mb0048 | Mathematical Optimization | Linear Programming

Operations Research Unit 1

Sikkim Manipal University Page No. 1
Unit 1 Introduction to Operations Research
Structure:
1.1 Introduction
Objectives
1.2 Historical Background
Definitions of operations research
1.3 Scope of Operations Research
1.4 Features of Operations Research
1.5 Phases of Operations Research
1.6 Types of Operations Research Models
A broad classification of OR models
1.7 Operations Research Methodology
Definition
Construction
Solution
Validation
Implementation
1.8 Operations Research Techniques and Tools
1.9 Structure of the Mathematical Model
1.10 Limitations of Operations Research
1.11 Summary
1.12 Glossary
1.13 Terminal Questions
1.14 Answers
1.15 Case Study
1.1 Introduction
Welcome to the unit on operations research management. Operations
research management focuses on the mathematical scoring of
consequences of a decision aiming to optimise the use of time, effort, and
resources to avoid blunders. The act of obtaining best results under any
given circumstances is known as optimising. The key purpose of Operations
Research (OR) is to do preparative calculations that aid the decision-making
process.
Operations Research Unit 1
Sikkim Manipal University Page No. 2
Hence, decision-making is a key part of our daily life. The ultimate goal of all
decisions is to maximise benefits and to minimise effort and time. OR gives
decision makers the power to make effective decisions and improve day-today
operations. Decision makers consider all the available options, study
the outcomes, and estimate the risks.
In simple situations, common sense and judgement can be used to take
decisions. For example, if you are buying a microwave or washing machine,
the decision-making process is not very complicated. You can simply
compare the price, quality, and durability of the well-known brands and
models in the market and take a decision based on it.
However, in complex situations, although it is possible to take decisions
based on one‘s common sense, a decision backed by mathematical
calculations reduces the risk factor and increases the probability of success.
Some such situations, where decision-makers have to depend on
mathematical scoring and reasoning, are finding an appropriate product mix
amidst competitor‘s products or planning a public transportation network in a
city.
Objectives:
After studying this unit, you should be able to:
- describe the historical background of OR
- list the significant features of OR
- describe the methodology of OR
- define the structure of a mathematical model in OR
- describe the significance of the function of OR
1.2 Historical Background
During the World War II, scientists from United Kingdom studied the
strategic and tactical problems associated with air and land defence of the
country. The aim of this study was to determine the effective utilisation of
limited military resources to win the battle. The technique was named
operations research. After World War II, operations research techniques
were developed and deployed in the decision-making process in
complicated situations in various fields, such as industrial, academic, and
government organisations.
Operations Research Unit 1
Sikkim Manipal University Page No. 3
1.2.1 Definitions of operations research
Churchman, Aackoff, and Aruoff defined operations research as “the
application of scientific methods, techniques and tools to the operation of a
system with optimum solutions to the problems” where 'optimum' refers to
the best possible alternative.
The objective of OR is to provide a scientific basis to the decision-makers
for solving problems involving interaction with various components of the
organisation. This can be achieved by employing a team of scientists from
different disciplines to work together for finding the best possible solution in
the interest of the organisation as a whole. The solution thus obtained is
known as an optimal decision.
You can also define operations research as “The use of scientific methods
to provide criteria for decisions regarding man, machine, and systems
involving repetitive operations.”
Self Assessment Questions
1. The main objective of OR is to provide a _______ ________ to the
decision-makers.
2. OR employs a team of _________ from _________ __________.
1.3 Scope of Operations Research
Any problem, either simple or complicated, can use OR techniques to find
the best possible solution. This section will explain the scope of OR by
analysing its application in various fields of everyday life.
- In defence operations – In modern warfare, the three major military
components namely, Air Force, Army, and Navy carry out the defence
operations. The activities in each of these components can be further
divided in four sub-components - administration, intelligence, operations,
training and supply. The applications of modern warfare techniques in
each of the components of military organisations require expert
knowledge in respective fields. Furthermore, each component works to
drive maximum gains from its operations and there is always a
possibility that the strategy beneficial to one component may be
unfeasible for another component. Thus in defence operations, there is
a requirement to co-ordinate the activities of various components. This
Operations Research Unit 1
Sikkim Manipal University Page No. 4
gives maximum benefit to the organisation as a whole, having maximum
use of the individual components. A team of scientists from various
disciplines gets together to study the strategies of different components.
After appropriate analysis of the various courses of actions, the team
selects the best course of action, known as the ‗optimum strategy‘.
- In industry – The system of modern industries is so complex that an
individual cannot intuitively judge the optimum point of operation in its
various components. The business environment is always changing and
any decision useful at one time may not be suitable some time later.
There is always a need to check the validity of decisions continuously
against the situations. The industrial revolution with increased division of
labour and introduction of management responsibilities has made each
component an independent unit having its own goals. For example,
production department minimises the cost of production but maximises
output. Marketing department maximises the output, but minimises cost
of unit sales. Finance department tries to optimise the capital investment
and personnel department appoints good people at minimum cost. Thus,
each department plans its own objectives and all these objectives of
various departments or components come to conflict with one another
and may not agree to the overall objectives of the organisation. The
application of OR techniques helps in overcoming this difficulty by
integrating the diversified activities of various components to efficiently
serve the interest of the organisation as a whole. OR methods in
industry can be applied in the fields of production, inventory controls and
marketing, purchasing, transportation, and competitive strategies.
- Planning – In modern times, it has become necessary for every
government to carefully plan, for the economic development of the
country. OR techniques can be fruitfully applied to maximise the per
capita income, with minimum sacrifice and time. A government can thus
use OR for framing future economic and social policies.
- Agriculture – With increase in population, there is a need to increase
agriculture output. However, this cannot be done arbitrarily. There are
several restrictions. Hence, the need to determine a course of action
that serves the best under the given restrictions. You can solve this
problem by applying OR techniques.
Operations Research Unit 1
Sikkim Manipal University Page No. 5
- In hospitals – OR methods can solve waiting problems in outpatient
department of big hospitals and administrative problems of the hospital
organisations.
- In transport – Different OR methods can be applied to regulate the
arrival of trains and processing times, to minimise the passengers
waiting time and reduce congestion, and to formulate suitable
transportation policy, thereby reducing the costs and time of transshipment.
- Research and development – OR methodologies can be applied in the
field of R&D for several purposes, such as to control and plan product
introductions.
Self Assessment Questions
3. A government can thus use OR for framing future ______ and _______
4. In hospital OR methods can solve waiting problems in ______
department of big hospitals and ______ problems of the hospital
organisations.
1.4 Features of Operation Research
Some key features of OR are as follows:
- OR is system-oriented. OR scrutinises the problem from an
organisation‘s perspective. The results can be optimal for one part of the
system, while the same can be unfavourable for another part of the
system.
- OR imbibes an inter–disciplinary team approach. Since no single
individual can have a thorough knowledge of all the fast developing
scientific know-how, personalities from different scientific and
managerial cadre form a team to solve the problem.
- OR uses scientific methods to solve problems.
- OR increases effectiveness of the management‘s decision-making
ability.
- OR uses computers to solve large and complex problems.
- OR offers a quantitative solution.
- OR also takes into account the human factors.
Operations Research Unit 1
Sikkim Manipal University Page No. 6
Self Assessment Questions
5. OR ________ inter-disciplinary approach.
6. OR increases the effectiveness of ________ ability.
1.5 Phases of Operations Research
The scientific method in OR study generally involves three phases.
Figure 1.1 depicts the three phases of OR.
Fig. 1.1: Phases of Operations Research
Let us now study the phases in detail.
Judgment phase
This phase includes the following activities:
- Determination of the operations
- Establishment of objectives and values related to the operations
- Determination of suitable measures of effectiveness
- Formulation of problems relative to the objectives
Research phase
This phase utilises the following methodologies:
- Operation and data collection for a better understanding of the problems
- Formulation of hypothesis and model
- Observation and experimentation to test the hypothesis on the basis of
additional data
- Analysis of the available information and verification of the hypothesis
using pre-established measure of effectiveness
- Prediction of various results and consideration of alternative methods
Operations Research Unit 1
Sikkim Manipal University Page No. 7
Action phase
This phase involves making recommendations for the decision process. The
recommendations can be made by those who identify and present the
problem or by anyone who influences the operation in which the problem
has occurred.
Self Assessment Questions
7. Action phase involves making recommendations for the decision
process. (True/False)
8. One of the OR phases is judgement phase. (True/False)
1.6 Types of Operations Research Models
A model is an idealised representation or abstraction of a real-life system.
The objective of a model is to identify significant factors that affect the reallife
system and their interrelationships. A model aids the decision-making
process as it provides a simplified description of complexities and
uncertainties of a problem in a logical structure. The most significant
advantage of a model is that it does not interfere with the real-life system.
1.6.1 A broad classification of OR models
You can broadly classify OR models into the types depicted in figure 1.2.
Fig. 1.2: Classification of Models
Operations Research Unit 1
Sikkim Manipal University Page No. 8
Let us now study the models in detail.
Physical models
These models include all forms of diagrams, graphs, and charts. They are
designed to tackle specific problems. They bring out significant factors and
interrelationships in pictorial form to facilitate analysis. There are two types
of physical models. They are:
- Iconic models
- Analogue models
Let us now study the two types of physical models in detail.
Iconic models are primarily images of objects or systems, represented on a
smaller scale. These models can simulate the actual performance of a
product. Analogue models are small physical systems having characteristics
similar to the objects they represent, such as toys.
Mathematical or symbolic models
These models employ a set of mathematical symbols to represent the
decision variable of the system. The variables are related by mathematical
systems. Some examples of mathematical models are allocation,
sequencing, and replacement models.
By nature of environment
These models can be further classified as follows:
- Deterministic models - These are the models in which everything is
defined and the results are certain, such as an EOQ model.
- Probabilistic models - These are the models in which the input and
output variables follow a defined probability distribution, such as the
games theory.
By the extent of generality
These models can be further classified as follows:
- General models – These are the models which you can apply in general
to any problem. For example, linear programming.
- Specific models - These are the models that you can apply only under
specific conditions. For example, you can use the sales response curve
or equation as a function in the marketing function.
Operations Research Unit 1
Sikkim Manipal University Page No. 9
Self Assessment Questions
9. Diagram belongs to physical models. (True/False)
10. Allocation problems are represented by iconic models. (True/False)
1.7 Operations Research Methodology
The basic dominant characteristic feature of operations research is that it
employs mathematical representations or models to analyse problems. This
distinct approach represents an adaptation of the scientific methodology
used by the physical sciences. The scientific method translates a given
problem into a mathematical representation which is solved and
retransformed into the original context. Figure 1.3 depicts the OR approach
to problem solving.
Fig. 1.3: Steps in the OR methodology
As shown in figure 1.3, OR methodology consists of five steps. They are -
defining the problem, constructing the model, solving the model, validating
the model, and implementing the result.
Let us now study the steps in detail.
1.7.1 Definition
The first and the most important step in the OR approach of problem solving
is to define the problem. One needs to ensure that the problem is identified
properly because this problem statement will indicate the following three
major aspects:
- Description of the goal or the objective of the study
- Identification of the decision alternative to the system
Operations Research Unit 1
Sikkim Manipal University Page No. 10
- Recognition of the limitations, restrictions, and requirements of the
system
1.7.2 Construction
Based on the problem definition, you need to identify and select the most
appropriate model to represent the system. While selecting a model, you
need to ensure that the model specifies quantitative expressions for the
objective and the constraints of the problem in terms of its decision
variables. A model gives a perspective picture of the whole problem and
helps in tackling it in a well-organised manner. Therefore, if the resulting
model fits into one of the common mathematical models, you can obtain a
convenient solution by using mathematical techniques. If the mathematical
relationships of the model are too complex to allow analytic solutions, a
simulation model may be more appropriate. Hence, appropriate models can
be constructed.
1.7.3 Solution
After deciding on an appropriate model, you need to develop a solution for
the model and interpret the solution in the context of the given problem. A
solution to a model implies determination of a specific set of decision
variables that would yield an optimum solution. An optimum solution is one
which maximises or minimises the performance of any measure in a model
subject to the conditions and constraints imposed on the model.
1.7.4 Validation
A model is a good representation of a system. However, the optimal solution
must work towards improving the system‘s performance. You can test the
validity of a model by comparing its performance with some past data
available from the actual system. If under similar conditions of inputs, your
model can reproduce the past performance of the system, then you can be
sure that your model is valid. However, you will still have no assurance that
future performance will continue to duplicate the past behaviour. Secondly,
since the model is based on careful examination of past data, the
comparison should always reveal favourable results. In some instances, this
problem may be overcome by using data from trial runs of the system. One
must note that such validation methods are not appropriate for non-existent
systems because data will not be available for comparison.
Operations Research Unit 1
Sikkim Manipal University Page No. 11
1.7.5 Implementation
You need to apply the optimal solution obtained from the model to the
system and note the improvement in the performance of the system. You
need to validate this performance check under changing conditions. To do
so, you need to translate these results into detailed operating instructions
issued in an understandable form to the individuals who will administer and
operate the recommended system. The interaction between the operations
research team and the operating personnel reaches its peak in this phase.
1.8 Operations Research Techniques and Tools
The different techniques and tools used in OR are as follows:
- Linear programming – You can use linear programming to find a
solution for optimising a given objective. The objective may be to
maximise profit or to minimise cost. You need to ensure that both the
objective function and the constraints can be expressed as linear
expressions of decision variables. You will learn about the various uses
of linear programming in Unit 2.
- Inventory control methods – The production, purchasing, and material
managers are always confronted with questions, such as when to buy,
how much to buy, and how much to keep in stock. The inventory model
aims at optimising these inventory levels.
- Goal programming – In linear programming, you take a single objective
function and consider all other factors as constraints. However, in real
life there may be a number of important objective functions. Goal
programming has several objective functions, each having a target
value. Programming models are developed to minimise deviations from
these targets.
- Queuing model – The queuing theory is based on the concept of
probability. It indicates the capability of a given system and the changes
possible in the system when you modify the system. In formulating a
queuing model, you need not take into account all the constraints. There
is no maximisation or minimisation of an objective function. Therefore,
the application of queuing theory cannot be viewed as an optimisation
process. You can use the queuing theory to estimate the required
balance between customer waiting time and the service capability of the
Operations Research Unit 1
Sikkim Manipal University Page No. 12
system. You need to first consider several alternatives, evaluate them
through queuing models, study their effect on the system, and then
make a choice. The criteria for evaluation will be measures of efficiency
of the system, such as the average length of a queue, expected waiting
time of a customer, and the average time spent by the customer in the
system. In this approach, your success primarily depends on the
alternatives considered and not much on the queuing models
developed.
- Transportation model – The transportation model is an important class
of linear programs. The model studies the minimisation of the cost of
transporting a commodity from a number of sources to several
destinations. The supply at each source and the demand at each
destination are known. The objective of the model is to develop an
integral transportation schedule that meets all the demands from the
inventory at a minimum total transportation cost.
The transportation problem involves m sources, each of which has ai (i =
1, 2, …..,m) units of homogeneous product and n destinations available,
and each of which requires bj (j = 1, 2…., n) units of products. Here ai
and bj are positive integers. The cost cij of transporting one unit of the
product from the ith source to the jth destination is given for each
i and j. It is assumed that the total supply and the total demand are
equal.
¿ ¿
= =
=
n
j 1
m
i 1
ai bj (1)
Condition (1) is guaranteed by creating either a fictitious destination with
a demand equal to the surplus if total demand is less than the total
supply or a (dummy) source with a supply equal to the shortage if total
demand exceeds total supply. The cost of transportation from the
fictitious destination to all sources and from all destinations to the
fictitious sources are assumed to be zero so that the total cost of
transportation will remain the same.
- In addition to the above, there are tools such as the sequence model,
the assignment model, and network analysis, which you will learn in
detail in the later units.
Operations Research Unit 1
Sikkim Manipal University Page No. 13
Self Assessment Questions
11. OR methodology consists of definition, solution, and validation only.
(True/False)
12. The interaction between the OR team and management reaches peak
level in the implementation phase. (True/False)
1.9 Structure of the Mathematical Model
Many industrial and business situations are concerned with planning
activities. In each case of planning, there are limited sources, such as men,
machines, material, and capital at the disposal of the planner. One has to
take decisions regarding these resources to maximise production, minimise
the cost of production, or maximise the profit. These problems are referred
to as the problems of constrained optimisation.
Linear programming is a technique for determining an optimal schedule of
interdependent activities, for the given resources. Therefore, you can say
that programming refers to planning and the process of decision-making
about a particular plan of action from a given set of alternatives.
Any business activity or production activity to be formulated as a
mathematical model can best be discussed through its parts, which are as
follows:
- Decision variables
- Objective function
- Constraints
Let us now study the parts in detail.
Decision variables
Decision variables are the unknowns, which you need to determine from the
solution of the model. The parameters represent the controlled variables of
the system.
Objective function
The objective function defines the measure of effectiveness of the system
as a mathematical function of its decision variables. The optimal solution to
the model is obtained when the corresponding values of the decision
variable yield the best value of the objective function whilst satisfying all
constraints. Therefore, you can say that the objective function acts as an
indicator for the achievement of the optimal solution.
Operations Research Unit 1
Sikkim Manipal University Page No. 14
While formulating a problem, the desire of the decision-maker is expressed
as a function of ‗n‘ decision variables. (This function is a linear programming
problem, that is, each of its items will have only one variable raised to the
power one). Some of the objective functions in practice are:
- Maximisation of contribution or profit
- Minimisation of cost
- Maximisation of production rate or minimisation of production time
- Minimisation of labour turnover
- Minimisation of overtime
- Maximisation of resource utilisation
- Minimisation of risk to environment or factory
Constraints
To account for the physical limitations of the system, you need to ensure
that the model includes constraints, which limit the decision variables to their
feasible range or permissible values. These are expressed as constraining
mathematical functions.
For example, in chemical industries, there are restrictions from the
government regarding the releasing of gases into the environment.
Restrictions from sales department about the marketability of some products
are also treated as constraints. Thus, a linear programming problem has a
set of constraints in practice.
The mathematical models in OR may be viewed generally as determining
the values of the decision variables x J, where J = 1, 2, 3, ------ n, which will
optimise Z = f (x 1, x 2, ---- x n).
Subject to the constraints:
g i (x 1, x 2 ----- x n) ~ b i, i = 1, 2, ---- m
And xJ > 0 j = 1, 2, 3 ---- n where ~ is s, >, or =.
The function f is called the objective function, where xj ~ bi, represent the ith
constraint for i = 1, 2, 3 ---- m where bi is a known constant. The constraint xj
> 0 is called the non-negativity condition, which restricts the variables to
zero or positive values only.
Operations Research Unit 1
Sikkim Manipal University Page No. 15
Example, Diet problem
Formulate the mathematical model for the following:
Vitamin–A and Vitamin–B are found in food–1 and food–2.
One unit of food–1 contains 5 units of vitamin–A and 2 units of vitamin–B.
One unit of food–2 contains 6 units of vitamin–A and 3 units of vitamin–B.
The minimum daily requirement of a person is 60 units of vitamin–A and 80
units of Vitamin–B.
The cost per one unit of food–1 is Rs.5 and one unit of food–2 is Rs.6.
Assume that any excess units of vitamins are not harmful. Find the minimum
cost of the mixture (food–1 and food–2) which meets the daily minimum
requirements of vitamins.
Mathematical model of the diet problem:
Suppose, x1 = the number of units of food–1 in the mixture
x2 = the number of units of food–2 in the mixture.
Let us formulate the constraint related to vitamin-A. Since each unit of food–
1 contains 5 units of vitamin– A, we have that x1 units of food–1 contains 5x1
units of vitamin– A. Since each unit of food– 2 contains 6 units of vitamin–A,
we have that x2 units of food–2 contains 6x2 units of vitamin–A. Therefore,
the mixture contains 5x1 + 6x2 units of vitamin-A. Since the minimum
requirement of vitamin– A is 60 units, you can say that 5x1 + 6x2 > 60.
Now let‘s formulate the constraint related to vitamin–B. Since each unit of
food–1 contains 2 units of vitamin–B we have that x1 units of food–1
contains 2x1 units of vitamin-B. Since each unit of food–2 contains 3 units of
vitamin–B, we have that x2 units of food–2 contains 3x2 units of vitamin–B.
Therefore the mixture contains 2x1 + 3x2 units of vitamin–B. Since the
minimum requirement of vitamin–B is 80 units, you can say that
2x2 + 3x2 > 80
Next let‘s formulate the cost function. Given that the cost of one unit of
food–1 is Rs.5 and one unit of food–2 is Rs.6. Therefore, x1 units of food–1
costs Rs.5x1, and x2 units of food–2 costs Rs.6x2.
Therefore, the cost of the mixture is given by cost = 5x1 + 6x2.
If we write z for the cost function, then you can write z = 5x1 + 6x2.
Since cost has to be minimised, you can write min z = 5x1 + 6x2.
Operations Research Unit 1
Sikkim Manipal University Page No. 16
Since the number of units (x1 or x2) are always non-negative, you have x1 >
0, x2 > 0.
Therefore, the mathematical model is:
5x1 + 6x2 > 60
2x1 + 3x2 > 80
x1 > 0, x2 > 0, min z = 5x1 + 6x2.
1.10 Limitations of Operations Research
The limitations are more related to the problems of model building, time, and
money factors. The limitations are:
- Magnitude of computation – Modern problems involve a large number
of variables. The magnitude of computation makes it difficult to find the
interrelationship.
- Intangible factors – Non–quantitative factors and human emotional
factors cannot be taken into account.
- Communication gap – There is a wide gap between the expectations of
managers and the aim of research professionals.
- Time and money factors – When you subject the basic data to frequent
changes then incorporating them into OR models becomes a costly
affair.
- Human factor – Implementation of decisions involves human relations
and behaviour.
Self Assessment Questions
13. OR imbibes _________ team approach.
14. Linear programming is tool of _______.
15. The three phases of OR are ________.
16. To solve any problem through OR approach, the first step is _______.
17. _________ represents a real life system.
18. _________ represents the controlled variables of the system.
Operations Research Unit 1
Sikkim Manipal University Page No. 17
1.11 Summary
Let us recapitulate the important concepts discussed in this unit:
- The objective of OR is to provide a scientific basis to the decisionmakers
for solving problems involving interaction with various
components of the organisation
- The scope of OR is in various fields such as defence, industry,
government, agriculture, hospitals, transport and research and
development.
- Some key features of OR are OR is system oriented, it imbibes inter–
disciplinary team approach and uses scientific methods to solve
problems.
- The three phases of OR are Judgement phase, research phase, and
action phase.
- You can broadly classify OR models according to physical models,
mathematical models , models by nature of environment and models by
extent of generality.
- The steps in OR methodology are problem definition, model
construction, model solution, model validation and result
implementation.
- Linear programming, inventory control methods, goal programming,
queuing model and transportation model are the different tools and
techniques used in OR.
- Any business activity or production activity has to be formulated as a
mathematical model.
- Some of the limitations of OR are magnitude of computation, intangible
factors, communication gap, time and money factors and human factors
1.12 Glossary
Probability: possible outcomes of an event
Hypothesis: unproved theory
Network analysis: a mathematical representation of the problem by lines
and nodes joined to form a network
Operations Research Unit 1
Sikkim Manipal University Page No. 18
1.13 Terminal Questions
1. Define OR.
2. What are the features of OR?
3. What is a model in OR? Discuss the different models available in OR.
4. Write short notes on the different phases of OR.
5. What are the limitations of OR?
1.14 Answers
Self Assessment Questions
1. Scientific basis
2. Scientists, different disciplines
3. economic , social policies
4. Outpatient, administrative
5. Imbibes
6. Decision making
7. True
8. True
9. True
10. False
11. False
12. False
13. Inter-disciplinary
14. OR
15. Judgement phase, research phase, and action phase
16. Define the problem
17. Model
18. Parameters
Terminal Questions
1. OR is defined as the application of scientific methods, techniques and
tools to the operation of a system with optimum solutions to the
problems - refer 1.2.1
2. Some key features of OR are OR is system oriented, it imbibes inter–
disciplinary team approach and uses scientific methods to solve
problems - refer 1.4
3. Types of operations research models - refer 1.6
Operations Research Unit 1
Sikkim Manipal University Page No. 19
4. Phases of operations research are judgement, research and action
phase - refer 1.5
5. Limitations of OR - refer 1.10
1.15 Case Study
IMB Optimiser System
IBM was considering integrating its national network of spare parts
inventories to improve service support for their customers. They developed
a model for their inventory system that improved customer service while
reducing the value of IBM‘s inventories by $ 250 million and saving an
additional $20 million per year through improved operational efficiency. A
particularly interesting aspect of the model validation phase of this study
was the way the future users of the inventory system had been incorporated
into the testing process. Because these future users were sceptical about
the system being developed, representatives were appointed to a user team
to serve as advisors to the OR team. After a preliminary version of the new
system had been developed, a pre-implementation test of the system was
conducted. Extensive feedback from the user team led to major
improvements in the proposed system.
Large computer system was used to apply this model. The system
developed was called ‗optimiser.‘ It provided optimal control of service levels
and spare-parts inventories throughout IBM‘s U.S. parts distribution
network, which included two central automated warehouses, dozens of field
distribution centres, and many thousands of outstation locations. The parts
inventory maintained in this network is valued in billions of dollars. Optimiser
consists of four major modules.
- A forecasting system module contains a few programs for estimating the
failure rates of individual types of parts.
- A data delivery system module consists of approximately 100 programs
that process over 15 gigabytes of data to provide the input for the model.
- A decision system module then solves the model on a weekly basis to
optimise control of the inventories.
- The fourth module includes six programs that integrate the optimiser into
IBM‘s Parts Inventory Management System (PIMS). PIMS is a
Operations Research Unit 1
Sikkim Manipal University Page No. 20
sophisticated information and control system that contains millions of
lines of code.
Careful planning was required to implement the complex optimiser system
for controlling IBM‘s national network of spare-parts inventories. Three
factors proved to be especially important in achieving a successful
implementation. By the time implementation phase was reached,
operational managers had a strong sense of ownership and had become
ardent supporters for installing optimiser in their functional areas. A second
success factor was a very extensive user acceptance test whereby users
could identify problem areas that needed to be rectified prior to
implementation. The third key was that a new system was phased in
gradually, with careful testing at each phase, so the major bugs could be
eliminated before the system went live nationally.
Discussion Questions:
1. Analyse the need for OR team at IBM.
2. Explain optimiser with its modules.
3. How did this new system help and what were the reasons for its
success.
Reference:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
- Cohen M., & Kamesan P. V., & Kleindorfer P., & Lee H., & Tekerian A.
(Jan-Feb. 1990). “Optimizer: IBM’s Multi-Echelon Invenentory System
for Managing Service Logistics‖.
E- References:
- newagepublishers.com.
http://www.newagepublishers.com/samplechapter/001012.pdf.
Operations Research Unit 2
Sikkim Manipal University Page No. 21
Unit 2 Linear Programming
Structure:
2.1 Introduction
Objectives
2.2 Linear Programming Problem
2.3 Requirements of LPP
Basic assumptions of LPP
2.4 Mathematical Formulation of LPP
General form of LPP
Canonical forms
2.5 Case Studies of LPP
Solved problem
2.6 Applications of LPP
2.7 Advantages of LPP
2.8 Limitations of LPP
2.9 Summary
2.10 Glossary
2.11 Terminal Questions
2.12 Answers
2.13 Case Study
2.1 Introduction
In the previous unit, we dealt with operations research. We discussed the
background, scope, features, phases, types, limitations, techniques, and
tools of operations research. We also analysed the operations research
methodology and the structure of the mathematical model. In this unit, we
will deal with linear programming. Linear programming focuses on obtaining
the best possible output (or a set of outputs) from a given set of limited
resources.
Minimal time and effort and maximum benefit coupled with the best possible
output or a set of outputs is the mantra of any decision-maker. Today,
decision-makers or managements have to tackle the issue of allocating
limited and scarce resources at various levels in an organisation, in the best
possible manner. Man, money, machine, time, and technology are some of
Operations Research Unit 2
Sikkim Manipal University Page No. 22
the common resources. The management‘s task is to obtain the best
possible output (or a set of outputs) from these given resources.
The output can be measured in terms of profit, cost, social welfare, and the
overall effectiveness. In several situations, you can express the output (or a
set of outputs) as a linear relationship among several variables. The amount
of available resources can be expressed as a linear relationship among
various system variables. The management‘s dilemma is to optimise
(maximise or minimise) the output or the objective function subject to the set
of constraints. Optimisation of resources in which both the objective function
and the constraints are represented in a linear form is known as Linear
Programming Problem (LPP).
Objectives:
After studying this unit, you should be able to:
- explain linear programming problem
- analyse a feasible region
- analyse the applications of LPP
2.2 Linear Programming Problem
Linear Programming (LP) is a mathematical technique designed to help
managers in their planning and decision-making. It is usually used in an
organisation that is trying to make the most effective use of its resources.
Resources typically include machinery, manpower, money, time, warehouse
space, and raw materials.
A few examples of problems in which LP has been successfully applied are:
- Developments of a production schedule that will satisfy future demands
for a firm‘s product and at the same time minimise total production and
inventory costs.
- Establishment of an investment portfolio from a variety of stocks or
bonds that will maximise a company‘s return on investment.
- Allocation of a limited advertising budget among radio, TV, and
newspaper spots in order to maximise advertising effectiveness.
- Determination of a distribution system that will minimise total shipping
cost from several warehouses to various market locations.
Operations Research Unit 2
Sikkim Manipal University Page No. 23
- Selection of the product mix in a factory to make best use of machine
and man hours available while maximising the firm‘s profit.
2.3 Requirements of LPP
The common requirements of LPP are as follows.
- Decision variables and their relationship
- Well-defined objective function
- Existence of alternative courses of action
- Non-negative conditions on decision variables
Let us now study the basic assumptions of LPP.
2.3.1 Basic assumptions of LPP
The basic assumptions of LPP are:
- Linearity - Both objective functions and constraints can be expressed
as linear inequalities.
- Deterministic - All co-efficients of decision variables in the objective
and constraints expressions are known and finite.
- Additivity - The value of the objective function and the total amount of
each resource used must be equal to the sum of respective individual
contributions by decision variables.
- Divisibility - The solution values of decision variables can be nonnegative
values including fractions.
Self Assessment Questions
1. Both the objective function and constraints are expressed in _____
forms.
2. LPP requires existence of _______, _______, of _______.
3. Solution of decision variables can also be ____________.
2.4 Mathematical Formulation of LPP
The procedure for mathematical formulation of a linear programming
problem consists of the following major steps:
Operations Research Unit 2
Sikkim Manipal University Page No. 24
Fig. 2.1: Procedure of mathematical formulation of a linear
programming problem
Solved problem 1:
A company has three operational departments (weaving, processing, and
packing) with a capacity to produce three different types of clothes - suits,
shirts, and woollens, yielding a profit of Rs.2, Rs.4, and Rs.3 per meter
respectively. One meter of suiting requires 3 minutes of weaving, 2 minutes
of processing, and 1 minute of packing. Similarly, one meter of shirting
requires 4 minutes of weaving, 1 minute of processing, and 3 minutes of
packing. One meter of woollen requires 3 minutes in each department. In a
week, total run time of each department is 60, 40, and 80 hours for weaving,
processing, and packing respectively.
Formulate the linear programming problem to find the product mix to
maximise the profit.
Operations Research Unit 2
Sikkim Manipal University Page No. 25
Mathematical formulation:
The data of the problem is summarised below in table 2.1:
Table 2.1: summary of the data of the problem
Departments Profit
Weaving
(in min)
Processing
(in min)
Packing
(in min)
(Rs. per
meter)
Suits 3 2 1 2
Shirts 4 1 3 4
Woollens 3 3 3 3
Availability (min) 60*60 40*60 80*60
Let us solve this problem according to procedure outlined in fig. 2.1
Step 1: The key decision is to determine the weekly rate of production for
the three types of clothes.
Step 2: Let us designate the weekly production of suiting, shirting, and
woollens as x1 meters, x2 meters, and x3 meters respectively.
Step 3: As it is not possible to produce negative quantities, feasible
alternatives are sets of values of x1, x2, and x3 satisfying x1, x2, and
x3 ≥ 0.
Step 4: The constraints are the limited availability of three operational
departments. One meter of suiting requires 3 minutes of weaving.
The quantity being x1 meters, the requirement for suiting alone will
be 3x1 units. Similarly, x2 meters of shirting and x3 meters of
woollen will require 4x2 and 3x3 minutes respectively. Thus, the
total requirement of weaving will be 3x1+4x2+3x3, which should not
exceed the available 3600 minutes. Therefore, the labour
constraint becomes 3x1+4x2+3x3≤ 3600.
Similarly, the constraints for the processing department and
packing departments are 2x1+x2+3x3≤ 2400 and x1+3x2+3x3≤ 4800
respectively.
Step 5: The objective is to maximise the total profit from sales. Assuming
that whatever is produced is sold in the market, the total profit is
given by the linear relation z= 2x1+4x2+3x3.
Operations Research Unit 2
Sikkim Manipal University Page No. 26
The linear programming problem can thus be put in the following algebraic
format:
Find x1, x2, and x3 so as to maximise
Z = 2x1+4x2+3x3
Subject to constraints: 3x1+4x2+3x3≤ 3600
2x1+x2+3x3≤ 2400
3x1+3x2+3x3≤ 4800
x1, x2, and x3 ≥ 0
2.4.1 General form of LPP
The LPP is a class of mathematical programming where the functions
representing the objectives and the constraints are linear. Optimisation
refers to the maximisation or minimisation of the objective functions.
The general linear programming model can be defined as follows:
Maximise or minimise
Z = c1 x1 + c2 x2 + - - - - + cn xn
Subject to the constraints,
a11 x1 + a12 x2 + —– + a1n xn ~ b1
a21 x1 + a22 x2 + —– + a2n xn ~ b2
——————————————-
am1 x1 + am2 x2 + ——- + amn xn ~ bm
and x1 ≥ 0, x2 ≥ 0, ——————– xn ≥ 0
Where cj, bi, and aij (i = 1, 2, 3… m, j = 1, 2, 3 … n) are constants
determined from the technology of the problem and xj (j = 1, 2, 3 —- n) is the
decision variable. Here ~ is either ≤ (less than), ≥ (greater than), or =
(equal). In terms of the above formulation, the coefficients cj, bi, and aij are
interpreted physically as follows. If bi is the available amount of resources i,
where aij is the amount of resource that must be allocated to each unit of
activity j, the ―worth‖ per unit of activity is equal to cj.
Operations Research Unit 2
Sikkim Manipal University Page No. 27
2.4.2 Canonical forms
You can represent the general LPP mentioned above in the canonical form
as follows:
Maximise Z = c1 x1+c2 x2 + —— + cn
Subject to
a11 x1 + a12 x2 + —— + a1n xn ≤ b1
a21 x1 + a22 x2 + —— + a2n xn ≤ b2
——————————————–
am1 x1+am2 x2 + …… + amn xn ≤ bm
x1, x2, x3, … xn ≥ 0
The following are the characteristics of the canonical form.
- All decision variables are non-negative.
- All constraints are of ≤ type.
- The objective function is of the maximisation type.
Any LPP can be represented in the canonical form by using five elementary
transformations, which are as follows:
1. The minimisation of a function is mathematically equivalent to the
maximisation of the negative expression of this function. That is,
Minimise Z = c1 x1 + c2x2 + ……. + cn xn
is equivalent to
Maximise – Z = – c1x1 – c2x2 – … – cn xn.
2. Any inequality in one direction (≤ or ≥) may be changed to an inequality
in the opposite direction (≥ or ≤) by multiplying both sides of the
inequality by –1. For example,
2x1+3x2 ≤ 5 is equivalent to –2x1–3x2 ≥ –5.
3. An equation can be replaced by two inequalities in opposite direction.
For example,
2x1+3x2 = 5 can be written as 2x1+3x2 ≤ 5 and 2x1+3x2 ≥ 5 or -2x1-3x2 ≥
5 and – 2x1 – 3x2 ≤ – 5.
Operations Research Unit 2
Sikkim Manipal University Page No. 28
4. An inequality constraint with its left hand side in the absolute form can
be changed into two regular inequalities. For example,
|2x1+3x2| ≤ 5 is equivalent to 2x1+3x2 ≤ 5 and 2x1+3x2 ≥ – 5 or – 2x1– 3x2 ≤ 5.
5. The variable, which is unconstrained in sign, is equivalent to the
difference between two non-negative variables. For example, if x is
unconstrained in sign then x = (x+ – x–) where x+ ≥ 0, x– ≤ 0.
Case-let
An automobile company has two units X and Y, which manufactures
three different models of cars - A, B, and C. The company has to supply
1500, 2500, and 3000 cars of type A, B, and C respectively per week
(6 days). It costs the company Rs.1,00,000 and Rs.1,20,000 per day to
run the units X and Y respectively. In a day, unit X manufactures 200,
250, and 400 cars and unit Y manufactures 180, 200, and 300 cars of A,
B, and C respectively. The operations manager has to decide on how
many days per week should each unit be operated to meet the current
demand at minimum cost.
The operations manager along with his/her team uses a LPP model to
arrive at the minimum cost solution.
2.5 Case Studies of LPP
Solved problem 1
A firm engaged in producing 2 models - model A and model B, performs
only 3 operations – painting, assembly, and testing. Table 2.1 depicts the
relevant data.
Table 2.1: Unit Sale Price and Hours Required for Each Unit
Unit sale price
Hours required for each unit
Assembly Painting Testing
Model A Rs. 50.00
Model B Rs. 80.00
1.0
1.5
0.2
0.2
0.0
0.1
Total numbers of hours available each week are as under assembly 600,
painting 100, and testing 30. The firm wishes to determine the weekly
product-mix to maximise revenue.
Operations Research Unit 2
Sikkim Manipal University Page No. 29
Solution: Let us first write the notations as under:
Z: Total revenue
x1: Number of Units of Model A
x2: Number of Units of Model B
x1, x2: Decision variables
Since the objective (goal) of the firm is to maximise its revenue, the
model can be stated as follows. The objective function Z = 50x1 + 80x2 is
to be maximised subject to the following constraints:
1.0x1 + 1.5x2 ≤ 600, (assembly constraints)
0.2x1+0.2x2 ≤ 100, (painting constraints)
0.0x1+0.1x2 ≤ 30, (testing constraints)
x1≥0, x2≥0, (the non-negativity constraints)
Solved problem 2
A milk distributor supplies milk in bottles to houses in three areas A, B,
and C in a city. His delivery charge per bottle is 30 paise in area A, 40
paise in area B, and 50 paise in area C. He has to spend on an average,
1 minute to supply one bottle in area A, 2 minutes per bottle in area B,
and 3 minutes per bottle in area C. He can spare only 2 hours 30 minutes
for milk distribution but not more than 1 hour 30 minutes for area A and B
together. He can deliver at the most 120 bottles.
Find the number of bottles that he has to supply in each area so as to
maximise the supply. Construct a mathematical model.
Operations Research Unit 2
Sikkim Manipal University Page No. 30
Solved problem 3
An oil company has two units A and B, which produce three different
grades of oil – super fine, medium, and low grade. The company has to
supply 12, 8, and 24 barrels of super fine, medium, and low-grade oil
respectively per week. It costs the company Rs.1000 and Rs.800 per day
to run the units A and B respectively.
In a day, unit A produces 6, 2, and 4 barrels and unit B produces 2, 2,
and 12 barrels of super fine, medium, and low grade respectively. The
manager has to decide on how many days per week should each unit be
operated in order to meet the requirement at minimum cost.
Formulate the LPP model.
Solution: Table 2.2 depicts the summary of the given data.
Table 2.2: Capacity and Requirements Details of an Oil Company
Product Capacity Requirements
Super fine
Medium
Low grade
Cost
Unit A Unit B
12
8
24
-
6
2
4
Rs. 1000
2
2
12
Rs. 800
Let x1 and x2 be the number of days that unit A and B must be operated
per week respectively. Then the objective of the manager is to:
Minimise the cost function
Z = 1000 x1 + 800 x2
Subject to the constraints 6x1 + 2x2 ≥ 12 (super fine)
2x1 + 2 x2 ≥ 8 (medium)
4x1 + 12 x2 ≥ 24 (low grade)
and x1 >0, x2 ≥ 0
2.5.1 Solved problem
A toy company manufactures two types of doll, a basic version-doll A and a
deluxe version-doll B. Each doll of type B takes twice as long to produce
one doll of type A. Thus, the company would have time to make a maximum
Operations Research Unit 2
Sikkim Manipal University Page No. 31
of 2000 types of A dolls per day. The supply of plastic is sufficient to
produce 1500 dolls per day (both A and B combined). The deluxe version
requires a fancy dress of which only 600 is available per day. If the company
makes a profit of Rs.3.00 and Rs.5.00 per doll, respectively on doll A and B,
then how many of each doll should be produced per day in order to
maximise the total profit. Formulate the problem.
Formulation
Let x1 and x2 be the number of dolls produced per day of type A and B
respectively.
Let doll A require t hrs.
So, doll B require 2t hrs.
The total time to manufacture x1 and x2 dolls should not exceed 2,000t hrs.
Therefore, tx1+2tx2 ≤ 2000t
Other constraints are simple. The linear programming problem becomes:
Maximise p= 3x1+5x2
Subject to restrictions
x1+2x2 ≤ 2000(time constraint)
x1+x2 ≤ 1500(plastic constraint)
x2 ≤ 600 (dress constraint)
and non-negativity restrictions
x1≥ 0, x2≥0
Self Assessment Questions
4. One of the characteristics of canonical form in the objective function
must be maximisation. (True/False)
5. 2x – 3y ≤ 10 can be written as -2x + 3y ≥-10. (True/False)
Operations Research Unit 2
Sikkim Manipal University Page No. 32
2.6 Applications of LPP
Linear programming is a powerful tool for selecting alternatives in a decision
problem. Consequently, it has been applied in a wide variety of problem
settings. We will indicate a few applications covering the major functional
areas of a business organisation.
- Finance - The problem of the investor could be a portfolio-mix selection
problem. In general, the number of different portfolios can be much
larger than what the example indicates. More and different kinds of
constraints can be added. Another decision problem involves
determining the mix of funding for a number of products when more than
one method of financing is available. The objective may be to maximise
total profits, where the profit for a given product depends on the method
of financing. For example, funding may be done with internal funds,
short-term debt, or intermediate financing (amortised loans). There may
be a limitation on the availability of each of the funding options as well
as financial constraints requiring certain relationships between the
funding options so as to satisfy the terms of bank loans or intermediate
financing. There may also be limitations on the production capacity for
the products. The decision variables would be the number of units of
each product to be financed by each funding option.
- Production and operations management - Quite often, in the process
industries, a given raw material can be converted into a wide variety of
products. For example, in the oil industry, crude oil is refined into
gasoline, kerosene, home-heating oil, and various grades of engine oil.
Given the present profit margin on each product, the problem is to
determine the quantities of each product that should be produced. The
decision is subject to numerous restrictions such as limits on the
capacities of various refining operations, raw-material availability,
demands for each product, and any government-imposed policies on the
output of certain products. Similar problems also exist in the chemical
and food-processing industries.
- Human resources - Personnel planning problems can also be analysed
with linear programming. For example, in the telephone industry,
demands for the services of installer-repair personnel are seasonal. The
problem is to determine the number of installer-repair personnel and
line-repair personnel required each month as part of the work force,
Operations Research Unit 2
Sikkim Manipal University Page No. 33
considering the total costs of hiring, layoff, overtime, and regular-time
wages are minimised. The constraints set includes restrictions on the
service demands that must be satisfied, overtime usage, union
agreements, and the availability of skilled people for hire. This example
runs contrary to the assumption of divisibility. However, the work-force
levels for each month would normally be large enough that rounding off
to the closest integer in each case would not be detrimental, provided
the constraints are not violated.
- Marketing - Linear programming can be used to determine the proper
mix of media to use in an advertising campaign. Suppose that the
available media are radio, television, and newspapers. The problem is to
determine how many advertisements to place in each medium. Of
course, the cost of placing an advertisement depends on the medium
chosen. We wish to minimise the total cost of the advertising campaign,
subject to a series of constraints. Since each medium may provide a
different degree of exposure of the target population, there may be a
lower bound on the total exposure from the campaign. Also, each
medium may have a different efficiency rating in producing desirable
results. Thus there may be a lower bound on efficiency. In addition,
there may be limits on the availability of each medium for advertising.
- Distribution - Another application of linear programming is in the area of
distribution. Consider a case in which there are m factories that must
ship goods to n warehouses. A given factory could make shipments to
any number of warehouses. Given the cost to ship one unit of product
from each factory to each warehouse, the problem is to determine the
shipping pattern (number of units that each factory ships to each
warehouse) that minimises total costs. This decision is subject to the
restrictions that demand at each factory cannot be more than it has the
capacity to produce.
Self Assessment Questions
6. Personnel planning problems can also be analysed with ___________.
7. Linear programming is a powerful tool for selecting alternatives in a
_________________.
8. Linear programming can be used to determine the _______________ to
use in an advertising campaign.
Operations Research Unit 2
Sikkim Manipal University Page No. 34
9. Quite often in the process industries, a given _________ can be made
into a wide variety of products.
2.7 Advantages of LPP
The advantages of linear programming techniques may be outlined as
follows:
- Linear programming technique helps in making the optimum utilisation of
productive resources. It also indicates how decision makers can employ
productive factors most effectively by choosing and allocating these
resources.
- The quality of decisions may also be improved by linear programming
techniques. The user of this technique becomes more objective and less
subjective.
- Linear programming technique provides practically applicable solutions
because there might be other constraints operating outside the problem.
These constraints must also be taken into consideration. Just because
so many units must be produced does not mean that all those can be
sold. So the necessary modification of its mathematical solution is
required for the sake of convenience to the decision maker.
- In production processes, high lighting of bottlenecks is the most
significant advantage of this technique. For example, when bottlenecks
occur, some machines cannot meet the demand while others remain idle
for some time.
2.8 Limitations of LPP
In spite of wide area of applications, some limitations are associated with
linear programming techniques. These are stated below:
- In some problems, objective functions and constraints are not linear.
Generally, in real life situations concerning business and industrial
problems, constraints are not linearly treated as variables
- There is no guarantee of integer valued solutions. For example, in
finding out how many men and machines would be required to perform a
particular job, rounding off the solution to the nearest integer will not give
an optimal solution. Integer programming deals with such problems.
- A linear programming model does not take into consideration the effect
of time and uncertainty. Thus the model should be defined in such a way
Operations Research Unit 2
Sikkim Manipal University Page No. 35
that any change due to internal as well as external factors can be
incorporated.
- Sometimes, large-scale problems cannot be solved with linear
programming techniques even when the computer facility is available.
Such difficulty may be removed by breaking the main problem into
several small problems and then solving them separately.
- Parameters appearing in the model are assumed to be constant. But, in
real life situations they are neither constant nor deterministic.
- Linear programming deals with only single objective, whereas in real life
situations, problems come across with multi-objectives. Goal
programming and multi-objective programming deal with such problems.
2.9 Summary
Let us recapitulate the important concepts discussed in this unit:
- In LPP, first identify the decision variables with economic or physical
quantities whose values are of interest to the management.
- The problems must have a well-defined objective function expressed in
terms of the decision variable.
- The objective is to maximise the resources when it expresses profit or
contribution. If the objective function indicates cost, then it has to be
minimised.
- The decision variables interact with each other through some
constraints. These constraints arise due to limited resources, stipulation
on quality, technical, legal, or a variety of other reasons.
2.10 Glossary
Decision variables - set of unknown quantities in LPP
Bottleneck - A point in the production process that causes delay.
2.11 Terminal Questions
1. Explain LPP by giving two examples.
2. Write any three applications of LPP.
3. What are the advantages of LPP?
4. Explain the basic assumptions of LPP.
5. State the general form of LPP.
Operations Research Unit 2
Sikkim Manipal University Page No. 36
6. Three grades of coal A, B, and C contain ash and phosphorous as
impurities. In a particular industrial process, a fuel obtained by blending
the above grades containing not more than 25% ash and 0.03%
phosphorous is required. The maximum demand of the fuel is 100 tons.
Percentage impurities and costs of the various grades of coal are shown
below. Assuming that there is an unlimited supply of each grade of coal
and there is no loss in blending, formulate the blending problem to
minimise the cost.
Coal grade Ash% Phosphorus% Cost per ton in Rs.
A 30 0.02 240
B 20 0.04 300
C 35 0.03 280
7. A firm manufactures two products. The net profit on product 1 is Rs.3
per unit and the net profit on product 2 is Rs.5 per unit. The
manufacturing process is such that each product has to be processed
in two departments D1 and D2. Each unit of product 1 requires
processing for 1 minute at D1 and 3 minutes at D2; each unit of product
2 requires processing for 2 minutes at D1 and 2 minutes at D2. Machine
time available per day is 860 minutes at D1 and 1200 minutes at D2.
How much of product 1 and product 2 should be produced every day so
that the total profit is maximum. Formulate this as a problem in LPP.
2.12 Answers
Self Assessment Questions
1. Linear
2. Alternate course of action
3. Fractions
4. True
5. True
6. Linear programming
7. Decision problem
8. Proper mix of media
9. Raw material
Operations Research Unit 2
Sikkim Manipal University Page No. 37
Terminal Questions
1. LPP is a mathematical technique for planning and decision making.
Examples: development of production schedule and establishment of an
investment portfolio. For more information refer 2.2
2. Applications of LPP in the area of finance, productions and operations
management, human resources, marketing and distributions. For more
information refer 2.6
3. Advantages of LPP: optimum utilisation of productive resources,
improves quality of decisions, provides practically applicable solutions.
For more information refer 2.7
4. Basic assumptions of LPP: Linearity, deterministic, additivity, divisibility
etc. For more information refer 2.3 and 2.3.1
5. The general linear programming model can be defined as follows:
Maximise or minimise Z = c1 x1 + c2 x2 + - - - - + cn xn
For more information refer 2.4.1
6. [Hint: x1= tons of grade A coal, x2= tons of grade B coal & x3= tons of
grade C coal]. For more information refer 2.4 and 2.5.
7. Maximise 3x1 + 5x2, subject to x1 + 2x2 ≥ 800 (minutes)
3x1 + 2x2 ≥ 1200 (minutes) x1, x2 ≥ 0 For more information refer 2.4
and 2.5.
2.12 Case Study
Airport Taxi Service
A transport company situated in New Delhi finds great demand for transport
services from the airport. Though there is taxi service from the airport, there
is yet heavy demand for more transport. Hence, the company is considering
the purchase of new vehicles for transportation between Delhi airport and
hotels in the city. After studying the different types of vehicles available,
there were three vehicles under consideration – station wagons, mini buses,
and large buses. The purchase price would be Rs. 2,45,000 for each station
wagon, Rs. 3,50,000 for a mini bus, and Rs. 5,00,000 for a large bus. The
Board of Directors have authorised a maximum amount of Rs. 50,00,000 for
these purchases. Because of the heavy air travel involved, the new vehicles
would be utilised at maximum capacity, regardless of the type of vehicles
Operations Research Unit 2
Sikkim Manipal University Page No. 38
purchased. The expected net annual profit would be Rs. 15,000 for the
station wagon, Rs. 35,000 for the mini bus, and Rs. 45,000 for the large bus.
The company has hired 30 new drivers for the new vehicles. They are
qualified drivers for all the three types of vehicles.
The maintenance department has the capacity to handle an additional 80
station wagons. A mini bus is equivalent to 5/3 station wagons and each
large bus is equivalent to two station wagons in terms of their use of the
maintenance department.
Discussion Questions:
1. Formulate a linear programming model, which would help to determine
the optimum number of each type of vehicle to be purchased in order to
maximise profit.
[Hint: x1= no. of station wagons to purchase, x2= no. of minibus to
purchase & x3= no. of large buses to purchase].
References :
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004). Operation
Research. Sultan Chand and Sons.
E-References:
- Linear Optimization.
http://www.mirrorservice.org/sites/home.ubalt.edu/ntsbarsh/Businessstat/
opre/partVIII.htm.
Operations Research Unit 3
Sikkim Manipal University Page No. 39
Unit 3 Graphical Analysis of
Linear Programming Problems
Structure:
3.1 Introduction
Objectives
3.2 Graphical Analysis
3.3 Some Basic Definitions
3.4 Graphical Methods to Solve LPP
Working rule
Classification of feasible points
Example on maximisation of LPP
Solved problems on mixed constraints LP problem
3.5 Some Exceptional Cases
Solved problem for alternative optimal solution
Solved problem for unbounded solution
Solved problem for inconsistent solution
3.6 Important Geometric Properties of LPP
3.7 Summary
3.8 Glossary
3.9 Terminal Questions
3.10 Answers
3.11 Case Study
3.1 Introduction
In the previous unit, we dealt with Linear Programming Problem (LPP). We
analysed the definition, requirements, mathematical formulation, case
studies, applications, advantages, and disadvantages of LPP. In this unit,
we will deal with the graphical analysis of LPP. Today, many resources
needed as inputs to operations are in limited supply. Operations managers
must understand the impact of this situation on meeting their objectives.
Linear Programming (LP) is one way through which the operations
managers can determine how best to allocate their scarce resources.
Once the linear programming model has been formulated on the basis of the
given objective and associated constraint functions, the next step is to solve
Operations Research Unit 3
Sikkim Manipal University Page No. 40
the problem and obtain the best possible or the optimal solution using
various mathematical and analytical techniques that can be employed for
solving the linear programming model.
The graphical solution procedure is one of the methods of solving two
variable linear programming problems.
Objectives:
After studying this unit, you should be able to:
- solve linear programming problems graphically
- identify some exceptional cases involved in LPP
- analyse the important geometric properties of LPP
3.2 Graphical Analysis
You can analyse linear programming with two decision variables graphically.
Example
Let us consider the following illustration:
Maximise z = 700 x1+500 x2
Subject to
4x1+3x2 ≤ 210
2x1+x2 ≤ 90
x1 ≥ 0, x2 ≥ 0
Let the horizontal axis represent x1 and the vertical axis x2 as shown in
figure 3.1. First, draw the line 4x1 + 3x2 = 210, (by replacing the inequality
symbols by the equality) which meets the x1-axis at the point A (52.50, 0)
(put x2 = 0 and solve for x1 in 4x1 + 3x2 = 210) and the x2–axis at the point B
(0, 70) (put x1 = 0 in 4x1 + 3x2 = 210 and solve for x2).
Operations Research Unit 3
Sikkim Manipal University Page No. 41
Fig. 3.1: Linear Programming with 2 Decision Variables
Any point on the line 4x1+3x2 = 210 or inside the shaded portion will satisfy
the restriction of the inequality, 4x1+3x2 <= 210. Similarly, the line
2x1+x2 = 90 meets the x1-axis at the point C(45, 0) and the x2 – axis at the
point D(0, 90) as shown in figure 3.2.
Fig. 3.2: Linear Programming with 2 Decision Variables
Operations Research Unit 3
Sikkim Manipal University Page No. 42
Figure 3.3 depicts how the area can be sketched by combining the two
graphs.
Fig. 3.3: Feasible Region
The three constraints including non-negativity are satisfied simultaneously in
the shaded region OCEB. This region is called feasible region.
3.3 Some Basic Definitions
The table 3.1 gives the some basic definitions of the terms in graphical
analysis.
Table 3.1: Basic Definitions in Graphical Analysis
No. Terms Definitions
1.
Feasible
Region
Any non-negative value of (x1, x2) (i.e., x1 ≥ 0, x2 ≥ 0) is a
feasible solution of the LPP if it satisfies all the constraints.
The collection of all feasible solutions is known as the
feasible region.
2. Convex
A set X is convex if for any point x1, x2 in X, the line
segment joining these points is also in X.
By convention, a set containing only a single point is also
a convex set.
Operations Research Unit 3
Sikkim Manipal University Page No. 43
3. Half Plane
A linear inequality in two variables is known as a half
plane. The corresponding equality or the line is known as
the boundary of the half plane.
4.
Convex
Polygon
A convex polygon is a convex set formed by the intersection
of finite number of closed half-planes. Refer figure
3.4 and figure 3.5
5.
Redundant
Constraint
A redundant constraint is a constraint which does not
affect the feasible region.
6.
Basic
Solution
A basic solution of a systems of m equations and n
variables (m < n) is a solution where at least n-m variables
are zero.
7.
Basic
Feasible
Solution
A basic feasible solution of a system of m equations and n
variables is where all basic variables are non-negative.
8.
Optimal
Feasible
Solution
Any feasible solution that optimises the objective function
is called an optimal feasible solution.
Note: The objective function is maximised or minimised at one of the
extreme points referred to as optimum solution. Extreme points are referred
to as vertices or corner points of the convex regions.
Self Assessment Questions
1. The collection of all feasible solutions is known as the ________
region.
2. A linear inequality in two variables is known as a _________.
Operations Research Unit 3
Sikkim Manipal University Page No. 44
3.4 Graphical Methods to Solve LPP
While obtaining the optimal solution to an LPP by the graphical method, the
statement of the following theorems of linear programming is used:
- The collection of all feasible solutions to an LPP constitutes a convex set
whose extreme points correspond to the basic feasible solutions.
- There are a finite number of basic feasible regions within the feasible
solution space.
- If the convex set of the feasible solutions of the system of simultaneous
equation is a convex polyhedron, then at least one of the extreme points
gives an optimal solution.
- If the optimal solution occurs at more than one extreme point, the value
of the objective function will be the same for all convex combination of
these extreme points.
3.4.1 Working rule
The method of solving an LPP on the basis of the above analysis is known
as the graphical method. The working rule for the method is as follows.
Step 1: Formulate the problem in terms of a series of mathematical
equations representing objective function and constraints of LPP.
Step 2: Plot each of the constraints equation graphically. Replace the
inequality constraint equation to form a linear equation. Plot the
equations on the planar graph with each axis representing
respective variables.
Step 3: Identify the convex polygon region relevant to the problem. The
area which satisfies all the constraints simultaneously will be the
feasible region. This is determined by the inequality constraints.
Step 4: Determine the vertices of the polygon and find the values of the
given objective function Z at each of these vertices. Identify the
greatest and the least of these values. These are respectively the
maximum and minimum value of Z.
Step 5: Identify the values of (x1, x2) which correspond to the desired
extreme value of Z. This is an optimal solution of the problem.
Operations Research Unit 3
Sikkim Manipal University Page No. 45
3.4.2 Classification of the feasible points
Figure 3.6 depicts the feasible points of any non-empty LP feasible region,
which can be classified as, interiors, boundaries, or vertices.
Source: http://www.mirrorservice.org/sites/home.ubalt.edu/ntsbarsh/
Business-stat/opre/partVIII.htm
Fig. 3.6: Classification of Feasible Region Points
The point B in the above two-dimensional figure, for example, is a
boundary point of the feasible set because every small circle is centred at
the point B. However small it may be, it contains both points, some inside
the set and some points outside the set. The point I is an interior point
because the I circle, all smaller circles, and some larger ones contain
exclusively points in the set. The collection of boundary points belongs to
boundary line (segment), e.g. the line segment cd. The intersections of
boundary lines (segments) are called the vertices; if feasible it is called the
corner point. In three-dimensional space and higher, the circles become
spheres and hyper-spheres.
It must be noted that the LP constraints provide the vertices and the cornerpoints.
A vertex is the intersection of 2-lines, or in general n-hyper planes in
LP problems with n-decision variables.
Operations Research Unit 3
Sikkim Manipal University Page No. 46
3.4.3 Example on maximisation of LPP
Solved problem 1
Solve the given LPP using the graphical method. Maximise z =50x1 + 80x2,
subject to the constraints
1.0x1 + 1.5x2 ≤ 600 1
0.2 x1 +0.2x2 ≤ 100 2
0.0x1 + 0.1x2 ≤ 30 3
and x1 ≥ 0, x2 ≥ 0
Solution
The horizontal axis represents x1 and the vertical axis x2. Plot the constraint
lines and mark the feasibility region as depicted in figure 3.7.
The shaded region is the feasibility region carried out by the constraints
operating on the objective function. This depicts the limits within which the
values of the decision variables are permissible.
The intersection point C and D can be solved by the linear equations
x2 = 30; x1 + 1.5 x2 = 600, and 0.2x1 + 0.2x2 = 100 and x1 + 1.5x2 = 600 that
is, C (150, 300) and D(300, 200).
The next step is to maximise revenues subject to the feasible region shown
in figure 3.7.
Fig. 3.7: Showing Feasible Region
Operations Research Unit 3
Sikkim Manipal University Page No. 47
Final solution: A(0,300) Z= 24000
B(500,0) Z= 25000
C(150, 300) Z= 31500
D(300,200) Z= 31000
We choose point C (150, 300) as the final solution which maximises the
given problem.
3.4.4 Solved problems on mixed constraints LP problem
Solved problem 2
Use the graphical method to solve the following LPP.
Minimise z = 20 x + 10 y, subject to the constraints
x + 2y ≤ 40 1
3 x + y ≥ 30 2
4x + 3y ≥ 60 3
and x, y ≥ 0
Solution: Plot each constraint on a graph by first writing it as a linear
equation. Then use the inequality condition of each constraint to mark the
feasible region as depicted in figure 3.8.
The coordinates of the extreme points of the feasible region are: A= (15, 0),
B= (40, 0), C= (4, 18), and D= (6, 12). The values of the objective function at
each of these extreme points are as depicted in table 3.2:
Table 3.2: Extreme Point Values of the Objective Function
Extreme
Point
Co-ordinates (x, y)
Objective Function Value,
z = 20x + 10 y
A (15, 0) 20 (1) + 10 (0) = 300
B (40,0) 20 (40) + 10 (0) = 800
C (4,18) 20 (4) + 10 (18) = 260
D (6,12) 20 (6) + 10 (12) = 240
The minimum value of the objective function Z = 240 occurs at the extreme
point D (6, 12). Hence the optimal solution is x = 6, y = 12, and min Z = 240.
Operations Research Unit 3
Sikkim Manipal University Page No. 48
Fig. 3.8: Showing Feasible Region
3.5 Some Exceptional Cases
In the section 3.4, we discussed some linear programming problems which
may be called as ‗well-behaved problems‘. In each case, a solution was
obtained, in some cases it took less effort while in some others it took a little
more. But a solution was finally obtained. It should not be taken as a rule.
There may be an LPP for which no solution exists or for which the only
solution obtained is an unbounded one. Though such problems seldom
occur in real situations, it will be an omission, if at this stage the reader is
not exposed to such exceptional cases.
Here we consider four different cases that arise in the application of
graphical method.
- A unique optimal solution - In this case, only one optimum solution will
be obtained in a graphical solution method.
- Multiple optimal solutions/Alternative solution -: When the objective
function is parallel to a binding constraint (i.e., a constraint that is
satisfied as an equation by the optimal solution), the objective function
will assume the same optimum value at more than one solution point.
For this reason, they are called alternative optima.
- An unbounded solution - When the values of the decision variables
may be increased indefinitely without violating any of the constraints, the
Operations Research Unit 3
Sikkim Manipal University Page No. 49
feasible region is unbounded. The value of objective function, in such
cases, may increase (for maximisation) or decrease (for minimisation)
indefinitely. Thus, both the solution space and the objective function
value are unbounded.
- Infeasible problem - When the constraints cannot satisfy
simultaneously, the linear programming problem has no feasible
solution.
3.5.1 Solved problem for alternative optimal solution
Solved problem 3
Use graphical method to solve the following LPP.
Maximise z = 2 x + 4 y, subject to the constraints:
x + 2 y ≤ 5 1
x + y ≤ 4 2
x, y ≥ 0
Solution
The problem is depicted graphically in figure 3.9. The extreme points of
the feasible region are O, A, B, and C. We observe that our objective
function is parallel to the line BC, which forms the boundary of the
feasible region. Thus, as we move the objective function away from the
origin, it coincides with the portion BC of the constraint line which forms
the boundary of the feasible region. This shows that any point including
the extreme points B and C on the same line between B and C is an
optimal solution. Therefore, in fact an infinite number of values of x and y
give the same values of objective function.
Now, the value of objective function at each of the extreme points is
evaluated as follows:
Table 3.3: Extreme Point Values of the Objective Function
Extreme
Point
Co ordinates
(x, y)
Objective Function Value,
z = 2x + 4 y
O (0, 0) 0
A (4,0) 8
B (3,1) 10 (Max)
C (0,2.5) 10 (Max)
Since any point on the line segment BC gives the maximum value (z=10)
of the objective function, there exists an alternative optima.
Operations Research Unit 3
Sikkim Manipal University Page No. 50
Fig. 3.9: Graphical Representation of Solved Problem 3
3.5.2 Solved problem for unbounded solution
Solved problem 4
Solve the given LPP using graphical method. Maximum z= 3x1+2x2
Subject to x1-x2≤1, x1+ x2≥3, and x1, x2≥0
Graphical solution: Figure 3.10 depicts the region of feasible solution,
which is the shaded area.
Fig. 3.10: Feasible Region of Two-dimensional LP
Operations Research Unit 3
Sikkim Manipal University Page No. 51
It is clear from this figure that the line representing the objective
function can be moved far even parallel to itself in the direction of
increasing z, and still has some points in the region of feasible
solutions.
Hence z can be made arbitrarily large, and the problem has no finite
maximum value of z. Such problems are said to have unbounded
solutions.
Infinite profit in practical problems of linear programming cannot be
expected. If LPP has been formulated by committing some mistake, it
may lead to an unbounded solution.
3.5.3 Solved problem for inconsistent solution
Solved problem
Solve the given LPP using graphical method
Maximise z= 3x1-2x2
Subject to constraints x1+x2 ≤1, 2x1+2x2≥ 4, and x1,x2≥0
Solution: Figure 3.11 depicts the problem graphically.
Fig. 3.11: Feasible Region of Two Dimensional LPP
The figure shows that there is no point in (x1, x2) which satisfies both the
constraints simultaneously. Hence the problem has no solution because
the constraints are inconsistent.
Operations Research Unit 3
Sikkim Manipal University Page No. 52
Self Assessment Questions
3. The feasible region is a convex set. (True/False)
4. The optimum value occurs anywhere in feasible region. (True/False)
5. An unbounded solution means that there exist a finite number of
solutions to the given problem. (True/False)
3.6 Important Geometric Properties of LPP
Geometric properties of LP problems, which are observed while solving
them graphically, are summarised below:
- The feasible region has an important property which is called the
convexity property of geometry, provided the feasible solution for the
problem exists.
- Convexity means that the region of feasible solutions has no holes in
them, that is, they are solids, and they have no cuts as on the boundary.
- This fact can be expressed more precisely by saying that the line joining
any two points in the region also lies in the region.
- The boundaries of the feasible regions are lines or planes.
- There are corners or extreme points on the boundary and there are
edges joining various corners.
- The objective function can be represented by a line or a plane for any
fixed value of Z.
- At least one corner of the region of feasible solutions will be an optimal
solution whenever the maximum or minimum value of z is finite.
- If the optimal solution is not unique, there are points other than corners
that are optimal but in any case at least one corner is optimal.
- A different situation is found when the objective function can be made
arbitrarily large. Of course, no corner is optimal in that case.
- Infeasible region.
Self Assessment Questions
6. The feasible region has an important property which is called the
convexity property of geometry (True/False)
Operations Research Unit 3
Sikkim Manipal University Page No. 53
7. The boundaries of the feasible regions are lines or planes.
(True/False)
8. If the optimal solution is not unique, there are points other than corners
that are optimal but in any case at least one corner is optimal.
3.7 Summary
Let us recapitulate the important concepts discussed in this unit:
- Any LPP with two decision variables can be solved graphically.
- Any non-negative solution satisfying all the constraints is known as a
feasible solution of the problem.
- The collection of all feasible solutions is known as a feasible region.
- The feasible region of an LPP is a convex set.
- The value of the decision variables, which maximises or minimises the
objectives function, is located on the extreme point of the convex set
formed by the feasible solutions.
- Sometimes the problem may be infeasible indicating that no solution
exists for the problem.
3.8 Glossary
Inequality constraints: condition of LPP that does not hold exactly
Constraint functions: function that describes the set of conditions of LPP
Optimal solution: best solution from all feasible solutions
3.9 Terminal Questions
1. Use the graphical method to solve the following LPP:
Maximise z= 5x1 + 3x2
Subject to:
3x1 + 5x2 ≥ 15
5x1 + 2x2 ≥ 10
x1, x2 ≥ 0
2. Use graphical method to solve the following LPP:
Maximise z= 2x1+4x2
Subject to constraints: x1+2x2 ≤ 5, x1+x2≤ 4, x1, x2≥ 0
Operations Research Unit 3
Sikkim Manipal University Page No. 54
3. Use graphical method to solve the following LPP:
Maximise z= 6x1+x2
Subject to constraints: 2x1+x2 ≥3, x1-x2≥0, and x1, x2≥ 0
3.10 Answers
Self Assessment Questions
1. Feasible
2. Half-plane
3. True
4. False
5. False
6. True
7. True
8. True
Terminal Questions
1. x1 = 20/19, x2 = 45/19, and max z = 235 / 19. Refer 3.4
2. Any point on the line segment BC gives the maximum value (z=10) of
the objective function. There exists an alternative solution. Refer 3.4
3. Unbounded solution. Refer 3.5.2
3.11 Case Study – Travel Agent’s Problem
A local travel agent is planning a charter trip to a famous sea resort. The
eight-day seven-night package includes the round-trip fare, surface
transportation, boarding and lodging, and selected tour options. The charter
trip is restricted to 200 persons and the past experience indicates that there
will not be any problem in arranging 200 passengers. The problem faced by
the travel agent is to determine the number of Deluxe, Standard, and
Economy tour packages to offer for this charter. All three plans differ in
terms of their seating and flight services, quality of accommodation, meal
plans, and tour options. The following table summarises the estimated
prices of the three packages and the corresponding expenses of the travel
agent. The travel agent has hired an aircraft for a flat fee of Rs.2, 00,000 for
Operations Research Unit 3
Sikkim Manipal University Page No. 55
the entire trip. The price per person and the cost for the three packages are
as follows:
Table 3.4: Summary of the Package Prices and Expenses of the Travel Agent
Tour plan Price(Rs.) Hotel cost(Rs.)
Meals and
expenses(Rs.)
Deluxe 10,000 4,750 3,000
Standard 7,000 2,500 2,200
Economy 6,500 2,200 1,900
In planning the trip, the following considerations must be taken into account:
- At least 10% of the packages must be of the deluxe type.
- At least 30% but not more than 70% must be of the standard type.
- At least 30% must be of the economy type.
- The maximum number of deluxe packages available in any aircraft is
restricted to 60.
- The hotel desires that at least 120 tourists should be for the deluxe and
standard packages together.
The travel agent wishes to determine the number of packages to offer in
each type of trip so as to maximise the total profit.
Discussion Questions:
1. Formulate the problem as LPP.
[Hint: x1= no. of deluxe packages, x2= no. of standard packages & x3=
no. of economy packages].
2. Find the optimum solution using graphical methods for the LPP and
interpret your results.
[Hint: Plot constraint lines and mark feasibility region]
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004). Operation
Research. Sultan Chand and Sons.
Operations Research Unit 3
Sikkim Manipal University Page No. 56
E- References:
- Linear Optimization http://www.mirrorservice.org/sites/
home.ubalt.edu/ntsbarsh/Business-stat/opre/partVIII.htm.
Operations Research Unit 4
Sikkim Manipal University Page No. 57
Unit 4 Simplex Method
Structure:
4.1 Introduction
Objectives
4.2 Standard Form of LPP
Fundamental theorem of LPP
4.3 Solution of LPP – Simplex Method
Initial basic feasible solution of an LPP
To solve an LPP in canonical form by simplex method
4.4 The Simplex Algorithm
Steps
4.5 Penalty Cost Method or Big M-method
4.6 Two Phase Method
4.7 Solved Problems on Minimisation
4.8 Summary
4.9 Glossary
4.10 Terminal Questions
4.11 Answers
4.12 Case Study
4.1 Introduction
In the previous unit we dealt with graphical analysis of linear programming
problems. We discussed some basic definitions, graphical methods to solve
LPP, some exceptional cases, advantages, limitations, and important
geometric properties of LPP.
In this unit, we will deal with the simplex method, which focuses on solving
an LPP of any enormity involving two or more decision variables.
The simplex algorithm is an iterative procedure for finding the optimal
solution to a linear programming problem. The objective function controls
the development and evaluation of each feasible solution to the problem. If a
feasible solution exists, it is located at a corner point of the feasible region
determined by the constraints of the system.
The simplex method simply selects the optimal solution amongst the set of
feasible solutions of the problem. The efficiency of this algorithm is because
it considers only those feasible solutions which are provided by the corner
points, and that too not all of them.
Operations Research Unit 4
Sikkim Manipal University Page No. 58
Objectives:
After studying this unit, you should be able to:
- create a standard form of LPP from the given problem
- apply the simplex algorithm to the system of equations
- find solution using big M-technique
- explain the importance of the two phase method
4.2 Standard Form of LPP
The characteristics of the standard form of LPP are:
- All constraints are equations. The right-hand side element of each
constraint equation is non-negative.
- All the variables are non-negative.
- The objective function is of maximisation type.
The inequality constraints of equations are obtained by adding or
subtracting the left-hand side of each such constraint by a non-negative
variable.
1. Introduce slack variables (Si) for ―s‖ type of constraint.
2. Introduce surplus variables (Si) and artificial variables (Ai) for ―>‖ type
of constraint.
3. Introduce only artificial variable for ―=‖ type of constraint.
4. Cost (Cj) of slack and surplus variables will be zero and that of artificial
variable will be ―M‖.
To make the right-hand side of a constraint equation positive, multiply both
sides of the resulting equation by -1. Use the elementary transformations
introduced with the canonical form to achieve the remaining characteristics.
Any standard form of the LPP is given by
Maximise or Minimise
z Cixi
n
i =1
= ¿
Subject to:
aij x j Si bi (bi 0)i 1,2...........m.
n
j 1
¿ ± = > =
=
and xj> 0, j = 1, 2, --- n
S i> 0, i = 1, 2, --- m
Operations Research Unit 4
Sikkim Manipal University Page No. 59
4.2.1 Fundamental theorem of LPP
A set of m simultaneous linear equations in n unknowns/variables, where
n > m, AX = b, with r (A) = m. If there is a feasible solution X > 0, then there
exists a basic feasible solution.
Self Assessment Questions
1. We add surplus variable for ―≤‖constraint. (True/False)
2. The right-hand side element of each constraint is non-negative in the
standard form. (True/False)
4.3 Solution of the LPP – Simplex Method
Consider an LPP given in the standard form.
To optimise z = c1 x1 + c2 x2 + ---+ cnxn
Subject to
a11 x1 + a12 x2 + -- + an x n± S1 = b1
a21 x1 + a22 x2 + ----+ a2nxn± S2 = b2
………………………………………………………………….
am1 x1 + am2 x2 + -- + amnxn±Sm = bm
x1, x2, --- xn, S1, S2 ---, Sm> 0
To each of the constraint equations, add a new variable called an artificial
variable on the left-hand side of every equation, which does not contain a
slack variable. Subsequently, every constraint equation will contain either a
slack variable or an artificial variable.
The introduction of slack and surplus variables does not alter the equations
of constraints or the objective function. Therefore, such variables can be
incorporated in the objective function with zero coefficients. However, the
artificial variables do change the constraints as these are added only to the
left-hand side of the equations.
The newly derived constraint equation is equivalent to the original equation,
only if all the artificial variables have a value zero. Artificial variables are
incorporated into the objective function with very large positive coefficient M
in the minimisation program and very large negative coefficient–M in the
maximisation program guaranteeing that the artificial variable takes the zero
Operations Research Unit 4
Sikkim Manipal University Page No. 60
value in optimal solutions. The large positive and negative coefficients
represent the penalty incurred for making a unit assignment to the artificial
variable.
Thus the standard form of LPP can be given as follows:
Optimise Z = CT X
Subject to AX = B,
and X > 0
Where X is a column vector with decision, slack, surplus, and artificial
variables, C is the vector corresponding to the costs; A is the coefficient
matrix of the constraint equations and B is the column vector of the righthand
side of the constraint equations.
Solved problem 1:
Minimise Z = 4x1 + x2
Subject to 3x1 + x2 = 3
4x1 + 3x2> 6
x1 + 2x2s 3
x1, x2> 0
Solution: Rewriting in the standard form,
Minimise Z = 4x1 + x2 + 0.S1 + 0.S2 + M (A1 + A2)
Subject to 3x1+ x2 + A1 = 3
4x1 + 3x2 – S1 + A2 = 6
x1 + 2x2 + S2 = 3
x1, x2, S1, S2, A1, A2> 0
Operations Research Unit 4
Sikkim Manipal University Page No. 61
When S2 is a slack variable, S1 is a surplus variable, and A1 and A2 are
artificial variables. Representing this program in matrixes, you have
Z = (4 1 0 0 M M)
|
|
|
|
|
|
|
|
|
|
|
|
.
|












