Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Distillation-How to Push a Tower to Its

Maximum Capacity
Simon X. Xu, Charles Winfield, John D. Bowman Tru-Tec Services, Inc.; Shelley, Suzanne . Chemical
Engineering ; New York  Vol. 105, Iss. 8,  (August, 1998): 100.

ProQuest document link

ABSTRACT
 
Final design of distillation equipment typically includes a certain degree of overdesign to compensate for
uncertainties in the parameters of the design model, and to assure operation over a wide range of operating
conditions. One of the more challenging engineering projects is to push a distillation tower to its maximum
hydraulic capacity - just to the brink of incipient flooding without exceeding that point. A generalized procedure for
testing a distillation tower in order to push it to its maximum capacity is discussed.

FULL TEXT
 
Final design of distillation equipment typically includes a certain degree of overdesign to compensate for
uncertainties in the parameters of the design model, and to assure operation over a wide range of operating
conditions. One of the more challenging engineering projects is to push a distillation tower to its maximum
hydraulic capacity -- just to the brink of incipient flooding without exceeding that point.
At the incipient-flood point, liquid starts to accumulate inside the tower and operating stability begins to decline.
Flooding results from a hydraulic imbalance between liquid and vapor flows, typically when the flowrate of liquid or
vapor exceeds the physical limit of the trays or packings. This is most often due to a design oversight, operating
excess, tray or packing whose integrity has been compromised by fouling, corrosion or damage.
Maximizing throughput capacity
The maximum capacity of a distillation tower is determined by its diameter, by the design of its internal elements
(trays, packings and other elements), and by other system properties. Distillation design involves defining the
tower diameter and the type and size of all trays, packings, distributors, collectors and other internals. With an
existing tower, the diameter, internals, and many of its auxiliary systems are fixed.
System properties -- which include surface tension, densities and viscosities of the vapor and liquid phases -- will
also be constant at specified temperature, pressure and product specifications, since the properties are
determined by temperature, pressure and stream compositions. Therefore, correlations between vapor and liquid
at the maximum capacity of an existing tower can be derived from various correlations of hydraulic limit
parameters. Flooding vapor velocity for sieve trays can be calculated via Fair [1]:
u subscript f = C subscript sb(sigma/20)0.2[(rhoL-rhoV) /rhoV]0.5
log C subscript sb = a + b log F subscript lv + c(log F subscript lv)2
F subscript lv = (L/V) (rhoV /rhoL)0.5
where:
u subscript f = Flooding velocity, m/s
C subscript sb= Capacity factor, m/s
F subscript lv= Flow parameter, unitless
sigma = Surface tension, dynes/cm

PDF GENERATED BY SEARCH.PROQUEST.COM Page 1 of 9


L = Liquid mass flowrate, kg/h
V = Vapor mass flowrate, kg/h
rhoV = Vapor density, kg/m superscript 3
rhoL = Liquid density, kg/m superscript 3
a, b and c = Correlation constants [2]
Assuming that tray sizes and system properties are fixed, one can derive an equation that relates vapor and liquid
rates at the incipient flood point:
V subscript max, subscript flood = F subscript flood(L)
Here F subscript flood(L) is a function of liquid rate, derived from the Fair flooding correlation. Limits to other
hydraulic parameters, including entrainment rate, downcomer backup flooding, and liquid residence time in
downcomers, can also be used to derive the equations to define maximum capacity. Figure 1 shows the capacity
boundaries for a typical sieve tray, with the coordinates expressed in volumetric flowrate.
The minimum (or tightest) value of all boundaries defines the actual maximum capacity curve (shown in Figure 1,
after correction with safety factors). A similar diagram can also be developed for other trays or a packed bed [3].
It should be noted that the hydraulic parameters used to define the boundaries depend on both the internal sizes
and system properties. Even for a tower with identical internals, property differences (e.g., density of the vapor
phase) between the top and bottom may affect the boundary significantly, because of changes in temperature,
pressure and stream composition.
Figure 2 shows the maximum capacity boundaries for top and bottom trays of a tower with identical sieve trays.
Downcomer capacity of the bottom is significantly less than that of the top, since density differences between
liquid and vapor are smaller at the bottom than the top; hence vapor-liquid disengagement in downcomers
becomes difficult toward the tower bottom.
In order to evaluate whether the existing tower throughput is close to its maximum capacity, the operating point --
which denotes the actual operating vapor and liquid rates on a tray -- can be marked on the same diagram as the
maximum capacity boundary (Figure 1). Various positions of the operating point relative to the capacity boundary
show how close to maximum capacity the existing loads are for this tray.
In theory, each tower tray should have its own capacity boundary and operating point, due to variations of
temperature, pressure, composition and flow rates. However, in practice, it is sufficient to analyze the operations
of key sections only. Taking a one-feed, two-product distillation tower with identical sieve trays as an example, the
``key'' trays would be the top and bottom ones, and any that are directly above and below the feed point. Table 1
lists the key sections for most towers.
Gauging key section behavior
Analysis of the capacity boundary and operating point for each key section will reveal which is closest to a
capacity boundary. However, making changes to the section closest to a capacity boundary may not necessarily
debottleneck the tower, because movement of the operating point will depend on how the tower is pushed to its
max capacity.
For a tray inside a distillation tower, vapor and liquid flowrates can increase independently or proportionally. In
other words, the operating point may move toward the boundary vertically, horizontally, or diagonally (Figure 3),
depending on adjustments made to operating variables and tray locations.
It is key to understand the movement tendencies of operating points near key sections. Quantitative analysis of
operating-point movement for various sections will require computer simulation, since predicting vapor-liquid
internal traffic and properties involves equations of heat balance, mass balance, vapor-liquid equilibria, hydraulics,
and mass- and heat-transfer models.
Fortunately, for most towers in the chemical process industries, it is possible to qualitatively analyze the behaviors
of the key sections, using basic rules. For a trayed distillation tower, the following general rules apply:
Feed temperature. The higher the feed temperature, the higher the vapor rate on the tray above the feed point and
the lower the liquid rate on the tray below the feed point.

