773-715

CHAPTER
15
Separation Equipment—
Design and Costs
Since the separation of most mixtures into their constituent chemical species is not
a spontaneous process, separation will require the expenditure of energy or the use
of external forces. Many techniques are available for separating either homoge-
neous or heterogeneous mixtures. This chapter examines several of the more widely
used separation techniques.
When the feed mixture is a homogeneous, single-phase solution, a second immis-
cible phase must often be developed or added before separation of the chemical species
can be achieved. The second phase is created by the use of an energy-separating agent
and/or the addition of a mass-separating agent. Some of the more important separation
processes besides distillation and absorption include azeotropic and extractive distil-
lation, stripping, extraction, and crystallization. Design procedures for most of these
separation operations have been incorporated as mathematical models into widely used
commercial computer-aided chemical process simulation and design programs for
steady-state operations.
The use of microporous and nonporous membranes as semipermeable barriers in
the separation process has been receiving considerable attention with numerous appli-
cations. Separation techniques that involve some form of barrier in the separation
process include microfiltration, ultrafiltration, nanofiltration, osmosis, gas separation,
and pervaporation. On the other hand, removal of certain components may be accom-
plished more successfully with the use of solid mass-separating agents. Separation
processes that operate in this fashion are adsorption, chromatography, and ion ex-
change. Finally, the use of external force fields can sometimes be used to provide the
separation driving force between dissimilar molecules and ions. The use of this concept
leads to another list of separation processes including centrifugation, thermal diffusion,
electrolysis, electrophoresis, and electrodialysis.
When the mixture is heterogeneous, it is often more practical to use some mechan-
ical process based on gravity, centrifugal force, pressure reduction, or electric and/or
magnetic field to separate the phases. Further separation techniques, if required, can
Selection of Suitable Separation Processes 755
then be applied to each phase. Some of the methods used for the separation of
heterogeneous mixtures include settling and sedimentation, flotation, centrifugation,
drying, evaporation, and filtration.
There are two measures, besides economics, that provide a good insight with
respect to the overall effect of a separation operation. One measure is the recovery per-
centage of key products obtained from the feed mixture. The other measure of separa-
tion is product purity. Both measures have served as guides for design engineers in
the selection of suitable separation processes required in industrial processes. These
measures have become even more critical as more recent separations involve the recov-
ery of very high-cost products requiring particularly high purities from temperature-
sensitive dilute feed streams as, for example, in many biotech and pharmaceutical
separations.
SELECTION OF SUITABLE SEPARATION

PROCESSES
The selection of the best separation process often must be made from a number of
possible separation processes. The important factors to be considered in making the
selection include exploitable property differences, feed and product conditions, and
characteristics of the separation process. The most important feed conditions are flw
rate and composition, particularly of the key components to be recovered or separated.
Generally, the more dilute the key components are in the feed mixture, the greater is
the cost of separation.1 However, some separation operations, such as those based
on the use of barriers or solid agents, perform very well even when feeds are dilute in
the component that is to be recovered. The most important product condition is the
product purity of the component or components that are recovered.
A guide to exploitable property differences and the separation processes utilizing
these property differences is provided in Table 15-1. In reviewing the suggested sepa-
ration processes that are based on these property differences, the designer must have
some understanding of the general regions of applicability for these processes. For ex-
ample, selecting chromatography to separate components from an industrial multicom-
ponent mixture would be unrealistic even though a large exchange equilibrium might
exist between the components. However, it would make some sense to use the latter
separation process to establish the presence of undesirable impurities in the feed or
product stream.
The achievement of very pure products either requires large differences in certain
properties between the components that are to be separated or requires a large num-
ber of stages for the separation. It is important to consider not only molecular proper-
ties but also the bulk thermodynamic and transport properties. A good source of
such data with estimation methods for many bulk properties is provided by Reid,
'G. E. Keller, AlChEMonogr. Set:. 83: 17 (1987).

756 chapter 15 Separation Equipment—Design and Costs
Table 15-1 Property differences associated with various separation

key processes1^
Exploitable property differences Related separation processes
Vapor pressure Distillation

Absorption and stripping
Drying
Evaporation
Solubility Crystallization
Leaching
Distribution coefficient Solvent extraction
Adsorption
Exchange equilibrium Ion exchange
Chromatography
Surface activity Froth fltalion
Molecular geometry Membrane
Dialysis
Molecular kinetic energy Mass diffusion
Thermal diffusion
Electrical field Precipitation
Particle size Filtration
Screening
Settling
Sedimentation
Particle size and density Centrifugation
Classification
Thickening
Decantation
Scrubbing
T
Modified from data presented by D. R. Woods, Process Design and Engineering Practice,
Prentice-Hall, Englewood Cliffs, NJ, 1995.
Prausnitz, and Poling/ Both molecular and bulk properties are given by Daubert and
Danner/
Generally, when a separation process is well understood, the process can readily
be designed using a mathematical model and/or scaled up to handle industrial flw
rates directly from laboratory data. Other separation processes may require pilot plant
data before a final design can be made with some assurance of operational success.
Operations based on a barrier separation are more expensive to stage than those based
on the creation or addition of a second phase. Since some separation processes are lim-
ited to a maximum size, the use of parallel units will become necessary. The choice of
single or parallel units can have an effect on the economics of the separation process.
'R. C. Reid, J. M. Prausnitz, and B. E. Poling, The Properties of Gases and Liquids, 4th ed., McGraw-Hill, New
York, 1987.
T. E. Daubert and R. P. Danner, Physical and Thermodynamic Properties of Pure Chemicals—Data Compilation,
DIPPR-AIChE, Hemisphere, New York, 1989.
For example, doubling the capacity of a single unit will normally increase the capital
investment by about 50 percent while the use of parallel units to handle the doubling
in capacity will require an additional investment of 100 percent.
Tables 15-2 and 15-3 present some guidelines to assist the design engineer in iden-
tifying one or more feasible processes that would be appropriate for the separation of
homogeneous and heterogeneous mixtures, respectively. The tables outline the normal
feed conditions and the separating agent, force field, or gradient that will be involved
in the separation process under consideration. Additional information in the table indi-
cates the approximate concentration range of key components that may be encountered
in the feed for each of the listed separation processes as well as an estimate of the range
in flw capacity of the equipment available for laboratory to commercial use associ-
ated with these processes. Finally, advantages and disadvantages listed for each of the
separation processes can provide insights for recognizing whether such alternative
processes could be utilized to perform the required separation. The next section pro-
vides additional information on several of the more widely used separation processes
identified in these tables.
Guidelines for the Separation Process Selection
Distillation The creation or addition of another phase in distillation is obtained by

