UNIT 10 AN INITIAL BASIC FEASIBLE SOLUTION OF THE TWANSPORATION PROBLEM

Structure

10.1
10.2

Introduction Objectives North-West Corner Method

'

f 0.3
10.4
10.5

Ma trix-Minima Met hod

Degenerate Solutions Summary

INTRODUCTION

Feasible SoPution of the T~~ansportstion ProbYem

In Unit 9, you were introduced to the TranspoiAtationProblem TP, its fornlulation as a EP? and its representation in a tabular dbrm. A few special characteristics of the TP were also described in sectioil 9.5 of IJnit 9 which give it a special structure. Being a special type of EPP, it could be solved by the Simplex Method. We already know that the initial basic feasible solution for this problem call be deter~ninedby adding an artificial variable to each constraint and then apglyiilg Phase 1 procedure. In this unit you will see that such an initial basic feasible solutioil can be obtained in a better way by exploiting the special structure of the problem. We shall discuss two methods to obtain an initial basic feasible solutioil of a TP namely North-West Corner Method and Matrix-Minima Method. Suppose you solve the TI? by sllnplex method. Observe that in ally TP with 'm' sources and '11' destiilations therc are m x n constraints in mil variables and each constraint is an equation. 'nus, the first step is to add an artificial variable to each of the constraints rcsulting in ~ n n + (m + 11) variables. Next stcp is to apply Phasc P Method to get an initial basic feasible solution to the problem. Thus, even for a moderate size problem where 'm' and 'hiare not very large, quite a large number of iterations are needed to obtain the initla1 basic fcasible solution. In this unit, we shall see that structure of the problem allows us to write dowt1 the initial basic feasible solutioil directly and cornlbrtably wathout adding artificial variable 2 1 1 d i~pplyiilgPhase 1.

Objectives
In this unit you should bc able to
Q)

North West Corner Method o r the Matrix-Millinla Metliod to obtain an initial basic feasible,solutions of a TP identify the degenerate solutio~is.

10.2 NORTH-WEST CORNER MErI'IIIOD

Let us come back to our Sewing Machi~lcs TP discussed in Unit 0. I n tabular , fomlitis:
Dl

D,

D,

il,

Special Linear
Prnernmmin~

We may first point out that this method does not take into account the costs of transportations Cii and the method is what the common sense dictates us. In this method, we start with extreme North West comer cell (1, 1) connecting source S, to destination Dl. In this route, 50 units are available at S, and 40 are needed by Dl. The minimum of 50 and 40 cell written as min [50,40] = 40 can be transported from source S, to the market Dl. Thus we decide to set x,, = 40, write it in the cell (1, 1) and encircle it as shown 40 resperesents the value of x,,. Now, demand of D,is exactly inet and therefore, it does not need any supply from S,. Therefore, we set x,, = 0. We do not write it in the table as it is a non basic variable. We can put a 'x'in the cell (2, 1). S lias slill 50 - - 4 0 = : 10 i ~ n i l s ava~lable with it. Dernand of marltct D-, is 30, h/liri [lo, 301 = 10. Thus PO units can be supplied by S, to D,. As before, we set x,, = 10 and encircle it. Now, availability at S, is exhausted. D, still needs 30 - PO = 20 units more. This can be supplied by S, only (as , , = Min [55, 201 = 20 and. availability of S, is exhausted). Thus set x encircle it. Now, S, has 55 - 20 = 35 units available and only 35 is the x e d of D, Why does it happen? = 35 and encircle it. You Clearly because the problem is balanced. So, set xZ3 see that you have very conveniently obtained a basic feasible solution of the T P with x,, = 40, x,, = PO, x,, = 20, x,, = 35. These are the basic variables. x ' . Count All other variables are non basic and zero where you have put a ' the number of basic variables. These are 4 viz x,,, x,,, x,,, x,,. You may like to verify what you have-learnt in Unit 9. Number of basic variables is (n+ ~ n) 1 = (2 + 3) -1 = 4. Having understood the technique of the method, let us take a 3 x 4 TP the f which is given as: tabular form o
Table I

We observe that &,= Cbj= 210, Thus, it is a balanced TP Start with N.W. comer cell (1, 1)
I

~easible ~olution of the TransportationProblem

Min [a,, b,] = Min [70,50] = 50 Set x,, = 50. Set x,, = x,, = 0 and cross 'x'. Why?

