APPLICATIONS OF MULTIOBJECTIVE

OPTIMIZATION IN CHEMICAL ENGINEERING

V. Bhaskar,1 Santosh K. Gupta2 and Ajay K. Ray1*

1
Department of Chemical and Environmental Engineering
National University of Singapore
10, Kent Ridge Crescent, Singapore 119260, Singapore
2
Department of Chemical Engineering
University of Wisconsin, Madison, Wl 53706-1691, USA

ABSTRACT

Multiobjective optimization involves the simultaneous optimization of

more than one objective function. This is quite commonly encountered in
Chemical Engineering. A considerable amount of research has been reported
in this area over the last twenty years. This is reviewed in the present paper.
The general background of this area is presented at the beginning, followed
by a description of how the results can be described in terms of Pareto sets.
We then present the several methods available for generating these optimal
solutions. Applications of optimization in Chemical Engineering wherein
multiple objectives are encountered, as well as special adaptations of the
basic algorithms required to solve these problems, are then discussed. Some
comments are also made on possible directions that future research may take
in this area.

Keywords: Multiobjective optimization, Pareto sets, ε-constraint technique,

Genetic Algorithm, Non-dominated sorting genetic algorithm.

1. INTRODUCTION

Present day Chemical Engineering is associated with core competencies

in four major areas: reaction engineering, transport phenomena, separations

2
On leave from: Department of Chemical Engineering, Indian Institute of
Technology, Kanpur 208016, India. E-mail: [email protected]
To whom correspondence should be addressed; E-mail: [email protected]

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications qfMultiobjective Optimization
In Chemical Engineering

science, and computational and systems science. Several paradigm shifts

have taken place in this discipline over the years. These include the
introduction of mathematical modeling in its various forms (including
process control, systems approach, etc.), the shift from unit operations to
transport phenomena, the recent transition towards biosystems, etc. However,
one facet of chemical engineers remains unchanged, namely, that they have a
responsibility of integrating the Chemical Engineering core constituencies
with economic parameters so as to achieve commercial success. In this
context, optimization of chemical processes plays a key role in Chemical
Engineering. Optimization techniques have been applied to problems of
industrial importance ever since the late 1940s and several excellent texts
(Beveridge and Schechter, 1970; Bryson and Ho, 1969; Edgar and
Himmelblau, 1988; Lapidus and Luus, 1967; Ray and Szekely, 1973)
describe the various methods with examples. The complexity of the problems
studied has increased as faster and more powerful computational hardware
and software have become available. We believe that solutions of even more
sophisticated optimization problems will become available as complex
biosystems are studied and are exploited commercially in the future. In this
decade, particularly, several industrial systems have been optimized, which
involve multiple objective functions and constraints, using a variety of
mathematical techniques and robust computational algorithms. In a few
cases, the optimal solutions have been implemented in industry with some
success. This short review describes these recent studies, and illustrates them
with their adaptations and applications. Some conjectures at a conceptual
level are presented thereafter, and it is hoped that these will mature to fruition
as more work gets reported in this interesting area of activity.
Optimization of chemical processes has been a fascinating field of study
for several decades. Until about 1980, virtually all problems in Chemical
Engineering were optimized using single objective functions. Often, the
objective function (also called the cost function) accounted for the economic
efficiency only, which is a scalar quantity. The reason for solving such
relatively simple optimization problems was possibly, the limitations posed
by the technology of computing systems. Most real-world Chemical
Engineering problems require the simultaneous optimization of several
objectives (multiobjective optimization) which cannot be compared easily
with each other (are non-commensurate), and so cannot be combined into a
single, meaningful scalar objective function. Examples include reliability,
safety, hazard analysis, control performance, environmental quality, etc.,

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

apart from the major goal of achieving economic efficiency. Until a few years
ago, these several objective functions were combined into a single scalar
objective function, using arbitrary weight factors, so that the problem could
become computationally tractable. This 'scalarization' of what is really a
vector objective function suffers from several drawbacks. One is that the
results are sensitive to the values of the weighting factors used, which are
difficult to assign on an α-priori basis. What is even more important is the
less-recognized fact that there is a risk of losing some optimal solutions
(Chankong and Haimes, 1983; Haimes, 1977). This happens if the
non-convexity of the objective function (a set is said to be convex if the line
joining two points of the set lies within that set) gives rise to a duality gap.
[Numerically, a duality gap is the difference between the primal and the dual
objective values (Farber, 1986; Goicoechea et al., 1982). Dual is another
mathematical program with the property that its objective is always a bound
on the original mathematical program, called the primal. If the primal is
feasible, the dual cannot be unbounded, and vice versa; if the dual is feasible,
the primal cannot be unbounded. A dual provides a sufficiency test for
optimality, for if feasible χ and y can be found such that f(x) = F(y), it
follows that χ is optimal in the primal and y is optimal in the dual].
Therefore, it is necessary that a vector form of all the objective functions be
used in formulating and solving real-life optimization problems. Such
multiobjective optimization problems form the subject of this review.
The concept of multiobjective optimization is attributed to the economist,
Pareto (1896). After several decades, this concept was recognized in
operations research and has recently become popular in engineering.
Extensive literature is now available on multiobjective optimization but we
limit ourselves to problems of Chemical Engineering interest only.
We start with a discussion of a simple example of a multiobjective
optimization problem to illustrate some elementary concepts. Most
techniques of solving such problems comprise of two phases - an objective
phase, which is more precise and mathematical, followed by a subjective
phase, which is statistical in nature and involves intuition and interaction with
human beings (called decision-makers, DMs) who are familiar with the
industrial operation being studied. In the first phase, a set of several optimal
solutions is generated. This is referred to as the Pareto set. For example, a
typical two-objective function minimization problem can be represented
mathematically as

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16. No. 1, 2000 Applications ofMultiobjective Optimization
In Chemical Engineering

MinI(x)([I,(x),I 2 (x)] (a)

subject to (s.t.):
Model equations; (b)
gj(x)<0, j = l , 2 , . . . , J (c)
hk(x) = 0, k =1,2,..., K (d) (1)

In Eq. 1, χ represents a p-dimensional set (vector) of ρ design or decision

variables. One needs to obtain the optimal values of x, which will minimize
the two individual objective functions, I ( and 12 (which are constituents of the
vector, I), while satisfying the requirements on g, and hk. g/x) and hk(x)
represent several inequality and equality constraints which limit the choices
ofx.
It is found that the solution of Eq. 1 often (but not always) comprises of
several sets of optimal values of x, and is not necessarily a unique, single
point in the p-dimensional x-space. These several solutions correspond to
different values of 11 and I2. Figure 1 shows the optimal solutions of Eq. 1
schematically, for a typical problem. Each point in the I2 vs. 11 plot in Figure
1 corresponds to an optimal solution, x (=[xi, x2,..., xp]) of Eq. 1. The curve
in Figure 1 is referred to as the Pareto set (Chankong and Haimes, 1983). If
we consider two points, A and B, on this Pareto set, we find that on moving
from one to the other, one objective function improves (decreases) while the
other one worsens (increases). It is, therefore, not possible for us to identify
which of these two points is superior to the other. Such points are called as
non-inferior or non-dominating points since they are equally good. More
formally and generally, non-inferior (or non-dominating) points are those for
which, on moving from one point to the other, an improvement in one
objective function cannot be obtained without deterioration in the other. Of
course, several additional solutions, x, also exist which satisfy all the
constraints in Eq. 1, but these are not optimal in any manner, and are
excluded from Figure 1, since these are inferior to (or are dominated over by)
the points shown in this diagram. We illustrate this using point C. If we
compare points A and C, we observe that these two points are
non-dominating. However, point C is not a member of the Pareto set because
point B is superior to C in terms of both the objective functions. The Pareto
set narrows down the choices available to a decision-maker, who has to
consider only these points for further evaluation. The Pareto set is obviously
the first step for an optimization problem and is extremely useful.
Point U, in Figure 1, is referred to as the Utopia. This is the point at which

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et αϊ. Reviews in Chemical Engineering

I,

Pareto-optimal set

I,' I,

Fig. 1: Schematic of a Pareto set for a typical two-objective function

minimization problem. A and B are non-inferior points; C is an
inferior point compared to B; U is the Utopia.

the two asymptotes of the. Pareto set, meet. The asymptote, I, = I; * =

constant, can be obtained by solving the single objective function
optimization problem in which we minimize only I,(x). Thus, Eq. la is
replaced by Min I,(x), with Eqs. Ib-d unchanged. It is clear that point U is not
a solution of Eq. 1 [else it would have dominated over all the points of the
Pareto set, and would have been the only (unique) solution to the problem],
but represents an 'ideal'. This point is a reference point used quite often for
comparing optimal solutions. Pareto sets where one objective function is to
be minimized while the other is to be maximized, or where both the objective
functions are to be maximized, can be drawn in a manner similar to Fig. I. In
this review, however, we will discuss problems involving the minimization of
all the objective functions only. This does not lead to any loss of generality
since it is known (Deb, 1995) that one can easily replace the maximization of
any objective function, say, I;, by the minimization of a function, Fj. Several

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

possible functions could be used (which do not change the location of the
optima), and a simple and popular transformation is F; s 1/(1 +1,).
A classic example of a 2-dimensional Pareto set (two-objective function
problem) in Chemical Engineering is provided by an isothermal batch reactor
in which the following series reaction is taking place:

(2)

We are interested in maximizing both the yield and the selectivity of the
desired product, B, simultaneously. These are defined by

molof desiredproduct.B, produced

mol of reac tan t fed
,~ mol of desired product, B, produced
Selectivity of B = (3)
mol of all products formed

Maximization of the yield is important since it leads to higher amounts of

B. Maximization of the selectivity is desired since it leads to a reduction in
the downstream separation costs. The yield and selectivity can be computed
quite easily as a function of time, t, for any feed and temperature, by
integrating the mass balance equations for this system. If we consider the
simple problem where t is the only decision variable, and select a feed of pure
A, and with temperature selected such that k2/k| = 0.1 and k ι = 0.1, we
obtain the results shown in Figure 2 (curve PQR, for 0 < t < 50). In region
PQ, point Q dominates over all the other points and so would be the natural
choice for the point of operation. However, the choice of the operating point
is not as easy in region QR. In this region, one objective improves and the
other worsens as we move from one point to another. Thus, QR is the Pareto
set of equally good, optimal points for this problem. The associated values of
the decision variable, t, for each of these points is also shown in Figure 2.
The choice of the operating point is to be decided based on some additional
criteria. Point U represents the Utopia. The existence of a Pareto set even in
such simple problems is not usually recognized or discussed in Chemical
Engineering texts. Some more classical multiobjective optimization problems
in Chemical Engineering have been presented by Seinfeld and MacBride,
1970; Nishitani and «workers, 1979, 1980, 1981; Shieh and Fan, 1980;
Umeda et al., 1980 and Grossmann et al., 1982.

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
 Reviews in Chemical Engineering

0
0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05
Selectivity
Fig. 2: Pareto optimal solutions (QR) for the yield-selectivity maximization
problem, for the reaction, A— l —>B 2
—»C, in an isothermal
batch reactor. Feed is pure A, k( = 0.1, k2/ki = 0.1. PQ represents the
inferior set. Corresponding plot for time vs. selectivity also shown.

The generation of the Pareto set (QR in Figure 2) was quite easy, and was
done using qualitative arguments once the yield vs. selectivity plot (PQR in

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

Figure 2) was obtained by integrating the mass balance equations. No

sophisticated optimization algorithm was required for this simple problem.
Several techniques are available to generate the Pareto sets for the more
general problems. These are discussed later in this review.
The generation of the Pareto set completes the more precise Phase I of the
optimization process. In the more subjective Phase II of the study, one has to
use additional information that is often intuitive and non-quantifiable in
nature, to choose an operating point from among the entire Pareto set, for
operation. This point is referred to as the preferred solution. A few methods
are available for obtaining this point. One is to ask several decision-makers to
rank the points on the Pareto set on a ΙΟ-point scale, using their intuition or
'gut' feeling. A weighted-average ranking is then computed and the 'best'
point is selected. This second phase of the optimization procedure is not
discussed here in too much detail.
It is to be noted that multiobjective optimization is becoming quite
popular in Chemical Engineering. In this review, we shall attempt to describe
the various techniques and their adaptations, which have been used to obtain
the Pareto solutions for several interesting and complex problems of interest
to chemical engineers. It is hoped that this review would spur more activity in
this area and also help in the recognition of Pareto sets where these have,
indeed, been obtained by experimentation or simulation. Table 1 summarizes
the several multiobjective optimization problems studied in Chemical
Engineering and classifies them into different areas. Table 2 provides more
details on the techniques used for optimization, methods used for the
generation of the non-inferior Pareto set and for decision-making, and also
gives some important comments on the problem studied.

