EXTRACTIVE DISTILLATION CALCULATIONS BY MODIFIED

RELAXATION METHOD'

TAKESHI ISHIKAWA AND MITSUHO HIRATA

Department of Industrial Chemistry, Faculty of Engineering,
Tokyo Metropolitan University, Tokyo, Japan

A modified relaxation procedure is developed for calculations of the multicomponent

distillation column. The method is applied to extractive distillation to calculate the product
distribution and the composition and temperature profile in the column when feed con-
ditions, flow rates, number of plates, locations of feed plates, and reflux ratio are
specified.
The method is simple and numerically stable. Some problems are discussed to demon-
strate the feasibility of the procedure.

Graphical methods of type (1) are intuitive and lend

Introduction themselves to an easy understanding of the idea of
Extractive distillation is an important process for the distillation calculations, but the procedures are rather
separation and purification of chemicals produced syn- complicated and time-consuming. Moreover, in these
thetically and in the separations of other closely boiling methods the calculations are carried out under the
components, and especially of azeotropic mixtures, special assumption that solvent concentration in the en-
which cannot be separated by ordinary distillation pro- riching section of the column is constant and determi-
cedure. nation of the locations of the feed plates is approximate,
Broadly, extractive distillation is distillation in which hence the results obtained by these methods are not
the volatilities of the key components are altered by always rigorous.
addition of a third component, the so-called solvent. On the other hand, the plate-to-plate calculation
Therefore, the system is composed of at least more than methods of type (2) are excellent. However, in these
three components. Such a system deviates markedly methods the values of the liquid compositions in the
rom an ideal mixture. enriching section of the extractive distillation column
Several methods have been proposed for solving non- are occasionally negative, because the initial values of
f
ideal multicomponent distillation problems and ex- the terminal compositions are not adequate. There-
tractive distillation problems. These methods may fore these methods often lead to no solution, unless a
be classified into following three categories : great deal of experience and judgement are applied in
(1) Graphical solutions, namely, the Benedict choosing initial values. Also, the results in each ite-
method3^, Bonilla method4), Atkins method2), Lee ration obtained from plate-to-plate calculation
ethod8^ and Tanaka method16). methods have no physical meanings, and only the
(2)
Plate-to-plate calculation procedures, namely, convergence values are useful.
The Lewis-Matheson method9) and the Thiele-Geddes Finally, the analytical method of type (3) is also an
method17), with a d method developed by Lyster10). interesting procedure. The method, however, is com-
m
Hirose6) presented a calculational procedure for plicated in comparison with others and involves the
solving non-ideal multicomponent distillation problems special assumption that solvent concentration in the
by combining the Lewis-Matheson method with the enriching section of the column is constant. Further-
odified 6 method. more, it is questionable whether this analytical
In recent years, Sadamoto13) developed a method for method can be applied to systems deviating largely
rom
an ideal mixture.
the calculation of number of plates in an extractive
istillation column by using the modified 6 method. In this paper, a new method is developed for cal-
f
Finally, (3) there is the analytical method proposed culation of the multicomponent distillation column.
by Sugie et al.15\ This procedure applies the analyt- The method was applied to extractive distillation for
ical method developed by Acrivos1) to extractive dis- calculating the product distribution and the composi-
m
d
tillation calculations. tion and temperature profile in the column when feed
conditions, flow rates, number of plates, locations of
Received on August 25, 1971
Presented at the 4th Autumn Meeting of the Soc. eed plates, and reflux ratio are specified.
of Chem. Engrs., Japan, Oct. 1970 This method employs the modified relaxation algo-
f
VOL.5 NO, 2
'27' 125
 f con d ens er

r-
¥
/ ¥
D. x D i /
/ ¥
2 / H ^ x. I
s, q s I J /
s I
si /
s *1 ¥
/
¥ /
¥
j
V
V H y i.M L j X 'J

