CHAPTER 3: TRANSPORTATION AND ASSIGNMENT PROBLEMS

Transportation Model
3.1. INTRODUCTION
In operations Research Linear programming is one of the model in mathematical programming, which is
very broad and vast. Mathematical programming includes many more optimization models known as Non -
linear Programming, Stochastic programming, Integer Programming and Dynamic Programming
- each one of them is an efficient optimization technique to solve the problem with a specific structure,
which depends on the assumptions made in formulating the model. We can remember that the general linear
programming model is based on the assumptions:

(a) Certainty
The resources available and the requirement of resources by competing candidates, the profit
coefficients of each variable are assumed to remain unchanged and they are certain in nature.

(b) Linearity
The objective function and structural constraints are assumed to be linear.

(c) Divisibility
All variables are assumed to be continuous; hence they can assume integer or fractional values.

(d) Single stage

The model is static and constrained to one decision only. And planning period is assumed to be
fixed.

(e) Non-negativity
A non-negativity constraint exists in the problem, so that the values of all variables are to be ³ 0,
i.e. the lower limit is zero and the upper limit may be any positive number.

(f) Fixed technology

Production requirements are assumed to be fixed during the planning period.
(g) Constant profit or cost per unit
Regardless of the production schedules profit or cost remain constant.
Now let us examine the applicability of linear programming model for transportation and
assignment models.

3.2. TRANSPORTATION MODEL

The transportation model deals with a special class of linear programming problem in which the objective is
to transport a homogeneous commodity from various origins or factories to different destinations
or markets at a total minimum cost.
To understand the problem more clearly, let us take an example and discuss the rationale of
transportation problem. Three factories A, B and C manufactures sugar and are located in different regions.
Factory A manufactures, b1 tons of sugar per year and B manufactures b2 tons of sugar per year and C
manufactures b3 tons of sugar. The sugar is required by four markets W, X, Y and Z. The requirement of the
 four markets is as follows: Demand for sugar in Markets W, X, Yand Z is d1, d2, d3 and d4 tons respectively.
The transportation cost of one ton of sugar from each factory to market is given in the matrix below. The
objective is to transport sugar from factories to the markets at a minimum total transportation cost.

Transportation cost per ton in Rs. Availability in tons

Markets
W X Y Z
A c 11 c 12 c 13 c 14 b1
Factories B c 21 c 22 c 23 c 24 b2
C c 31 c 32 c 33 c 34 b3
Demand in d1 d2 d3 d4 Σ b j / Σd j
Tons.

For the data given above, the mathematical model will be:
Minimize Z = c11 x11 + c12 x12 + c13 x13 + c14 x14 + c21 x21 + c22 x22 + c23 x23 + c24 x24 +
c31 x31 + c32 x32 + c33 x33 + c34 x34 subject to a condition: OBJECTIVE FUNCTION.
a11 x11 +a12 x12 + a13 x13 + a14 x14 £ b1 (because the sum must be less than or equal to the
available capacity)
+ a23 x23 + a24 x24 £ b2
a31 x31 + a32 x32 + a33 x33 + a34 x34 £ b3 MIXED STRUCTURAL CONSTRAINTS.

a11 x11 + a21 x21 + a31 x31 ³ d1

(because the sum must be greater than or equal to the demand
a12 x12 + a22 x22 + a32 x32 ³ d2 of the market. We cannot send less than what is required)
a13 x13 + a23 x23 + a33 x33 ³ d3
a14 x14 + a24 x24 + a34 x34 ³ d4 and
All xij and xji are ³ 0 where i = 1,2,3 and j = 1,2,3,4. (This is because we cannot
supply negative elements). NON-NEGATIVITY
CONSTRAINT.
 The above problem has got the following properties:
1. It has an objective function.
2. It has structural constraints.
3. It has a non-negativity constraint.
4. The relationship between the variables and the constraints are linear.
We know very well that these are the properties of a linear programming problem. Hence the
transportation model is also a linear programming problem. But a special type of linear programming
problem.
Once we say that the problem has got the characteristics of linear programming model, and then we
can solve it by simplex method. Hence we can solve the transportation problem by using the simplex
method. As we see in the above given transportation model, the structural constraints are of mixed type.
That is some of them are of £ type and some of them are of ³ type. When we start solving the transportation
problem by simplex method, it takes more time and laborious. Hence we use transportation algorithm or
transportation method to solve the problem. Before we discuss the transportation algorithm, let us see
how a general model for transportation problem appears. The general problem will have 'm' rows and 'n'
columns i.e., m × n matrix.
n m

Minimize Z = å åc ij x j
j =1 i =1
s.t. where i = 1 to m and j = 1 to n.

