FIN-3304 Management Science Applications-2019

5. Sensitivity Analysis with Graphical Solution

Method
The optimum solutions we obtain for a given linear programming model represent a solution under
a set of restrictive assumptions, such as the certainty of the decision environment. In real-world
situations, the decision environment is dynamic rather than static. Thus, the decision maker is vitally
interested in the “what if” questions concerning the effect of change in model parameters on the
optimum solution.

Sensitivity analysis is a post-optimality analysis which attempts to evaluate the sensitivity of the
optimum solution to discrete changes in the model parameters 1. If the optimum solution is very
sensitive to change in certain parameters, special efforts should be directed to accurately forecasting
the future values of those parameters. However, if the optimum solution is no very sensitive to
changes in parameter, estimating the value of that parameter more accurately would waste time
and effort.

However, if we need to perform sensitivity analysis of many small changes in the parameters, the
interactive approach may be both costly and time consuming. Thus, the simplex based sensitivity
analysis can be a very useful tool. In this session we discuss the sensitivity analysis step by steps
using graphs since there is no short-cut in graphical method.

Theoretically, we can focus sensitivity analysis into five sub-topics as follows 2:

(1). Change in the unit contribution rate (cij)

(2) Change in the available resources (bi)
(3) Change in the technological coefficients (aij)
(4) Addition of a new constraint: and
(5) Addition of a new variable (This is not valid for Graphical method)

By way of introducing sensitivity analysis, let us examine the following problem:

Maximize Z: = 40X1 + 50X2

Subject to: X1 + 2X2 <= 40 Labor (Hrs)

4X1 + 3X2 <= 120 Material (Kg)

X1 , X2 >= 0

1
In other word sensitivity analysis is the analysis of the erect of parameter changes on the optimal solution.
2
However, we limit our discussion to first, second and last sub topic because calculation process of other two
will take time and difficult. A Computer package has to be used for those.
33
 FIN-3304 Management Science Applications-2019

CHANGE IN THE UNIT CONTRIBUTION RATE:

In many real world problems, contribution rates (unit profit or unit cost of the decision variables in the
objective function) change frequently. In our example suppose the unit profit of X1 change from Rs.40/=
to Rs.66.66/= while the unit profit of X 2 remains same the only change is the unit contribution rate of X 1 in
the objective function. On the other hand the main thing is to note that there will be no change in the
feasible region, it remains exactly the same. Thus, the only thing we have to check is the optimality of
the present solution.

The graphical solution for this original problem is shown in figure 1. According to the figure 1 as well as
the Table-01 , optimal point is shown to be at point B. (X 1=24, X2=8, Z= 1360).

Maximize Z: = 40X1 + 50X2

Table-01

Points X1 X2 Z
Figure- 01
0 0 0 0
A 0 20 1000
B 24 8 1360
C 30 0 1200

If we change the profit of X1 from Rs. 40/= to Rs. 66.66/= A

what would happen to the present optimum solution.?

B
How would that effect the solution identified in figure 1?
B

CC
As shown in the figure 2, the new solution is:

Maximize Z: = 66.66 X1 + 50 X2
Table-02

Points X1 X2 Z
0 0 0 0
A 0 20 1000
B 24 8 2000
C 30 0 2000
Figure-02
Figure-02

Then, both points “B and C” in the feasible range become optimum; the solution is a multiple optimum.
In other words the objective function line became steeper, so much so that the optimal solution point
changed from B to B and C. What this change tells us? What would happen
AA if we further increase the
contribution rate of X1 up to 100?

BB

C
C

34
 FIN-3304 Management Science Applications-2019

The solution shows in figure 3 and the Table No.03.

Maximize Z: = 100 X1 + 50X2

Figure-03

35
 Points X1 X2 Z
0 0 0 0
A 0 20 2000
B 24 8 2800
C 30 0 3000

The new solution became point C. The objective function line became more steeper, so
C much so that
the optimal solution point changed to point C.

On the other hand, it is interesting to see what would happen if we decrease the contribution rate of X1
from original Rs.40/= to Rs. 25/= The results shows in the figure 4. and the following table No.04
Figure - 04

Maximize Z: = 25 X1 + 50 X2
A
points X1 X2 Z
0 0 0 0
A 0 20 1000
B 24 8 1000
C 30 0 750

Now points A and B both become optimum.

Similarly, if we further reduce the contribution rate of X 1 from

original Rs.40/= up to Rs. 20/= you will note the point A
become the optimum. (See figure 5) and the following Table
No.05

Maximize Z: = 20x1 + 50 X2
Figure-05
Points X1 X2 Z
0 0 0 0
A 0 20 1000
B 24 8 880
C 30 0 600
A
What this story tells us? According to the discussion we had so for, this story says us that:

 when we increase the contribution rate of X 1 from original 40 to any amount less than 66.66,
the present optimum solution point B continues as optimum.
 when we increase the contribution rate of X 1 from original 40 to exactly 66.66, the present
optimum solution point become multiple optimum.
 when we increase the contribution rate of X 1 from original 40 to any amount beyond 66.66, the
new optimum solution emerged.