Because demand of Dl is exactly met and it does not need any supply from
S, or s,.

I
I

I
I

availability at S, = 70 - 50 = 20 Demand for D, = 40 Min [20,40] = 20 Set x,, = 20. Set x,, = x,, = 0 and put 'x'. Why? Because, supply at Sl is exhausted. It cannot supply to markets D, and D,.
I

i/

D ,still needs 40 - 20 = 20 units which are to be supplied by next source S,.

I

.I
i

Min [20, 501 = 20

S , can supply all the 20 units needed 20.

Set %z = 20. Set x,, = 0 and put a 'x'. Why?

i
I

,is exactly met. Because demand of D

I

I
I

Next consider demand of D, which is 50 units. S, is left with 50 - 20 = 30 units,? Min [30,50] = 30.
'

S, can supply 30 units needed by D,. Set x,= 30.

:
I

Set x, = 0 and put a 'x'. Why? Demand of D, is 50 - 30 = 20 units more.

S, has 90 units.

25

'

I I
I I
I

Special Linear Programming Problems

Min [90,20] = 20 Set x , , = 20. Demand of D, is exactly met. Next consider the clemancl of D,.

'i

Demand of D, = 70 availability of S , = 90 - 20 = 70 .
Why it so happens? Set x , , = 70 and get a basic feasible solution in which basic variables have I values

These u e 6 in number. What do you expect. This should be (rn + n) -1 = (3 + 4) -1'= 6 in number.

, , =0 Now, non-basic variables are x,, = x,, = x,, = x,,= x,, = x

and these me indicated by d cross in the table.

t1
I 1

Yon may note in the ribove two examples that the circled basic variables move along I~orizonttll and vertical lines or cells alternately. These could also move along vertical and horizontal lines if Min [a,, b,] is equal to a,. In the above two cases it was b, in each case. You may note that we have talked only about the number s f positive varixblcs in a bzsic feasible solution. If you recall the definition of a basic feasible solution to the system of equations Ax = b, x 2 0 where A is a rn x n matrix, and rank of A = rn, then you require that:
1) 2)

number of positive variables is at the most 'm'. Columns associated with the 'm'variables . . labeled as basic variables are linearly independent.

We are yet to establish the truth o i t h e statement (2). In a TP, we shall say that basic variables are linearly independent or the basic cells (cells in which basic'variables are circled) are in linearly independent position. We sl~alldiscuss this stateinent aAer outlining Matrix Minima Method to find the initial basic feasible solution of a '1'P. Matrix'Minima Method to find the initial basic feasible solution of a TP.
Y
f

EXERCISE I:
Using North-West Corner Method, find a basic feasible solution of the following TP.

26
A .'

Feasibtc Solution of the Transportation Problem

(ii)

10.3 MATRIX-MINIMA METHOD

You have observed in the section 10.2, that we have given no consideration to the data cv Now,we discuss method in which we transport starting w i t h cheapest route and using the routes in ascending order of costs of

Special Linear ProgramningPrOb'ems

I

For demonstration, let us come back tc, two problems discussed in the section 10.2. The Sewing Machine TP IS: Table

- 11

The collection of cells we can call a matrix. Thus it is a 2 x 3 matrix i.e. two rows and three colurnns. The cheapest route is the one connecting source S1to market Dl where cost per unit is 5. i.e. C,, = 5. availability at S, is 50 Demand at Dl is 40

Min [50,40] = 40
Thus, S, can supply all the 40 units needed by Dl. Set x,, = 40 and circle it.
1

..

Set x,, = 0 and cross 'x'. Why? Next look for next cheapest cost among uncrossed cells. (Crossed routes are no more needed). This route is (2, 2) connecting source S, to miirket D, with C,, = 8 availability at S, = 55 Demand at D, = 30
I

,
1

Min [55,30] = 30 Set x,, = 30 and circle 30.

