). The gas component with
solution flows past one side of the membrane and a lower molecular weight flows more rapidly
pure water flows on the other side. The sodium through the barrier, so that the gas beyond the
hydroxide is forced through the membrane because barrier is richer in that component. Only a fraction
of the concentration difference between the solu- of the gas flowing to the barrier passes through.
tion and the pure water, but the hemicellulose is The remainder, which is depleted in the lighter
in colloidal suspension in the solution and is too component, is removed for further processing.
large to pass through the pores in the membrane. Power costs for gas compression are very high, so
Dialysis is also used to recover sulfuric acid from gaseous diffusion has only one major industrial
copper refinery waste liquors and in the purifi- application: the separation of uranium-235 from
cation of blood in the artificial kidney. uranium-238. Over a billion dollars worth of gas-
Electrodialysis utilizes an electrical potential eous diffusion plants make this separation to
difference as a driving force for ions in solu- produce enriched U 235 for nuclear reactor fuel.
tions (5). Positive pass through an appro-
ions
priate membrane they are attracted to the neg-
as Other Separation Processes
ative terminal, and negative ions travel in the oppo- A number of important unit operations involve the
site direction through a membrane to the positive separation of components of a mixture, but the
terminal. In this way, both positive and negative operations are not customarily grouped with the
ions can be removed. usual separation processes.
Electrodialysis is being used to recover waste Evaporation, crystallization, and drying in-
acid and to remove salt from brackish well water. volve simultaneous heat and mass transfer. In evap-
It may also be useful in seawater purification, if oration a liquid solution is concentrated by vapor-
membrane and power costs can be reduced. izing part of the solvent. The heat requirements are
Dialysis and electrodialysis both involve mass large because the latent heat of vaporization of the
transfer across a membrane from one phase to a solvent must be supplied. Evaporation has many
second phase. The two phases are completely mis- industrial applications wherever solutions must be
cible, so that the membrane also serves to keep concentrated. What is commonly referred to as
them from mixing together. seawater distillation is in fact an evaporation pro-
A more recently developed membrane separ- cess to recover potable water.
ation process uses a pressure difference to force If a solution is evaporated until it becomes
molecules of solvent through a membrane, while saturated with solute, further evaporation or cool-
molecules of solute remain behind. This process ing results in precipitation of solid crystals. This is
does not involve mass transfer between phases. It is the physical basis for crystallization a unit oper-
,
more analogous to filtration, where suspended ation used to separate solutes from solution.
solid particles are separated from a liquid by for- Crystallization is used industrially in the manu-
cing the mixture through a porous filter medium so facture of many inorganic salts. It can also be used
that the liquid passes through and the solid remains to separate mixtures of salts by fractional crystal-
behind. The use of a membrane permits a similar lization. Evaporation and crystallization are dis-
the stage. At equilibrium no further net change of a liquid or a finely divided solid; the V-phase may
composition of the phases is possible for a given set be a gas, vapor, or liquid, depending on the sepa-
of operating conditions. In actual industrial equip- ration process. Gases and liquids flow easily from
ment, it is usually not practical to allow sufficient stage to stage, but solids are more difficult to
time with thorough mixing to attain equilibrium. transfer. For this reason, stage operations where
Therefore, an actual stage does not accomplish as the solid moves countercurrent to a fluid phase are
large a change in composition as an equilibrium less common. Very finely divided solids can be
stage. The stage efficiency is defined as the ratio of made to flow like fluids, so this technique is used
a composition change in an actual stage to that in to move solids. In other cases, the solid must be
an equilibrium stage. Stage efficiencies for indust- mechanically transferred from stage to stage.
rial equipment range between a few percent and For economic reasons, stages must accomplish
about 100 percent. Because an equilibrium stage the contact and separation as simply as possible.
gives the greatest composition change possible for a For example, the force of gravity is used to separ-
given set of operating conditions, it is also referred ate phases and promote flow whenever possible, to
to as an ideal or theoretical stage. Even though 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
s**+
Liquid flow
S^4
l¥ll
:£WW\
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. rates, so relatively constant efficiencies are poss-
Bubble-cap columns can be operated over a wide ible. Because of their relatively high cost, bubble-
range of phase flow rates with high stage effici- cap trays have given way to newer types of trays,
ency. Little liquid can leak through the gas pass- described below.
ages, and the height of liquid on each tray can be Perforated plates have become very popular in
maintained at the desired level. As a result, gas- recent years because of their simplicity and low
liquid contact is good over a wide range of flow 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
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
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
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
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-
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
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
quired size of the operating unit, and hence its cost Gibbs defined chemical potential in terms of
and operability, can be calculated from a know- fundamental thermodynamic functions (11). Quali-
ledge of the phase equilibrium limitations and the tatively, it expresses the tendency of the com-
rate of transfer obtainable. This chapter considers ponent to leave the phase or chemical state being
the basic theoretical framework used in predicting described. Thus, it is a measure of component
the composition of phases in equilibrium, presents instability.
some of the empirical relations used to correlate Workers later showed that the fugacity (de-
phase equilibrium data, and shows some typical 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
29
30 STAGE OPERATIONS
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)
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)
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
\ ^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
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
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
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.
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
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
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.
PHASE RELATIONS 35
Nonideal solutions may have temperature- The enthalpy of a binary mixture is relative to
composition curves substantially different from some arbitrary reference condition chosen for the
those of Figure 3.2. For example, Figure 3.3 pure components. For example, in Figure 3.4 the
presents two nonideal systems, isopropanol- enthalpies of pure liquid ethanol at 32° F and pure
propylene chloride and acetone-chloroform. Each liquid water at 32°F were chosen equal to zero.
system forms an azeotrope — a mixture with an Mixtures of ethanol and water may have enthalpies
equilibrium vapor of the same composition as the different from zero at 32° F because of enthalpy
liquid. At the azeotropic composition, the dew changes on solution and dilution. The changes are
point and bubble point are equal, and the mixture small for the ethanol— water system. Liquid mix-
vaporizes at a single temperature. This conclusion tures at temperatures other than 32° F will have
can be reached by applying the phase rule. In a enthalpies different from zero because of the enth-
binary liquid— vapor system there can be two alpy change on heating, as determined by the heat
independent variables. Usually the pressure is capacity and the temperature change. If a mixture
specified and so there remains one variable. The is vaporized, its enthalpy will increase by an
definition of the azeotrope sets the compositions amount equal to the latent heat of vaporization.
of the two phases as equal to each other, and this All enthalpy changes are considered in constructing
restriction eliminates the final independent vari- an enthalpy-composition diagram for a binary sys-
able. Therefore, the temperature is set at an unique tem using experimental data on heat capacities,
value. For example, in the isopropanol-propylene heats of solution, and heats of vaporization.
chloride system, the equilibrium liquid and vapor On Figure 3.4, the curve labeled H represents
compositions become equal at 0.68 mole fraction vapors at the dew point (saturated vapors), and the
isopropanol. Since the dew point and bubble point curve labeled h represents liquids at the bubble
are also equal at this composition, the mixture will point (saturated liquids). The vertical distance
vaporize at a single temperature, as does a pure between the curves is therefore the latent heat of
liquid. For this reason, azeotropes are often called vaporization of a mixture. Between the two curves
constant-boiling mixtures. This azeotrope boils at a is a liquid-vapor two-phase region. To determine
temperature lower than the boiling point of either points on the saturated vapor and liquid curves,
pure isopropanol or pure propylene chloride. which are in equilibrium with each other, the
Systems of this type are referred to as minimum- equilibrium curve on the lower part of the diagram
boiling mixtures. On the other hand, the system is used. A value of the vapor composition y a may
acetone-chloroform has a maximum-boiling azeo- be chosen on the vapor enthalpy curve (H). From
trope at 0.34 mole fraction acetone (Figure 3.3/?). the equilibrium curve, the value of x a in equili-
The presentation of x, y, T information at a brium with the chosen y a may be determined. The
constant pressure, as in Figure 3.2, is frequently a value of x a may then be plotted on the liquid
useful choice; however, in phase-relation material enthalpy curve (h). A tie line then can be drawn
balance computations done at a specific pressure, connecting the two points representing composi-
the temperature may not be specifiable. For these tions that are in equilibrium. The temperature-
calculations, a plot of x versus y as in Figure 3.1 is labeled lines in the two-phase region are tie lines.
much more useful. If necessary a T versus x or y If the equilibrium temperature is given, the
locus can be added. vapor and liquid compositions are obtained from a
In distillationand other mass-transfer process temperature-composition plot, such as Figure 3.3a
calculations, energy balances as well as mass for the C3H7OH— 3 H 6 CI 2 system. The region be-
balances are required. Thus, a measure of the low the saturated- liquid curve represents subcooled
system energy is needed. This is usually the liquids. Lines of constant temperature (isotherms)
enthalpy, defined as the internal energy plus the are given in this region. In the ethanol-water sys-
PV product (see Chapter 20). Enthalpy-concentra- tem, the heats of solution and dilution are small, so
tion data are given for ethanol— water in Figure 3.4 that the isotherms are nearly straight lines, indi-
and for ammonia— water and nitrogen— oxygen in cating constant enthalpy. Above the saturated-
Appendix D. The molar enthalpy is plotted on the vapor curve is the region of superheated vapors.
vertical axis and the mole fraction of the more Typically, increasing the pressure increases
volatile component on the horizontal axis. Any the temperature at which vapor-liquid equilibria
point on the diagram defines both the enthalpy and occur, and also decreases the sharpness of separa-
the composition of a binary mixture. Similar plots tion between vapor and liquid compositions. This
can be made using mass-fraction compositions. is illustrated for the ethanol-water system in Figure
36 STAGE OPERATIONS
c: 4 0.6 : B
-
PHASi: RLLATIONS 37
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
•
<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
,
,
y^ (e) Ammonia-water at 100 psia, H = 20,000,
-5,000 1
0.8 overall mole fraction ammonia = 0.4.
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.
-
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 / /
0.4 0.6 10
xa y
, , mass fraction acetic acid
tion of the solute may influence the mutual need not be plotted. The diagonal side of the
solubility of the solvents. In such cases, equilibrium triangle represents mixtures where x b = or
data must be plotted on a three-component yb = 0. The coordinate axis for component b
diagram. Application of the phase rule to two- actually is a perpendicular to the diagonal side of
phase, three-component systems shows that there the triangle from the right angle. Since the third
can be three independent variables. In this case, coordinate axis is not used, it is omitted.
there are four possible variables: temperature, Any point within the triangle represents a
pressure, and the concentrations of two compo- three-component mixture. Certain overall composi-
nents. The concentration of the third component is tions exist as one liquid phase, whereas others may
not independent because it is set by stipulating the split into two liquid phases. The curve nopqr
other two. But the concentrations of the two (Figure 3.8) separates the two-phase region (to the
components are not entirely independent because left) from the one-phase region (to the right) and is
it was stipulated that the two phases are in called the phase envelope. Compositions of phases
equilibrium. Therefore, if temperature, pressure, in equilibrium must lie on the phase envelope. A
and one concentration are specified, the other two mixture on the lower part of the envelope (such as
concentrations are set. q) contains a high concentration of the un-
Data for ternary systems are often reported extracted component and is called the raffinate
on equilateral-triangular diagrams; however, for phase. A mixture on the upper part of the envelope
engineering calculations a right triangle is more (such as o) is rich in the solvent and is referred to
convenient for reporting the data, since ordinary as the extract. Compositions of phases which are in
rectangular-coordinate graph paper may be used in equilibrium may be connected by a tie line, such as
constructing the diagram. If the calculation in- o—q. Alternatively, equilibrium compositions may
volves only relatively low solute concentrations, be determined by use of the equilibrium curve on
only that part of the diagram should be plotted, the lower part of the diagram. At the plait point
with an expanded horizontal scale for greater (p) the tie line has been reduced to a point where
graphical accuracy. the extract and raffinate are identical. Note that
The ternary system isopropyl ether-acetic the equilibrium curve ends at the plait point at
acid-water (Figure 3.8) is representative of systems 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
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
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
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
0.20
\ 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).
800
i+
0.6 0.8 I ]
f . ya < lc
|
m oles
propane/lb mole
proppne and propylene
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)
gas phase
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). (
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 )
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
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
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
.
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)
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
Rearrangement yields
Lo_z_ Va7
"
(4.7)
V2 x a0
Lo_
=
Zc - Vc7
(4.8)
V2 x c0
x c0 ~ z c Vcl
(4.9)
x a0 — Za Val
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
E
V is determined by Equation 4.10.
o Zc
TO z - y2 200
2 = 2
yc2 fv2 100
*o
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
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.
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
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 )
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 considered on a solvent-free basis. Thus, L' is the
rule (Equation 4.12): 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
V2 length of linex z compositions. Thus, X
c is the mass of c per unit
= 0.78
2 length of linex K2 mass of a +b. (Note the use of capital letters for
mass ratio.)
or by Equation 4. 1 2 for component a: The notation and relationships then become:
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
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
-,
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.
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.
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
.
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 :
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.
4.2. A liquid containing 40 mole percent n-pentane, (d) 300 lb of the solution in the following mixture is
salt.
(c) 40,000 N/m 2 .
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.
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
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
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.
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
,
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
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
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
L + \Z/v + 1
= Z = LN + V y I 5.1)
(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
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
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
XA =
i [
1 =
(5.6) A L -V, Ln -Vn + }
Ax A = L x - V^y, = L n x n - Vn + ^y n + -
[ L-n — \//v + i
A/7 A =L h -V,H =L n h n -Vn + ,Hn + The numerators of Equation 5.9 are the net flow
: ,
of the cascade, the total net flow (A) is equal to This is shown in Figure 5.4. For the case shown,
Ln — Wv + 1 By Equation 5.1 • x Aa and x Ab are negative, and x Ac is positive and
greater than 1.0. This indicates that the net flows
^-o
— V-\ = Ln — V/v + i
of the solute and major raffinate component are in
Therefore, the total net flow is the same at either a direction opposite to the total net flow, but the
end of the cascade. The total net flow at any stage net flow of solvent is in the same direction as the
(L n — Vn _
is easily shown to be equal to that at
1 ) total net flow and is larger than the total net flow.
either end by writing a mass balance between Even for the fictitious stream A, x a + x b + x c must
either end and stage n. Thus, the total net flow is equal 1.0. Therefore, if x a and x b are negative, x c
constant at any point in the cascade. Equations 5.7 must be greater than 1.0. The delta point located
and 5.8 can be justified in a similar manner. The by using the concentrations at the ends of the
concept of net flow is useful because the net flows cascade will be used to calculate from stage to
are constant through the cascade. Once they are stage, as shown in the following discussion.
evaluated at either end of the cascade, the values Equations 5.6, 5.7, and 5.8 also define the net
can be used at any point. There is usually no need flow at any stage n in the cascade. It can be seen
to determine the actual values of net flows in that the component net flow is constant through
typical graphical calculations, but the concept is all stages at steady state, so that the delta point
64 STAGE OPERATIONS
tion of the compositions of the phases in succes- stages fixed and the recovery is to be deter-
is
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 .
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
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-
(a) Extract purity (the concentration of sol- gave more than five stages, and Trial 2 gave less
Trial 1
v*i
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 Areduction in V/L moves the delta point
the recovery approaching 100 percent and the farther from the triangular diagram; an increase
purity approaching percent as V/L approaches shifts it closer. The minimum V/L may be deter-
infinity. In actual operation, an economic balance mined by extending the tie lines in the direction of
must be made between recovery and purity. A high the delta point. The extended tie line on which the
extraction recovery is desirable, but the solute then delta point first falls as it moves away from the dia-
must be recovered from a dilute extract — by distil- gram determines the minimum V/L. It is difficult
lation, for example. This secondary recovery is to predict generally exactly which tie line will
more difficult with dilute solutions. Obviously, it is determine the minimum V/L. It may be a tie line
impossible to obtain both high recovery and high through the feed composition, through the exit
purity in simple countercurrent flow for this sys- raff inate composition, or at any intermediate point,
tem. The next chapter considers techniques for as demonstrated in the following example.
obtaining both high recovery and high purity. The fraction unrecovered of component a is
given by L N x Na IFz Fa so that the fraction re-
,
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)
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 1:
Final xA
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
X& or (h^ x s
. )
Vwior-
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
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
)
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-
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.
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?
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
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 .
Enriching
section
D A-^*- F A-
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
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
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 (
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).
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 )
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)
5 = A + F (6.5)
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 rene, and the two products are to contain 90
feed may be /.-phase or l/-phase or a mixture. percent and 5 percent styrene (on a solvent-free
Therefore, its composition is denoted by z instead basis). The ratio of extract reflux to product is to
of x or y. Because x N is the composition of a be 10/1. The solvent from the solvent separator
stream flowing from an equilibrium stage, it must contains a negligible quantity of styrene and ethyl-
lie on the phase envelope, as shown in Figure 6.4. benzene. How many equilibrium stages are re-
The compositions x Dl x c and x are identical, , quired? What is the solvent feed rate per 1000 kg
since L c is simply split into D and L If the . of fresh feed?
solvent separation is not complete, L c contains a
Solution. The extraction cascade with extract
small quantity of solvent, as indicated by x c on
reflux is shown in Figure 6.1. The solution to the
Figure 6.4. By a mass balance around the solvent
problem is given in Figure 6.5, which is a replotting
separator, V-\ = L c + VCl and y^ may be located at
of the equilibrium diagram of Figure 3.10. The
the intersection of the straight line / c x c and the
solvent-free coordinates are more convenient than
phase envelope, since V } is flowing from an equi-
the triangular coordinates (see Problem 6.3). All
librium stage.
compositions and masses must be given on a
The delta point (x A shown by Equations 6.
) is
solvent-free basis.
and 6.3 to be on line x /i or Wne x D y c which are ,
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 : ,
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
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
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)
XA C -
= -23.8 A/? A =L hQ - V,H y
(6.11
A' = -410 kg
Figure 6.6. Enthalpy and mass nomenclature
X Ac =34.2 around the condenser.
82 STAGE OPERATIONS
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
(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-
(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 less or water. The choice of medium is based upon the
than one equilibrium stage. intended operating temperatures. For example,
Figures 6.2 and 6.7 show the bottoms with- liquid nitrogen (— 300° F, 40 psig) is used as a
drawn before the reboiler, so that the stream enter- cooling fluid in the condenser of a methane purifi-
ing the reboiler is totally vaporized. In most distil- cation column in a liquid hydrogen plant.
lation towers, the bottoms are withdrawn from the 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 )
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
/?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
.
^*mm
-5,000
-10,000 -
-15,000
-20,000
-25,000
Number of stages
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
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.
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
v v
•
c •
.v+
*6->»=0
three terms on the right-hand side of Equation streams, including naptha, kerosine, light diesel fuel, and
heavy dieselfuel. The heaviest components (tars and/or
6.25, the addition must be carried out in two steps.
asphalt) are withdrawn from the bottom of the column,
First, / and F are graphically added to give the sum
barely visible below the concrete support ring.
2. The 2 and A may be added to give A (i.e., x A
The column, which is 197 ft tall and 28 ft in diameter,
will lie on the straight line x A zv ). The intersection processes 200,000 barrels per day of crude oil. Columns of
of this line with x N y N + i isx A .
thistype typically use valve trays, as shown in Figure 2.10.
The sum I may be considered as a "fictitious The scale of the picture can be appreciated by identifying
feed." The sections of the cascade between 1 and / two maintenance workers on a catwalk about halfway up
and between F and N behave 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,
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.
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
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
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
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?
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-
(e) What is the minimum reflux ratio for this separ- (d) How many stages would be required if the reflux
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
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.
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-
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-
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?