å aij xij £ bi where i = 1 to m and j = 1 to n

i =1

åa
j =1
ij x ji ³ d j
where i = 1 to m and j = 1 to n
3.3. COMPARISON BETWEEN TRANSPORTATION MODEL AND GENERAL LINEAR
PROGRAMMING MODEL

Similarities
1. Both have objective function.
2. Both have linear objective function.
3. Both have non - negativity constraints.
4. Both can be solved by simplex method. In transportation model it is laborious.
5. A general linear programming problem can be reduced to a transportation problem if (a) the aij's
(coefficients of the structural variables in the constraints) are restricted to the values 0 and/or 1
and (b) There exists homogeneity of units among the constraints.

Differences
1. Transportation model is basically a minimization model; where as general linear programming
model may be of maximization type or minimization type.

2. The resources, for which, the structural constraints are built up is homogeneous in
transportation model; where as in general linear programming model they are different. That is
one of the constraint may relate to machine hours and next one may relate to man-hours etc. In
transportation problem, all the constraints are related to one particular resource or commodity,
which is manufactured by the factories and demanded by the market points.
3. The transportation problem is solved by transportation algorithm; where as the general linear
programming problem is solved by simplex method.
4. The values of structural coefficients (i.e. xij) are not restricted to any value in general linear
programming model, where as it is restricted to values either 0 or 1 in transportation problem. Say
for example:
Let one of the constraints in general linear programming model is: 2x –3y +10z £ 20. Here the
coefficients of structural variables x, y and z may negative numbers or positive numbers of zeros.
Where as in transportation model, say for example x11 + x12 + x13 + x14 = bi = 20. Suppose the
value of variables x11, and x14 are 10 each, then 10 + 0. x12 + 0. x13 + 10 = 20. Hence the
coefficients of x11 and x14 are 1 and that of x12 and x13 are zero.

3.4. APPROACH TO SOLUTION TO A TRANSPORTATION PROBLEM BY USING

TRANSPORTATION ALGORITHM
The steps used in getting a solution to a transportation problem is given below:

3.4.1. Initial Basic Feasible Solution

Step 1. Balancing the given problem. Balancing means check whether sum of availability constraints
must be equals to sum of requirement constraints. That is Sbi = Sd j . Once they are equal,
go to step two. If not by opening a Dummy row or Dummy column balance the problem. The
cost coefficients of dummy cells are zero. If Sb is greater than Sd j , then open a dummy
i
 column, whose requirement constraint is equals to Sbi – Sd j and the cost coefficient of
the cells are zeros. In case if Sd j is greater than Sbi , then open a dummy row, whose
availability constraint will be equals to Sd j – Sbi and the cost coefficient of the cells are
zeros. Once the balancing is over, then go to second step. Remember while solving general linear
programming problem to convert an inequality into an equation, we add (for maximization problem)
a slack variable. In transportation problem, the dummy row or dummy column, exactly similar to
a slack variable.
Step II. A .Basic feasible solution can be obtained by three methods, they are
(a) North - west corner method.
(b) Least - cost cell method. (Or Inspection method Or Matrix minimum - row minimum
- column minimum method)
(c) Vogel's Approximation Method, generally known as VAM.
After getting the basic feasible solution (b.f.s.) give optimality test to check whether the
solution is optimal or not.
There are two methods of giving optimality test: (a)
Stepping Stone Method.
(b) Modified Distribution Method, generally known as MODI method
3.4.2.Properties of a Basic feasible Solution
1. The allocation made must satisfy the rim requirements, i.e., it must satisfy availability
constraints and requirement constraints.
2. It should satisfy non negativity constraint.
3. Total number of allocations must be equal to (m + n – 1), where 'm' is the number of rows and 'n'
is the number of columns. Consider a value of m = 4 and n = 3, i.e. 4 × 3 matrix. This will have
four constraints of £ type and three constraints of ³ type. Totally it will have 4 +
3 (i.e m + n) inequalities. If we consider them as equations, for solution purpose, we will have 7
equations. In case, if we use simplex method to solve the problem, only six rather than seen
structural constraints need to be specified. In view of the fact that the sum of the origin capacities
(availability constraint) equals to the destination requirements (requirement
constraint) i.e., Sbi = S dj, any solution satisfying six of the seven constraints will
automatically satisfy the last constraint. In general, therefore, if there are 'm' rows and 'n'
columns, in a given transportation problem, we can state the problem completely with m + n – 1
equations. This means that one of the rows of the simplex tableau represents a redundant
constraint and, hence, can be deleted. This also means that a basic feasible solution of a
transportation problem has only m + n – 1 positive components. If Sbi = S dj, it is always
possible to get a basic feasible solution by North-west corner method, Least Cost cell method
or by VAM.