t- 1
F ,qP f
Fig. 2 An ideal equilibrium stage in the column
Zfi
f +1
yth stage and V and L refer to the vapor and liquid
flow rates
N , respectively.
The vapor holdup, H^, in Fig. 2 is neglected in the
material balance because it is usually small relative to
the
liquid holdup.
By use of the mean value theorem of integral and
the mean value theorem of differential calculus, the
Fig. 1 A model for the extractive distillation column left-hand and right-hand sides, respectively, of Eq.(l)
may be stated in the following form.
rithm for the solution of non-linear simultaneous t+Jt[{Vj+1yitj+1JrLj-1xitj-1)-(Vjyij+LjXij)'
}dt
equations. Therefore, the method cannot be clas-
sified into any one of the three categories described =[(Vj+1yi,j+1+Lj-lxi)j-l)-{Vjyij+LjXij)lavdt (2)
bove. Weconsider that it belongs to a fourth type.
a This procedure, as described later, makes use of un-
steady state equations for the determination of the Substituting Eqs.(2) and (3) into Eq.(l) yields
steady state solution. Therefore, this method is equi-
valent to having a pilot plant column and actually
operating it to obtain the answers. Also, problems -1 td =[(VJ+1yitj+t+Lj-iXi,y_!)
-{Vjytj+Ljxij)'\av
in the sellection of initial values are simple, and no
special consideration for the least amount of compo- It is assumed that the liquid holdup on each stage
remains constant with respect to time, that is,
nents involved in distillate and/or bottoms is necessary.
dHl=,
Mathematical Model dt
Furthermore, by taking the limit of each term of
The extractive distillation column shown in Fig. 1 Eq.(4) as At approaches zero, following differential
is
considered as an ideal model. This column has N
equilibrium plates, including a solvent feed plate and equation is obtained.
a feed plate, and has a condenser at the top and a Hi~rr-) =t<yj+iVt.j+i+Lj-i*i.j-i)
dt reboiler at the bottom. For the convenience of cal- /t
culations the stages are numbered from top to bottom, . -(VjVij+LjXi^l
with the condenser as zeroth stage and reboiler as the If all the compositions and the flow rates at time t
AM- lth stage. are
known, it is then possible to calculate all the con-
Fig. 2 shows such an ideal equilibrium stage, ex- centration gradients, dXij/dt.

cluding condenser, solvent feed plate, feed plate and However, an implicit form of the finite difference
reboiler. approximation for the differential equation is
For the time period or interval from t to t-\-Jt, the ij dx
component-material balance on the yth stage as ~~dt

t At
shown in Fig. 2 is given by the integral-difference
or the convenience of calculation, consider time t
equation with the law of conservation of mass.
to beiteration n and t-\-At to be iteration n-\-l. Then
a combination ofEqs.(6) and (7) leads to
F
^[HPtjl^-lHjXtjl (1) c»+i=xfj+ftjKVj+iyt,j^+Lj^Xt,y_!

where j refers to the stage number and i refers to the -{Vjytj+LjX^T

component number. where pj is equal to (AtjH1-) and the relaxation
Hfj refers to the total moles of liquid holdup on the factor. Eq.(8) is also the working equation of the
126 (28) JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
procedure proposed by Rose13). For the condenser, method may be called mathematically a simultaneous
the solvent feed plate, the feed plate and the reboiler, point relaxation method.
respectively, similar but slightly different equations This method is stable, but the convergence rate is
canbe written as follows, generally too slow for practical use, and if feed compo-
xTi^xlt+PJL Vtfto-LoVu-DxDtY (9) sitions are used as the initial values of the liquid compo-
x1+,1 =xrl,+ftsUSzSi+Vt+iyt,l+1+L,-1xitS.i)
sitions for all stages of the column, further calculation
~(Vsyis+Lsxis)T (10) cannot proceed because the correction term of the
xni}1 =xntf+(2/[(FzFi+Vf+1yi,f+1+Lf.1xiif-1) right-hand side in Eqs.(8)~(12) becomes zero.

-{Vfyif+Lfxif)T (ll) Accordingly, to increase the rate of convergence

and to be able to apply for distillation calculations of
-
WxmT (12) the non-ideal multicomponents, wepropose the follow-
ng procedures.
If the compositions and flow rates at any iteration Instead of x7l>j^1 in Eq.(8), the values of the liquid
(n) are known, Eqs.(8)~(12) may be used to calculate compositions of the above stage calculated at (^+l)th
the compositions at iteration (n+1), and the process

i
iteration, that is, #?*/-i> were used as the composi-
repeated to obtain values for iteration (w+2), and so ions of the j-lth stage. Consequently, Eq.(8) may
on.In this calculation, the values of vapor composi- be written as follows.

