See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/313939623

A new approach for solving fully fuzzy linear fractional programming

problems using the multi-objective linear programming

Article in RAIRO - Operations Research · January 2017

DOI: 10.1051/ro/2016022

CITATIONS READS

65 391

3 authors:

Sapan Kumar Tarni Mandal

National Institute of Technology, Jamshedpur National Institute of Technology, Jamshedpur
19 PUBLICATIONS 260 CITATIONS 37 PUBLICATIONS 631 CITATIONS

SEE PROFILE SEE PROFILE

S A. Edalatpanah
Ayandegan Institute of Higher Education
206 PUBLICATIONS 2,597 CITATIONS

SEE PROFILE

All content following this page was uploaded by Sapan Kumar on 06 April 2017.

The user has requested enhancement of the downloaded file.

RAIRO-Oper. Res. 51 (2017) 285–297 RAIRO Operations Research
DOI: 10.1051/ro/2016022 www.rairo-ro.org

A NEW APPROACH FOR SOLVING FULLY FUZZY LINEAR FRACTIONAL

PROGRAMMING PROBLEMS USING THE MULTI-OBJECTIVE LINEAR
PROGRAMMING

Sapan Kumar Das1, Tarni Mandal1 and S.A. Edalatpanah2

Abstract. This paper deals with developing an efficient algorithm for solving the fully fuzzy linear
fractional programming problem. To this end, we construct a new method which is obtained from
combination of Charnes−Cooper scheme and the multi-objective linear programming problem. Fur-
thermore, the application of the proposed method in real life problems is presented and this method is
compared with some existing methods. The numerical experiments and comparative results presented
promising results to find the fuzzy optimal solution.

Mathematics Subject Classification. 90C70, 90C90, 90C05, 90C08.

Received July 20, 2015. Accepted February 27, 2016.

1. Introduction
Consider the following linear fractional programming:

ct x + q F (x)
Max t
= (1.1)
d x+r G(x)
s.t. ⎛ ⎞

Ax ⎝ = ⎠ b,


x  0.
Where x is the decision variables, A is the coefficients of the technological matrices, ct , dt are the coefficients
associated with the objective function. q, r ∈ R. For some values of x, G(x) may be equal to zero. To avoid
such cases, one requires that either
⎧ ⎛ ⎞ ⎛ ⎞ ⎫
⎨   ⎬
x  0, Ax ⎝ = ⎠ b, ⇒ G(x) > 0} or {x  0, Ax ⎝ = ⎠ b, ⇒ G(x) < 0 .
⎩   ⎭

Keywords. Fully fuzzy linear programming, linear fractional programming, linear programming, multi-objective linear
programming; triangular fuzzy number.
1 Department of Mathematics, National Institute of Technology Jamshedpur, 831014 Jamshedpur India.
[email protected]
2 Department of Applied Mathematics, Ayandegan Institute of Higher Education, Tonekabon, Iran.

Article published by EDP Sciences 

c EDP Sciences, ROADEF, SMAI 2017
286 S. KUMAR DAS AND T. MANDAL

Nowadays, the problem of linear fractional programming has significant application in different real life areas
such as production planning, financial sector, health care and all engineering fields. However, in real world
applications, certainty, reliability and precision of data is often illusory. The optimal solution of an LP only
depends on a limited number of constraints; therefore much of the collected information has little impact on
the solution. It is useful to consider the knowledge of experts about the parameters as fuzzy data.
The fuzzy linear fractional programming problems in which all the parameters and variables are represented
by fuzzy numbers are known as fully fuzzy linear fractional programming (FFLFP) problems. In the actual
cases the parameters may be uncertain or a vague estimation about the variables is known as those are found
in general by some experiment. To overcome the uncertainty and vagueness, one may use the fuzzy numbers
in the place of the crisp numbers. Thus the crisp system of linear fractional programming problem becomes a
fuzzy system of FLFP problem or FFLFP problem. In the FFLFP problem all the parameters and variables
are considered to be fuzzy numbers. It is an important issue to develop mathematical models and numerical
techniques which apply in real life circumstances that would appropriately treat the general fuzzy or fully fuzzy
linear fractional programming because subtraction and division of fuzzy numbers are not the inverse operations
to addition and multiplication respectively. Nowadays, the problem of FLFP problems has significant application
in different real life areas such as production planning, financial sector, health care and all engineering fields.
For this reason, this is an important area of research in the recent years. In this paper, we consider the FFLFP
problem. In recent years, many methods currently exist for solving FFLFP problem under nonnegative fuzzy
variables.
The concept of fuzzy set and fuzzy numbers was first introduced [13]. In [7], the authors proposed a method
to solve multi-objective linear fractional programming (MOLFP) problem under a fuzzy satisfied. A general
concept of Pareto optimal solution and use two types of fuzzy goals (called fuzzy equal and fuzzy min) was
introduced in [8]. In [13] the concept of linguistic variation was introduced. The FFLP problem by establishing
all the coefficients and variables of a linear program as being fuzzy quantities was introduced in [1].
Recently Pop and Minasian [6], proposed a method for solving fully falsified linear fractional programming
problems where all the parameters and variables are triangular fuzzy numbers. In [10, 11], they considered the
same problem of [6] for solving fully fuzzy linear fractional programming problem.
In this paper, we modified the methods of [6, 11]. First we transform the FFLFP problem into a FFLP
problem with the help of Charnes−Cooper method [3]. Then using a new technique, the FFLP problem will be
converted into a multi-objective linear programming (MOLP) problem. We also prove that this solution can be
considered as an exact solution of FFLFP problem. Finally, we show that advantages of the proposed method
over the existing method [6, 11].
This paper is organized as follows: some basic definitions and notations are presented in Section 2. In Section 3,
the general form of linear fractional programming (LFP) problem is presented. In Section 4, the general form
of FFLFP problem is presented and a new method is proposed for solving FFLFP problems. Advantages of the
proposed method over the existing method with the help of a simple example are provided in Section 5. To
show the application of the proposed method a real life problem is solved in Section 6. In Section 7, we discuss
the importance of our results. Finally, conclusion has been drawn in the last section.