PDF GENERATED BY SEARCH.PROQUEST.COM Page 2 of 9


Feed composition. The greater the percentage of light components in the feed, the higher the vapor load on the
next tray above, and the lower the liquid load on the tray below the feed point.
Reflux temperature. The lower the reflux temperature, the heavier the vapor and liquid loads on the top tray.
Reflux rate. The higher the reflux rate, the heavier the vapor and liquid loads on the top reflux tray.
Boil-up rate. The higher the boil-up rate, the heavier the vapor load on the tray above the reboiler return nozzle.
Operating pressure. The higher the operating pressure, the lower the vapor volume rates on all trays in the tower.
Keep in mind that all variables are not independent of one another. For instance, heat balance will require a higher
boil-up rate for a higher reflux rate. Therefore a higher reflux rate will increase both vapor and liquid traffic on all
the trays in the tower.
It should also be stressed that tower throughput is often limited by product specifications (or by the separation
efficiency of the tower). Poor disengagement between vapor and liquid phases reduces mass-transfer efficiency as
the loads increase. The relative volatility between components affects the separation, as well.
An adjustment of some operating variables can improve vapor-liquid disengagement and increase efficiency by
reducing the internal loads. However the total tower capacity will not necessarily increase, because variables such
as operating pressure and reflux ratio have an opposite effect on the relative volatility or phase-equilibrium.
The relationship between operating pressure and minimum reflux ratio is shown in Figure 4 for a propylene-
propane splitter. It is clear that the higher the operating pressure, the more reflux is necessary to produce the same
separation. In this case, reducing vapor volumetric flowrates (by increasing pressure) can easily be offset by the
increased reflux ratio (or internal vapor-liquid traffic).
For most towers operating at or above atmospheric pressure, it is not practical to expect a significant reduction of
vapor load by raising pressure. However, in vacuum towers, vapor volumetric flowrates can be significantly
decreased by increasing operating pressure. For a tower with a top pressure of 10 mm Hg, an increase of 3 mm Hg
will reduce vapor volumetric flowrate by 23%, based on the Ideal Gas Law:
(V-V')/V = (1/P-1/P')/(1/P) = (1/10-1/13)/(1/10) = 23%
where:
V = Vapor flowrate at the lower pressure, m superscript 3/h
V' = Vapor flowrate at the higher pressure, m superscript 3/h
P = Lower pressure, mm Hg
P' = Higher pressure, mm Hg
In general, interactions between pressure and reflux ratio, and other factors, such as thermal degradation of
bottoms at higher pressure and temperature, should be considered before adjusting operations.
Achieving maximum capacity
As discussed, internal vapor-liquid traffic can be varied by adjusting operating conditions. In the real world, the
range of operating adjustments are often defined by other practical limits. For example, operating units that are
either upstream or downstream of the process may not be able to tolerate pressure or temperature changes. The
operative questions should be ``How can we increase this feed to its maximum rate?'' or ``Where is the flood
origination point, and what is its root cause?''
Conducting a baseline gamma scan for a tower (Box, p. 104) with known integrity at design rates is always the first
step (Box, above). This is performed to document a column's online performance under a given set of test
conditions. Quite commonly, baseline scans are used to get a snapshot of tray or packed-bed loadings during
startup and commissioning of a revamp or design modification.
Standard industry practice is to conduct a baseline scan when a column is known to be operating well, to establish
benchmarks that will be used as standards to compare against, should the column have problems later. It is highly
recommended that columns that are susceptible to fouling or tray damage be ``baselined'' shortly after a startup --
before questions concerning mechanical integrity have an opportunity to develop.
Such a scan is particularly helpful in applications where nozzles, electrical conduits, or other sources of external
interference can obscure scan results. Referring to baseline scans for ``before'' and ``after'' comparisons