t
tions, yij, are obtained from the corresponding values
of liquid compositions, Xtj, by use of the vapor-liquid '>iV
<j+PjU=x'"Vj+1ylj+1 +LJ-1x'itj1-l)
equilibrium relation. -( VjV*tJ+LjX*tt)] (19)
This is the basic equation for our method. Relative
Method of Computations to Rose's method this may be called a successive point
relaxation
method. It is as stable as Rose's method,
Assuming constant molal overflow and that the
but the rate of convergence is more rapid. For the
liquid holdup is equal for all stages of the column, the
condenser, the feed plates and the reboiler, an equation
vapor and liquid flow rates in each section of the col-
of this same kind can be written.
umnare the following,
Alternatively, when the calculations are begun at
for the solvent recovery section,
the bottom of the column and proceed upward, by
L=RD (13)
means ofa manner similar to that described above, but
V=(R+1)D (14)
slightly different, the basic equations were easily deriv-
for the enriching section, ed. Briefly, instead of y\j+1 in Eq.(8), y7Cj'+i are
L" =RD+Sqs (15) used as the vapor compositions of the y+lth stage.
V"=(R+l)D+S(l-qs) (16) Therefore, Eq.(8) may be written as

for the recovery section, ;?}l =*?y+Ao[(V',+iy":y+i+£/-i*" y-i)

L' =RD+Sqs+FqF (17) -<yjvi,+Ljx«ti)]

V'=(R+l)D-S(l-q8)-F(l-qI) (18) In this case, it is emphasized that the vapor compo-
sitions, 2/i^+i, in equilibrium with the liquid, a??*/+i,
where R is reflux ratio.
When the thermal conditions, the compositions and on the i+lth stage at (n+l)th iteration must be
flow rates offeeds, the amounts of the distillate stream, evaluated before Eq.(20) is applied to the 7th stage.
number of plates, locations of feed plates, and reflux Thus, for simplicity, calculation from top to bottom of
ratio are specified, calculations may be started using he column may be recommended.
feed composition as the initial values of the liquid com- For the calculation of the vapor compositions to be
positions for all stages of the column. The first step in equilibrium with the liquid, relation for the non-
t
is the use of the basic equation for the condenser to ideal multicomponent vapor-liquid equilibrium has
calculate the compositions of the liquid on this stage to be used, because in extractive distillation a solvent
at the first iteration. The second step is to use the is added to increase the nonideality of the system. The
basic equation to calculate the compositions of the calculations of such a system may be carried out by
irst stage at the first iteration. he following procedure.
In the same manner the compositions at first itera- In this system, the vapor-liquid equilibrium ratio
t
f

tion are calculated for each of the other stages from depends on the temperature, the pressure and the
top to bottom. Alternatively the calculations can compositions. Hence when the liquid compositions
egin at the bottom and proceed upward. are given, by use of the applicable activity coefficient-
b In this method, as the calculation for each stage is liquid composition relation, for example, the Wilson
independent of that for all other stages, the above equation18^ or the Margules equationll), the activity
calculations can start from any arbitrary stage and to coefficients were computed.
roceed upward or downward. Then assuming a temperature and proceeding until
Now consider calculation from top to bottom. In (22/tj-']_) is less than a specified tolerance, the tem-
p
Rose's method, when the compositions of the 7th perature was corrected by the Newton-Raphson itera-
stage at nth iterations, a??,y_i, had been used as the tion method. For the purpose of decreasing the num-
compositions of the above stage. Therefore, the ber ofiterations in this calculation, the following device
127
VOL.
5 NO. 2

[29]
 STA R T Table 1 Specifications of Example 1
R EA D
F 100.0 NT 10 V-L relation

SPECI
V ^ ^ l Sｻ F IC A -
D40.0
W60.0
NF
Total
4 "CompV
1
~au
3.0
TFi
0.3
by Eq sS 14 ￨ T IO N S q 1.0 condenser 2 2.0 0.4
R 3.0 3 1.0 0.3

K= 1 NT: number of total plates

NF : feed plate numbered from top
X jj b y E q .19
Comp.: component
a: relative volatility
IS
!
fri ｻ

i= W xfp
1
n
b y E q .1 5
Step4. Evaluate vapor composition yá"j, that is
in equilibrium with the liquid, x^j.
0 J> N *1 Step 5. Solve the basic equations, Eq.(19), for x%y\
ye s using the successive relaxation method.
N O R M A L IZ E Step 6. If the values of the liquid compositions deter-
o F x K+1i K = K *1
mined in Step (5) are negative because of over-correc-
tion, to decrease the values of the relaxation factor the
｣ i: by E q .2 0
applicable constant is multiplied by it and then the
& 2 b y E q .2 1
calculation proceeds, using the new value in Step (5).
On the other hand, if the value is not negative, go
｣.i-tr<S.
y es on to the succeeding step.
W R IT E
RESU Step 7. Normalize the liquid compositions deter-
mined in Step (5).
S TO P
Step 8. Repeat Step (4) through Step (7) unitil the
following equations are satisfied.
Fig. 3 Flow diagram for the computer calculations FzPi+Szsi-Dxlt- Wxlj 1