\
|
2
1
2
1
2
1
A
A
S
S
x
x
Subject to
|
|
|
.
|



\
|
÷
1 2 0 1 0 0
4 3 1 0 0 1
3 1 0 0 1 0
|
|
|
|
|
|
|
|
|
|
|
|
.
|












\
|
2
1
2
1
2
1
A
A
S
S
x
x
=
| |
|
|
|
.
|




\
|
3
6
3
and
|
|
|
|
|
|
|
|
|
|
|
|
.
|












\
|
2
1
2
1
2
1
A
A
S
S
x
x
> 0
4.3.1 Initial basic feasible solution of an LPP
Consider a system of m equations in n unknowns x1, x2 - - xn,
a11 x1 + a12 x2 + - - + a1nxn = b1
a21 x1 + a22 x2 + - - + a2nxn = b2
………………………………………………………
am1 x1 + am2 x2 + - - + amnxn = bn
Where m s n
To solve this system of equations, you must first assign any of n – m
variables with value zero. The variables assigned the value zero are called
Operations Research Unit 4
Sikkim Manipal University Page No. 62
non-basic variables, while the remaining variables are called basic
variables. Solve the equation to obtain the values of the basic variables. If
one or more values of the basic variables are valued at zero, then the
solution is said to degenerate, whereas if all basic variables have non-zero
values, then the solution is called a non-degenerate solution. A basic
solution satisfying all constraints is said to be feasible.
Consider an LPP given in the standard form
Optimise Z = CTX
Subject to AX = B,
X > 0
To arrive at the initial solution of such a problem denoted by X0, treat all
decision and surplus variables as non-basic variables. Assign value zero; all
slack and artificial variables as basic variables are assigned values, which
are on the right-hand side of the corresponding constraint equations.
Solved problem 2
To obtain basic solution of the system of equations 2x1 + x2 – x3 = 2
3x1 + 2x2 + x3 = 3
where x1, x2, x3 > 0
Solution: At the onset, assign 3 – 2 = 1 variable the value zero since
there are three variables and two equations.
Case (i)
Let x3 = 0 i.e., x3 be a non-basic variable. Then, the equation becomes
2x1 + x2 = 2
3x1 + 2x2 = 3
Solving, we get
x1 = 1, x2 = 0.
Therefore, the solution degenerates, but is feasible.
Case (ii)
Let x2 be a non-basic variable i.e., x2 = 0, then solution is
x1 = 1 and x3 = 0.
Here also the solution degenerates but is feasible
Operations Research Unit 4
Sikkim Manipal University Page No. 63
Case (iii)
Let x1 be non-basic i.e., x1= 0
Solution is x2 = 5/3 , x3 = –
3
1
The solution non-degenerates, but is not feasible.
4.3.2 To solve an LPP in canonical form by simplex method
To solve an LPP problem in canonical form by the simplex method, follow
the steps below.
1. Zj = sum of [cost of variable x its coefficients in the constraints –
profit or cost coefficient of the variable].
2. Convert the problem into standard form.
3. Find Zj - Cj for each variable.
4. Slack and artificial variables will form basic variable for the first
simplex table. Surplus variable will never become basic variable for
the first simplex table.
5. Select the most negative value of Zj - Cj. That column is called key
column. The variable corresponding to the column will become basic
variable for the next table.
6. Divide the RHS quantities by the corresponding values of the key
column to get ratios; select the minimum ratio. This becomes the key
row. The basic variable corresponding to this row will be replaced by
the variable found in step 5.
7. The element that lies both on key column and key row is called
pivotal element.
8. Once an artificial variable is removed as basic variable, its column
will be deleted from the next iteration.
9. For maximisation problems, decision variables coefficient will be
same as in the objective function. For minimisation problems,
decision variables coefficients will have opposite signs as compared
to objective function.
10. Values of artificial variables will always be M for both maximisation
and minimisation problems.
11. The process is continued till all Zj - Cj > 0.
Operations Research Unit 4
Sikkim Manipal University Page No. 64
Self Assessment Questions
3. A basic solution is said to be a feasible solution if it satisfies all
constraints. (True/False)
4. If one or more values of basic variable are zero then solution is said to
be degenerated. (True/False)
5. The element that lies both on key column and key row is called pivotal
element. (True/False)
4.4 The Simplex Algorithm
To test for optimality of the current basic feasible solution of an LPP uses
the following algorithm called simplex algorithm. Let‘s also assume that
there are no artificial variables existing in the program.
4.4.1 Steps
Figure 4.1 depicts the steps to solve the simplex algorithm.
Fig. 4.1: Steps to Solve Simple Algorithm
Operations Research Unit 4
Sikkim Manipal University Page No. 65
Let us now study the steps in detail.
1) Locate the negative number in the last row of the simplex table. Do not
include the last column. The column that has negative number is called
the pivot column.
2) Next, form ratios by dividing each positive number in the pivot column,
excluding the last row into the element in the same row and last column.
Assign that element to the pivot column to yield the smallest ratio as the
pivot element. If more than one element yields the same smallest ratio,
choose the elements randomly. The program has no solution, if none of
the element in the pivot column is non-negative.
3) To convert the pivot element to unity (1) and then reduce all other
elements in the pivot column to zero, use elementary row operations.
That is to divide each element in key row by key element, to find new
values for that row.
4) To make other elements in key column zero, use the formula:
Number in new row = Number in old row ± [ Number above or below key
element * Corresponding number in the new row]
5) Replace the x-variable in the pivot row and first column by x-variable in
the first row pivot column. The variable to be replaced is called the
outgoing variable and the variable that replaces it is called the incoming
variable. This new first column is the current set of basic variables.
6) Repeat steps 1 through 4 until all the negative numbers in the last row
excluding the last column are exhausted.
7) You can obtain the optimal solution by assigning the value to each
variable in the first column corresponding to the row and last column. All
other variables are considered as non-basic and are assigned value
zero. The associated optimal value of the objective function is the
number in the last row and last column for a maximisation program, but
the negative of this number for a minimisation problem.
Operations Research Unit 4
Sikkim Manipal University Page No. 66
Case-let
A manufacturing company discontinued production of an unprofitable
product line. This created excess production capacity. The company‘s
management is considering devoting this excess capacity to one or more
of the other ongoing products.
Knowing the capacity of the machines, the number of machine hours
required to produce one unit of each product and the unit profit per
product, the management needs to find out how much of each product
the company should produce to maximise profit.
The management can use the simplex method to arrive at the required
solution.
Solved problem 3
Maximise z = x1+ 9x2 + x3
Subject to x1 + 2x2 + 3x3s 9
3x1 + 2x2 + 2x3s 15
x1, x2, x3 > 0
Rewriting in the standard form
Maximise z = x1 + 9x2 + x3 + 0.S1 + 0.S2
Subject to the conditions
x1 + 2x2 + 3x3 + S1 = 9
3x1 + 2x2 + 2x3 + S2 = 15
x1, x2, x3, S1, S2> 0
Where S1 and S2 are the slack variables.
Solution: The initial basic solution is x1, x2, x3 =0, S1 = 9, S2 = 15
Therefore, X0 =
| |
|
.
|


\
|
2
1
S
S
, C0 =
| |
|
.
|


\
|
0
0
Table 4.1 depicts the initial simplex table.
Operations Research Unit 4
Sikkim Manipal University Page No. 67
Table 4.1: Initial Simplex Table
X1
1
X2
9
X3
1
S1
0
S2
0
Ratio
S1 0
S2 0
1
3
2*
2
3
2
1
0
0
1
9
15
2
9
= 4.5 ÷
2
15
= 7.5
Zj – Cj –1
–9
|
–1 0 0
Work column* – pivot element
S1 – outgoing variable, x2 – incoming variable
Since the three Zj – Cj are negative, the solution is not optimal. Choose
the maximum negative value that is –9. The corresponding column vector
x2 enters the basis replacing S1, since ratio is at minimum. You can use
the elementary row operations to reduce the pivot element to 1 and other
elements of work column to zero.
Divide row 1 with the key element 2 and get the new row. Now make all
other elements in the key column zero.
First iteration – The variable x2 becomes a basic variable replacing
S1 and you obtain the table 4.2.
Table 4.2: First Iteration Table, Variable x2 Becomes A Basic Variable
X1 X2 X3 S1 S2 Ratio
1 9 1 0 0
X2 9
2
1
1
2
3
2
1
0
2
9
S2 0 2 0 – 1 – 1 1 6
Zj – Cj 7/2 0
2
25
2
9
0
2
81
Since all the elements of the last row are non-negative, the optimal
solution is obtained. The maximum value of the objective function Z is
2
81
,
achieved for x2=
2
9
S2 = 6, are the basic variables (refer the shaded
column of table 4.2.). All other variables are non-basic.
Operations Research Unit 4
Sikkim Manipal University Page No. 68
Solved problem 4
Use simplex method to solve the LPP
Maximise Z = 2x1 + 4x2 + x3 + x4
Subject to x1 + 3x2 + x4 s 4
2x1 + x2 s 3
x2 + 4x3 + x4 s 3
x1, x2, x3, x4 > 0
Rewriting in the standard form
Maximise Z = 2x1 + 4x2 + x3 + x4 + 0.S1 + 0.S2 + 0.S3
Subject to x1 + 3x2 + x4 + S1 = 4
2x1 + x2 + S2 = 3
x2 + 4x3 + x4 + S3 = 3
x1, x2, x3, x4,
S1, S2, S3 > 0
Solution:
The initial basic solution is x1, x2, x3, x4 = 0, S1 = 4, S2 = 3, S3 = 3
Therefore, X0 =
|
|
|
.
|



\
|
3
2
1
S
S
S
=
|
|
|
.
|



\
|
3
3
4
C0 =
|
|
|
.
|



\
|
0
0
0
Table 4.3 depicts the initial table.
Table 4.3: The Initial Table
X1 X2 X3 X4 S1 S2 S3 Ratio
2 4 1 1 0 0 0
S1 0 1 3* 0 1 1 0 0 4
3
4
÷
S2 0 2 1 0 0 0 1 0 3
1
3
S3 0 0 1 4 1 0 0 1 3
1
3
Zj – Cj – 2 – 4 – 1 – 1 0 0 0 0
|
pivot column * pivot element
Operations Research Unit 4
Sikkim Manipal University Page No. 69
S1 is the outgoing variable and x2 is the incoming variable to the basic set.
The first iteration gives the table 4.4:
Table 4.4: The First Iteration Table - X2 is the Incoming Variable
X1 X2 X3 X4 S1 S2 S3 Ratio
2 4 1 1 0 0 0
X2 4
3
1
1 0
3
1
3
1
0 0
3
4
S2 0
3
5
0 0 –
3
1

3
1
1 0
3
5
S3 0 –
3
1
0 4*
3
2

3
1
0 1
3
5
12
5
÷
Zj – Cj –
3
2
0 – 1
3
1
3
4
0 0
3
16
|
pivot column
X3 enters the new basic set replacing S3, the second iteration gives the
table 4.5:
Table 4.5: Second Iteration Table – X3 Replaces S3
X1 X2 X3 X4 S1 S2 S3 Ratio
2 4 1 1 0 0 0
X2 4
3
1
1 0
3
1
3
1
0 0
3
4
4
S2 0
3
5
* 0 0 –
3
1

3
1
1 0
3
5

x3 1 –
12
1
0 1
6
1

12
1
0
4
1
12
5
Zj –Cj –
4
3
0 0
2
1
4
5
0
4
1
4
23
|
pivot column
X1 enters the new basic set replacing S2, the third iteration gives the
table 4.6:
Operations Research Unit 4
Sikkim Manipal University Page No. 70
Table 4.6: Third Iteration Table – X1 Replaces S2
X1 X2 X3 X4 S1 S2 S3 Ratio
2 4 1 1 0 0 0
X2 4 0 1 0
5
2
5
2

5
1
0 1
X1 2 1 0 0 –
5
1

5
1
5
3
0 1
X3 1 0 0 1
20
3

10
1
20
1
4
1
2
1
Zj –Cj 0 0 0 27/20
10
11
20
9
4
1
2
13
Since all the elements of the last row are non-negative, the optimal
solution is Z =
2
13
, which is achieved for x2 = 1, x1 = 1, x3 =
2
1
, and x4 = 0.
Case Study
A manufacturing firm has discontinued production of a certain
unprofitable product line. This created considerable excess of production
capacity. The management is considering devoting this excess capacity
to one or more of the three products, called product 1, 2, and 3. Table 4.7
depicts the available capacity on the machines, which might limit output.
Table 4.7: Capacity of the Machine for Production
Machine Type
Available Time
(in machine hours per week)
Milling Machine 250
Lathe 150
Grinder 50
Table 4.8 depicts the number of machine-hours required for each unit of
the respective product.
Operations Research Unit 4
Sikkim Manipal University Page No. 71
Table 4.8: Machine Hours for Each Unit
Productivity (in Machine hours/Unit)
Machine Type Product 1 Product 2 Product 3
Milling Machine 8 2 3
Lathe 4 3 0
Grinder 2 0 1
The unit profit would be Rs.20, Rs.6, and Rs.8 for products 1, 2, and 3.
Find how much of each product the firm should produce in order to
maximise profit?
Let x1, x2, and x3 units of products 1, 2, and 3 be produced in a week.
Then the total profit from these units is
Z = 20 x1 + 6 x2 + 8 x3
To produce these units, the management requires
8x1 + 2x2 + 3x3 machine hours of milling machine
4x1 + 3x2 + 0 x3 machine hours of lathe
and 2x1 + x3 machine hours of grinder
Since the time available for these three machines are 250, 150, and 50
hours respectively, we have
8x1 +2x2 + 3x3 s 250
4x1 + 3x2 s 150
2x1 + x3 s 50
Obviously x1, x2, x3 > 0
Thus the problem is to
Maximise Z = 20x1 + 6x2 + 8x3
Subject to 8x1 + 2x2 + 3x3 s 250
4x1 + 3x2 s 150
2x1 + x3 s 50
x1, x2, x3 > 0
Operations Research Unit 4
Sikkim Manipal University Page No. 72
Rewriting in the standard form,
Maximise Z = 20x1 + 6x2 + 8x3 + 0S1 + 0S 2 + 0S 3
Subject to 8x1 + 2x2 + 3x3 + S1 = 250
4x1 + 3x2 + S2 = 150
2x1 + x3 + S3 = 50
x1, x2, x3, S1, S2, S 3 > 0
Solution: The initial basic solution is x1, x2, x3 =0,
X0 =
|
|
|
.
|



\
|
3
2
1
S
S
S
=
|
|
|
.
|



\
|
50
150
250
Table 4.9 depicts the initial simplex table.
Table 4.9: The Initial Simplex Table
X1 X2 X3 S1 S2 S3 Ratio
20 6 8 0 0 0
S1 0 8 2 3 1 0 0 250 250/8 = 31.25
S2 0 4 3 0 0 1 0 150
4
150
= 37.5
S3 0 2* 0 1 0 0 1 50
2
50
= 25 ÷
ZJ – Cj

20
– 6 – 8 0 0 0 0
|
pivot column* pivot element
Operations Research Unit 4
Sikkim Manipal University Page No. 73
X1 enters the basic set of variables replacing the variable S3. The first
iteration gives the table 4.10:
Table 4.10: The First Iteration Table- X1 Replaces S3
X1 X2 X3 S1 S2 S3 Ratio
20 6 8 0 0 0
S1 0 0 2 – 1 1 0 – 4 50
2
50
= 25
S2 0 0 3* – 2 0 1 – 2 50 ÷
3
50
X1 20 1 0
2
1
0 0
2
1
25
Zj – Cj 0 – 6 2 0 0 10 500
|
pivot column * pivot element
X2 enters the basic set of variables replacing the variable S2. The second
iteration gives the following table:
Table 4.11: The Second Iteration Table- X2 Replaces S2
X1 X2 X3 S1 S2 S3 Ratio
20 6 8 0 0 0
S1 0 0 0
3
1
1
3
÷2

3
8
3
50
50
X2 6 0 1 –
3
2
0
3
1

3
2
3
50

X1 20 1 0 *
2
1
0 0
2
1
25 50 ÷
Zj – Cj 0 0 – 2 0 2 6 600
|
Work column * pivot element.
X3 enters the basic set of variables replacing the variable X1. The third
iteration yields the following table:
Operations Research Unit 4
Sikkim Manipal University Page No. 74
Table 4.12: The Third Iteration Table- X3 Replaces X1
X1 X2 X3 S1 S2 S3 Ratio
20 6 8 0 0 0
S1 0 –
3
2
0 0 1 –
3
2
– 3 0
X2 6
3
4
1 0 0
3
1
0 50
X3 8 2 0 1 0 0 1 50
Zj – cj 4 0 0 0 2 8 700
Since all zj – cj> 0 in the last row, the optimum solution is 700.
i.e., the maximum profit is Rs.700, which is achieved by producing
50 units of product 2 and 50 units of product 3.
Self Assessment Questions
6. The element that lies both on key column and key row is called pivotal
element (True/False)
7. Pivotal element lies on the crossing of key column and key row.
(True/False)
8. In the simplex algorithm, the variable to be replaced is called the
outgoing variable and the variable that replaces it is called the
incoming variable (True/False)
4.5 Penalty Cost Method or Big-M Method
Consider an LPP in which at least one of the constraints is of type > or =.
While expressing in the standard form, add a non-negative variable to each
of such constraints. These variables are called artificial variables.
Addition of artificial variables causes violation of the corresponding
constraints as they are added to only one side of an equation. The new
system is equivalent to the old system of constraints only if the artificial
variables are valued at zero. To guarantee such assignments in the optimal
solution, you can incorporate artificial variables into the objective function
with large positive or negative coefficients in minimisation and maximisation
programs respectively. You denote these coefficients by ± M. Whenever
artificial variables are a part of the initial solution X0, the last row of simplex
Operations Research Unit 4
Sikkim Manipal University Page No. 75
table will contain the penalty cost M. You can make the following
modifications in the simplex method to minimise the error of incorporating
the penalty cost in the objective function. This method is called big Mmethod
or penalty cost method.
Fig. 4.2: Steps in Big M-method
Let us now study the steps in detail.
1) The last row of the simplex table is decomposed into two rows, the first
of which involves those terms not containing M, while the second
involves those containing M.
2) The step 1 of the simplex method is applied to the last row created in the
above modification and followed by steps 2, 3, and 4 until this row
contains no negative elements. Then step 1 of simplex algorithm is
applied to those elements next to the last row that are positioned over
zero in the last row.
3) Whenever an artificial variable ceases to be basic, it is removed from the
first column of the table as a result of step 4. It is also deleted from the
top row of the table as is the entire column under it.
4) The last row is removed from the table whenever it contains all zeroes.
Operations Research Unit 4
Sikkim Manipal University Page No. 76
5) If non-zero artificial variables are present in the final basic set, then the
program has no solution. In contrast, zero valued artificial variables in
the final solution may exist when one or more of the original constraint
equations are redundant.
Solved problem 5
Use penalty cost method to
Maximise z = 2x1 + 3x2
Subject to x1 + 2x2s 2
6x1 + 4x2> 24
x1, x2> 0
Solution: Rewriting in the standard form, we have
Maximise Z = 2x1 + 3x2 + 0S1 + 0S2 – M A1
Subject to x1 + 2x2 + S1 = 2
6x1 + 4x2 – S2 + A1 = 24
x1, x2, S1, S2, A1 > 0
Table 4.13 depicts the initial simplex table.
Table 4.13: The Initial Simplex Table
X1 X2 S1 S2 A1
2 3 0 0 – M Ratio
S1 0 1* 2 1 0 0 2 2
1
2
=
A1 – M 6 4 0 –1 1 24 4
6
24
=
–2 –3 0 0 0 0
– 6M – 4M 0 M –24M
|
Work column
The first iteration gives the following table:
Operations Research Unit 4
Sikkim Manipal University Page No. 77
Table 4.14: The First Iteration Table
X1 X2 S1 S2 A1
2 3 0 0 – M
X1 2 1 2 1 0 0 2
Ai – M 0 – 8 -6 – 1 1 12
0 1 2 0 0 4
0 8M 6M 1M 0 – 12M
The process is complete as all the elements of the last two rows are nonnegative.
But existence of non-zero artificial variables in the basic
solution shows that the problem has no solution (infeasible).
Self Assessment Questions
9. The value of artificial variable is ―M‖. (True/False)
10. Artificial variables are entered as basic variables. (True/False)
4.6 Two Phase Method
The drawback of the penalty cost method is the possible computational error
resulting from assigning a very large value to the constant M. To overcome
this difficulty, a new method is considered, where the use of M is eliminated
by solving the problem in two phases.
Phase I: Formulate the new problem. Start by eliminating the original
objective function by the sum of the artificial variables for a minimisation
problem and the negative of the sum of the artificial variables for a
maximisation problem. The simplex method optimises the objective of
maximisation or minimisation with the constraints of the original problem. If a
feasible solution is arrived, the optimal value of the new objective function is
zero (suggestive of all artificial variables being zero). Subsequently proceed
to phase II. If the optimal value of the new objective function is non-zero, it
means there is no solution to the problem and the method terminates.
Phase II: Start phase II using the optimum solution of phase I as the base.
Then take the objective function without the artificial variables and solve the
problem using the simplex method.
Operations Research Unit 4
Sikkim Manipal University Page No. 78
Solved problem 6
Use the two phase method to
Maximise z = 3x1 – x2
Subject to 2x1 + x2 > 2
x1 + 3x2 s2
x2 s4,
x1, x2 > 0
Rewriting in the standard form,
Maximise z = 3x1 – x2 + 0S1 – MA1 + 0.S2 + 0.S3
Subject to 2x1 + x2 – S1 + A1 = 2
x1 + 3x2 + S2 = 2
x2 + S3 = 4,
x1, x2, S1, S2, S3, A1 > 0
Phase 1: Consider the new objective,
Maximise Z* = – A1
Subject to 2x1 + x2 – S1 + A1 = 2
x1 + 3x2 + S2 = 2
x2 + S3 = 4,
x1, x2, S1, S2, S3, A1 > 0
Table 4.15 depicts the initial simplex table arrived through the simplex
method.
Table 4.15:The Initial Simplex Table
X1 X2 S1 A1 S2 S3
0 0 0 –1 0 0 Ratio
A1 –1 2* 1 –1 1 0 0 2
2
2
= 1 ÷
S2 0 1 3 0 0 1 0 2
1
2
= 2
S3 0 0 1 0 0 0 1 4
–2 –1 1 0 0 0 –2
|
Work column * pivot element
X1 enters the basic set replacing A1.
Operations Research Unit 4
Sikkim Manipal University Page No. 79
The first iteration gives the following table:
Table 4.16: The First Iteration Table
X1 X2 S1 A1 S2 S3
0 0 0 –1 0 0
X1 0 1
2
1

