Certificate Course on

Operations Management and Analytics

(August 2022-January, 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
09-01-2023 1
 Welcome address by Program
coordinator & faculty colleagues

09-01-2023 2
Program faculty/industry expert
• Dr. Surya Prakash Singh, IIT Delhi

• Dr. Abhijit Majumdar, IIT Delhi

• Dr. Pankaj Dutta, IIT Bombay

• Dr. Akhilesh Kumar, IIT Kharagpur

• Dr. Sunil Agrawal, IIITDM Jabalpur

• Dr. Rajesh Matai, BITS Pilani

• Mr. Ram, Director, TATA CLIQ

09-01-2023 3
Program overview
• Entire program would be covered in 11 modules across
25 weeks

• 25 weeks includes 21 weeks of live sessions and 04

weeks reserved for assessments/tutorials

• Program will be delivered majorly by IIT Delhi faculty

• Some sessions would be taken by experts from outside

IIT Delhi

09-01-2023 4
• Sessions would be conducted over weekends
especially on Sunday 10 AM - 1.30 PM with 30
minutes of bio break.

• Total 42 sessions would be conducted on 21 weeks.

• Each session would be of 90 minutes including

Q&A.

09-01-2023 5
Program content
Dr. Surya

Dr. Surya

Dr. Surya

Dr. Abhijit

Dr. Abhijit

09-01-2023 6
 Dr. Surya

Dr. Surya

Dr. Sunil/
Mr. Ram

Dr. Surya

Dr. Rajesh

Dr. Rajesh/
Dr. Surya

09-01-2023 7
 Program weekly schedule
Module I: Foundations to Operations Management 1 Week 2 Sessions
Analytics (07 August)
Module II: Production and Capacity Planning 2 Weeks 4 Sessions
(13-20 August)
Module III: Maintenance and Queue Management 2 Weeks 4 Sessions

Evaluation 6th Week

Module IV: Quality and Six Sigma in Operations 2 Weeks 3 Sessions

Module V: Inventory Management 2 Weeks 3 Sessions

Module VI: Materials Resource Planning 2 Weeks 4 Sessions

09-01-2023 8
 Evaluation 13th Week
Module VII: Forecasting and Demand 2 Weeks 4 Sessions
Management
Module VIII: Predictive Analytics 2 Weeks 4 Sessions

Module IX: Descriptive Analytics 3 Weeks 5 Sessions

Evaluation 20th Week

Module X: Facility Planning and Project 2 Weeks 4 Sessions
Management

Module XI: Prescriptive Analytics 3 Weeks 5 Sessions

Evaluation 25th Week

09-01-2023 9
09-01-2023 10
 Certificate Course on
Operations Management and Analytics
(August 2022-January, 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
09-01-2023 11
Module 1

• Introduction to concepts, tools, techniques and

Analytics

• Traditional and advanced production system

• Need for production process

09-01-2023 12
 Session 1

Introduction to Operations
Management & Analytics

09-01-2023 13
What is Operations Management
and Analytics?
• First, let’s understand what is OM and how it is
evolved?

• Production Management

• Operations Management

09-01-2023 14
In early 30s, the term “production management”
was widely referred as the objective was to only
increase production to meet demand without much
paying any attention on after sales services or
customer’s service level.

09-01-2023 15
 Production Management

09-01-2023 16
After few decades, a new concept of “Operations
Management” came which focus not only on
production but also on after sales service.

To enable this, feedback is kept active so as to

modify the processes and design of products to
meet the customers need.

09-01-2023 17
 Operations Management

09-01-2023 18
Operations management is defined as to
produce a right product and to deliver it
at right location in right time in right
quantity with right price having right
quality.

Operations management deals with high

service level.

09-01-2023 19
 Some of prominent names in Production and
Operations Management

09-01-2023 20
 An
application
in daily life

(from “Cheaper by
the Dozen”)

09-01-2023 21
 Long-range plans
(over one year)
Research and Development
New product plans
Capital investments
Facility location/expansion

Top
executives Intermediate-range plans
(3 to 18 months)
Sales planning
Production planning and budgeting,
Operations Inventory management,
managers subcontracting levels

Short-range plans
(up to 3 months)
Job assignments
Operations Sequencing
managers, Job scheduling
supervisors, Dispatching
foremen Overtime

Responsibility Planning tasks and horizon

09-01-2023 22
Areas of concern in Operations
Management
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 23
 1. Demand forecasting
Forecasts create estimates What we will learn?
that help organization in
developing & implementing
production plan. • Simple moving average

Operations managers are • Simple Exponential method

responsible for the
processes using the
forecasts to deliver the • Trend Adjusted method
product at the most optimal
way. • Forecasting error
09-01-2023 24
 Resource Production
Planning Planning

Marketplace Demand Mgt.

Master
Production
Scheduling

09-01-2023 25
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 26
2. Production & Capacity Planning
It is a production activity, in advance of few
months, to give an idea to management:

(i) What quantity of materials/resources to be

procured?

(ii) When to produce so that the total cost of

production can be minimized?

09-01-2023 27
 What we will learn?

• Job Sequencing

• Production level

• Chase strategy

• Prescriptive analytics for production planning

• CPOF and Bill of Material method

09-01-2023 28
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 29
 3. Maintenance Management
Because machines malfunctions
have a direct impact on:

Production capacity

Production costs

Product and service quality

Employee or customer safety

Customer satisfaction
09-01-2023 30
 What we will learn?
Need of maintenance.
Total preventive maintenance (TPM).
Optimal tradeoff between Repairs & PM.
How Speedy Should Repairs Be?

09-01-2023 31
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 32
 4. Queue management

Also known as
Waiting line
management

09-01-2023 33
 What we will learn?

Some well known queuing models.

How to calculate waiting time?

How to reduce the waiting time?

How to know the queue length?

09-01-2023 34
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 35
 5. Inventory Management
What we will learn?
Inventory carries
about 50-60% cost of • Various inventory costs
the final product.
• EOQ & POQ model
Therefore, proper
management of • How much to order?
inventory becomes
essential. • When to order?

• How to classify various inventories?

09-01-2023 36
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 37
6. Materials Management Planning

09-01-2023 38
 What we will learn?
How to construct MRP table?
What to order?
How much to order?
When to order?
09-01-2023 39
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 40
7. Six sigma and Quality control
UCL

Target

LCL
Process is “in control.”

What we will learn?

• Six sigma concept
• Control charts for manufacturing industries
• Control charts for services industries
09-01-2023 41
Areas of concern in Operations
Management…contd
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 42
 8. Facility planning

What we will learn?

Facility planning refers
to the optimal Types of layout
arrangement of How to design manufacturing layout?
machines within the How to design services layout?
shop floor. How to design cellular layout?

09-01-2023 43
Areas of concern in Operations
Management…
1. Demand forecasting
2. Production & capacity Planning
3. Maintenance Management
4. Queue management
5. Inventory Management
6. Materials Management Planning
7. Six Sigma and Quality control
8. Facility planning
9. Project Management

09-01-2023 44
9. Project Management

09-01-2023 45
 • What we will learn?

• Defining Project
• Construction of project network
• How to construct a project using AOA & AON approach?
• How to compute project completion time?
• How to estimate project completion time under
uncertainty?
• Project crashing

09-01-2023 46
Classification of Analytics
• Predictive analytics

• Descriptive analytics

• Prescriptive analytics

09-01-2023 47
09-01-2023 48
Operations Management &
Analytics
• Predictive analytics in demand forecasting

• Prescriptive analytics in production management

and facility planning

• Descriptive analytics in vendor selection in

operations & supply chain

09-01-2023 49
 Operations Management Analytics is
applicable to…..

Procurement
Operations Distribution

Suppliers Manufacturer Distributors Consumers

Benefits of an efficient operations
• Effective utilization of resources
• Reduced work-in-process inventory
• Increased responsiveness to customers
• Increased through put
• Reduced non-value adding activity
• Reduced production cost
• Reduced production time

09-01-2023 51
Classification of production system

09-01-2023 52
Applications of operations
management and analytics
• Mass production industry

• FMCG industry

• Logistics industry

• Services industry such as hospital, hotel etc

• Automotive industry

• Small and medium enterprises

09-01-2023 53
Future of Operations Management
and Analytics
• Manufacturing sectors such as automotive, textile,
cement, steel, construction etc.

• Services sections such as hospital, hotel, airlines,

tourism, entertainment industry etc.

• Logistics service providers

• Construction and Project management

09-01-2023 54
 Operations management in Practice

09-01-2023 55
 Session 2

Operations Sequencing and

Scheduling

09-01-2023 56
Sequencing
Sequencing establishes the order for doing the jobs at
each machine.

Sequencing reflects job priorities according to the way

that jobs are arranged in the queues.

Assume Jobs x, y, and z have been assigned to machine 1.

Jobs x, y, and z are in a queue. Sequencing rules

determine which job should be first in line, which second,
etc.
57
A good sequence provides less waiting time, decreased
delivery delays, and better due date performance.

There are costs associated with waiting and delays.

There are other costs associated, such as, set up cost
and in-process inventory costs.

The objective function can be to minimize total

production time.

58
Scheduling
A production schedule is the time table that specifies
the times at which the jobs in a production department
will be processed on various machines.

The schedule gives the starting and ending times of each

job on the machines on which . the job has to be
processed.

59
 Example
Suppose there are three jobs (A, B, and C) in a shop floor that are to be
processed on four types of machines (M1, M2, M3, and M4).

These three jobs consist of 4, 3, and 4 operations respectively; and there

are four machines-one machine of each operations.

The operations for job A are A1, A2, A3, and A4. The operations of job B
are B1, B2, and B3. Similarly, the four operations of job C are C1, C2, C3,
and C4.

Each job is characterized by its routing that specifies the information

about the number of operations to be performed, the sequence of these
operations, and the machines required for processing these operations.

The times required for processing these operations are also required for
developing a production schedule.

60
The table gives the machine required for each
operation of each job. For example, the first Operation Machine Processing Time
Job
Number Number (Days)
operation of job A, A1, is processed on machine
A A1 M1 5
M1; second operation, A2, is processed on A2 M3 3
machine M3 and so on. A3 M4 7
A4 M2 4

The operations of all jobs have to follow their B B1 M2 2

processing sequences. For example operation A3 B2 M3 6
of job A can not be processed before operation B3 M4 8

A2.
C C1 M1 4
C2 M2 6
The processing time for each operation is also C3 M3 8
C4 M4 2
given in the table.

61
What should be an objective?

The objective is to schedule these jobs so as to

minimize the time to complete all jobs.

This time is called make-span or the schedule

time.

Make-span is referred in operations management.

62
 Let’s take a random solution
One of the schedules for this example is presented below in the form
of a Gantt Chart.

The Gantt chart, for each machine, shows the start and finish times of
all operations scheduled on that machine.

63
 Can we have better schedules?
Several alternative schedules can be generated for this example. The
schedules differ in the order in which the jobs are processed on the four
machines.

Three of these schedules are:

o The first schedule orders jobs as: A first, then B and then C (A-B-C).
o The second schedule orders jobs as: B first, then A, and then C (B-A-C).
o The third schedule orders jobs as: C first, then A, and then B (C-A-B).

The Gantt charts for these schedules are shown in next slide.

The values of make-span for these three schedules are 25, 27 and 30 days
respectively. Schedule A-B-C is the best of these three schedules.

64
 Time (Days)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
M1 A1 A1 A1 A1 A1 C1 C1 C1 C1 Sequence A-B-C
M2 B1 B1 C2 C2 C2 C2 C2 C2 A4 A4 A4 A4 (Make-span = 25
M3 A2 A2 A2 B2 B2 B2 B2 B2 B2 C3 C3 C3 C3 C3 C3 C3 C3
days)
M4 A3 A3 A3 A3 A3 A3 A3 B3 B3 B3 B3 B3 B3 B3 B3 C4 C4

Sequence B-A-C
(Make-span = 27
days)

Time (Days)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
M1 C1 C1 C1 C1 A1 A1 A1 A1 A1 Sequence C-A-B
M2 B1 B1 C2 C2 C2 C2 C2 C2 A4 A4 A4 A4 (Make-span = 30
M3 B2 B2 B2 B2 B2 B2 A2 A2 A2 C3 C3 C3 C3 C3 C3 C3 C3
days)
M4 A3 A3 A3 A3 A3 A3 A3 C4 C4 B3 B3 B3 B3 B3 B3 B3 B3

65
Is sequence A-B-C the best solution (global optimal)? Can we
find a better sequence than this? So, the scheduling techniques
attempt to answer it.

It should be noted that there are different effectiveness

measures of a schedule in different situations. Minimizing
make-span is the widely considered as it optimize production
time (product lead time i.e. make span).

66
Assumptions in operations scheduling
Once a job is started on a machine, its processing can
not be interrupted, that is, preemption is not allowed.

The machines are continuously available and will not

break down during the operations (This assumption is
rather unrealistic but we make this assumption to avoid
complexity in discussing scheduling concepts).

A machine is not kept idle if a job is available for

operation.

Also, each machine can process only one job at a time.

67
Classification of Scheduling Problems
The scheduling problems can be classified based on the
following criteria:

• Sequence of machines
• Number of machines
• Processing times
• Job arrival time
• Objective function

68
 • Sequence of Machines

The sequencing problems, based on the sequence of

machines, are classified as:

Flow Shops

Job Shops

69
Flow Shop
In a flow-shop , processing of all jobs require machines in the same order.

The following table gives an example of a flow-shop in which three jobs,

A, B, and C are processed on four machines, M1, M2, M3, and M4.

The sequences of machines to process these jobs are same (M1-M3-M4-

M2).

Example of a Flow Shop

Operation # Operation # Operation # Operation # Machine for Machine for Machine for Machine for
Job
1 2 3 4 Operation # 1 Operation # 2 Operation # 3 Operation # 4

A A1 A2 A3 A4 M1 M3 M4 M2
B B1 B2 B3 B4 M1 M3 M4 M2
C C1 C2 C3 C4 M1 M3 M4 M2

70
 Job Shop
In a job shop the sequence of machines will be mixed,
that is, the jobs may require machines in different
sequences.

Example of a Job Shop

Operation # Operation # Operation # Operation # Machine for Machine for Machine for Machine for
Job
1 2 3 4 Operation # 1 Operation # 2 Operation # 3 Operation # 4

A A1 A2 A3 A4 M1 M3 M4 M2
B B1 B2 B3 M2 M3 M4
C C1 C2 C3 C4 M1 M2 M3 M4

71
• Number of Machines
Based on the number of machines, the scheduling
problems are classified as:

Referred as
1. Single machine problems Sequencing
problem

2. Two-machine problems

managed by
3. Multiple (>=3 machine) problems prescriptive
analytics

72
 • Processing Times
Deterministic: If processing times of all jobs are
known and constant the scheduling problem is
called a deterministic problem.

Probabilistic: The scheduling problem is called

probabilistic (or stochastic) if the processing times
are not fixed; i.e., the processing times must be
represented by a probability distribution.

Subject of
research

73
• Job Arrival Times
Based on this criterion, scheduling problems are
classified as static and dynamic problems.

Static: In the case of static problems the number of

jobs is fixed and will not change until the current set of
jobs has been processed.

Dynamic: In the case of dynamic problems, new jobs

enter the system and become part of the current set of
unprocessed jobs. The arrival rate of jobs is given in the
case of dynamic problems.
Subject of
research
74
• Objective Functions
Operations scheduling problems has several objectives.

Minimize make-span (mostly considered)

Minimize average flow time (or job completion time)
Average number of jobs in the system
Minimize average tardiness
Minimize maximum tardiness
Minimize number of tardy jobs

75
1. Single Machine Scheduling (or
Sequencing problem)

There is one machine on which several jobs have

to be processed.

The order in which these jobs will be processed

needs to be specified. This schedule will not be
changed until all jobs have been processed. This is
the “static” version of the problem.

76
Few sequencing rules

There are several rules that can be used to find the

order of processing. We will study the following
three rules.

First Come First Served (FCFS)

Shortest Processing Time (SPT) or Shortest

Operation Time.

Earliest (shortest) Due Date (EDD)

77
Objective Functions

There are several objective functions that can be minimized in a

single machine problem. We will study the following objective
functions.

Minimize average completion (flow) time.

Minimize average number of jobs in the system.
Minimize average tardiness.
Minimize maximum tardiness
Minimize number of tardy jobs.

Note: A job is tardy if it is not completed by its due date.

78
 Example

Let’s consider an Example.

Days

There are five jobs A, B, C, D, and E. A is the Job Time Due Date
first job that arrived in the production
department. B,C, D, and E followed A in this A 17 45
order. B 12 35
C 22 27
The processing times and due dates of all jobs
are also given. D 18 54
E 26 47
The order in which these jobs have to be
processed needs to be specified.

79
 Let’s apply FCFS rule
First Come First Served
A is the first job to be processed and Job Time Due Date
Completion
Tardiness
Time
will be completed at time 17. Its due A 17 45 17 0
date is 45. So job A is not late (tardy); B 12 35 29 0
tardiness is zero. C 22 27 51 24
D 18 54 69 15
E 26 47 95 48
Job B starts after job A, and is
completed at time 29 (17+12). This is
also not tardy. Tardiness = Completion Time – Due Date

In this way the completion time and Make it zero if you get a
tardiness of all jobs are completed. negative value.

80
FCFS: Calculation of Objective
Functions
Average Completion Time: Add
completion times of all jobs and divide by
the number of jobs (261/5). It is 52.5.
Average Number of Jobs in the System:
This is obtained by dividing the total of
completion times of all jobs by the
completion time of the last job (261/95). It
is 2.75.
Average Tardiness : This is obtained by
adding the tardiness of all jobs and
dividing it by the number of jobs (87/5). It
is 17.4.
Maximum Tardiness: This is the maximum
of all tardiness values, It is 48.
Number of Tardy Jobs: Count the number
of jobs that are tardy. Three jobs (C, D,
and E) are tardy for this problem. It is 3.

81
 Let’s apply SPT rule
Due Completion
Job Time Tardiness
Date Time
A 17 45 17 0
B 12 35 29 0
C 22 27 51 24
The jobs are processed in the D 18 54 69 15
E 26 47 95 48
increasing order of their
processing times.
Shortest Processing Time
Completion
The job with the minimum Job Time Due Date
Time
Tardiness
processing time (B) is processed B 12 35 12 0
first. B is followed by A, D, C, A 17 45 29 0
and E. D 18 54 47 0
C 22 27 69 42
E 26 47 95 48
The calculations of the objective Total 252 90
functions follow the same
procedure as for the FCFS rule. Average Completion Time 50.4
Average Number of Jobs in System 2.65
Average Tardiness 18
Maximum Tardiness 48
Number of Tardy Jobs 2

82
 Let’s apply EDD Due Completion
Job Time Tardiness
Date Time
A 17 45 17 0
B 12 35 29 0
C 22 27 51 24
The jobs are processed in the D 18 54 69 15
E 26 47 95 48
increasing order of their due
dates.
Earliest Due Date
The job with the minimum due Completion
Job Time Due Date Tardiness
date (C) is processed first; and is Time
followed by B, A, E, and D. C 22 27 22 0
B 12 35 34 0
A 17 45 51 6
The calculations of the objective E 26 47 77 30
functions follow the same D 18 54 95 41
procedure as for the FCFS rule Total 279 77

Average Completion Time 55.8

Average Number of Jobs in System 2.94
Average Tardiness 15.4
Maximum Tardiness 41
Number of Tardy Jobs 3
83
 Shortest Processing Time
Completion
Job Time Due Date Tardiness
Time
B 12 35 12 0
A 17 45 29 0
D 18 54 47 0
C 22 27 69 42
E 26 47 95 48
Total 252 90

Average Completion Time 50.4

Average Number of Jobs in System 2.65
Average Tardiness 18
Maximum Tardiness 48
Number of Tardy Jobs 2

Earliest Due Date

Completion
Job Time Due Date Tardiness
Time
C 22 27 22 0
B 12 35 34 0
A 17 45 51 6 Which rule is the best?
E 26 47 77 30
D 18 54 95 41
Total 279 77

Average Completion Time 55.8

Average Number of Jobs in System 2.94
Average Tardiness 15.4
Maximum Tardiness 41
Number of Tardy Jobs 3

09-01-2023 84
09-01-2023 85
 2. Two Machines Problem
Consider a problem with five jobs (A, B, C, D, and E); and two
machines as M1 and M2.

All five jobs consist of two operations each. The first operation of
each job is processed on machine M1; and the second operation is
processed on machine M2.

Data for a 5-Job 2-Machine Flow Shop Problem

Machine for Machine for Time for Time for
Operation Operation
Job Operation # Operation # Operation # 1 Operation #
#1 #2
1 2 (Days) 2 (Days)
A A1 A2 M1 M2 8 3
B B1 B2 M1 M2 5 7
C C1 C2 M1 M2 6 9
D D1 D2 M1 M2 7 1
E E1 E2 M1 M2 4 6
86
The scheduling objective is to find an optimal sequence
that gives minimum make span for these five jobs on two
machines.

Once we know an optimal sequence, the time to

complete all jobs can be determined i.e. make span.

For example, A-B-C-D-E is a sequence order that tells us

that A is the first job to be processed; B is the second job
and so on. E is the last job to be processed. Is it optimal?

Another sequence could be B-C-A-E-D. Is it optimal?

What is the optimal sequence?

87
Number of Sequences
For this 5-job problem there are 120 (5!) different
sequences. Similarly, for a six-job problem, the number
of sequences will be 720 (6!).

In general, for a “n” job problem there are n! (n-

factorial) sequences.

Our goal is to find the best sequence that minimizes

make-span.
88
 Gantt Chart Sequence A-B-C-D-E
The Gantt chart for the sequence A-B-C-D-E is given below.

The value of make-span (time to complete all jobs) is 36 days.

Our objective is to identify the sequence that minimizes the value of make-
span.

Gantt Chart for Sequence A-B-C-D-E

Time (Days)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

M1 A1 A1 A1 A1 A1 A1 A1 A1 B1 B1 B1 B1 B1 C1 C1 C1 C1 C1 C1 D1 D1 D1 D1 D1 D1 D1 E1 E1 E1 E1

M2 A2 A2 A2 B2 B2 B2 B2 B2 B2 B2 C2 C2 C2 C2 C2 C2 C2 C2 C2 D2 E2 E2 E2 E2 E2 E2

89
 Johnson’s Rule
to Minimize (optimal) Make-span
We use following four steps to find the optimal sequence.

Step 1: Find the minimum processing time on both machines.

Step 2: Identify the job & machine for minimum time identified at Step 1.

Step 3: Scheduling Rule

(a) If the machine identified in Step 2 is the first machine i.e. M1
then the job will be scheduled in the first available position.
(b) If the machine identified in Step 2 is the second machine i.e.
M2 then the job will be scheduled in the last available position.

Step 4: Remove the job whose position is fixed in Step 3; and go to Step 1.

Continue this process until all jobs have been scheduled.

90
 Iteration 1

Machine for Operation # 1

Machine for Operation # 2

Time for Operation # 1

Time for Operation # 2

Operation # 1

Operation # 2

(Days)

(Days)
Job
Step 1: The minimum time is 1.

Step 2: The job is D and the machine A A1 A2 M1 M2 8 3

is M2.
B B1 B2 M1 M2 5 7
Step 3: The machine identified at C C1 C2 M1 M2 6 9
Step 2 is machine M2, job D will be D D1 D2 M1 M2 7 1
assigned to last available position i.e. E E1 E2 M1 M2 4 6
position 5 as given below.

Position 1 Position 2 Position 3 Position 4 D

Step 4: Delete job D .

91
 Time for Operation # 1 (Days)

Time for Operation # 2 (Days)

Iteration 2

Machine for Operation # 1

Machine for Operation # 2

Operation # 1

Operation # 2
Job
Step 1: The next minimum time is 3.

Step 2: The job is A and the

A A1 A2 M1 M2 8 3
machine is M2.
B B1 B2 M1 M2 5 7
Step 3: The job A will be assigned to C C1 C2 M1 M2 6 9
last available position i.e. position 4 D D1 D2 M1 M2 7 1 Scheduled

as given below. E E1 E2 M1 M2 4 6

Position 1 Position 2 Position 3 A D

Step 4: Delete job A.

92
 Time for Operation # 1 (Days)

Time for Operation # 2 (Days)

Machine for Operation # 1

Machine for Operation # 2

Iteration 3

Operation # 1

Operation # 2
Job
Step 1: The minimum time is 4.

Step 2: The job is E and the machine

is M1. A A1 A2 M1 M2 8 3 Scheduled

B B1 B2 M1 M2 5 7
Step 3: The job E will be assigned to C C1 C2 M1 M2 6 9
the first available schedule position
D D1 D2 M1 M2 7 1 Scheduled
i.e. position 1 as given below.
E E1 E2 M1 M2 4 6

E Position 2 Position 3 A D

Step 4: Delete job E.

93
 Time for Operation # 1 (Days)

Time for Operation # 2 (Days)

Machine for Operation # 1

Machine for Operation # 2

Iteration 4

Operation # 1

Operation # 2
Job
Step 1: The minimum time is 5.

Step 2: The job is B and the

machine is M1.
A A1 A2 M1 M2 8 3 Scheduled

Step 3: The job B will be assigned B B1 B2 M1 M2 5 7

to the first available position i.e. C C1 C2 M1 M2 6 9
position 2 as given below.
D D1 D2 M1 M2 7 1 Scheduled

E E1 E2 M1 M2 4 6 Scheduled

E B Position 3 A D

Step 4: Delete job B.

94
 Iteration 5

Time for Operation # 1 (Days)

Time for Operation # 2 (Days)

Machine for Operation # 1

Machine for Operation # 2

Operation # 1

Operation # 2
Job
The only unscheduled job at this
stage is C and; it will be assigned
to the remaining unassigned
position 3.
A A1 A2 M1 M2 8 3 Scheduled
The final sequence is given below.
B B1 B2 M1 M2 5 7 Scheduled

The value of make-span for this C C1 C2 M1 M2 6 9

sequence will be determined by D D1 D2 M1 M2 7 1 Scheduled

drawing the Gantt chart. E E1 E2 M1 M2 4 6 Scheduled

E B C A D

95
 Sequence E-B-C-A-D
The Gantt chart for the sequence E-B-C-A-D is
given below. The value of make-span is 31 days.

The best (global optimal or minimum) value of

make-span for this problem is therefore, 31 days.
Gantt Chart for Sequence E-B-C-A-D

Time (Days)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

M1 E1 E1 E1 E1 B1 B1 B1 B1 B1 C1 C1 C1 C1 C1 C1 A1 A1 A1 A1 A1 A1 A1 A1 D1 D1 D1 D1 D1 D1 D1

M2 E2 E2 E2 E2 E2 E2 B2 B2 B2 B2 B2 B2 B2 C2 C2 C2 C2 C2 C2 C2 C2 C2 A2 A2 A2 D2
96
Case problem:

Try to construct Gantt Chart to know

the make span.

09-01-2023 97
09-01-2023 98
 3. Three Machines Problem
Consider a problem with five jobs (A, B, C, D, and E); and three
machines as M1, M2, & M3.
All five jobs consist of three operations each. The first operation of
each job is processed on machine M1; the second operation is
processed on machine M2, and the third operation is processed on
machine M3.

Data for a 5-Job 2-Machine Flow Shop Problem

Machine Machine Machine Time for Time for Time for
Opera
Operat Operat for for for Operatio Operatio Operatio
Job tion #
ion # 2 ion # 3 Operatio Operatio Operatio n#1 n#2 n#3
1
n#1 n#2 n#3 (Days) (Days) (Days)
A A1 A2 A3 M1 M2 M3 8 5 4
B B1 B2 B3 M1 M2 M3 10 6 9
C C1 C2 C3 M1 M2 M3 6 2 8
D D1 D2 D3 M1 M2 M3 7 3 6
E E1 E2 E3 M1 M2 M3 11 4 5
99
 If any of the following condition holds true
Min{T1j} ≥ Max{Tij}
and/or Min{T3j} ≥ Max{Tij}

Then we can convert 3-Machine problem to a 2-Machine and

can apply Johnson’s method to get an optimal sequence.
This will give a minimum production time i.e. make span.

Time for Time for

Time for Time for Time for
Dummy Dummy
Job Operation # 1 Operation # 2 Operation # 3
Operation Operation #
(Days) (Days) (Days)
# 1 (Days) 2 (Days)
A 13 9 8 5 4
B 16 15 10 6 9
C 8 10 6 2 8
D 10 9 7 3 6
E 15 9 11 4 5

100
 Applying Johnson’s rule we can get an optimal sequence

C B E D A C B D E A

09-01-2023 101
 C B E A D C B A E D

C B A D E C B D A E

So, we get total six optimal sequence for the

3-Machine problem
C B E D A C B D E A

C B E A D C B A E D

C B A D E C B D A E

09-01-2023 102
All these six sequence will give an
optimized production schedule for these
five jobs on three different machines.

Let’s take first sequence to construct

production schedule. Similarly, it can
done for other sequence.
C B E D A

09-01-2023 103
 C B E D A

09-01-2023 104
For more than three machines
and/or conditions do not met for
three machine problem, we need
to apply Prescriptive analytics.
Prescriptive analytics-we will see in
Module 11.

09-01-2023 105
09-01-2023 106
 Certificate Course on
Operations Management and Analytics
(August 2022-January 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
09-01-2023 1
 Please email course
specific queries at

[email protected]

09-01-2023 2
 Module 2

• Introduction to Production & Capacity Planning

• Chase and Production Level Strategies for Production

Planning

• CPOF and Capacity Bill Methods for Capacity Planning

• Analytics in Production and Capacity Planning (Module XI)

09-01-2023 3
 Session 1-2

Production Planning

09-01-2023 4
Where Production Planning is in OMA?

Demand Forecast

yes No
Start Increase
Capacity?
Production Capacity

09-01-2023 5
 Higher Production Cost
Inefficient Higher Product Cost
Production Plan Productivity Down
Reduced Sales
Start Low Profit

Production
Minimized Production Cost
Efficient Resources Utilized
Production Plan Enhance Sales
Higher Profit

Increased Productivity

09-01-2023 6
 Therefore, Production Plan Need
to be Optimized.

How to optimized?

Pure Strategies Mixed Strategies

Capacity Option Analytical models
(Chase & Level Strategy) (Prescriptive Analytics)
Demand Option Will discuss in Module XI

09-01-2023 7
 Chase Strategy Services Industries

Production Level Manufacturing

Strategy Industries

Services &
Mixed Strategy Manufacturing
Industries

09-01-2023 8
 The Production Planning Process

 What is Aggregate Production Planning?

 Aggregate Planning Strategies

 Capacity Options
 Demand Options
 Mixing Options to Develop a Plan

09-01-2023 9
 What is Production Planning?
It is an production planning activity which is done in
advance of 3 to 18 months, to give an idea to management
to what quantity of materials/resources are to be procured
and when, so that the total cost of production can be
minimized.

 Objective is to minimize cost over the period by

 Production rates
 Labor levels
 Inventory levels
 Overtime work
 Subcontracting rates
 Other controllable variables

09-01-2023 10
 Planning period
Long-range plans
(over one year)
Research and Development
New product plans
Capital investments
Facility location/expansion

Top
executives Intermediate-range plans
(3 to 18 months)
Sales planning
Production planning and
Operations budgeting, inventory,
managers subcontracting levels

Short-range plans
(up to 3 months)
Job assignments
Operations Ordering
managers, Job scheduling
supervisors,
foremen

Responsibility Planning tasks and horizon

09-01-2023 11
What is Aggregate Production Planning?

Quarter 1
Jan Feb Mar
150,000 120,000 110,000

Quarter 2
Apr May Jun
100,000 130,000 150,000

Quarter 3
Jul Aug Sep
180,000 150,000 140,000

09-01-2023 12
 Aggregate Production Planning

09-01-2023 13
 Competitors’ Raw material Market
behavior availability demand External
to firm
External Planning Economic
capacity for conditions
production

Current Current Inventory Activities Internal

physical workforce levels required to firm
capacity for prod.

09-01-2023
14
 Few Production Planning Strategies
1. Adjust changes by varying workforce size

2. Use part-time, overtime, idle time to absorb changes

3. Use subcontractors and maintain a stable workforce

4. Change prices or other factors to influence demand

5. Use inventories to absorb changes

09-01-2023 15
 Classification of Production Planning
Strategies

1. Capacity Option (Chase or Production level)

Pure

2. Demand Option

3. Mixed Option (Prescriptive Analytics-Module XI)

09-01-2023 16
Capacity Options
1. Changing inventory levels

 Increase inventory in low demand

periods to meet high demand in the
future

09-01-2023 17
2. Varying workforce size by hiring or layoffs

 Match production rate to demand

 Training and separation costs for hiring and

laying off workers

 Laying off workers may lower morale and

productivity

09-01-2023 18
3. Varying production rate through overtime
or idle time

 Allows constant workforce

 May be difficult to meet large increases in

demand

 Overtime can be costly and may drive down

productivity

 Absorbing idle time may be difficult

09-01-2023 19
4. Subcontracting

 Temporary measure during periods of peak

demand

 May be costly

 Assuring quality and timely delivery may be

difficult

 Risk for a possible competitor

09-01-2023 20
 5. Using part-time workers
 Useful for filling unskilled or low
skilled positions, especially in
services

09-01-2023 21
Demand Options
6. Influencing demand
 Use advertising or promotion to
increase demand in low periods
 Attempt to shift
demand to slow
periods
 May not be
sufficient to
balance demand
and capacity
09-01-2023 22
7. Back ordering during high- demand
periods

 Requires customers to wait for an order

without loss of goodwill or the order

 Most effective when there are few if any

substitutes for the product or service

 Often results in lost sales

09-01-2023 23
 Chase strategy
 Match output rates to demand forecast for
each period

 Vary workforce levels (by hiring or firing) or

vary production rate (by means of overtime,
idle time, sub-contracting)

 Favored by many service organizations

(because inventory option is difficult or
impossible to adopt)

09-01-2023 24
 Level strategy (or level Scheduling)
 Daily production is uniform
 Use inventory or idle time as buffer
 Stable production leads to better quality
 Ex. Toyota, Nissan, Maruti, Tata Motors keep
production at uniform levels and may (1.
Inventory go up/ down to buffer the difference,
2. find alternate work for employees.)
 Some combination of capacity options, a
mixed strategy, might be the best solution

09-01-2023 25
 Example 1
ABC a manufacturer of roofing tiles has developed
monthly Forecasts for roofing tiles and presented the
period January-June in the table 1.

To represent the projected demand, ABC also draws

a graph (figure 1) that charts the daily demand each
month. The dotted line across the chart represents
the production rate required to meet average
demand which is computed by dividing the total
expected demand by number of production days.
 Table 1: Expected demand and number of production days.