2. Preliminaries
In this section, we have presented some basic concept of fuzzy triangular number, which was very useful in
this paper.

Definition 2.1 ([10]). Let X denotes a universal set. Then a fuzzy subset A of X is defined by its membership
function μà : X → [0, 1]; which assigned a real number μà (X) in the interval [0, 1], to each element x ∈ X,
 SOLVING FFLFP PROBLEMS USING MOLP PROBLEMS 287

where the values of μà (X)x shows the grade of membership of x in Ã. A fuzzy subset à can be characterized
as a set of ordered pairs of element x and grade μà (X) and is often written à = (x, μà (x)) : x ∈ X is called a
fuzzy set.

Definition 2.2 ([10]). A fuzzy number à = (b, c, a) is said to be a triangular fuzzy number if its membership
function is given by
⎧
⎪ (x − b)
⎪
⎪ , b  x  c,
⎪
⎨ (c − b)
⎪
μà (X) = (a − x)
⎪
⎪ , c  x  a,
⎪
⎪ (a − c)
⎪
⎩
0, else.

Definition 2.3. A triangular fuzzy number (b, c, a) is said to be non-negative (non-positive) triangular fuzzy
number if and only if b  0 (a  0).

Definition 2.4 ([10]). Two triangular fuzzy number à = (b, c, a) and B̃ = (e, f, d) are said to be equal if and
only if b = e, c = f, a = d.

Definition 2.5 ([6]). A ranking is a function R : F (R) → R where F (R) is a set of fuzzy number defined on set
of real numbers, which maps each fuzzy number into the real line, where a natural order exists. Let à = (b, c, a)
is a triangular fuzzy number then (Ã) = b + 2c
4
+a
.

Definition 2.6 ([6]). Letà = (b, c, a), B̃ = (e, f, d) be two triangular fuzzy numbers then the fuzzy arithmetic
is defined as follows:

(i) Ã + B̃ = (b, c, a) + (e, f, d) = (b + e, c + f, a + d);

(ii) −Ã = (−a, −c, −b);
(iii) Ã − B̃ = (b, c, a) − (e, f, d) = (b − d, c − f, a − e);
(iv) Let à = (b, c, a) be any triangular fuzzy number and B̃ = (e, f, d) be a non-negative triangular fuzzy
number then one may have,

⎧
⎪ (be, cf, ad) if b  0,
⎪
⎪
⎨
à ⊗ B̃ = ÃB̃ = (bd, cf, ad) if b < 0, a  0,
⎪
⎪
⎪
⎩
(bd, cf, cd) if c < 0,

Definition 2.7. Let à = (b, c, a), B̃ = (e, f, d) be two triangular fuzzy numbers. We say that à is relatively
less than B̃, if and only if:

(i) b < e or
(ii) b = e and (b − c) > (e − f ) or
(iii) b = e, (b − c) = (e − f ) and (a + b) = (d + e).

Note. It is clear from Definition 2.7 that b = e, (b − c) = (e − f ) and (a + b) = (d + e) if and only if à = B̃.
288 S. KUMAR DAS AND T. MANDAL

3. Linear fractional programming problem and its fully fuzzy version

In this section the general form of linear fractional programming problem, converting linear fractional pro-
gramming into the linear programming problem and the general form of multi-objective linear fractional pro-
gramming problem is discussed. For satisfaction, we assume that (1.1) satisfies the condition that:
⎧ ⎛ ⎞ ⎫
⎨  ⎬
x  0, Ax ⎝ = ⎠ b, ⇒ G(x) > 0 . (3.1)
⎩  ⎭

Now, by the following theorem we show that the linear fractional programming problem can be transform into
a LP problem.