the repeated vaporization and condensation of the Aid. The separation process ex-
ploits the differences in vapor pressure of the key components in the mixture to initi-
ate the separation. The advantages of distillation are its simple flwsheet, low capital
investment, and low risk. The separation process has the ability to handle wide ranges
of feed concentrations and throughputs while producing a high-purity product.
This separation process can easily be staged. It is possible to have more than
100 theoretical stages in one distillation column. Accordingly, relative volatilities be-
tween key components to be separated can be as low as 1.2, and the separation can still
be achieved. In addition, mass-transfer rates between phases are generally quite high,
and phase contacting and disengaging are easily accomplished so that equipment
dimensions needed to obtain high degrees of separation are usually not excessive.
The equipment requirements for distillation are small compared to those for most
other separation processes. All that is generally required is a column, a reboiler, a con-
denser, and some relatively small ancillary equipment. By comparison, separation
processes based on the use of mass-separating agents typically involve two or more
separate mass-transfer zones, or require semibatch operation, or do both to perform the
same separation.
The economies of scale, as noted earlier, favor distillation for large-scale separa-
tions. Process information on many of the other separation processes indicates that
these have higher scale-up factors. Thus, the economic improvement associated with
higher throughputs is not normally as pronounced with alternative separation
processes as it is with distillation. This lowered potential for cost reduction is due to
the added flwsheet complexity and/or the necessity of using equipment in parallel to
meet the increased throughput. Membrane separations often encounter this design
limitation.
I ^ >, P CO 1/2
0.3 go P "O •3 2 CO CO -a 3
cn > •-; UJ ■ P p P
3 p
^5 S
■•C 3 ^ cs
"3
a C c_ — P P
iS rj -
^ ^ '— -y- Q> - a
£ « a -3
cu
c
ti.
b.tc x .3 3
cS -3 -3
C/3
O cr — C c.
CJ p p
p C- O H p cz . cr.3
3 >. p b
z 00 X p a 3 o p p p p 00 a- o
cz
co
p P_ w P
v 1 CO 3 ^ _p c 2 p p
ex: •^ 3 p 3 s cd fa O
CO .3
p. p 5
3 o.
E"
CO P P 3 a. (D ■£ cp 3 a o 3
3
P a i= _3 0 fa.
s _3 O P
O o i*l p
s _o P 0
1 "3 —- "3 vc 3 ■S S
p P P (/2 P bO p
3 p' 2 co 2
fa- P CO b S O B- 3 x
p ■p p P 5 a p >> a
p 3 O £ = B
CD pp ■ac/:—' 1 e
5 b
X E • •-- o ;3
« "co cz -h
CO t/^ 8 g — CO
CO p CO
p O c p *3 p p rp P'.B 3 3
p CO p — B H) s
3 rB P
P S & a a
u 5 > E N —| - , 3 ~
2 1 p >>
P .£ S co a fa
w £ co E •—> £
H P- co C > 3 33 C >..3
3 a. fa "3 p X; CO P • •*—• p 3 CO o fa ^ 0
o £ "a
p • ■—«
-2 3
p Oc co -2 3 V
0 O .s- s p 3
p ^
> b0 o E 8
c 3 c p u — p > - s 00
O •— g ^
<D 00 0 CJ CQ 3 > 3 D- o^ 3 .S 00 -3 3 o S cn 3 p u
D>
O
£ .P ^ CO O
o 0 O 0 o O o 0
JZ X p in in in.
CD _ n. P3 I I I l
c CD S3 c <N •J -t T -a-
£ 11 1 1 I 1
'■3 p ^ ~
T3 •= 1 c o o o O
CD o
s- ^ ^ .i s-
ro Z P. s P
a
o
«
P
p 'p
P p
p CO C—< 3
3 p O
c
3 3 3 cop ■p 3
'3 §" p
o C4^ p 3p « 3
3 i=
o
CO > _> .p
■p
CO 3 •*—> o 3 CO
o •5 ^
C
3
3 P 3 co pc3 3 C30
3
o
co 3
3
o p P P P P p P
p « c ® •-3 3 3 3 "a
CO fa CO 3 -o 3 a 3 3
g"^ 2 P 3" 3 •^3 3 ■£ 3 P
«
(0 C/3 CD S C£ X fa a -J a 00 fa X
p
o
o
0 P
p a
sz o g- p 7? 3
s > 3 >
o c fa-
p ® 0-0
-fa 3 p
V) p 1 in i ^ m
3s
in
p c £ ® 3 3 fa -2 3 ^
c o
• =- 2 3 O 3 p a 3 O b d
p —— P I
p _ CD [/j O fa b" fa • P rî |
P p 1 •pT-H 1 P •p
3 0 © 271 3 co '5 3
■g p
P
p ¥
c ^ 3" 3- P cr 1 pa 0 3-1 3- 3"
■3 c ® 52 -- 3 N faS
fa CD S fa > 3 i> •— fa s fa fa
D)
ro
i-
p
c
p
O 3
p co
3 O
CMI c =2 O 3 3
c CO P ,0 .2 3 p 3 _o 3 3
N
m p 3p coP a o > .2 "fa-* bl) _o
T' ?5 y? BJD b 3 o P p 3 _s
p ai O 3 p 3 CJ 3 & 'a 3
P 3 .—i a p
CO
Q. 3 S .£ o p P fa fa 8 2 >,
Si p p O 3 N co X .fa p fa X p
tr faX! 3 Q ^ - t-g
« « op c tu 3 ^ =
tU i-
tu o
o
^ ^ 2 « c P 3 ©
T3 cr _© 3 2 3^
•B -a u .l
« B "© _o Sfos
N to —- 1 a- >, ^ d
ol'i d 3
^., • —
^ (D "S s 3 u- - - 3
3
^ 60 ©
> • F-" £ 3 2 S 1
O 3 3 © S .2° 2-
|i§
^ -C "" (U ^ c/D
v- (U CJ
© 60 ^ X « tu r--
2 u tu a. bp-u i— to © u- pS Ot
S 3 >> to 3 U3
„ oc c
c3 -23 to © >.
© tu C
2 as 60 1— C to
• —H
©
3 rt
u _ c6
v. c
T3 0)- c 3 © (U ■- 3
£ "B E U U a CT" 60 3
o cr -u © C 3 g J to s I
J E S B --2 C2 • — 3 ro ri—
U 3 Po o
3 CJ
z
b tu 3
c u -a O CJ
o __ '•2
c .2 -td • o 2 tu 5
.2 2 $ 60 .2 -
a — C c d I
a cc oo o i/i O 3
C 4J Zo S
« c ^ d o
CJ
S. a 2 S H ■-2 2 .2 c
- a.| x; > -3 "X tu u oo
oc
S
w ^ a g
3 - = ■r c o g G-
C3 3 5 8
3 C2
-C - OJ
to-, to
•3
O
U
2
3
2 3 b f
^ -
a
O <+_ U B cu > - ■o
i—H I
•2I1 >. O ^ a 3
tu S g «
"O c (U .2 3, o 60 ,8
2^2 3 c3
G. -7-
Z, ^
>5
o d 5 b c OS ^C 3 O. ^
a - 2 5 2 o 3 gog -o
dJ Z
S2
c 5
rG ■S- §.
I I I fl
Ti •n ^r ^ .
I I I c
&• >
ci^ —*
^ ^"£}
o o O
a ^ u.' 2
<U '~*
s
a.
d
a
to to to g © E Id
_3 G 3 to tu ■e ©
i 1 ■— tU
IU 3 2 S ^3
■£. a tu
c 3 c o to 3 U-.
60 o (3
tu o a o e O a O t3 ■o
—
3 c
tu 3 UH 3 a
O i- d, a -©
— c
tu
§- «J
Co -M
to • —H Ul 3 © 3 3 ■ll
to 3 a £ u
P a CJ (U 3 £ II
(U O <u
II
o - o
O
5 3 ^J (U
3 — Z c on C — LJ
60 i E S Z £ w 8 "© s ^ .,
3 UK
w —-0 <uCJ
-a -f
x in 3 x
ccd 00 -z:c
ox OJ
X ■b- r~ •—
. 00
J '—J cn 3 fP 3 m 1 tu -g
tu
tu
IF .g
C g . So
0 g
D- . d
c c o c a 3
1 0 to a
G.
1)
0 £ 0 3 -o o ao a
to tU E s
4J 1 1 1 c 3 c d .60
C 6)
to o _o Pa
- 3 ■s 3 >>
a o i-i §■ -—'
l-H —
CJ 1
 a.
0-t
o
'C 2 -2
« tu Ca Q
-o __
(U a> O —- M cc c S
E
OJ £2 o- to o
'S
D TJ S3
__ ■3 00 tu
3 o 3 2
< ^
Q a. £
759
a >> If1 >.x -a a 5f J.
-a So 50 a m 0 0 ^2 ■S
0=0 x
c
.2? E." 8 2 0 "a
— 0 X 0
• -- <£ O
W) -a js g. 0 a — 0 3 o
C a a 6a 6-, a-
_o <: if ^ x o 0 ■0 O
'■S = y <1>
-—
0 = a a 2 f. o 2 =?£
a 1/2 o | N
• —— X a 0 "£
■—(
•0
4-0
e o o S
c3 a
0 S T3 a C/5 a 9- a 3 0 _a 0
o I -a a a 0 ?X
£ 0 > £ "o 60 3 O a- CZ
X c
a 0 0 o .2 -2 2
<D • 4—•
— 2 X o £
if)
ir ■3 I •H HH0 O a 0
4-0 a X a a N
0 2 | f -> a £ 11 Cj o^.
u — — 0 Ji §- 0 CO a X s a .2 0 S x
Q, CO -3
^ •S
c —■ 0 O I •5 OO0
s 0 2 ~ ir3 60 0 60 » ■£
a a ■ —-
•— 0 1 ii
0 -a X ^3 1 X x -3
O 0 -0 60 C/} E?
a 60 3
a a"
2°
0 o s-
a. 60 2 - £
S !>■
o — 2 -2 '3 O X a •x a
Qh ta '-a ^ on w — If X O 0 £ 2 0 -J < X x
X
o x-
sa
- if. u0 a x
o a
•w
—< 0 c a oj
0
0 ■ X a Q I
> —' a 0 a 0 0 a if
a
V) 2 0 0 o _a 0 2 £ d o u
0) Ml ^ a '0 f a "o _a 0 y o o. a 00 S
a a a S TZ S- 2 0 > o O Co 0 a
a o O C3 £ x 0 0 > §-2 X g
>■ TD o 0 O X if
T3 0 0 2°
S
a)
CZ (J > if > o 2 m S 2 f
£ c
03 fa § o £ _a ^0 . a
w W/ X) § *1
o o 2 _ 0- a o
til -a
0
3 0 a 2 a 1 0
a
C —
-^0
a •— a
o 0 0 ~~ 4-> «
a 0co
-H c- - cr-o a0 p*~» (D C3 s li £
f
0)
c
E •a §-
0
if t;x 0 x 0 § S O if i-H ~~
•a
 .. 0f
a) a 0 « X
'-0 — 2 a 0 X a 0 0
1 a a | os a £ <6
O) 01 0a. 0 a
If Q M a o a " 2 § i"® o £ x
._ o o
on 4-0 — O D -a A 01 8 8 (D o if 1 â
2 X £ S 2
0) •£ -2
0 if a
£ 0 0
D) £Vj |
C a
>—• >•
— -' ^ (N
4-' f-J
I I
a
I ■S
0 ^ ia,.
Si - >£ X x 0 o c 0 a, — on
s a. 0 CN a- »0 a-, fa,
ro « ^ I a 0' 2
a I I O I 0—
0 a v-
w ■c •- I
vC
1 X
•o
1
u ? £ o O O
Z £ := 3" m a^d
£
C I^D
o
00 a^ ^
o CZ a 0 0 S. 2 8
0 0£ E M
0 C 0 .c
a ■0 0 0 a -0 0 .0
o
£
if £
if X 0
cox ^ 2 g
If) —' g .a 00 2 a a 0 — X
= CZ t: >. ,3 0 £ a a a "a = « £^|op a
If)
If) b -- fe .a • 4-0
•—> 2 a X £ y= £ S £ 3 a = uj
5 c ®-a a 2 o
If a 0 if "O o
0 a W) a a a
U a- 0 —■
0 ojsS t. ■o 2 a ■■0
:::;
0 00 a 0 0 P a '—
0
O C/3 C8 B C£ O x < a a U £ O x X X 0- M O, U S
!_ uj > 0
a
0 a 06
d '=
p
JZ 11 2 £ 2
0
»—H 0 Qj
0H £ ■
a c X
•— X
v) o a a rn 0 0 C 0
0 0 "o ^Î
c x .2 a a If X o c
■o 2 o M X 1 X | 2 vn Q 5 S
0 a I a 0 "3 Tl" a ^r X 0 • z
0 o,
x
-q o al E a 1 "0 7 On ? 2 d
0 X C X ^
"5 « in
■a T5 S 21° £ ^
X —H "o
If X
o x£
a> '3 HH '3 1
0.5-
0 f g-57 1
0 a a a- X ■|x cr cr a
-- f O If*
E b 03 = a « j & 2
^ £ U") X X 1 £ £ M— O , •
0 <D - ?
c D..0 d
0
O if a a "S
0 a a a a -§ § 5
_o a
f? a =£ 0
o
12 -i a .2 a ^0
O if a a .2 a X^2 -Q 0
10 -00 a 2 a 00 —0 if a
M a o c3 a o x a Q
* £ • »-H
0 0 % 0 i
0 • 4--
i-H
a
M
a O x
0
a ■x 2
0
a ^
a |.§ 0 • —4 a X Of) 0
a- c S x a & £
n 0 ■0 0 'a _o 0 &■ > o • -0 0
x- 0 £ a
cr a.x: E 5- U Q m U U- .£ Q Q,
760
On the other hand, distillation can become very expensive when the relative
volatility between key components is decidedly below 1.2 over part of or all the distil-
lation operating curve in the column. In fact, some separations can involve azeotrope-
forming systems in which the relative volatility for the key components actually
becomes less than unity. In cases of low relative volatility, other separation processes
should be examined to attain less expensive options.
Fixed composition and product requirements for purity can sometimes create
additional problems. For example, separation of a high-boiling component from a 10
to 20 percent or less concentration in the feed will increase the energy costs per unit of
desired product and make it excessive. Also, when the boiling range of one set of com-
ponents overlaps the boiling range of another set of components from which it must be
separated, distillation will require a number of columns to complete the separation. In
such a case, another separation process might exploit other property differences and
achieve the separation in fewer steps.
Distillation can also be at a disadvantage when extremes of temperature and pres-
sure are encountered. For temperatures less than —40CC or greater than 250GC, the cost
for the added refrigeration or heating involved escalates rapidly from typical cooling
water and steam costs. If required operating pressures are less than about 2 kPa, col-
umn size and vacuum costs also escalate rapidly. Likewise, pressures greater than
5 MPa will result in an escalation of column costs.
Earlier we noted that distillation provides a relatively low scale-up factor. Unfor-
tunately, it does not scale down very well. Thus, alternative processes that may not
compete economically in the separation of large production rates may compete quite
well when production rates are quite low.
Finally, distillation may be unsuitable when undesirable reactions occur at column
temperatures. This can result in significant product loss or in the formation of by-
products that can further complicate the separation process. For example, if only a
small fraction of the feed reacts to form a by-product that precipitates on the tray open-
ings, fouling may require intermittent cleaning with considerable loss in production or
the requirement of a duplicate separation system to maintain continuous operation
during the cleanup period.
Absorption A widely used alternative to distillation for the separation of a solute

from a gas stream is absolution. In this separation process, the gas mixture is contacted
with a liquid solvent which preferentially absorbs one or more components from the
gas stream. Liquid flw rate, temperature, and pressure are the variables that must be
set in this separation process.
The absorption factor L/KV for a relatively dilute component in the gas phase
determines how readily that component will absorb in the liquid phase. Here L and
V are the liquid and vapor mass flw rates, respectively, and K is the vapor/liquid
equilibrium value for that component. When the absorption factor for a specified
component is large, its absorptivity in the liquid is increased. When the absorption
factor is increased by increasing the liquid flw rate, the number of trays in the col-
umn required to obtain a specified separation decreases. However, at high values of
the absorption factor, an increase in the liquid flw rate achieves diminishing returns.
Accordingly, an economic optimum for process design has been established, namely,
\.2 < L/KV < 2.0. A value of 1.4 is often used.'
In an absorber, the transfer of solute from the gas to the liquid phase can bring
about a heating effect that will generally lead to a temperature rise down the column.
If the component being absorbed by the liquid is dilute, the heating effect will be small,
resulting in a small temperature rise. However, with a large absorption the temperature
rise down the column can be significant. Unfortunately, an increase in temperature
decreases the solubility of the absorbent and reduces the separating efficiency of the
absorption process. To minimize the effect of the temperature rise in absorbers, the liq-
uid flwing down the column is sometimes cooled through heat exchange with cooling
water or refrigeration at intermediate locations in the column.
Greater solubility of the desired component or components can be obtained by
operating the absorber at higher pressures. However, operating at higher pressure tends
to be more expensive, particularly if this requires the use of a gas compressor. Thus,
economically there is an optimal pressure for the absorption process.
Stripping Once a solute is dissolved in a solvent, it is often necessary to separate the

solute from the absorbent in a stripping operation to recycle the solvent back to the ab-
sorber. The stripping factor KV / L should be large, to permit stripping the solute and
concentrating the latter in the vapor phase. For a stripping column, the stripping factor
should be in the range of 1.2 < KV/L < 2.0 and economically is often close to 1.4.
With strippers a temperature decrease is noted as the liquid flws down the column.
The reasoning is analogous to that developed earlier for absorbers. Thus, to enhance
the stripping process requires an increase in operating temperature and/or a decrease in
column pressure.
Extraction Liquid-liquid extraction is a process often selected to separate azeotropes

and components with overlapping boiling points. The extraction process requires the
addition of a solvent in which one or more of the components in the mixture are solu-
ble. Components in the liquid mixture are separated based on their different solubilities
in the solvent.
The extraction process can offer energy savings, particularly when solvent recovery
is relatively easy. The process can be operated at low to moderate temperatures for re-
covery of thermally sensitive products in the food and pharmaceutical industries. How-
ever, the requirement of a solvent increases the complexity of the process. The latter
consists of an extractor plus at least another distillation column to recover the solvent. If
the solvent has some miscibility in both components that are being separated, a second
distillation column will be required to complete the recovery of the solvent.
One of the more recent advances in liquid-liquid extraction has been the use of a
supercritical Hid such as carbon dioxide as a solvent. Supercritical solvents of fer
enhanced transport properties because solutes diffuse more rapidly through a super-
critical solvent than through a liquid solvent. Supercritical Aids also provide other
desirable solvent properties, with equilibrium ratios and separation factors that are
J. M. Douglas, Conceptual Design of Chemical Processes, McGraw-Hill, New York, 1988.