3.4.3. Basic Feasible Solution by North - West corner Method

Let us take a numerical example and discuss the process of getting basic feasible solution by
various methods.
Example 4.1. Four factories, A, B, C and D produce sugar and the capacity of each factory is given below:
Factory A produces 10 tons of sugar and B produces 8 tons of sugar, C produces 5 tons of sugar and
that of D is 6 tons of sugar. The sugar has demand in three markets X, Y and Z. The demand of market X is
7 tons, that of market Y is 12 tons and the demand of market Z is 4 tons. The following matrix gives the
transportation cost of 1 ton of sugar from each factory to the destinations. Find the Optimal Solution for
least cost transportation cost.

Cost in Rs. per ton (× 100)

Factories. Markets. Availability in tons.
X Y Z

A 4 3 2 10
B 5 6 1 8
C 6 4 3 5
D 3 5 4 6

Requirement in tons. 7 12 4 Σ b = 29, Σ d = 23

 Here Sb is greater than Sd hence we have to open a dummy column whose requirement constraint is
6, so that total of availability will be equal to the total demand. Now let get the basic feasible solution by
three different methods and see the advantages and disadvantages of these methods. After this let us give
optimality test for the obtained basic feasible solutions.

a) North- west corner method

(i) Balance the problem. That is see whether Sbi = Sd j . If not open a dummy column or
dummy row as the case may be and balance the problem.
(ii) Start from the left hand side top corner or cell and make allocations depending on the
availability and requirement constraint. If the availability constraint is less than the requirement
constraint, then for that cell make allocation in units which is equal to the availability constraint. In
general, verify which is the smallest among the availability and requirement and allocate the
smallest one to the cell under question. Then proceed allocating either sidewise or down- ward to
satisfy the rim requirement. Continue this until all the allocations are over.
(iii) Once all the allocations are over, i.e., both rim requirement (column and row i.e., availability and
requirement constraints) are satisfied, write allocations and calculate the cost of
transportation.
Solution by North-west corner method:

For cell AX the availability constraint is 10 and the requirement constraint is 7. Hence 7 is smaller than
10, allocate 7 to cell AX. Next 10 – 7 = 3, this is allocated to cell AY to satisfy availability requirement.
Proceed in the same way to complete the allocations. Then count the allocations, if it is equals to m + n – 1,
then the solution is basic feasible solution. The solution, we got have 7 allocations which is = 4 + 4 – 1 = 7.
Hence the solution is basic feasible solution.
 Now allocations are:
From To Units in tons Cost in Rs.

A X 7 7 × 4 = 28

A Y 3 3 × 3 = 09

B Y 8 8 × 6 = 48

C Y 1 1 × 4 = 04

C Z 4 4 × 3 = 12

D Z 1 1 × 4 = 04

D DUMMY 5 5 × 0 = 00

Total in Rs. 105

3.4.4. Solution by Least cost cell (or inspection) Method: (Matrix Minimum method)
(i) Identify the lowest cost cell in the given matrix. In this particular example it is = 0. Four cells of
dummy column are having zero. When more than one cell has the same cost, then both the cells are
competing for allocation. This situation in transportation problem is known as tie. To break the tie, select
any one cell of your choice for allocation. Make allocations to this cell either to satisfy availability
constraint or requirement constraint. Once one of these is satisfied, then mark crosses (×) in all the cells in
the row or column which ever has completely allocated. Next search for lowest cost cell. In the given
problem it is cell BZ which is having cost of Re.1/- Make allocations for this cell in similar manner and
mark crosses to the cells in row or column which has allocated completely. Proceed this way until all
allocations are made. Then write allocations and find the cost of transportation. As the total number of
allocations are 7 which is equals to 4 + 4 – 1 = 7, the solution is basic feasible solution.
 (Note: The numbers under and side of rim requirements shows the sequence of allocation and the units
remaining after allocation)
Allocations are:

From To Units in tons Cost in Rs.