The methods developed in the three previous chap- equilibrium stage, but a new relationship replaces
ters can be applied to the calculation of multistage the delta point in calculating from stage to stage.
separation processes when sufficient physical data The relationship is obtained from a mass balance
are available for the system to be separated. In over a section of the cascade. The following para-
many cases alternative approximate calculation graphs develop the relationship and then indicate
methods may be used. The approximate methods under what conditions it is useful in calculations of
are usually more rapid than those considered pre- specific mass-transfer operations.
viously, and they often require less physical data. A mass balance around any stage n and the
But the methods are based upon simplified physical left end of the simple countercurrent cascade in
models of the systems to be separated, so that the Figure 5.2 gives the following.
methods are applicable only to systems that more Total mass balance:
or less the physical models. Some of the approxi-
fit
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-
)
Xn
solvent (c)
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 /-, ,
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
(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
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
- 0.04
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.
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
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 .
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
J
K '
(7.14)
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
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
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.
y* >. + !
v„ v„*l
Stage n
J..-1 Ln
(a)
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
,
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
SIMPLIFIED CALCULATION METHODS 107
may be determined in the same manner, and the the enthalpy-composition diagram; they are also
curved operating line may then be drawn in, as the equilibrium stages determined for this case,
shown. It is not necessary to use streams actually where L and V are not constant. The variation of
flowing between equilibrium stages to determine the latent heat of vaporization is large for the
the operating line. Any straight line from the delta system in Figure 7.10, but the operating lines are
point cuts the /.-phase and \/-phase curves at com- only slightly curved. Assumption of constant molar
positions that represent a point on the operating overflow to give straight operating lines for this
line. system would have yielded about one less equi-
The complete determination of curved oper- librium stage than was obtained with the curved
ating lines is shown in Figure 7.10 for a case in operating lines in Figure 7.10. An error of this
distillation where the latent heat varies, the con- magnitude may be tolerable in some engineering
struction lines for each stage are shown to illustrate applications, so the assumption of constant molar
how the curved operating lines are located. The overflow is often made, especially when little or no
stepsbetween the equilibrium curve and the curved enthalpy data are available.
operating lines are not only construction lines for
H, h
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.
x a>y a
UNSTEADY-STATE OPERATION
Figure 7.10. Determination of curved operating The calculation methods developed up to this
lines. point have applied to steady-state operation in
108 STAGE OPERATIONS
which the compositions and flow rates do not vary extraction or absorption. The length of this period
with time. In such cases, the feed is added, willdepend upon the flow rates of the two phases,
products are withdrawn continuously, and flow the hold-up volume of the equipment, and the rate
rates may change from point to point in the equip- of approach to equilibrium. The transient period
ment, but at any specified point there is no change may range from minutes in small equipment,
with time. An unsteady-state or transient process through several hours in large-scale installations, to
involves changes of variables with time. There are several days in some special applications.
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)
Batch Distillation
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 (a)
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
-
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 /
is continuously increased during the run. Calcula- xa , mole fraction benzene in liquid
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
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.
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
*
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)
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 curve, where x n is the composition of the liquid
physical variables that determine its value. A leaving the actual stage. Thus, E v is the ratio of
theoretically more fundamental efficiency is likely the actual change in average vapor composition to
to lead to more adequate empirical correlations. the change that would occur if the total vapor were
The Murphree stage efficiencies (5) describe in equilibrium with the liquid actually leaving the
E, = (7.34)
The stage shown is a perforated plate where the *n- 1
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,
Ev =—~ie - 1. (7.36)
mv
where
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
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
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-
/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)
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 */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)]
(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-
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
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-
(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
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 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.
(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?
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.
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
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
MULTICOMPONENT STAGE OPERATIONS 123
(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 ,
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 fore considered together. For these combined
remaining variables (T,P,h). In Figure 3.4, the streams P = 2, so that by Equation 8.4 there are C
pressure is fixed in a vapor-liquid equilibrium, so independent intensive variables associated with the
that specifying a single composition and the two equilibrium mixture leaving the stage. The total
intensive state variables, P and T defines the sys- number of independent intensive variables associ-
tem in the one-phase region. On the other hand, in ated with the streams flowing to and from the
the two-phase region for the two-component sys- stage then (C + 1 + (C + 1 + C = 3C + 2.
is ) )
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)
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
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
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
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
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
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
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
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
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 ,
for the specified methane recovery. operating line lies below the equilibrium curve, as
132 STAGE OPERATIONS
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)
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 result, Yjcannot be evaluated directly from X,, but
it must be determined by trial and error.
limited by the approach to equilibrium between
y 1a and x 0a On the other hand, y 1c =0, because
.
As a first step, an approximate equilibrium
component c is much less volatile than component curve is located using the slope at the origin and
b. one point of high concentration taken at the
bottom of the column (X N Y N ). At the origin ,
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 + :
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
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
(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
,
N=-
Fritx N -yN + y
/K S
(8.13)
logS Example 8.7. Solve Example 8.5 analytically.
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
/
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
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
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)
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
/',
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
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,
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
K„ + 1
* = 0.75x„ 6 + 0.0987 (A) L = L + F and V=V
STAGE 1
Ki/a a *1
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
.
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
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
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
{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 ]
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
) 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)
\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
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'
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
*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:
Stage
No. 7"°F iC 4 H 10 nC 4 H 10 iC 5 H 12 nCsH 12 nC6H 14
Stage
No. r°F iC 4 H 10 nC4Hio 1C4H10 nC 5 H 12 nC 6 H 14
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
Fz F B(x B lx D ca Fz F
)
f'(6) = -0.15918
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 the distillate is to be high, and the concentration of
(starting where the linear temperature profile was other components in the distillate is to be as low as
first assumed). possible. To isolate the effect of each variable, all
The value of 6 for this first iteration other variables are held constant at the values of
(114.9599) is far removed from 1.0 because the Example 8.11. For example, to study the effect of
initial assumed temperature profile is quite inac- varying the total number of stages, the other vari-
curate. This large value of 6 corrects the [x Bi l ables (reflux ratio, DIF ratio, and feed tray loca-
x Di)ca t0 9' ve more reasonable compositions from tion) were held constant. In an actual design all of
which to calculate the second temperature profile. these variables and others would be considered
The 6 for the second complete iteration is found to together to determine the best economic design.
be 0.3623; for further iterations, d is 1.1648
(third), 0.9881 (fourth). Five complete iterations Number of Stages
are required to give 6 = 1.0001275, which is close As shown in Figure 8.16, the recoveries of both
enough to 1 .0. As an alternative, it is possible to butanes and their combined concentration in-
continue iterating until temperature profiles check creases as the total number of stages increases. This
within an arbitrary accuracy (such as 0.01°F). A is as expected, since more stages give greater frac-
third alternative is to require that the stage com-
100
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.
Design Variables
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
_
Feed-Tray Location
Reflux Ratio
Figure 8.19 shows that an optimum exists in feed-
Both recovery and concentration of butanes in- tray location. The optimum occurs near the middle
creases as reflux ratio is increased, as shown in of the column. Location near either end reduces
Figure 8.17. In this figure, all conditions are the both purity and recovery of butanes.
same as Example 8.11, except that L Q /D is varied. This brief discussion of design variables is
Higher reflux ratios for the same distillate produc- intended to show the complexity involved in
tion increase the internal liquid and vapor flow choosing the best design. The design engineer
rates. This in turn increases steam and cooling searches for an optimum in a multidimensional
water costs and may raise equipment costs by system.
requiring a greater column diameter to handle the
greater flows. Here again, an economic balance
must be reached.
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
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.
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.
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
(a) How many equilibrium stages are required at a 8.26. A mixture containing 25 mole percent n-
(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.
(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 .
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
Part Two
Nearly all unit operations involve the transport of port results from the motion of large groups or
mass, heat, or momentum. Such transport may clusters of molecules.
occur within one phase or between phases. In The following six chapters will consider the
many cases, it is necessary to know the rate of fundametal mechanisms of molecular and turbu-
transport of mass, heat, or momentum in order to lent mass, heat, and momentum transport. The
design or analyze industrial equipment for unit three transport processes are closely related, and
operations. For example, in the stage operations they will be developed in a general way wherever
covered in the preceding chapters the rate of mass possible.
transfer between phases was assumed to be great The terms transport and transfer are used
enough to permit the rapid establishment of equi- interchangeably in many cases. If a distinction is
librium. In reality, however, the rate of mass trans- made, transport usually is used when referring to
fer is an important factor in determining the stage the fundamental mechanism within a single phase,
efficiency, which must be known to determine the whereas transfer
refers to the overall process.
actual number of stages necessary for a given sepa- After the transport mechanisms have been
ration. developed, a number of applications will be con-
A knowledge of the principles of momentum sidered. Methods of evaluating and correlating
transport is a prerequisite to the proper design of transport properties will be discussed. Application
the pumps and piping systems so essential to the of the transport principles to interphase transfer
chemical process industry. Similarly, the design of will be developed. Integration of the rate equation
industrial heat exchangers rests on an understand- will be demonstrated where concentration varies
ing of the fundamental transport mechanism for along the length of the contacting path. Finally,
thermal energy. after the fundamentals of transport systems have
The fundamental mechanisms of mass, heat, been thoroughly developed, applications to the
and momentum transport are closely related. The design of industrial equipment will be considered
rate equations for the three transport systems are in Part Three of this book.
of the same form, and in certain simple physical A table of the notation used in Part Two
situations the mechanisms of transport are iden- together with dimensions and typical units, is given
tical. Molecular transport depends upon the after Chapter 14.
motion of individual molecules. Turbulent trans-
159
Nine
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 mum, while that at the wall is zero. Momentum is
the system than in another region, molecular transported from the region of higher flow velocity
motion leads to a net transport of the species from to the region of lower velocity by the mechanisms
the region of higher concentration to the region of of migration and collision. In a gas the intermol-
lower concentration. For example, if a bottle of ecular forces are weak, so that momentum trans-
perfume is opened in the middle of a room of still port occurs through migration and subsequent col-
air, some perfume vaporizes and the perfume mol- lision of molecules. In a liquid, intermolecular
ecules are then transported by molecular motion forces are strong, so that a substantial force must
from the region of higher perfume concentration be applied to induce flow of molecules past each
near the bottle to a region of lower concentration other. As a result, the rate of momentum transport
farther away. Such molecular transport of mass is is higher in than in gases for the same
liquids
usually referred to as molecular diffusion. Because velocity pattern. Molecular momentum transport
molecular mass transport requires actual migration occurs in laminar flow, the two terms are some-
of molecules, the rate of mass transport is much times used interchangeably.
greater in gases than in liquids; diffusion in solids is The practical applications of mass, heat, and
extremely slow for the same difference in concen- momentum transport developed long before the
tration. basic mechanisms were understood. The engineers
The molecular transport of heat occurs from a who applied the principles of the three types of
region of higher temperature to a region of lower transport were not primarily interested in the simi-
temperature. Because temperature is a measure of larities among the mechanisms of transport. As a
the internal energy of a molecule, a higher tem- result, three distinct systems of notation and
perature indicates a greater internal energy. Thus, nomenclature developed. Thus, fluid mechanics or
heat is internal energy being transported from fluid flow is the traditional name for the unit
higher-energy molecules to lower-energy molecules. operation based upon momentum transport. Heat
Because heat is energy in transit, some scientists transfer and mass transfer are also traditional unit
and engineers prefer to use the term energy trans- operations. Although it might be desirable from a
port instead of heat transport. Not all of the fundamental point of view to use one set of nota-
energy of a molecule can be transported across the tion in discussing the three systems, the engineer
available temperature difference. The portion of using a unified nomenclature would not be under-
the internal energy that can be transferred across a stood by the majority of engineers in industry. For
temperature difference is sometimes referred to as this reason, after the transport-rate equation is
thermal energy. The molecular transport of heat is derived using a completely general notation to
called conduction. The transport of heat occurs show the interrelation, it will then be rewritten in
not only by molecular migration but even more the traditional terminology of mass, heat, and
effectively by collision of molecules vibrating momentum transport for application to industrial
around relatively fixed positions. Energy is trans- problems. In this discussion it is well to remember
mitted by collision of vibrating molecules in liquids that there are not only fundamental similarities in
and solids where little or no migration occurs. As a mechanism, but also fundamental differences.
result, many and solids are excellent con-
liquids
ductors of heat, whereas gases are relatively poor The General Molecular-Transport Equation
conductors. A poor conductor is a good insulator, Derivation of the general rate equation for mol-
by definition. In metals, heat is also transported by ecular transport first requires that a physical model
electron migration, as discussed in a following be postulated. The previous discussion indicates
section. that no single physical model is likely to be
Momentum transport occurs in flowing gases adequate to describe transport in gases, liquids, and
and liquids. Under extreme forces some solids may solids, because of differences in structure and
be considered to "flow," but not in the same transport mechanism in the three states. In this
manner as liquids and gases. A crystalline solid may section a very simple physical model based upon
fracture along crystal planes, instead of flowing molecular migration alone will be used to derive
uniformly. This discussion is limited to the fluid the general rate equation. Although the resultant
mechanics of gases and liquids. In a flowing fluid, transport equation is strictly applicable only to the
one region may possess a higher velocity than model, it may be extended to real gases, liquids,
another. For example, with flow in a pipe, the and solids. Because the model considers only mol-
velocity at the center can be shown to be a maxi- ecular migration, ignoring turbulent transport and
MOLECULAR TRANSPORT MECHANISM 163
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)
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
(MLy/6)
Flux of transferent property V net Na lA(MlL y L z d) qlA (HlL y L z d) Ty9c
—f
LyL
-x'-y'-z
Typical Units
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
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
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)
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
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 ,
(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 model gas mc is uniform throughout the volume
the arrow on each molecule indicates the magni- element, and the vectorial sum of all mc is zero.
tude of its velocity, and the direction of the arrow Consequently, it does not contribute to the excess
indicates the direction of the velocity. Each of the of momentum necessary for momentum transport.
molecules in the three slabs shown is moving at the If the flow velocities of the slabs are not equal, an
same speed and consequently possesses a momen- excess of momentum can exist in one slab com-
tum equal in magnitude to those of all other mole- pared to another. The momentum caused by the
cules. Since the direction of movement is random, bulk flow of the gas in the /-direction will be
the sum of the momenta is zero.
all transferred between slabs in the x-direction, as
On the other hand, the model gas in Figure indicated in Figure 9.46.
9Ab is shown to have in addition a flow velocity v The model gas is considered to be flowing in a
in the -(-/-direction. Each molecule then may be regular manner in the +/-direction. Average groups
considered to have two components of momen- of molecules flow parallel with the xy and yz
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 gas flow is concerned. Of course, individual mol-
all three slabs, the momentum concentration of ecules still move randomly in any direction, but
each slab would be the same, and there would be the average movement is uniformly in the flow
no net transport of momentum from slab to slab. direction. This uniform macroscopic movement is
In the gas pictured, the flow velocities in the slabs called laminar flow. A more random macroscopic
differ, as indicated by the length of the arrows flow is referred to as turbulent flow; it will be
representing the flow momentum. This is an discussed in Chapter 12.
idealized picture; the flow velocities within any 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
(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 .
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
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 )
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.
Momentum flux = r v g c = 50 —m —
kg m/s
~
s
. c n 9 m/s
k plete understanding of the theoretical background,
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
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 )
=
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 .
-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 .
(b) Predict the viscosity of argon at 400°C tions in the model, but it is not sufficiently
and atmospheric pressure. accurate for most engineering calculations. More
(c) Predict the mass diffusivity of argon at rigorous kinetic theories must be used to predict
0°C and atmospheric pressure. transport properties, as discussed in the following
(d) Predict the thermal conductivity of argon section on real gases.
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
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
%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
+ )
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 )
Then
9=
T3b *= 298/123 = 2.42
= 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
-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
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
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
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 ,
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
'
)
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
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
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
The stress-shear rate diagram (Figure 9.11) shows applied. Bentonite clay suspensions and some sols
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-
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
Table 9.2 DIFFUSIVITIES IN SOLIDS AT 1 atm* of energy; lattice vibration is an important mechan-
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
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
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 .
(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-
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
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
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 being heated as it passes through a steam-jacketed
position and with time. Thus, it is necessary to pipe. The volume element might be a section of
integrate Equations 9.1 1, 9.21, and 9.31, taking pipe one foot long. Heat is transferred from the
into account variations within the volume of the hot pipe to the water by molecular transport when
system under consideration. In addition, mechan- the water is in laminar flow. In addition, the water
isms other than molecular transport may contrib- flowing into the one-foot section brings with it a
189
190 MOLECULAR AND TURBULENT TRANSPORT
uranium nucleus. The fission of the nucleus liber- fication of the physical model requires a thorough
ates not only neutrons and fission fragments but understanding of all mechanisms that may in-
also energy that appears as heat. Thus, the nuclear fluence the concentration of transferent property
reactor core is also an example of transfer with in the system. The model should adhere closely to
internal generation of heat. physical reality, although certain physical phenom-
The rate of generation as given in Equation ena that have negligible influence are often omit-
10.4 may have a negative value. For example, in ted. For example, in a mass-transport model ther-
diffusion with chemical reaction, the reactants are mal diffusion and forced diffusion are usually
being consumed (to give other chemical species), so omitted because they are negligible compared to
that the rate of "generation" of reactants is nega- ordinary diffusion resulting from a mass concen-
tive.On the other hand, the rate of generation of tration gradient. In practice, the physical model is
the products of the reaction would be positive. frequently a compromise between physical rigor
The rate of accumulation given in Equation and mathematical tractability. A physical model
10.4 takes into account the unsteady-state including all possible mechanisms often leads to
behavior of the transfer within the volume mathematical expressions that are difficult to use.
DIFFERENTIAL MASS, HEAT, AND MOMENTUM BALANCES 191
The model may be simplified to make the math- 1 actually described physical reality with sufficient
ematics more feasible, but the simplified model accuracy. For example, suppose it was assumed
may no longer lead to an accurate description of that the thermal conductivity and heat capacity of
the actual system. Therefore, it is always desirable a metal bar being heated were constant, when in
to check the final expressions describing the sys- fact they varied considerably over the range of
tem by using experimental data obtained from the temperature through which the bar was being heat-
system. ed. The final equation relating temperature to time
2. Balances are written for the pertinent trans- and position would not fit experimental data on
ferent properties. Mass, heat, and momentum bal- temperature taken at various times and positions in
ances are obtained for a small, finite volume ele- the bar. If the final expressions do not fit the
ment, including all mechanisms that appear in the experimental data, the physical model and its
physical model. Because concentration and mech- mathematical description must be examined to de-
anisms vary within the volume of the system, it is termine how they can be improved. This procedure
usually necessary to obtain the balances for an often leads to a better understanding of the actual
infinitesimal volume element. This leads to differ- physical mechanisms involved. It is possible that
ential equations, which are then integrated to de- the model may never fit the data, however, because
scribe the entire system. To obtain the differential of a significant lack of understanding of physical
equations, the limits of the balances on a small, reality. If the data are accurate, they are a better
finite volume element are taken as the dimensions description of physical reality than is any inference
of the volume element approach zero, in the usual drawn from the physical model.
procedure of differential calculus. If the transfer-
The five steps outlined above are applied to
ee property concentration is a function of more molecular-transport systems of varying complexity
than one independent variable, a partial differential
in the remainder of this chapter. Only a few simple
equation is necessary. For example, in the electric
systems are described to demonstrate the pro-
heating element being heated, the temperature is a
cedures. Many more complex systems have been
function of both time and position in the element,
analyzed (1,5). This method of analysis is ex-
so that a partial differential equation is needed to
tremely important to an engineer, who may call
relate temperature to time and position.
upon an applied mathematician to assist in the
3. The flux terms in the differential-balance
integration in step 4, but will have to do the other
equations are replaced by concentration-gradient
steps alone.