Production Demand Per Day

Month Expected Demand Days (computed)
Jan 900 22 41

Feb 700 18 39

Mar 800 21 38

Apr 1,200 21 57

May 1,500 22 68

June 1,100 20 55

6,200 124
 Average Total expected demand
requirement =
Number of production days
6,200
= = 50 units per day
124
Forecast demand
Production rate per working day

70 –
Level production using average
monthly forecast demand
60 –

50 –

40 –

30 –

0 –
Jan Feb Mar Apr May June = Month
     
Figure 1 22 18 21 21 22 20 = Number of
working days
09-01-2023 28
Possible Pure Strategy 1
Constant Workforce

Cost Information
Inventory carrying cost $ 5 per unit per month
Subcontracting cost per unit $10 per unit
Average pay rate $ 5 per hour ($40 per day)
$ 7 per hour
Overtime pay rate
(above 8 hours per day)
Labor-hours to produce a unit 1.6 hours per unit
Cost of increasing daily production rate $300 per unit
(hiring and training)
Cost of decreasing daily production rate $600 per unit
(layoffs)
Table 2

09-01-2023 29
 Monthly
Cost Information
Production at Demand Inventory Ending
Inventory 50 Units
Month carry cost per Day Forecast $ 5Change
per unit per Inventory
month
Jan
Subcontracting 1,100
cost per unit 900 $10 +200
per unit 200
Feb pay rate 900
Average 700 +200
$ 5 per 400
hour ($40 per day)
Mar 1,050 800 +250
$ 7 per hour 650
Overtime pay rate
(above 8 hours per day)
Apr 1,050 1,200 -150 500
Labor-hours to produce 1.6 hours per unit
May 1,100 a unit 1,500 -400 100
Cost of increasing daily production rate $300 per unit
June 1,000
(hiring and training)
1,100 -100 0
Cost of decreasing daily production rate $600 per unit 1,850
(layoffs)
Total units of inventory carried over from one
Table 13.3 month to the next = 1,850 units
Workforce required to produce 50 units per day = 10 workers
09-01-2023 30
 Monthly
Costs
Cost Information
Production at Demand Calculations
Inventory Ending
Month carry
Inventory
Inventory 50 Units
cost per Day $9,250
carrying Forecast $ 5Change
perunits
(= 1,850 unit per Inventory
month
carried x $5
Jan
Subcontracting 1,100
cost per unit 900 per $10unit)
+200
per unit 200
Regular-time
Feb pay rate
Average labor
900 49,600
700 (= 10
$ 5 workers
+200
per x $40per
hour ($40 per
400
day)
Mar 1,050 800 day 124 days) 650
x+250
$ 7 per hour
Overtime pay rate
Other (above 8 hours per day)
Apr costs (overtime,
1,050 1,200 -150 500
hiring, layoffs,
Labor-hours to produce a unit 1.6 hours per unit
May
subcontracting) 1,100 1,500
0 -400 100
Cost of increasing daily production rate $300 per unit
June
Total cost
(hiring
1,000
and training)
1,100
$58,850 -100 0
Cost of decreasing daily production rate $600 per unit 1,850
(layoffs)
Total units of inventory carried over from one
Table 13.3 month to the next = 1,850 units
Workforce required to produce 50 units per day = 10 workers
09-01-2023 31
 7,000 –

6,000 – Reduction
of inventory
Cumulative demand units

5,000 – 6,200 units

Cumulative level
production using
4,000 – average monthly
forecast
requirements
3,000 –

2,000 – Cumulative forecast

requirements
1,000 –
Excess inventory

–
Jan Feb Mar Apr May June
Figure 2
09-01-2023 32
Possible Pure Strategy 2
Subcontracting

Production Demand Per Day

Month Expected Demand Days (computed)
Jan 900 22 41
Feb 700 18 39
Mar 800 21 38
Apr 1,200 21 57
May 1,500 22 68
June 1,100 20 55
6,200 124

Minimum requirement = 38 units per day

09-01-2023 33
 Forecast demand
Production rate per working day

70 –
Level production
60 – using lowest
monthly forecast
50 – demand

40 –

30 –

0 –
Jan Feb Mar Apr May June = Month
     
22 18 21 21 22 20 = Number of
working days
09-01-2023 34
Cost Information
Inventory carrying cost $ 5 per unit per month
Subcontracting cost per unit $10 per unit
Average pay rate $ 5 per hour ($40 per day)
$ 7 per hour
Overtime pay rate
(above 8 hours per day)
Labor-hours to produce a unit 1.6 hours per unit
Cost of increasing daily production rate $300 per unit
(hiring and training)
Cost of decreasing daily production rate $600 per unit
(layoffs)

09-01-2023 35
Cost Information
Inventory carry cost $ 5 per unit per month
In-housecost
Subcontracting production
per unit = 38$10units per day
per unit
Average pay rate x $124
5 perdays
hour ($40 per day)

Overtime pay rate

= 4,712
$ 7 perunits
hour
(above 8 hours per day)
Subcontract
Labor-hours units
to produce a unit = 6,200 - 4,712
1.6 hours per unit

=
Cost of increasing daily production rate
(hiring and training)
1,488
$300 per
units
unit

Cost of decreasing daily production rate $600 per unit

(layoffs)

09-01-2023 36
Cost Information
Inventory carry cost $ 5 per unit per month
In-housecost
Subcontracting production
per unit = 38$10units per day
per unit
Average pay rate x $124
5 perdays
hour ($40 per day)

Overtime pay rate

= 4,712
$ 7 perunits
hour
(above 8 hours per day)
Costs Subcontract
Labor-hours units
to produce a unit = 6,200 - 4,712
Calculations
1.6 hours per unit
Regular-time
Cost labor
of increasing $37,696
daily production =
rate (= 7.6 workers
1,488
$300 unit x $40 per
units
per
(hiring and training) day x 124 days)
Cost
Subcontracting 14,880rate (= $600
of decreasing daily production 1,488per unitx $10 per
units
(layoffs)
unit)
Table 13.3
Total cost $52,576

09-01-2023 37
Possible Pure Strategy 3 Hiring and firing

Production Demand Per Day

Month Expected Demand Days (computed)
Jan 900 22 41
Feb 700 18 39
Mar 800 21 38
Apr 1,200 21 57
May 1,500 22 68
June 1,100 20 55
6,200 124

Production = Expected Demand

09-01-2023 38
 Forecast demand and
Production rate per working day

monthly production
70 –

60 –

50 –

40 –

30 –

0 –
Jan Feb Mar Apr May June = Month
     
22 18 21 21 22 20 = Number of
working days

09-01-2023 39
Cost Information
Inventory carrying cost $ 5 per unit per month
Subcontracting cost per unit $10 per unit
Average pay rate $ 5 per hour ($40 per day)
$ 7 per hour
Overtime pay rate
(above 8 hours per day)
Labor-hours to produce a unit 1.6 hours per unit
Cost of increasing daily production rate $300 per unit
(hiring and training)
Cost of decreasing daily production rate $600 per unit
(layoffs)

09-01-2023 40
 Basic
Production
Cost Extra Cost of Extra Cost of
Daily (demand x Increasing Decreasing
Forecast Prod 1.6 hrs/unit x Production Production
Month (units) Rate $5/hr) (hiring cost) (layoff cost) Total Cost
Jan 900 41 $ 7,200 — — $ 7,200
$1,200
Feb 700 39 5,600 — 6,800
(= 2 x $600)
$600
Mar 800 38 6,400 — 7,000
(= 1 x $600)
$5,700
Apr 1,200 57 9,600 — 15,300
(= 19 x $300)
$3,300
May 1,500 68 12,000 — 15,300
(= 11 x $300)
$7,800
June 1,100 55 8,800 — 16,600
(= 13 x $600)
$49,600 $9,000 $9,600 $68,200

09-01-2023 41
Comparison of Three Strategies
Cost Plan 1 Plan 2 Plan 3
Inventory carrying $ 9,250 $ 0 $ 0
Regular labor 49,600 37,696 49,600
Overtime labor 0 0 0
Hiring 0 0 9,000
Layoffs 0 0 9,600
Subcontracting 0 14,880 0
Total cost $58,850 $52,576 $68,200

Plan 2 is the lowest cost option

09-01-2023 42
Mathematical Approaches
 Transportation Method of Linear
Programming
 Produces an optimal plan. But its application is
limited.

 Other Mathematical Models

 Prescriptive Analytics (We will discuss at Module XI)
 Simulation

09-01-2023 43
 Example 2:
Transportation Method
Sales Period
Mar Apr May
Demand 800 1,000 750
Capacity:
Regular 700 700 700
Overtime 50 50 50
Subcontracting 150 150 130
Beginning inventory 100

tires

Costs
Regular time $40 per tire
Overtime $50 per tire
Subcontracting $70 per tire
Carrying $2 per tire per month

09-01-2023 44
 Important points
1. Carrying costs are $2/tire/month. If goods
are made in one period and held over to
the next, holding costs are incurred

2. Supply must equal demand, so a dummy

column called “unused capacity” is added

3. Because back ordering is not viable in this

example, cells that might be used to satisfy
earlier demand are not available

09-01-2023 45
 4. Quantities in each column designate the
levels of inventory needed to meet demand
requirements

5. In general, production should be allocated

to the lowest cost cell available without
exceeding unused capacity in the row or
demand in the column

09-01-2023 46
 By
Applying
100
100
700
Heuristic
700
50
Method
50

150 150

700 700
50
Production
50

50 100 150 Cost=

650 50 700 108600
(=40*700+52*50+7
50 50 2*150+40*700+50*
130 50+70*50+72*100+
130
800 1000 750 230 2780 40*650)

Is Production Plan Best?

09-01-2023 47
 By
Applying
100 Analytics
700

50 Production
150
Cost=
700
105700
50 (=40*700+52*50+72*15
0+40*700+50*50+70*50
150 +40*700+50*50)

700

50
Will
130
800 1000 750 230 2780
discuss in
Module XI
09-01-2023 48
Summary of Aggregate Planning
Methods
Solution
Techniques Approaches Important Aspects
Graphical Trial and Simple to understand and
methods error easy to use. Many
solutions; one chosen
may not be optimal.
Transportation Optimization LP software available;
method of linear permits sensitivity
programming analysis and new
constraints; linear
functions may not be
realistic.

09-01-2023 49
 Solution
Techniques Approaches Important Aspects
Prescriptive Optimization Knowledge of LP modeling
Analytics is required. Knowledge of
LINGO, Excel Solver
required.
Simulation Change Complex; may be difficult
parameters to build and for managers
to understand.

09-01-2023 50
Few Service Scenarios
 Restaurants (highly variable demand)
 Smoothing the production process (or
production rate)
 Determining the optimal workforce size

 Hospitals
 Responding to patient demand

09-01-2023 51
  Airline industry

 Extremely complex planning problem

 Involves number of flights, number of

passengers, air and ground personnel,
allocation of seats to fare classes

 Resources spread through the entire

system

09-01-2023 52
 Example 3:
Prescriptive Analytics (Exact Method)
(Discuss in detailed at Module XI)
Item Cost
Materials $10/unit
Inventory holding cost $2/unit/month
Marginal cost of a stockout $5/unit/month
Hiring and training costs $300/worker
Layoff cost $500/worker
Labor hours required 4/unit
Regular time cost $4/hour
Over time cost $6/hour
Cost of subcontracting $30/unit

09-01-2023 53
 Month Demand Forecast
January 1,000
February 3,000
March 3,800
April 4,800
May 2,000
June 1,400

09-01-2023 54
Prescriptive Analytics in Production Planning
(Define Decision Variables)
Wt = Workforce size for month t, t = 1, ..., 6

Ht = Number of employees hired at the beginning of month t, t = 1, ..., 6

Lt = Number of employees laid off at the beginning of month t, t = 1, ..., 6

Pt = Production in month t, t = 1, ..., 6

It = Inventory at the end of month t, t = 1, ..., 6

St = Number of units stocked out at the end of month t,t = 1, ..., 6

Ct = Number of units subcontracted for month t, t = 1, ..., 6

Ot = Number of overtime hours worked in month t, t = 1, ..., 6

09-01-2023 55
 (Define Objective Function)
6 6
Min  640W t   300 H t
t 1 t 1
6 6 6
  500 Lt   6 Ot   2 I t
t 1 t 1 t 1
6 6 6
  5 S t  10 Pt   30 C t
t 1 t 1 t 1

09-01-2023 56
(Define Constraints Linking Variables)
• Workforce size for each month is based on hiring
and layoffs

W t  W t 1  H t  Lt, or
W t  W t 1  H t  Lt  0
for t  1,...,6, where W 0  80.

09-01-2023 57
 • Production for each month cannot exceed
capacity

Pt  40W t  Ot 4 ,
40W t  Ot 4  Pt  0,
for t  1,...,6.

09-01-2023 58
 • Inventory balance for each month

I t 1  Pt  C t  Dt  S t 1  I t  S t ,
I t 1  Pt  C t  Dt  S t 1  I t  S t  0,
for t  1,...,6,where I 0  1,000,
S 0  0,and I 6  500.
09-01-2023 59
• Over time for each month

Ot  10W t,
10W t Ot  0,
for t  1,...,6.

09-01-2023 60
09-01-2023 61
 Module 2

• Introduction to Production & Capacity Planning

• Chase and Production Level Strategies for Production

Planning

• CPOF and Capacity Bill Methods for Capacity Planning

• Analytics in Production and Capacity Planning (Module XI)

09-01-2023 62
 Session 3-4

Capacity Planning

09-01-2023 63
 Capacity Planning
• Capacity Planning: Role in Operations Management

• Capacity Planning Techniques/Methods

• CPOF Method

• Capacity Bill Method

• Capacity Requirements Planning Method

• Capacity Definitions

• Technique Advantages/Limitations
09-01-2023 64
Capacity Planning in Manufacturing

Demand Forecast

yes No
Start Increase
Capacity?
Production Capacity

How to calculate capacity?

09-01-2023
 How to calculate capacity?
Three Techniques we will discuss to calculate capacity?

• Approximate Calculation Using Historical

Info (CPOF)

• Detailed Calculation Using BOM, Routing

Info etc (Capacity Bill Method)

• Exact Calculation Using BOM, Routing Info,

MRP, Lead Time (CRP Method) Module VI
09-01-2023 66
09-01-2023 67
 Problem Data

Assumption: 40 hrs in a week

There are three work centers viz. 100, 200, 300

09-01-2023 68
09-01-2023 69
Based on Bill of Material and Routing data, the time required
for A and B on three work centres can be calculated.

09-01-2023 70
 1. Using CPOF

09-01-2023 71
09-01-2023 14
Example 1

09-01-2023 73
Solution:

09-01-2023 74
 2. Using Capacity Bill

09-01-2023 75
Calculations of Bill
of Capacity Table:

09-01-2023 76
09-01-2023 77
09-01-2023 78 12
09-01-2023 79
09-01-2023 80
09-01-2023 81
Similarly, capacity requirements on WC 100,
200 & 300 for rest of the periods can be
calculated and shown below.

09-01-2023 82
 Comparing Capacity Plan of
CPOF & Capacity Bill

09-01-2023 83
 3. Using CRP
(Will discuss/continue once MRP is covered in
Module VI)

09-01-2023 84
09-01-2023 85
 Technique Inputs
1 CPOF Percent utilization of Work Centre
2 Capacity Bill Bill of Material and Routing data
3 CRP In addition to 2nd technique, MRP for all
end items is also needed

Demand data is a basic information and required for all

techniques

09-01-2023 86
Capacity Definitions
• Required capacity: Estimated using capacity planning
techniques

• Rated capacity: Available capacity considering equipment

capabilities

• Demonstrated capacity: Actual (or realized) capacity based

on actual production output

09-01-2023 24
 Capacity Planning Techniques: Pros and Cons
CPOF
Pros Cons
Simple to use Do not give right information
Take care of historical information
Capacity Bill
Pros Cons
Easy to use Doesn’t consider lead time information of all
inventory items
Minimal computational
requirements
Consider product mix Less accurate as compared to CRP

Capacity Requirements Planning

Considers: Requires MRP system,
Time-phased Inventory information,
All primary & secondary data Extensive computational
Consider lead times
Considers product mix

09-01-2023 25
Production Planning Problem for Practice # 1

09-01-2023 89
Production Planning Problem for Practice # 2

09-01-2023 90
09-01-2023 91
Capacity Planning Problems for Practice # 1

09-01-2023 92
Capacity Planning Problems for Practice # 2

09-01-2023 93
09-01-2023 94
 Certificate Course on
Operations Management and Analytics
(August, 2022-January, 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
09-01-2023 1
 Module 3

• Need of maintenance.
• Types of maintenance activities.
• Optimal tradeoff between Repairs & PM.
• How Speedy Should Repairs Be?
• Need of queue management.
• Types of queue management models.
• M/M/1, M/M/S, M/D/1, M/G/1

2
 Session 1-2

Maintenance Management

3
 Why Maintenance?
• Equipment malfunctions have a direct impact on:

• Production capacity

• Production costs

• Product and service quality

• Employee or customer safety

• Customer satisfaction
 Maintenance Departments
• A maintenance manager typically is a plant engineer
who reports to a plant or manufacturing manager

• Maintenance departments are usually split into two

groups:

• Buildings and Grounds

• Equipment (This is our concern in OMA)

 Types of Maintenance Activities

Repair
Activity (Case I)
[breakdown
maintenance]

Maintenance Periodic
Preventive [Time based
Activity maintenance]
Activity (Case II)
Predictive
[Condition based
maintenance]

Corrective Activity
(Case III)
• Repairs
• Repair activities are reactive.

• Breakdowns and malfunctions typically occur when

equipment is in use.

• Standby machines and parts can speed repairs.

• Regularly scheduled inspections are performed.

• PM activities are performed before equipment fails.

• PM is usually performed during idle periods.

Neither Repair nor Preventive
Maintenance alone is the best
maintenance policy.

Therefore, Repair and Preventive

Maintenance need to be balanced
and optimized.
 Trade-off Between Repairs and PM
• At minimum level of PM, it is a remedial policy
• fix machines only when they break
• the cost of breakdowns, interruptions to production, and
repairs is high

• As the PM effort is increased, breakdown and repair cost

is reduced

• At some optimal point, the total maintenance cost (PM,

breakdown, and repair) reach a minimum
 How Speedy Should Repairs Be?

Cost

Minimum
Total Cost
of Repairs Total Costs
of Repairs
Cost of Repair
Crews & Shops,
Spare Parts, and
Standby Machines
Cost of
Interruptions to
Production
0
Slow Speed of Making Repairs Fast
 How Frequently the Maintenance Check is required?

Annual Cost

Minimum Total
Maintenance Cost
Total
Maintenance
Costs
Minimum
Level of Preventive
Preventive Maintenance
Maintenance Cost
Breakdown
and Repair
Cost
Degree of Preventive Maintenance
Case I:
Repair
 Repair Programs
• Objectives
• Get equipment back into operation as quickly as possible.

• Control cost of repairs crews.

• Control cost of the operation of repair shops.

• Control the investment in replacement spare parts.

• Control the investment in standby or backup machines.

• Perform required amount of repairs at each malfunction.

 Repair Crews and Standby Machines

• Repairs often performed on an emergency basis to:

• Minimize interruptions to production
• Correct unsafe working conditions
• Improve product/service quality

• In emergency situations:
• Specialists may work overtime
• Supervisor/engineers are nearby to collaborate
• Standby machines may be quickly put in operation
 How Speedy Should Repairs Be?

Cost

Minimum
Total Cost
of Repairs Total Costs
of Repairs
Cost of Repair
Crews & Shops,
Spare Parts, and
Standby Machines
Cost of
Interruptions to
Production
0
Slow Speed of Making Repairs Fast
 Decision Analysis in Repair Programs

• Determining the size of repair crews

• This is a repair-capacity decision
• Queuing method is often used
• Computer simulation is used when the assumptions of
queuing formulas do not apply

• Determining the number of standby machines to have

• Trade-off between cost of lost production time and cost of
machine storage, handling, ….
Example 1: Determining the Number of Repairmen
Solution:
Since Wt is 2 hours and λ is 12 machines per hour in the
problem, the equation can be applied to determine the
average service rate i.e. µ. This will give necessary average
service rate for repairing the entire welding machines.
1
2=
µ−12
µ=12+0.5=12.5

Now, we need to calculate the average service time repair a

specialist would take to repair the welding machine. This
can be easily done as
1
Average service time (St)=
ℎ𝑜𝑢𝑟𝑠 𝑝𝑒𝑟 𝑚𝑎𝑐ℎ𝑖𝑛𝑒 𝑝𝑒𝑟 𝑠𝑝𝑒𝑐𝑖𝑎𝑙𝑖𝑠𝑡
=1/0.75=1.333
So, using µ and St we can determine the number of repair
specialist the operations manager need to keep.

µ 12.5
Number of repair specialist= = = 9.375
𝑆𝑡1.333
= 10 (𝐴𝑝𝑝𝑟𝑜𝑥)
 Advantages of Letting Workers Repair
Their Own Machines

• May be more sensitive to potential malfunctions

• Increase flexibility

• Can make minor repairs faster

• Can avoid minor repairs by cleaning, lubricating, adjusting

& servicing machines

• Operate machines more carefully

 Case II:
Preventive Maintenance
[Condition Based Maintenance]
 Reasons for Preventive Maintenance
• Reduce the frequency and severity of interruptions due to
malfunctions

• Extend the useful life of equipment

• Reduce the total cost of maintenance by substituting PM

costs for repair costs

• Provide a safe working environment

• Improve product quality by keeping equipment in proper

adjustment
 Determining the Frequency of Performing PM

• First, compute the expected number of

breakdowns for each PM policy.

• Next, compute the expected breakdown cost,

preventive maintenance cost, and total cost for
each PM policy.

• Finally, identify the policy that minimizes the total

cost per unit of time (say, per week).
 How Frequently the Maintenance Check is required?

Annual Cost

Minimum Total
Maintenance Cost
Total
Maintenance
Costs
Minimum
Level of Preventive
Preventive Maintenance
Maintenance Cost
Breakdown
and Repair
Cost
Degree of Preventive Maintenance
 Expected Number of Breakdown
n
Bn  N( p n )  B(n 1) p1  B(n  2) p 2  ...  B1p (n 1)
1

where:
Bn = expected number of breakdowns for each of the
PM policies
pn = probability that a breakdown will occur between
PM inspections when PM is performed every n
periods
N = number of machines in group
Example 2: Determining the Frequency of PM
Solution:
Here, in the problem the value of n = 4. Based on the data given in the table,
we can calculate the expected number of breakdowns. The detailed calculation
is shown below.
Expected number of breakdown for week 1 (B1):
Here, N = 4 and p1=0.2
Expected number of breakdown for week 2 (B2):
Expected number of breakdown for week 3 (B3):

4.672
The calculated values are shown in the table below. We can see that PM
policy of every two weeks comes out to be the most economical which
minimize breakdown and weekly preventive maintenance i.e. Rs. 11640.

09-01-2023 32
Example 3: Determining the number of standby machines

09-01-2023 33
09-01-2023 34
Solution:

09-01-2023 35
09-01-2023 36
09-01-2023 37
 Case III:
Corrective Maintenance
• It is applied to improve reliability and
maintainability of machines.

• It is achieved by the implementation of

kaizens on:
• MTTR reduction

• MTBF improvement

• Uptime

09-01-2023 39
 Mean Time to Repair (MTTR)
• It is a basic measure of the maintainability of
the repairable items.

• It is expressed as the average time required to

repair machine.

MTTR = Total Breakdown Time / Total Breakdown

Incidences

09-01-2023 40
• It can be measured in minutes, hours or days.

• It should be very low or ideally it should be 0.

• While calculation of MTTR idle time is not

considered.

09-01-2023 41
Example 4: Calculating MTTR

Calculate MTTR for the month of December,

2021 in which there were 2 breakdowns of 65
and 75 minutes.

09-01-2023 42
Solution

As per the equation,

MTTR = Total Breakdown Time / Total Breakdown

Incidences

MTTR = (65 + 75 ) Minutes/ 2 = 140 Minutes/2

MTTR = 70 minutes = 1.16 hours

09-01-2023 43
Mean Time Between Failure (MTBF)
• It is nothing but the average time between
failures of the system/ machine/ equipment.

• It is given by
MTBF = (Total working time – Total Breakdown
Time)/ Total Breakdown Incidences

09-01-2023 44
• It can be measured in minutes, hours or
days.

• It should be very high and ideally it should be

infinite (∞).

• While calculation of MTBF idle time is not

considered.

09-01-2023 45
Example 5: Calculating MTBF

Calculate MTBF for the month of December

2021 in which there were 2 breakdowns of
111 and 222 minutes. Total working days in
December, 2021 were 24.

09-01-2023 46
Solution

Working time = Working days * 24 hours

= 24 days * 24 hours

= 576 hours

09-01-2023 47
Breakdown time = 111 + 222 minutes

= 333 minute

= 5.55 hours

09-01-2023 48
Therefore, as per the equation,

MTBF = (Total working time – Total Breakdown

Time)/ Total Breakdown Incidences

MTBF = (576-5.55)/ 2
= 285.22 hours

09-01-2023 49
 Uptime
• It is a measure of the time a machine has
been up without any downtime or
breakdown

• It is measured in %

• It is given by

Uptime = (Available Time – Breakdown Time)

* 100/ Available Time
09-01-2023 50
Example 6: Calculating Uptime
Suppose available time is 24 days and
breakdown time is 4 hours, then Uptime can
calculated as

Uptime = (Available Time – Breakdown Time)

* 100 /Available Time
Uptime = [(24*24)-4]*100/24*24
= 93.31%

09-01-2023 51
 Maintenance Policy and Operations Strategies

• PM program is essential to the success of a product-

focused positioning strategy

• If a machine breaks down, all downstream operations

will soon run out of parts to work on
 Scheduling Maintenance Activities

• PM and production are increasingly viewed as

being equally important

• In some plants, two 8-hour shifts are devoted to

production and one 4-hour mini-shift is devoted to
PM

• In other plants, three shifts are used for

production, but time allowances are factored into
production schedules for PM activities
 Maintenance Database Requirements
• Detailed records, or an ongoing history, must be
maintained on each machine
• Dates and frequency of breakdowns
• Descriptions of malfunctions
• Costs of repairs

• Machine specifications/checklists for PM inspection

• Computers generally used to maintain a database

• Also, data can be kept in plastic pocket on a machine

 Modern Approaches to Maintenance

• Maintenance at the source - workers have the

fundamental responsibility for preventing machine
breakdowns by conducting PM on their own machines

• Workers listen for indications of potential equipment

malfunction

• Maintenance-related records maintained by workers

• Use of quality circles

 Trends in Maintenance
• Production machinery is becoming more and more
complex and maintenance personnel must keep pace

• Special training programs to maintain worker skill

level

• Subcontracting service companies

• Production workers maintain own equipment

• Computer assistance in maintenance

 Maintenance Issues in Service Organizations

• Maintenance issues are not limited to manufacturing

• Transportation firms (airlines, trucking companies, package delivery

services, railroads) must keep their vehicles in top operating condition

• Highway departments must maintain roadways

• Office personnel are reliant on printers, computers, and fax machines

working properly

• As services become increasingly automated, service firms face more

and more maintenance issues
 Problems for Practice # 1

09-01-2023 58
 Problems for Practice # 2

09-01-2023 59
09-01-2023 60
 Problems for Practice # 3

09-01-2023 61
09-01-2023 62
09-01-2023 63
09-01-2023 64
09-01-2023 65
09-01-2023 66
09-01-2023 67
09-01-2023 68
 Session 3-4

Queue Management
 Need of queue management

• Direct impact on the organization’s performance.

• Product/Services costs may go up.

• Reputation in market may go down.

• Firm may lost business in long run.

Application of queue management

 Often called queuing theory

 Waiting lines are common situations

 Useful in both
manufacturing
and service
industries
Situation Arrivals in Queue Service Process
Supermarket Grocery shoppers Checkout clerks at cash
register
Highway toll booth Automobiles Collection of tolls at booth
Doctor’s office Patients Treatment by doctors and
nurses
Computer system Programs to be run Computer processes jobs
Telephone company Callers Switching equipment to
forward calls
Bank Customer Transactions handled by teller
Machine Broken machines Repair people fix machines
maintenance
Harbor Ships and barges Dock workers load and unload
Queue characteristics

 Characteristics of a Waiting-Line

 Arrival Characteristics

 Waiting-Line Characteristics

 Service Characteristics
1. Arrivals or inputs to the system
 Population size, behavior, statistical distribution

2. Queue discipline, or the waiting line itself

 Limited or unlimited in length, discipline of people or items
in it

3. The service facility

 Design, statistical distribution of service times
 Arrival characteristics

1. Size of the population

 Unlimited (infinite) or limited (finite)
2. Pattern of arrivals
 Scheduled or random, often a Poisson
distribution
3. Behavior of arrivals
 Wait in the queue and do not switch
lines
 No balking or reneging
Population of Arrivals Queue Service Exit the system
dirty cars from the (waiting line) facility
general
population …
Dave’s
Car Wash

Enter Exit

Arrivals to the system In the system Exit the system

Arrival Characteristics Waiting Line Service Characteristics

 Size of the population Characteristics  Service design
 Behavior of arrivals  Limited vs.  Statistical distribution
 Statistical distribution unlimited of service
of arrivals  Queue discipline
 Poisson Distribution

e-x
P(x) = for x = 0, 1, 2, 3, 4, …
x!

where P(x) = probability of x arrivals

x = number of arrivals per unit of time
 = average arrival rate
e = 2.7183 (which is the base of the
natural logarithms)
 e-x
Probability = P(x) =
x!

0.25 – 0.25 –

0.02 – 0.02 –
Probability

Probability
0.15 – 0.15 –

0.10 – 0.10 –

0.05 – 0.05 –

– –
0 1 2 3 4 5 6 7 8 9 x 0 1 2 3 4 5 6 7 8 9 10 11 x
Distribution for  = 2 Distribution for  = 4
 Waiting-line characteristics

 Limited or unlimited queue length

 Queue discipline - first-in, first-out

(FIFO) is most common

 Other priority rules may be used in

special circumstances
 Service characteristics

 Service time distribution

 Constant service time

 Random service times, usually a

negative exponential distribution
 Negative Exponential Distribution

Probability that service time is greater than t = e-µt for t ≥ 1

µ = Average service rate
1.0 –
Probability that service time ≥ 1

e = 2.7183
0.9 –
0.8 – Average service rate (µ) = 3 customers per hour
0.7 –  Average service time = 20 minutes per customer
0.6 –
0.5 –
0.4 –
Average service rate (µ) =
0.3 – 1 customer per hour
0.2 –
0.1 –
0.0 |– | | | | | | | | | | | |
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50 2.75 3.00
Time t (hours)
 Queue Systems
A doctor’s OPD office

Queue
Service Departures
Arrivals facility after service

Single-queue, single-server system

A visa office dual window drive-through

Queue
service service Departures
Arrivals facility facility after service

Single-queue, multi-server system

Situation at bank, post office and check-in counter service windows

Service
facility

Queue
Service Departures
Arrivals facility after service

Service
facility

Single-queue, Multi-server system

Some college registrations, electronics industry, heavy industries etc.

service service
Queue facility facility
Departures
Arrivals after service
service service
facility facility

Multi-server, multi-queue system

 Measuring Queue Performance
1. Average time each customer/job spends in the queue (Wq)

2. Average time each customer/job spends in the system (Ws)

3. Average number of customers/jobs in the queue (Lq)

4. Average number of customers/jobs in the system (Ls)

5. Probability that the service facility will be idle (P0)

6. Utilization factor for the system (p)

7. Probability of specific number of customers in system (Pn)

 Queuing Costs

Cost

Minimum
Total Total expected cost
cost
Cost of providing service

Cost of waiting time

Low level Optimal High level

of service service level of service
 Queuing Models

Deterministic Little’s
Queuing Law
Models
Queuing
Models M/M/1
Probabilistic M/M/S
Queuing M/D/1
Models M/G/1
 Deterministic Queuing Model: Little’s Law

09-01-2023 88
09-01-2023 89
 Probabilistic Queuing Models
All queuing models here assume:

 Arrival follows poisson distribution

 FIFO discipline

 A single-server or multi-server

 Server follows negative exponential

 Model Name Example
A Single-channel Information counter
system at department store
(M/M/1)

Number Number Arrival Service

of of Rate Time Population Queue
Queue Server Pattern Pattern Size Discipline
Single Single Poisson Exponential Unlimited FIFO
Model Name Example
B Multi-server Airline ticket
(M/M/S) counter

Number Number Arrival Service

of of Rate Time Population Queue
Queue Server Pattern Pattern Size Discipline
Single Multiple Poisson Exponential Unlimited FIFO
 Model Name Example
C Constant- Automated car
service wash
(M/D/1)

Number Number Arrival Service

of of Rate Time Population Queue
Queue Server Pattern Pattern Size Discipline
Single Single Poisson Constant Unlimited FIFO
 Model Name Example
D Constant- Emergency
service
(M/G/1)

Number Number Arrival Service

of of Rate Time Population Queue
Queue Server Pattern Pattern Size Discipline
Single Single Poisson General Unlimited FIFO
 Queuing Model-A: M/M/1
1. Arrivals are served on a FIFO basis and
every arrival waits to be served regardless
of the length of the queue

2. Arrivals are independent of preceding

arrivals but the average number of arrivals
does not change over time

3. Arrivals are described by a Poisson

probability distribution and come from an
infinite population
4. Service times vary from one customer to
the next and are independent of one
another, but their average rate is known

5. Service times occur according to the

negative exponential distribution

6. The service rate is faster than the arrival

rate
Example 1
Solution
Example 2

09-01-2023 100
Solution

09-01-2023 101
09-01-2023 102
09-01-2023 103
Queuing Model-B: M/M/S
09-01-2023 105
Example 3

09-01-2023 106
Solution

09-01-2023 107
09-01-2023 108
Example 4
Bank tellers and customers
 = 18, µ = 20

Solution
Lq
Utilization factor ρ = /µ = .90 Wq = 

Number of Number
service windows S in queue Time in queue
1 window 1 8.1 .45 hrs, 27 minutes
2 windows 2 .2285 .0127 hrs, ¾ minute
3 windows 3 .03 .0017 hrs, 6 seconds
4 windows 4 .0041 .0003 hrs, 1 second
Queuing Model-C: M/D/1
Example 5
Solution

09-01-2023 113
09-01-2023 114
 Queuing Model-D: M/G/1

09-01-2023 115
09-01-2023 116
Example 6

09-01-2023 117
09-01-2023 118
 Some other queuing models

09-01-2023 119
Other Queuing Approaches

 The single-queue models cover many queuing

situations

 Variations of the three queue systems are possible

 Multi-server models
exist for more
complex situations
 Practice Questions

1) A waiting line meeting the M/M/1 assumptions has an

arrival rate of 4 per hour and a service rate of 12 per hour.
What is the probability that the waiting line is empty?
Po = 1 - / = 1 - 4/12 = 8/12 or 0.667.

