GENERAL DESIGN ALGORITHM BASED ON PSEUDO-

EQUILIBRIUM CONCEPT FOR MULTISTAGE MULTI-

COMPONENT LIQUID-LIQUID SEPARATION
PROCESSES
Takeshi TSUBOKAand Takashi KATAYAMA*
Department of Chemical Engineering, Faculty of Engineering
Science, Osaka University, Toyonaka, Osaka, 560

Anew general method for mathematical simulation of multicomponent equilibrium stage pro-
cesses is presented. The procedure consists of solving componentmaterial balance equations
with a tridiagonal matrix algorithm and performing equilibrium calculation based on the "pseudo-
equilibrium" concept, instead of solving simultaneous equilibrium equations by some optimization
method. The present method is simple, fast and numerically stable.
The application of the method to multistage, multicomponent extraction problems is described.
Moreover, the feasibility of the modified Wilson equation to extraction problems is shown.

procedure which constitutes their "error equation",

Introduction and material balance and equilibrium equations
The calculation of liquid-liquid separation in com- similar to those of Lewis and Matheson in distillation
plex columns has become more and more important calculation. Later, Smith and Brinkley's method
in modernprocess designs. The calculation method, was modified by Hanson et al1] and Smith7\ Re-
however, has not been developed to the same degree cently, Roche5 6), Holland2}, and Tierney et al.9) used
as the multicomponent distillation calculation method. the Newton-Raphson method for solving simul-
This reason is mainly attributed to the following three taneous equations of extraction systems. The plate-
differences between liquid-liquid extraction and vapor- to-plate calculation method lacks adaptability to
liquid processes in several respects. complex columns. The procedure with the Newton-
Firstly, since two liquid phases are involved in Raphson method requires a considerable amount of
extraction, a calculation of the distribution coefficient computer storage.
between the extract and raffinate phases involves This paper presents a general calculation method
activity coefficients in two liquid phases instead of for the multistage, multicomponent liquid-liquid ex-
one liquid phase. Therefore, raffinate compositions traction problem with multiple feeds and also a solvent
in equilibrium with knownextract compositions or recovery section. The proposed method uses the
the inverse should be calculated by an iterative method. tridiagonal matrix algorithm for solving the material
This procedure is fairly complicated. Secondly, balance equations, and the "pseudo-equilibrium"
solubilities rather than enthalpies control the phase concept for the convergence of the procedure. The
rates, and enthalpy balances cannot be used to computation is simple, fast, and numerically stable.
calculate the rate profiles. Thirdly, an activity co- In addition, good applicability of the modified Wilson
efficient equation that can accurately express multi- equation12) to liquid-liquid extraction problems is
component liquid-liquid equilibrium relations is demonstrated by use of numerical examples.
seldom available.
For less than four-component systems, extraction Mathematical Model
calculations can be handled graphically using tri- For the convenience of deriving the general working
angular or mass-ratio diagrams4'7 10). For systems of equations, a hypothetical model column as shown in
more than four components, however, it is necessary Fig. 1 is considered. This column has N stages,
to resort to analytical techniques to obtain a valid which are numbered from top to bottom. Each
solution of liquid-liquid extraction problems. Smith stage in the model column is assumed to be an equilib-
and Brinkley8) proposed a plate-to-plate calculation rium stage, that is, the extract stream leaving a
Received August 25, 1975. stage is in equilibrium with the raffinate stream leav-
40 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
ing the same stage. * EXTRACT OR

In obtaining the desired flow rates and composi- M RECOVERY SOLVENT

tions, one must solve the appropriate material balance

and equilibrium relations. For each ideal stage the
following equations, which are analogous to Wang's 1 *"f J--^ui
for distillation calculations13}, are derived. 2 ] TOP PRODUCT

