Fuzzy Sets and Systems 139 (2003) 395 – 405

www.elsevier.com/locate/fss

A goal programming procedure for fuzzy multiobjective linear

fractional programming problem
B.B. Pala;∗ , B.N. Moitraa , U. Maulikb
a
Department of Mathematics, Faculty of Science, University of Kalyani, Kalyani 741235, West Bengal, India
b
Kalyani Govt. Engineering College, Kalyani 741235, India
Received 17 August 2000; received in revised form 18 July 2002; accepted 19 July 2002

Abstract

This paper presents a goal programming (GP) procedure for fuzzy multiobjective linear fractional program-
ming (FMOLFP) problems.
In the proposed approach, which is motivated by Mohamed (Fuzzy Sets and Systems 89 (1997) 215), GP
model for achievement of the highest membership value of each of fuzzy goals de9ned for the fractional
objectives is formulated. In the solution process, the method of variable change on the under- and over-
deviational variables of the membership goals associated with the fuzzy goals of the model is introduced
to solve the problem e:ciently by using linear goal programming (LGP) methodology. The approach is
illustrated by two numerical examples.

c 2002 Elsevier B.V. All rights reserved.
Keywords: Multiobjective linear fractional programming; Fuzzy multiobjective linear fractional programming; Fuzzy
programming; Fuzzy goal programming; Goal programming; Linear goal programming

1. Introduction

During the mid-1960s and early 1970s of the last century fractional programming (FP) was studied
extensively [3,1] in the literature. In contrast to the single objective FP, multiobjective fractional
programming (MOFP) has not been discussed that extensively and only a few approaches have
appeared in the literature [4,13,14,18] concerning MOFP. It may be pointed out that in most of the
MOFP approaches, the problems are converted into single objective FP problems and then solved
employing the method of Charnes and Cooper [3] or Bitran and Novaes [1].

∗
Corresponding author.
E-mail address: [email protected] (B.B. Pal).

0165-0114/03/$ - see front matter  c 2002 Elsevier B.V. All rights reserved.
PII: S 0 1 6 5 - 0 1 1 4 ( 0 2 ) 0 0 3 7 4 - 3
396 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405

To overcome the computational di:culties of using conventional FP approaches to MOFP prob-

lems, the theory of fuzzy sets has been introduced in the 9eld of FP. Linguistic variable approach
of Zadeh [24] to FMOLFP problem has been proposed by Luhandjula [15] in 1984. Luhandjula’s
approach has been further developed by Rao et al. [6,7]. Other approaches in this area have also
been investigated [5,20]. However, works in this 9eld are still not quite satisfactory and development
of e:cient computational procedure is one of the emerging areas of investigations.
In the past few years, adaptation of existing multiobjective programming methodologies
[2,11,12,19,22] to fuzzy programming problems has been studied [9,21,25,26]. Among all the ap-
proaches developed so far, GP [12] has appeared as a robust tool for solving multiobjective fuzzy
programming problems in [9,10,17,23,25,26].
In this article, the GP approach to fuzzy programming problems introduced by Mohamed [16]
is extended to solve FMOLFP problems. In the GP model formulation of the problem, 9rst the
objectives are transformed into fuzzy goals by means of assigning an aspiration level to each of
them. Then achievement of the highest membership value (unity) to the extent possible of each of
the fuzzy goals is considered.
In the solution process, the under- and over-deviational variables of the membership goals asso-
ciated with the fuzzy goals are introduced to transform the proposed model into an equivalent LGP
model to solve the problem e:ciently in the decision situation.

2. Problem formulation

The general format of a classical multiobjective linear fractional programming problem can be
stated as

ck X +  k
Optimize Zk (X) = ; k = 1; 2; : : : ; K
dk X +  k
   
 6 
subject to X ∈ S = X ∈ Rn | AX  =  b; X ¿ 0; b ∈ Rm ; (2.1)
 
¿

where ck ; dk ∈ Rn ; k ; k are constants and S = .