FzFt +SzSi
\
ATC N+l
is used. The initial values of the temperature for the (2 2 [(^;i-iy/ii;i]!^! (25)

first stage calculated, for example at top or bottom of

the column, is taken equal to the temperature defined where £H and £2 are tolerances. Fig.3 shows the flow
by the relation diagram for computer calculations of such a procedure.
NC
i1hi
In Fig.3, the repetitions for the component and the
stage were omitted, to avoid complexity on the dia-
where Tbi is a boiling point for the pure components gram. Also, the flow diagram for calculation of the
consisting of the system at the pressure and x\ is the vapor compositions was omitted because of space limi-
initial values of the liquid compositions. If the con- tations.
verged value of the temperature was determined by
means of the procedure interpreted before, that value Illustrative Examples
is used as the initial value of the stage below and same This method has been programed for a FACOM
procedure is continued. 270-30 digital computer at Computer Center ofTokyo
Metropolitan University, and many test problems
Computational Procedure have been successfully solved without relying on spe-
Computational procedure may be written as follows. cial experience.
Step 1. Relaxation factor, fij, is given as the recipro- The computer running time for the method is pro-
cal number of two to five times the summation of portional to the number of components and the num-
amounts of solvent feed and feed per unit time, from ber of stages in the column. For the problems tested,
a large number of trials to avoid divergence and in- the computing time was about 0.04 seconds per com-
stability
in calculations, that is, ponent per stage per iteration.

Pj~ (2~5)x(S+F) (22) Example 1

In confirmation of availability of the new method,
Step 2. Calculate the vapor and liquid flow rates of firstly, the same problem as was described in our pre-
each section by use ofEqs.(13) to (18).
Step 3. Assume an initial composition on each stage. vious paper14) was attempted.
It is convenient to use overall feed composition present- The specifications of the problem are shown in
ed by the following equation as the initial values, Table
1. Briefly, consider the distillation of an ideal
system consisting of three components. The column
has 12 stages, including a total condenser and a re-
iJ S+F
boiler.
128 [30) JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
 Table 2 Results of Example 1
S t a e re N o . j Composition p r o fi le x S um x Composition p r o fi l e x S u m x
1 2 3 1 2 3