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- Geometrical Considerations
able. Systems to be analyzed may take any size or shape.
4. The differential equations are integrated, Many of these shapes are not easy to describe
using appropriate boundary conditions. The re- analytically, so that approximate numerical meth-
quired integration may be quite complex. Some- ods may be necessary. A few simple shapes that
times the several differential equations are inter- may be treated analytically are presented here.
related, so that they must be solved simul- The flat-slab geometry (Figure 10.1a) is based
taneously. In many cases, the equations cannot be upon the rectangular coordinate system. In this
integrated analytically, so that numerical methods chapter, molecular transport is presumed to occur
must be used. The advent of digital computers has only in the x-direction. The area across which the
made numerical methods much more feasible. transport occurs is constant and equal to the area
Bounday conditions are chosen from known data of the plane perpendicular to the transport direc-
on the system. For example, when a fluid flows tion. The finite volume element has an area A and a
through a pipe, the velocity at the pipe wall is thickness Ax.
assumed zero. This information could be used as a For the cylinder (Figure 10.16), the direction
boundary condition to evaluate one integration of molecular transport is assumed to be only along
constant in an expression relating fluid velocity a radius. It is also assumed that there is radial
and position in the pipe. symmetry — that the properties at a given r are the
5. The expressions are checked with
final same regardless of the direction of r. This elim-
available experimental data. This step is necessary inates the need to consider the angular component
to determine whether the model formulated in step in the cylindrical coordinates.
192 MOLECULAR AND TURBULENT TRANSPORT
i> = -5 — 10.5)
dr
(\}/A) = -8A —
dT
dr
= constant (10.6)
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
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 )
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)
(>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)
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
-F2 + (t v A x + Ax = (10.24)
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
d2 T
x = 1 .5 in., v = 1 - (45)
(-j-J
= -23.75 ft/min -kA —
dx
7
2
= (10.31b)
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
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)\
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 .
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
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
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, 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 .
Diffusion of 6
Boundary
Na = _
A
^ dx
0.1 1:
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
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
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
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
/
(?).-£
Na
K"?)
Na
2
J
ft
Stationary air
\ ic_
— (10.59)
(t)-£)(-?)-©(3 Methanol
because the bulk flow balances the molecular trans- The tank is tapered, as shown in Figure 10.5. The
*= 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
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
-
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
(0.62)(1: / 760-0 \
'760- 135/
and
(/Va ) f
= 0.00286 lb moles/hr
Na =j(Na )t
Figure 10.6. Transfer with internal generation.
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
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
->(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
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-\
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)
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
— 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
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
-*(^)=^
\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
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.
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)
(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
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 '/ '
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)
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,
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).
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
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
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.
5 in. -T
cp = 0.032 Btu/lb°F
sphere.
24 hr?
(b) What is the temperature halfway through the (b) Calculate the temperature at r = % in.
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.
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.
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^-
Equations of Change
The mass, heat, and momentum transport equa- applied to the simple physical systems analyzed in
tions developed in Chapter 10 may be extended to Chapter 10, they will, of course, simplify to the
the more general case involving three dimensions same equations developed directly in that chapter.
and time. These general equations, sometimes
called equations of change, are partial
the Time Derivatives
often shows that certain terms in the equations The partial derivative describes the increase in
may be neglected, resulting in an equation that can T (thedependent variable) with respect to any one
be solved without complex mathematics. This of the four independent variables. Thus, 81790 is
chapter serves as an introduction to a very compli- the increase in F with time at a fixed position and
cated subject, focusing upon the concepts behind is called the partial time derivative. Sometimes the
the development of the equations. Applications to partial derivative is written {dF/dd x ,y,z where the
some commonly encountered problems will subscripts indicate which variables are fixed. In this
demonstrate the utility of the equations of change. text the subscripts will not be used. The partial
The approach to the analysis of mass, heat, time derivative (81790) is the time rate of increase
and momentum transport presented in this chapter of the concentration that would be seen by a
contrasts with that developed in Chapter 10. In the stationary observer.
earlier chapter, a physical system was analyzed to If is not stationary, then the time
the observer
develop a transport equation for that specific situa- rate ofchange of concentration varies with
tion. In this chapter, equations are derived that position. The time rate of increase of F as viewed
describe transport systems in general. These general by a moving observer is called the total time deriva-
equations may then be applied to specific physical tive. From Equation 11.1 the total derivative is
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
-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
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
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
.
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
+ 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 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 '
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,
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)
+ \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
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
.
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
= 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
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)
MK = pv :i i.29)
Mz = pw z*U. + Ay
x-direction is
b(pu)
Ax Ay Az !
11.36)
_p _
Dv
=v .
T+v/,_ pg (11.44)
bd
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)
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
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
— — = 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
'
rate of energy generated is bq x bq y bq z
-
9q Ax Ay Az dx by dz -
: 11.6O)
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 )
- - 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
is
v)
ox ay q )
+ — 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
= 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
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.
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
.
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
(b)
Line m n m n m '
Specific Notations
Parameter General
Symbol Notation Heat Mass Momentum
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
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
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
>'x + Ax + (Tx- A;
+ Tr =
A0'x (11.81)
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
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
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. Therefore 2/0.4 = 5.0 time increments will be re-
When the surface temperature of a slab is quired. The initial condition of the slab is shown in
suddenly changed, the question of what temper- Figure 11.7a. The notation ca2 refers to the con-
ature applies at 8 = must be answered before the centration at zero time at 2 distance increments
graphical construction of the Schmidt method can from the surface. The presubscript will in general
proceed. The problem is that at 9 = 0, the surface give the time increment, and the postsubscript, the
temperature may be considered equal to the initial distance increment from the surface of the slab.
temperature of the slab, to the new temperature The initial concentration gradient is horizontal at 3
after the sudden exposure, or to some value be- percent alcohol. However, on exposure to dry air
tween the two. the surface concentration drops to zero. Just be-
One recommendation (effective, though fore exposure it is 3 percent. Thus, c a0 ma y be
without theoretical justification) is to use an arith- either or 3 percent. For the first time increment
metic average of the two initial conditions — that 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
' 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 =
the slab and ca2 in the mirror image. Since (a) An observer standing on the train platform sees
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.
Heat Transfer, McGraw-Hill, New York (1958), Equation 11.44 differ physically from Equation 11.40?
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
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.
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.
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
TURBULENT-TRANSPORT MECHANISM 233
the qualitative difference between laminar and tur- any external vibration, and supersmooth finish on
bulent flow. the duct work. This evidence supports the notion
Consider a glass tube in which water flows at that the incidence of eddies is brought about by
any desired mean velocity. At the center of the external imperfections of the systems of the type
tube a fine jet of water-soluble dye is introduced that cannot be practically eliminated. In those
through a capillary tube so that a thin filament of experiments in which high velocity was maintained
dye is injected into the stream of water; the veloc- with absence of eddies, pressure-drop data were
ity of the dye stream is equal to that of the water correlated with Equation 10.94. A system oper-
at the point of introduction. Reference to Figure ating at abnormally high velocity in viscous flow
12.1a shows that at low water velocity the dye could be made to revert to turbulent flow by
filament retains its identity in the water stream, simply physically jarring the pipe. This indicates
tending to widen very slightly during the down- that high-velocity viscous flow is an abnormal or
stream passage because of molecular diffusion of metastable state.
dye. At a slightly higher mean velocity, as shown in It should also be noted that an upper limit of
Figure 12.1b the dye filament breaks up into finite viscous flow as well as a lower limit of turbulent
large eddies. Farther downstream the eddies break flow seems to exist and that the limits are sep-
up further, and the dye that has been introduced arated by a transition region. The Reynolds appara-
tends to become homogeneously dispersed. At tus and other means of examination (8) also indi-
much higher mean velocity (Figure 12.1c) the eddy cate that the eddy formation begins at the center
activity becomes extremely violent, and the region of the tube to form a central core of eddy activity.
of homogeneous dye color approaches the point of The diameter of the core increases with increase in
dye entry. From visual observation it is evident mean velocity.
that the eddies in normal pipe flow are of the order The conclusions that can be drawn from the
of one-tenth the pipe diameter, and that the eddies Reynolds experiment are: (1) above a certain mean
move completely random pattern. In some
in velocity for a given system, relatively large eddies
experimental studies, the eddy formation has been form that flow cross-stream in some random be-
suppressed by careful control so that high mean havior; (2) these eddies are larger and more abun-
velocities were maintained with absence of eddies, dant at the center of the tube; (3) an increase in
but these conditions demand, for example, deaer- mean velocity of the fluid widens the turbulent
ation of the fluid, insulation of the apparatus from core until the tube is essentially filled with the core
of eddy activity.
-Dye-injection needle
Homogeneous-color zone-
much larger in turbulent flow and that additional
Dye-injection needle
variable dependencies exist, in the Reynolds
experiment the eddies are distinguishable by eye,
Flow —b which is indicative of the magnitude of the eddies.
"
i i
Since the eddy has a random cross-flow component
superimposed upon the bulk flow, it is reasonable
(O
to suppose that momentum can be transferred by
Figure 12.1.The Reynolds experiment (vertical scale
exaggerated), (a) Flow pattern at low mean velocity
eddy cross-flow motion as well as by molecular
with dye injection, (b) Flow pattern at higher velocity cross-flow motion. The visible examinations are
with dye injection, (c) Flow pattern at high velocity sufficient to indicate that the cross-flow motion
with dye injection. exists on a macroscopic scale.
234 MOLECULAR AND TURBULENT TRANSPORT
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
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
+ —+—=
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)
( 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
(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
,
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 )
-\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,
| Mir r
2
p = density of the fluid, M/L r Ly,
= absolute viscosity, M/L Y8
I Laminar Buffer fi
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
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 of a tube; therefore, Equation 12.25 is incorrect
this region can be approximated by Equation for the center of a tube and probably in the region
12.23; therefore, in a practical sense, the assump- near the center. Away from the center the equa-
tion of laminar flow in the small region would lead tion is a reasonable empirical representation of the
to an error in theory rather than to an error in velocity at various points. Actually, it is unreason-
prediction of rate of transport. The buffer layer able to suppose that the velocity profile should
+ +
extends from y = 5 to y = 30. The Reynolds ex- follow three distinct equations; more likely one
periment and others have shown that eddy activity very complex function is correct, but this function
is most pronounced at the center of a tube and is has not been proposed as yet. In spite of obvious
absent? in the laminar sublayer, so that, as the mathematical inconsistencies, the three equations
name implies, the buffer layer is a region of tran- 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:
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-
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
From the definition of u*. velocity distribution, which utilizes three distinct
equations.
{Ty9c)^
u* 2 The nextradial position that will be con-
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
This equation is indeterminate, so that no light can The initial production of eddies has not been
be shed on the eddy diffusivity at the center of the discussed. It is generally supposed and supported
tube by this method. experimentally that eddy activity is caused by mi-
nute imperfections in the system (5). The experi-
All the foregoing indicates that the velocity- ments of Reynolds and others confirm this. If
distribution equations are inconsistent in some re-
eddy activity begins, secondary eddies are formed
spects. The velocity data derived from these equa- by the primary eddies so that unless some mecha-
tions are useful empirical values, but the equation nism exists to cause decay, the eddy activity will
forms are apparently incorrect. If the equations multiply indefinitely. The viscous shear of the fluid
were correct, the derivatives of the equations and especially the high shear rate near the walls
would be valid at all points, and dvldr would be tend to cause eddy decay so that the system of
zero at the center of the tube. eddies comes to steady state as the rate of eddy
Example 12.1 offers some indication of the generation equals the rate of eddy decay.
difficulties encountered when working in a flow
Extensive investigations have been made of
regime that is not rigorously described mathemati-
the size of eddies and their energy content in
cally. The velocity profile equations result from the
relation to that of the overall stream. It is to be
correlation of experimental data without having
hoped that the elaborate studies of turbulence, its
available an adequate theory as a foundation. Con-
origin, and other factors will give complete under-
sequently, the data are reported in a convenient standing sometime in the future. The over-
+ +
orderly manner (as u and y that is not rigorous.
)
simplified developments of the counterpart quan-
The general aim of the experiments was to describe tities in molecular and turbulent transport should
the velocity profile with special attention focused
present an adequate picture of the radical differ-
upon the condition near the tube wall. No partic- ences between them. Simultaneously, the similarity
ular effort was made to describe conditions at the
+
of the mean free path of the molecule and the
center of the tube, and therefore a term in y at
distance traveled by an eddy before it loses its
the center was omitted. All the above emphasizes
identity and of the molecular speed and cross-flow
that the present empirical mathematical form-
velocity should be appreciated.
ulations of turbulence are certainly not theoreti-
It is necessary to appreciate the existence of a
cally adequate. In order to complete the picture of
laminar-flowing layer and a turbulent core, both
turbulence, a rigorous mathematical representation
vaguely bounded and overlapping in a buffer or
is required.
transition layer.When the flow pattern is known,
From the foregoing material, a model has
extension of the same concepts to transport of heat
been described that results in a mathematical defi-
and mass follows simply and logically.
nition of turbulent flow. There are intuitive objec-
tions to a model that considers gross particles of
fluidmaking small, sporadic motions. Although the HEAT AND MASS TRANSFER IN TURBULENT FLOW
model is probably oversimplified, it can be visual- The eddy activity in turbulence serves as a transfer
ized and analyzed, and it is consistent with obser- mechanism for momentum. In addition, since gross
vation. particles of fluid move in cross-flow, this eddy
A more may be derived if each
rigorous model activity also serves as a medium for physical mix-
pulsation in recognized and treated
velocity is ing. The physical mixing of fluids is important in
mathematically as an unsteady-state case. An mass and heat transfer. Consider boundary
a wall
attempt in this direction is presented in Chapter 13 that is capable of delivering heat or mass to a
under "Penetration Theory." Other models of tur- flowing stream. If the concentration of transferent
bulence have been described (1 1,12,14,15) property at the wall is significantly higher than in
From Equation 12.15 it is evident that at any the main stream of fluid, a gradient will be set up
point at which turbulence occurs, molecular trans- at the wall, and transfer will occur into the fluid.
port occurs simultaneously and by a parallel mech- After steady-state operation prevails, the concen-
anism. In Example 12.1 the value of ET was usually tration gradients in the various sections of the fluid
much greater than v. Consequently, although the streams will be shown in Figure 12.6 and as de-
molecular mechanism existed, the magnitude was scribed below.
almost negligible. In systems with high momentum
diffusivities, v may be of such magnitude that a The gradient across the laminar sublayer is
1.
significant part of the momentum transport is by with most of the difference in concentration
large,
the molecular mechanism. between the wall and center of the fluid occurring
242 MOLECULAR AND TURBULENT TRANSPORT
Boundary -
—
Radial position (r)~*
Figure 12.7. Transfer in turbulent flow.
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
A Vi 12
dr
d(pcp T)
(AItH^ 1
dr
Therefore,
dT
or = -5.78°F/ft
dr
+ <12 361
(D 1 K)-* '**»* -
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).
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.
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.
Thirteen
Fundamentals of Transfer
Mechanisms
In Chapter 12 turbulence was studied in detail in ing through the equation by any one term; each
an effort to describe the individual mechanisms quotient so generated indicates in a dimensionless
that are operating at each radial position at steady form the ratio of the contributions of the mecha-
state. The mathematical manipulation of velocities, nisms in the numerator and denominator.
gradients, and diffusivities at points in a system is If such differential equations can be inte-
cumbersome. would be advantageous for design
It grated, there is no profit gained by investigating
purposes if the same information were available in these dimensionless ratios; if they cannot be inte-
the form of simpler functions. The simple func- grated, but each contributing mechanism can be
tions should respond readily to mathematical identified, then these ratio terms are useful in
handling. In addition, the data required to evaluate interpreting the contribution of each mechanism.
the functions should not be difficult to determine If the phenomenon is so complex that negligi-
and, if possible, should be in terms of gross dimen- ble knowledge can be gained from investigation of
sions, mean velocities, and the physical properties the differential equations, then empirical processes
of the system. Equations expressed in gross dimen- are available for evolving dimensionless groupings
sions and mean velocities are the result of integra- of the involved variables. The groupings are
tion between boundaries of equations describing extremely helpful in planning and correlating
point conditions such as Equation 12.15. The experimental programs of investigation, or in inter-
integrated equations describe overall conditions at preting the results of experiments. If the pheno-
some position along flow duct and define transfer
a mena can be analyzed adequately to yield such
cannot be
coefficients. Certain of the integrations dimensionless groupings expressing the contribu-
performed by ordinary analytical procedures, in tion as in Table 13.1 these can be very valuable. If
,
which case the integrated form may be indicated or only the individual variables can be
defined and then evaluated or correlated by identified — with some uncertainty, perhaps —
empirical methods based upon experimental data. then these identified variables can be grouped into
dimensionless terms expressing certain ratios in one
Mechanism Ratio Analysis of the techniques of dimensional analysis, as
The form of expressions for momentum
general described in Appendix A, and resort must be had
transport derivesfrom differential equations to these techniques.
indicating the contribution of each mechanism. In addition to the definition and evaluation of
For certain cases, these differential equations were transfer coefficients, the analogous behavior of
derived in Chapter 11. Even though these equa- mass, heat, and momentum transfer is examined in
tions cannot be integrated in the general case, they this chapter. Further applications of mass, heat,
yield valuable information. Since they are rigor- and momentum transport are dependent upon con-
ously derived from fundamentals, the terms must ditions other than those heretofore described. For
be dimensionally homogeneous. Each term example, in certain momentum-transfer phenom-
expresses the contribution of a particular mecha- ena, the velocity distributions vary with axial
nism. Valuable information can be gotten by divid- position as well as with radial position as in the
246
FUNDAMENTALS 01 TRANSFER MLCHAMSMS 247
P pressure forces
Euler number
pu 2 inertia forces
inertia forces
Froude number
i-9 gravity forces
*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
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
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
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.
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
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
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
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
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
p )
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-
FUNDAMENTALS OF TRANSFER MECHANISMS 253
etry in the friction factor and Reynolds number Example 13.3. Calculate the equivalent diam-
is defined by the diameter of the tube. For shapes eter for a rectangular duct 3 ft high and 5 ft wide.
other than circular, the diameter must be replaced
Solution. For a rectangular duct the stress at
by an appropriately chosen variable or group of
every point on the wall is not constant. Since the
variables that describes the system with a single
dimensions of the duct are of about the same
linear dimension that is equivalent in behavior to
order, in the absence of more exact information
D. If were chosen that was not equiva-
a variable
lent, curves A and B in Figure 13.1 would be
Equation 13.20 may be used as an approximation
for the equivalent diameter.
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
D™
eq = —
4Si
b
=
(4)(5x3)
,„ r,
2(3 + 5)
= 3.75 ft
=
Mry hL 4S, earlier refers tothe condition that occurs after a
D ;13.18)
fluid has progressed through a duct far enough so
(-AP) b
that no further change will take place in the
where b is the wetted perimeter of the duct = irD. velocity pattern with further progress through the
For any other shape a more general form of duct. Fully developed turbulence is assumed in all
Equation 13.18 must be developed. If more than work previous to this topic. This section will
one surface is present, as in the annulus, the vari- examine the behavior of fluids prior to the full
ous surfaces probably exert different stresses on development of turbulence. Examples of this area
the fluid so that the force balance is of study are entry sections of long tubes and large-
diameter short ducts. Outside the field of chemical
-AP)S l
=t A +t 2 A 2 +
1 1 . (13.19] engineering, those involved in the design of ship
hulls and airfoils are more concerned with the
If the stresses are not equal, Equation 13.19 does
study of this flow behavior.
not reduce to any simple function. If as an approx-
Consider a flat plate with an edge facing the
imation the mean stress is used to replace the
direction of flow. The flat plate can be visualized
different stresses on the various surfaces, Equation
as suspended in an infinite duct so that the volume
13.19 reduces to Equation 13.18. For the shapes
of the plate has no effect upon the fluid velocity in
normally encountered, Equation 13.18 is a reason-
the duct; that is the volume of the plate is negli-
able approximation except in extremes, such as
gible with respect to the volume of the duct. The
annuli made up of small wires running through the
velocity of the bulk of the fluid (free stream) is
center of large tubes.
designated as v fs At the instant a differential
.