It is customary to assume that dk X + k ¿0; ∀X ∈ S.
In MOFP if an imprecise aspiration level is introduced to each of the objectives then, these fuzzy
objectives are termed as fuzzy goals.
Let gk be the aspiration level assigned to the kth objective Zk (X ). Then, the fuzzy goals appear
as:

(a) Zk (X )&gk (for maximizing Zk (X ));

(b) Zk (X ).gk (for minimizing Zk (X )),

where & and . indicate the fuzziness of the aspiration levels, and is to be understood as “essentially
more than” and “essentially less than” in the sense of Zimmermann [25].
 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405 397

Hence, the fuzzy linear fractional goal programming can be stated as follows:
Find X
so as to satisfy Zk (X ) & gk ; k = 1; 2; : : : ; k1
Zk (X ) . gk ; k = k1 + 1; : : : ; K
 
6 (2.2)
subject to AX  =  b;
¿
X ¿ 0:
Now, in the 9eld of fuzzy programming, the fuzzy goals are characterized by their associated
membership functions. The membership function k for the kth fuzzy goal Zk &gk can be expressed
algebraically according to Rao et al. [23] as

 1 if Zk (X ) ¿ gk ;

 Z (X ) − l
k k
k (X ) = if lk 6 Zk (X ) 6 gk ; (2.3)

 g k − l k

0 if Zk (X ) 6 lk ;
where lk is the lower tolerance limit for the kth fuzzy goal.
On the other hand, the membership function k for the kth fuzzy goal Zk (X ).gk can be de9ned
as

 1 if Zk (X ) 6 gk ;

 u − Z (X )
k k
k (X ) = if gk 6 Zk (X ) 6 uk ; (2.4)

 uk − g k

0 if Zk (X ) ¿ uk ;
where uk is the upper tolerance limit.
Now, in a fuzzy decision environment, the achievement of the objective goals to their aspired
levels to the extent possible is actually represented by the possible achievement of their respective
membership values to the highest degree. Regarding this aspect of fuzzy programming problems, a
GP approach seems to be most appropriate for the problem considered in this paper.

2.1. Goal programming formulation

In fuzzy programming approaches, the highest degree of membership function is 1. So, as in

Mohamed [16], for the de9ned membership functions in (2.3) and (2.4), the Oexible membership
goals with the aspired level 1 can be presented as:
Zk (X ) − lk
+ d− +
k − dk = 1; (2.5)
gk − l k
uk − Zk (X )
+ d− +
k − dk = 1; (2.6)
uk − g k
398 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405

where dk− (¿0) and dk+ (¿0) with dk− dk+ = 0 represent the under- and over-deviations, respectively,
from the aspired levels.
In conventional GP, the under- and=or over-deviational variables are included in the achieve-
ment function for minimizing them and that depend upon the type of the objective functions to be
optimized.
In this approach, only the under-deviational variable dk− is required to be minimized to achieve
the aspired levels of the fuzzy goals. It may be noted that any over-deviation from a fuzzy goal
indicates the full achievement of the membership value (see [8]).
Now it can be easily realized that the membership goals in (2.5) and (2.6) are inherently non-
linear in nature and this may create computational di:culties in the solution process. To avoid such
problems, a linearization procedure is presented in the following section.

2.1.1. Linearization of membership goals

The kth membership goal in (2.5) can be presented as
1
Lk Zk (X ) − Lk lk + d− +
k − dk = 1; where Lk =
gk − l k
Introducing the expression of Zk (X ) from (2.1), the above goal can be presented as
Lk (ck X + k ) + d− + 
k (dk X + k ) − dk (dk X + k ) = Lk (dk X + k );

where Lk = 1 + Lk lk , or
Ck X + d− +
k (dk X + k ) − dk (dk X + k ) = Gk ; (2.7)
where Ck = Lk ck − Lk dk ; Gk = Lk k − Lk k .
Similar goal expressions for the membership goal in (2.6) can also be obtained.
However, for model simpli9cation, the expression in (2.7) can be considered as a general form
of goal expression for any type of the stated membership goals.
Now, using the method of variable change as presented by Kornbluth and Steuer [14], the goal
expression in (2.7) can be linearized as follows:
Letting Dk− = dk− (dk X + k ) and Dk+ = dk+ (dk X + k ), the linear form of the expression in (2.7)
is obtained as
Ck X + Dk− − Dk+ = Gk (2.8)
with Dk− ; Dk+ ¿0 and Dk− Dk+ = 0 since dk− ; dk+ ¿0 and dk X + k ¿0.
Now, in making decision, minimization of dk− means minimization of Dk− =(dk X + k ), which is
also a non-linear one.
It may be noted that when a membership goal is fully achieved, dk− = 0 and when its achievement
is zero, dk− = 1 are found in the solution.
So, involvement of dk− 61 in the solution leads to impose the following constraint to the model
of the problem:
Dk−
6 1;
dk X + k
 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405 399

i.e.,

−dk X + Dk− 6 k :