2. TECHNIQUES OF GENERATING PARETO SETS

Extensive research has been reported on the algorithms used for

generating the non-inferior Pareto solutions. These are described in several
textbooks (Zeleny, 1974, 1982; Cohon, 1978; Hwang and Masud, 1979;
Steuer, 1986; Haimes et al., 1975, 1990; Chankong and Haimes, 1983;
Goicoechea et al., 1982) and research and review articles (Geoffrion,
1967a-c; Geoffrion et al., 1972, Zionts and Wallenius, 1976, 1980, Hwang et
al., 1980, Clark and Westerberg, 1983, Srinivas and Deb, 1995). We present
a short summary of the different techniques used.

8
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et αϊ. Reviews in Chemical Engineering

Table 1
Multiobjective Optimization in Chemical Engineering:
Based on Areas in Chemical Engineering

PROBLEM DESCRIPTION CONTRIBUTORS

PROCESS DESIGN AND CONTROL

Multicriteria optimization in chemical processes Seinfeld and MacBride, 1970

Optimal process systems synthesis Nishidaetal., 1974

Multiple effect evaporator design Nishitani et al., 1979

Aeration vessel for waste water treatment Nishitani et al., 1980

Optimal synthesis of methanation process Shieh and Fan, 1980

Toluene -steam dealkylation process Umeda et al., 1980

Modeling petrochemical industry Sophosetal., 1980

Energy resource conservation -Evaporator system Nishitani et al., 1981

Process design with dynamic aspects Lenhoff and Morari, 1982

Chemical complexes Grossmann et al., 1982

Design of chemical plants with robust dynamic operability Palazoglu and Arkun, 1986

characteristics

Complex chemical process system: Benzene chlorination K llneretal., I989

process

Modular multivariable controller Meadowcraft et al., 1992

Binary distillation synthesis Luyben and Floudas, 1994a

Reactor-separator-recycle system Luyben and Floudas, 1994b

Minimum bias control of a paper machine process Liu et al., 1997

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1. 2000 Applications of Multiobjective Optimization
In Chemical Engineering

Table 1 continued

POLYMERIZATION REACTION ENGINEERING

Multiobjective dynamic optimization of semibatch Tsoukas et al, 1982.

copolymerization reactors/process. Butalaetal., 1988,

Choi and Butala, 1991

Optimization of chain propagation with monomer termination Fanetal., 1984

in a batch reactor

Steady state optimization of continuous co-polymerization Farber. 1986

reactors
Dynamic optimization of a Nonvaporizing nylon 6 batch Wajge and Gupta, 1994

reactor

Optimization of an industrial semibatch nylon 6 batch reactor Sareen and Gupta, 199S

Dynamic optimization of an industrial nylon 6 semibatch Mitraetal., 1998

reactor using genetic algorithm

Free radical bulk polymerization reactor using genetic Garg and Gupta, 1999

algorithm

Industrial nylon 6 semibatch reactor system using genetic Gupta and Gupta, 1999

algorithm

Industrial wiped film PET finishing reactor Bhaskaretal.,2000

PMMA reactors and film production Zhou et al., 2000

BIOCHEMICAL ENGINEERING / WASTE TREATMEN'Γ

Waste water treatment system Takama and Umeda, 1980

Design and operation of anaerobic digesters using Videlaetal., 1990

multiobjective optimization criteria

10
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
K Bhaskar et al. Reviews in Chemical Engineering

Table 1 continued

Fuzzy multiobjective optimization for self-sucking impellers Kraslawskietal., 1991b

in a bioreactor

Waste treatment costs in discrete process synthesis and Ciricand Huchette, 1993

optimization problems

Carboxy-methylation of starch Tijsenetal., 1999

MIXING

Kenics static mixers Kraslawski and Pustelnik,

1991a

CATALYTIC REACTORS

Steam reformers optimization using GA Rajeshetal.,2000

ELECTROCHEMICAL PROCESS

Electrochemical reduction process of maleic-acid to succinic Smigielskietal., 1992

acid

AIR POLLUTION CONTROL

Cyclone separators Ravietal.2000a

Venturi scrubbers Ravietal.2000b

MEMBRANE SEPARATION MODULES

Beer dialysis Chanetal. 2000

GENERAL

Tools for multicriteria decision making Clark and Westerberg, 1983

Parametric optimization approach for multiobjective Dimkou and Papalexandri,

engineering problems involving discrete decisions 1998

11
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

ction pro m has also been solved.

ective fu on case. Two objective
eto surfa d e n e d for
optimiza
A multil

cess.
thod
lied

is
w
M
Technique used fo
decision making

ll
||||J 1
l Hi
•2 e ζ- αe·
•S §8
ο _ I «
Ά
Is
0 3
II
ls

il
5s
ε s
I|

; ·-

Z t,

12
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
 Λ
JO
«s
υ
o
T3
υ
V. Bhaskar et al.

Mi

_ 'S
S ts
° 's
we
S.-Π Μ

«—
™
is

la l
1. Maximization of co
o: > t a
SIJi

g^ii
2

2 S SS
J£S:S
si-s
ί Ss

i Mi
-2 2 ε s

molecular weight
Iι_ Iυ S«ο ρl

|
w
O
1. Maximization of co
molecular weight
Styrene-acrylonitrile syst
1. Maximization of co
molecular weight
Opt design o the Discretization techniques empli

Authenticated | 134.99.128.41
withrobustdynamic op circumvent [he problem of inflr

Download Date | 9/24/13 12:39 PM

number of constraints in the
optimization problem
Reviews in Chemical Engineering

13
Brought to you by | Heinrich Heine Universität Düsseldorf
 Vol. 16, No. 1, 2000 Applications ofMultiobjective Optimization
In Chemical Engineering

g 11 i ·=
3! S-i |:
!S|&f-g~
sii*u*
ΐΙ|ϊ·5ΐ|
li|5ll|
Iff§!£!
gJ g« 3 3g

I!
*·£
iε
1-2
Ξ «η

.3·
.Ξ
*j
c l
.2

JU i
IS

l
!|
li
i!
13
.ε <t
o

III
II
11!

II

14
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et αϊ. Reviews in Chemical Engineering

24*
a 3ο -β
Ο ·* u
<= €
5o —P.a
« ο

i
oill*
· S.J H o S

o.

J
f
•σ
\Λϊ Ι
S o ** ,S β
ω
*·«ΐ
Έ|ϊί £§·!
ug Σ Is .Ε 5 « !

ee
H

l
li
15
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
 16
ι2
,C

Ξ
υ
T3

0
υ
3

"
.C

β>
0

-—
cu
(Λ

.C

ΐέ
^

ο
ώ
Ι
S

Ο
Equality constraint on the desired

Ν
Ι

I
1
if
ο

Ο
number average molecular weight ι
1

II
1ft
J-jj.2
lit
«|J
Stopping condition on the desired

•-ε

i -i -s
ι||
ϊ|!
conversion imposed. Smoothening
'S

•nil
1 »1 8
Tg
1=
the Pareto sets to yield suboptimal
Ko/. 7r5, No. 1. 2000

solution sets carried out.

| Sι
_

.2 .a E .g >. § Ε

<« ΟΊ£ S 3 β > 7.

-·|1·ί1
^C

•e
.£

w
Ο

??
E

CO
C

as
ON

Ϊ
•S

ε
-5
I
eft
Equality constraint on the desired

η
^

i
1

-i S
ο—
ll
number average molecular weight ι

Η Isis t ll
£·!
Έ

.S 15 α
.S "5 £
**· c c
8«
Stopping condition on the desired

S S-8
8 -| §

ll!
-8a

conversion imposed. Adaptation m

to give optimal vapor release rate
history (continuous variable rather
a discrete point). Also, optimal vat
J* SS "Sa

the jacket fluid temperature genera

.a
^^
.C

•β
M
E

O
n
J

a
j^i
Endpoint constraints on the numb«

i
1
1
Έ

If
If
?g
"So
0 <

ll
II SS
average chain length and the mono

Ί ^-,

-il
8.
,1

|ί

11
!
conversion incorporated. Unique
optimal solution obtained.
Ε

V
-H
^c

ο
CO

*i

a
j;

S
Equality constraint on the desired

SP
I
1

fa
ll

O2
number average molecular weight ι

Jif

Authenticated | 134.99.128.41
.Is«

if"
's § §
.lit
11.
Stopping condition on the desired

111
1 e1

|||

Download Date | 9/24/13 12:39 PM

|
conversion imposed. Optimal tract
control valve opening and the optil
value of the jacket fluid temperatui
1 Is,

generated.

Brought to you by | Heinrich Heine Universität Düsseldorf

In Chemical Engineering
Applications of Multiobjective Optimization
 ο

es
«S
.c

υ
T3

2
υ
f

.a
E

C
"c
Q
c

v
•s
'S
Ο

r
00

S
Mapping of the decision varial

J,

S
imization ofthe methane feed rate

S
1

1
I«

Ό Ο
3 ?

S ΐ
C ^j
done inside the NSGA code tc

<"
imization ofthe flowrate of CO in

zS.
[as for a fixed production rate of the bounds chromosome .§

c
Ogen in an existing side-fired steai speciflc(dcpending on the cho

.ε 9 ρ -5 £
rmer other decision variables).

ο
.a

Si
ε

C
<S
'c

m
c
I

Γ*
•3
NSGA adapted to use inequali

0
0
00
£·

s
°
imization ofthe residence time of

5
1
Ή

1
M

„
constraints on certain product

·§ α<
mer melt and the concentrations ο

!
«

£C„
isirable side products formed in th properties by "artificial killing
«trial continuous wiped film PET chromosomes in the process ο
?w^
.2

'S
tor.Equality constraint on the desil optimization. Unique optimal

Σ i S·!· e Σ & i l S-S

ee of polymerization imposed. obtained.

£ £-3 -g a a
•J

.Ξ
.0
E

Γ»
c

'S
·§

0
S

c
&
c

S
<a
Additional bounds and over-ri

is
Limization ofthe overall collection

1
1

JS_ ">Ο
^
constraints needed to be impo;

Sz,
;iency and minimization ofthe pre

If

ΣΪ Ι
* IP e
sg
ι / overall cost ofthe cyclone sepai Bounds were made chromosol
specific.

.a
ο

1"»
(Λ
^

W
c

g
.··

•i

I
ee
limization ofthe percentage remo\ Procedure to develop useful d

8
a
1

1
Έ

•g Ο
charts representing trade-offs ι

1<
llcohol from beer and minimizatio

ζ ε.

J •5
!·-

«
U s f II

ϊ2*
and few other optimization pn

•s^-S
•emoval ofthe 'extract' (taste chem |
studied.

—
_c
ε

TS
-S

.2
g

e
g
s

3
00

s
s
limization ofthe monomer conven Constraints on the number avi

1
1
1

oOT
·§1<
o
g

fi it

ss-section average value) and molecular weight and the tem|

a S=
imization ofthe length ofthe film imposed

S
£
tor in the continuous casting proce

E S£
5
vIA

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Reviews in Chemical Engineering

Brought to you by | Heinrich Heine Universität Düsseldorf

Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

2.1 Utility Functions

In this technique, one defines a scalar utility function, u(I), of the
different objective functions, Ii, li,..., l„:

u(l) = f(I„l2,..,I„) (4)

The utility function is optimized to obtain the preferred solution. The utility
function is an abstract variable indicating goal attainment. For a chemical
engineer, the objectives in a typical industrial setting could either be
quantifiable quantities, as, for example, minimization of the cost of
production while meeting specifications on the products and effluents, etc.,
or, a decision maker (DM) could define the utility to include intuitive
quantities, e.g., operability and controllability of the plant, reliability of the
process, etc. In this interpretation, the possibility of measuring the utility does
exist, though formal techniques are not available. Thus, the utility is a way of
allowing the decision maker to describe his goals in an abstract way. The
drawback of this technique is that the utility function, usually, cannot be
determined systematically (Haimes et al., 1975). No Chemical Engineering
examples have been solved using this method, to the best of our knowledge.

