Linear Programming: Sensitivity Analysis and Interpretation of Solution

Sensitivity analysis is the study of how the changes in the coefficients of an

optimization model affect the optimal solution. Because sensitivity analysis is concerned
with how these changes affect the optimal solution, the analysis does not begin until the
optimal solution to the original linear programming problem has been obtained. For that
reason, sensitivity analysis is sometimes referred to as post optimality analysis.

Sensitivity analysis is important to decision makers because real-world problems exist in

a changing environment. Prices of raw materials change, product demand changes,
companies purchase new machinery, stock prices fluctuate, employee turnover occurs,
and so on. If a linear programming model has been used in such an environment, we
can expect some of the coefficients to change over time. We will then want to determine
how these changes affect the optimal solution to the original linear programming
problem.

Using sensitivity analysis, we can answer questions such as the following:

1. How will a change in a coefficient of the objective function affect the optimal solution?
2. How will a change in the right-hand-side value for a constraint affect the optimal
solution?

Graphical Sensitivity Analysis

The objective of Graphical Sensitivity Analysis is to determine the optimal solution's

sensitivity to certain changes that may occur in the data.

There are two graphical solution methods can be used to perform sensitivity analysis

1. Change In Objective Function Coefficients

2. Change In The Right-Hand-Side Values for the Constraints

Change In Objective Function Coefficients

The objective function is a means to maximize (or minimize) something. The range of
optimality for each objective function coefficient provides the range of values over
which the current solution will remain optimal.

As long as the objective function line stays within the shaded region, the solution will
remain optimal.
Change In The Right-Hand-Side Values for the Constraints

Let us now consider how a change in the right-hand side for a constraint may affect the
feasible region and perhaps cause a change in the optimal solution to the problem.
Adding or reducing the value of a constraint, the result will either expand or reduce the
feasibility region of the solution and movement in the optimal solution. With either of the
result, we now want to determine whether one of the new feasible solutions provides an
improvement in the value of the objective function.

Dual Value is the change in the value of the optimal solution per unit increase in the
right-hand side.
Sensitivity Analysis: Computer Solution

Sensitivity analysis can be easily done through the help of systems already available for
use such CPLEX, Gurobi, LINGO, MOSEK, COIN-OR, Excel Solver, and Analytic
Solver for Excel.

Computer software packages for solving linear programs are readily available. Most of
these provide the optimal solution, dual value or shadow price information, the range of
optimality for the objective function coefficients, and the range of feasibility for the right-
hand sides.
Limitations Of Classical Sensitivity Analysis

Classical sensitivity analysis obtained from computer output can provide useful
information on the sensitivity of the solution to changes in the model input data.
However, classical sensitivity analysis provided by most computer packages does have
its limitations.

Three Limitations

1. Simultaneous Changes
2. Changes In Constraint Coefficients
3. Nonintuitive Dual Values

Simultaneous Changes

The sensitivity analysis information in computer output is based on the assumption that
only one coefficient changes; it is assumed that all other coefficients will remain as
stated in the original problem. Thus, the range analysis for the objective function
coefficients and the constraint right- hand sides is only applicable for changes in a
single coefficient. In many cases, however, we are interested in what would happen if
two or more coefficients are changed simultaneously.
Changes In Constraint Coefficients

Classical sensitivity analysis provides no information about changes resulting from a

change in the coefficient of a variable in a constraint. Even though this is a single
change in a coefficient in the model, there is no way to tell from classical sensitivity
analysis what impact the change in the coefficient of S will have on the solution.

Nonintuitive Dual Values

Constraints with variables naturally on both the left-hand and right-hand sides often lead
to dual values that have a nonintuitive explanation.

OPERATIONS MANAGEMENT APPLICATIONS

Linear programming (LP) is a mathematical modeling technique used by operations

managers to determine optimal solutions to decision problems. Linear programming
applications developed for production and operations management include scheduling,
staffing, inventory control, and capacity planning.

Many decisions faced by an operations manager are centered around the best way to
achieve the objectives of the firm subject to the constraints of the operating
environment. These constraints can be limited resources, such as time, labor, energy,
materials, or money, or they can be restrictive guidelines, such as a recipe for making
cereal, engineering specifications, or a blend for gasoline.
A Make-or-Buy Decision

The make-or-buy decision is an important one for many organizations. Traditionally it

has been made using standard cost accounting methods. The make-or-buy issue
involves determining whether a particular component should be made in-house or
purchased. It is an important one for many manufacturing organizations today as they
rationalize their supply-chain for improved productivity and profit.

Production Scheduling

One of the most important applications of linear programming deals with multiperiod
planning such as production scheduling. The solution to a production scheduling
problem enables the manager to establish an efficient low-cost production schedule for
one or more products over several time periods (weeks or months). Essentially, a
production scheduling problem can be viewed as a product-mix problem for each of
several periods in the future. The manager must determine the production levels that
will allow the company to meet product demand requirements, given limitations on
production capacity, labor capacity, and storage space, while minimizing total
production costs.

One advantage of using linear programming for production scheduling problems is that
they recur. A production schedule must be established for the current month, then again
for the next month, for the month after that, and so on. When looking at the problem
each month, the production manager will find that, although demand for the products
has changed, production times, production capacities, storage space limitations, and so
on are roughly the same. Thus, the production manager is basically re-solving the same
problem handled in previous months, and a general linear programming model of the
production scheduling procedure may be applied frequently. Once the model has been
formulated, the manager can simply supply the data—demand, capacities, and so on—
for the given production period and use the linear programming model repeatedly to
develop the production schedule.

Workforce Assignment

Workforce assignment problems frequently occur when production managers must

make decisions involving staffing requirements for a given planning period. Workforce
assignments often have some flexibility, and at least some personnel can be assigned
to more than one department or work center. Such is the case when employees have
been cross-trained on two or more jobs or, for instance, when sales personnel can be
transferred between stores. In the following application, we show how linear
programming can be used to determine not only an optimal product mix, but also an
optimal workforce assignment.