You are on page 1of 14

This article was downloaded by: [University of Sydney]

On: 30 December 2014, At: 05:20


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Chemical Engineering Communications


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcec20

FIBONACCI SEARCH FOR OPTIMAL FEED LOCATION


J.C. WANG

Engineering Department , Union Carbide Chemicals and Plastics , P.O. Box 8361, South
Charleston, WV, 25303
Published online: 24 Apr 2007.

To cite this article: J.C. WANG (1980) FIBONACCI SEARCH FOR OPTIMAL FEED LOCATION, Chemical Engineering
Communications, 4:6, 651-663, DOI: 10.1080/00986448008935937
To link to this article: http://dx.doi.org/10.1080/00986448008935937

PLEASE SCROLL DOWN FOR ARTICLE


Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained in the
publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations
or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any
opinions and views expressed in this publication are the opinions and views of the authors, and are not the
views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be
independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses,
actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever
caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Chern. Eng. Commun. Vol. 4, pp. 651-663


0098-6445/80/0406-0651 S04.50/0

<!l Gordon

and Breach, Science Publishers Inc., 1980


Printed in the U.S.A.

FIBONACCI SEARCH FOR OPTIMAL


FEED LOCATION
J.e. WANG*

Downloaded by [University of Sydney] at 05:20 30 December 2014

Union Carbide Chemicals and Plastics,


Engineering Department,
P.o. Box 8361, South Charleston, WV 25303
(Received July 23,1979; in final form October 11,1979)
A Fibonacci search technique is used in conjunction with a rigorous multicomponent distillation computer
module to find the optimal feed location within a section of a distillation column. The function to be
minimized can be one of the following:

*key componentratio difference


"reflux ratio or reboiler ratio
"condenser duty or reboilerduty
This technique has been used successfully in the- relocation of feed stages of many existing distillation
columns and thus saved energy consumption. We shall describe this technique with a sample problem.

INTRODUCTION
Feed location is an important factor in the design of a new distillation column or in
evaluating the performance of an existing column. A mislocated feed not only can
seriously impair separation efficiency but also affect the energy consumption. The
subject of optimal feed location has been studied by many investigators. McCabe and
Thiele,' Gilliland,' and others define the optimum feed stage based on matching the
ratio of the light and heavy key components in the feed and on the feed stage. Scheibel
and Montross," Floyd and Hipkin;' and Maas,' have developed procedures to find the
optimum feed stage that produces the maximum separations. Shipman!' gave a
criterion of the optimal feed stage based on the minimization of the entropy
production. Although his approach is thermodynamically sound, the method is
difficult to apply and has not much practical value. Recently, Waller and Gustafsson"
treated the feed stage problem in their optimization of the steady state distiIJation
operation.
Distillation processes are the most energy intensive operation in the petroleum and
petrochemical industries. Prengle, et at," reported that up to 40% of the energy
consumed in a petrochemical plant is taken by the distillation columns. It is therefore
more logical to define the optimal feed location based on the energy consumption.
Luyben" has given such a criterion by stating "The minimum energy consumption is
achieved in a column with a fixed number of total trays when not only X D and X B are
held at their specified values but also the feed is introduced onto the optimum feed
"Dr. Wang is now with Simulation Science, Inc., 1400 N. Harbor Blvd., Fullerton, CA 92635.

652

J.C. WANG

tray." In this paper we shall show that the energy consumption of a distillation column
with a fixed number of stages for a specified separation is an unimodal function or the
feed stage location. This function can be evaluated by using a steady state computer
simulation program. The problem of finding the optimal feed location is then to search
for the minimum point of this function. A Fibonacci search procedure is then
developed for solving this problem. The procedure is simple and can be readily
incorporated into any distillation simulation program.