appreciably larger than with normal solvents. Since the solvent is in the supercritical
state, it can be recovered as a gas by simply reducing the pressure and changing the
temperature.
The primary disadvantage of supercritical Aid extraction is that the extractor
must be maintained at a pressure sufficient to keep the solvent in the supercritical state.
Since supercritical Aid extraction generally requires operating pressures between 7
and 35 MPa, the capital and operating costs are higher. Nevertheless, supercritical Aid
extraction has found favor in the purification of high-boiling specialty products where
molecular distillation previously had been used. In such applications this separation
process has the advantage because multiple stages can be used, whereas in molecular
distillation staging is difficult and expensive.
Crystallization Under certain operating conditions, dissolved materials in solution

are recoverable in solid form by precipitation upon cooling, removal of the solvent, or
the addition of precipitating agents. The design of crystallizers is based on the knowl-
edge of phase equilibria, solubilities, rates and amounts of nuclei generated, and rates of
crystal growth. Each crystallization system is unique in most of these design attributes,
and these are not predictable. Since theoretical advances have been slow in this sepa-
ration process, pilot-plant evaluation is normally required to establish critical design
parameters.
Various crystallizers presently employed in a number of industrial processes have
been identified by Walas.' For example, batch crystallizers are primarily used in the
production of fine chemicals and pharmaceuticals at a rate of 0.0015 to 0.15 kg/s. On
the other hand, crystallizers are also of the continuous type utilizing circulating evap-
orators and/or circulating cooling crystallizers. Many of these crystallizers provide
some control of crystal size with the aid of special design features.
Membrane Separation This type of separation is accomplished with the use of spe-
cially prepared membranes that selectively permit one or more components of a feed
stream to pass through the barrier while retarding the passage of other components in the
feed stream. The stream passing through the membrane is referred to as the permeate
while the stream retained by the membrane is the retentate. In most membrane processes,
a pressure difference between the feed and permeate streams provides the driving force
for the separation.
Membrane processes are classified by the size of particles or molecules being sep-
arated as defined in Table 15-4. Filtration, discussed under heterogeneous separation,
was not included in this classification since the barrier in filtration separates particles
with dimensions greater than 5 x 10-6 m.
Since membrane processes can separate at the molecular and very small particle
levels, a large number of difficult separations are now possible. Membrane processes,
except for pervaporation, generally do not require a phase change to make a separa-
tion. Energy requirements, as a rule, will be low unless energy is required to increase
the pressure of the feed stream to obtain a suitable pressure driving force. Suitable
t
S. M. Walas, Chemical Process Equipment, Butterworth-Heinemann, Newton, MA, 1988.
periodic operation can introduce a great deal of process complexity, which can in some
cases result in unacceptably high capital costs for the separation.
Adsorption processes find applicability throughout a broad band of industries in
both bulk separations and gas purification. When adsorption processes are used in bulk
separation applications, the design engineers must recognize that large quantities of
heat are liberated during the adsorption process and that conversely large quantities of
heat are consumed during the desorption process. For gas adsorptions, this release
of energy is often 2 to 3 times the heat of vaporization. For liquid absorptions, this
energy release is somewhat smaller. For gas purification, which involves adsorption of
small amounts of gas, this energy release can usually be neglected. However, for bulk
separations, the energy release becomes an important issue in the process design of
such equipment.
Chromatographic processes have a technological similarity to adsorption processes
since both use selective concentration on a solid surface to effect a separation. The chro-
matographic separations generally involve either laboratory work or the separation
of very small, high-value chemicals for use by the pharmaceutical and biotechnology
industries.
Ion exchange is also similar to adsorption in that solid particles are used in the
separation process and regeneration is necessary. However, a chemical reaction is in-
volved. For example, in water softening, the solid organic polymer in its sodium form
removes calcium ions from the water and replaces them chemically with sodium ions.
After prolonged use, the polymer is regenerated to remove the calcium ions.
External Field or Gradient Separation As noted earlier, separations utilizing ex-

ternal force fields or thermal gradients can be used to take advantage of the response of
molecules and ions to such forces. Since the use of these separation processes is lim-
ited to a few specialized industrial applications, the emphasis in this chapter will not be
directed to these specialized separation processes.
Settling and Sedimentation In settling processes, solid particles or liquid drops are
separated from a Aid by gravitational forces acting on the particles or drops. The Aid
can be a liquid or a gas. Flash drums are often used to separate a vapor-liquid mixture
by gravity. In such a separation the velocity of the vapor within the drum must be less
than the settling velocity of the liquid drops. In a simple gravity settler for removing a
dispersed liquid phase from another liquid phase, the horizontal velocity of the Aid
within the settler must be low enough to allow the low-density droplets to rise to the in-
terface and the high-density droplets to move away from the interface and coalesce.
The prime function of a sedimentation device is to produce a more concentrated
slurry. In one type of thickener the slurry is fed to a cone-shaped tank with a slowly re-
volving rake to scrape and move the thickened slurry toward the center of the cone for
removal. Often in such operations a Acculating agent is added to the slurry to aid in
the settling process.
Flotation In this process of separating specific solid-solid mixtures, gas bubbles are
generated in a liquid and become attached to selected solid particles, allowing them
to rise to the liquid surface. The solid particles are removed from the surface by an
overflw weir or a mechanical scraper . The separation of the solid particles is depen-
dent on differing surface properties which permit preferential attachment to the gas
bubbles. A number of additives can be added to meet various requirements of the flta-
tion process. For example, modifiers control the pH of the liquid-solid mixture, acti-
vators increase the activity of the solid surface, and frothers create a stable froth and
assist in the separation. The bubbles of gas can be generated by gaseous dispersion,
dissolution, or electrolysis of the liquid.
Centrifugation Separation by applying a centrifugal force field was noted earlier

under external field separation. It is mentioned again since it is a useful separation tool
when gravity separations of heterogeneous mixtures may be too slow because of the
close densities of the particles and the Aid, the small particle size of the particle
resulting in low settling velocities, or the formation of a stable emulsion. Use of cen-
trifugal force increases the force acting on the particle and permits more rapid separa-
tion times. Besides solids separation from liquids, it is also sometimes used to separate
two liquids with different densities by centrifugation. In this case, the denser Aid
occupies the outer periphery of the centrifuge, since the centrifugal force is greater on
the denser liquid.
Drying The main purpose of this separation process is to remove liquid from a
liquid-solid system to produce a dry solid. Generally, the liquid removed is water, but
drying can also be used to separate organic liquids from solids. Most dryers obtain the
heat required to vaporize the liquid from the solid by employing a series of gas-solid
contacting devices. The specific type of contacting device selected is based on the con-
dition of the feed, the desired form of the product, and the temperature sensitivity of
the solid. For example, if the feed is a dilute mixture of solids in a liquid, the dryer se-
lected to remove the liquid could be either a single-drum or a spray dryer. On the other
hand, for a moist solid the drying choice could be a tunnel, screw, or rotary dryer. The
variety of equipment available for drying service is quite large, particularly when addi-
tional requirements of product form and heat sensitivity need to be considered.
Drying devices are generally subdivided into two groups: those that depend on
mechanical means for gas-solid contact and those that depend on Aid motion. Within
each group there is a second subdivision, namely, those that utilize either direct or
indirect heating. Direct heating occurs when hot gases or combustion products come in
direct contact with the solid being dried. In indirect heating, the energy transfer is
through a conductive surface that separates the process stream from the heating media.
In the processing of solutions or slurries, successful heating devices provide a
transition zone at the entrance, atomize the Aid, or premix it with recycled solids to
enhance the Aw . From an energy conservation viewpoint, total evaporization of the
liquid from the solution or slurry should be avoided unless other separating or concen-
trating techniques are impractical or inappropriate. For gummy pastes, a pretreatment
step is often necessary to convert the paste to a form that it can be successfully
processed. With dusty materials, special equipment may be required to maintain a
dust-free environment. Heat-sensitive solids will require dryers with precise tempera-
ture control and may even require drying under vacuum conditions.
Evaporation In separation technology, evaporators may be considered as liquid dry-

ers or concentrators that separate solvents from a solution by evaporation. In most
cases the solvent is water. Evaporation, as a separation process, is distinct from distil-
lation in that the solute is nonvolatile and a high degree of separation can be achieved
in one stage.
Evaporators are essentially reboilers with adequate provisions for separating liq-
uid and vapor phases and for removing solids when they are precipitated or crystal-
lized from the solution. However, because of the varied did property characteristics
of the solutions that need to be concentrated by evaporation of the solvent, there are
presently many specialized types of evaporators in use. For those evaporators employ-
ing tubes to circulate the heating Aid, the tubes may be horizontal or vertical, long or
short. The liquid may be inside or outside of the tubes, and circulation may be natural
or forced.
The forced circulation type of heater is the most versatile, particularly for viscous
and fouling operation, but such units are more expensive than the natural circulation
evaporators. Long-tube vertical evaporators, with either natural or forced convection,
are the most widely used units. The tubes in these units range from 0.019 to 0.063 m in
diameter with a length of 4 to 12 m. Thus, there is a wide range in capacity for this type
of evaporator.
Since vaporization is energy-intensive, thermal economy is a major considera-
tion. If a stream must be concentrated, separation processes such as precipitation and
filtration or reverse osmosis that do not require a heat input should be considered. If
there is no practical alternative to evaporation, then it may be possible to improve on
the thermal economy by recovering and reusing the latent heat of the vaporized sol-
vent. In practice, individual evaporators are staged successively so that each unit op-
erates at a lower pressure than the preceding one. In this way, the vapor from one
stage can be used as a source of heat to evaporate the liquid in the next stage. There
are a number of different arrangements available when multistage systems are being
considered for performing the separation process. Such arrangements make it possi-
ble to recover energy even from low-level process steam that is used as the heat
source.
Filtration In this mode of separation, solid particles in a liquid or gas are removed
by passing the mixture through a porous medium that retains the particles and passes
the Aid. Filtration occurs by either cake filtration in which the particles are retained
on the surface of the filter medium or depth filtration in which the particles are retained
within the filter medium. The filter medium in cake filtration is generally a cloth of nat-
ural or artificial fibers or even metal. The filter cloth most often is supported between
plates in an enclosure. To separate solid particles from a gas, the filter cloth is arranged
in the shape of a large thimble and is designated as a bag filter. Filtration of a slurry is
best done with either a porous belt or drum that rotates through the slurry to remove
particles by direct contact. If purity of the filter cake is of no concern, filter aids can be
added prior to the filtration step to assist with the filtration process.
Rather than using a cloth, the deep-bed filters use a granular medium in which the
solid particles are trapped within the medium. The trapped particles can be released by
periodically reversing the flw of the filtrate. Approximately 3 percent of the through-
put is required for the periodic backwashing.
EXAMPLE 15-1 Identification of Feasible Options to Obtain a High-Purity

