Introduction

eny of operations Research methods

Simulation
building models we have to framea number of
depend upon model building and for
assumptions. In linear programming
nodel some of theassumptions are (1) the objective
constraints are also linear. In function is linear and (ii) the
inventory models, to determine the Economic
order quantity we assume that the demand is
instanteneous and there 1S no lead time. Even if uniform, production is
there is lead time we assume it to
he a constant time. In queuing models we assume that
the
service time is exponential and service rate is greater than arival
is poisson,
the arrival rate to
imate the expected queue length and the expected waiting time of a
These are some ofthe examples of OR Techniques which depend upon customer.
a number
of assumptions and in many of these models the assumption are not realistic.
To amicable to analytical results these assumptions are made and these formulae
may not give reasonable decison making results. In such situation we can not
depend upon the models developed. There are also many real life situations
where in model building is so complex and it may not be possible to build
suitable analytical models. In such cases the idea called simulation comes as an
altermative technique to get the designed results in areasonably satisfied way.
Simulation analysis is a natural and logical extension to the analytical and
mathematical modets to many problems of operations research. For many
Situations defying mathematical formulation, simulation is the only toolthat
might be used to obtain relevent answers concerning the problem. For example
in designing an airplane the designer can solve various equations describing
the aerodynamics of the plane. If those equations are difficult to solve,a scale
model can be built and its behaviour be observed in a wind tunnel. In simulation
We build mathematical models while we cannot solve and run them on trial and
error data to simulate the behaviour of the system.
ne word simulation has been used so far without describing what it is.
Therefore we need to develop a working definition of this term. The following
w definitions to
definition has been adopted as one of the popular and familiar
resolve many of the complex problems that appear in real life situation.
DefinSimulation
ition is anumerical technique rconducting experiments
which involve.
describe the behaviour
logical and Imathematical relationships that interact to
and structuring of time.
extended period of
a complex real-world system over
 18.2
Simulation has often been'described as the process of
Opercreating
ationstheResearch
ofresults without even actually attaining that results itself.
Simulation essence
the construction, experimentation and manipulation ofacomplex model \towil forecast
involve
viewed as a method of
future data. Although simulation is sometimes f
often to be employed when all other model, fails,
simulation is one oflast resort
widely used and accepted technique in system analysis and the most
In management science it is
consideredlaas one ofthe powerful operations research.
tools
it is one of the hardset to apply properly and perhaps the most
which to draw accurate conclusion, difevent
ficulthough
from
Simulation can be used in the following situations.
1. It can be used to experiment with new situations about which we
no information so as toprepare for what may happen, have
2 Simulation analysis can be used to verify analytical solutions.
3. When new elements are introduced into asystem, simulation can be
toanticipate used
bottlenecks and other problems that may arise in the behaviore
of the system.
4. Simulation can serve as a pre service test to try out new
policies and
decision rules of a dynamical system.
5. If can be used as an experimental device for testing theoretical analvsis
and statistical analysis of a dynamical system.
6. Simulation enables one to study dynamical systems in either
real time,
compressed time or expanded time.
Limitation of simulation
Eventhough simulation is a powerful techniques where model building for
analytical results fails, it has its own disadvantages.
1. It isnot precise in many situations.
2. It is not a method of optimisation.
3. A good simulation model is very expensive and often it takes a
to developa useful model. long period
4. Simulation generates a way of
solutions themselves.
evaluating solution but does not generate
3. In many case, lack of precision make simulation
apply. models more difficult to
Simulation Procedure
Aeffective simulatior requires a good
amounttoof all
Though there is no single simulation procedure planning and organiza
simulation problemsthe
following general procedure can be adopted in
1. Define the problem to be general.
2. Formulate the simulated.
model intended to be used
3. Test the model, compare it behaviour with the behaviour of the actual