Downloaded by [University of Sydney] at 05:20 30 December 2014

THE SIMULATION MODULE


Ever since the invention of the digital computer, numerous methods have been
proposed for the computer solution of the distillation problem. It is beyond our scope to
review the various methods in this paper. By selecting some of the methods we have
developed a sophisticated computer program not only for distillation but also for
absorption (with or without reboiler) and liquid-liquid extraction. These operations
can be classified as the Multicomponent Multistage Separation Processes and therefore we have designated our program as the MMSP program." The program is
modular and it consists of various subprograms. These include the thermodynamics
subprograms, the mathematical subprograms, the initiation subprograms, and the
separation subprograms. It can also be readily incorporated to any process flowsheet
simulator such as Union Carbide's IPES (Integrated Process Engineering System)
program." The input can either come from the standard 8 1/2 x II data sheets or from
a stored file. The output can either be printed or stored in a file to be used for column
design programs such as our APT (All Purpose Tray) program. Figure I shows the
information flow of such a program system.
The methods used in the MMSP program are: The Wang-Hcnkc'" method
(DSTLWH) for distillation; the Sujata's Sum Rate!' method (DSTLSR) for absorbers or liquid-liquid extractors; and the Block Tridiagonal matrix method (DSTLBT)
developed by Naphthali and Sandholrn" and modified by Tang. I ' In the last method,
the MESH (Material Balance, Equilibrium, Summation, and Heat balance) equa-.
tions are solved simultaneously using the Newton Raphson technique. The MESH
equations are arranged in such a way that the Jacobian matrix is in a simple block
tridiagonal form. Using this method, various top and bottom specifications can be
made. These include condenser and reboiler temperature, condenser and reboiler heat
duty, reflux and reboiler ratio, and most importantly, the product purities in terms of
mole fraction, moles, or percent recovery of the key components.
For the design problems of determining the stage and reflux requirements in a
distillation column for a specified key component separation, we employed VanWinkle's" shortcut method (SHTCUT). The number of equilibrium stages are first
computed from SHTCUT and the reflux requirement and energy consumption are
computed by use of the Block Tridiagonal matrix method (DSTLBT).
The MMSP program has been used within Union Carbide for the simulation,
design, control studies, and energy audit of many distillation columns, absorbers, and
liquid-liquid extractors. The column can be a simple one with one feed and two
products or a complex one with multiple feeds, multiple sided raws, and multiple

OPTIMAL FEED LOCATION


rFEED AND COLUMN
INPUT DATA FORMS

PHYSICAL PROPERTY
INPUT DATA FORMS

PHYSICAL PROPERTY
DATASET (PP-COMMON)

Downloaded by [University of Sydney] at 05:20 30 December 2014

~(DATA
BANK

-.

FEED AND COLUMN


INPUT DATA SET

-..

THERMODYNAMICS
SUBPROGRAMS

COLUMN INITIATION
SUBPROGRAMS

MATHEMATICAL
SUBPROGRAMS

Ul

...J
IJJ

8~

--

~
~

z
~

0::
IJJ

J:

GFLASH
DSTLBT

XFLASH
DSTLSR

SHTCUT
DSTLWH

Ul

\PES

IIIz

STAND ALONE PROGRAM

PROCESS FLOWSHEET

I-

SEPARATION SUBPROGRAMS

653

MMSP

!i

>

Ul
Ul

tl

o0::

J:

D.
~I

PRINTER
OUTPUT

FIGURE I

PROBLEMSL
OR PROP. \
FILES

III
z
-c
J:

t
TRAY DESIGN PROGRAM
APT~ALL PURPOSE TRAY

Information flow for MMSP simulation and design.

interstage heaters (or coolers). The mixture to be separated can be of any degree of
nonideality ranging from those obeying Raoult's law to highly non ideal systems such
as azeotropes. We have incorporated in our Thermodynamics Subprogram Library a
spectrum of equilibrium models. These include Chao-Seader correlation, RobinsonChao correlation, Poynting correction, and others for liquid phase fugacity; RedlichKister, Regulat Solution, NRTL, Wilson equation, Margules, and Uniquac for liquid
activity coefficients; Redlich-Kwong, Second Virial, Prausnitz-Chueh modified

654

J.C. WANG

Redlich-Kwong and other equations of state to compute vapor phase fugacity


coefficient and vapor enthalpy departure. We also can use the BWR, Starling BWR,
and Soave modified Redlich-Kwong equation to calculate K-values and enthalpy
departures for hydrocarbon mixtures. The data required for all these models can be
retrieved from a comprehensive Data Bank in which the pure component physical
properties of more than 1000 compounds are stored.
The MMSP program is used for the evaluation of the energy consumption of the
distillation column as a function of the location of the main feed stream. The input
data to the program include:

Downloaded by [University of Sydney] at 05:20 30 December 2014

total number of stages including condenser and reboiler


feed conditions, rates and compositions
sided raw locations and rates, if any
purity specifications (in terms of mole fraction or percent recovery of the
specified components in the distillate and bottom products)
the codes to indicate the models selected for calculating the K-values and the
enthalpies. The pure component data can be retrieved from the Data Bank but
binary interaction parameters must be supplied in the input sheets.
Most of the input data items listed above are provided with the item numbers in the
data sheets. Any of these items can be changed by supplying its item number and its
new value. At the end of each case, the computed results are stored on the computer.
The program will then proceed to read the change cards, change the values of the
specified items, and perform the computation for the new case. Since the new case
starts with a converged solution of the previous case, the new case converges very fast.
Finally, the results of all the cases are printed in a single output. This change case
procedure can be overseen by a control subroutine such as the one described in this
paper for the optimal feed location.
The output from the program includes the column profiles of stage temperatures,
internal vapor and liquid rates, the stage vapor and liquid compositions and the
following items which are related to the energy consumption of the distillation
column:
the reflux ratio or reflux rate,
the reboiler ratio or reboiler vapor rate
the condenser duty
the reboiler duty.
In addition the following two items related to matching the compositions in the feed
and on the feed stage can also be computed:

the absolute value of the difference between the key components ratio in the feed
and that on the feed stage, i.e.,

OPTIMAL FEED LOCATION

655

Downloaded by [University of Sydney] at 05:20 30 December 2014

the sum of the absolute values of the difference between the feed composition and
the feed stage composition for all the components.

where Z - feed composition, m.f.; X ~ feed stage composition, m.f. Subscripts i for the
ith component, lk for the light key, hk for the heavy key.

EFFECT OF FEED LOCATION ON DISTILLATION


To show the effect of the feed stage location on the performance of the distillation
column, let us consider the problem of benzene-toluene-xylene separation. As shown in
Fig. 2, 100 Kg molesjhr of the mixture is fed to a distillation column to be separated to
a distillate containing 0.01 mole % of toluene and a bottom product containing 0.01
mole % of benzene. A shortcut calculation indicated that 40 stages are required at a
reflux ratio equal to 1.2 times the minimum. Therefore, 42 stages including condenser
and reboiler are used in the rigorous calculation. The stages are numbered from top to
bottom with the condenser as the first stage and the reboiler as the 42nd stage. The
effect of the feed stage in the middle section of the column (Stage 16 to Stage 29) is to
be studied by using the change case procedure.
The simulation was done by using the Block Tridiagonal matrix method with the
purity specifications for the distillate (0.0001 rn.f. of toluene) and the bottom product
(0.0001 m.f. of benzene). The Wang-Henke method was used for initiating the column
profiles to be used in the base case. The Chao-Seader correlation' was used to
calculate the K-values and the Yen-Alexander correlation" was used for calculating
the vapor and liquid enthalpies. All the physical properties required were retrieved
from the Data Bank. The computed results are plotted in Figure 3. The smoothed
curves for the fourteen case studies with feed stage at 16 to 29 include the values of Cl. k
and Cl. c ' the reflux ratio, the reboiler ratio, the condenser duty, and the reboiler duty.
These results show clearly that all the quantities listed are unimodal functions of the
feed stage location with only one minimum point within the section of the column. It
was also noticed that the minimum points for all the quantities with the exception of
Cl. k fall on the same point. This has been true for all the problems we have encountered.
Therefore, the criterion of optimal feed stage based on the matching of the key
component ratio is not correct. The criterion based on the energy consumption should
be used. Since the energy consumption is related to the reflux rate, the boil-up rate,
the condenser duty, arid the reboiler duty, the optimal feed location can also be defined
as the one that minimized one of these quantities (or the weighted sum of these
quantities). The optimal feed location problem can then be determined by applying a
unidimensional search technique.

J.e. WANG

656

p= 101

KPA

___3

Downloaded by [University of Sydney] at 05:20 30 December 2014

--~----

16
-------

F, 100 Kc;MoL.ES /HR


SATUARED LIQUIDI
0
AT

364 K

BENzENE, 0.6 M.F.]


TOL.UENE, 0.3
_XVL.ENE,
0.1

[
FIGURE 2 The sample problem.

~ItKGMoL.ES/ HR

E~E NZEN. E I 0,0001 MIF.j


TOL.UENE, 0.7499
)(VL.ENE 1 0.2500

OPTIMAL FEEO LOCATION

657

~ t--\-\--4--1~-4-+--------4-----.----+-------+---+--+~I_+__I_l
5.5

0
i0(=

0 5 4 t--++t-~-;-...