(1) Summation equation

Zxi3-1=0

2>«,-
(2)
(1)
i =o F . tJI
(2) Equilibrium equation Fj-T-T-
The equilibrium equation for a component/ on a Fig. 1 Model for
mathematical simula-
stage j is tion of liquid-liquid
XijJij-yafi^
(3)
or yij=KijxiS extraction column VN Ln.,
where Kij=yyyJj. The distribution coefficient Ktj is Fn I1_l_L
expressed as a function of temperature and composi-
tions in both phases.
(3) Material balance equation
I ^-N RAFFINATE
Lj_1xij_1+ Vj+tfi^+FjZij-LjXij-
Vj-yt^O (4)
Substitution of Eq. (3) into Eq. (4) results in the solvent recoveries as
following material balance equation for each com- ~(U+ U,)Xil+ F2(l
(10) -ReCi)Knxi2=0
ponent, which involves only raffinate-phase composi-
where U1=L1/RLl. So, the first-line elements in
tions.
Lj_1xij-1-(LjJr VjKi3)Xij+ Vj+1Kij+1Xij+1= -FjZij
Eq. (9) are revised as follows:
B^-iJU+UJ, C1= V,(l-Reci)Ki2, A=0
(5)
2<j<N-l The over-all material balance in Eq. (8) is also changed
to
Notice that in the top stage (y=l), L0=0, thus L^Vj+r-iVi+UO+t k=2 Fu (ll)
-(L1+ V1Kil)xil+ V,Ki2xi2= -FlZil (6)

In the bottom stage (j=N), VN+1=0, thus With above manipulation of the material balance
I^N-lXijsf_i (x^jv+VN&iNjXiN- J^NZi and equilibrium equations, system equations for the
(7)
multicomponent separation in a complex column
The over-all material balance equation from the top become 2N(M+l) simultaneous equations. The pro-
stage through a stagey is blem now is to find a set of 2N(M+l) variables of
Lj= Vi+1- Vy xij9 yi3-9 L3- and V, so as to satisfy the system equations
3 (8)
k=l
(Eqs. (1), (2) and (9)). A direct simultaneous solu-
Combining Eqs. (5)~(7), we obtain a tridiagonal tion is impossible for their nonlinear nature. Con-
matrix form as sequently, the solution is generally obtained by using
~BX d "Irxtl 1 VD1 - some iterative procedures.
A % £$2 C> 2 %i 2 -*-^2 Calculation Procedure
Aj Bo C3- xi3- =Dj
The calculation begins with the assumption of
AN_1 BN_i Gjy-i XiN_j DN_i internal phase flow rates and their compositions. It
AN BN J_xiN J [_DN __ is assumed that the key components in the raw solu-
(9)
tion feed are separated completely into extract and
raffinate phases. Thus, the initial end flow rates of
where both phases, LN and Vl9 are estimated and the re-
B1=-(L1+ V1Kn), C1= VtKti, A=-^z41 maining flow rates are linearly interpolated between
Aj=Lj_x, Bj= -{Lj+ VjKi0), Cj= Vi+1Kii+1, these assumed flow rates and end feed rates. The
initial compositions are also all provided by specify-
ing the compositions of both 1st and iV-th stages
through the same assumption above, and by allowing
If the top stage has no feed stream and is offered for linear interpolation for compositions of the remaining
a solvent recovery section, Eq. (6) is rewritten with stages.

VOL. 9 NO. 1 1976 41

 1) The raffinate-phase activity coefficients yfj are
Q START^)
^ __
READ INPUT DATA estimated with the normalized new raffinate-phase
i FOR PROBLEM
INITIALIZE
compositions xri5, thus the distribution coefficients
STREAM RATES
COMPOSITIONS
AND Kij are modified approximately, leaving the old ex-
1 tract-phase activity coefficients jjj intact,
CALCULATE Kjj Kli ^ rimdlrTWl1)
(14)

2) The extract compositions y\$ are recalculated

as follows :
,,r Jf' -y,r
(15)
| r=r+1 \* 1
3) The extract-phase activity coefficients yl3- are
|~SOLVE Eg.(10)~| now estimated with the normalized extract-phase
compositions y\h and consequently the distribution
<^Jx^j-x;-1|^^> !-^MODIFY K^-l coefficients are revised as
Kij = rtixW/rUyh)
(16)