A Y 8 8 × 3 = 24

A Z 2 2 × 2 = 04

B Z 3 3 × 1 = 03

B DUMMY 5 5 × 0 = 00

C X 1 1 × 6 = 06

C Y 4 4 × 4 = 16

D X 6 6 × 3 = 18

Total in Rs. 71

3.4.5. Solution by Vogel’s Approximation Method: (Opportunity cost method)

(i) In this method, we use concept of opportunity cost. Opportunity cost is the penalty for not taking
correct decision. To find the row opportunity cost in the given matrix deduct the smallest
element in the row from the next highest element. Similarly to calculate the column
opportunity cost, deduct smallest element in the column from the next highest element. Write
row opportunity costs of each row just by the side of availability constraint and similarly write
the column opportunity cost of each column just below the requirement constraints. These are
known as penalty column and penalty row.
The rationale in deducting the smallest element form the next highest element is:
Let us say the smallest element is 3 and the next highest element is 6. If we transport one unit
 through the cell having cost Rs.3/-, the cost of transportation per unit will be Rs. 3/-. Instead we
transport through the cell having cost of Rs.6/-, then the cost of transportation will be Rs.6/- per
unit. That is for not taking correct decision; we are spending Rs.3/- more (Rs.6 – Rs.3 = Rs.3/-).
This is the penalty for not taking correct decision and hence the opportunity cost. This is the
lowest opportunity cost in that particular row or column as we are deducting the smallest
element form the next highest element.
Note: If the smallest element is three and the row or column having one more three, then we
have to take next highest element as three and not any other element. Then the
opportunity cost will be zero. In general, if the row has two elements of the same magnitude
as the smallest element then the opportunity cost of that row or column is zero.
(ii) Write row opportunity costs and column opportunity costs as described above.
(iii) Identify the highest opportunity cost among all the opportunity costs and write a tick (Ö )
mark at that element.
(iv) If there are two or more of the opportunity costs which of same magnitude, then select any one of
them, to break the tie. While doing so, see that both availability constraint and requirement
constraint are simultaneously satisfied. If this happens, we may not get basic feasible solution i.e
solution with m + n – 1 allocations. As far as possible see that both are not satisfied
simultaneously. In case if inevitable, proceed with allocations. We may not get a solution with, m
+ n – 1 allocations. For this we can allocate a small element epsilon (Î) to any one of the empty
cells. This situation in transportation problem is known as degeneracy. (This will be discussed
once again when we discuss about optimal solution).
In transportation matrix, all the cells, which have allocation, are known as loaded cells and those,
which have no allocation, are known as empty cells.
(Note: All the allocations shown in matrix 1 to 6 are tabulated in the matrix given below:)
 Consider matrix (1), showing cost of transportation and availability and requirement constraints. In
the first row of the matrix, the lowest cost element is 0, for the cell A-Dummy and next highest element is
2, for the cell AZ. The difference is 2 – 0 = 2. The meaning of this is, if we transport the load through the
cell A-Dummy, whose cost element is 0, the cost of transportation will be = Rs.0/- for
each unit transported. Instead, if we transport the load through the cell, AZ whose cost element is Rs.
2/- the transportation cost is = Rs.2/- for each unit we transport. This means to say if we take decision to
send the goods through the cell AZ, whose cost element is Rs.2/- then the management is going to loose Rs.
2/- for every unit it transport through AZ. Suppose, if the management decide to send load through the cell
AX, Whose cost element is Rs.4/-, then the penalty or the opportunity cost is Rs.4/-. We write the minimum
opportunity cost of the row outside the matrix. Here it is shown in brackets. Similarly, we find the column
opportunity costs for each column and write at the bottom of each corresponding row (in brackets). After
writing all the opportunity costs, then we select the highest among them. In the given matrix it is Rs.3/- for
the rows D and C. This situation is known as tie. When tie exists, select any of the rows of your choice.
At present, let us select the row D. Now in that row select the lowest cost cell for allocation. This is
because; our objective is to minimize the transportation cost. For the problem, it is D-dummy, whose cost
is zero. For this cell examine what is available and what is required? Availability is 6 tons and requirement
is 5 tons. Hence allocate 5 tons to this cell and cancel the dummy row from the problem. Now the matrix is
reduced to 3 × 4. Continue the above procedure and for every allocation the matrix goes on reducing, finally
we get all allocations are over. Once the allocations are over, count them, if there are m + n – 1 allocations,
then the solution is basic feasible solution. Otherwise, the degeneracy occurs in the problem. To solve
degeneracy, we have to add epsilon (Î), a small element to one of the empty cells. This we shall discuss,
when we come to discuss optimal solution. Now for the problem the allocations are:

From To Load Cost in Rs.

A X 3 3 × 4 = 12

A Y 7 7 × 3 = 21

B X 3 3 × 5 = 15

B Z 5 5 × 1 = 05

C Y 5 5 × 4 = 20

D X 1 1 × 3 = 03

D DUMMY 5 5 × 0 = 00

Total Rs. 76

Now let us compare the three methods of getting basic feasible solution:
 Inspection or least cost cell
North – west corner method. Vogel’s Approximation Method.
method
1. The allocation is made The allocations are made The allocations are made
from the left hand side top depending on the cost of the depending on the opportunity
corner irrespective of the cost cell. Lowest cost is first cost of the cell.
of the cell. selected and then next highest
etc.
2. As no consideration is As the cost of the cell is As the allocations are made
given to the cost of the cell, considered while making depending on the opportunity
naturally the total allocations, the total cost of cost of the cell, the basic feasible
transportation cost will be transportation will be solution obtained will be very
higher than the other comparatively less. nearer to optimal solution.
methods.
3. It takes less time. This The basic feasible solution, we It takes more time for getting
method is suitable to get get will be very nearer to basic Feasible solution. But the
basic feasible solution optimal solution. It takes more solution we get will be very
quickly. time than northwest coroner nearer to Optimal solution.
method.
4. When basic feasible When optimal solution is VAM and MODI is the best
solution alone is asked, it is asked, better to go for option to get optimal solution.
better to go for northwest inspection method for basic
corner method. feasible solution and MODI for
optimal solution.

In the problem given, the total cost of transportation for Northwest corner method is Rs. 101/-. The
total cost of transportation for Inspection method is Rs. 71/- and that of VAM is Rs. 76/-. The total cost got
by inspection method appears to be less. That of Northwest coroner method is highest. The cost got by
VAM is in between.
Now let us discuss the method of getting optimal solution or methods of giving optimality test for
basic feasible solution.