0(
III

0::

Downloaded by [University of Sydney] at 05:20 30 December 2014

J: 5 3

3.7

t----+-ll----++

II:
III

II:
III

~ 5 2 r----tt-\--~-I--__+

3.6 0
lD

III

III

3.5

~ 5. 1 r--~t-\--t__-_":~-I__~r__-

0::

c
~ 5.0 +-+----11--+-++
II:
III

:! 4. 9 -t--+--I--l-1~
o

lD
III

0:: 4. 8

1.8

r---\f--\.~

III

0:: 1.6

1.2 4.6

II:
III

~ 1.0 4.5

c
~ 0.8

i=o(

0::

t ,5

::I

4.4

1.4

0::

i.a

II.
III

0(

l:t:

~ 0
:iC

1.7

06

~ 0.4
z
o

E
0.2
rJl
o

13 0.0

29

28

27
NUMBER

FIGURE 3

Effect of feed stage on distillatiion (benzene-toluene-xylene separation).

THE SEARCH ALGORITHM


There are various methods for the search of the minimum point of a function. One of
these is the graphical method that we just described in the previous section. Beveridge
and Schechter,' Wilde,19 and others have reviewed one dimensional search techniques.
Himmelblau' and Beveridge and Schechter' compare the efficiency of the various
unidimensional search techniques. They show that the Golden Search and the
Fibonacci Search are the two most efficient methods. Since the Fibonacci search

J.e. WANG

658

technique is based on the Fibonacci numbers, it is especially convenient for integer


search problems such as our optimal feed stage problem. In this case, the Fibonacci
number F. is the number of stages within a section of columns in which the optimal
feed stage lies and n is the required number of function evaluations (column
simulations).
Using the Fibonacci search technique, we have developed a computational procedure for locating the optimum feed stage with a section of the distillation column. The
function to be minimized can be one of the following:

1 = s,
Downloaded by [University of Sydney] at 05:20 30 December 2014

lk
-I (Zlk)
Zhk - (Xhk ) I,
X

1- I: I z, -

(X;)iF I,

the key ratio difference

(1)

ir

The composition difference

(2)

1-

R/ D,

the reflux ratio

(3)

1~

V/ B,

the boilup ratio

(4)

1 ~ Qc>

1=

the condenser heat duty

QR, the reboiler heat duty

(5)
(6)

where J F indicates the feed stage number.


The computational procedure for the optimal feed stage search algorithm includes
the following steps:

Step I
Supply the input data for the column simulation except the feed stage location. The
data requirements are outlined in the previous section. Also specify the function to be
minimized.

Step 2
Specify the section of the column to be searched by giving the stage number J F, near
the column top and the stage number J F b near the bottom. The difference between J F b
and J F, must be a Fibonacci number F. Then find the number of function evaluations
n corresponding to F. from the table of Fibonacci number (Table I).

Step 3
Determine the first two stage numbers to be checked as follows:
JF, - JF,

+ F._ 2

OPTIMAL FEED LOCATION

659

TABLE I
The First 14 Fibonacci Numbers

F,

0
I
2
3
4
5

1
I
2
3
5
8
13
21
34
55
89
144
233
377

Downloaded by [University of Sydney] at 05:20 30 December 2014

6
7
8
9
10
11
12
13

Step 4
Evaluate the function to be minimized as I, and 12 corresponding to the feed stage
numbers JF, and JF 2 by performing the column simulation with the purity specifications for the top and bottom product streams. Also set the search number I - 2.

Step 5
Checkj; vs./2' If II </2, go to Step 6, otherwise set:

and evaluate a new 12 corresponding to the new JF 2 by performing the column


simulation, then go to Step 7.

Step 6
Set

J.C. WANG

660
and evaluate a new
simulation.

/1

corresponding to the new JF I by performing the column

Step 7
Update the search counter I as

I + I

Downloaded by [University of Sydney] at 05:20 30 December 2014

Step 8
Cheek I vs. n. If I = n stop the iteration and go to output. Otherwise repeat Step 5
through Step 8.
This procedure has been programmed into the MMSP program as the fBSRCH
subprogram. The results from all the column simulations are stored and printed in the
final output along with the column profiles corresponding to the optimal feed stage
location. These results including ~b ~C> reflux ratio, reboiler rate, condenser duty,
reboiler duty, number of iterations used in the column simulation, and the convergence
of the column simulations are printed in tabular form.

