Operations Research Unit 13

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.

Manipal University Jaipur Page No. 252

Operations Research Unit 13

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

Manipal University Jaipur Page No. 253

Operations Research Unit 13

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.

Manipal University Jaipur Page No. 254

Operations Research Unit 13

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 1 cm 2 cm 3 cm 4 cm 5 cm
rain Rain Rain Rain Rain 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 2 cm 3 cm
Rain Rain 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 RN range
probability
No rain 0.50 0.50 00-49
1 cm Rain 0.25 0.75 50-74
2 cm rain 0.15 0.90 75-89
3 cm Rain 0.05 0.95 90-94
4 cm Rain 0.03 0.98 95-97
5 cm Rain 0.02 1.00 98-99

Manipal University Jaipur Page No. 255

Operations Research Unit 13

Step 2: Previous day no rain distribution is as depicted in table 13.4:

Table 13.4: Distribution Table
Event Probability Cumulative probability RN range
No rain 0.75 0.75 00-74
1 cm Rain 0.15 0.15 75-89
2 cm rain 0.06 0.96 90-95
3 cm Rain 0.04 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 cm
7 70 1 cm Rain 2 cm
8 06 No rain 2 cm
9 78 1 cm Rain 3 cm
10 76 2 cm Rain 5 cm

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
Manipal University Jaipur Page No. 256
Operations Research Unit 13

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.

Manipal University Jaipur Page No. 257

Operations Research Unit 13

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 Random Number of Chef Number Fraction
week Digit customers ‘in’ or of days of days
‘out’ ‘in’ to ‘in’ to
date date
1 Monday 3 Average In 1 1.00
2 Tuesday 5 Average In 2 1.00
3 Wednesday 1 Lots In 3 1.00
4 Thursday 8 Very few Out 3 0.75
5 Friday 6 Very few In 4 0.80
6 Saturday 6 Very few Out 4 0.67
7 Sunday 4 Average Out 4 0.57
8 Monday 4 Average Out 4 0.50
9 Tuesday 0 Lots In 5
10 Wednesday 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.

Manipal University Jaipur Page No. 258

Operations Research Unit 13

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.

Manipal University Jaipur Page No. 259

Operations Research Unit 13

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 0.2
4 0.5
8 0.3
1-3 2 0.3
4 0.5
7 0.2
2-4 2 0.3
4 0.3
6 0.4
3-4 3 0.3
6 0.4
8 0.3
4-5 2 0.2
3 0.2
4 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:

Manipal University Jaipur Page No. 260

Operations Research Unit 13

Table 13.10: Simulation Table

Activity
Simul 1-2 1-3 2.4 3-4 4-8 Project
Critical Paths
ation RN D RN D RN D RN D RN D Duration
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 Time (in days) and its
predecessor probability
3 4 5
A – 0.20 0.60 0.20
B – 4 5 6 7 8
0.10 0.30 0.30 0.20 0.10
C A 1 3 5
0.15 0.75 0.10
D B,C 4 5
0.80 0.20
E D 3 4 5 6
0.10 0.30 0.30 0.30
F D 5 7
0.20 0.80
G E,F 2 3
0.50 0.50

Manipal University Jaipur Page No. 261

Operations Research Unit 13

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

Manipal University Jaipur Page No. 262

Operations Research Unit 13

Table 13.12: Random Number Coding

Cumulative Random
Activity Time Probability number
Probability
Interval
A 3 0.20 0.20 00-19
4 0.60 0.80 20-79
5 0.20 1.00 80-99
B 4 0.10 0.10 00.09
5 0.30 0.40 10-39
6 0.30 0.70 40-69
7 0.20 0.90 70-89
8 0.10 1.00 90-99
C 1 0.15 0.15 00-14
3 0.75 0.90 15-89
5 0.10 1.00 90-99
D 4 0.80 0.80 00-79
5 0.20 1.00 80-99
3 0.10 0.10 00-09
E 4 0.30 0.10 10-39
5 0.30 0.70 40-69
6 0.30 1.00 70-99
F 5 0.20 0.20 00-19
7 0.80 1.00 20-99
G 2 0.50 0.50 00-49
3 0.50 1.00 50-99

Simulation worksheet
Activity
Table 13.13 depicts the simulation worksheet.
Table 13.13: Simulation Worksheet
Simulation A B C D E F G
RN D RN D RN D RN D RN D RN D 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

Manipal University Jaipur Page No. 263
Operations Research Unit 13

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

Manipal University Jaipur Page No. 264

Operations Research Unit 13

The processing times at the two machines are as follows:

Table 13.15: Simulation Table
Machine M1 Machine M2
Time Probability
Probability Time(minutes)
(minutes)
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.
Manipal University Jaipur Page No. 265
Operations Research Unit 13

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:

Manipal University Jaipur Page No. 266

Operations Research Unit 13

 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.

Manipal University Jaipur Page No. 267

Operations Research Unit 13

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.

Manipal University Jaipur Page No. 268

Operations Research Unit 13

 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

Manipal University Jaipur Page No. 269

Operations Research Unit 13

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

Manipal University Jaipur Page No. 270

Operations Research Unit 13

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.

Manipal University Jaipur Page No. 271

Operations Research Unit 13

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

Manipal University Jaipur Page No. 272

Operations Research Unit 13

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.

Manipal University Jaipur Page No. 273