For shapes other than those circular in cross
element of moving fluid contacts the leading edge
section, assuming that the mean stress is a satisfac-
of the flat plate, the velocity of that fluid element
tory replacement for the actual wall stresses, a
immediately decreases to zero, which is consistent
replacement term D eq for the geometrical factor D
with the concept of no slip at a boundary. Any
may be written
acceleration or deceleration is associated with a
4Si force; therefore, the deceleration of the fluid
D eq = 13.20) element generates a stress in the flow direction on
the surface. At the leading edge of the plate, an
where D eq
= the equivalent diameter. element of fluid of differential volume is deceler-
other words, the equivalent diameter is
In ated instantaneously by a differential area of the
equal to four times the cross-sectional area of the surface. Instantaneous deceleration requires an
duct divided by the wetted perimeter. By substitu- infinite force. Mathematically, (r / ;) = = °°. The ,
v
tion of appropriate terms for cylindrical geometry, next element of fluid (measured in the x-direction)
it can be shown that for cylinders D
eq =D. The would still be moving at v = v fs Successive layers
.
friction factor measured for the special case of from the plate will be retarded as / increases and
round pipe cannot be expected to fit rigorously for the stress distributes itself over a layer of some
noncircular conduits of various configurations. thickness.
254 MOLECULAR AND TURBULENT TRANSPORT
/ 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 )
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
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>,)
^ P '
the position y, dimensionless.
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
.
tion 13.27.
(t ynv
v) 0.0228
2 0.25
13.24)
Vfs P (/V Re
e'6
)
(13.27)
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
m
(0.037) v fs p (10 ) the calculated stress on a plate is safely high.
{Tyh = 2
1 13.30)
'
_
^ 1=
(0.037) vfs p
2
x 10~ 5
)
5 )0 .2 \ p ! 1.8
(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
^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 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
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 )
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
— = 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 Form drag appears in many chemical engineering
that the fluid is moving at a velocity greater than applications. Metering devices are designed to max-
v fSl so that the linear speed of the fluid has in- imize or minimize form drag, according to require-
creased in that region. Beyond the widest part of ments. Roughness in pipe is an example of form
the body, the fluid is subject to a directional drag. The finite constant rate at which particles
acceleration opposite to that at the leading face of settle in fluids is another. Other engineering appli-
the body and to a deceleration of linear speed as cations are the stress on structures in wind and
the fluid returns to the normal free-stream pattern flowing water.
downstream of the body. The summation of all Correlations of flow characteristics and geom-
forces on the body due to acceleration and deceler- etry for bodies in a free stream utilize dimension-
ation constitutes the form drag of the body. less groups that are analogous in concept to the
In Figure 13.4a, a shape is chosen such that friction factor and Reynolds number described in
no sharp discontinuities exist along the length. The the sections of this chapter covering flow through
boundary layer is continuous over the length of the pipe and momentum transfer through the bound-
body but obviously must adopt a shape and char- ary layer. The drag coefficient (CD is defined as
)
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
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 .
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
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 -\
at the wall.
tration of transferent property between the wall
** For momentum transfer the integration may be per-
and the mean value of the fluid to the maximum
formed by writing E T as a function of r and dvldr as in Equation
difference between the wall and the center. The 12.22. The derivative dvldr can be written as a function of rby use
ratio 7 is a function of the distribution of Y with 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, |
\
(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
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-
Consider first the ratio for the process under consideration. Why should
there be a dependence upon whether the fluid is
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
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.
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
: )
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
{» \
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
)
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
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
Table 1 3.3 MASS-TRANSFER COEFFICIENTS FOR VARIOUS EXPRESSIONS FOR DRIVING FORCES
Na (driving force)
General equation:
" [-4-A
\ coefficient /
Coefficients
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
hr
moles of a transferred
ft 2 (mole fraction)
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)
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
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 problem
Pblm ~
o -0 = AA7 atm
(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 )
= 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
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 ,
/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
!
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
)
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 )
113.82)
v h (13.86)
When Equation 13.82 was tested experimentally, it This equation is a statement of the Colburn
was found to correlate data approximately for analogy between heat and momentum transfer.
gases in turbulent flow but did not correlate data Equation 13.86 is derived from two equations
for liquids in turbulent flow nor for any fluids in that correlate experimental data over the range
e < 300,000 and 0.6 </V Pr < 100.
laminar flow. Gases are characterized by values of 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
8
transfer processes differ some manner func-
in
tionally related to the Prandtl or Schmidt number. which the Reynolds analogy.
is
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)
(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 )
"
(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 !
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
d(pCpT)
The Martinelli
/V Nu = 7 + 0.025(/V Pe 08 )
* (13.96)
= 0.625 - 4
A/ Nu (A/p e ) (13.97)
(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
,
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
,
turbulence.
The Martinelli analogy is a valuable starting
point for the prediction of transfer behavior of 0.8
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
/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/
/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)
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
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
/V Nu
= 105(47
= 10,500
. 5) (^84)
/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
/V N u = 800
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 '
Note that over the entire range, the Martinelli HEAT AND MASS TRANSFER IN THE BOUNDARY
analogy is in good agreement with the empirical LAYER
correlation of experimental data. The Colburn
It was shown earlier that a flat plate in a stream
analogy agrees perfectly, as it should, for air,
produces a boundary layer of fluid of velocity
water, and oil, because the Colburn analogy for
lower than the free-stream velocity and that this
these substances is the same as the empirical
layer lies close to and is affected by the surface. If
equation. The discrepancy for oil between the
the same shape at some temperature T is placed in
Martinelli analogy and the empirical equation is :
transport. Test of the analogy shows this assump- changes from Tfs to Ty, the temperature boundary
tion to be true at N Pr = 1 and approximately true layer begins to form at yy Meanwhile, the velocity
.
at N Sc = 1. The Colburn analogy is based entirely boundary layer would have begun to form at / = 0.
upon similarity of empirical equations, and there- The two layers develop according to the appro-
fore, over the range of application, it is in good priate fluid and system properties. An analogous
agreement with experiment Critical examination mass-transfer boundary layer can also occur. Con-
of the analogy shows that it does not represent a sider a pan of water subject to a gas flow parallel to
mathematical statement of the mechanism. The the water surface. If the surface stress of the wind
Colburn analogy when applied to mass transfer does not cause ripples on the surface, the smooth
results in an error of 1.32 to 1.46, as mentioned in surface is geometrically similar to the flat plate,
an earlier footnote. The reason for the discrepancy and if the partial pressure of the water in the air is
is attributed at least in part to surface effects on different from the vapor pressure of water at the
the descending film of fluid. The discrepancy then liquid a boundary layer of water-vapor
surface,
is geometrical, and the implication may be that the concentration will form along the water surface.
systems are not exactly analogous in a geometrical The entrance to a heat exchanger and the entrance
sense. The Martinelli analogy is based upon to a wetted-wall tower are other examples of
equality of eddy diffusivities of heat and momen- thermal or mass-boundary layers combined with a
tum as a statement of mechanism. The Martinelli momentum boundary layer. As in momentum
analogy shows remarkable agreement over the transfer, when the edges of the thermal or mass
entire range of Prandtl number, indicating that the boundary layer meet at the center of the tube, the
analogy is a reasonable statement of the mechan- flow is fully developed, and no further growth of
ism. A similar analogy for mass transfer agrees with boundary layer takes place with progress through
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 "f>
Vfs Tfs
dT
,
rP = -{6 + E)
dx
where in the light of present knowledge
Figure 13.8. Simultaneous formation of momentum
ET = Eg = E/\/ 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
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
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 )
2
v fs P
0.037
(/VRe),
02
is in the order of 10
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 /
This latter values is probably more nearly correct Figure 13.4, the boundary layer separates and an
than that obtained using the Reynolds analogy. eddy forms behind the roughness element The
eddy circulates fluid within a very small finite part
More rigorous analogies have been written for of the free volume of the tube as a result of the
the boundary layers in a manner similar to the position of the roughness element. The local turbu-
work of Martinelli described above. The differences lence intensity in the eddy is greater than in the
between the analogies are dependent upon the region nearby. The eddy transfers momentum from
choice of assumptions and the velocity-distribution the flowing stream with an increase in measured
function chosen for the derivation. The reader may stress. The intense circulation within The eddy does
refer to textbooks in fluid mechanics for the not contribute proportionately to the bulk mixing.
development of these equations (54). It should be remembered that the local eddy is
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
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
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
= — 2/3 = 0.007
*= — (r,-n (13.111)
//V (/V Sc )
1*T
r„sr rft
=r
Turbulent Turbulent
core core
Buffer layer
Stagnant film
_ Laminar
flowing film Laminar
c
f h- '/ h- sublayer
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
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 .
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)
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
WnuU =1.86(/V Pe )
1/3
(-^j (-) (Nt4u) am
(13.127) W Re /V Pr )
1/3 14
ID \ / u \
-
=/„.
Hi
wall temperature (7"i
= 1.86(A/ Re )- 2/3 ^-j (13.131)
k, c P p, h = fluid properties evaluated at
,
/Q
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
(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
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 sity,with the gas motion upward along the surface.
called natural or free convection. A review of free An increase in velocity for a given element of fluid
convection is given in Reference 36. means that the element sweeps out a greater
A common example of natural convection is volume per unit time, which tends to bring new air
the steam-heated radiator. The air near the radia- into the element in diagonal path from the outside
tor's surface is heated, thereby reducing its density cold-air supply. The process is shown qualitatively
so that it is displaced upward by the cooler, more in Figure 13.13. All phenomena described here
dense, surrounding air. A steady-state flow pattern take place within a momentum and thermal bound-
is established, with cooler air flowing into the ary layer established near the wall, with conse-
bottom of the radiator and warmer air flowing out quent velocity and thermal gradients developing in
of the top. The warmed air is eventually cooled in the process. At the start, the velocity pattern of
the room, so it settles and flows back to the base the boundary layer is characteristic of laminar
of the radiator, thereby completing the flow cycle. flow, but with progress up the surface, turbulent
The relatively low flow velocities and heat- flow may occur. After some progress up the sur-
transfer rates of free convection contrast with the face, the original element that is diluted somewhat
higher velocities and transfer rates of forced con- with additional "outside" air reaches a temperature
vection where a pump or fan maintains fluid flow, closer to the surface of the wall, and the thermal
as discussed earlier in this chapter. In cases of driving force is diminished. Further qualitative
relatively low flow velocity, as in laminar flow of analysis of the boundary-layer behavior can be
the previous section, both free and forced convec- made by examining the effect of the various trans-
tion may play a role. Free convection occurs port mechanisms on the rate of acceleration of the
wherever a temperature gradient exists in a fluid, 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
10-
o Vertical plane
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
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
=
P
RT
and
1 dp 1
P-- pdT T
where T is absolute temperature.
(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)
Pi = liquid viscosity
AH V = latent heat of vaporization Q = AH v w = nhDL(Tv - T2 ) (13.144)
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
/? = 0.725
where the total condensate flows smoothly from H,D 2 N
2/3
(T V -T2 )
J
column.
Example 13.14. A condenser
is fitted with
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
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.
=0.011(/V Rec )
1
'^^ /VPr,
/? = 934 Btu/hr ft
2
°F
::-
-
^ ^^^ -
i
7 -
6
5
4
-""-
^^ Recommended
3 ^^-^Eq. 13.142
^-<>
Eq. 13.14CT"-^^^\.
::"
It is necessary to check on the assumption of every condensing apparatus must have associated
laminar flow. with it some boiling apparatus. In spite of the
many necessary applications, boiling of liquids
4A 4w 4 x 30.7
" ec
= 347 remains phenomenon about which little is under-
a
ft irDft (3.14)(tM(0.56x2.42) stood. The apparatus under discussion consists of a
Thus, the laminar flow assumption is valid. submerged solid, usually in the form of tube or a
nearly flat plate, as in a kettle jacket, whose tem-
primary importance. Types of apparatus in which temperature (TB but increasing surface tempera-
)
boiling may occur are many and varied in applica- ture (T-\), the response of the system could be
tion; for that matter, it is almost safe to say that traced. As T^ increases, formation of bubble nuclei
296 MOLECULAR AND TURBULENT TRANSPORT
duces turbulence near the surface with consequent ing bubbles makes it impossible for liquid to reach
increase in heat-transfer coefficient. The increase in the surface, hence starving the interface of liquid
heat-transfer coefficient favored by the decrease
is and generating a relatively stagnant layer of vapor
in viscosity in the high-temperature region near the on the hot surface.
surface. In this range the bubbles originate at Any alteration of the conditions of surface
favored sites on the surface. These sites are surface and fluid can affect the heat-transfer coefficient.
imperfections at which the shape of the imperfec- For example, if the pressure of the system is
tion, and perhaps a residual fragment of vapor or increased, the bubble volume per mole of vapor is
dissolved or adsorbed gas, results in conditions decreased. This decrease will tend to suppress the
favorable for bubble formation and growth. As T-\ incidence of film boiling by decreasing vapor
is increased, additional less-favorable sites become volume and permit increased heat-transfer
active until the entire surface is actively forming coefficients and increased values of the maximum
bubbles. This condition is called nucleate boiling. temperature difference. The vapor blanket in film
If the increase in driving force is continued by boiling is present because the density difference
increasing 7^ at constant TBl the heat-transfer between liquid and vapor in the gravitational field
coefficient decreases after passing through a 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 surface, the liquid will spread under vapor bubbles
heat transfer in the vicinity of the wall increases tending to "snip" them off and away from the
sharply because of the vapor blanket. Figure 13.19 solid surface, whereas, if the liquid does not wet
is a plot of h as a function of 7", — TB and shows the surface, there will be less tendency for the
the maximum value. vapor bubbles to disengage. (Contrast this with
One well-supported theory proposes that as dropwise condensation.) Thus, a low surface
the generation of vapor increases with increasing tension or tendency to wet a solid is favorable to
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)
)
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
(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
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
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:
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
(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
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.
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
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 As an example, situations arise in heat transfer
being transferredfrom a hot fluid through a tube where radiation, conduction, and convection occur
wall to a cold fluid. Gas absorption involves the simultaneously. The conduction part of the transfer
transfer of mass from a solute-rich gas phase to a process has one resistance while parallel to the
solute-poor liquid phase. In the case of a fluid conduction resistance; there is a different resist-
flowing through a pipe, momentum is transferred ance along the radiation path.
through the fluid to the pipe wall. A further complication occurs if the heat-
In Chapter 13, the rate equation was transfer process is considered at another point in
integrated in the transfer direction at a given posi- the exchanger where the temperatures of both
tion in an apparatus to define a transfer co- phases have changed. Although the same qualita-
efficient. It is the purpose of this chapter to show tive arrangement of resistances is present, the
the interrelation of resistances and driving forces in changes in temperature have resulted in different
an interphase transfer process at a particular point temperature gradients and different resistances at
in a piece of equipment and to integrate the rela- this point.
tionships over the entire transfer area. In addition,
a design equation relating the rate of transfer to Series Resistances at a Point: Steady State
the total required transfer area will be developed.
It has already been established that the rate of
transfer, be it heat, mass, or momentum, is equal
resistance
(14.1!
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
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)
-(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)
2 (ADy
\pA=-
7=1
(14.7)
(^)totai=(-Ar> , orfl )(—
\ Ha
+—
Ho
)
'
(14.14)
/=n
2 R,
/-i
-AT, - Ar ? UjAj)
(iM)totai = 's 7T - /
(14.15)
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
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
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)
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
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
;
Cold-phase temperature *-
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
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
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
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
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
0.236 + 0.262
Example A
double pipe heat exchanger
14.2. A = 0.249 ft
2
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)
Total resistance = —
UA
that the resistance of the copper wall
for two reasons: first, and, second,
it is thin,
is negligible
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
(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
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
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 )
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 )
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
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
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)
CaL
Figure 14.9. Phase relationships for soluble Figure 14.10. Phase relationships for in-
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.
-
^ 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*
— 0.01565 -
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
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.
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 , )
= 39.1
KL mkc k L [0.01955M2.09) 0.068
Total resistance, both phases = {ca * - ca )
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 )
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
=
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
2 cr(IVV) 2 d(T L L)
! 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
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
•-=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
(d) Repeat part (c) for the same size pipe, but made 14.15. A vapor mixture of ethyl alcohol and air is
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
(lb moles/ft 3 )
1 1
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-
temperature distribution, what is the heat loss from this 14.18. Solve Problem 11.18 such that the sphere is
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?
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)
latent heat of vaporization (H/M), Btu/lb mole Q rate of heat transfer (H/8), Btu/hr
*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 )
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
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
In the next several chapters, the fundamentals of experimental data taken in the laboratory or pilot
heat, mass, and momentum transfer will be applied plant. An engineer asked to design a plant must do
to the design of industrial equipment. As it is used it with the data available or recommend what
here, design in most cases will be limited to the additional data are needed. The engineer cannot
evaluation of the transfer area required. This expect a complete and precise theoretical basis for
evaluation is only the first step in a complete the proposed process therefore and must be willing
design. The detailed process and plant design that and able to use empirical methods and approxi-
must follow the determination of transfer area will mations based upon experience whenever the theor-
not be considered here. Such a design must include etical background is inadequate to do the job.
complete specification of mechanical details, ma- There is no such thing as the permanently
terials of construction, electrical components, pro- perfect design, because of the possibility of new
cess control and operation, and many other factors. and cheaper materials of construction or because of
In addition, a detailed economic analysis must be a shift in either the technical or economic speci-
made on the many process alternatives to deter- fications under which the equipment was originally
mine which is the optimum — that is, which will designed. It is necessary that the design be tech-
produce the product at the lowest unit cost. nically sound, economically close to the optimum,
The engineer may be asked to design new and cognizant of safety and working conditions for
equipment or to evaluate and suggest improve- operating personnel. In many plants, the value of
ments in the performance of existing equipment. continuing production justifies reasonable expendi-
The calculation methods considered in the succeed- tures to avoid interruptions.
ing chapters may be applied to either the synthesis The design and operation of any chemical
of new processes or the analysis of existing pro- plant require the close cooperation of specialists in
cesses. Although the fundamentals covered in part many technical and nontechnical fields. Engineers
two are completely appropriate in describing the must understand and appreciate the problems of
transfer phenomena under study, it is in their their coworkers whenever they influence her or his
practical application that many compromises must own area of specialization.
be made. Often the theoretical development under- This section will consider the unit operations
lying a particular application is incomplete. This based upon heat transfer, mass transfer, simul-
lack of complete understanding leads to the use of taneous heat-and-mass transfer, and momentum
empirical equations most commonly based upon transfer.