2.2 Indifference Functions

This method is based on a relative comparison of the objectives. The DM
is expected to specify the preference of one objective over the other.
Indifference functions/curves (Chankong and Haimes, 1983) are usually
plotted to obtain the relative worth of the individual objective functions. The
indifference trade-off or the marginal rate of substitution (which is always
present for any two points on the indifference curve) is defined as the amount
of degradation of one objective the DM is willing to tolerate in exchange for
the amount of improvement in the other objective while the preferences for
the two points on the indifference curve remain the same. The application of
this technique is limited to a maximum of two or three objectives, since a
geometrical representation and analysis is required. Some form of reduction
of the dimensionality is necessary when there are more than three objective
functions, which makes this procedure quite cumbersome. Again, this
technique has not been applied to problems in Chemical Engineering.

18
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

2.3. Lexicographic Approach

The objective functions are classified (by 'priority') by the DM and the
preferred solution is defined as that which simultaneously optimizes as many
of the objectives as possible, hierarchically. The DM is asked to rank the
criteria in terms of decreasing importance (say, I b I2,..., ln). Only the
alternatives that yield the most 'preferred' value for I| are used for further
evaluation. This solution set is used for the next round to get the most
preferred value of I2. This procedure is repeated until all the criteria are
satisfied to yield the best solution. If the best solution is not unique, then
additional attributes are used to choose the best solution. This approach
generated considerable interest because it mimics the process of human
decision-making. But the problem with this approach is that the solution is
very sensitive to the ranking of the objectives by the DM and, therefore, one
should be careful in applying this technique to objectives having almost equal
importance. Different results may be obtained for such cases depending upon
the ranking of such objectives by the DM [Haimes et al., 1975; Chankong
and Haimes, 1983]. Recently, the lexicographic technique was used by
Meadowcraft et al., 1992, for the design of a modular multi variable controller
(MMC). Both steady state and dynamic characteristics were considered. The
controller thus designed could handle the list of desired operational
objectives, re-prioritization of the control objectives and constraints. The
controller also provided explicit fault-tolerant control in the presence of
instrument failures and allowed manual control of the failed subsystems. The
controller finds the solution where as many objectives are satisfied as
possible, in order of importance, by using a combination of lexicographic
technique and the generalized goal programming approach (discussed in Sec
2.6). Inequality constraints are classified into hard and soft constraints
depending on their relative importance to the other objectives and are
incorporated into the design procedure.

2.4. Parametric Approach

This is the simplest of all the techniques. The objective functions, I,, are
combined into a single overall scalar objective, I:

19
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16. No. I, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

(5)

where χ e X, X being the feasible optimum region. A non-inferior point is

generated by changing the relative weights, Wj. The solutions may be viewed
as points in the optimum search-space that offer the least conflict among the
objectives. Least conflict is possible when the weights used are equal for all
the objectives, which is not a practical, satisfying solution for a real system.
The relative weights need to be assigned to each of the objective functions
based on the individual optimum solution. The major drawback of this
technique is that it cannot generate the entire non-inferior set when
"duality-gaps" exist, which means that the problem is non-convex (Haimes et
al., 1975; Chankong and Haimes, 1983; Srinivas and Deb, 1995). It is well
known that verifying the convexity requirement for large-scale systems is
quite difficult. Also, for such systems, α-priori knowledge of the weights is
practically not possible. Though this method gives a solution to the overall
objective, the insight into the conflicts among the objectives is lost in the
process. Some of the classical examples in Chemical Engineering have used
the parametric approach in solving the multiobjective problem. Seinfeld and
MacBride, 1970 have used a modification (Geoffrion, 1967c) of this method
to minimize the parametric sensitivities in a refinery. A simplified refinery
model was formulated and the objectives considered were (i) maximization
of the total yearly profit and (ii) minimization of the sensitivity of the profit
to variations in refinery conditions. Ten constraints (both equality and
inequality) were imposed and twelve decision variables were considered in
this study. Fan et al., 1984, have used this approach in the optimization of
batch polymerizations involving chain propagation with monomer
termination. Trade-offs among the monomer conversion and the mean and
the standard deviation of the molecular weight distribution (MWD) of the
polymer, have been explained using a triangular graph. Though the method is
fairly simple to use, the accuracy of the solution is not guaranteed.

20
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

2.5 The ε-Constraint Approach

This approach is one of the most popular methods that have been applied
for solving multiobjective optimization problems. This technique was
proposed by Haimes et al., 1971 and is described in detail by Haimes et al.,
1975 and Chankong and Haimes, 1983. This technique does not require the
existence of supporting hyper-planes and overcomes duality-gaps in
non-convex sets. These advantages have made this approach very useful and
popular.
In this technique, we optimize a single objective function (chosen from
among the original ones) while treating all the remaining objectives as
inequality constraints. A multiobjective optimization problem defined similar
to the two-objective function problem defined in Eq. 1 is, thus, reformulated
in this technique as
Min I, (x)
s.t.:

hk(x) = 0 k =1,2,..., K (6)

It is to be mentioned that any one of the n objective functions, I b I2,..., !„,

can be selected as the single objective function to be minimized (and is
referred to as 1,) [we shall assume that all the objective functions are to be
minimized, since maximization problems can easily be transformed into
minimization problems, as discussed earlier].
By varying the constraint-levels, EJ, one can generate the entire Pareto
optimal hyper-surface in the η-dimensional I-space. The minimum values of
the individual 8j have to be obtained by solving simpler, 1 -objective function
optimization problems (Chankong and Haimes, 1983). Any appropriate
technique can be used to solve the optimization problem in Eq. 6. It is clear
that this simpler optimization problem (Eq. 6) has to be solved several times
to generate the Pareto set, and so this method is quite compute-intensive.
Three different adaptations of this technique are available: the equality
constraint approach, the inequality constraint approach and the hybrid
(weighting-constraint) approach. The equality constraint has the added
advantage that analytical solutions for small-size problems can be obtained.
Wajge and Gupta, I994a. used this form to optimize a non-vaporizing nylon

21
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. I, 2000 Applications ofMulliobjective Optimization
In Chemical Engineering

6 batch reactor. They minimized the (i) reaction time, tf, and (ii)
concentration, [C2]f, of the undesired side product, the cyclic dimer, in the
product, while requiring the monomer conversion and the number average
chain length in the product to lie at design or desired values (xm.d and μη,α,
respectively). The temperature history, T(t), was used as the decision
variable. Pontryagin's minimum principle was used with the first-order
control vector iteration method (Ray and Szekely, 1973) to solve the problem
with the ε-equality constraint incorporated. These workers reported some
numerical oscillations in the Pareto sets. They also encountered convergence
problems when using larger values of ε. Sareen and Gupta, 1995, used the
same objective functions and end-point constraints (on the product) to
optimize an industrial semi-batch nylon 6 reactor. In this reactor, monomer
and water vaporize and build up the pressure in the vapor space above the
liquid reaction mass, at a rate depending on the release of the vapor mixture
through a control valve. They used the jacket-fluid temperature, Tj (a scalar
value), as well as the pressure history, p(t) (a continuous function of time, t),
in the reactor as the decision variables. The pressure history was
parametrized (shape fixed) and was represented in terms of five parameters
(constants, rather than a function). Sequential quadratic programming (SQP),
developed by Gill et αϊ., 1981 (which has a first-order convergence), was
used for solving the single objective function problem obtained with the ε-
equality constraint method. Some amount of scatter was obtained in the
Pareto set. A 'smoothening procedure' was used to get sub-optimal but
smoother Pareto sets. Substantial improvements in the operation of the
industrial reactor were indicated when operating the reactor at any of the
points on the Pareto set.
The inequality form in Eq. 6 is very useful when the DM wishes to solve
the problem interactively. The Kuhn-Tucker (positive) multipliers associated
with the optimization problem reflect the sensitivity of the primary objective,
Ii, to the active constrained objectives, Ij, and so give an indication of the
trade-offs among the objectives. The values of the Kuhn-Tucker multipliers
indicate the price that one needs to pay in terms of one objective, if one
wishes to improve another, while still remaining on the non-inferior set. The
drawback of this technique is that the test of non-inferiority needs to be
performed. This leads to an increase in the computational time. This method
also locates some of the points that are inferior (Chankong and Haimes,
1983; Clark and Westerberg, 1983). This method has been used extensively
(Sophos et σ/., 1980; Tsoukas et al., 1982; Fan et al., 1984; Farber, 1986;

22
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

Palazoglu and Arkun, 1986; Videla et al., 1990; Luyben and Floudas, 1994 a,
b) in Chemical Engineering.
Sophos et al., 1980, optimized a petrochemical industry using multiple
objectives. These were to optimize (i) the energy utilization (which included
maximization of the change in the thermodynamic availability and
minimization of lost work) and (ii) carbon utilization efficiency (which is
equivalent to minimization of the feed stock consumption). Compromise
'bounding structures' (optimal solutions of each objective) were generated
for the system studied. The two objectives given above were considered first
and a Pareto set was obtained between the feed consumption and the lost
work. Pareto surfaces were also generated for a three objective function
problem using linear programming. Two preferred solutions were identified
from the Pareto surface. In one case, the method used was goal programming
(discussed in Section 2.6) while in another, commensuration of the objective
functions for a specific case was used before solving a single objective
function optimization problem.
Tsoukas et al., 1982, were the first to introduce the concept of multi-
objective optimization in polymerization reactors, where the presence of
conflicting objectives is common. Pareto optimal solutions were determined
for the dual objectives of (i) narrowing the copolymer composition and (ii)
narrowing the molecular weight distribution for styrene-acrylonitrile (SAN)
copolymerization in a semi-batch reactor. Temperature, initiator con-
centration and monomer addition histories (functions of time) were used as
the decision variables. Farber, 1986, extended this work to continuous
copolymerization reactors. They considered both methyl methacrylate-vinyl
acetate (MMA-VA) as well as SAN systems. Pareto sets were obtained
[composition vs. molecular weight; conversion vs. molecular weight] for the
MMA-VA system using the temperature history of the reactor as the decision
variable. In the SAN system, a Pareto optimal set was obtained for
conversion vs. molecular weight by using both temperature history and
residence time as decision variables. Fan et al., 1984, also used this technique
for chain polymerizations in a batch reactor (the same as discussed in the
previous study) and compared the results with those obtained using the
parametric approach.
Palazoglu and Arkun (1986) studied the problems of operability
associated with improper design, modeling and control system design in the
presence of uncertainties. They proposed a process design methodology to
find optimal operating regions for chemical units. This approach guides the

23
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

designer towards process designs with improved dynamic operability and

economics. Characterization of dynamic operability was done using an index
for robustness. The ellipsoid algorithm (Ecker and Kupferschmidt, 1983) was
used for solving the optimization problem interactively. The optimization
problem consists of non-linear problems with an infinite number of
constraints that are frequency dependent, and are handled using a
discretization procedure developed by these workers. A symbolic logic
language, MACSYMA, was used to get the design variables transferred from
the non-linear plant equations to the transfer function, thereby avoiding
successive linearizations. A system of three isothermal CSTRs was used as
an example.
A modified ε-constraint approach has been used for the optimization of an
anaerobic digester treating the leachate from a controlled solid urban waste
landfill. Videla et al., 1990, considered the (i) maximization of the net
production of energy from biogas, (ii) maximization of the percentage
removal of chemical oxygen demand (COD), and (iii) minimization of the
capital outlay, as the three objective functions. The constraints imposed were
(i) the percentage COD removal must be at least 50%;
(ii) the digester must treat a specified flow rate of leachate per day; and
(iii) all the individual objective functions must be positive.
The decision variables were the residence time, the temperature of operation
and the diameter to height ratio. The weighted sum of the deviations from the
respective optima of the individual objective functions was minimized.
Pareto optimal solutions were generated for the net biogas production vs. the
percent COD removal, for different volumes of the digester. An interactive,
iterative optimization procedure, based on the solution of the non linear
programming (NLP) problems was used to obtain the preferred solution for
the two cases. The first of these involved the minimization of the cost while
the other involved the maximization of the net energy production with at least
75% of COD removal.
Luyben and Floudas (1994 a, b) analysed the interaction between process
design and control using a multiobjective framework for two example
systems, a binary distillation system and a reactor-separator-recycle system.
They proposed a systematic approach to determine both the steady-state
economic and open-loop controllability objectives within the mathematical
programming framework of process synthesis. The possible design
alternatives were translated to a multiobjective mixed-integer non-linear
optimization problem. A non-inferior solution set was determined by