Theorem 3.1. Assume that (y, 0) with y  0 is feasible for the following LP problem:

Max ct y + qα

s.t dt y + rα = 1,
Ay − bα = 0,

α  0, y  0. (3.2)
Then following the condition (3.1), the LFP (3.1) is equivalent into LP (3.2).

Proof. Let x be the feasible solution of LFP (3.1). Define q(x) = (y, α), with α = (dt x + r)−1 and y = αx. Then
y  0, α > 0, Ay − bα = α(Ax − b) = 0, dt y + rα = α(dt x + r) = 1, Thus (y, α), is feasible for LP (3.2).
Conversely, if (y, α), is feasible for LP (3.2) and (y, 0) is feasible for LP (3.2), then α > 0, and x = y/t satisfies
x> = 0, Ax − b = (Ay − bα)/α = 0. Therefore, the feasible set of LFP problem is one-one onto the feasible set
for LP (3.2).
Moreover, the objective functions are related by

(ct x + q)/ (dt x + r) = (ct y + qα)/ (dt y + rα) = (ct y + qα)/1.

Thus, the theorem is proved. 

Next, we will model the fully fuzzy linear fractional programming problem where all the variables and all
the parameters are triangular fuzzy numbers. Let us consider a general format of fully fuzzy linear fractional
programming problem as follows:
c̃t x̃ + q̃
Max Z̃ =
d˜t x̃ + r̃
s.t. Ãx̃  b̃

x̃  0. (3.3)
First, we transform problem (3.3) in to a fully fuzzified LP problem using Charnes−Cooper transformation [3],
and obtain the following problem.
Max c̃t ỹ + q̃ t̃
subject to Ãỹ − b̃t̃  0̃,
˜ + r̃ t̃  1̃,
dỹ

ỹ, t̃  0. (3.4)
 SOLVING FFLFP PROBLEMS USING MOLP PROBLEMS 289

Let us assume that all the parameters x̃, c̃, q̃, d, ˜ r̃, b̃ and z̃ are represented by triangular fuzzy numbers
((x) , (x) , (x) ), ((c ) , (c ) , (c ) ), ((q) , (q) , (q)r ), ((dt )p , (dt )q , (dt )r ), ((b)p , (b)q , (b)r ), ((r)p , (r)q , (r)r ) and
p q r t p t q t r p q

((a)p , (a)q , (a)r ), respectively. After computing the fuzzy quantities with respect to Definitions 2.2 and 2.3 we
modify the maximization of the objective function and constraints function described by a triangular fuzzy
number with the maximization of three values of the fuzzy number. Then we can rewrite the mentioned FFLP
as follows:
Max (z1 , z2 , z3 ) = ((ct y)p , (ct y)q , (ct y)r ) + ((q)p , (q)q , (q)r ) (3.5)
s.t. ((a)p , (a)q , (a)r ) ⊗ ((y)p , (y)q , (y)r ) − ((bt)p , (bt)q , (bt)r )  0̃,
((dt y)p , (dt y)q , (dt y)r + (rt)p , (rt)q , (rt)r )  1̃,
((y)p , (y)q , (y)r )  0. ((t)p , (t)q , (t)r )  0.

Remark 3.2. If ỹ ∗ = ((y ∗ )p , (y ∗ )q , (y ∗ )r ) is said to be an exact optimal solution of problem (3.4) then it will
satisfies in the following conditions:

(i) Ãỹ ∗ = b̃;

(ii) ∀ỹ = ((y)p , (y)q , (y)r ), we have that, c̃t ỹ + q̃ t̃  c̃t ỹ ∗ + q̃ t̃ (in the case of minimization problem c̃t ỹ + q̃ t̃ 
c̃t ỹ ∗ + q̃ t̃).

Remark 3.3. Let ỹ ∗ be an exact optimal solution of the problem (3.4) then y is called the alternative exact
 
optimal solution of the problem (3.4) if there exists c̃t ỹ + q̃ t̃ = c̃t ỹ ∗ + q̃ t̃, ∃ ỹ ∈ s̃.