PDF GENERATED BY SEARCH.PROQUEST.COM Page 3 of 9


eliminates ambiguity and unnecessary inference.
Conducting a baseline scan can reveal if a tower was over-designed or is being under-utilized. And, a complete set
of operating data documenting the column's performance while the scan was in progress is invaluable for
subsequent tests and analysis.
In general, a minimum evaluation package should include the following:
-- A clearly defined objective
-- A process flowsheet
-- A tower vessel drawing
-- Condenser and reboiler information
-- Temperature and pressure data for the top of the tower
-- Temperature, pressure, composition and flowrate of all streams entering and leaving the tower
-- Product specifications
-- Type and size of trays and packings
More data will required if the system is complex. Any system having multiple feed points or product draws -- such
as a crude distillation tower that has side strippers and pump-arounds -- can be construed as complex.
Testing procedure
A generalized procedure for testing distillation towers is discussed below. Similar methods apply to absorption or
stripping towers.
1. Define adjustable ranges (or allowances) of operation variables. These include condenser reflux rate, reboiler
boil-up rate, upstream and downstream pressure requirements, preheater feed temperature, and heat transfer
duties of the intercooler, interheater or pumparounds.
2. Clearly define the overall objectives for optimization. The objective should be attainable in terms of material
balance, heat balance and phase equilibrium. For towers with multi-feeds or side-draws (for instance, crude
towers), the internal vapor-liquid traffic will vary significantly with sections and ratios between feeds and draws. It
is important to specify which feeds and side-draws are to be increased, and to be able to anticipate the tower's
response, before the tests.
3. Estimate optimum operation variables and possible bottlenecks. While a rigorous computer simulation can be
very helpful for many existing towers, possible performance responses can be predicted using distillation basics
and operation experience. For instance, vapor volumetric traffic will be reduced by increasing operating pressure,
which will reduce entrainment. At the same time, relative volatility decreases make separation more difficult due to
the increase in pressure. Therefore an optimum pressure may exist within the operable pressure range. For a tower
that has not suffered any internal damage, most of the performance bottlenecks will be located at the key sections
listed in Table 1.
4. Identify the incipient flooding point. A normal baseline test and scan should always be the first step for
gathering a complete set of data. All tests should be performed according to a well-defined plan. An approximate
flooding point can be found by increasing feedrates and adjusting other variables -- reflux, boil-up and so on --
proportionally in large steps. Several scans may be needed to define the location of the incipient flood point.
5. Verify a stable maximized operation. After the flooding point has been determined, the tower should be slowly
unloaded to roughly 5% below the flooded condition. Then one final test and scan should be done to verify that the
tower is completely operable, and to determine steady-state values of the operating conditions.
Figure 5 shows a light-hydrocarbon absorption process where propane and heavier feed components are absorbed
by a lean oil, and lighter components leave the top as dry gas. The 80-tray tower is 12 ft in dia. Trays 1-34 are
identical 4-pass sieves; Trays 35-45 and 47-80 are identical 2-pass sieves; Tray 46 is a chimney tray for liquid draw
and return of an intercooler.
Feed enters a flash drum, and then the separate liquid and vapor streams enter the absorber above Tray 34. Lean
oil is mixed with overhead condensate and sent to the top tray (Tray 80). Overhead light gas leaves the system
from the reflux drum. Bottoms-rich oil feeds a neighboring stripper. The stripout vapor then returns to the absorber