24
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

applying the ε-constraint technique within the framework of the generalized

Benders' decomposition (Geoffrion, 1972; Paules and Floudas, 1989). A
method based on the cutting plane algorithm (Geoffrion et al., 1972) was
used to determine a best-compromise solution by the knowledge of trade-off
weights among the economic and control objectives. As explained earlier,
this information is obtained by the partial derivatives from the non-inferior
set. This procedure was applied for binary distillation synthesis involving
multiple objectives, control configuration selection and possible heat
integration. They also applied this procedure for the multiobjective
optimization of a recycle system consisting of a reactor followed by a
distillation column, where the distillate is recycled to the reactor. A
non-linear model incorporating all possible design variables was used for
optimization. It was found how specific changes in design affect both the
economic and control objectives. The conflict between the process and
disturbance gains was very well exhibited by the non-inferior solutions.
The third type of the ε-constraint approach is the hybrid form that
combines the principles of the parametric approach with the ε-constraint
technique. The hybrid form can be represented as

Max/Min wTI (x)

subject to:
0 < Wj < 1

gj(x)<0 j = l,2 ..... J

hk(x) = 0 k =1,2,..., K (7)

The hybrid form does not require any non-inferiority tests and is, therefore,
more efficient when the user is interested in generating only the Pareto
optimal solutions numerically. Interactive decision making is not possible
with this method.
Though the ε-constraint technique can also be used to solve problems
which are non-convex, the difficulty with this method is that an a-priori
knowledge of the appropriate range of values of εί is required.

25
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

2.6. Goal Programming (Charnes and Cooper, 1961)

A preferred solution is obtained by minimizing a weighted average

deviation of the objective functions from the goals set by the DM. This is
represented mathematically as

n
Minl = (8)

where χ e X (the feasible region), and y; is the goal or the demand-level of

the ith objective. Usually, r takes the value of 2 with yj as the individual
optimum of the ith objective. This represents the root mean square deviation
from the goals. The arbitrary selection of the demand-level vector leads to a
non-Pareto optimal solution. Equal weights for all the deviations of the
objectives are normally assumed, as was assumed in Eq. 8. If weights are
introduced, then the problem of a duality gap arises when the problem is
non-convex. Also, the DM has to be aware of the individual optimum of each
objective prior to the selection of the values of the demand levels.
A variation of this technique is the min-max formulation to solve
multiobjective optimization problems. Nishida et a/., 1974, formulated and
solved the synthesis problem as a min-max formulation within the framework
of optimal control. The objective of this method was to minimize the relative
deviations from the individual optimum of the single objective functions,
when r = oo. The method tries to minimize the objective conflict. The best
compromise solution is obtained when objectives with equal importance are
optimized. This method also suffers from the drawback of duality gaps and
sometimes it locates solutions which are inferior.
The technique of goal programming has been applied in Chemical
Engineering by Kraslawski et α/., 1991, Kraslawski and Pustelnik, 1991 and
Smigielski et α/., 1992. Kraslawski et al., 1991, have used this technique for
optimizing the operating conditions for self-sucking impellers in a bioreactor.
The objectives were to (i) minimize the mixing power requirement and (ii) to
maximize the mass transfer coefficient. The decision variables were the rate
of rotation of the impeller, and the diameter to height ratio of the tank. The
problem was transformed to a min-max formulation and the fuzzy weights
determined based on the 'linguistic' variables (qualitative information such
as small, large, very small, very large, etc.) used in the problem.

26
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

The min-max problem-formulation was also used by Kraslawski and

Pustelnik in determining the optimum values of the number of elements and
the flow velocity for a kenics static mixer. The two objective functions used
were: (i) minimization of the pressure drop and (ii) the maximization of the
degree of mixing. The optimal solution was obtained by a sequentially-built
compromise function method, which prioritizes the objective functions (a
form of lexicographic approach) α-priori. Both the methods used
experimental data and the simplex method to arrive at the optimal solution.
The min-max formulation uses the weights decided by the DM whereas the
latter method considers the solution to be feasible if the degree of mixing
(first priority) attains 99% of the maximum degree of mixing. A complete
analysis needs to be made by changing the weights and the priority levels to
give an understanding of the non-inferior solutions. The reader does not get
any insight about the nature and extent of the conflict between the two
objectives in this case.
A multiobjective optimization study (using experiments) was carried out
by Smigielski et al., 1992, on the electrochemical reduction of maleic acid to
succinic acid. The objectives were to (i) maximize the degree of inversion
and (ii) minimize the wastes produced in the process. The decision variables
used were the concentrations of maleic acid and sulphuric acid, current
intensity, temperature, and the electric charge. The optimum operating point
was achieved by formulating a goal-programming problem, where the
demand-level vector consists of values of the Utopia of the individual
objectives. Fuzzy set theory was applied to get the weights in the objective
function. The method is termed as the utopia-point method. The simplex
method was used to arrive at the best compromise solution. Based on
investigations on a pilot plant, a design was suggested for the continuous
production of succinic acid. This work did not use any mathematical model
for the optimization. It used, instead, data obtained from experiments. The
data did not cover the entire feasible region.

2.7 Genetic Algorithm (GA)

G A is a search technique developed by Holland (1975) that mimics the

process of natural selection and natural genetics. In this algorithm, we code a
set of values of the decision variables (a solution, x) in terms of a 'string' (or
chromosome) of binary numbers, generated using random numbers. A
'population' (gene pool) of such binary strings is first generated. Each

27
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

chromosome is then mapped into a set of real values of the decision

variables, using the upper and lower bounds of each of these. This ensures
that the decision variables lie within their bounds. Then, a model of the
process is used to provide values of the objective function for each
chromosome. The value of the objective function of any chromosome reflects
its 'fitness'. The Darwinian principle of 'survival of the fittest' is used to
generate a new and improved gene pool (new generation). This is done by
preparing a 'mating pool', comprising of copies of chromosomes, the number
of copies of any chromosome being proportional to its fitness (Darwin's
principle). Pairs of chromosomes are then selected randomly, and pairs of
daughter chromosomes generated using operations similar to those in genetic
reproduction. The gene pool evolves, with the fitness improving over the
generations.
GA is noted for its robustness. This algorithm is superior to traditional
optimization algorithms in many aspects, and has become quite popular in
recent years. It is better than calculus-based methods (both direct and indirect
methods) that generally seek out the local optimum, and which may miss the
global optimum. Most of the older techniques require values of the
derivatives of the objective functions, and quite often, numerical
approximations of the derivatives are used for optimization. In most real-life
problems, the existence of derivatives is questionable and often, the functions
are discontinuous, multi-modal and noisy. In such cases, calculus-based
methods fail. Enumerative schemes, which are based on the point-by-point
comparison of the values of the objective function in a discretized infinite (or
even a finite) search space, are inefficient for large problems since the search
space is often, too large. Random search techniques, too, suffer from a
similar disadvantage since they work like enumerative techniques in the long
run. GA is superior to these techniques since it is conceptually different from
these traditional algorithms in several respects. It uses a population of several
points simultaneously, as well as works with probabilistic (instead of
deterministic) operators. In addition, GA uses information on the objective
function and not its derivatives, nor does it require any other auxiliary
knowledge.
Three common operators are used in GA [called simple GA (SGA), to
distinguish it from its various adaptations] to obtain an improved (next)
generation of chromosomes. These are referred to as reproduction, cross-over
and mutation. Reproduction, as described above, is the generation of the
mating pool, where the chromosomes are copied probabilistically, based on

28
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

their fitness values. Then, a pair of daughter chromosomes are produced by

selecting a cross-over site (chosen randomly) and exchanging the two parts of
the pair of parent chromosomes (selected randomly from the mating pool), as
illustrated below for two chromosomes carrying information about three
decision variables, each represented by four binary digits:

1 0 0 1 1 1 0 0 0 1 1 1 1 0 0 1 1 1 0 1 1 1 0 0
(9)
0 0 1 1 0 1 0 1 1 1 0 0 0 0 1 1 0 1 0 0 0 1 1 1
parent chromosomes daughter chromosomes
In Eq. 9, the crossover site for this pair of parent chromosomes is taken just
after the fifth binary digit. It is hoped that the daughter strings are superior. If
they are worse than the parent chromosomes, they will slowly die a natural
death over the next few generations (the Darwinian principle at work).
The mutation operator is required for the following reason. In Eq. 9, let us
assume that all the chromosomes in the gene pool have a 0 at the second
position. There is a finite probability of this happening, since the generation
of the binary numbers is done probabilistically. It is obvious that cross-over
will never be able to generate chromosomes with a 1 at this position, and if it
so happens that the optimum is, indeed, located at a point described by such a
chromosome, GA will be unable to reach this solution. The mutation operator
looks at each binary digit in every daughter chromosome in the gene pool,
and transforms a 0 into a 1 (or vice versa) with a small probability. In effect,
it moves the chromosome locally in the x-space, to create a better
chromosome. The entire process is repeated till some termination criterion is
met (the specified maximum number of generations is attained, or the
improvements in the values of the objective functions become lower than a
specified tolerance). A more elaborate description of SGA is available in
Holland, 1975, Goldberg, 1989, and Deb, 1995.
We now turn our attention to extensions of SGA to solve problems
involving multiobjective optimization. Since GA uses a population of points,
it seems very natural to use GA for such problems to capture a number of
solutions simultaneously. Schaffer (1984) was the first to apply an adapted
vector evaluated genetic algorithm (VEGA) to solve multiobjective
optimization problems. Though it was simple to implement, the method has a
bias towards some Pareto-optimal solutions (Goldberg, 1989; Srinivas and
Deb, 1995). In order to overcome this problem of bias with some of the
optimal solutions, Goldberg proposed a non-dominated sorting procedure.