2
1
2
1
0 0 1
S2 0 0
2
5
2
1
–(1/2) 1 0 1
S3 0 0 1 0 0 0 1 4
0 0 0 1 0 0 0
Phase I is complete since there are no negative elements in the last row.
The optimal solution of the new objective is Z* = 0.
Phase II:
Consider the original objective function,
Maximise z = 3x1 – x2 + 0S1 + 0S2 + 0S3
Subject to x1 +
2
x2 –
2
S1 =1
2
5
x2 +
2
S1 + S2 =1
x2 + S3 = 4
x1, x2, S1, S2, S3> 0
Table 4.17 depicts the corresponding simplex table with the initial
solution x1 = 1, S2 = 1, S3 = 4.
Table 4.17: The Corresponding Simplex Table
X1 X2 S1 S2 S3
3 –1 0 0 0 Ratio
X1 3 1
2
1

2
1
0 0 1
S2 0 0
2
5
2 *
1
1 0 1
2
1
1
= 2
S3 0 0 1 0 0 1 4
0
2
5
-3/2
0 0 3
Operations Research Unit 4
Sikkim Manipal University Page No. 80
| Work column * pivot element
Proceeding to the next iteration, we get the following table:
Table 4.18: The Phase II Iteration Table
X1 X2 S1 S2 S3
3 –1 0 0 0
X1 3 1 3 0 1 0 2
S1 0 0 5 1 2 0 2
S3 0 0 1 0 0 1 4
0 10 0 3 0 6
Since all the elements of the last row are non-negative, the current
solution is optimal.
The maximum value of the objective function is Z = 6, which is attained
for X1 = 2, X2 = 0.
Solved problem 7
Maximise z = 3x1 + 2x2,
Subject to 2x1 + x2 s2,
3x1 + 4x2 – S2 + A1> 12,
x1, x2 > 0
Rewriting in the standard form,
Maximise z = 3x1 + 2x2 + 0S1 + 0.S2 – MA1
Subject to 2x1 + x2 + S1 = 2
3x1 + 4x2 – S2 + A1 = 2
x1, x2, S1, S2, A1 > 0
Solve the above using the two phase method.
Phase I:
Consider the new objective function
Maximise z* = – A1
Subject to 2x1 + x2 + S1 =2
3x1 + 4x2 – S2 + A1 = 12,
Operations Research Unit 4
Sikkim Manipal University Page No. 81
x1, x2, S1, S2, A1 > 0
Table 4.19 depicts the initial simplex table.
Table 4.19: The Initial Simplex Table
X1 X2 S1 S2 A1 Ratio
0 0 0 0 –1
S1 0 2 1* 1 0 0 2
1
2
= 2
A1 – 1 3 4 0 –1 1 12
4
12
= 3
–3 – 4 0 1 0 –12
| Work column
The first iteration gives the following table:
Table 4.20: The First Iteration Table
X1 X2 S1 S2 A1
0 0 0 0 –1
X2 0 2 1 1 0 0 2
A1 – 1 – 5 0 – 4 – 1 1 4
5 0 4 1 0 – 4
Since all the elements of the last row are non-negative, the procedure is
complete. But the existence of non-zero artificial variables in the basic
set indicates that the problem is infeasible
4.7 Solved Problems on Minimisation
Solved problem 8
Minimise = Z = 3x1 + 8x2
Subject to
x1 + x2 = 200
x1> 80
x2s 60
x1, x2 > 0
Solution:
In the standard form
Operations Research Unit 4
Sikkim Manipal University Page No. 82
Minimise Z = 3x1 + 8x2 + MA1 + OS1 + MA2 + OS2
Subject to
x1 + x2 +A1 = 200
x1 – S1 + A2 = 80
x2 + S2 = 60
S1, S2, A1, A2> 0
Table 4.21: Simplex Table 1
Cj - 3 - 8 0 0 - M - M
C.B B.V X1 X2 S1 S2 A1 A3 Qty Ratio
- M A1 1 1 0 0 1 0 200 200
- M A2
1
P.E
0 - 1 0 0 0 80 80 ÷ K R
0 S2 0 1 0 1 0 1 60 o
Zj - Cj -2M +3 | K C -M + 8 M 0 0 0
Table 4.22: Simplex Table 2
Cj - 3 - 8 0 0 - M
C.B B.V X1 X2 S1 S2 A1 Qty Ratio
Transformation
- M A1 0 1 1 0 1 120 120 R1
1 = R1-R2
1
- 3 x1 1 0 -1 0 0 80 o R2
1 = R2
0 S2 0
1
P.E
0 1 0 60 60 ÷ K R R3
1 = R3
Zj - Cj 0 -M + 8| K C -M + 3 0 0
Table 4.23: Simplex Table 3
Cj - 3 - 8 0 0
C.B B.V X1 X2 S1 S2 A1 Qty Ratio Transformation
- M A1 0 0
1
P.E
- 1 1 60 60 R1
1 = R1
- 3 x1 1 0 - 1 0 0 80 -ve R2
1 = R2
- 8 x2 0 1 0 1 0 60 o R3
1 = R3 – R1
1
Zj - Cj 0 0
- M + 3 |
KC
M - 8 0 0
Operations Research Unit 4
Sikkim Manipal University Page No. 83
Table 4.24: Simplex Table 4
Cj - 3 - 8 0 0
C.B B.V X1 X2 S1 S2 Qty Ratio Transfor-mation
0 S1 0 0 1 - 1 60 - ve R1
1 = R1
- 3 x1 1 0 0 - 1 140 -ve R2
1 = R2 + R1
1
- 8 x2 0 1 0
1
P.E
60 80 ÷ K R R3
1 = R1
1
Zj - Cj 0 0 0 - 5
Table 4.25: Simplex Table 5
Cj - 3 - 8 0 0
C.B B.V X1 X2 S1 S2 Qty Ratio Transformation
0 S1 0 1 1 0 120 R1
1 = R1 + R3
1
- 3 x1 1 1 0 0 200 R2
1 = R2 + R3
1
0 S2 0 1 0 1 60 R3
1 = R3
Zj - Cj 0 5 0 0
Since all Zj - Cj > 0, the optimum solution is x1 = 200, x2 = 0, and min
Z = 600.
4.8 Summary
Let us recapitulate the important concepts discussed in this unit:
- The simplex algorithm is an iterative procedure for finding the optimal
solution to a linear programming problem
- The simplex method simply selects the optimal solution amongst the set
of feasible solutions of the problem.
- The inequality constraints of equations are obtained by adding or
subtracting the left-hand side of each such constraint by a non-negative
variable
- The introduction of slack and surplus variables does not alter the
equations of constraints or the objective function.
4.9 Glossary
Slack variable: variable added to the left of ―s‖ type of constraint to change
it to equality.
Operations Research Unit 4
Sikkim Manipal University Page No. 84
Surplus variable: variable subtracted from the left of the right ―>‖ type of
constraint to change it to equality.
Simplex Table: table used in the simplex algorithm to keep trace of
computations made at every iteration.
4.10 Terminal Questions
1. Maximise z = 3x1 – x2
Subject to the constraints
2x1 + x2 > 2
x1 + 3x2 s 3
x2 s 4
x1, x2 > 0
2. Minimise z = 6x1 + 7x2
Subject to the constraints
x1 + 3x2 > 12
3x1 + x2 > 12
x1 + x2 > 8
x1, x2 > 0
4.11 Answers
Self Assessment Questions
1. False
2. True
3. True
4. True
5. True
6. True
7. False
8. True
9. True
10. True
Operations Research Unit 4
Sikkim Manipal University Page No. 85
Terminal Questions
1. Rewrite in the standard form. The maximum value of the objective
function is Z = 9 which is attained for X1 = 3, X2 = 0. Refer 4.6
2. In the standard form minimize Z, Optimum solution is Z = 5.8, X1= 8/3,
X2 = 6. Refer 4.7
4.12 Case Study
New Plant Construction
The state electricity board is planning to construct a new plant for the next
10 years. It is possible to construct four types of electric power facilities –
steam plants using coal for energy, hydro electric plants with no reservoir,
hydroelectric plants with small reservoirs, and hydroelectric plants with large
reservoirs.
Consumption of electricity is based on three characteristics. The first is the
total annual usage – the requirement in the area is estimated to be 4,000
billion kilowatt-hours by the 10th year. The second characteristic is the peak
usage of power – usually on a hot summer day at about 2 P.M. Any plan
should provide enough peaking capacity to meet a projected peak need of
3,000 million kilowatts in the 10th year. The third characteristic is guaranteed
power output measured as the average daylight output in midwinter when
the consumption is high and water levels for hydroelectric power are low.
The 10-year requirement is for 2,000 million kilowatts of guaranteed power.
The various possible power plants vary in terms of how they satisfy
characteristics. For example, hydro electric plants with reservoirs provide
substantial peaking capacity, whereas steam plants and hydroelectric plants
with no reservoirs are poor in this respect.
The characteristics of the various types of plants are shown in the table
below. Each is measured in terms of a unit of capacity. The unit of capacity
is defined to be the capacity to produce 1 billion kilowatt-hours per year. The
annual operating costs of the various types of plants also vary considerably.
The final column in the table shows the discounted total costs, including
both investments costs and the discounted annual operating costs.
Operations Research Unit 4
Sikkim Manipal University Page No. 86
Characteristics of electric plants per unit (1 bn KWh) of annual output.
Type
Guaranteed
Output
(millions of
kilowatt)
Peak Output
(million of
kilowatt)
Investment
Cost
(Rs. ‘000)
Discounted
Total cost
(Rs. ‘000)
Steam 0.15 0.20 1200 2600
Hydroelectric:
no reservoir
0.10 0.10 1600 1680
Hydroelectric:
small reservoir
0.10 0.40 2400 2560
Hydroelectric :
large reservoir
0.80 0.90 4000 4400
Discussion Questions:
1. Help the company in developing a 10-year plan that would detail the
capacity of each type of plant that it should build.
2. Develop an LP model and solve it to minimise the total discounted cost.
However, there is a restriction that no more than Rs.14,000 million can
be used for investment in plants over the 10 years.
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
E- Reference :
- The Design of Manufacturing Systems. http://www.springerlink.com
Operations Research Unit 5
Sikkim Manipal University Page No. 87
Unit 5 Duality in Linear Programming Problem
Structure:
5.1 Introduction
Objectives
5.2 Importance of Duality Concepts
5.3 Formulation of Dual Problem
5.4 Economic Interpretation of Duality
Economic interpretation of dual variables
5.5 Sensitivity Analysis
Changes in Cj of a non-basic variable
Change in Cj of a basic variable
Change in available resources
Calculating the range
5.6 Summary
5.7 Glossary
5.8 Terminal Questions
5.9 Answers
5.10 Case Study
5.1 Introduction
In the previous unit, simplex method, we studied about the standard form
and the solution for linear programming problem. The simplex method
focuses on solving linear programming problem of any enormity involving
two or more decision variables. The simplex algorithm is an iterative
procedure for finding the optimal solution to a linear programming problem.
The objective function controls the development and evaluation of each
feasible solution to the problem. If a feasible solution exists, it is located at a
corner point of the feasible region determined by the constraints of the
system.
In this unit, duality in linear programming problem, we will study about the
importance of duality concepts and the formulation of dual concepts. We will
also learn about the sensitivity analysis. Every Linear Programming Problem
(LPP) is associated with another linear programming problem involving the
same data and optimal solutions. The two problems are said to be duals of
each other. One problem is called the primal, while the other problem is
called the dual.
Operations Research Unit 5
Sikkim Manipal University Page No. 88
The dual formulation is derived from the same data and solved in a manner
similar to the original 'primal' formulation. In other words, you can say that
dual is the 'inverse' of the primal formulation due to the following reasons:
- If the primal objective function is 'maximisation' function, then the dual
objective function is 'minimisation' function and vice-versa.
- The column coefficient in the primal constraint is the row coefficient in
the dual constraint.
- The coefficients in the primal objective function are the Right Hand Side
(RHS) constraint in the dual constraint.
- The RHS column of constants of the primal constraints becomes the
coefficient of the dual objective function.
The concept of duality is useful to obtain additional information about the
variation in the optimal solution. These changes could be affected in the
constraint coefficient, in resource availabilities and/or objective function
coefficient. This effect is termed as post optimality or sensitivity analysis.
Objectives:
After studying this unit, you should be able to:
- describe the duality concept
- solve a dual problem
- describe the economical interpretation
- apply sensitivity analysis
5.2 Importance of Duality Concepts
The importance of duality is due to two main reasons:
- If the primal contains a large number of constraints and a smaller
number of variables; the labour of computation can be considerably
reduced, by converting it into the dual problem and then solving it.
- The interpretation of the dual variable from the loss or economic point of
view proves extremely useful in planning future decisions.
Characteristics of dual solutions
If the primal problem possesses a unique non-degenerate, optimal solution,
then the optimal solution to the dual is unique. However, dual solutions arise
under a number of other conditions which are as follows:
Operations Research Unit 5
Sikkim Manipal University Page No. 89
- When the primal problem has a degenerate optimal solution, the dual
has multiple optimal solutions.
- When the primal problem has multiple optimal solutions, the optimal dual
solution is degenerate.
- When the primal problem is unbounded, the dual is infeasible.
- When the primal problem is infeasible, the dual is unbounded or
infeasible.
Self Assessment Questions
1. Dual linear programming problem always reduces the amount of
computation. (True/False)
2. It is possible to reverse the dual linear programming problem to primal
linear programming problem. (True/False)
5.3 Formulation of Dual Problem
Consider the following linear programming problem:
Maximise Z = c1x1 +c2x2 + . . . +cnxn
Subject to the constraints
a11 x1 + a12 x2 + . . . + a1n xn ≤ b1
a21 x1 + a22 x2 + . . . + a2n xn ≤ b2
am1 x1 + am2 x2 + . . . + amn xn ≤ bm
x1, x2, . . ., xn ≥ 0
To construct a dual problem, we must adopt the following guidelines:
- The maximisation problem in the primal becomes a minimisation
problem in the dual and vice versa.
- (≤) type of constraints in the primal becomes (≥) type of constraints in
the dual and vice versa.
- The coefficients c1, c2,...,cn in the objective function of the primal
become b1, b2,…,bm in the objective function of the dual.
- The constants b1, b2,…,bm in the constraints of the primal become
c1, c2,...,cn in the constraints of the dual.
- If the primal has n variables and m constraints the dual will have m
variables and n constraints.
- The variables in both the primal and dual are non-negative.
Operations Research Unit 5
Sikkim Manipal University Page No. 90
Thus the dual problem will be as follows:
Minimise W = b1 y1 + b2 y2 + . . . +bmym
Subject to the constraints
a11 y1 + a21 y2 + . . . + am1ym ≥ c1
a12 y1 + a22 y2 + . . . + am2ym ≥ c2
a1n y1 + a2n y2 + . . . + amnym ≥ cn
y1, y2, . . .,ym≥ 0
Formation of dual linear programming problem is easier; when the canonical
form of linear programming problem for maximisation problem must contain
―≤‖ type of constraints, while for minimisation problem, it must contain ―≥”
type of constraints.
Solved problem 1
Write the dual of Max Z = 4x1 + 5x2 Subject to 3x1 + x2≤ 15
x1 +2 x2 ≤ 10
5x1 + 2x2≤ 20
x1, x2, ≥ 0
Solution
The given problem is in its standard form. Therefore, its dual is:
Min W = 15y1 + 10 y2 + 20 y3
Subject to 3y1 + y2 + 5 y3 ≥ 4
y1 + 2y2 + 2y3 ≥ 5
y1, y2, y3, ≥ 0
Solved problem 2
Write the dual of
Min Z = 10x1 + 12x2
Subject to 2x1 +3 x2≥ 10
5x1 +6 x2≥ 20
x1 + 2x2≥ 15
2x1 + 3x2 ≥ 12
x1, x2, ≥ 0
Solution
The given problem is in its standard form. Therefore, its dual problem is:
Max W = 10y1 + 20 y2 + 15 y3 + 12y4
Operations Research Unit 5
Sikkim Manipal University Page No. 91
Subject to 2y1 + 5y2 + y3 + 2y4 ≤ 12
6y1 + 2y2 + 3 y3 ≤ 10
y1, y2, y3,y4 ≥ 0
Solved problem 3
Write the dual of
Max Z = 100x1 + 200x2
Subject to 3x1 - 10 x2≥ 15
+4x1 +15 x2 ≤ 20
x1, x2, ≥ 0
Solution
First we convert
3x1 - 10x2 ≥ 15
to ―≤” type as shown here.
-3x1 + 10x2 ≤ -15
Therefore, its dual is:
Min Z = -15y1 + 20 y2
Subject to -3y1 + 4y2 ≥ 100
10y1 + 15y2 ≥ 200
y1, y2, ≥ 0
Solved problem 4
When the constraints contain ―=‖ sign, write the dual of
Max Z = 40x1 + 30x2
Subject to 10x1 +6 x2≤ 15
5x1 +7x2 ≥ -10
x1 + x2 = 9
x1 + x2 ≥ 10
x1, x2, ≥ 0
Solution
Firstly,
5x1 – 7x2 ≥ -10
Operations Research Unit 5
Sikkim Manipal University Page No. 92
is rewritten as
-5x1 +7x2 ≤ 10
Secondly,
x1 + x2 = 9
is written as
x1 + x2 ≤ 9
Also
x1 + x2 ≥ 9 this is same as -x1 - x2 ≤ -9
Therefore, the given problem is:
Max Z = 40x1 + 30x2
Subject to 10x1 + 6 x2 ≤ 15
- 5x1 +7 x2 ≤ 10
x1 +x2 ≤ 9
- x1 – x2<=- 9
-x1 –x2 ≤ -10
Therefore, its dual is:
Min W = 15y1 + 10 y2 + 9y3
1 – 9y3
11 -10y4
Subject to 10y1 - 5y2 + y3
1 – y3
11 –y4 ≥ 40
6y1 + 7y2 + y3
1 – y3
11-y4 ≥ 30
y1, y2, y3
1 – y3
11,y4 ≥ 0
Let y3
1 – y3
11= y3
Then the dual becomes
Min W =- 15y1 +10 y2 +9 y3 -10y4
Subject to 10y1 - 5y2 + y3-y4 ≥ 40
6y1 + 7y2 + y3 +y4≥ 30
y1, y2,y4 ≥ 0
y3 is unrestricted in sign
Operations Research Unit 5
Sikkim Manipal University Page No. 93
Solved problem 5
Write the dual of
Min Z = 12x1 + 15x2
Subject to 5x1 +3 x2≥ 10
x1 + x2 ≤ 5
x1 ≥ 0
x2 is unrestricted in sign
Solution
Let x2 =x2
1 – x2
11≥ 0. Therefore, its standard form is:
Min Z = 12x1 + 15x2
1 -15x2
11
Subject to 5x1 +3(x2
1- x2
11) ≥ 10
- [x1 +(x2
1- x2
11)] ≥ -5
Or - x1 - x2
1+ x2
11≥ -5
x1, x2
1, x2
11≥ 0
Therefore, dual is:
5y1 – y2 ≤ 12 ………….…(1)
3y1 – y2 ≤ 15 ………….…(2)
-3y1 + y2 ≤ - 15 ……………(3)
No.3 constraint is 3y1 – y2 ≥ 15 …………(4)
Constraints 2 and 4 give
3y1 – y2 = 15
The dual problem is:
Max W = 10y1 –5 y2
Subject to 5y1 – y2 ≤ 12
3y1 – y2 =15
y1 ≥ 0 y2 unrestricted in sign
Solved problem 6
Write the dual of the following linear programming problem:
Minimise Z = 3x1 – 2x2 + 4x3
Subject to 3x1 + 5x2 + 4x3 ≥ 7
6x1 + x2 + 3x3 ≥ 4
7x1 - 2x2 - x3 ≤ 10
x1 - 2x2 + 5x3 ≥ 3
4x1 + 7x2 - 2x3 ≥ 2,
x1, x2, x3 ≥ 0
Operations Research Unit 5
Sikkim Manipal University Page No. 94
Solution
Since the problem is of minimisation, all constraints should be of ≥ type.
Therefore, you have to multiply the third constraint throughout by -1 so
that -7x1 + 2x2 + x3 ≥ -10.
Let y1, y2, y3, y4 and y5 be the dual variables associated with the above
five constraints. Then the dual problem is given by:
Maximise W = 7y1 + 4 y2 – 10 y3 + 3 y4 + 2y5
Subject to 3y1 + 6y2 – 7 y3 + y4 + 4y5 ≤ 3
5y1 + y2 + 2y3 – 2y4 + 7y5 ≤ -2
4y1 + 3y2 + y3 + 5y4 – 2y5 ≤ 4
y1, y2, y3, y4, y5≥ 0
Solved problem 7
Find the dual of:
Maximise 12x1 + 10x2
Subject to 2x1 + 3x2≤ 18
2x1 + x2≤ 14
x1, x2≥ 0
Solution
The 'dual' formulation for this problem will be as follows:
Minimise 18y1 + 14y2
Subject to 2y1 + 2y2≥ 12
3y1 + y2≥ 10
y1>= 0, y2≥ 0
Points to be noted
- The column coefficients in the primal constraint namely (2, 2) and
(3, 1) have become the row coefficients in the dual constraints.
- The coefficient of the primal objective function namely 12 and 10
have become the constants in the right hand side of the dual
constraints.
- The constants of the primal constraints, namely 18 and 14, have
become the coefficient in the dual objective function.
- The direction of the inequalities has been reserved.
- While the primal is a ‗maximisation‘ problem, the dual is a
‗minimisation‘ problem.
Operations Research Unit 5
Sikkim Manipal University Page No. 95
Solved problem 8
Obtain the dual problem of the following primal formulation:
Maximise Z = 2x1 + 5x2 + 6x3
Subject to 5x1 + 6x2 -x3≤ 3
-2x1 + x2 + 4x3 ≤ 4
x1 - 5x2 + 3x3 ≤ 1
-3x1 - 3x2 + 7x3 ≤ 6
x1, x2, x3≥ 0
Solution
Step 1 - Write the objective function of the dual. As there are four
constraints in the primal, the objective function of the dual will have
4 variables.
Minimise r* = 3 w1 + 4 w2 + w3 + 6 w4
Step 2 - Write the constraints of the dual. As all the constraints in the
primal are '<', the constraints in the dual will be '>'. The column COM+
efficient of the primal becomes the row coefficient of the dual.
Constraints 5w1 - 2w2 + w3 - 3w4 ≥ 2
6w1 + w2 - 5w3 -3w4 ≥ 5
-w1 + 4w2 + 3w3 + 7w4≥ 6
Step 3 - Therefore the dual of the primal is:
Minimise Z = 3w1 + 4w2 + w3 + 6w4
Subject to: 5w1 - 2w2 + w3 - 3w4 ≥ 2
6w1+ w2 - 5w3 - 3w4 ≥ 5
- w1 + 4w2 + 3w3 + 7w4 ≥ 6
w1, w2, w3, w4 >= 0
Solved problem 9
Obtain the dual of the following linear programming problem:
Minimise Z = 5x1 -6x2 + 4x3
Subject to the constraints: 3x1 + 4x2 + 6x3 ≥ 9
x1 + 3x2 + 2x3 ≥ 5
7x1 - 2x2 - x3 ≤ 10
x1 - 2x2 + 4x3 ≥ 4
2x1 + 5x2 - 3x3≥ 3
x1, x2, x3 ≥ 0
Operations Research Unit 5
Sikkim Manipal University Page No. 96
Solution
As you can see, one of the primal constraints is a "≤‖ constraint while the
others are all "≥‖ constraints. The dual cannot be worked out unless all
the constraints are in the same direction. To convert this into "≥‖
constraint, multiply both the sides of the equation by "-" sign. After
multiplying the constraint by "-" sign, it will become 7x1 + 2x2 + x3≥ -10.
Now that all the constraints are in the same direction and the dual can be
worked out, the dual formulation is:
Maximise Z = 9w1 + 5w2 - 10w3 + 4w4 + 3w5
Subject to: 3w1 + w2 - 7w3 + w4 + 2w5 ≤ 5
4w1 + 3w2 + 2w3 - 2w4 + 5w5 ≤ -6
6w1 + 2w2 + w3 + 4w4 + -3w5 ≤ 4
w1, w2, w3, w4, w5 ≥ 0
Self Assessment Questions
3. The coefficients of decision variables in the objective function become
quantities on the right hand side of ___________________.
4. ―s‖ constraints changes to ________ type in dual linear programming.
5. For every linear programming problem, there exists a unique
________ problem.
5.4 Economic Interpretation of Duality
The linear programming problem is thought of as a resource allocation
model, where the objective is to maximise revenue or profit subject to limited
resources. The associated dual problem offers interesting economic
interpretations of the linear programming resource allocation model.
Consider a representation of the general primal and dual problems where
primal takes the role of a resource allocation model. Table 5.1 depicts
general primal and dual problems in a resource allocation model.
Operations Research Unit 5
Sikkim Manipal University Page No. 97
Table 5.1: Resource Allocation Model
Primal Dual
Maximise
¿
=
=
n
j 1
z c j. x j
Subject to
¿
=
s =
n
j 1
aij x j bi , i 1, 2, ...,m.
xj ≥ 0, j = 1,2,…., n
Minimise
¿
=
=
m
i 1
w bi. yi
Subject to
¿
=
> =
m
i 1
aij yi ci , j 1, 2, ..., n.
yj ≥ 0, i = 1,2,…., m
In this resource allocation model, the primal problem has ‗n‘ economic
activities and ‗m‘ resources. The coefficient cj in the primal represents the
profit per unit of activity j. Resource i, whose maximum availability is bi, is
consumed at the rate aij units per unit of activity j.
5.4.1 Economic interpretation of dual variables
For any pair of feasible primal and dual solutions, (objective value in the
maximisation problem) ≤ (objective value in the minimisation problem).
At the optimum, the relationship holds as a strict equation.
Note: Here, the sense of optimisation is very important.
Hence, for any two primal and dual feasible solutions, the values of the
objective functions, when finite, must satisfy the following inequality:
¿ ¿
= ÷
= s =
n
j 1
m
i 1
z c j x j bi y i w
The strict equality, z = w, holds true when both the primal and dual solutions
are optimal. Consider the optimal condition z = w. Given that the primal
problem represents a resource allocation model, you can think of z as
representing the profit in rupees. bi represents the number of units available
of the resource i. Therefore, we can express the equation z = w as
profit (Rs) = ¿ (units of resource i) x (profit per unit of resource i).
This means that the dual variables yi, represent the worth per unit of
resource i.
Operations Research Unit 5
Sikkim Manipal University Page No. 98
Note: Variables yi are also called as dual prices, shadow prices and simplex
multipliers.
With the same logic, the inequality z < w associated with any two feasible
primal and dual solutions is interpreted as (profit) < (worth of resources).
This relationship implies that as long as the total return from all the activities
is less than the worth of the resources, the corresponding primal and dual
solutions are not optimal. Optimality is achieved only when the resources
have been exploited fully, which can happen when the input equals the
output (profit). Economically, the system is said to be disequilibrium (non
optimal) when the input (worth of the resources) exceeds the output (return).
Equilibrium occurs only when the two quantities are equal.
Self Assessment Questions
6. Dual variables represent the worth or unit of a resource. (True/False)
7. Optimality is reached when the resources are not fully utilised.
(True/False)
8. At optimum level, the relationship holds as a strict equation.
(True/False)
5.5 Sensitivity Analysis
The management of a company rarely restricts its interest to the numerical
values of an optimal solution. It is interested in knowing the impact of
changes in the input parameter values on the optimal solution. This process
is called sensitivity analysis.
In linear programming, all model parameters are assumed to be constant.
However; in real-life situations, the decision environment is always dynamic.
Therefore, it is important for the management to know how profit would be
affected; by an increase or decrease in the resource level, by change in the
technological process, and by change in the cost of raw materials. Such an
investigation is known as sensitivity analysis or post optimality analysis.
The results of sensitivity analysis establishes upper and lower bounds for
input parameter values within which they can vary without causing major
changes in the current optimal solution.
Sensitivity analysis allows us to figure out which data offers a significant
impact on the results. This in turn allows us to concentrate on getting
accurate data for those items, or at least running through several scenarios
Operations Research Unit 5
Sikkim Manipal University Page No. 99
with various values of the crucial data, to get an idea of the range of
possible outcomes.
Sensitivity analysis is important due to the following reasons:
- values of linear programming parameters might change
- linear programming parameters have an uncertainty factor attached to
them
Any change in a linear programming‘s parameters affects the optimality of
the basic variable. This is due to the following two reasons:
- When a variable (or number of variables) in row 0 has a negative
coefficient, we can obtain a better basic feasible solution of larger
z-value, by pivoting a non-basic variable with a negative coefficient in
row 0. If this occurs, the basic variable becomes a sub-optimal basis.
- A constraint (or number of constraints) may have a negative RHS. In
such a case, at least one member of basic variable will become negative
and will no longer yield a basic feasible solution. This is an infeasible
basic.
The following six types of changes in linear programming‘s parameters
cause changes in the optimal solution:
- Changing the objective function coefficient of a non-basic variable.
- Changing the objective function coefficient of a basic variable.
- Changing the right-hand side of a constraint.
- Changing the column of a non-basic variable.
- Adding a new variable or activity.
- Adding a new constraint.
Solved problem 10
Luminous lamps produce three types of lamps A, B and C. These lamps
are processed on three machines X, Y and Z.
To find out the optimal solution in order to maximise profit and minimise
cost, first we find out whether a previously determined optimal solution
remains optimal if the contribution rate is changed. An increase in Cj of a
variable would mean that, resources from other products should be
diverted to this more profitable product.
The reverse is true for a minimisation problem.
Operations Research Unit 5
Sikkim Manipal University Page No. 100
Table 5.2 depicts the full technology and input restrictions.
Table 5.2: Lamp Production and Profit Per Unit
Product
Machine
Profit Per Unit
M1 M2 M3
A 10 7 2 12
B 2 3 4 3
C 1 2 1 1
Available time 100 77 80
Solution
The linear programming model for this problem is as follows:
Maximise Z = 12 x1 + 3 x2 + x3
Subject to10 x1 + 2 x2 + x3 s 100
7 x1 + 3 x2 + 2x3 s 77
2 x1 + 4 x2 + x3 s 80
x1, x2 , x3 > 0
On solving this problem by the simplex method, we get
x1 =
8
73
, x2 = ,
8
35
x3 = 0
And the optimal value of the objective function is
8
981
.
5.5.1 Changes in Cj of a non-basic variable
A non basic variable can be brought on the basis only if its contribution rate
is attractive. Therefore, we need to determine the upper limit of the profit
contribution (Cj) of each non basic variable. The reverse is true for a
minimisation problem. Table 5.3 depicts the final simplex table of this
problem.
Operations Research Unit 5
Sikkim Manipal University Page No. 101
Table 5.3: Final Simplex Table
Cj 12 3 1 0 0 0
CB
Basic
variables B
x1 x2 x3 s1 s2 s3
Solution values
b (=xB)
12 x1 1 0 –
16
1
16
3

8
1
0
8
73
3 x2 0 1
16
13

16
7
8
5
0
8
35
0 s3 0 0
8
17
÷
8
11
4
9
÷ 1
4
177
zj 12 3
16
27
16
15
8
3
0
zj–cj 0 0
16
11
16
15
8
3
0
From this final simplex table, we can see that profit contribution for product
C is Re 1, which is not greater than its zj. Therefore, to bring x3 into the
basis, its profit contribution rate cj must exceed Rs.
16
27
to make zj–cj value
negative or zero that is zj–cj s 0. Specifically,
- If cj* –cj>zj – cj,, then a new optimal solution must be derived.
- If cj* – cj = zj – cj, then an alternative optimal solution exists.
- If cj* –cj<zj – cj, then the current optimal solution remains unchanged.
In case of c3 = 1 and z3 – c3 =
16
11
,
C3* –1 >
16
11
C3* >
16
11
+1 =
16
27
x3 can be introduced in the basis, if its contribution rate c3 increases to at
least Rs
16
27
. If it increases beyond that, then the current solution will no
longer be optimal.
Operations Research Unit 5
Sikkim Manipal University Page No. 102
5.5.2 Change in Cj of a basic variable
Consider the case of product A (x1 column) and divide each zj - cj entry in the
index row by the corresponding coefficient in the x1 row as shown here:
- Minimum
|
|
.
|


\
|
<
÷
÷
sA s
|
|
.
|


\
|
>
÷
;y 0
y
Z C
;y 0 C Minimum
y
Z C
ij
ij
j j
ij i
ij
j j
Referring to the final simplex table, we observe that corresponding to the
non basic variables x3 and x5, y13, y15<0.
Therefore,
Minimum
(
(
(
(
¸
(




¸

|.
|
\
|
÷ ÷ |
.
|
\
|
÷ ÷
8
1
8
3
,
16
1
16
11
= minimum (11, 3) = 3
Corresponding to the non- basic variable x4 , y14>0, therefore,
Minimum
( (
(
(
¸
(



¸

16
3
16
15
= 5
Hence,
–5 s c1
* -12 s 3 i.e. 7 s c1
* s 15
Thus, we can say that the optimal solution is insensitive as long as the
changed profit coefficient c1
* varies between Rs. 7 and Rs. 15.
5.5.3 Change in available resources
If the available resources change, we need to investigate whether a
previous optimal solution remains feasible or not. For long term planning, it
is important to know the bounds within which each available resource (for
instance machine hours) can vary, without causing drastic changes in the
current optimal solution. To illustrate this, divide each quantity in the XB
column by the corresponding coefficient in the X4 column of the table 5.4.
Table 5.4 depicts the resources table.
Operations Research Unit 5
Sikkim Manipal University Page No. 103
Table 5.4: Resources Table
XB X4
X
X
4
B
8
73
16
3
3
146
8
35
16
÷ 7
– 10
4
177
8
11
11
354
The least positive ratio
| |
.
|