325
Fifteen
Heat Transfer
The chemical process industries all utilize to a very Double-Pipe Heat Exchangers
large extent the transfer of energy as heat. The The simplest type of heat exchanger is the double-
purpose of this chapter is to apply the design equa- pipe heat exchanger shown in Figure 15.1. The
tion as it has been developed to the practical solu-
double-pipe heat exchanger is essentially two con-
tion of industrial heat-transfer problems. Two
centric pipes with one fluid flowing through the
mechanisms of heat transfer have already been center pipe while the other fluid moves cocurrently
established:
or countercurrently in the annular space. The
length of each section is usually limited to standard
Molecular — the transfer of heat by molecular pipe length, so that if an appreciable heat-transfer
action; this is referred to as conduction.
surface is required, banks of sections are frequently
Turbulent — the transfer of heat by a mixing used. If the required area is too large, a double-pipe
process; this is usually referred to as convection.
exchanger is not recommended. The use of a double-
pipe heat exchanger is not limited to liquid— liquid
It has been established that these two mech- heat exchange but may also be used for gas— liquid
anisms can occur simultaneously or individually. exchange and for gas-to-gas exchange. Materials of
A mechanism of heat transfer that is
third construction may vary, depending upon the fluids
common but has not yet been discussed is radiation, being handled. Either fluid may be moved through
which is the transfer of heat by emission and the tube or annulus at relatively high velocities,
absorption of energy without physical contact. thereby aiding in the heat-transfer process.
Unlike conduction or convection, which depend
upon physical contact for thermal-energy transfer, Shell-and-Tube Exchangers
radiation depends upon electromagnetic waves as a
When the required heat-transfer surface is large, the
means for transferring thermal energy from a hot
recommended type of exchanger the shell-and-
is
source to a low- temperature sink. Radiation may
tube variety. In this type of heater or cooler, large
occur simultaneously with, or independent of, the
heat-transfer surface can be achieved economically
other two mechanisms of transfer.
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
INDUSTRIAL HEAT-EXCHANGE EQUIPMENT
end of the tube into a close-fitting hole in the tube
The heat exchangers used by chemical engineers sheet by a process called "rolling." The resultant
cannot be characterized by any one design; indeed, tube bundle is then enclosed by a cylindrical casing
the varieties of such equipment are endless. How- (the shell), through which the second fluid flows
ever, the one characteristic common to most heat around and through the tube bundle.
exchangers is the transfer of heat from a hot phase The simplest form of the shell-and-tube ex-
to a cold phase with the two phases being separ- changer is shown in Figure 15.2— a single-pass
ated by a solid boundary. tubular heat exchanger. The fluid flowing through
327
328 APPLICATIONS TO EQUIPMENT DESIGN
Annulus
fluid
—c
Tube
fluid
1
out
the tubes enters a header or channel where it is of the tubes at high velocity, which gives good
distributed through the tubes in parallel flow and heat-transfer coefficients. The number of tube
leaves the unit through another header. Either the passes employed depends upon the economics of
hot or the cold fluid may flow in the shell of the the design and operation and upon the space avail-
exchanger surrounding the tubes. able. Complexity in design sometimes results in
Parallel flow through all tubes at a low veloc- expense in fabrication that must be balanced
ity gives a low heat-transfer coefficient and low against improved performance. Another disadvan-
pressure drop. For higher rates of heat transfer, tage of the multipass exchanger is the added
multipass operation may be used. Such a heat frictional loss due to higher linear velocities and
exchanger is illustrated in Figure 15.3. In this type the entrance and exit losses in the headers. Only an
of construction, the fluid in the tubes is diverted economic balance can indicate the most judicious
by baffles within the distribution header. The design.
liquid passes back and forth through some fraction Note that in Figure 15.3 baffles are placed
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.
HEAT TRANSFER 329
Shell
fluid
out
Figure 15.3. Heat exchanger with four tube passes and one shell pass.
• Floating head
(a)
Floating head
(b)
Shell cover
•
Floating tube sheet
Shell cover gasket
Floating cover
(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
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
Type of Exchanger Fluid Velocity, ft/s Fluid Velocity, ft/s Btu/hrft 2 °F Ref.
(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)
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.
Specific heat = 0.45 + 0.000625(7° F - 60), Btu/lb right-hand side of Equation could be integrated (/?)
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- .
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
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
= 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
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 )
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
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 :
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
1.38
12
= 0.115ft
= 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
/V Pr = —
cpp
k
=
(0.45X0.39x2.42)
r-—:
0.085
= 5.0
= 9180 Ib/hr. Thus,
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
^^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
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 .
{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 *)
'
—
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
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
400° -» 200°
TA y-TA2 t
Ar =200°
1 AT2 = 100
C
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
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
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
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
^
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
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
The deposit that increases resistance may be Within the plant is an unused shell-and-tube,
simple dirt, mechanically deposited; it may be a floating-head exchanger have one shell pass and
true crystalline scale deposited because of the in- two tube passes. The shell is mild steel, and the
verted solubility curve of some dissolved constit- tubes are type 304 stainless. Specifications of the
uent present; or it may be a deposit originating surplus exchanger reveal a 20-in. I.D. shell with 158
from chemical reactions within the fluid or be- 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 square pitch. Baffles are placed 6 in. apart.
a complex function of heat duty, flow rate, geo- Will the surplus exchanger be suitable for the
metry, and the particular characteristics of the fluid. cooling requirements? Assume the inorganic liquid
Because scale buildup varies with time and its has the physical properties of water.
increase accompanied by a drop in the transfer
is
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 tubes of 304 stainless will probably be acceptable
exponential decrement in coefficient has been used for this purpose. The river water will in all prob-
to express buildup in other cases. Scales may ap- ability be dirty, and the exchanger should have a
proach an asymptote, or they may "mushroom" removable tube bundle. The square pitch of the
rapidly once they start (29). tube layout is good for cleaning the outside of the
The absence of a definitive theory regarding tubes. Consequently, from a point of view of
scale formation usually requires actual perform- materials of construction and ease of maintenance,
ance data to help in determining length of service the surplus exchanger seems adequate. But, this is
before cleaning. The limit in permissible fouling not enough, because the heat load on the ex-
may be expressed in resistance units or as a per- changer will demand a certain transfer area. It will
centage of the clean coefficient below which opera- therefore be necessary to estimate the cooling sur-
tion becomes impossible or uneconomical. face required and check this against the available
HEAT TRANSFER 349
- 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
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
0.301 u (0.68x2.42
350 APPLICATIONS TO EQUIPMENT DESIGN
\ 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/
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
ature may be calculated by the usual manner of The required area is.
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
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.
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)
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-
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
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
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- forces more comfortable.
cation of new equipment, with conventional With the hot stream assumed the minimum,
geometry and known coefficients, these tools offer the operating line is always above and steeper than
little if any advantage over conventional techniques the equilibrium (45°) line. Its span on the ordinate
using UAAT calculations. For a true counterflow represents the temperature increment of the hot
exchanger, with no change in phase, the advantage stream, the span on the abcissa represents the tem-
appears only in evaluating q and the outlet temper- perature increment of the cooler stream, and the
ature, thus avoiding a trial-and-error solution for slope (m) represents the ratio of C max /Cmjn The .
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
(b)
j
UdA = — 15.38) 1 15.35a/?)
^m in
and a rate equation
For the case where essentially all resistance is
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,
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 '
= = 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
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^--*-
,
- - '
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 )
becomes Cmin
Cf,
Cc becomes Cmax n r 12 3 4 5
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\>.
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
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
HEAT TRANSFER 359
because the flue-gas travel is uniformly vertical, that for electrons and atoms. It has gradually
eliminating the flow-direction changes that cause become understood that these properties are dif-
much of the pressure drop in horizontal furnaces. ferent aspects of the same phenomenon and that
The monorail attached to the top of the stack the aspects are necessary and complementary. The
permits the removal of the heating-element tubes uncertainty principle of Heisenberg states that it is
vertically through the header-box doors. impossible to determine simultaneously the
velocity (or associated energy or momentum) and
The Nature of Radiation position of a particle. Thus, experiments designed
Although the mechanism of radiant-energy transfer to determine particle velocity find that radiant
is not completely understood, the associated energy has the characteristics of waves, whereas
phenomena are explainable in terms of a dualistic experiments designed to determine particle posi-
tion find that the energy has particle properties.
theory. This theory deals separately with the
emission and reception of radiation, and with its The wavelike appearance of energy probably arises
transmission. Radiation is emitted and received in
from the identical form of equations describing
discrete particles or pulses called photons. The
wave motion and those describing the probability
of finding a photon at a given location.
energy transmitted in each photon is a function of
the frequency of emission but is otherwise con-
Upon photons travel
striking a receiver, the
into it they strike an electron or nucleus that
until
stant. The emission of energy must lower the
is susceptible to the energy level of the photon.
energy of the emitting body. In terms of the Bohr
model of the atom, which pictures electrons orbit- The collision results in an increase in energy of the
receiving atom, as an increased amplitude of vibra-
ing around the atom nucleus at various energy
tion of the nucleus, as an increasing rotational or
levels, emission would occur as electrons jump
translational energy of the molecule, or as an
from a high-energy orbit to a lower-energy orbit.
increase in energy level of one of the orbiting
Emission also occurs by lowering the vibrational
electrons. Since solids have closely spaced atoms,
energy of the nucleus with respect to other nuclei
radiation is usually absorbed very close to the
in the molecule or by decreasing the rotational or
molecule. The surface. For liquids, the penetration would be
translational motion of the
greater, whereas for gases, much greater penetra-
frequency of emission is a function of the energy
tions are obtained. The depth of penetration will
levels of the emitting atoms, and, since any finite
body contains atoms in a wide range of energy depend upon the characteristics of the incident
states, the emitted radiation will have a complete
photons as well as upon the characteristics of the
In transmission, radiation
receiving body.
spectrum of frequencies.
has all the properties of an electromagnetic
— that Thermal Radiation
wave is, a vibrating electric field and a
similarly vibrating magnetic field. The planes of All solids, liquids,and most gases receiva and give
vibration of these fields are perpendicular to each off energy in the form of electromagnetic waves,
other, and the directions of vibration are perpendi- which are similar whether they are perceived in the
cular to the direction of radiant transmission. The form of thermal energy, light, X-rays, radio waves,
speed of the beam is fixed by the medium through or radar signals. These various forms differ only in
which it travels. In a vacuum, the speed is about wavelength. Light waves, those perceived by the
186,000 miles/s, whereas in denser media it is human eye, have wave lengths ranging between
slightly less. In any case, 0.35 and 0.75 micron, whereas electromagnetic
waves received as thermal energy have wavelengths
C=A/7 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.
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
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,
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.
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 .
follows that a black body emits radiation at the energy is absorbed and consequently emitted by
maximum rate possible of any body of equal size the nonblack body per unit of surface (Btu/hr ft 2 ).
and shape at the same temperature. Consider two
e = ae b = ee b ! 15.58)
bodies identical in every characteristic except that
one is black, whereas the other has an absorptivity which defines e, the emissivity, as the ratio of the
(a) less than 1. If each body is put into a furnace rate at which energy is radiated from a unit surface
held at a constant temperature (7" ), both bodies and that at which it would be radiated from a
7"
will eventually reach This will occur because
.
black surface of equal area. From Equation 15.58,
thermal energy will be transferred to each of them
a = e (15.59)
by radiation from the furnace until such time as
there is no longer a temperature difference which is known as Kirchhoff's law. Since a is
between the furnace and the enclosed bodies. Once temperature-dependent, e must be also. Thus,
they reach T the bodies must maintain this
, Kirchhoff's law holds only at constant tempera-
temperature, growing neither hotter nor cooler ture, or, in other words, thermal equilibrium
than the furnace. This is necessary by the second law between emitter and receiver is required.
of thermodynamics, which explicitly prohibits the Although black-body characteristics cannot
spontaneous generation of a temperature differ- be found in naturally occurring surfaces, they are
ence between two bodies that can transfer thermal very closely approximated by a cavity in a solid
energy only to each other. At this condition of substance that is open to the atmosphere only
thermal equilibrium, then, each body must be through a pinhole in the solid surface. Any radiant
absorbing and emitting thermal radiation at equal beam coming to this opening will pass into the
rates; otherwise, the temperature would not stay cavity. There it will bounce from wall to wall,
constant. The rate of absorption and emission for being partly absorbed at each contact. Because its
the black body will be different from that for the chances of finding the opening again are almost nil,
nonblack body. The situation may be clarified by the beam's energy will be completely absorbed,
writing thermal-energy balances around each body. 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
= 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
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
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
- >
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)
.
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.
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
and
- oAT2
(<7nb Absorbed
(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. (
rounding walls.
60
1
|
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
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
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
'
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
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
"*"
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 '
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)
—
_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 y
'
=dA, cos0i
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
,
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 .
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
(15.92)
or
Equation 15.92 gives the total radiation from a ~
= 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)
-
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.
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
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=
10'
x = 40
"
(a) (b)
100°F.
Wall 4
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
*~
?
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 =
^^^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
]
a 2
= 3580 Btu/hr transferred from the fur-
where the factors b, c, B, and C are as defined on
nace door to the wall
Figure 15.31, and Fi-»(i+u) = 0.22.
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
qjA = aT *t
x l x
= t
Effect of Zero-Flux Surfaces
e l
P 2 "i 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
c7 ,= a^ 1
e1 _ 2 (r 1
4
-T 2
4
) (15.105) or
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 )
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 '
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
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
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.
the resulting surface emissivity factor is approx- 9. Chilton, T. C, and A. P. Colburn, Ind. Eng. Chem.,27,
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)
mp., °F c P Btu/lb/°F
, U, Btu/hr (c) Use arithmetic average of h at 80°F and h at
100 0.50 540 (d) Compare answers with the rigorous solution of
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.
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
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
(
— )
-
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
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.
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
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.
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
384
MASS TRANSFER 385
•Packing istics.
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
Specific
Surface Packing Dumped
Percent (a v ) Factor Weight
Packing Voids (e) ft
2 /ft 3 F lb/ft 3
2 63 111 580 54
1 73 58 155 40
2 74 28 65 38
Metal Rasch g rings (1/16 in. wall)
i
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
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
the V-phase, whereas the downward-flowing phase and a component balance over the same section
will be the /.-phase. gives
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
.
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 :
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 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
'
,
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,
.
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
//
a i
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
,
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
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)
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)
=
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
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
and
/.' = /.(! -x) (c)
(d)
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 - 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
.
relative magnitudes of resistances encountered in decreases as liquid rate increases, but varies little
the transfer process. Unfortunately, this is not so with gas rate (except at very low gas rates, where
simple because a constant value of is not usually m H G approaches zero as the gas rate approaches
encountered outside the range of dilute solutions. zero). H G increases as tower height increases, but is
Data for heights of transfer units in either 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
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
)
3 3
p v = gas density, lb/ft (kg/m
— o = liquid surface tension, dynes/cm
oc = critical surface tension of packing
)
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)
= 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
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
/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
)
•
——
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.
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
i i
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'
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
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
'
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
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
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
transfer unit (H c) was found to be a linear func- thrown to the outside wall by centrifugal force,
tion of the ratio of the velocities of the two phases. where they coalesce and run down to a collector.
Vc The scrubbed gas exits by a central duct, essen-
H oc =a +b (16.49) tially free of liquid droplets. The Venturi scrubber
mVD disperses a liquid absorbent into a high-speed gas
where
moving through a narrow constriction in a flow
H c
= height of the overall continuous-phase
duct, thereby obtaining a fine dispersion and large
transfer unit, ft interfacial transfer area. The resulting mixture is
Vc
= superficial velocity of the continuous then separated in a settling tower or cyclone. Both
water phase, based upon the empty the cyclone scrubber and the Venturi scrubber are
tower cross-sectional area, ft/hr used to collect aerosols and dusts, as well as for gas
VD = superficial velocity of the dispersed absorption. They require considerable energy for
toluene phase, based upon the empty gas and liquid flow compared to other contacting
tower cross-sectional area, ft/hr devices.
m = slope of the equilibrium curve, = m
dy* Idx (for this system, m = 0.865)
a, b = empirical constants for a particular CYCLIC TRANSFER
packing size The mass
unit operations involving the transfer of
For 1/2-in. packing, a = 3.8 and b = 2; whereas for between and a solid phase include adsorp-
a fluid
1 -in. packing, a = 8.2 and b = 2; so larger packing is tion, ion exchange, drying, and leaching. Adsorp-
less effective. 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
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 ions. Anion-exchange resins may have fixed ionic
rigors of In addition, adsorbents are
handling. groups of primary, secondary, or tertiary amines,
selective in their ability to adsorb specific solutes. phosphonium, and arsonium. Many types of
Consequently, the chemical nature of the solid exchange materials in various physical states have
must be considered to assure satisfactory perform- been synthesized. For example, spherical beads,
ance. Commercial adsorbents include bentonite, membranes, fibers, and liquids have been prepared.
bauxite, alumina, bone charcoal, fuller's earth, The "zeolites" are naturally occurring clays that
carbon, and silica gel. are capable of exchanging cations.
Equilibrium data for adsorption are usually Typical of the operation that occurs during
presented in the form of adsorption isotherms (see 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)
The rate of transfer in ion exchange is compli- solid adsorbent. The rate at which adsorption
cated by the complex mechanism of the transfer occurs is influenced by one or more of several
process. A postulated mechanism of ion exchange factors:
consists of the following steps:
1. Diffusion of solute from the flowing gas
1. Diffusion of ions from the bulk liquid to the external surface of the solid particle.
phase to the surface of the resin. 2. Diffusion of the solute into the gas-filled
2. Diffusion of the ions from the surface of internal pores of the particle.
the resin into the resin to the site of exchange. 3. Reaction at the gas-solid interface.
3. Exchange of the ions at the active site. 4. Diffusion of the adsorbed solute within
4. Diffusion of the replaced ions from within solvent phase.
the resin to its surface.
5. Diffusion of the replaced ions from the If none of these steps are slow, it is possible to
resin surface to the bulk liquid. imagine instantaneous and complete adsorption
upon the first increment of the bed as the first
Any one of the five steps may be a rate
increment of gas enters. The solid adsorbent will
limiting although too few data exist to
factor,
immediately adsorb solute until it is in equilibrium
permit any generalizations. The rate for step 3 is with the entering gas stream. Thereafter, the enter-
best described by a kinetic reaction rate, whereas ing gas flows unchanged through this first incre-
the rate for the diffusion steps may be described ment, and onward to have its solute adsorbed in
by the usual rate equation for counter diffusion as the next increment. This action continues until
indicated by Equation 16.14. Once again, a suit-
enough feed has been introduced to saturate the
able mass-transfer coefficient is needed. If the
entire bed to equilibrium with the gas, at which
exchange of ions (step 3) is very rapid compared to time the effluent gas instantaneously increases
the diffusion steps, then ion-exchange equipment
from zero concentration of solute to the concen-
may be designed by employing Equation 16.11. If tration of the entering stream.
the rate-controlling step is the exchange step, then During the time the adsorption front is pro-
kinetic procedures are needed.
gressing through the bed, the unadsorbed compo-
nent has been flowing through the remainder of
Analysis of Fixed-Bed Adsorption the bed and out as the effluent. If the concentra-
Because of the several interacting physical mechan- tion of solute in the effluent is measured, it will be
isms involved in adsorption or ion exchange, and found to be zero until the time at which the bed
because such calculations are usually applied to a becomes saturated; then it suddenly increases to
finite quantity of adsorbent in a cyclic operation, the entering composition.
the design of fixed beds for adsorption can be an The time required to saturate the bed is a
involved procedure. Extensive summaries of the function of the adsorptive capacity of the bed, the
principles and procedures are available in Refer- flow rate, and the concentration of adsorbable
ences 1 2 and 1 3. A simple procedure for the analy- constituent in the feed. Accordingly, a direct pro-
sis of fixed beds is presented in the following portionality exists between the time required to
paragraphs, as applied to certain limited situations. 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 )
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
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,
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
rium with the gas entering (of composition Y, ). Figure 16.18. Compositions in
Equilibrium also exists between the vapor leaving the adsorption zone.