5 t h I t e r a ti o n 3 0 th I t e r a t io n
D is t i ll a t e 0 0.63572 0 .3 1 8 1 4 0 .0 4 6 1 4 1. 0 0 0 0 0 0.661 19 0 . 324 88 0. 0 1 3 9 3 1.0 0 0 0 0
1 0.50803 0 .3 8 1 3 6 0. 1 10 6 2 1. 0 0 0 0 1 0.55548 0.40941 0. 0 3 5 1 1 1.0 0 0 0 0
2 0.43034 0 . 40 69 8 0. 16 2 6 8 1. 0 0 0 0 0 0.46414 0.46490 0. 0 7 0 9 6 1.0 0 0 0 0
3 0.37729 0 . 4 143 7 0. 2 0 8 3 4 1. 0 0 0 0 0 0.38661 0.48600 0. 1 2 7 3 9 1.0 0 0 0 0
F eed 0.33940 0 . 4 0 84 0 0. 2 5 2 2 1 .0 0 0 0 1 0 .3 2 0 5 6 0 .4 7 1 1 7 0 .2 0 8 2 6 0 .9 9 9 9 9
5 0.33678 0 . 4 0 88 3 0. 2 5 4 3 9 1. 0 0 0 0 0 0.29002 0 .4 9 8 2 5 0 .2 1 1 7 3 1 .0 0 0 0 0
6 0.33430 0 . 40 9 16 0. 2 5 6 5 3 0.99999 0.25597 0 .5 2 6 7 1 0 .2 1 7 3 2 1 .0 0 0 0 0
7 0.30002 0. 4 3 6 0 0 0 .2 6 3 9 8 1. 0 0 0 0 0 0.21901 0 .5 5 3 5 2 0 .2 2 7 4 7 1 .0 0 0 0 0
8 0.24841 0 .4 7 0 5 2 0 .2 8 1 0 7 1. 0 0 0 0 0 0.17978 0. 5 7 2 7 4 0 .2 4 7 4 8 1 .0 0 0 0 0
9 0.18694 0. 4 9 3 8 4 0. 3 2 0 2 2 1. 0 0 0 0 0 0.138 97 0 .5 7 3 2 7 0 .2 8 7 7 6 1 .0 0 0 0 0
10 0.122 8 3 0 .4 7 6 9 1 0 .4 0 0 2 6 1. 0 0 0 0 0 0.09768 0 .5 3 7 4 0 0 .3 6 4 9 2 1 .0 0 0 0 0
B o tto m s 11 0.06877 0 . 39 67 6 0. 5 3 4 4 6 0.999 99 0.05857 0. 4 4 6 6 2 0 .4 9 4 8 1 1 .0 0 0 0 0
1 0 t h I te r a ti o n 4 2 n d I t e r a t io n ( F in a l r e s u l ts )
D is t ill a t e 0 0.68720 0. 2 8 8 6 8 0 .0 2 4 1 2 1. 0 0 0 0 0 0.66105 0. 3 2 5 2 5 0 .0 1 3 7 0 1 .0 0 0 0 0
1 0.57623 0 .3 6 3 1 0 0 .0 6 0 6 7 1. 0 0 0 0 0 0.55551 0.40997 0. 0 3 4 5 2 1.0 0 0 0 0
2 0.48362 0 .4 1 2 3 6 0 .1 0 4 0 2 1. 0 0 0 0 0 0.46443 0. 4 6 5 4 2 0 .0 7 0 1 4 0 .9 9 9 9 9
3 0.40077 0 .4 4 0 9 8 0. 1 5 8 2 5 1. 0 0 0 0 0 0.38701 0 .4 8 6 4 1 0 .1 2 6 5 8 1 .0 0 0 0 0
F e ed 0 .3 2 7 8 6 0 .4 4 5 8 6 0 .2 2 6 2 9 1. 0 0 0 0 1 0.32091 0. 4 7 1 3 5 0 .2 0 7 7 5 1 .0 0 0 0 1
5 0.29284 0 .4 7 5 7 1 0 .2 3 1 4 5 1. 0 0 0 0 0 0.29047 0 .4 9 8 4 2 0 .2 1 1 1 1 1 .0 0 0 0 0
6 0.25217 0. 5 0 8 4 6 0 .2 3 9 3 7 1. 0 0 0 0 0 0.25652 0 .5 2 6 9 6 0 .2 1 6 5 3 1 .0 0 0 0 1
7 0.20873 0. 5 3 8 3 7 0 .2 5 2 8 9 0.9 9999 0.2 1962 0 .5 5 3 9 8 0 .2 2 6 4 0 1 .0 0 0 0 0
8 0.16499 0. 5 5 6 9 9 0. 2 7 8 0 1 0.9 9999 0. 180 44 0 .5 7 3 6 1 0 .2 4 5 9 5 1 .0 0 0 0 0
9 0 . 12 2 7 0 0 .5 5 1 7 1 0 .3 2 5 6 0 1. 0 0 0 0 1 0. 13 96 4 0 .5 7 4 8 0 0 .2 8 5 5 6 1 .0 0 0 0 0
10 0.08344 0 . 5 05 86 0. 4 10 7 0 1. 0 0 0 0 0 0.09832 0 .5 3 9 7 5 0 .3 6 1 9 3 1 .0 0 0 0 0
B o tto m s 1 1 0.04852 0 .4 0 7 6 6 0 .5 4 3 8 2 1. 0 0 0 0 0 0.05907 0.44957 0. 4 9 1 3 6 1.0 0 0 0 0
2 0 th I te r a tio n Results of S h on o
D is t i ll a t e 0 0.66445 0. 3 2 0 3 1 0 .0 1 5 2 4 1. 0 00 00 0.6613 0 .3 2 5 0 0 .0 1 3 7 1 .0 0 0 0
1 0.55806 0 . 40 3 53 0. 0 3 8 4 1 1. 0 00 00 0.5558 0 .4 0 98 0 .0 3 4 4 1 .0 0 0 0
2 0.46468 0. 4 5 9 8 2 0 .0 7 5 5 0 1. 0 00 00 0.4647 0. 4 6 53 0 .0 7 0 0 1 .0 0 0 0
3 0.38604 0. 4 8 2 10 0 .1 3 1 8 6 1. 0 00 0 0 0.3872 0 .4 8 63 0 .12 6 5 1 .0 0 0 0
F eed 0.31997 0 .4 6 9 0 2 0 .2 1 1 0 0 0.9 9999 0.3210 0. 4 7 13 0 .2 0 7 7 1 .0 0 0 0
5 0.28918 0. 4 9 5 9 2 0 .2 1 4 9 0 1. 0 0 0 0 0 0.2906 0 .4 9 8 3 0 .2 1 1 0 1.0 0 0 0
6 0.25487 0 .5 2 3 9 8 0 .2 2 1 1 5 1. 0 0 0 0 0 0.2567 0 .5 2 6 9 0 .2 1 6 4 1 .0 0 0 0
7 0.21761 0 . 55 00 2 0. 2 3 2 3 7 1. 0 0 0 0 0 0.2198 0. 5 5 3 9 0 .2 2 6 3 1 .0 0 0 0
8 0 . 17 8 0 5 0 .5 6 7 8 5 0 .2 5 4 1 0 1. 0 0 0 0 0 0. 1806 0. 5 7 3 6 0 .2 4 5 7 1 .0 0 0 0
9 0.13697 0. 5 6 6 1 2 0 .2 9 6 9 1 1. 0 0 0 0 0 0. 1398 0 .5 7 4 9 0 .2 8 5 2 1 .0 0 0 0
10 0.09563 0 .5 2 7 3 6 0 .3 7 7 0 1 1. 0 0 0 0 0 0.0985 0 . 5 40 0 0. 3 6 1 5 1. 0 0 0 0
B o tto m s 1 1 0.05686 0. 4 3 4 5 8 0 .5 0 8 6 7 1. 0 0 0 0 1 0 .0592 0 .4 5 0 0 0 .4 9 0 7 1.0 0 0 0