Propylene Product from a Propane/Propylene Mixture
This separation process is presently carried out using a 150-180 stage distillation tower with a large re-
flx ratio. Operating conditions of such a separation require the use of two columns in series, as detailed
in Example 15-2. Because of the large capital investment and operating costs for achieving this separa-
tion, what are some other feasible separation options that could be used to obtain a high-purity propy-
lene product similar to that achieved with the conventional separation process?
■ Solution
An evaluation of the properties associated with the two components is necessary to determine what
properties might be exploited to effect a separation. Key properties are listed below.
Property Propane Propylene
Molecular weight, kg/kg mol 44.096 42.081

Normal melting point, K 85.5 87.9
Normal boiling point, K 231.1 225.4
Critical temperature, K 369.8 364.8
Critical pressure, kPa 4250 4610
Dipole moment, debyes 0.0 0.4
Van der Waal's volume, nrVkg mol 0.03757 0.03408
Van der Waal's area, m2/kg mol x 10-8 5.59 5.060
Acentric factor 0.152 0.142
The only property listed above that might be exploited is the dipole moment. Because of the asym-
metric location of the double bond in propylene, its dipole moment is considerably greater than that
of propane. A review of the literature reveals that the use of facilitated transport membranes with
silver ions selectively transporting the propylene across the membrane has provided a promising sep-
aration process.+ The economic benefits of using a hybrid system that combines the separation as-
pects of the membrane systems coupled with a considerably smaller distillation column have been
examined by other investigators.^ One configuration investigated by these investigators was shown
in the design problem presented in Chap. 3.
Another separation process that exploits the polarity difference between propylene and propane
uses extractive distillation with a polar solvent, such as furfural or an aliphatic nitrile, to reduce the
volatility of the propylene.^ The literature also reports that either silica gel or a zeolite can be used to
selectively recover the propylene by adsorption.'11
'R. D. Hughes, J. A. Mahoney, and E. F. Steigelmann, in Recent Developments in Separation Science, vol. 9, N. Li
and J. M. Calo, eds., CRC Press, Cleveland, Ohio, 1982, p. 173.
Â. A. Al-Rabiah, K. D. Timmerhaus, and R. D. Noble, in Proceedings 5th World Congress of Chemical Engi-
neering, American Institute of Chemical Engineers, New York, 1996, vol. 4, pp. 335-340.
fij.S. Patent 2,588,056, March 1952.
'C. M. Shu, S. Kulvaronan, M. E. Findley, and A. T. Liapis, Separation Technology. 1: 18 (1990).
Equipment Design and Costs for Separating Homogeneous Mixtures 769
EQUIPMENT DESIGN AND COSTS FOR

SEPARATING HOMOGENEOUS MIXTURES
As noted earlier, the separation of homogeneous mixtures requires the formation or

addition of another phase. The transfer of mass from one phase to another is involved
in the operations of distillation, azeotropic distillation, extractive distillation, absorp-
tion, stripping, extraction, and crystallization. The principal function of the associated
equipment used for these operations is to permit efficient contact between the phases.
Even though many different types of equipment have been developed for these separa-
tion processes, only a few of the more widely used separation devices can be included
in this equipment design and cost summary.
SEPARATION BY DISTILLATION
There are two basic approaches used in distillation column design, namely, design and
rating. The former approach involves the design of a new column to determine the
column diameter and height required to achieve a specified separation. This approach
utilizes stage-to-stage calculations for determining the number of equilibrium stages.
The rating approach involves the retrofit of an existing column in which the column
diameter and height are fixed and the flw capacity and separation are to be deter-
mined. Since the rating approach has convergence advantages, it is most often used in
computer algorithms which do stage-to-stage calculations.
In the design of a new column or rating of an existing one, many design engineers
obtain final designs and price quotes from equipment vendors. However, to evaluate
such vendor responses, a preliminary design should be made. Accordingly, the proce-
dures involved in the design of a distillation column are summarized in Fig. 15-1 and
briep outlined with supporting equations in this section of the chapter .
Distillation Design Procedures for Columns with Sieve Trays
Designation of Design Bases The first step in the design procedure (see Fig. 15-1)
is to establish the composition and physical properties of the feed and products, the
feed rate, and any special limitations placed on the separation. The latter could include
maximum temperature and pressure drop restrictions, presence of toxic materials or
reactive components, etc.
Selection of Design Variables The second step involves selecting adequate design
variables. The operating pressure is generally the first variable to be addressed. An
increase in operating pressure is often reflcted by an increase in separation difficulty,
an increase in reboiler and condenser temperatures, an increase in vapor density, and a
decrease in the latent heat of vaporization. As the pressure is lowered, these effects are
reversed. The lower limit is often set by the desire to avoid vacuum operation and use
of external refrigeration for the condenser because of capital and operating cost penal-
ties. It is usually adequate, if process constraints permit, to fix the operating pressure at
as low a pressure above ambient that permits cooling water or air cooling to be used in
the condenser. In other words, the operating pressure should be selected so that the bub-
ble point of the overhead product is at least 5 to 10oC above the summer cooling water
temperature, or to atmospheric pressure if the latter would introduce vacuum operation.
Establish
separation basis
Select operating
conditions
Obtain or develop
equilibrium data
Calculate equilibrium
stages or transfer units
Select
column
Random internals Standard
packing packing
Tray type
Calculate column Calculate column Calculate column

diarrleter diameter dianteter
Calculate column Calculate column Calculate column

height height height
Figure 15-1
Procedural steps involved in the design of a distillation column
The second variable that needs to be set for the distillation process is the reflux
ratio. With an increase in reflux ratio from its minimum, the capital cost decreases ini-
tially because of a reduction in the number of plates required for the separation. How-
ever, the utility costs increase as more reboiling and condensation duties are required.
If the capital costs involved in the separation process are annualized and combined with
the annual cost of the utilities, an optimal reflux ratio can be obtained. Most designers
select a ratio of actual reflux to minimum reflux of at least 1.2 or above, except in spe-
cial cases such as ultralow-temperature separation, since a small error in the design
data or a small change in the operating conditions might lead to a process that does not
meet the original design specifications.
A design variable of lesser importance is the vapor/liquid condition of the feed. If
the feed is subcooled, the number of trays in the rectifying section is decreased and the
number of trays in the stripping section is increased. This feed condition requires more
heat in the reboiler and less cooling in the condenser. Partially vaporized feed reverses
these effects. For a given separation, the feed conditions can be optimized.
Establishment of Physical Equilibria Data Once the bases and operating con-
ditions have been established, the appropriate phase equilibria must be determined.
Either experimental or predicted phase equilibria can be used. Since most experimental
Separation by Distillation 771
vapor/liquid equilibria are for binary mixtures, and not for multicomponent mixtures,
it is necessary to use the data for binary pairs and combine these data with a model to
predict multicomponent behavior. The Wilson, NRTL, and UN1QUAC models1 use
this approach to predict vapor/liquid equilibria. When no binary data are available, the
UNIFACT model can be used for the prediction based only on functional groups. In
either case, most commercial simulators are capable of providing reasonably accurate
thermodynamic bases to predict phase equilibria.
Determination of Number of Equilibrium Stages Even though shortcut and stage-

by-stage methods for calculating the number of equilibrium stages are available in
many commercial computer software programs, it is necessary to understand the fun-
damentals involved to determine whether the computer results are realistic. Of the
shortcut methods, the Fenske-Underwood-Gilliland method is the most widely used.
To determine the number of equilibrium stages, first the minimum number of stages
and the minimum reflux must be evaluated. The minimum number of stages Nmin is
obtained from the Fenske relation^
in[(XLK/XHK)D(XHK/XLK)B] /lc 1N
Wmin = — (15-1)
ini<XLK/HK)a\
where x^k is the mol fraction of the light key, xHK the mol fraction of the heavy
key, (o/lk/hk)™ the average geometric relative volatility of the light key to the heavy
key, while the subscripts D and B refer to the distillate and bottom products, respec-
tively. The geometric average value for the relative volatility is calculated by using the
dew point temperature of the assumed overhead product and the bubble point temper-
ature of the assumed bottoms product. Thus,
(^LK/HK)a\ = l((*LK/HK)D(<XLK/HK)ls]{/~ (15-lfl)
To determine the minimum reflux ratio requires Eqs. (15-2) and (15-3) developed by
Underwood,1" namely,
■A OliXfj _
1
2^ 77 = (15-2)
where a,- is the average geometric relative volatility of component i in the mixture rel-
ative to the heavy key, x rj the mol fraction of component i in the feed, and q the mols
of saturated liquid on the feed tray per mol of feed. The value of 0 is determined by
trial and error and lies between the relative volatilities of the two key components. The
minimum reflux Rmin is obtained from
«mi„ +1 = 2 (15-3)
77 - ®
t
G. M. Wilson. J. Am. Chem. Soc., 86: 127 (1964), H. Renon and J. M. Prausnitz, AlChE Journal. 14: 135 (1968),
J. M. Prausnitz et ai., Computer Calculations for Multicomponent Vapor-Liquid and Liquid-Liquid Equilibria,
Prentice-Hall. Englewood Cliffs, NJ, 1980.
'tA. Fredenslund et al., Vapor-Liquid Equilibria Using UNIEAC, Elsevier, Amsterdam, Netherlands, 1977.
§
M. R. Fenske, Ind. Eng. Chem., 24: 482 (1932).
"A. J. V. Underwood. Chem. Eng. Prog., 44: 603 (1948).
chapter 15 Separation Equipment—Design and Costs
where n is the number of individual components in the feed and a/x, the mol fraction
of component i in the distillate. Gilliland related the number of equilibrium stages N
as a function of the number of equilibrium stages and the minimum reflux ratio with a
plot that was transformed by Eduljee1^ into the relation
\ 0.566"
N — N mm R - R mm
= 0.75 1 - (15-4)
N + 1 R+ 1
where R is the operating reflux selected by the designer. Distillate and bottoms distri-
butions of any nonkey components in the feed may be evaluated after the minimum
number of stages has been calculated from
^ = (J5.5)
XB,i (XHK)B
where xbj is the mol fraction of component i in the bottoms and (a,)av is the average
geometric relative volatility of component i relative to the heavy key as given by
Eq. (15-la). The Kirkbride5 method is used to determine the ratio of trays above and
below the feed point.
2"
nd B xhk (Xlk)I
log -f = 0.206 log (15-6)
Nr n xlk -{Xhk)d-
where B and D are the mol flow rates of the bottoms and distillate, respectively, and
No and Nr are the number of equilibrium stages above and below the feed tray,
respectively.
Selection of Column Internals At this point the designer must make a selection
based on performance and cost whether a tray, random packing, or structural packing
is best for the separation process being considered. Typically, trays are favored when
the operating pressure and liquid flow rate are high and when the column diameter is
large. Random packings are more often recommended when the column diameter is
small, corrosion and foaming are present, or batch columns are to be used. Structural
packings, on the other hand, are considered for low-pressure and vacuum operation.
Additionally, they are often selected when low pressure drop across the column is
required or low liquid holdup is desired.
Sieve, valve, and bubble-cap trays are examples of traditional crossflow trays.
The development of newer trays over the past decade to provide improved perfor-
mance has been significant. Descriptions of some of these newer trays, including the
Nye,™ Max-Frac,™ some of the improved multiple-downcomer trays, such as ECMB
and EEMD, as well as the Ultra-Frac,™ P-K,™ and Trutna trays, are presented by
Humphrey and Keller.''1
Figure 15-2 provides a cross-sectional view of a traditional distillation column in
operation showing an example of a sieve, valve, and bubble-cap tray. Of these tray types,
the sieve tray is the choice in many distillation separations since its tray fundamentals
+
E. R. Gilliland, hid. Eng. Chem., 32: 1220 (1940).
Tl. E. Eduljee, Hydrocarbon Proc.. 54: 120 (1975).
5
C. G. Kirkbride, Petrol. Refiner, 23: 32 (1944).
i!
' J. L. Humphrey and G. E. Keller, II, Separation Process Technology, McGraw-Hill, New York, 1997.
Gas out
Shell
Liquid in
Flooding Overflow weir
check line
Downcomer
Tray supporting ring
Tray stiffening bar
Tray Sample or
sidestream
a withdrawal
Liquid
y >( -o ■' n ,
Feed in
:--f7rrrr^ rrzzrv
Bubble-cap contactor r*.mnnm !□□□□! \uaouj
Valve-tray contactor
(Open valves)
(Example of closed
valve = —)
Foam, froth, or
y v (Sy (
) )')) liquid droplets
Sieve-tray contactor c
Vapor risers s / */ / s
Gas in
Liquid out
Figure 15-2
Cross-sectional view of a finite-stage contactor column in operation
showing an example of a sieve tray, a valve tray, and a bubble-cap tray
are well established, entailing low risk. In addition, the trays are low in cost relative to
many other tray types while handling wide variations in flow rates. Overall tray effi-
ciencies of sieve trays generally are between 60 and 85 percent. Disadvantages of these
trays are their higher pressure drops and lower capacities relative to some of the newer
Table 15-5 Sieve tray geometry
Parameter Range
Hole diameter, m 0.005-0.025