The proposed approach for solving FFLFPP can be summarized as follows:

Step 1. With respect to Definitions 2.2 and 2.3 the problem (3.5), can be rewritten as:

Max ((ct y + qt)p , (ct y + qt)q , (ct y + qt)r )

s.t.
((ay)p , (ay)q , (ay)r ) − ((bt)p , (bt)q , (bt)r )  0̃,
((dt y + rt)p , (dt y + rt)q , (dt y + rt)r )  1̃,

(y)p , (y)q , (y)r  0. ((t)p , (t)q , (t)r )  0. (3.6)

Step 2. Equivalently, with regard to Definition 2.4 the problem (3.6) may be written as follows:

Max ((ct y + qt)p , (ct y + qt)q , (ct y + qt)r )

s.t.
(ay)p − (bt)r  0,
(ay)q − (bt)q  0,
(ay)r − (bt)p  0,
(dt y + r)p + tp  1,
(dt y + r)q + tq  1,
(dt y + r)r + tr  1,

(y)q − (y)p  0, (y)r − (y)q  0, (y)p  0, (t)q − (t)p  0, (t)r − (t)q  0, (t)p  0. (3.7)
290 S. KUMAR DAS AND T. MANDAL

Step 3. Regarding Definition 2.6 the problem (3.7) is converted to the MOLP problem with three crisp objective
functions and the constraints are changed as follows:

Max (ct y + qt)p

Max (ct y + qt)p − (ct y + qt)q

Max (ct y + qt)p + (ct y + qt)r
s.t
(ay)p − (bt)r  0.
(ay)p − (bt)r − ((ay)q − (bt)q )  0,
(ay)p − (bt)r + ((ay)r − (bt)p )  0,
(dt y + r)p + tp  1,
(dt y + r)p + tp − (dt y + r)q − tq  0,
(dt y + r)p + tp + (dt y + r)r + tr  2,

(y)q − (y)p  0, (y)r − (y)q  0, (y)p  0, (t)q − (t)p  0, (t)r − (t)q  0, (t)p  0. (3.8)
Step 4. In objective functions the lexicographic method will be used to obtain the optimal solution of the
problem (3.8). So, we get:
Max (ct y + qt)p
s.t.
(Ay)p − (bt)r  0,
(Ay)p − (bt)r − ((Ay)q − (bt)q )  0,
(Ay)p − (bt)r + ((Ay)r − (bt)p )  0,
(dt y + r)p + tp  1,
(dt y + r)p + tp − (dt y + r)q − tq  0,
(dt y + r)p + tp + (dt y + r)r + tr  2,

(y)q − (y)p  0, (y)r − (y)q  0, (y)p  0, (t)q − (t)p  0, (t)r − (t)q  0, (t)p  0. (3.9)
If we get a unique optimal solution then it is an optimal solution of the problem (3.5) and stop. Otherwise go
to next step.
Step 5. Solve the following problem over the optimal solutions that are achieved in Step 4 as follows:

Max (ct y + qt)p − (ct y + qt)q

s.t.
(ct y + qt)p = l∗
(Ay)p − (bt)r  0,
(Ay)p − (bt)r − ((Ay)q − (bt)q )  0,
(Ay)p − (bt)r + ((Ay)r − (bt)p )  0,
(dt y + r)p + tp  1,
 SOLVING FFLFP PROBLEMS USING MOLP PROBLEMS 291

(dt y + r)p + tp − (dt y + r)q − tq  0,

(dt y + r)p + tp + (dt y + r)r + tr  2,

(y)q − (y)p  0, (y)r − (y)q  0, (y)p  0, (t)q − (t)p  0, (t)r − (t)q  0, (t)p  0, (3.10)
∗
where l is the optimal value of problem (3.9). If the problem (3.9) has a unique optimal solution, then it is an
optimal solution of problem (3.5) and stop. Otherwise go to next step.
Step 6. Solve the following problem over the optimal solutions that are achieved in Step 5 as follows:

Max (ct y + qt)p + (ct y + qt)r

s.t.
(ct y + qt)p − (ct y + qt)q = k ∗
(ct y + qt)p = l∗
(Ay)p − (bt)r  0,
(Ay)p − (bt)r − ((Ay)q − (bt)q )  0,
(Ay)p − (bt)r + ((Ay)r − (bt)p )  0,
(dt y + r)p + tp  1,
(dt y + r)p + tp − (dt y + r)q − tq  0,
(dt y + r)p + tp + (dt y + r)r + tr  2,

(y)q − (y)p  0, (y)r − (y)q  0, (y)p  0, (t)q − (t)p  0, (t)r − (t)q  0, (t)p  0, (3.11)
∗
where k is the optimal value of problem (3.10). Hence, the optimal solution of the problem (3.5) is obtained
by solving problems (3.11).
Now by theorem we clarify the lexicographic optimal solution of the problem (3.9) can be considered as an
exact optimal solution of the problem (3.5).
Theorem 3.4. If ỹ ∗ = (y ∗ )p , (y ∗ )q , (y ∗ )r be an optimal solution of problems (3.9)−(3.11), then it is also an
exact optimal solution of the problem (3.10).
Proof. By the method of contradiction, let ỹ ∗ = (y ∗ )p , (y ∗ )q , (y ∗ )r be an optimal solution of (3.9)−(3.11), but
it is not the exact optimal solution of the problem (3.5). Let us we considerỹ 0 = (y 0 )p , (y 0 )q , (y 0 )r , such that
in the case of maximization:

(ct y ∗ )p , (ct y ∗ )q , (ct y ∗ )r ≺ (ct y 0 )p , (ct y 0 )q , (ct y 0 )r . 

