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

net/publication/267163398

Fourier method for solving transportation problems with mixed constraints

Article · January 2010

CITATIONS READS

32 223

2 authors:

Pa.Shanthi Pandian Geethanjali Natarajan

Sri Sairam Engineering college United Institute of Technology
57 PUBLICATIONS   791 CITATIONS    26 PUBLICATIONS   584 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Geethanjali Natarajan on 02 September 2021.

The user has requested enhancement of the downloaded file.

 Int. J. Contemp. Math. Sciences, Vol. 5, 2010, no. 28, 1385 - 1395

Fourier Method for Solving Transportation

Problems with Mixed Constraints

P. Pandian and G. Natarajan
Department of Mathematics, School of Advanced Sciences
V I T University, Vellore-632 014, Tamil Nadu, India
[email protected]

Abstract
A new method called, Fourier transportation algorithm based on Modified
Fourier Elimination method is proposed for finding an optimal solution of
transportation problems with mixed constraints. This method is very easy to
understand and apply which can serve managers by providing the optimal solution to
a variety of distribution problems with mixed constraints. The proposed method is
illustrated with a numerical example.

Mathematics Subject Classifications: 90C08, 90C90

Keywords: Transportation problem, Mixed constraints, Optimal solution, Fourier

transportation method.

1 Introduction

The transportation problem is a special class of linear programming problem,

which deals with shipping commodities from sources to destinations. The objective
of the transportation problem is to determine the shipping schedule that minimize
that total shipping cost while satisfying supply and demand limits. The
transportation problem has an application in industry, communication network,
planning, scheduling, transportation and allotment etc. In literature, a good amount
of research has available to obtain an optimal solution for transportation problems
with equality constraints [2,4,7,12]. In real life, however, most problems have mixed
constraints accommodating many applications that go beyond transportation related
problems to include job scheduling, production inventory, production distribution,
allocation problems, and investment analysis. The transportation problems with
mixed constraints are not addressed in the literature because of the rigor required to
solve these problems optimally. A literature search revealed no systematic method
for finding an optimal solution of transportation problems with mixed constraints.
Gupta et al.[3] and Arsham [1] obtained the more-for-less solution for the
1386 P. Pandian and G. Natarajan

transportation problem with mixed constraints by relaxing the constraints and by

introducing new slack variables. While yielding the best more-for-less solution, their
method is tedious since it introduces more variables and requires solving sets of
complex equations. Perturbed method was used for solving the transportation
problems with constraints in [5,8,9].

Williams [11] developed a method called the Fourier elimination method

for solving linear programming problems based on the Fourier elimination
technique which generates more additional and redundant constraints. Kanniappan
and Thangavel [6] modified the method of Fourier elimination method developed
by Williams [11] for solving linear programming problems by choosing a variable
for elimination, which generates minimum number of constraints in each step
compared to that of the method of Fourier elimination method.

Recently, Veena et al.[10] have proposed a heuristic method for solving

transportation problems with mixed constraints which is based on the theory of
shadow price. The solution obtained by the heuristic method introduced by Veena et
al.[10] is an initial solution of the transportation problems with mixed constraints.

In this paper, we develop a new method called Fourier transportation

method for finding an optimal solution of transportation problems with mixed
constraints based on the Modified Fourier’s Elimination method [6] in single stage.
This method is very easy to understand and apply which can serve managers by
providing the optimal solution to a variety of distribution problems with mixed
constraints. The proposed method is illustrated with a numerical example.

2 Transportation problems with mixed constraints

Consider the mathematical model for a TP with mixed constraints .

m n
(P) Minimize z = ∑ ∑ cij xij
i =1 j=1
subject to
n
∑ xij ≥ ai , i ∈ Q (1)
j=1
n
∑ xij ≤ ai , i ∈ T (2)
j=1
n
∑ xij = ai , i ∈ S (3)
j=1
m
∑ xij ≥ bj , j∈ U (4)
i =1
Fourier method for solving transportation problems 1387

m
∑ xij ≤ bj , j∈ V (5)
i =1
m
∑ xij = bj , j∈ W (6)
i =1
xij ≥ 0, i = 1,2,..., m and j = 1,2,..., n and integers (7)
where Q, T and S are pairwise disjoint subsets of {1,2,3,..., n } such that
Q ∪ T ∪ S = {1,2,3,..., n };
U , V and W are pairwise disjoint subsets of {1,2,3,..., m } such that
U ∪ V ∪ W = {1,2,3,..., m };
cij is the cost of shipping one unit from supply point i to the demand point j ;
ai is the supply at supply point i ;
b j is the demand at demand point j and
x ij is the number of units shipped from supply point i to demand point j.

