0% found this document useful (0 votes)
334 views102 pages

DM Mid Term

The document outlines the components and evaluation criteria for the Decision Modelling course. It will include individual assignments, quizzes, a mid-term exam, end-term exam, and a group project presentation, with varying weightages and durations. The group project will utilize course concepts to formulate and solve a decision-making problem. Decision making involves finding the best course of action and can be normative, aiming for optimal solutions, or descriptive, considering real-world influences. Mathematical models can be classified as static or dynamic, deterministic or probabilistic, and with single or multiple objectives. The problem-solving process involves defining the problem, identifying alternatives, evaluating criteria, and choosing a solution.

Uploaded by

Sourya Mitra
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
334 views102 pages

DM Mid Term

The document outlines the components and evaluation criteria for the Decision Modelling course. It will include individual assignments, quizzes, a mid-term exam, end-term exam, and a group project presentation, with varying weightages and durations. The group project will utilize course concepts to formulate and solve a decision-making problem. Decision making involves finding the best course of action and can be normative, aiming for optimal solutions, or descriptive, considering real-world influences. Mathematical models can be classified as static or dynamic, deterministic or probabilistic, and with single or multiple objectives. The problem-solving process involves defining the problem, identifying alternatives, evaluating criteria, and choosing a solution.

Uploaded by

Sourya Mitra
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 102

09/11/2020

EVALUATION COMPONENTS
S. No. Component Weightage Duration of Nature of
(%) Examination Examination
1 Individual 10% - Open Book
Assignments &
DECISION MODELLING Class Participation
TERM-II 2 Quizzes (n-1 quiz 10% Between 15- Closed Book
marks will be 25 minutes
MBA (2020-21) counted)
3 Mid-term Exam 30% 1.5 hour Closed Book
4 End-term Exam 30% 2 hour Closed Book
SESSION 1 5 Group Project 20% - Presentation
Presentation
INTRODUCTION TO DECISION MODELLING
Group project (preferably six students in a group)
The group project will provide an opportunity to utilize the course
Dr. Devendra Kumar Pathak concepts and employ the discussed tools/ techniques to formulate and
(Ph.D., IIT Delhi)
solve a decision-making problem.
Assistant Professor, An Introduction to Management Science: 2
Operations Management & Decision Sciences, Quantitative Approaches to Decision Making
by Anderson, Sweeney, Williams, and Martins.
Indian Institute of Management (IIM) Kashipur
2012. Cengage Learning. (13th Edition)

INTRODUCTION: DECISION MAKING NORMATIVE VS. DESCRIPTIVE DECISION MAKING

 Decision Making is the most generic term in ‘Management’.


Normative (Prescriptive) decision theory models the most ideal
 Decision making precedes any activity- personal or decision for a given situation.
organizational. • In normative theory, an actor is assumed to be fully rational.
 Forecasting and Decision making are two fundamental building • Normative decisions always try to find the highest expected
blocks of any managerial function. value outcome (optimal solution).
Forecasting / • Economic Man/ Rationality-based model
Prediction about Decision making Action
future situation
Descriptive decision theory is more about what will occur in a
situation, not what should.
 The prime focus of Decision Making is to find the best • Descriptive decision theory takes into consideration outside
course of action in every situation. factors that influence an actor’s decisions toward less optimal,
less rational ends.
3 • Administrative man model (Bounded4
Rationality)

1
09/11/2020

CLASSIFICATION OF MATHEMATICAL MODELS TYPES OF MATHEMATICAL MODEL

Mathematical Model
• Descriptive vs. Prescriptive Inventory
Theory

Math. Programming Queuing Replacement


• Static vs. Dynamic Model
Simulation
Static models are at an equilibrium or steady state, as Statistical &
Game theory
Econometric
opposed to dynamic models which change with respect to Linear
time Programming

•Transportation NLP Dynamic Stochastic


•Deterministic vs. Probabilistic •Assignment Programming Programming
Integer
Programming
•Single objective vs. Multiple Objective

All integer Mixed integer 0-1 Programming

5 6
Multiple Objective Decision Making (MODM) – A special Class of quantitative methods

PROBLEM SOLVING QUANTITATIVE ANALYSIS AND DECISION MAKING

 7 Steps of Problem Solving  Decision-Making Process


(First 5 steps are the process of decision making)
Structuring the Problem Analyzing the Problem
1. Define the problem.
2. Determine the set of alternative solutions. Define Identify Determine Choose
Evaluate
the the the an
3. Determine the criteria for evaluating Problem Alternatives Criteria
Alternatives
Alternative
alternatives.
4. Evaluate the alternatives. • Problems in which the objective is to find the best solution
5. Choose an alternative (make a decision). with respect to one criterion are referred to as single-
criterion decision problems. [Car selection based on cost]
-------------------------------------------------------------------
-- • Problems that involve more than one criterion are referred
to as multi-criteria decision problems. [Car selection based on cost,
6. Implement the selected alternative. comfort, fuel efficiency, mileage, resale value, etc.]
7 8
7. Evaluate the results.