2) A waiting line meeting the M/M/1 assumptions has an

arrival rate of 10 per hour and a service rate of 12 per hour.
What is the average time a unit spends in the system and
the average time a unit spends waiting?
3) A crew of mechanics at the Highway Department garage
repair vehicles that break down at an average of  = 7.5
vehicles per day (approximately Poisson in nature). The
mechanic crew can service an average of  = 10 vehicles per
day with a repair time distribution that approximates an
exponential distribution.
a. What is the utilization rate for this service system?
b. What is the average time before the facility can return a
breakdown to service?
c. How much of that time is spent waiting for service?
d. How many vehicles are likely to be in the system at any one
time?
4) At the order fulfillment center of a major mail-order firm,
customer orders, already packaged for shipment, arrive at
the sorting machines to be sorted for loading onto the
appropriate truck for the parcel's address. The arrival rate at
the sorting machines is at the rate of 100 per hour following a
Poisson distribution. The machine sorts at the constant rate
of 150 per hour.
a. What is the utilization rate of the system?
b. What is the average number of packages waiting to be
sorted?
c. What is the average number of packages in the sorting
system?
d. How long must the average package wait until it gets
sorted?
e. What would Lq and Wq be if the service rate were
exponential, not constant?
09-01-2023 124
09-01-2023 125
09-01-2023 126
09-01-2023 127
 Certificate Course on
Operations Management and Analytics
(August 2022-January, 2023)

Program Faculty:
Dr. Abhijit Majumdar
R. Jaikumar Chair Professor for Decision Sciences
Department of Textile & Fibre Engineering
E-Mail: [email protected]
25-09-2022 1
 Module IV

o Process Capability
o Standard Normal Distribution
o Standard Deviation
o Processes in various Sigma levels
o Situations of Process Capability
o Process Capability Indices
o Root of Six Sigma Philosophy
o The Six Sigma Concept of Motorola
o Lean Six Sigma (LSS)
o Control charts for Quality.
o Control charts for Variables (X-bar and R Charts.
o Control charts for Attributes (c, u, p and np charts).
Six Sigma & Process Capability
(Session 1)
 Process Capability
o Process Capability can be defined as-
“Ability of a process to make a feature within its
tolerance”.

o Tolerance is the difference between specifications.

o Specifications are Upper Specification Limit (USL) and

Lower Specification Limit (LSL).

o Process Capability enables-

o successful manufacturing and sales
o prevents scrap, sorting and rework
o allows job to run smoothly
o major impact on cost and schedule
 What is a Standard Normal Distribution?
o A Normal Standard Distribution Curve is a symmetric
distribution where the area under the normal curve is 1 or
100%.
o The standard normal distribution have a mean (µ) of 0 and a
standard deviation of 1.
 Standard Deviation or Sigma
o The distance from the centre of the bell to each individual
measurement that makes up the bell curve is termed as the
Standard Deviation or Sigma.
 Why it is Six Sigma?
o The more number of standard deviations between
process average and acceptable process limits fits-
o The less likely that the process performs beyond the acceptable
process limits

o It causes a defect

o This is the reason why a 6σ (Six Sigma) process performs better

than 1σ, 2σ, 3σ, 4σ, 5σ processes.

o A 6σ process is reliable in almost all major situations

except some systems whose defects can cause
unrepairable consequences.
Processes in various sigma levels
 Processes in various sigma levels
o The red curve indicates a 2σ level of performance
o peak is very low
o fewer outputs are around the desired average
o the variation is from extreme left to extreme right of the figure

o The green curve indicates improvement from 2σ to 3σ

o the process variation reduces
o the process has a larger peak
o more outputs are around the desired average

o The blue curve shows performance increase from 3σ to 6σ

o the process centres between USL and LSL
o does not have much variation
o majority of process outputs are around the desired average
o causes less defects beyond the LSL and USL
 Situations

Three situations are possible-

(1) Process capability less than the tolerance

(2) Process capability equal to the tolerance

(3) Process capability greater than the tolerance

Case I: Process capability < tolerance
Case II: Process capability = tolerance
Case III: Process capability > tolerance
 Illustration
Location pins for work holding devices are ground to a diameter of
12.50 mm with a tolerance of ±0.05 mm. If the process is centered at
12.50 mm and the dispersion is 0.02 mm, what percent of the product
must be scrapped and what percent can be reworked?
 Illustration…continued

o USL = 𝝁 + 𝟎. 𝟎𝟓 = 𝟏𝟐. 𝟓𝟓 𝐦𝐦;

o LSL = 𝝁 − 𝟎. 𝟎𝟓 = 𝟏𝟐. 𝟒𝟓 𝐦𝐦;

𝑿𝒊 − 𝝁 𝟏𝟐.𝟒𝟓−𝟏𝟐.𝟓𝟎
o Z= = = - 2.50;
𝝈 𝟎.𝟎𝟐

o From the table “Area under normal curve” for Z value of -2.50,
area = 0.0062 or 0.62% scrap

o Since the process is centered between the specifications and a

symmetrical distribution is assumed, the rework percentage is
equal to scrap percentage = 0.62%
 Process Capability Indices
Capability indices express process capability and
performance.

o CP = Process Capability (simple indicator of process capability)

o Cpk = Process Capability Index

(adjustment of Cp for the effect of non-centred distribution)

o Pp = Process Performance (indicator of process performance)

o Ppk = Process Performance Index

(adjustment of Pp for the effect of non-centred distribution)

o k stands for ‘centralizing factor’ taking into consideration that data

may not be centred (hence, index shall be smaller).
 Mathematical formulations of
Process Capability Indices
𝐓𝐨𝐥𝐞𝐫𝐚𝐧𝐜𝐞 𝐔𝐒𝐋 −𝐋𝐒𝐋
Cp = =
𝟔𝛔 𝟔𝛔
where 𝟔𝝈 is natural process tolerance

ന 𝐗ന −𝐋𝐒𝐋
𝐔𝐒𝐋 − 𝐗
Cpk = minimum { ; }
𝟑𝛔 𝟑𝛔
where,
ഥ
𝑹
𝝈 = ; s.d. of the process wrt the observed characteristic;
𝒅𝟐

ഥ = Average Range;
𝑹

𝒅฀𝟐 = co-efficient depending on sample size

 Relation between Cp and Cpk

o The Cp value does not change as the process centre

changes.

o Cp = Cpk when the process is centred.

o Cpk is always ≤ than Cp.

o Cpk = 1 indicates that the process is producing products

conforming to specifications.

o Cpk < 1 indicates that the process is producing products

not conforming to specifications.
 Relation between Cp and Cpk …continued

o Cp < 1 indicates that the process is not capable.

o Cp = 𝟎 indicates the average is equal to one of the

specification limits.

o Negative Cpk indicates that the average is outside the

specifications.
Cpk – measure of a process within specification
Example
Short-term & Long-term Process Capability Indices
CP and Cpk
o are short term indices
o compute the index wrt the samples of the data (average
values of the samples)
o useful in order to know if, under the best conditions, the
process is capable of fitting into the specifications or not.

Pp and Ppk
o are used for long term analysis
o take into account whole process (all individual data within all
samples)
ഥ − 𝑿𝒊 )𝟐
σ𝒊 (𝑿
o 𝝈= ഥ = average value of characteristic;
where, 𝑿
𝑵−𝟏
𝑿𝒊 = individual measuring value;
N = total no. of measurements (i = 1,…,N)
 Root of Six Sigma Philosophy
o The roots of Six Sigma as a measurement standard can be
traced back to Carl Friedrich Gauss (1777-1855) who
introduced the concept of the normal curve.

o Six Sigma as a measurement standard in product variation can

be traced back to the 1920’s.

o Walter Shewhart showed that three sigma from the mean is

the point where a process requires correction.

o Credit for coining the term “Six Sigma” goes to a Motorola

engineer named Bill Smith (1929-1993) winning Malcolm
Baldrige Excellence Award in 1988.

o “Six Sigma” is a federally registered trademark of Motorola.

 The Six Sigma
o Standard deviation is the best measure of process
variability.

o The smaller the standard deviation, the less

variability in the process.

o If sigma ( 𝝈 ) is reduced to the point that the

specifications are at ±6, then 99.9999998% of the
product will be between specifications.

o Non-conformance rate will be 0.002 parts per

million(PPM) and Cp equals to 2.
Non-conformance rate when process is centered
Non-conformance rate when process is centered
Table: Non-conformance Rate and Process Capability
when the process is centered
Specification Percent Nonconformance Process
Limit Conformance rate Capability
(PPM) (Cp)
± 1𝝈 68.7 317300 0.33
± 2𝝈 95.45 485500 0.67
± 3𝝈 99.73 2700 1.00
± 4𝝈 99.9937 63 1.33
± 5𝝈 99.999943 0.57 1.67
± 6𝝈 99.9999998 0.002 2.00
Source: Besterfield D.H., “ Quality Control”, 7th Edition, Pearson Education, pp:236
 Non-conformance rate when process is off-
centered

Figure: Non-conformance rate when process is off-centered at ±1.5 𝝈

Non-conformance rate when process is centered
Table: Non-conformance Rate and Process Capability
when the process is off-centered at ±1.5 𝝈
Specification Percent Nonconformance Process
Limit Conformance rate Capability
(PPM) (Cp)
± 1𝝈 30.23 697700 -0.167
± 2𝝈 69.13 308700 0.167
± 3𝝈 93.32 66810 0.500
± 4𝝈 99.3790 6210 0.834
± 5𝝈 99.97670 2330 1.167
± 6𝝈 99.9996600 3.4 1.500
Source: Besterfield D.H., “ Quality Control”, 7th Edition, Pearson Education, pp:236
 The Six Sigma Concept of Motorola
o Six Sigma is a method for improving quality by removing defects
and their causes in business process activities.

o The method concentrates on those outputs which are important

to customers.

o The method translates these customer needs into measurable

requirements, the so called CTQs (Critical To Quality).

o A robust measurement system is established to obtain clean and

precise data relating to the process.

o Six Sigma is a business transformation methodology focusing on

the reduction of variation and elimination of defects by using
various statistical, data-based tools and techniques.
The Robust Design
 Six sigma DMAIC methodology
Motorola developed a five phase approach called ‘DMAIC
Model’ to achieve the highest level in the Six Sigma, i.e., 3.4
defects per million.

The five phases are:

o Define process goals in terms of key critical parameters (i.e. CTQ) on
the basis of customer requirements or Voice of Customer (VOC)
o Measure the current process performance in context of goals
o Analyze the current scenario in terms of causes of variations and
defects
o Improve the process by systematically reducing variation and
eliminating defects
o Control future performance of the process
 Six sigma DMAIC methodology
Strategic Steps Deliverables Tools used
Define Statement of Work o Gantt Chart/Time Line
(SoW) o Flow Chart/Process Map
o Quality Function Deployment (QFD)
Measure Base Line Figures o SIPOC (Suppliers, Inputs, Process, Outputs,
and Customers )
o IPO (Input Process-Output) diagram
Analyse Identified Root Causes o Cause-and-Effect Diagram
o 5-Why
o Scatter Diagram
o Regression
o ANOVA
Improve Selected Root Causes o Affinity Diagram
and Improvement o Hypothesis Testing
Implementation Plan o DoE
o Failure Mode Effect Analysis (FMEA)
Control Control Plan Charts & o Control Charts
Monitor Standard o Poka-Yokes
Operating Procedures o Standardization, Documentation
(SOP) Corrective Actions o Final Report and Presentation
 Lean Six Sigma
o Six Sigma when coupled with ‘Lean Principles’ is called ‘Lean Six
Sigma’ .

o This principle eliminates waste in process steps by using ‘Lean Tools’

which is based on Toyota Production System(TPS).

o Lean identifies three different kinds of wastes using Japanese

terminology-
o muda (waste of time and materials),
o mura (unevenness/variation), and
o muri (the overburdening of workers or systems).

o This enhances productivity gain and improvement of speed and

flow in the value stream.

o Kaizen is the process of continuous improvement involving

brainstorming sessions to fix problems.
 Six Sigma versus Lean
o Lean Principle gives a holistic view.

Lean uses
o value-stream mapping,
o balancing of workflow,
o streamline and improve the efficiency of processes, and
o increase the speed of delivery

o Six Sigma takes a more data-based and analytical approach.

o Six sigma employs the following tools-

o Voice Of the Customer (VoC)
o Measurement Systems Analysis (MSA)
o Statistical hypothesis testing
o Design of Experiments (DoE)
o Failure Modes and Effects Analysis (FMEA)
25-09-2022 35
Quality Control Charts
(Session 2 and 3)
 Contents
o Process and Quality Control
o Control Charts
o Types of Control Charts
o Variable Control Charts
o Attribute Control Charts

o Important terms of Control Charts

o Calculations of Control Limits
o Different types of control chart (X bar, R, np and p)
o Benefits of using Control Charts
Process Control
 Process

o Critical Factors
o Defect-free products
o Reliability and durability of parts & processes
o Customers’ expectations change rapidly
o More precise specifications
o Upgrading of existing processes and modern
technologies
 Process Control Tools & Techniques

o Brain storming
o Process mapping, flowcharting
o Statistical process control
o Control charts
o Pareto chart
o Cause and effect diagrams (Fishbone / Ishikawa)
o Check sheets
o Bar charts
o Histograms
o Scatter diagrams
Main objective of Process Control
 Objective of Control Chart

o Control charts that determine whether the process is

performing “in” or “out” of control.

o To distinguish between random (chance) variation and

assignable variation
 Random and Assignable Variations

o Random: Many factors cause small variations

o Assignable: One or a few factors cause large
variations
 Controlled Variation

Controlled variation is characterized by –

o A stable and consistent pattern of variation over

time, and is associated with common causes.

o A process operating with controlled variation has

an outcome that is predictable within the bounds
of the control limits.
Example of Controlled Variation
 Uncontrolled Variation

Uncontrolled variation is characterized by –

o A variation that changes over time and is associated

with special causes.

o The outcomes of this process are unpredictable.

o A customer may be satisfied or unsatisfied given this

unpredictability.
Example of Uncontrolled Variation
 Process Control in Decision Making

Process control can add value if used effectively in

decision making, for example:

o If the process is under control, the desired product

quality will be achieved.

o If process analysis shows a technology is not serving

its purpose, it can be modified or replaced to
improve product quality, save costs or improve
productivity.
Normal Distribution

 f ( x)dx  1

Normal Distribution
 Control Chart

o A statistical tool to determine if a process is in control.

o Control charts are the graphical tool for Statistical

Process Monitoring (SPM).

o Control charts are also known as Shewhart charts

or Process-Behaviour charts.
 History of Control Chart

o Developed in 1920’s.

o Developed by Walter Andrew Shewhart (1891-1967)-

an American physicist, engineer and statistician.

o Often referred to as the father of statistical control.

 Elements of Control Chart

o Horizontal axis for sample number or sub-group value.

o Vertical axis for sample statistics e.g. Target or Avg.

mean line, UCL and LCL Line.

o Target or Avg. Mean or Central Line – X bar.

o Upper control Limit line-UCL Line.

o Lower control Limit line-LCL Line.

o Plotting of the sample sub-group avg. value.

Example of Control Chart
 Idea behind Control Chart

If the observed value 𝛉 is such that LL < 𝛉 < UL,

o process is assumed to be in control
o Otherwise, the process is out of control

𝝈𝜽
∝ ∝
( ) ( )
𝟐 𝟐

𝝁𝜽
 Action and Warning Limits

o If the distribution of 𝜽 is normal, then ∝Τ𝟐 = 0.001

o 𝝁0.001 =3.09 (from Table entitled “Probability points

of the standard normal distribution”)

o Control limits are 𝝁𝜽 ± 3.09𝝈𝜽 , also known as

Action Limits.

o When ∝Τ𝟐 = 0.025, then Control limits are

𝝁𝜽 ± 1.96𝝈𝜽 , also known as Warning Limits.
Action and Warning Limits
 Out of Control and In Control Processes

o Process is “Out of control” if

one or more than one points lie outside the control
points.

o Process is “In control” if

- the sample points lie within the control limits.
- the points vary above and below the average value.
Is the Process Under Control?
Is the Process Under Control?
Is the Process Under Control?
Is the Process Under Control?
Is the Process Under Control?
 Attributes and Variables
o Let us consider Two articles supplied by a producer
to a consumer in lots or batches.

T-shirt Cotton yarn

Sampling by Attributes Sampling by Variables

(No of defects in T-shirts/
(A specific property of the item, like
No of defective T-shirt in a pack of
diameter is measured )
100 )
 Types of Control Chart

Two Types

o Variable Control Charts

o Attribute Control Charts

 Variable Control Charts

o Deals with measurable items.

o Examples-

(1) Weight;

(2) Height;

(3) Speed;

(4) Volume
 Types of Variable Control Charts

o X-Bar Chart
deals with an average value in a process

o R Chart
takes into count the range of the values.

o MA Chart
deals with the moving average of a process.
 Types of Attribute Control Charts

o np- Chart
deals with no of defectives in each set of samples.

o p - Chart
deals with percent/ fraction defectives in each set of
samples.

o c - Chart
deals with number of defects per unit in sample set.
 Calculation of Control Limits

o Control Limits are calculated by-

o Estimating the standard deviation, σ, of the sample

data

o Multiplying that number by three

o Adding (3 x σ to the average) for the UCL and

o subtracting (3 x σ from the average) for the LCL

o Mathematically,
CL = Average ± 3 x σ
How to select a Control Chart
Control Chart for Discrete Data
 Phase I:
Control Charts for Continuous Data
 Control Chart for Continuous Data

o x̅̅ Chart
X-bar Chart is used to evaluate consistency of
process averages by plotting the average of each
subgroup.

o R Chart
Range Chart plot the ranges of each subgroup.
 Example of X-bar and R Chart
Example
In a garment manufacturing Industry, the R&D team analysed and
collected 20 sets of T-Shirt sizes with a subgroup size of 4. Compute
X-bar and R values.
Example of X-bar and R Chart
 Example of X-bar Chart

Solution:
X-bar Chart
o 𝐂𝐋 = 𝟐𝟗. 𝟗
ഥ
𝝈
ന+𝟑 ×
o UCL = 𝒙 ഥ
ന + 𝑨𝟐 × 𝑹
= 𝒙
√𝒏
= 29.9 + 0.729 × 11.85 = 38.538
ഥ
𝝈
ന−𝟑 ×
o LCL = 𝒙 = 𝒙ന − 𝑨𝟐 × 𝑹ഥ = 21.261
√𝒏
(0.729 is the value of A2 for subgroup 4; table provided in slide no. 30)
Constants
Example of X-bar Chart
 Example of R Chart

R Chart
o UCL𝑹ഥ = D4 × 𝑹 = 2.282 × 11.85 = 27.041

o LCL𝑹ഥ = D3 × 𝑹 = 0
 Example of R Chart

Graphical Interpretation
From the both X bar and R charts it is clearly evident that most of
the values are out of control, hence the process is not under control.
 Phase II:
Control Charts for Discrete Data
 Properties of Binomial Distribution

1. Experiment consist of n identical trials

2. The trials are independent
3. Each trial has only two outcomes
4. The probability of one outcome is p and the other is
q=1-p
5. We are interested in r, the number of success
observed during the n trials.
6. A binary outcome: generally called “success” and
“failure”
 Binomial Distribution

Note the general pattern emerging  if you have only two

possible outcomes in n independent trials, then the probability
of exactly r “successes”=
n = number of trials

n r nr
 r  p (1  p)
  1-p = probability
of failure
X=#
p=
successes probability of
out of n success
trials
 Binomial Distribution
Take the example of 5 coin tosses. What’s the probability
that you get exactly 3 heads in 5 coin tosses?

Outcome Probability
THHHT (1/2)3 × (1/2)2
HHHTT (1/2)3 × (1/2)2
TTHHH (1/2)3 × (1/2)2
HTTHH (1/2)3 × (1/2)2 The probability
ways to
5
HHTTH (1/2)3 × (1/2)2 of each unique
 
arrange 3 HTHHT (1/2)3 × (1/2)2 outcome (note:
heads in THTHH (1/2)3 × (1/2)2
 3 5 trials HTHTH (1/2)3 × (1/2)2
they are all
equal)
HHTHT (1/2)3 × (1/2)2
THHTH (1/2)3 × (1/2)2 10
arrangements × (1/2)3 × (1/2)2 =0.3125

5C3 = 5!/3!2! = 10
Binomial Distribution

P ( x)  Crn p r q n  r
n!
C 
n

r !(n  r )!
r

  np
  npq
 Poisson Distribution

• When there is a large number of

trials, but a small probability of
success, binomial calculation becomes
impractical

o Example: Number of deaths of

football player during a match.

o Project delayed but actual cost at

completion is less than budgeted
cost Simeon D. Poisson (1781-
1840)
 Poisson Distribution

r 
e
P(r ) 
r!
mean  
var :   
2

Where, r = number of rare events per unit of time

and e = 2.71828…
 Control Chart for Discrete Data

o c – Chart (no of defects)

identifies the total count of defects per unit (c) that occurred
during the sampling period (constant no/amount of samples).

It assumes the underlying data approximating the Poisson

distribution.

o u - Chart
tracks a sample having more than one defect when the number
of samples of each sampling period may vary significantly.
 Example of c – Chart (no of defects)
Example
Fabric lot of 500 m length is being teste everyday. Defects in each
same are recorded. Based on the given data, draw the appropriate
control chart and comment on the state of control.

Sample No. 1 2 3 4 5 6 7 8 9 10
No. of defects in 12 14 16 18 16 14 12 12 32 16
the sample (c)
Sample No. 11 12 13 14 15 16 17 18 19 20
No. of defects in 18 16 14 12 16 18 12 19 18 21
the sample (c)
 Example of c - Chart
Solution:
o No. of lots (k) = 20

o σ 𝒄 = 𝟑𝟐𝟔
𝑻𝒐𝒕𝒂𝒍 𝒏𝒐.𝒐𝒇 𝒅𝒆𝒇𝒆𝒄𝒕𝒔 σ𝒄 𝟑𝟐𝟔
o 𝒄ത = = = = 16.3
𝑻𝒐𝒕𝒂𝒍 𝒏𝒐.𝒐𝒇 𝒍𝒐𝒕𝒔 𝒌 𝟐𝟎

o UCL = 𝒄ത + 𝟑 × 𝒄ത = 28.412

o LCL = 𝒄ത − 𝟑 × 𝒄ത = 4.188
 Example of c - Chart

Graphical Interpretation
The average number of defects per lot is 16.3. Sample 9 is outside of
the control limit. Hence the process is out of control.
 Example of u - Chart
Example
Twenty samples of carpets are inspected for varying sample size and the
number of defects in each sample is noted in the following table:
Sample No. of No. of Sample No. of No. of
No. carpets defects No. carpets defects
inspected inspected
1 25 12 11 20 9
2 20 5 12 20 12
3 25 7 13 15 14
4 15 7 14 25 6
5 25 10 15 20 7
6 15 4 16 25 12
7 20 6 17 15 5
8 15 2 18 25 6
9 15 4 19 15 8
10 25 10 20 25 4
 Example of u - Chart
Solution:
o No. of carpets inspected (n) = 405
o No. of defects found (c) =150
𝑻𝒐𝒕𝒂𝒍 𝑵𝒐.𝒐𝒇 𝒅𝒆𝒇𝒆𝒄𝒕𝒔 𝒇𝒐𝒖𝒏𝒅 σ𝒌
𝒊=𝟏 𝒄𝒊 𝟏𝟓𝟎
ഥ =
o 𝒖 =σ𝒌 = = 0.370
𝑻𝒐𝒕𝒂𝒍 𝑵𝒐.𝒐𝒇 𝒄𝒂𝒓𝒑𝒆𝒕𝒔 𝒊𝒏𝒔𝒑𝒆𝒄𝒕𝒆𝒅 𝒊=𝟏 𝒏𝒊 𝟒𝟎𝟓

𝟏 𝒌 𝟏
ഥ =
o 𝒏 σ𝒊=𝟏 𝒏𝒊 = × 405 = 20.25
𝒌 𝟐𝟎

o 𝐂𝐋 = 𝟎. 𝟑𝟕𝟎
ഥ
𝒖
ഥ+𝟑 ×
o UCL = 𝒖 = 0.7755
ഥ
𝒏
ഥ
𝒖
ഥ−𝟑 ×
o LCL = 𝒖 = - 0.0355 ≈ 0
ഥ
𝒏
 Example of u - Chart

Graphical Interpretation
The point corresponding to sample 13 lies outside the upper control
limit. Therefore, the process is not under statistical control with
respect to the number of defects per carpet.
 Control Chart for Defectives

o np - Chart (For defectives)

identifies the total count of defective units (one unit
may have one or more defects) with a constant
sampling size.

Mathematically,

𝑻𝒐𝒕𝒂𝒍 𝑵𝒐.𝒐𝒇 𝒅𝒆𝒇𝒆𝒄𝒕𝒊𝒗𝒆 𝒊𝒕𝒆𝒎𝒔

np =
𝑵𝒐.𝒐𝒇 𝒍𝒐𝒕𝒔 𝒔𝒂𝒎𝒑𝒍𝒆𝒅
 Control Chart for Fraction Defectives

o p – Chart (for fraction defectives)

shows the number of tracked failures (np) divided by the
number of total units (n).

Mathematically,
𝑵𝒐.𝒐𝒇 𝒅𝒆𝒇𝒆𝒄𝒕𝒊𝒗𝒆 𝒊𝒕𝒆𝒎𝒔
Fraction or Proportion Defective (p) =
𝑻𝒐𝒕𝒂𝒍 𝒏𝒐.𝒐𝒇 𝒊𝒕𝒆𝒎𝒔 𝒊𝒏𝒔𝒑𝒆𝒄𝒕𝒆𝒅
 Example of np – Chart (for defectives)
Example
In a LED bulb manufacturing factory the supervisor draws randomly
constant sample size of 200 bulbs every hour and reported the
number of defective bulbs for each lot. Based on the given data,
prepare the control chart for the number of defectives and
determine process is in statistical control.

Sample No. 1 2 3 4 5 6 7 8 9 10
No. of 4 8 6 6 4 8 2 1 9 6
defectives (np)
Sample No. 11 12 13 14 15 16 17 18 19 20
No. of 8 1 2 9 4 3 9 6 2 7
defectives (np)
 Example of np - Chart
Solution:
o No. of lots (k) = 20

o σ 𝒏𝒑 = 𝟏𝟎𝟓; σ 𝒏 = 𝟒𝟎𝟎𝟎
𝑻𝒐𝒕𝒂𝒍 𝒏𝒐.𝒐𝒇 𝒅𝒆𝒇𝒆𝒄𝒕𝒊𝒗𝒆𝒔 σ 𝒏𝒑 𝟏𝟎𝟓
ഥ =
o 𝒑 = = = 0.02625
𝑻𝒐𝒕𝒂𝒍 𝒏𝒐.𝒐𝒇 𝒔𝒂𝒎𝒑𝒍𝒆𝒔 σ𝒏 𝟒𝟎𝟎𝟎
𝑻𝒐𝒕𝒂𝒍 𝑵𝒐.𝒐𝒇 𝒅𝒆𝒇𝒆𝒄𝒕𝒊𝒗𝒆 𝒊𝒕𝒆𝒎𝒔 σ 𝒏𝒑 𝟏𝟎𝟓
𝒑 =
o nഥ = = = 5.25
𝑵𝒐.𝒐𝒇 𝒍𝒐𝒕𝒔 𝒔𝒂𝒎𝒑𝒍𝒆𝒅 𝒌 𝟐𝟎

𝒑 + 𝟑 × nഥ
o UCL = nഥ ഥ) = 12.033
𝒑(𝟏 − 𝒑

𝒑 − 𝟑 × nഥ
o LCL = nഥ ഥ) = -1.533 ≈ 0
𝒑(𝟏 − 𝒑
 Example of np - Chart

Graphical Interpretation
In the above example all the points or number of defective bulbs
within each lot is between the UCL and LCL.
 Example of p – Chart (Fraction defectives)
Example

A factory manufactures cricket balls. To check the quality of the balls, the
manufacturer selected 20 samples of same size 100 from the manufacturing
process time to time. He/she visually inspected each selected ball for certain
defects. After the inspection, he/she obtained the following data:

Sample No. 1 2 3 4 5 6 7 8 9 10
Proportion 0.10 0.04 0.08 0.15 0.08 0.00 0.01 0.05 0.05 0.08
defective
Sample No. 11 12 13 14 15 16 17 18 19 20
Proportion 0.10 0.00 0.06 0.05 0.03 0.20 0.05 0.07 0.01 0.08
defective

Estimate the proportion defective of the process. Does the process appear to be
under control with respect to the proportion of defective balls?
 Example of p - Chart
Solution: 𝒑 + 𝟑 × nഥ
nഥ ഥ)
𝒑(𝟏 − 𝒑

𝟏 𝟏
ഥ=
o 𝒑 σ 𝒑= × 1.29 = 0.065
𝒌 𝟐𝟎

ഥ = 0.065;
o CL = 𝒑

ഥ (𝟏−ഥ
𝒑 𝒑)
ഥ+𝟑 ×
o UCL = 𝒑 = 0.140;
𝒏

ഥ (𝟏−ഥ
𝒑 𝒑)
𝒑−𝟑 ×
o LCL =ഥ = - 0.01
𝒏
Contd….
 Example of p - Chart

Graphical Interpretation
The points corresponding to sample numbers 4 and 16 lie outside the control
limits. So, the process is not under statistical control with respect to the proportion
of defective balls.
 Benefits of using a Control Chart
o Control Chart is a powerful tool that can be used in
Problem Solving Techniques –

o Seven Quality Control (7 QC Tools)

o 8-Disciplines Problem Solving Methodology

o Various phases of Six Sigma-DMAIC to reduce process

variations

o Lean Manufacturing for 3M Waste elimination from

processes.

o PDCA Deming Cycle for Continuous improvement in

product and processes.
7 QC Tools
DMAIC-Model
PDCA-Model (Deming Wheel)
25-09-2022 108
 Certificate Course on
Operations Management and Analytics
(August 2022-January, 2023)

Program Faculty:
Dr. Abhijit Majumdar
R. Jaikumar Chair Professor for Decision Sciences
Department of Textile & Fibre Engineering
E-Mail: [email protected]

22-09-2022 1
 Module V

• Need of Inventory Management.

• Inventory Management Policies .
• P-Q and EOQ Inventory Models.
• Newspaper Vendor Inventory Model.
• ABC and VED Inventory Classification.
Inventory Policy and EOQ Model
(Session 1)
 Review Policies

• Continuous review policy (Q policy)

• Periodic review policy (P policy)

P policy Q policy
Constant Review period Order quantity
Variable Order quantity Order placement time
Calculate Base stock level Reorder point
 Continuous review policy:
constant demand & constant lead time

IP IP IP

Order Order Order Order

received received received received
On-hand inventory

Q Q Q

OH OH OH

R
Order Order Order
placed placed placed

L L L Time
 Continuous review policy:
constant demand & constant lead time

• Lead time has NO impact if the demand is deterministic and at a

constant rate.
• Safety stock is not needed
• Lead time under uncertain demand requires even a larger safety
stock.
 Continuous review policy:
variable demand & constant lead time
IP
IP
Order
Order
Order received
received
received Order
received
On-hand inventory

Q
Q Q
OH

R
Order Order Order
placed placed placed

Time
L L L
 Continuous review policy:
variable demand & constant lead time

Reorder level R has two components

o Average inventory needed during lead time
o Safety stock: to meet the variability of demand during lead time
 Continuous review policy:
variable demand & constant lead time

AVG  average demand/ day

SDD = standard deviation of daily demand
L = lead time (days)
Assumption: Demand variation in different days are independent of each other.

Reorder level =R
= Average inventory needed duirng lead time
+ safety stock
=AVG  L  Z  SDD  L

The second tem becomes zero when there is no demand variation.

Service level and safety factor

Service level (%) Z

50 0
90 1.28
95 1.65
99 2.33
99.9 3.1
 Continuous review policy:
variable demand & constant lead time
 Continuous review policy:
variable demand & constant lead time

The order quantity = Q

The quanity of inventory before the receipt = safety stock
=Z  SDD  L

The quanity of inventory after the receipt

= Order quantity + safety stock
=Q + Z  SDD  L

Q
So, average inventory =  Z  SDD  L
2
 Continuous review policy:
variable demand & variable lead time
AVG  average demand
L  average lead time
SDD = standard deviation of daily demand
SDL = standard deviation of lead time

The safety stock

 
2
=Z  AVG  SDL  SDD 2
L

Reorder level =R

 AVG  SDL 
2
= AVG  L +Z   SDD 2  L
 Continuous review policy: Cost model

Calculating total systems costs

Total cost = Annual cycle inventory holding cost

+ Annual ordering cost
+ Annual safety stock holding cost

D Q
TC   Co   Cc  Safety stock  Cc
Q 2
Periodic review policy
 Periodic review policy

• Inventory is reviewed after a fixed period.

• Inventory policy is characterized by a single parameter
• Base stock level
• Order is made so that to raise inventory position to
base stock level.
• When the order is placed, inventory level is increased
to protect any shortages till the next order arrives.
 Periodic review policy

AVG  average demand/ day

SDD = standard deviation of daily demand
L = lead time (days)
R  review period (days)

Base stock level = AVG   L  R   safety stock

=AVG   L  R   Z  SDD  L  R

When the order is placed, inventory level is increased to protect any

shortages till the next order arrives.
 Periodic review policy

Base stock level = AVG   L  R   safety stock

=AVG   L  R   Z  SDD  L  R

The inventory is received after L days.

So, invenory level after the receipt =AVG  R  Z  SDD  L  R

Inventory level before receipt =Z  SDD  L  R

AVG  R
Therefore, average level of inventory =  Z  SDD  L  R
2
 P-Q Problem

1. The annual demand for a product is 15600 units. The weekly

demand is 300 units with standard deviation of 90. The cost to
place order is 31.2 Rs and lead time is 4 weeks. The annual
inventory carrying cost is 0.10 Rs per unit. Find the reorder
point to provide 98% service level (Z=2.05). (Answer: 1569).
Suppose the production manager is asked to reduce the safety
stock by 50%. What will be the new service rate (Answer:
83%)
2. Daily demand for a product is 100 units with a standard
deviation of 25 units. The review period is 10 days and the
lead time is 6 days. At the time of review, there are 50 items
in the stock. If 98% service rate is desired, how many units
should be ordered? (Ans 1750-1755).
EOQ Models
 Introduction

o Inventory is a stock of items kept on hand to

meet customer demand.
o A certain level of inventory is maintained to
meet anticipated demand.
o For uncertain demand, safety (buffer) stocks
are kept on hand.
o Additional stocks are sometimes built up to
meet seasonal or cyclical demand.
 Inventory carrying cost

It is the Costs of holding items in storage.

o Vary with level of inventory and sometimes
with length of time for which the inventory is
held.
o Include facility operating costs, record keeping,
interest, etc.
o Assigned on a per unit basis per time period, or
as percentage of average inventory value
(usually estimated as 10% to 40%).
 Ordering cost

It is the costs of replenishing stock of inventory.

o Expressed in rupees/ dollars/ euros per order,
independent of order size.
o Total ordering cost vary with the number of
orders made.
o Include purchase orders, shipping, handling,
inspection, etc.
 Stock-out or shortage cost

It is the estimated costs associated with insufficient

inventory.
o Result in permanent loss of sales and profits.
o Loss of reputation or brand value.
o Sometimes penalties involved (service level
agreement SLA)
o Delivery may be delayed if customer is internal.
 Inventory control

o Two types of systems are followed –

o continuous monitoring (fixed-order quantity)
o periodic monitoring (fixed-time).
o In a continuous monitoring system, an order is
placed for the same constant amount (EOQ) when
inventory decreases to a specified level (reorder
point).
o In a periodic monitoring system, an order is placed
for a variable amount after a specified period of
time.
 EOQ model
o Economic order quantity, or economic lot size, is the
quantity ordered when inventory decreases to the
reorder point.
o Purpose of the EOQ model is to determine the optimal
order size that will minimize total inventory costs.

o Three variants of EOQ:

o Basic EOQ model (purchase or instantaneous receipt)
o EOQ model with non-instantaneous receipt
(manufacturing model)
o EOQ model with shortages
 Basic EOQ model

o Demand is known with certainty and is

relatively constant over time.
o No shortages are allowed.
o Lead time for the receipt of orders is constant.
o The order quantity is received all at once and
instantaneously.
Basic EOQ model
 Basic EOQ model (purchase model)

Let annual demand is = D

Economic order quantity = Q
Average inventory = Q/2
Carrying cost per unit per year = Cc
Annual inventory carrying cost (Tc)= Cc×Q/2
 Basic EOQ model
Ordering cost= Co
Number of orders = D/Q
Total ordering cost = Number of orders × cost of one order
Total ordering cost (To)= (D/Q) × Co

Total cost (TC) = Total ordering cost (To) +

Total inventory carrying cost (Tc)
= (D/Q) × Co + Cc×Q/2
Basic EOQ model
D Q
TC  Co  Cc
Q 2

TC D Co  Cc  0

Q Q2 2

or, D Co  Cc
Q2 2

or, Q  2 DCo
Cc
 Basic EOQ model
Optimum number of orders: Minimum total inventory cost:
D D
 
Q 2 D.Co 1 D
=  Q  Cc   Co
Cc 2 Q

D.Cc 1 2 D  Co D
    Cc   Co
2Co 2 Cc 2 D  Co
Cc
Time between orders:
D  Co  Cc D  Co  Cc
 
Q 2Co 2 2
 
D D.Cc
 2 D  Co  Cc
Basic EOQ model
Non instantaneous receipt (manufacturing model)

p is the rate of receipt and

d is the rate of demand

 d  d
Q 1-  Q 1- 
Order receipt period  t1  
p Q
Depletion period  t2  
 p
 p-d p d

 d
Q 1- 
1 1 1 Q
Cycle time  t1  t2   
Q p
 Q    
p d  p d p d
 Non instantaneous receipt

 
Maximum inventory level  Q 1 dp 
 

Q  
Average inventory level  1 p 
 d
2 

Q  
Total carrying cost  Cc  1 p 
 d
2 

 
Total annual inventory cost  Co  D  Cc  Q 1 dp 
Q 2 

p is the rate of receipt and

d is the rate of demand
 Non instantaneous receipt

Q  
Cc 1 p   Co D at lowest point of total cost curve
 d
2  Q

2Co D
Optimal order size: Qopt 
Cc (1 d / p)
 Non instantaneous receipt: example
p is the rate of receipt and
Co  $150 d is the rate of demand
Cc  $0.75 per unit
D  11680 m per year  11680/365  32 m per day
p  160 m per day

2Co D 215011680
Optimal order size: Qopt   
  
Cc 1 dp  0.751 32 
   160 
 2416.6 m

Calculate: Total inventory cost

Production run period
Depletion period
 Non instantaneous receipt: example

Q  
Total minimum annual inventory cost  Co   Cc  1 p 
D  d
Q 2 

2,416.6  
150 11680  0.75  1
 32   $1450
2416.6 2  160 

Production run length  Q  2416.6 15.1 days

p 160
 
Q 1 d   32 
2416.71 
Depletion time period =  p
=  160  =60.4
d 32
   
Maximum inventory level  Q 1 dp   2416.61 32  1932.8 yd
   160 

p is the rate of receipt and

d is the rate of demand
EOQ with planned shortages
EOQ With Price Breaks
 Price breaks

• Often, the supplier encourages larger order sizes by

providing price discounts. Such quantity discounts are
common for many consumer products.