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
\
y dY
Yb
£ i
x. Zy
za
Y*
dY
(16.68)
~1 .Y
Y* - Y
t
vb
M-Mb
Zy = :16.69)
~ ~
Ms _(1 -f)Ma
V T
T where M the cumulative mass of effluent cor-
is
0.35
from the operating line relationship
0.342
caL = CaS
1.94
200
CaL.s =(0.95X0.342)
50 -
= 0.325
E F G
F= —
100- (0.426) (22.7)
100
= 0.903
_ (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
This expression shows that f is equal to the area exchange resin to remove the Cl~. In practice,
above the breakthrough curve in Figure 16.21. The mixed resins are used to remove both ions simul-
area below the curve is easily obtained from the taneously.
data of columns F and G by numerical integration. Although several figures were used to aid in
The area is found to be 0.574; hence, f = 0.426. the visualization of the solution of this example,
5. Calculation of H OL Since the overall
:
none of them were necessary for the calculation.
liquidphase mass-transfer coefficient is given, it is The operating line and the equilibrium curve were
convenient to calculate H 0L using a modified ver- expressed by analytical functions, and both
integrations were carried out using a numerical
sion of Equation 16.66
method. Hence, problems of this type can be
za = HolNol 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
16.75)
point under observation mole fraction
,
Liquid
M -<>----- ^=^U plate {E v
efficiency for reasons discussed in
Chapter 7. The molar flow rate of the bubble
)
Gas I
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_.
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
.
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.
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
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
(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
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
SIMULTANEOUS HEAT AND MASS TRANSFER - HUMIDIFICATION 421
be used as part of a solvent-recovery system, but in of liquid. The scrubbers are often applied to
these applications the condensable vapor is not removing dusts from a gas stream or for reacting
water but instead a solvent such as trichlorethy- the liquid with a component of the gas stream as
lene, benzene, or methanol. Any one of these well as for removal of a condensable component
applications can, in principle, be carried out in from the gas phase.
similar pieces of equipment. The directions of mass As water for process cooling has become
transfer and heat transfer are determined by the more precious, the recooling and reuse of this
relation between the humidity and temperature of water have become common. Most often mechan-
the inlet gas phase and the temperature of the ical or natural draft cooling towers are used,
contacting liquid. The unit size, the convenience of although this cooling can also be done in spray
recovering gas or liquid phases, and the materials of ponds or in large holding ponds. Spray ponds have
construction used all limit the applicability of a the disadvantage of small capacity for a given pond
single piece of apparatus. area and large water losses. Typical spray nozzles
The most obvious form of humidification for these applications deliver 50 gal/min each at
equipment the spray chamber. Here, the contact-
is about 7 psig. They are located about 6 ft above the
ing liquid sprayed as a mist into the gas stream.
is pond surface and spaced about every 4 ft. In
Gas velocity is kept low so that the contact time is exposed areas the pond might be surrounded by a
high and so that there will be only a small amount louvered fence about 12 ft high to reduce wind
of liquid physically entrained in the gas stream. losses. Empirical correlations are available to allow
These units are usually restricted to small-scale estimation of the size of spray pond necessary for a
operations and are frequently used in humidity given duty at known air conditions and spray
control of a room or plant where either humidi- nozzle spacing and design (9). Typical water load-
2
fication or dehumidification of the inlet air is ing rates for spray ponds are about 500 Ib/hr ft of
required. pond area.
The fume scrubber shown Figure 17.1 is a
in More typically, and especially for larger cool-
variation of a spray chamber. Here, throat pieces ing duties, cooling towers are used. These towers
are used to assure close contact of gas and liquid are most often of wood construction with multiple
streams, and the alternate open areas allow separa- wood-slat decks. Aluminum, steel, brick, concrete,
tion of the two phases. Nozzle capacities are norm- and asbestos-board casings have also been used.
ally 5—10 gal/min of water with throat pieces The water is sprayed above the top deck and
handling 300ft 3 /min of gas per gallon per minute trickles down through the various decks to a
Stack
Water inlet
Spray nozzle
Figure 17.1. Spray chamber type of fume scrubber. (Courtesy Schutte & 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
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 this book. Figure 17.4 is a cutaway drawing of a
Further information is available in References 3, 4, double-flow, induced-draft cooling tower. Here air
5, 12, and 24. enters both sides of the tower and flows across the
intermediate-sized water-cooling applica-
In path of the falling water to a central duct. A fan
tions, forced or induced draft towers are usually pulls the air up the duct and blows it out the top
used. Induced-draft towers are preferred because of the tower. Water is distributed initially by flood-
they prevent the recirculation of the humidified ing a distributor plate containing porcelain nozzles
air. In these towers, fans are located on the top of through which the water flows onto the packing.
the tower. Air is pulled into louvers around the Packing is of wood-slat construction assembled
bottom of the tower and up through the decks without nails. The drift eliminators, which are
countercurrent to the water flow. In the largest wood baffles through which the air passes before
towers (capacities up to 100,000 gal/min of water), entering the fan, reduce entrainment of water and
air may be pulled in through two open sides of the thus reduce mist content in the air discharged from
tower and up to a central fan through inclined the fan. Such a tower would be about 48 ft wide
baffles. Empirical correlations are available to by 30 ft high with a fan 14 ft in diameter. Several
facilitate cooling-tower selection and design (1). 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 17.5. 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
100p a 100/
Percent relative saturation =
(Pa)s Ys
(17.4)
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-
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
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
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
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
mm
.
V 17.15 becomes
Figure. 17.7. General humidification
process. c L (T2 -T )(Y2 -Y,)=[ch2 (T2 -T )
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 ,
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
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)
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 ,
should be obtainable from conveniently measured mass diffusivity (£?) varies only slightly over a wide
quantities such as the dry- and wet-bulb tempera- concentration range. Thus, the variation in A/ Sc is
tures. slight and may be considered negligible when satura-
The saturation locus and lines of constant tion occurs at low molal humidity. For the lines to
percent or relative humidity are of primary import- be straight and parallel, the ratio k y \ w lh c must be
ance on the humidity diagram. As discussed constant throughout the entire range of T and Y
previously, the saturation locus is directly related covered by the chart. Thus, /V Sc and \ w must
to the vapor-pressure curve, and it will appear either be constant or vary in a compensating way.
roughly as shown in Figure 17.10. Of similar curva- For the air— water system, the lines are approx-
ture will be lines of constant percent saturation or imately straight and parallel and are often so rep-
of constant relative saturation. If lines of constant resented.
percent saturation are plotted, they will fall at Curves connecting all points having the same
ordinates directly proportional to the ordinate of adiabatic-saturation temperature are also plotted
the saturation locus at the same temperature. For on the humidity chart and are useful in calcula-
example if Ys = 0.10 at T u Y at T^ for the 30 tions concerning humidification and drying pro-
percent saturation curve will be 0.03. If lines of cesses. From Equation 17.21a, the slope of the line
constant relative saturation are plotted, no such connecting any point and its adiabatic-saturation
direct proportion exists. This is evident from the temperature will be —c h -[/\sa If c h is constant
.
definition of relative saturation, Equation 17.4. over the range of temperature and humidity
The relative-saturation curves will always be lower covered by the locus of all points having the same
than the equivalent percent saturation curves. Only adiabatic-saturation temperature, c h l\ sa will also
one of the measures of partial saturation is norm- be the slope of this locus. Generally, c h is not
ally presented on a single chart. constant but increases as the humidity increases.
In order to allow immediate determination of Then, the locus of points of fixed adiabatic-
the absolute humidity from wet-bulb temperature 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^ ). ,
Temperature [T)
Figure 17.11. Constant wet-bulb- temperature and adiabatic-saturation-temperature
curves on a humidity chart.
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"
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
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
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
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
,
Temperature (Tv )
transfer
Figure 17.15. Conditions in a cooling tower Figure 17.16. Conditions for latent- and
(humidifier). 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)
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
400
70 80 90 100 110
/ V'dH v = I L av c L dTL (17.54)
Liquid temperature <TL or T )
t
V'(H V 2-H V :) = L aw c L (TL2 -TL ,) have the values of TLn and H Vn The interface .
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
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.
235
= 118
V
and the line CB results. The
position of point B
givesH V2 =3550, but nothing about T V2
tells .
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
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,
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/ :
(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
=
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 \
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
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, )
.
- °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
—
i
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.
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
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
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,
Slope = 50 = —
Lc v V 18
=
-;
z 8ft S 50
= 18lbmoles/hrft 2
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
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 .
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
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
Figure 17.29. Conditions of mercial towers based upon the experimental results
Example 17.6(b). will be oversized if they are shorter than the
experimental column but undersized if they are
Thus, the Mickley construction forms a tool longer. This is probably one of the reasons for the
by which rate coefficients may be easily deter- generally held belief that increasing the height of a
mined for full-size operating cooling towers and packed tower decreases its efficiency. Another
dehumidifiers. Unfortunately, almost no data of cause of this effect is the fact that bed porosity at
this sort have been reported in the technical litera- the walls is always higher than that at the center of
ture. the tower. Therefore, the liquid tends to flow
If coefficients obtained by measurements on down the column wall instead of being distributed
operating-plant equipment are unavailable, the evenly throughout the packing. This channeling
designer must use coefficients obtained on experi- effect is greater in tall columns, and elaborate
mental or pilot-plant equipment or coefficients redistribution systems have sometimes been used.
obtained on the basis of analogies between heat, The mass-transfer coefficients for humidifica-
mass, and momentum transfer. Despite the tion are not fundamentally different from those
simplicity of humidification systems where all the obtained on absorption equipment. Only gas-phase
mass-transfer gradient occurs in the gas phase, data 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
0.0025
0.0020
0.0019
0.0018
1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500
Hv , Btu/lb mole of dry air
other gas-liquid mass-transfer coefficients, they will The methods discussed here apply to the air-
depend upon the gas flow rate, viscosity, and dif- water system or to any system where h c al
fusivity and upon the geometry of the packing. k Y ac h = 1. If this restriction cannot be met but
They will also be dependent upon liquid flow rate, h c alkYac h = r, the development must be based
primarily because at low liquid rates the packing is upon Equation 17.51a. Instead of enthalpy, the
not completely wet. Thus, gas-absorption mass- graph would have (c h rT+\,Y) plotted against T
transfer rates, such as those calculated from Equa- and used as the "driving force" in the rate-equation
tion 16.74, should be usable for humidification or solution. Simplified procedures for this general
dehumidification operations if the geometries are case are available (10). Such procedures retain the
identical. graphical method used here but adjust the
Part of the mass-transfer data available from interface-locus curve to account for the fact that r
experimental work are summarized in Table 17.2 is not unity.
(21). Some understandable inconsistencies among
Natural-Draft Cooling Tower Design
data taken from a variety of sources are present.
Also, since the relations shown are empirical results In the design of natural-draft cooling towers, such
of experimental data, they should not be extra- as those shown in Figures 17.1 and 17.2, the design
is complicated by the fact that the gas velocity is
polated beyond the range of the data without full
consciousness of the uncertainty that results. not directly controlled. The warm, humid air leav-
A similar situation exists in relation to appro- ing the tower packing is less dense than the cool,
priate heat-transfer coefficients. For the air— water dryer air that enters the packing. Thus, it is dis-
system at high liquid rates, h c a may be calculated placed up through the hyperbolic chimney by the
through the Lewis relation. At low liquid rates, force of gravity acting upon the air surrounding the
incomplete wetting of the packings makes the area tower. Airflow velocity thus depends upon the air
for heat transfer (a h considerably larger than that
)
density decrease caused by the transfer processes
for mass transfer (a M). The result is that measured
occurring in the tower and by the tower height. Of
lkyaM may be two or three times as
values of h c a h course, the extent of the transfer processes, in
large asc h depending upon the liquid flow rate (7). turn, depends upon the airflow velocity.
The liquid-phase heat-transfer coefficient may not The development of the simultaneous equa-
be so readily calculated. Here, experimental data tion set describing 1) the effect of humidity and
on the packing to be employed must be used, or temperature on air density, 2) the heat and mass
analogies between heat and mass transfer must be transfer in the column packing (Equations 17.51,
utilized. 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
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
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
between tiers
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,
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
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
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
(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
(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-
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
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
As used in this chapter, the term drying applies to of droplet sizes that, of course, dry to different
the transfer of liquid from a wet solid into an extents under the same conditions. In the deter-
unsaturated gas phase. The removal of moisture gent drying, exposure of the droplet to the drying
from gases, also frequently referred to as drying, has medium produces a tough, partly dry shell around
been discussed under the names of dehumidification the droplet that inhibits moisture escape. Heat
and adsorption. flows readily through the shell, however, and
The factors discussed in relation to humidifi- evaporates the liquid inside the droplet. The vapor
cation and dehumidification and the mass- and produced swells the droplet, often rupturing it and
heat -transfer rate equations written in Chapter 17 sometimes even blowing new "balloons" on the
apply also to drying operations. The drying sides of the original droplet. Thus, the dried
process, then, is identical with a humidification particles are mixtures of hollow spheres, sphere
process except for the influence exerted by the fragments, and clusters of hollow spheres. The size
solid itselfon the drying process. This influence is distribution cannot be readily predicted from the
considerable. The study of drying and the calcula- size distribution of the spray itself. The rate of fall
tion of required dryer size must take into account of the particles through the hot air, and therefore
a host of problems in the areas of fluid mechanics, the drying time, is very difficult to predict, and the
surface chemistry, and solid structure as well as the initial droplet-size distribution is itself only
transfer rate problems dealt with in the discussion occasionally calculable. Again, in drum drying
of
of humidification. In many cases, these physico- such products as milk and baby cereal, the slurry
chemical phenomena are so complicated and so coats a hot drum and adheres to it during drying.
incompletely understood that the quantitative The thickness of the drying layer is a function of
design of the dryer is impossible. For example, in the surface tension and cohesive characteristics of
the drying of timber some of the liquid is held the slurry, but the exact relationship is still un-
within the wood fibers. This moisture can migrate known. Of course, the thickness controls the
into the drying air only by diffusion through the extent of drying produced and hence the final
fiber walls. Since this moisture diffusion through moisture content. These factors take on added
the wood is slow, the wood may dry nearly com- importance because, in general, the dried solid is
pletely at the surface before the liquid escapes. The the valuable product. Its shape, color, stability,
uneven drying can cause cracking and warping of stickiness, and hence its salability all depend upon
the wood. The whole problem has been attacked the drying process to which it has been subjected.
empirically so that conditions can be set to dry
timber successfully, but the basic mechanism of
liquid movement remains doubtful. Again, in spray
GENERAL DRYING BEHAVIOR
drying of detergents and of many other materials, In drying a wet solid with a gas of fixed tempera-
the required drying time depends to a large extent ture and humidity, one general pattern of behavior
upon the size of droplets sprayed into the hot always appears. Immediately after contact between
chamber. Any atomizing device produces a range the sample and the drying medium, the solid
456
SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 457
content. Beyond this point, the surface tempera- Figure 18.2. Typical drying rate curve for constant dry-
ture rises, and the drying rate falls off rapidly. The ing conditions, drying rate as a function of moisture
falling-rate period may take a far longer time content.
than the constant-rate period, even though the
moisture removal may be much less. The drying period represented by segment AB
of the curves of
rate approaches zero at some equilibrium moisture Figure 18.1 and Figure 18.2 the unsteady-state
is
content which is the lowest moisture content period during which the solid temperature reaches
obtainable with this solid under the drying condi- its steady-state value. Although the shape shown is
tions used. Figures 18.1 and 18.2 show typical typical, almost any shape is possible, and AB may
drying curves, one on the basis of moisture content occur at decreasing rate as well as the increasing
versus time and the other on the basis of rate of rate shown. During the constant-rate period (seg-
drying versus moisture content. The moisture- ment BC of the drying curves of Figures 18.1 and
content— time plot (Figure 18.1) is the form in 18.2), the entire exposed surface is saturated with
which drying test data might be obtained. Figure water. Drying proceeds as from a pool of liquid
18.2, the rate-of-drying— moisture-content plot, is with the solid not directly influencing the drying
much more descriptive of the drying process. It is rate. It is possible that the roughness of the solid
obtained by differentiating data in the form of surface over which the liquid film extends may
Figure 18.1 and thus is subject to considerable increase mass- and heat-transfer coefficients, but
scattering of data and resulting uncertainty. this effect has not been firmly established. The
These typical drying curves are related to the surface temperature reaches the wet-bulb tempera-
mechanism by which drying occurs. The drying 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
Time (0), hr
a measurable constant-rate period. At point C, the
Figure 18.1. Typical drying curve for constant drying moisture content of the solid is barely adequate to
conditions, moisture content as a function of time. supply the entire surface.
458 .APPLICATIONS TO EQUIPMENT DESIGN
During the drying period between points C catalysts,zinc sulfate monohydrate, and sodium
and D of Figure 18.2, called the "first falling-rate phosphates. For these substances, equilibrium
period," the surface becomes more and more moisture contents are usually very close to zero.
depleted in liquid because the rate of liqiud move- Most organic solids are either amorphous,
ment to the surface is slower than the rate of mass fibrous, or gellike and constitute the second of the
transfer from the surface, until at point D there is major classes. These materials hold moisture as an
no significant area of liquid-saturated surface. The integral part of the solid structure or trapped
part of the surface that is saturated dries by con- within fibers or fine interior pores. In these sub-
vective transfer of heat from and mass to the stances, moisture movement is slow and probably
drying gas stream. Vapor from lower levels in the occurs by diffusion of the liquid through the solid
sample diffuses to the part of the surface that is structure. As a result, the drying curves of the
not saturated and then continues its diffusion into substances show only very short constant-rate
the gas stream. This mechanism is very slow com- periods, ending at high values of critical moisture
pared to the convective transfer from the saturated content. For the same reasons, the first falling-rate
surface. period is much reduced, and most of the drying
At moisture contents lower than that at point process is liquid-diffusion controlled; that is, the
D of Figure 18.2,all evaporation occurs from the drying rate is controlled by the rate of diffusion of
interior of the solid. As the moisture content liquid through the solid. The bulk of the drying
continues tofall, the path for diffusion of heat and occurs in the second falling-rate period. The
mass grows longer, and the concentration potential equilibrium moisture contents are generally high,
eventually decreases until at XE '
the equilibrium
t
indicating that a significant quantity of water is
moisture content, there is no further drying. The held so intimately in the solid structure, or in such
equilibrium moisture content is reached when the fine pores, that its vapor pressure is significantly
vapor pressure over the solid is equal to the partial reduced. Because the water present is such an
pressure of vapor in the incoming drying gas. This intimate part of the solid structure, the solids are
period is called the "second falling-rate period." affected by moisture removal. The surface layers
tend to dry more rapidly than the interior. If the
Gasses of Materials According to Drying Behavior drying rate is high, it may cause such differences in
Materials may be divided into two major classes on moisture content through the sample that warping
the basis of their drying behavior. Granular or or cracking occurs. In other cases, it may cause
crystalline solids that hold moisture in the inter- formation of a relatively impervious partly dry
stices between particles or in shallow, open surface shell that further inhibits interior drying and may
pores constitute the first of these classes. In these accentuate the unevenness of moisture content
materials, moisture movement is relatively un- through the sample and consequent tendency for
hindered and occurs as a result of the interplay of solid deterioration. Because of these reactions, the
gravitational and surface tension or capillary conditions under which drying occurs are critical.
forces. The constant-rate period continues to The conditions must be chosen with primary con-
relatively low moisture contents. Although the sideration for the effect of the conditions on
falling-rate period divides into the two regions product quality, and process economy or operating
mentioned above, it usually approximates a convenience must be subordinated. Examples of
straight line on a rate-versus-moistu re-content such materials include eggs, detergents, glues,
basis. The which is usually inorganic, is
solid itself, soluble coffee extract, cereals, starch, animal
relatively unaffected by the presence of the liquid blood, and soybean extract.
and therefore is unaffected by the drying process.