Here, on the basis of the previous discussion, it may be pointed out that any such constraint
corresponding to dk+ does not arise in the model formulation.
Now, if the most widely used and simplest version of GP (i.e., minsum GP) [12] is introduced to
formulate the model of the problem under consideration, then the GP model formulation becomes:

Find X so as to
K
Minimize Z= wk− Dk−
k=1
and satisfy Ck X + Dk− − Dk+ = Gk
 
6 (2.9)
subject to AX =  b

¿
and −dk X + Dk− 6 k ;
X ¿ 0;
Dk− ; Dk+ ¿ 0; k = 1; 2; : : : ; K;

where Z represents the fuzzy achievement function consisting of the weighted under-deviational
variables, where the numerical weights wk− (¿0); k = 1; 2; : : : ; K represent the relative importance
of achieving the aspired levels of the respective fuzzy goals subject to the constraints set in the
decision situation. To assess the relative importance of the fuzzy goals properly, the weighting
scheme suggested by Mohamed [16] can be used to assign the values to wk− (k = 1; 2; : : : ; K). In the
present formulation, wk− is determined as


 1
 for the de9ned k in (2:3);
gk − l k
wk− = (2.10)

 1
 for the de9ned k in (2:4):
uk − g k
The minsum GP method [12] can then be used to solve the problem in (2.9).

Remark. With regard to the use of the above weighting scheme to the goals of the problem in
(2.9), it can easily be shown that the theorem given by Mohamed [16] on the equivalence of fuzzy
multiobjective linear programming and weighted LGP is also applicable to FMOLFP and non-linear
GP.

The situation is discussed as follows:

The 9nal goal expression in the model is

Ck X + Dk− − Dk+ = Gk ;
400 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405

which is actually the linearized form of the original goal expression

Lk (ck X + k ) + d− + 
k (dk X + k ) − dk (dk X + k ) = Lk (dk X + k ):

This expression can be rewritten as

ck X +  k
Lk + d− + 
k − d k = Lk ;
dk X +  k
where Lk = 1=(gk − lk ), and Lk = 1 + Lk lk and where (gk − lk ) represents the admissible violation
constant.
The generalized fractional form of it is

Zk (X ) + (gk − lk )d− +
k − (gk − lk )dk = gk :

Now, the conventional form of a goal in GP is:

Zk (X ) + nk − pk = bk ;

where Zk (X ) may be linear=non-linear including fractional and where nk and pk represents the under-
and over-deviational variables, respectively.
Comparing the above two forms of goals it is found that

nk = (gk − lk )d−
k ;

that is, dk− = nk =(gk − lk ).

Now, from the point of view of minimizing the weighted goal deviational variables in GP, we
see that the GP formulations for the two forms of goals are equivalent with dk− = nk =(gk − lk ).
Therefore, the weights of the goals in the proposed model are reciprocal to the admissible violation
constants.

3. Numerical examples

Example 1. The following numerical example studied by Luhandjula [15] is considered to illustrate
the above approach:
x1 − 4
Maximize Z1 =
−x2 + 3
−x1 + 4
and Maximize Z2 =
x2 + 1 (3.1)
subject to −x1 + 3x2 6 0;
x1 6 6;
x1 ; x2 ¿ 0:
Let the fuzzy aspiration levels of the two objectives be (2; 4), respectively. Then the problem can
be designed as
 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405 401

Find X (x1 ; x2 ) so as to satisfy the following fuzzy goals:

x1 − 4
Z1 = & 2;
−x2 + 3
−x1 + 4
Z2 = &4
x2 + 1
subject to the given set of constraints in (3.1).
Now, let the tolerance limits of the two fuzzy objective goals be (−1; −2), respectively.
The membership functions of the goals are obtained as