r r r i
D a c a a a. c ^jm a

1 o c a A > V^ c m r xi

2 jK h C o m p o n e n t Ite ra t io n
Num ber
3 蝣jjt t i a a 5 th
10 th
t 4 (Feed ) ^> o o ｫ fc a a A u a 30 th
n F ina l
E A > R es u lt
=> 5
z:
5 6 a m

n 7
all o o

8 a r

9 a jT

Fig. 4 Liquid composition 10 n ｫ^ r

profile for Example 1
w agio a a *- 一 oo
I

1 0 2 0 3 0 A O 5 0 6 0 0
L i q u id C o m p o s itio n ( m o le p e r c e n t )

VOL. 5 NO. 2 1972 (3i; 129

 Starting with feed composition used as initial values he results obtained by Shono14) are also shown in the
of the liquid compositions for all stages of the column, last columns of Table 2.
a solution was obtained in 42 iterations. Table 2 It can be seen from this table that the results of this
and Fig.4 give the computational results of this prob- work agree very well with those of Shono. Conse-
lem. In Table 2 liquid composition profiles xij9 uently, it is confirmed that this new method is useful.
computed at the end of the 5th, 10th, 20th, 30th, and Table 3 shows the variation for the errors of each

T
q
last iterations. It is observed that xis changed only component in this problem. There are convergence
slightly after the 30th iteration. differences depending on the component. As shown
n Fig.4, however, the results obtained at the end of
the 30th iteration could be considered sufficiently ac-
Table 3 Variation for errors of the material balance of curate in practical use.
each component for Example 1 In this problem, whenSxj = 10~3and £2 = 1CT6, the

i
I te ra tio n E r ro r s computing running time, including the printout of
n u m b er ｣2 input data and detailed results of the final iteration,
5 3.89xlO"2 7. 23 x lO" 2 1 .4 2 x lO"1 1 .5 7 x lO " 2 was about 30 seconds. When the same problem was
10 3 .9 1 x lCT 3 9 .5 5 x lO -2 1 .2 3 x lO"1 2 .2 0 x lO 3 olved by Rose's method, 349 iterations were required
20 1. 3 4 x lO - 3 2 .9 4 x lO " 2 3 .7 8 x lO"2 8 .9 3 x lO " 5
30 1 .14 x lO - 3 5. 3 4 x lO " 3 8 .2 6 x l O "3 3 .6 8 x lO"
for a converged solution and the computing time was

s
40 6.13 x lO " 4 7 .5 6 x lO " 4 1 .6 2 x lO-3 1 .3 6 x lO"7 about 300 seconds.
42 4.65xl(T4 3 .9 1 x lO " 4 9 .8 7 x lO-4 5 .18 x lO"8
xample 2