\
|
11
354
indicates the number of hours of machine M1
that can be decreased. The least negative ratio (–10) indicates the increase
in the number of hours of machine M1.
5.5.4 Calculating the range
Lower limit = 100 –
11
354
=
11
746
Upper limit = 100 – (–10) = 110
Therefore, the range of hours for M1 is
11
746
to 110.
We can calculate the range of hours for machine M2 and M3 in the same
manner.
Self Assessment Questions
9. Sensitivity analysis is carried out on _______ simplex table.
10. Sensitivity analysis helps us to study the effect of changes in _______
_______ in objective function.
11. The results of sensitive analysis establish ___________ and
________ ___________ for input parameters value.
Operations Research Unit 5
Sikkim Manipal University Page No. 104
5.6 Summary
Let us recapitulate the important concepts discussed in this unit:
- For every linear programming problem there exists a dual linear
programming problem.
- The dual linear programming problem helps us to reduce the amount of
calculation involved in original linear programming problem.
- The dual linear programming problem also helps us to interpret the
economic variables more effectively.
5.7 Glossary
Optimality: most desirable solution subject to constraints
Canonical form: standard form
Primal problem: original linear programming problem
5.8 Terminal Questions
Write the dual of the following linear programming problems:
1. Maximise 2 = 7x1 + 5x2
a. Subject to constraints x1 + 2x2s 6
b. 4x1 + 3x2s 12
c. x1 x2 > 0
2. Maximise z = 3x1 + 4x2
a. Subject to constrains 5x1 + 4x2s 200;
b. 3x1 + 5x2s 150;
c. 5x1 + 4x2> 100;
d. 8x1 + 4x2> 80,
e. x1> 0, x2> 0
3. Maximise z = 2x1 + x2
a. Subject to constraints 4x1 + 3x2s 12,
b. 4x1 + x2s 8,
c. 4x1 – x2s 8,
d. x1, x2> 0
Operations Research Unit 5
Sikkim Manipal University Page No. 105
5.9 Answers
Self Assessment Questions
1. False
2. True
3. Dual
4. ≥
5. Dual
6. True
7. False
8. True
9. Final
10. Resource levels
11. Upper, lower, bounce
Terminal Questions
1. Minimise W = 6y1 + 12y2
Subject to constraints y1 + 4y2 ≥ 7
2y1 + 3y2 ≥ 5
y1 + y2 ≥ 0
2. Minimise w = 200y1 + 150y2 – 100y3 – 80y4
Subject to constraints 5y1 + 3y2 – 5y3 – 8y4 ≥ 3
4y1 + 5y2 – 4y3 – 4y4 ≥ 4
y1, y2, y3, y4 ≥ 0
3. Minimise w = 12y1 + 8y2 + 8y3
Subject to constraints 4y1 + 4y2 + 4y3 ≥ 2
3y1 + y2 - y3 ≥ 1
y1, y2, y3, ≥ 0
5.10 Case Study
Managing a Farm
The Oxman family owns and operates a 640-acre farm that has been in the
family for several generations. The Oxmans always have had to work hard
to make a decent living from the farm and have had to endure some
occasional difficult years. Stories about earlier generations overcoming
Operations Research Unit 5
Sikkim Manipal University Page No. 106
hardships due to droughts, floods, etc., are an important part of the family
history. However, the Oxmans enjoy their self-reliant lifestyle, and gain
considerable satisfaction from continuing the family tradition of successfully
living off the land during an era, when many family farms are being
abandoned or taken over by large agricultural corporations.
John Oxman is the current manager of the farm while his wife Eunice runs
the house and manages the farm‘s finances. John‘s father, grandpa Oxman,
lives with them and still puts in many hours working on the farm. John and
Eunice‘s older children, Frank, Phyllis, and Carl, also are given heavy
chores before and after school. The entire family can produce a total of
4,000 person-hours worth of labour during the winter and spring months,
and 4,500 person-hours during the summer and fall. If any of these personhours
are not needed, Frank, Phyllis, and Carl will use them to work on a
neighbouring farm for $5 per hour during the winter and spring months and
$5.50 per hour during the summer and fall.
The farm supports two types of livestock: dairy cows and laying hens, as
well as three crops: soybeans, corn, and wheat. (All three are cash crops,
but the corn is also a feed crop for data per acre planted and wheat is also
is used for chicken feed.) The crops are harvested during the late summer
and fall. During the winter months, John, Eunice, and grandpa make a
decision about the mix of livestock and crops for the coming year.
Currently, the family has just completed a particularly successful harvest,
which has provided an investment fund of $20,000 that can be used to
purchase more livestock. (Other money is available for ongoing expenses,
including the next planting of crops). The family currently has 30 cows
valued at $35,000 and 2,000 hens valued at $5,000. They wish to keep all
this livestock and perhaps purchase more. Each new cow would cost
$1,500, and each new hen would cost $3.
Over a year‘s time; the value of a herd of cows will decrease by about 10
percent, and the value of a flock of hens will decrease by about 25 percent
due to aging. Each cow will require 2 acres of land for grazing and 10
person-hours of work per month, while producing a net annual cash income
of $850 for the family. The corresponding figures for each hen are: no
significant acreage, 0.05 person-hours/month, and an annual net cash
income of $4.25. The chicken house can accommodate a maximum of
Operations Research Unit 5
Sikkim Manipal University Page No. 107
5,000 hens, and the size of the barn limits the herd to a maximum of
42 cows.
For each acre planted in each of the three crops, table 5.5 depicts the
number of person-hours of work that will be required during the first and
second halves of the year, as well as a rough estimate of the crop‘s net
value (in either income or savings in purchasing feed for the livestock).
Table 5.5: Corn’s Net Value
Data per acre planted
Soybeans Corn Wheat
Winter and spring, person-hours 1.0 0.9 0.6
Summer and fall, person-hours 1.4 1.2 0.7
Net value Rs. 70 Rs. 60 Rs. 40
To provide much of the feed for the livestock; John wants to plant at least
1 acre of corn for each cow in the fund. Plus the value of the livestock at the
end of the coming year, plus any income from working on a neighbouring
farm, minus living expenses of $40,000 for the year coming year‘s herd, in
addition, at least 0.05 acre of wheat for each hen in the coming year‘s flock.
John, Eunice, and grandpa are now discussing how much acreage should
be planted in each of the crops and how many cows and hens to have for
the coming year. Their objective is to maximise the family‘s monetary worth
at the end of the coming year; (the sum of the net income from the livestock
for the coming year plus the net value of the crops for the coming year plus
what remains from the investment).
Discussion Questions:
1. Identify verbally the components of a linear programming model for this
problem.
2. Formulate this model. (Either an algebraic or a spreadsheet formulation
is acceptable.)
3. Obtain an optimal solution and generate the additional output provided
for performing post optimality analysis (for example, the sensitivity report
when using excel). What does the model predict regarding the family‘s
monetary worth at the end of the coming year?
Operations Research Unit 5
Sikkim Manipal University Page No. 108
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
E-References:
- http://www.psnacet.edu.in/courses/MBA/Linearprogramming/lecture-
11.pdf
- http://agecon2.tamu.edu/people/faculty/mccarl-bruce/mccspr/new04.pdf
- http://www.sce.carleton.ca/faculty/chinneck/po/Chapter6.pdf.
- http://www-ee.eng.hawaii.edu/~jyee/491.06/lectures/chapter06.pdf
Operations Research Unit 6
Sikkim Manipal University Page No. 109
Unit 6 Transportation Problem
Structure:
6.1 Introduction
Objectives
6.2 Formulation of Transportation Problem (TP)
6.3 Transportation Algorithm (MODI Method)
6.4 The Initial Basic Feasible Solution
North-west corner rule
Matrix minimum method
Vogel‘s approximation method
6.5 Moving Towards Optimality
Improving the solution
Modified distribution method/MODI method/U – V Method
Degeneracy in transportation problem
6.6 Summary
6.7 Glossary
6.8 Terminal Questions
6.9 Answers
6.10 Case Study
6.1 Introduction
In the previous unit, ‗duality in linear programming problem‘, we studied
about how every linear programming problem (LPP) is associated with
another linear programming problem involving the same data and optimal
solutions. We also learnt about formulation of dual problem, economic
interpretation of duality and sensitivity analysis. In this unit, ‗transportation
problem‘, we will study about the formulation of transportation problem. We
will learn about the initial basic feasible solution and the various methods.
We will also study how to improve the solution and about the modified
distribution method. To facilitate the presentation and solution, the general
transportation problem is normally portrayed in a tabular form.
Transportation model is an important class of linear programs. For a given
supply at each source and a given demand at each destination, the model
studies the minimisation of the cost of transporting a commodity from a
number of sources to several destinations.
Operations Research Unit 6
Sikkim Manipal University Page No. 110
To illustrate the background of a common transportation problem, a typical
situation is portrayed here. Assume that petrol is imported and supplied from
three ports, that is, Mumbai, Kolkata and Chennai. The destination cities
where petrol needs to be supplied are Delhi, Bangalore, Hyderabad and
Ahmedabad. Petrol can be transported from any of the ports to any of the
destination cities; however the transportation costs vary from one place to
another. The transportation problem in this unit helps in determining the
quantity of petrol that needs to be supplied from each of the ports to the
destination cities, such that the total transportation cost is minimised after
satisfying the supply and demand conditions. Figure 6.1 depicts a
transportation problem.
Port Cities
(Source) (Destination)
Fig. 6.1: A Transportation Problem
Where,
Si = Quantity of petrol available at source i, (i=1, 2, 3)
Dj = Quantity of petrol required at destination j, (j = 1, 2, 3, 4)
Cij = The cost of shipping one barrel of petrol from source i to destination j
for each route
Xij =The number of barrels of petrol per route from source i to destination j
Mumbai
S1
Kolkata
S2
Chennai
S3
Delhi
D1
Hyderabad
D2
Bangalore
D3
Ahmedabad
D4
Operations Research Unit 6
Sikkim Manipal University Page No. 111
To facilitate the presentation and solution, the general transportation
problem is normally portrayed in a tabular form by means of a transportation
tableau as discussed later in this unit.
Objectives:
After studying this unit, you should be able to:
- identify the transportation problem
- locate the initial basic feasible solution
- compare the advantages of various methods of finding initial basic
feasible solution
- apply the model to minimise the cost of transporting a commodity
- solve the degeneracy in the transportation problem
6.2 Formulation of Transportation Problem (TP)
The transportation problem involves m sources, each of which has available
ai (i = 1, 2… m) units of homogeneous product and n destinations, each of
which requires bj (j = 1, 2…., n) units of products. Here ai and bj are positive
integers. The cost cij of transporting one unit of the product from the
ith source to the jth destination is given for each i and j. The objective is to
develop an integral transportation schedule that meets all demands from the
inventory at a minimum total transportation cost. Table 6.1 depicts a
transportation problem represented in the form of a transportation table.
Table 6.1 Transportation Table
D1 D2 Dn Supply
S1
x11
x12
x1n
a1
S2
x21
x22
x2n
a2
S3
x31
x32
x3n
a3
Sm
xm1
xm2
xmn
am
Demand b1 b2 bn ¿ai = ¿bi
C11
C21
C31
Cm
1
C12
C22
C32
Cm
2
C1n
C2n
C3n
Cm
n
Operations Research Unit 6
Sikkim Manipal University Page No. 112
Each source has one row and each destination has one column. The
capacity of each source is shown at the end of each row, and each
destination is written under its corresponding column. These capacities and
demands are known as rim conditions. The unit cost to ship from each
source to each destination is written in the upper-right-hand corner of the
cell in the matrix.
If the total demand is equal to the total supply it is a balanced transportation
problem. When the demand is not equal to the supply it is an unbalanced
transportation problem.
It is assumed that the total supply and the total demand are equal.
¿ ¿
= =
=
n
j 1
m
i 1
ai bj (1)
The condition (1) is guaranteed by creating; either a fictitious destination
with a demand equal to the surplus if total demand is less than the total
supply, or a (dummy) source with a supply equal to the shortage if total
demand exceeds total supply. The cost of transportation from the fictitious
destination to all sources and from all destinations to the fictitious sources
are assumed to be zero, so that total cost of transportation will remain the
same. If the total demand is greater than total supply, then there is no
feasible solution.
The standard mathematical model for the transportation problem is as
follows:
Let xij be number of units of the homogenous product to be transported from
source i to the destination j
Then objective is to minimise z = ij
m
i 1
n
j 1
¿ ¿ Cij x
= =
Subject to
¦
¦
¦
)
¦ ¦
¦
`
¹
= =
= =
¿
¿
=
=
x bj;ij 1,2,.......... .,n
x a ,i 1,2,......,m
m
i 1
ij
n
j 1
ij i
(2)
With all xij > 0 and integers
Operations Research Unit 6
Sikkim Manipal University Page No. 113
Theorem: A necessary and sufficient condition for the existence of a
feasible solution to the transportation problem (2) is:
¿ ¿
= =
=
n
j 1
m
i 1
ai bj
Self Assessment Questions
1. Transportation problems are a special type of ___________.
2. The number of rows and columns need not always be ___________.
3. If total demand is greater than total supply then there is _____ ____
solution.
6.3 Transportation Algorithm (MODI Method)
A feasible solution has to be found always. Rather than determining a first
approximation by a direct application of the simplex method, it is more
efficient to work with the transportation table. The transportation algorithm is
the simplex method specialised to the format of table involving the following
steps:
i) Finding an initial basic feasible solution
ii) Testing the solution for optimality
iii) Improving the solution, when it is not optimal
iv) Repeating steps (ii) and (iii) until the optimal solution is obtained
The solution to transportation problem is obtained in two stages
In the first stage, we find the basic feasible solution using any of the
following methods:
- North-west corner rule
- Matrix minima method or least cost method
- Vogel‘s approximation method.
In the second stage, we test the basic feasible solution for its optimality by
MODI method.
Self Assessment Questions
4. Basic feasible solution In transportation problem can be found using
North-west corner rule. (True/False)
5. Transportation problems can also be solved by simplex method.
(True/False)
Operations Research Unit 6
Sikkim Manipal University Page No. 114
6.4 The Initial Basic Feasible Solution
Let us consider a transportation problem involving m-origins and ndestinations.
Since the sum of origin capacities equals the sum of
destination requirements, a feasible solution always exists. Any feasible
solution satisfying m + n – 1 of the m + n constraints is a redundant one and
hence, it can be deleted. This also means that a feasible solution to a
transportation problem can have only m + n – 1 positive component;
otherwise, the solution will degenerate.
It is always possible to assign an initial feasible solution to a transportation
problem, satisfying all the rim requirements. This can be achieved either by
inspection or by following some simple rules. We can begin by imagining
that the transportation table is blank that is initial xij = 0. The simplest
procedures for initial allocation are discussed in the following section.
6.4.1 North-west corner rule
Step 1 - The first assignment is made in the cell occupying the upper left
hand (north-west) corner of the transportation table. The maximum feasible
amount is allocated here is:
x11 = min (a1, b1)
Either the capacity of origin O1 is used up or the requirement at destination
D1 is satisfied or both. This value of x11 is entered in the upper left hand
corner (small square) of cell (1, 1) in the transportation table.
Step 2 - If b1> a1, the capacity of origin O is exhausted and the requirement
at destination D1 is still not satisfied, then, at least one variable in the first
column will have to take on a positive value. Move down vertically to the
second row and make the second allocation of magnitude:
x21 = min (a2, b1 – x21) in the cell (2, 1)
This either exhausts the capacity of origin O2 or satisfies the remaining
demand at destination D1.
If a1> b1, the requirement at destination D1 is satisfied, but the capacity of
origin O1 is not completely exhausted. Move to the right in a horizontal
position to the second column to make the second allocation of magnitude:
x12 = min (a1 – x11, b2) in the cell (1, 2)
Operations Research Unit 6
Sikkim Manipal University Page No. 115
This either exhausts the remaining capacity of origin O1 or satisfies the
demand at destination D2.
If b1 = a1, the origin capacity of O1 is completely exhausted as well as the
requirement at destination is completely satisfied, then there is a tie at the
second allocation. An arbitrary tie breaking choice is made. Make the
second allocation of magnitude:
x12 = min (a1 – a1, b2) = 0 in the cell (1, 2)
or
x21 = min (a2, b1 – b2) = 0 in the cell (2, 1)
Step 3 - Start from the new north-west corner of the transportation table;
satisfying the destination requirements and exhausting the origin capacities
one at a time, moving down towards the lower right corner of the
transportation table, until all the rim requirements are satisfied.
Application
- It is used for transportation to satisfy such obligations where cost is not
the criteria. For example, in the case of Food Corporation of India Ltd.
Solved problem 1
Table 6.2 depicts a transportation problem. Determine an initial basic
feasible solution using the north west corner rule:
Table 6.2 Transportation Problem
D1 D2 D3 D4 Supply
O1 6 4 1 5 14
O2 8 9 2 7 16
O3 4 3 6 2 5
Demand 6 10 15 4 35
Where Oi and Dj represent the ith origin and the jth destination respectively.
Operations Research Unit 6
Sikkim Manipal University Page No. 116
Solution
1. Start with the top most left corner
2. Allocate the maximum possible amount here.
3. Move to the right hand cell if there is still any available quantity,
otherwise move to down cell. Again perform step 2 and continue
until all the available quantity is exhausted.
Following north west corner rule, the first allocation is made in the
cell (1,1), the magnitude being x11 = min (14, 6) = 6
The second allocation is made in the 6 10 15 4 cell (1, 2) and
the magnitude of allocation is given by x12= min (14 – 6, 10) = 8
The third allocation is made in the cell (2, 2), the magnitude being
x22 = min (16, 10 – 8) = 2.
The magnitude of fourth allocation, in the cell (2, 3) is given
by x23 = min (16 – 2, 15) = 14.
The fifth allocation is made in the cell (3, 3), the magnitude
being x33 = min (5, 15 –14) =1.
The sixth allocation in the cell (3, 4) is given by x34 = min
(5 –1, 4) = 4.
Now all the rim requirements have been satisfied and hence, an initial
feasible solution to the transportation problem has been obtained.
Table 6.3 depicts the solution.
Table 6.3 Initial Feasible Solution to the Transportation Problem
D1 D2 D3 D4
O1
O2
O3
6
6
4
8
1
5
14
8
9
2
2
14
7
16
4
3
6
1
2
4
5
6 10 15 4
Clearly, this feasible solution is non-degenerate basic feasible solution as
the allocated cells do not form a loop. The transportation cost according
to the loop is given by:
Z = x11 c11 + x12 c12 + x22 c22 + x23 c23 + x33 c33 + x34 c34
= 6×6+8×4+ 2×9 +14×2 + 1×6+4×2= 128
Operations Research Unit 6
Sikkim Manipal University Page No. 117
6.4.2 Matrix minimum method
Step 1 - Determine the smallest cost in the cost matrix of the transportation
table. Let it be cij. Allocate xij= min ( ai, bj) in the cell ( i, j )
Step 2 - If xij= ai cross the ith row of the transportation table, decrease bj by ai
and proceed to step 3.
If xij= bj cross the ith column of the transportation table, decrease ai by bj and
proceed to step 3.
If xij= ai= bj cross either the ith row or the ith column, but not both.
Step 3 - Repeat steps 1 and 2 to reduce transportation table until all the rim
requirements are satisfied. Whenever the minimum cost is not unique, make
an arbitrary choice among the minima.
Solved problem 2
Obtain an initial basic feasible solution to the following transportation
problem using the matrix minima method. Table 6.4 depicts the initial
table.
Table 6.4 Initial Table
D1 D2 D3 D4 Capacity
O1 1 2 3 4 6
O2 4 3 2 0 8
O3 0 6 8 6 10
Demand 4 6 8 6 24
Where Oi and Di denote ith origin and jth destination respectively.
Solution
The transportation table of the given transportation problem has 12 cells.
Following the matrix minima method,
The first allocation is made in the cells (3, 1), the magnitude being x31 = 4.
This satisfies the requirement at destination D1 and thus we cross the first
column from the table 6.5. The second allocation is made in the cell
(2, 4), the magnitude being x24 = min (6, 8) =6. Cross the fourth column of
the table 6.5. This yields the table 6.5 as depicted.
Operations Research Unit 6
Sikkim Manipal University Page No. 118
Table 6.5: First Iteration Table 6.6: Second Iteration
There is again a tie for the third allocation. Arbitrarily choose the cell
(1, 2) and allocate x12 = min (6, 6) = 6. Cross the second column of the
first row. Next, choose to cross off the first row of the table 6.5. The next
allocation of magnitude x32 = 0 is made in the cell (3, 2), cross the second
column. Table 6.7 depicts the result
Table 6.7: Third Iteration Table 6.8: Final Solution
Again you choose arbitrarily to make the next allocation in cell (2, 3) of
magnitude x23 = min (2, 8) = 2, cross the second row in table 6.7. The last
allocation of magnitude x33 = min (6, 6) = 6 is made in the cell (3, 3). Now
that all the rim requirements have been satisfied, an initial feasible
solution has been determined. This solution is depicted in table 6.8.
Since the cells do not form a loop, the solution is basic and degenerate.
The transportation cost according to the above route is given by:
Z=6×2+2×2+6×0+4×0+0×2+6×2=28
6.4.3 Vogel’s approximation method
The Vogel‘s approximation method (VAM) takes into account not only the
Operations Research Unit 6
Sikkim Manipal University Page No. 119
least cost cij, but also the cost that just exceeds cij. The steps of the method
are given as follows:
Step 1 - For each row of the transportation table, identify the smallest and
the next to smallest costs. Determine the difference between them for each
row. Display them alongside the transportation table by enclosing them in
parenthesis against the respective rows. Similarly, compute the differences
for each column.
Step 2 - Identify the row or column with the largest difference among all the
rows and columns. If a tie occurs, use any arbitrary tie breaking choice. Let
the greatest difference correspond to the ith row and let Cij be the smallest
cost in the ith row. Allocate the maximum feasible amount xij = min (ai, bj) in
the (i, j)th cell and cross off the ith row or the jth column in the usual manner.
Step 3 - Recompute the column and row differences for the reduced
transportation table and go to step 2. Repeat the procedure until all the rim
requirements are satisfied.
Remarks
- A row or column ―difference‖ indicates the minimum unit penalty incurred
by failing to make an allocation to the least cost cell in that row or
column.
- It is clear that VAM determines an initial basic feasible solution, which is
very close to the optimum solution, but the number of iterations required
to reach the optimal solution is small.
Area of application
- It is used to compute transportation routes in such a way as to minimise
transportation cost for finding out location of warehouses
- It is used to find out locations of transportation corporations depots
where insignificant total cost difference may not matter.
Operations Research Unit 6
Sikkim Manipal University Page No. 120
Solved problem 3
A company has three production facilities S1, S2 and S3 with production
capacity of 7,9 and 18 units (in 100s) per week of a product respectively.
These units are to be shipped to four warehouses D1, D2, D3 and D4
with requirement of 5, 8, 7 and 14 units (in 100s) per week, respectively.
Table 6.9 depicts the transportation costs (in rupees) per unit between
factories to warehouses.
Table 6.9: Transportation Costs
D1 D2 D3 D4
S1 19 30 50 10
S2 70 30 40 60
S3 40 8 70 20
Solution
The differences (penalty costs) for each row and column have been
calculated as depicted in table 6.10. In the first round, the maximum
penalty, 22 occurs in column D2. Thus the cell (S3, D2) having the least
transportation cost 8 is chosen for allocation. The maximum possible
allocation in this cell is 8 and it satisfies demand in column D2. Adjust the
supply of S3 from 18 to 10 (18-10).
The new row and column penalties are calculated except column D2
because its demand has been satisfied. The second round allocation is
made in column D1 with target penalty 21 in the same way as in the first
round as depicted in cell (S1, D1) of the table 6.10..
In the third round, the maximum penalty 50 occurs at S3. The maximum
possible allocation of 10 units is made in cell (S3, D4) having least
transportation cost 20 as depicted in table 6.10.
The process is continued with new allocations till a complete solution is
obtained. The initial solution using Vogel‘s approximation method is
depicted in table 6.10.
The total transportation cost associated with this method is calculated as
follows:
Operations Research Unit 6
Sikkim Manipal University Page No. 121
Total cost = 5 x 9 + 2 X 10 + 7 X 40 + 2 X 60 + 8 X 8 + 10 X 20 = Rs. 779
Table 6.10: Initial Solution using Vogel’s Approximation Method
D1 D2 D3 D4 Supply row difference
S1 19 (5) 30 50 10 (2) 7 9 9 40 40
S2 70 30 40 (7) 60 (2) 9 10 20 20 20
S3 40 8 (8) 70 20 (10) 18 12 20 50 X
Demand 5 8 7 14 34
Column
difference
21 22 10 10
21 X 10 10
X X 10 10
10 50
Self Assessment Questions
6. In matrix-minima method, you start allocating from the upper left hand
(north-west) corner of the table. (True/False)
7. In Vogel‘s approximation method, you first construct penalty and then
start allocating. (True/False)
8. North-west corner rule gives optimum solution. (True/False)
9. Vogel‘s approximation method gives solution near to the optimum
solution. (True/False)
6.5 Moving Towards Optimality
After evaluating an initial basic feasible solution to a transportation problem,
the next question is how to get the optimum solution. The basic techniques
are illustrated as follows:
1. Determine the net evaluations for the non–basic variables (empty cells)
2. Determine the entering variable
Operations Research Unit 6
Sikkim Manipal University Page No. 122
3. Determine the leaving variable
4. Compute a better basic feasible solution
5. Repeat steps (1) to (4) until an optimum solution has been obtained
6.5.1 Improving the solution
Definition - A loop is the sequence of cells in the transportation table such
that:
- Each pair of consecutive cells lie either in the same row or same column
- No three consecutive cells lie in the same row or same column
- The first and the last cells of the sequence lie in the same row or column
- No cell appears more than once in the sequence
Consider the non-basic variable corresponding to the most negative of the
quantities cij – ui – vj, calculated in the test for optimality; it is made the
incoming variable. Construct a loop consisting exclusively of this incoming
variable (cell) and current basic variables (cells). Then allocate the incoming
cell to as many units as possible, after appropriate adjustments have been
made to the other cells in the loop. Avoid violating the supply and demand
constraints, allow all allocations to remain non-negative and reduce one of
the old basic variables to zero.
6.5.2 Modified distribution method/MODI method/U–V method
Step 4 - You repeat steps 1 to 3 to till all allocations are over.
Step 5 - For allocating all forms of equations ui+ vj = cj, set one of the dual
variable ui / vj to zero and solve for others.
Step 6 - Use this value to find Aij = cij - ui - vj. If all Aij> 0, then it is the optimal
solution.
Step 7 - If any Aijs 0 select the most negative cell and form loop. Starting
point of the loop is positive and alternative corners of the loop are
negative and positive. Examine the quantities allocated at negative
places. Select the minimum, add it to the positive places and
subtract from the negative places.
Step 8 - Form a new table and repeat steps 5 to 7 till Aij> 0
Operations Research Unit 6
Sikkim Manipal University Page No. 123
Balanced transportation problem
Solved problem 4
Table 6.11 depicts the cost coefficients demands and supplies for a
transportation problem. Solve the transportation problem.
Table 6.11 Cost Coefficients Demands and Supplies
A B C Supply
I
II
III
6 8 4 14
12
5
4 9 8
1 2 6
Demand 6 10 15
Solution
Since total demand = 31 = total supply, the problem is balanced. The
initial basic feasible solution is obtained by Vogel‘s approximation method.
Table 6.12 depicts the initial solution.
Table 6.12 Initial Solution
Supply
6
8
4
14 14
12
5
4
6
9
5
8
1
1 2
5
6
Demand 6 10 15
For allocated cells
u1 + v3 = 4 u1 = -4
Set u2 = 0
u1 + v3 = 4 u3= -4
u2 + v2 = 9 v1 = 4
Operations Research Unit 6
Sikkim Manipal University Page No. 124
u2 + v3 = 8 v2 = 9
u3 + v2 = 5 v3 = 8
Note: Variable ui/vi is repeated very often for easy calculation. Here u2 is
repeated often. For unallocated cells Aij = cij–ui–vj
A11 = 6 – (-4) – 4 = 6
A12 = 8 – (-4) – 9 = 3
A31 = 1 – (-4) – 4 = 1
A33 = 6 – (-4) – 8 = 2
The optimum allocations are x13 = 14, x21 = 6, x22 = 5, x23 = 1, x32 = 5
The minimum transportation cost is 14×4+6×4+5×9+1×8+5×2 = 143
Unbalanced transportation problem
Solved problem 5
A car company is faced with an allocation problem resulting from a rental
agreement, which allow cars to be returned to locations other than those
from where they were originally rented. At the present time there are two
locations with 15 and 13 cars respectively and 4 locations requiring 9, 6,
7 and 9 cars respectively. Table 6.13 depicts the unit transportation costs
(in dollars) between the locations.
Table 6.13 Unit Transportation Costs between Locations
Destinations
D1 D2 D3 D4
Sources S1 45 17 21 30
S2 14 18 19 31
Obtain a minimum cost schedule.
Operations Research Unit 6
Sikkim Manipal University Page No. 125
Solution
Since the supply and requirements are not equal it is called an
unbalanced transportation problem. In general, if Eai=Ebj then it is called
an unbalanced transportation problem. We introduce either a dummy row
or a column with cost zero quantities and Ebi=Eaj respectively. Applying
Vogel‘s approximation method we find the basic feasible solution.
Table 6.14 depicts the basic feasible solution.
Table 6.14: Basic Feasible Solution using Vogel’s Approximation Method
D1 D2 D3 D4 Supply row difference
S1 45 17 (6) 21(3) 30 (6) 15/9/6 4 4 4 11
S2 14(9) 18 19 (4) 31 13/4 4 4 4 12
S3 0 0 0 0 (3) 3 0 0 0 0
Demand 9/0 6 7/3 9/6 31
Column
difference
14 17 19 30
31 1 2 1
- 1 2 1
- - 2 1
X34 = Min [3, 9] = 3 x 0 = 0
X21 = Min [13, 9] = 9 x 14 = 126
X12 = Min [15, 6] = 6 x 17 = 102
X23 = Min [4, 7] = 4 x 19 = 76
X14 = Min [6, 6] = 6 x 30 = 180
X13 = Min [6, 3] = 3 x 21 = 63
Total cost 547
Testing for optimality
u1 + v2 = 17 Set u1 = 0
u1 + v3 = 21 u2= -2
u1 + v4 = 30 u3 = -30
u2 + v1 = 14 v1 = 16
u2 + v3 = 19 v2 = 17
u3 + v4 = 0 v3 = 21
v4 = 30
Operations Research Unit 6
Sikkim Manipal University Page No. 126
For unallocated cells
Aij = cij - ui - vj
A11 = 45 – 0 – 16 = 29
A22 = 8 + 2 – 17 = 3
A24 = 31 +2 – 30 = 3
A31 = 0 + 30 – 26 = 14
A32 = 0 + 30 – 17 = 13
A33 = 0 + 30 – 21 = 9
For non-allocated cells, determine cij– uI– vi. Since all the quantities are
non-negative, the current solution is optimal. The minimum transportation
cost is equal to:
6×17+3×21+6×30+9×14+4×19+3×0 = 547.
This is achieved by transporting x12 = 6 cars from source 1 to destination
2, x13 = 3, x14 = 6 cars from sources 1 to destinations 3 and 4
respectively; x21 = 9 and x33 = 4 cars from sources 2 to destinations 1 and
3 respectively.
6.5.3 Degeneracy in transportation problem
A basic solution to an m-origin, n destination transportation problem can
have at the most m+n-1 positive basic variables (non-zero), otherwise the
basic solution degenerates. It follows that whenever the number of basic
cells is less than m + n – 1, the transportation problem is a degenerate one.
The degeneracy can develop in two ways:
Case 1 - The degeneracy develops while determining an initial assignment
via any one of the initial assignment methods discussed earlier.
To resolve degeneracy, you must augment the positive variables by as
many zero-valued variables as is necessary to complete the required
m + n – 1 basic variable. These zero-valued variables are selected in such a
manner that the resulting m + n – 1 variable constitutes a basic solution.
The selected zero valued variables are designated by allocating an
extremely small positive value ε to each one of them. The cells containing
these extremely small allocations are then treated like any other basic cells.
The ε‘s are kept in the transportation table until temporary degeneracy is
Operations Research Unit 6
Sikkim Manipal University Page No. 127
removed or until the optimum solution is attained, whichever occurs first. At
that point, we set each ε = 0.
Case 2 - The degeneracy develops at the iteration stage. This happens
when the selection of the entering variable results in the simultaneous drive
to zero of two or more current (pre-iteration) basic variables.
To resolve degeneracy, the positive variables are augmented by as many
zero-valued variables as it is necessary to complete m+n-1 basic variables.
These zero-valued variables are selected from among those current basic
variables, which are simultaneously driven to zero. The rest of the
procedure is exactly the same as discussed in case 1.
Note - The extremely small value ε is infinitely small and it never affects the
value it is added to or subtracted from. Introduce ‗e‘ in unallocated minimum
cost cell to avoid forming a loop.
Self Assessment Questions
10. All the values of Aij = Cij - ui - vj should be __________ or _________ for
the solution to be optimum.
11. In unbalanced transportation problem ¿ai is ________ ______ to ¿bj.
12. If the number of allocation is less than _________ then it is said to be a
degenerate transportation problem.
6.6 Summary
Let us recapitulate the important concepts discussed in this unit:
- The transportation problem is a special type of linear programming
problem in which the objective is to transport a homogeneous product
manufactured at several plants (origins) to a number of different
destinations at a minimum total cost.
- There are several different techniques for computing an initial basic
feasible solution to a transportation problem, such as North-west corner
rule, Matrix minimum method and Vogel‘s approximation method.
- The degeneracy in transportation problem can develop in two ways
o The degeneracy develops while determining an initial assignment via
any one of the initial assignment methods discussed.
o The degeneracy develops at the iteration stage.
Operations Research Unit 6
Sikkim Manipal University Page No. 128
6.7 Glossary
Algorithm: sequence of steps to solve complex problems
Degeneracy: In transportation problem the number of occupied cells is less
than m+n-1 where m is number of rows and n is number of columns
6.8 Terminal Questions
1. Table 6.15 depicts a transportation problem. Find the solution.
Table 6.15 Transportation Problem
2. A company has three cement factories located in cities 1, 2, 3 which
supply cement to four projects located in towns 1, 2, 3, 4. Each plant can
supply daily 6, 1, 10 truck loads of cement respectively and the daily
cement requirements of the projects are respectively 7, 5, 3, 2 truck
loads. Table 6.16 depicts the transportation cost per truck load of
cement (in hundreds of rupees) from each plant to each project site.
Table 6.16: Transportation Cost
Factories
Determine the optimal distribution for the company so as to minimise the
total transportation cost.
3. Table 6.17 depicts a transportation problem. Solve the transportation
problem using all three methods to obtain the initial feasible solution.
1 2 3 4
1 2 3 11 7
2 1 0 6 1
3 5 8 15 9
Operations Research Unit 6
Sikkim Manipal University Page No. 129
Table 6.17: Transportation Problem
A B C D E F supply
O1 9 12 9 6 9 10 5
O2 7 3 7 7 5 5 6
O3 6 5 9 11 3 11 2
O4 6 8 11 2 2 10 9
Demand 4 4 6 2 4 2 22
6.9 Answers
Self Assessment Questions
1. Linear programming problem
2. Equal
3. No feasible
4. True
5. True
6. True
7. True
8. False
9. True
10. Zero or positive
11. Not equal
12. m + n – 1
Terminal Questions
1. The optimal transportation cost is Rs. 796
2 The optimal transportation cost is Rs. 10, 000
3 The minimum transportation cost is Rs. 112 as c ÷0
Operations Research Unit 6
Sikkim Manipal University Page No. 130
6.10 Case Study
Transporting Wood to the Market
Amber Wood is a lumber company that has three sources of wood and
five markets to be supplied. The annual availability of wood at
sources 1, 2 and 3 is 15, 20, and 15 million board feet, respectively. The
amount that can be sold annually at markets 1, 2, 3, 4 and 5 is 11, 12, 9, 10
and 8 million board feet respectively.
In the past, the company has shipped the wood by train. However, because
shipping costs have been increasing, the alternative of using ships to make
some of the deliveries is being investigated. This alternative would require
the company to invest in some ships. Except for these investment costs, the
shipping costs in thousands of rupees per million board feet by rail and by
water would be as depicted in table 6.18 for each route:
Table 6.18: Rail Costs and Shipping Costs
Unit Cost by Rail (Rs. 1,000's) Unit Cost by Ship (Rs. 1,000's)
Market Market
Source 1 2 3 4 5 1 2 3 4 5
1 61 72 45 55 66 31 38 24 - 35
2 69 78 60 49 56 36 43 28 24 31
3 59 66 63 61 47 - 33 36 32 26
Table 6.19 depicts the capital investment in ships required for each million
board feet to be transported annually by ship along each route.
Table 6.19: Capital Investment for Ships
Investment for Ships ($1,000's)
Market
Source 1 2 3 4 5
1 275 303 238 - 285
2 293 318 270 250 265
3 - 283 275 268 240
Considering the expected useful life of the ships and the time value of
money, the equivalent uniform annual cost of these investments is one-tenth
the amount depicted in the table 6.19. The objective is to determine the
overall shipping plan that minimises the total equivalent uniform annual cost.
Operations Research Unit 6
Sikkim Manipal University Page No. 131
You are the head of the operations research team that has been assigned
the task of determining this shipping plan for each of the following three
options:
- Option 1: Continue shipping exclusively by rail
- Option 2: Switch to shipping exclusively by water
- Option 3: Ship by either rail or water, depending on which is less
expensive for the particular route
Discussion Questions:
1. Present your results for each option and compare.
2. Consider the fact that these results are based on current shipping and
investment costs, so the decision on the option to adopt now should take
into account, management‘s projection of how these costs are likely to
change in the future. For each option describe the scenario of future
cost changes that would justify adopting the option now.
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
E-References:
- http://www.rajalakshmi.org/Department/MECH/2009ES/ME1018
NOL.doc. accessed on April 28, 2010.
Operations Research Unit 7
Sikkim Manipal University Page No. 132
Unit 7 Assignment Problem
Structure:
7.1 Introduction
Objectives
7.2 Mathematical Formulation of the Problem
7.3 Hungarian Method Algorithm
7.4 Routing Problem
Unbalanced AP
Infeasible assignments
Maximisation in AP
7.5 Travelling Salesman Problem
7.6 Summary
7.7 Glossary
7.8 Terminal Questions
7.9 Answers
7.10 Case Study
7.1 Introduction
In the previous unit we dealt with the formulation of Transportation Problem
(TP), transportation algorithm (MODI method), and the initial basic feasible
solution. We also discussed about how to get the optimum solution. In this
unit, we will deal with assignment problem. The assignment problem is a
special case of transportation problem, where the objective is to minimise
the cost or time of completing a number of jobs by a number of persons, and
to maximise revenue and sales efficiently.
In other words, when the problem involves the allocation of ‗n‘ different
facilities to ‗n‘ different tasks, it is often termed as an assignment problem.
This model is mostly used for planning. The assignment model is useful in
solving problems, such as assignment of machines to jobs, assignment of
salesman to sales territories, travelling salesman problem, and many more
similar situations. It may be noted that with ‗n‘ facilities and ‗n‘ jobs, there
are ‗n‘ possible assignments.
One way of finding an optimal assignment is to write all the ‗n‘ possible
arrangements, evaluate their total cost, and select the assignment model
Operations Research Unit 7
Sikkim Manipal University Page No. 133
offering the minimum cost. This method can be unfeasible due to
involvement of computational procedures. In this unit, we will study an
efficient method for solving assignment problems.
Let‘s say there are ‗n‘ jobs in a factory having ‗n‘ machines to process the
jobs. A job i (=1… n), when processed by machine j (=1… n) is assumed to
incur a cost Cij. The assignment is to be made in such a way that each job
can be associated with one and only one machine. You can then determine
an assignment of jobs to the machines to minimise the overall cost.
The cost data is given as a matrix where rows correspond to jobs and
columns to machines and there are as many rows as the number of
columns. The number of jobs and number of machines should be equal.
Assignment becomes a problem because each job requires different skills
and the capacity or efficiency of each person with respect to these jobs can
be different. This gives rise to cost differences. If each person is able to do
all the jobs with the same efficiency then all costs will be the same and each
job can be assigned to any person. When assignment is a problem it
becomes an optimisation problem. Therefore, you can compare an
assignment problem to a transportation problem. The cost element is given
and is a square matrix and the requirement at each destination is one and
the availability at each origin is also one.
Additionally, you have a number of origins, which equal the number of
destinations. Therefore, the total demand is equal to the total supply. There
is only one assignment in each row and each column. However, if you
compare this to a transportation problem, you will find that a general
transportation problem does not have the above mentioned limitations.
These limitations are peculiar to assignment problems only.
An assignment problem can be either balanced or unbalanced. Let‘s first
focus on a balanced assignment problem. A balanced assignment problem
is one where the number of rows = the number of columns (comparable to a
balanced transportation problem where total demand =total supply).
Minimisation case for an assignment problem
Figure 7.1 depicts the steps that must be performed to solve the
minimisation case for an assignment problem (AP).
Operations Research Unit 7
Sikkim Manipal University Page No. 134
Fig. 7.1: Steps to Solve the Minimisation Case for an Assignment Problem
Step 1: Determine the total opportunity cost matrix.
a) Arrive at a column opportunity cost matrix by subtracting the lowest entry
of each column of the given payoff matrix from all the entries in the
column.
b) Subtract the lowest entry of each row of the matrix obtained in
(a) from all the entries in its row.
The result of step 1b) gives the total opportunity cost matrix.
Step 2: Determine whether an optimal assignment can be made.
a) Cover all the zeros of the current total opportunity cost matrix with the
minimum possible number of horizontal and vertical lines.
b) If the number of lines in step 2a) equals the number of rows (or
columns) of the matrix, the problem can be solved. Make a complete
assignment so that the total opportunity cost involved in the assignment
is zero.
c) If the number of lines drawn in step 2a) is less than the number of rows
(or columns) of the matrix, proceed to step 3.
Step 3: Revise the total opportunity cost matrix.
a) Subtract the lowest entry in the uncovered cells of the current total
opportunity cost matrix from all the uncovered cells.
b) Add the same lowest entry to only those cells in which the covering lines
of step 2 cross.
Operations Research Unit 7
Sikkim Manipal University Page No. 135
The result of steps 3a) and 3b) is a revised total opportunity cost matrix.
Step 4: Repeat steps 2 and 3 until an optimal assignment having a total
opportunity cost of zero can be made.
Figure 7.2 depicts the various methods used to solve the assignment
problem by any of the following four methods.
- Enumeration method
- Simplex method
- Transportation problem
- Hungarian method
However, in this unit, we will focus on the most commonly used method, that
is the Hungarian method, to solve the assignment problems.
Fig. 7.2: Methods to Solve Assignment Problem
Objectives:
After studying this unit, you should be able to:
- interpret an assignment problem mathematically
- explain a routing problem
- describe the travelling salesman problem
- state the significance of the assignment problem
- compute the problem using the Hungarian method
- explain practical problems, such as routing problems and travelling
salesman problems
Operations Research Unit 7
Sikkim Manipal University Page No. 136
7.2 Mathematical Formulation of the Problem
Let xij be a variable defined by
¹ ´ ¦
=
if the i job is assigned to the j machine
job is not assigned to the j machine
x
th th
th
ij
1
th 0 if the i
Since only one job is assigned to each machine, you have
¿
=
n
i 1
xij = 1 and ¿
=
n
j 1
xij = 1
Hence, the total assignment cost is given by
z = ¿ ¿
= =
n
i 1
n
j 1
xij cij
Thus the assignment problem takes the following mathematical form:
Determine xij ≥ 0 (i, j =1… n)
Minimise z = ¿ ¿
= =
n
j 1
n
i 1
xij cij
Subject to the constraints
¿
=
n
i 1
xij = 1 j =1, 2, n
And ¿
=
n
j 1
xij = 1 i = 1, 2… n
With xij = 0 or 1
Note: In an assignment problem, if you add a real number to or subtract a
real number from each element of a row or column of the cost matrix, then
the optimum assignment for the modified matrix is also optimum for the
original one.
Operations Research Unit 7
Sikkim Manipal University Page No. 137
Self Assessment Questions
1. In an AP, the constraints are of equality type. (True/False)
2. The number of facilities should be equal to the number of resources.
(True/False)
3. A balanced assignment problem is one where the number of rows ≠ the
number of columns. (True/False)
7.3 Hungarian Method Algorithm
Hungarian method algorithm is based on the concept of opportunity cost
and is more efficient in solving assignment problems. The following steps
are adopted to solve an AP using the Hungarian method algorithm.
Step 1: Prepare row ruled matrix by selecting the minimum values for each
row and subtract it from the other elements of the row.
Step 2: Prepare column-reduced matrix by subtracting minimum value of
the column from the other values of that column.
Step 3: Assign zero row-wise if there is only one zero in the row and cross
(X) or cancel other zeros in that column.
Step 4: Assign column wise if there is only one zero in that column and
cross other zeros in that row.
Step 5: Repeat steps 3 and 4 till all zeros are either assigned or crossed. If
the number of assignments is equal to number of rows present, you have
arrived at an optimal solution, if not, proceed to step 6.
Step 6: Mark () the unassigned rows. Look for crossed zero in that row.
Mark the column containing the crossed zero. Look for assigned zero in that
column. Mark the row containing assigned zero. Repeat this process till all
the makings are done.
Step 7: Draw a straight line through unmarked rows and marked column.
The number of straight line drawn will be equal to the number of
assignments made.
Step 8: Examine the uncovered elements. Select the minimum.
- Subtract it from the uncovered elements.
- Add it at the point of intersection of lines.
Operations Research Unit 7
Sikkim Manipal University Page No. 138
- Leave the rest as is.
- Prepare a new table.
Step 9: Repeat steps 3 to 7 till optimum assignment is obtained.
Step 10: Repeat steps 5 to 7 till number of allocations = number of rows.
The assignment algorithm applies the concept of opportunity costs. The cost
of any kind of action or decision consists of the opportunities that are
sacrificed in taking that action.
Example 1
Consider the problem of assigning five jobs to five persons. The
assignment costs are depicted in table 7.1.
Table 7.1: Assignment Costs Table
Determine the optimum assignment schedule.
Solution: Applying Hungarian method
Table 7.2: Row Reduced Matrix (Optimum Assignment Schedule)
Column reduced matrix will be the same as each column‘s minimum
value is zero. Then you can start assigning the jobs as depicted in table
7.3.
Operations Research Unit 7
Sikkim Manipal University Page No. 139
Table 7.3: Column Reduced Matrix
Since the number of assignments is 5.
Therefore, the optimum assignment schedule is A to 5, B to 1, C to 4,
D to 3, and E to 2.
Self Assessment Questions
4. In Hungarian method, you prepare row-reduced matrix. (True/False)
5. The number of assignments should be equal to the number of rows for
an optimum solution. (True/False)
6. There can be more than one allocation in a row. (True/False)
7.4 Routing Problem
7.4.1 Unbalanced AP
Unbalanced assignment problem is an assignment where the number of
rows is not equal to the number of columns and vice versa. For example,
the number of machines may be more than the number of jobs or the
number of jobs may be more than the number of machines. In such a
situation, you have to introduce dummy rows or columns in the matrix. The
dummy rows or columns will contain all cost elements as zero. This
balances the problem and then you can use Hungarian method to find the
optimal assignment.
Unbalanced assignment problem: No. of rows ≠ No. of columns.
Operations Research Unit 7
Sikkim Manipal University Page No. 140
Solved problem 1
Solve the following AP.
Table 7.4: Assignment Operations Table
Solution: Introducing a dummy row and applying Hungarian method,
you have:
Table 7.5: Reduced Matrix
Table 7.6: Optimum Assignment Solution
Hungarian method leads to multiple solutions. Selecting (03, M2)
arbitrarily.
Therefore, the optimum assignment schedule is O1 to M1, O2 to M3, O3
to M2, and O4 to M4.
Operations Research Unit 7
Sikkim Manipal University Page No. 141
7.4.2 Infeasible assignments
It is sometimes possible that a particular person is incapable of doing certain
work or a specific job cannot be performed on a particular machine. The
solution of the assignment problem should take into account these
restrictions so that the infeasible assignment can be avoided. This can be
achieved by assigning a very high cost (say ∞ or M) to the cells where
assignments are prohibited, thereby, restricting the entry of this pair of job –
machine or resource – activity into the final solution. After inserting a high
value o at the cell we need to apply Hungarian method to solve the problem.
Solved problem 2
Solve the following A.P.
Table 7.7: Assignment Table
Solution: Introducing ―o‖ in places having dashes and applying
Hungarian method, you have:
Table 7.8: Assignment Table
Table 7.9: Row-reduced Matrix
Operations Research Unit 7
Sikkim Manipal University Page No. 142
Table 7.10: Column Reduced Matrix
Therefore, the optimum assignment schedule is T1 to P1, T2 to P4, T3 to
P2, and T4 to P3.
7.4.3 Maximisation in AP
Some assignment problems are phrased in terms of maximising the profit or
effectiveness or payoff of an assignment of people to tasks or of jobs to
machines. You cannot apply the Hungarian method to such maximisation
problems. Therefore, you need to reduce it to a minimisation problem.
It is easy to obtain an equivalent minimisation problem by converting every
number in the table to an opportunity loss. To do so, you need to subtract
every value from the highest value of the matrix and then proceed as usual.
You will notice that minimising the opportunity loss produces the same
assignment solution as the original maximisation problem.
Operations Research Unit 7
Sikkim Manipal University Page No. 143
Solved problem 3
Solve the following maximisation assignment problem.
Table 7.11: Assignment Table
Solution: Since it is a maximisation problem, subtract every value from
the maximum value of 90. Thus you have:
Table 7.12: Opportunity Loss Table
Table 7.13: Row Reduced Matrix
Table 7.14: Column Reduced Matrix
Operations Research Unit 7
Sikkim Manipal University Page No. 144
Therefore, the optimum assignment schedule is C1 to P2, C2 to P1, C3 to
P3, and C4 to P4.
Case-let
A tailoring unit has four sewing machines of different makes. Each
machine is capable of stitching all the required designs and patterns.
However, the profit factor differs for each assignment. The unit is looking
at maximising profit.
To do so, the unit needs to carry out an optimal assignment exercise of
assigning the right jobs to the right machines.
Solved problem 4
Five different machines can do any of the required five jobs with
different profits resulting from each assignment as depicted in table
7.15.
Table 7.15: Five Different Machines and Their Jobs
Find out the maximum profit possible through optimal assignment.
Solution: To start with, convert an assignment problem of profit
maximisation type into an assignment problem of cost minimisation type.
To do so, form a new modified matrix by subtracting each entry of the
matrix from the greatest entry 62 of the matrix. Use the resulting matrix
to solve the problem of cost minimisation type.
Operations Research Unit 7
Sikkim Manipal University Page No. 145
Table 7.16: Opportunity Matrix
Now apply the Hungarian method to this problem and get the row and
column reduced matrices as depicted in table 7.17 and 7.18.
Table 7.17: Row Reduction Table 7.18: Column Reduction
Then draw the minimum number of lines to suppress the zeros as
depicted in table 7.19.
Table 7.19: Lines Suppressing Zeros
Here, number of lines drawn N = 4 < number of rows (= 5).
The smallest uncovered entry is 4. Therefore, subtract 4 from all
uncovered entries and add it to all the entries on the intersection of the
lines as depicted in table 7.20.
Operations Research Unit 7
Sikkim Manipal University Page No. 146
Table 7.20: Lines not Suppressing Zeros
Again draw minimum number of lines to cover all the zeros as depicted
in table 7.21.
Table 7.21: Number of Lines = Number of Rows
Here the number of lines drawn N = 5 = number of rows.
Mark the zero entry with □ that occurs exactly once in a row and cancel
that row or column as depicted in table 7.22.
Table 7.22: Number of Lines ≠ Number of Rows
After the above steps, row 2 has zero as the last entry, mark it with □.
Now row 1 will have the third entry zero uncovered. Mark it with □. So
row 4 will have only one entry in cell (4, 4) with zero. This gives you the
optimal assignment as depicted in table 7.23.
Operations Research Unit 7
Sikkim Manipal University Page No. 147
Table 7.23: Optimum Assignment
Thus, x13 = x25 = x31 = x44 = x52 = 1 and rest all xij = 0. The optimal
assignment is depicted in table 7.24.
Table 7.24: Optimal Assignment
Therefore, maximum profit is 214.
Self Assessment Questions
7. In unbalanced AP, the number of rows ________ to number of
columns.
8. Hungarian method cannot be applied directly to _________ problem.
9. If some jobs cannot be assigned to some machines, then it is called
_________ assignment problem.
7.5 Travelling Salesman Problem
Routing problem
Network scheduling is a technique for planning and scheduling of large
projects. It has successfully been applied in transportation and
Operations Research Unit 7
Sikkim Manipal University Page No. 148
communication problems. A typical network problem involves finding route
from one node (origin) to another (destination) between which alternative
paths are available at various stages of the journey. The problem is to select
the route that yields minimum cost. A number of different constraints may be
placed on acceptable routes for instance, not returning to the node passed
or passing through each node just once. These kinds of problems are called
as routing problems.
A wide variety of problems other than routing may be developed in
connection with the construction and utilisation of networks. Here, you are
going to consider the special type of routing problem that occurs frequently
in OR – the travelling salesman problem.
Suppose a salesman has to visit ‗n‘ number of cities. He wishes to start from
a particular city, visit each city once, and then return to his starting point.
The objective is to select the sequence to visit the cities in such a way that
his total travelling time is minimised. Starting from a given city, the salesman
will have total of (n-1) different sequences. Further, since the salesman has
to visit all the ‗n‘ number of cities; the optimal solution remains independent
of selection of the starting point.
The problem can be represented as a network where the nodes and arcs
represent the cities and the distance between them respectively. Assume
that in a five city problem, a round trip of the salesman is given by the
following arcs.
(3,1), (1,2), (2,4), (4,5), (5,3)
These arcs in order are the first, second, third, fourth, and fifth directed arcs
for the trip. Generally the kth directed arc represents the kth leg of the trip that
is on leg k, the salesman travels from city i to city j.
(i, j = 1, 2… n; i= j )
To formulate the problem, whose solution will yield the minimum travelling
time, let the variables xijk be defined as:
¹ ´ ¦
=
0, otherwise
directed from city i to city j
th
1, if k is
ijk
x
Where, i, j, and k are integers that vary between 1 and n.
Operations Research Unit 7
Sikkim Manipal University Page No. 149
Following are the constraints of the problem.
a) Only one directed arc may be assigned to a specific k, thus
¿ ¿
j k
xijk = 1 k =1, 2, 3…n
i= j
b) Only one other city may be reached from a specific city i, thus
¿ ¿
j k
xijk =1, i = 1, 2…, n
c) Only one other city can initiate a direct arc to a specified city j, thus
¿¿
i k
xijk =1, j =1, 2... n
d) Given the kth directed arc ends at some specific city j, the (k+1)th directed
arc must start at the same city j; thus
¿i
xijk = ¿
r
xjr ( k +1) for all j and k.
i= j r = j
These constraints ensure that the round trip will consist of connected
arcs. The objective function is to minimise
z = ¿
i
¿¿
j k
dij xijk i= j
Where dij is the distance from city i to city j.
Self Assessment Questions
10. In travelling salesman problem, the objective is to visit each cities
________ __________.
11. Salesman has ________ different sequences if n is the number of
cities to be visited.
7.6 Summary
Let us recapitulate the important concepts discussed in this unit:
- This unit on assignment problems focuses on a special type of
transportation problem, where the objective is to allocate ‗n‘ number of
different facilities to ‗n‘ number of different tasks.
Operations Research Unit 7
Sikkim Manipal University Page No. 150
- Although an assignment problem can be formulated as a linear
programming problem, it is solved by a special method know as
Hungarian method.
- If the number of persons is the same as the number of jobs, the
assignment problem is said to be balanced. The unit also explains the
travelling salesman problem in brief.
7.7 Glossary
Opportunity cost matrix: A cost matrix that depicts the ‗lost opportunity‘
when allocating resource to activity.
7.8 Terminal Questions
1. Four jobs are to be done on four different machines. The cost in rupees
for producing ith on the j th machine is given below:
Machines
M1 M2 M3 M4
J1 15 11 13 15
J2 17 12 12 13
J3 14 15 10 14
J4 16 13 11 17
Assign the jobs to different machines to minimise the total cost.
2. A marketing manager has 5 salesmen and 5 sales districts. Considering
the capabilities of the salesman and the nature of districts, the marketing
manager estimates that the sales per month (in hundred rupees) for
each salesman in each district would be as follows.
Sales districts
A B C D E
1 32 38 40 28 40
2 40 24 28 21 36
3 41 27 33 30 37
4 22 38 41 36 36
5 29 33 40 35 39
Find the assignment of salesman to districts that will result in maximum
sales.
Job
s
Salesman
Operations Research Unit 7
Sikkim Manipal University Page No. 151
3. In a plant layout, there are five vacant places. The plant orders four
machines to be installed in the vacant places. The cost of installing is as
follows:
M/G A B C D E
M1 9 11 15 10 11
M2 12 9 - 10 9
M3 - 11 14 11 7
M4 14 8 12 7 8
Find the optimum assignment.
4. Find the assignment that maximises the total sale.
Zone
Sales men 1 2 3 4
M1 42 35 28 21
M2 30 25 20 15
M3 30 25 20 15
M4 24 20 16 12
7.9 Answers
Self Assessment Questions
1. True
2. True
3. False
4. True
5. True
6. False
7. ≠
8. Maximisation problem
9. Infeasible
10. Only once
11. (n-1)
Terminal Questions
1. The optimum assignment policy is
Job1 to machine 2, Job 2 to machine 4
Operations Research Unit 7
Sikkim Manipal University Page No. 152
Job 3 to machine1, Job 4 to machine 3
And the minimum assignment cost = Rs. (11+13+14+11) = Rs. 49.
Refer 7.3
2. Optimal assignment policy is salesman 1 to district B, 2 to A, 3 to E,
4 to C, and 5 to D. Hence the maximum sales = Rs.
(38+40+37+41+35) ×100 = Rs. 19,100. Refer 7.4.3
3. M1 – A2; M2 – B; M3 – E; M4 – D Total 38. Refer 7.4.2
4. M1 – 1; M2 – 2; M3 – 3; M4 – 4 Total 99. Refer 7.4.3
7.10 Case Study
Move-It Company
The Move-It company has two plants producing fork-lift trucks that are then
shipped to three distribution centres. The production costs are the same at
the two plants, and the cost of shipping for each truck is shown for each
combination of plant and distribution centre:
Distribution Centre
1 2 3
Plant
A Rs. 800 Rs. 700 Rs. 400
B Rs. 600 Rs. 800 Rs. 500
A total of 60 fork-lift trucks are produced and shipped per week. Each plant
can produce and ship any amount up to a maximum of 50 trucks per week,
so there is considerable flexibility on how to divide the total production
between the two plants so as to reduce shipping costs. However, each
distribution centre must receive exactly 20 trucks per week.
Assume that distribution centres 1, 2, and 3 must receive exactly 10, 20,
and 30 units per week respectively. For administrative convenience,
management has decided that a single plant will supply each distribution
centre totally, so that one plant will supply one distribution centre and the
other plant will supply the other two distribution centres. The choice of
assignments of plants to the distribution centres is to be made solely on the
basis of minimising total shipping cost.
Operations Research Unit 7
Sikkim Manipal University Page No. 153
Discussion Questions:
1. Formulate this case as an assignment problem by constructing the
appropriate cost table, including identifying the corresponding assignee
and tasks.
2. Obtain an optimal solution.
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
- Sharma S.C. Operation Research: Simulation and Replacement Theory.
Operations Research Unit 8
Sikkim Manipal University Page No. 154
Unit 8 Integer Programming Problem
Structure:
8.1 Introduction
Objectives
8.2 Types of Integer Programming Problems
8.3 Gomory‘s All-IPP Method
Construction of Gomory‘s constraints
8.4 All IPP Algorithm
8.5 Branch and Bound Technique
Branch and bound algorithm
8.6 Summary
8.7 Glossary
8.8 Terminal Questions
8.9 Answers
8.10 Case Study
8.1 Introduction
In the previous unit, you learnt about mathematical formulation of the
problem and Hungarian method algorithm. You also learnt the routing
problem and the travelling salesman problem. In this unit you will study
about Integer Programming Problem (IPP). The IPP is a special case of
Linear Programming Problem (LPP), where all or some variables are
constrained to assume non-negative integer values. In LPP, the decision
variables as well as slack or surplus variables were allowed to take any real
or fractional value. However, there are certain real life problems in which the
fractional value of the decision variables has no significance. For example, it
does not make sense saying 1.5 men working on a project or 1.6 machines
in a workshop. The integer solution to a problem can, however, be obtained
by rounding off the optimum value of the variables to the nearest integer
value. This approach can be easy in terms of economy of effort, time and
cost that might be required to derive an integer solution but this solution may
not satisfy all the given constraints. Secondly, the value of the objective
function so obtained may not be an optimal value. Integer programming
techniques come to our rescue during such scenarios.
Operations Research Unit 8
Sikkim Manipal University Page No. 155
Integer LP problems are those in which some or all of the variables are
restricted to integer (or discrete) values. An integer LP problem has
important applications. Capital budgeting, construction scheduling, plant
location and size, routing and shipping schedule, batch size, capacity
expansion, fixed charge, etc are few problems which demonstrate the areas
of application of integer programming.
In this unit, you will learn about Integer Programming Problem (IPP) and the
Gomory‘s all-IPP method. You will also learn all IPP algorithm and the
branch and bound technique.
Objectives:
After studying this unit, you should be able to:
- analyse an Integer Programming Problem (IPP)
- solve an IPP using Gomory‘s method
- describe the branch and bound technique
8.2 Types of Integer Programming Problems
Linear integer programming problems can be classified into three
categories:
1. Pure (all integer programming problems) - Here, all decision
variables are restricted to integer values.
2. Mixed integer programming problems - Here some, but not all, of the
decision variables are restricted to integer values.
3. Zero – one integer programming problems - Here, all decision
variables are restricted to integer values of 0 and 1.
An integer programming problem(IPP) can be described as follows:
Determine the value of unknowns x1, x2, … , xn
So as to optimise z = c1x1 + c2x2 + . . .+ cnxn
Subject to the constraints
ai1 x1 + ai2 x2 + . . . + ainxn = bi , i = 1,2,…,m
and xj > 0 j = 1, 2, … ,n
Where xj being an integral value for j = 1, 2, … , k s n.
Operations Research Unit 8
Sikkim Manipal University Page No. 156
If all the variables are forced to take only integral value that is k = n, it is
called an all (or pure) integer programming problem. If some of the variables
are restricted to take integral value and the remaining (n – k) variables take
any non-negative value, then the problem is known as a mixed integer
programming problem.
Self Assessment Questions
1. Integer programming is applied to problems that involve discrete
variables. (True/False)
2. If some variables take on non-negative values, then it is known as pure
IPP. (True/False)
8.3 Gomory’s All – IPP Method
In the previous section, you learnt the types of integer programming
problems. You will now learn the Gomory‘s all-IPP method. An optimum
solution to an IPP is obtained by using the simplex method, ignoring the
restriction of integral values. In the optimum solution, if all the variables have
integer values, the current solution will be the required optimum integer
solution.
Otherwise, the given IPP is modified by inserting a new constraint called
Gomory‘s constraint or secondary constraint. This constraint represents
necessary conditions for integrability and eliminates some non-integer
solution without losing any integral solution. On addition of the secondary
constraint, the problem is solved using dual simplex method to obtain an
optimum integral solution.
If all the values of the variables in the solution are integers, then an optimum
integer-solution is obtained, or else a new constraint is added to the
modified LPP and the procedure is repeated till the optimum solution is
derived. An optimum integer solution will be reached eventually after
introducing enough new constraints to eliminate all the superior non-integer
solutions. The construction of additional constraints, called secondary or
Gomory‘s constraints is important and needs special attention.
Operations Research Unit 8
Sikkim Manipal University Page No. 157
8.3.1 Construction of Gomory’s constraints
Consider a LPP or an optimum non–integer basic feasible solution. With the
usual notations, let the solution be displayed as depicted in table 8.1.
Table 8.1: Simplex Table
yB xB Y1 y2 y3 y4
y2 y10 y11 y12 y13 y14
y3 y20 y21 y22 y23 y24
y00 y01 y02 y03 y04
The optimum basic feasible solution is given by
xB = [x2, x3 ] = [y10, y20]; max z = y00
Since xB is a non-integer solution, we can assume that y10 is fractional. The
constraint equation is:
y10 = y11 x1 + y12 x2 + y13 x3+ y14 x4 (x1 = x4 = 0)
It reduces to:
y10 = y11 x1 + x2 + y14 x4 (x1 = x4 = 0)_____ (1)
Because x2 and x3 are basic variables (which implies that y12 = 1 and
y13 = 0). The above equation can be rewritten as:
x2= y10 - y11 x1 - y14 x4
This is a linear combination of non-basic variables.
Now, since y10 > 0 the fractional part of y10 must also be non-negative. You
can split each of yij in equation (1) into an integral part Iij, and a non-negative
fractional part, f1j for j = 0,1,2,3,4. After the breakup of equation (1), you can
write it as:
I10 + f10 = (I11 + f11) x2 + (I14 + f14) x4
Or
f10 - f11 x2 - f14 x4 = x2 + I11 x1 + I14x4 - I10 _____ (2)
If you compare equations (1) and (2), you will see that if you add an
additional constraint in such a way that the left-hand side of equation (2) is
an integer, then you will be forcing the non-integer y10 towards an integer.
Operations Research Unit 8
Sikkim Manipal University Page No. 158
The desired Gomory‘s constraint is:
f10 – f11 x1 – f11 x4 ≤ 0
It is possible to have f10 – f11 x1 – f11 x4 = h where h > 0 is an integer. Then
f10 = h + f11 x1 + f14 x4 is greater than one. This contradicts that 0 <fij< 1 for
j = 0, 1, 2, 3, 4.
Thus Gomory‘s constraint is
sla 10
j 1,4
ij j
j 1, 4
ij j 10
j 1, 4
ig j 10
or f x G (1) f
f x f or f . x f
÷ + = ÷
> ÷ s ÷
¿
¿ ¿
=
= =
Where Gsla (1) is a slack variable in the above first Gomory‘s constraint.
The additional constraint to be included in the given LPP is towards
obtaining an optimum all integer solution. After adding the constraint, the
optimum simplex table is depicted as in table 8.2.
Table 8.2: Optimum Simplex Table
yB xB y1 y2 y3 y4 Gsla(1)
y1 y10 y11 y12 y13 y14 0
y2 y20 y21 y22 y23 y24 0
Gsla(1) – f10 – f11 0 0 – f14 1
y00 y01 y02 y03 y04 y05
Since – f10 is negative, the optimal solution is unfeasible. Thus the dual
simplex method is to be applied for obtaining an optimum feasible solution.
After obtaining this solution, the above referred procedure is applied for
constructing second the Gomory‘s constraint. The process is to be
continued till all the integer solution has been obtained.
Self Assessment Questions
3. An optimum solution to IPP is first obtained by using _____________.
4. With the addition of Gomory‘s constraint, the problem is solved by
____________________.
Operations Research Unit 8
Sikkim Manipal University Page No. 159
8.4 All IPP Algorithm
In the previous section, you learnt the Gomory‘s all-IPP method. You will
now learn all the IPP algorithm.
The iterative procedure for the solution of integer programming problem is
as follows:
Figure 8.1 depicts the iterative procedure of IPP.
Fig. 8.1: Iterative Procedure of IPP
Operations Research Unit 8
Sikkim Manipal University Page No. 160
Step 1: Convert the minimisation IPP into maximisation form. Ignore the
integrality condition.
Step 2: Introduce the slack or surplus variables, if needed to convert the
inequations into equations and obtain the optimum solution of the given LPP
by using simplex algorithm.
Step 3: Test the integrality of the optimum solution:
a) If the optimum solution contains all integer values, an optimum basic
feasible integer solution has been obtained.
b) If the optimum solution does not include all integer values, then proceed
to the next step.
Step 4: Examine the constraint equations corresponding to the current
optimum solution. Let these equations be represented by:
y , x b , (i 0, 1, 2, ........, m ) 1
j i
n1
j 0
ij
¿ = =
=
Where n' denotes the number of variables and m' the number of
equations. Choose the largest fractional part of bis to find { }bi i
i
max
Let it be