Based on Definition 2.7, we have three conditions as follows:

Case (i). In case of maximization, we consider (ct y ∗ )p ≺ (ct y 0 )p . Also, with respect to the assumption we have:

(Ay 0 )p = (b̃1 ),
(Ay 0 )p − (Ay 0 )q = (b̃1 ) − (b̃2 ),
(Ay 0 )p + (Ay 0 )r = (b̃1 ) + (b̃3 ),
(y 0 )p − (y 0 )q  0, (y 0 )r − (y 0 )q  0, (y 0 )p  0.

Therefore, (y 0 )p , (y 0 )q , (y 0 )r is a feasible solution of problem (3.9) in which the objective value in

(y 0 )p , (y 0 )q , (y 0 )r is greater than the objective value in (y ∗ )p , (y ∗ )q , (y ∗ )r . But it is contradiction.
292 S. KUMAR DAS AND T. MANDAL

Case (ii). In case of maximization, let us we consider (ct y ∗ )p = (ct y 0 )p , and (ct y ∗ )p −(ct y ∗ )q (ct y 0 )p −(ct y 0 )q
also, with respect to the assumption we have:

(Ay 0 )p = (b̃1 ),

(Ay 0 )p − (Ay 0 )q = (b̃1 ) − (b̃2 ),

(Ay 0 )p + (Ay 0 )r = (b̃1 ) + (b̃3 ),
(y 0 )p − (y 0 )q  0, (y 0 )r − (y 0 )q  0, (y 0 )p  0.
Therefore, (y 0 )p , (y 0 )q , (y 0 )r is a feasible solution of problem (3.5) in which the objective value in
(y 0 )p , (y 0 )q , (y 0 )r is greater than the objective value in (y ∗ )p , (y ∗ )q , (y ∗ )r . But it is contradiction.
Case (iii). In case of maximization, let us we consider (ct y ∗ )p = (ct y 0 )p , (ct y ∗ )p − (ct y ∗ )q = (ct y 0 )p − (ct y 0 )q
and (ct y ∗ )p + (ct y ∗ )r ≺ (ct y 0 )p + (ct y 0 )r . Also, with respect to the assumption we have:

(Ay 0 )p = (b̃1 ),

(Ay 0 )p − (Ay 0 )q = (b̃1 ) − (b̃2 ),

(Ay 0 )p + (Ay 0 )r = (b̃1 ) + (b̃3 ),
(y 0 )p − (y 0 )q  0, (y 0 )r − (y 0 )q  0, (y 0 )p  0.
Therefore, (y 0 )p , (y 0 )q , (y 0 )r is a feasible solution of problem (3.11) in which the objective value in
(y ) , (y 0 )q , (y 0 )r is greater than the objective value in (y ∗ )p , (y ∗ )q , (y ∗ )r . But it is contradiction.
0 p

Therefore ỹ ∗ = (y ∗ )p , (y ∗ )q , (y ∗ )r is an exact optimal solution of problem (3.5).

4. Numerical example
In order to illustrate the proposed method to solve fully fuzz linear fractional programs, consider the following
example:

Example 4.1 ([6]). Let us consider the following linear fractional program as follows:
x1 − x2 + 1
Max z =
x1 + x2 + 2
s.t. x1 + x2  2,
x1 − x2  1,
x1 , x2  0.

The optimal solution of the problem is x1 = 1, x2 = 0 and the optimal value of z is 2/3 = 0.66667.
We attach now to this problem a fully fuzzified problem. Thus the fully fuzzy linear fractional programming
problem which we want to solve is as follows:

(0, 1, 2)x̃1 − (−2, −1, 0)2x̃2 + (0, 1, 2)

Max z = (4.1)
(0, 1, 2)x̃1 + (0, 1, 2)x̃2 + (1, 2, 3)
s.t.
(1, 2, 3)x̃1 + (1, 2, 3)x̃2  (1, 2, 3),
(0, 1, 2)x̃1 − (−2, −1, 0)x̃2  (0, 1, 2),
x̃1 , x̃2  0.
 SOLVING FFLFP PROBLEMS USING MOLP PROBLEMS 293

Now the problem is as follows:

(0, 1, 2) ⊗ ((x1 )p , (x1 )q , (x1 )r ) − (−2, −1, 0) ⊗ ((x2 )p , (x2 )q , (x2 )r ) + (0, 1, 2)
Max
(0, 1, 2) ⊗ ((x1 )p , (x1 )q , (x1 )r ) + (0, 1, 2) ⊗ ((x2 )p , (x2 )q , (x2 )r ) + (1, 2, 3)
s.t.
(0, 1, 2) ⊗ ((x1 )p , (x1 )q , (x1 )r ) + (0, 1, 2) ⊗ ((x2 )p , (x2 )q , (x2 )r )  (1, 2, 3),
(0, 1, 2) ⊗ ((x1 )p , (x1 )q , (x1 )r ) − (−2, −1, 0) ⊗ ((x2 )p , (x2 )q , (x2 )r )  (0, 1, 2),
(x1 )p , (x1 )q , (x1 )r , (x2 )p , (x2 )q , (x2 )r  0.
According to Steps 2 and 3, we get the following multiple objective linear programming problem:

Max (−2(y2 )p , (y1 )q − (y2 )q + t2 , 2(y1 )r + 2t3 )

s.t.

−3t3  0,
(y1 )q + (y2 )q − 2t2  0,
2(y1 )r + 2(y2 )r − t1  0,
−2(y2 )r − 2t3  0,
(y1 )q − (y2 )q − t2  0,
2(y1 )r  0,
t1  1,
q q
(y1 ) + (y2 ) + 2t2  1,
2(y1 )r + 2(y2 )r + 3t3  1,
(y)q − (y)p  0, (y)r − (y)q  0, (y)p  0, t1 , t2 , t3  0.

Using Steps 3, 4, 5 and Step 6, the optimal solution of the problem is:

ỹ ∗ = {ỹ1∗ = ((y1∗ )p , (y1∗ )q , (y1∗ )r ) = (0, 0.2, 0.2),

ỹ ∗ = {ỹ2∗ = ((y2∗ )p , (y2∗ )q , (y2∗ )r ) = (0, 0, 0),
t∗ = {(t1 , t2 , t3 ) = (0.1, 0.2, 0.2).

Then the solution of the problem is:

x̃∗ = {x̃∗1 = ((x∗1 )p , (x∗1 )q , (x∗1 )r ) = (0, 1, 1),

x̃∗ = {x̃∗2 = ((x∗2 )p , (x∗2 )q , (x∗2 )r ) = (0, 0, 0).

The triangular fuzzy number z̃ ∗ = (0, 0.66667, 4)

The obtained result is exactly the optimal value of the problem which start with the original problem.
By comparing proposed method results with existing method [6, 11], based on Definition 2.7, we conclude
that our result is more efficient than other existing method.

0 = (z1∗ )proposed method=(z1∗ ) method of [3 ] > (z1∗ ) method of [2 ] = −0.21

0.66667 = (z1∗ + z2∗ )proposed method = (z1∗ + z2∗ ) method of [2] = 0.66667>(z1∗ + z2∗ ) method of [3 ] = 0.55
(0, 0.66667, 4) = (z ∗ )proposed method >(z ∗ ) method of [2 ] = (−0.21, 0.66667, 5.822)>(z1∗ ) method of [3 ] = (0, 0.55, 1.09).
294 S. KUMAR DAS AND T. MANDAL

5. Application of the proposed method in real life problems

In the existing method [6,11] it is difficult to solve for the fully fuzzy transportation problem in real life case.
In the existing methods [6, 11], problem reduced to the deterministic MOLFP problem subject to conjunctive
system of disjunctive non-linear constraints. So, it is very difficult to solve a transportation problem. So, we
applied in our proposed method for solving fully fuzzy transportation problem.
In this section, to show the application of proposed method the real life problem is solved by the proposed
method and it is concluded that the proposed method applied any real life problems.

Example 5.1 (Production planning in Taiwan).

Dali Company is the leading producer of soft drinks and low-temperature foods in Taiwan. Currently, Dali
plans to develop the South-East Asian market and broaden the visibility of Dali products in the Chinese
market. Notably, following the entry of Taiwan to the World Trade Organization, Dali plans to seek strategic
alliance with prominent international companies and introduced international bread to lighten the embedded
future impact. In the domestic soft drinks market, Dali produces tea beverages to meet demand from four
distribution centers in Taichung, Chiayi, Kaohsiung, and Taipei, with production being based at three plants in
Changhua, Touliu, and Hsinchu. According to the preliminary environmental information, Table 1 summarizes
the potential supply available from the given three plants. The forecast demand from the four distribution
centers as is shown Table 2. The profit of the company gained by each route is presented in Table 3. Table 4
summarizes the unit shipping cost for each route for the upcoming season. The environmental coefficient and
related parameters generally are imprecise numbers with triangular possibility distributions over the planning
horizon due to incomplete or unobtainable information. For example, the unavailable supply of the Changhua
plant is (7.2, 8, 8.8) thousand dozen bottles, the forecast demand of the Taichung distribution center is (6.2,
7, 7.8) thousand dozen bottles, profit per dozen bottles for Changhua to Taichung is (8, 10, 10.8) dollars and
the transportation cost per dozen bottles for Changhua to Taichung is (8, 10, 10.8)dollars. The management of
Dali is initiating a study to maximize the profit as much as possible.