e -\ft-(dt)ea-
U ft Nowconsider separation of acetone from acetone(l)
NC NT+1,X^J1-^jV -acetonitrile(2) mixture by extractive distillation using
)2j e2-2j I t^
water(3) as solvent. The problem specifications and
hysical data are given in Table 4.
Briefly, the column has 15 equilibrium plates with
Table 4 Specifications and physical data of Example 2
a total condenser and a reboiler. The solvent enters
F=1 . 0 qF = ¥ ,0 N T = 15 F e e d s c o m p o s it i o n s
at the 3rd plate, and the feed enters at the 10th plate.
E
p
S = 2 .0 Gs= 1 .0 N S = 3 2 ^ = 0 .9 ^ ^ 1 = 0 .0
D=0.5 R = 3 .0 N F = 10 zF2= O A z s 2 = 0 .0 The column is operated at atmospheric pressure.
W =2 . 5 > = 0 .1 z F 3 = 0 .0 2 5 3= 1 .0 Starting with feed composition used as the initial
Marseilles p a r a m e t e r s a n d A n t o in e 's c o n s t a n t s u s e d
values of the liquid compositions for all stages of the
C o m p . A n t o i n e 's M a r seil le s column, and assuming relaxation factor is 0.1, a solu-
No. c o n s t a n t 1 7* p a r a m e te rs17^ tion was obtained in 42 iterations. Table 5 gives the
B i computational results of this problem.
7 . 2 3 9 6 7 1 2 7 9 .8 7 0 1 2 3 7 .5 A 12 = A 21 =
0.2885 0 .2 5 0 1
7 .2 4 2 9 9 1 3 9 7 .9 2 9 0 2 3 8 .9
0.3197 0. 0 8 2 8
Conclusions
7.96681 . 20 9 9 2 2 8 .0 A 23= A 32 =
0.0019 0. 0 2 4 2
A newmethod for calculation of the multicomponent
^ 12 3= distillation column has been developed. The calcula-
0.2454 tion solves all unsteady-state equations simultaneously
by a modified relaxation method. The method was

Table 5 Final results of Example 2

Sta ge N 3o . T em p . C o m p o s i t i o n p r o f i lx tje 2 V tj 3
[-C ] C o m p o s i t i o n p r o f i l e2

(D ) 0 0.96723 0 .0 0 0 1 4 0. 0 3 2 7 3
1 6 0. 1 0 0.84949 0. 0 0 0 19 0. 15 0 3 2 0.96723 0 .0 0 0 1 4 0.03283
2 67.65 0.59301 0. 0 0 0 2 1 0. 4 0 6 7 8 0.87902 0 .0 0 0 1 7 0.120 8 1
(S ) 7 9. 4 5 0.25436 0 .0 0 0 2 0 0. 7 4 5 4 4 0.6 8681 0. 0 0 0 18 0.3130 2
4 7 9 . 42 0.25489 0 .0 0 0 4 1 0. 7 4 4 7 0 0.68720 0 .0 0 0 3 7 0.31243
5 79.36 0.25613 0 .0 0 0 8 5 0. 7 4 3 0 2 0.68815 0. 0 0 0 7 5 0.31111
6 7 9 . 22 0.25898 0 .0 0 1 7 0 0. 7 3 9 3 2 0.69035 0. 0 0 1 5 0 0.30816
7 7 8 . 90 0.26552 0 .0 0 3 4 3 0. 7 3 10 5 0.69537 0. 0 0 2 9 9 0.30168
8 78.19 0.28091 0 .0 0 7 0 0 0. 7 12 0 9 0.70687 0. 0 0 6 0 1 0.287 34
9 76.44 0.32086 0 .0 1 4 8 4 0. 6 6 4 3 0 0.73375 0. 0 1 2 2 7 0.25397
(F ) 10 7 1. 5 8 0 .44945 0. 0 33 80 0. 5 16 7 5 0.80369 0. 0 2 5 9 8 0.170 34
l l 7 1. 5 8 0.44918 0. 0 33 89 0. 5 16 9 3 0.80334 0. 0 2 6 0 5 0 . 17 0 3 7
12 7 1 .6 2 0.44798 0. 0 34 16 0. 5 1786 0.80283 0 .0 2 6 2 7 0 . 17 0 9 0
13 7 1. 7 7 0.44308 0. 0 34 88 0. 5 22 04 0.80013 0. 0 2 6 8 7 0. 17 3 0 2
14 7 2 .4 3 0.42296 0. 0 3 6 6 9 0. 5 4 0 3 5 0.78918 0. 0 2 8 4 8 0.182 52
15 7 5 . 07 0.34736 0. 0 4 0 2 5 0. 6 12 3 9 0.74383 0. 0 3 2 5 6 0.223 62
(W ) 16 83.84 0 . 16 62 4 0. 0 3 9 9 9 0. 7 9 3 7 7 0 .57 37 3 0. 0 4 0 5 7 0.38571