• The text discusses two types of discounts.

o All units: discount is applied to all units.

o Incremental units: discount is applied to the
additional units beyond the breakpoint.

41
 EOQ with Price breaks
• Suppose that the charge for photocopying is 1 rupee
per copy for 0-9 copies and 0.8 rupee per copy for 10-
49 copies. If 12 copies are made the total charge is
computed as follows:

– All units: each of the 12 units is charged 0.8 rupee.

So, the total charge = 0.8 12 = 9.6 rupee

– Incremental units: the first 9 copies is charged 1

rupee each and the remaining 3 units are charged
0.8 rupee each. So, the total charge = 1 9 + 0.8 3
=11.24 rupee.

42
 EOQ Model with Price Breaks (All Units)

Total Charge (All Units)

4.5
4
3.5
Total Charge $

3
2.5 $0.10/unit
2 $0.08/unit
1.5
1
0.5
0
0 10 20 30 40
Quantity Ordered 43
 EOQ with Price breaks (all units)
• Assumptions
– Demand occurs at a constant rate of D items per
year.
– Ordering Cost is Oc per order.
– Purchase Cost is c1 per item if the quantity ordered is
between 0 and x1, c2 if the order quantity is between x1
and x2, etc.
– Holding Cost is Cc per item in inventory per year
(note holding cost is based on the cost of the item,
).
– Delivery time (lead time) is constant.
 EOQ with Price breaks (all units)

Total cost = Ordering cost +

Inventory holding cost +
product purchase cost

D Q
TC   Co   Cc  D  Ci
Q 2
 EOQ with Price breaks (all units)

• The total annual cost function is computed and plotted

in the next slide for the following example:

D  1, 092
Cc  25%  0.25ci
Oc  20

Order size Unit cost

0-399 c1  $3.20
400-899 c2  $2.976
900  c3  $2.88
 Total cost graph with price breaks
Total Annual Cost (All units)

4200
Total Annual Cost

4000
3800 $3.2/unit
$2.976/unt
3600
$2.88/unit
3400 Feasible cost
3200
3000
100 400 700 1000 1300
Quantity Ordered 47
 Procedure

1. If the first feasible EOQ is for the lowest price, then

this quantity is best.

2. Otherwise, calculate the total cost for the first

feasible EOQ and also calculate the total cost at
EACH price break lower than the price break
associated with the first feasible EOQ.
 Procedure

• Beginning with lowest price, calculate the EOQ

for each price level until a feasible EOQ is found.
It is feasible if it lies in the range corresponding
to its price.
• If the first feasible EOQ found is for the lowest
price level, this quantity is best.
• Otherwise, calculate the total cost for the first
feasible EOQ and for the larger price break
quantity at each lower price level. The quantity
with the lowest total cost is optimal.
 Example

D= 10000 units/ year

Oc= Rs 20 per order
Cc= 20% of Ci (storage, interest, obsolescence)

Ci=
Quantity 0-499 Unit price: 5.00 Rs
Quantity 500-999 Unit price: 4.50 Rs
Quantity >=1000 Unit price: 3.90 Rs
 Example: feasible EOQ

2 D  Co
EOQ 
Cc

2 10000  20
EOQ3.9   716 (Not feasible)
3.9  0.2

2 10000  20
EOQ4.5   667 (Feasible)
4.5  0.2

2 10000  20
EOQ5   632 (Not feasible)
5  0.2
 Example: total cost calculation

Q=632 when Q=667 when Q=716 when Price break

Ci=5.00 Ci=4.50 Ci=3.90 1000

Holding cost = Not feasible 667/2×4.5×0.2 Not feasible 1000/2×3.9×0.

Q/2×Cc =Rs 300.15 2
=Rs 390
Ordering cost = 10000/667×20 10000/1000×2
D/Q×Oc =Rs 299.85 0
=Rs 200
Item purchase 10000×4.5 10000×3.9
cost = D×Ci =Rs 45000 =Rs 39000

Total cost Rs 45600 Rs 39590

22-09-2022 53
Newspaper Vendor Model
(Session 2)
 News vendor problem

• Demand for the product is uncertain

(probabilistic)
• How much product should be stocked
• Marginal profit (for each unit sold)
• Marginal loss (for the ones that are bought
and not sold)
• Goal: Maximize expected profit
 Application areas

• Perishable items
• Valentine day’s flower
• Christmas gift
• Fashion cloths
• Newspapers
• Cafeteria food
 Application areas

Newsvendor selling Times of India

Sells the papers for Rs. 1.00
Buys them for 70 paise.
Leftover papers sold at 20 paise each.
He will definitely sell at least 35 papers, but no more than
40.
35 papers with probability 0.10
36 papers with probability 0.15
37 papers with probability 0.25
38 papers with probability 0.25
39 papers with probability 0.15
40 papers with probability 0.1
 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36

37

38

39

40

>40
 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36 0.15 0.9 0.10 0.27 0.05 0.22

37

38

39

40

>40
 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36 0.15 0.9 0.10 0.27 0.05 0.22

37 0.25 0.75 0.25 0.225 0.125 0.10

38

39

40

>40
 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36 0.15 0.9 0.10 0.27 0.05 0.22

37 0.25 0.75 0.25 0.225 0.125 0.10

38 0.25 0.5 0.50 0.15 0.25 -0.10

 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36 0.15 0.9 0.10 0.27 0.05 0.22

37 0.25 0.75 0.25 0.225 0.125 0.10

38 0.25 0.5 0.50 0.15 0.25 -0.10

39 0.15 0.25 0.75 0.075 0.375 -0.30

 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36 0.15 0.9 0.10 0.27 0.05 0.22

37 0.25 0.75 0.25 0.225 0.125 0.10

38 0.25 0.5 0.50 0.15 0.25 -0.10

39 0.15 0.25 0.75 0.075 0.375 -0.30

40 0.10 0.1 0.9 0.03 0.45 -0.42

 Marginal analysis

Demand = X Probability P= (1-P) = Marginal Marginal Expected

that probability probability profit loss NET profit
demand is = of selling of of NOT @0.3 Rs @0.5 Rs from storing
X xth unit selling xth Xth unit
unit
<35 0

35 0.10 1 0 0.30 0 0.30

36 0.15 0.9 0.10 0.27 0.05 0.22

37 0.25 0.75 0.25 0.225 0.125 0.10

38 0.25 0.5 0.50 0.15 0.25 -0.10

39 0.15 0.25 0.75 0.075 0.375 -0.30

40 0.10 0.1 0.9 0.03 0.45 -0.42

>40 0 0 1 0 0.50 -0.50

Marginal profit declines as no. of units increases

 Marginal analysis model
c: cost per unit
r: selling price per unit
s: salvage value per unit
MP: marginal profit (cost of understocking) from selling a
stocked unit = r-c
ML: marginal loss (cost of overstocking) from NOT selling a
stocked unit = c-s
x: the number of units stocked
P(x): the cu. probability that the xth unit is sold

P( x)  MP  1  P( x)  ML
ML
or, P( x) 
ML  MP
 Marginal analysis model

ML
P( x) 
ML  MP
0.50
  0.625
0.50  0.30

x    z 
 37.5  z  1.44
 37.5  0.32  1.44
 37.04
The value of z can be found from z table.

When x is lower than mean, probability of selling xth unit is higher than 50%.
So, z is negative in this case.
z table
22-09-2022 68
ABC & VED Analysis
(Session 3)
 Pareto principle

VILFREDO PARETO (1848-1923)

– 20% of population owns 80% of nations wealth

– 20% of students cause 80% of problems

– 20% of items accounts for 80% of firms

expenditure
 ABC analysis

• ABC analysis is an inventory categorization method

which consists in dividing items into three categories.

– A being the most valuable items

– B being the items with moderate value
– C being the least valuable ones.

• Critical few and trivial many

 ABC analysis

• The annual consumption value is calculated with

the formula:

Annual demand × item cost per unit

• The inventory manager can identify inventory hot

spots, and separate them from the rest of the
items, especially those that are numerous but not
that costly.
 ABC analysis: levels

Percentage of items Percentage value of Inventory policy

annual usage

Close day to day

Class A items About 20% About 80%
control

Class B items About 30% About 15% Regular review

Infrequent
Class C items About 50% About 5%
review

About 10 % of materials consume 70 % of resources

About 20 % of materials consume 20 % of resources
About 70 % of materials consume 10 % of resources
 Step 1: calculate annual spending

Item number Unit cost Annual demand Total cost per year

101 5 48,000 240,000

102 11 2,000 22,000

103 15 300 4,500

104 8 800 6,400

105 7 4,800 33,600

106 16 1,200 19,200

107 20 18,000 360,000

108 4 300 1,200

109 9 5,000 45,000

110 12 500 6,000

Total usage 737,900

 Step 2: calculate % usage

Item number Unit cost Annual demand Total cost per Usage as a %
year of total usage

101 5 48,000 240,000 32,5%

102 11 2,000 22,000 3%
103 15 300 4,500 0,6%
104 8 800 6,400 0,9%
105 7 4,800 33,600 4,6%
106 16 1,200 19,200 2,6%
107 20 18,000 360,000 48,8%
108 4 300 1,200 0,2%
109 9 5,000 45,000 6,1%
110 12 500 6,000 0,8%
Total usage 737,900 100%
 Step 3: sort items by usage

Item Cumulative Unit Annual Total cost per Usage as a % of Cumulative % of

number % of items cost demand year total usage total

107 10% 20 18,000 360,000 48,8% 48,8%

101 20% 5 48,000 240,000 32,5% 81,3%
109 30% 9 5,000 45,000 6,1% 87,4%
105 40% 7 4,800 33,600 4,6% 92%
102 50% 11 2,000 22,000 3,0% 94,9%
106 60% 16 1,200 19,200 2,6% 97,5%
104 70% 8 800 6,400 0,9% 98,4%
110 80% 12 500 6,000 0,8% 99,2%
103 90% 15 300 4,500 0,6% 99,8%
108 100% 4 300 1,200 0,2% 100%
Total usage 737,900 100%
 Step 4: classification

Percentage of Percentage usage

Cathegory Items Action
items (%)

Class A 107, 101 20% 81,6% Close control

109, 105, 102, Regular

Class B 40% 16,2%
106 review

104, 110, 103, Infrequent

Class C 40% 2,5%
108 review
 Inventory policy for A items

Small in number, but consume large amount of

resources

o Tight control
o Secured storage
o Rigid estimate of requirements
o Strict & closer watch
o Low safety stocks
o Frequent reorder
o Managed by top management
 Inventory policy for B items

Moderate in number and consume moderate

amount of resources

o Moderate control
o Purchase based on rigid requirements
o Reasonably strict watch & control
o Moderate safety stocks
o Managed by middle level management
 Inventory policy for C items

Larger in number, but consume lesser amount of resources

o Ordinary control measures
o Purchase based on usage estimates
o High safety stocks
o Managed by lower management
 VED analysis

Done based on shortage cost

o Vital: shortage can not be tolerated

o Essential: shortage can be tolerated for a short

period

o Desirable: shortage does not have adverse effect,

may require more resources
 ABC and VED: service rates

V E D

A 90 85 80

B 95 90 85

C 100 95 90
Inventory Classification Schemes
 Conclusions

• The boundary between class A and class B might

not be as sharply defined.

• The purpose of this classification is to ensure that

inventory manager use resources to maximum
efficiency by concentrating on those items that
have the greatest potential savings.

• Selective control will be more effective than an

approach that treats all items identically.
22-09-2022 85
 Certificate Course on
Operations Management and
Analytics
(August 2022-January 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
Module VI

• Independent demand inventory (This covered in Module V)

• Dependent demand inventory

• Basic purpose of MRP

• Construction of MRP table

– BOM/Product Structure Tree
– Outputs of MRP
(Session 1 and 2)
– Benefits of MRP

• MRP Scheduling Terminologies

• MRP Nervousness

• MRP-II (Optimizing Ordering Policy using EOQ, POQ, PPB) (Session 3 and 4)

• ERP (Much broader, integrated information system including inventory)

Session 1 and 2
 Independent and Dependent
Inventory Items

• Independent items are those which are either directly

consumed in the market or consumed in a
department/production unit to complete a product
which eventually consumed in a market.

• Dependent items are those which are required as per

the demand of the independent items in the market/
production unit.
 Information flows

Capacity Management
& Production Plan
Item # 1

Demand forecasting
Item # 2

Independent
Items (FGI)
Inventory Management
Tool & Techniques

Item # 3
Item # 4
.
.
.
Item # n
So far, we learnt EOQ, POQ etc for deterministic/
stochastic demand. Let’s see how to make use of
these inventory management techniques for actual
management of inventory items in a industry?

Item # 1
Item # 2 Minimize

Independent
Items (FGI)
Item # 3 Total
Item # 4 Procurement
. Cost
.
.
Item # n
 Basic purpose of MRP

• How much to procure

Module V

• When to procure Module VI

• How to procure

Module IX
• Whom to procure (Descriptive Analytics)
 Construction of MRP
Bill of Material/Product Structure Tree

Let’s try to understand the Bill of Material

and its role in MRP with the help of an
example.
Bicycle (A finished good inventory)
BOM of select inventory items
Example 1

If the demand for

Bicycle = 50

then requirement for

• Nut = ____?
• Spokes=_____?
Example 2
If the demand for
Bicycle is 50, 45, 60,
100 for first four
periods

Then the
requirement for Nut
& Spokes in
respective periods
would be?
To answer this, we
need MRP table
and few other
inputs.
 Material Requirements Planning
• MRP is a means for determining the number of parts,
components, and materials needed to produce a product
in the entire planning period.

• MRP provides time scheduling information specifying

when each of the materials, parts, and components
should be ordered or produced

• Dependent demand drives MRP

• MRP is a software system

 Material Requirements Planning
System
• Based on a master production schedule, a material
requirements planning system:
– Creates schedules identifying the specific parts
and materials required to produce end items

– Determines exact unit numbers needed

– Determines the dates when orders for those

materials should be released, based on lead times
 Aggregate Forecasts
Firm orders
product of demand
from known
plan from random
customers
customers

Engineering Master production

Schedule (MPS) Inventory
design
transactions
changes

Material
planning
Bill of (MRP Inventory
material computer record file
file program)
Secondary reports
Primary reports
Exception reports
Planned order schedule for Planning reports
inventory and production Reports for performance
control control
©The McGraw-Hill Compaies, Inc., 2004
MRP Scheduling Terminologies
• Gross requirements

• Scheduled receipts

• Projected available balance

• Safety stock

• Net requirements

• Planned order receipt

• Planned order release

• Gross Requirement (GR)

It is forecasted/estimated demand of an independent

item, or

It is a planned order release of a dependent item

required as per the BOM.
Example 3
X
If the marketing team
estimated the demand
of X as 100, 120, 110, A(2) B(1)
90 for four periods then
this would be GR for
the MRP table. C(3) C(2) D(5)

Solution:

X 1 2 3 4
GR 100 120 110 90
• Scheduled Receipt (SR)

It represents the number of respective items

scheduled to receive and the receiving is almost
treated as confirm .

SR is taken into account for calculation of net

requirement of an inventory item.

SR is taken as the beginning of the period.

Example 4
The marketing team estimated the X
demand of X as 100, 120, 110, 90.
The production unit confirmed the
extra production of 25 units to be A(2) B(1)
available in 3rd quarter due of
consideration to safety stock to
tackle variation. This information C(3) C(2) D(5)
should be given as an input to MRP
table.
Solution:
X 1 2 3 4
GR 100 120 110 90
SR 25
NR 100 120 85 90
• Projected Available Balance (PAB)

It represents the actual number of respective item left

at the end of the period after meeting demand, or

Actual number of respective item carried forward

from the previous period after meeting the demand
to the next period.
Example 5
X
The marketing team estimated the demand of X
as 100, 120, 110, 90. The production unit
confirmed the extra production of 25 units to be A(2) B(1)
available in 3rd period due to consideration of
safety stock to tackle variation. There is unsold
unit of 175 from the previous planning period. C(3) C(2) D(5)

Solution: This information should be given as an input to MRP table.

X 1 2 3 4
GR 100 120 110 90
On-hand=175
SR 25
PAB 75
NR 0 45 85 90
• Net Requirement (NR)

It represents the actual number of respective item

need to be procured after due consideration of all
inputs such as SR, SS, and on-hand inventory.
Example 6
X
The marketing team estimated the demand of X
as 100, 120, 110, 90. The production unit
confirmed the extra production of 25 units to be A(2) B(1)
available in 3rd period due to consideration of
safety stock to tackle variation. There is unsold
unit of 85 from the previous planning period. The C(3) C(2) D(5)

safety stock of 20 need to be maintained in all

periods.
Solution: This information should be given as an input to MRP table.
X 1 2 3 4
GR 100 120 110 90 On-hand=85
SR 25 SS= 20
SS 20 20 20 20
NR 15+20=35 120 85 90
• Planned Order Receipt (POR)

It represents the actual number of respective item

need to be received at the beginning of the period.

It is equal to the NR of the respective item

(POR=NR).
• Planned Order Release (PORel)

It represents the actual number of respective item

need to be released at the beginning of the period
after due consideration of lead time.

It is always equal to the POR & NR of the respective

item (PORel=POR=NR).
Example 7
The marketing team estimated the demand of X as 100, 120, 110, 90.
The production unit confirmed the extra production of 25 units to be
available in 3rd period due to consideration of safety stock to tackle
variation. There is unsold unit of 85 from the previous planning period.
The safety stock of 20 need to be maintained in all periods. The lead
time for X is 2 periods.

A(2) B(1)

C(3) C(2) D(5)

Solution: This information should be given as an input to MRP table.

X -2 -1 1 2 3 4
GR 100 120 110 90
SR 25
On-hand=85
SS 20 20 20 20
SS= 20
NR 15+20=35 120 85 90
LT= 2

POR 35 120 85 90
PORel 35 120 85 90
Example 8

Item On-Hand Lead Time (Weeks)

X X 50 2
A 75 3
B 25 1
A(2) B(1) C 10 2
D 20 2

C(3) C(2) D(5)

Requirements include 95 units (80 firm orders and 15 forecast) of X

in week 10
Solution:
Day: 1 2 3 4 5 6 7 8 9 10
X Gross requirements 95
X LT=2 Scheduled receipts
Proj. avail. balance 50 50 50 50 50 50 50 50 50 50
On- Net requirements 45
hand Planned order receipt 45
50 Planner order release 45
A Gross requirements 90
A(2) LT=3 Scheduled receipts
Proj. avail. balance 75 75 75 75 75 75 75 75
On- Net requirements 15
hand Planned order receipt 15
75 Planner order release 15
B Gross requirements 45
It takes LT=1 Scheduled receipts
Proj. avail. balance 25 25 25 25 25 25 25 25
2 A’s for On-
hand
Net requirements
Planned order receipt
20
20
each X 25 Planner order release 20
C Gross requirements 45 40
LT=2 Scheduled receipts
Proj. avail. balance 10 10 10 10 10
On- Net requirements 35 40
hand Planned order receipt 35 40
10 Planner order release 35 40
D Gross requirements 100
LT=2 Scheduled receipts
Proj. avail. balance 20 20 20 20 20 20 20
On- Net requirements 80
hand Planned order receipt 80
20 Planner order release 80
 Day: 1 2 3 4 5 6 7 8 9 10
X Gross requirements 95
X LT=2 Scheduled receipts
Proj. avail. balance 50 50 50 50 50 50 50 50 50 50
On- Net requirements 45
hand Planned order receipt 45
50 Planner order release 45
A Gross requirements 90
A(2) B(1) LT=3 Scheduled receipts
Proj. avail. balance 75 75 75 75 75 75 75 75
On- Net requirements 15
hand Planned order receipt 15
75 Planner order release 15
B Gross requirements 45
LT=1 Scheduled receipts
It takes Proj. avail. balance 25 25 25 25 25 25 25 25
On- Net requirements 20
1 B for hand Planned order receipt 20

each X 25
C
Planner order release
Gross requirements 45
20
40
LT=2 Scheduled receipts
Proj. avail. balance 10 10 10 10 10
On- Net requirements 35 40
hand Planned order receipt 35 40
10 Planner order release 35 40
D Gross requirements 100
LT=2 Scheduled receipts
Proj. avail. balance 20 20 20 20 20 20 20
On- Net requirements 80
hand Planned order receipt 80
20 Planner order release 80
 Day: 1 2 3 4 5 6 7 8 9 10
X Gross requirements 95
X LT=2 Scheduled receipts
Proj. avail. balance 50 50 50 50 50 50 50 50 50 50
On- Net requirements 45
hand Planned order receipt 45
50 Planner order release 45
A Gross requirements 90
A(2) B(1) LT=3 Scheduled receipts
Proj. avail. balance 75 75 75 75 75 75 75 75
On- Net requirements 15
hand Planned order receipt 15
75 Planner order release 15
C(3) B Gross requirements 45
LT=1 Scheduled receipts
Proj. avail. balance 25 25 25 25 25 25 25 25
On- Net requirements 20
hand Planned order receipt 20
25 Planner order release 20
It takes 3 C
LT=2
Gross requirements
Scheduled receipts
45 40

C’s for On-

Proj. avail. balance
Net requirements
10 10 10 10 10
35 40
each A hand
10
Planned order receipt
Planner order release 35
35
40
40

D Gross requirements 100

LT=2 Scheduled receipts
Proj. avail. balance 20 20 20 20 20 20 20
On- Net requirements 80
hand Planned order receipt 80
20 Planner order release 80
 Day: 1 2 3 4 5 6 7 8 9 10
X Gross requirements 95
X LT=2 Scheduled receipts
Proj. avail. balance 50 50 50 50 50 50 50 50 50 50
On- Net requirements 45
hand Planned order receipt 45
50 Planner order release 45
A Gross requirements 90
A(2) B(1) LT=3 Scheduled receipts
Proj. avail. balance 75 75 75 75 75 75 75 75
On- Net requirements 15
hand Planned order receipt 15
75 Planner order release 15
C(3) C(2) B Gross requirements 45
LT=1 Scheduled receipts
Proj. avail. balance 25 25 25 25 25 25 25 25
On- Net requirements 20
hand Planned order receipt 20
25 Planner order release 20
It takes 2 C
LT=2
Gross requirements
Scheduled receipts
45 40

C’s for On-

Proj. avail. balance
Net requirements
10 10 10 10 10
35 40
each B hand
10
Planned order receipt
Planner order release 35
35
40
40

D Gross requirements 100

LT=2 Scheduled receipts
Proj. avail. balance 20 20 20 20 20 20 20
On- Net requirements 80
hand Planned order receipt 80
20 Planner order release 80
 Day: 1 2 3 4 5 6 7 8 9 10
X Gross requirements 95
X LT=2 Scheduled receipts
Proj. avail. balance 50 50 50 50 50 50 50 50 50 50
On- Net requirements 45
hand Planned order receipt 45
50 Planner order release 45
A Gross requirements 90
A(2) B(1) LT=3 Scheduled receipts
Proj. avail. balance 75 75 75 75 75 75 75 75
On- Net requirements 15
hand Planned order receipt 15
75 Planner order release 15
C(3) C(2) D(5) B Gross requirements 45
LT=1 Scheduled receipts
Proj. avail. balance 25 25 25 25 25 25 25 25
On- Net requirements 20
hand Planned order receipt 20
25 Planner order release 20
It takes 5 C
LT=2
Gross requirements
Scheduled receipts
45 40

D’s for each On-

Proj. avail. balance
Net requirements
10 10 10 10 10
35 40
B hand
10
Planned order receipt
Planner order release 35
35
40
40

D Gross requirements 100

LT=2 Scheduled receipts
Proj. avail. balance 20 20 20 20 20 20 20
On- Net requirements 80
hand Planned order receipt 80
20 Planner order release 80
Example 9
Solution
 Example 10
Given the following product structure tree diagram, complete
the MRP records for A, B, and C.

C B (2)

C
 Week 4 5 6 7 8 9
Gross 5 15 18 8 12 22
Solution
Scheduled PART A
Receipt
Q=20; LT=1;
PAB (21)
SS=0
P.O.Receipt
P.O. Release
Gross
Scheduled 32 PART B
Receipt
Q=40; LT=2;
PAB (20) SS=0
P.O.Receipt
P.O. Release
Gross

Scheduled PART C
Receipt Q=L-4-L; LT=1;
PAB (50)
SS=10
P.O.Receipt
P.O. Release
 Example 11
Draw the time-phased product structure tree.

A Parts L.T.
A 1

B(2) C (3) B 2
C 1
D 1
D(2) E(2)
E(2) F(2) E 2
F 3
G 2
G(1) D(2)
 Time-phased product structure of A
Solution
D
B

E A

E
G C
F

1 2 3 4 5 6 7
Time in week
Practice Question # 1
Practice Question # 2
Practice Question # 3
Practice Question # 4
Practice Question # 5
Practice Question # 6
Session 3 and 4
 Manufacturing Resource Planning
(MRP II)
• Goal: Plan and monitor all resources of a manufacturing firm (closed
loop):
– manufacturing
– marketing
– finance
– engineering

• Optimize the ordering policy of respective items obtained from MRP

considering ordering and inventory carrying cost (Sessions 3 & 4)
• Optimize the ordering policy of respective items obtained from MRP
considering ordering, inventory carrying cost & supplier’s capacity
etc. (Prescriptive analytics- Module XI).
 Lot Sizing in MRP Programs
• Economic order quantity (EOQ)

• Periodic order quantity (POQ)

• Part Period Balancing (PPB)

• Lot-for-lot (L4L)
Example 12
From the MRP table, the requirement of one of the dependent item
say “D17” of an independent item “X” is shown below for 12
weeks (one quarter). The item D17 is a B-class type item from
rupee value of the inventory but also a critical item. Therefore, the
inventory manager wish to optimize the total procurement cost
considering the fact of ordering and carrying cost.

Co=300 per order and Cc=2 per unit per week.

Item D171 2 3 4 5 6 7 8 9 10 11 12
PO Rel. 10 10 15 20 70 180 250 270 230 40 0 10
Solution

To understand the use of EOQ, POQ and PPB concepts of

inventory management, we will see the detailed calculations
and steps involved in the excel.

It is assumed that the formula to calculate EOQ is known to

you. Shortage (or back ordering cost) is not considered.
However, the same can be considered using the EOQ formula
with shortage cost.
Practice Question # 1

Given the following weekly requirements from MRP record for the item, determine the
sequence of planned order release using the EOQ, POQ & PPB procedures. It is assumed
that the on-hand inventory is 30. Calculate the inventory carrying cost on the basis of
weekly ending inventory values. Which procedure provides the lowest total cost for the
eight-week period? Ordering cost is Rs. 32 per order, Inventory carrying cost is Rs. 2 per
week per unit.

Week 1 2 3 4 5 6 7 8
Requirements 45 2 10 12 6 0 14 5
Practice Question # 2
A company has estimated the requirements for a part as given in
the table below.

The ordering cost is Rs. 300 and inventory carrying cost is Rs 2

per unit per month. There are 50 parts available on hand. The
company plans to determine the ordering policy over the next 12
months to meet the demand requirement.

•Apply EOQ, PQO and PPB to handle the situation.

Month 1 2 3 4 5 6 7 8 9 10 11 12
Requir 150 10 15 20 70 250 250 250 250 40 0 100
ement
 Certificate Course on
Operations Management and
Analytics
(August 2022-January, 2023)

Prof. Surya Prakash Singh

PhD (IIT Kanpur), PDF (NUS Singapore, MIT USA)
Dhananjaya Chair Professor
Department of Management Studies
Indian Institute of Technology Delhi 1
E-Mail: [email protected]
QUALITATIVE
– use management judgment, expertise, and opinion to
predict future demand

QUANTITATIVE
– Time series
statistical techniques that use historical demand data to predict
future
– Regression methods
attempt to develop a mathematical relationship between demand
and factors that cause its behavior
2
 Why Forecasting?
Decisions Based on Forecasts
Production Personnel
– Aggregate planning, – Workforce planning,
inventory control, hiring, layoff
scheduling
Marketing
– New product
introduction, sales-
force allocation,
promotions
Finance
– Plant/equipment
investment,
budgetary planning
3
 Why Forecasting?
Forecasting is an important for all strategic
and planning decisions in a business
organization.

Forecasts of product demand, materials,

labor, financing are an important inputs to
scheduling, acquiring resources, and
determining resource requirements.

4
Generally speaking the word
forecast…
There is only one forecast, a Sales forecast.

Financial Plan

Sales Forecast Production Plan

(Unconstrained)

Marketing Plan

Most companies have a hard time distinguishing the

difference between the forecasting process and the
planning process…
– Dr. John (Tom) Mentzer, University of Tennessee at
Knoxville
5
Everyone wants to touch the
forecast…
But no one wants to be accountable for the results
“Cultural” Issue
Requires change management driven by a
“Champion”

6
Law of Universal Forecasting…
The more people who touch the forecast,
the more inaccurate the forecast, and

The more fact-based (information\data

supported) and mathematically derived
the forecast, the more accurate the
forecast…

7
 Forecasting Horizons.
Short Term (0 to 3 months): for inventory
management and scheduling.
Medium Term (3 months to 2 years): for
production planning, purchasing, and
distribution.
Long Term (2 years and more): for
capacity planning, facility location, and
strategic planning.

8
 Principles of Forecasting
Forecasts are almost always wrong.
Every forecast should include an estimate
of the forecast error.
The greater the degree of aggregation, the
more accurate the forecast.
Long-term forecasts are usually less
accurate than short-term forecasts.

9
Factors Affecting Forecasting
Methods

Time frame

Demand behavior

10
 Demand Behavior
Trend
– a gradual, long-term up or down movement of
demand
Cycle
– an up-and-down repetitive movement in demand
Seasonal pattern
– an up-and-down repetitive movement in demand
occurring periodically
Random variations
– movements in demand that do not follow a pattern

11
Theory of Forecasting
Methods...
Forecast = Pattern (s) + Randomness

12
 Forecast = Pattern (s) + Randomness

This simple equation is really saying that

when the average pattern of the underlying
data has been identified some deviation will
occur between the forecasting method
applied and the actual occurrence.

That deviation is called “Error”, or

unexplainable variance.
13
 Forecast = Pattern (s) + Randomness

This simple equation is really saying that

when the average pattern of the underlying
data has been identified some deviation will
occur between the forecasting method
applied and the actual occurrence.

That deviation is called “Error”, or

unexplainable variance.
14
 Forecast = Pattern (s) + Randomness
Maximize Minimize

This simple equation is really saying that

when the average pattern of the underlying
data has been identified some deviation will
occur between the forecasting method
applied and the actual occurrence.

That deviation is called “Error”, or

unexplainable variance.
15
 Forecast = Pattern (s) + Randomness

Two Broad mathematical categories

1. Time Series
2. Causal

16
Forecast = Pattern (s) + Randomness

Trend
Seasonality
Cyclical

17
 Forms of Forecast
Movement

Demand
Demand

Random
movement

Time Time
(a) Trend (b) Cycle

Demand
Demand

Time Time
(c) Seasonal pattern (d) Trend with seasonal pattern
18
 One methodology fits all
philosophy…
Most systems are driven by this philosophy
– Always, time series methods
Exponential Smoothing
Winter’s most widely used mathematical method
– Easy to systematize

We have > 90 different models

19
 Bottom Line…
There is no Best Method
– The best method depends on the data, the
purpose, the organizational environment and
the perspective of the forecaster.

– Your market, products, goals, and constraints

should be considered when selecting the
forecasting tools best for you...