(
¸
1
k b or write it is as
o
fk
Step 5: Express the negative fractions if any, in the kth row of the optimum
simplex table as the sum of a negative integer and a non-negative fraction.
Step 6: Find the Gomorian constraint
¿
=
>
n 1
j 0
kj j ko
f , x f
And add the equation
¿
=
= ÷ +
n1
j 0
sla ko kj j G (1) f f . x
To the current set of equation constraints
Operations Research Unit 8
Sikkim Manipal University Page No. 161
Step 7: Start with a new set of equation constraints. Find the new optimum
solution by dual simplex algorithm, that is, choose a variable to enter into
the new solution having the smallest ratio:
{(Cj– Zj)/ yij; yij<0} so that Gsla (1) is the initial leaving basic variable.
Step 8: If the new optimum solution for the modified LPP is an integer
solution, it is also feasible and optimum for the given IPP. If it is not an
integer solution, then return to step 4 and repeat the process until an
optimum feasible integer solution is obtained.
Solved Problem: 1
Find the optimum integer solution of the following all IPP.
Maximise z = x1 + x2
Subject to the constraints
3x1 +2 x2 ≤5;
x2 ≤2
x1, x2 > 0 and are integers
Solution: Step 1: Introduce the slack variables s1>0 and s2 > 0, using
simplex method an optimum non–integer solution is depicted in the
table 8.3:
Initial iteration. Non- integer optimum solution
Table 8.3: Optimal Non-integer Solution
cJ --> 1 1 0 0
CB
Variable in
Basis B
Solution Values
b (= xB)
X1 X2 S1 S2
1 X1 1/3 1 0 1/3 -2/3
1 X2 2 0 1 0 1
Z = 7/2 cj-zj 0 0 -1/3 -1/3
Here, since all cj-zj ≤ 0, the optimal solution is : x1=1/3 and x2= 2 and
max Z = 7/2
Step 2: Since x1 is the only basic variable whose value is a non-negative
fraction, we consider the first row for generating the Gomory Cut.
Considering x1 equation as the source row we write:
1/3 = x1+ 0 x2 + 1/3 s1 – 2/3 s2 (x1- source row)
Operations Research Unit 8
Sikkim Manipal University Page No. 162
The factoring of solution of the x1 – source row gives:
(0 + 1/3) = ( 1+0 ) x1 + (0 + 1/3) s1 + (-1 + 1/3) s2
We note that each of the non-integer coefficient is factored into integer
and fractional parts in such a manner that the fractional part is strictly
positive.
Rearranging the equation so that all of the integer coefficient appear on
the left hand side. This gives:
1/3 + (s2 – x1) = 1/3 s1+ 1/3 s2
Since x1 and s2 are non-negative, left hand side must satisfy:
1/3 ≤ 1/3 s1+ 1/3 s2
1/3 + sg1 = 1/3 s1+ 1/3 s2
- 1/3 =sg1- 1/3 s1- 1/3 s2
Where sg1 is the new non negative (Integer) slack variable.
By adding this equation (also called Gomory cut) at the bottom of the
table 8.4 The new table obtained is depicted in table 8.5.
Table 8.4: Optimal but Feasible Solution
cJ --> 1 1 0 0 0
CB
Variable in
Basis B
Solution
Values b
(= xB)
X1 X2 S1 S2 Sg1
1 X1 1/3 1 0 1/3 -2/3 0
1 X2 2 0 1 0 1 0
0 Sg1 -1/3 0 0 -1/3 -1/3 1 
Z = 7/2cj-zj
Ratio: Min(cj-zj) / y3J(<0)
0
-
0
-
-1/3
1

-1/3
1
0
-
Step 4: Since solution depicted in table 8.4 is infeasible, apply the dual
simplex method to find feasible as well as an optimal solution. The key
row and key columns are marked in table 8.4. The new solution is
obtained by applying following row operations and is depicted in
table 8.5.
Operations Research Unit 8
Sikkim Manipal University Page No. 163
R3(new)R3(old) X -3 ; R1(new)R1(old) X – (1/3) R3(new)
Table 8.5: Optimal Integer Solution
cJ --> 1 1 0 0 0
CB
Variable in
Basis B
Solution Values b
(= xB)
X1 X2 S1 S2 Sg1
1 X1 0 1 0 0 -1 1
1 X2 2 0 1 0 1 0
0 Sg1 1 0 0 1 1 -3
Z = 2cj-zj 0 0 0 0 -1
The solution provided in the table 8.5 is X1 = 0, X2 = 2, Sg1 = 2 and
Max Z = 2.
This also satisfies the integer requirement.
Self Assessment Questions
5. For Gomory‘s constraint select the variable whose fractional value is
more? (True/False)
6. Optimum values in a pure IPP can be x=2 and y=3.5. (True/False)
8.5 Branch and Bound Technique
In the previous section, you learnt the IPP algorithm. You will now learn the
branch and bound technique. Sometimes a few or all the variables of an IPP
are constrained by their upper or lower bounds or by both. The most general
technique for a solution of such constrained optimisation problems is the
branch and bound technique. The technique is applicable to both all (or
pure) IPP as well as mixed IPP. The technique for a maximisation problem
is discussed as follows:
Let the IPP be
Maximise ¿
=
=
n
j 1
z c j x j _______________________ (1)
Subject to the constraints
a x b i 1,2,....,m
n
j 1
¿ ij j s i =
=
___________________ (2)
Operations Research Unit 8
Sikkim Manipal University Page No. 164
xj is integer valued, j = 1, 2, …….., r (< n) ___________ (3)
xj > 0 …………………. j = r + 1, …….., n ______________ (4)
Further let us suppose that for each integer valued xj, we can assign lower
and upper bounds for the optimum values of the variable by
Lj ≤ xj ≤ Uj j = 1, 2, …., r ______________________ (5)
This is the main idea behind ―the branch and bound technique.‖
Consider any variable xj, and let k be some integer value satisfying
LjsksUj – 1. Then clearly an optimum solution (1) through (5) also satisfies
either linear constraint.
xj > k + 1 _________________________________ (6)
Or the linear constraint xj ≤ k ______________________ (7)
To explain how this partitioning helps, let us assume that there were no
integer restrictions (3), and it yields an optimal solution to LPP – (1), (2), (4)
and (5). This indicates x1 = 1.66 (for example).
Then you formulate and solve two LPPs each containing (1), (2) and (4). But
(5) for j = 1 is modified to be 2 ≤ x1 ≤ U1 in one problem and L1 ≤ x1 ≤ 1 in the
other. Further to each of these problems, process an optimal solution
satisfying integer constraint (3).
Then the solution having the larger value for z is clearly the optimum for the
given IPP. However, it usually happens that one (or both) of these problems
have no optimal solution satisfying (3) and thus some more computations
are required. Now let us discuss, stepwise, the algorithm that specifies how
to apply the partitioning (6) and (7) in a systematic manner to finally arrive at
an optimum solution.
Let‘s start with an initial lower bound for z, say z(0) at the first iteration, which
is less than or equal to the optimal value z*. This lower bound may be taken
as the starting Lj for some xj.
In addition to the lower bound z(0), you also have a list of LPPs (to be called
master list) differing only in the bounds (5). To start with (the 0th iteration)
the master list contains a single LPP consisting of (1), (2), (4) and (5). Let us
now discuss the procedure that specifies how the partitioning (6) and (7) can
Operations Research Unit 8
Sikkim Manipal University Page No. 165
be applied systematically to eventually get an optimum integer-valued
solution.
8.5.1 Branch and bound algorithm
At the tth iteration (t = 0, 1, 2 …), apply the following steps to get an optimum
integer solution.
Step 0: If the master list is not empty, choose an LPP from it. Otherwise
stop the process, go to step 1.
Step 1: Obtain the optimum solution to the chosen problem. If either
i) It has no feasible solution or
ii) The resulting optimum value of the objective function z is less than
or equal to z(t), then let z(t+1) = z(t) and return to step 0; otherwise go
to step 2.
Step 2: If the obtained optimum solution satisfies the integer constraints (3)
then record it. Let z(t+1) be the associated optimum value of z; return to
step 0. Otherwise move to step 3.
Step 3: Select any variable xj, j = 1, 2, …., p, that does not have an integer
value in the obtained optimum solution to the LPP, chosen in step 0. Let *
x j
denote this optimal value of xj. Add two LPPs to the master list. These LPPs
are identical to the LPPs chosen in step 0, except that, the lower bound
on xj is replaced by (
¸
(
¸
*
x j + 1. Let z(t+1) = z(t); return to step 0.
Note: At the termination of the algorithm, if a feasible integer valued solution
yielding z(t) has been recorded, it is optimum, or else no integer valued
feasible solution exists.
Solved Problem 2:
Use Branch and bound method to solve the following LPP:
Maximise Z = 7x1 + 9x2 subject to the constraints:
-x1 + 3x2 ≤ 6, 7x1 + x2 ≤ 35, x2 ≤ 7, x1 , x2 ≥ 0 and are integers.
Solution:
Step 1: Ignoring the integer restriction, the optimum solution to the given
LPP can easily be obtained as:
x1= 9/2 , x2 = 7/2 and maximise z = 63.
Operations Research Unit 8
Sikkim Manipal University Page No. 166
Step 2: Since the solution is non-integral, let us choose x1, i.e., x1* = 9/2
being the largest fractional value.
Step 3: Considering the value of z as initial upper bound, i.e., z = 63; the
lower bound is obtained by rounding off the values of x1 and x2 to
the nearest integers, i.e., x1 = 4 and x2 = 3. Then the lower bound
is z = 55.
Step 4: Since [x1*] = [9/2] = 4; We have,
Sub-Problem 1: Maximise Z = 7x1 + 9x2 subject to:
-x1 + 3x2 ≤ 6, 7x1 + x2 ≤ 35, 0 ≤ x1 ≤ 4, 0 ≤ x2 ≤ 7.
Sub-Problem 2: Maximise Z = 7x1 + 9x2 subject to:
-x1 + 3x2 ≤ 6, 7x1 + x2 ≤ 35, x1 ≥ 5, 0 ≤ x2 ≤ 7.
Step 5: Optimum solutions for the sub-problems are obtained as follows:
Sub-Problem 1: x1= 4, x2=10/3 and maximum z = 58.
Sub-Problem 2: x1= 5, x2=0 and maximum z = 35.
Since the solution for sub-problem 1 is not in integers, we
subdivide it into the following two sub-problems:
Sub-Problem 3: Maximise Z = 7x1 + 9x2
subject to
-x1 + 3x2 ≤ 6,
7x1 + x2 ≤ 35 ,
and 0 ≤ x1 ≤ 4, 0 ≤ x2 ≤ 3.
Sub-Problem 4: Maximise Z = 7x1 + 9x2
subject to
-x1 + 3x2 ≤ 6,
7x1 + x2 ≤ 35,
and 0 ≤ x1 ≤ 4, x2 ≥ 4.
Step 6: The optimum solutions for the sub-problem 3 and 4 are:
Sub-Problem 3: x1 = 4, and x2= 3 and maximum z=55.
Sub-Problem 4: No feasible solution.
Step 7: Among the recorded integer valued solutions, since the largest
value of z is 55; the required optimum solution is v1 = 4, x2 = 3
and maximum z = 55.
Operations Research Unit 8
Sikkim Manipal University Page No. 167
The entire branch and bound procedure for the given problem is depicted in
figure 8.2:
Fig. 8.2: The Flow of Sub-Problems in the Branch and Bound Problems
Self Assessment Questions
7. Branch and bound technique is applied when some variables have
upper or lower bounds. (True/False)
8. We start the branch and bound technique with the lower bound.
(True/False)
8.6 Summary
Let us recapitulate the important concepts discussed in this unit:
- Linear integer programming problems can be classified into pure, mixed
integer and zero- one interger programming problems.
- Gomory‘s constraint represents the necessary conditions for integrability
and eliminates some non-integer solution without losing any integral
solution
- The most general technique for a solution of such constrained
optimisation problems is the branch and bound technique.
- This unit also describes two algorithms to determine the optimal solution
for an integer programming problem. One is the All IPP algorithm and
the other is the branch and bound algorithm.
Operations Research Unit 8
Sikkim Manipal University Page No. 168
8.7 Glossary
Non-basic variables: variables set to value zero in order to solve the value
of basic variables.
Equation constraints: A constraint written in the form of an equation
containing a =.
8.8 Terminal Questions
1. Use branch and bound technique to solve the following problem:
Maximise z = 3x1 + 3x2 + 13 x3
Subject to
– 3x1 + 6x2 + 7x3 ≤ 8
6x1 – 3x2 + 7x3 ≤ 8
0 ≤ xj ≤ 5, j = 1, 2, 3
2. What is integer programming?
3. Explain the Gomory‘s all integer algorithm of an IPP?
8.9 Answers
Self Assessment Questions
1. True
2. False
3. Simplex method
4. Dual simplex method
5. True
6. False
7. True
8. False
Terminal Questions
1. At the end of the 8th iteration we get the optional solution to the IPP is x1
= x2 = 0, x3 = 1, z* = 13. For more details refer Section 8.5
2. An integer programming problem(IPP) can be described as follows:
Determine the value of unknowns x1, x2, … , xn
Operations Research Unit 8
Sikkim Manipal University Page No. 169
So as to optimise z = c1x1 + c2x2 + . . .+ cnxn¬
Subject to the constraints
ai1 x1 + ai2 x2 + . . . + ainxn = bi , i = 1,2,…,m
and xj 0 j = 1, 2, … ,n
Where xj being an integral value for j = 1, 2, … , k n.
If all the variables are forced to take only integral value that is k = n, it is
called an all (or pure) integer programming problem.
For more details refer to Section 8.2
3. The given IPP is modified by inserting a new constraint called Gomory‘s
constraint or secondary constraint. For more details refer to Section 8.3
8.10 Case Study
Managing Stocks
Mr. Don Fredricks, an expeditor at the local warehouse for Wood World
sighed as he moved boxes and boxes of inventory to the side in order to
reach the shelf where the particular item he needed was located. He
dropped to his hands and knees and squinted at the inventory numbers
lining the bottom row of the shelf. He did not find the number he needed. He
worked his way up the shelf until he found the number matching the number
on the order slip. Just his luck! The item was on the top row of the shelf!
Don walked back through the warehouse to find a ladder, stumbling over
boxes of inventory littering his path. When he finally climbed the ladder to
reach the top shelf, his face crinkled in frustration. Not again! The item he
needed was not in stock! All he saw above the inventory number was an
empty space covered with dust! Don trudged back through the warehouse to
make the dreadful phone call. He dialled the number of Annie, the
saleswoman on the kitchen showroom floor of Wood World, and informed
her that the particular light fixture, the customer had requested, was not in
stock. He then asked her if she wanted him to look for the rest of the items
in the kitchen set. Annie told him that she would talk to the customer and
call him back.
Annie hung up the phone and frowned. Mr. David, her customer, would not
be happy. Ordering and receiving the correct light fixture from the regional
warehouse would take at least two weeks. Annie then paused to reflect
Operations Research Unit 8
Sikkim Manipal University Page No. 170
upon business during the last month and realised that over 80 percent of the
orders for kitchen sets could not be filled because items needed to complete
the sets were not in stock at the local warehouse. She also realised that
Wood World was losing customer goodwill and business because of stock
outs. The furniture mega store was gaining a reputation for slow service and
delayed deliveries causing customers to turn to small competitors that sold
furniture directly from the showroom floor. Annie decided to investigate the
inventory situation at the local warehouse. She walked the short distance to
the building next door and gasped when she stepped inside the warehouse.
What she saw could only be described as chaos. Spaces allocated for some
items were overflowing into the aisles of the warehouse while other spaces
were completely bare. She walked over to one of the spaces overflowing
with inventory to discover the item that was overstocked. She could not
believe her eyes! The warehouse had at least 30 rolls of pea-green
wallpaper! No customer had ordered pea-green wallpaper since 1973! Annie
marched over to Don demanding an explanation. Don said that the
warehouse had been in such a chaotic state since his arrival one year ago.
He said the inventory problems occurred because management had a policy
of stocking every furniture item on the showroom floor in the local
warehouse. Management only replenished inventory every three months,
and when inventory was replenished, management ordered every item
regardless of if it had been sold. Don also said that he had tried to make
management aware of the problems with overstocking unpopular items and
under stocking popular items, but that management would not listen to him
because he was simply an expeditor.
Annie understood that Wood World required a new inventory policy. Not
only was the mega store losing money by making customers unhappy with
delivery delays, but it was also losing money by wasting warehouse space.
By changing the inventory policy to stock only popular items and replenish
them immediately when they are sold, Wood World would ensure that the
majority of customers receive their furniture immediately and that the
valuable warehouse space was utilised effectively.
Annie needed to sell her inventory policy to the management. Using her
extensive sales experience, she decided that the most effective sales
strategy would be to use her kitchen department as a model for the new
inventory policy. She would identify all kitchen sets comprising 85 percent of
Operations Research Unit 8
Sikkim Manipal University Page No. 171
customers orders. Given the fixed amount of warehouse space allocated to
the kitchen department, she would identify the items Wood World should
stock in order to satisfy the greatest number of customer orders. She would
then calculate the revenue from satisfying customer orders under the new
inventory policy, using the bottom line to persuade management to accept
her policy. Annie analysed her records over the past three years and
determined that 20 kitchen sets were responsible for 85 percent of the
customer orders. These 20 kitchen sets were composed of up to eight
features in a variety of styles. Annie listed each feature and its popular
styles as depicted in table 8.6.
Annie then created a table showing the 20 kitchen sets and the particular
features comprising each set. To simplify the table, she used the codes
shown in parentheses above to represent the particular feature and style.
The table is given on the next page. For example, kitchen set 1 consists of
floor tile T2, wallpaper W2, light fixture L4, cabinet C2, counter top O2,
dishwasher D2, sink S2 and range R2. Notice that sets 14 through 20 do not
contain dishwashers. Annie knew she had only a limited amount of
warehouse space allocated to the kitchen department. The warehouse could
hold 50 square feet of tile and 12 rolls of wallpaper in the inventory bins. The
inventory shelves could hold two light fixtures, two cabinets, three
countertops and two sinks. Dishwashers and ranges are similar in size, so
Wood World stored them in similar locations. The warehouse floor could
hold a total of four dishwashers and ranges. Every kitchen set always
includes exactly 20 square feet of tile and exactly five rolls of wallpaper.
Therefore, 20 square feet of a particular style of tile and five rolls of a
particular style of wallpaper are required for the styles to be in stock.
Floor Tile Wallpaper Light Fixtures Cabinets
(T1) White textured
tile
(W1) Plain ivory
paper
(L1) One large
rectangular
frosted fixture
(C1) Light solid
wood cabinets
(T2) Ivory textured
tile
(W2) Ivory paper
with dark brown
pinstripes
(L2) Three small
square frosted
fixtures
(C2) Dark solid
wood cabinets
Operations Research Unit 8
Sikkim Manipal University Page No. 172
(T3) White
checkered tile with
blue trim doors
(W3) Blue paper
with marble
texture
(L3) One large
oval frosted
fixture
(C3) Light wood
cabinets with
glass
(W4) Light yellow
paper with
marble texture
(L4) Three small
frosted globe
fixtures
(C4) Dark wood
cabinets with
glass
Countertops Dishwashers Sinks Ranges
(O1) Plain light
wood countertops
(D1) White
energy- saving
dishwasher
(S1) Sink with
separate hot and
cold water taps
(R1) White
electric oven
(O2) Stained light
wood countertops
(D2) Ivory
energy- saving
dishwasher
(S2) Divided sink
with separate hot
and cold water
taps and garbage
disposal
(R2) Ivory electric
oven
(O3) White lacquercoated
countertops
(S3) Sink with
one hot and cold
water tap
(R3) White gas
oven
O4) Ivory lacquercoated
countertops
(S4) Divided sink
with one hot and
cold water tap
and garbage
disposal
(R4) Ivory gas
oven
Discussion Questions:
1. Formulate and solve an IPP model to maximise the total number of
kitchen sets (and thus the number of customer orders) Wood World
stocks in the local warehouse. Assume that when a customer orders a
kitchen set, all the particular items composing that kitchen set are
replenished at the local warehouse immediately.
2. How many of each feature and style should Wood World stock in the
local warehouse? How many different kitchen sets are in stock?
3. Annie convinces the management that the kitchen department should
serve as a testing ground for future inventory policies. To provide
adequate space for testing, management decides to allocate all the
space freed by the nursery department to the kitchen department. The
Operations Research Unit 8
Sikkim Manipal University Page No. 173
extra space means that the kitchen department can store not only the
dishwashers and ranges from part (c), but also all sinks, all countertops,
three of the four light fixtures and three of the four cabinets. How much
does the additional space help?
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
Operations Research Unit 9
Sikkim Manipal University Page No. 174
Unit 9 Infinite Queuing Models
Structure:
9.1 Introduction
Objectives
9.2 Queuing Theory
9.3 Operating Characteristics of a Queuing System
9.4 Constituents of a Queuing System
Arrival pattern
Completely random arrivals
9.5 Service Facility
9.6 Queue Discipline
Customer behaviour
Server behaviour
9.7 Summary
9.8 Glossary
9.9 Terminal Questions
9.10 Answers
9.11 Case Study
9.1 Introduction
In the previous unit, you learnt the Integer Programming Problem (IPP) and
the Gomory‘s all-IPP method. You also learnt all IPP algorithm and the
branch and bound technique. This chapter introduces the idea of queuing
theory. The formation of waiting lines or queues is common and is usually
formed by elements, people or events that are awaiting some form of
service. Who isn‘t familiar with queues in our daily existence? On the way to
your workplace, you have to wait at the bus stop, wait for the traffic light to
turn green, wait for your turn to enter the lift. You come across cars in line at
petrol pumps for service, queues to deposit cash at the bank, mobile
subscriber waiting for a new connection, customers waiting at the checkout
counter, goods in production/warehouse waiting to be shipped and even
aircrafts waiting for a free runway to take off.
The one thing common to all the above mentioned examples is that
customers arrive at a service centre and wait for their turn to receive the
service. Though the arrival of customers is irregular and the time taken for
Operations Research Unit 9
Sikkim Manipal University Page No. 175
the service is not consistent, queues build up during hours of demand and
disappear during the lull period.
In commercial or industrial situations, it may not be economical to have
waiting lines. On the other hand, it may not be feasible or economical to
totally avoid queues. An executive dealing with the system then would like to
find the optimal facilities to be provided. In this unit, you will learn about
queuing theory, operating characteristics and constituents of a queuing
system. You will also learn about service facility and queue discipline.
Objectives:
After studying this unit, you should be able to:
- describe the queuing process
- evaluate the Queuing Theory
- differentiate the customer‘s behaviour.
9.2 Queuing Theory
Queuing theory had its beginning in the research work of a Danish engineer
named A. K. Erlang. In 1909, Erlang experimented with fluctuating demand
in telephone traffic. Eight years later he published a report addressing the
delays in automatic dialling equipment. At the end of World War II, Erlang‘s
early work was extended to more general problems and to business
applications of waiting lines.
Queuing theory is a collection of mathematical models of various queuing
systems. It is based on probability concepts. It gives an indication of the
capability of a given system and the possible changes in its performance
with modification to the system. All the constraints of the process are not
taken into account in the formulation of a queuing model. The application of
queuing theory cannot be viewed as an optimisation process as there is no
maximisation or minimisation of an objective function.
Formation of Queues
Queues or waiting lines arise when the demand for a service facility
exceeds the capacity of that facility, that is, the customers do not get service
immediately upon request but must wait, or the service facilities stand idle
and wait for customers.
Operations Research Unit 9
Sikkim Manipal University Page No. 176
For example
- Supermarkets must decide how many cash register checkout positions
should be opened.
- Gasoline stations must decide how many pumps should be opened and
how many attendants should be on duty.
- Manufacturing plants must determine the optimal number of mechanics
to have on duty in each shift to repair machines that break down.
- Banks must decide how many teller windows to keep open to serve
customers during various hours of the day.
Self Assessment Questions
1. Customers arrive at a bank at regular intervals. (True/False)
2. Queuing identifies the optimal service facilities to be provided.
(True/False)
3. Queuing theory is based on the deterministic model. (True/False)
9.3 Operating Characteristics of a Queuing System
In the previous section, you learnt about queuing theory. You will now learn
the operating characteristics of a queuing system. A queuing model has the
following operating characteristics which enables us to understand and
efficiently manage a queue:
- Queue length: The number of customers in the waiting line reflects one
of the two conditions. Short queues could mean either good customer
service or too much capacity. Similarly, long queues could indicate
either low server efficiency or the need to increase capacity
- Number of customers in system: The number of customers in queue
and also those being served in the queue relates to the service
efficiency and capacity. Large values imply congestion, potential
customer dissatisfaction and a need for more capacity.
- Waiting time in queue: Long lines do not reflect long waiting times if
the service rate is fast. However, when waiting time seems long to
customers, they perceive that the quality of service is poor. Long waiting
times may indicate a need to adjust the service rate of the system or
change the arrival rate of customers.
Operations Research Unit 9
Sikkim Manipal University Page No. 177
- Waiting time in system: The total elapsed time from entry into the
system until exit from the system may indicate problems with customers,
server efficiency or capacity. If some customers are spending too much
time in the service system, there may be a need to change the priority
discipline, increase productivity or adjust capacity in some way.
- Service facility utilisation: The collective utilisation of the service
facilities reflects the percentage of time the facilities are busy.
Management is interested in maintaining high utilisation but this
objective may adversely impact the other operating characteristic.
A queuing system is said to be in transient state when its operating
characteristics are dependent upon time. If the operating characteristics
become independent upon time, the queuing system is said to be in a
steady state.
Self Assessment Questions
4. _________ represents number of customers waiting in the queue.
5. ________ ________ times may indicate a need to adjust the service
rate of the system or change the arrival rate of customers
6. _____________ represents the percentage of time the facilities are
busy.
9.4 Constituents of a Queuing System
In the previous section, you learnt the operating characteristics of a queuing
system. You will now learn the constituents of a queuing system. The
constituents of a queuing system include arrival pattern, service facility and
queue discipline.
- Arrival pattern: It is the average rate at which the customers arrive.
- Service facility: Examining the number of customers served at a time
and the statistical pattern of time taken for service at the service facility.
- Queue discipline: The common method of choosing a customer for
service amongst those waiting for service is ‗First Come First Serve‘.
Operations Research Unit 9
Sikkim Manipal University Page No. 178
Figure 9.1 depicts the components of a queuing system.
Fig. 9.1: Components of a Queuing System
Figure 9.2 depicts an example showing the components of a queuing
system.
Fig. 9.2: Car Wash System Showing Components of a Queuing System
9.4.1 Arrival pattern
The arrival of customers can be regular as in case of an appointment
system of a doctor or flow of components on a conveyor belt. The regular
pattern of arrivals is neither very common nor very easy to deal with
mathematically. The following are the important arrival characteristics:
Operations Research Unit 9
Sikkim Manipal University Page No. 179
1. Size of the population: Unlimited (infinite) or limited (finite)
2. Pattern of arrivals (statistical distribution)
3. Behaviour of arrivals
Our primary concern is the pattern of completely random arrivals.
9.4.2 Completely random arrivals
If the number of potential customers is infinitely large, then probability of an
arrival in the next interval of time will not depend upon the number of
customers already in the system. (The assumption is valid by and large,
except for queues involving a small finite number of customers.) When the
arrivals are completely random, they follow the Poisson distribution, which
equals to the average number of arrivals per unit time.
Sometimes it is necessary to distinguish between groups of customers, such
as male and female callers, or large and small aircrafts during the arrivals.
There are several other types of arrival patterns which shall not be dealt with
due to their restricted applications.
Self Assessment Questions
7. Every queuing process has an arrival pattern, a service facility and a
queue discipline as its constituents.(True/False)
8. If the arrivals are completely random, then it follows Poisson
distribution. (True/False)
9.5 Service Facility
In the previous section, you learnt the constituents of a queuing system. You
will now learn the service facility. Service Facility is based on three
parameters – Availability of service, number of service centres and duration
of service.
i) Availability of service
It is necessary to examine if there are any constraints that reduce the
number of customers to be served at a time, apart from specifying the time
span of the availability of service. For example, in a waiting line for a
suburban train, apart from the timings of the train services, the probability
distribution of the number of passengers that can be accommodated in a
train that arrives is relevant.
Operations Research Unit 9
Sikkim Manipal University Page No. 180
ii) Number of service centres
If only one service centre is referred to as a service channel, obviously only
one customer can be served at a time. There will definitely be more than
one service centre and the behaviour of the queues will vary with the
number of channels available for service. Multiple service channels may be
arranged in series or in parallel.
Multiple service channels are arranged in series when a customer has to go
through several counters one after another with each providing a different
part of the service. For instance, bank counters where a customer has to go
to at least two counters to withdraw is an example of arrangement in series.
On the other hand ticket booths in a railway station have multiple channels
with parallel arrangement.
iii) Duration of service
This is the length of time taken to serve a customer. This can be constant or
varying.
(a) Constant service time: Though not in practice, an assumption that
service time is constant holds true, if the pattern of arrivals is very
irregular.
(b) Completely random service time: The service time can be considered
completely random when:
- The server does not distinguish between the various arrivals.
- The server does not deliberately change the duration of service on
the basis of the time taken to serve the previous arrival.
- The server forgets the time for which he/she has been serving a
customer.
Under the above mentioned conditions, the service time follows
exponential distribution which means it is equal to reciprocal of the
average rate of service.
(c) Service time following Erlang distribution
Sometimes, the assumption of an exponential distribution for service time is
not valid. Hence Erlang family of service time distributions is used.
Table 9.1 depicts examples of queuing discipline in daily life.
Operations Research Unit 9
Sikkim Manipal University Page No. 181
Table 9.1: Examples of Queuing Discipline in Daily Life
Sr.
No.
Situation
Arriving
Customers
Service Facility
1. Passage of customers through
supermarket checkout
Shoppers Checkout counters
2. Flow of automobile traffic through
a road network
Automobiles Road Network
3. Transfer of electronic messages Electronic
messages
Transmission lines
4. Banking transactions Bank patrons Bank tellers
5. Flow of computer programmers
through a computer system
Computer
Programmers
Central processing
unit
6. Sale of theatre tickets Theatre visitors Ticket booking
windows
7. Arrival of trucks to carry fruits and
vegetables from a central market
Trucks Loading crews and
facilities
8. Registration of unemployed at
employment exchange
Unemployed
personnel
Registration
assistants
9. Occurrences of fires Fires Firemen and
equipment
10. Flow of ships to the seashore Ships Harbour and
docking facilities
11 Calls at police control room Service calls Policemen
Self Assessment Questions
9. Multiple service channels may be arranged in ______________ or in
_________.
10. The service time can be __________ or _________.
9.6 Queue Discipline
In the previous section, you learnt the service facility. You will now learn
about queue discipline. The queue discipline is the order or manner in which
customers from the queue are selected for service. There are a number of
ways in which customers in the queue are served:
Static queuing disciplines are based on the individual customer‘s status in
the queue like:
Operations Research Unit 9
Sikkim Manipal University Page No. 182
I. First In First Out (FIFO) – If the customers are served in the order of
their arrival. E.g., Toll road
II. Last In First Out (LIFO) – Where the last in items are operated first.
E.g., Cargo handling situations, getting down from the bus, etc.
Dynamic queuing disciplines are based on the individual customer attributes
in the queue:
I. Service in Random Order (SIRO): Here customers are selected for
service at random irrespective of their arrivals in the service system.
II. Priority service: Under this rule, customers are grouped in priority
classes on the basis of some attributes such as service time or
urgency, and FIFO rule is used within each class to provide service.
E.g., premium queues in temples like Tirupati.
III. Pre-emptive priority (or Emergency): Under this rule, an important
customer is allowed to enter into the service immediately after entering
into the system even if a customer with lower priority is already in
service. E.g., Emergency service in hospitals, ambulance and fire
brigades in traffic signals
9.6.1 Customer behaviour
a) Balking: Arriving customers are said to ―balk‖ if they do not join a queue
because of their reluctance to wait or some customers even before
joining the queue get discouraged by seeing the number of customers
already in service system or estimating the excessive waiting time for a
desired service, decide to return for service at a later time. In queuing
theory this is known as balking.
b) Collusion: Customers may be in collusion, meaning that only one
person would join the queue, but demand service on behalf of several
customers.
c) Reneging: Impatient customers who would not wait beyond a certain
time and leave the queue are said to renege.
For example, a customer who has just arrived at a grocery store and
finds that the salesmen are busy in serving the customers already in the
system, will either wait for service till his patience is exhausted or
estimates that his waiting time may be excessive and so leaves
immediately to seek service elsewhere.
Operations Research Unit 9
Sikkim Manipal University Page No. 183
d) Jockeying: Some customers keep on switching over from one queue to
another in a multiple service centres. This is called jockeying.
9.6.2 Server behaviour
Although the service timings have been specified, the server may not be
available through the entire span of time. For instance, in every one hour the
server may disconnect from the service centre for 5 minutes so that it can
go through its daily updates or cleanup routine. The server is characterised
by:
- The arrangement of service facilities
- The distribution of service times
- Server‘s behaviour
- Management policies
Self Assessment Questions
11. When customers keep on switching over from one queue to another
then it is called ________.
12. _____ ________ ______ ______ are the types of customer behaviour.
9.7 Summary
Let us recapitulate the important concepts discussed in this unit:
- The waiting line theory or queuing theory analysis suggests the number
of facilities required and the cost of customer‘s waiting time and the
optimum service level.
- The queuing theory contributes vital information required for balancing
the cost of service and cost associated with waiting time of the
customer.
- A queuing system is said to be in transient state when its operating
characteristics are dependent upon time.
9.8 Glossary
Erlang family of service time distributions: two constraint generalization
of exponential distribution.
Statistical pattern: set of probability distributions
Operations Research Unit 9
Sikkim Manipal University Page No. 184
Probability distribution: a table containing the outcomes of an experiment
with its likelihood of occurrence.
9.9 Terminal Questions
1. Explain the operating characteristics of a queuing system.
2. Which are the important arrival characteristics?
3. State the ways in which customers in a queue are served.
9.10 Answers
Self Assessment Questions
1. False
2. True
3. False
4. Queue length
5. Long waiting
6. Collective utilisation
7. True
8. True
9. Series, parallel
10. Constant, varying
11. Jockeying
12. Balking, Collusion, reneging, jockeying
Terminal Questions
1. The operating characteristics of a queuing system are queuing length,
number of customers in system,waiting time in queue, waiting time in
system and service facility utilisation. For more details, refer section 9.3
2. Size of population, pattern of arrivals and behavior of arrivals. For more
details refer section 9.4.1
3. Static queuing disciplines and dynamic queuing disciplines. For more
details refer section 9.5
Operations Research Unit 9
Sikkim Manipal University Page No. 185
9.11 Case Study
Drive-in Photo Shop
Sunder is planning to open a drive-through photo developing booth with a
single service window that will be open approximately 200 hours per month
in a busy commercial area. Space for a drive-through lane is available for a
rental of Rs. 8000 per month per car length. Sunder needs to decide how
many car lengths of space to provide for his customers.
Excluding this rental cost for the drive-through lane, Sunder believes that he
will average a profit of Rs 160 per customer served(nothing for a drop off of
film and Rs 320 when the photographs are picked up). He also estimates
that customers will arrive randomly at a mean rate of 20 per hour, although
those who find the drive-through lane full will be forced to leave. Half of the
customers who find the drive-through lane full wanted to drop off the film
and the other half wanted to pick up the photographs. The half who wanted
to drop off film will take their business elsewhere instead. The other half of
the customers who find the drive-through lane full will not be lost because
they will keep trying later until they can get in and pick up their photographs.
Sunder assumes that the time required to serve a customer will have an
exponential distribution with a mean of 2 minutes.
Discussion Questions:
a) Find L(expected no. of customers in queuing system) and the mean rate
at which customers are lost when the number of car lengths of space
provided is 2, 3, 4 and 5.
b) Calculate W (waiting time in the system) and L for the cases considered
in (a).
c) Use the results from part(a) to calculate the decrease in the mean rate
at which customers are lost when the number of car lengths of space
provided is increased from 2 to 3, from 3 to 4, from 4 to 5. Then
calculate the increase in expected profit per hour(excluding space rental
costs) for each of these three cases.
d) Compare the increases in expected profit found in part(c) with the cost
per hour of renting each car length of space. What conclusion do you
draw about the number of car lengths of space that Sunder should
provide?
Operations Research Unit 9
Sikkim Manipal University Page No. 186
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
E-References:
- http://www.mrt.ac.lk/maths/TMJAC/Operational%20Research
%20Techinques%20II/OR%20Lecture%203.pdf, accessed on Nov 9,
2009
Operations Research Unit 10
Sikkim Manipal University Page No. 187
Unit 10 Mathematical Analysis of Queuing Theory
Structure:
10.1 Introduction
Objectives
10.2 Mathematical Analysis of Queuing Process
10.3 Properties of Queuing System
Arrivals
Service
Queuing discipline
10.4 Notations
10.5 Service System
10.6 Single Channel Models
10.7 Multiple Service Channels
10.8 Erlang Family of Distribution of Service Times
10.9 Applications of Queuing Theory
10.10 Limitations of Queuing Theory
10.11 Summary
10.12 Glossary
10.13 Terminal Questions
10.14 Answers
10.15 Case Study
10.1 Introduction
In the previous unit, you learnt queuing theory, operating characteristics and
constituents of a queuing system. You also learnt about service facility and
queue discipline. Queuing theory is a collection of mathematical models of
various queuing systems that takes in input parameters and provides
quantitative parameters describing the system performance.
Because of the random nature of the processes involved, the queuing
theory is rather demanding and all models are based on very strong
assumptions (not always satisfied in practice). Many systems (especially
queuing networks) are not soluble at all, so the only technique that may be
applied is simulation.
Operations Research Unit 10
Sikkim Manipal University Page No. 188
Nevertheless queuing systems are practically very important because of the
typical trade-off between various costs of providing service and the costs
associated with waiting for the service (or leaving the system without being
served). High quality fast service is expensive, but costs incurred by
customers waiting in the queue are minimum. On the other hand long
queues may cost a lot because customers (e.g., machines) do not work
while waiting in the queue or customers leave because of long queues. So a
typical problem is to find an optimum system configuration (e.g., the
optimum number of servers). The solution may be found by applying
queuing theory or by simulation. In this unit, you will learn about
mathematical analysis of queuing theory, notations and service systems
involved in queuing theory. You will also learn about Erlang family of
distribution of service times and the applications and limitations of queuing
theory.
So, let us start with the mathematical analysis of queuing process.
Objectives:
After studying this unit, you should be able to:
- describe the mathematical analysis of queuing theory
- classify the different notations involved in queuing theory
- recognise different service systems involved
- explain Erlang family of distribution of service times
10.2 Mathematical Analysis of Queuing Process
Statistical equilibrium
While analysing the queuing process, our main interest is in developing a
mathematical model which represents the system. This implies that changes
occurring in the characteristics of the system are not to be considered if they
are of short durations. When we specify the statistical distributions of the
arrivals or service times, we want to ensure an equilibrium state.
In a queuing process, at each point of time, there is a probability distribution
for the length of the queue. Compare the number of customers arrived at the
post office, 15 minutes after the counter is opened to that after one hour.
After the initial rush, you‘ll find the system with the same type of probabilities
of arrivals.
Operations Research Unit 10
Sikkim Manipal University Page No. 189
The probability distribution of the arrivals will then be different from that of
the initial state. The state, in which the probability distribution remains the
same, is called the steady state and the system is said to have acquired a
state of statistical equilibrium.
In the steady state, there will be variations in the queue from time to time but
the probability distributions representing the queuing process will remain the
same and are independent of the time at which the system is examined.
10.3 Properties of Queuing System
In the previous section, you learnt the mathematical analysis of queuing
process. You will now learn the properties of the system.
While developing queuing models, we will confine our discussion to queuing
systems with the following properties.
10.3.1 Arrivals
- Customers are discrete entities.
- Population – Finite/Infinite.
- No simultaneous arrivals.
- Pattern of arrivals in a time period t0 follows Poisson distribution with
average arrival rate λ.
Probability (Number of arrivals = K) = λt(λt)x/ x !
Where x = 0, 1, 2 ……..
10.3.2 Service
- Single serve channel/multiple channels
- Single queue/infinite capacity
10.3.3 Queuing discipline
- First come first serve
Note: When the number of arrivals follows Poisson distribution (discrete),
the inter-arrival time, that is, the time between arrivals follows exponential
distribution (continuous).
Operations Research Unit 10
Sikkim Manipal University Page No. 190
10.4 Notations
In the previous section, you learnt the properties of the system. You will now
learn about the notations. The queuing systems with which you are
concerned are denoted by M/M/1 and M/M/c where M stands for exponential
inter arrival and exponential service time distributions and the third character
indicates the number of channels (or servers) available (1 or c).
Table 10.1 depicts a list of notations that are commonly used.
Table 10.1: List of Notations Used
Notation
Z
H
p
c
m
n
Pn
E(m)
E(m/m > 0)
E(n)
W
E(w)
P(w = 0)
E(w/w > 0)
V
F(v)
E(v)
Meaning
= Average number of arrivals per time
= Average number of customers served per unit time
= Traffic intensity
= Number of service channels
= Number of customers in the queue or the length of the queue
= Number of customers in the System (Number of customers in the
queue + Number of customers being served )
= Steady state probability of finding n people in the system
= Average length of queue
= Average length of a non-empty queue
= Average number of customers in the system
= Steady state waiting time of a customer
= Probability of not having to waiting time of a customer
= Average waiting time of a customer given that there is a queue
= Time spent by a customer in the system in steady state
= Waiting time in the queue + Service time
= Probability density function of the time spent by a customer in the
system
= Average time spent by a customer in the system
Self Assessment Questions
1. E (m) refers to ________ _________ ________ _________.
2. Probability density function of the time spent by a customer in the
system is denoted by _____________.
3. ___________ arrivals are allowed.
Operations Research Unit 10
Sikkim Manipal University Page No. 191
10.5 Service System
In the previous section, you learnt about notations. You will now learn about
service system. The service is provided by a service facility (or facilities).
This may be a person (a bank teller, a barber, a machine (elevator, gasoline
pump), or a space (airport runway, parking lot, hospital bed) to mention a
few. A service facility may include one person or several people operating
as a team.
There are two aspects of a service system
(a) the configuration of the service system and
(b) the speed of the service.
a) Configuration of the service system
The customers‘ entry into the service system depends upon the queue
conditions. If at the time of customers‘ arrival, the server is idle, then the
customer is served immediately. Otherwise the customer is asked to join the
queue, which can have several configurations. By configuration of the
service system we mean how the service facilities exist. Service systems
are usually classified in terms of their number of channels, or numbers of
servers.
i. Single server – Single queue – The models that involve one queue –
one service station facility are called single server models where customer
waits till the service point is ready to take him for servicing. Students‘
arriving at a library counter is an example of a single server facility.
Figure 10.1 depicts the single server – single queue model.
Fig. 10.1: Single Server – Single Queue Model
ii. Single server – Several queues – In this type of facility there are several
queues and the customer may join any one of these but there is only one
service channel. Figure 10.2 depicts the single server – several queues
model.
Operations Research Unit 10
Sikkim Manipal University Page No. 192
Queues
Service facility Customer leave
Arrivals
Fig. 10.2: Single Server – Several Queues Model
iii. Several (parallel) servers – Single queue – In this type of model there
is more than one server and each server provides the same type of facility.
The customers wait in a single queue until one of the service channels is
ready to take them in for servicing. Figure 10.3 depicts the several (parallel)
servers – single queue model.
Service stations
Fig. 10.3: Several (Parallel) Servers – Single Queue Model
iv) Several servers – Several queues – This type of model consists of
several servers where each of the servers has a different queue. Different
cash counters in an electricity office where the customers can make
payment in respect of their electricity bills provide an example of this type of
model. Figure 10.4 depicts the several (parallel) servers – several queues
model.
Operations Research Unit 10
Sikkim Manipal University Page No. 193
Fig. 10.4: Several (Parallel) Servers – Several Queues Model
v) Service facilities in a series –
In this, a customer enters the first station and gets a portion of service and
then moves on to the next station, gets some service and then again moves
on to the next station and finally leaves the system having received the
complete service. For example, machining of a certain steel item may
consist of cutting, turning, knurling, drilling, grinding and packaging
operations, each of which is performed by a single server in a series.
Figure 10.5 depicts multiple servers in a series.
Fig. 10.5: Multiple Servers in a Series
b) Speed of the service
In a queuing system, the speed with which the service is provided can be
expressed in either of the two ways – as service rate or as service time.
- The service rate describes the number of customers serviced during a
particular time period.
- The service time indicates the amount of time needed to service a single
customer.
- Service rates and times are reciprocal of each other and either of them
is sufficient to indicate the capacity of the facility.
Operations Research Unit 10
Sikkim Manipal University Page No. 194
Thus if a cashier can attend, on an average 5 customers in an hour, the
service rate would be expressed as 5 customers/hour and service time
would be equal to 12 minutes/customer.
Generally, we consider the service time only. If these service times are
known exactly, the problem can be easily handled. If these are different and
not known with certainty, we have to consider the distribution of the service
times in order to analyse the queuing system. Generally, the queuing
models are based on the assumption that service times are exponentially
distributed about some average service time.
10.6 Single Channel Models
In the previous section, you learnt about service system. You will now learn
about single channel models. The formulae are depicted in figures 10.6 and
10.7. The examples illustrate the computation of various measures of
efficiency of a queuing system, but give an idea of the areas of application
as well.
( ) ( ) ( )
( ) (1 ) ( )
1
( )
( ) 1
1
( / 0)
( )
( )
( )
( / 0)
( )
( )
1
(1 )
2
µ ì µ ì
ì µ µ ì
µ ì
µ
µ ì
µ µ ì
ì
µ ì
ì
µ ì
µ
µ µ ì
ì
µ
µ
= ÷ ÷ ÷
= ÷ ÷ ÷
÷
=
= = ÷
÷
> =
÷
=
÷
=
÷
> =
÷
=
= ÷
= ÷
f v e v
f w e w
E v
P w o
E w w
E w
E n
E m m
E m
P
P pn
n
n
Fig. 10.6: Formulae for Poisson Arrivals, Exponential Service, Single Channel
Queuing Models – Infinite Population
Operations Research Unit 10
Sikkim Manipal University Page No. 195
Fig. 10.7: Formulae for Poisson Arrivals, Exponential Service, Single Channel
Queuing Models – Number of Customers Linked to N
Solved problem 1:
Patrons arrive at a small post office at the rate of 30 per hour. Service by the
clerk on duty takes an average of 1 minute per customer.
a) Calculate the mean customer time:
i) Spent waiting in line
ii) Spent receiving or waiting for service
b) Find the mean number of persons:
i) In line
ii) Receiving or waiting for service
Solution:
Mean arrival rate λ = 30 customers per hour
= customer per minute
Mean service rate μ = 1 per minute
Traffic intensity
(a) (i) Mean customer time spent waiting in line
Operations Research Unit 10
Sikkim Manipal University Page No. 196
(ii) Mean customer time receiving or waiting for service
= 2 minutes
(b) (i) Mean number of persons in line
= 0.5 customer
(ii) Mean number of persons receiving or waiting for service
Customer
Solved problem 2:
The Tool Company‘s quality control department is manned by a single clerk,
who takes an average of 5 minutes in checking parts of each of the
machines coming for inspection. The machines arrive once in every 8
minutes on an average. One hour of machine is valued at Rs. 15 and a
clerk‘s time is valued at Rs. 4 per hour. What is the average hourly queuing
system costs associated with the quality control department?
Solution:
Mean arrival rate l = 1/8 per minute = per hour
Mean service rate per minute = 12 per hour
Average time spent by a machine in the system E (v) = = hours
Average queuing cost per machine =
An average arrival of machines per hour costs
= Rs. 25 per hour
Average hourly queuing cost = Rs. 25
Average hourly cost for the clerk = Rs. 4
Hence total hourly cost for the department = Rs. 29
Operations Research Unit 10
Sikkim Manipal University Page No. 197
Solved problem 3:
A television repairman finds that the time spent on his jobs has an
exponential distribution with a mean of 30 minutes. If he repairs sets in the
order in which they came in and if the arrival of sets follow a Poisson
distribution with an average rate of 10 per 8 hour day, what is the
repairman‘s expected idle time each day? How many jobs are ahead of the
average set just brought in?
Solution:
λ = 10/8 = 5/4 sets/hr; μ = (1/30) 60 = 2 sets/hr.
Number of hours for which the repairman remains busy in an 8-hour day =
8* (λ / μ) = 8(5/8) = 5 hrs.
Repairman‘s expected idle time in an 8 hour day = 8-5 = 3 hrs.
Expected number of T.V sets in the system = (Ls) = λ / (μ - λ)
= (5/4)/ [2-(5/4)]
= 5/3
= 2 sets (approx.)
Solved problem 4:
Arrivals at a telephone booth are considered to follow Poisson distribution
with an average time of 10 minutes between one arrival and the next. The
length of phone call is assumed to be distributed exponentially, with a mean
of 3 minutes.
a) What is the probability that a person arriving at the booth will have to
wait?
b) The telephone department will install a second booth when convinced
that an arrival would expect to wait for at least 3 minutes for a phone
call. By how much should the flow of arrivals increase in order to justify
a second booth?
c) What is the average length of the queue that forms from time to time?
d) What is the probability that it will take him/her more than 10 minutes
altogether to wait for the phone and complete his call.
Operations Research Unit 10
Sikkim Manipal University Page No. 198
Solution:
λ = 1/10 person/minute; μ = 1/3 = 0.33 person/minute.
a) Probability that a person arriving at the booth will have to
wait = 1- P0 = 0.3
b) The installation of the 2nd booth will be justified only if the arrival rate is
more than the waiting time.
Let λ‘ be the increased arrival rate. Then expected waiting time in the
queue will be (Wq) = λ‘ / μ (μ - λ‘)
= λ‘ / 0.33(0.33- λ‘)
Or λ‘ =0.16
Hence the increase in the arrival rate is 0.16-0.10 = 0.06 arrivals
/minute
c) Average length of non-empty queue = μ / (μ - λ). = 0.33/0.23 = 1.42857
Customers (approx.)
d) Probability of waiting for 10 minutes or more = P (t>=10)