!
1

28

Set x,, = 0 and put a 'x'. Why?

I

i,
a,

! i i

Now, look up for cheapest route in uncrossed or unused routes. It is (1,3) with C,, = 10. availability at Sl = 50 - 40 = 10

Feasible Solution of the TransportationProblem

I

Demand at D, = 35. Min [lo, 351 = 10.

--

set x,, = 10 and circle it. Next cheapest and only unused route is (2,3) with q, = 20. availability at S, = Dernand;bC~, = 25. Why? Set x,, = 25 and circle it. This gives a basic feasible solution with x,, = 40, x,, = 10, x,, = 30, x, = 25. Other variables are non basic. Note that number of basic variables is (2 + 3) -1 = 4. Also, note that this basic feasible solution is different from corresponding solution that we obtained in section 10.2. It is because, now costs have been taken into consideration. However, we have yet to establish Linear Independence of these basic variables, We next find a basic feasible solution by Matrix Method for the second problem discussed in section 10.2. The problem is
Table I11
1

I I I I

I
I

Minimum cost in the matrix is in the cell (3, 1) linking third source S, to Destination Dl and in the cell (2,4) linking source S, to destinatiop D4. We can choose any one of the two the cell. Let us choose (3, 1)

29

Special Linear
Programming Problerns

Min [go, 501 = 50

Set x,, = 50 and circle it as a basic variable. As demand of D, is exactly met, we put a cross in the cells ( I , 1) and (2, 1) indicating that these are non basic i.e. x,, = x,, = 0 and these are non basic variables. Next equally cheapcost route is (2, 4) with c, = 1.

Set , x = 50 and circle it as a basic variable

Set x , , = x,,

= 0 as

these are non basic variables and put a 'x'.

We now look for the cheapest route among open routes i.e. routes which have not been crossed. These are (3, 3) aid (1'4). We find that c,, = 4 and c,, = 4. Choose any one of these say (3,3). Capacity available at S, is 90 - 50 = 40 Demand of D, = 50
Min [40, SO] = 40
I

Set x , , = 40 and circle it as a basic variable.

, , = 0 and put a 'x'. Why? Set x,, = x

Choose next equally cheaper cost route (1,4)

I
I

demand for D, is 70 - 50 = 20

1
I

Min [70,20] = 20
Set x,, = 20 and circle it as-a basic variable.

, .

Next higher cost in the matrix with routes unused is c, = 6.

.

. .

,.
I

availability at S, = 70 - 20 = 50 demand at D , = 50 - 40 = 10

j
I

Min [50, 101 = 10

30

I
11,

I
I

Set x,, = 10 and encircle it as a basic variable Last unused route is (I, 2) with c,, = 7 Availability with S., = 70 - (20 + 10) = 40 demand at D, = 40. S.etx,, = 40 and circle it.

Feasible Solution of the TransportationProblem

'

:
'
.
I

.I
!I,

:
I

1.: 11 !I
1.;
,

T h i s gives you a basic feasible solution with (4 + 3) - ,1= 6 basic variables

viz.
'

1:.
:!

1
,
I

/I!
ili

xI2= 40, x13 = 10, x14 = 20

~~4

= 50, x3, = 50, ~ 3 . 3 = 40

/I/ );i j!!

ill

1;

1 i:
/I1

Other variables are non basic. Observe that the basic feasible solution to the problem obtained by NorthWest Comer Method and one obtained by Matrix Mini~na Method arc distinct. You may also observe that if you were to solve this problem as a LPP, you would have added artificial va iables and apply Phase I Method to reach the initial basic feasible solution. Imagine the amount of labour saved. ' This is one of the reasons to discuss TP as a seperate unit. r , EXERCISES 2:
I

1 -

I
t

1 Using Matrix Minima Methgd, find the initial basic feasible solution to the following TP.

Special Linear Programming Problems

(ii)
Dl

D 2

D 3

45
I
I
I

40

35

I
I

So far, w e have discirssed two neth hods to find all initial basic feasiblc solution to a given TP. As already i~~el~tioned, we have yet to verify the I'act that the solution so o h t a i n ~ d is a basic feasible solution. We now state a ~.ule or procedul.~ lbl-dieel(i11gwhether a given feasible solution is basic, i.e. celk are in linearly i ~ ~ d e p e ~ i dpositions. cnt I'liis is called the chain ~*ule.
Closed Chain
L

Consider the 2 x 3 TP discussed in the beginning of this unit.