Moisture Movement - Diffusion Mechanism
As a result, drying conditions can be chosen on a
basis of convenience and economic advantage with In relatively homogeneous solids such as fibrous
little concern over the effect of the conditions on organics, gel like substances, or porous cakes, mois-
the properties of the dried products. In the case of ture probably moves toward the surface mainly by
hydrates, the drying conditions will affect the pro- molecular diffusion. The rate of moisture move-
duct by changing the hydrate obtained, but other- ment is then expressed by Ficks's law (Equation
wise the materials are not affected by the drying 11.13) altered to apply to this particular case.
conditions over wide temperature and humidity 2
d X'
ranges. Examples of this class of materials would = 9, (18.1:
be crushed rock, titanium dioxide, chrome yellow, dd dx 2
SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 459
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
Drying-gas flow
approaches the temperature of the drying medium, ward the end of the drying process.
but the surface from which evaporation actually
occurs remains at the wet-bulb temperature. The CALCULATION OF DRYING-TIME
second critical point is hard to find experimentally,
In calculations involving drying, the drying-rate
curve must be considered in its major sections, for
Drying-gas flow
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)
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 )
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
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)
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
1
iii
- paper, newsprint
1
1/
i
// 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.,
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
!
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"
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
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
1
1
1
1
1
1
0.038
shown in Figure 18.11. From the humidity chart, 1
=
*o = 1.0^
0.4
is 20 percent for the conditions given. Thus, for
•\
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
,
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
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- some heat is always transferred to the sides and
ture and humidity are considered to be constant. bottom of any pan or conveyor system used. Only
The calculation of drying rate and time in in through dryers, where the drying air flows
continuous dryers requires enthalpy and material through the solid being dried, is this assumption
balance calculations as well as the use of the heat- relatively good.
and mass-transfer rate equations. Complete and Third, no evaporation occurs during the initial
rigorous calculations involve successive approx- warm-up or cool-down period. This assumption
imation solutions of the proper equations written brings in a varying error depending upon the condi-
for each small segment of the dryer. To reduce the tion of the feed relative to its steady-state tempera-
labor involved, a number of approximations are ture maintained during the constant-rate period.
usually made, although they are not essential to By using it, the condition of stock and drying
dryer calculations. medium at the start of the constant-rate period
The first of the approximations is the assump- may be more readily calculated.
tion that the heat transferred is transferred from
all Fourth, moisture removal occurs entirely at
air to stockby convection mechanisms. This the adiabatic-saturation temperature. The heating
assumption requires that the apparatus operate of the sample that occurs during the falling-rate
nearly adiabatically and that heat transfer by con- period is thus assumed to occur after the last of the
duction and radiation be negligible. moisture has been removed. This final condition is
Second, the area for mass transfer is assumed closely followed in the drying of a granular sand-
SIMULTANEOUS HEAT AND MASS TRANSFER - DRYING 469
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)
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 = .
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
,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
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.
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
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
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
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,, -
=
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
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
-4 2
dryer operates at 4 rpm with a slope of 0.5 in. in 50 x 10 v
1
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/- chance of entraining significant amounts of solid
hr ft 2 given above probably applies to much larger into the airstream. Here an air temperature of
particle sizes than are involved here, so that 1000°R is assumed. Of course, some entrainment
another method of estimating the limit of possible may occur at lower gas-flow rates because of the
airflow must be used. Dusting must occur when the complex flow pattern and the interference of
particles are picked up by the airstream. Actually, particle against particle. Therefore, a gas-flow rate
the airstream does not flow directly opposite to [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)
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 )
Blower
Cyclone
separator
Solids,
feed
Heater
IlhQ- Air
feed
Blower
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
-
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
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
.
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 8,35. Twin-drum dryer with splash feed. Splashing the feed onto the drum
1
.=::' :.'="
Figure 18 3-6
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 the wet scrubber may be returned to the feed.
evaporation of liquid from the spray keeps the Wide variations on this system are possible, usually
product temperature low even in the presence of dictated by the properties of the feed and the
very hot gases. Spray dryers are relatively large desired product characteristics. The drying
units and may be inefficient in their energy use. chamber may operate with airflow countercurrent
Because of their wide range of utility and to the product flow — or in a complex flow path.
convenient product form, spray dryers are used Secondary air may be fed into the drying chamber
with an enormous range of products. A partial list as shown here, or may be omitted, or may enter at
would include coffee, milk, detergents, dyestuffs, several points. The product-collection system often
pesticides, polymers, tile slips, blood plasma, uses only one cyclone collector, but it may also
enzymes, penicillin, starch, metal 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
a feed delivery and atomization system, a hot-gas erosion of the nozzle orifice.
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 nozzles must be completely homogeneous. If the
throughput and give larger and more uniform feed is formed by mixing several ingredients, agi-
droplet size than does the two-fluid nozzle shown tation must be thorough enough to eliminate any
in Figure 18.40. Thus, they are often used in of these small particles. Even so, fine screens are
production-sized spray dryers. A pressure nozzle usually placed in the feed line ahead of the nozzle
built for spray-dryer service is shown in Figure and may be actually built into the nozzle. Homo-
18.41. The nozzle produces a high-velocity tangen- genizers are often incorporated in the high-pressure
tial motion the liquid being sprayed. The result-
in positive-displacement pumps used to feed pressure
ing centrifugal force makes the fluid swirl around nozzles. Homogenizers force the feed through
the circumference of the nozzle hole forming an air short capillary tubes at high velocity, thus breaking
core along the axis of the hole. The fluid then swirls up the feed stream and violently remixing it.
out into a hollow, cone-shaped sheet that finally Centrifugal disk atomizers may be used to
breaks into droplets. Figure 18.42 shows this form spray fluids that cannot be made homogeneous
of atomization from a swirl type of pressure nozzle enough to pass through a nozzle. Moreover, they
operating at 30 psig (20) spraying water. To get the produce an extremely uniform droplet size and do
same degree of atomization with a viscous liquid, a not require a high-pressure feed or impart an axial
pressure of several thousand pounds per square velocity to the sprayed droplets. Centrifugal
inch might be necessary. atomizers are also less affected by variations in the
Both pneumatic and pressure nozzles require feed properties such as percent solids or viscosity,
the fluid being sprayed to flow through narrow or even by variations in feed flow rate. Such an
passages. Thus, any lumps, crystals, or other solids atomizer is shown in Figure 18.43. In the atomizer,
suspended in the fluid will clog the spray nozzle. the feed liquid is delivered to a wheel rotating at
Moreover, even the finest grit will wear the nozzle 6,000-20,000 rpm. The fluid is accelerated to a
tip, causing enlargement and uneven atomization. high centrifugal speed on the disk and forced off
For these reasons, the material delivered to the 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
Core of the entire spray chamber. Nozzles spray axially
in the spray chamber, and hence require a tall,
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- and whole spheres. Figure 18.44a shows a photo-
ing air until it completely disappeared. In drying micrograph of a spray-dried organic acid. Here, the
solutions or emulsions, the drying particle reaches rupturing of droplets was relatively nonviolent,
a temperature somewhat higher than the wet-bulb producing secondary hollow spheres attached to
temperature as drying proceeds. Initially, the liquid the original ones. In Figure 18.446, a similar
evaporates from the droplet surface. The relatively photomicrograph is shown of a spray-dried mold-
dry surface may form a tough shell through which ing powder. Here, rupture of the droplets resulted
liquid from the interior of the droplet must diffuse in the production of fragments.
in order to escape. This diffusion is a much slower It is obvious that the time required for drying
process than the transfer of heat through the drop- cannot be calculated by the simple mechanism for
let shell to the interior, so the liquid tends to drying beds of porous material. The time will
evaporate in place. As a result, the droplet swells, depend upon the temperature, humidity, and flow
making the shell thinner and diffusion through it conditions of the drying gas, the size of droplets
faster. the shell is relatively inelastic and
If produced by the atomizer, and the properties of
impermeable, the internal evaporation is usually the material being dried. Moreover, the properties
great enough to cause rupture of the shell, either of the finished product will depend upon the same
producing fragments or making a new hollow bulb factors. The importance of these factors has long
beside the original one. Thus, the typical spray- been recognized, and a considerable amount of
dried product consists of ruptured hollow spheres theoretical and experimental work has been done
486 APPLICATIONS TO EQUIPMENT DESIGN
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
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 . .
100
v^ ^^^.^^
^L
>. 80
». 2_
60
100 150 200 250 300
1
P. lb/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)
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
1649 (1951).
p.
adiabatic, tunnel dryer in sheets 1 m by 3 m by 2 cm in
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-
'
2.25 40.57 7.50 36.22 the belt if this belt moves at a velocity of 1 ft/min.
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
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
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
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
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
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.
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
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)
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
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
Vapor
w
Figure 19.5. Cross-sectional diagram of vertical-
lowering the heater relative to the disengaging
space. It cannot be prevented so easily when the
heat-exchange surface is inside the evaporator
tube evaporator with forced circulation. body.
498 .APPLICATIONS TO EQUIPMENT DESIGN
^^B^n
SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 499
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
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
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 *•
between the boiling point of the solution and that culations necessary to fix the heating-area require-
of the pure solvent at the same total pressure. This ment of a single evaporator from a knowledge of
simple relation holds only under stringent restric- its duty are straightforward. The heat-transfer-rate
tions. The range of boiling point must be narrow, equation need only be solved simultaneously with
and the solution must obey Raoult's law. This the overall heat balance and the total and solute
restriction implies that the solution be dilute and 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 violently mixed and therefore homogeneous in
found to be roughly straight and parallel to this composition. This assumption is realistic except in
line but displaced above it. Such a plot for the the case of long-tube vertical evaporators and
NaOH— 2 system is shown in Figure 19.11. This special types such as the turbulent-film evaporator.
system is highly nonideal, as indicated by its large Since 50 percent NaOH solution is being removed
heat of solution. Nevertheless, the Duhring lines from the evaporator, the solution within the
are reasonably straight and parallel up to 60 mass evaporator must be uniformly 50 percent NaOH.
percent NaOH. Thus, for most systems, the (b) The boiling point of the solution is that
Duhring lines can be plotted if the boiling point is 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
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 )
40 50
NaOH, percent by weight
pressure increases tends to make the ratio of vapor sequences that combine these two (i.e., mixed
produced per pound of steam condensed less than feed), or they may be fed in parallel, with fresh
unity. feed evaporating to final concentrate in each
effect. multibody operation, the number of
In
Multiple-Effect Evaporation stages may
be different for steam and liquor. Three
In any evaporation operation, the major process common feed arrangements for multiple-effect
cost is in the steam consumed, therefore, methods evaporation are illustrated by schematic flow
of reducing steam consumption (or of increasing sheets in Figure 19.14.
economy, defined as mass of vapor produced per In addition to the economy increase in
unit mass of steam consumed) are very attractive. multiple-effect evaporation, a capacity variation
The most common of the available methods is to would be expected. Note, however, that the
use the vapor generated in the first evaporator as temperature difference from initial steam to the
the heating fluid for a second evaporator. Ideally, final condenser, that was available for a single-
this method should produce almost 2 lb of vapor effect evaporator will be unchanged by inserting
for every pound of steam consumed. The method any additional effects between the steam supply
is feasible if the second evaporator is operated at a and the condenser. For the simplest case, where
lower pressure than the first, so that a positive each effect has area and coefficient equal to that of
value of —AT is obtained across the steam-chest every other effect and where there are no boiling-
surface of the second evaporator. Obviously, point rises,
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)
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 ,
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-
,
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.
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)
,
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 , , , , ,
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.
,
or
Solids balance around effect 1 \
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,
) )
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
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 ,
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
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 )
To condenser
101.7' F
Vt Ib/hr V2 Ib/hr V, Ib/hr
Steam
344.3° F
125 psia
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)
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
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
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
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 -
-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 '
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 , , ,
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)
(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)
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
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.
o
result, there are several large recompression-
a.
3|n
>
evaporation units in operation in European
1o
Di
</)
_ chemical plants.
T^C
/
Figure 19.22.
Enthalpy
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
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
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 517
Of the evaporating crystallizers, the salting used as a cooler, in which case crystallization
evaporator is the most common. In older forms, occurs by solution cooling. Flow rises in the ex-
this crystallizer consisted of an evaporator below ternal heater shown in Figure 19.24. The flow tube
which were chambers into which the salt
settling that pierces the crystallizer body continues to the
settled. The chambers were connected in tandem bottom of the crystal collector and classifier.
so that salt could be removed from one of them Hence, flow is upward in this classifier, allowing
while the other was connected to the system. simultaneous contact between the crystals and the
Modern practice no longer has these settling slightly supersaturated solution and classification
chambers directly connecting to the body. They of the crystals. This is the unique feature of this
are most commonly replaced by a classifying crystallizer. The largest particles are thus the ones
settler, through which a stream of the liquor from that progress to the bottom of the classifier and are
the evaporator body is pumped continuously. The withdrawn in the product magma. Fine crystals
51 APPLICATIONS TO EQUIPMENT DESIGN
^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 solid phase in equilibrium with saturated solution
chemical species that separates can be determined will be ice. This system exhibits an eutectic at
from a temperature-composition phase diagram, about 4.5 percent Na 2 S0 4/ at which concentration
which is often called a solubility diagram. Data on the minimum freezing point temperature of
the solubility of solid compounds in solvents as a 29.84° F occurs. Although the data for this diagram
function of temperature can be found in most are presented at a total pressure of 1 atm, the
handbooks, and a very complete collection is given diagram can be used over a reasonable pressure
by Seidell (10). In Figure 19.27, data for the range because only liquid and solid phases are
Na 2 S04-H 2 system are plotted to show solu- involved. For crystallizer calculations in which the
bility as a function of temperature for a total final temperature is known and in which there has
SIMULTANEOUS HEAT AND MASS TRANSFER - EVAPORATION AND CRYSTALLIZATION 521
been no evaporation, the saturated-solution con- its features are in order. This system has four
centration, the weight of crystal per given weight hydrates and a eutectic at point b. Points c, d, e,
of feed of known concentration, and the species and f represent transition points from one hydrate
precipitated can be determined from a solubility in equilibrium with solution to another. As with
diagram. These problems require only a material the Na 2 S0 4 — 2 diagram, most of the data used
balance for their solution. here apply at 1 atm pressure. In addition to the
For crystallization problems involving energy 1-atm data, the vapor-liquid saturation lines for
balances, such as those in which evaporation occurs 0.5 atm and 0.2 atm are given. All other lines on
or in which the final temperature of an adiabatic the chart apply at 1 atm.
crystallizer is unknown enthalpy data as well as It may be instructive to follow on Figure
solubility data are needed. In such cases, the 19.29 the course of a process in which a mixture of
enthalpy-composition diagram is convenient. CaCI 2 *6H 2 and CaCI 2 *4H 2 having an overall
Figure 19.28 shows the enthalpy-composition mixture composition of 60 percent CaCI 2 40 per-
,
diagram for the Na 2 S0 4 — 2 system. This is a cent H 2 is heated under 1-atm total pressure. At
relatively simple diagram, and the principles low temperatures, the system consists of a mixture
involved in its construction and interpretation of CaCI 2 -6H 2 and CaCI 2 -4H 2 0. Use of the
should be familiar to the reader from a study of inverse lever-arm principle shows the mix to con-
Chapter 3. Within any field, the relative quantities tain about 90 weight percent CaCI 2 -4H 2 0. As the
are calculated from the familiar lever-arm ratio. The mixture is warmed, no phase change occurs until
CaCl2— 2 diagram shown in Figure 19.29 is the temperature reaches 86° F. At this point, solu-
much more complicated, and some explanation of tion of composition and enthalpy represented by
-100
-150 -150
40 50 60
Concentration, weight percent Na 2 S0 4
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
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
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
Fx F '
= C
Mh
+ F(1-x F ')
V-C h a
X' 4H 2 product from
is
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
'
—
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-
, )
is about the most concentrated that is recom- embryo. As the supersaturation increases, the addi-
mended by most manufacturers. tion more molecules to the embryo be-
of still
The material and energy balances may not comes more likely, the embryo grows and stabil-
always be made in as direct a way as is indicated izes, and a crystal nucleus grows. Study of the
above. There is some evidence that the assumption thermodynamics of surfaces shows that the solu-
of equilibrium in the interface inherent in the use bility of crystals decreases as the size increases.
of the energy balance is not accurate in the case of Thus, once a crystal nucleus is well formed, it
vacuum crystallizers. With these units, rates of tends to grow.
mass transfer from the liquid into the vapor are so Primary nucleation, which includes homo-
great that the "absolute vaporization rate" is geneous nucleation as described above, and nuclea-
approached (20). This phenomenon is somewhat tion on minute insoluble particles, called hetero-
analogous to the sonic velocity phenomenon in geneous nucleation, occur only to a minor extent
that increasing the driving force no longer propor- in a crystallizer magma. In these situations,
tionally increases the rate of transfer. As a result of secondary the precipitation onto
nucleation,
this phenomenon, and of the accumulation of crystal surfaces, accounts for most solids forma-
crystals at the interface, the pressure in the vapor tion. Additional crystals also form by attrition
space may be significantly less than would be pre- when the crystals break in the violent agitation of
dicted on the basis of the liquid phase temperature. the crystallizer magma. Secondary crystallization
For example, the products from a vacuum crystal- occurs through the formation of new nuclei at
lizer operating at 40° F may have to be pumped out surfaces and corners of "seed" crystals already
from a pressure that would correspond to a 35°F present in the magma and the breaking away of
liquid temperature. This behavior cannot now be these nuclei, and through the formation of nuclei
predicted without experimental data. in areas of low-energy content close to crystal sur-
faces or to surfaces of the crystallizer. Crystals may
Crystallization Mechanism also form via heterogeneous crystallization where
The mechanism by which crystallization occurs is foreign matter forms a site for nucleation and
of more than passing interest because it influences crystal growth. The situation is complex, but most
the conditions within the crystallizer and the of these mechanisms result in increased nucleation
properties of the product obtained. As would be in situations of high supersaturation and in concen-
expected, the deposition of a solid crystal can trated magmas (19).theories concluded
Earlier
occur only as the result of a concentration driving that nucleation could occur only at supersatura-
force from the bulk solution to the interface of the tions greater than some minimum value and that in
solid surface. Because the interface concentration the region of lower supersaturation no nucleation
must be the equilibrium concentration for solid could occur (13,14). This is an oversimplified view,
and solution coexistence — that is, because the even for homogeneous nucleation, and clearly is
interface concentration is that of saturated not realistic for practical crystallizer conditions.
solution — the concentration of the bulk fluid Crystal growth is at least as complex a process
must be greater than saturation. The extent of the as nucleation. The most probable of growth-
super-saturation depends upon the number and mechanism theories pictures crystal growth occur-
shape of crystals upon which precipitation occurs, ring at dislocations on the surface. The dislocation is
the temperature level, the solution concentration, self-perpetuating, and the crystal grows in a spiral
and the violence of the agitation present. as a result of surface forces. Screwlike surface
Crystallization begins with the nucleation irregularities have been found on photomicro-
mechanism by which a small crystal forms. In a graphs of crystal surfaces, thus substantiating this
homogeneous liquid, the process starts with a theory. Despite the complex growth mechanism, it
random association of solute molecules occurring is instructive to approach crystal growth using the
by normal molecular motion. In most cases, this classical rate-equation forms. Rate equations for
cluster dissassociates, again through normal crystal growth have been written using the two-
molecular activity. In some cases, additional molec- film theory. Separate equations were written for
ules join the cluster and it begins to assume a the diffusion of solute torn the edge of the
regular spacing of molecules and form a new phase. laminar layer to the crystal face (15) and for the
At this point, it is called an embryo. The embryo is turbulent transfer of solute from the body of the
also usually short-lived, redissolving easily because solution to the laminar interfaced). These equa-
the concentration gradient favors transfer from the 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
-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
<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
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 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)
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
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
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
+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
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 — ,
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
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.