I Yes
The criterion of Eq. (13) being satisfied, extract-phase
^^^r ^^^-^ N°I 1 flow rates Vd are renewed by the following formula:
<^^ ^l?x^j " 1l<e2^^> åº CORRECT ^
^^^^^^1" ^^^^^ STREAM RATES
ViZrti - Vj
(17)

I Yes
and raffinate flow rates Lj are corrected by Eq. (8) or
print final I Jisn Eqs. (ll) and (12), with the corrected Vd. This pro-
RESULTS ! 1
cess terminates as soon as the following criterion is
Fig. 2 Flow chart for liquid-liquid extraction calculation satisfied :
ZIS*?,~-l|<e»
Whenthe separation column has a solvent separa- 3 i (18)

tion section, all extract stream rates Vj are assumed The flow chart for this proposed method is given in
in the manner mentioned above, but V± is revised, Fig.2.

say, as 0.9 times of F2. Raffinate flow rates Lh ex- In brief, as seen in Fig. 2, this procedure has two
cept Ll9 are assumed by Eq. (ll) instead of Eq. (8), convergence routines; the inner routine mends dis-
and Li is evaluated through the following equation tribution coefficients under the condition of fixed
taking into account the reflux ratio RL 9 internal flow rates, and the outer one revises the
L1=(V2- V1W+l/RLl) (12) internal flow rates.
The initial set of activity coefficients is computed on
the basis of the assumed compositions of xi3- and yi3. Illustrative Examples
To illustrate the design algorithm, four problems
After this preparation, Eq. (9) is solved
manner for xi3- with fixed-phase
in the iterative
flow rates and suc- are considered. Example 1 is the same problem
cessively improved distribution coefficients, until the examined by Hanson et al.1] and Holland2}. Ex-
difference between the last compositions xlj1 and the ample 2 is the modified form of Example 1, where
new ones xri3-on each stage satisfies the following the separation column with a solvent separation
criterion : section is treated. These two problems are used to
show that the proposed argorithm is effective in
Ti \x\Jx-xrij\ <e1 (13) extraction design calculations. The latter two pro-
The improvement of distribution coefficients is blems, Example 3 and Example 4, demonstrate that
effected based on the concept of "pseudo-equilibrium". the modified Wilson equation is applicable with suf-
The concept is derived from understanding that a ficient accuracy to the calculation of liquid-liquid
computer is skilful in evaluating function values by separation systems.
setting variables but is weak in the reverse. The Example 1
"pseudo-equilibrium" concept amends distribution
A binary solution of acetone and ethanol is to be
coefficients through the following three steps : separated by extraction in a countercurrent column
As Oliver3) pointed out, the Margules equation of Hanson et using two solvents. The upper solvent is chloro-
alP involves an error in subscripts of A-term. For Example 1 form and the lower solvent is water. The column
there is no significant difference between the results by the
proposed method with the correct equation and with Hanson's has 15 theoretical stages, and the binary solution is
equation. According to the correct equation proposed by fed to the 6th stage from the top. The activity co-
Oliver, the equation used in Holland's work2) also has an efficients are computed by use of the three-suffix
error in subscripts of A-term different from Hanson's. We
adopt the correct equation. Margules equation*. The complete problem specifi-
42 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