.I*

The tabularrepresentation is given in table I1 viz.

Dl
5

D 2
15

D3
10

a,

's,
10
s2

50

20

55
Q

bj -t 40

30

'

35

Consider the set of cells (1, I), (1,3), (2, 31, (2, 1). Starting with any one of the cells say (1, I) draw alternately horizontal and vertical lines to reach the

other cells. If doing so, we come back to the starting cell, we say that a closed chain is formed. Thus in this case, the circuit is as follows:
*>

Feasible Solution of the

TransportationProblem

(1, 1)

(193)

Similarly, we observe that (1,2),'(1,3), (2, 3), (2,2) forms a closed chain as given below:

Likewise, (1, I), (1,2), (2, 2), (2, 1) from a closed chain . Next let us consider the cells.

In this case, drawing alternates horizontal and vertical line, we do not come back to the starting cell (1, 1). Thus, these cells do not form a closed chain and therefore, form simply a chain. Similarly, (1, I), (1, '3), (2,3) form a chain, (1, I), (1,3), (2,2) form a chain.

We now give you a method or rule to test if given any number of cells, these are in linearly independent positions or in linear dependent positions.

Closed Chain Rule

Given a feasible solution, if you can foim a closed circuit involving all the basic cells or any subset or part of basic cells, then the cells are in linearly dependent position. If you can form no circuit involving the basic cells above, the cells are in linearly independent positions.
I

Applying this rule, you may observe that initial feasible solution obtained by North.West Corner Method or by Matrix Minima Method is indeed a basic

S~ecial~inear~rogrnmmingfeasible solution satisfying the.criterion of cells being in independent Problems positions.

As an illustration, the solution from the 2 x 3 TP is: ai

basic cells are (1, l), (1,2), (2,2), (2,3).

You observe that these cells do not from a closed chain. Thus these are in linearly Independent positions.

..

Next, consider the solution obtained for the second TP in section 10.2

(I, 1) with allocation 50, (1,2) with allocation 20, (2,2) with allocation 20, (2, 3) with allocation 30.

~
, I

(3,3) with allocation 20, (3'4) with allocation 70. These cells do not form a closed chain as shown below: (1,l)
' I 1 1

1
I

2)

(2-2)

(2,3) 3)
(374)

(39

Thus, these cells are in linearly independent positions.

'
I

34

I*

You III;IY co~isider tlie solution obt;~illcd for t l ~ c 'I'P discussed in the tleginning of section 10.3.

Fcasiblc Solulion of L h c Transportalion Problcm

b, -+

40

30

35

I3asic cells are ( I , l), (1, 3), (2, 3), (2,2)

'I'hcy do not from a closed chain ns shown below:

(1, 1)

(1,3)

Similarly, for solution of the second problem discussed by Matrix Minima Method, basic cells are (1,2), (1, 3), (1,4), (2,4), (3, I), (3, 3).
We may even try to form closed chains involving all the,cells. We do not succeed :is shown below:

We shall extend this rule in the next section to identify the cells at zero level i l l case, [he solution ob;ained by any of the methods or even an any stage (after goirig through dle next unit) is a degenerate solution.

Special Linear Programming Problems

10.4 DEGENERATE BASIC FEASIBLE SOLUTION

You may recall that in a LPP, a basic feasible solution is said to be degenerate if certain basic variables (one or more) are at zero level, i.e. number of positive variables is less then m for the system of equation Ax = b, x 2 0,A being m x n and rank of A is m.However, in a LPP you were not faced with the problem of identifying the basic variable at zero level. You got these variables at zero level in the process of solving the LPP.
-

Let us see what happens in a Transporation Problem(TP) while determining basic feasible solution by North West Corner Method or by Matrix Minima Method. In certain cases, you may discover that the number of basic n) - 1 was number. 23is is variables circled are less than required (m -Iillustrated by the following example:
Table I1