29
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

The idea was implemented in different ways by Fonseca and Fleming (1993),
Horn et al. (1994) and Srinivas and Deb (1995). The algorithm implemented
by Srinivas and Deb (1995) is called the non-dominated sorting genetic
algorithm (NSGA). It is to be noted that NSGA overcomes the pitfalls of the
previous two techniques (Fonseca and Fleming, 1993, Horn et al. 1994).
Fonseca and Fleming (1998 a, b) have recently extended their algorithm,
applying the principles of niche and sharing. The concept of non-dominated
sorting is first discussed and then we turn our attention to the applications of
this technique in Chemical Engineering. It is interesting to note that most of
the recent applications in Chemical Engineering have used NSGA (adapted
as required by the problem).
NSGA uses a ranking selection method to emphasize the good points and
a niche method to create diversity in the population without losing a stable
sub-population of good points. Principally, NSGA differs from SGA only in
the way of selection. A check for non-dominance is first carried out among
all the chromosomes in the gene pool before reproduction is performed. All
the non-dominated chromosomes from the entire population are first
identified and assigned a front number (Front No. = 1). These non-dominated
chromosomes are then assigned a dummy fitness value (which is usually the
number of chromosomes, Np, but could be any other arbitrarily selected, large
value instead). The dummy fitness value of any chromosome in this front is
then modified according to a sharing procedure (Goldberg and Richardson,
1987; Deb, 1989; Deb and Goldberg, 1991) by dividing it by the niche count
of the chromosome. The niche count of a chromosome represents the number
of neighbors around it, with distant neighbors contributing less than those
nearby. The niche count, thus, gives an idea of how crowded the
chromosomes are in the x-space. This is obtained, say, for the ith
chromosome, by computing its distance, dy, from another, say, jth
chromosome in the x-space, and using a sharing function, Sh, as given below

d
Sh(djj) = l - (— —ü T , if dij<a , ;0, otherwise
$h re (10)
l CT shareJ

In Eq. 10, ashlire, a computational parameter, is the maximum distance

allowed between two chromosomes to qualify as neighbors. Obviously, if dy
is larger than a^m, its contribution to Sh is zero (the jth chromosome is then
not considered to be a neighbor of the ith), while if dy = 0, its contribution to

30
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

Sh is 1. For intermediate values of the distance between the two

chromosomes, Sh lies between 0 and 1. Thus, by summing up Sh(dy) for all
values ofj in any front comprising of non-dominated chromosomes, one can
get an idea of how crowded the ith chromosome really is. This summation is
referred to as the niche count of chromosome i. The shared fitness value of
chromosome, i, is the ratio of the common dummy fitness value assigned
initially to each member of the front, and its niche count, Use of the shared
fitness value for reproduction, thus, helps spread out the chromosomes in the
front since crowded chromosomes are assigned lower fitness values. This
procedure is repeated for all the members of the first front. Once this is done,
these chromosomes are temporarily removed from consideration, and all the
remaining ones are tested for non-dominance. The non-dominated
chromosomes in this round are classified into the next front (Front No. = 2).
These are all assigned a dummy fitness value that is a bit lower than the
lowest shared fitness value of the previous front. Sharing is performed
thereafter. This procedure is continued till all the chromosomes in the gene
pool are assigned shared fitness values. The usual operations of reproduction,
cross-over and mutation are now performed. It is clear that the
non-dominated members of the first front that have fewer neighbors, will get
the highest representation in the mating pool. Members of later fronts, which
are dominated, will get lower representations (they are still assigned some
low fitness values, rather than 'killed', in order to maintain the diversity of
the gene pool). Sharing forces the chromosomes to be spread out in the
x-space. The population is found to converge very rapidly to the Pareto set. It
is to be noted that any number of objectives (both minimization and
maximization problems) can be solved using this procedure. A flowchart
describing this technique is presented in Figure 3 (Mitra et al., 1998; Garg
and Gupta, 1999). Additional details about the algorithm, its efficiency over
other techniques and some comments on the choice of the computational
parameters to be used in NSGA, are described in Srinivas and Deb, 1995 and
Deb, 1989.
NSGA has been used to solve a variety of multiobjective optimization
problems in Chemical Engineering in recent years, as for example, in the
areas of polymer reaction engineering, catalytic reactors, membrane modules,
cyclone separators and venturi scrubbers. Various adaptations of the basic
NSGA were made in order to be able to obtain meaningful solutions, and
these are described below along with the example problems studied. It is
hoped that such a discussion would enable a reader to be able to devise his

31
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

identify nondominated individuals

reproduction according to assign dummy fitness

dummy fitness values

sharing in current front

front =front + 1

Fig. 3: A flowchart describing NSGA (Mitra et al., 1998; Garg and Gupta,
1999).

own adaptations for the problem of interest.

The first application of NSGA in Chemical Engineering was for an
industrial nylon 6 semi-batch reactor by Mitra et al. (1998). The industrial
reactor model was validated (Wajge et al., 1994b) before it was used for
multiobjective optimization. The two objective functions used were to
minimize (i) the total reaction time, tf, and (ii) the concentration, [C2]f, of the
cyclic dimer (an undesirable by-product) in the product. Equality constraints
were imposed on the monomer conversion, xmmf, in the product stream, as well
as on the number average chain length, μη,(·, of the product, so that design

32
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

values, xm.d and μ^, are attained for these. The first constraint was taken care
of by adding it in the form of a penalty function (Deb, 1995) to both the
objective functions. The objective functions were, thus, modified to

= t f +w 1 tl-(x m yx II1 , d )] 2 (a)

Min I2 = [C2]f + w,[l - (xmy xmjf (b) (1 1)

The second equality constraint, μη?Γ = μη><ι, was used as a stopping condition,
i.e., integration of the model equations, a set of ordinary differential
equations (ODEs), was stopped when this constraint was met. The decision
variables used in this study were (i) the rate of release, V-Kt), of the vapor
from the semi-batch reactor (a function of time, t) which influenced the
pressure in the reactor, and (ii) the jacket fluid temperature, Tj (a scalar).
NSGA had to be adapted to apply to decision variables that were continuous
functions of time. This was achieved by discretizing the continuous function
into several, equi-spaced (in time) scalar values, VTii; i = 1, 2, ..., Q, and
constraining the value, VTi, to lie within a small range of the previous value,
VTin. This ensured that the final continuous function, Vj(t) [obtained by
curve-fitting the digitized values] was implementable. Pareto-optimal
solutions were obtained for a specified value of the feed water concentration
(see Figure 4). These results were compared with the solutions obtained
earlier by Sareen and Gupta, 1995, who parameterized the pressure history,
and so introduced some artificial constraints into the problem (since the
shape of the pressure history was assumed to be fixed). Mitra et al. (1998)
found that the solutions obtained by NSGA were superior. Interestingly,
considerable improvement in the operation of the industrial reactor was
indicated by this study, and we understand that these results were
implemented on the industrial reactor.
Gupta and Gupta (1999) extended this work on the industrial nylon-6
reactor system to consider the multiobjective optimization of the
reactor-cum-control valve assembly. They considered the fractional-opening
of the control valve as one of the decision variables (again, a function of
time), instead of the rate of release of vapor from the reactor. The second
decision variable was the temperature of the jacket fluid, a scalar value. The
Pareto optimal solutions obtained for this system were found to be worse
when compared to the solutions obtained by Mitra et al., who had studied the
reactor alone (see Figure 4). This was because the operation of the control
valve, which released the vapors, excluded certain sets of values of V-r(t),

33
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. I, 2000 Applications ofMultiobjective Optimization
In Chemical Engineering

0.55

0.45

n
0.35

3.45
iNO.25
U

•—•0.15
CM
Ο
'3.45
0.05
0.40 0.50 0.60 0.70 0.80 0.90
VW
Fig. 4: Comparison of the Pareto solutions for an industrial semi-batch
nylon 6 reactor for two different choices of the decision variables
(Mitra et al., 1998; Gupta and Gupta, 1999). Subscript, ref, indicates
values being used in industry before changeover to near-optimal
conditions.

which were permitted in the study of Mitra et al. (1998). It is clear that for
industrial systems, the optimization of the entire system is more valuable than
that of its major parts (Aatmeeyata and Gupta, 1998).
Garg and Gupta, 1999, applied NSGA to the multiobjective optimization
of free radical bulk polymerization reactors, wherein diffusional effects (the
Trommsdorff, cage and glass effects) are manifested. The two objective
functions used were the minimization of (i) the total reaction time, tf, and (ii)
the polydispersity index, Qf, of the product. The manufacture of polymethyl
methacrylate in a batch reactor was chosen as the example system. Equality
constraints were used on the value of the number average chain length, μΠιη
and the monomer conversion, xm>r, in the final product, similar to that for the

34
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

nylon 6 reactor. Optimal temperature histories, T(t), were generated for a

given initiator concentration in the feed. Interestingly, a unique optimal
solution, instead of a Pareto set of several optimal solutions, was obtained.
This was so for all the cases studied. This inference was of considerable
importance since a controversy had existed on this point for several years,
based on earlier optimization studies that used a scalar objective function
comprising of a weighted sum of the two objectives.
Rajesh et al. (2000) carried out the multiobjective optimization of a
side-fired steam reformer using NSGA. This is a very important industrial
process and there is an inherent trade-off between the objectives. Two
objective functions were considered: (i) minimization of the methane feed
rate (input to the reformer), FCn4.im for a specified rate of production of
hydrogen, FH2ilu,i(, from the industrial unit, and (ii) maximization of the rate of
production of export steam (which was equivalent to maximization of the
flow rate, FCO-OU|, of CO in the syngas). The optimization problem was solved
using a pre-validated model. An equality constraint was used on the rate of
production of hydrogen (equated to a desired value), and an inequality
constraint was imposed on the maximum wall temperature of the reformer
tubes. The decision variables used were: the temperature of the gas mixture at
the reformer inlet, pressure at the inlet of the reformer, steam to carbon (in
the form of CH4) ratio in the feed, hydrogen (recycled H2) to carbon (as CH4)
ratio in the feed, the total molar flow rate of the feed, and the temperature of
the furnace gas. Pareto optimal solutions for specified rates of production of
hydrogen were obtained (see Figure 5). Comparison of the current operating
conditions of the industrial reformer studied (point Q in Figure 5, for FH2,unii
= 3450 kg/hr) with an arbitrary point, P, on the Pareto set (for the same
Fn2,unii) indicated scope for considerable improvement. In this problem, it was
found that feasible values of one of the decision variables, the steam to
carbon ratio in the feed, were within the bounds specified α-priori, and were,
in fact, influenced by the values selected for the other decision variables. The
code for NSGA had, therefore, to be adapted for this problem also. Values of
all the decision variables except the steam to carbon ratio were first generated
for every chromosome, following the normal procedure of mapping with the
α-priori bounds. The bounds for the steam to carbon ratio were then
generated for each chromosome. The substring (of binary digits generated
randomly) corresponding to this last decision variable was then mapped into
a real number, using these chromosome-specific bounds for each
chromosome. Similar procedures can be used whenever the bounds on

35
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

26.0

- - FH2. WM
i-- i>«. «m-3450 Igftr

25.5
m
Ό
16.0

12.0

8.0
10 11 12 13 14 15

w~3\,
Fig. 5: Pareto sets for three different values of FH2,unit- Ii (minimized) =
Few.™, t (minimized) = 108/FCOiOUt. Point Q represents the current
operating conditions of the industrial reformer studied, with FH2,unit=
3450 kg/hr. P is an arbitrarily selected point on the Pareto for the
same value of FH2,unit, which is better than Q in terms of both the
objective functions (Rajesh et al., 2000).

decision variables are chromosome-specific, and need to be adapted

continuously during the computation.
Polyethylene terephthalate (PET) is another commercially important
polymer, mostly used in the manufacture of synthetic fibers. Bhaskar et al.
(2000a) formulated a multiobjective optimization problem on the industrial,
third stage wiped film reactor used to produce this polymer. The objective
functions used in this problem were to minimize the concentrations of two
undesirable side products, namely, that of the (i) acid end groups, and (ii) the
vinyl end groups, in the output stream. An equality constraint was imposed
on the degree of polymerization, DPout, of the product, so as to produce PET
having a desired value, DPd (i.e., DP^t = DPd). The acid end group
concentration in the product was further restricted to lie below a certain value
(an inequality constraint), while the concentration of diethylene glycol end
group in the product was restricted to lie in a specified range (two inequality

36
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
K Bhaskar et al. Reviews in Chemical Engineering

constraints). This was consistent with current industrial requirements. The

three inequality constraints were taken care of by 'penalty-killing' of the
chromosomes that violated these constraints, by adding on a heavy penalty
value (104) to both the objective functions [similar to those in Eq. 11] when
these constraints were violated. The solution of the problem was found to be
a unique point, and no Pareto set of optimal solutions was obtained. It must
be emphasized that it is almost impossible to predict on an α-priori basis
whether a Pareto set would be obtained or not in any complex multiobjective
optimization problem. In fact, our experiences with the simulation of this
reactor (Bhaskar et al., 2000b) had suggested opposing effects and trends,
and we were almost sure that a Pareto set would be obtained! The unique
optimal solution was found to be superior to the current operating conditions
in the industrial reactor studied. Solutions were generated for a few other
choices of the objective functions, as for example, the minimization of the
residence time and of the acid end group concentration in the product (which
minimizes the concentration of the vinyl end group simultaneously), but
Paretos were still not obtained.
The technique of instantaneously killing the chromosomes violating the
inequality constraints, referred to as penalty killing, could lead to some
amount of numerical scatter in the final results. Such scatter is less likely to
be present in solutions obtained using the slow killing of such chromosomes
using the more accepted penalty function method (Eq. 11). This is
particularly true for complex problems where some amount of multiplicity
and/or insensitivity (of optimal solutions) is present. This was encountered in
the PET reactor optimization problem, and the unique solutions obtained for
different values of DPd showed some scatter. The optimal results obtained in
such problems can be smoothened out (Bhaskar et al., 2000a; Sareen and
Gupta, 1995) for use in industry by solving an optimization problem having a
smaller degree of freedom (e.g., one with fewer decision variables). Indeed,
Goldberg (1989) has mentioned that G A (and so its various adaptations)
could have premature convergence when used on complex problems, but that
all the results still lie in the optimal 'region'.
Ravi et al. (2000a) performed multiobjective optimization of a train of
identical cyclone separators operating in parallel, to process a specified flow
rate of dust-laden gases. They considered two objective functions: (i)
maximization of the overall collection efficiency, η0, and (ii) minimization of
the pressure drop, Δρ (related to the operating cost). The decision variables
used for this problem are the number of cyclones, the diameter of the