 Table 1 Specifications of Example 1 Table 3 Specifications of Example 2
Total number of stages 1 5 To tal n umbe r of stag es 16
Feed stages : upper solvent (in c lu d in g so lv e n t rec o v er y sy stem )
h y d ro c a r b o n f e e d F ee d s ta g e s : u p p e r so lv e n t
l o w er s o l ve n t 1 5 h y d ro c a r b o n f ee d
F ee d r at e s an d a na l ys is l ow e r s o lv e n t 1 6
F I F 1 5 F ee d r at e s an d a na l ys is
1 A c eto n e o.0 0. 1 0. 0
2 E th a n o 1 0.0 0. 1 0. 0 i 7" F1 6
Ac et on e 0 .0 0 . 1 0 .0
3 C h lo ro fo rm o. 8 0.0 0.0 2 E th a n o 1 0.0 0. 1 0 0
4 W a ter 0.0 0.0 1. 0 3 C h lo r o fo rm 0. 8 0.0 0.0
T o ta l 0. 8 0. 2 1. 0 4 W a ter 0.0 0.0 1. 0
V al u es o f t hr e e su f fi x M ar g ul es bi na r y c on st a nt s T o ta l 0. 8 0.2 1. 0
A n S o lv en t r ec o v eries
0. 0 0. 5 4 4 6 - 0.94 17 1. 8 7 2 V iy n/ V 2yi 2
0. 599 0. 0 1. 6 1 1. 4 6 1 A c e to n e 0. 0 0 5 0
- 0. 6 7 4 0.50 1 0.0 5. 9 1 2 E th a n o 1 0. 0 0 1 8
1. 3 3 8 0. 877 4.76 0.0 3 C h lo r o fo rm 0. 0029
4 W a ter 0. 9 9 9 3
Table 2 Solution of Example 1 Extr act refl ux ratio (LJU,) 4.0
F in a l c o mp o s it i o ns an d fl o w r a t es o f s t re a ms L N a n d V i
x iN yn T he sa m e v al u es o f t hr e e s uf f ix Ma r gu l e s b in a ry co n st a nt s a s
H anson H o i- T h is H a n so n H o i- T h is i n E xa m p le 1 a re us e d .
la n d w ork la n d w o rk
1 0 . 1 0 5 14 0. 10506 0. 10249 0. 0 0.0 0.0 Table 4 Solution of Example 2
2 0. 0 4 4 1 9 0. 0 4 5 3 1 0. 0559 8 0. 05534 0. 0542 5 0.0 443C F in a l c o m p o sitio n s a n d fl o w ra te s o f s tr ea m s U u L N
a n d V^
3 0.83 653 0.836 13 0. 8160 1 0. 0 0 4 0 4 0. 0039 7 0. 00372
4 0 .0 14 2 4 0. 0 1 3 50 0. 02552 0. 9 4 0 8 7 0 . 9 4 17 8 0. 95184 x tl x iN yn
R a te 0.9 51 - 0.97 6 1. 0 4 9 1. 0 4 8 1. 0 2 4 H o i- T h is H o i- T h is H o i- T h is
/ lan d w ork land wor k land work
I I A ve r ag e d d if f er e nc e b e tw e en ac t iv i ti e s i n b ot h p h as e s
Avera ged diff erenc e* 3. 857 x lO "4 1 0. 0 0. 0 0. 10479 0. 10028 0. 0 0.0
(b y t hi s wo r k) 2 0 . 9 2 9 7 6 0 . 9 4 8 9 1 0 . 0 4 7 0 0 0 . 0 7 2 0 2 0 . 0 0 0 12 0 . 0 0 0 2 6
3 0 . 0 6 0 8 8 0 . 0 4 7 0 1 0 . 8 3 4 5 4 O . 8O O 8 O 0 . 0 0 0 0 1 0 . 0 0 0 0 2
I ll N um b e r o f it e r at i o ns o f p r o po s e d c a l cu l a ti o n p r o c ed u r es 4 0 . 0 0 9 3 6 0 .0 0 4 6 3 0 . 0 1 3 5 9 0 . 0 2 6 8 9 0 . 9 9 9 8 7 0 . 9 9 9 7 3
Calc ulation proced ure No. of iterat ions R a te - 0 .0 2 9 - 0. 997 0. 987 0. 973
Tierney et al. 24**
H a ns o n e t al . 42* II Av e ra g ed di f fe r en c e b et w ee n a c ti v it i es in bo t h p ha s es
Singled method 1 7* * Aver aged dif fere nce* 5. 365 x 10 "
Dir ect i ter ation (0= 1) 23* (b y t hi s w or k)
M u lt i - 0 m et h o d 1 l* I ll C o n v e rg en c e c h a ra c teristic s o f p ro p o se d c a lc u la tio n
T h i s wo r k 7 (33 [sec])*** p ro ce d u re s
C al cu l at io n p ro c ed ur e No . o f Co mp u ta ti o na l
* Z\(a?j-aJj)Ia?MNx M) ite ra tio n s tim e [se c ]
Si ngl e- ^ m eth od ** 1 5 49
** The results are reported by Holland with the convergence D ir e c t i te r a ti o n ( 0 = 1 )* * 21 39
criterion \JV/V\<10~* by use of IBM 360/65. Multi-0 method** 1 1 36
*** The result indicates the outer iteration number which
was obtained with the same convergence criterion as This wor k*** 38
that of Holland by use of NEAC-2200 MODIV EX/VII. * EI(^-^)Kil/[(^-i) xM]