20
 Forecasting Methods

QUALITATIVE
– use management judgment, expertise, and opinion to
predict future demand
QUANTITATIVE
– Time series
statistical techniques that use historical demand data to predict
future demand
– Regression methods
attempt to develop a mathematical relationship between
demand and factors that cause its behavior

21
 QUALITATIVE METHODS
Management, marketing, purchasing, and
engineering are sources for internal
qualitative forecasts

Delphi method
– involves soliciting forecasts about
technological advances from experts

22
 Forecasting Process
1. Identify the 2. Collect historical 3. Plot data and identify
purpose of forecast data patterns

6. Check forecast 5. Develop/compute 4. Select a forecast

accuracy with one or forecast for period of model that seems
more measures historical data appropriate for data

7.
Is accuracy of No 8b. Select new
forecast forecast model or
acceptable? adjust parameters of
existing model
Yes
9. Adjust forecast based 10. Monitor results
8a. Forecast over
on additional qualitative and measure forecast
planning horizon
information and insight accuracy

23
QUANTITAIVE METHODS
Time Series
Assume that what has occurred in the past
will continue to occur in the future
Relate the forecast to only one factor - time
Include
– moving average
– exponential smoothing
– Trend exponential smoothening
– linear trend line
– ARIMA
– Deep learning
– Random Forest
– Reinforcement Learning 24
 Moving Average
Naive forecast
– demand the current period is used as next
period’s forecast
Simple moving average
– stable demand with no pronounced behavioral
patterns
Weighted moving average
– weights are assigned to most recent data

25
Example:

Moving Average:
Naïve Approach
ORDERS
MONTH PER MONTH FORECAST

Jan 120 -
Feb 90 120
Mar 100 90
Apr 75 100
May 110 75
June 50 110
July 75 50
Aug 130 75
Sept 110 130
Oct 90 110
Nov - 90
26
 Simple Moving Average
n

i=1
Di where
MAn = n = number of periods in
n the moving average
Include n most recent observations Di = demand in period i
Weight equally
Ignore older observations

1/n

n ... 3 2 1
today 27
Example:
3-month Simple Moving Average
3
ORDERS MOVING  Di
i=1
MONTH PER MONTH AVERAGE MA3 =
Jan 120 – 3
Feb 90 –
Mar 100 – 90 + 110 + 130
= 3
Apr 75 103.3
May 110 88.3
June 50 95.0
= 110 orders
July 75 78.3
Aug 130 78.3
for Nov
Sept 110 85.0
Oct 90 105.0
Nov - 110.0

28
Example:
5-month Simple Moving Average

ORDERS MOVING
5
MONTH PER MONTH AVERAGE
Jan 120 –

i=1
Di

Feb 90 – MA5 =
Mar 100 –
5
Apr 75 –
90 + 110 + 130+75+50
May 110 – =
June 50 99.0
5
July 75 85.0
Aug 130 82.0 = 91 orders
Sept 110 88.0 for Nov
Oct 90 95.0
Nov - 91.0

29
 Smoothing Effects
150 –

125 – 5-month

100 –
Orders

75 –

50 – 3-month

Actual
25 –

0– | | | | | | | | | | |
Jan Feb Mar Apr May June July Aug Sept Oct Nov
Month
30
 Weighted Moving Average

 Adjusts WMAn =  Wi Di
i=1
moving
where
average
method to Wi = the weight for period i,
between 0 and 100 percent
more closely
reflect data  W = 1.00
i
fluctuations

31
Example:
Weighted Moving Average
Example
MONTH WEIGHT DATA
August 17% 130
September 33% 110
October 50% 90
3
November Forecast WMA3 = 
i=1
Wi Di

= (0.50)(90) + (0.33)(110) + (0.17)(130)

= 103.4 orders

32
 Exponential Smoothing

 Averaging method

 Weights most recent data more strongly

 Reacts more to recent changes

 Widely used, accurate method

33
 Include all past observations
Weight recent observations much more heavily than very old
observations:

weight
Decreasing weight given
to older observations

today

34
35
36
37
38
weight
0  1
Decreasing weight given 
to older observations
 (1   )
 (1   ) 2
 (1   ) 3
today 

39
Exponential Smoothing: Analytics

40
New forecast = weighted sum of last period
actual value and last period forecast
– : Smoothing constant
– Ft : Forecast for period t
– Ft-1: Last period forecast
– Dt-1: Last period actual value
41
Thus, new forecast is weighted sum of old forecast and actual demand

Notes:
– Only 2 values (Dt and Ft-1 ) are required, compared with n for moving
average
– Parameter a determined empirically (whatever works best)
– Rule of thumb:  < 0.5
– Typically,  = 0.2 or  = 0.3 work well

42
 Ft +1 =  Dt + (1 - )Ft
where:
Ft +1 = forecast for next period
Dt = actual demand for present period
Ft = previously determined forecast for present
period
 = weighting factor, smoothing constant

43
 Effect of Smoothing Constant
0.0    1.0
If  = 0.20, then Ft +1 = 0.20 Dt + 0.80 Ft

If  = 0, then Ft +1 = 0 Dt + 1 Ft = Ft
Forecast does not reflect recent data

If  = 1, then Ft +1 = 1 Dt + 0 Ft = Dt
Forecast based only on most recent data
44
Example:
Exponential Smoothing (α=0.30)

PERIOD MONTH DEMAND Sample calculations for 2nd period

1 Jan 37 F2 = D1 + (1 - )F1
2 Feb 40 = (0.30)(37) + (0.70)(37)
3 Mar 41
= 37
4 Apr 37 F3 = D2 + (1 - )F2
5 May 45 = (0.30)(40) + (0.70)(37)
6 Jun 50
= 37.9
7 Jul 43
8 Aug 47
F13 = D12 + (1 - )F12
9 Sep 56
10 Oct 52 = (0.30)(54) + (0.70)(50.84)
11 Nov 55 = 51.79
12 Dec 54

45
Exponential Smoothing
(cont.)
FORECAST, Ft + 1
PERIOD MONTH DEMAND ( = 0.3) ( = 0.5)
1 Jan 37 – –
2 Feb 40 37.00 37.00
3 Mar 41 37.90 38.50
4 Apr 37 38.83 39.75
5 May 45 38.28 38.37
6 Jun 50 40.29 41.68
7 Jul 43 43.20 45.84
8 Aug 47 43.14 44.42
9 Sep 56 44.30 45.71
10 Oct 52 47.81 50.85
11 Nov 55 49.06 51.42
12 Dec 54 50.84 53.21
13 Jan – 51.79 53.61
46
 Exponential Smoothing (cont.)
70 –

60 – Actual  = 0.50

50 –

40 –
Orders

 = 0.30
30 –

20 –

10 –

0– | | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
Month
47
Simple Exponential Smoothing
Properties of Simple Exponential Smoothing
– Widely used and successful model

– Requires very little data

– Larger , more responsive forecast; Smaller ,

smoother forecast

– “best”  can be found by Solver

– Suitable for relatively stable time series

48
 Complicating Factors
Simple Exponential Smoothing works well with
data that is “moving sideways” (stationary)

Must be adapted for data series which exhibit a

definite trend

Must be further adapted for data series which

exhibit seasonal patterns

49
 Holt’s Method:
Double Exponential Smoothing
Ideas behind smoothing with trend:
– ``De-trend'' time-series by separating base from trend
effects
– Smooth base in usual manner using 
– Smooth trend forecasts in usual manner using 

Smooth the base forecast Lt

Ft   D t  (1   )( Ft 1  Tt 1 )
Smooth the trend forecast Tt
Tt   ( Ft  Ft 1 )  (1   )Tt 1

50
 AFt +1 = Lt +1 + Tt +1 or

AFt = Lt + Tt
Forecast k periods into future Ft+k with base and trend
AFt  k  Lt  kTt
where
T = an exponentially smoothed trend factor

Tt +1 = (Lt +1 - Lt) + (1 - ) Tt or

Tt = (Lt – Lt-1) + (1 - ) Tt-1

where
AF = Adjusted forecast
Tt = the last period trend factor
 = a smoothing constant for trend
 = a smoothing constant for level
51
Example:
Adjusted Exponential
Smoothing (β=0.30)
PERIOD MONTH DEMAND
Sample calculations for 3rd period
1 Jan 37 T3 = (L3 - L2) + (1 - ) T2
2 Feb 40 = (0.30)(38.5 - 37.0) + (0.70)(0)
3 Mar 41
4 Apr 37 = 0.45
AF3 = L3 + T3 = 38.5 + 0.45
5 May 45
= 38.95
6 Jun 50
7 Jul 43 T13 = (L13 - L12) + (1 - ) T12
8 Aug 47 = (0.30)(53.61 - 53.21) + (0.70)(1.77)
9 Sep 56 = 1.36
10 Oct 52
11 Nov 55 AF13 = L13 + T13 = 53.61 + 1.36 = 54.96
12 Dec 54

52
 Adjusted Exponential Smoothing:
Example
FORECAST TREND ADJUSTED
PERIOD MONTH DEMAND Lt +1 Tt +1 FORECAST AFt +1

1 Jan 37 37.00 – –
2 Feb 40 37.00 0.00 37.00
3 Mar 41 38.50 0.45 38.95
4 Apr 37 39.75 0.69 40.44
5 May 45 38.37 0.07 38.44
6 Jun 50 38.37 0.07 38.44
7 Jul 43 45.84 1.97 47.82
8 Aug 47 44.42 0.95 45.37
9 Sep 56 45.71 1.05 46.76
10 Oct 52 50.85 2.28 58.13
11 Nov 55 51.42 1.76 53.19
12 Dec 54 53.21 1.77 54.98
13 Jan – 53.61 1.36 54.96
53
 Adjusted Exponential Smoothing
Forecasts
70 –

Adjusted forecast ( = 0.30)

60 –
Actual
50 –

40 –
Demand

30 – Forecast ( = 0.50)

20 –

10 –

0– | | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
Period
54
 Holt-Winter’s Method:
Exponential Smoothing w/ Trend & Seasonality

Ideas behind smoothing with trend and seasonality:

– “De-trend” and “de-seasonalize” time-series by separating
base from trend and seasonality effects
– Smooth base in usual manner using 
– Smooth trend forecasts in usual manner using 
– Smooth seasonality forecasts using g

Assume m seasons in a cycle

– 12 months in a year
– 4 quarters in a month
– 3 months in a quarter
– et cetera
55
Smooth the base forecast Lt

Smooth the trend forecast Tt

Smooth the seasonality forecast St

Adjusted Forecast Ft with trend and seasonality

Forecast k periods into future Ft+k with trend & seasonality

56
There are following three ways to adjust seasonality:

1. Multiplicative seasonality: (Level + Trend) *

Seasonality

2. Additive Seasonality: Level + Trend + Seasonality

This is a
multiplicative one

57
Example Using Holt-Winter Model

DEMAND (1000’S PER QUARTER)

YEAR 1 2 3 4
2002 12.6 8.6 6.3 17.5
2003 14.1 10.3 7.5 18.2
2004 15.3 10.6 8.1 19.6
2005 ?

Let’s GO TO EXCEL

58
Example:
DEMAND (1000’S PER QUARTER)
YEAR 1 2 3 4 Total
2002 12.6 8.6 6.3 17.5 45.0
2003 14.1 10.3 7.5 18.2 50.1
2004 15.3 10.6 8.1 19.6 53.6
Total 42.0 29.5 21.9 55.3 148.7

Di
Seasonal factor = Si = D
D1 42.0 D3 21.9
S1 = = = 0.28 S3 = = = 0.15
D 148.7 D 148.7
D 29.5 D 55.3
S2 = 2 = = 0.20 S4 = 4 = = 0.37
D 148.7 D 148.7
59
 Seasonal Adjustment (cont.)

For 2005

y = 40.97 + 4.30x = 40.97 + 4.30(4) = 58.17

SF1 = (S1) (F5) = (0.28)(58.17) = 16.28

SF2 = (S2) (F5) = (0.20)(58.17) = 11.63
SF3 = (S3) (F5) = (0.15)(58.17) = 8.73
SF4 = (S4) (F5) = (0.37)(58.17) = 21.53

60
 Comparison of
Holts & Winter Model

61
Regression: Linear Trend Line

 xy - nxy
y = a + bx b =
 x2 - nx2
where a = y-bx
a = intercept
b = slope of the line where
x = time period n = number of periods
y = forecast for
demand for period x  x
x = = mean of the x values
n
 y
y = n = mean of the y values

62
 Least Squares Example
x(PERIOD) y(DEMAND) xy x2
1 73 37 1
2 40 80 4
3 41 123 9
4 37 148 16
5 45 225 25
6 50 300 36
7 43 301 49
8 47 376 64
9 56 504 81
10 52 520 100
11 55 605 121
12 54 648 144
78 557 3867 650

63
 Least Squares
Example (cont.)
78
x = = 6.5
12
557
y = = 46.42
12
xy - nxy 3867 - (12)(6.5)(46.42)
b = = =1.72
x - nx
2 2
650 - 12(6.5) 2

a = y - bx
= 46.42 - (1.72)(6.5) = 35.2

64
 Linear trend line y = 35.2 + 1.72x
Forecast for period 13 y = 35.2 + 1.72(13) = 57.56 units

70 –

60 –
Actual

50 –
Demand

40 –
Linear trend line
30 –

20 –

10 – | | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
0– Period

65
 Forecasting Accuracy or
Performance
How good is the forecast?
Mean Forecast Error (MFE or Bias): Measures
average deviation of forecast from actuals.

Mean Absolute Deviation (MAD): Measures

average absolute deviation of forecast from
actuals.

Mean Absolute Percentage Error (MAPE):

Measures absolute error as a percentage of the
forecast.

Standard Squared Error (MSE): Measures

variance of forecast error 66
Forecasting Performance Measures

67
 Mean Forecast Error
(MFE or Bias)

Want MFE to be as close to zero as possible -- minimum

bias
A large positive (negative) MFE means that the forecast
is undershooting (overshooting) the actual observations
Note that zero MFE does not imply that forecasts are
perfect (no error) -- only that mean is “on target”
Also called forecast BIAS

68
Mean Absolute Deviation (MAD)

Measures absolute error

Positive and negative errors thus do not cancel out (as
with MFE)
Want MAD to be as small as possible
No way to know if MAD error is large or small in relation
to the actual data

69
Mean Absolute Percentage Error
(MAPE)

Same as MAD, except ...

Measures deviation as a percentage of actual data

70
Mean Squared Error (MSE)

Measures squared forecast error -- error variance

Recognizes that large errors are disproportionately more
“expensive” than small errors
But is not as easily interpreted as MAD, MAPE -- not as
intuitive

71
 Comparison of
Forecast

FORECAST MAD MAPD E (E)

Exponential smoothing ( = 0.30) 4.85 9.6% 49.31 4.48
Exponential smoothing ( = 0.50) 4.04 8.5% 33.21 3.02
Adjusted exponential smoothing 3.81 7.5% 21.14 1.92
( = 0.50,  = 0.30)
Linear trend line 2.29 4.9% – –

72
 Tracking Signal Plot

3 –
Tracking signal (MAD)

2 –
Exponential smoothing ( = 0.30)
1 –

0 –

-1 –

-2 – Linear trend line

-3 –
| | | | | | | | | | | | |
0 1 2 3 4 5 6 7 8 9 10 11 12
Period

73
 Certificate Course on
Operations Management and Analytics
(August 2022-January 2023)

Program Faculty:
Ramgopal Prajapat
Director Data Science, TATA CLiQ
Affiliation :
https://ramgopalprajapat.com/

14-11-2022 1
 Session 1 and 2
(Demand Forecasting Using R)

14-11-2022 2
 Agenda

Session 1

1 Getting started with R

2 Basic Data Manipulations using R

3 Time Series Data and Package

Session 2
Forecasting Scenario – Moving Average and Exponential
4
Smoothing

5 Forecasting Scenario – Regression Model

14-11-2022 3
 Agenda
Session 1

1 Getting started with R

2 Basic Data Manipulations using R

3 Time Series Data and Package

Session 2
Forecasting Scenario – Moving Average and Exponential
4
Smoothing

5 Forecasting Scenario – Regression Model

14-11-2022 4
 R for Statistical Computing

R is an open source statistical computing environment used by researchers, data scientists,

statistician and analytics professionals

R is case sensitive programming environment

14-11-2022 5
 Getting started on using R for Statistical
Analysis, Visualization and Text Mining

Install & Load

Download Install R
Packages

• Click on downloaded R • R works on functions and objects

• Download R from installation file
• http://cran.r-project.org • For each functionalities (e.g. Forecast), we
• http://cran.r- may have to refer a list of R packages
project.org/bin/windows/base
• Download version based on your system • Follow the instruction to install • User have to search to find relevant
configurations R packages for each of the functionalities
• Operating System (Window /Linux etc) • Steps to install R packages listed in the
next slide
• System Type - 32/64 bit

• A system can have multiple

version of R installed on a
system

14-11-2022 6
 Installation of R Packages (File Menu)

Step 1: Set CRAN Step 2: Select CRAN location

Set CRAN

Select
Step 4: Select and Load R Package Step 3: Select and Install R Package
Package

Install and
Load
Package

14-11-2022 7
 Install an R packages (R Console)
• An R package is installed by supplying the name of the package to the function “install.
package”
• For example the “tm” package can be downloaded and installed using R Console:
install.packages("tm")
• Load an R package to use functions/objects of the package

• Example of loading R package:

library(“tm")
• Difference between package and library. A package is a standardized collection of R
material (providing code, data, or documentation) and a library is a place or directory
where R knows to find packages and load into working environment

14-11-2022 8
 R Studio
IDE for Predictive Modeling in R

14-11-2022 9
 R Studio: Installation
• Go to http://www.rstudio.com/
• Follow the link and then choose the desktop application
• Download the R base package (see also next page) & R-Studio
• Follow the instructions of the Installer

First download & install

the R base package
Then download & install
R Studio
Follow the link

14-11-2022 10
R Studio: Installation cont..

14-11-2022 11
First Session with R Studio

Memory window: list of objects

Currently in the memory

Command-line window Multi-function Window

(for file browsing, plotting,

Software package management …)

14-11-2022 12
Action Items

• Install – R as per instructions

• Install R Studio as per instructions

• Some of the required R packages are listed on the next slide, can you please
install using either File Menu or R Program?

• R code to install a package “forecast” is install.package(“forecast"), so install the

list of packages (mentioned from next slide) one by one

• R is case sensitive, so be careful when you are installing R packages using R code

14-11-2022 13
 Packages and Descriptions
Package Name Description
foreign read data files from other stats packages like SAS
dplyr package for various data management tasks
reshape2 Package to easily melt data to long form
ggplot2 Data visualization using graphic
sqldf Connect sql database and perform sql query.
plyr It’s split and apply combine function in R
stringr stringr is a set of simple wrappers that make R's string functions more
simpler and easier to use
tm Provides a framework for text mining application in R.
forecast Generic function for forecasting from time series or time series models
 Agenda

Session 1

1 Getting started with R

2 Basic Data Manipulations using R

3 Time Series Data and Package

Session 2
Forecasting Scenario – Moving Average and Exponential
4
Smoothing

5 Forecasting Scenario – Regression Model

14-11-2022 15
Data Types & Objects

• Data Type
• Numeric, Integer, Character, Logical,
complex
• R Objects
• Vector, matrix, array, list, data frame

14-11-2022 16
Reading & Writing Data

• Reading data from program editor

• Reading Data from files
• CSV, SAS, .TXT, SPSS
• Reading data from social media
• Writing Data
• CSV, Tab delimed

14-11-2022 17
Commonly used functions in R
• Numeric functions
• Character functions
• Statistical functions
• Date
• User defined

14-11-2022 18
Hands on Exercise

1. Install a package - data.table and give a

name of function starting character "d“
2. Install and Load package -datasets. Use
“women” data frame available in this
package. Find number of woman with
height more than average but weight
less than average
3. Create a data frame of 15 Indian cities
and their population size. You could refer
cities and population from this page.
http://www.worldatlas.com/articles/the-
biggest-cities-in-india.html

14-11-2022 19
 Agenda

Session 1

1 Getting started with R

2 Basic Data Manipulations using R

3 Time Series Data and Package

Session 2
Forecasting Scenario – Moving Average and Exponential
4
Smoothing

5 Forecasting Scenario – Regression Model

14-11-2022 20
Forecasting Packages
• “TTR” R package can be used to smooth time
series data using a simple moving average.
• We can make forecasts for further time points by
using “forecast” package.
• ‘fpp2’: Data for "Forecasting: Principles and
Practice"

14-11-2022 21
 Agenda

Session 1

1 Getting started with R

2 Basic Data Manipulations using R

3 Time Series Data and Package

Session 2
Forecasting Scenario – Moving Average and Exponential
4
Smoothing

5 Forecasting Scenario – Regression Model

14-11-2022 22
Simple Moving Average
Forecast Ft+1 is average of n previous observations Dt, Dt-1, …
All observations are equally weighted

𝟏
Ft+1 = 𝒏(Dt + Dt-1 + Dt-2 + … + Dt+1-p)

𝟏
Ft+1 = 𝒏 σ𝒕𝒊=𝒕+𝟏−𝒏 𝑫𝒊
Issues with Simple Moving Average
• All observations treated equally
• Restricted to few observations
• Does not capture trend effectively

14-11-2022 23
Simple Moving Average
Forecast Ft+1 is average of n previous observations Dt, Dt-1, …
All observations are equally weighted

𝟏
Ft+1 = 𝒏(Dt + Dt-1 + Dt-2 + … + Dt+1-p)

𝟏
Ft+1 = 𝒏 σ𝒕𝒊=𝒕+𝟏−𝒏 𝑫𝒊
Issues with Simple Moving Average
• All observations treated equally
• Restricted to few observations
• Does not capture trend effectively

14-11-2022 24
 Simple Moving Average
• Include a set of previous observations: n Stock market analysts will
often use a 50 or 200 day
• Weight all observations equally moving average to help
• Completely ignore observations prior to n them see trends in the
stock market and
time periods
(hopefully) forecast where
the stocks are headed
Dt-1

Dt Observations
Ft+1 Forecast

1/n Weightage

14-11-2022 25
 Simple Moving Average
• Include a set of previous observations: n Stock market analysts will
often use a 50 or 200 day
• Weight all observations equally moving average to help
• Completely ignore observations prior to n them see trends in the
stock market and
time periods
(hopefully) forecast where
the stocks are headed
Dt-1

Dt Observations
Ft+1 Forecast

1/n Weightage

14-11-2022 26
Simple Exponential Smoothing
Forecast Ft+1 is a weighted average of previous observations Dt, Dt-1, …
More recent observations receive higher weights in comparison to earlier
observations

• It is a form of weighted
Dt-1 moving average

Dt Observations • Information from all data

Ft+1 Forecast points can be captured

Increasing Weightage Issues with Simple Moving

Average
• Does not capture trends and
seasonality effectively
• Forecasts from weighted
moving average

14-11-2022 27
Simple Exponential Smoothing
Exponential moving average can be re-written as a function of previous forecast

Ft+1 = 𝜶 Dt + 𝜶 (𝟏 − 𝜶) Dt-1 + 𝜶 (𝟏 − 𝜶)2 Dt-2 + …

Ft+1 = 𝜶 Dt + (𝟏 − 𝜶) [𝜶 Dt-1 + 𝜶 (𝟏 − 𝜶) Dt-2 + …]
Ft+1 = 𝜶 Dt + (𝟏 − 𝜶) 𝑭𝒕
The new forecast is weighted sum of old forecast and actual demand

• The parameter α is determined empirically (whatever works best)

• Rule of thumb is α <= 0.5
• Typically α = 0.2 or α = 0.3 works well
• Future forecasts are just the last forecast. That is, Ft+k = 𝐅𝐭
• Exponential moving average suffers from same issues as weighted moving
average: lack of ability to capture trends and cyclicities

14-11-2022 28
Hands on in R

14-11-2022 29
 Agenda

Session 1

1 Getting started with R

2 Basic Data Manipulations using R

3 Time Series Data and Package

Session 2
Forecasting Scenario – Moving Average and Exponential
4
Smoothing

5 Forecasting Scenario – Regression Model

14-11-2022 30
 Business Scenario and Application of Multiple
Regression
• Problem Statement or Scenario: An organization selling services and cash
collection can depend on various factors – cash due in the month, cash past
due and cash future due. Accurate forecasting of cash collection helps the
organization in managing cash flow requirements

Over Cash
Due Collection
Due in
Month Future
Due

What is % of Cash Collection variance is driven by each of these variables?

14-11-2022 31
Forecasting using Regression

• Define business problem to solved using Predictive Model

Problem • Define Performance window and create response variable
Statement • Define Observation window and crease independent variables

• Univariate and bivariate analysis to understand the input variables

• Variable Transformation and created derived variables
Data Preparation • Missing and outlier treatment

• Split the data into at least two samples – development for developing model
on and validation for validating the model developed
Model • Develop the model
Development • Model development checks such as residual, co-linearity, and normality

• Validate the model developed on validation sample

Model Validation • Create documentation on the final model

14-11-2022 32
Hands on in R

14-11-2022 33
14-11-2022 34
 Certificate Course on
Operations Management and
Analytics
(August 2022-January 2023)
Business Simulation and Modeling
Prof. Pankaj Dutta
Professor of Decision Sciences
SJM School of Management
IIT Bombay
2022 1
2022 2
What is Objective of this Module/Session?

This session introduces you to the Business simulation and risk modeling of
business problems.
 After providing a brief concept of basic business simulation, (a) the
concept of Monte Carlo Simulation with examples pertaining to different
domains will be discussed, and then (b) risk analysis would be carried out
in the context of business simulation using advanced excel spreadsheet.
 @Risk software would be used to illustrate the business problems under
risk and subsequent statistical analysis would be drawn to address various
real world business problems/case study on supply chain.
 Generally, optimization models ignore uncertainty but Risk optimizer helps to find risk
associated with each strategy; accordingly we can seek out strategies that will help to
minimize risks while achieving desired goals.
 It uses the Monte Carlo Simulation, which is one of the widely used techniques by
corporate managers and scientists as an aid for decision making and performance
analysis.

2022 3
 Managerial Approach to Decision Making

Manager analyzes
situation (alternatives)

Makes decision to
resolve conflict

Decisions are
implemented

2022 4
 THE MODELING PROCESS

The Role of Managerial Judgment in the Modeling Process:

as applied to the first two stages of decision making.

Analysis
Model Results

Interpretation
Abstraction

Symbolic
World

Real
World

Management Intuition
Decisions
Situation

2022 5
 Business Simulation
 Strategic thinking and decision modeling of business problems
using simulation…
 Study of business dynamics/system behavior…
 Uncertainty and risk modeling of business problems…
 Simulation and statistical inference using excel spreadsheet…
 What if analysis…
 Examples…

What can be simulated?

Almost anything can and almost everything has...

Simulation’s greatest strength is its ability to answer “what

if” questions...
2022 6
 Business Simulation and Risk Modeling

- Simulation is the process of designing a

Simulation: model of a real system and conducting
experiments with this model for the
Monte Carlo Simulation purpose of understanding the behaviour
How does it works? (within the limits imposed by a criterion or
set of criteria) for the operation of the
system.

Risk Modeling:
What is Uncertainty?
What is risk?
Risk in Optimization Problems?

2022 7
 Applications:

• MANUFACTURING: material handling systems, assembly lines, automated

production facilities, inventory control systems, plant layout, etc..
• COMPUTER SYSTEMS: hardware components, software systems, networks,
data base management, information processing, etc..
• FINANCE and BUSINESS: stock and commodity analysis, pricing policies,
marketing strategies, cash flow analysis, forecasting, etc..
• GOVERNMENT: military weapons and their use, military tactics, population
forecasting, land use, health care delivery, fire protection, criminal justice,
traffic control, etc..
 Telecommunications: almost all type of problems can be handled
 Applications at Disney World…

And the list goes on and on...

2022 8
 An experiment : Steady-state supply chain

 Suppose a simple supply chain that has been in steady-

state for some time. The Retailer’s inventory has been
constant at some level for a long time,

2022 9
 Steady-state supply chain

# A retailer maintains an inventory of product that is shipped to

customers on demand.
# Upon shipping, the retailer always orders immediately from his
supplier the same amount of product just shipped.
# The supplier also is very regular. He always deliveries the
product to retailer 7 days after the he places the order.
# The supplier has never been out-of-stock (and never will be!).
# No product shipped by the supplier is ever, or will ever be,
defective, damaged or lost in transit.

2022 10
 Demand changes

 Suppose, all of a sudden, the volume of demand from

customer coming into the retailer steps up to a new
higher level, and then remains there.

Sketch the new behavior:

On the axes provided in Figure, sketch the
pattern you think will be traced by the level of
the retailer's inventory, over time, following
the one- step-increase to customer demand.

2022 11
 The retailer's inventory behavior

 following the step-increase in demand, the Retailer's inventory

will decline in a straight-line manner for 7 days; it then will level
off and remain at the new, lower level.
12
 Steps of Simulation Process

Identify the problem: The simulation process is used to solve any problem
only when the assumptions required for analytical models are not satisfied or
there is no appropriate model developed for a system under study. For
example, a queuing situation may be of interest but the arrival and/or service
pattern do not meet the assumptions required to use queuing theory.

Identify the decision variables and decide the performance criterion

(objective): In the context of an inventory control situation, the demand
(consumption rate), lead time and safety stock are identified as decision
variables. These variables shall be responsible to measure the performance of
the system in terms of total inventory cost under the decision rule – when to
order.

2022 13
 Construct a simulation model: For developing a simulation model, an
intimate understanding of the relationships among the elements of the
system being studied is required. For this purpose the influence diagram
(drawn in a variety of different ways) is useful because simulation models
for each of these diagrams may be formulated until one seems better or more
appropriate than the other. Even after one has been chosen, it may be
modified again and again before a final version is acceptable.

2022 14
 Testing and validating the model: The validation process requires (i)
determine whether the model is internally correct in a logical and
programming sense called internal validity and (ii) determine whether it
represents the system under study called external validity. The first step
involves checking the equations and procedures in the model for accuracy,
both in terms of mistakes (or errors) and in terms of properly representing the
system under study. The verification of internal validity can be simplified if
the model is developed in modules and each module is tested as it is
developed.
After verifying internal validity the model is tested by substituting historical
values into the model and seeing if it replicates what happens in reality. If the
model passes this test, extreme values of the input variables are entered and
the model is checked for the expected output.

2022 15
 Designing of the experiment: Experimental design refers to controlling the
conditions of the study, such as the variables to include.

It requires to determine factors considered fixed and variable in the model

(ii) levels of the factors to use, (iii) what the resulting dependent measures
are going to be, (iv) how many times the model will be replicated, and
length of time of each replication, and so on. For example, in a queuing
simulation we may consider arrival and service rates constant but vary the
number of servers and the evaluate the customer waiting times. (dependent
variable).

2022 16
 Identify the Problem

Run the simulation model: Run the

Identify Decision Variables,
model on the computer to get the Performance Criterion and
Decision Rules
results in the form of operating
characteristics.
Construct Simulation Model
Evaluate the resultant correct-ness. Modify the
Validate the Model model by
If the simulation process is changing the
complete, then select alternatives: input data, i.e.
Design Experiments (specify values of
Examine the results of problem as values of decision variables decision
to be tested) variables
well as their reliability) otherwise
make desired changes in model Run or Conduct the Simulation
decision variables, parameters e
Is simulation process completed?
best course of action (or design,
and return to Step 3.
Examine the results and
Select the best course of action

Steps of Simulation Process

2022 17
 Modeling Simulation Process

2022 18
 What is Monte Carlo simulation?

• In a Monte Carlo simulation we attempt to follow the ‘time dependence’ of

a model for which change, or growth, does not proceed in some rigorously
predefined fashion but rather in a stochastic manner which depends on a
sequence of random numbers which is generated during the simulation.
• A large proportion of the applications of simulations are for probabilistic
models.
• The Monte Carlo technique is defined as a technique for selecting numbers
randomly from a probability distribution for use in a trial (computer run) of
a simulation model.
• The partitioned roulette wheel in Monte Carlo replicates the probability
distribution for demand if the values of demand occur in a random manner.

2022 19
 Monte Carlo Simulation

 The Monte Carlo simulation technique

involved conducting repetitive experiments on the model of the
system under study
with some known probability distribution to draw random samples
(observations)
using random numbers.

 If a system cannot be described by a standard probability distribution, an

empirical probability distribution can be constructed.

2022 20
 Steps of Monte Carlo Simulation

Setting up a probability distribution for variables to be analyzed.

Building a cumulative probability distribution for each random
variable.
Establishing an interval of random numbers for each variable.
Generate random numbers and then assign an appropriate set of
random numbers to represent value or range (interval) of values for
each random variable.
Conduct the simulation experiment using random sampling.
Repeat Step 4 until the required number of simulation runs has
been generated.
Design and implement a course of action and maintain control.

2022 21
How to design the scheme for Monte Carlo Simulation?

Let Y=Objective function = function(x) be a real-valued function

X= input variables  stochastic
Y= output variables  stochastic/deterministic
X 1 2 3 4
Prob. .2 .1 .4 .3
Calculate F(x)  Cumulative distribution function
X: f(x) F(x) Range of U(two digit random number)
1 .2 .2 00 - 19
2 .1 .3 20 - 29
3 .4 .7 30 - 69
4 .3 1 70 - 99

2022 22
 Discrete probability case:

Let Y=Objective function = function(x) be a real-valued function

X= input variables  stochastic
Y= output variables  stochastic/deterministic
X 1 2 3 4
Prob. .2 .1 .4 .3

Calculate F(x)  Cumulative distribution function

X: f(x) F(x) Range of U(two digit random number)
1 .2 .2 00 - 19

2 .1 .3 20 - 29
3 .4 .7 30 - 69
4 .3 1 70 – 99

Let us generate the random numbers: 17, 67, 34, 56, 20, 01, 98, 76, 25, …
Actual Input data: 1, 3, 3, 3, 2, 1, 4, 4, 2

2022 23
Probabilistic case: Continuous distribution

For Exponential distribution

𝑓 𝑥 = 𝜆 𝑒 −𝜆𝑥 ; 𝑥 ≥ 0 & 𝜆 > 0
𝐹 𝑥 = 1 − 𝑒 −𝜆𝑥
Let, F(x) = u and the random number generating scheme is
−1
𝑥= ln 𝑢 ; 𝑢 ∈ [0,1)
𝜆
For example: The inter-arrival rate in a waiting line system (Restaurant/call center, etc.)
follows exponential distribution

For Triangular distribution?

For Normal distribution?
For Uniform distribution?

2022 24
How to simulate values of a normal random
variable?
If you enter into any cell the formula NORMINV(rand(), mu , sigma), you
will generate a simulated value of a normal random variable having a mean
mu and standard deviation sigma. For example, let mu = 40000 (E1) and
sigma = 10000 (E2), then the simulated mean and sigma is illustrated as:

2022 25
 Example (Discrete Case): Simulation of Queuing
Problems
A dentist schedules all his patients for 30-minute appointments. Some of the
patients take more or less than 30 minutes depending on the type of dental work
to be done. The following summary shows the various categories of work, their
probabilities and time actually needed to complete the work:

Category of Time Required Probability

Service (minutes) of Category
Filling 45 0.40
Crown 60 0.15
Cleaning 15 0.15
Extraction 45 0.10
Checkup 15 0.20

 Simulate the dentist’s clinic for four hours and determine the average waiting time for the patients
as well as the idleness of the doctor. Assume that all the patients show up at the clinic at exactly
their scheduled arrival time starting at 8.00 a.m. Use the following random numbers for handling
the above problem: 40 82 11 34 25 66 17 79

2022 26
 Example (Discrete Case): Simulation of Queuing
Problems

Simulate the dentist’s clinic for four hours and determine the average
waiting time for the patients as well as the idleness of the doctor. Assume
that all the patients show up at the clinic at exactly their scheduled arrival
time starting at 8.00 a.m. Use the following random numbers for handling
the above problem:
40 82 11 34 25 66 17 79

2022 27
 Example (Discrete Case): Simulation of Queuing
Problems
Solution: The cumulative probability distribution and random number
interval for service time are shown below:

Category Service Time Required Probability Cumulative Random Number

of Service (minutes) Probability Interval

Filling 45 0.40 0.40 00–39

Crown 60 0.15 0.55 40–54
Cleaning 15 0.15 0.70 55–69
Extraction 45 0.10 0.80 70–79
Checkup 15 0.20 1.00 80–99

The various parameters of a queuing system such as arrival pattern of

customers, service time, waiting time in the context of the given problem
are shown below:

2022 28
 Arrival Pattern and Nature of Service

Patient Scheduled Random Category of Service Time

Number Arrival Number Service (minutes)

1 8.00 40 Crown 60
2 8.30 82 Checkup 15
3 9.00 11 Filling 45
4 9.30 34 Filling 45
5 10.00 25 Filling 45
6 10.30 66 Cleaning 15
7 11.00 17 Filling 45
8 11.30 79 Extraction 45

2022 29
 Simulate the Process

Computation of Arrivals, Departures and Waiting of Patients

Time Event Patient Number Waiting
(Patient Number) (Time to Exit) (Patient Number)
08.00 1 arrive 1 (60) –
08.30 2 arrive 1 (30) 2
09.00 1 departs; 3 arrive 2 (15) 3
09.15 2 depart 3 (45) –
09.30 4 arrive 3 (30) 4
10.00 3 depart; 5 arrive 4 (45) 5
10.30 6 arrive 4 (15) 5, 6
10.45 4 depart 5 (45) 6
11.00 7 arrive 5 (30) 6, 7
11.30 5 depart; 8 arrive 6 (15) 7, 8
11.45 6 depart 7 (45) 8
12.00 End 7 (30) 8
2022 30
 Average Waiting Time for Dentists

Patient Arrival Time Service Starts at Waiting Time

(minutes)

1 8.00 8.00 0
2 8.30 9.00 30
3 9.00 9.15 15
4 9.30 10.00 30
5 10.00 10.45 45
6 10.30 11.30 60
7 11.00 11.45 45
8 11.30 12.30 60
280
The average waiting time = 280/8 = 35 minutes.

2022 31
 Example (Continuous Case): A Model for
Profit on a Special Sale Promotion

A large catalog merchandiser is planning to have a special furniture

promotion a year from now. To do this the company must place its order
for the furniture now. It plans to sign a contract with the manufacturer for
3000 chairs at a cost of $175 per unit, which the company plans to offer
initially for $250 per unit. The promotion will last for eight weeks, after
which all remaining units will be offered for sale at a half the initial price,
or $125 per unit. The company believes that 2000 units will be sold during
the first eight weeks.
Represent the profit from this sale.

2022 32
 Example (Continuous Case): A Model for
Profit on a Special Sale Promotion

P = profit
C = per unit price ($175)
R = initial price ($250)
S = units ordered (3000)
V = no of units sold during the first 8 weeks of the promotion (2000)

R-C = net profit per unit sold for first 8 weeks

R/2-C = net profit or loss per unit after 8 weeks.
S-V = no of units sold at the discounted price R/2.

Profit(P) = (R-C)V + (R/2-C)(S-V)

Thus the profit will be exactly $100,000.

2022 33
 Example (Continuous Case): A Model for Profit on a
Special Sale Promotion

 Demand is uncertain and follows Triangular distribution

 Sales price is also uncertain and follows Uniform distribution
 Simulation: Excel Spreadsheet Approach?
 Excel: furniture promotion

 Post Spreadsheet Analysis: Confidence interval for profit?

2022 34
Example (Continuous Case): A Model for Profit on a
Special Sale Promotion

Lets simulate the model using excel ?