Table 1. Supply of the plants.

Source Changhua Touliu Hsinchu

Supply (Thousand dozen bottles) (7.2, 8, 8.8) (12, 14, 16) (10.2, 12, 13.8)

Table 2. Demand of the destinations.

Destination Taichung Chiayi Kaohsiung Taipei

Demand (thousand dozen bottles) (6.2, 7, 7.8) 8.9, 10, 11.1) (6.5, 8, 9.5) (7.8, 9, 10.2)

Table 3. Profit of the company.

Source Destination
Taichung Chiayi Kaohsiung Taipei
Changhua (8, 10, 10.8) (20.4, 22, 24) (8, 10, 10.6) (18.8, 20, 22)
Touliu (14, 15, 16) (18.2, 20, 22) (10, 12, 13) (6, 8, 8.8)
Hsinchu (18.4, 20, 21) (9.6, 12, 13) (7.8, 10, 10.8) (14, 15, 16)
 SOLVING FFLFP PROBLEMS USING MOLP PROBLEMS 295

Table 4. Shipping costs.

Source Destination
Taichung Chiayi Kaohsiung Taipei
Taichung (1.5, 2, 2.5) (4, 5, 6) (1.3, 2, 2.5) (3, 4, 5)
Touliu (2.5, 3, 4) (2, 3, 4) (2.3, 3, 4) (1.5, 2, 2.5)
Hsinchu (3, 4, 5) (2, 3, 4) (1.5, 2, 2.7) (2, 3, 4)

The real world problem can be modeled to the following FFLFP problem:
⎧ ⎫
⎨ (8, 10, 10.8)x̃11 + (20.4, 22, 24)x̃12 + (8, 10, 10.6)x̃13 + (18.8, 20, 22)x̃14 + ⎬
(14, 15, 16)x̃21 + (18.2, 20, 22)x̃22 + (10, 12, 13)x̃23 + (6, 8, 8.8)x̃24 +
⎩ (18.4, 20, 21)x̃ + (9.6, 12, 13)x̃ + (7.8, 10, 10.8)x̃ + (14, 15, 16)x̃ ⎭
31 32 33 34
Max Z = ⎧ ⎫
⎨ (1.5, 2, 2.5)x̃11 + (4, 5, 6)x̃12 + (1.3, 2, 2.5)x̃13 + (3, 4, 5)x̃14 + ⎬
(2.5, 3, 4)x̃21 + (2, 3, 4)x̃22 + (2.3, 3, 4)x̃23 + (1.5, 2, 2.5)x̃24 +
⎩ (3, 4, 5)x̃ + (2, 3, 4)x̃ + (1.5, 2, 2.7)x̃ + (2, 3, 4)x̃ ⎭
31 32 33 34

s.t.

x̃11 + x̃12 + x̃13 + x̃14  (7.2, 8, 8.8)

x̃21 + x̃22 + x̃23 + x̃24  (12, 14, 16)
x̃31 + x̃32 + x̃33 + x̃34  (10.2, 12, 13.8)
x̃11 + x̃21 + x̃31  (6.2, 7, 7.8)
x̃12 + x̃22 + x̃32  (8.9, 10, 11.1)
x̃13 + x̃23 + x̃33  (6.5, 8, 9.5)
x̃14 + x̃24 + x̃34  (7.8, 9, 10.2)
x̃ij  0, where x̃ij = (x1ij , x2ij , x3ij ), i = 1, 2, 3; j = 1, 2, 3, 4.

After solving this problem by using Steps 1, 2, 3, 4, 5 and 6, we have

⎡ ⎤
(0, 0, 0),
⎢ (0, 0, 0.25), ⎥
⎢ ⎥
⎢ (0, 0, 0.625), ⎥
⎢ ⎥
⎢ ⎥
⎢ (7.64, 8, 8), ⎥
⎢ ⎥
⎢ (0, 0, 0), ⎥
⎢ ⎥
⎢ (0, 0, 0), ⎥
⎢ ⎥
x̃ = ⎢ ⎥.
⎢ (0, 0, 1.45 ), ⎥
⎢ ⎥
⎢ (12, 14, 14), ⎥
⎢ ⎥
⎢ (5.8, 7, 7.5), ⎥
⎢ ⎥
⎢ ⎥
⎢ (8.8, 10, 11.25), ⎥
⎢ ⎥
⎣ (6.4, 8, 8), ⎦
(5.8, 10, 12.5).