130 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN

:m
 applied to the calculation of extractive distillation ( Superscripts )
problems. n - iteration number
T L = referstoliquid
(1) hisComputational
method has theprocedure
following is advantage;
simple and does not V = referstovapor
require a very large and fast computer. 0 = initial value
(2) Each trial gives the exact answer for distillation / = refers to recovery section
under the chosen conditions. Accordingly, using this // =refers to enriching section
method is equivalent to obtaining the answers. (Subscripts)
(3) For multicomponent problems, the choice of 1 =componentnumber
j =stagenumber
successive trial value is very simple, and no special t =time
consideration for very small quantities of a compo- F -refers to feed
nent in distillate or bottoms stream is necessary. / =referstofeedplate
(4) If the real starting composition, holdup and cor- S =refers to solvent feed
responding flow rate and time lags are used, the calcu- 5 =refers to solvent feed plate
lations describe the approach to steady state.
(5) This method can be used to solve multiple- Literature Cited
column problems where the columns are interrelated.
(6) The method converges relatively rapidly and 1) Acrivos, A. and N. R. Amundson: Chem. Eng. Sci., 4,
stably. 29 (1955)
2) Atkins, G. T. and C. M. Boyer: Chem. Eng. Progr., 45,
553 (1949)
3) Benedict, 3VI. and L. Rubin: Trans. Am. Inst. Chem.
Engrs., 41, 353 (1945)
Nomenclature 4) Bonilla, C. F.: Chem. Eng. Progr., 37, 669 (1941)
5) Hala, E.: "Vapor-Liquid Equilibrium Data", Per-
Aij - Margulesparameters [-] gamon Press (1968)
Ait Bit Ct~ Antoine constants [-] 6) Hirose, Y. and H. Hiraiwa: Kagaku Kogaku, 32, 998
D å= distillate [mole/hr] (1968)
F = feed rate [mole/hr] 7) Ishikawa, T. and M. Hirata: Kagaku Kogaku, submitted
HJ = vaporholdup [mole] 8) Lee, K. T. and K. Kammermeyer: Chem. Eng. Progr.
H*f = liquid holdup [mole] Symp. Ser., 49, 99 (1953)
L = liquid flow rate [mole/hr] 9) Lewis, W. K. and G. L Matheson: Ind. Eng. Chem., 24,
N = numbersof plates [-] 496 (1932)
10) Lyster, W.N., S.L. Sullivan, D. S. Billingsley and
R - refluxratio [-] C. D. Holland: Petrol. Refiner, 38, 221 (1959)
S - solvent feed rate [mole/hr] ll) Margules, M.,: Akad. Wiss. Wien, Math. Naturw. Kl.
T = temperature [°C] II, 104, 1243 (1895)
V = vaporflowrate [mole/hr] 12) Rose, A., R. F. Sweeny and V.N. Schrodt: Ind. Eng.
W = bottom product [mole/hr] Chem., 50, 737 (1958)
q = thermal conditions offeeds [-] 13) Sadamoto, S., M. Kodama and M. Yorizane: Kagaku
x =liquid composition, mole fraction [-] Kogaku, 33, 557 (1969)
y =vaporcompostion, mole fraction [-] 14) Shono, H. and M. Hirata: Kagaku Kogaku, 33, 1236
z =feed composition, mole fraction [-] (1969)
<Greek letters) 15) Sugie, H. and I. Yamada: Kagaku Kbgakur 32, 273
(1968)
y = activity coefficient 16) Tanaka, S. and I. Yamada: Kagaku Kogaku {Chem.
Aij = Wilsonparameters Eng.. Japan), 28, 661 (1964)
p. = relaxation factor 17) Thiele, E. W. and R. L. Geddes: Ind. Eng. Chem., 25,
A - increment 290 (1933)
£ = tolerance errors 18) Wilson, G.M.: J. Am. Chem. Soc, 86, 127 (1964)

VOL. 5 NO. 2 1972 (33) 131