2
09/11/2020

QUANTITATIVE ANALYSIS AND DECISION MAKING QUALITATIVE ANALYSIS AND DECISION MAKING

 Analysis Phase of Decision-Making Process

Qualitative Analysis
• based largely on the manager’s judgment and
experience
• includes the manager’s intuitive “feel” for the
problem
• is more of an art than a science

9 10

QUANTITATIVE ANALYSIS AND DECISION MAKING QUANTITATIVE ANALYSIS AND DECISION MAKING

 Analysis Phase of Decision-Making Process  Quantitative Analysis Process

Quantitative Analysis  Model Development


• analyst will concentrate on the quantitative  Data Preparation
facts or data associated with the problem  Model Solution
 Report Generation
• analyst will develop mathematical expressions
that describe the objectives, constraints, and
other relationships that exist in the problem

• analyst will use one or more quantitative


methods to make a recommendation
11 12

3
09/11/2020

MODEL DEVELOPMENT MATHEMATICAL MODELS

 Models are representations of real objects or situations  Objective Function – a mathematical expression that
describes the problem’s objective, such as maximizing
profit or minimizing cost
 Three forms of models are:
 Iconic models - physical replicas (scalar representations) of  Consider a simple production problem.
real objects (e.g., model airplane and toy truck)
 Suppose x denotes the number of units produced and sold each
 Analog models - physical in form, but do not physically week,
resemble the object being modeled (e.g., speedometer,  the firm’s objective is to maximize total weekly profit.

 With a profit of $10 per unit,


thermometer)
 the objective function is 10x.
 Mathematical models - represent real world problems
through a system of mathematical formulas and
expressions based on key assumptions, estimates, or
13 14
statistical analyses

MATHEMATICAL MODELS MATHEMATICAL MODELS

 Constraints – a set of restrictions or limitations, such


as production capacities  Uncontrollable Inputs – environmental factors that
are not under the control of the decision maker
 To continue our example, a production capacity
constraint would be necessary if, for instance, 5  In the preceding mathematical model, the profit per
hours are required to produce each unit and only 40 unit ($10), the production time per unit (5 hours),
hours are available per week. The production and the production capacity (40 hours) are
capacity constraint is given by 5x < 40. environmental factors not under the control of the
manager or decision maker.
 The value of 5x is the total time required to produce x
units; the symbol indicates that the production time
required must be less than or equal to the 40 hours
available.
15 16

4
09/11/2020

MATHEMATICAL MODELS MATHEMATICAL MODELS

 Decision Variables – controllable inputs; decision  A complete mathematical model for our simple
alternatives specified by the decision maker, such as production problem is:
the number of units of a product to produce.
Maximize 10x (objective function)
 In the preceding mathematical model, the subject to: 5x < 40 (constraint)
production quantity x is the controllable input to the
x>0 (non-negativity constraint)
model.

[The second constraint reflects the fact that it is not


possible to manufacture a negative number of units.]

17 18

MATHEMATICAL MODELS MATHEMATICAL MODELS

 Deterministic Model – if all uncontrollable inputs to the


model are known and cannot vary Uncontrollable Inputs
(Environmental Factors)

 Stochastic (or Probabilistic) Model – if any


uncontrollable are uncertain and subject to variation Controllable
Output
Inputs Mathematical
(Projected
 Stochastic models are often more difficult to analyze. (Decision Model
Results)
 In our simple production example, if the number of
Variables)
hours of production time per unit could vary from 3 to
6 hours depending on the quality of the raw material,
the model would be stochastic.
19 20

5
09/11/2020

MODEL SOLUTION MODEL SOLUTION

 The analyst attempts to identify the alternative (the set of  Trial-and-Error Solution for Production Problem
decision variable values) that provides the “best” output
for the model. Production Projected Total Hours Feasible
Quantity Profit of Production Solution
 The “best” output is the optimal solution. 0 0 0 Yes
2 20 10 Yes
 If the alternative does not satisfy all of the model 4 40 20 Yes
constraints, it is rejected as being infeasible, regardless 6 60 30 Yes
of the objective function value.
8 80 40 Yes
10 100 50 No
 If the alternative satisfies all of the model constraints, it is
12 120 60 No
feasible and a candidate for the “best” solution.
21 Maximize 10x (objective function) 22
subject to: 5x < 40 (constraint)
x>0 (constraint)