3.4.6. Optimality Test: (Approach to Optimal Solution)

Once, we get the basic feasible solution for a transportation problem, the next duty is to test whether the
solution got is an optimal one or not? This can be done by two methods. (a) By Stepping Stone Method, and
(b) By Modified Distribution Method, or MODI method.
(a) Stepping stone method of optimality test
To give an optimality test to the solution obtained, we have to find the opportunity cost of empty cells.
As the transportation problem involves decision making under certainty, we know that an optimal solution
must not incur any positive opportunity cost. Thus, we have to determine whether any positive opportunity
cost is associated with a given progarmme, i.e., for empty cells. Once the opportunity cost of all empty
cells are negative, the solution is said to be optimal. In case any one cell has got positive opportunity
cost, then the solution is to be modified. The Stepping stone method is used for finding the opportunity costs
of empty cells. Every empty cell is to be evaluated for its opportunity cost. To do this the methodology is:
1. Put a small ‘+’ mark in the empty cell.
 2. Starting from that cell draw a loop moving horizontally and vertically from loaded cell to loaded
cell. Remember, there should not be any diagonal movement. We have to take turn only at
loaded cells and move to vertically downward or upward or horizontally to reach another loaded
cell. In between, if we have a loaded cell, where we cannot take a turn, ignore that and proceed to
next loaded cell in that row or column.
3. After completing the loop, mark minus (–) and plus (+) signs alternatively.
4. Identify the lowest load in the cells marked with negative sign.
5. This number is to be added to the cells where plus sign is marked and subtract from the load of
the cell where negative sign is marked.
6. Do not alter the loaded cells, which are not in the loop.
7. The process of adding and subtracting at each turn or corner is necessary to see that rim
requirements are satisfied.
8. Construct a table of empty cells and work out the cost change for a shift of load from loaded cell
to loaded cell.
9. If the cost change is positive, it means that if we include the evaluated cell in the programme, the
cost will increase. If the cost change is negative, the total cost will decrease, by including the
evaluated cell in the programme.
10. The negative of cost change is the opportunity cost. Hence, in the optimal solution of
transportation problem empty cells should not have positive opportunity cost.
11. Once all the empty cells have negative opportunity cost, the solution is said to be optimal. One
of the drawbacks of stepping stone method is that we have to write a loop for every empty
cell. Hence it is tedious and time consuming. Hence, for optimality test we use MODI method rather
than the stepping stone method.
Let us take the basic feasible solution we got by Vogel's Approximation method and give optimality
test to it by stepping stone method.
Basic Feasible Solution obtained by VAM:
 Table showing the cost change and opportunity costs of empty cells:
Table I.

S.No. Empty Evalution Opportunity cost

Cost change in Rs.
Cell Loop formation -(Cost change)

1. AZ +AZ – AX + BX – BZ +2 – 4 + 5 – 1 = + 2 –2
2 A Dummy + A DUMMY – AX + BX – B DUMMY +0 – 4 + 3 – 0 = – 1 +1
3 BY + BY – AY + AX – BX +6 – 3 + 4 – 5 = +2 -2
4 B DUMMY + B DUMMY – BX + DX – D DUMMY +0 – 5 +3 – 0 = –2 +2
5 CX +CX – CY + AX – AY 6 – 4 + 3 – 4 = +1 –1
6 CZ +CZ – BZ + BX –AX + AY – CY +2 –1 +5 – 4 +5 – 4 =+1 –1
7 C DUMMY + C DUMMY – D DUMMY + DX – + 0 – 0 +3 – 4 +3 – 4 = +2
AX + AY – CY –2
8 DY +DY – DX + AX – AY +5 – 3 +4 – 3 = +3 –3
9 DZ +DZ – DX +BX – BZ +4 – 3 + 5 – 1 = +5 –5

In the table 1 cells A DUMMY, B DUMMY, C DUMMY are the cells which are having positive
opportunity cost. Between these two cells B DUMMY and C DUMMY are the cells, which are having
higher opportunity cost i.e Rs. 2/ - each. Let us select any one of them to include in the improvement of the
present programme. Let us select C DUMMY.
 Table II.

S.No. Empty Evalution Opportunity

Cell Loop formation Cost change in Rs. Cost

1 AX +AX –DX + D DUMMY – C DUMMY +4–3+0–0+4–3=+2 –2

+ CY – AY
2 AX AZ – AY + CY – C DUMMY + +2–3+4–0+0–3+ –4
D DUMMY – DX+ BX – BZ 3–0=+4
3 ADUMMY + A DUMMY – AY + DX – +0–4+3–0=–1 +1
D DUMMY
4 BY +BY – BX + DX – D DUMMY + + 6 –5 + 3 –0 + 0 – 4 = 0 0
C DUMMY – CY
5 B DUMMY + B DUMMY – BX + DX – D DUMMY + 0 – 5 + 3 – 0 = –2 +2
6 CX + CX – DX + D DUMMY – C DUMMY + 6 –3 + 0 – 0 = +3 –3
7 CZ + CZ – C DUMMY + D DUMMY +2–0+0–3+5–1=+3 –3
– DX + BX – BZ
8 DY DY – CY + C DUMMY – D DUMMY +5–4+0–0=1 –1

9 DZ + DZ – DX + BX – BZ + 4 – 3 +5 – 1 = + 5 –5

Cells A DUMMY and B DUMMY are having positive opportunity costs. The cell B DUMMY is
having higher opportunity cost. Hence let us include this cell in the next programme to improve the solution.
Table III.