Remark 1 : If Q, T, U and V are empty, then the problem (P) becomes the TP with
equality constraints.

The transportation problem with mixed constraints (P) can be represented

by a matrix of order m x n called the cost matrix. For m = 3 and n = 3 , the
structure of cost matrix is

1 2 3 Supply
1 l h k =e
2 p q r ≥f
3 x y z ≤g
Demand =a ≥b ≤c

The maximization problem corresponding to the problem (P) is given below.

⎛ m n ⎞
(XP) Maximize w = − ⎜ ∑ ∑ cij x ij ⎟
⎜ i =1 j =1 ⎟
⎝ ⎠
subject to (1) to (7) are satisfied where w = − z .

The equivalent problem to the problem (XP) is given below.

(EP) Maximize w
m n
subject to ∑ ∑ c ij x ij + w ≤ 0
i =1 j =1
and also, (1) to (7) are satisfied .
1388 P. Pandian and G. Natarajan

Theorem 1: If { xijo : i = 1,2,..., m and j = 1,2,..., n } is an optimal solution for the

problem (EP), then { xijo : i = 1,2,..., m and j = 1,2,..., n } is also an optimal solution of
m m
(P) and the minimum value of z , zo = ∑ ∑ cij xijo .
i =1 j=1
Proof. Since (EP) is an equivalent problem of (XP), then
{ xijo : i = 1,2,..., m and j = 1,2,..., n } is an optimal solution for the problem (XP).
Since (XP) is the maximization type problem corresponding to the problem (P), then
{ xijo : i = 1,2,..., m and j = 1,2,..., n } is an optimal solution for the problem (P) and
m m
the minimum value of z , zo = ∑ ∑ cij xijo .
i =1 j=1
Hence the theorem.

Dominance property: Consider the following system of inequalities

a1x1 + a2 x2 + a3 x3 ≤ d1 (n1)
b1x1 + b2 x2 + b3 x3 ≤ d 2 (n2 )
d d d 2 d1 d d
If 2 ≤ 1 , ≤ and 2 ≤ 1 where each pair d 2 and d1 , b1 and a1 ,
b1 a1 b2 a2 b3 a3
b2 and a2 and b3 and a3 is of same sign or zero, we say that the inequality (n1)
is dominated by the inequality (n2 ) . Also, inequality (n1) is called the redundant
constraint of the system.

Theorem 2: Consider the system

a11 x1 + a12 x 2 + ... + a1n x n ≤ b1
a 21 x1 + a 22 x 2 + ... + a 2n x n ≤ b2
M . (I)
a m1 x1 + a m 2 x 2 + ... + a mn x n ≤ bm
x1, x2 ,..., xn ≥ 0 .
If aij ≥ 0, for all i, then x j = 0.
Proof: Now, eliminating x j from the given system using Fourier elimination
method, we have the following system
a11x1 + a12 x2 + ... + a1 j −1x j −1 + a1 j +1x j +1 + ... + a1n xn ≤ b1
a21x1 + a22 x2 + ...... + a2 j −1x j −1 + a2 j +1x j +1 + ... + a2n xn ≤ b2
M . (II)
am1x1 + am 2 x2 + ...... + amj −1x j −1 + amj +1x j +1 + ... + amn xn ≤ bm
x1, x2 ,..., x j −1, x j +1,..., xn ≥ 0 .
Fourier method for solving transportation problems 1389

From the system (I) and the system (II), we observe that there is no role of the
variable x j , that is , x j is an inactive variable. Therefore, x j = 0.
Hence the theorem.

For easy to understand and to compute quickly, we use a new type of

computation table called Fourier elimination table. The Fourier elimination table
for a linear system AX ≤ B is a rectangular arrangement which contains variables
X and its coefficients, the values of B and equation numbers.