cations are tabulated in Table 1.
The results for this example are given in Table 2,
together with the results of Hanson et al.^ and Hol-
land2>. The agreement between the results of the
three groups is good in general, but the values are
not necessarily consistent. Referring to the averaged
difference between both phase activities
however, assures good accuracy of the results by the
proposed calculation method.
Example 2
This problem is the same as Example 1 except that
the separation column under consideration has a
solvent separation section. This column has 16
theoretical stages, and the 1st stage is used for the
solvent separation section. The reflux ratio LJUx is
fixed as 4.0. The three-suffix Margules equation is
also used. Table 3 summarizes the input flow rates
and operating conditions.
VOL 9 NO. 1 1976
** The results are reported by Holland with the convergence
criterion |JF/F|<10"3 by use of IBM 360/65.
*** The results were obtained with the same convergence
criterion as that of Holland by use of NEAC-2200 MOD
IV EX/VII, and the iteration number is the outer one.
The results for the problem are compared with
in Table 2, Holland's results2) in Table 4. The comparison of
these results provides poor agreementin the values of
compositions and flow rates. However, the reason-
ably lower value of the arithmetic averaged difference
between both phase activities guarantees the accuracy
of the results by the proposed method.
Example 3
This problem is a transform of the design problem
which was solved graphically by Treybal10). The top
feed solution, composed of water and acetone, is to
be separated with 1,1,2-trichloroethane as a bottom
solvent in a countercurrent multiple-contact opera-
43
 Table 5 Specifications of Example 3 Table 7 Specifications of Example 4
T o t al n um b e r o f st a g es To t a l n u m be r of s ta g e s
F e e d s t a g e s : h y d r o c a r b on f e e d Feed stages : upper hydrocarbon feed
s ol vent l o w e r h y d r o c ar b o n f e ed
F e ed r a te s a nd a n al ys i s s o lv e n t
Fi Fe ed ra t es a n d a na ly s is
1 W a te r 2. 775 0. 0 Fi
2 1, 1,2 -T r ic h lo ro eth a n e 0.0 > 0. 225 1 W a te r 2.77 5 4 . 1 62 0.0
3 A ce to n e o. 861 0. 0 2 1, 1,2 -T r ic h lo ro e th a n e 0.0 0.0 0.37 5
T o ta l 3. 6 3 6 0. 225 3 A ce to n e 0.86 1 0 . 4 30 0.0
V al u es o f m o di f ie d Wi l s on c on s ta n t s T o ta l 3. 6 3 6 4.592 0.37 5
A ij- h Th e s am e v a lu e s of m o di f i e d W i l so n c o ns t a n ts a s i n E x am p l e
0. 0 1405.69 - 7 7 1. 8 7
2 9 9 7. 5 9 0.0 - 3 1 2. 8 3 a r e u s e d .
14 84 . 7 9 8 3 1. 4 4 0. 0
Pn = O A 94 0 18 = O .2 4 4 iO 28 = 1 .2 5 8
(a t 2 5 -C )