Now the optimal value of the problem may be written as:

(Z̃) proposed method = ((Z1 ), (Z2 ), (Z3 )) = (2.26, 4.64, 9.48).
296 S. KUMAR DAS AND T. MANDAL

Figure 1. Membership functions of the optimal solution by the present method and existing
method [6, 11].

By comparing proposed method results with existing method [6, 11], based on Definition 2.7, we conclude
that our result is more efficient than other existing method.

2.26 = (z1∗ ) proposed method = (z1∗ )method of [3 ] >(z1∗ ) method of [2 ] = 2

4.64 = (z1∗ + z2∗ )proposed method = (z1∗ + z2∗ ) method of [2 ] = 4.64>(z1∗ + z2∗ ) method of [3 ] = 4.2
(2.26, 4.64, 9.48) = (z ∗ ) proposed method >(z
∗
) method of [2 ] = (2, 4.64, 9)>(z1∗ ) method of [3 ] = (2, 4.2, 11).

6. Results and discussion

The FFLFP problem, chosen in Example 4.1 and one real life transportation problem can be solved by
using the proposed method. In the existing method [6, 11] it is difficult to solve for fully fuzzy transportation
problem in real life case. Obtained result by the present method has been compared with the results of existing
method [6, 11]. It is worth mentioning that one may check that the results obtained by the existing method
may not be satisfied the constraints properly where the results obtained by the present method satisfied those
constraints exactly. In the proposed methodology the FFLFP problem turns into a crisp LP problem and that
problem is solved by using LINGO Version 11.0. According optimal solution by both the methods is depicted
in Figures 1 and 2.

7. Conclusions
In this paper, a new solving procedure has been suggested to solve the FFLFP problem. Using
Charnes−Cooper transformation method, we transformed the fully fuzzy linear fractional programming prob-
lem into fully fuzzy linear programming problem. After that, the fully fuzzy linear programming problem is
converted into its equivalent MOLP problem. It is our belief that the proposed method for solution of FFLFP
 SOLVING FFLFP PROBLEMS USING MOLP PROBLEMS 297

Figure 2. Membership functions of the optimal solution by the present method and existing
method [6, 11].

problem in real life problem as well as simple problem may be of considerable interest for mathematician working
in this field.

References
[1] J.J. Buckley and T. Feuring, Evolutionary algorithm solution to fuzzy problems: fuzzy linear programming. Fuzzy Sets Syst.
109 (2000) 35–53.
[2] M. Chakraborty and S. Gupta, Fuzzy mathematical programming for multi objective linear fractional programming problem.
Fuzzy Sets Syst. 125 (2002) 335–342.
[3] A. Charnes and W.W. Cooper, Programming with linear fractional functionals. Nav. Res. Logist. Q. 9 (1962) 181–186.
[4] D. Dutta, R.N. Tiwari and J.R. Rao, Multiple objective linear fractional programming- A fuzzy set theoretic approach. Fuzzy
Sets Syst. 52 (1992) 39–45.
[5] R. Ezzati, E. Khorram and R. Enayati, A new algorithm to solve fully fuzzy linear programming problems using MOLP
problem. Appl. Math. Model. 39 (2015) 3183–3193.
[6] B. Pop and I.M. Stancu-Minasian, A method of solving fully fuzzified linear fractional programming problems. J. Appl. Math.
Comput. 27 (2008) 227–242.
[7] M. Sakawa and H. Yano, An interactive fuzzy satisficing method for multiobjective linear fractional programming problems.
Fuzzy Sets Syst. 28 (1988) 129–144.
[8] M. Sakawa, H. Yano and J. Takahashi, Pareto optimality for multiobjective linear fractional programming problems with fuzzy
parameters. Inf. Sci. 63 (1992) 33–53.
[9] I.M. Stancu-Minasian and B. Pop, On a fuzzy set approach to solving multiple objective linear fractional programming problem.
Fuzzy Sets Syst. 134 (2003) 397–405.
[10] B. Stanojevic and I.M. Stancu-Minasian, On solving fuzzified linear fractional programs. Adv. Model. Optim. 11 (2009) 503–
523.
[11] B. Stanojevic and I.M. Stancu-Minasian, Evaluating fuzzy inequalities and solving fully fuzzified linear fractional programs.
Yugoslav J. Oper. Res. 1 (2012) 41–50.
[12] M.D. Toksari, Taylor series approach to fuzzy multi objective linear fractional programming. Inf. Sci. 178 (2008) 1189–1204.
[13] L.A. Zadeh, Fuzzy Sets. Inf. Control 8 (1965) 338–353.

View publication stats