The Fourier elimination table for the following linear system of inequalities
a1 x1 + a2 x2 + a3 x3 ≤ d1 (n3 )
b1 x1 + b2 x2 + b3 x3 ≤ d 2 (n4 )
is given below :
x1 x2 x3 B Equation number
a1 a2 a3 d1 (n3 )
b1 b2 b3 d2 (n4 )

3 Fourier transportation method

Now, we propose a new method called Fourier transportation method for
finding an optimal solution of a transportation problem with mixed constraints.

The Fourier transportation method proceeds as follows.

Step 1: Write the given transportation problem with mixed constraints in the form
of a pure integer linear programming problem.
Step 2: Convert the pure integer linear programming problem obtained from the
Step 1.into a maximization problem.
Step 3: Write the maximization problem obtained from the Step 2. having only one
type of the inequality ‘ ≤ ’ by eliminating one variable in an equality
constraint in the given problem.
Step 4: Write the equivalent pure integer linear programming problem to the
modified maximization problem obtained from the Step 3. and then, construct the
Fourier elimination table for the equivalent problem..
Step 5: Select and remove a variable from the Fourier elimination table by the
Modified Fourier’s elimination method [6].
Step 6: Form a reduced Fourier elimination table from the Fourier Elimination
table after deleting the true statement(s) and redundant constraints (if any) and
also, using the Theorem 2.
1390 P. Pandian and G. Natarajan

Step 7: Repeat the Step 5 to the Step 6 until all variables of xij ' s are eliminated
except the objective function variable w .
Step 8: Find the least upper bound of all maximum possible values of w . Say d o .
The optimum solution w is d o . The values of all xij ' s are computed using
backward substitution method and basic algebraic method. Say xijo , i = 1,2,…,m
and j = 1,2,…,n.
Step 9: The optimal solution for the transportation problem with mixed constraints
is xij = xijo , i = 1,2,…,m and j = 1,2,…, n and the minimum value of z , z o = d o .

4 Numerical Examples
The proposed method is illustrated by the following examples.

Example 1. Consider the following transportation problem with mixed constraints.

1 2 3 Supply
1 10 1 7 =5
2 5 7 1 ≥6
3 8 9 2 ≤9
Demand =8 ≥ 10 ≤5
Now, the pure integer linear programming problem of the given problem is given
below.
(P) Minimize z = 10 x11 + x12 + 7 x13 + 5 x21 + 7 x22 + x23 + 8 x31 + 9 x32 + 2 x33
subject to x11 + x12 + x13 = 5 (8)
x21 + x22 + x23 ≥ 6 (9)
x31 + x32 + x33 ≤ 9 (10)
x11 + x21 + x31 = 8 (11)
x12 + x22 + x32 ≥ 10 (12)
x13 + x23 + x33 ≤ 5 (13)
x11, x12 , x13 , x21, x22 , x23 , x31, x32 , x33 ≥ 0 and integers. (14)

Now, the maximization problem corresponding to (P) is given below.

(XP) Maximize w = −10 x11 − x12 − 7 x13 − 5 x21 − 7 x22 − x23 − 8 x31 − 9 x32 − 2 x33
subject to (8) to (14) are satisfied where w = − z .

Now, using ( 8) and (11) , (XP) can be written as follows.

(XP1) Maximize w = −45 − 4 x11 − 6 x13 − 7 x22 − x23 − 3x31 − 9 x32 − 2 x33
subject to x11 + x13 ≤ 5 ; x31 + x32 + x33 ≤ 9 ; x13 + x23 + x33 ≤ 5 ;
x11 + x31 ≤ 8 ; x11 + x13 − x22 − x32 ≤ −5 ; x11 + x31 − x22 − x23 ≤ 2 ;
Fourier method for solving transportation problems 1391

x11 , x13 , x22 , x23 , x31 , x32 and x33 ≥ 0 and integers.

Now, the equivalent problem to the problem (XP1) is given below.