(x1 − 4)=(−x2 + 3) + 1
1 = (3.2)
3
and
(−x1 + 4)=(x2 + 1) + 2
2 = : (3.3)
6
Then the membership goals can be expressed as

(x1 − 4)=(−x2 + 3) + 1
+ d− +
1 − d1 = 1; (3.4)
3
(−x1 + 4)=(x2 + 1) + 2
+ d− +
2 − d2 = 1; (3.5)
6
where di− ; di+ ¿0 with di− di+ = 0; i = 1; 2.
Following the procedure, the membership goals are restated as:

x1 + 2x2 + D1− − D1+ = 10 (3.6)

and

− x1 − 4x2 + D2− − D2+ = 0; (3.7)

where

D1− = 3d−
1 (−x2 + 3); D1+ = 3d+
1 (−x2 + 3);

D2− = 6d−
2 (x2 + 1); D2+ = 6d+
2 (x2 + 1):

Now, the restrictions d− −

1 61 and d2 61 give

3x2 + D1− 6 9; (3.8)

− 6x2 + D2− 6 6: (3.9)

402 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405

Thus the equivalent GP formulation is obtained as

Find X (x1 ; x2 ) so as to

Minimize ( 13 D1− + 16 D2− )

and satisfy x1 + 2x2 + D1− − D1+ = 10

−x1 − 4x2 + D2− − D2+ = 0

subject to 3x2 + D1− 6 9;

−6x2 + D2− 6 6;

−x1 + 3x2 6 0;

x1 6 6;

x1 ; x2 ¿ 0;

Di− ; Di+ ¿ 0; i = 1; 2:
The problem is solved by using minsum GP method and the optimal solution obtained is

x1 = 6; x2 = 2;

Z1 = 2; Z2 = − 23
and the membership values achieved are
1 = 1; 2 = 0:222:

Example 2. Suppose under the same set of constraints of the previous example, the decision-maker
(DM) desires to
x1 − 4
Maximize Z1 =
−x2 + 3
−x1 + 4
and Minimize Z2 = :
−x2 + 1
The aspiration levels of the fuzzy objectives are given as (2; 1), respectively.
Then the fuzzy goals of the problem become
x1 − 4
satisfy &2
−x2 + 3
−x1 + 4
and satisfy . 1:
x2 + 1
Now let the tolerance limits of the goals be (0; 2), respectively.
 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405 403

The resultant minsum GP formulation becomes

Find X (x1 ; x2 ) so as to
Minimize (0:5D1− + D2− )
and satisfy x1 + 2x2 + D1− − D1+ = 10
x1 + x2 + D2− − D2+ = 3
subject to 2x2 + D1− 6 6;
−x2 + D2− 6 1;
x1 + 3x2 6 0;
x1 6 6;
x1 ; x2 ¿ 0
Di− ; Di+ ¿ 0; i = 1; 2:

The optimal solution to the problem is

x1 = 6; x2 = 2:

The achieved objective function values are

2
Z1 = 2; Z2 = − :
3
The resulting membership values are

1 = 1; 2 = 1 with the over-deviation value d+ 5

2 = 3:

The result shows that the optimal solution of the previous problem remains same. But it may be
noted that the second objective goal is overly satis9ed here, i.e., its value is now satisfactorily lower
than the given upper limit. The happening of this situation is actually exhibited by the over-deviation
value of the associated membership goal.

4. Conclusion

In this paper a linearization process of solving FMOLFP problems via minsum GP is investigated.
In the decision environment, if the right hand side values of the system constraints are also fuzzy,
then that can also be accommodated in the framework of the GP model presented here.
Regarding the task of the DM to assess the relative importance of the objective goals, the primary
diRerence between the conventional (crisp) GP and the fuzzy approach presented in this paper is
that in the former case the values to wk− (k = 1; 2; : : : ; K) are given on the basis of the DMS needs
404 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405

and desires in the decision context, whereas in the later case that solely depends on the permissible
tolerance values of violation of the aspired levels of the fuzzy objective goals.

Acknowledgements

The authors are thankful to the Editors-in-Chief and the Referees for their helpful suggestions and
comments which have improved the presentation of the paper.