Fractional free area, nr 0.06-0.16
Fractional downcomer area, m2 0.05-0.30
Pitch/hole diameter ratio 2.5-4.0
Tray spacing, m 0.305-0.915
Weir height, m 0.025-0.075
Traditional random packings
151
Raschig ring Intalox saddle Berl ring
(ceramic) (ceramic) (ceramic)
Newer random packings
^3
Pall® ring Intalox® saddle Cascade® MiniRing

(metal) (metal) (metal)
Figure 15-3
Examples of random packings
trays and structured packings. Typical sieve tray geometries, including guidelines for
hole dimensions, free areas, and downcomer areas, are provided in Table 15-5.
Examples of random packings are shown in Fig. 15-3. The newer random packings
have thinner elements and therefore require less of the column volume than traditional
packings. A variety of materials are available for random packings with both advantages
and disadvantages. For example, stoneware is susceptible to attack by alkali and hydro-
fluoric acid. When metals are used, there is a concern for wettability of the surface and
the possibility of high corrosion rates. Plastic packings are lightweight, are easy to
install, provide a low pressure drop per theoretical stage, are low-cost, and corrosion is
not a problem. However, such packings can exhibit wettability problems and normally
experience an upper temperature range. The geometric characteristics and the height
Table 15-6 Geometry and efficiency of several random packings+
Surface area per Approx.

Type of packing Void fraction volume, m2/m3 HETP, m
25-mm Ceramic Raschig rings 0.73 190 0.6-0.12

25-mm Ceramic Intalox saddles 0.78 256 0.5-0.9
25-mm Ceramic Bed saddles 0.69 259 0.6-0.9
25-mm Plastic Pall rings 0.90 267 0.4-0.5
25-mm Metal Pall rings 0.94 207 0.25-0.3
50-mm Norpac® 0.94 102 0.45-0.6
50-mm Highflow® rings 0.93 108 0.4-0.6
'Additional values are available in M. S. Peters and K. D. Timmerhaus, Plant Design and Economics for
Chemical Engineers, 4th ed., McGraw-Hill, New York, 1991, p. 690.
Table 15-7 Geometry and HETP of several structured packings1
Surface area per Approx.

Type of packing Void fraction volume, m2/m3 HETP, m
Intalox 2T (Norton) 0.96 213 0.2-0.3

Flexipac® 1 (Koch) 0.91 558 0.2-0.3
Flexipac® 2 (Koch) 0.93 249 0.3-0.4
Gempak® 4A (Glitsch) 0.91 525 0.2-0.3
Gempak® 2A (Glitsch) 0.93 262 0.3-0.4
Sulzer BX (Sulzer) 0.90 499 0.2-0.3
' Additional values are available in H. Z. Kister, Distillation Design, McGraw-Hill, New York, 1992, p. 446.
equivalent to a theoretical plate (HETP) of several random packings are presented in

Table 15-6.
Structured packings with a geometric structure of corrugated sheets positioned in
a parallel arrangement are fabricated to fit the dimensions of the column. When placed
in the column, successive elements are oriented at 90° from each other, as shown in
Fig. 15-4. The structured packing occupies about 60 to 70 percent of the column
volume while the remaining volume is used for flow distribution and phase disengage-
ment. Structured packings are available in sheet or gauze form and in a variety of
materials. The geometric characteristics and HETP of structured packings are high-
lighted in Table 15-7.
Structured packings can provide higher mass-transfer efficiencies and capacities
than traditional tray contactors for vacuum and low-pressure applications. The effect
of this increased efficiency and capacity is a smaller required column size. The perfor-
mance of structured packings remains unaffected even at gas flow rates as low as
10 percent of the design load. In the case of absorption, another possible benefit of this
packing is a reduction in the solvent/gas feed ratio that again permits the use of smaller
equipment. However, disadvantages of structured packings are higher costs relative to
traditional trays and the difficulty of maintaining good liquid and vapor distribution
throughout the column.
The choice between use of a tray column or a packed column for a given mass-
transfer operation should, theoretically, be based on a detailed cost analysis for the two
Overhead vapor
Reflux from
condenser
Structured
packing
Liquid
collectors
Feed
Liquid
distributor
Inlet pipe
Column
sump
Circulation pipe
to the reboiler
Figure 15-4
\7 Distillation column equipped with struc-
Bottom product tured packing
types of contactors. Thus, the optimum economic design for each type would be de-
veloped in detail, and the final choice would be based on a consideration of costs and
profits at the optimum conditions. In many cases, however, the decision can be made
on the basis of a qualitative analysis of the relative advantages and disadvantages,
eliminating the need for a detailed cost comparison. The following general advantages
and disadvantages of tray and packed columns should be considered when a choice
must be made between the two types of contactors:
1. Stage efficiencies for packed towers must be based on experimental tests with each
type of packing. The efficiency varies not only with the type and size of packing,
but also with the fluid rates, fluid properties, column diameters, operating pressure,
and, in general, extent of liquid dispersion over the available packing surface.
2. Because of liquid dispersion difficulties in packed columns, the design of tray
columns is considerably more reliable and requires a lower safety factor when the
ratio of liquid mass velocity to gas mass velocity is low.
3. Tray columns can be designed to handle wide ranges of liquid rates without
flooding.
4. If the operation involves liquids that contain dispersed solids, use of a tray column
is preferred because the plates are more accessible for cleaning.
5. Tray columns are preferred if interstage cooling is required to remove heats of re-
action or solution because cooling coils can be installed on the plates, or the liquid-
delivery line from plate to plate can be passed through an external cooler.
6. The total weight of a dry tray column is usually less than that of a packed column
designed for the same duty. However, if liquid holdup during operation is taken
into account, both types of columns have about the same weight.
7. When large temperature changes are involved, as in distillation operations, tray
columns are often preferred because thermal expansion or contraction of the
equipment components may crush the packing.
8. Design information for tray columns is generally more readily available and more
reliable than that for packed columns.
9. Random-packed columns generally are not designed with diameters larger than
1.5 m, and diameters of commercial tray columns are seldom less than 0.67 m.
10. Packed columns prove to be less expensive and easier to construct than tray
columns if highly corrosive fluids must be handled.
11. Packed columns are usually preferred if the liquids have a large tendency to foam.
12. The amount of liquid holdup is considerably less in packed columns.
13. The pressure drop through packed columns may be less than the pressure drop
through tray columns designed for the same duty. This advantage, plus the fact that
the packing serves to lessen the possibility of column wall collapse, makes packed
columns particularly desirable for vacuum operation.
Diameter Evaluation for Columns with Sieve Trays Determination of the column
diameter first requires calculation of the net vapor (gas) velocity at flood conditions Vnf
in the column from
(15-7)
where Csh is the Souders and Brown factor at flood conditions in m/s, a is the surface
tension in dyne/cm, and pL and py are the densities of the liquid and vapor streams in
the column, respectively. The Souders and Brown factor is obtained from Fig. 15-5
after specifying a reasonable tray spacing. Standard tray spacings for large-diameter
columns are generally either 0.46 or 0.61 m, but 0.3- and 0.91-m spacings are also
used.
J
—r |
["ray spacin 8
/ u0 - n m <36 n.
/(). 51 in (24 n.)
/(). 46 m (18 ir!.)_
—
0.31 m (12 in.) // -1

0.23 m (S in.)
0.15 m (t in. /
10~2 to-1 1 10
m
(L/V)(pv-p,)
Figure 15-5
Chart for estimating values of Csb (±10 percent) in Eq. (15-7). [Adapted from
J. R. Fair, Petro/Chem. Eng., 33(10): 45 (1961) with permission.]
The actual vapor velocity Vn is selected by assuming that it is 50 to 90 percent of

the net vapor velocity at flood conditions. (The lower and upper values are only se-
lected when some unusual flow conditions are expected in the separation process.) The
net column area is then obtained from
An = (15-8)
'n
and
Ac = An + Ad (15-8r/)
where An is the net column area used in the separation process, A(l the downcomer
area, Ac the cross-sectional area of the column, m'v the volumetric flow rate of the
vapor, and Vn the actual vapor velocity. The column diameter is then calculated from
D = (4A£./7r)1/2.
Height Evaluation of Columns with Sieve Trays Conversion of the equilibrium

stages to actual stages requires the use of an overall tray efficiency. The most rigorous
method begins with point efficiencies and then converts these efficiencies to an overall
tray efficiency. Since this approach is not practical for multicomponent mixtures, a
simple analytical expression, developed by Lockett,1 is recommended when no sup-
porting experimental data are available, namely,
E0 = QA92[^L(cLK/HK)„rOM5 (15-9)
+
M. J. Lockett, Distillation Design Fundamentals, Cambridge University Press, Cambridge, United Kingdom,
1986.
where E0 is the overall tray efficiency, {oiLk/hk)w the average relative volatility be-
tween the light and heavy keys, and /x/. the liquid viscosity of the feed mixture. Since
the plate spacing was selected in a prior design step, all the information is now avail-
able to determine the height of the column exclusive of the height required for phase
disengagement and required internal hardware. Thus,
Hc = (/Vact - \)HS + A// (15-10)
where Hc is the actual column height, rVact the actual number of trays, Hs the plate spac-
ing, and A// the additional height required for column operation as noted above.
Determination of Diameter and Height of a Distillation Column
A sieve tray distillation column is to be used to separate a propane/propylene mixture. The multi-
component feed to the column is completely vapor at a temperature of 43.30C, a pressure of
1668 kPa, and a composition as shown below.
Component /h, kg/s Component m, kg/s
Acetylene 1.3 x lO-6 MAPDt 5.4 x lO-4

Ethylene 1.3 x lO-6 1,3-Butadiene 3.7 x lO-4
Ethane 3.3 x lO-4 /-Butene 1.9 x lO-4
Propylene 6.41 1-Butene 1.2 x lO-4
Propane 4.62
'MAPD refers to the C, acetylenes (Cjff,), methylacetylene and propadiene.
It is desired to obtain a product with a propylene concentration of 99.6 mol percent. Determine the
diameter and height of the column if a tray spacing of 0.762 m is specified.
■ Solution
A review of the feed composition indicates that the separation is essentially a separation between
propane and propylene. However, use of the McCabe-Thiele procedure could introduce considerable
inaccuracies because of the difficulty in the graphical evaluation. Therefore, the relations for separating
multicomponent mixtures will be used with a number of assumptions to minimize the calculations.
The first step is a material balance around the column. Since the other components in the feed
are negligible relative to the two key components, assume only these two components in the material
balance.
F= B+ D
^(C3H6)F = £*(C3H6)fi + f>>'(C3H6)D
^V(C,H8)F = flV(C3H8)B + O.V(C3Hs)d
where F = 0.257 kg mol/s, V(c3h6)f = 0.5924. and *(c3H6)fl = 1 - -*(c3H8)s . also
y(C3H6)D = 0.996 y(c3H8,0 = 0.004 B — F— D
Solving these equations simultaneously provides the following results:
D = 0.1527 kg mol/s B = 0.1043 kg mol/s
V(c3H6)fi = 0.0022 V(C3H8)b = 0.9978

To obtain the relative volatility between the light key (propylene) and the heavy key (propane),
assume an average temperature of 43.30C or 316.4 K. From vapor pressure data provided by the
International Critical Tables, the relative volatility of propylene to propane at this temperature is
1.136. A slightly more accurate value could have been obtained with Eq. (15-la) by using the bubble
point temperature of the bottoms and the dew point temperature of the distillate to calculate the geo-
metric mean of the relative volatility.
Determine the minimum number of stages with Eq. (15-1).
ln[(xLK /xhk)D(XHK /XLK)B]

Nmin —
lnaor/tf/r
ln[(0.996/0.004)(0.9978/0.0022)]
= 91.2 stages
In 1.136
To obtain Rmin, use Eqs. (15-2) and (15-3).
CjXF.i
— I —q let t? = 0 since feed is all vapor
E a, - 0
i=i
1.136(0.5924) + 1(04076) =:J 0 = ^
1.136-0 1 -0
CtiXpj
^min + ' — ^
i = l Cf, - 0
1.136(0.996) + 1(0.004) = ^
1.1336 -1.0555 1-1.0555
tfmin = 12.98 assume R — \.2Rm\r
R 1.2(12.98) = 15.58
The actual number of theoretical stages is determined with Eq. (15-4).