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
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
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
Condensate inlet
Outlet
High-vacuum
ejector
Steam inlet
Water
inlet
Water-inlet for
conical
sprays
Condenser
Vapor inlet
Manhole
Handhole
Barometric leg
Hotwell
force the valvebody down against the seat. As long valve. Steam enters the trap, forcing the discharge
as steam is present the valve remains closed, but a check valve to open and thus emptying the trap.
small leak blows through the valve passage. When the level drops to a preset point, the float
The orifice trap is small and light It will operate returns the condensate valve to the open position
under a wide range of pressures and handle non- and closes the live-steam valve.
condensables as well as steam. It is, however, af- Another, and even more common, solution to
fected by corrosion and by clogging of the orifices the problem is through a barometric condenser.
with dirt or scale. Here, a surface, or more commonly
a jet, con-
Other types of steam traps that operate on denser is on
top of a long tail pipe.
located
energy balance principles are available, but they are Condensate stands in the tail pipe high enough so
not as common as the types discussed above. that the hydrostatic head balances the pressure
difference between the vacuum space and the
Condensate Discharging Methods atmosphere. Figure 19.41 shows two types of baro-
In multiple-effect evaporators and in vacuum metric condensers. In the two-stage barometric
crystallizers,condensate must be discharged from a condenser, vapors are contacted countercurrently
vacuum space to the atmosphere. The discharging with a falling-water shower. Noncondensables are
is often done with surface condensers followed by handled in a jet ejector after each stage of the
a pressure return trap in conjunction with check condenser. In the jet barometric condenser, vapors
valves. Figure 19.40 is a schematic diagram of this are entrained into a high-speed water jet. This jet
kind of trap. Condensate enters the trap but is acts as both steam condenser and entrainer of
prevented from leaving because the trap is at sub- noncondensables. The jet condenser eliminates the
atmospheric pressure. As the liquid level rises in steam use of a steam-jet ejector with a barometric
the trap, a ball float actuates two valves, closing condenser, but operates best at pressure greater
the condensate-feed valve and opening a live-steam than 2 in. Hg. The water required by a barometric
Cleaned
air
Air plus
entrained
liquid
Separated liquid
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.
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
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.
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)
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
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-
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
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
541
542 APPLICATIONS TO EQUIPMENT DESIGN
©
The or
1. fluid flowing into the system is
-.0.9
3
/
^ :s
/
/
5 07
/
/
- :f /
/
/
15
0.4
::-" 4 ::~
10 1C"
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
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
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)
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.
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, />,
English engineering system, these dimensions yield of an open elevated tank 25 ft above the reservoir
units of ft lbf/lb. In this system of dimensions with surface through a 3-in. I.D. pipe. The inlet to the
its definition of the pound force (lb^) and the
pump is located 10 ft below the water surface, and
pound mass (lb), there is a temptation to cancel the water level in the tank is constant at 160 ft
the two, thereby producing units of simply "feet" above the reservoir surface. The pump delivers
for each term of Equation 20.18. Each term is then 150gal/min. If the total loss of energy due to
referred to as a "head," expressed in feet of fluid 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-
,
Reservoir
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 may be placed anywhere, several different sets will
similar symbols are traditionally used to represent be used. The only requirement is that all the
them. They cannot be canceled with each other. energy between the points must be accounted for.
The power supplied to the fluid in a flow (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
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)
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
(2)(32.2)(a)
:
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 ft-lb,/lb
schedule number « 1000 —
s
where
Again, tne same theoretical work requirement
results. Other reference points could be used, and P= internal working pressure, force/area
exactly the same result would be found, providing s = allowable stress, force/area
all energies at each reference point are considered.
Ten schedule numbers are in use: 10, 20, 30, 40,
It may be concluded that although any set of
60, 80, 100, 120, 140, and 160. The pipe-wall
reference points may be used, wise selection of
thickness increases with the schedule number. For
reference points can simplify calculations.
steel pipe, schedule 40 is the number corre-
sponding to "standard" pipe. In Appendix C-6 are
PIPE, TUBING, AND FITTINGS
tabulated the dimensions and other useful data for
Before considering the 2£ term in Equation 20.14 schedule 40 and 80 threaded steel pipe. For other
in more detail, it is important to give some thought schedule numbers and other types of pipe, Refer-
to the means of containing the fluid in a given flow ence 8 is recommended.
system. Some demarcation between pipe and tubing
Chemical-process applications frequently 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
Nipples Bushing
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
(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
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
.
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
(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.
/•/»,
= v2 = Thus, the
So
remain constant,
liquid levels v-\ 0.
energy balance becomes 2F for the 6-in. pipe = 2F 6in .
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
/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
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
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.
= v2 = (20.24)
v-\ ?
Wf = ' l 2 fv
2
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
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
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
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
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
) )
Manometers Pi P-2
manometer orientation, presence of intermediate wire changes its electrical resistance, which can be
fluids, or any other variation in manometer design. measured and related to the pressure.
Note that the manometer reading is multiplied by
the difference in density of the two fluids to get The General Meter Equation
the pressure drop. Consider a flow system containing a fluid meter
As indicated above, variations of the man- about which an energy balance is to be made. This
ometer shown in Figure 20.8 may be made. It may meter produces a change in the kinetic energy of
be inverted so as to use a metering fluid of lighter the flowing fluid that can be sensed as a pressure
density than the flowing fluid. For small readings, difference between a point 1 just upstream of the
the tube may be inclined at some angle to the meter and a point 2 at the meter. Points 1 and 2
horizontal so as to magnify the linear displacement represent locations about which an energy balance
of fluid. Another technique used to magnify small 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)
Fluid C fT =
Fluid C
The bracketed term within the square root relative to the orifice plate is a function of the
represents a decrease attributable to
pressure fluid velocity as well as the relative diameters of
changes in kinetic energy between points 1 and 2. the orifice and the duct. The position of the
The friction term will include losses due to form downstream pressure tap should approximate the
friction and skin friction, as has been shown in vena contracta to ensure a maximum reading on
Equation 20.21. The manometer connecting points the pressure-difference indicator.
1 and 2 will give a reading that is indicative of the The application of Equation 20.40 to the
pressure difference between these two points. This orifice permits calculation of the fluid
plate
pressure difference is produced by two effects: the velocity and hence of the mass-flow rate through
change in kinetic energy caused by a change in the duct. Some modification of this equation is
velocity because of a change in flow area, and necessary to make it more useful. Consider again
permanent pressure loss due to skin and form the bracketed term of Equation 20.40. The term
friction. The manometer indicates only the total (—AP/p) represents the total pressure difference
pressure difference between these points and does between points 1 and 2, and the SFterm includes
not differentiate between the two effects. all between the same two points. The
fluid friction
Equation 20.40 is referred to as the general meter difference between these terms represents the
equation. It could be written in terms of v2 or in kinetic-energy change between points 1 and 2 as
terms of the mass rate of flow by the appropriate indicated by Equation 20.38. This difference can
use of the continuity equation. The general therefore be expressed as a fraction of the total
equation will now be applied to specific types of pressure difference (— AP), or
flow meters.
AP\
The Orifice Meter (20.41
P
An orifice meter is an extremely simple apparatus;
it normally consists of a flat plate with a centrally where C^ 2
is a proportionality factor always ess ...
i/! =da
M=r) (20.44)
%=0
- 1
H=f) = Cia
*«ra
-1
C2 S
(20.45)
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.
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
0.8 1.0
:• ;e Jiarrerer
Ratio.
t
pipe diameter
Orifice Reynolds number:
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
=
-AP 161
Example 20.6. A lube oil is flowing through a /? = 0.203 ft
.
THE ENERGY BALANCE IN FLOW SYSTEMS 563
The Venturi Meter tubes arranged parallel to flow. The outer tube is
A properly designed venturi meter (Figure 20.13) perforated with small holes perpendicular to the
is one in which form friction is minimized. The
flow direction that lead into the annular space. The
streamlined shape of the meter virtually eliminates annular space is otherwise sealed except for a
boundary-layer separation so that form drag is manometer lead. The inner tube has a small
negligible. The converging cone is about 25—30 opening pointed into the flow. This tube is
degrees, and the diverging cone should not exceed connected to the other side of the manometer. At
7 degrees. Venturi meters are difficult and expen- equilibrium there is no fluid motion within the
sive to manufacture and in large sizes are very 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 Figure 20.14 illustrates a pitot tube. An
is replaced by C v which has a value of about 0.98
,
energy balance may be written between points 1
for commonly encountered conditions. A venturi and 2. At point 2, the fluid velocity is zero so the
meter also measures an average velocity. energy balance becomes
Flow
("i)' = 2a
ra- SF (20.48)
Flow duct
i' 9 =
(D "i
Flow
^\P"
Static pressure
^
openings
Sectional view
of tip
of pitot tube
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
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
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 )
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
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
Electrode Figure 20.22. Vortex flow meters. Two sizes are shown.
Electrode Vortices are shed from the nonstreamlined vertical
3. Production of a slurry of a finely divided stabilizers at the tip is shown. This impeller allows
-
THE ENERGY BALANCE LN FLOW SYSTEMS S71
-ZF=W f (20.60)
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.
Baffle t- -
'
f Agitator
(b)
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
Marine propeller
3 2.7- -3.9 0.75- -1.3 4 0.10 5
3 blades, pitch = D'
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
Turbine 1
Ch1 D
6 blades
Flat paddle
2 blades
10.000
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)
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
(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/)
-
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
1.0ft-lb/7lb.
tings) of drawn steel tubing (30 mm O.D., 2.6 mm wall
a nuclear reactor to study corrosion problems. The equip- 20.11. A copper smelter is located in a small city
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
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?
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
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
580
FLUID MOTIVE DEVICES 581
Air
»^~ chamber
Driving
gear
Cylinder
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
^
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.)
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
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)
-Volute
Drive shaft
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
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
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 ,
— + 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
(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
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
D<"i_5* Si
;.APHRAGM
CONNECTING
SHAFT DISCHARGE
MANIFOLD
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
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 expressed terms of the height of a column of
in
engineer should consult with the manufacturer, fluid it support - the head. For example, an
will
perhaps asking for a recommendation (and guaran- oil of 200 cp viscosity reduces pressure and ca-
tee) of apump for the application. pacity about 10 percent and increases power by
Whether the design engineer selects the pump about 25 percent relative to water. If high-viscosity
or asks for a manufacturer's recommendation, liquids are to be pumped, appropriate performance
detailed information on the capacity and head data should be obtained from the manufacturer.
required and on the properties of the liquid to be Corrections for high viscosity are available in
pumped must be immediately at hand. A general Reference 3.
consideration of these factors is given below. If the liquid contains solids in suspension, as
Capacity. The flow rate required by the in sewage, paper stock, slurries, and some foods,
process itself determines the capacity of the pump special pump designs may be necessary. For
needed. Not only steady-state flow, but also example, one of the major problems encountered
expected surges must be considered. Some safety in the development of the artificial heart (which is
factor for uncertainties in design should be in- a small, specialized pump) is the destruction of red
cluded. Possible increase in flow rate because of blood corpuscles, The corpuscles are suspended in
process expansion might be taken into account. the liquid plasma, and many conventional pump
Care should be taken not to overestimate needed designs lead to corpuscle attrition; hence, special,
capacity by putting in too large a safety factor. 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
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.
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
1750 rpm
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 necessary to use a throttling valve in the line to use
by checking the specific speed. No pump speed is up this extra head. This is, of course, wasteful of
given, so a reasonable value of 1800 rpm will be energy.
used.
COMPRESSING GASES
1800-
/V5 = = 1288 Gases are frequently involved in chemical pro-
0.75
(70) cesses; they must be moved through pipelines and
Examination of Figure 21.19 shows this specific process equipment using appropriate fluid motive
speed to be at the upper end of the range for the devices, just as process liquids must be pumped.
standard centrifugal pumps, so the catalog of a Fans, blowers, and compressors are used to impel
manufacturer of this type of pump is consulted gases. Fans produce a small pressure increase (up to
(Figure 21.23). The actual speed of 1750 for 0.4 psi); blowers produce a larger increase (up to
these pumps would lower the calculated specific 4.0 psi), and compressors produce the largest in-
speed somewhat. Examination of the figure shows crease (from 4.0 psi up to 60,000 psi in routine
that size 3x4—10 can do the job. This is a pump applications). The distinction among these three
with a 3-in. discharge line, a 4-in. suction line, and terms not always clear, nor should the pressures
is
a maximum impeller size of 10 in. Figure 21.17a is given above be viewed as rigid boundaries among
now consulted to choose the proper impeller size. the three types of equipment. Fans are sometimes
The required head and capacity give a point that called blowers, and vice versa. Higher-pressure
falls slightly above the curve for the 9-in. impeller; blowers are frequently called compressors. In this
hence, that size could not quite handle the job. book, the term compressing gases refer to the
The 10-in. impeller is selected, even though it can action of any gas motive device, regardless of the
deliver a head of 90 ft at 300 gal/min. It may be pressure increase involved.
FLUID MOTIVE DEVICES 597
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
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.
£ c
i S" 2
hp
^
^J*
^>
-
cylinder through valves set to be actuated
pressure difference between cylinder contents and
when the
(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
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 21.31. 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
Inlet
Intermediate- Discharge
pressure
impeller
(a)
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
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
Figure 21.35. Operating characteristics for centrifugal and axial compressors, (a)
INLET
PORT
ROTATING
LIQUID
COMPRESSANT
w
C\ THIS SECTOR,
IN
COMPRESSED GAS DISCHARGE
ESCAPES AT DISCHARGE PORTS CONNECTIONS
KEY
ROTOR-one moving part
High-pressure
steam inlet
2nd stage
Discharge
High-pressure
steam inlet
1st stage
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
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
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).
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.
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.
PROBLEMS
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,
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)
(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
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.
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
(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
15 ft
20 ft
6ift
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
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
APPLICATIONS OF THE MECHANICS OF PARTICLE where F is the resultant force acting on any body,
MOVEMENT THROUGH A FLUID dvldd is the acceleration of the body, and is the m
mass of the body.
General Principles In Figure 22.2, the forces acting on the falling
In Chapter 13 the concept of form drag was intro- body are the external force (FE ), a buoyant force
duced. For steady flow of a fluid past a solid, {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
c "^T
611 \. \
il i/)
8 o
V E
QO __ y- Slope = -2
Q .
nj
Q0
= °
<J -
14
1.0
CD v fs 2 pS
Fb)9c =m ~^ (22.2)
Fd9c = « ! 13.40)
dd
m — = ™/l
2 dv
dv CD v pS — i p<
ml --\g-3iTD
dd -K) 2m
(22.6)
Since m=
dd
p pv
\
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-
.
—
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
.
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 )
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
= 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
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
'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
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
/V- Re /i (9X0.001
DP = = 0.000225 m
logCD = -2logA/ Re + vp (4.0X1000)
sizes
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-
Hydraulic water
velocity while a solid mixture, whose separation is
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 mixed when the stroke is completed, the rakes are lifted
and then flow through the annulus whose cross- and returned to the starting position. The slurry is
sectional area varies. Classification action occurs in thus kept in continuous agitation. The time of the
this annular space with the small particles leaving raking stroke is adjusted so the heavy particles have
with the liquid and the larger particles settling to time to settle while the fines remain near the
the bottom for removal. surface of the slurry in the rake compartment.
Another piece of equipment for solid separa- Thus, the heavy material is moved upward along
tion whose mechanism is based upon settling the floor of the tank and removed at the top of the
velocity is the rake classifier (Figure 22.8). The apparatus as a dense slurry.
rake classifier is a tank with an inclined bottom in For classifying fines material, a bowl classifier
which are provided movable rakes. Feed is intro- (Figure 22.9) is used. It consists of a shallow
duced near the middle of the tank. The lower end cylinder with a gentle conical bottom. The feed
of the tank has a weir overflow over which the slurry is fed at the center near the surface, and the
fines that have not settled leave in dilute suspen- 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
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 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,
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
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
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
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
"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
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
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
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 )
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
')
2 Values, ft
2
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*
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.
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
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;
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)
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 22.22. Batch-settling results. of the layer having the lowest capacity for the
632 APPLICATIONS TO EQUIPMENT DESIGN
1
VL
W///////,
1k the clear-liquid interface.
the limiting layer
the interface is
is
If
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)
dv Zi = zL +6 L vL [22.56)
vL = c—
dc
-v (22.50)
vL =cf(c)-f(c) (22.51)
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
c L Z; =c z (22.57)
40
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
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
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 .
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
Concentration, g/l
and solving for {d u — 6C )
Figure 22.28. Solution to Example 22.6.
=
A{zc -zu \
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 ,
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
= 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.
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
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.
D p vs p
= =
f
v
2
pL^-e)A p Mm
<
*\{1-e)ATT~)
\(1 - e)A p pf
(2Z71)
which rearranges into
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)
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
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-
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)
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,
= 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
150 —^
(1 -e) pv,
+ 1.75^— (p,-pfc
(22.90)
/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 expansion depends upon the distributor design and
particles above the bed. Here, the bed has many of
bed diameter, and hence is difficult to predict on
the characteristics of a liquid with the fluid phase
the basis of laboratory or pilot results (29).
acting as a gas bubbling through it. This is called Equation 22.92 gives an estimate of this expansion
aggregative fluidization. In addition to these two for particulate fluidization in a bed where wall
main classifications, fluidized beds show consider- effects and recirculation are absent. In designing a
able bypassing and slugging — that is, intermittent fluidized bed, the engineer must allow not only for
and unstable flow of large gas bubbles through the this expansion but also for the height above the
bed. Both these behavior patterns are strongly top of the bed required to get adequate solid-fluid
affected by the design of the distributor that holds separation, the transport disengaging height (TDH).
up the packed bed and distributes gas flow into it. This also depends upon the characteristics of
The presence of particulate or aggregative the fluid-solid system and has not yet been
fluidization must be the result of the influence of adequately correlated. Figure 22.32 gives an
gravitational forces on the phases present in a empirical correlation for estimating TDH for fluid
fluidized bed as well as the bulk fluid mechanics of catalytic cracking catalyst (64) and shows typical
the system. Thus, the Froude number (v2 IDp g), TDH values. At heights above TDH, there will still
which is the ratio of kinetic to gravitational energy, be catalyst where there is particle
losses, especially
would be expected to be one of the criteria. Orig-
inally, the Froude number was taken as the entire 40 ^ 1 !N^ 1
indicated that if
10
_
^>$% \V^ ^t)^^
^Ss,
s.
l uq
OvW.22
= 1.53
92
cm/sec -
:
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
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
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 *
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 =0.190 the average of the together by static electricity. Such agglomerates
e at vsm ft/s,
have been observed and seem to be particularly
previously determined velocities.
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
L_ 1-e (
0.58
= 4.00
1-e 0.145 5 loo
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. Dense
Phase Slugging Flow. Here the
pressure dropalmost entirely a result of the
is
/ / 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.)
2fs Lp s vs 2 D pe
ArL/D = APfg + AP fs = fL v2 /2g c D + -e
1 -e 1
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
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
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
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
— 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
P
Fluid-Solid Conveying, Slurry Transport. In
addition to the pneumatic transport described It has been well known that the constants in
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
Equation Regime
Designation (m) a b c d e
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
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
Boundary
Equation Between
Designation Regions a b' c d'
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
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
(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