(EP) Maximize w
subject to 4 x11 + 6 x13 + 7 x22 + x23 + 3x31 + 9 x32 + 2 x33 + w ≤ −45;
x31 + x32 + x33 ≤ 9 ; x13 + x23 + x33 ≤ 5 ; x11 + x13 − x22 − x32 ≤ −5 ;
x11 + x31 ≤ 8 ; x11 + x13 ≤ 5 ; x11 + x31 − x22 − x23 ≤ 2 ;
x11 , x13 , x22 , x23 , x31 , x32 and x33 ≥ 0 and integers.

By the Theorem 2., we have x11 = 0 , x13 = 0, x31 = 0, x23 = 0 and x33 = 0 and
therefore, problem (EP) reduces to

(EP1) Maximize w
subject to 7 x22 + 9 x32 + w ≤ −45; x32 ≤ 9 ; − x22 − x32 ≤ −5 ;
x22 and x32 ≥ 0 and integers.
Now, we form the starting Fourier elimination table for the problem (EP1).

x22 x32 w B Equation Number

7 9 1 -45 (15)
0 1 0 9 (16)
-1 -1 0 -5 (17)
-1 0 0 0 (18)
0 -1 0 0 (19)
Now, eliminating x22 from the table using Modified Fourier elimination method,
we have the following Fourier elimination table after removing the true statements
and the redundant constraints.

x32 w B Equation Number

1 0 9 (16)
2 1 -80 (20) [(15)+7(17)]
-1 0 0 (19)

By the Theorem 2., x32 = 0. Therefore, we remove the column of x32 from the
above table and also, remove the true statements and the redundant of the result
table. Then, we have the following table.

w B Equation Number
1 -80 (20)
The least upper bound of the maximum possible values of w is - 80. Therefore, the
minimum transportation cost , z = 80.
Now, using backward substitution method and basic algebraic method , we have
x11 = 0 , x12 = 5 , x13 = 0 , x21 = 8 , x22 = 5 , x23 = 0 , x31 = 0 , x32 = 0 , x33 = 0 .
1392 P. Pandian and G. Natarajan

Therefore, the optimal solution is x11 = 0 , x12 = 5 , x13 = 0 , x21 = 8 , x22 = 5 ,

x23 = 0 , x31 = 0 , x32 = 0 , x33 = 0 and z = 80.

Example 2. Consider the following balanced transportation problem.

01 02 03 Supply
01 4 1 7 80
02 3 2 2 20
03 5 3 4 35
Demand 60 40 35

Now, the purely integer linear programming problem for the above balanced
transportation problem is given below.
(P) Minimize z = 4 x11 + x12 + 7 x13 + 3x 21 + 2 x 22 + 2 x 23 + 5 x31 + 3x32 + 4 x33
subject to x11 + x12 + x13 = 80 (21)
x 21 + x 22 + x 23 = 20 (22)
x31 + x32 + x33 = 35 (23)
x11 + x 21 + x31 = 60 (24)
x12 + x 22 + x32 = 40 (25)
x13 + x 23 + x33 = 35 (26)
x11 , x12 , x13 , x 21 , x 22 , x 23 , x31 , x32 , x33 ≥ 0 and integers. (27)

Now, the maximization problem corresponding to (P) is given below.

(EP) Maximize w = −4 x11 − x12 − 7 x13 − 3 x 21 − 2 x 22 − 2 x 23 − 5 x31 − 3x32 − 4 x33
subject to (21) to (27) are satisfied where w = − z .

Now, using (21) to (27) , (EP) can be written as follows.

(EP1) Maximize w = −540 − 2 x 22 + 4 x 23 − x32 + 4 x33
subject to
x 22 + x 23 ≤ 20 ; x32 + x33 ≤ 35 ; x 22 + x32 ≤ 40 ;
x 23 + x33 ≤ 35 and x 22 , x 23 , x32 , x33 ≥ 0 and integers .

The equivalent problem to the problem (EP1) is given below.

(EP2) Maximize w
subject to
2 x 22 − 4 x 23 + x32 − 4 x33 + w ≤ −540 ; x22 + x23 ≤ 20 ; x32 + x33 ≤ 35 ;
x 22 + x32 ≤ 40 ; x 23 + x33 ≤ 35 and x 22 , x 23 , x32 , x33 ≥ 0 and integers .