= ∫ (λ / μ) (μ - λ) e -(μ - λ)tdt

= ∫(0.3)(0.23)e-0.23tdt = 0.069 e -0.23t = 0.03
10 -0.23
This shows that 3 percent of the arrivals on an average will have to wait for
10 minutes or more before they can use the phone.
Solved problem 5:
A hospital emergency room can accommodate at most M = 5 patients. The
Patients arrive at a rate of 4 per hour. The single staff physician can only
treat 5 patients per hour. Any patient overflow is directed to another hospital.
a) Determine the probability distribution for the number of patients either
waiting for or receiving treatment at any given time
b) Determine the mean values for the number of patients in the emergency
room. And the number of patients waiting to see the doctor.
Solution:
M = Maximum number of customers in the system = 5
λ = 4 per hour and = 5 per hour
= 8
Operations Research Unit 10
Sikkim Manipal University Page No. 199
The probability distribution for the number of patients in the system
Pr =
n
| |
.
|

\
|
µ
ì
n P0 for 0 M
The probability distribution for the number of patients either waiting for
treatment or receiving treatment is as shown in table 10.2.
Table 10.2: Probability Distribution
N Pn
0 0.271
1 0.217
2 0.173
3 0.139
4 0.111
5 0.089
(b) (1) The mean value for the number of patients in the emergency room
= 0 x (0.271) + 1 x (0.217) + 2 x (0.173) + 3 x (0.139) + 4 x (0.111) + 5
x (0.089)
= 1.868332032.
Solved problem 6:
Cars arrive at a toll gate on a frequency according to Poisson distribution
with mean 90 per hour. Average time for passing through the gate is 38
seconds.
Drivers complain of long waiting time. Authorities are willing to decrease the
passing time through the gate to 30 seconds by introducing new automatic
devices. This can be justified, if under the old system, the number of waiting
cars exceeds 5 and the percentage of gate‘s idle time under the new system
should not exceed 10.
In the above scenario, can the new device be justified?
Operations Research Unit 10
Sikkim Manipal University Page No. 200
Solution:
In the old system,
Meaning arrival rate = 90 per hour = 1.5 per minute
Mean service rate = per minute
Traffic intensity = =
Expected number of the waiting cars
=
Hence the new system is justified on the basis of the expected number of
waiting cars.
Under the new system, the probability of device being idle
= P 0 = 1 –
= 1 - .75
= .25 which is greater than 10% required for the new device to be
introduced.
As only one of the stipulated conditions is fulfilled, the new automatic device
is not justified.
Solved problem 7:
Customers arrive at a one – window drive-in bank according to a Poisson
distribution with mean 10 per hour. Service time per customer is exponential
with mean 5 minutes. The space in front of the window, including that for the
serviced car can accommodate a maximum of 3 cars. Other cars can wait
outside this space.
a) What is the probability that an arriving customer can drive directly to
the space in front of the window?
b) What is the probability that an arriving customer will have to wait
outside the indicated space?
c) How long is an arriving customer expected to wait before starting
service?
Operations Research Unit 10
Sikkim Manipal University Page No. 201
d) How many spaces should be provided in front of the window so that all
the arriving customers can wait in front of the window at least 20% of
the time?
Solution:
(a) λ = 10 per hour; 10/12 = 5/6 per hour
(b) p = λ/μ = 10/12 = 5/6
The probability that an arriving customer can drive directly to the
space in front of the window
= P0+P1+P2
= (1-p)(1 + p +p2 )
= required probability = .42
(c) Probability that an arriving customer has to wait outside the indicated
space = probability that there are at least 3 customers in the space in
front of the window
= 0 + p1 + p2 + p3)
=1 – (1- p) (1 + p + p2 + p3)
= .48
(d) The average waiting time of a customer in queue.
(e) The existing space in front of the window can accommodate up to
three cars. Probability that a customer can wait in front of the window
must be at least 0.20. This is possible if the number of customers in
the system is 0, 1 or 2 for at least 20% of the time.
i.e., 0 + p1 + p2 ≥0.20
It is seen from (a) that 0 + p1 + p2 = 0.42
As even with one space in front of the window, 42 percent of the times an
arrival waits in the space, the number of spaces required is one or more.
Self Assessment Questions
4. The expected number of customers in non-empty queue is given by
___________.
Operations Research Unit 10
Sikkim Manipal University Page No. 202
5. The probability that an arriving customer has to wait for receiving
service is given by ___________.
10.7 Multiple Service Channels
In the previous section, you learnt about single channel models. You will
now learn about multiple channel models. The analysis of systems involving
several service channels is more complex. The major practical utility of
these models are the ways of improving to provide additional service
facilities. The list of formulae to be used is depicted in figure 10.8, followed
by few examples.
Fig. 10.8: Formulae for Poisson Arrivals, Exponential Service Multi channel
Queuing Models – Infinite Population
Solved problem 8:
Ships arrive at a port at a rate of one in every 3 hours, with a negative
exponential distribution of inter–arrival times. The time a ship occupies a
Operations Research Unit 10
Sikkim Manipal University Page No. 203
berth for unloading and loading has a negative exponential distribution with
an average of 12 hours. If the average delay of ships waiting for berths is to
be kept below 6 hours, how many berths should there be present at the
port?
Solution:
For multichannel queues, 1/ ensure that the queue does not
explode.
Let us calculate the waiting time when c = 5
Substituting for n, c and p
P 0 =
Average waiting time of a ship
E (w) = 0
rp
Pe 1/(c 1)!/(c p)21 ÷ ÷ ÷
= 6.65
This is greater than 6 hours and inadequate.
When P0
Moreover, average waiting time of a ship E(w) = 1.71 hours.
Hence 6 berths should be provided at the port.
Solved problem 9:
A bank has two counters for withdrawals. Counter 1 handles withdrawals of
value less than Rs. 300 and counter 2 handles withdrawals of Rs. 300 and
above. Analysis of service time shows a negative exponential distribution
with mean service time of 6 minutes per customer for both the counters.
Operations Research Unit 10
Sikkim Manipal University Page No. 204
Arrival of customers follows Poisson distribution with mean 8 per hour for
counter 1 and 5 per hour for counter 2.
a) What are the average waiting times per customer of each counter?
b) If each counter could handle all withdrawals irrespective of their value
how would the average waiting time change?
Solution:
(a) For counter one 1 = 10 per hour, λ1 = 8 per hour.
For counter two, = 10 per hour, λ2 = 5 per hour
(b) When each counter can handle all withdrawals irrespective of their
value with an average arrival rate of 13 per hour, mean service time
of 10 per hour at each counter and number of channels = 2,
= 4. 71P0 = 0.21
Average waiting time E(w)
= P0 = 4.38961039 minutes
Self Assessment Questions
6. Write the formula for calculating the expected number of customers in
the system.
7. Write the formula for calculating the average waiting time of the
customer in the queue.
Operations Research Unit 10
Sikkim Manipal University Page No. 205
10.8 Erlang Family of Distribution of Service Times
In the previous section, you learnt about multiple channel models. You will
now learn the Erlang family of distribution of service times. The queuing
processes discussed till now are mathematically simple. Assuming that the
service time follows negative exponential distribution, you are also taking
the standard deviation equal to its mean. There would be situations, where
the mean and the standard deviation substantially differ. At these instances
the models have to be made more general by using a distribution system,
which coincide closely to the practical problems, but yet retains the
simplicity of the properties of negative exponential distribution. A.K. Erlang
first studied such a distribution system.
Consider the distribution of a service time involving a fixed number of
phases, k, and each phase has a negative exponential distribution. If there
are k phases and the average time taken by a customer through each phase
is
(1/ kμ).units, the service time distribution f (t) is given by
The mean of this distribution is (1 / μ) and standard deviation is (1/ μ√k).
If k = 1, f (f) = me–m1 which is the negative exponential distribution; same as
the models considered earlier.
For k = 2, f (f) = 4m2te–2mf, etc.
The mode is located at t = (k-1 / μk).
If k = ¥, the variance is zero and this corresponds to a case where the
service time is constant and has the value (1/μ).
Figure 10.9 depicts the way the density functions vary as ‗k‘ increases.
Operations Research Unit 10
Sikkim Manipal University Page No. 206
Fig. 10.9: Variation in Density Functions as ‘k’ Increases
The measures of efficiency should take into account the number of
customers getting service, number having entered during any one or more
of the phases and the number yet to join. For arrivals following Poisson
distribution, with mean l and service time following the kth Erlang distribution
with mean, (1/μ) the formulae applicable are given below.
Average queue length
Average number of units in the system
Average waiting time
Average time spent in the system
For constant service time, limiting k to ∞
Operations Research Unit 10
Sikkim Manipal University Page No. 207
Solved problem 10:
At the bus depot‘s cafeteria, the customer has to pass through three
counters. At the first counter, customer buys coupons, at the second they
collect coffee or tea. The server at each counter takes an average two
minutes although the distribution of the time of service is approximately
exponential. If the arrival of customers to the cafeteria is approximately
Poisson at an average rate of six per hour, what is the average time spent
by customer waiting in the cafeteria? What is the average time of getting the
service? What is the most probable time spent in getting the service?
Solution:
This is a queuing process with service time following Erlang distribution.
No. of phases k = 3
persons per minute
= 6 persons per hour
= 1/10 person per minute
(1) Average waiting time = 6 minutes
(2) Average time spent in collecting coupons, snacks, etc.
(3) Most probable time spent in collecting coupons, snacks. etc/
Self Assessment Questions
8. Write the formula for E (m)
9. Write the formula for E (w)
Operations Research Unit 10
Sikkim Manipal University Page No. 208
10.9 Applications of Queuing Theory
The applications of queuing theory are:
1. Traffic control
2. Determining the sequence of computer operations
3. Predicting computer performance
4. Health services (e.g., control of hospital bed assignments)
5. Airport traffic, airline ticket sales
6. Layout of manufacturing systems.
7. Telecommunications
10.10 Limitations of Queuing Theory
In the previous section, you learnt the applications of queuing theory. You
will now learn the limitations of queuing theory. Classic queuing is too
mathematically restrictive to be able to model all real world situations. This
restriction arises because the underlying assumptions of the theory do not
always hold well in the real world. Alternative means of analysis have been
devised in order to provide some insight into problems which do not fall
under the scope of queuing theory, though they are often scenario-specific
since they generally consist of computer simulations and/or of analysis of
experimental data.
10.11 Summary
Let us recapitulate the important concepts discussed in this unit:
This unit discusses about the mathematical analysis of Queuing process &
how the service system is useful in different process.
10.12 Glossary
Queuing discipline: It refers to the decision rule by which the customers
are selected from queue for service
10.13 Terminal Questions
1. Discuss about Erlang family of distribution of service times.
2. A stenographer has 5 persons for whom she performs stenographic
work. Arrival rate follows Poisson distribution and service times are
Operations Research Unit 10
Sikkim Manipal University Page No. 209
exponential. Average arrival rate is 4 per hour with an average service
time of 10 minutes. Cost of waiting is Rs. 8 per hour while the cost of
servicing is Rs. 2.50 each. Calculate:
i) The average waiting time of an arrival
ii) The average length of the waiting line
iii) The average time which an arrival spends in the system
3. A ticket issuing office is being manned by a single server. Customers
arrive to purchase tickets according to Poisson distribution with a mean
rate of 30 per hour. The time required to serve a customer has an
exponential distribution with a mean of 90 seconds. Find the value of pn,
Ls and Ws. Denote the expected line length and waiting time in the
system respectively.
(Hint: =54/6 * 60, ų= 60/90 = 2/3. Use formulae for Pn, Ls, Ws)
4. Discuss different types of service systems?
10.14 Answers
Self Assessment Questions
1. Average length of queue
2. F (v)
3. Simultaneous
4. m / m – l
5. λ / μ
6. E(m/m > 0) = μ/(μ – λ)
7. P(w = 0) = 1-P
8.
2k ( )
k 1 2
µ µ÷ì
+ ì
9.
2k ( )
k 1
µ µ÷ì
+ ì
Terminal Questions
1. Refer 10.8
2. i) 12.4 min ii) 0.79= one stenograph iii) 22.4 min.
3. Pn=1/4(1/4)n, Ls = 3 customers , Ws= 6 min.
4. Refer 10.5
Operations Research Unit 10
Sikkim Manipal University Page No. 210
10.15 Case Study
Queuing Solution
Never dull. That is how you would describe your job at the centralised
records and benefits administration centre for Always First, a large company
manufacturing computers and computer peripherals. The company provides
a call centre at the administration centre for the 60,000 Always First
employees throughout the United States. Employees contact the call centre
to obtain information about dental plans and pension options, to change tax
forms and personal information and to process leaves of absence and
retirements. The centralised records and benefits administration centre and
its call centre opened just six months ago to replace 35 small regional
administration centres across the country. Since this recent opening, you
and John Ducker, the director of human resources, have endured one long
roller coaster ride. Receiving the go-ahead from corporate headquarters to
establish the centralised records and benefits administration centre was
definitely an up. Getting caught in the crossfire of angry customers (all
employees of Always First) because of demand overload for the records and
benefits call centre was definitely a down. Accurately forecasting the
demand for the call centre provided another up. And today you are faced
with another down. John approaches your desk with a not altogether
attractive frown on his face. He begins complaining immediately, ―I just don‘t
understand. The forecasting job you did for us two months ago really
allowed us to understand the weekly demand for the centre, but we still
have not been able to get a grasp on the staffing problem. We used both
historical data and your forecasts to calculate the average weekly demand
for the call centre. We transformed this average weekly demand into
average hourly demand by dividing the weekly demand by the number of
hours in the working week. We then staffed the centre to meet this average
hourly demand by taking into account the average number of calls a
representative is able to handle per hour. But something is horribly wrong.
Operational data records show that over thirty percent of the customers wait
over four minutes for a representative to answer the call! Customers are still
sending me numerous complaints and executives from corporate
headquarters are still breathing down my neck! I need help!‖ You calm John
down and explain to him that you think you know the problem: the number of
calls received in a certain hour can be much greater (or much less) than the
Operations Research Unit 10
Sikkim Manipal University Page No. 211
average because of the stochastic nature of the demand. In addition, the
number of calls a representative is able to handle per hour can be much
less (or much greater) than the average depending upon the types of calls
received. You then tell him to have no fear; you have the problem under
control. You have been reading about the successful application of queuing
theory to the operation of call centres, and you decide that the queuing
models you learned in school will help you determine the appropriate
staffing level.
Discussion Questions:
(a) You ask John to describe the demand and service rate. He tells you that
calls are randomly received by the call centre and that the centre
receives an average of 70 calls per hour. The computer system installed
to answer and hold the calls is so advanced that its capacity far exceeds
the demand. Because the nature of a call is random, the time required to
process a call is random, where the time frequently is small but
occasionally can be much longer. On an average, however,
representatives can handle 6 calls per hour. Which queuing model
seems appropriate for this situation? Given that slightly more than 35
percent of customers wait over 4 minutes before a representative
answers the call, use this model to estimate how many representatives
John currently employs.
(b) John tells you that he will not be satisfied unless 95 percent of the
customers wait only 1 minute or less for a representative to answer the
call. Given this customer service level and the average arrival rates and
service rates from part (a), how many representatives should John
employ?
(c) Each representative receives an annual salary of US $30,000 and John
tells you that he simply does not have the resources available to hire the
number of representatives required to achieve the customer service
level desired in part (b). He asks you to perform sensitivity analysis. How
many representatives would he need to employ to ensure that 80
percent of customers wait 1 minute or less? How many would he need to
employ to ensure that 95 percent of customers wait 90 seconds or less?
How would you recommend John choose a customer service level?
Would the decision criteria be different if John‘s call centre were to serve
Operations Research Unit 10
Sikkim Manipal University Page No. 212
external customers (not connected to the company) instead of internal
customers (employees)?
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
Operations Research Unit 11
SikkimManipalUniversity Page No. 213
Unit 11 Finite Queuing Models
Structure:
11.1 Introduction
Objectives
11.2 Finite Queuing Models
Finite queuing tables
Measures of system efficiency
Use of finite queuing tables
11.3 Summary
11.4 Glossary
11.5 Terminal Questions
11.6 Answers
11.7 Case Study
11.1 Introduction
In the previous unit, you learnt about mathematical analysis of queuing
theory, notations and service systems involved in queuing theory. You also
learnt about Erlang family of distribution of service times and the
applications and limitations of queuing theory. The models discussed so far
relate to situations involving infinite population of customers, that is, the
queue can increase indefinitely. However, you may come across scenarios
where the possible number of arrivals is limited and small. In this unit, we
will discuss such scenarios.
In a production shop, if the machines are considered as customers requiring
service from repair crews or operators, the population is restricted to the
total number of machines in the shop. In a hospital ward, the probability of
the doctors or nurses being called for service is governed by the number of
beds in the ward. Similarly, in an aircraft the number of seats is finite and
the number of stewardesses provided by the airlines will be based on the
consideration of the maximum number of passengers who can demand
service.
Operations Research Unit 11
SikkimManipalUniversity Page No. 214
In case of a queuing system with infinite population, you can improve the
efficiency of the system in terms of reduced average length of queues,
average waiting time and time spent by the customer in the system. To do
this, you need to increase the number of service channels. However, such
increases mean additional cost and will have to be balanced with the
benefits likely to accrue. If the queuing system in a machine shop is under
study, the cost of providing additional maintenance crews or operators can
be compared with the value of additional production due to reduced
downtime of the machines. Sometimes it is not possible to quantify the
benefits–the management will have to base its decisions on the desired
standards for customer service. In this unit, you will learn the finite queuing
models, finite queuing tables and the use of these models and tables.
Objectives:
After studying this unit, you should be able to:
- describe queuing models and various queuing disciplines
- interpret and apply finite queuing tables to solve problems
11.2 Finite Queuing Models
Basic concept of queuing models
You can use a queuing model to approximate a real queuing situation or
system, so that you can analyse the queuing behaviour mathematically.
Queuing models allow you to determine a number of useful steady state
performance measures. These measures are:
- Average number in the queue, or the system
- Average time spent in the queue, or the system
- Statistical distribution of those numbers or times
- Probability of the queue being empty or full
- Probability of finding the system in a particular state
Analysis of the relevant queuing models allows you to identify the cause of a
queuing issue and assess the impact of the proposed changes. Typically,
you need to analyse a queuing model to minimise the sum of:
- Customer waiting cost
- Service capacity cost
Operations Research Unit 11
SikkimManipalUniversity Page No. 215
In a finite queuing model, the system processes incoming requests or
messages at a certain rate. Any newly arrived request or message is added
to the queue only if the number of requests or messages waiting in the
queue is less than the maximum.
The model is based on the following assumptions:
- A queue is characterised by the maximum permissible number of
customers that it can contain. This also could be finite or infinite. In most
of the practical situations, it is finite.
- Queuing discipline determines the manner in which the system handles
calls from the customers. It defines the way in which they are served, the
order in which they are served and the way in which resources are
divided between the customers. A finite queuing process may follow any
of the following queuing disciplines:
Figure 11.1 depicts the queue discipline route in a finite queuing process.
Fig. 11.1: Queue Discipline Route in a Finite Queuing Process
First Come - First Served (FCFS):
This principle states that customers are served one at a time and that the
customer that has been waiting the longest is served first. In most queuing
models, the order is usually first come-first served.
Priority:
Priority-discipline models give priority to rush jobs and important customers
over others. For instance, machines of high cost may be given priority for
maintenance while others may be kept waiting even if they had broken down
earlier.
Operations Research Unit 11
SikkimManipalUniversity Page No. 216
Random:
This principle states that the system may randomly select a customer for
service. For example, in a machine shop if a single operator is attending to
several machines and several machines call for his attention at a time, he
may attend first to the one nearest to him.
Last Come-First Served (LCFS):
This discipline is based on the stack method.
Most quantitative parameters, such as the average queue length and
average time spent in the systemdo not depend on priority queues. That is
the reason most models either do not take the queuing discipline into
account at all or assume the normal FCFS queue. The only parameter that
depends on the queuing discipline is the variance (or standard deviation) of
the waiting time.
The analysis of finite queuing models is more complex than those with
infinite population although the approach is similar.
Simple waiting line model
The simplest waiting line model assumes that arrivals join a queue that is of
unlimited size, waiting in line until their turn for service comes on a first-cumfirst-
serve basis and then enter a service facility consisting of a single
channel.
If, W = Average time spent in queue (i.e., sum of the expected waiting time
and expected service time) = Average rate of units passing through the
system per unit λ tim e L = Average number of units in the system.
Then, L= λ W
If μ = Average service rate
1/λ = Expected inter arrival time
1/μ = Expected service time
Further, the utilisation factor, ρ (i.e., the expected fraction of time the server
is busy) is given by:
ρ = λ / sμ
Where, (sμ) is the fraction of the system's service capacity that being
utilised on the average by arriving customers (Cλ)
Operations Research Unit 11
SikkimManipalUniversity Page No. 217
The crux of the queuing theory is to 'achieve a trade-off between excessive
waiting by customers (i.e., too much demand) and cost due to excessive idle
time at the service facility (i.e., too little demand).‘ Lending system in a
library is an example of single facility channel waiting line.
11.2.1 Finite queuing tables
Figure 11.2 depicts the different notations used in finite queuing tables.
Consider a machine shop with N machines. The inter breakdown time of
these machines follows a negative exponential distribution with mean U.
The number of breakdowns follows Poisson distribution with mean
U
1
ì =
N = Population (machines, customer etc.)
M = Service channels (repairmen, telephone linesetc.)
T = Average service time (repair time, length of conversation on a telephone etc.)
W = Average waiting time
U = Average running time (of machines) or meantime between calls for service per
unit
H = Average number of units being serviced
L = Average number of units waiting for services
J = Average number of units in operation
F = Efficiency Factor
X = Service Factor
D = Probability that if a unit calls for service, it will have to wait.
Fig. 11.2: Notations Used in Finite Queuing Tables
It is assumed that machines are kept running (or in operation) all the time,
except when in repairs or waiting for repair crew to attend. If M repair crews
are available, the time taken by any crew follows a negative exponential
distribution with mean T. Naturally, a machine which has broken down will
have to wait for repairs if all the repair crews are busy.
11.2.2 Measures of system efficiency
For a given set of machines, the efficiency of the repair system may be
judged by the extent to which machines have to wait for repairs. If W is the
average time for which a machine has to wait, the efficiency factor F is
defined as,
Operations Research Unit 11
SikkimManipalUniversity Page No. 218
T U W
T U
F
+ +
+
=
At any point of time, a machine will either be running or under repair or
waiting for repairs. Therefore, the total number of machinesN= J + H + L.
N
L
and
N
J
,
N
H
, correspond to the probability that a machine is being
repaired, running or waiting for repairs respectively. In the finite queuing
tables, service factor X is defined as
T U
T
X
+
= . X is an indicator of the
utilisation of repair crew.
The formulae for other properties of the system are given below:
T U W
T U
H L J
H J
F
NF (1 X)
T W U
NU
J
N (1 F)
T W U
NW
L
FNX
T W U
NT
H
+ +
+
=
+ +
+
=
= ÷
+ +
=
= ÷
+ +
=
=
+ +
=
Operations Research Unit 11
SikkimManipalUniversity Page No. 219
Table 11.1: Finite queuing Tables
Source: Shim, K. Jae, & Siegel, G. Joel. Operations Management . Barron's Educational Series
11.2.3 Use of finite queuing tables
The above table gives values of F and D for different values of N, M and X.
They are arranged in the ascending order of the values of the population.
For each N, the value of X increases from .001 to .950. For a given service
factor X, several values of M can be found. For each value of X and M,
values of D and F are tabulated. The steps for finite queuing tables are as
follows:
- Find mean service time T and mean running time U.
Operations Research Unit 11
SikkimManipalUniversity Page No. 220
- Compute the service factor.
- Select the table corresponding to the population N.
- For the given population, locate the service factor value.
- Read from tables, values of D and F for the number of service crews M.
- If necessary, these values may be interpolated between relevant values
of X.
- Calculate the other measures L, W, H and J from the formulae given.
Figure 11.3 depicts the steps for finite queuing tables.
Fig. 11.3: Finite Queuing Tables Steps
The overall efficiency F of the system will increase with the number of
service channels (M) provided. As mentioned earlier, addition of service
crews involves cost, which should be justified by the increase in the
efficiency of the system that is additional running time of machines.
However, it will be seen from the table that as M increases the rate of
increase in efficiency decreases. The practical significance is that beyond a
certain value of M, it is not worthwhile increasing M as there would be no
appreciable increase in the efficiency of the system.
Operations Research Unit 11
SikkimManipalUniversity Page No. 221
Problem 1:
In a chemical factory, there are five hoppers of identical size which feed
material to grinding mills. Due to changes in the requirement of material,
there are variations in the time taken for emptying the hoppers. On the
basis of past experience, it was found to follow negative exponential
distribution with an average of 10 hours between getting emptied.
Whenever a hopper gets empty, it has to be filled by a pay loader.
Although the capacity of the hopper is the same, the time taken to fill the
hoppers varies due to different locations from where the material is to be
loaded.
The time for filling the hopper also was found to follow negative
exponential distribution with an average of 2.5 hours. The company hires
the pay loaders at a cost of Rs. 100 per hour irrespective of whether it is
operated or not. If the mill has to be stopped due to the empty hopper, it
costs Rs. 1000 per hour in terms of loss of profits. To determine the
number of pay loaders which the company should engage to minimise
overall cost you need to create a finite queuing table.
Solved Problem 1:
For the above scenario, determine the number of pay loaders the company
should engage to minimise overall cost.
Solution: Since T = 2.5 and U = 10
0.200
2.5 10
2.5
T U
T
X =
+
=
+
=
For N = 5, X = 0.200 we have the following values from the tables:
M D F
3 0.028 0.998
2 0.194 0.976
1 0.689 0.801
We now prepare a table as depicted in table 11.1:
Operations Research Unit 11
SikkimManipalUniversity Page No. 222
Table 11.2: Number of Pay Loaders to be Engaged to Minimise Cost
1. No. of pay loaders 3 2 1
2. Overall efficiency of system (F) 0.998 0.976 0.801
3. N(1 – X) 4.00 4.00 4.00
4.
Average number of mills running
per hour J = NF(1 – X)
3.992 3.904 3.204
5. Expected profits atRs. 1000 per hour 3,992 3,904 3,204
6.
Loading cost at Rs. 100 per
hour per pay loader
300 200 100
7. Expected net profits per hour (5 – 6) 3,692 3,704 3,104
As increasing the number of pay loaders beyond two reduces the profits,
the company should engage only two pay loaders.
Self Assessment Questions
1. When the possible number of arrivals is limited, then we apply infinite
queuing model. (True/False)
2. The queue discipline in a finite queuing process can be random.
(True/False)
3. The efficiency factor for this model is
T W
T U
F
+
+
= . (True/False)
4. __________ principle states that customers are served one at a time
and that the customer that has been waiting the longest is served first.
5. ______ discipline models give priority to rush jobs and important
customers over others.
11.3 Summary
Let us recapitulate the important concepts discussed in this unit:
- This unit on finite queuing theory deals with situations where customers
arrive, wait for the service, get the service and leave the system.
- Customers may come individually or in groups from large/small
population, at known/variable times, form one or more queues and move
in a certain order to the service station(s)to avail of services whose
speed may be fixed or variable.
Operations Research Unit 11
SikkimManipalUniversity Page No. 223
- Queuing systems are analysed for determining the optimal service level,
where the total cost of providing service and waiting, is minimised.
- An increase in the service level increases the cost of providing service,
but reduces the cost of waiting, while a decrease in the service level
induces opposite changes.
11.4 Glossary
Stack: Last in first out collection
Variance: how far the waiting time in a queue is spread out
Trade-off: compromise at the expense of one or more objectives
11.5 Terminal Questions
1. Customers arrive at the first class ticket counter of a theatre at a rate of
12 per hour. There is one clerk serving the customers at a rate of 30
per hour.
i) What is the probability that there is no customer in counter (for
instance, the system is idle)?
ii) What is the probability that there are more than 2 customers in the
counter?
iii) What is the probability that there is no customer waiting to be
served?
iv) What is the probability that a customer is being served and nobody
is waiting?
2. Assume that at a bank teller window, the customers arrive in their cars
at the average rate of twenty per hour according to thePoisson
distribution. Also assume that the bank teller spends an average of two
minutes per customer to complete a service and the service time is
exponentially distributed. Customers, who arrive from an infinite
population, are served on a FCFS basis and there is no limit to the
possible queue length.
i) What is the expected waiting time in the system per customer?
ii) What is the mean number of customers waiting in the system?
iii) What is the probability of zero customers in the system?
iv) What value is the utilisation factor?
Operations Research Unit 11
SikkimManipalUniversity Page No. 224
11.6 Answers
Self Assessment Questions
1. False
2. True
3. False
4. First Come - First Served (FCFS)
5. Priority
Terminal Questions
1. i. P (the system is idle) = 1- µ = 1-0.4 = 0.6.
ii. P (n > 2) = 1 – P (n s 2) = 1 – [P (0) + P (1) + P (2)] = 1- 0.936 =
0.064.
iii. P (no customer waiting to be served) = P (0) + P (1) = 0.84.
iv. P (a customer is being served and none is waiting) = P (1) = 0.24.
For more details, refer section 11.2
2. i. Expected waiting time in the system per customer,
Ws = 1/ (µ-ì) = 1/ (30-20) =1/10 hr.
ii. Mean number of customers waiting in the system,
Lq =
3
4
3
2
1
3
2
1
2
=
÷
=
÷ µ
µ
.
iii. Probability of zero customers in the system,
P (0) =
3
1
3
2
1÷ µ = 1÷ = .
iv. Utilisation factor,
3
2
µ =
For more details, refer section 11.2
11.7 Case Study
Warehousing Corporation
A single crew is provided for unloading and/or loading each truck that
arrives at the loading deck of a warehouse. These trucks arrive according to
Operations Research Unit 11
SikkimManipalUniversity Page No. 225
a Poisson input process at a mean rate of one per hour. The time required
by a crew to unload and/or load a truck has an exponential distribution. The
expected time required by a one man crew would be two hours.
The cost of providing each additional member of the crew is Rs 10 per hour.
The cost that is attributable to having a truck not in use is estimated to be
Rs 40 per hour. Assume that the mean service rate of the crew is
proportional to its size, what should the size be in order to minimise the
expected total cost per hour?
The general routine is that order pickers assemble orders placed by stores
and load them.Oneof the company trucks wait at the single loads, on a
FCFS basis. They then proceed to the store where the order is destined,
unload and return for another order. Because of the many different routes
and distances, and traffic problems at different times of the day, the time
between arrivals of trucks at the dock is random–averaging 30 minutes. The
loading time also follows an exponential distribution–averaging 15
minutes.Truck drivers are paid Rs. 10 per hour and the crew of two loaders
are each paid Rs. 4 per hour. Truckers have complained about the long
waiting, so a sample was taken showing that truckers did indeed wait for an
average of 32 minutes. The warehouse manager knows that a second truck
dock would probably solve the problem, but the large capital expenditure
plus the disruption of operations during construction are deterrents to the
solution. Tests are made with different crew patterns and it is found that a
crew of 3 can be used to advantage, reducing the loading time to
10 minutes. As management trainee working with the corporation, help the
corporation in finding suitable answers to the following questions:
Discussion Questions:
a. Is the crew of 3 loaders more economical than the crew of 2?
b. What is the probability that an arriving truck will find at least one truck
already in the system?
c. How long is the crew idle?
d. If another truck dock is available with the present operating
characteristics, how will it affect the total expected cost per hour?
Operations Research Unit 11
SikkimManipalUniversity Page No. 226
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
E-References:
- http://www.scribd.com/doc/14160718/Operations-Management-Ch19-
Waiting-Lines. accessed on Dec 20, 2011.
- http://staff.um.edu.mt/jskl1/simweb/intro.htm. accessed on Dec 30, 2011.
- http://docs.google.com/gview?a=v&q=cache%3AuNc0DtLpqBMJ%3Aw
ww.stsc.hill.af.mil%2Fcrosstalk%2F2007%2F12%2F0712LauAppendix.
pdf+finite+queuing+model&hl=en&gl=uk&sig=AFQjCNHDjm-
WKl7g8LD_DztsUvEJxyr9Ig&pli=1. accessed on Dec 27, 2011.
Operations Research Unit 12
Sikkim Manipal University Page No. 227
Unit 12 Simulation
Structure:
12.1 Introduction
Objectives
12.2 Simulation
12.3 Methodology of Simulation
12.4 Basic Concepts
12.5 Simulation Procedure
Allocation of random numbers
Use of random number tables
12.6 Sample Size
12.7 Summary
12.8 Glossary
12.9 Terminal Questions
12.10 Answers
12.11 Case Study
12.1 Introduction
In the previous unit, you learnt the finite queuing models, finite queuing
tables and the use of these models and tables. Simulation is a numerical
technique for conducting experiments that involve certain types of
mathematical and logical relationships necessary to describe the behaviour
and structure of a complex real world system over an extended period of
time. In other words, it is a quantitative technique that utilises a
computerised mathematical model in order to represent actual decision
making under conditions of uncertainty for evaluating alternative courses of
action based upon facts and assumptions. In this unit, you will learn about
simulation, methodology and basic concepts of simulation. You will also
learn simulation procedure and sample size.
A definition of simulation as given by Shannon:
―Simulation is the process of defining a model of a real system and
conducting experiments with this model for the purpose of understanding
the behaviour (within the limits imposed by a criterion or a set of criteria) for
the operation of a system.”
Operations Research Unit 12
Sikkim Manipal University Page No. 228
Objectives:
After studying this unit, you should be able to:
- define simulation
- explain the application of simulations in business problems
- describe the use of random number table
12.2 Simulation
Using simulation, an analyst can introduce the constants and variables
related to the problem, set up the possible courses of action and establish
criteria which act as measures of effectiveness. The major reasons for
applying simulation technique to OR problems may be listed as below:
1. It is an appropriate tool to use in solving a problem when experimenting
on the real system:
a) Would be disruptive.
b) Would be too expensive.
c) Does not permit replication events.
d) Does not permit control over key variables.
2. It is desirable tool for solving a business problem when a mathematical
model is:
a) Complex to solve.
b) Beyond the capacity of available personnel.
c) Not detailed enough to provide information on all important decision
variables.
3. The major reasons for adopting simulation in place of other
mathematical techniques are:
a) It may be the only method available, because it is difficult to observe
the actual reality.
b) Without appropriate assumption, it is impossible to develop a
mathematical solution.
c) It may be too expensive to actually observe the system.
d) There may not be sufficient time to allow the system to operate for a
very long time.
4. It provides a trial-and-error movement towards the optimal solution. The
decision maker selects an alternative to experience the effect of the
selection and then improves the selection. In this way, the selection is
adjusted until it approximates the optimal solution.
Operations Research Unit 12
Sikkim Manipal University Page No. 229
12.3 Methodology of Simulation
In the previous section, you learnt why simulation technique is applied. You
will now learn the methodology of simulation. The methodology developed
for simulation process consists of seven steps depicted in figure 12.1:
Fig. 12.1: Steps in Methodology of Simulation
Self Assessment Questions
1. Simulation is useful to analyse problems where analytical solution is
difficult. (True/False)
2. Simulation cannot be used where mathematical techniques can be
used. (True/False)
3. Simulation solutions would be identical to those using mathematical
techniques. (True/False)
Operations Research Unit 12
Sikkim Manipal University Page No. 230
12.4 Basic Concepts
In the previous section, you learnt the methodology of simulation. You will
now learn the basic concepts. Simulation is also called experimentation in
the management laboratory. While dealing with business problems,
simulation is often referred to as ‗Monte Carlo Analysis‘. Two American
mathematicians, Von Neumann and Ulan, in the late 1940s, found a
problem in the field of nuclear physics too complex for analytical solution
and too dangerous for actual experimentation. They arrived at an
approximate solution by sampling. The method they used had resemblance
to the gambler‘s betting systems on the roulette table, hence the name
‗Monte Carlo‘ has stuck.
Imagine a betting game where the stakes are based on correct prediction of
the number of heads, which occur when five coins are tossed. If it were only
a question of one coin, most people know that there is an equal likelihood of
a head or a tail occurring, that is the probability of a head is ½. However,
without the application of probability theory, it would be difficult to predict the
chances of getting various numbers of heads, when five coins are tossed.
Why don‘t you take five coins and toss them repeatedly? Note down the
outcomes of each toss after every ten tosses, approximate the probabilities
of various outcomes. As you know, the values of these probabilities will
initially fluctuate, but they would tend to stabilise as the numbers of tosses
are increased. This approach in effect is a method of sampling, but is not
very convenient. Instead of actually tossing the coins, you can conduct the
experiment using random numbers. Random numbers have the property
that any number is equally likely to occur, irrespective of the digit that has
already occurred.
Let us estimate the probability of tossing of different numbers of heads with
five coins. We start with set random numbers depicted in table 12.1:
Table 12.1: Random Number Set
78466 71923
78722 78870
06401 61208
04754 05003
97118 95983
Operations Research Unit 12
Sikkim Manipal University Page No. 231
By following a convention that ―even‖ digits signify a head (H) and the ―odd‖
digits represent a tail (T), the tossing of a coin can be simulated. The
probability of occurrence of the first set of digits is ½ and that of the other
set is also ½ – a condition corresponding to the probability of the occurrence
of a head and the probability of occurrence of a tail respectively.
It is immaterial as to which set of five digits should signify a head. The rule
could be that the digits 0, 1, 2, 3 and 4 represent a head and the digits 5, 6,
7, 8 and 9 a tail. It is only necessary to take care that the set of random
numbers allotted to any event matches with its probability of occurrence. For
instance, if you‘re interested in allotting random numbers to three events A,
B and C with respective probabilities 0.24, 0.36 and 0.40, choose two digit
random numbers from 00 to 99.
The numbers 00 to 23 signify event A, 24 to 59 signify B and 60 to 99 signify
C. The first set of five random digits in the list of random numbers implies
that the outcome of the first toss of 5 coins is as depicted in table 12.2:
Table 12.2: Outcome of First Toss of 5 Coins
Coin 1 2 3 4 5
Random Number 7 8 4 6 6
Outcome T H H H H
Hence it is 4 heads and 1 tail.
Proceeding in the same way, you can tabulate the results of the first ten
tosses as depicted in table 12.3.
Table 12.3: Results of First Ten Tosses
Toss No. Number of
Heads Tails
1
2
3
4
5
6
7
8
9
10
4
3
4
3
1
1
3
4
3
1
1
2
1
2
4
4
2
1
2
4
Operations Research Unit 12
Sikkim Manipal University Page No. 232
Based on the ten tosses of the coins, the estimates of probabilities of
occurrence of different numbers of heads are as depicted in figure 12.2:
Fig. 12.2: Probabilities of Occurrence of Different Numbers of Heads
Continue experimenting further, as these estimates come closer to the
theoretical value with increasing sample size.
The results for obtaining 2 heads and 3 tails for 100 throws are depicted in
table 12.4:
Table 12.4: Results for 100 Throws
10 throws 0
20 throws 6
30 throws 11
40 throws 14
50 throws 18
60 throws 19
70 throws 21
80 throws 22
90 throws 24
100 throws 27
Table 12.5 compares the final results at the end of 100 throws with the
theoretical probabilities.
Operations Research Unit 12
Sikkim Manipal University Page No. 233
Table 12.5: Final Results at the End of 100 Throws
No. of
heads
Estimated
Probability
Theoretical
Probability
0
1
2
3
4
5
0.03
0.21
0.27
0.33
0.12
0.04
0.03
0.16
0.31
0.31
0.16
0.03
It is observed that the results obtained with the large sample of 100, when
compared are more in favour with the theoretical values than with a sample
of ten sets of numbers.
Self Assessment Questions
4. Simulation may be called experimentation in the _________ _______.
5. Random numbers have the property that any number is ____ to occur.
6. The totality of probability assigned to the variable should always be
equal to _______.
12.5 Simulation Procedure
In the previous section, you learnt the basic concepts. You will now learn the
simulation procedure. The approach to solve a gambling problem can be
extended to decision making in business where risk is a common feature.
The probabilities associated with the variables can be estimated on the
basis of availability of previous data or by entering subjective values.
In any simulation problem, the variables to be studied will be given with
associated probabilities. The initial conditions will also be specified. You can
choose random numbers from table. However, to get uniform results, the
random numbers will be specified. The first step involves coding the data
that is, you assign random numbers to the variable. Then you identify the
relationship between the variables and run the simulation to get the results
Let us illustrate this by a simple example of a queuing process.
Operations Research Unit 12
Sikkim Manipal University Page No. 234
Solved problem 1:
A sample of 100 arrivals of customers at a retail sales depot is according to
the following distribution:
Table 12.6: Samples of 100 Arrivals of Customers
Time between Arrivals (min) Frequency
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
2
6
10
25
20
14
10
7
4
2
Study of the time required to service customers by adding up the bills,
receiving payment, making change and placing packages in hand trucks,
yields the distribution depicted in table 12.7:
Table 12.7: Study of the Time Required to Service Customers
Service time (min) Frequency
.5
1.0
1.5
2.0
2.5
3.0
12
21
36
19
7
5
Estimate the average percentage customer waiting time and average
percentage idle time of the server by simulation for the next 10 arrivals.
Solution:
Step 1: Convert the frequency distributions of time between arrivals and
service time to cumulative probability distributions.
Step 2: Allocate random numbers 00 to 99 for each of the values of time
between arrivals and service time. The range allocated to each value
corresponds to the value of cumulative probability.
Operations Research Unit 12
Sikkim Manipal University Page No. 235
Step 3: Using random numbers from table, sample the random time of
arrival and service time for ten sets of random numbers.
Step 4: Tabulate waiting time of arrivals and idle time of servers.
Step 5: Estimate the percentage waiting time of arrivals and percentage idle
time of servers corresponding to the ten samples.
12.5.1 Allocation of random numbers
The tables 12.8, 12.9 and 12.10 depict the allocation of random numbers
based on time between arrivals and service time.
Table 12.8: Allocation of Random Numbers – Time Between Arrivals
Time
Between
Arrivals
(1)
Frequency
(2)
Cumulative
Frequency
(3)
Cumulative
Probability
(3)/100
(4)
Random
Number
Allocated
(5)
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
2
6
10
25
20
14
10
7
4
2
2
8
18
43
63
77
87
94
98
100
0.02
0.08
0.18
0.43
0.63
0.77
0.87
0.94
0.98
1.00
00 to 01
02 to 07
08 to 17
18 to 42
43 to 62
63 to 76
77 to 86
87 to 93
94 to 97
98 and 99
Table 12.9: Allocation of Random Numbers – Service Time
Service
Time
(1)
Frequency
(2)
Cumulative
Frequency
(3)
Cumulative
Probability
(4)
Random Nos.
Allocated
(5)
0.5
1.0
1.5
2.0
2.5
3.0
12
21
36
19
7
5
12
33
69
88
95
100
0.12
0.33
0.69
0.88
0.95
1.00
00 to 11
12 to 32
33 to 68
69 to 87
88 to 94
95 to 99
Operations Research Unit 12
Sikkim Manipal University Page No. 236
Note that the upper bound of random numbers allocated for each value of
the parameter is one less than the corresponding cumulative frequency,
since you have chosen a range of random numbers from 00 to 99
Table 12.10: Arrival and Service Time
Arrivals Service
Arrival No.
Random No.
Time
Between
Arrivals
Time of
Arrival
Random
No.
Service Time
Time
of
Start
Time
of
Finish
Waiting
Time of
Arrival
Idle
Time
of
server
1
2
3
4
5
6
7
8
9
10
78
78
06
04
97
71
78
61
05
95
3.5
3.5
1.0
1.0
4.5
3.0
3.5
2.5
1.0
4.5
3.5
7.0
8.0
9.0
13.5
16.5
20.0
22.5
23.5
28.0
54
24
51
45
46
84
58
58
60
24
1.5
1.0
1.5
1.5
1.5
2.0
1.5
1.5
1.5
1.0
3.5
7.0
8.0
9.5
13.5
16.5
20.0
22.5
24.0
28.0
5.0
8.0
9.5
11.0
15.0
18.5
21.5
24.0
25.5
29.0
-
-
-
0.5
-
-
-
-
0.5
-
3.5
2.0
-
-
2.5
1.5
1.5
1.0
-
2.5
Total 1.0 14.5
The service facility is made available at clock time zero and the server has
to be idle for 3.5 minutes, when the service for first arrival starts. The service
is completed at 5.0 minutes and again the server is idle for 2 minutes till the
second arrival joins the system. The first three arrivals get immediate
service and they don‘t have to wait, as the server is idle when they arrive.
The fourth arrival that joins at 9.0 minutes has to wait for 0.5 minute, while
the service to the third is completed. Similarly the waiting time and idle time
can be computed for further arrivals.
Total elapsed time = 29 minutes
Waiting time of arrival = 1 minute
Percentage of waiting time = ( 1 x 100) / 29 = 3.4
Idle time for server = 14.5 minutes
Percentage of idle time = ( 14.5 x 100) / 29 =50%
Operations Research Unit 12
Sikkim Manipal University Page No. 237
12.5.2 Use of random number tables
The random numbers can be selected by any random process, similar to
drawing numbered chips from a hat. However, it is convenient to use a table
of random numbers prepared on the basis of some physical phenomenon.
The grouping of random numbers in the tables has no significance and one
should be concerned with individual digits only. The first random number
could be picked at random from any point in the tables and the subsequent
ones are to be selected preceding sequentially either in vertical or horizontal
direction. Depending upon the number of digits required, the random
numbers will be chosen in sets of single digit or two digit numbers, that is
pseudorandom numbers.
True random numbers cannot be produced by an algorithm and hence
random numbers generated using recursive equations are referred to as
pseudorandom numbers.
There are several methods of generating pseudorandom numbers, but you
will briefly learn only the mid square method. Operation starts with an
arbitrary four digit integer called the seed. To obtain the first random
number, the seed is squared and all digits except the middle four are
ignored. This process is repeated each time using the previous random
number as the new seed.
Seed Uo = 8695
Uo
2 = 75603025
Taking the middle 4 digits,
U1 = 6030
U1
2 = 36360900
U2 = 3609
Repeating the above procedure:
U3=0248, U4=0615, U5= 3182, U6=3035 U7= 2112, U8 = 4805, U9 =2063
One of the disadvantages of the mid square method is that the generated
numbers start cycling after a short set of random numbers is obtained.
There are methods by which the seed can be chosen to obtain a fairly long
sequence of numbers before cycling starts. Statistical tests check whether
the generated sequence is truly random.
Operations Research Unit 12
Sikkim Manipal University Page No. 238
Self Assessment Questions
7. In any simulation problem, initial conditions are stated. (True/False)
8. Random numbers are assigned for cumulative probability values.
(True/False)
9. Without identifying any relationship between variables, you can solve
the simulation problem. (True/False)
12.6 Sample Size
In the previous section, you learnt the simulation procedure. You will now
learn the sample size. As you have seen with the coin-tossing experiment,
the larger the number of trials, the more confident you can be in your
estimates. The question that arises is how many trials for simulation. If the
experiment is as simple as tossing a coin involving only one variable, you
can work out the sample size required for a given confidence level.
Usually, a simulation model involves several variables making it impossible
to determine the number of trials required to obtain the desired accuracy at
a specified confidence level.
One can only say that the accuracy associated with simulation improves as
the square root of the number of trials, which results in large number of
trials.
This requires a great deal of computational effort and the use of computer
becomes inevitable. In fact, special simulation languages such as GPSS
and SIMSCRIPT have been developed to save time and effort required to
structure and debug simulation models.
A practical indicator of when to stop simulation trials is when the results that
violently fluctuate initially tend to stabilise if the simulation is continued. If the
successive cumulative results tally reasonable well, the simulation may be
stopped. The degree of accuracy required varies with the problem on hand
and calls for analyst judgment.
Self Assessment Questions
10. A simulation model involves several variables making it impossible to
determine the number of trials required to obtain the desired accuracy
at a specified confidence level. (True/False)
Operations Research Unit 12
Sikkim Manipal University Page No. 239
11. Large number of trails require a great deal of computational effort and
the use of computer becomes optional. (True/False)
12. A practical indicator of when to stop simulation trials is when the results
that violently fluctuate initially tend to stabilise if the simulation is
continued. (True/False)
Solved problem 2:
Every morning a bread vendor buys loaves of bread at Re. 0.45 each by
placing his order one day in advance (at the time of receiving his previous
order) and sells them at Re. 0.70 each. Unsold bread can be sold the next
day at Re. 0.20 per loaf and thereafter should be treated as of no value. The
pattern of demand for bread is depicted in table 12.11:
Table 12.11: Pattern of Demand for Bread
Fresh Bread One Day old Bread
Daily Sales Probability
of demand
Daily
Sales
Probability of
demand
50 0.01 0 0.10
51 0.03 1 0.20
52 0.04 2 0.08
53 0.07 3 0.02
54 0.09
55 0.11
56 0.15
57 0.21
58 0.18
59 0.09
60 0.02
The vendor adopts the following order rule, if there is no stock with him at
end of the previous day. He orders 60 units. Otherwise he orders 50 or 55
whichever is nearest the actual fresh bread sale on the previous day.
Starting with zero stock and a pending order for 55 loaves simulate for 10
days and calculate the vendor‘s profits.
Operations Research Unit 12
Sikkim Manipal University Page No. 240
Solution:
Table 12.12 depicts the allocation of random numbers.
Table 12.12: Allocation of Random Numbers
Fresh Bread One day old bread
Daily
Sale
Prob
of
dema
nd
Cum Prob
of
demand
Rand
No.
allocated
Dail
y
Sale
Prob of
deman
d
Cum
Prob of
demand
Rand
No.
Alloca
-ted
50
51
52
53
54
55
56
57
58
59
60
0.01
0.03
0.04
0.07
0.09
0.11
0.15
0.21
0.18
0.09
0.02
0.01
0.04
0.08
0.15
0.24
0.35
0.50
0.71
0.89
0.98
1.00
00
01 – 03
04 – 07
08 – 14
15 – 23
24 – 34
35 – 49
50 – 70
71 – 88
89 – 97
98 - 99
0
1
2
3
0.70
0.20
0.08
0.02
0.70
0.90
0.98
1.00
00 to
69
70 to
89
90 to
97
98 to
99
You can now construct a table through simulation to see how the stocks and
sales fluctuate. Table 12.13 depicts the results of simulation.
Operations Research Unit 12
Sikkim Manipal University Page No. 241
Table 12.13: Results of Simulation
FRESH BREAD ONE DAY BREAD
Day
Receipt
of the
start day
Random
No
Sale
Closing
stock
Order
for
next
day
Opening
stock
Random
No.
Sale
1
2
3
4
5
6
7
8
9
10
55
60
60
50
55
60
60
55
55
60
72
06
12
14
79
70
85
71
21
98
58
52
53
50
58
55
58
57
58
55
58
54
60
0
8
7
0
0
3
2
0
1
0
60
60
50
55
60
60
55
55
60
55
0
0
8
7
0
0
3
2
0
1
86
54
88
58
48
0
0
1
0
0
0
1
0
0
0
Total 570 549 21 2
Table 12.13 represents lost sales, as stock is limited as previous day‘s
closing stock is zero
Estimated profit = (549 ×0.70 +2 ×20) -570 ×0.45 = Rs. 128.20
Solved problem 3:
The maintenance manager of a chemical company is interested in
determining a rational policy for maintenance of pneumatic conveying
equipment. The equipment is a part of the process line and affects
production. It has one bearing each on the inlet side (A) and the outlet side
(B). Whenever there is a failure of any bearing, it has to be replaced
immediately. The company has a good system of maintaining records on
performance of the equipment and the failure records are available.
Given the cost of the bearings, the cost of downtime of equipment and the
time taken to replace either one bearing or both the bearings, you can use a
simulation model to find out the best alternative solution for a given set of
constraints. For the above scenario, the following failure data is available:
Operations Research Unit 12
Sikkim Manipal University Page No. 242
Table 12.14 depicts the equipment performance data.
Table 12.14: Equipment Performance Data
The cost of bearing is Rs. 300 each for ‗A‘ and Rs. 500 each for ‗B‘. The
cost of downtime of equipment is Rs. 700 per hour and it takes 2 hours to
replace one bearing either at inlet or outlet side and 3 hours to replace both
the bearings. The three maintenance policies to be evaluated are:
1. Replace a bearing only when it fails.
2. Replace both the bearings if one fails.
3. Replace the bearing which fails plus the other one if it has been in use
for more than its estimated average services life – 600 hours for bearing
A and 860 hours for bearing B
Find the best alternative through simulation.
Solution:
It is assumed that the failure of a bearing is independent of the maintenance
policy followed. Random numbers are allocated for different failure times:
(Table 12.15 depicts the random numbers)
No. of failures
Age at failure
(hours)
Inlet side
Bearing (A)
Outlet side
Bearing (B)
150
300
450
600
750
900
1,050
1,200
1,350
7
16
18
23
14
10
7
5
0
0
3
9
12
16
18
33
7
2
Operations Research Unit 12
Sikkim Manipal University Page No. 243
Table 12.15: Random Numbers
Bearing A Bearing B
Life (hrs) Cumulative
Probability
Random Nos. Cumulative
Probability
Random
Nos.
150
300
450
600
750
900
1050
1200
1350
0.07
0.23
0.41
0.64
0.78
0.88
0.95
1.00
-
00 to 96
07 to 22
23 to 40
41 to 63
64 to 77
78 to87
88 to 94
95 to 99
-
0.00
0.03
0.12
0.24
0.40
058
0.91
0.98
1.00
-
00 to 02
03 to 11
12 to 23
24 to39
40 to57
58 to 90
91 to 97
98 and99
We can now select random numbers from the tables and generate a set of
12 bearings for each of the bearings. Table 12.16 depicts the set of 12
bearing for each of the bearings generated.
Table 12.16: Set of 12 Bearings for Each of the Bearings Generated
1
2
3
4
5
6
7
8
9
10
11
12
10
22
24
42
37
77
99
96
89
85
28
63
300
300
450
600
450
750
1200
1200
1050
900
450
600
300
600
1050
1650
2100
2850
4050
5250
6300
7200
7650
8250
99
96
18
36
50
79
80
96
31
07
62
77
1350
1200
600
750
900
1050
1050
1200
750
450
1050
1050
1350
2550
3150
3900
4800
5850
6900
8100
8850
9300
10,350
11,400
Let us compare the costs of three policies for the first 7200 hours.
Policy 1: Replace a bearing only when it fails.
‗A‘ requires replacement 10 times, and ‗B‘ 7 times during this period as seen
from the lives of successive bearings.
Operations Research Unit 12
Sikkim Manipal University Page No. 244
Total cost = (300 × 10 + 500 × 7) + (17 × 2 × 700) = Rs. 30, 300
Policy 2: Replace both the bearing if one fails.
Table 12.17 depicts bearing failure information.
Table 12.17: Bearing Failure
Elapsed time
(hrs)
Bearing fails
first
Life (hrs) of
newly fitted
bearing A (From
table 12.16)
Life (hrs) of
newly fitted
bearing B (From
table 12.16)
300 A 300 1200
600 A 450 600
1050 A 600 750
1650 A 450 900
2100 A 750 1050
2850 A 1200 1050
3900 B 1200 1200
5100 A & B 1050 750
5850 B 900 450
6300 B 450 1050
6750 A 600 1050
7350 A - -
Policy 3 is the cheapest.
The simulation is limited to 7200 hours of operation since the purpose is
only to illustrate the method. With such small number of trials the results
may turn out to be erratic. Consider policy 1 again. With a different set of
random numbers and extended simulation, there may be occasions of both
bearing failing at the same time, thus affecting the cost of downtime for
replacement. Simulation with sufficiently large number of trials can only lead
to dependable decisions.
Operations Research Unit 12
Sikkim Manipal University Page No. 245
Table 12.18 depicts the bearing record.
Table 12.18: Bearing Record
Start
of
the
period
Life bearing
in service
(hrs)
Be
aring
wh
ich
fail
s
Time
elapsed
(hrs)
Age
surviving
bearing
Did the
surviving
bearing
(A/B)
Complete
its est.
avg life?
If “No”
in col
7
balance
life
of
surviving
Replace
ment
policy
A B
(1) (2) (3) (4) (5) (6) (7) (8) (9)
0 300 1350 A 300 300 B - No 1050 Replace
A
300 300 1050 A 600 600 B - No 750 Replace
A
600 450 750 A 1050 1050 B - Yes -- Replace
A & B
1050 600 1200 A 1650 600 B - No 600 Replace
A
1650 450 600 A 2100 1050 B - Yes -- Replace
A & B
2100 750 600 B 2700 600 B - Yes -- Replace
A & B
2700 1200 750 B 3450 750 B - Yes -- Replace
A & B
3450 1200 900 B 4350 900 B - Yes -- Replace
A & B
4350 1050 1050 A&
B
5400 -- -- Replace
A & B
5400 900 1050 A 6300 900 B - Yes -- Replace
A & B
6300 450 1200 A 6750 450 B - No 750 Replace
A & B
6750 600 750 A 7350 1050 B - Yes -- Replace
A & B
This will be the life (from Table 12.18) if newly fitted. For survivors from
previous replacement, this will be the balance life (col. 8).
Operations Research Unit 12
Sikkim Manipal University Page No. 246
Solved problem 4:
A factory produces 150 scooters. But the production rate varies with the
distribution depicted in table 12.19.
Table 12.19: Production Rate and Probability
Production
Rate
147 148 149 150 151 152 153
Probability 0.05 0.10 0.15 0.20 0.30 0.15 0.05
At present the track will hold 150 scooters. Using the following random
numbers determine the average number of scooters waiting for shipment in
the factory and average number of empty space in the truck. Random
Numbers 82, 54, 50, 96, 85, 34, 30, 02, 64, 47.
Solution:
Table 12.20 depicts the production rate and probability.
Table 12.20: Production Rate and Probability
Production
rate
Probability
Cumulative
Probability
Random No.
Assigned
147 0.05 0.05 00 – 04
148 0.10 0.15 05 - 14
149 0.15 0.30 15 - 29
150 0.20 0.50 30 - 49
151 0.30 0.80 50 - 79
152 0.15 0.95 80 - 94
153 0.05 1.00 95 - 99
Table 12.21 depicts the simulation worksheet.
Table 12.21: Simulation Worksheet
Trial
No.
Random
No.
Simulated
Production
Rate
Scooter
Waiting in
the factory
Number of
example
spaces in
the truck
1 82 152 2 -
2 54 150 2 -
3 50 150 2 -
4 96 153 5 -
5 85 152 7 -
6 34 150 7 -
Operations Research Unit 12
Sikkim Manipal University Page No. 247
7 30 150 7 -
8 02 147 4 3
9 64 151 5 -
10 47 150 7 -
Total 3
Therefore, average number of scooters waiting = 7/10 =0.7/day
Average number of empty space = 3/10 = 0.3/day
Solved problem 5:
Dr. Strung is a dentist. He gives appointments to patients every half an hour.
However, he does not know the nature of illness of patients arriving at his
clinic. From past records, he has the following probability distribution and
also knows the exact treatment timings. He starts his clinic at 8.00 A.M.
Using the following information determine the average waiting time of the
customers and idle time of the doctor. Random Numbers 56, 40, 26, 66, 87,
48, 17, 22, 04, 15.
Table 12.22 depicts the probability distribution in past record.
Table 12.22: Probability Distribution in Past Record
Nature of illness Probability Time taken for
treatment (min)
Filling 0.10 50
Check up 0.30 15
Crowning 0.15 40
Cleaning 0.30 15
Extraction 0.15 30
Solution:
Table 12.23 depicts the assigning of random number.
Table 12.23: Assigning Random Number
Illness Probability Cum. Prob.
Random No.
Assigned
Filling 0.10 0.10 00 -09
Check up 0.30 0.40 10 -39
Crowning 0.15 0.55 40 -54
Cleaning 0.30 0.85 55 -84
Extraction 0.15 1.00 85 -99
Operations Research Unit 12
Sikkim Manipal University Page No. 248
Table 12.24 depicts the simulation worksheet.
Table 12.24: Simulation Worksheet
Trial
No
Random
No.
Nature of
illness
Time
Taken
Patients
Arrival
Time
Treatment
Doctor
Idle time
Patients
Waiting
Time
Starts Finishes
1 56 Cleaning 15 8.00 8.00 8.15 -
2 40 Crowning 40 8.30 8.30 9.10 15 -
3 26 Check up 15 9.00 9.10 9.25 - 10
4 66 Cleaning 15 9.30 9.30 9.45 5 -
5 87 Extraction 30 10.00 10.00 10.30 15 -
6 48 Crowing 40 10.30 10.30 11.10 - -
7 17 Check up 15 11.00 11.10 11.25 - 10
8 22 Check up 15 11.30 11.30 11.45 5 -
9 04 Filling 50 12.00 12.00 12.50 15 -
10 15 Check up 15 12.30 12.50 13.05 - 20
Total 45 40
Doctor idle time = 45 minutes
Patients average waiting time = 40/10 = 4 minutes
12.7 Summary
Let us recapitulate the important concepts discussed in this unit:
- Using simulation, an analyst can introduce the constants and variables
related to the problem, set up the possible courses of action and
establish criteria which act as measures of effectiveness.
- The methodology developed for simulation process consists of seven
steps.
- In any simulation problem, the variables to be studied will be given with
associated probabilities.
- The mid square method is used for generating pseudorandom numbers.
One of the disadvantages of the mid square method is that the
Operations Research Unit 12
Sikkim Manipal University Page No. 249
generated numbers start cycling after a short set of random numbers is
obtained.
- The accuracy associated with simulation improves as the square root of
the number of trials, which results in large number of trials.
12.8 Glossary
Sample size: Number of trials in a simulation
GPSS (General Purpose Simulation System): a discrete event simulation
programming system developed in 1950‘s for application in queuing,
manufacturing, transportation and defense systems
SIMSCRIPT: a discrete event simulation programming system developed in
1962 by Harry Markowitz .It can be programmed with simple English like
statements to generate the simulation model.
12.9 Terminal Questions
1. Explain the different steps involved in simulation methodologies?
2. Briefly write down the basic concepts concerned with simulation.
3. Explain the simulation procedure?
4. Briefly explain the use of random number tables.
5. Two components have to be produced in M/c A and M/c B and then
finally assembled. The time taken to assemble on two machines varies
with the probability distribution depicted in table 12.25. Using simulation
technique and the ordered pair of random numbers, first for M/c A and
second for M/c B, find the average time taken for production.
Table 12.25: Assembling Time for the Two Machines
M/c - A M/c - B
Production
Time in
min
Probability Production
Time in
min
Probability
22 0.15 30 0.05
23 0.20 31 0.15
24 0.30 32 0.25
25 0.20 34 0.25
26 0.15 35 0.20
36 0.10
Operations Research Unit 12
Sikkim Manipal University Page No. 250
Random Numbers (10, 92); (25, 83); (36, 76); (44, 15); (57, 25); (62, 67);
(04, 99); (72, 53); (81, 35); (94, 07)
12.10 Answers
Self Assessment Questions
1. True
2. False
3. False
4. Management laboratory
5. Equally likely
6. 1
7. True
8. True
9. False
10. True
11. False
12. True
Terminal Questions
1. The methodology developed for simulation process consists of seven
steps. For more details refer to section 12.3
2. Concepts of simulation: simulation of complex problems, sampling
allotting random numbers to events and theoretical probability. For more
details refer to section 12.4
3. In any simulation problem, the variables to be studied will be given with
associated probabilities. The initial conditions will also be specified. You
can choose random numbers from table. For more details refer to
section 12.5
4. To generate a list of number and allot it to events. For more details refer
to section 12.5.2
5. 57.7 minutes. For more details refer to section 12.6
Operations Research Unit 12
Sikkim Manipal University Page No. 251
12.11 Case Study
Coin-Flipping Game
You are the lucky winner of a sweepstakes contest. Your prize is an allexpense-
paid vacation at a major hotel in Las Vegas, including some chips
for gambling in the hotel casino.
Upon entering the casino, you find that, in addition to the usual games
(blackjack, roulette, etc.), they are offering an interesting new game with the
following rules:
a) Each play of the game involves repeatedly flipping an unbiased coin
until the difference between the heads tossed and the number of tails is
three.
b) If you decide to play the game, you are required to pay US $1 for each
flip of the coin. You are not allowed to quit during a play of the game.
c) You receive US $8 at the end of each play of the game.
Thus, you win money if the number of flips required is fewer than 8, but you
lose money if more than 8 flips are required. Here are some examples (H
denotes head and T denotes Tail)
HHH 3 flips You can win US $5
THTTT 5 flips You can win US $3
THHTHTHTTTT 11 flips You lose US $3
How would you decide whether to play this game?
This can be simulated because you are imitating the actual play of the game
without actually winning or losing any money.
Discussion Question
1. Generate a sequence of random observations and give your suggestion.
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
Operations Research Unit 13
Sikkim Manipal University Page No. 252
Unit 13 Simulation Monte-Carlo Method
Structure:
13.1 Introduction
Objectives
13.2 Monte-Carlo Simulation
Problems on Monte-Carlo method
13.3 Applications of Simulation
Simulation in networks
Simulation in job sequencing
13.4 Advantages of Simulation
13.5 Limitations of Simulation
13.6 Summary
13.7 Glossary
13.8 Terminal Questions
13.9 Answers
13.10 Case Study
13.1 Introduction
In the previous unit, you learnt the basic concepts and methodology of
simulation. You also learnt simulation procedure and sample size. Monte-
Carlo technique has become such an important part of simulation models
that the terms are often assumed to be synonymous. However, it is only a
special technique of simulation. The technique of Monte-Carlo involves the
selection of random observations within the simulation model.
The technique is restricted for application involving random numbers to
solve deterministic and stochastic problems. The principle of this technique
is replacement of actual statistical universe by another universe described
by some assumed probability distribution and then sampling from this
theoretical population by means of random numbers.
In fact, this process involves the generation of simulated statistics (random
variables) that can be explained in simple terms as choosing a random
number and substituting this value in it.
In this unit, you will learn the Monte-Carlo simulation method and the
application of simulation methods. You will also learn the advantages and
limitations of simulation.
Operations Research Unit 13
Sikkim Manipal University Page No. 253
Objectives:
After studying this unit, you should be able to:
- describe Monte-Carlo simulation method
- analyse the applications of simulation
- list the advantages and limitations of simulation
13.2 Monte-Carlo Simulation
The Monte-Carlo method is a simulation technique in which statistical
distribution functions are created by using a series of random numbers. This
approach has the ability to develop many months or years of data in a
matter of few minutes on a digital computer.
The method is generally used to solve the problems that cannot be
adequately represented by mathematical models or where solution of the
model is not possible by analytical method.
The Monte-Carlo simulation procedure can be summarised in the steps
depicted in figure 13.1:
Fig. 13.1: Monte-Carlo Simulation Procedure
Operations Research Unit 13
Sikkim Manipal University Page No. 254
Let us now describe each step in detail.
Step 1: Define the problem:
a) Identify the objectives of the problem.
b) Identify the main factors that have the greatest effect on the objectives
of the problem.
Step 2: Construct an appropriate model:
a) Specify the variables and parameters of the model.
b) Formulate the appropriate decision rules, i.e., state the conditions under
which the experiment is to be performed.
c) Identity the type of distribution that will be used. Models use either
theoretical distributions or empirical distributions to state the patterns of
occurrence associated with the variables.
d) Specify the manner in which time will change.
e) Define the relationship between the variables and parameters.
Step 3: Prepare the model for experimentation:
a) Define the starting conditions for the simulation.
b) Specify the number of runs of simulation to be made.
Step 4: Using steps 1 to 3, experiment with the model:
a) Define a coding system that will correlate the factors defined in step 1
with the random numbers to be generated for the simulation.
b) Select a random number generator and create the random numbers to
be used in the simulation.
c) Associate the generated random numbers with the factors identified in
step1 and coded in step 4(a).
Step 5: Summarise and examine the results obtained in step 4.
Step 6: Evaluate the results of the simulation.
Step 7: Formulate proposals for advice to management on the course of
action to be adopted and modify the model, if necessary.
Operations Research Unit 13
Sikkim Manipal University Page No. 255
13.2.1 Problems on Monte-Carlo method
Solved problem 1:
The occurrence of rain in a city on a day is dependent upon whether or not it
rained on the previous day. If it rained on the previous day, the rain
distribution is as depicted in table 13.1:
Event:
Table 13.1: Distribution Table
Event No
rain
1 cm
Rain
2 cm
Rain
3 cm
Rain
4 cm
Rain
5 cm
Rain
Probability 0.50 0.25 0.15 0.05 0.03 0.02
If it did not rain on the previous day, the rain distribution is as depicted in
table 13.2:
Table 13.2: Rain Distribution Table
Event No rain 1 cm
Rain
2 cm
Rain
3 cm
Rain
Probability 0.75 0.15 0.06 0.04
Simulate the city‘s weather for 10 days and determine the total days without
rain as well as the total rainfall during the periodby simulation. Use the
following random numbers:
67 63 39 55 29 78 70 06 78 76
For simulation assume that for the first day of the simulation it had not
rained the day before.
Solution: We simulate the city‘s weather with and without rainfall in the
following steps:
Step 1: Previous day rain distribution is as depicted in table 13.3:
Table 13.3: Rain Distribution Table
Event Probability Cumulative
probability
RN range
No rain 0.50 0 . 50 0 0 - 4 9
1 cm Rain 0.25 0 . 7 5 5 0 - 7 4
2 cm rain 0.15 0 . 9 0 7 5 - 8 9
3 cm Rain 0 . 0 5 0.95 90-94
4 cm Rain 0.03 0 . 9 8 9 5 - 9 7
5 cm Rain 0.02 1 . 0 0 9 8 - 9 9
Operations Research Unit 13
Sikkim Manipal University Page No. 256
Step 2: Previous day no rain distribution is as depicted in table 13.4:
Table 13.4: Distribution Table
Event Probability Cumulative probability R N r a n ge
No rain 0 . 7 5 0 .75 0 0 - 7 4
1 cm Rain 0 . 1 5 0 . 15 7 5 - 8 9
2 cm rain 0 . 0 6 0.96 90-95
3 cm Rain 0 . 0 4 1.00 96-99
Step 3: Simulation for 10 days using the given random numbers is as
depicted in table 13.5:
Table 13.5: Random Table
Day RN Events Cumulative rain
1 67 No rain
2 63 No rain
3 39 No rain
4 55 No rain
5 29 No rain
6 78 1 cm Rain 1 c m
7 70 1 cm Rain 2 c m
8 06 No rain 2 c m
9 78 1 cm Rain 3 c m
10 76 2 cm Rain 5 c m
During the simulated period it did not rain on 6 out of 10 days. The total
rainfall during the period is 5 cm.
Solved problem 2:
The number of customers at a restaurant each evening is distributed as
depicted in table 13.6:
Table 13.6: Distribution of Customers
Number of customers Lots Average Very few
Probability 0.2 0.4 0.4
The chef refuses to work on an evening when there are very few customers
and walks out. He will not return to work until an evening when there are lots
Operations Research Unit 13
Sikkim Manipal University Page No. 257
of customers although he always comes on Friday because he gets paid
then. We are interested in fraction of evenings that the chef is at the
restaurant.
Solution: In this example, the trial is whether the number of customers is
‗lots‘, ‗average‘ or ‗very few‘ and it is repeated for each evening when the
restaurant is open. Also, whether the chef is ‗in‘ or ‗out‘ depends not only on
the outcome of this trial but also on his whereabouts the previous evening
and whether it is a Friday or not.
Here we wish to simulate the level of attendance at the restaurant on each
evening. To determine whether he is ‗in‘ or ‗out‘ on that particular evening,
we use the rules governing the chef‘s behaviour as stated in the question.
Here the sequence of in‘s and out‘s forms a simulated history of the system
which helps us in determining the fraction of evenings for which the chef is
‗in‘. Now we can proceed as follows:
Step1: To simulate the number of customers each evening
Using the random digits and the given (prescribed) probability distribution,
we simulate the number of customers each evening as depicted in
table 13.7:
Table 13.7: Simulation of Customers
Number of customers Lots Average Very few
Probability 0.2 0.4 0.4
Digits 0.1 2,3,4,5 6,7,8,9
Step 2: To generate attendance level
The attendance on a particular evening can be simulated by selecting a
digit, from 0 to 9 inclusive, at random. Using table 13.7, we can interpret the
digits as state of the restaurant. On account of the number of digits assigned
to each one, the outcomes in the simulation will have the same probability of
occurrence as those in the real system.
Step 3: To determine whether the chef works or not
After generating an attendance level, we can determine whether the chef
works or not by the rules provided in the question. Table 13.8 depicts a
specimen run of the simulation.
Operations Research Unit 13
Sikkim Manipal University Page No. 258
The following specimen run was started with ‗Monday‘ and with the chef
being ‗in‘. Keeping the view in mind that we are trying to reproduce the long
term behaviour of the restaurant, the initial conditions of the simulation will
not affect the results we obtain.
Table 13.8: Distribution Table
Day Day of the
week
Random
Digit
Number of
customers
Chef
‘in’ or
‘out’
Number
of days
‘in’ to
date
Fraction
of days
‘in’ to
date
1
2
3
4
5
6
7
8
9
10
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Sunday
Monday
Tuesday
Wednesday
3
5
1
8
6
6
4
4
0
Average
Average
Lots
Very few
Very few
Very few
Average
Average
Lots
In
In
In
Out
In
Out
Out
Out
In
1
2
3
3
4
4
4
4
5
1.00
1.00
1.00
0.75
0.80
0.67
0.57
0.50
0.56
Step 4: When to stop the simulation run
As far as determination of any average effects is concerned, the more time
periods simulated the better. However, in actual practice we require a
criterion for deciding when the simulation should stop. In the specimen run
we could continue until the fraction at the end of each line does not change
very much. At this stage there is no point in going further and we have the
desired answer. In this example, it approaches 0.56 and the chef is ‗in‘ 56%
of all days.
Self Assessment Questions
1. Which of the following is not correct?
a) Randomness is a key requirement of Monte-Carlo simulation.
b) Monte-Carlo simulation involves modelling a deterministic system.
c) Simulation is a statistical experiment as such its results are subject
to statistical error.
d) In Monte-Carlo simulation, a problem is solved by simulating the
original data with random number generators.
Operations Research Unit 13
Sikkim Manipal University Page No. 259
2. The technique of Monte-Carlo involves the selection of ____________
observations within the ____________ model.
13.3 Application of Simulation
The range of application of simulation in business is extremely wide. Unlike
other mathematical models, simulation can be easily understood by the
users and thereby facilitates their active involvement. This makes the results
more reliable and also ensures easy acceptance for implementation. The
degree to which a simulation model can be made close to reality is
dependent upon the ingenuity of the OR team who identifies the relevant
variables as well as their behaviour.
You have already seen, by means of an example, how simulation could be
used in a queuing system. It can also be employed for a wide variety of
problems encountered in production systems–the policy for optimal
maintenance in terms of frequency of replacement of spares or preventive
maintenance, number of maintenance crews, number of equipment for
handling materials, job shop scheduling, routing problems, stock control,
etc. The other areas of application include dock facilities, facilities at airports
to minimise congestion, hospital appointment systems and even
management games.
In case of other OR models, simulation helps the manager to strike a
balance between opposing costs of providing facilities (usually meaning long
term commitment of funds) and the opportunity costs of not providing them.
Let us find some of the examples how simulation is used in different
situations.
13.3.1 Simulation in networks
A number of network simulation models have also been developed, e.g.,
simulation of probabilistic activity times in PERT network. The critical path
and the project duration can be found out with a randomly selected activity
times for each activity. The probability distribution of project completion time
and the probability that each of given activity is on the critical path can be
obtained by repeating the process a number of times.
Operations Research Unit 13
Sikkim Manipal University Page No. 260
Solved problem 3:
A project schedule consists of five activities and the duration of these
activities is non-deterministic with the probability distribution as depicted in
table 13.9:
Table 13.9: Probability Distribution Table
Activity Days Probability
1-2 1
4
8
0.2
0.5
0.3
1-3 2
4
7
0.3
0.5
0.2
2-4 2
4
6
0.3
0.3
0.4
3-4 3
6
8
0.3
0.4
0.3
4-5 2
3
4
0.2
0.2
0.6
Simulate the duration of the project 10 times and estimate the chances of
various paths to be critical. What is the average duration of the project?
Solution:
From the PERT network, we observe that there are two paths from the start
to end of the project, namely 1245 and 1345. Using one-digit
random number corresponding to the probability distribution of various
activities, we obtain the simulated duration of each path and hence the
simulated duration of various activities and the simulated duration of the
project. Table 13.10 depicts the results of the simulation:
Operations Research Unit 13
Sikkim Manipal University Page No. 261
Table 13.10: Simulation Table
Activity
Simul
ation
1-2
RN D
1-3
RN D
2.4
RN D
3-4
RN D
4-8
RN D
Project
Duration
Critical Paths
1 8 8 9 2 8 6 3 6 6 4 18 1245
2 5 4 5 4 8 6 1 3 7 4 14 1245
3 5 4 6 4 7 6 4 6 7 4 14 1245
1345
4 7 8 5 4 1 2 9 8 5 4 16 1345
5 0 1 6 4 8 6 1 3 8 4 11 1245
6 8 8 3 4 0 2 4 6 3 3 13 125
7 4 4 7 4 4 4 4 0 3 3 11 145
8 5 4 8 7 2 2 6 6 6 4 17 145
9 2 4 5 4 9 6 0 3 3 3 13 145
10 6 4 1 2 3 4 7 8 0 2 12 145
RN: Random Number; D: Duration
On the basis of the above simulation, we observe that there is a 70 percent
chance of the path 145 being critical. Further, the average duration of
the project is 13.9 days.
Solved problem 4:
A project consists of 7 activities. The time for performance of each of the
activities is a random variable with the respective probability distribution as
depicted in table 13.11:
Table 13.11: Probability Distribution Table
Activity Immediate
predecessor
Time (in days) and its
probability
A