APPLICATION Of THE SEARCH PROCEDURE


We shall illustrate the fibonacci search procedure by applying it to the sample
problem of benzene-toluene-xylene separation discussed previously. The section of the
column between stage 16 and 29 was searched for the optimal feed stage location that
produces the minimum reboiler heat duty. The total number of stages within the
section is the fibonacci number 13. from Table I we found that n - 6 and therefore a
maximum of six (6) column simulations were required. The search procedure was
started as follows. figure 4 is the graphical representation of this procedure.
The first two feed stages to be checked were
JF I

JF,

+ Fn_ ,

16

+ 5 = 21

JF 2

J F,

+ Fn_ I

16

+ 8 - 24

Column simulations (using DSTLBT) were performed with these two feed stages. The
corresponding reboiler duties were obtained as:

f, - 4.7702 GJ/hr

/2 =

4.5997 GJ/hr

Since f, > /" the minimum point must lie in the section between stage 21 to stage
29. Therefore we eliminated the section from stage 16 to stage 21 from further
consideration. We then followed Step 5 to set the top limit of the new section as JF, -

OPTIMAL FEED LOCATION

661

5.9
5.8

1\

5.7
5.6
5.5
5.4

\
\
\
\

--_.

_.- --_ ..-

i
1-

-_..-.- --._._-

I
II

-- f--+-

--- -

Downloaded by [University of Sydney] at 05:20 30 December 2014

5.3

~5.2

i
i

1\

!;(

\ !
\

Q
~

.J

lJJ

.. / 1/

It:

It:

I-

g 4.7

lJJ
.It:

i:

1\

4. 8

4. 6

r,,~

SECOND
4. 5 -ELIMINATION

JF.

4. 4
4 3
(JF.i 29

'"'
'

28

26

27

FIGURE 4

..

1!"IRST ELIMINATION _
JF,

JFz JF.
(MIN)

24

25

FEED

23

22

STAGE

21

20

19

18

NUMBER

Graphical representation of the Fibonacci search procedure.

JF, - 21. We also setf! - f2 - 4.5997 GJ/hr, and JF, - JF2 ~ 24. A new JF2 was
obtained as
JF2

JFt

+ JFb -

JF,

21

29 - 24 - 26

Another column simulation with feed stage at 26 was performed. The reboiler duty
was obtained as

f,

4.7046 GJ /hr

Since f, < f2' the optimal feed location must lie above stage 26. The column section
between stage 26 and stage 29 was eliminated from further consideration. Following
Step 6, we set JFb - 26'/2 ~ f, - 4.5997 GJ/hr, and JF2 - JF, ~ 24. A new JF, was
obtained as

J.e. WANG

662
JF I

JF,

+ JFb

The column simulation with JF I

+ 26 - 24

23

23 yielded the reboiler duty

II -

Downloaded by [University of Sydney] at 05:20 30 December 2014

JF, - 21

4.6242 GJ/hr

Since f > I,. the minimum point must lie in the section from stage 24 to stage 26. The
section to be searched was finally reduced to three stages. The reboiler duties
corresponding to two of these stages as feed stage (24 and 26) were already obtained.
The final column simulation performed with feed stage J F I - 25 yielded a reboiler
duty II = 4.6168 GJ /hr. This was larger than the reboiler duty (I, ~ 4.5997) obtained
for feed stage J F, = 24. Therefore the optimal feed location was found to be stage 24
from the top.
Table II lists the results obtained from the MMSP program for the above search
procedure. The total number of column simulations used in this search procedure was
5 as compared to 13 used in the case study. In Table III we compared the computer
times used and the computer costs for these two procedures. An IBM 370/168
computer was used for the simulations. The advantage of using the Fibonacci Search
procedure becomes more obvious when larger numbers of stages are in the section to
be searched.

TABLE II
Results of Fibonacci Search Applied to the Sample Problem
Feed
Stage No.

a,

a,

Reflux
Ratio

Boilup
Ratio

Condenser Duty
GJjhr

Reboiler Duty
GJjhr

2\
26
23
25
24

0.31\6
0.6\80
0.0695
0.3422
0.\088

0.\024
0.\789
0.0610
0.0986
0.0589

1.4771
1.4417
1.3983
1.3942
\.3838

3.5292
3.48\4
3.42\ 0
3.4156
3.40\3

4.6349
4.5703
4.4887
4.4818
4.4625

4.7702
4.7056
4.6242
4.6\68
4.5977

TABLE III
Comparison Between Fibonacci Search and Case Study

Fibonacci
Search
No. of Column Simulations

Total No. of Iterations


(DSTLBT)
Total CPU Time. Sec.
(IBM 370/168)
Total Cost. $
(Printing & CPU)