37
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

cyclones, and seven geometric ratios (shape) of the cyclones. The bounds on
two of the geometric ratios were adapted (depending on the values selected
for some of the other decision variables, chromosome-specific bounds) to
ensure that the cyclone shape was physically meaningful. The technique used
was similar to what was used in the optimization of the steam reformer. An
upper limit was imposed on the inlet velocity of gases (an inequality
constraint). This constraint was implemented using penalty killing, as was
done for the PET wiped film reactor optimization problem. Δρ vs. η0 Pareto
sets were obtained (see Figure 6). It was found that the optimal values of the
decision variables could not, at first sight, be explained physically. This
problem is quite common in multiobjective optimization problems associated
with a considerable amount of 'freedom'. One then needs to solve several

ι ι ι
6500 -

5500
/
4500 ί9
03 i
^3500 ί

2500 f

1500 S
_
500
I I I
0.6 0.7 0.8 0.9 1.0

Fig. 6: Pareto set (Δρ vs. η0) for a train of identical cyclone separators in
parallel (Ravi el al., 2000a). 165mVs of dust-laden gases from a
paper mill are processed.

38
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

simpler multiobjective optimization problems, each with only one, or, at most
two decision variables, to find out which of the several decision variables in
the original problem are the most sensitive and important in deciding the
Pareto set. Such a study leads to considerable physical insight and, at times,
even helps in developing appropriate bounds of the decision variables, as
well as in selecting appropriate constraints to be used in solving the
multiobjective optimization problem, which may not be known too precisely
at the beginning. Such detailed studies also help in explaining the scatter that
may be present in the final results - both in the Pareto set obtained as well as
in the plots of the individual decision variables over the range of objective
functions covered by the Pareto set. A similar study is being carried out (Ravi
et al., 2000b) on the multiobjective optimization of venturi scrubbers. Pareto
optimal solutions are being obtained in the preliminary studies completed till
now.
Multiobjective optimization of hollow fiber membrane modules was
carried out by Chan et al. (2000), using the dialysis of beer to produce
low-alcohol beer, as an example. The two objective functions considered
were (i) the maximization of the removal of alcohol from the beer, and (ii)
the minimization of the removal of 'taste chemicals (called extract)' by the
module. Inequality constraints were imposed on the pressure drops on both
the shell and the tube sides, as well as on the inner diameter of the shell. The
decision variables considered were the flow rate of water on the shell side,
the length of the fiber, the fractional free area in the shell (related to the tube
pitch), the internal radius of the individual hollow fibers and the thickness of
the hollow fiber membranes. Again, this problem is associated with an
immense amount of freedom, and several simpler one- and two- decision
variable problems have been studied to develop physical insight. It was found
that the internal radius of the membrane was the most important decision
variable influencing the Pareto set.
Another application of considerable industrial importance is the
optimization of the continuous casting of polymethyl methacrylate (PMMA)
films. In this process, a prepolymer is first produced in an isothermal plug
flow tubular reactor (PFTR). The product from this reactor flows as a thin
film through a furnace. The temperature, Tw(z), of the upper and lower
surfaces of the polymer film varies with the axial location, z, in the furnace.
The two objective functions (Zhou et al., 2000) are (i) the maximization of
the cross section-average value of the monomer conversion at the end of the
furnace, xm.av.f, and (ii) the minimization of the length, Zf, of the furnace. The

39
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications ofMultiobjective Optimization
In Chemical Engineering

end-point constraint used was that the section-average value of the number
average chain length in the product, μη,»ν,Γ, should be equal to a desired value,
μη>(1. In addition, a 'local' constraint is to be satisfied in this problem. This
takes the form that the temperature at any point in the film must be below an
upper safe value, to prevent degradation (discoloration) of the polymer film.
The decision variables used were the temperature of the isothermal PFTR,
concentration of the initiator in the feed to the PFTR, monomer conversion at
the end of PFTR, film thickness (all scalars), and the temperature
programming, Tw(z), in the furnace (a continuous function). In order to
simplify the problem, the temperature of the surface of the film, Tw(z), was
assumed to be a cubic function of z. Thus, the continuous function was
replaced by four constants. The local constraint was taken care of by using
the penalty killing procedure. In this study, the requirement, μη,,ν,Γ ~ un,d> was
taken care of by using it both as a penalty function in the two objective
functions, and also by using it as a stopping condition. This leads to faster
convergence to the Pareto solution.

2.8 Other Techniques

Several other workers have used algorithms developed by them or
adapted for their particular work to solve multiobjective optimization
problems, which cannot be grouped into the above classifications. These are
described here. Nishitani and Kunugita (1979) developed a code to determine
the optimal flow pattern for multiple effect evaporator systems using a
multiobjective approach. A modified parametric method is used in this work.
All the objective functions are first represented in terms of a common scale.
The problem is split into four simpler problems and solved. Pareto sets are
generated between the energy consumption (steam consumption) and an
economic objective (heat transfer area). The relationship between the
solutions to the four simpler problems and the overall solution to the problem
are discussed. Results are obtained for a process for concentrating milk. The
conflict between the objectives is not quantitatively understood by this
technique, though it is easy to implement for simple systems.
A process design problem was solved by Umeda et al. (1980) using the
simplex method (Beveridge and Schechter, 1970; Edgar and Himmelblau,
1988), a pattern search technique, interactively. The pattern search algorithm
helps in adjusting the weights of the objectives to find the preferred solution.
The proposed method starts with a set of Pareto optimal solutions obtained

40
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

using a combination of the parametric method and goal programming (the

goal being the Utopia point). Then, a more preferred Pareto optimal solution
is generated than the existing non-inferior set using the simplex method.
Sorting of the members of the subset is done by minimum pair-wise
comparison in every iteration. Merging of the new solution with the ordered
sub-set is done by the minimum comparison merging technique. Optimal
design of a toluene-steam dealkylation process is performed to illustrate the
usefulness of the procedure. The objectives considered were (i) the
investment cost, (ii) the annual operating cost, and (iii) the reliability of the
process. The major advantage of this procedure is that the decision maker
(DM) has to put the current set of solutions into an order. The method
requires the Utopia to be determined before the multiobjective optimization
problem can be solved.
Takama and Umeda (1980) extended this approach for locating the
non-inferior points by decomposing the original (upper-level or interior)
problem into simpler (lower-level) sub-problems involving single objective
functions. The algorithm generates alternative solutions by solving the
sub-problems at lower levels, and performs a search to identify the best
compromise solutions from among the alternatives. The DM adjusts the
coupling variables at the upper-level and passes the values to each of the
lower-level optimization problems, that are then solved. The process is
repeated until the DM is satisfied with the solution. The choice of the
coupling variables could pose problems because there is no guarantee that all
the sub- problems will remain feasible for any set of values of the coupling
variables. In addition, it is not necessary that all engineering problems can be
decomposed. The authors have applied this algorithm to the optimal design of
a water treatment system.
Lenhoff and Morari (1982) proposed a design approach which considers
the economic as well as the dynamic aspects simultaneously in the design of
resilient processing plants. They used a bounding technique based on the
Lagrangian theory, taking advantage of the typical modular structure of
processing units. They defined two performance indices: an economic
performance index (EPI) and a dynamic performance index (DPI), to
characterize a problem. Both of these were minimized and a Pareto set (called
the performance index diagram) between the EPI and the DPI was obtained
for a sample problem involving three configurations of a heat-integrated
distillation column for the methanol-water system. Certain configurations
could be eliminated since these were inferior solutions. The effect of

41
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

changing the steady state design on the performance indices was studied
using the bounding technique. A trade-off was found to exist between them,
and three non-inferior designs were determined.
Clark and Westerberg (1983) reviewed the mathematical tools used for
multi-criteria decision making while addressing problems in design
optimization where an Outer' optimization problem is constrained by an
'inner' one (embedded optimization). They proposed a single level solution
to locate the local optimum quickly. Two approaches to solve this problem
were suggested. The first was based on the active set strategy for solving the
inner problem, while the second involved the relaxation of complimentary
conditions. The concept of embedded optimization was illustrated on the
calculation of the equilibrium composition for multiphase, reacting systems.
An interactive CAD tool, CONSOLE, was used by Butala et al. (1988)
for the multiobjective optimization of a semi-batch reactor for the
manufacture of styrene-acrylonitrile (SAN) copolymer. Dynamic
optimization was used to obtain the two decision variables, the temperature
and the monomer feed rate histories, to produce copolymer having specified
values of the average copolymer composition and the average molecular
weight. A third-order polynomial was assumed for the two decision variables
and their coefficients were obtained using the feasible direction algorithm of
Nye and Tits (1986). The problem specifications (both functions and
numbers) were classified into objectives, as well as hard and soft constraints.
The initial solution had to satisfy the hard constraints first. The soft
constraints and the objectives compete simultaneously in the second phase of
optimization. In the final phase, the objectives were improved till the DM is
satisfied. This process was achieved by scaling of the specification space
(GOOD/BAD) based on the DM's knowledge of the process. It is to be
mentioned here that this process is highly intuitive and the DM should have a
very good knowledge about the problem, before he starts to solve it. Choi and
Butala (1991) extended this work to develop open loop optimal control for a
semi-batch methyl methacrylate-vinyl acetate (MMA-VA) co-polymerization
reactor. The feed rates of the monomers and the temperature of the reactor
(both functions of time) were used as decision variables. Copolymer having
specified values of the average composition and molecular weight was to be
produced. Hie optimal control policies were tested on an experimental stirred
tank polymerization reactor. These workers experienced some plant-model
mismatch. The need for improving the optimization method to minimize the
plant-model mismatch was indicated.

42
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

Ciric and Huchette (1993) tried to minimize the waste treatment costs and
maximize the profits of a chemical process, by employing the basic
relationship between the sensitivity problem and a multiobjective
optimization problem. The proposed approach uses a modified form of the
outer approximation method to identify discretely different regions of the
Pareto set. The complete solution set is generated using a sequential
approximation algorithm (Ciric and Jia, 1992). The proposed algorithm was
applied to the production of ethylene glycol from ethylene oxide and water.
This approach is better than the conventional sensitivity analysis techniques
such as linear sensitivity analysis and exhaustive search.
Liu et al. (1997) applied the concepts of multiobjective optimization to
the multimodel-based minimum bias control of a benchmark paper machine
process. They used modified goal programming for solving the problem.
Köllner et al. (1989 a, b) determined the optimal process flow structure
for the chlorination of benzene. They used the modified simplex method
(Neider and Mead; Box complex method; Beveridge and Schechter, 1970;
Edgar and Himmelblau, 1988) incorporating the inequality constraints on the
decision variables, with a stochastic global-search method (Schammler,
1988). They obtained compromise solutions between the total investment
cost and variable costs per annum for the optimal process flow structure.
They could eliminate four heat exchangers in the process using their study
involving multiobjective optimization.
Dimkou and Papalexandri (1998) proposed a decomposition-based
algorithm for solving convex MINLP (mixed integer non-linear
programming) problems and identifying the Pareto sets. Parametric nonlinear
optimization is used to identify the upper bounds of the optimal parametric
solution. Lagrangian information from the deterministic non-linear
programming (NLP) problems is used to determine the lower bounds to the
solution, as is done in Benders' decomposition method (Geoffrion, 1972).
The algorithm is illustrated using two simple examples. The assumption of
convexity limits the applicability of the proposed algorithm for real life
problems.
Tijsen et al. (1999) generated Pareto sets for the factors affecting the
degree of substitution of the hydroxyl groups in starch by ether groups, using
experimental data. They optimized the process conditions for the
modification of starch. They found the important factors by a backward
elimination strategy (Kleinbaum, 1994).
Methods used for generating non-inferior solutions and their applications

43
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
 Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

in Chemical Engineering have been discussed. This forms the objective phase
of the decision-making process. A few methods are available to select the
best (or preferred) solution from among the Pareto optimal points, using
additional, and often non-quantifiable, information. This forms the second,
subjective phase. In addition to the technique presented earlier, involving
consultation with several DMs to obtain their rankings of the Pareto
solutions, the surrogate worth trade-off method (Haimes and Hall, 1974)
seems to be popular in Chemical Engineering (Nishitani et al., 1980; Wajge
and Gupta, 1994a; Sareen and Gupta, 1995). This method uses the
Lagrangian multipliers obtained while generating the Pareto sets, to analyze
the trade-offs between the non-commensurate objectives. The preferred
solution is usually the one at which the improvement in one of the objective
functions is equivalent to the degradation that results in the other objectives.