2022 35
 Decision Making under Uncertainty with
RISK Optimizer

 @Risk – Risk Analysis using Monte Carlo Simulation

 Methodology:
 The models and techniques combining with the power of Microsoft Excel and
Palisade Corporation’s RISK Optimizer add-in to find the best answers to problems
affected by uncertainty.
 Genetic optimization combines with Monte Carlo simulation to provide optimal
answers for many difficult, often previously unsolvable, real-world business
problems.

 Simulation with the Excel Add-in @Risk

2022 36
 Thousands of companies worldwide in all
areas of business and industry, including a
majority of Fortune 500 companies -
amounting to hundreds of thousands of
users, make productive use of Palisade
software every day.

2022 37
 Decision Making under Uncertainty using MC
Simulation

 @Risk – Risk Analysis using Monte Carlo Simulation

 @Risk – Discounted Cash Flow Analysis
 @Risk – Insurance – Loss Reserves Estimation
 @Risk – Product Mix Problems
 @Risk – Case Study on Supply Chain Network Design
 @Risk – Portfolio Optimization
 @Risk – Project Management/Project Scheduling under Risk
 @Risk – Inventory Management
 @Risk – Supply Chain Analysis
 @Risk – Finance, Marketing, Manufacturing, etc.
 Precision Tree – Decision Analysis
 Evolver- Optimization using Solver- What’s new?

2022 38
 A Resilient Supply Chain Network Design for
Perishable Products:
A Case Study of a Milk Industry

 Perishable products are known as short life cycle products and have
obsolete or deteriorating in nature.
 As milk supply chain is one of the most untouched but extremely relevant issue
in developing countries like India due its poor transport, infrastructure and
refrigerating facilities which cause huge economic loss, we thought of considering
the case of milk supply chain.
 The case company selected under study was from Chhattisgarh state
of India.
 Why Milk Industry?
 Very less research on perishable products
 In 2014-15, a total of 146.3 million tons of milk produced by India;
which accounts to 18.5% world’s milk production
 It is estimated that the milk production would rise to 200 million tons
by 2022
 Major challenges faced are (Problems in the case):
 inadequate storage and chilling facilities at the processing units
 lack of infrastructure during transportation and distribution which leads to
disruption
 State’s poor road network
 Lack of investment in cold chain
 The company often encounters the leakages in the tin/cans and around 3% of the
quantity to be received by the processing units is lost as spillage and pilferage
 During transportation of packets from processing units to retailers on an average 4%
of the packets are spoiled due to improper refrigeration
 Questions?

 Q1. Why is there a need for managing the supply chain of ‘perishable
products’ in contrast to the ‘conventional products’ especially in
developing countries, where proper infrastructure and transportation
facilities are not adequate?
 Q2. How to design an optimal supply chain network and to develop a
suitable distribution planning under disruption risks for the supply chain of
perishable products?
 Q3. If there is an order mismatch between demand and supply at the
downstream supply chain, then what is the suitable way to handle such
issues in the context of supply chain of perishable products under various
disruption scenarios?
 Proposed supply chain network under study

Case background
 IB group is an old Indian company located in
Chhattisgarh which has cow’s milk as its main
product
 Procures from 2 suppliers with a total of
5000+ cows
 Milk is transported from the suppliers to one
of the 3 processing units for processing
 The packaged milk is then sent from
processing units to 5 different retailers located
across four districts in the Chattisgarh state
 There are two available routes from each
supplier to each processing unit thus giving
rise to 12 unique combinations
 From each of the processing facility, the
product is distributed to each retailer by a
single path only, thus giving rise to 15 unique
paths
 Transportation of packaged milk is done either
in refrigerated vans or in insulated milk tanker
vans at a temperature of 4 degree Celsius
 Supply chain cost components

Supply chain cost from supplier s’ to manufacturer m through route f’:

….(1)
Gs’ : Unit material/production cost (which includes material cost plus holding cost) at supplier s
Cs’mf’: Unit cost of shipping of from supplier s to manufacturer m through route f’
Decision variables:
βs‘mf’s: Fraction of supply disruption at transportation route f’ (between supplier s’ and
 xsmf’s : quantity or
manufacturer m) amount of raw material
𝜎′ : Unit penalty cost of disruption in transportation link between the supplier and the shipped from supplier s
to manufacturer m
manufacturer
through route f’
 xmrfs : quantity of final
product shipped from
manufacturer m to
Supply chain cost from manufacturer m to retailer r: retailer r through rote f
….(2)  xms : quantity
outsourced from third
Gm : Sum of unit processing, packaging and unit holding cost at manufacturer m
party manufacturer
Cm: Unit cost of outsourcing (+transportation cost) from third party manufacturer  to primary  ys : Binary variable
manufacturer m equals to 1 if supplier s is
open (or selected at
Cmrf: Unit cost of shipping from manufacturer m to retailer r through route f
candidate location) and 0
βmrfs: Fraction of supply disruption in the transportation link f (between manufacturer m and otherwise
retailer r)  ym : binary variable
equals to 1 if
𝜎 : Unit penalty cost of disruption in transportation link between the manufacturer and the
manufacturer m is open
retailer (or selected at candidate
location) and 0 otherwise
 Supply chain cost components

Supply chain cost at Retailer r

= Ordering/Handling cost + Excess cost (of Overstocking) + Shortage cost (of Understocking)

The Newsvendor Model

 For illustration, two

….(3) cases arises here

 D > Q  stockout, at a
CE: Unit excess cost to retailer cost of: cu (D – Q)+ =
cu max{D –Q, 0}
Cs: Unit shortage cost to retailer
Or: Unit ordering/handling cost at retailer r  D < Q  overstock, at
a cost of co (Q–D)+ =
X’mrfs : quantity of final product received at retailer r through route f co max{Q – D, 0}
Dr: Demand at retailer r
 Last two parts of
equation (3) form a
Convex function

# The Total Supply Chain Cost is the sum of the equations (1), (2) and (3).
# The demand considered in our study as normally distributed with mean a and variance b.
# On putting the cumulative demand distribution and on rearranging the terms, we obtain a mixed integer stochastic programming model.
 Total Supply Chain Cost
 The Total Supply Chain Cost is the sum of the equations (1), (2) and (3) + the fixed
charged cost.

Fixed charge cost Manufacturer level cost Supplier level cost Retailer level cost

H’s’: Supplier's fixed opening cost at candidate location s’

Hm: Manufacturer's fixed opening cost at candidate location m
(Dr): Cumulative distribution function of Dr
E(Dr): Expected demand at retailer r
 List of Constraints
Descriptions: Constraints:

Manufacturer’s
capacity constraint

The amount of capacity that is missed by the manufacturer, m, as a result of disruptions and the quantity that is shipped
from the third party manufacturer should be greater or equal to the total quantity of goods that is to be shipped to the
retailer.

Supplier’s capacity
constraint

It ensures that the quantity to be shipped from supplier, s’ , should not exceeds its capacity.

Budget constraint

It imposes a budget constraint on suppliers and manufacturers in opening facilities.

Flow balance
constraint

It imposes a flow balance between the higher and lower echelons.

Fill Rate
constraint
It ensures that the desired fill rate level is greater or equal to Z
Non-Negativity
constraint

These two are for non-negativity and binary requirements, respectively.

Binary
constraint
The Simulation Framework
 @Risk Modeling
 In order to take into account the uncertainty of the
disruptions, we model the problem in @Risk analysis
software
 We analyze the effect of the disruptions in the transportation
link between the manufacturers, retailers and suppliers on
the total supply chain cost
 We used @Risk v7.5 for running the simulation where the
disruptions are considered to follow a normal distribution
curve of identified mean and variance
 We analyse the effect of disruption at routes:
 Supplier to Manufacturer
 Processing Unit to Retailer
 Total Disruption at Routes (from supplier to processing unit as well as
processing unit to retailer)
 Input data

Costs involved • Demand for each retailer is uncertain and it is observed

that demand follows a normal distribution
• The parameters associated with retailers like per unit
 Cost of production at the Supplier handling cost and demand and its standard deviation are
 Transportation cost – Supplier to Processing Unit given below:
 Per unit cost of Production at Processing Unit Procurement
Retailer Expected demand Standard Deviation
 Transportation Cost – Processing Unit to Retailer Cost
 Procurement Cost for Retailer Retailer 1 1.5 4000 2-3%
 Overstocking Cost (excess products at the retailer) Retailer 2 1 3000 2-3%
 Understocking Cost (unable to cater the demand) 1.3 4000 2-3%
Retailer 3
 Fixed Cost – Supplier and Processing Unit
Retailer 4 1.25 4500 2-3%

Retailer 5 1.3 2300 2-3%

• The parameters associated with suppliers and manufacturing/processing units are given below:

Processing Per unit processing Per unit

Capacity Fixed cost Supplier Capacity Fixed cost
Unit cost production cost

2.7 7000 250000 Supplier 1 17.5 15000 150000

Unit 1
2.5 4000 200000 Supplier 2 17 10000 100000
Unit 2
2.1 11000 350000
Unit 3
 Input data
Transportation costs

 There are two possible routes from each supplier • The per unit transportation costs from the
to Processing unit and the per unit cost for each manufacturers to the retailers are as follows:
route are mentioned in the below table:
Transportation
r1 r2 r3 r4 r5
Transportation Cost f1 f2 Cost
s1-m1 5 6 m1 8 8.5 9 7.5 8.5
s1-m2 5.5 6.5 m2 7.5 7 7 8.5 8
s1-m3 6 5.5 m3 8 9 7.5 8 9
s2-m1 6.5 6
s2-m2 5.5 7
s2-m3 6 7

• Disruption percentages: We have included a factor for disruption, β, which signifies the amount of disruption that can take place
in the routes during transportation. This disruptions can occur in the routes from the supplier either to the manufacturer or vice
versa. If disruption occurs at a route, the remaining quantity of (1-β) is received at the destination

f1 f2 r1 r2 r3 r4 r5

s1-m1 0.03 0.04 m1 0.02 0.04 0.03 0.04 0.01

s1-m2 0.02 0.02
m2 0.04 0.02 0.04 0.05 0.03
s1-m3 0.04 0.04
s2-m1 0.05 0.05 m3 0.05 0.02 0.04 0.05 0.03

s2-m2 0.03 0.03

β from Processing unit to supplier
s2-m3 0.02 0.04

β from supplier to Processing unit Disruption scenario and the probability of occurrence
Prob. Of
Scenarios All routes m1 m2 m3
occurrence
1 1 0 1 1 0.625
2 1 0 1 0 0.15
3 1 1 0 0 0.1
4 1 0 0 1 0.05
5 1 1 1 0 0.075
 Results:

Disruption at all routes

 Disruptions with uncertainty at all routes Location Allocation
from processing units to retailers as well as m1 1 s1 1
from suppliers to processing units 1
m2 s2 1
 Fill rate of 95% is considered m3 1
 The solution of optimal placements is given
below: Decision Variables for Processing Unit and
suppliers

Total Cost of Supply Chain (for 90 days) = ₹ 5,67,14,210

Complete disruption case Comparative Results Disruption free case

Quantity shipment decisions: Complete disruption case
The total cost of the supply chain is ₹ 5,67,14,210
Sensitivity Analysis and variation of total cost at different fill rate
DISRUPTION AT ALL ROUTES - RESULTS

 The total cost of the supply chain follows a normal distribution as shown in the above
figure
 It is observed that for 90% certainty the cost would lie between ₹ 6,09,97,320 and
₹6,10,05,340 with a mean of ₹6,10,08,881
 There is a 5% chance that cost is more than ₹6,10,05,340 and less than ₹ 6,09,97,320 .
 DISRUPTION AT ALL ROUTES - INSIGHTS

Tornado graph: Effect of disruptions on the total cost Scatter plot diagram

 From the tornado graph, we can find out which are the most critical routes that influence and drive up the costs
 This information is very useful and actions can be taken to enhance those routes thereby driving down the costs
 In the case, the route from s2-m1/f2 is the most critical route. If the disruptions in this route can be reduced or
made under control, there can be significant reductions in cost
 We can observe the top 10 critical routes/factors that influence cost from the tornado graph

• The scatter plots also reinstate that s2-m1/f2 is the most critical path

• In addition, it is positively correlated (Pearson correlation coefficient is 0.6439) to the cost function. R2 value for this
link is |0.6439|2 = 0.4146. This means 41.4% variability in the total cost can be explained by the disruptions at
route s2/m1. Similar observations can be made for all the routes as well.
Thank you!
@Risk by Palisade
Uses Monte Carlo simulation

- Some operation/ supply chain problems will be

discussed/ illustrated in the class/ session (This is just a
reference file)

1
Introduction

Risk Analysis
 Any method - qualitative and/ or quantitative
for assessing the impacts of risk on decisions.

 Many Risk Analysis methods blend both

qualitative and quantitative techniques.

 The goal of these methods is to help the

decision-maker choose a course of action.

2
Features

Uses Monte Carlo simulation to take all

possible outcomes into account.
The result is a distribution of possible
outcomes and the probabilities of getting
those results.
One of the strengths of Monte Carlo
simulation is that it produces enough data to
create accurate graphs. Histograms,
cumulative curves, area and line graphs, etc.

3
Features +
Create a summary report of results using
the Quick Report command.
Report in Excel containing a histogram, cumulative
curve, Tornado graph, and summary statistics.
Sensitivity Analysis
Determines which input distributions have the biggest
impact on the outputs.

Scenario Analysis
I dentifies combinations of inputs - or scenarios - which
lead to output target values.
4
Risk Analysis with @RISK
 Risk Analysis in @RI SK is a quantitative method that seeks to
determine the outcomes of a decision as a probability distribution.

I n general, Risk Analysis with @RI SK encompasses four steps:

 1. Develop a Model - Define the problem or situation in an Excel
worksheet format.

 2. Identifying Uncertainty - Determine which model inputs are

uncertain, and represent those using ranges of values with @RI SK
probability distribution functions. I dentify which result or output of
the model to analyze.

3. Analyzing the Model with Simulation - Run the simulation
to determine the range and probabilities of all possible outcomes
for the outputs identified.

 4. Make a Decision - With complete information from analysis

and preferences, make a decision.
5
Developing a Model
Model construction is based on
spreadsheet data

 Examples:
Launching a new product - impact of potential profits

Potential pollution effects of a new factory on a river

Effectiveness of a new drug therapy on an illness

Exploratory site for oil wells

6
Identifying Uncertainty

Determine which inputs in your model are

uncertain

Probability distribution functions to represent

a range of possible values

 @RI SK takes uncertainty into account

7
Risk Distribution

8
RISKview

9
Best Fit

10
Analyzing the Model with
Simulation
Capabilities
Specify the number of iterations
Update the spreadsheet in real-time
numerically or graphically

Control the convergence

Use the default settings - @RI SK will automate
 Provides a range of possible outcomes and
probabilities of occurrence
11
Simulation-Optimization

12
Sensitivity Analysis

13
Make a Decision

Allows multiple simulations to be run back-

to-back.

Correlate I nputs for More Accurate Models.

Statistics and Graphing

Custom Applications

14
Correlation

15
Graphing and Reporting

16
Model - Results

17
 Certificate Course on
Operations Management and Analytics
(August, 2022-January, 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
09-01-2023 1
 Module IX

• Introduction to Multi-Criteria Decision Making (MCDM) techniques

• Analytical Hierarchy Process (AHP) (Session 1)

• Interpretive Ranking Process (IRP) (Session 2)

• Interpretive Structural Modelling-I (ISM) (Session 3)

• Interpretive Structural Modelling-II (ISM) (Session 4)

• Case Study (Session 5)

09-01-2023 2
Data Analytics Methodology

3
Venn Diagram for Source of Data

4
 Module IX

Module VIII

Module XI

Decision Science Analytics

5
Data and Decision Sciences = Data + Analytics = DATA ANALYTICS

6
Data and Decision Sciences = Big Data + Analytics
= BIG DATA ANALYTICS

7
Source: Gilvan C. Souza, Supply chain analytics, Business Horizons (2014) 57, 595—605

8
 Models and Analytics
o Simulation Model (Predictive Analytics-Module VII / VIII)

o Qualitative Model (Descriptive Analytics-Module IX)

o Quantitative Model (Prescriptive Analytics-Module XI)

AI/ML
o Heuristics/ Meta-heuristics Model

o Hyper-heuristic Model
9
o Simulation Based Decision Models

(Predictive Analytics)
 Monte Carlo model
 Discrete (event) based model
 Combined Discrete/Continuous model

Module VII / VIII

10
o Qualitative Models
 Delphi method
 AHP (covered)
 IRP (covered)
 ANP (Descriptive Analytics)
 ISM (covered)
 DEMATEL
 TOPSIS Module IX
 PROMETHEE
 ELECTRE
 EATWOS
 HYBRID models
 Etc.
11
 (Prescriptive Analytics)
o Quantitative Models
Operations Research Models
L.P./N.L.P. (Single Obj.) & Goal Prog Models (Multi Obj.)
Binary/ Integer/Mixed Integer models
 Static/ Dynamic models
 Stochastic/ Probabilistic models
 Statistical (Forecasting) Models

 Use of EXCEL/ LINGO

Module XI
12
 Generic Analytical Model
Y = f (X1, X2, …, Xn)

Y = dependent variable

Xi = independent variables (inputs impacting Y)

f(.) = function defining the relationship Xi & Y

13
 Model Form of Independent Techniques
Category f(.) variables
Predictive Unknown, Known Regression, Time
(Module VII ill defined series,
& VIII) discriminant
Descriptive Know well Unknown or MCDM tools &
(Module IX) defined uncertain techniques
Prescriptive Know well Known or LP,GP, MOLP
(Module XI) defined certain

14
 Session 1
Analytical Hierarchy Process (AHP)
 Introduction
• Conflicting Objectives and Tradeoffs in Decision Problems
– e.g. higher returns vs. lower risks in investment, better
performance vs. lower price of computer
• Objectives with Incomparable Attribute Scales
– e.g. maximize profits vs. minimize impacts on environments
• Multi-Criteria Decision Making (MCDM)
– A study of methods and procedures that handle multiple
attributes/criteria
– Usages
• Identify a single most preferred alternative
• Rank alternatives
• Shortlist a limited number of alternatives
• Distinguish acceptable from unacceptable possibilities

1/9/2023 16
Most Decision Problems are Multi-criteria

• Maximize profits
• Satisfy customer demands
• Maximize employee satisfaction
• Satisfy shareholders
• Minimize costs of production
• Satisfy government regulations
• Minimize taxes
• Maximize bonuses

1/9/2023 17
 The Analytic Hierarchy
Process (AHP) for Decision
Making

Decision Making involves setting priorities

and the AHP is the methodology for doing
that.

1/9/2023 18
 What is AHP?
• A Process that Leads One to (Saaty, 1980)
– Structure a problem as a hierarchy or as a system with
dependence loops
– Elicit judgments that reflect ideas, feelings, and emotions
– Represent those judgments with meaningful numbers
– Synthesize results
– Analyze sensitivity to changes in judgments

• AHP also uses a weighted average approach idea, but it

uses a method for assigning ratings and weights that is
considered more reliable and consistent

1/9/2023 19
 Purposes of AHP
• To structure complexity in gradual steps from the
large to the small, or from the general to the
particular, so we can relate them with greater
accuracy according to our understanding

• To improve our awareness by richer synthesis of

our knowledge and intuition; AHP is a learning tool
rather than a means to discover the TRUTH

1/9/2023 20
 Steps in AHP
• Step 1: Decompose the problem into a hierarchy
– Start with an identification of the criteria to be used in evaluating
different alternatives, organized in a tree-like hierarchy

• Step 2: Collect input data by pairwise comparisons of criteria

at each level of the hierarchy and alternatives

• Step 3: Estimate the relative importance (weights) of criteria

and alternatives and check the consistency in the pairwise
comparisons

• Step 4: Aggregate the relative weights of criteria and

alternatives to obtain a ranking of each alternative with
regards to the goal
1/9/2023 21
 Hierarchy (Cont.)
• How to Structure a Hierarchy?
– Identify the overall objective or goal
– Identify criteria to satisfy the goal
– Identify, where appropriate, sub-criteria under each criterion
– Identify alternatives to be evaluated in terms of the sub-
criteria at the lowest level
– If the relative importance of the sub-criteria can be assessed
and the alternatives can be evaluated in terms of the sub-
criteria, the hierarchy is finished
– Otherwise, continue inserting levels until it is possible to link
levels and set priorities (relative weights) on the elements at
each level in terms of the elements at the level above it

1/9/2023 22
 Pairwise Comparison (Cont.)
• Scale for Pairwise Comparisons
– 1. Equally preferred
– 3. One is moderately preferred over the other
– 5. One is strongly preferred over the other
– 7. One is very strongly preferred over the other
– 9. One is extremely preferred over the other
– 2,4,6,8 intermediate values
– Reciprocals for inverse comparison

1/9/2023 23
 Hierarchy

GOAL

CRITERIA

ALTERNATIVES

1/9/2023 24
1/9/2023 25
1/9/2023 26
 Maximize Overall
GOAL
Job Satisfaction

Research Growth Benefits Colleagues Location Reputation CRITERIA

Job A Job B Job C ALTERNATIVES

Hierarchy for a Job Selection Decision

1/9/2023 27
1/9/2023 28
 Goal More General

C1 C2 C3

C11 C12 C13 C21 C22 C31 C32 C33

Sub-criteria at More Specific

the lowest level

Alternatives

Structure of a Hierarchy

1/9/2023 29
 Hierarchy (Cont.)

• How Large Should a Hierarchy Be?

– Large enough to capture decision maker’s major
concerns
– Small enough to remain sensitive to change in what is
important

1/9/2023 30
 Maximize Society’s
Overall Benefit

National Health, Safety Political

Economy Environment Factors

Accidents Political
Cheap Foreign Capital Natural Unavoidable Independ- Centra-
Long-Term Cooperative-
Electricity Trade Resources Resources Pollution ence lization
Risks ness

No big Coal-fired Nuclear

power plants power plant power plant
Energy Decision in the Parliament of Finland (Saaty, 1980)

1/9/2023 31
 Pairwise Comparison Matrix
Criterion Criterion Criterion Approximated Consistency
(Alternative) (Alternative) … (Alternative) Weight Measure
1 2 n
n W /W n
Criterion ~
~  1Ti i  (W1/Wi ) Wi
(Alternative)
1
W1/W1
T1
W1/W2
T2
… W1/Wn
Tn W1  i 1
n
CM1= i 1
~
W1
n W /W
Criterion  2Ti i
~ n ~
 (W2 /Wi ) Wi
(Alternative)
2
W2/W1
T1
W2/W2
T2
… W2/Wn
Tn W2  i 1
n
CM2= i1
~
W2

… … … … … … …
n W /W n ~
Criterion  nTi i  (Wn /Wi ) Wi
~
… Wn  CMn=
i 1 i 1
(Alternative) Wn/W1 Wn/W2 Wn/Wn ~
n Wn
n T1 T2 Tn
n n n
  
n
Wi Wi Wi ∑ CM i
Total T1= i 1 T2= i 1 … Tn= i 1 CI ≈ ( i=1 n - n) /(n - 1)
W1 W2 Wn

CI = 0 or λmax = n implies perfect consistency

1/9/2023 32
CI = 0.1 is the generally accepted threshold value
Example 1:
Understanding AHP from School Selection Problem

Parents wish to prioritize the selection of school for their kid. They identified
three schools based on their past knowledge and plan to prioritize these three
schools. They want to apply AHP to take the decision.

1/9/2023 33
Solution
Step 1:

Satisfaction With School GOAL

School Vocational College Music

Learning Friends
Life Training Preparation Classes
CRITERIA

School A School B School C ALTERNATIVES

1/9/2023 34
Step 2 & Step 3:
Pairwise Comparison Matrix among criteria

L F SL VT CP MC
L 1 4 3 1 3 4
F 1/4 1 1 1/3 1 2
SL 1/3 1 1 1/5 1/2 2
VT 1 3 5 1 1 3
CP 1/3 1 2 1 1 2
MC 1/4 1/2 1/2 1/3 1/2 1
Sum 3.17 10.50 12.50 3.87 7.00 14.00

Pairwise Comparison Matrix of Six Criterion

1/9/2023 35
 Approximated Eigenvector
L F SL VT CP MC Weight Weight
1/3.17 4/10.5 3/12.5 1/3.87 3/7 4/14
L 0.32 0.32
=0.32 =0.38 =0.24 =0.26 =0.43 =0.29
(1/4)/3.17 1/10.5 1/12.5 (1/3)/3.87 1/7 2/14
F 0.10 0.10
=0.08 =0.10 =0.08 =0.09 =0.14 =0.14
(1/3)/3.17 1/10.5 1/12.5 (1/5)/3.87 (1/2)/7 2/14
SL 0.09 0.09
=0.11 =0.10 =0.08 =0.05 =0.07 =0.14
1/3.17 3/10.5 5/12.5 1/3.87 1/7 3/14
VT 0.27 0.27
=0.32 =0.29 =0.4 =0.26 =0.14 =0.21
(1/3)/3.17 1/10.5 2/12.5 1/3.87 1/7 2/14
CP 0.15 0.15
=0.11 =0.10 =0.16 =0.26 =0.14 =0.14
(1/4)/3.17 (1/2)/10.5 (1/2)/12.5 (1/3)/3.87 (1/2)/7 1/14 0.07 0.07
MC
=0.08 =0.05 =0.04 =0.09 =0.07 =0.07

Normalized Pairwise Comparison Matrix of Six Criteria

1/9/2023 36
 Approximated Consistency
L F SL VT CP MC Weight Measure
L 1 4 3 1 3 4 0.32 6.22

F 1/4 1 1 1/3 1 2 0.10 6.5

SL 1/3 1 1 1/5 1/2 2 0.09 6.29

VT 1 3 5 1 1 3 0.27 6.30
CP 1/3 1 2 1 1 2 0.15 6.31

MC 1/4 1/2 1/2 1/3 1/2 1 0.07 5.86

(1•0.32 + 4•0.10 +3•0.09 +1•0.27 +3•0.15+ 4•0.07 )

CM1=
0.32 = 6.22
((1 / 4 )•0.32 +1•0.10 +1•0.09 + (1/3)•0.27 +1•0.15 + 2•0.07 )
CM2= 0.10 = 6.50
Likewise, we can calculate CM3, CM4, CM5, and CM6
CI =0.048 from
( 6.22 + 6.5+ 6.29 + 6.30 + 6.31+5.86 )
CI ≈ [ 6 - 6] /(6 - 1) = 0.049 eigenvalue
1/9/2023 37
 Pairwise Comparison Matrix among alternatives for each criteria
Approximated
L A B C A B C Weight

A 1 1/3 1/2 1/6 (1/3)/1.67 (1/2)/4.5 0.16

A
=0.17 =0.20 =0.11 [.17+.2+.11]/3
B 3 1 3
3/6 1/1.67 3/4.5 0.59
B
C 2 1/3 1 =0.50 =0.60 =0.67
Sum 6 1.67 4.5 2/6 (1/3)/1.67 1/4.5
C 0.25
=0.33 =0.20 =0.22
Pairwise Comparison Matrix
of Three Alternative Schools Normalized Pairwise Comparison
With Respect to Learning Matrix of Three Alternative
Schools With Respect to Learning

1/9/2023 38
 Approximated Consistency
L A B C Weight Measure CM
A 1 1/3 1/2 0.16 3.01
B 3 1 3 0.59 3.08

C 2 1/3 1 0.25 3.07

(10.16 1 / 30.59 1 / 20.25 )

CM1=
0.16  3.01
( 30.16 10.59 30.25)
CM2= 0.59  3.08
CM3= ( 20.161/ 30.5910.25)  3.07
0.25
Consistency
Index CI ≈ [ (3.013.3083.07 )  3] /(3  1)  0.027

1/9/2023 39
 Approximated Consistency
F A B C Weight Measure CM
A 1 0.25 1/3 0.12 3.00
B 4 1 2 0.56 3.02

C 3 1/2 1 0.32 3.02

Consistency
Index CI ≈  0.0091

1/9/2023 40
 Approximated Consistency
SL A B C Weight Measure CM
A 1 3 4 0.62 3.03
B 0.33 1 2 0.24 3.01

C 0.25 1/2 1 0.14 3.00

Consistency
Index CI ≈  0.0091

1/9/2023 41
 Approximated Consistency
VT A B C Weight Measure CM
A 1 5 2 0.59 3.56
B 0.2 1 2 0.22 3.22

C 0.5 1/2 1 0.19 3.12

Consistency
Index CI ≈  0.151

1/9/2023 42
 Approximated Consistency
CP A B C Weight Measure CM
A 1 0.167 0.11 0.07 3.02
B 6 1 2 0.54 3.19

C 9 0.5 1 0.39 3.19

Consistency
Index CI ≈  0.068

1/9/2023 43
 Approximated Consistency
MC A B C Weight Measure CM
A 1 0.11 0.167 0.06 3.02
B 9 1 5 0.71 3.34

C 6 0.2 1 0.23 3.13

Consistency
Index CI ≈ 0.08428

1/9/2023 44
Step 4:

L F SL VT CP MC
0.32 0.1 0.09 0.27 0.15 0.07
A 0.16 0.12 0.62 0.59 0.07 0.06 0.29

B 0.59 0.56 0.24 0.22 0.54 0.71 0.46

C 0.25 0.32 0.14 0.19 0.39 0.23 0.25

1/9/2023 45
The ranking of three alternatives i.e. schools would
be

Rank 1: School B

Rank 2: School A

Rank 3: School C

1/9/2023 46
 Composition and Synthesis
• Combine the relative importance of criteria and alternatives to
obtain a global ranking of each alternative with regards to the goal

Criteria Cj (j=1,2,…,n) and

their corresponding weights

wC1 wC2 … wCn

Composite impact
C1 C2 … Cn
A1 wA1C1 wA1C2 … wA1Cn OA1 = wC1wA1C1+wC2wA1C2+…+wCnwA1Cn
A2 wA2C1 wA2C2 … wA2Cn OA2 = wC1wA2C1+wC2wA2C2+…+wCnwA2Cn
… … … … … …

Am wAmC1 wAmC2 … wAmCn OAm = wC1wAmC1+wC2wAmC2+…+wCnwCnAm

Weights of alternatives Ai
(i=1,2,…,m) w.r.t. criteria Cj
1/9/2023 (j=1,2,…,n) 47
1/9/2023 48
 Session 2
Interpretive Ranking Process (IRP)

49
 Why IRP when we have AHP?

For “n” criteria and “m” alternatives, in AHP we need

“n+1” matrices to be constructed through questionnaires.
And, each matrix need to be checked for consistency to
finally get rankings.

where as

For “n” criteria and “m” alternatives, in IRP we need only

one matrix to be constructed through questionnaires and
there is no need for consistency check.
Four matrices in AHP for n=4 & m=3
C1 C2 C3 C4
Step 1:

C1 --
C2 --
C3 --
C4 -- Single matrix in IRP for n=4 & m=3
C1 A1 A2 A3
A1 -- A1 A2 A3
A2 --
A3 -- A1 --
A2 --
Step 2:

C2 A1 A2 A3
A1 --
A3 --
A2 --
A3 --
C3 A1 A2 A3 Criteria (C1, C2,C3,C4) to
A1 -- be written in each cell
A2 --
A3 --
C4 A1 A2 A3
A1 --
A2 --
A3 --
Steps of the Basic Process
 Case
Example

Source: Multiple-attribute decision making methods for plant layout design problem,
Robotics and Computer-Integrated Manufacturing, Volume 23, Issue 1, February 2007, Pages 126-137
 Table: Pair-wise comparison matrix for IRP
(derived from questionnaires used to get inputs for AHP)

55
56
Table: Net-Dominance matrix for IRP

57
 Let’s apply IRP for school example in excel and
compare the results of AHP and IRP.

1/9/2023 58
 Strengths
• It is easy to compare by the impact of interactions rather
than the variables in abstract sense.

• It is comparatively easy to judge the dominance of one

interaction over the other rather than the extent of
dominance.

• It is based on the strength of paired comparison as it does

not create any cognitive overload.

• It is not necessarily dependent on weightage of criteria,

which is a debatable issue in MCDM methodologies.
 Limitations
• It is based on interpretive and judgmental processes and at times
may be highly subjective.

• It usually treats all the criteria equally ignoring their relative

importance, as given in the illustration. However, this limitation can
be overcome by assigning ordinal weights to various criteria and
carrying out sensitivity/Scenario analysis.

• It is difficult to be validated in terms of objective validation tests.

• It is difficult to interpret a matrix of size beyond 10x10 as the

number of paired comparisons would exponentially increase, and
thus only modest sized problems can be effectively implemented
with this process.
 Session 3-4

Introduction to ISM and Risk

Mitigation in Operations Using ISM

09-01-2023 62
 Session 3
Introduction to ISM
MCDM tools and techniques are classified into
two main groups.

Alternatives (Options) are ranked considering

set of criteria. Examples: AHP, IRP, ANP,
TOPSIS, EATOWS, PROMETHEE etc.

Criteria are ranked or classified into driving

(causing effect) or dependent (being effected).
Examples: ISM, DEMATEL

1/9/2023 64
So, we will see how ISM works. Considering school
example problem or Facility Layout problem from
previous session.

We have following criteria in school problem.

• Learning
• Friendship
• School Life
• Vocational Training
• College Preparation
• Music Classes

1/9/2023 65
Similarly, we have following criteria in layout selection.

• Distance
• Adjacency
• Shape Ratio
• Flexibility
• Accessibility
• Maintenance

In AHP, we use scale 1-9 which generally bring in

consistency among decision makers. However, in ISM
only binary scale is used which reduced to large extent
inconsistencies.
1/9/2023 66
 In AHP In ISM

L F SL VT CP MC L F SL VT CP MC
L 1 4 3 1 3 4 L 1 0,1 0,1 0,1 0,1 0,1
F 1/4 1 1 1/3 1 2 F 1 0,1 0,1 0,1 0,1
SL 1/3 1 1 1/5 1/2 2 SL 0,1 0,1 1 0,1 0,1 0,1
VT 1 3 5 1 1 3 VT 0,1 0,1 0,1 1 0,1 0,1
CP 1/3 1 2 1 1 2 CP 0,1 0,1 0,1 0,1 1 0,1
MC 1/4 1/2 1/2 1/3 1/2 1 MC 0,1 0,1 0,1 0,1 0,1 1

1/9/2023 67
 INTERPRETIVE STRUCTURAL MODEL

• ISM is a process that is based on relational

mathematics, which clarifies and transforms
ill-structured and poorly articulated mental
models about the system under study into a
clear and interrelated structured set of
system elements.
 ORIGIN OF THE CONCEPTS
• J.N. Warfield, Binary matrices in system modeling, IEEE
Trans. Syst. Man Cyber., SMC-3 (September 1973), p. 5

• J.N. Warfield, Developing subsystem matrices in structural

modeling, IEEE Trans. Syst. Man Cyber., SMC-4 (January
1974), p. 1

• J.N. Warfield, Developing interconnection matrices in

structural modeling, IEEE Trans. Syst. Man Cyber., SMC-4
(January 1974), p. 1

• J.N. Warfield, Toward interpretation of complex structural

models, IEEE Trans. Syst. Man Cyber., SMC-4 (September
1974), p. 5
 ISM BACKGROUND

• Directed graph (“digraph”)

1 2

3 4
 • Binary matrix representation (“adjacency matrix”): can
we go from one node to another in one step or directly?

Ending Node:
1 2 3 4

1 1 1 1 0
 
2
1 0 1 0
1 0
Starting Node:
3 0 0
 
0 1 
4
 1 0
This is also referred as Initial Reachability Matrix
 • Connections between elements (“reachability matrix”): can
we go from one node to another in any number of steps?