PDF GENERATED BY SEARCH.PROQUEST.COM Page 4 of 9


as vapor inlet below Tray 1. The intercooler removes some of the absorption heat at Tray 46.
In this example, plant management required a higher vapor throughput to the absorber. Although the tower is
operating as per its specifications, vapor throughput is restricted by flooding at 85% of the original design
capacity.
The objective is to define the location of the absorption tower's hydraulic limit. Increasing the vapor feed at a
constant reflux rate will mainly affect tower operation above Tray 34, by increasing vapor traffic on Trays 35 to 80.
The key trays that are affected in this situation are: Tray 35, Trays 45, 46, 47, which have tray-type transitions and
flowrate changes, and Tray 80.
A process simulation showed that the section between feed and intercooler (Trays 35-45) was more heavily loaded
than the section between intercooler and the top (Trays 46-80). In other words, Trays 35-45 appeared to be the
bottleneck when vapor feed was increased. A test run was performed first at normal feedrates. The tower was
scanned completely from top to bottom through the tray active areas, to verify tray integrity and establish a
baseline condition.
Note that the scan performed at normal rates (shown as a the yellow dashed line on Figure 6) indicated that all 80
trays were in place with no indication of damage. Establishing mechanical integrity is an absolute requirement
before further progress toward optimizing performance can be achieved. It is impossible to accurately assess and
predict the behavior of a tower with damaged internals.
At a higher vapor feedrate, the second scan was performed for the affected section (Trays 35-80), with the results
(solid solid line) overlaid on the normal scan plot (Figure 6). This scan shows that Trays 45-62 were flooded at high
rates -- not Trays 35-45 as the simulation predicted. The flood initiated from the active area of Tray 45 (just below
the chimney tray) and propagated up to Tray 62. The tower did not flood completely during the scan.
The incipient flood location (Tray 45) can be used further to test the vapor feedrate at the incipient flood point by a
stationary monitoring test [4]. Tray 45 is identical to all the trays below it, but is adjacent to the intercooler
chimney tray. Based on simulation and scan results, one could concentrate tower-debottlenecking efforts on
altering the designs of the chimney tray, transition configuration, and intercooler nozzles.
Given today's market conditions, most manufacturers are finding it economical to maximize throughput by
optimizing the performance of existing distillation equipment through various methods that do not require
significant shutdown or capital investment. To do so, one must accurately discern hydraulic limits and locate the
precise bottleneck locations within the column. Gamma scan technology can be easily used to detect the incipient
flood point on an operating distillation tower, and this approach provides reliable data for safely pushing a column
to its maximum capacity.
References
1. Fair, J. R., Section 14: Gas Absorption and Gas-Liquid System Design, ``Perry's Chemical Engineers' Handbook,''
7th ed., McGraw-Hill, New York, 1997, pp. 14-27.
References
2. Kessler, D.P., and Wankat, P.C., Chem. Eng., September 26, 1988, pp. 72-74.
References
3. Bravo, J.L., Chem. Eng. Prog., Volume 7, p. 36, 1997.
References
4. Bowman, J.D., Chem. Eng. Prog., Volume 2, p. 25, 1991.
AuthorAffiliation
Simon X. Xu (Xu Xiaomin) is the principal engineer for Tru-Tec Services, Inc. (11005 West Fairmont Parkway,
LaPorte, TX 77571; Phone: 281-471-8715; Fax: 281-471-8607; E-mail: xus@kochind.com). He has authored more
than 30 articles on various aspects of distillation, including process simulation, hydraulics, non-destructive testing
and troubleshooting. A member of AIChE, Xu earned B.S. and Ph.D. degrees in petroleum processing and organic
industry from the Petroleum University (Beijing). Previously, he was the director of the Separation Engineering
Center for the Petroleum University of Beijing.

