Module - Exercises In Quantitative Techniques

Linear Programming

Tell Me

Linear programming is used primarily in relation to production, but can also be used for other
specific project-related activities.

As an example of where it might be used, think of a business that makes axles and gearbox
components. The manager will know:

How much profit each item creates

In any decision-making process, there is likely to be a need to determine the most suitable
solution. This will generally involve complying with certain key constraints, such as cost limits,
maximum stock availability or optimum production levels. Where there is a single constraint, it
is normally very simple to arrive at an appropriate decision. If a machine can produce an output
of 25 units per day, then it should be possible to work at that rate, assuming that there are
sufficient raw materials ready for use. This is the optimum use of that machine in these
circumstances. However, if there are additional constraints, such as a restriction on the labour
hours that can be worked or a limit to the available raw materials (or a combination of both), it
will be necessary to seek a compromise that will prove most profitable. This is known as
optimisation.

For each process involved, it will be possible to state the constraints in a mathematical way.
For example, if the number of components (C) available on any one day is limited to 36, we
could state that:

C < 36 (where the symbol < denotes less than)

Similarly, we might use the following signs to provide accurate mathematical descriptions:

> more than

≥ more than or equal to

≤ less than or equal to

= equal to

Thus, if we are told that, using the same machine, it takes 2 hours to produce component A, or
3 hours to produce component B, and that there is a total a working time of 15 hours available
per day for that machine, we could state that

2A + 3B ≤ 15

This becomes our first constraint.

A ≤ 23

Linear programming is a way of creating equations linking those various factors. So for
example, you can create three equations based on the total amounts of steel, time and
machine time available. One of these will be:

Weekly amount of steel ≥ total steel for gearbox components + total steel for axles

(where ≥ stands for ‘greater than or equal to’)

You can then create straight-line graphs showing the three equations, and look for the values
that represent the larger level of profit per week.

In this way, a series of constraints can be produced which, when plotted on a graph, will allow
us to identify a region where all constraints can be satisfied. This region is known as the
feasible region. All points within this region will provide an acceptable solution to the problem,
and it will then be possible to choose which single one is the optimum solution.

In most cases, the appropriate solution will be the one that maximises profit. We will, most
likely, need to look for any production constraints and any assembly constraints.

In order to fully understand how linear programming works, we can follow an example to
investigate a problem and obtain a solution. Once the optimum solution has been determined,
we can evaluate this in terms of profit through the use of a simple profit relationship calculation.
This is often referred to as the objective function.

Show Me

For more detail on how linear programming can be used to solve problems graphically, study
at least one of the examples on the following website:

J E Beasley’s OR-Notes: http://people.brunel.ac.uk/~mastjjb/jeb/or/morelp.html

Once you have formed your equations, you can also use computer software to find the best
solution. The following video, and the later ones in this series, show how you can create the
equations, and then use Excel to calculate the optimal outcome.

Linear Programming Formulation in Excel: https://www.youtube.com/watch?

Involve Me

In the preceding Involve Me activities, you identified a future project relating to your own work,
and created a network diagram and a Gantt chart for the project (or a part of it).
It is likely that you project will involve the use of materials and resources. You can often identify
the most effective way of using those resources by using linear programming.

Create equations based on the various resources and outcomes. Many of them will involve the
idea of ‘less than or equal to’. Using those equations graphically, or using tools like Excel, find
the outcome that will use those inputs most effectively.
Feedback
The usefulness of this approach will vary, depending on your project. However, it is a very
useful tool in many wider organisational situations, answering questions like:

How many of each product should we create (rather like the gearbox and axle
example you studied)?
How can we best categorise our call centre staff so that they each deal with a limited
number of caller questions?

Show Me

Certain constraints will be found to be of little or no importance. This may occur as a result of
the influence of other constraints that have a greater significance. Such constraints are known
as slack constraints, whereas those that are key to the formulation of the feasible region are
termed tight constraints. For example, if a machine has a production capacity of 320 units per
day, but there are only raw materials for 200 units, we can see that the machine’s production
constraint is irrelevant – a slack constraint. In this case, the availability constraint is the tight
constraint.

Not all linear programming exercises are concerned with maximising profit. In certain cases, it
may be more important to reduce costs of production. Such a problem is referred to as a
minimization problem.

Show Me More

An example of the use of linear programming can be found at

http://www.purplemath.com/modules/linprog3.htm

Show Me

Special cases

The linear programming technique may not always identify an appropriate solution. This will
occur if either of the following two special cases arises:

No feasible region – this occurs when the constraints are mutually exclusive and do not
define a single, clear region. This may be the result of one constraint (perhaps defining a
minimum production level) which is unobtainable when related to other constraints
Multiple optimum solutions – this occurs when lines drawn on the graph are parallel, and
therefore do not contain an area. In this case, any solution found within the line segment
will provide an optimum solution but will not be unique. It would be advisable to consider
further factors to distinguish between the conditions before the final solution is selected

Sensitivity

An important benefit of linear programming is the ease with which the sensitivity of the
conditions defined by the constraints can be investigated. Slight variations in the constraints
can easily be observed diagrammatically, with the resultant effects clearly apparent. In this
way, it may be possible to identify slight modifications to either the production process or the
stock availability which could enhance the final optimised solution.

Any additional potential profit, calculated from the objective function, which may arise from a
marginal change to the resources, is known as an opportunity cost (or a shadow price).
However, it should be noted that shadow prices apply only if the constraints continue to
operate in the original way. Any significant changes may have the effect of making other
constraints more critical to the overall linear programming problem.

Up to now, we have considered problems with a relatively small number of variables. Where
real-life situations have a large number of variables with a complex array of associated
constraints, it will be necessary to use a computer-based method, which looks at a larger
number of ‘degrees of freedom’ for its mathematical solution. This often follows a technique
known as the Simplex method, which systematically eliminates the effect of each constraint in
turn until a single solution remains.

The Simplex approach is based on eight stages that take place sequentially. These stages are
(i) problem finding, (ii) fact finding, (iii) defining the problem, (iv) idea finding, (v) selecting and
evaluating, (vi) planning, (vii) selling the idea, and (viii) acting.

An alternative approach is through appreciative enquiry which emphasises the importance the
environment has on the problem domain. This approach is based on investigating what works
well in the surrounding environment of a problem to be solved.

Furthermore, the Soft Systems Methodology (SSM) focuses on analysing different aspects of a
problem and identifying the real causes of the problem before attempting any solutions. One of
the techniques used by SSM is the creation of root definitions by considering the problem as a
series of elements, including the concepts of customer, owner, environment and
transformation.

Show Me More

A series of video clips describing the Simplex method can be found at

https://www.youtube.com/watch?v=zDch07vVBxI