By the Theorem 2., we have, x 22 = 0, x32 = 0 and therefore, the problem (EP2)
reduces to
(EP3) Maximize w
Fourier method for solving transportation problems 1393

subject to
− 4 x 23 − 4 x33 + w ≤ −540 ; x 23 ≤ 20 ; x33 ≤ 35 ;
x 23 + x33 ≤ 35 and x23 , x33 ≥ 0 and integers.

Now, we form the starting Fourier elimination table for the problem (EP3).

x 23 x33 w B Equation Number

-4 -4 1 -540 (28)
1 0 0 20 (29)
0 1 0 35 (30)
1 1 0 35 (31)
-1 0 0 0 (32)
0 -1 0 0 (33)

Now, eliminating x 23 from the first iteration table using Modified Fourier
elimination method, we have the following Fourier elimination table after
removing the true statements and the redundant constraints.

x33 w B Equation Number

0 1 -400 (34) [4(31)+(28)]
-4 1 -460 (35) [4(27)+(28)]
1 0 35 (30)
-1 0 0 (33)

Now, eliminating x33 from the first iteration table using Modified Fourier
elimination method, we have the following Fourier elimination table after
removing the true statements and the redundant constraints.

w B Equation Number
1 -400 (34)

The least upper bound of the maximum possible value of w = −400 . Therefore,
total minimum transportation cost z = 400.

Now, using the backward substitution method and the basic algebraic method, we
obtain the values of variables: x11 = 40 ; x12 = 40 ; x13 = 0 ; x 21 = 0 ; x 22 = 0 ,
x 23 = 20 ; x31 = 20 ; x32 = 0 and x33 = 15 .
Therefore, the optimal solution is x11 = 40 , x12 = 40 , x13 = 0 , x 21 = 0 ,
x 22 = 0 , x 23 = 20 , x31 = 20 , x32 = 0 and x33 = 15 and z = 400.
1394 P. Pandian and G. Natarajan

5 Conclusion
We have provided a method called the Fourier transportation method to find
an optimal solution for transportation problems with mixed constraints. This method
is very easy to understand and apply. So, the Fourier transportation method can
serve managers by providing the optimal solution to a variety of distribution
problems with mixed constraints. In near future, we extend this method to
generalized transportation problems and multiobjective transportation problems.

REFERENCES

[1] H. Arsham , Postoptimality analyses of the transportation problem, Journal of

the Operational Research Society, 43 (1992), 121–139.

[2] M.S.Bazaraa , J.J.Jaruis and H.D. Sherali, Linear programming and Network
Flows, John Wiley and Sons, New York, 1997.

[3] A. Gupta, S. Khanna and M. Puri, Paradoxical situations in transportation

problems, Cahiers du Centre d’Etudes de Recherche Operationnell, 34 (1992),
37–49.

[4] S.K.Gupta, Linear programming and network model, Affiliated East-West

Press (PVT), New Delhi, 1989.

[5] Y. Intrator and B. Lev, Methods for identification of vanishing variables in

transportation problems and its possible applications, Computers and
Operations Research, 3 (1976), 259–304.

[6] P.Kanniappan and K.Thangavel, Modified Fourier’s method for solving

linear programming problems, OPSEARCH, 35(1998), 45-56.

[7] H.S. Kasana and K.D. Kumar, Introductory Operations Research Theory
and Applications, Springer International Edition, New Delhi, 2005.

[8] B. Lev, A non iterative algorithm for tridiagonal transportation problems and
its generalization, Journal of Operations Research Society of America,
20(1972), 109–125.

[9] B. Lev and Y. Intrator, Applications of vanishing variables methods to

special structured transportation problems, Computers and Operations
Research, 4(1977), 21–26.
 Fourier method for solving transportation problems 1395

[10] Veena Adlakha, Krzysztof Kowalski and Benjamin Lev, Solving

transportation problems with mixed constraints, International Journal of
Management Science and Engineering Management, 1 (2006), 47-52.

[11] H.P.Williams, Fourier’s method of linear Programming and its Dual,

American Mathematical Monthly, 93 (1986), 681-695.

[12] W.L.Winston, Introduction to Mathematical Programming Applications and

Algorithms, Duxbury Press, California, 1991.

Received: January, 2010

View publication stats