0.566
n - yvmin R - Rmm
= 0.75 1 -
N+ 1 R+ 1
0.566
N — 91.2 15.58 - 12.98
= 0.75 1 -
N+\ \ 15.58+1
N -91.2 = (A + 1)(0.487)
N = 179 theoretical stages
The location of the feed is obtained from Eq. (15-6).
Np X'HK (xlk)b
log — = 0.206 log
XLK / F .(xhk)D.
0.1043 0.4076 \ / 0.0022

= 0.206 log
0.1527 /V 0.5924 0.004
N = Nd + NI)
Solving these two equations results in
No — 107 theoretical stages NB = 72 theoretical stages
Thus, the feed enters the column 72 theoretical stages above the bottom stage.
Equation (15-7) is used to obtain the net vapor velocity at flood conditions. Normally, the net
vapor velocity should be obtained at both ends of the column to determine the maximum column
diameter that will be required. In this design, the maximum diameter will occur at the top of the col-
umn, and only this calculation is pursued. Equation (15-7) requires information on the liquid and
vapor densities as well as the surface tension. Since the distillate is essentially pure propylene, the
liquid and vapor densities of this component are required. Liquid density is obtained from Table 2-30
of Perry's Chemical Engineers' Handbook, 7th edition, as 11.309 kg mol/m3 or 476 kg/m3. Vapor
density is obtained with the aid of the ideal gas relationship as 27 kg/m3. The surface tension for
propylene has been estimated as 6 dyne/cm. Once the function (L/ V)(pv/pL)0-5 has been estab-
lished, Fig. 15-5 can be used to obtain Csb.
0.5 / ic co\ / n \ 0.5
o© ysxii) —
From Fig. 15-5, Csb = 0.075 m/s.

With Eq. (15-7)
v 0.2 / \ 0.5
« \ / PL - PV
V,,/ = Csh
20/ V Pv
0.2 / 4-7^ n\ 0.5
, _ , ,'476 — 27V
= (0.075) ( — I ( —— j = 0.240 m/s
•m
Assume 80 percent of flooding.
Vn = 0.8V,,/ = 0.8(0.240) = 0.192 m/s
From Eq. (15-8),
mv
A,, =
V,,
(16.58) (0.1527) (42.08)/27 ^^ 2

— = 20.55 nr
0.192
Assume the downcomer occupies 15 percent of the cross-sectional area of the column. Thus,
A,, 20.55 ,
Ac — —— = = 24.2m2
0.85 0.85
and
1/2 1/2
'4 A, A 4(24.2)
D = = 5.55 m
7T 7t
Since the number of theoretical stages is too large for one column, two columns will be required, each
with 90 theoretical stages and a column diameter of 5.6 m.
782
EXAMPLE 15-3 Computer Determination of Diameter and Height

of a Distillation Column
With the data provided in Example 15-2 for separating a propane/propylene mixture, use one of the
commercially available simulators to determine the required diameter and height of the distillation
column that will provide the same high-purity product as specified.
■ Solution
The operation of the propane/polymer distillation column was simulated with the SCDS (Chemstations)
model using the Soave-Redlich-Kwong equation of state. The reflux ratio used will also be 1.2/?min.
The feed location has been optimized to provide the lowest number of trays required. A tray spacing of
0.762 m was specified to maintain the flood percent below 80 percent (similar to that selected in
Example 15-2). For this separation, the relative volatility varies between 1.132 and 1.14. It was
assumed that the separation required with this low a relative volatility and high product purity would
necessitate a two-column arrangement. This was verified in the simulation results.
The temperature profile of the propane/propylene two-column system as a function of the theo-
retical stage number prepared by the simulation program is shown in Fig. 15-6 to emphasize the small
temperature changes that occur from one stage to the next. The design data provided by the program
are shown here. '
50
49
48
47
46
u
o
1) 45
5
C3
•_ 44
D. Feed stage Second column
£ 43
.u
H
42
41
40
39 First column
38 _L
30 60 90 120 150 180
Theoretical stage number
Figure 15-6
Temperature profile for the propylene/propane two-column
system as a function of theoretical stage number. (From R. E.
Treybal, Mass Transfer Operation, McGraw-Hill, 1980. Re-
produced with permission of The McGraw-Hill Companies.)
Design data
Selected design parameters from simulation
Number of columns 2
Number of theoretical trays 180 (valve trays)
Average tray efficiency 0.85
Tray spacing 0.762 m
Column internal diameter 5.33 m
Column reflux ratio 15.563
Column cross-sectional area 21.86 m2
Column active area 17.85 m2
Percentage of flood at bottom stage 74.7%
Percentage of flood at top stage 79.4%
Pressure drop per tray 0.77 kPa per tray
It is apparent that the results from the computer simulation provide excellent agreement with the
hand calculations. Generally such good agreement is not obtained; rather, there can be as much as a
±25 percent difference in the height and diameter calculations, particularly with the separation of
multicomponent mixtures exhibiting a large degree of nonideality. In such cases, a good simulation
program will provide a much better approach to actual operating conditions.
Distillation Design Procedures for Columns with

Random Packing
As noted in Fig. 15-1, the initial procedures for designing a new distillation column are
similar when either a trayed column or a packed column is selected for the separation
process. Differences in the design calculations become apparent when the column di-
ameter and height need to be established. Accordingly, the following will describe the
design procedures associated with the diameter and height aspects of those columns
utilizing random packing.
Diameter Evaluation for Columns with Random Packing In contrast to distilla-

tion columns with trays, performance in packed columns is strongly affected by both
liquid and vapor rates in the column. Not only is flow limited, but at high throughputs
the gas flow impedes the liquid flow, which can eventually lead to flooding of the col-
umn. Thus, packed columns also operate at a vapor velocity that is 70 to 90 percent of
the flooding velocity. A simplified shortcut method that has been the standard of the in-
dustry for several decades utilizes the generalized pressure drop correlation chart orig-
inally developed by Sherwood' and improved by several other investigators.§ One
Considerable additional design parameters are provided by this simulation program, including extensive design
data on liquid and vapor velocities, weir details, downcorner specifications, etc.
^ E. Treybal, Mass Transfer Operations, 3d ed., McGraw-Hill, New York, 1980, p. 195.
Tor example, see M. Leva, Ghent. Eng. Prog.. 88(1); 65 (1992).
Packing Packing
size, mm factor Fp
Approximate flooding 16 100

25 50
200
38 40
50 25
400
o. 75 18
KOI)
200
S
Q. 100
10"
Ml
o
Gas pressure drop Pa/m
10"
10-'
(LMl/VMv) \pv/{pL - pv)\[1/2
where
G = VMv/Ac, superficial gas mass flux, kg/(s*m2) L = liquid flow rate, kg mol/s
Fp = packing factor, dimensionless; see insert Ac = column cross-sectional area, m2
l±i = liquid viscosity, Pa-s Mv = molecular weight of vapor, kg/kg mol
pv = gas density, kg/m3 ML — molecular weight of liquid, kg/kg mol
Pi = liquid density, kg/m3 V = gas flow rate, kg mol/s
Figure 15-7
Flooding and pressure drop correlations for packed columns. (From R. E. Treybal,
Mass Transport Operations, 3d ed., McGraw-Hill, New York, 1980, p. 195 with
permission of McGraw-Hill.)
version of this correlation chart, shown in Fig. 15-7, permits the designer to estimate
the cross-sectional area of the column after selecting a recommended pressure drop per
unit height of packing. Recommended pressure drops in packed columns for atmos-
pheric and high-pressure separations range from 400 to 600 Pa/m, for vacuum opera-
tion between 4 and 50 Pa/m, and for absorption and stripping columns between 200
and 400 Pa/m.
The parameters involved in this estimation of the cross-sectional area Ac, are L, F,
pL, pv, pui, and Fp. In Fig. 15-7, L and V are the liquid and vapor mass flow rates,
respectively, pi and py are the liquid and vapor densities, respectively, pci is the liquid
viscosity in Pa-s, and F,, is the packing factor associated with the random packing
selected. The latter parameter is normally available from the supplier of the packing.
If such information is not available, an approximation relating packing size (diameter
in mm) to Fp for some of the more widely used packings is displayed in the upper
right-hand comer of Fig. 15-7. Raschig rings and Berl saddles are excluded from this
approximation.
The cross-sectional area for the column can be evaluated directly from the value
obtained for the ordinate. The column diameter again is given by D = (4Ac./7r)1/2.
Height Evaluation for Columns with Random Packing Determination of the

height of packing required in a column to achieve a specific separation involves use of
either the height of a transfer unit (HTU) or the height equivalent to a theoretical plate
(HETP) approach. To obtain the total height using the first approach also requires one
to evaluate the number of transfer units (NTU) to satisfy the relation
Z = (HTU) (NTU) (15-11)
where Z is the total height of the mass-transfer zone. HTU and NTU are defined as
m"
HTU = ^ (15-12)
(KGCie)Ac
and
dy
NTU = /r — (15-13)
J
4 y-
where tn" is the molar flow rate of vapor, KG the overall mass-transfer coefficient, ae
the area of interfacial contact between the liquid and vapor phase per unit volume of
contactor, Ac the cross-sectional area of the column, y the mol fraction of the compo-
nent in the vapor phase, and y* the mol fraction of the component in the vapor phase
that would be in equilibrium with that component in the liquid phase.
To include the contribution of both the gas- and liquid-phase resistances, the HTU
is defined as
Uv Vi
HTU = HTU v + A.HTU i = —-—h A —— (15-14)
kyae kiae
where Vv and V, are the superficial velocities of the vapor and liquid, respectively, kv
and k/ the individual mass-transfer coefficients for the vapor and liquid expressed in
units of velocity, respectively, and A. the ratio of the slope of the equilibrium line to the
slope of the operating line L/V.
A number of design procedures for evaluating the HTU for packed columns are
available in the literature. For example, the approach by Bravo and Fair' involves esti-
mating the individual mass-transfer coefficients and an effective interfacial area for
mass transfer. The mass-transfer coefficients established by these investigators, in units
of m/s, are obtained from
O.ll^DvRe^Sc0/33
1J. L. Bravo and J. R. Fair, Ind. Eng. Chem. Proc. Des. & Dev., 21: 162 (1982).
and
/ X 0.333 / D X 0.667
5
^ = 3.72x 10- (^) (^) Scfs(apDpf* (15- 16)
V Pl / V aw /
where
0.75 -1
a,
1 — exp -1.45 Re^Fr-^WeJ2^ (15-17)
a,
1488/? Vv
Rev = (15-18)
dplLy
1488 plFZ.
Re/ = (15-19)
Cpl-lL
a_pVl
Fr, = (15-20)
g
453.2pLVl
We, = (15-21)
a pa
0.000672/1;
Sc/ = (15-22)
PLOl
Note that subscripts L and V refer to the liquid and vapor phases. Other terms with their
units include the packing surface area ap m2/m, the respective diffusion coefficients Dv
and Dl m2/s, the packing diameter Dp m, the area of wetted packing aw m2, the
respective densities pp and pv kg/m3, the respective viscosities and pv cP, the
surface tension a dyne/cm, and the critical surface tension oy, which is assumed to be
61 dyne/cm for ceramic packing, 75 dyne/cm for structured packings, and 33 dyne/cm
for PEB packing. The terms Fr, We, and Sc refer to the Froude, Weber, and Schmidt
dimensionless numbers, respectively.
Determination of the individual mass-transfer coefficients now permits evaluation
of the effective area a,, for mass transfer from
ae = apaa5(CaLRev)a392HTUa4 (15-23)
where the dimensionless capillary number Ca, is obtained from
0.304 VLixL
Ca/ = (15-23a)
a
The HTU^, HTU^, and HTU can now be determined from Eq. (15-14).
Even though the HTU approach is more rigorous, many designers elect to use
the HETP approach because it provides a comparison with the number of theoretical
stages determined with trayed columns. The relationship between HETP and HTU is
given by
HETP = HTU ^ (15-24)