Transitive links:
Ending Node:
2-2 (2-1-2)
1 2 3 4 3-2 (3-1-2)
3-3 (3-1-2-3)
1 1 1 0 4-1 (4-2-1)
 
1 4-3(4-2-1-3)

Starting Node: 2 1 1 1 0
1 1 1 0

3

1 1 1 1 

4

This is also referred as Final Reachability Matrix

 Restructured hierarchical digraph

1 2 3

1/9/2023 73
ISM provides two outputs
1) ISM Digraph (Hierarchy)
2) MICMAC Diagram

Ending Node:
1 2 3 4 P (Driving Power

1
1 1 1 0 3
 
Starting Node: 2 1 1 1 0 3

1 1 1 0 This is Final Reachability Matrix


3 3

1 1 1 1 

4 4
D (Dependence) 4 4 4 1
ISM Digraph (Hierarchy) is developed using Matrix
Partitioning Method

Iteration 1:

Variable Reachability Set Antecedent Intersection Level

Set Set
1 1,2,3 1,2,3,4 1,2,3 I
2 1,2,3 1,2,3,4 1,2,3 I
3 1,2,3 1,2,3,4 1,2,3 I
4 1,2,3,4 4 4
Iteration 2:

Variable Reachability Set Antecedent Intersection Level

Set Set
1 1,2,3 1,2,3,4 1,2,3 I
2 1,2,3 1,2,3,4 1,2,3 I
3 1,2,3 1,2,3,4 1,2,3 I
4 1,2,3,4 4 4 II
Based on the iteration 1 and iteration 2, variable 1, 2, & 3
are levelled at the first level, and the fourth variable
levelled at the second.

The hierarchy based on levelling partitioning can be

developed shown in next slide.

1/9/2023 77
 ISM hierarchical digraph

This is the first output from ISM which inform the

decision makers about the driving and dependent
variables.
 ISM MICMAC is developed using P and D values from the
binary matrix
Driving Power (P)

4 Independent Linkage

1, 2, 3
Autonomous Dependent

Dependent (D)

This is the second output from ISM which informs about

the magnitude & location of variables.
1/9/2023 80
 TYPES OF FACTORS/VARIABLES
Autonomous factors:
These factors have weak drive power and weak dependence. They are
relatively disconnected from the system, with which they have few links,
which may be very strong.

Dependent factors:
These factors have weak drive power but strong dependence.

Linkage factors:
These factors have strong drive power as well as strong dependence.
These factors are unstable in the fact that any action on these factors will
have an effect on others and also a feedback effect on themselves.

Independent factors:
These factors have strong drive power but weak dependence. A factor
with a very strong drive power, called the ‘key factor’ falls into the
category of independent or linkage factors.
 OBJECTIVE OF ISM
• To identify and rank driving and dependent
variables or elements.

• To establish the interrelationship among the

variables.

• To understand the managerial implication of

the business problem or research.
 STEPS OF ISM
1. Variables affecting the system under
consideration are listed, which can be objectives,
actions etc.

2. From the variables identified in first step,

contextual relationship is established between
pairs of variables.

3. A structural self-interaction matrix (SSIM) is

developed which indicates pairwise relationships
among variables.
V for the relation from element i to element j
and not in both directions (row to column)

A for the relation from element j to element i

but not in both directions (column to row)

X for both directional relations from element

i to j and j to i

0 (zero), no relation between the elements.

4. Reachability matrix is developed from the SSIM
and the matrix is checked for transitivity. The
transitivity of the contextual relation is a basic
assumption made in ISM. It states that if a
variable A is related to B and B is related to C, then
A is necessarily related to C.

5. The reachability matrix obtained in fourth step is

partitioned into different levels.

Reachability matrix is partitioned in rachability,

antecedent and intersection sets. Similar
reachability and intersection set defines a level.
6. A directed graph is drawn and the transitive links
are removed. The resultant digraph is converted
into an ISM, by replacing variable nodes with
statements.

7. The ISM model developed in seventh step is

reviewed to check for conceptual inconsistency
and necessary modifications are made.
 APPLICATIONS OF ISM
Year Area Research paper
1978 Technology R. H. Watson, Interpretive structural modeling -A useful
assessment tool for technology assessment? Technological
Forecasting and Social Change, 11(1978) 165-185

2005 Reverse V. Ravi, Ravi Shankar, Analysis of interactions among the

logistics barriers of reverse logistics, Technological Forecasting
and Social Change, 72 (2005) 1011-1029

2006 Supply chain M. N. Faisal, D.K. Banwet, Ravi Shankar, Supply chain risk
risk mitigation: modeling the enablers, Business Process
management Management Journal, 12 (2006) 535 – 552

2007 Supply chain A. Agarwal, Ravi Shankar, M.K. Tiwari, Modeling agility
agility of supply chain, Industrial Marketing Management, 36
(2007) 443–457
Year Area Research paper
2009 Reverse G. Kannan, S. Pokharel, P. S. Kumar, A hybrid approach
logistics using ISM and fuzzy TOPSIS for the selection of reverse
logistics provider, Resources, Conservation and
Recycling, 54 (2009) 28-36

2011 Green supply A. Diabat, K. Govindan, An analysis of the drivers

chain affecting the implementation of green supply chain
management, Resources, Conservation and Recycling, 55
(2011) 659-667
2012 3 PL K. Govindan, M. Palaniappan, Q. Zhu, D. Kannan, Analysis
of third party reverse logistics provider
using interpretive structural modeling, International
Journal of Production Economics, 140 (2012) 204-211

2013 Tourism L. Z. Lin, H. R. Yeh, Analysis of tour values to develop

management enablers using an interpretive hierarchy-based model in
Taiwan, Tourism Management, 34(2013) 133-144

89
Year Area Research paper
2014 Sustainable A .Diabat, D. Kannan, K. Mathiyazhagan, Analysis of
supply chain enablers for implementation of sustainable supply chain
management-A textile case, Journal of Cleaner
Production, 83 (2014) 391-403
2015 Eco-efficiency V. Ravi, Analysis of interactions among barriers of eco-
in packaging efficiency in electronics packaging industry, Journal of
Cleaner Production, 101(2015) 16-25

2015 Humanitarian Devendra K. Yadav, Akhilesh Barve, Analysis of critical

supply chain success factors of humanitarian supply chain: An
application of Interpretive Structural Modeling,
International Journal of Disaster Risk Reduction, 12
(2015) 213-225

90
 Session 4
Risk Mitigation in
Operations Using ISM

92
 EXAMPLE 1:
MODELING RISK MITIGATION IN SUPPLY CHAINS

93
 VARIABLES OR ENABLERS

1. Information sharing
2. Agility
3. Trust among SC partners
4. Collaborative relation among SC partners
5. Information security
6. Corporate social responsibility

94
 VARIABLES OR ENABLERS

7. Incentive & revenue sharing in supply chain

8. Strategic risk planning
9. Risk sharing in supply chain
10. Knowledge about risk in supply chain
11. Continual risk analysis and assessment

95
 STRUCTURAL SELF-INTERACTION MATRIX (SSIM)

Information sharing influences all other elements and influenced by trust (3) and
collaborative relations (4)

Agility does not influenced any other element.

 REACHABILITY MATRIX

Trust (3) influences information sharing (1) and information sharing improves
knowledge about various risks (10). So, trust and knowledge about various risks have a
transitive link.
 REACHABILITY MATRIX

Reachability set of Information sharing (1) have slements 1-11, whereas the antecedent
set have three elements i.e. 1, 3 and 4.
 MATRIX PARTITIONING: ITERATION 1

Elements 2 (agility) and 11 (continual risk analysis) have same reachability and
intersection sets. So, they will form the level I (at the top).
 MATRIX PARTITIONING: ITERATION 2

Elements 2 and 11 have been deleted from the reachability matrix. Now, elements 5
(information security) and 9 (risk sharing) have the same reachability and intersection
sets. So, they will form the level II.
 MATRIX PARTITIONING: ITERATION 3

Elements 5 and 9 have been deleted from the reachability matrix. Now, elements 6
(CSR) and 8 (strategic risk planning) have the same reachability and intersection sets. So,
they will form the level III.
ITERATIONS

102
ISM MODEL
MICMAC ANALYSIS

Independent Linkage

Dependent
Autonomous
 TYPES OF FACTORS/VARIABLES
Autonomous factors:
These factors have weak drive power and weak dependence. They are
relatively disconnected from the system, with which they have few links,
which may be very strong.

Dependent factors:
These factors have weak drive power but strong dependence.

Linkage factors:
These factors have strong drive power as well as strong dependence.
These factors are unstable in the fact that any action on these factors will
have an effect on others and also a feedback effect on themselves.

Independent factors:
These factors have strong drive power but weak dependence. A factor
with a very strong drive power, called the ‘key factor’ falls into the
category of independent or linkage factors.
 STRENGTHS OF ISM

o Helps in presenting a complex system in a

simplified way.

o Transforms unclear and poorly articulated

mental models of systems into visible,
well-defined models.

o Identifies the driver and driven elements.

 LIMITATIONS OF ISM
o Technique of ISM can be used only by
persons who are knowledgeable
about it and are trained to interpret
the data.

o Computer program is required to

handle problems with large number of
elements.
 Session 5

Case Study on Vendor Selection

09-01-2023 109
Falcon Industries Limited
About FIL
 Threat from domestic market

Response to domestic market

1/9/2023 111
 FIL’s inhouse capability

Bonnet Stud

1/9/2023 112
 Challenge with Bonnet Stud

Selecting right vendor for bonnet stud

1/9/2023 113
 Information about prospective vendors

1/9/2023 114
1/9/2023 115
1/9/2023 116
 Other issues

What decision Mr Joshi would like to take?

1/9/2023 117
 Solution
AHP/IRP techniques can be applied to help Mr. Joshi in
resolving question b & c. Let’s see how to apply these tools.
Step 1: Identify factors
1. Basic price
2. Location
3. Credit
4. Delivery
5. Quality certification
6. RM Testing
7. Quality inspection
8. MOQ

1/9/2023 118
Step 2: Identify alternatives (i.e. vendors)

1. Creative Engineering Pvt Limited

2. Super Fasteners Limited
3. Atlas Engineering Products Limited

Step 3a: Construct pairwise comparison matrices

among factors (i.e. criteria)

1/9/2023 119
Step 3b: Construct pairwise comparison matrices
among all vendors for each selected criteria

Step 4: Combine matrices output to rank vendors

This will handle Mr. Joshi’s b & c issues.

1/9/2023 120
 Certificate Course on
Operations Management and
Analytics
(August 2022-January 2023)

Program Faculty:
Dr Rajesh Matai
Birla Institute of Technology and Science Pilani
 Module X

Strategic Importance of Facility Layout

Types of Facility Layout

Process Layout (Session 1)

• Quantitative Design
• Qualitative Design
• QAP Design (Prescriptive Analytics Module XI)

Product Layout and Assembly Line (Session 2)

08-01-2023 2
 Session 1-2

Facility Planning

08-01-2023 3
Innovations at McDonald’s
 Indoor seating (1950s)
 Drive-through window (1970s)
 Adding breakfast to the menu (1980s)
 Adding play areas (late 1980s)
 Redesign of the kitchens (1990s)
 Self-service kiosk (2004)
 Now three separate dining sections

08-01-2023 4
Innovations at McDonald’s
 Indoor seating (1950s)
 Drive-through window (1970s)
Six out of the
 Adding breakfast to the menu (1980s)
seven are layout
 Adding play areas (late 1980s)
decisions!
 Redesign of the kitchens (1990s)
 Self-service kiosk (2004)
 Now three separate dining sections

08-01-2023 5
McDonald’s New Layout
 Seventh major innovation
 Redesigning all 30,000 outlets around the
world
 Three separate dining areas
 Linger zone with comfortable chairs and Wi-Fi
connections
 Grab and go zone with tall counters
 Flexible zone for kids and families
 Facility layout is a source of competitive
advantage

08-01-2023 6
Strategic Importance of Layout
Decisions

The objective of layout strategy is to

develop a cost-effective layout that
will meet a firm’s competitive needs

08-01-2023 7
Layout Design Considerations

 Higher utilization of space, equipment, and

people
 Improved flow of information, materials, or
people
 Improved employee morale and safer
working conditions
 Improved customer/client interaction
 Flexibility
08-01-2023 8
Types of Layout

1. Process-oriented layout

2. Product-oriented layout

3. Fixed-position layout

08-01-2023 9
Good Layouts Consider

1. Material handling equipment

2. Capacity and space requirements
3. Environment and aesthetics
4. Flows of information
5. Cost of moving between various
work areas

08-01-2023 10
Process-Oriented Layout

 Like machines and equipment are

grouped together
 Flexible and capable of handling a
wide variety of products or services
 Scheduling can be difficult and setup,
material handling, and labor costs
can be high

08-01-2023 11
 Patient A - broken leg
ER
triage Emergency room admissions
room
Patient B - erratic heart
Surgery pacemaker

Laboratories

Radiology ER Beds Pharmacy Billing/exit

Figure 9.3
08-01-2023 12
Methods to Design Process Layouts

 Goal: minimize material handling costs

Quantitative
 Block Diagramming Method
 minimize nonadjacent loads
 use when quantitative data is available
 Relationship Diagramming
 based on location preference between areas Qualitative
Method
 use when quantitative data is not available

08-01-2023 13
  Arrange work centers so as to
minimize the costs of material
handling

 Basic cost elements are

 Number of loads (or people) moving
between centers
 Distance loads (or people) move
between centers

08-01-2023 14
 n n

Minimize cost = ∑ ∑ Xij Cij

i=1 j=1

where n = total number of work centers or departments

i, j = individual departments
Xij = number of loads moved from department i to
department j
Cij = cost to move a load between department i and
department j

08-01-2023 15
Example 1
Arrange six departments in a factory to
minimize the material handling costs. Each
department is 20 x 20 feet and the building
is 60 feet long and 40 feet wide.
1. Construct a “from-to matrix”
2. Determine the space requirements
3. Develop an initial schematic diagram
4. Determine the cost of this layout
5. Try to improve the layout
6. Prepare a detailed plan

08-01-2023 16
 Number of loads per week
Department Assembly Painting Machine Receiving Shipping Testing
(1) (2) Shop (3) (4) (5) (6)

Assembly (1) 50 100 0 0 20

Painting (2) 30 50 10 0

Machine Shop (3) 20 0 100

Receiving (4) 50 0

Shipping (5) 0

Testing (6)

Figure 9.4
08-01-2023 17
 Solution
Area 1 Area 2 Area 3

Assembly Painting Machine Shop

Department Department Department
(1) (2) (3)
Initial
Random
Layout 40’

Receiving Shipping Testing

Department Department Department
(4) (5) (6)

Area 4 Area 5 Area 6

Figure
60’
08-01-2023 18
 Interdepartmental Flow Graph
100

50 30
1 2 3

10
100

4 5 6
50
Figure
08-01-2023 19
 n n

Cost = ∑ ∑ Xij Cij

i=1 j=1

Cost = $50 + $200 + $40

(1 and 2) (1 and 3) (1 and 6)
+ $30 + $50 + $10
(2 and 3) (2 and 4) (2 and 5)
+ $40 + $100 + $50
(3 and 4) (3 and 6) (4 and 5)

= $570

08-01-2023 20
 Revised Interdepartmental Flow Graph

30

50 100
2 1 3

50 100

50
4 5 6

Figure
08-01-2023 21
 n n

Cost = ∑ ∑ Xij Cij

i=1 j=1

Cost = $50 + $100 + $20

(1 and 2) (1 and 3) (1 and 6)
+ $60 + $50 + $10
(2 and 3) (2 and 4) (2 and 5)
+ $40 + $100 + $50
(3 and 4) (3 and 6) (4 and 5)

= $480

08-01-2023 22
 Area 1 Area 2 Area 3

Painting Assembly Machine Shop

Department Department Department
(2) (1) (3)

Improved 40’
Layout
Receiving Shipping Testing
Department Department Department
(4) (5) (6)

Area 4 Area 5 Area 6

60’
Figure
08-01-2023 23
Qualitative Layout Design-Relationship
Chart

Figure 9.1

08-01-2023 24
 A Absolutely necessary
Example 2 E
I
Especially important
Important
O Okay
U Unimportant
X Undesirable
Production
O
Offices A
U I
Stockroom O E
A X A
Shipping and U U
receiving
U O
Locker room O
O
Toolroom

08-01-2023 25
Solution
(a) Relationship diagram of original layout

Offices Locker Shipping

room and
receiving

Key: A
E
I
Stockroom Toolroom Production
O
U
X

08-01-2023 26
 (b) Relationship diagram of revised layout

Stockroom

Offices Shipping
and
receiving

Locker Key: A
Toolroom Production
room E
I
O
U
X

08-01-2023 27
Computer Software For Process Layout

 Graphical approach only works for

small problems
 Computer programs are available to
solve bigger problems
 CRAFT
 ALDEP
 CORELAP
 Factory Flow
08-01-2023 28
Example 3 (Using CRAFT)
PATTERN PATTERN
1 2 3 4 5 6 1 2 3 4 5 6

1 A A A A B B 1 D D D D B B

2 A A A A B B 2 D D D D B B

3 D D D D D D 3 D D D E E E

4 C C D D D D 4 C C D E E F

5 F F F F F D 5 A A A A A F

6 E E E E E D 6 A A A F F F

TOTAL COST 20,100 TOTAL COST 14,390

EST. COST REDUCTION .00 EST. COST REDUCTION 70.
ITERATION 0 ITERATION 3

08-01-2023
(a) (b) 29
08-01-2023 30
 Session 2

Product-Oriented Layout

Organized around products or families of similar

high-volume, low-variety products

08-01-2023 31
 Fabrication line
 Builds components on a series of machines
 Machine-paced
 Require mechanical or engg. changes to balance

 Assembly line
 Puts fabricated parts together at a series of
workstations
 Paced by work tasks
 Balanced by moving tasks

Both types of lines must be balanced so that the time to

perform the work at each station is the same

08-01-2023 32
 Advantages
1. Low variable cost per unit
2. Low material handling costs
3. Reduced work-in-process inventories
4. Easier training and supervision
5. Rapid throughput

Disadvantages
1. High volume is required
2. Work stoppage at any point ties up the whole operation
3. Lack of flexibility in product or production rates

08-01-2023 33
 Figure

08-01-2023 34
 Example 4
Suppose you load work into the three work stations below
such that each will take the corresponding number of
minutes as shown. What is the cycle time of this line?

Station 1 Station 2 Station 3

Minutes
per Unit 6 7 3

Solution
The cycle time of the line is always determined by the work station taking the
longest time. In this problem, the cycle time of the line is 7 minutes. There
is also going to be idle time at the other two work stations.

08-01-2023 35
Example 5
You’ve just been assigned the job a setting up an
electric fan assembly line with the following tasks:

Task Time (Mins) Description Predecessors

A 2 Assemble frame None
B 1 Mount switch A
C 3.25 Assemble motor housing None
D 1.2 Mount motor housing in frame A, C
E 0.5 Attach blade D
F 1 Assemble and attach safety grill E
G 1 Attach cord B
H 1.4 Test F, G

08-01-2023 36
 Solution

Task Task
Predecessors
A None Predecessors
E D
B A F E
C None G B
D A, C H E, G

A B G
H

C D E F
08-01-2023 37
Question 1: Which process step defines the maximum rate of
production?

2 1 1
A B G 1.4
H

C D E F
3.25 1.2 .5 1

Answer: Task C is the cycle time of the line and therefore, the
maximum rate of production.

08-01-2023 38
 Question 2: Suppose we want to assemble 100
fans per day. What would our cycle time have
to be?

Answer:

Production time per day

Required Cycle Time, C =
Required output per period
420 mins / day
C= = 4.2 mins / unit
100 units / day

08-01-2023 40
Question 3: What is the theoretical minimum
number of workstations for this problem?

Answer:

Theoretical Min. Number of Workstations, N t

Sum of task times (T)

Nt =
Cycle time (C)

11.35 mins / unit

Nt = = 2.702, or 3
4.2 mins / unit

08-01-2023 41
 Rules To Follow for Loading Workstations

• Assign tasks to station 1, then 2, etc. in sequence. Keep assigning to a

workstation ensuring that precedence is maintained and total work is
less than or equal to the cycle time. Use the following rules to select
tasks for assignment.

• Primary: Assign tasks in order of the largest number of following tasks

• Secondary (tie-breaking): Assign tasks in order of the longest operating

time

08-01-2023 42
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

08-01-2023 43
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2)

08-01-2023 44
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2)
B (2.2-1=1.2)

08-01-2023 45
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2)
B (2.2-1=1.2)
G (1.2-1= .2)

Idle= .2
08-01-2023 Module: X Facility Planning 46
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2) C (4.2-3.25)=.95
B (2.2-1=1.2)
G (1.2-1= .2)

Idle= .2
08-01-2023 Module: X Facility Planning 47
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2) C (4.2-3.25)=.95
B (2.2-1=1.2)
G (1.2-1= .2)

Idle= .2 Idle = .95

08-01-2023 Module: X Facility Planning 48
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2) C (4.2-3.25)=.95 D (4.2-1.2)=3

B (2.2-1=1.2)
G (1.2-1= .2)

Idle= .2 Idle = .95

08-01-2023 Module: X Facility Planning 49
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2) C (4.2-3.25)=.95 D (4.2-1.2)=3

B (2.2-1=1.2) E (3-.5)=2.5
G (1.2-1= .2)

Idle= .2 Idle = .95

08-01-2023 Module: X Facility Planning 50
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2) C (4.2-3.25)=.95 D (4.2-1.2)=3

B (2.2-1=1.2) E (3-.5)=2.5
G (1.2-1= .2) F (2.5-1)=1.5

Idle= .2 Idle = .95

08-01-2023 Module: X Facility Planning 51
 Task Followers Time (Mins)
A 6 2
2 1 1
1.4 C 4 3.25
A B G
H D 3 1.2
B 2 1
E 2 0.5
C D E F
F 1 1
3.25 1.2 .5 1
G 1 1
H 0 1.4

Station 1 Station 2 Station 3

A (4.2-2=2.2) C (4.2-3.25)=.95 D (4.2-1.2)=3

B (2.2-1=1.2) E (3-.5)=2.5
G (1.2-1= .2) F (2.5-1)=1.5
H (1.5-1.4)=.1
Idle= .2 Idle = .95 Idle = .1
08-01-2023 Module: X Facility Planning 52
 Question 4: Which station is the bottleneck? What
is the effective cycle time?

Answer: Work station is 3 and effective cycle time is

4.1

08-01-2023 53
 Question 5: What is the efficiency of assembly line?

Answer:

Sum of task times (T)

Efficiency =
Actual number of workstations (Na) x Cycle time (C)

11.35 mins / unit

Efficiency = =.901
(3)(4.2mins / unit)

08-01-2023 54
08-01-2023 55
 Certificate Course on
Operations Management and
Analytics
(August 2022-January 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
 Module X

Session 3-4
• What is Project Management ?

• Critical Path Method

• Crashing the Project

• Project Evaluation Review Technique

57
An Overview of Project Management

Social Cost Benefit

Analysis SCM in Project
Management
4
9

Quality Strategic
Bidding issues
3 14
13

Project Financing Environmental

Selection 12 Issues
2 15

58
 Difference between Project & Production?

Done once Done repetitively

• Automobile factory • Produce Automobiles
• Build a house • Operate house
• Construct hospital • Treat patients
• Conceive new product • Manufacturing the product
• Developing a Business plan • Executing the Business plan

59
WHAT IS A PROJECT?
PURPOSE
• An undertaking or venture to accomplish some
objective or goal
STRUCTURE
• A set of interrelated jobs whose accomplishment
leads to the completion of the project
COMPONENTS
• Jobs or activities consume time and resources and
are governed by precedence relations

60
PROJECTS AS
AGENTS OF CHANGE

State B

Alternative
Projects
State A (Paths)

61
 INPUT(S) OUTPUT(S)
Transformation
-Men Useful
-Machines Economic/Social/Political Goods or
Environment Services
-Materials -dynamic
-Money -uncertain
-Information FEED BACK
-Energy
---------- Undesirable
outputs
(Effluents, Fumes etc.)

A PRODUCTION SYSTEM AS
AN INPUT-OUTPUT SYSTEM

62
 A Project as a Production System
Mass production

Batch
Production

Job
Production

Q
Projects
Quantity
to be
Made
1

P (No. of Products or “VARIETY”)

63
VARIETY OF PROJECTS
• Projects at personal level
• Projects in local neighbourhood
• Organisational projects
• National projects
• Global projects

64
PERSONAL PROJECTS
• Preparing for an examination
• Writing a book
• Getting dressed
• Wedding in the house
• A birthday function
• A family vacation

65
PROJECTS IN LOCAL
NEIGHBOURHOOD

• A school function
• Cleanliness drive
• Construction of clubs
• Tree plantation exercise
• Establishment of a park
• Welcoming a dignitary to the colony

66
 ORGANISATIONAL PROJECTS
• Construction of building, highway
• Planning & launching a new product
• A turnaround in a refinery
• A training for managers in the organization
• Conducting a marketing survey
• Completing a financial audit
• Disposal of dead stock

67
NATIONAL PROJECTS

• Launching a new satellite

• Literacy campaign

• Poverty removal drive

• Organizing general elections

• Preparation of annual budget

68
GLOBAL PROJECTS

• Organising peace missions (UN)

• Space exploration

• Conducting World Trade

• Environment protection

69
 Introduction
• Many organizations today have a great interest in
project management.
• Computer hardware, software, networks, and the use
of interdisciplinary and global work teams have
radically changed the work environment.
• The U.S. spends $2.3 trillion on projects every year, or
one-quarter its gross domestic product, and the world
as a whole spends nearly $10 trillion of its $40.7 gross
product on projects of all kinds.*

*PMI, The PMI Project Management Fact Book, Second Edition, 2004. 70
 Advantages of Using Formal
Project Management
• Better control of financial, physical, and human resources.
• Improved customer relations.
• Shorter development times.
• Lower costs.
• Higher quality and increased reliability.
• Higher profit margins.
• Improved productivity.
• Better internal coordination.
• Higher worker morale (less stress).

71
What Is a Project?
• A project is “a temporary endeavor undertaken to
create a unique product, service, or result.”*
• Operations is work done to sustain the business.
• A project ends when its objectives have been
reached, or the project has been terminated.
• Projects can be large or small and take a short or
long time to complete.

*PMI, A Guide to the Project Management Body of Knowledge

72
(PMBOK® Guide) (2004)
Examples of few Projects
• Construction of National Highways.

• A small software development team adds a new

feature to an internal software application.

• Implementation of ERP in a company.

• A college campus upgrades its technology

infrastructure to provide wireless Internet access.

73
• Set-up of new oil refinery or set up of new
manufacturing plant.

• Construction of temporary bridge during Maha-mela

such as Kumbh.

• A television network develops a system to allow

viewers to vote for contestants and provide other
feedback on programs.

• A government group develops a system to track child

immunizations.

74
 Project Attributes
• A project:
• Has a unique purpose.
• Is temporary.
• Is developed using progressive collaboration.
• Requires resources, often from various areas.
• Should have a primary customer or sponsor.
• The project sponsor usually provides the direction and funding for
the project.
• Involves risk/uncertainty.

75
 The Triple Constraint
• Every project is constrained in different ways by its:
• Scope goals: What work will be done?

• Time goals: How long should it take to complete?

• Cost goals: What should it cost?

• It is the project manager’s duty to balance these

three often-competing goals.

76
The Triple Constraint of Project
Management

Successful project
management means
meeting all three
goals (scope, time,
and cost) – and
satisfying the
project’s sponsor!

77
What is Project Management?
• Project management is “the application of
knowledge, skills, tools and techniques to project
activities to meet project requirements.”*

*PMI, A Guide to the Project Management Body of Knowledge

(PMBOK® Guide) (2004).
78
FEATURES OF PROJECTS

• Well defined collection of jobs

• Generally non-repetitive, one time effort

• Jobs interrelated through precedence

• Jobs otherwise independent

79
• Jobs consume time and resources
• Coordination needed between individuals,
groups & organisations
• Constant pressure of conformance to time/cost
/performance goals

80
LIFE CYCLE OF A PROJECT
1. Project Selection
2. Project Planning
Scope of work & network development
Basic Scheduling
Time Cost tradeoffs
Resource Considerations in projects
3. Project Implementation (Production…)
4. Project Completion and Audit (Quality control etc…)

81
1. PROJECT SELECTION

• Project Identification

• Project Appraisal

• Project Selection

82
PROJECT IDENTIFICATION
• Receptive to new ideas

• Vision of future growth

• Long term objectives
• SWOT analysis
• Preliminary project analysis

83
PROJECT IDENTIFICATION
Objectives
Internal
SWOT Brainstorming
External
Project possibilities
Criteria Screening
Candidate project proposals
84
PROJECT APPRAISAL
• Market Appraisal
• Technical Appraisal
• Financial Appraisal
• Economic Appraisal
• Ecological Appraisal
• A Feasibility Report considers all these issues prior
to project adoption.

85
MARKET APPRAISAL
• Aggregate future demand
• Market share
• Current and future competition
• Location and accessibility of consumers
• Technological scenario /Obsolescence
• Possible pricing options

86
TECHNICAL APPRAISAL
• Engineering aspects
• Locations
• Size
• Production process

87
FINANCIAL APPRAISAL
• Cash flows over time
• Profitability
• Break even point
• Net present value
• Internal rate of return
• Payback period
• Risk

88
ECONOMIC APPRAISAL
• Benefits and costs (in shadow prices)
• Distribution of income in society
• Level of savings & investment in society
• Self sufficiency, employment and social order

89
ECOLOGICAL APPRAISAL
• Environmental damage
Air
Water
Noise
Other
• Restoration measures and cost

90
PROJECT SELECTION
CRITERIA
• Investment • Similarity to existing
• Rate of return business
• Risk • Expected life
• Likely profit • Flexibility
• Payback • Environmental impact
• Competition

91
PROJECT SELECTION
CRITERIA (Continued)
• Similarity to existing business
• Expected life
• Flexibility
• Environmental impact
• Competition

92
MULTI-CRITERIA
EVALUATION
CRITERIA
C1 C2 … CN
P1 S1
PROJECTS P2 S2

PM SM
WEIGHTS W1 W2 ... WN

93
2. PROJECT PLANNING
• Forming a project team with a leader
• Defining scope and terms of reference
• Work breakdown structure
• Basic Scheduling
• Time cost tradeoffs
• Resource Considerations

94
BASIC SCHEDULING
• Project representation as a network
• Estimation of activity durations
• Forward and backward pass
• Determination of activity floats
• Critical path for selective control and minimum
project duration

95
Basic Rules to Follow in Developing
Project Networks
• Networks typically flow from left to right.
• An activity cannot begin until all of its activities are complete.
• Arrows indicate precedence and flow and can cross over each
other.
• Identify each activity with a unique number; this number
must be greater than its predecessors.
• Looping is not allowed.
• Conditional statements are not allowed.
• Use common start and stop nodes.

96
Activity-on-Node Fundamentals

97
Activity-on-Node Fundamentals (cont’d)

98
Example 1:

99
Koll Business Center—Partial Network

100
Koll Business Center—Complete Network

101
 Network Computation Process
• Forward Pass—Earliest Times
• How soon can the activity start? (early start—ES)
• How soon can the activity finish? (early finish—EF)
• How soon can the project finish? (expected time—ET)
• Backward Pass—Latest Times
• How late can the activity start? (late start—LS)
• How late can the activity finish? (late finish—LF)
• Which activities represent the critical path?
• How long can it be delayed? (slack or float—SL)

102
103
Activity-on-Node Network

104
105
 Forward Pass Computation

• Add activity times along each path in the network

(ES + Duration = EF).
• Carry the early finish (EF) to the next activity where
it becomes its early start (ES) unless…
• The next succeeding activity is a merge activity, in
which case the largest EF of all preceding activities is
selected.

106
107
Backward Pass Computation
• Subtract activity times along each path in the
network (LF - Duration = LS).
• Carry the late start (LS) to the next activity where it
becomes its late finish (LF) unless
• The next succeeding activity is a burst activity, in
which case the smallest LF of all preceding activities
is selected.

108
Determining Slack (or Float)
• Slack (or Float)
• The amount of time an activity can be delayed after the
start of a longer parallel activity or activities.
• Total slack
• The amount of time an activity can be delayed without
delaying the entire project.
• The critical path is the network path(s) that has
(have) the least slack in common.

109
110
Hammock Activities
• Hammock Activity
• An activity that spans over a segment of a project.
• Duration of hammock activities is determined after the
network plan is drawn.
• Hammock activities are used to aggregate sections of
the project to facilitate getting the right amount of detail
for specific sections of a project.

111
Activity-on-Arrow Network Building Blocks

112
113
114
Example 2:

115
116
117
118
119
120
121
122
123
PERT—PROGRAM EVALUATION REVIEW TECHNIQUE

• Assumes each activity duration has a range that

statistically follows a beta distribution.
• PERT uses three time estimates for each activity:
optimistic, pessimistic, and a weighted average to
represent activity durations.
• Knowing the weighted average and variances for each
activity allows the project planner to compute the
probability of meeting different project durations.

124
125
Activity Time Calculations
The weighted average activity time is computed by
the following formula:

(1)

126
Activity Time Calculations (cont’d)
The variability in the activity time estimates is
approximated by the following equations:
The standard deviation for the activity:

(2)

The standard deviation for the project:

(3)

Note the standard deviation of the activity is squared in this equation;

this is also called variance. This sum includes only activities on the
critical path(s) or path being reviewed.

127
Example 3:

128
Probability of Completing the Project
The equation below is used to compute the “Z” value
found in statistical tables (Z = number of standard
deviations from the mean), which, in turn, tells the
probability of completing the project in the time specified.

(4)

129
Hypothetical Network

130
131
Possible Project Duration

132
Z Values

TABLE 6
133
 Explanation of Project Costs
• Project Indirect Costs
• Costs that cannot be associated with any particular work
package or project activity.
• Supervision, administration, consultants, and interest
• Costs that vary (increase) with time.
• Reducing project time directly reduces indirect costs.
• Direct Costs
• Normal costs that can be assigned directly to a specific work
package or project activity.
• Labor, materials, equipment, and subcontractors
• Crashing activities increases direct costs.

134
Reducing Project Duration to
Reduce Project Cost
Identifying direct costs to reduce project time

Gather information about direct and indirect costs

of specific project durations.

Search critical activities for lowest direct-cost

activities to shorten project duration.

Compute total costs for specific durations and

compare to benefits of reducing project time.

135
Project Cost—Duration Graph

136
Constructing a Project Cost—Duration
Graph
• Find total direct costs for selected project durations.
• Find total indirect costs for selected project
durations.
• Sum direct and indirect costs for these selected
project durations.
• Compare additional cost alternatives for benefits.

137
Constructing a Project Cost—Duration
Graph
• Determining Activities to Shorten
• Shorten the activities with the smallest increase in cost per unit
of time.
• Assumptions:
• The cost relationship is linear.
• Normal time assumes low-cost, efficient methods to complete the
activity.
• Crash time represents a limit—the greatest time reduction possible
under realistic conditions.
• Slope represents a constant cost per unit of time.
• All accelerations must occur within the normal and crash times.

138
 Example 4:
Activity Graph

FIGURE 2
139
Cost—Duration Trade-off Example

140
Cost—Duration Trade-off Example (cont’d)

141
Cost—Duration Trade-off Example (cont’d)

142
Cost—Duration Trade-off Example (cont’d)

143
Cost—Duration Trade-off Example (cont’d)

144
Summary Costs by Duration

145
Project Cost—Duration Graph

146
Practical Considerations
• Using the Project Cost—Duration Graph
• Crash Times
• Linearity Assumption
• Choice of Activities to Crash Revisited
• Time Reduction Decisions and Sensitivity

147
148
Operations Management and
Analytics

August 2022-January 2023

Prof. Surya Prakash Singh

PhD (IIT Kanpur), PDF (NUS Singapore, MIT USA)
Dhananjaya Chair Professor
Department of Management Studies
Indian Institute of Technology Delhi 1
E-Mail: [email protected]
 Session 1 and 2

(Introduction to Decision Science

Using OR: Linear Programming
Techniques)

2
 Topics

• Introduction to OR (Decision Science)

• Terminology used in OR (Decision Science)
• Linear programming
• Model Formulation
• A Maximization Model Example
• A Minimization Model Example
• Irregular Types of Linear Programming Models
• Characteristics of Linear Programming Problems

3
 Introduction to OR

Decision Science Models are classified as:

• Unconstrained Problem

• Constrained Problem

4
If a boat is moving in a upstream side
of the river at 28 km/hr and
downstream side at 40 km/hr. What
would be the speed of boat and river?