3
0.20
4
0.60
5
0.20
B –
4
0.10
5
0.30
6
0.30
7
0.20
8
0.10
C A 1
0.15
3
0.75
5
0.10
D B,C 4
0.80
5
0.20
E D 3
0.10
4
0.30
5
0.30
6
0.30
F D 5
0.20
7
0.80
G E,F 2
0.50
3
0.50
Operations Research Unit 13
Sikkim Manipal University Page No. 262
A) Draw the network diagram and identify the critical path using the
expected activity times.
B) Simulate the project using random numbers to determine the activity
times. Find the critical path and the project duration.
C) Repeat the simulation four more times. Is the same path critical in all the
simulations?
Solution:
a) using the given information, the resulting network is depicted in figure
13.2. The time for each of the activities is the expected time obtained by the
summation of the products of the activity time and the corresponding
probability. For example, for activity A,
The Expected Time = 3*0.20 + 4*0.60 + 5*0.20 = 4 days.
Fig. 13.2: PERT Network
Network design
Observe that the critical path of the project is 123467, with an
expected duration of 20.2 days.
To use the simulation for obtaining the project length, we first determine the
random number intervals for each activity. Random number coding is shown
in the table. The five simulation runs, by taking random numbers, are as
depicted in table 13.12. For each run, the critical path and the project
durations are obtained and the project time is obtained as follows:
Total time = Larger of times for activities A and C, and B + Time for activity
D + Larger of times for activities E + Time for activity G
Operations Research Unit 13
Sikkim Manipal University Page No. 263
Table 13.12: Random Number Coding
Activity Time Probability
Cumulative
Probability
Random
number
Interval
A 3
4
5
0.20
0.60
0.20
0.20
0.80
1.00
00-19
20-79
80-99
B 4
5
6
7
8
0.10
0.30
0.30
0.20
0.10
0.10
0.40
0.70
0.90
1.00
00.09
10-39
40-69
70-89
90-99
C 1
3
5
0.15
0.75
0.10
0.15
0.90
1.00
00-14
15-89
90-99
D 4
5
3
0.80
0.20
0.10
0.80
1.00
0.10
00-79
80-99
00-09
E 4
5
6
0.30
0.30
0.30
0.10
0.70
1.00
10-39
40-69
70-99
F 5
7
0.20
0.80
0.20
1.00
00-19
20-99
G 2
3
0.50
0.50
0.50
1.00
00-49
50-99
Simulation worksheet
Activity
Table 13.13 depicts the simulation worksheet.
Table 13.13: Simulation Worksheet
Simulation A
RN D
B
RN D
C
RN D
D
RN D
E
RN D
F
RN D
G
RN D
1 68 4 13 5 09 1 20 4 73 6 07 5 92 3
2 99 5 93 8 18 3 20 4 22 4 07 5 29 2
3 57 4 33 5 49 3 65 4 92 6 98 7 00 2
4 57 4 12 5 31 3 96 5 86 6 92 7 91 3
5 77 4 37 5 34 3 11 4 27 4 10 5 59 3
RN: random number; D: Duration
Operations Research Unit 13
Sikkim Manipal University Page No. 264
We get the information as depicted in figure 13.3 from the above table:
Fig. 13.3: Simulation of Customers
It may be observed that the same path is not critical in each of the runs.
13.3.2 Simulation in job sequencing
In this section, we illustrate the use of simulation in the sequencing
problems.
Example:
A job has to be processed over two machines, machine M1 and machine
M2 in that order. The distribution of inter-arrival time of the jobs at the first
machine is as depicted in table 13.14:
Table 13.14: Distribution of Arrival Time
Time(minutes) 1 2 3 4
Probability 0.2 0.2 0.2 0.4
Operations Research Unit 13
Sikkim Manipal University Page No. 265
The processing times at the two machines are as follows:
Table 13.15: Simulation Table
Machine M1 Machine M2
Time
(minutes)
Probability Time(minutes) Probability
1 0.1 4 0.2
2 0.2 5 0.3
3 0.3 6 0.4
4 0.3 7 0.1
5 0.1
On the basis of 10 simulation runs, find out the average queue length before
machine M1 and the average queue length before machine M2.
Solution:
The various steps for simulation are:
1. Simulate the inter-arrival time of the ith job on machine M1 with the help
of a random number (i= 1, 2…….).
2. Arrival time of ith job.
= Arrival time of (i-1)th job (i=2,3, ….) + Inter-arrival time of ith job.
Arrival time of the first job= Inter-arrival time of the first job
3. Simulate the processing time on machine M1 by random numbers.
4. Departure time on machine M1
= Arrival time for processing on machine M2
= Max (Arrival time of ith job, process completion time of (i-1)thjob
on machine M1) + processing time of ith job on machine M1
5. If arrival time of ith job < Process completion time of (i-1)th, ith arrival
waits.
If arrival time of ith job < Process completion time of (i-2)th, both ith and (i-
1)th arrivals wait.
6. Simulate the processing time of machine M2 with the help of random
numbers.
7. Process completion by machine M2 of ith arrival
= Max [Arrival time of ith job, process completion time of (i-1)th job]
+ Process time by machine M2.
Operations Research Unit 13
Sikkim Manipal University Page No. 266
8. If arrival time of ith< process completion time of (i-1)th, the ith arrival waits.
If arrival time of ith< process completion time of (i-2)th , both ith and (i-1)th
arrivals wait, etc.
Using the procedure, the simulated processing data of 10 arrivals are
displayed in the simulation worksheet as depicted in figure 13.4:
Fig. 13.4: Simulation of 10 Arrivals
On the basis of the simulation study, the average queue length before
machine M1 is 0.5 and the average length before machine M2 is 1.2.
Self Assessment Questions
3. The ___________ and the project duration can be found out with a
randomly selected activity times for each activity.
4. The range of application of simulation in _________ is extremely wide.
13.4 Advantages of Simulation
Simulation is the process of experimenting on the model rather than on the
operation which the model represents. Following are some of the
advantages of simulation:
Operations Research Unit 13
Sikkim Manipal University Page No. 267
- The study of very complicated systems or sub-systems can be done with
the help of simulation. Simulation has been described as ‗What to do
when all else fails.‘
- We can investigate the consequences for a system of possible changes
in parameters in terms of the model.
- The knowledge of a system obtained in designing and conducting the
simulation is very valuable.
- Simulation enables us to assess the possible risks involved in a new
policy before actually implementing it.
- Simulation of complicated systems helps us to locate which variables
have the important influences on system performance.
- Simulation methods are easier to apply than pure analytical methods.
13.5 Limitations of Simulation
Notwithstanding with the above advantages, following are some of the
limitations of simulation:
- Simulation generates a way of evaluating solutions but it does not
generate the solution techniques.
- Sometimes, simulation models are expensive and take a long time to
develop. For example, a corporate planning model may take a long time
to develop and prove expensive also.
- The simulation model does not produce answers by itself. The user has
to provide all the constraints for the solutions which he/she wants to
examine.
- Not all situations can be evaluated using simulation. Only situations
involving uncertainty are considered.
- It is the trial-and-error approach that produces different solutions in
repeated runs. This means it does not generate optimal solutions to
problems.
- Simulation is a time-consuming exercise.
Self Assessment Questions
5. An advantage of simulation as opposed to optimisation is that:
a) Several options of measure of performance can be examined.
b) Complex reallife problems can be studied.
Operations Research Unit 13
Sikkim Manipal University Page No. 268
c) It is applicable in cases where there is an element of randomness
in a system.
d) All the above.
6. Large complicated simulation models are appreciated because:
a) Their average costs are not well-defined.
b) It is difficult to create the appropriate events.
c) They may be expensive to write and use as an experimental
device.
d) All of the above.
7. Simulation should not be applied in all the cases because it:
a) Requires considerable talent for model building and extensive
computer programming efforts.
b) Consumes much computer time.
c) Provides at best approximate solution to problem.
d) All of the above.
8. Analytical results are taken into consideration before a simulation study
so as to:
a) Identify suitable values of the system parameters.
b) Determine the optimal decision.
c) Identify suitable values of decision variables for the specific
choices of system parameters.
d) All of the above.
13.6 Summary
Let us recapitulate the important concepts discussed in this unit:
- The Monte-Carlo method is a simulation technique in which statistical
distribution functions are created by using a series of random numbers.
- Simulation could be used many systems such as in a queuing,
production, dock facilities, facilities at airports to minimise congestion,
hospital appointment systems and even management games.
- A number of network simulation models have also been developed, such
as simulation of probabilistic activity times in PERT network.
- Simulation enables us to assess the possible risks involved in a new
policy before actually implementing it.
Operations Research Unit 13
Sikkim Manipal University Page No. 269
- Not all situations can be evaluated using simulation. Only situations
involving uncertainty are considered.
13.7 Glossary
Job sequencing: the arrangement of the tasks required to be carried out.
job shop: Organization of work in a manufacturing organization by job
functions.
13.8 Terminal Questions
1. Explain the use of simulation in networks?
2. What are the advantages of using simulation?
3. What are the Limitations of using Simulation?
4. Write a short note on Monte-Carlo simulation?
5. For a project comprising activities A, B …..… H, the following information
is available:
Precedence relationships:
A and B are the first activities of the project. C succeeds A while B
precedes D. Both C and D precede E and F. Activity G follows activity F
while H is the last activity of the project and succeeds E and G. Time
estimates and Probabilities:
Simulate the project and determine, using random numbers, the activity
times. Find the critical path and the project duration. Repeat five times.
Does the critical path change? State the estimated duration of the project in
each of the trials.
Activity Time days
1 2 3 4 5 6
A 0.02 0.04 0.4
B 0.3 0.7
C 0.3 0.3 0.4
D 0.2 0.6 0.2
E 0.6 0.4
F 0.8 0.2
G 0.3 0.5 0.2
H 0.1 0.2 0.3 0.4
Operations Research Unit 13
Sikkim Manipal University Page No. 270
13.9 Answers
Self Assessment Questions
1. (b) Monte-Carlo simulation involves modelling a deterministic system.
2. Random, Simulation.
3. Critical path
4. Business
5. (d) All of the above.
6. (c) They may be expensive to write and use as an experimental device.
7. (d) All of the above.
8. (c) Identify suitable values of decision variables for the specific choices
of system parameters.
Terminal Questions
1. To find the critical path and project duration. For more details refer
13.3.1
2. Advantages of simulation: to studying complicated systems, investigate
the consequences for a system and assess the possible risks involved.
For more details refer 13.4
3. Limitations of simulation: does not generate the solution techniques,
simulation models are expensive and does not produce answers by
itself. For more details refer 13.5
4. The Monte-Carlo method is a simulation technique in which statistical
distribution functions are created by using a series of random numbers.
This approach has the ability to develop many months or years of data
in a matter of few minutes on a digital computer. For more details
refer 13.2
5. Using random numbers from the first six columns for six trials:
Trial Critical paths Duration
1 BDFGH 16
2 BDFGH 19
3 BDFGH 18
4 BDFGH 17
5 ACFGH
BDFGH 16
6 ACFGH and
BDFGH 18
Operations Research Unit 13
Sikkim Manipal University Page No. 271
13.10 Case Study
Workload problems in planer department
This was the first time that Carl Schilling had been summoned for a meeting
with the management in the fancy executive offices upstairs and he hopes it
will be the last time. Carl doesn‘t like the pressure. He has had enough
pressure just dealing with all the problems he has been encountering as the
foreman of the planer department on the factory floor. What a nightmare this
last month has been! Fortunately, the meeting had gone better than Carl
had feared. The management actually had been quite nice. They explained
that they needed to get Carl‘s advice on how to deal with a problem that was
affecting the entire factory. The origin of the problem is that the planer
department has had a difficult time keeping up with its workload. Frequently
there are a number of work pieces waiting for a free planer. This waiting has
seriously disrupted the production schedule for subsequent operations,
thereby greatly increasing the cost of in-process inventory as well as the
cost of idle equipment and resulting lost production. They understood that
this problem was not Carl‘s fault. However, they needed to get his ideas on
what changes were needed in the planer department to relieve this
bottleneck. Imagine this! The management, with graduate degrees from the
fanciest business schools in the country, asking advice from a poor working
slob like him who had barely made it through high school. He could hardly
wait to tell his wife that night. The meeting had given Carl an opportunity to
get two petpeeves off his chest. One peeve is that he has been telling his
boss for months that he really needs another planer, but nothing ever gets
done about this. His boss just keeps telling him that the planers he already
has aren‘t being used 100 percent of the time, so how can adding even
more capacity be justified. Doesn‘t his boss understand about the big
backlogs that build up during busy times? Then there is the other peeve – all
those peaks and valleys of work coming to his department. At times, the
work just pours in, and a big backlog builds up. Then there might be a long
pause when not much comes in, so the planers stand idle, part of the time. If
only those departments that are feeding castings to his department could
get their act together and even out the work flow, many of his backlog
problems would disappear. Carl was pleased that the bigwigs were nodding
their heads in seeming agreement as he described these problems.
Operations Research Unit 13
Sikkim Manipal University Page No. 272
They really appeared to understand. And they seemed very sincere in
thanking him for his good advice. Maybe something is actually going to get
done this time. Here are the details of the situation that Carl and his
―management‖ are addressing. The company has two planers for cutting flat
smooth surfaces in large castings. The planers currently are being used for
two purposes. One is to form the top surface of the platen for large hydraulic
lifts. The other is to form the mating surface of the final drive housing for a
large piece of earth-moving equipment. The time required by a planer to
perform each job varies somewhat, depending largely upon the number of
passes that must be made. In particular, for each platen or housing, the time
required has a translated exponential distribution, where the minimum time
is 10 minutes and the additional time beyond 10 minutes has an exponential
distribution with a mean of 10 minutes. Castings of both types arrive one at
a time to the planer department. For each type, the arrivals occur randomly
with a mean rate of 2 per hour. Based on Carl Schilling‘s advice, the
management has asked an OR analyst (you) to analyse the following three
proposals for relieving the bottleneck in the planer department:
Proposal 1: Obtain one additional planer. The total incremental cost
(including capital recovery cost) is estimated to be $30 per hour. (This
estimate takes into account the fact that, even with an additional planer, the
total running time for all the planers will remain the same.)
Proposal 2: Eliminate the variability in the inter-arrival times of the castings,
so that the castings would arrive regularly, one every 15 minutes, alternating
between platen castings and housing castings. This would require making
some changes in the preceding production processes, with an incremental
cost of $60 per hour.
Proposal 3: Make a change in the production process that would reduce the
variability in the time required by a planer to perform each job. In particular,
for either type of casting, the time required now would have an Erlang
distribution with a mean of 20 minutes and shape parameter k _ 10. The
incremental cost in this case would be $20 per hour.
These proposals are not mutually exclusive, so any combination can be
adopted. It is estimated that the total cost associated with castings having to
wait to be processed (including processing time) is $200 per hour for each
platen casting and $100 per hour for each housing casting, provided the
Operations Research Unit 13
Sikkim Manipal University Page No. 273
waits are not excessive. To avoid excessive waits for either kind of casting,
all the castings are processed as soon as possible on a first come, firstserve
basis.
Discussion Question:
Management‘s objective is to minimise the expected total cost per hour. Use
simulation to evaluate and compare all the alternatives, including the status
quo and the various combinations of proposals. Then make your
recommendation to the management.
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
Operations Research Unit 14
Sikkim Manipal University Page No. 274
Unit 14 Project Scheduling and PERT-CPM
Structure:
14.1 Introduction
Objectives
14.2 Basic Difference between PERT and CPM
PERT
CPM
Project scheduling by PERT-CPM
14.3 PERT/CPM Network Components and Precedence Relationship
Critical path calculations
Determination of the critical path
Determination of floats
14.4 Project Management – PERT
14.5 Summary
14.6 Glossary
14.7 Terminal Questions
14.8 Answers
14.9 Case Study
14.1 Introduction
In the previous unit, simulation Monte-Carlo method, we studied about the
Monte-Carlo simulation. We also learnt about the various problems of this
method, the applications of the simulation, the advantages and the
limitations of the simulation. In this unit, project scheduling and PERT-CPM
in operations research management, we will study about the basic
difference between PERT and CPM. We will also learn about PERT and
CPM network components, determination of floats and project management.
Project management has evolved as a new field with the development of
two analytic techniques for planning, scheduling and controlling projects.
These are the project evaluation and review technique (PERT) and the
critical path method (CPM). PERT and CPM are basically time-oriented
methods in the sense that they both lead to the determination of a time
schedule.
Projects may be defined as a collection of interrelated activities (or tasks)
which must be completed in a specified time according to a specified
Operations Research Unit 14
Sikkim Manipal University Page No. 275
sequence and require resources, such as personnel, money, materials,
facilities, etc. Few examples of such projects are construction of a bridge, a
highway, a power plant, repair and maintenance of an oil refinery or
designing of an air plane, development and marketing of a new product, and
research and development.
The growing complexities of today‘s projects have demanded more
systematic and more effective planning techniques with the objective of
optimising the efficiency of executing the project. Efficiency here refers to
the utmost reduction in the time required to complete a project while
ensuring optimum utilisation of the available resources.
Objectives:
After studying this unit, you should be able to:
- define a project
- distinguish between PERT and CPM
- apply PERT-CPM method to network analysis
- describe project management
14.2 Basic Difference between PERT and CPM
There are no essential differences between PERT and CPM as both of them
share in common the determination of a critical path. Both are based on the
network representation of activities and their scheduling, which determines
the most critical activities to be controlled in order to meet the completion
date of the project.
14.2.1 PERT
Some key points of PERT are as follows:
- PERT was developed in connection with an Research and Development
(R&D) work. Therefore, it had to cope with the uncertainties that are
associated with R&D activities. In PERT, the total project duration is
regarded as a random variable. Therefore, associated probabilities are
calculated in order to characterise it.
- It is an event-oriented network as in the analysis of a network, emphasis
is given on the important stages of completion of a task rather than the
activities required to be performed to reach a particular event or task.
- PERT is normally used for projects involving activities of non-repetitive
nature in which time estimates are uncertain.
Operations Research Unit 14
Sikkim Manipal University Page No. 276
- It helps in pinpointing critical areas in a project, so that necessary
adjustment can be made to meet the scheduled completion date of the
project.
14.2.2 CPM
- CPM was developed in connection with a construction project, which
consisted of routine tasks whose resource requirements and duration
were known with certainty. Therefore, it is basically deterministic.
- CPM is suitable for establishing a trade-off for optimum balancing
between schedule time and cost of the project.
- CPM is used for projects involving activities of repetitive nature.
14.2.3 Project scheduling by PERT-CPM
Project scheduling consists of three basic phases: planning, scheduling and
controlling. Figure 14.1 depicts the three basic phases of PERT-CPM.
Fig. 14.1: Phases of PERT-CPM
- Project planning – Figure 14.2 depicts the activities to be performed in
the project planning phase.
Fig. 14.2: Activities in Project Planning Phase
Operations Research Unit 14
Sikkim Manipal University Page No. 277
- Identify various tasks or work elements to be performed in the project.
- Determine requirement of resources, such as men, materials, and
machines, for carrying out the activities listed.
- Estimate costs and time for various activities.
- Specify the inter-relationship among various activities.
- Develop a network diagram showing the sequential inter-relationships
between the various activities.
- Project scheduling - Once the planning phase is over, scheduling of
the project is when each of the activities required to be performed, is
taken up for execution. The various steps involved during this phase are
given as follows:
1. Estimate the durations of activities. Take into account the resources
required for these executions in the most economic manner.
2. Based on the time estimates, prepare a time chart showing the start
and finish times for each activity. Use the time chart for the following
exercises:
- To calculate the total project duration by applying network
analysis techniques, such as forward (backward) pass and floats
calculation
- To identify the critical path
- To carry out resource smoothing (or levelling) exercises for
critical or scarce resources, including re-costing of the schedule
taking into account resource constraints
- Project controlling – Project controlling refers to comparing the actual
progress against the estimated schedule. If significant differences are
observed then we need to re-schedule the project to update or revise
the uncompleted part of the project.
Self Assessment Questions
1. Project consists of interrelated activities. (True/False)
2. Project activities are to be completed in a specified time according to a
specified sequence. (True/False)
3. PERT and CPM identifies non-critical activities. (True/False)
4. PERT is an activity-oriented network. (True/False)
5. CPM is used for projects that are repetitive in nature.(True/False)
Operations Research Unit 14
Sikkim Manipal University Page No. 278
14.3 PERT/CPM Network Components and Precedence
Relationship
PERT/CPM networks consist of two major components which are as follows:
- Events – An event represents a point in time that signifies the
completion of some activities and the beginning of new ones. The
beginning and end points of an activity are thus described by two events
usually known as the tail and the head events. Events are commonly
represented by circles (nodes) in the network diagram. They do not
consume time and resource.
- Activities – Activities of the network represent project operations or
tasks to be conducted. An arrow is commonly used to represent an
activity, with its head indicating the direction of progress in the project.
Activities originating from a certain event cannot start until the activities
terminating at the same event have been completed. They consume
time and resource.
Events in the network diagram are identified by numbers. Numbers are
given to events such that the arrow head number is greater than the arrow
tail number. Activities are identified by the numbers of their starting (tail)
event and ending (head) event.
Figure 14.3 depicts the network where the arrow (P.Q) extended between
two events represents the activity. The tail event P represents the start of
the activity and the head event Q represents the completion of the activity.
Fig. 14.3: Basic PERT-CPM Network
Figure 14.4 depicts an example of another PERT-CPM network with
activities (1, 3), (2, 3) and (3, 4). As the figure indicates, activities (1, 3) and
(2, 3) need to be completed before activity (3, 4) starts.
Operations Research Unit 14
Sikkim Manipal University Page No. 279
Fig. 14.4: A PERT-CPM Network
The rules for constructing the arrow diagram are as follows:
1. Each activity is represented by one and only one arrow in the network.
2. No two activities can be identified by the same head and tail events.
3. To ensure the correct precedence relationship in the arrow diagram, we
need to answer the following questions as we add every activity to the
network:
- What activities must be completed immediately before these
activities can start?
- What activities must follow this activity?
- What activity must occur concurrently with this activity?
This rule is self-explanatory. It actually allows for checking (and rechecking)
the precedence relationships as one progresses in the development of the
network.
Solved problem 1
Construct the arrow diagram comprising activities A, B, C …….. and L
such that the following relationships are satisfied:
1) A, B and C the first activities of the project, can start simultaneously.
2) A and B precede D.
3) B precedes E, F and H.
4) F and C precede G.
5) E and H precede I and J.
6) C, D, F and J precede K.
7) K precedes L.
8) I, G and L are the terminal activities of the project.
Operations Research Unit 14
Sikkim Manipal University Page No. 280
Solution
Figure 14.5 depicts an analysis network.
Fig. 14.5: Analysis Network
The dummy activities D1 and D2 are used (dotted lines) to establish
correct precedence relationships. D3 is used to identify activities E and H
with unique end events. The events of the project are numbered such
that their ascending order indicates the direction of the progress in the
project.
Note: A dummy activity in a project network analysis has zero duration.
14.3.1 Critical path calculations
The application of PERT/CPM should ultimately yield a schedule specifying
the start and completion time of each activity. The arrow diagram is the first
step towards achieving that goal. The start and completion timings are
calculated directly on the arrow diagrams using simple arithmetic. The end
result is to classify the activities as critical or non-critical.
An activity is said to be critical if a delay in the start of the course makes a
delay in the completion time of the entire project.
A non-critical activity is such that the time between its earliest start and its
latest completion time is longer than its actual duration. A non-critical activity
is said to have a slack or float time.
14.3.2 Determination of the critical path
A critical path defines a chain of critical activities that connects the start and
end events of the arrow diagram. In other words, the critical path identifies
all the critical activities of a project.
Operations Research Unit 14
Sikkim Manipal University Page No. 281
The critical path calculations are done in two phases.
The first phase is called the forward pass. In this phase, all calculations
begin from the start node and move to the end node. At each node a
number is computed representing the earliest occurrence time of the
corresponding event. These numbers are shown in squares. Here we note
the number of heads joining the event. We take the maximum earliest timing
through these heads.
The second phase is called the backward pass. It begins calculations from
the ―end‖ node and moves to the ―start‖ node. The number computed at
each node is shown in a triangle A near the end point, which represents the
latest occurrence time of the corresponding event. In backward pass, we
see the number of tails and take the minimum value through these tails.
Let ESi be the earliest start time of all the activities emanating from event i.
Then ESi represents the earliest occurrence time of event i.
If i = 1 is the ―start‖ event then conventionally for the critical path
calculations, ESi = 0.
Let Dij be the duration of the activity (i, j).
Then the forward pass calculations for all defined (i, j) activities with ESi=0 is
given by the formula:
ESi = maxi {ESi+Dij}
Therefore, to compute ESj for event j, we need to first compute ESi for the
tail events of all the incoming activities (i, j).
With the computation of all ESj, the forward pass calculations are
completed. The backward pass starts from the ―end‖ event. The objective of
the backward pass phase is to calculate LCi, the latest completion time for
all the activities coming into the event i.
Thus, if i = n is the end event, LCn = ESn initiates the backward pass.
In general for any node i, we can calculate the backward pass for all defined
activities using the formula:
LCi = min {LCj-Dij}
Operations Research Unit 14
Sikkim Manipal University Page No. 282
We can now identify the critical path activities using the results of the
forward and backward passes. An activity (i, j) lies on the critical path if it
satisfies the following conditions:
- ESi = LCi
- ESj = LCj
- ESj-ESi = LCj-LCi = Dij
These conditions actually indicate that there is no float or slack time
between the earliest start and the latest start of the activity. Thus, the
activity must be critical.
In the arrow diagram these are characterised by same numbers within
rectangles and triangles at each of the head and tail events. The difference
between the numbers in rectangles or triangles at the head event, and the
number within rectangles or triangles at the tail event is equal to the duration
of the activity. Thus, we will get a critical path, which is a chain of connected
activities, spanning the network from start to end.
Solved problem 2
Consider a network which stands from node 1 and terminates at node 6,
the time required to perform each activity is indicated on the arrows.
Figure 14.6 depicts an analysis network.
Fig. 14.6: Analysis Network
Solution
Let us start with the forward pass with ESi = 0.
Since there is only one incoming activity (1, 2) to event 2 with D12 = 3.
ES2 = ES1+ DS2 = 0+3=3.
Operations Research Unit 14
Sikkim Manipal University Page No. 283
Let us consider the end 3, since there is only one incoming activity (2, 3)
to event 3, with D23 = 3.
ES3 = ES2+ D23 = 3+3 = 6.
To obtain ES4, since there are two activities A (3, 4) and (2, 4) to the
event 4 with D24 = 2 and D34 = 0.
ES4= maxi=2, 3 {ESi + De4}
= max {ES2 +D24, ES3 + D34}
= max {3+2, 6+0} = 6
Similarly, ES5 = 13 and ES6 = 19.
This completes the forward pass.
In the backward pass we have
LC6= 19 = ES6
LC5 = 19-6 = 13
LC4= min J = 5, 6 {LCj – D4j} = 6
LC3 = 6, LC2 = 3 and LC1 = 0
Therefore, activities (1, 2), (2, 3) (3, 4) (4, 5) (5, 6) are critical and (2, 4)
(4, 6), (3, 6), are non-critical.
Thus, the activities (1, 2), (2, 3), (3, 4), (4, 5) and (5, 6) define the critical
path which is the shortest possible time to complete the project i.e., 19
days.
14.3.3 Determination of floats
Once the determination of the critical path is done, we have to compute the
floats for the non-critical activities. For the critical activities this float is zero.
Before showing how floats are determined, it is necessary to define two new
times that are associated with each activity. These are as follows:
- Latest start (LS) time and
- Earliest completion (EC) time
We can define activity (i, j) for these two types of time by:
LSij= LCj – Dij
ECij = ESi + Dij
There are two important types of floats namely:
- Total float (TF)
- Free float (FF)
Operations Research Unit 14
Sikkim Manipal University Page No. 284
The total float TFij for activity (i, j) is the difference between the maximum
time available to perform the activity (= LCj – ESi) and its duration (= Dij)
TFij = LCj – ESi– Dij = LCj – ECij = LSij – ESi
The free float is defined by assuming that all the activities start as early as
possible. In this case FFij for activity (i, j) is the excess of available time
(= ESi – ESi) over its deviation (= Dij); i.e.
FFij = ESi – ESi - Dij
Note: For critical activities float is zero. Therefore, the free float must be
zero when the total float is zero. However, the converse is not true, that is, a
non-critical activity may have zero free floats.
Let us consider the example taken before the critical path calculations. The
floats for the non-critical activities can be summarised as depicted in the
table 14.1.
Table 14.1 Floats for Non-critical Activities
Activity
(i j)
Duration
Dij
Earliest Latest Total
Float
TFij
Free
Float
FFij
Start
ESi
Completion
ECij
Start
LSij
Completion
LCj
(1, 2)
(2, 3)
(2, 4)
(3, 4)
(3, 5)
(3, 6)
(4, 5)
(4, 6)
(5, 6)
3
3
2
0
3
2
7
5
6
0
3
3
6
6
6
6
6
13
3
6
5
6
9
8
13
11
19
0
3
4
6
10
17
6
14
13
3
6
6
6
13
19
13
19
19
0*
0*
1
0*
4
11
0*
8
0*
0
0
1
0
4
11
0
8
0
Note: Total float = ESij = LFij - ESij
Free float = Total float - - Head slack
* Critical activity *
Operations Research Unit 14
Sikkim Manipal University Page No. 285
Solved problem 3
A project consists of a series of tasks A, B, C, – D, – E, F, G, H, I with the
following relationships:
- W < X,Y means X and Y cannot start until W is completed
- X,Y < W means W cannot start until both X and Y are completed
With this notation construct the network diagram having the following
constraints A < D, E; B, D < F; C < G, B < H; F, G < I.
Also find the minimum time of completion of the project, the critical path,
and the total floats of each task, when the time (in days) of completion of
each task is as follows:
Task: A B C D E F G H I
Time: 23 8 20 16 24 18 19 4 10
Solution
Figure 14.7 depicts an analysis network.
Fig. 14.7: Analysis Network
ES1 = 0, ES2 = 20, ES3 = 23, ES4 = 59, ES5 = 39, ES6 = 57, ES7 = 67
Table 14.2 depicts an activity table.
Table 14.2 Activity Table
Activity
(i, j)
Duration
Dij
Earliest Latest Table
Float
TFij
Free
Float
FFij
Start
ESe
Finish
Eeij
Start
Lj-Dij
Finish
Lj
(1, 2)
(1, 3)
(1, 4)
(2, 5)
(3, 4)
20
23
8
19
16
0
0
0
20
23
20
23
8
39
39
18
0
31
38
23
38
23
39
57
39
18
0*
31
18
0*
0
0
31
0
0
Operations Research Unit 14
Sikkim Manipal University Page No. 286
(3, 7)
(4, 5)
(4, 6)
(5, 6)
(5, 7)
(6, 7)
24
0
18
0
4
10
23
39
39
39
39
37
47
39
57
39
43
67
43
57
39
57
63
57
67
57
57
57
67
67
20
10
0*
18
24
0*
20
0
0
18
24
0
Critical path is 1 – 3 – 4 – 6 – 7.
Self Assessment Questions
6. Events do not consume ________ and _________.
7. Arrow‘s head number is _________ than its tail number.
8. Dummy activity is introduced in a network to keep proper _________
relationship.
9. Critical path calculation includes both _________ and _________.
14.4 Project Management – PERT
Probability and cost consideration in project scheduling
The analysis in CPM does not take into account the cases where time
estimates for the different activities are probabilistic. It also does not
consider explicitly the cost of schedules. Here we will consider both
probability and cost aspects in project scheduling.
Probability considerations are incorporated in project scheduling by
assuming that the time estimate for each activity is based on three time
estimates. They are as follows:
a = the optimistic time, which will be required if the execution of the project
goes extremely well.
b = the pessimistic time, which will be required if everything goes bad.
m = the most likely time, which will be required if execution is normal.
The most likely estimate m need not coincide with the mid-point
2
a + b
of a
and b.
Then the expected duration of each activity D can be obtained as the mean
of
2
a + b
and 2 m.
Operations Research Unit 14
Sikkim Manipal University Page No. 287
Therefore,
6
a b 4m
3
2m
2
a b
D
+ +
=
+
+
=
We can use this estimate to study the single estimate D in the critical path
calculation. These formulae are derived from beta distribution.
The variance of each activity denoted by V is defined by,
Variance V =
2
6
b a
|.
|
\
| ÷
The earliest expected times for the node i is denoted by E (µi). For each
node i, E(µi) is obtained by taking the sum of the expected times of all
activities leading to the node i, when more than one activity leads to a node
i, then the greatest of all E(µi) is chosen. Let µi be the earliest occurrence
time of the event i, we can consider µi as a random variable. Assuming that
all activities of the network are statistically independent, we can calculate
the mean and the variance of µi as follows:
E {µi } = ESi and Var{µi } = ¿
k
Vk
Where, k defines the activities along the largest path leading to i.
For the latest expected time, we consider the last node. Now for each path
move backwards and substitute the D iJ for each activity (i, j).
Thus we have,
E (Lj) = E (µa)
E (µi) = L (Lj) – Dij
if only one path events from j to i or if it is the minimum of {E [Lj) – Dij] for all j
for which the activities (i, j) are defined.
Note: The probability distribution of times for completing an event can be
approximated by the normal distribution due to central limit theorem.
Since µi represents the earliest occurrence time, event will meet a certain
schedule time STi (specified by an analyst) with probability
Operations Research Unit 14
Sikkim Manipal University Page No. 288
Pr (µi s STi) = Pr
|
|
.
|