References

[1] G.R. Bitran, A.G. Novaes, Linear programming with a fractional objective function, Oper. Res. 21 (1973) 22–29.
[2] A. Charnes, W.W. Cooper, Management Models of Industrial Applications of Linear Programme (Appendix B),
vol.-I, Wiley, New York, 1961.
[3] A. Charnes, W.W. Cooper, Programming with linear fractional functions, Naval Res. Logist. Quart. 9 (1962) 181–
186.
[4] B.D. Craven, Fractional Programming, Heldermann Verlag, Berlin, 1988.
[5] D. Dutta, J.R. Rao, R.N. Tiwari, ERect of tolerance in fuzzy linear fractional programming, Fuzzy Sets and Systems
55 (1993) 133–142.
[6] D. Dutta, R.N. Tiwari, J.R. Rao, Multiple objective linear fractional programming—a fuzzy set theoretic approach,
Fuzzy Sets and Systems 52 (1992) 39–45.
[7] D. Dutta, R.N. Tiwari, J.R. Rao, Fuzzy approaches for multiple criteria linear fractional optimization: a comment,
Fuzzy Sets and Systems 54 (1993) 347–349.
[8] R.G. Dyson, Maximin programming, fuzzy linear programming and multicriteria decision making, J. Oper. Res. Soc.
31 (1980) 263–267.
[9] E.L. Hannan, Linear programming with multiple fuzzy goals, Fuzzy Sets and Systems 6 (1981) 235–248.
[10] E.L. Hannan, On fuzzy goal programming, Decision Sci. 12 (1981) 522–531.
[11] C.L. Hwang, A.S.M. Masud, Multiple Objective Decision Making—Methods and Applications, Springer, Berlin,
1979.
[12] J.P. Ignizio, Goal Programming and Extensions, Lexington D.C. Health, MA, 1976.
[13] J.S.H. Kornbluth, R.E. Steuer, Multiple objective linear fractional programming, Management Sci. 27 (1981) 1024–
1039.
[14] J.S.H. Kornbluth, R.E. Steuer, Goal programming with linear fractional criteria, European J. Oper. Res. 8 (1981) 58
–65.
[15] M.K. Luhandjula, Fuzzy approaches for multiple objective linear fractional optimization, Fuzzy Sets and Systems 13
(1984) 11–23.
[16] R.H. Mohamed, The relationship between goal programming and fuzzy programming, Fuzzy Sets and Systems 89
(1997) 215–222.
[17] R. Narasimhan, Goal programming in a fuzzy environment, Decision Sci. 11 (1980) 325–336.
[18] B.B. Pal, I. Basu, A goal programming method for solving fractional programming problems via dynamic
programming, Optimization 35 (1995) 145–157.
[19] C. Romero, A survey of generalized goal programming, European J. Oper. Res. 25 (1986) 183–191.
[20] M. Sakawa, K. Kato, Interactive decision-making for multiobjective linear fractional programming problems with
block angular structure involving fuzzy numbers, Fuzzy Sets and Systems 97 (1988) 19–31.
[21] M. Sakawa, T. Yumine, Interactive fuzzy decision-making for multiobjective linear fractional programming problems,
Large Scale Systems 5 (1983) 105–114.
[22] R.E. Steuer, An interactive multiple objective linear programming procedure, in: M.K. Starr, M. Zeleny (Eds.),
Multiple Criteria Decision Making, North-Holland, Amsterdam, 1977, pp. 225–239.
[23] R.N. Tiwari, S. Dharmar, J.R, Rao, Fuzzy goal programming—an additive model, Fuzzy Sets and Systems 24 (1987)
27–34.
 B.B. Pal et al. / Fuzzy Sets and Systems 139 (2003) 395 – 405 405

[24] L.A. Zadeh, The concept of linguistic variable and its application to approximate reasoning, Part I, II, III, Inform.
Sci. 8 (1975) 199–244, 301–352;
L.A. Zadeh, The concept of linguistic variable and its application to approximate reasoning, Part I, II, III, Inform.
Sci. 9 (1975) 43–80.
[25] H.-J. Zimmermann, Fuzzy programming and linear programming with several objective functions, Fuzzy Sets and
Systems 1 (1978) 45–55.
[26] H.-J. Zimmermann, Fuzzy Sets, Decision Making and Expert Systems, Kluwer Academic Publishers, Boston,
Dordrecht, 1987.