Case Study

\3

20

50

3.27

8.78

4.36

20.00

OPTIMAL FEED LOCATION

663

Downloaded by [University of Sydney] at 05:20 30 December 2014

CONCLUSION
A computational procedure using the Fibonacci search technique has been developed
for the location of the optimal feed stage within a section of the column such that the
minimum energy consumption of the distillation column can be achieved. The
procedure is simple and can be readily incorporated into any rigorous distillation
simulation module. This procedure has been used in Union Carbide for relocation of
the feed stage in many existing columns. During 1978 over 25,000 column simulations
have been performed by use of the MMSP computer program. Eight percent of these
simulations used the Fibonacci search procedure for optimal feed location. Most of
them were for feed stage relocation in existing distillation columns to reduce energy
consumption in the plants. The saving of energy cost was estimated to be over a million
dollars per year.
LITERATURE CITED
I. Beveridge, G.S.G .. and Schechter, R.S., Optimization Theory and Practice. McGraw-Hili Book Co..

1970.
2. Chao, K.C .. and Seader, J.D., "A General Correlation of Vapor-Liquid Equilibrium in Hydrocarbon
Mixtures," A./.Ch.E. Journal, 1,598 (1961).
3. Floyd, E.R., and Hipkin, H.G., "Locating Feed Trays in Fractionators," Ind. Eng. Chem., 55 (6), 34
(1963).
.
4. Gilliland, E.R .. "Multicomponent Rectification-Optimum Feed Plate Composition," Ind. Eng.
Chern.. 32, 918 (1940).
5. Himmelblau. D.M., Applied Nonlinear Programming, McGraw-Hili Book Co., 1972.
6. Luyben, W.L., "Steady-State Energy Conservation Aspects of Distillation Control System Design,"
Ind. Eng. Chem. Fundamentals, 14 (4), 321 (1975).
7. Maas, J.H., "Optimum Feed Stage Location in Multicomponent Distillations," Chern. Eng., April 16,
1973, page 96.
8. McCabe, W.L., and Thiele, E.W., Ind. Eng. Chem., 17,605 (1925).
9. Naphtali, L.M., and Sandholm, D.P., "Multicomponent Separation Calculation by Linearization,"
A./.Ch.E. Journal, 17 (1),148 (1971).
10. Prengle, H.W., Crump, J.R., Fang, e.S., Frupa, M., Henley, J., and Worley, T., "Potential for Energy
Conservation of Industrial Operation in Texas," Final Report on Project SID-10, Governor's Energy
Advisory Council, The State of Texas (1974).
II. Scheibel, E.G., and Montross, C.F., "Optimum Feed Tray in Multicomponent Distillation Calculations," Ind. Eng. Chem .. 401 (8), 1398 (1948).
12. Shipman, e.W., "On the Optimum Choice of Feed Stage in Staged Equilibrium Processes," A.I.Ch.E.
Journal, 18 (6),1253 (1972).
13. Sujata, A.D., "Absorber-Stripper Calculation Made Easier," Hydrocarbon Processing and Pet. Ref.
40 (12),137 (1961).
14. Tang, Y.P., "Multicomponent Distillation Calculation by Newton's Method," Union Carbide Chemicals and rlastics. Engineering Department Report. June 7,1972.
15. Union Carbide Chemicals and Plastics, "User's Manual for the Integrated Process Engineering
System," 1976.
16. Waller, K.V., and Gustafsson, T.K., "On Optimal Steady-State Operation in Distillation," Ind. Eng.
Chern. Process Des. Dev., 17 (3), 313 (1978).
17. Wang, J.e. and Henke, G.E., "Tridiagonal Matrix for Distillation," Hydrocarbon Processing," 45 (8),
155 (1966).
.
18. Wang, J.e., "User's Manual for Simulation of Multicomponent Multistage Separatian Processes,"
Union Carbide Chemicals and Plastics, Engineering Department, 1977.
19. Wilde, DJ., "Optimum Seeking Method," Prentice Hall, Inc., Englewood, N.J., 1964.
20. VanWinkle, M., and Todd, W.G., "Optimum Fractionator Design by Simple Graphical Methods,"
Chern. Eng.. September 20,1971, pp. 136-148.
21. Yen, L.e., and Alexander, R.E., "Estimation of Vapor and Liquid Enthalpies," A.I.Ch.E. Journal. II
(2),334 (1965).

You might also like