PDF GENERATED BY SEARCH.PROQUEST.COM Page 5 of 9


Charles Winfield is president of Tru-Tec Services, Inc. (11005 West Fairmont Parkway, LaPorte, TX 77571; Phone:
281-471-8715; Fax: 281-471-8607; E-mail: winfieldc@kochind.com). He was co-founder of Tru-Tec, Inc., a company
specializing in industrial problem solving using an array of non-destructive evaluation techniques. Trained in
nuclear technology, Winfield graduated from Texas State Technical Inst., and studied health physics at the Nuclear
Science Center Research Reactor at Texas A&M University.
John D. Bowman is vice-president of sales and marketing for Tru-Tec Services, Inc. (11005 West Fairmont
Parkway, LaPorte, TX 77571; Phone: 281-471-8715; Fax: 281-471-8607; E-mail: bowman0j@kochind. com). He is
editor of Tru-News, Tru-Tec's quarterly publication, which focuses on process diagnostic applications and how
they are used to enhance performance or increase reliability in CPI operations. Bowman has authored or co-
authored 12 technical papers. He has been with Tru-Tec for over 10 years, in a range of engineering capacities.
Previously, he worked for Bently Nevada Corp. and Phillips Petroleum Co. He holds a B.S. in mechanical
engineering from Kansas State University.
Illustration
Illustration: Graph: FIGURE 1 (left). Separation capacity boundaries for a typical sieve tray can be shown
graphically, with the coordinates expressed in volumetric flowrate
Illustration
Illustration: Graph: FIGURE 2 (right). In a tower with identical sieve trays, the downcomer capacity of the bottom
tray is much lower than that of the top tray, since the density difference between liquid and vapor is smaller at the
bottom. As a result, vapor-liquid disengagement becomes more difficult toward the bottom
JAEGER
Illustration
Illustration: Graph: FIGURE 3. For a given distillation tray, vapor and liquid flowrates can increase either
independently or proportionally. Thus, the operating point may move toward the boundary vertically, horizontally,
or diagonally, depending on adjustments made to the operating variables and tray locations
Illustration
Illustration: Graph: FIGURE 4. In this propylene-propane splitter, the higher the operating pressure, the more reflux
is necessary to produce the same separation. In this case, reducing vapor volumetric flowrates (by increasing
pressure) can easily be offset by the increased reflux ratio (or internal vapor-liquid traffic)
Illustration
Illustration: Diagram: FIGURE 5. Simulation identified the bottleneck in this light-hydrocarbon absorber, allowing
vapor throughput to be raised
Illustration
Illustration: Graph: FIGURE 6. For the light-hydrocarbon absorption shown in Figure 5, a test run was first
performed at the normal feed rate, then at a higher vapor feed rate. The results of both scans for the affected
section (Trays 35-80) are shown here. The scan for the higher vapor feedrate (shown with a solid orange line)
shows that Trays 45-62 were flooded at the higher rate -- not Trays 35-45 as the simulation had initially predicted.
The flooding initiated from the active area of Tray 45 (just below the chimney tray) and propagated up to Tray 62
Illustration
Illustration: FIGURE 7. At most design rates, a clear vapor region exists between the top of the froth and the
bottom of the tray immediately above
Illustration
Illustration: FIGURE 8 (BELOW). The size of the liquid-vapor disengagement zone decreases as the feedrate rises.
As feed increases from 120% of design to 130% of design, entrained liquid droplets continue to encroach upon the
liquid-vapor disengagement zone, until a clear vapor space no longer exists
Illustration
Illustration: Graph: FIGURE 9. Shown here is the operation-capacity diagram of the tray directly above the feed tray
(note the movement of the operating points)

PDF GENERATED BY SEARCH.PROQUEST.COM Page 6 of 9


(available online)

TABLE 1. KEY TOWER SECTIONS FOR EVALUATING CAPACITY INCREASES

Top section -- Reflux tray or packed bed

Feed sections -- Trays or packings above or below

any inlet stream, including:

- vapor feed

- liquid feed

- two-phase feed

- pump-around liquid return

- side-stripping vapor return

- additives

- solvents

Draw sections -- Draw trays (i.e., trays or packings above

and below any exit streams, including side

products, interexchanger draw and pump-around

draw)

Chimney trays -- Liquid collection for re-distribution in a

packed tower

Transition -- The first and last tray

sections (marking the beginning and the end) where

there is a change in design or orientation

of trays or packings

PDF GENERATED BY SEARCH.PROQUEST.COM Page 7 of 9


Distribution -- Distributors or redistributors, and

sections inlet-distribution internals

Bottom section -- Tray or packed bed above the reboiler

return nozzle

DETAILS

Subject: Chemical engineering; Processes; Hydraulics; Studies

Location: US

Classification: 9190: US; 8640: Chemical industry, includes rubber &plastics; 5310: Production
planning &control; 9130: Experimental/theoretical treatment

Publication title: Chemical Engineering; New York

Volume: 105

Issue: 8

Pages: 100

Number of pages: 0

Publication year: 1998

Publication date: August, 1998

Section: Engineering Practice

Publisher: Access Intelligence LLC

Place of publication: New York

Country of publication: United States, New York

Publication subject: Chemistry, Engineering--Chemical Engineering

ISSN: 00092460

Source type: Trade Journals

Language of publication: English

Document type: PERIODICAL

PDF GENERATED BY SEARCH.PROQUEST.COM Page 8 of 9


Accession number: 01692941

ProQuest document ID: 194413077

Document URL: https://search.proquest.com/docview/194413077?accountid=53364

Copyright: Copyright 1998 The McGraw-Hill Companies, Inc.

Last updated: 2016-11-19

Database: ProQuest Central

Database copyright  2018 ProQuest LLC. All rights reserved.

Terms and Conditions Contact ProQuest

PDF GENERATED BY SEARCH.PROQUEST.COM Page 9 of 9

You might also like