3. FUTURE DIRECTIONS

Several approaches have been used to obtain solutions of multiobjective

optimization problems in Chemical Engineering. It appears that the
evolutionary algorithms (e.g., GA, NSGA) used in recent years are quite
robust for generating non-inferior solutions for large-scale complex
problems, and will, we expect, become even more popular in the future.
These offer several advantages as described below (Deb, 1999):

• Earlier methods have to be applied several times over to (hopefully)

obtain a Pareto-optimal solution.
• Some of the methods are sensitive to the shape of the Pareto-optimal
front.
• The efficiency of the scalar objective optimization method determines
the spread of the Pareto-optimal solutions.
• Uncertainties and stochasticities cannot be handled efficiently using
classical methods.
• Problems with a discrete search space cannot be handled efficiently
using classical single objective optimization methods (Deb, 1995).

Though NSGA has been tested on several, reasonably large optimization

problems in Chemical Engineering, it needs improvement in the area of
constraint-handling for still larger and more complex problems. The equality
constraints have been handled using the penalty function technique, while the

44
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

inequality constraints have used penalty killing. In case of problems

involving a very large number of constraints, the latter may not be too
efficient since there exists a possibility of losing diversity in the gene pool. A
better understanding of the values of the computational parameters used in
G A and their effects is required to increase their speed of convergence. These
have limited the use of NSGA to problems that are not as complex as
otherwise possible.
Simulated annealing (SA) is another emerging non-traditional algorithm
(Kirpatrick et al., 1983; Aarts and Korst, 1989) which has been used in
solving optimization problems in engineering in the recent years, and we
expect, will become popular as newer developments take place. SA mimics
the cooling of molten metals in its search procedure. The procedure begins
with the selection of an initial solution (called a point). A neighbouring point
is then created and compared with the current point. The algorithm of
Metropolis et al. (1953) is used to determine whether the new point is
accepted or not. This technique, thus, works with a single point at a time, and
a new point is created at each iteration according to the Boltzmann
probability distribution. The method is found to be effective in finding global
optimum solutions when a slow cooling procedure is used (Deb, 1995).
Adaptations can be made in SA to solve multiobjective optimization
problems using the concept of non-dominance, somewhat along the lines of
NSGA. We expect these adaptations to compete with NSGA in terms of
speed of convergence and robustness.
In the next several years, even more complex problems in which the
constraints are not known in a very precise manner (as discussed in this
review) will be solved. In fact, one could easily envisage a situation where a
DM looks at two Pareto sets simultaneously, a Pareto between the objective
functions, and another Pareto between the extents of constraint-violation, to
decide upon the preferred solution. Obviously, NSGA will need adaptations
to solve such problems, by classifying chromosomes into finer sub-fronts.
The only conclusion we can make is that the future holds exciting promises.

ACKNOWLEDGEMENTS
For their contributions to the literature on the multiobjective optimization
applications using NSGA, the works of K. Deb, K. Mitra, S. Garg, R.R.
Gupta, J.K. Rajesh, Dr. G.P. Rangaiah, C.Y. Chan, Aatmeeyata, R. Gopalan,
Dr. M.B. Ray and F.B. Zhou are gratefully acknowledged. We thank Dr.
M.V. Garland for helping us with the translation of the article from German
to English.
45

Brought to you by | Heinrich Heine Universität Düsseldorf

Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications ofMultiobjective Optimization
In Chemical Engineering

NOMENCLATURE

Symbol Description

[C2] Cyclic dimer concentration in nylon 6 manufacture, mol/kg

djj Dimensionless distance between the ith and jth chromosomes in
the x-space
DP Degree of polymerization (= μη)
FcH4.in Flow rate of methane in the input stream, kg/hr
Fco,out Flow rate of CO in the exit stream of the first reactor in the
reformer plant (in the syngas), kg/hr
FH2,unit Flow rate of H2 in the exit stream from the reformer plant, kg/hr
g(x) Vector of inequality constraints, g;(x)
h(x) Vector of equality constraints, h;(x)
I Vector of objective functions, I;
ki, k2 Rate constants in Eq. 2
Ng Generation number in GA
Np Number of chromosomes in the population in G A
p Pressure
Δρ Pressure drop in cyclone, Pa
Q Polydispersity index of polymer
Sh Sharing function (Eq. 10)
t Time
T Temperature
u(I) Utility function
VT Rate of release of vapor mixture from nylon 6 reactor through
control valve, mol/hr
w Vector of weightage factors, Wj
xm Monomer conversion
χ Vector of decision variables, x,
X Feasible region of χ
y Vector of goals, yj
z Axial position in furnace reactor

Subscripts / Superscripts
av Cross-section average value
d Desired or design value
f, out Final, outlet value

46
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

J Jacket-fluid value
ref Reference value (currentlyused in industrial reactor)
Τ Transpose
w Wall or surface value

Greek

ε; Constants in ε-constraint method

η0 Overall collection efficiency in cyclone
μη Number average chain length of polymer
Computational parameter in Eq. 10

REFERENCES

Aarts, E. and J. Korst. "Simulated Annealing and Boltzmann Machines: A

Stochastic Approach to Combinatorial Optimization and Neural
Computing", Wiley, Chichester, 1989.
Aatmeeyata and S. K. Gupta. "Simulation and optimization of an industrial
nylon 6 reactor: a review", Polym. Plastics Tech. Eng., 37 (1998):
201-239.
Beveridge, G.S.G. and R. S. Schechter. "Optimization: Theory and Practice",
McGraw Hill, New York, 1970.
Bhaskar, V., S. K. Gupta and A. K. Ray. "Multiobjective optimization of an
industrial wiped film PET reactor", AIChEJ. (2000a), in press.
Bhaskar, V., S. K. Gupta and A. K. Ray. "Modeling of an industrial wiped
film poly (ethylene terephthalate) reactor", Polym. Reac. Eng. (2000b), in
press.
Bryson, A.E. and Y. C. Ho. "Applied Optimal Controt\ Blaisdell, Waltham,
MA, 1969.
Butala, D., K. Y. Choi and M. K. H. Fan. "Multiobjective dynamic
optimization of a semibatch free-radical co-polymerization process with
interactive CAD tools", Comput. Chem Eng., 12 (1988): 1115-1127.
Chan, C.Y., Aatmeeyata, S. K. Gupta and A. K. Ray. "Multiobjective
optimization of membrane separation modules", J. Memb. Sei. (2000),
submitted.

47
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications ofMultiobjective Optimization
In Chemical Engineering

Chankong, V. and Y. Y. Haimes. "Multiobjective Decision Making - Theory

and Methodolog^, Eisevier, New York, 1983.
Charnes, A. and W. Cooper. "Management Models and Industrial
Applications of Linear Programming", Wiley, New York, vol. 1, 1961
Choi, K. Y. and D. N. Butala. "An experimental-study of multiobjective
dynamic optimization of a semibatch copolymerization process", Polym.
Eng. Sei., 31(1991), 353-364.
Ciric, A. R. and S. G. Huchette. "Multiobjective optimization approach to
sensitivity analysis - waste treatment costs in discrete process synthesis
and optimization problems", Ind. Eng. Chem. Res., 32 (1993): 2636-2646.
Ciric, A. R. and T. Jia." Multiobjective optimization and nonlocal sensitivity
in process source reduction problems: Continuous optimization
problems", / & EC Special Symposium on Emerging Technologies in
Hazardous Waste Management, Atlanta, G A, Sept 21-22, 1992.
Clark, P. A. and A. W. Westerberg. "Optimization for design problems
having more than one objective", Comput. Chem. Eng., 7 (1983):
259-278.
Cohon. J. L. "Multiobjective Programming and Planning", Academic, New
York, 1978.
Deb, K. "Evolutionary Algorithms for Multi-Criterion Optimization in
Engineering Design", Evolutionary Algorithms in Engineering and
Computer Science: Recent Advances in Genetic Algorithms, Evolution
Strategies, Evolutionary Programming, Genetic Programming and
Industrial Applications, K. Miettinen, P. Neittaanmäki, M. M. Mäkelä
and J. Periaux Edition, John Wiley, June 1999.
Deb, K. "Genetic algorithms in multimodel function optimization (TCGA
report no. 89002)" Tuscaloosa, AL: University of Alabama, The Clearing
house for Genetic Algorithms, 1989.
Deb, K. "Optimization for Engineering Design: Algorithms and Examples",
Prentice Hall of India, New Delhi, 1995.
Deb, K. and D. E. Goldberg. "An investigation of niches and species
formation in genetic function optimization", In J.D. Schaffer, Ed.,
Proceedings of the Third International Conference on Genetic
Algorithms, San Mateo, CA, Morgan Kaufmann, (1991): 42-50.
Dimkou, T. I. and K. P. Papalexandri. "A parametric optimization approach
for multiobjective engineering problems involving discrete decisions",
Comput. Chem. Eng., 22 (1998): S951-S954.
Ecker, J. G. and M. Kupferschmidt. "An ellipsoid algorithm for nonlinear
programming", Math. Program., 27 (1983): 83-106.
48
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

Edgar, T. F. and D. M. Himmelblau. "Optimization of Chemical Processes",

McGraw Hill, New York, 1988.
Fan, L. T., C. S. Landis and S. A. Patel, in "Frontiers in Chemical Reaction
Engineering", L. K. Doraiswamy and R. A. Mashelkar, Eds., Wiley
Eastern, New Delhi, 1984,609-623.
Farber, J. N. "Steady state multiobjective optimization of continuous
co-polymerization reactors", Polym. Eng. Sei., 26 (1986): 499-507.
Fonseca, C. M and P. J. Fleming. "Genetic algorithms for multiobjective
optimization: formulation, discussion and generalization", in S. Forrest,
Ed., Proceedings of the Fifth International Conference on Genetic
Algorithms. San Mateo, CA, Morgan Kaufmann, (1993): 416-423.
Fonseca, C. M and P. J. Fleming. "Multiobjective optimization and multiple
constraint handling with evolutionary algorithms - Part II: Application
example", IEEE Trans. Sys. Man. Cy., A., 28 (1998a): 38-47.
Fonseca, C. M. and P. J. Fleming. "Multiobjective optimization and multiple
constraint handling with evolutionary algorithms - Part I: A unified
formulation", IEEE Trans. Sys. Man. Cy., A., 28 (1998b): 26-37.
Garg, S., and S. K. Gupta. "Multiobjective optimization of a free radical bulk
polymerization reactor using genetic algorithm", Macromol. Theor.
Simul, 8 (1999): 46-53.
Geoffrion, A. M. "Strictly concave parametric programming, Part I (basic
theory", Mgmt. Sei., 13 (1967a): 244-253.
Geoffrion, A. M. "Strictly concave parametric programming, Part II
(additional theory and computational consideration", Mgmt. Sei., 13
(1967b): 359-370.
Geoffrion, A. M. "Solving bicriterion mathematical programs", Ops. Res., 15
(1967c): 39-54.
Geoffrion, A. M. "Generalized Benders decomposition", J. Opt. Theory
Applic., 10 (1972): 237-260.
Geoffrion, A. M., J. S. Dyer and A. Feinberg. "An interactive approach for
multi-criterion optimization, with an application to the operation of an
academic department", Mgmt. Sei., 19 (1972): 357-368.
Gill, P. E., W. Murray and M. H. Wright. "Practical Optimization",
Academic, New York, 1981.
Goicoechea, A., D. R. Hansen, and L. Duckstein. "Multiobjective Decision
Analysis with Engineering and Business Applications", Wiley, New York,
1982.