S.No. Empty Evaluation Opportunity

Cell Loop formation Cost change in Rs. Cost

1 AX +AX – AY + CY – C DUMMY + +4 – 3 + 4 – 0 + 0 – 5 = 0 0
B DUMMY – BX
2 AZ + AZ – BZ + B DUMMY – C DUMMY +2 –1 + 0 – 0 + 4 – 3 =+ 2 –2
+ CX – AX
3 A DUMMY + A DUMMY – C DUMMY + CY – AY +0 – 0 + 4 – 3 = +1 –1
4 BY + BY – B DUMMY + C DUMMY – CY +6 – 0 + 0 – 4 = + 2 –2
5 CX + CX – BX + B DUMMY – C DUMMY +6 – 5 + 0 – 0 = +1 –1
6 CZ + CZ – BZ + B DUMMY – C DUMMY +2 – 1 + 0 – 0 = +1 –1
7 DY +DY – CY + C DUMMY – B DUMMY +5 – 4 + 0 – 0 + 5 – 3 = +3 –3
+ BX – DX
8 DZ + DZ – BZ + BX – DX +4 – 1 + 5 – 3 = +5 –5

9 D DUMMY + D DUMMY – DX + BX – B DUMMY + 0 – 3 + 5 – 0 = +2 –2

 All the empty cells have negative opportunity cost hence the solution is optimal. The allocations
are:

S.No Loaded cell Load Cost in Rs.

1 AY 10 10 × 3 = 30

2 BX 01 01 × 5 = 05

3 BZ 05 05 × 1 = 05

4 B DUMMY 02 02 × 0 = 00

5 CY 02 02 × 4 = 08

6 C DUMMY 03 03 × 0 = 00

7 DX 06 06 × 3 = 18

Total in Rs. 66

Total minimum transportation cost is Rs. 66/-

Optimal allocation.

(b) Modified Distribution Method of Optimality test

In stepping stone method, we have seen that to get the opportunity cost of empty cells, for every cell
we have to write a loop and evaluate the cell, which is a laborious process. In MODI (Modified DIstribution
method, we can get the opportunity costs of empty cells without writing the loop. After
getting the opportunity cost of all the cells, we have to select the cell with highest positive opportunity cost
for including it in the modified solution.
Steps in MODI method:
1. Select row element (ui) and Column element (vj) for each row and column, such that ui + vj
= the actual cost of loaded cell. In MODI method we can evaluate empty cells simultaneously
and get the opportunity cost of the cell by using the formula (ui + vj) – Cij, where Cij is the actual
cost of the cell.
2. In resource allocation problem (maximization or minimization method), we have seen that once
any variable becomes basis variable, i.e., the variable enters the programme; its opportunity
cost or net evaluation will be zero. Here, in transportation problem also, once any cell is loaded,
its opportunity cost will be zero. Now the opportunity cost is given by (ui
+ vj) – Cij, which is, equals to zero for a loaded cell.
i.e. (ui + vj) – Cij = 0 which means, (ui + vj) = Cij. Here (ui + vj) is known as implied cost of
the cell. For any loaded cell the implied cost is equals to actual cost of the cell as its
opportunity cost is zero. For any empty cell, (implied cost – actual cost) will give
opportunity cost.
3. How to select ui and vj? The answer is:
(a) Write arbitrarily any one of them against a row or against a column. The written ui or vj
may be any whole number i.e ui or vj may be £ or ³ to zero. By using the formula (ui + vj)
= Cij for a loaded cell, we can write the other row or column element. For example, if the
actual cost of the cell Cij = 5 and arbitrarily we have selected ui = 0, then vj is given by ui +
vj = 0 + vj = 5. Hence vj = –5. Like this, we can go from loaded cell to loaded cell and
complete entering of all ui s and vj s.
(b) Once we get all ui s and vj s, we can evaluate empty cells by using the formula (ui + vj)
– Actual cost of the cell = opportunity cost of the cell, and write the opportunity cost
of each empty cell at left hand bottom corner.
(c) Once the opportunity costs of all empty cells are negative, the solution is said
to be optimal. In case any cell is having the positive opportunity cost, the
programme is to be modified.
Remember the formula that IMPLIED COST OF A CELL = ui + vj Opportunity
cost of loaded cell is zero i.e (ui + vj) = Actual cost of the cell. Opportunity cost of
an empty cell = implied cost – actual cost of the cell = (ui
+ vj) – Cij
(d) In case of degeneracy, i.e. in a basic feasible solution, if the number of loaded cells
are not equals to m + n – 1, then we have to add a small element epsilon (Î), to any
empty cell to make the number of loaded cells equals to m + n –
1. While adding 'Î' we must be careful enough to see that this Î should not form a
closed loop when we draw horizontal and vertical lines from loaded cell to loaded
cell. In case the cell to which we have added Î forms a closed loop, then if we
cannot write all ui s and vj s.
Î is such a small element such that a + Î = a or a – Î = a and Î – Î = 0.
 Implied cost Actual cost Action