and the total height of the packing is obtained from
Z = (HETP)/V (15-25)
where N is the number of equilibrium stages required. Since evaluation of HETP gen-
erally involves determining the HTU, the preferred method is to obtain HETP values
from experimental or plant data. For example, Kister' presents HETP values for a wide
range of packing used in industrial separations. If such data are not available, the same
author has also provided some helpful rules of thumb for predicting the HETP values
for columns with random packings in terms of the column diameter obtained earlier,
namely,
D for D < 0.5 m (15-26a)
HETP = 0-5D03 for D > 0.5 m (15-26/?)
D03 for absorption columns (15-26c)

with D > 0.5 m
For vacuum distillation, it is recommended that an extra 0.15 m be added to these pre-
dicted HETP values.
Distillation Design Procedures for Columns

with Structured Packing
Diameter Evaluation for Columns with Structured Packing A procedure similar

to that used for determining the diameter of a column with random packing may be
used to obtain the column diameter with structured packing. If Fig. 15-7 is used to ob-
tain the cross-sectional area of the column, the designer will need to obtain a suitable
packing factor from the vendor for the structured packing selected.
Another procedure, developed by Kister and Gill, ' involves calculating the pres-
sure drop over a unit length of the column under flood conditions. A correlation, sim-
ilar in format to Fig. 15-7, is used to determine the flood velocity in columns with
structured packing. The operating velocity in the column is designated as a fraction of
the flood velocity. The cross-sectional area of the column is then obtained by dividing
the volumetric flow rate of the vapor by the operating velocity. Column diameter is
obtained from D = (4y\(./7r)1/2.
Height Evaluation for Columns with Structured Packing The total height of
packing required to achieve a specific separation again necessitates determination of
HTU and HETP. Several options are available. A rule of thumb for quick estimation of
HETP for structured packing has been presented by Harrison and France5
9.29
HETP = + 0.10 (15-26J)
t
H. Z. Kister, Distillation Design, McGraw-Hill. New York, 1992.
t
H. Z. Kister and D. R. Gill, Chem. Eng. Prog., 87(2): 32 (1991).
§
M. E. Harrison and J. J. France, Chem. Eng., 96(4): 121 (1989).
where ap is the packing surface area per unit volume in m2/m3 and the HETP obtained
is in meters. A more accurate approach would be to use the interpolation or extrapola-
tion of packing efficiency data of Kister noted in the previous section.
Rating of Distillation Equipment
As noted earlier, rating a distillation column can entail either examining performance
under new operating conditions or determining whether a retrofit of an existing column
with perhaps improved column internals could prove to be economically attractive.
For example, Humphrey and Seibert1 have shown that in vacuum distillation, the re-
placement of sieve trays with structured packing with its increased capacity is a good
return on the investment.
The steps involved in the rating of a distillation column are not unlike those used in
the design of a new distillation column. The first step must carefully detail the objective
of the rating study and provide precise information about the column including existing
internal hardware. Accurate information must also be available about feed composition,
rates, and physical properties. Equilibrium data must be obtained from experimental or
literature sources. If none are available, such data will have to be developed by using
commercially available software programs.
If the internal configuration of the column is not to be modified, the next step is to
calculate the number of equilibrium stages required for the new design specifications.
However, if the objective is to increase the capacity or the purity of the product, the
internal configuration will probably have to be changed. For example, improved purity
could be achieved by removing sieve trays and replacing them with commercial struc-
tured packing with smaller spacings between packing elements. Calculation of the
equilibrium stages can be initiated once the column modifications have been finalized.
Next, the designer must establish if the diameter of the existing column is ade-
quate to handle the new vapor and liquid rates. The procedures for evaluating the re-
quired diameter of the column are similar to those described for the design of a new
column. Likewise, the procedures for evaluating the required height of the column are
similar to those discussed earlier. If the resulting values of column diameter and height
match those for the existing column, the proposed separation can be achieved. Obvi-
ously, if the required diameter and height are larger than those available for the exist-
ing column, modifications will have to be made either in the separation objectives or
in the column configuration and operation.
Rigorous Design Methods for Multicomponent Distillation
The preceding section has outlined shortcut methods for the design of distillation
columns using either trays or packing. Since these methods included many simplifying
assumptions to reduce the calculation efforts, the results often can only be considered
as reasonable estimates. To obtain a more accurate picture of the separation process re-
quires a rigorous, stage-by-stage evaluation. The advent of high-speed digital computers
+
J. L. Humphrey and A. F. Seibert, "Separation Technologies—Advances and Priorities," Final Rept., Gas
Research Institute, ID No. 89/0005, February 1989.
has made possible solutions of equilibrium-stage models for multicomponent, multi-

stage distillation-type columns to an exactness limited only by the accuracy of the
phase equilibria and enthalpy data used. Time and cost requirements for obtaining such
solutions are low compared with the cost of hand calculations.
Over the past several decades, methods have also become available that, with ap-
propriate computer programs, can accurately solve most distillation problems quickly
and efficiently. Some of the more widely used models, namely, the BP (boiling-point)
method for distillation, the SR (sum rates) method for absorption and stripping, the
ISR (isothermal sum rates) method for liquid-liquid extraction, the SC (simultaneous
correction) procedures for distillation of wide-boiling mixtures, and the 10 (inside-
out) method that reduces the computational time of earlier methods are detailed in
Seader and Henley.1
Many of the simulators presently available offer various options which allow the
user to select not only a particular algorithm for stage calculations, but also the model
or approach to determine the vapor/liquid equilibria. Some simulators also allow the
user to incorporate tray efficiencies as a function of composition, so that the actual
number of trays can be established as well. Again, users of such simulators must have
a good understanding of distillation fundamentals to appreciate the limitations of the
methods and models used, and to recognize errors should they occur.
The deficiencies associated with the equilibrium stage models enumerated above
have been recognized for some time, and a new model has been developed for distilla-
tion and other staged processes by Krishnamurty and Taylor.5 The feature of the RBD
(rate-based design) method is that component material and energy balances for both
the liquid and vapor phases, together with mass- and energy-transfer rate equations,
and equilibrium equations for the phase interface are solved directly to determine the
actual separation.
In this design method, calculations are performed on an incremental basis as one
proceeds from stage to stage through the column. The uncertainties in the computa-
tions using average tray efficiencies in prior models based on individual components
are avoided. A major advantage of this approach is that it replaces the two-step method
of first determining the number of equilibrium stages and then using an average tray or
packing efficiency to determine the required column height. A review of this method,
as well as a comparison with the traditional equilibrium approach, has been presented
by Seader.55
Design Procedures for Other Distillation Processes
Batch Distillation Most distillation processes operate in a continuous fashion, but

there is a growing interest in batch distillation, particularly in the food, pharmaceutical,
and biotechnology industries. The advantage of this separation process is that the dis-
tillation unit can be used repeatedly, after cleaning, to separate a variety of products.
7. D. Seader and E. J. Henley, Separation Process Principles, J. Wiley, New York, 1998.
t
R. Krishnamurty and R. Taylor, A/OfE 7., 31(3): 449 (1985).
7. D. Seader, in Perry's Chemical Engineers' Handbook, 7th ed., R. H. Perry and D. W. Green, eds., McGraw-
Hill, New York, 1997, pp. 13-52 to 13-54.
Cooling fluid
Vapor
Overhead
product
tank
Steam
Condensate - I
Figure 15-8
Simple batch distillation
The unit generally is quite simple, but because concentrations are continually changing,
the process is more difficult to control.
In single-stage distillation, like that shown in Fig. 15-8, it is assumed that the
vapor and liquid are in equilibrium. If L' is the number of mols of liquid in the column
at any given time and dL' is the differential amount vaporized, a material balance on
component i yields
L'xi = (L' - dL'){xi - dxi) + (y,- + dyi)dL' (15-27)
Neglecting products of differentials and integrating over the change in L' from the ini-
tial to the final condition and from the initial to the final concentration of liquid estab-
lishes the Rayleigh equation for batch distillation, given as
dxi
In ^ = r (15-28)
yi - Xi
where L\ and Lj are the initial and final mols of liquid in the column, respectively; x^
and Xi2 are the initial and final mol fractions of component i in the column, respectively;
and xi and yi are the mol fractions of component i in the liquid and vapor, respectively,
during the batch distillation process. If the mixture to be separated is a binary one and
equilibrium data are available, the vapor mol fraction can be expressed in terms of
Xj and oiij, the relative volatility between components i and j. For a constant relative
volatility between these two components, Eq. (15-28) can be integrated to give
L' 1 Xi20 -Xn) 1 - xn

In — = In h In (15-29)
OCjj - 1 X/|(l - Xi2) 1 - xn
From this relation the designer can estimate the quantity of liquid that must be distilled
to change the composition of the remaining liquid from xn to xn-
Azeotropic Distillation A separation more sophisticated than simple distillation is

required when the components to be separated form an azeotrope. If the composition
of the azeotrope is sensitive to pressure, this property can be used to carry out the
Cooling fluid
Cooling fluid
Makeup solvent
(furfural)
X v . -V
Isobutane
product 1-Butene
Extractive product
distillation Solvent
Fresh feed tower
(isobutane plus recovery
1 -butene) tower
Steam
Recycled furfural
Steam
Pump
Figure 15-9
Example of extractive distillation separating 1-butene from isobutane
separation provided no material decomposition occurs. A change in azeotropic com-