On the other hand:

 when we decrease the contribution rate of X 1 from original 40 to any amount grater than 25,
the present optimum solution point B continues as optimum
 when we decrease the contribution rate of X 1 from original 40 to exactly 25, the present
optimum solution point become multiple optimum.
 when we decrease the contribution rate of X 1 from original 40 to any amount less than 25, the
new optimum solution emerged.

Based on this argument we can conclude that in case of X 1, the maximum point we can increase the
contribution rate is Rs. 66.66 and the minimum point of which the contribution rate can be decreased is
Rs. 25/- in order to continue the present optimum solution as optimum. Beyond these limits there will be
a new solution. Therefore, Rs. 66.66 is called as Upper Limit and Rs. 25 is called as Lower Limit3. With
in this range the present optimum solution point continue as optimum. Similarly you can calculate
upper and lower limits for X2 also.

CHANGE IN THE AVAILABLE RESOURCES:

The second type of sensitivity analysis we will discuss is the sensitivity ranges for the constraints
quantity values i.e the value to the right of the inequality sings in the constraints. (Right Hand Side
values-RHS) In order to discuss this aspect, we will use the same example used earlier.

Maximize Z: = 40X1 + 50X2

Subject to : X1 + 2X2 <= 40 Labor (Hrs)

4X1 + 3X2 <= 120 Material (Kg)

X1 , X2 >= 0

Consider the company can increase the labor hours from 40 to 60. The effect of this change in the
model is graphically displaced in Figure 6.

3
The sensitivity range for an objective coefficient is the range of values over which the current optimum solution
will remain optimal.
 By increasing the available labor hours from 40 to 60, the
feasible region is changed. It was originally OABC and now it is
Fiigure-06
OA1B1C. B1 is the new optimum solution instead of B.
However, the important consideration in this type of sensitivity
analysis is that the solution mix (variables that do not have zero
A1 values) including slack variables, do not change even though
B1
the values of variables of x1 and x2 change. (from x1 =24,
x2=8 to x1 =12, x2 =24) The focus of sensitivity analysis for
constraint quantity values is to determine the range over which
the constraint quantity values can changed without changing
C the solution variable mix, specially including the slack variables.
If the quantity value for the labor constraint is increased from
40 to 80 hours, the new solution space is OA 1C, and a new
solution variable mix occurs at A1 as shown in figure 7 (a).

Whereas at the original optimum point, B, both x1 and X2 are in the solution, at the new optimal point ,
A1, only x2 is produced (ie. X1= 0, x2=40, s1=0, s2=0). Thus the upper limit of the sensitivity range for
the quantity value for the first constraint, which we will refer
to as q1, is 80 hours. At this value the solution mix changes Figure- 07 (a)
such that X1 no longer produced. Further as q1 increases
past 80 hours, solution will be infeasible.

A1
Similarly, if the value of q1 is decreased to 30 hours, then the
new feasible solution space is OA 1C, as shown in figure 7 (b).
The new optimum point is at C, where no X2 are produced.
The new solution is x1=30 X2= 0,S1 =0, S2=0 and Z = 1200.
C
Again the variable mix is changed.

What this story tells us?

Figure -7(b)  When we increase available labor up to the maximum point of 80, X2 increase up to maximum
40 and X1 decrease up to 0.

 When we decrease the available resources up to the

A1 minimum 30, X1 increase up to 30 and X2 decrease up to 0.

C  Beyond this available resources ranges, solutions will be

infeasible.

Therefore, summarizing the sensitivity range for the constraint

quantity value for labor hour is 30 <= q1 <= 80 hours.

The sensitivity range for material also can be determined

graphically in a similar manner.
Summarizing the sensitivity range for q1 and q2 are:

30 <= q1 <= 80 hours

60 <=q2 <=160 Kgs.

Based on the above discussion we can summarize the effect of change in the right hand side values
(resources) of the constraints on graphical solution as follows:

 The change of RHS values means that the variables in the optimum solution will remain feasible
as long as the resources are within this range. It does not mean that the solution values will
remain the same. There will be new solution values. But production of x1 and x2 continuous.
 Slope of the constraint line will not be changed. But the constraint line will shift either to right
or to the left. In other word feasible area is changed and also the optimum solution point
changed.
 The marginal values of the resources in question is valid only up to the upper limit of that
resources. (You will realize this after discussing this adjustment with the simplex method)