49
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

Goldberg, D. E. "Genetic Algorithms in Search, Optimization and Machine

Learning", Addison-Wesley, Reading, MA, 1989.
Goldberg, D. E. and J. Richardson. "Genetic algorithms with sharing for
multimodal function optimization", In J. J. Grefenstette, Ed., Proceedings
of the Second International Conference on Genetic Algorithms. Princeton,
NJ, Lawrence Erlbaum, (1987): 41-49.
Grossmann, I. E., R. Drabbant and R. K. Jain. "Incorporating toxicology in
the synthesis of industrial chemical complexes", Chem. Eng. Commun.,
17(1982): 151-170.
Gupta, R. R., and S. K. Gupta. " Multiobjective optimization of an industrial
nylon 6 semibatch reactor system using genetic algorithm", J. Appl.
Polym. Sei., 73 (1999): 729-739.
Haimes, Υ. Υ. "Hierarchical Analyses of Water Resources Systems:
Modeling and Optimization of Large-scale Systems", McGraw Hill, New
York, 1977.
Haimes, Y. Y. and W. A. Hall. "Multiobjectives in water resources systems
analysis: The surrogate worth trade-off method", Water Resources Res.,
10 (1974): 615-624.
Haimes, Y. Y., D. A. Wistner and L. S. Ladson. "On bicriterion formulation
of the integrated systems identification and system optimization", IEEE
Trans. Sys. Man. Cy., SMC-1 (1971): 296-297.
Haimes, Y. Y., K. Tarvainen, T. Shima and J. Thadathil. "Hierarchical
Multiobjective Analysis of Large-scale Systems", Hemisphere, New York,
1990.
Haimes, Y. Y., W. A. Hall and H. T. Freedman. "Multiobjective Optimization
in Water Resources Systems: the Surrogate Worth Trade-off Method',
Elsevier, Amsterdam, 1975.
Holland, J. H. "Adaptation in Natural and Artificial Systems", University of
Michigan Press, Ann Arbor, MI, 1975.
Horn. J., N. Nafpliotis and D. E. Goldberg. "A niched Pareto genetic
algorithm for multiobjective optimization", Proceedings of the First IEEE
Conference on Evolutionary Computation, (1994): 82-87.
Hwang, C. L., and A. S. M. Masud. "Multiple Objective Decision Making -
Methods and Applications", Lecture notes in economics and mathematical
systems, No. 164, Springer, Berlin, 1979.
Hwang, C. L., S. R. Paidy, K. Yoon and A. S. M. Masud. "Mathematical
programming with multiple objectives: a tutorial", Comput. & Ops. Res.,
7 (1980): 5-31.

50
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
V. Bhaskar et al. Reviews in Chemical Engineering

Kirpatrick, S., C. D. Gelatt and M. P. Vecchi. "Optimization with simulated

annealing", Science, 220 (1983): 671-680.
Kleinbaum, D. G. "Logistic Regression: a Self-learning Texf, Springer, New
York, 1994.
Köllner, H., G. Schammler and M. Müller. "Multiobjective optimization as a
help in decision-making rationalization of complex chemical-processing
systems. 2. Mathematical-modeling and optimization", Chem. Tech., 41
(1989a): 11-14.
Köllner, H., G. Schammler and M. Müller. "Multiobjective optimization as a
help in decision-making rationalization of complex chemical-processing
systems. 3. Discussion of the results", Chem. Tech., 41 (1989b): 54-57.
Kraslawski, A. and P. Pustelnik "An experimental multiobjective
optimization", Hung. J. Ind. Chem., 19 (1991): 137-141.
Kraslawski. A., E. Rzyski and J. Stelmach. "Application of fuzzy
multiobjective optimization for self-sucking impellers in a bioreactor",
Bioproc. Eng., 6 (1991): 109-116.
Lapidus, L. and R. Luus. "Optimal Control of Engineering Processes",
Blaisdell, Waltham, MA, 1967.
Lenhoff, A. M. and M. Morari. "Design of resilient processing plants—I.
Process design under consideration of dynamic aspects", Chem. Eng. Sei.,
37 (1982): 245-258.
Liu, D. Y., S. L. Shah, D. G. Fisher and X. Liu. "Multimodel-based minimum
bias control of a benchmark paper machine process", Can. J. Chem. Eng.,
75(1997): 152-160.
Luyben, M. L. and C. A. Floudas. "Analyzing the interaction of design and
control - 1. A multiobjective framework and application to binary
distillation synthesis", Comput. Chem. Eng., 18 (1994a): 933-969.
Luyben, M. L. and C. A. Floudas. "Analyzing the interaction of design and
control--2. Reactor-Separator-Recycle system", Comput. Chem. Eng., 18
(1994b): 971-994.
Meadowcraft, . ., C. Brosilow and G. Stephanopoulos. "The modular
multivariable controller: 1. Steady-state properties", AIChEJ., 38 (1992):
1254-1277.
Metropolis, N., A. Rosenbluth, M. Rosenbluth, A. Teller and E. Teller.
"Equation of state calculations by fast computing machines", J.
Chem.Phys., 21 (1953): 1087-1092.

51
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. 1, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

Mitra, K., K. Deb and S.K. Gupta. "Multiobjective dynamic optimization of

an industrial nylon 6 semibatch reactor using genetic algorithm", J. Appl.
Polym. Sei., 69 (1998): 69-87.
Murtagh, B. A. and M. A. Saunders." M1NOS5.0 User's Guide-Appendix A:
Minos5.2. Technical Report SOL 83-20R", Systems Operations
Laboratory, Department of Operations Research, Stanford University
(1988).
Nishida, N., A. Ichikawa and E. Tazki. "Synthesis of optimal process systems
with uncertainty", Ind. Eng. Ghent., Proc. Des. Dev., 13 (1974): 209-214.
Nishitani, H. and E. Kunugita. "The optimal flow pattern of multiple effect
evaporator systems", Comput. Chem. Eng., 3 (1979): 261-268.
Nishitani, H. and E. Kunugita. "Multiobjective analysis for energy and
resource conservation in an evaporator system", Paper presented at the
Second World Congress of Chemical Engineering, Montreal, Canada,
October 1981.
Nishitani, H., Y. Ogaki, E. Kunugita and L. T. Fan. "Multiobjective
optimization of an aeration vessel for waste water treatment", Chem. Eng.
Commun., 5 (1980): 135-147.
Nye, W. T. and A. L. Tits. "An application-oriented, optimization-based
methodology for interactive design of engineering system", Int. J.
Control,*! (1986): 1693-1721.
Palazoglu, A. and T. Arkun. "Design of chemical plants in the presence of
uncertainty: a multiobjective approach", Comput. Chem. Eng., 10 (1986):
567-575.
Pareto, V. "Cows d'economie Politique", F. Rouge, Lausanne, Switzerland,
1896.
Paules, G. E. and C. A. Floudas. "APROS: Algorithmic development
methodology for discrete continuous optimization problems", Ops. Res.,
37 (1989): 902-915.
Rajesh, J. K., S. K. Gupta, G. P. Rangaiah and A. K. Ray. "Multiobjective
optimization of steam reformer performance using genetic algorithm",
Ind. Eng. Chem. Res., 39 (2000): 706-717.
Ravi, G., S. K. Gupta and M. B. Ray. "Multiobjective optimization of
cyclone separators", Ind. Eng. Chem. Res., submitted (2000a).
Ravi, G., S. K. Gupta, S. Viswanathan and M. B. Ray. "Multiobjective
optimization of venturi scrubbers", Ind. Eng. Chem. Res., in preparation
(2000b).

52
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
K Bhaskar et al. Reviews in Chemical Engineering

Ray, W. H. and J. Szekely. "Process Optimization, with Applications in

Metallurgy and Chemical Engineering", Wiley, New York, 1973.
Sareen, R. and S. K. Gupta. "Multiobjective optimization of an industrial
semibatch nylon 6 reactor", J. Appl. Polym. Sei., 58 (1995): 2357- 2371.
Schaffer, J. D. "Some experiments in machine learning using vector
evaluated genetic algorithms", Doctoral Dissertation, Vanderbilt
University, Nashville, TN, 1984.
Schammler, G. "Zur Anwendung von Suchverfahren bei der Lösung von
Polyoptimierungsproblemen in der Verfahrenstechnik", Dissertation (A),
TH Köthen, Germany, 1988
Seinfeld, J. H. and W. L. MacBride. "Optimization with multiple
performance criteria", Ind. Eng. Chem., Proc. Des. Dev., 9 (1970): 53-57.
Shieh, J. H. and L. T. Fan. "Multiobjective optimal synthesis of methanation
process", Paper presented at 73rf annual meeting of AIChE, Chicago, 1L,
Nov. 1980.
Smigielski, K., J. Gora and J. Kula. "The fuzzy experimental multiobjective
optimization of the electrochemical reduction process of maleic-acid to
succinic acid", Hung. J. Ind. Chem., 20 (1992): 117-120.
Sophos, A., E. Rotstein and G. Stephanopoulos. "Multiobjective analysis in
modeling the petrochemical industry", Chem. Eng. Sei., 35 (1980):
2415-2426.
Srinivas, N. and K. Deb. "Multiobjective function optimization using
nondominated sorting genetic algorithms", Evol. Comput., 2 (1995):
221-248.
Steuer, R. E. "Multiple Criteria Optimization: Theory, Computation and
Application", Wiley, New York, 1986.
Takama, N. and T. Umeda. "Multi-level, multi-objective optimization in
process engineering", Chem. Eng. Sei., 36 (1980): 129-136.
Tijsen, C. J., H. J. Scherpenkate, E. J. Stamhuis and A. A. C. M. Beenackers.
"Optimization of the process conditions for the modification of starch",
Chem. Eng. Sei., 54 (1999): 2765-2772.
Tsoukas, ., . V. Tirrell and G. Stephanopoulos. "Multiobjective dynamic
optimization of semibatch copolymerization reactors", Chem. Eng. Sei.,
37 (1982): 1785-1795.
Umeda, T., T. Kuriyama, S. Kobayashi and A. Ichikawa. "Interactive
solution to multiple criteria problems in chemical engineering design",
Comput. Chem. Eng., 4 (1980): 157-165.

53
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM
Vol. 16, No. I, 2000 Applications of Multiobjective Optimization
In Chemical Engineering

Videla. S., R. Chamy, E. Navarrete and J. M. Lema. "Design and operation of

anaerobic digesters using multiobjective optimization criteria", J. Chem.
Tech. Biotech., 49 (1990): 223-231.
Wajge, R. M. and S. K. Gupta. "Multiobjective dynamic optimization of a
non-vaporizing nylon 6 batch reactor", Polym. Eng. Sei., 34 (1994a):
1161-1172.
Wajge, R. M., S. S. Rao and S. K. Gupta. "Simulation of an industrial
semi-batch nylon 6 reactor: Optimal parameter estimation", Polymer, 35
(1994b): 3722-3734.
Zeleny, M. "Linear Multiobjective Programming", Lecture notes in
economics and mathematical systems series, Springer, New York, 1974.
Zeleny, M. "Multiple Criteria Decision Making', McGraw Hill, New York,
1982.
Zhou, F. B., S. K. Gupta and A. K. Ray. "Multiobjective optimization of the
continuous casting process for poly (methyl methacrylate) using adapted
genetic algorithm", J. Appl. Polym. Sei., (2000): in press.
Zionts, S. and J. Wallenius. "An interactive programming method for solving
the multiple criteria problem", Mgmt. Sei., 22 (1976): 652-663.
Zionts, S. and J. Wallenius. "Identifying efficient vectors: some theory and
computational results", Ops. Res., 24 (1980): 785-793.

54
Brought to you by | Heinrich Heine Universität Düsseldorf
Authenticated | 134.99.128.41
Download Date | 9/24/13 12:39 PM