Table 6 Solution of Example3

I F i n a l c o m p os i t io n s a nd f l ow r at e s
X lj X Bj yu J 2i ys.i
3. 6 6 8 0 . 7 54 2 5 0.0 0672 0.2 3910 1. 1 2 3 0 . 19 32 6 0. 19718 0.60 975
3. 5 4 6 0.769 79 0.0 0598 0.2240 8 1. 1 5 5 0. 1 80 68 0 . 2 1 3 15 0. 6 06 95
3. 3 7 4 0.794 56 0.0 0489 0.20 032 1. 0 3 3 0 . 16 6 1 3 0. 2349 1 0.5990 1
3. 1 6 6 0 . 8 30 3 9 0. 0036 1 0. 16585 0. 86 1 0. 1 42 9 0 0 . 27 60 2 0.5804 3
2.947 0. 87733 0.00235 0. 12025 0.653 0 . 10 8 8 5 0 . 35 62 2 0. 5 34 25
2. 7 3 8 0 . 9 34 2 6 0.0 0132 0.0644 0 0. 434 0. 0 6 4 2 5 0. 52526 0. 4 10 0 5
II O th e r c o nv e rg e n ce c ha r ac t e ri s ti c s
A ve r ag e d d if f er e nc e b e tw e en a ct i vi t ie s i n b o th ph a se s =5 . 74 6 x 1 0" 4 *
N um b er o f o ut e r i te r at i o ns =4 1
C o m p u t a t i o n al t i m e [ s e c ] = 50
w h e re c on v e rg e n c e c r it e r i on i s E I S ** i - 1￨ < 1 0 "
3 i

Z Kaij -aTj)la}j\KNx M)

Table 8 Solution of Example 4

I F in a l co m p o si t i on s a nd f l ow r a te s
X IJ X 2j X %j yu yv ys.i
3. 4 5 8 0.7 8373 0. 005 35 0. 2106 1 1. 6 3 1 0. 17 20 2 0. 2 2 54 0 0 . 60 25 8
3. 2 14 0. 82333 0.00 382 0. 17270 1. 4 5 3 0. 14846 0 . 2 6 5 30 0. 58453
3.042 0. 85630 0.00285 0. 1 4 0 9 7 1.209 0 . 12 54 4 0. 3 1 3 6 8 0. 5589 1
2. 9 3 2 0.8 8030 0. 0022 8 0. 1175 1 1. 0 3 7 0 . 10 665 0. 3 6 2 5 7 0. 53007
7 .5 1 0 0 . 89 7 1 8 0 .0 0 1 94 0 . 1 00 9 2 0 . 9 27 0.0 9354 0. 4 0 3 7 3 0 . 50 25 7
7. 4 5 1 0.90 271 0 . 0 0 18 3 0.0 9550 0. 9 1 3 0.0 8926 0. 4 18 7 1 0.4 9192
7. 3 5 5 0 . 9 12 1 4 0 . 0 0 16 7 0.0 8624 0. 8 5 4 0.0 8188 0. 4 4 6 5 1 0.4 7154
7. 203 0 . 9 2 80 5 0 . 0 0 14 1 0.070 57 0. 7 5 8 0.0 6917 0. 5 0 1 3 8 0.4 2951
6. 972 0.95470 0.00105 0.04426 0. 606 0.04 725 0. 6235 3 0. 32935
I I Ot h er c on v er g e nc e c h a ra c t er i st i c s
Av e ra g ed di f fe r en c e b et w ee n a c t iv it i es i n b ot h p h as e s= 7 .1 8 2 x 1 0 "4 *
N um be r o f o ut er it er at i on s = 23
C o m p u ta tio n a l tim e [sec ] = 78
w h e r e c o n v e rg e n c e c r i t e r io n i s Z￨ S * ^ - l l < 1 0 "
5 i

H \(afj -aTj)la}j\l(Nx M)