The initial basic feasible solution has been obtained using North-West Corner Method. Observe that the data of the Sewing Machine TP has been slightly altered to get such a degenerate solution. What do you observe? The number of basic variables is 3 which is less than (3 + 2) - 1 = 4. Such a solution in LPP you call a degenerate basic feasible solution. We must now identify the additional basic variable at zero level to make it a complete basic feasible solution having (m + n - 1) variables. To do it, we give the following rule:

Rule

.i

Select a variable as a candidate for basic variable at zero level. Suppose it is x,, in the above example. i.e. cell (1,2). It is called a candidate cell. Start joining it by horizontal lines and vertical lines alternately to basic variable , cells with circled allocation,

I
I

In this process, if we can come back to the candidate cell, this is disqualified. Why? Because they form a linearly dependent set. Otherwise, we can put a zero in the cell and circle it as @ .
. -.

I
I

I(

j
i

II

..

To illustrate the rule, (1, 2) can be joined only to (1, 1) by a horizontal line and then (1, 1) cannot be joined to any basic cell by a vertical line. So, we cannot come back to (1,2). Thus (1,2) qualifies and we can put a 0 in the cell (1,2) and circle it. You may observe that (1,3) and (2, 1) also qualify to become basic variables at zero level. Remember only one cell is to be marked to make (3 + 2) - 1 = 4 basic variables. To make it more clear, let us take another TP as follows:

Feasible Solution of Ihe Transportation Problem

Note that the basic feasible solution has been written using Matrix Minima I Method. Number of basic variables needed is (5 + 3) - 1 = 7. We find only 6 of these, in the process. Thus an addition basic variable at zero level is to be identified. Following the rule, we see that (1, 1) does not qualify as we have the following closed chain.

Thus (1, 1) forms a closed chain with basic cells or variables at (1,3), (3, 3) and (3, 1).
4

(1,2) also does not qualify as it forms a closed chain as shown below.

Special Linear Programming Problems

You may note that (1,5) qualifies as it does not form a closed chain.

You may similarly discover that (2, I), (2,2), (2,3), (2, 4) and (3,s)qualify and (3,4) does not. Thus you can put a zero or @ in any of these cells (1,5), (2, I), (2,2), (2, 3), (2,4),but not in (1, I), (1, 2) and (3,4). Remember you have to choose only one to make a total of (5 + 3) - 1 = 7 basic variables or cells.

EXERCISE 3: Using North West comer Method obtain a basic feasible solution of the following TP. Indicate also the basic variables at zero level.

EXERCISE 4: Determine a basic feasible solution by Matrix Minima Method Indicate thebasic variable at zero level.

10.5 SUMMARY
Having gone through this unit, you are able to appreciate that although a TP is a LPP, you can find its initial basic feasible solution very conveniently and comfortably without adding artificial variables and applying Phase I Method. Also, you are at ease finding a basic feasible solution either using NorthWest Corner Method or Matrix Minima Method. In case you encounter degeneracy at any stage, you can identify the basic variable to be set equal to zero and thus write the complete degenerate basic feasible solution.

Feasible Solution of the Transportation Problem

i I
j I

i;
!

1,

j:

1:
I,

!:I

I1':'
I,
i

j:
It
I

1,
I:.
i.
!,I !'
i

1~

Special Linear Programming Problems

10.6 ANSWERSIWINTSISOLUTIoNS
El
(i)
X I I = 10,
X23 = x2l =

30,

x22 =

10.

10,

(ii)

xl1=25,

E2

(i)

xll = 10,
x 2 1 = 40,

XI?=

10,

x 3 3 = 30. x 2 2 = 10, x23 =

(ii)

x,1=40,

20,

E3

(i)

x, I

= 40,

x 1 2 = 10,

x 2 2 = 30,

(Other basic variables at zero level also possible, e.g. X23 = 0 or x3? = 0.)

E4

X I ! = 30,

x 1 2 = 50,

x 2 3 = 50,

(Other basic variables at zero level are also possible, e.g. x13 = 0,
X33 = 0.)