problem environment.
Simulation
18.3
4. Identify and Collect the data needed to testthe model
simulation
5, Run the
the results of the simulation
6. Analyse
test for testing the validity of simuBation
7. Apply a
Monte-CarloSimulation
Monte-Carlo method is a Simulation modelin asimplified version of real life
situations which gives solution of a problem to be solved by trial and error
approach. The origin of this technique of simnulation is due to great
mathematicians.John Von Newmann and Stanislaw Alam. During the world war
Ilphysicists of Los Alamos Scientific Laboratory were puzzled bythe behaviour
ofneutrons. How far would neutrons travel through different materials? The
Omathematicians suggested asolution which amount to submiting the problem
toa rouletle wheel. They gave it the name Monte-Carlo" for their secret work.
The Monte-Carlo method, which is actually the study of the laws of change was
SOpopularly spread so much and this has become an important technique in
variety of problems
Operations Research. This technique has been used to solve ausing mathematical
involving stochasticsituations which cannot be solved by
tools.
random numbers and is used to solve
The Monte-Carlo method employes
problems that depend upon probability where physical experimentation is
model is impossible. This is a
impracticable and formulation of a mathematical probability distribution of the
based on sampling techniques. First
the numbers is
method determined. Then a set of random
variable under consideration is the actual
values thathave the same distribution as
used to generate a set of involved in
deviced to simulate. The following are the steps
experience it is
simulation.
carrying out Monte-Carlo know the past behaviour of the
preliminary survey to
Step I: Conduct a
system. appropriate model
observations are taken to choose some probability
Step I1:Sample sampling procedure and determine the
correct
Ior the system. Use the distribution
distributions for the yalues of interest. intocumulative
distribution
Convert the probability
Drep 11 :
function. random numbers
Select a sequence of variable ofinterest with the
Step 1V :
sequence of values of
Step V: Determine the obtained in Step 1V.
of Step V.
quence of random variables mathematical function to,the values
Step VI: Fit some standard follows:
mathematically as distríbutionfunction F(x).
This can be described cumulative
variable X has its sample space by using the
Suppose the random
randomly drawnfromthe
Then the item x may be
following procedure.
 Operations Research
184
againstX.
(1) Plot F(x) decimal X fromthe uniform distribution on (0, 1) with the
arandomnumber tables.
(2) Select random
deternine the Xvalue
help of
(3) Usethe curve
obtained
value
in step (1)to
Yofstep (2). of X
corresponding
tothe decimal repeated to get required values
be
The above steps can
Generatingrandom numbers
considered as a value obtained from
may be uniform
Arandom number 1]. A physical example for this distribution isthe
distribution on the interval [0,circumference of which is marked with the values
modified roulette wheel, the a free rotation, comes to rest the arrow
s1
the wheel, after
from 0to 1. When assuned to be a random number
number, which may be
point to the real distribution.
number drawn from uniform
alsoother methods of generatingrandom numbers. One method of
There are
such as spinning adise
generating random numbers is to use a physical device
table containing one milion
B this method Rands Corporation produced a
random numbers.
Square
Another method of generating random numbers is known as "Mid
is 235
Method". Here we choose a number of threedigits. Suppose this number
This is called a seed random number. We consider its square.
(235)' = 55225.Here we consider the middle three digits 522 and this is the
second random number. (522) 0272484. The middle three digit number is 724.
This is the third random numbet. This way we proceed toobtain as many random
numbers as we want. A number of methods of generating random numbers
using computer is evolued. The most comnonly used method is the congruence
method or residue methodwhich isdescribed by

th
where a, b, mare constants and r, and r- are the th and (i+1)" random
numbers. This formula can be expressed as follows. Choose two positive numbers
a and b. Let r be a random number chosen initially. r, is called the randoni
number seed. Multiply r, by a and add bto this and then divide by m. Callthe
remainder by r . Then . is called the nd random number. Repeat this
process to get the successive random numbers. If b0, this procedure of
generating random numbets is called mixed generato.
Ifb=0, the procedureis called multiplication congruential generator. Ifa=1,
the procedure is called additive congruential generator. Let us considerexamples
of generating random numbers by the
above methods.