So, such problems have variables

and data but do not have any
constraints and/or any objective.

5
On the other hand, all business problems
comes with constraint. Such problems
have variables, data, constraints and
objective(s). Objective can be single or
multiple.
We will see how Operations Research
(OR) can be applied to solve such
business problems.

OR is also termed as Prescriptive

Analytics.
6
 • OR involves “research on operations”. Thus OR
applied to problems that concern how to conduct
and co-ordinate the operations with in the
organization.

• OR has been applied widely in manufacturing,

transportation, construction, telecommunication,
financial planning, health care etc.

• Using OR techniques, decision science models are

developed to help decision makers to take the
most appropriate (i.e. best/ optimal) decision. This
is what known as Prescriptive Analytics.
7
 Decision Science Model Formulation Using LP
Example 1:
Product mix problem: Beaver Creek Pottery Company

Product resource requirements and unit profit:

Resource Requirements
Labor Clay Profit
Product
(hr/unit) (lb/unit) ($/unit)
Bowl 1 4 40
Mug 2 3 50

8
How many bowls and mugs
should be produced to
maximize profits given
labor and materials
constraints?
9
 Decision Science Model Formulation
A Maximization Example

Resource 40 hrs of labor per week

Availability: 120 lbs of clay per week
Decision x1 = number of bowls to produce per week
Variables: x2 = number of mugs to produce per week
Objective Maximize Z = 40*x1 + 50*x2
Function: Where Z = profit per week
Resource 1*x1 + 2*x2  40 hours of labor
Constraints: 4*x1 + 3*x2  120 pounds of clay
Non-Negativity x1  0; x2  0
Constraints:
10
 Complete Decision Science Model Formulation

Complete Linear Programming Model:

Maximize Z = 40*x1 + 50*x2

subject to: 1*x1 + 2*x2  40 (Labour Constraint)

4*x1 + 3*x2  120(Raw material Constraint)
x 1, x 2  0 (Non-negativity constraint)

11
 Feasible Solutions

• A feasible solution does not violate any of the constraints:

Example x1 = 5 bowls
x2 = 10 mugs
Z = 40*x1 + 50*x2 = 700
Labor constraint check:
1(5) + 2(10) = 25 < 40 hours, within constraint
Clay constraint check:
4(5) + 3(10) = 70 < 120 pounds, within constraint

12
 Infeasible Solutions

• An infeasible solution violates at least one of the

constraints:
Example x1 = 10 bowls
x2 = 20 mugs
Z = 1400
Labor constraint check:
1(10) + 2(20) = 50 > 40 hours, violates constraint

13
 Basic Terminologies

Feasible solution:
• Any solution LPP/ NLP which donot violate constraints are
called feasible solution.
• Feasible solution may be optimal (best) solution.
• Any LPP/NLP can have more than one feasible solution.
• Optimal solution to LPP/NLP must be a feasible solution.

Infeasible Solution:
• Any solution violate at least one constraint is called infeasible
solution.
• Any LPP/NLP have infinite number of infeasible solution.
• Infeasible solution lies outside the bounded region.
14
Optimal solution:
• It is a feasible solution that has the most favorable value of the
objective function.
• The most favorable mean the largest possible objective value
if the objective is to maximize and smallest value if the
objective is to minimize.
• A LPP/ NLP can have more than one Optimal solution.
Corner Point Feasible Solution:
• CPF is a solution that lies at the corner of the feasible region
• Every LPP with feasible solution and bounded feasible region
must possess CPF solutions and at least one optimal solution.
• The best CPF solution must be an optimal solution.
• If LPP has exactly one one optimal solution, it must e a CPF
solution.
15
Bounded feasible Region:

• A bounded feasible region may be enclosed in a circle.

• A bounded feasible region will have both a maximum value
and a minimum value.

Unbounded Region:

• An unbounded feasible region can not be enclosed in a circle.

• If the coefficients on the objective function are all positive, then
an unbounded feasible region will have a minimum but no
maximum.

16
 Linear Programming: An Overview

• Linear programming is an analytical technique in which

linear algebraic relationships represent a firm’s decisions,
given a business objective, and resource constraints.
• Objectives of business decisions frequently involve
maximizing profit or minimizing costs.
• Steps in application of LPP:
– Identify problem as solvable by linear programming.
– Formulate a mathematical model of the unstructured
problem.
– Solve the model.
– Implementation

17
 Decision Model Formulation

Decision variables - mathematical symbols

representing levels of activity of a firm.
Objective function - a linear mathematical relationship
describing an objective of the firm, in terms of decision
variables - this function is to be maximized or minimized.
Constraints – requirements or restrictions placed on the
firm by the operating environment, stated in linear
relationships of the decision variables.
Parameters - numerical coefficients and constants used
in the objective function and constraints.

18
 Summary of Model Formulation Steps

Step 1 : Clearly define the decision variables

Step 2 : Construct the objective function

Step 3 : Formulate the constraints

19
Model…….. …………………….

Example 2: Product mix problem

20
For example product P consists of two subassemblies. To manufacture the
first subassembly, one unit of RM1 passes through machine A for 15
minutes. The output of machine A is moved to machine C where it is
processed for 10 minutes. The second subassembly starts with RM2
processed in machine B for 15 minutes. The output is taken to machine C
for 5 minutes of processing. The two subassemblies are joined with a
purchased part in machine D. The result is a finished unit of P. Product Q is
manufactured by a similar process as indicated in the figure.

The rectangle at the upper left indicates that one machine of each type is
available. Each machine operates for 2400 minutes per week. OE stands
for operating expenses. For this case the operating expenses, not including
the raw material cost is $6000. This amount is expended regardless of
amounts of P and Q produced.

Our problems include the following: Find the product mix that maximizes
profit.

21
Decision Model Formulation

There are many problems that might be posed regarding the PQ

situation, but we choose the problem of allocating the times available
on the machines to the manufacture of the two products. The decisions
involve the amounts of the two products.
P= amount of product p is produced
Q= amount of product q is produced
Maximize Z = 45*P + 60*Q - 6000
Subject to:
15*P+10*Q<=2400 (Constraint for M/C A)
15*P+30*Q<=2400 (Constraint for M/C B)
15*P+5*Q<=2400 (Constraint for M/C C)
10*P+5*Q<=2400 (Constraint for M/C D)
P<=100 (Marketing Constraint)
Q<=50 (Marketing Constraint )

22
Example 3:

Resource Allocation problem:

In this example, we consider a manufacturing facility that

produces five different products using four machines. The
scarce resources are the times available on the machines
and the alternative activities are the individual production
volumes. The machine requirements in hours per unit are
given. The unit profits are also shown in the table.

23
The facility has four machines of type 1, five of type 2, three
of type 3 and seven of type 4. Each machine operates 40
hours per week. The problem is to determine the optimum
weekly production quantities for the products. The goal is to
maximize total profit.

Machine data and processing requirements (hrs./unit)

M/C Quantity Product 1 Product 2 Product 3 Product 4 Product 5
M1 4 1.2 1.3 0.7 0.0 0.5
M2 5 0.7 2.2 1.6 0.5 1.0
M3 3 0.9 0.7 1.3 1.0 0.8
M4 7 1.4 2.8 0.5 1.2 0.6
Profit —— 18 25 10 12 15 24
Decision Model Formulation

Pj : quantity of product j produced, j = 1,...,5

Objective Function:
Maximize Z = 18P1+ 25P2 + 10P3 + 12P4 + 15P5

Subject to constraint:
1.2P1 + 1.3P2 + 0.7P3 + 0.0P4 + 0.5P5 < 160 (Constraint for M1)
0.7P1 + 2.2P2 + 1.6P3 + 0.5P4 + 1.0P5 < 200 (Constraint for M2)
0.9P1 + 0.7P2 + 1.3P3 + 1.0P4 + 0.8P5 < 120 (Constraint for M3)
1.4P1 + 2.8P2 + 0.5P3 + 1.2P4 + 0.6P5 < 280 (Constraint for M4)

Non-negativity Constraint:
Pj > 0 for j = 1,...,5

25
Irregular Types of Linear Programming Problems

For some linear programming models, the general rules

do not apply.

Special types of problems include those with:

– Multiple optimal solutions
– Infeasible solutions
– Unbounded solutions

26
 Bounded Feasible Region Area

Max. Z = 40x1 + 50x2

s.t. 1x1 + 2x2  40
4x2 + 3x2  120
x1, x2  0

Figure: Feasible Solution Area 27

 An Infeasible Problem

Every possible solution

violates at least one
constraint:
Maximize Z = 5x1 + 3x2
subject to: 4x1 + 2x2  8
x1  4
x2  6
x1, x2  0

Figure: Graph of an Infeasible Problem

28
 An Unbounded Problem

Value of objective function

increases indefinitely:
Maximize Z = 4x1 + 2x2
subject to: x1  4
x2  2
x1, x2  0

Figure: Graph of an Unbounded Problem 29

Characteristics of Linear Programming Problems

A linear programming problem requires a decision - a

choice amongst alternative courses of action.
The decision is represented in the model by decision
variables.
The problem encompasses a goal, expressed as an
objective function, that the decision maker wants to
achieve.
Constraints exist that limit the extent of achievement of
the objective.
The objective and constraints must be definable by
linear mathematical functional relationships.

30
Properties of Linear Programming Models

Proportionality - The rate of change (slope) of the

objective function and constraint equations is constant.
Additivity - Terms in the objective function and
constraint equations must be additive.
Divisibility -Decision variables can take on any
fractional value and are therefore continuous as opposed
to integer in nature.
Certainty - Values of all the model parameters are
assumed to be known with certainty (non-probabilistic).

31
 Graphical Solution of LP Models

Graphical solution is limited to linear programming

models containing only two decision variables (can be
used with three variables but only with great difficulty).

Graphical methods provide visualization of how a

solution for a linear programming problem is obtained.

32
 Coordinate Axes
Graphical Solution of Maximization Model (1 of 12)

Maximize Z = $40x1 + $50x2

subject to: 1x1 + 2x2  40
4x2 + 3x2  120
x1, x2  0

Coordinates for Graphical Analysis

33
Solving decision model
using Graph / Excel/
LINGO

34
Solving Decision Models Using Graph

Graphical solution is limited to linear programming

models containing only two decision variables (can be
used with three variables but only with great difficulty).

Graphical methods provide visualization of how a

solution for a linear programming problem is obtained.

35
 Decision Model 1: A Maximization Case

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40

4x1 + 3x2  120
x1, x2  0

Decision Model of
Beaver Creek
Pottery Company
Figure: Coordinates for Graphical Analysis
36
Labor Constraint

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Graph of Labor Constraint 37

Labor Constraint Area

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Labor Constraint Area 38

 Clay Constraint Area

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Clay Constraint Area 39

40
Both Constraints

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Graph of Both Model Constraints 41

Feasible Solution
Area

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Feasible Solution Area

42
 Objective Function
Solution = Rs. 800

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Objection Function Line for Z = 800 43

Maximize Z = 40x1 + 50x2
subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Alternative Objective Function Lines

44
 Optimal Solution

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Identification of Optimal Solution 45

Maximize Z = 40x1 + 50x2
subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Optimal Solution Coordinates 46

 Corner Point
Solutions

Maximize Z = 40x1 + 50x2

subject to:1x1 + 2x2  40
4x2 + 3x2  120
x1, x2  0

Figure: Solutions at All Corner Points 47

 Optimal Solution
for New Objective
Function

Maximize Z = 70x1 + 20x2

subject to:1x1 + 2x2  40
4x2 + 3x2  120
x1, x2  0

Figure: Optimal Solution with Z = 70x1 + 20x2 48

 Decision Model 2: A Minimization Case

• Two brands of fertilizer available - Super-Gro, Crop-Quick.

• Field requires at least 16 pounds of nitrogen and 24 pounds
of phosphate.
• Super-Gro costs $6 per bag, Crop-Quick $3 per bag.
• Problem: How much of each brand to purchase to minimize
total cost of fertilizer given following data ?
Chemical Contribution

Nitrogen Phosphate
Brand
(lb/bag) (lb/bag)
Super-gro 2 4

Crop-quick 4 3 49
Figure : Fertilizing Farmer’s field 50
LP Model Formulation
A Minimization Example
Decision Variables:
x1 = bags of Super-Gro
x2 = bags of Crop-Quick
The Objective Function:
Minimize Z = 6x1 + 3x2
Where: 6x1 = cost of bags of Super-Gro
3x2 = cost of bags of Crop-Quick
Model Constraints:
2x1 + 4x2  16 lb (nitrogen constraint)
4x1 + 3x2  24 lb (phosphate constraint)
x1, x2  0 (non-negativity constraint)

51
 LP Model Formulation and Constraint Graph
A Minimization Example

Minimize Z = 6x1 + 3x2

subject to:2x1 + 4x2  16
4x1 + 3x2  24
x1, x2  0

Figure: Graph of Both Model Constraints 52

 Feasible Solution Area
A Minimization Example

Minimize Z = 6x1 + 3x2

subject to:2x1 + 4x2  16
4x1 + 3x2  24
x1, x2  0

Figure: Feasible Solution Area 53

 Optimal Solution Point
A Minimization Example

Minimize Z = 6x1 + 3x2

subject to:2x1 + 4x2  16
4x2 + 3x2  24
x1, x2  0

Figure: Optimum Solution Point 54

 Case of Multiple Optimal Solutions
Beaver Creek Pottery Example

Objective function is
parallel to to a constraint
line.
Maximize Z=40x1 + 30x2
subject to: 1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0
Where:
x1 = number of bowls
x2 = number of mugs

Figure: Example with Multiple Optimal Solutions 55

 Practice question # 1

Solve the following model by graphical method:

Problem:
Maximize Z = 4x1 + 5x2
subject to: x1 + 2x2  10
6x1 + 6x2  36
x1  4
x1, x2  0

56
Step 1:
Plot all constraints

Figure: Constraint Equations 57

Maximize Z = 4x1 + 5x2
subject to: x1 + 2x2  10
6x1 + 6x2  36
x1  4
x1, x2  0
Step 2: Determine the
feasible solution space

Figure: Feasible Solution Space and Extreme Points 58

Maximize Z = 4x1 + 5x2
subject to: x1 + 2x2  10
6x1 + 6x2  36
x1  4
x1, x2  0
Step 3 and 4: Determine
the solution points and
optimal solution

Figure: Optimal Solution Point

59
Solving Decision Models Using
EXCEL SOLVER

60
99
100
Click “Go” button to
manage Excel Add-ins

101
102
Now Solver add-in will appea
under “Data” tab

103
66
 Session 2

Sensitivity and Dual Analysis of

Linear Decision Models:
Managerial Analysis
(Post-Optimality Analysis)

67
 Introduction
For any decision problem represented as an
Optimization problem, sensitivity analysis gives several
vital information and insights about the nature of the
decision scenario

Sensitivity analysis presents insights mainly about the

input side of the problem specifically like,
– How the firm should price its inputs in case it is leasing its
assets?
– Where to invest to attain maximum profit/ revenue?
– Where to disinvest?
– What are the critical resources that can impact profitability?
– And many more….
 Beaver Creek Pottery Company
revisited
How many bowls and mugs should be
produced to maximize profits given labor
and materials constraint?
Product resource requirements and unit
profit:
Product Resource Requirements
Labor (hr/unit) Clay (lb/unit) Profit ($/unit)
Bowl 1 4 40
Mug 2 3 50

Labor Availability : 40hours

Clay Availability : 120lb
Maximize Z = 40x1 + 50x2
subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0

Figure: Solutions at All Feasible Corner Points 70

 Sensitivity Analysis in Linear Program

• Sensitivity analysis determines the effect on the optimal

solution of changes in parameter values of the
objective function and constraint equations.

• Changes may be reactions to anticipated uncertainties in

the parameters or to new or changed information
concerning the model.
 Beaver Creek Pottery Example

Maximize Z  $40 x1  $50 x2

subject to : x1  2 x2  40
4 x1  3 x2  120
x1, x2  0

Figure: Optimal solution point

 Change x1 Objective Function Coefficient

Maximize Z  $100x1  $50 x2

subject to : x1  2 x2  40
4 x1  3 x2  120
x1, x2  0

Figure Changing the objective

function x1 coefficient
 Change x2 Objective Function Coefficient

Maximize Z  $40 x1  $100x 2

subject to : x1  2 x2  40
4 x1  3 x2  120
x1, x2  0

Figure Changing the objective

function x2 coefficient
Objective Function Coefficient: Sensitivity
Range
• The sensitivity range for an objective function coefficient is
the range of values over which the current optimal solution
point will remain optimal.
• The sensitivity range for the xi coefficient is designated as ci.
objective function Z  $40 x1  $50 x2
The slope of the objective function is 4 given by:
5
50 x2  Z  40 x1
Z 4 x1
x2  
50 5
If the slope of the objective function changes to 4 , the line is
3
parallel to the constraint line (next slide).
 Objective Function Coefficient:
Sensitivity Range for c1 and c2
Figure Determining the sensitivity range for c1
Analysing LP Using
EXCEL SOLVER
Maximize Z  $40 x1  $50 x2
subject to : x1  2 x2  40
4 x1  3 x2  120
x1, x2  0
 Beaver Creek Pottery Company
revisited
• The corner feasible points are (0,0), (0,20), (24,8), (30,0).
• Evaluating the objective function at each point gives profit attained by choosing these
product combinations.
• The maximum amongst them is the optimal profit

Corner Point Objective Function Value

(40X1+50X2)
(0,0) 40(0) + 50(0) = 0
(0,20) 40(0) + 50(20) = 1000
(24,8) 40(24) + 50(8) = 1360
(30,0) 40(30) + 50(0) = 1200
Optimal Solution
 Beaver Creek Pottery Company
Sensitivity Analysis

Q 1. Let’s say that the labour capacity in the

problem is for a specific period say a week.
Skunk Creek Pottery is a competitor firm
and wants to lease our firm for the coming
week due to a sudden surge in demand of
clay pots. How much should we charge ?
Q 2. Badger Creek Pottery is our competitor
firm and has an immediate order of mugs,
However, there is unavailability of clay in
market due to sudden floods. The
procurement manager of Badger wishes to
purchase 10lbs of clay from us due to
unavailability of clay in the market. We have
procured the clay at $3 per lb. How much we
should charge for 1lb of clay in that respect ?
Q 3. For the coming week, our management
wants to increase production and for that
they can invest 50$ more than the current
week. One pound of clay and one hour of
labor costs 5$ in the coming week. The
management is confused where to invest
(raw material (clay) or labor) or combination
of these to get maximum increase in profit.
Q 4. Recently, during stock taking store employees
have discovered 80lbs of clay in an old godown
owned by us. Procurement Manager Mr Chandraul
suggests we sell some amount of clay since he
feels that the whole amount of clay will not be used
in this week production. He also suggest that we
sell some to Badger Creek company who are
offering $5 per lb of clay. Should the management
go with this advice?
Q 5. Because of heavy competition in pottery
market, we have been forced to reduce the selling
price of bowls such that the profit of selling a bowl
is zero. Production Manager Mr Aluwalia suggests
that we stop the production of bowls entirely.
However, for Beaver Creek Pottery the bowl is a
legacy product and the management wants to
produce as little bowl as possible. What amount of
profit justifies the production of such small
amount?
So, sensitivity analysis is a tool
which can be applied to
optimally handle such
situations.

90
Q1. Skunk Creek Pottery is a competitor firm
and wants to lease our firm for the coming week
due to a sudden surge in demand of clay pots.
How much should we charge ?

Ans: The opportunity cost of leasing the pottery

plant for a week is 1360$. Therefore, the
amount charged should be greater than 1360$
Corner Point Objective Function Value
(40X1+50X2)
(0,0) 40(0) + 50(0) = 0
(0,20) 40(0) + 50(20) = 1000
(24,8) 40(24) + 50(8) = 1360
(30,0) 40(30) + 50(0) = 1200
Q 2. Badger Creek Pottery is our competitor firm and has an
immediate order of mugs, However, there is unavailability of
clay in market due to sudden flood. The procurement
manager of Badger wishes to purchase 10 lbs of clay from us
due to unavailability of clay in the market. We have procured
the clay at $3 per lb. How much we should charge for 1 lb of
clay in that respect?

Ans: The price to be charged for 1lb of clay should depend

on the contribution of 1lb of clay in the total profit.

So, the Shadow Price informs us about the contribution of a

unit quantity of the resource in the objective function.
Beaver Creek Pottery Company
Sensitivity Analysis
In the “Solver Results” before clicking OK,
select “Sensitivity” in reports.
In the “Solver Results” before clicking OK,
select “Sensitivity” in reports.

Shadow Price for labor

Shadow Price for clay

Q 3. For the coming week, our management wants to increase
production and for that they can invest 50$ more than the
current week. One pound of clay and one hour of labor costs 5$
in the coming week. The management is confused where to
invest (raw material (clay) or labor) or combination of these to
get maximum increase in profit.

Ans: 50$ can buy individually 10 hours of labor and 10 lbs of

clay. 10 hours of labor gives us a profit of 160$ more than the
previous profit (1360$) while 10 lbs of clay gives us only a 60$
increase.
Since by substituting 1 hr of labour with 1 hr of clay always
results in a loss of 10$.
For maximum increase in profit, we have to invest all the amount
(50$) in increasing our labor hour capacity by 10 units.
By doing this, the new profit would be 1360$ + 160$ = 1520$
Q 4. Recently, during stock taking store employees have
discovered 80lbs of clay in an old godown owned by us.
Procurement Manager Mr Chandraul suggests we sell
some amount of clay since he feels that the whole amount
of clay will not be used in this week production. He also
suggest that we sell some to Badger Creek company who
are offering $5 per lb of clay. Should the management go
with this advice?

Ans: With this new discovery, our stock of clay increases to

120 + 80 = 200lbs.
Since, 1 lb of clay has a shadow price of 6$, it might be not
a good idea not to sell at 5$/lb of clay.
 However, the shape of the feasible region has
changed now

• The new optimal point is (40,0) with

Obj. function value 1600.
(0,66.67)

• Reducing 10lb of clay from the stock

4x1 + 3x2  200 to 190lb doesn’t change the optimal
X2
objective function value

• Hence, selling at any value greater

(0,20) 1x1 + 2x2  40 than zero makes sense.

• Shadow price for clay constraint for

this optimization problem is zero.
X1
(40,0) (50,0)
In the original optimization problem, the validity of
the original shadow price is given with respect to
limits.

Allowable
increase/decrease says the
limits within which the
shadow price remains
valid.

In the original problem, the

shadow price remains valid
for the limits (120-60,
120+40), i.e, (60,160).
Q 5. Because of heavy competition in pottery
market, we have been forced to reduce the selling
price of bowls such that the profit of selling a bowl
is zero. Production Manager Mr Aluwalia suggests
that we should stop the production of bowls
entirely. However, for Beaver Creek Pottery the
bowl is a legacy product and the management
wants to produce as little bowl as possible. What
amount of profit justifies the production of such
small amount?
 This becomes a
new problem
Maximize Z = 0x1 + 50x2
subject to:1x1 + 2x2  40
4x1 + 3x2  120
x1, x2  0
$ 400

$0

Figure: Solutions at All Feasible Corner Points

100
Looking at the sensitivity analysis,
For a variable that takes zero value in
optimal solution, reduced cost gives
the amount of increase necessary to
produce at least one unit of the
variable.

In this case, adding 25 to the co-

efficient of X1, gets the new objective
function 25X1 + 50X2. In this case, the
objective function value is 1000 with
two optimal values (0,20) and (25,8)

Hence, the profit should be brought to

a value greater than or equal to 25 to
produce at least one unit of X1.
 Duality in Linear Program
 With every linear programming problem, there is associated
another linear programming problem which is called the dual of
the original (or the primal) problem.

The primal-Dual Relationship

Similarly, DUAL Solution
and its Sensitivity can be
analysed Using EXCEL
SOLVER
 PRIMAL Solution

DUAL Solution
107
 Module XI

Session 3

(Prescriptive Analytics Module)

 Introduction to Lingo (Solving LP using Lingo)

 QAP Design

 QAP Example

Module: XI Facility Planning- 1

Prescriptive Analytics
Module: XI Facility Planning- 2
Prescriptive Analytics
Module: XI Facility Planning- 3
Prescriptive Analytics
Module: XI Facility Planning- 4
Prescriptive Analytics
Module: XI Facility Planning- 5
Prescriptive Analytics
Module: XI Facility Planning- 6
Prescriptive Analytics
Module: XI Facility Planning- 7
Prescriptive Analytics
Module: XI Facility Planning- 8
Prescriptive Analytics
Module: XI Facility Planning- 9
Prescriptive Analytics
Module: XI Facility Planning- 10
Prescriptive Analytics
 LINGO FORMULATION USING SETS
Model:
Sets:
decision/1 2/: X, C; ! There are two decision variables;
constraint / 1..3/: B; ! There are three constraints;
matrix (constraint, decision): A; ! LHS matrix of constraint;
Endsets

Max = @ Sum (decision (j): C(j) * X (j));

@For(constraint (i): @sum (decision(j): A(i,j)*X(j)) <= B (i));

DATA:
C = 3 5;
B = 4 12 18;
A= 10
02
3 2;
Enddata
END Module: XI Facility Planning- 11
Prescriptive Analytics
Module: XI Facility Planning- 12
Prescriptive Analytics
Module: XI Facility Planning- 13
Prescriptive Analytics
 Solving given LP by LINGO

Solving another problem by LINGO

Maximize Z = 2X1 +4X2 + 3x3

Subject to ……………..
3X1 + 4x2 + 2x3 <= 60
2X1 + x2 + 2x3 <= 40
X1 + 3x2 + 2x3 <=80
XXI
Module: 1, X2, Facility
X3>=0Planning- 14
Prescriptive Analytics
 Single Objective Facility Layout Problem
(SOFLP) /QAP
 SOFLP is generally formulated as a quadratic
assignment problem (QAP) introduced by Koopmans
and Beckman (1957).
 According to Garey and Johnson (1979), and Sahni and
Gonzalez (1976) it is an NP-hard problem and one of the
frequently used formulation to solve FLP.

Module: XI Facility Planning- 15

Prescriptive Analytics
 MATHEMATICAL FORMULATION (QAP)

QAP: (Koopmans & Beckman, 1957)

n n n n

Min F(X)= 
i 1
  F
j 1 k 1 l 1
ik * Djl * Xij * Xkl (1)

ik j l

X
j 1
ij 1 for all i=1…n (2)

X
(3)
ij 1 for all j=1…n
i 1

(4)
X ij  {0,1}  i and j
Module: XI Facility Planning- 16
Prescriptive Analytics
Xij= 1 if Facility “i” is located/assigned to location “j”.

Xij= 0 if Facility “i” is not located/assigned to location

“j”.

Fik is the flow between two facility i and k.

Djl is the distance between two locations j and l.

n is the number of facilities (departments)

Module: XI Facility Planning- 17

Prescriptive Analytics
 Complexity of SOFLP (QAP)

 For n sized QAP, n facilities are to be assigned at n

locations. Therefore the numbers of possible
permutations are n! or possible feasible solutions are n!
(n factorial).
 To get optimal solution by exact algorithms all n!
permutations are to be enumerated that requires huge
computational effort. Therefore, exact algorithms for
solving QAP are limited to instances of size 20 to 30.

Module: XI Facility Planning- 18

Prescriptive Analytics
Process Layout (QAP)Example
Arrange six departments in a
factory to minimize the
material handling costs. Each
department is 20 x 20 feet and
the building is 60 feet long
and 40 feet wide.

Module: XI Facility Planning- 19

Prescriptive Analytics
 Flow Matrix
Number of loads per week
Department Assembly Painting Machine Receiving Shipping Testing
(1) (2) Shop (3) (4) (5) (6)

Assembly (1) 50 100 0 0 20

Painting (2) 30 50 10 0

Machine Shop (3) 20 0 100

Receiving (4) 50 0

Shipping (5) 0

Testing (6)

Module: XI Facility Planning- 20

Prescriptive Analytics
Process Layout (QAP)Example
Area 1 Area 2 Area 3

Assembly Painting Machine Shop

Department Department Department
(1) (2) (3)

40’

Receiving Shipping Testing

Department Department Department
(4) (5) (6)

Area 4 Area 5 Area 6

60’ 21
Module: XI Facility Planning-
Prescriptive Analytics
 Distance Matrix

Department Assembly Painting Machine Receiving Shipping Testing

(1) (2) Shop (3) (4) (5) (6)

Assembly (1) 1 2 1 1 2

Painting (2) 1 1 1 1

Machine Shop (3) 2 1 1

Receiving (4) 1 2

Shipping (5) 1

Testing (6)

Module: XI Facility Planning- 22

Prescriptive Analytics
 Optimal Solution by Lingo
• Objective=430
Area 1 Area 2 Area 3

5 3 1

40’

4 2 6

60’
Area 4 Area 5 Area 6 23
 QAP Example
Interdepartmental Flow Graph

100
5 3 1

30 20
50

4 2 6
50

Module: XI Facility Planning- Figure 9.6 24

Prescriptive Analytics
 Optimal Cost Recheck
n n

Cost = ∑ ∑ Xij Cij

i=1 j=1

Cost = $50 + $100 + $20

(1 and 2) (1 and 3) (1 and 6)
+ $30 + $50 + $10
(2 and 3) (2 and 4) (2 and 5)
+ $20 + $100 + $50
(3 and 4) (3 and 6) (4 and 5)

= $430

30-12-2022 25
Module: XI Facility Planning- 26
Prescriptive Analytics
 Certificate Course on
Operations Management and Analytics
(August 2022-January 2023)

Program Coordinator:
Dr. Surya Prakash Singh
PhD (IIT Kanpur), PDF (NUS Singapore-MIT USA)
Dhananjaya Chair Professor & Chairperson
Operation & Supply Chain Management Group
Department of Management Studies
E-Mail: [email protected], [email protected]
30-12-2022 1
 Session 4 & 5

Case study on Production Planning (Session 4)

Case Study on Distribution (Session 5)

30-12-2022 2
Case problem 1:

30-12-2022 3
Solution:

1. Selection of decision variables

2. Objective function modelling

3. Constraint modelling

4. Solving in EXCEL/LINGO SOLVER

30-12-2022 6
 1. Selection of decision variables
Indices
i = 1 to 4 for all quarters

Decision Variables
Xi, Yi, Ii, Hi, Fi, Ei

Xi= Number of products made using regular time in quarter i.

Yi= Number of products made using over time in quarter i.
Ii= Number of products left in quarter i.
Hi=Number of employees hired in quarter i.
Fi= Number of employees fired in quarter i.
Ei= Number of employees in quarter i.
30-12-2022 7
 2. Objective function modelling

Z (Minimze) = Production cost from regular time +

Production cost from over time +
Inventory carrying cost +
Employees firing cost +
Employees hiring cost

Z = ∑(320*Xi) + ∑(380*Yi) + ∑(20*Ii) + ∑(200*Hi) +

∑(200*Fi)

= 320(X1+X2+X3+X4) + 380(Y1+Y2+Y3+Y4)+
20(I1+I2+I3+I4)+ 200(H1+H2+H3+H4) +
200(F1+F2+F3+F4)
30-12-2022 8
 2. Constraint modelling
X0= 2400 Produciton level in last quarter

I0 + Xi +Yi = Ii + Di for Quarter i

I0= 300 Initial inventory level

E0= 600 Initial employees

I4=100 Desired closing inventory level

Ei=E0+Hi-Fi Employees in Quarter i

4Ei=Xi+Yi Production level for Quarter i

30-12-2022 Xi, Yi, Ii, Hi, Fi, Ei are integers 9
We can move either to EXCEL or
LINGO SOLVER to optimally solve
the case study.

30-12-2022 10
 3. LINGO Model

30-12-2022 11
 LINGO Solution

30-12-2022 12
 SENSITIVITY ANALYSIS
FROM LINGO Model

30-12-2022 13
 EXCEL MODEL

30-12-2022 14
30-12-2022 15
 EXCEL SOLVER SOLUTION

30-12-2022 16
 SENSITIVITY ANALYSIS FROM EXCEL SOLVER

30-12-2022 17
 Cost Data

30-12-2022 19
 Conditions or
constraints

30-12-2022 20
30-12-2022 21
30-12-2022 22
Solution using Prescriptive Analytics:

1. Selection of decision variables

2. Objective function modelling

3. Constraint modelling

4. Solving in EXCEL/LINGO SOLVER

30-12-2022 23
 1. Selection of decision variables

i j k
Indices
i = 1 to 2
J = 1 to 4
K = 1 to 6

30-12-2022 24
 Xij Yjk

Decision Variables
Xij
Yjk Total Variables
= 2*4+4*6+2*6
Zik = 8+24+12= 44

Zik
Xij = Quantity shipped from i to j (i.e. from Factories to Depots)

Yjk = Quantity shipped from j to k (i.e. from Depots to Customers)

Zik = Quantity shipped from i to k (i.e. from Factories to Customers)

30-12-2022 25
 2. Objective function modelling
i

j Xij

k Zik Yjk

Z (Minimize) = ∑ ∑ (Cij*Xij) + ∑ ∑ (Cjk*Yjk) + ∑ ∑ (CiK*ZiK)

Z =(0.5X11+ 0.5X12+ 1.0X13 + 0.2X14 + 1000X21+ 0.3X22+ 0.5X23 + 0.2X24)+
(1000Y11+ 1.5Y12+ 0.5Y13 + 1.5Y14 + 1000Y15+ 1.0Y16+
1.0Y21+ 0.5Y22+ 0.5Y23 + 1.0Y24 + 0.5Y25+ 1000Y26 +
1000Y31+ 1.5Y32+ 2.0Y33 + 1000Y34 + 0.5Y35+ 1.5Y36 +
1000Y41+ 1000Y42+ 0.2Y43 + 1.5Y44 + 0.5Y45+ 1.5Y46)+
(1.0Z11+ 1000Z12+ 1.5Z13 + 2.0Z14 + 1000Z15+ 1.0Z16+
2.0Z21+ 1000Z22+ 1000Z23 + 1000Z24 + 1000Z25+ 1000Z26 )
30-12-2022 26
 3. Constraints modelling
X11+X12+X13+X14+Z11+Z12+Z13+Z14+Z15+Z16<=150000
X21+X22+X23+X24+Z21+Z22+Z23+Z24+Z25+Z26<=200000

X11+X21= Y11+Y12+Y13+Y14+Y15+Y16 (Depot 1)

X12+X22= Y21+Y22+Y23+Y24+Y25+Y26 (Depot 2)
X13+X23= Y31+Y32+Y33+Y34+Y35+Y36 (Depot 3)
X14+X24= Y41+Y42+Y43+Y44+Y45+Y46 (Depot 4)
X11+X21<=70000 (Depot capacity 1)
X12+X22 <=50000 (Depot capacity 2)
X13+X23 <=100000 (Depot capacity 3)
X14+X24 <=40000 (Depot capacity 4)

Z11+Z21+Y11+Y21+Y31+Y41>=50000 (C1 Customer)

Z12+Z22+Y12+Y22+Y32+Y42 >=10000 (C2 Customer)
Z13+Z23+Y13+Y23+Y33+Y43 >=40000 (C3 Customer)
Z14+Z24+Y14+Y24+Y34+Y44 >=35000 (C4 Customer)
Z15+Z25+Y15+Y25+Y35+Y45 >=60000 (C5 Customer)
Z16+Z30-12-2022
26+Y16+Y26+Y36+Y46 >=20000 (C6 Customer) 27
 Z11 <= 50000 (C1 Customer)
Y12 <= 10000 (C2 Customer)
Y25 <= 60000 (C5 Customer)
Y36+Y46<=20000 (C6 Customer)

Xij>=0 for all i & j

Yjk >=0 for all j & k
ZiK >=0 for all i & K

30-12-2022 28
 We can move either to EXCEL or
LINGO SOLVER to optimally solve
the case study.

30-12-2022 29
 3. LINGO Model

30-12-2022 30
 LINGO Solution

30-12-2022 31
 Sensitivity Analysis

30-12-2022 32
 LINGO Model using sets

30-12-2022 33
30-12-2022 34
30-12-2022 35