ui + vj > C ij A better programme can be designed by including this cell

in the solution.
ui + vj = C ij Indifferent; however, an alternative programme with same
total cost can be written by including this cell in the
programme.
ui + vj < C ij Do not include this cell in the programme.

Now let us take the basic feasible solution obtained by VAM method and apply MODI method of
optimality test.
Basic feasible solution got by VAM method.

The cell C DUMMY is having a positive opportunity cost. Hence we have to include this cell in the
programme. The solution has m + n – 1 allocations.
 The cell B DUMMY is having a positive opportunity cost. Thïs is to be included in the modified
programme.

As the opportunity cost of all empty cells are negative, the solution is optimal. The solution has m
+ n – 1 allocations. The
allocations are:
S.No Loaded Cell Load Cost in Rs.
1 AY 10 10 × 3 = 30
2. BX 01 01 × 5 = 05
3. BZ 05 05 × 1 = 05
4. B DUMMY 02 02 × 0 = 00
5. CY 02 02 × 4 = 08
6. C DUMMY 03 03 × 0 = 00
7. CX 06 06 × 3 = 18
Total Cost in Rs. 66

Readers can verify the optimal solution got by Stepping stone method and the MODI method they are
same. And they can also verify the opportunity costs of empty cells they are also same. This is the
advantage of using MODI method to give optimality test. Hence the combination of VAM and MODI can
be conveniently used to solve the transportation problem when optimal solution is asked.

3.4.7. Alternate Solutions

By principle, we know that the opportunity cost of a loaded cell or a problem variable is always equals
to zero. In case any empty cell of the optimal solution of a transportation problem got zero as the
opportunity cost, it should be understood that it is equivalent to a loaded cell. Hence by including that cell,
we can derive another solution, which will have same total opportunity cost, but different allocations. Once
one alternate solution exists, we can write any number of alternate solutions. The methodology is:
1. Let the Optimal solution is matrix A with one or more empty cells having zero as the opportunity
cost.
 2. By including the cell having zero as the opportunity cost, derive one more optimal solution, let it
be the matrix B.
3. The new matrix C is obtained by the formula: C = d A + (1– d) B, where 'd' is a positive
fraction less than 1.
It is better to take always d = 1/2, so that C = 1/2 A + 1/2 B.
Now we shall take the optimal solution of the problem above and write the alternate optimal
solutions.
Matrix A (First optimal Solution).

The cell AX, having zero opportunity cost is included in revised solution. The loop is:
+ AX – BX + B DUMMY – C DUMMY + CY – AY = + 4 – 5 + 0 – 0 + 4 – 3 = 0
Allocation:

S.No Loaded Cell Load Cost in Rs.

1. AX 01 01 × 4 = 04
2. AY 09 09 × 3 = 18
3. BZ 05 05 × 1 = 05
4. B Dummy 03 03 × 0 = 00
5. CY 03 03 × 4 = 12
6. C Dummy 02 02 × 0 = 00
7. DX 06 06 × 3 = 18
Total cost in Rs. 66
 Matrix B (First alternative solution):

Matrix C (Second alternate solution)

The total cost is 0.5 × 4 + 9.5 ×3 + 0.5 × 5 + 5 × 1 + 2.5 × 0 + 2.5 × 0 + 2.5 × 0 + 6 × 3 = Rs.
66/-
Once we get one alternate solution we can go on writing any number of alternate solutions until we
get the first optimal solution.