position of at least 5 percent with a change in pressure is generally required.7
The design process for separating an azeotrope sensitive to pressure changes re-
quires the use of two distillation columns in which the azeotrope product from the
high-pressure column is fed to the low-pressure column, as shown in Fig. 15-9. The
azeotrope product from this column is recompressed and added to the fresh feed of
the high-pressure column. A major problem in using the pressure change scheme is that
the smaller the change in azeotropic composition with change in pressure, the larger
is the recycle back to the high-pressure column.
If the azeotrope is not sensitive to changes in pressure, it may be possible to add
another component to the mixture to favorably change the relative volatility of the key
components. This technique is also useful when a mixture is difficult to separate be-
cause of a fairly low relative volatility between the key components. The addition of a
relatively volatile mass-separating agent to the mixture that is to be separated forms
either a low-boiling binary azeotrope with one of the key components or, more often,
a tenary azeotrope containing both key components. Separation by the latter process is
only feasible if condensation of the overhead vapor results in two liquid phases, one of
which contains the majority of one of the key components and the other contains
most of the added component. This type of a separation process will require the use of
a decanter located after the overhead condenser to separate the two liquid phases.
t
C. D. Holland, S. E. Gallum, and M. J. Lockett, Chem. Eng., 88(3): 185 (1981).
Extractive Distillation This type of distillation operation also utilizes the addition of
a mass-separating agent to accomplish the desired separation. Here the mass-separating
agent is a relatively nonvolatile solvent. As with azeotropic distillation, the separation
is possible because the solvent interacts more strongly with one of the components than
with the other. This stronger interaction improves the relative volatility between the key
components. However, note that extractive distillation is different from azeotropic dis-
tillation since the mass-separating agent is withdrawn from the bottom of the column
with one of the key components and does not form an azeotrope with any of the other
components.
Generally, extractive distillation is more useful than azeotropic distillation since
the former does not depend on the formation of an azeotrope. This provides a greater
choice of mass-separating agents to aid in the separation process. For greatest success
with the separation, the separating agent should have a chemical structure similar to
that of the less volatile component being separated. It will then tend to form a near
ideal mixture with the less volatile component and a nonideal mixture with the more
volatile component. This, in turn, has the effect of increasing the volatility of the more
volatile component.
In those cases where the flow rate of the solvent can be varied within limits, a larger
solvent flow rate often results in a better separation. However, since the solvent is part
of the reboil in both the extraction and stripping columns, there will also be a greater
energy demand with a larger solvent flow rate. Additionally, higher solvent flow rates
increase the distillate temperature.
During the separation process there will inevitably be losses if a mass-separating
agent is used. Even if these losses in the agent are insignificant in terms of the cost of
the agent, they could create environmental problems later in the process. The best
way to solve the effluent problems caused by the loss of mass-separating agents is to
eliminate them from the design unless practical difficulties and excessive cost dictate
their use.
Equipment Costs for Tray and Packed Columns
The purchased cost for tray and packed columns can be divided into the following com-
ponents: (1) cost for column shell, including heads, skirts, manholes, and nozzles; (2) cost
for internals, including trays and accessories, packing, supports, and distributor plates;
and (3) cost for auxiliaries, such as platforms, ladders, handrails, and insulation.
The cost for fabricated column shells is estimated on the basis of either weight or
the diameter and height of the column. The former estimate is often preferred to more
accurately account for the effect of pressure on the wall thickness of the column shell.
Figure 15-10 presents the cost based on column weight and Fig. 15-11 on the basis of
diameter and height. This cost is for the column shell only, without trays, packing, or
connections.1 The cost for installed connections, such as manholes and nozzles, may
be approximated from the data presented in Fig. 15-12.
+
See also the information presented in Chap. 12 on costs for tanks, pressure vessels, and storage equipment along
with information on design methods for tanks and pressure vessels.
Weight, lb
1(1 104 10
10 I - I I -I
316 Stainless steel
10
304 Stainless steel
' i i Carbon steel
—
^ ,04
10
Jan. 2002
1(1 10 10
Weight, kg
Figure 15-10
Purchased cost of columns and lowers. Costs are for shell with two heads and
skirt, but without trays, packing, or connections.
Vertical height of column, ft

0
106
Pressure adjustment factors Material adjustment factors
1,035 kPa (150 psia) Carbon steel 1.0
5,000 kPa (725 psia) 316 Stainless steel 3.0
10,000 kPa (1,450 psia) Nickel alloy 7.4
20,000 kPa (2,900 psia)
30,000 kPa (4,350 psia) 12.2
10 40,000 kPa (5,800 psia) 15.8
D = 4 in
o
o
-o
0)
(/,
C3
4=
P D=2m
10 t) = 1 m
— D = 0.3 m
Jan. 2002
J 1 _l
10
Vertical height of column, m
Figure 15-11
Purchased cost of vertical columns. Price does not include trays, packing, or connections.
Column wall thickness, in.

10
105
c
o
•s
o
<D
c
C
O Flanged manholes.
o
E 0.76-m (30-in,) internal diameter
Flanacd manliol 10
"o 0.46-m (18-in.) internal diameter
-a
Flanged nozzles
|
o
<D
C 1(1
B —
O :Couplings, 0.025- to 0.051-m (I - to 2-in.)
o
internal diameter H 10
&
w
C/3
o
u
Jan. 2002
102
10 -3 I0-2 10 -I
Column wall thickness, m
Figure 15-12
Installed cost of steel column connections. Values apply to 136-kg (300-lb)
connections. Multiply costs by 0.9 for 68-kg (150-lb) connections and by 1.2 for
272-kg (600-lb) connections.
Column diameter, ft
I 10
10 s
Quantity factors apply
to all types of trays
1 3.00 8 1.80 15 1.25
2 2.80 9 1.65 16 1.20
3 2.65 10 1.50 17 1.15
4 2.50 11 1.45 18 1.10 Bubble-cap trays
I04 5 2.30 12 1.40 19 1.05 (stainless steel)!
C- 6 2.15 13 1.35 20-29 1.00
7 2.00 14 1.30 30-39 0.98
40+ 0.97 Valve trays
(stainless steel)
Sieve trays Bubble-cap trays
or
(stainless steel) (carbon steel)
= 10 Sieve trays ■
Stamped turbognd
trays(stainless steel) (carbon steel)
Valve trays
(carbon steel)
Jan. 2002
Column diameter, m
Figure 15-13
Purchased cost of trays in tray columns. Price includes tray deck, bubble caps,
risers, downcomers, and structural-steel parts.
Separation by Distillation
Tray costs are shown in Fig. 15-13 for conventional installations. The purchased
costs for slotted-ring and high-efficiency saddle packings including column internal
supports and distributors are given in Fig. 15-14. For rough estimates, the cost for a
distributor plate in a packed column may be assumed to be the same as that for one
bubble-cap tray. The complete installed costs of various types of distillation columns
are presented in Fig. 15-15. Figure 15-16 provides complete installed costs for packed
columns. Both cost estimates cover the costs of all components normally associated
with both types of columns, as outlined above. The installed cost of several industrial-
type insulations, all with aluminum jackets, for columns and tanks may be approxi-
mated from Fig. 15-17.
Packed height, ft
10 10
i of' i i i i i i
Inside diameter of column, m

5
io
oe
c/}
C
0
1 io4
c3
sz
CJ
Im
3
CL 1.5
ID3
Packing material adjustment factors

Carbon steel 1.2
Polyethylene 1.0
Porcelain 1.0
304 Stainless steel 2.2
316 Stainless steel 4.2
Jan. 2002
102 L
1 10 102
Packed height, m
Figure 15-14
Purchased cost of stacked-ring and high-efficiency saddle packing (price includes
column internal support and distribution)
Diameter, in.
>>
C3
C- /
Bubble-plate lowers
« 10'
— Stainless steel
c
T3 Copper
Carbon steel
c
u Steel shell, stainless-steel trays
Sieve-tray lowers
Turbognd tray towers
Jan. 2(K)2
Diameter, m
Figure 15-15
Purchased cost of distillation columns including installation and auxiliaries
Diameter, in.
i ^ i
7
1
^4-
o
s
Packed towers
o Glass-lined steel
-a
Stainless steel _
o Haves I
U
Carbon stee
Jan. 2002
10
Diameter, m
Figure 15-16
Purchased cost of packed columns including installation and auxiliaries
Separation by Absorption and Stripping
Thickness, in.
10
103 1 III 1 1
1 —r i ill • i— i
-
-
-
-
^ Cork _ 60
■a -o
Foamglass o
1(SI
.= ^ Fiberglass
u-
O
O 0
£ I02 10 >2
£ - 1
c-: -
§- - o-
-
u - b
a. a.
_ c
o U
U
All with al uminum acket

(use for eq uipmenl only) J an. 20 02
10
10 -2 10-'
Thickness, m
Figure 15-17
Installed cost of various industrial insulations for columns, towers, and tanks
SEPARATION BY ABSORPTION AND STRIPPING
Absorption and stripping processes are similar to the distillation process. In the latter,
the mass transfer is between the liquid and the vapor in each stage throughout the col-
umn. In absorption and stripping, the mass transfer occurs between a gas phase and the
liquid phase on each stage throughout the column. In absorption the solute, or compo-
nent to be absorbed, is transferred from the gas phase to the liquid phase. In stripping,
the reverse is true; that is, the transfer is from the liquid phase to the gas phase.
Absorbers are often coupled with strippers to provide recovery of the adsorbent, as
shown in Fig. 15-18. Since the stripping process is not perfect, the absorbent recovered
contains components that were present in the fresh feed to the absorber. When water
is used as the absorbent, it is many times more common to recover the absorbent by
distillation than by stripping.
The equipment used for absorption and stripping is very similar to that used for
distillation except that no reboiler or condenser is required in the process. The separa-
tion process is conducted with trayed columns, packed columns, as well as spray
towers and centrifugal contactors. The column internals for trayed columns preferably
include sieve and valve trays. The packed towers use random and structured packing
similar to the geometric configurations noted earlier in Figs. 15-3 and 15-4.
In most applications the choice of the absorber is between a trayed column and a
packed column. The latter is generally favored when the requirements indicate a col-
umn diameter less than 0.65 m and a packed height of less than 6 m. Packed towers
should also be considered if the separation process involves corrosive fluids or if there

773-715

Uploaded by

Copyright:

Available Formats

773-715

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

773-715

Uploaded by

Copyright:

Available Formats

CHAPTER

Design and Costs

SELECTION OF SUITABLE SEPARATION

'G. E. Keller, AlChEMonogr. Set:. 83: 17 (1987).

Table 15-1 Property differences associated with various separation

Exploitable property differences Related separation processes

Vapor pressure Distillation

Guidelines for the Separation Process Selection

Distillation The creation or addition of another phase in distillation is obtained by

Absorption A widely used alternative to distillation for the separation of a solute

Stripping Once a solute is dissolved in a solvent, it is often necessary to separate the

Extraction Liquid-liquid extraction is a process often selected to separate azeotropes

J. M. Douglas, Conceptual Design of Chemical Processes, McGraw-Hill, New York, 1988.

Crystallization Under certain operating conditions, dissolved materials in solution

External Field or Gradient Separation As noted earlier, separations utilizing ex-

Centrifugation Separation by applying a centrifugal force field was noted earlier

Evaporation In separation technology, evaporators may be considered as liquid dry-

EXAMPLE 15-1 Identification of Feasible Options to Obtain a High-Purity

Property Propane Propylene

Molecular weight, kg/kg mol 44.096 42.081

EQUIPMENT DESIGN AND COSTS FOR

As noted earlier, the separation of homogeneous mixtures requires the formation or

Distillation Design Procedures for Columns with Sieve Trays

Calculate column Calculate column Calculate column

Calculate column Calculate column Calculate column

Determination of Number of Equilibrium Stages Even though shortcut and stage-

(^LK/HK)a\ = l((*LK/HK)D(<XLK/HK)ls]{/~ (15-lfl)

Table 15-5 Sieve tray geometry

Hole diameter, m 0.005-0.025

Traditional random packings

Newer random packings

Pall® ring Intalox® saddle Cascade® MiniRing

Table 15-6 Geometry and efficiency of several random packings+

Surface area per Approx.

25-mm Ceramic Raschig rings 0.73 190 0.6-0.12

Table 15-7 Geometry and HETP of several structured packings1

Surface area per Approx.

Intalox 2T (Norton) 0.96 213 0.2-0.3

equivalent to a theoretical plate (HETP) of several random packings are presented in

0.31 m (12 in.) // -1

The actual vapor velocity Vn is selected by assuming that it is 50 to 90 percent of

Height Evaluation of Columns with Sieve Trays Conversion of the equilibrium

Hc = (/Vact - \)HS + A// (15-10)

Determination of Diameter and Height of a Distillation Column

Component /h, kg/s Component m, kg/s

Acetylene 1.3 x lO-6 MAPDt 5.4 x lO-4

'MAPD refers to the C, acetylenes (Cjff,), methylacetylene and propadiene.

^(C3H6)F = £*(C3H6)fi + f>>'(C3H6)D

^V(C,H8)F = flV(C3H8)B + O.V(C3Hs)d

where F = 0.257 kg mol/s, V(c3h6)f = 0.5924. and *(c3H6)fl = 1 - -*(c3H8)s . also

y(C3H6)D = 0.996 y(c3H8,0 = 0.004 B — F— D

Solving these equations simultaneously provides the following results:

D = 0.1527 kg mol/s B = 0.1043 kg mol/s

V(c3H6)fi = 0.0022 V(C3H8)b = 0.9978

ln[(xLK /xhk)D(XHK /XLK)B]

To obtain Rmin, use Eqs. (15-2) and (15-3).

tfmin = 12.98 assume R — \.2Rm\r

The actual number of theoretical stages is determined with Eq. (15-4).

N = 179 theoretical stages

The location of the feed is obtained from Eq. (15-6).

where F = 0.257 kg mol/s, V(c3h6)f = 0.5924. and (c3H6)fl = 1 - -(c3H8)s . also