\
|
µ
÷ µ
s
µ
µ ÷ µ
V( )
ST E( )
V( )
E( )
i
i i
i
i i
= Pr (Z s Ki)
Where, Z ~N (01) and Ki =
V( )
ST E( )
i
i i
µ
÷ µ
It is a common practice to compute the probability that event i will occur no
later than its LCe. Such probability will represent the chance that the
succeeding events will occur within the (ESe, LCe) duration.
Solved problem 4
A project is represented by the network depicted in figure 14.8 and has
the following data:
Fig. 14.8: Analysis Network
Table 14.3 depicts a data table.
Table 14.3 Data Table
Task A B C D E F G H I
Optimistic time 5 18 26 16 15 6 7 7 3
Pessimistic time 10 22 40 20 25 12 12 9 5
Most likely time 8 20 33 18 20 9 10 8 4
Determine the following:
1. Expected task times and their variances.
2. The earliest and latest expected times to reach each event
3. The critical path
4. The probability of an event occurring at the proposed completion
data if the original contract time of completing the project is 41.5
weeks.
5. The duration of the project that will have 97% channel of being
completed.
Operations Research Unit 14
Sikkim Manipal University Page No. 289
Solution
Using the formula we can calculate expected activity times and variance
as depicted in table 14.4.
2
6
b a
(a b 4m) V
6
1
D |
.
|
\
| ÷
= + + =
Table 14.4 Earliest and Latest Expected Time for Events
Activity A B m v
1-2
1-3
1-4
2-5
2-6
3-6
4-7
5-7
6-7
5
18
26
16
15
6
7
7
3
10
22
40
20
25
12
12
9
5
8
20
33
18
20
9
10
8
4
7-8
20-00
33-0
18-0
20-0
9-0
9-8
8-0
4-0
0.696
0.444
5.429
0.443
2.780
1.000
0.694
0.111
0.111
Forward pass
E1 = 0 E2 = 7.8 E3 = 20 E4 = 33 E5 = 25.8 E6 = 29
E7 = 42.8
Backward pass
L7 = 42.8 L6 = 38.8 L5 = 34.8 L4 = 33.0 L3 = 29.8 L2 = 16.8
L1 = 0.
The E-values and L-values are depicted in figure 14.9.
Fig. 14.9: Analysis Network
- The critical path is shown by a thick line in the figure. The critical
path is 1-4-7 and the earliest completion time for the project is 42.8
weeks.
- The last event 7 will occur only after 42.8 weeks. For this we require
only the duration of critical activities. This will help us in calculating
the standard duration of the last event.
Operations Research Unit 14
Sikkim Manipal University Page No. 290
Expected length of critical path = 33 + 9.8 = 42.8
Variance of article path length = 5.429 + 0.694 = 6.123
Probability of meeting the schedule time is given by
Pi (Z s Ki) = Pi (Z – 0.52) = 0.30 (from normal distribution table)
Thus, the probability that the project can be completed in less than or
equal to 41.5 weeks is 0.30. In other words, probability that the project
will get delayed beyond 41.5 weeks is 0.70.
- Given that P (Z s Ki) = 0.95. But Z0.9S = 1.64, from normal distribution
table. Then 1.6 u = or
2.47
ST 42.8
is 1.6 u
V( )
ST E( )
i
i
i i
÷
=
µ
÷ µ
Sji = 1.64×2.47+42.8 = 46.85 weeks.
Self Assessment Questions
10. In a project network, a sequence of activities may form a loop.
(True/False)
11. A critical activity must have its total and free floats equal to zero.
(True/False)
12. A non-critical activity cannot have zero total float. (True/False)
13. The critical path of project network represents the minimum duration
needed to complete the network. (True/False)
14. The analysis in CPM does not take into account the cases where time
estimates for the different activities are probabilistic. (True/False)
14.5 Summary
- Critical path computations are quite simple, yet they provide valuable
information that simplifies the scheduling of complex projects.
- The result is that PERT-CPM techniques enjoy tremendous popularity
among practitioners in the field.
- The usefulness of the techniques is further enhanced by the availability
of specialised computer systems for executing, analysing and controlling
network projects.
14.6 Glossary
Uncertainty: volatility.
Task: activity that has to be completed within the specified time.
Operations Research Unit 14
Sikkim Manipal University Page No. 291
Latest start (LS) time: the latest time that an activity can begin without
delaying the project completion time.
14.7 Terminal Questions
1. Write down the basic difference between PERT and CPM.
2. Explain project management (PERT).
3. A project has 10 activities. The following table shows the information
about the activities.
Table 14.5: Activities information
Activity Preceding activity Duration in weeks
A – 6
B – 3
C A 5
D A 4
E A 3
F C 3
G D 5
H B, D, E 5
I H 2
J I, G, F 3
Draw the network
Find the project duration
Identify the CPM
Prepare the schedule
4. A small project consisting of eight activities has the following
characteristics:
Table 14.6: Time estimates (in weeks)
Activity
Preceding
activity
Most
optimistic
time (a)
Most likely
time (m)
Most
pessimistic
time (b)
A None 2 4 12
B None 10 12 26
C A 8 9 10
D A 10 15 20
E A 7 7.5 11
F B, C 9 9 9
G D 3 3.5 7
H E, F, G 5 5 5
Operations Research Unit 14
Sikkim Manipal University Page No. 292
1. Draw the PERT network for the project.
2. Determine the critical path.
3. If a 30-week deadline is imposed, what is the probability that the project
will be finished within the time limit?
4. If the project manager wants to be 99% sure that the project is
completed on this schedule date, how many weeks before that date
should he start the project work?
14.8 Answer
Self Assessment Questions
1. True
2. True
3. True
4. False
5. True
6. Time, resource
7. Greater than
8. Precedence
9. Forward pass & backward pass
10. False
11. True
12. True
13. True
14. True
Terminal Questions
1. There are no essential differences between PERT and CPM as both of
them share in common the determination of a critical path. For more
details refer to section 14.2
2. Probability considerations are incorporated in PERT by assuming that
the time estimate for each activity is based on three time estimates of
optimistic time, pessimistic time and most likely time. For more details
refer to section 14.4
3. b) 20 weeks c) A – D – H – I - J . For more details refer section 14.3
4. b) A – D – G – H c) 0.6591 d) 34.7 weeks. For more details refer
section 14.4
Operations Research Unit 14
Sikkim Manipal University Page No. 293
14.9 Case Study
Kaushik Mills – production capacity building
Kaushik Mills decided to increase its production capacity by building a new
feed mill. The project consisted of a number of separate tasks, some of
which could not be started before others were complete. The table below
shows the lists of activities, along with the times expected for each. The
management wanted to advance the schedule as much as possible in order
to save valuable time in getting the new mill into operation. The president of
the mils commented: ‖Every week saved is worth Rs 70,000 in lost
contribution if we can get going‖.
Some of the construction activities could be sped up. For example, the firm‘s
architects could by working overtime, design the new plant in 10 weeks
instead of the originally estimated 12 weeks. This advancement would cost
the mills an additional Rs 25,000 per week that is advanced.
The president of mills had already held discussions with mill contractors, an
independent firm which was a potential contractor for one of the projects
major tasks, for building the plant. Kaushik Mills intended to do the other
tasks either itself or through its agents. During the talks with the mill
contractors, the management had explored a number of bonus and penalty
clauses. One of these was that for every week that the plant was built ahead
of 10 weeks, the mills would pay contractors an additional Rs 75,000.
Activity Expected Precedent Minimum Crash
Activities Time(weeks) (rs/week)
A 12 - 10 24,000
B 8 A 18 -11
C 6 A 11 3,000
D 13 A 13 -11
E 4 B,C 4 -11
F 101 C 8 30,000
G 6 C 4 1500
H 101 E,F 6 75,000
I 8 D,G 8 -11
J 8 H,I 8 -11
K 4 J 4 -11
Operations Research Unit 14
Sikkim Manipal University Page No. 294
1. The Management of Kaushik Mills desires to know which activities it
would crash? How should they schedule their workers?
References:
- Kapoor V. K. (2005). Operations Research. Sultan Chand and Sons.
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Kanti Swarup & Gupta P. K., & Hira D. S., & Manmohan (2004).
Operation Research. Sultan Chand and Sons.
Operations Research Unit 15
Sikkim Manipal University Page No. 295
Unit 15 Game Theory
Structure:
15.1 Introduction
Objectives
15.2 Competitive Situations
Marketing different brands of a commodity
Campaigning for elections
Fighting military battles
15.3 Characteristics of Competitive Games
N-person game
Zero-sum game
Two-person zero-sum game (Rectangular game)
Strategy
Pure strategy
Mixed strategy
15.4 Maximin – Minimax Principle
Saddle point
Solution to a game with a saddle point
15.5 Dominance
Solving games using dominance
15.6 Summary
15.7 Glossary
15.8 Terminal Questions
15.9 Answers
15.10 Case Study
15.1 Introduction
In the previous unit, we learnt about project management and basic
difference between PERT and CPM. We also learnt the PERT/CPM network
components and precedence relationship.John Von Newman developed the
Game Theory in the year 1928. The theory gained importance after 1944,
when John Von Newman along with Morgenstern published their work
on‗Theory of games and economic behaviour.‘ This highly resourceful field
of study is developing fast.
Operations Research Unit 15
Sikkim Manipal University Page No. 296
In this unit, you will learn about game theory, competitive situations and
characteristics of competitive games. You will also learn about Maximin–
Minimax principle and dominance.
Objectives:
After studying this unit, you should be able to:
- analyse competitive situations
- apply game theory to such situations
- evaluate a solution using saddle point and dominance principle
- define zero-sum game
15.2 Competitive Situations
Competitive situations arise when two or more parties with conflicting
interests operate. The competitive situations depicted in figure 15.1 may
occur.
Fig. 15.1: Types of Competitive Situations
Let us now describe the competitive situations in detail.
15.2.1 Marketing different brands of a commodity
Two or more brands of detergents/soaps trying to capture the market by
adopting various methods (courses), such as advertising through electronic
media, providing cash discounts to consumers or offering larger sales
commission to dealers are said to be in a competitive situation.
15.2.2 Campaigning for elections
Two or more candidates contesting in elections are in a competitive
situation. They adopt various methods (courses), such as campaigning
through TV, door to door campaigning or campaigning through public
meetings to capture more votes.
Operations Research Unit 15
Sikkim Manipal University Page No. 297
15.2.3 Fighting military battles
Another example of a competitive situation is where two forces are fighting a
war to gain supremacy over one another. They adopt various courses of
action, such as direct ground attack on enemy camp, ground attack
supported by aerial attack or playing defensive by not attacking. The above
mentioned situations can be considered as a competitive game where the
parties (players) adopt a course of action (play the game).
Self Assessment Questions
1. Competitive situations arise when _______ or ________ parties with
___________ operate.
15.3 Characteristics of Competitive Games
A competitive game has the following characteristics:
1. The number of players or competitors is finite.
2. Each player has finite number of courses of action or moves.
3. A game is played when each player adopts any one course of action.
4. Every time a game is played, the corresponding combination of courses
of action leads to a transaction or payment to each player. The payment
is called pay-off or gain. The pay-off may be monetary (money) or some
benefit, such as increased sales.
5. The players do not communicate with each other.
6. The players are aware of the rules before starting the game.
15.3.1 N-person game
A game having ‗n‘ players participating is called ‗n-person‘ game.A twoplayer
game is called 2-person game (two person game).
15.3.2 Zero-sum game
If the sum of the gains (pay-off) of the players in a gameis zero, the game is
called zero-sum game.
A zero-sum game with two players is called two-person zero-sum game. It is
also called rectangular game.
In a two-person zero-sum game, the gain of one player is the loss of the
other.
Operations Research Unit 15
Sikkim Manipal University Page No. 298
15.3.3 Two-person zero-sum game (Rectangular game)
A two-person zero-sum game is a game where:
- Two players participate.
- The gain of one player is the loss of the other.
Consider a two-person zero-sum game with the players A and B. Let
A1 ,A2 ,...Am be the m courses of action for player A. Let B1,B2, ...Bn be the
n courses of action for player B. Let aij (i=1,2,.....m;j=1,2,....n)be the pay-off
(gain) of player A when he plays the course of action, i A and player B plays
the course of action j B . Then, the matrix depicted in figure 15.2 is the payoff
(gain) matrix of player A.
Fig. 15.2: Pay-Off (Gain) Matrix of Player A
This is a m×n (read as m by n) game. Here, aij is A‘s gain and B‘s loss.
Therefore, (-aij ) is the gain of B. To obtain the pay-off matrix of B, write
(- ij a ) in the place of ij a in the above matrix and then write the transpose of
the matrix.
15.3.4 Strategy
In a game, the strategy of a player is predetermined. The player uses this
strategy to select a course of action during the game.The strategy of a
player may be ‗pure‘or ‗mixed‘ as depicted in figure 15.3
Operations Research Unit 15
Sikkim Manipal University Page No. 299
Fig. 15.3: Types of Strategy
15.3.5 Pure strategy
During the game, if a player‘s strategy is to adopt a specific course of action,
irrespective of the opponent‘s strategy, the player‘s strategy is called pure
strategy.
15.3.6 Mixed strategy
If a player chooses his course of action according to pre-assigned
probabilities, then the player‘s strategy is called mixed strategy. Thus, if
player A decides to adopt courses of action 1 2 A and A with perspective
probabilities 0.4 and 0.6, it is mixed strategy.
Case-let
2-Finger Morra Game
Two persons A and B play a game. They simultaneously raise their hand
and exhibit either one finger or two fingers. If both A and B show one
finger or if both show two fingers, then A should pay Rs. 10 to B. On the
other hand, if one player shows one finger and the other player shows two
fingers, B should pay Rs. 5 to A.
The pay-off matrix of A is:
Player B
B1 (one finger) B2 (two fingers)
Player A
A1 (one finger)
A2 (two fingers)
-10 5
5 -10
If the player A decides to show one finger ( A1 ) , his strategy is pure
strategy. On the other hand, if A decides to play A1 with probability
0.5 and A2 with probability 0.5, his strategy is mixed strategy. This
means, if he is repeatedly playing, he should play both A1 and
A2 randomly almost equal number of times.
Operations Research Unit 15
Sikkim Manipal University Page No. 300
Self Assessment Questions
2. The number of players or competitors is finite (True/False).
3. If the sum of the gains of the players in a game is 1, the game is called
zero-sum game (True/False)
4. A zero-sum game with two players is called a ______ game.
5. During the game, if a player‘s strategy is to adopt a specific course of
action, irrespective of the opponent‘s strategy is called ______ strategy
6. If a player chooses his course of action according to pre-assigned
probabilities, then the player‘s strategy is called ______ strategy.
15.4 Maximin – Minimax Principle
Solving a two-person zero-sum game
Player A and player B are to play a game without knowing the other player‘s
strategy. However, player A would like to maximise his profit and player B
would like to minimise his loss. Also each player would expect his opponent
to be calculative.
Suppose player A plays 1 A .
Then, his gain would be a11 ,a12 ,...a1n accordingly B‘s choice would
beB1 ,B2 ,...Bn .Leto1 =min{ a11 ,a12 ,...a1n } .
Then, o 1 is the minimum gain of A when he plays A1 ( 1 o is the minimum
pay-off in the first row.)
Similarly, if Aplays A2 , his minimum gain iso 2 , the least pay-off in the
second row.
You will find corresponding to A‘s play A1 ,A2 ,......Am , the minimum gains
are the row minimumso1 ,o 2 ,..o m .
Suppose A chooses the course of action where i o is maximum.
Then the maximum of the row minimum in the pay-off matrix is called
maximin.
The maximin is:
) ` ¹
¹ ´ ¦
o = (aij )
j
min
i
max
Operations Research Unit 15
Sikkim Manipal University Page No. 301
Similarly, when B plays, he would minimise his maximum loss.
The maximum loss to B is when B j is ( aij )
max
| j = i .
This is the maximum pay-off in the th j column.
The minimum of the column maximums in the pay-off matrix is called
minimax.
The minimax is:
Ifo =| =v( say ) , the maximin and the minimax are equal and the game is
said to have a saddle point. Ifo < | , then the game does not have a saddle
point.
Note: o cannot be greater than| .
15.4.1 Saddle point
In a two-person zero-sum game, if the maximin and the minimax are equal,
the game has a saddle point.
Saddle point is the position where the maximin (maximum of the row
minimums) and minimax (minimum of the column maximums) coincide.
If the maximin occurs in the th r row and if the minimax occurs in the
th s column, the position (r, s) is the saddle point.
Here, v =ars is the common value of the maximin and the minimax.
It is called the value of the game.
The value of a game is the expected gain of player A, when both the players
adopt optimal strategy.
Note 1: If a game has a saddle point, (r, s), the player‘s strategy is pure
strategy. For players A and B, the suggested solutions are Ar and Bs
respectively.
) ` ¹
¹ ´ ¦
|= (aij )
i
max
j
min
Operations Research Unit 15
Sikkim Manipal University Page No. 302
Note 2: If a game does not have a saddle point, the solution offers a mixed
strategy.
Note 3: A game is said to be fair if its value is zero.
15.4.2 Solution to a game with a saddle point
Consider a two-person zero-sum game with players A and B. Let
A1,A2 ,...Am be the courses of action for player A. Let B1,B2 ,.....Bn be the
courses of action for player B.
The saddle point of the game is as follows:
1. The minimum pay-off in each row of the pay-off matrix is encircled as
shown in solved problem 2.
2. The maximum pay-off in each column is written within a box as shown in
solved problem 2.
3. If any pay-off is circled as well as boxed, that pay-off is the value of the
game. The corresponding position is the saddle point.
Let (r, s) be the saddle point. Then, the suggested pure strategy for
player A is r A . The suggested pure strategy for player B isBs . The value of
the game isars .
Note: However, if none of the pay-offs is circled or boxed, the game does
not have a saddle point. Hence, the suggested solution for the players is
mixed strategy.
Solved problem 1:
Two persons A and B, without showing each other, place a coin each on the
table. If the coins happen to be of the same denomination, player A will take
both of them. If they happen to be of different denominations, player B will
take both of them. Player A has a few one-rupee coins and two-rupee coins.
Player B has a few one-rupee, two-rupee and five-rupee coins.
i) Write down the pay-off matrix of A. Does the game have a saddle
point? If so, write down the solution.
ii) What happens to the game if both the players play only with one-rupee
and two-rupee coins?
Operations Research Unit 15
Sikkim Manipal University Page No. 303
Solution:
i) The pay-off matrix of A is as follows.
1. The minimum pay-off in each row is circled. (Here, the minimums
repeat.)
2. The maximum pay-off in each column is boxed.
3. The pay-off ‗-1‘ is circled as well as boxed. Therefore, the game has a
saddle point. It is the position (1, 3).
The solution to the game is:
a) Strategy for A is 1 A .
b) Strategy for B is 3 B .
c) Value of the game is v = -1 rupees.
ii) If both the players play only with one-rupee and two-rupee coins, the
pay-off matrix of A is:
Here, the game does not have a saddle point. Therefore, the suggested
solution is adoption of mixed strategy for the players.
Operations Research Unit 15
Sikkim Manipal University Page No. 304
Solved problem 2:
Verify whether the 2-finger morra game explained earlier has a saddle
point. If so, write down the solution for the game.
Solution:
The pay-off matrix of player A is as follows:
1.. The minimum pay-off in each row is circled.
2. The maximum pay-off in each column is boxed.
3. Since none of the pay-off is circled as well as boxed, the game does
not have a saddle point. Hence, solution for the game is mixed
strategy for both the players.
Solved problem 3:
A labour union of a firm is negotiating a new five-year settlement
regarding payments with the management. The union has the following
options: ‗ A1 : Aggressive bargaining‘, ‗ A2 : Bargaining with reasoning‘ and
‗ A3 : Conciliatory approach.‘ The likely mode of responses from the
management are: ‗B1 : Aggressive bargaining‘, ‗ 2 B : Bargaining with
reasoning‘, ‗B3 : Legalistic approach‘ and ‗B4 : Conciliatory approach.‘
The gains to the union in each case are as follows:
Operations Research Unit 15
Sikkim Manipal University Page No. 305
What strategy would you suggest for the two sides? What is the value of
the game?
Solution:
1.. The minimum pay-off in each row is circled.
2. The maximum pay-off in each column is boxed.
3. The value 12 is circled as well as boxed. So, the game has a saddle
point. It is the position (1, 3).
Therefore, the solution to the game is as follows:
a) Strategy for the union is ‗ A1 : Aggressive bargaining.‘
b) Strategy for the management is ‗B3 : Legalistic approach.‘
c) Value of the game is v = 12.
Operations Research Unit 15
Sikkim Manipal University Page No. 306
Solved problem 4:
Solve the game.
Is the game fair?
Solution:
1. The minimum pay-off in each row is circled.
2. The maximum pay-off in each column is boxed.
3. The value 7 is circled as well as boxed. So, the game has a saddle
point. It is the position (2, 1).
Therefore, the solution to the game is –
a) Strategy for the player is 2 A (second course of action)
b) Strategy for the opponent isB1 (second course of action)
c) Value of the game is v = 7
Operations Research Unit 15
Sikkim Manipal University Page No. 307
Solved problem 5:
A two-person zero-sum game, has the following pay-off matrix. Solve the
game.
Is the game fair?
Solution:
1. The minimum pay-off in each row is circled.
2. The maximum pay-off in each column is boxed.
3. The value 0 is circled as well as boxed. So, the game has a saddle
point at position (2, 2).
Therefore, the solution to the game is:
a) Strategy for the player is 2 A (second course of action).
b) Strategy for the opponent is 2 B (second course of action).
c) Value of the game is v = 0.
d) Since the value of the game is 0, the game is fair.
Self Assessment Questions
7. Saddle point occurs at row minimum and column maximum. (True/False)
8. If the value of the game is zero, then it is called zero-sum
game.(True/False)
9. The pay off matrix represents the gain for top player.(True/False)
Operations Research Unit 15
Sikkim Manipal University Page No. 308
15.5 Dominance
In a rectangular game, the pay-off matrix of player A is pay-off in one
specific row ( r row ) th exceeding the corresponding pay-off in another
specific row( s row ) th . This means thatwhatever course of action is
adopted by player B, for A, the course of action Ar yields greater gains than
the course of action As . Therefore, Ar is a better strategy than As
irrespective of B‘s strategy. Hence, you can say that Ar dominates As .
Alternatively, if each pay-off in a specific column ( p column ) th is less than
the corresponding pay-off in another specific column( q column ) th , it means
strategy Bp offers minor loss than strategy Bq irrespective of A‘s strategy.
Hence, you can say that Bp dominatesBq . Therefore, you can say that:
a) In the pay-off matrix, if each pay-off in r row th is greater than
(or equal to) the corresponding pay-off in thes row th , Ar dominates As .
b) In the pay-off matrix, if each pay-off in p column th is less than
(or equal to) the corresponding pay-off in theq column th ,
Bp dominatesBq .
At times, a convex combination of two or more courses of action may
dominate another course of action. Whenever a course of action (say
As orBq ) is dominated by others, then that course of action ( As orBq ) can
be deleted from the pay-off matrix. Such a deletion will not affect the choice
of the solution, but it reduces the order of the pay-off matrix. Successive
reduction of the order using dominance property helps in solving games.
15.5.1 Solving games using dominance
Consider a two-person zero-sum game with players A and B.
Let A1 ,A2 ,......Am be the courses of action for player A. Let B1 ,B2 , ...Bn be
the courses of action for player B.
Operations Research Unit 15
Sikkim Manipal University Page No. 309
Suppose the game has a saddle point. Use the dominance property in
sequence to delete the courses of action of A as well as B till the pair
comprising the saddle point remains alone. The procedure to arrive at the
saddle point is as follows:
a) In the pay-off matrix, if each pay-off in r row th is greater than (or equal
to) the corresponding pay-off in thes row th , Ar dominates As , hence
s A is deleted.
b) In the pay-off matrix, if each pay-off in p column th is less than (or equal
to) the corresponding pay-off in theq column th , Bp dominatesBq ,
henceBq is deleted.
c) Repeat the above steps in succession to achieve the saddle point.
Note: Sometimes, a convex combination of two or more courses of action
may dominate another course of action.
Solved problem 6:
Solve the following game using dominance property.
Solution: In the pay-off matrix, each pay-off in the first row exceeds the
corresponding pay-off in the third row. Therefore, A1 dominates A3 . So,
A3 is deleted. Hence the reduced matrix is as follows:
Here, each pay-off in the third column is less than the corresponding
pay-off in the first column. Therefore, B3 dominatesB1 . Similarly,
Operations Research Unit 15
Sikkim Manipal University Page No. 310
B3 dominates B2 . Also, B3 dominates B4 . Thus, the matrix reduces to
Here, since 12>8, 1 A dominates 2 A . And so, finally the matrix reduces
to –
Thus, (1, 3) is the saddle point and the solution to the game is as follows:
a) Strategy for A is A1 .
b) Strategy for B isB3 .
c) Value of the game is v = 12
Solved problem 7:
Solve the following zero-sum game and find its value.
Solution: In the pay-off matrix, each pay-off in the second column is less
than (or equal to) the corresponding pay-off in the third column. And so,
the course of action Q dominates R. Similarly, Q dominates S. After
deleting R and S, the reduced matrix is as follows.
Here, pay-off in the second row is greater than (or equal to) the
corresponding pay-offs in the first, third as well as fourth rows. Therefore,
Operations Research Unit 15
Sikkim Manipal University Page No. 311
B dominates A, C and D. After deleting A, C and D, the reduced matrix is
as follows:
Here, 1<6. Therefore, Q dominates P. After deleting P, the reduced matrix
is as follows:
Therefore the solution to the game is:
a) Strategy for Company X is B.
b) Strategy for Company Y is Q.
c) Value of the game is v = 1.
Solved problem 8:
In a two-person zero-sum game, the pay-off matrix of A is –
Write down the pay-off matrix of player B.
Solution: The pay-off matrix of B is as follows:
Operations Research Unit 15
Sikkim Manipal University Page No. 312
Self Assessment Questions
10. In the pay-off matrix, if each pay-off in rth row is greater than (or equal
to) the corresponding pay-off insth row the , Ar dominates As , hence As
is __________.
11. In the pay-off matrix, if each pay-off in pthcolumnis less than (or equal
to) the corresponding pay-off in qth column, Bp, dominates Bq, hence Bq
is _________.
15.6 Summary
Let us recapitulate the important concepts discussed in this unit:
- Competitive situations arise when two or more parties with conflicting
interests operate.
- Competitive situations arise when two or more parties with conflicting
interests operate.
- In a game, the strategy of a player is predetermined. The player uses
this strategy to select a course of action during the game.The strategy of
a player may be ‗pure‘or ‗mixed‘.
- The maximum of the row minimum in the pay-off matrix is called
maximin. The minimum of the column maximums in the pay-off matrix is
called minimax.
- Saddle point is the position where the maximin (maximum of the row
minimums) and minimax (minimum of the column maximums) coincide.
- Whenever a course of action is dominated by others, then that course of
action can be deleted from the pay-off matrix. The deletion will not affect
the choice of the solution, but it reduces the order of the pay-off matrix.
Successive reduction of the order using dominance property helps in
solving games.
15.7 Glossary
Matrix: rectangular arrangement of numbers in rows and columns
Transpose of a matrix:the matrix that results from interchanging rows into
and columns into rows
Strategy: course of action during the game for achieving gains
Operations Research Unit 15
Sikkim Manipal University Page No. 313
15.8 Terminal Questions
1. Solve the following rectangular game.
2. In a rectangular game, pay-off matrix of player A is as follows:
i) Solve the game.
ii) Write down the pay-off matrix of B and then, solve the game.
3. Briefly describe the characteristics of competitive game.
4. Explain maximin-minimax principle.
15.9 Answers
Self Assessment Questions
1. Two, more, conflicting interest
2. True.
3. False
4. Rectangular
5. Pure
6. Mixed
7. True
8. True
9. False
10. Deleted
11. Deleted
Operations Research Unit 15
Sikkim Manipal University Page No. 314
Terminal Questions
1. Value of the game is 1. A‘s strategy is (0, 1, 0, 0), while B‘s strategy is
(0, 0, 1, 0, 0).For more details refer section 15.3.3.
2. Value is 5. A‘s strategy is (1, 0), while B‘s strategy is (1, 0).For more
details, refer section 15.3.3.
3. Number of players is finite, each player has finite moves, player can
adapt any one course of action. For more details refer to section 15.3 for
answer.
4. The maximin is:
) ` ¹
¹ ´ ¦
o = (aij )
j
min
i
max
The minimax is:
) ` ¹
¹ ´ ¦
|= (aij )
i
max
j
min
For more details refer to section 15.4 for answer.
15.10 Case Study
Swimming Teams
The ABC swim team will soon have an important meet with the XYZ team.
Each team has a star swimmer (Ram and Shyam respectively) who can
swim very well in the 100-yard butterfly, backstroke and breaststroke
events. However the rules prevent them from being used in more than two
of these events. Therefore, their coaches now need to decide how to use
them to maximum advantage.
Each team will enter three swimmers per event. For each event, the
following table gives the best time previously achieved by Ram and Shyam
as well as the best time for each of the swimmers who will definitely enter
the event.(Whichever event Ram or Shyam does not swim, his team‘s third
entry for that event will be slower than the two shown in the table).
Operations Research Unit 15
Sikkim Manipal University Page No. 315
ABC Swim Team XYZ Swim Team
Entry Entry
1 2 Ram Shyam 1 2
Butterfly stroke 1:01.6 59.1 57.5 58.4 1:03.2 59.8
Backstroke 1:06.8 1:05.6 1:03.3 1:02.6 1:04.9 1:04.1
Breaststroke 1:13.9 1:12.5 1:04.7 1:06.1 1:15.3 1:11.8
The points awarded are 5 points for first place, 3 points for second place, 1
point for third place, and none for lower places. Both coaches believe that all
swimmers will essentially equal their best times in this meet. Thus Ram and
Shyam each will definitely be entered in two of the three events.
Discussion Questions
a) The coaches must submit all their entries before the meet without
knowing the entries for the other team, and no changes are permitted
later. The outcome of the meet is very uncertain. Formulate this as a
two-person, zero-sum game.
b) Eliminate dominated strategies, find the strategies for both the teams
according to minimax criterion.
References:
- Sharma J. K. (2006). Operations Research. Macmillan India Limited.
- Taha H. Operations Research. Prentice Hall.
- Gupta P. K. Problems in Operations Research (Methods and Solutions).
Sultan Chand and Sons Publishers.
- Bronsoon, R. Theory and Problems of Operations Research. Schaum‘s
Outline Series.
E-References:
- http://www.springerlink.com/content/g7ph5p3k4128t608/?MUD=MP,
retrieved on Dec 20, 2011.
–––––––––––––––––––––

Sign up to vote on this title
UsefulNot useful