OPERATIONS RESEARCH

LECTURE SIX
Linear programming (2)
Lecturer: Dr. Emily Roche
INTRODUCTION
This lecture will focus on graphical solution to linear programming problems.

Intended learning outcomes

At the end of this lecture, you will be able to Solve LLP using graphical method

References
These lecture notes should be supplemented with relevant topics from the books listed
in the Bibliography at the end of the lecture

Solving mathematical models for linear programming

There are two main methods of solving linear programming problems. In this lecture, we
will learn the first method which is graphical solution to the problems.
This method is used to solve a linear programming problem with only two decision
variables.
Procedure
1. Formulate the linear optimization model.
2. Graph the feasible space.
3. Determine the coordinates of each of the corner points within the feasible space.
4. Find the value of the objective function at each of the corner points.
5. For a bounded region, the solution is given by the corner point producing the
optimum value of the objective function.
6. For an unbounded region, check that the solution exists. If it does, it will occur at
a corner point.
Example
A manufacturer produces two different cloth products Silk and Cotton. Silk has the
contribution of three dollars per unit, and Cotton four dollars per unit. The
manufacturer wishes to establish the weekly production plan to optimize the
contribution. The production data are as follows:

Per unit
Machine (hours) Labour (hours) Material (Kg)
Silk 4 4 1
Cotton 2 6 1
Total availability/week 100 180 40

Because of a trade agreement, sales of Silk are limited to a weekly maximum of 20 units
and to honour an agreement with an old established customer at least 10 units of
Cotton must be sold per week. Formulate and solve the linear programming program.

Solution
Let 𝑥𝑖 , 𝑖 = 1, 2 be the units of Silk and Cotton to be produced respectively, then the
optimization model will be:
Maximize 𝑥0 = 3𝑥1 + 4𝑥2
4𝑥1 + 2𝑥2 ≤ 100
4𝑥1 + 6𝑥2 ≤ 180
𝑥1 + 𝑥2 ≤ 40
Subject to
𝑥1 ≤ 20
𝑥2 ≥ 10
𝑥1 ≥ 0
Since this problem has only two decision variables, we will use graphical method to
find the solution.

We plot all the constraints on a single cartesian plane, with inequality signs changed
into equal signs. The direction in which each constraint holds is then determined by
observing the direction of the inequality sign. We then shade the unwanted region. The
area which is satisfied by all the constraints including all the boundaries is known as the
solution space or the feasible space.

We are interested in determining the points in the solution space which yields the
maximum value of 𝑥0 . This is determined by identifying the coordinates of the corner
points within the feasible space i.e., (0,10), (0,30), (20,10) and (15,20), and then
determining the corresponding value of the objective function at each of the corner
points. Thus;
𝑥1 𝑥2 Objective function 𝑥0
0 10 40
0 30 120
20 10 100
15 20 125
The coordinates (15, 20) represent the optimal value of the objective function 𝑥0 = 125.

Effects of constraints
Constraints are the conditions which limit the decision variables to their feasible values.
A constraint is binding if altering it also alters the optimal solution.
When using graphical solution method, binding constraints (scarce resources) will pass
through the optimum solution point. In the above example, the binding constraints are

4𝑥1 + 2𝑥2 ≤ 100 (machime hours)

4𝑥1 + 6𝑥2 ≤ 180 (labour hours)
Less severe constraints that do not affect the optimum solution are known as non-binding
constraints (abundant resources). These constraints do not pass through the optimum
solution point in graphical method. In the above example, the non-binding constraint is
𝑥1 + 𝑥2 ≤ 40 (materials (kg))
The management should be given information about the value of the scarce resources.
These valuations are known as the shadow/dual prices. They are obtained from the
amount of increase/decrease in in the optimal contribution that will arise if 1 more/less
unit of the scarce resource was made available.

Determining the shadow prices

There are two methods of finding the shadow prices when dealing with graphical
solution to linear optimization models:

a. Arithmetic method
The process here is to determine the magnitude of change on the optimal solution when
one constraint is increased by a single unit as the other is held constant and vice versa.
That is, assume that one more machine hour is available, but the labour hours remain
constant at 180. The resulting difference in the optimal contribution will be the shadow
price corresponding to the machine hour.
Hence,
4𝑥1 + 2𝑥2 = 101
4𝑥1 + 6𝑥2 = 180
Solving the two equations simultaneously, we obtain
𝑥2 = 19.75 𝑥1 = 15.35
The corresponding value of the objective function then becomes
𝑥0 = 3𝑥1 + 4𝑥2 = 3(15.375) + 4(19.75) = 125.125
The change in the optimal contribution is
125.125 − 125 = 0.125
Therefore, the shadow price per machine hour is 0.125 dollars.
Similarly,
4𝑥1 + 2𝑥2 = 100
4𝑥1 + 6𝑥2 = 181
Solving the two equations simultaneously, we obtain
𝑥2 = 20.25 𝑥1 = 14.875
The corresponding value of the objective function then becomes
𝑥0 = 3𝑥1 + 4𝑥2 = 3(14.875) + 4(20.25) = 125.625
The change in the optimal contribution is
125.625 − 125 = 0.625
Therefore, the shadow price per machine hour is 0.625 dollars.

b. Dual formulation
This involves formulating a dual optimization model for a primal sub-model of the
binding constraints, then solving for the dual variables.
From the above example,
The primal sub-model of the binding constraints is:
Maximize 𝑥0 = 3𝑥1 + 4𝑥2
4𝑥1 + 2𝑥2 ≤ 100
Subject to
4𝑥1 + 6𝑥2 ≤ 180
𝑥1 ≥ 0, 𝑥2 ≥ 0
The dual problem will be
Minimize 𝑦0 = 100𝑦1 + 180𝑦2
4𝑦1 + 4𝑦2 ≥ 3
Subject to
2𝑦1 + 6𝑦2 ≥ 4
𝑦1 ≥ 0, 𝑦2 ≥ 0
Solving the two constraint equations simultaneously results give the values of the dual
variables as
𝑦1 = 0.125, 𝑦2 = 0.625
Which are the shadow prices corresponding to the machine hours and labour hours
respectively
The contribution of the binding constraint to the optimal solution can be confirmed by
multiplying the available units by the contribution per unit (shadow prices).
Since the available machine hours was 100, and each machine hour contribute
0.125 dollars, then the total contribution to the optimal solution from machine hours is
100(0.125) = 12.5
Similarly, the total contribution from the labour hour having aa maximum availability of
180 will be:
180(0.625) = 112.5
Therefore, the total contribution to the optimal solution from the binding constraints is
12.5 + 112.5 = 125

Bibliography
Lucey, T. (2002). Quantitative Techniques (6th ed.). Cengage Learning.

Taha, H. A. (2017). Operation Research An introduction (10th ed.). Prentice-Hall, Inc.