tion at 25°C. Total number of theoretical
The modified Wilson equation is applied for calculat-
ing activity coefficients. The parameter values are
obtained by numerically fitting the tie-line data for the
1 , 1 ,2-trichloroethane-water-acetone system at 25°Cn).
Table 5 clearly specifies the problem.
The solution is listed in Table 6. The convergence
is reasonably good and the modified Wilson equation
is successfully applied to the extraction problem.
Example 4
This problem is also a transform of the extraction
column design problem which was handled graphically
44
stages is 6. by Treybal10). Both the top feed and the middle
feed solutions, composedof water and acetone, are to
be separated by extracting with pure 1,1,2-trichloro-
ethane. The cascade is assumed to operate iso-
thermally at 25°C. Total number of theoretical
stages is 9, and feed plates for the binary solutions are
1st and 5th stages. The activity coefficients are com-
puted by use of the modified Wilson equation. The
complete specifications for this problem are shown in
Table7.
The complete solution is given in Table 8. Both
the convergence of the proposed procedure and the
JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
 applicability of the modified Wilson equation are leaving y-th stage [kg-mol/hr]
satisfactory. Xu = mole fraction of component / of raffinate-
phase stream leaving y-th stage [-]
Conclusion yij = mole fraction of component i of extract-
phase stream leaving y-th stage [-]
A new general calculation method, based on the zij = mole fraction of component i of feed
tridiagonal matrix operation and the convergence stream to y-th stage [-]
procedure by the "pseudo-equilibrium" concept, has Tij = activity coefficient of component / of
been developed for multistage, multicomponent liquid- phase flow leaving y-th stage [-]
liquid extraction problems. This technique has the e± = tolerance of error in inner convergence
routine [-]
advantage of yielding stable solutions for different £2 = tolerance of error in outer convergence
types of tower processes, including multi-feeds with a rou tine [-]
solvent recovery section. The iterative method has hj = energy of interaction between i-j pair [cal/mol]
an outer convergence routine that modifies phase <Superscripts>
flow rates and an inner convergence one that corrects L = raffinate phase
distribution coefficients based on the concept of r = numberof iterations
''pseudo-equilibrium' '. V = extract phase
' =approximation
The applicablity of the proposed method to the -=
normalize
calculation of liquid-liquid separation processes is
Literature Cited
demonstrated through numerical computation. More-
over, the feasibility of the modified Wilson equation 1) Hanson, D. N., J. H. Duffin, and G. F. Somerville: "Com-
in expressing equilibrium relations for design calcula- putation of Multistage Separation Processes", Reinhold,
Co. (1962).
tions of liquid-liquid extraction is verified. 2) Holland, C D. : "Fundamentals and Modeling of Separa-
Nomenclature tion Processes-Absorption, Distillation, Evaporation, and
Extraction", Prentice-Hall, Inc., Englewood Cliffs, N.J.
Aj,Bj,Cj,Dj = coefficients in tridiagonal matrix (1975).
equation [kg-mol/hr] 3) Oliver, E. D. : Ind. Eng. Chem., Fundamentals, 7, 335 (1968).
atj = activity of component i at y-th stage [-] 4) Powers, J. E.: Chem. Eng. Progr., 50, 291 (1954).
Fj = flow rate of feedstream to y-th stage 5) Roche, E. C: Brit. Chem. Eng., 14, 1394 (1969).
[kg-mol/hr] 6) Roche, E. C: Brit. Chem. Eng. Process Technology, 16, 821
Kn = distribution coefficient of component i (1971).
at y-th stage [-] 7) Smith, B. D.: "Design of Equilibrium Stage Processes",
Lj = flow rate of raffinate-phase stream McGrow-Hill, New York (1963).
leaving y-th stage [kg-mol/hr] 8) Smith, B. D. and W. K. Brinkley: AIChEJ., 6, 451 (1960).
M = number of components [-] 9) Tierney, J. W. and J. A. Bruno: ibid., 13, 556 (1967).
N = number of theoretical stages [-] 10) Treybal, R. E.: "Liquid Extraction", McGraw-Hill, Inc.,
Rect = solvent recovery of component / New York (1963).
(JW PW [-] ll) Treybal, R. E., L. D. Weber, and J. F. Daley: Ind. Eng.
RLl = reflux ratio of raffinate-phase side stream Chem., 38, 817 (1946).
from 1st stage (LJUO [-] 12) Tsuboka, T. and T. Katayama: /. Chem. Eng. Japan, 8, 181
Ui = flow rate of raffinate side stream from (1975).
1 st stage [kg-mol/hr] 13) Wang, J. C. and G. E. Henke: Hydrocarbon Process., 45,
Vj = flow rate of extract-phase stream 155 (1966).

VOL.9 NO. 1 1976 45