MODEL SOLUTION REPORT GENERATION

 A variety of software packages are available for solving  A managerial report, based on the results of the model,
mathematical models. should be prepared.
 The report should be easily understood by the decision
 Microsoft Excel maker.
 LINGO  The report should include:
(https://www.lindo.com/lindoforms/downlingo.html)
 the recommended decision
 other pertinent information about the results (for example,
how sensitive the model solution is to the assumptions and
data used in the model)

23 24

6
09/11/2020

EXAMPLE: PONDEROSA DEVELOPMENT CORP. EXAMPLE: PONDEROSA DEVELOPMENT CORP.

PDC leases office space for $2,000 per month.


Ponderosa Development Corporation (PDC) is a small The cost of supplies, utilities, and leased
real estate developer that builds only one style house.
equipment runs another $3,000 per month.
The selling price of the house is $115,000.

The one salesperson of PDC is paid a


Land for each house costs $55,000 and lumber,
commission of $2,000 on the sale of each house.
supplies, and other materials run another $28,000 per
house. Total labor costs are approximately $20,000 per
house. PDC has seven permanent office employees whose
monthly salaries are given on the next slide.

25 26

EXAMPLE: PONDEROSA DEVELOPMENT CORP. EXAMPLE: PONDEROSA DEVELOPMENT CORP.

Employee Monthly Salary Question


President $10,000 Identify all costs and denote the marginal cost and
VP, Development 6,000 marginal revenue for each house.
VP, Marketing 4,500
Project Manager 5,500
Controller 4,000
Office Manager 3,000
Receptionist 2,000
Marginal cost is defined as the rate of change of the total cost with respect to
production volume. That is, it is the cost increase associated with a one-
unit increase in the production volume. (Variable cost)

Marginal revenue is defined as the rate of change of total revenue with respect
27 28
to sales volume. That is, it is the increase in total revenue resulting from a
one-unit increase in sales volume.

7
09/11/2020

EXAMPLE: PONDEROSA DEVELOPMENT CORP. EXAMPLE: PONDEROSA DEVELOPMENT CORP.

Question
Write the monthly cost function c (x), revenue
Answer
function r (x), and profit function p (x).
The monthly salaries total $35,000 and
monthly office lease and supply costs total another $5,000.
This $40,000 is a monthly fixed cost.
Answer
The total cost of land, material, labor, and
sales commission per house, $105,000, is the marginal c (x) = variable cost + fixed cost = 105,000x + 40,000
cost for a house. r (x) = 115,000x
p (x) = r (x) - c (x) = 10,000x - 40,000
The selling price of $115,000 is the
marginal revenue per house.

29 30

EXAMPLE: PONDEROSA DEVELOPMENT CORP. EXAMPLE: PONDEROSA DEVELOPMENT CORP.

Question
What is the breakeven point for monthly sales
of the houses?

Answer
r (x ) = c (x )
115,000x = 105,000x + 40,000
Solving, x = 4.

31 32

8
09/11/2020

EXAMPLE: PONDEROSA DEVELOPMENT CORP. EXAMPLE: PONDEROSA DEVELOPMENT CORP.

Question 1200 Total Revenue =


115,000x

Thousands of Dollars
What is the monthly profit if 12 houses per month are 1000
built and sold?
800
Answer
600
p (12) = 10,000(12) - 40,000 = $80,000 monthly profit Total Cost =
400 40,000 + 105,000x

200
Break-Even Point = 4 Houses
0
0 1 2 3 4 5 6 7 8 9 10
Number of Houses Sold (x)
33 34

MANAGEMENT SCIENCE TECHNIQUES MANAGEMENT SCIENCE TECHNIQUES

 Linear programming is a problem-solving approach


 Linear Programming  Simulation developed for situations involving maximizing or
 Integer Linear  Decision Analysis minimizing a linear function subject to linear
Programming  Goal Programming constraints that limit the degree to which the objective
 Nonlinear Programming can be pursued.
 Analytic Hierarchy Process
 PERT/CPM  Integer linear programming is an approach used for
 Forecasting
problems that can be set up as linear programs with
 Inventory Models  Markov-Process Models the additional requirement that some or all of the
 Waiting Line Models  Distribution/Network Models decision recommendations be integer values.
 Network models are specialized solution procedures
for problems in transportation system design,
information system design, project scheduling, …..
35 36

9
09/11/2020

MANAGEMENT SCIENCE TECHNIQUES MANAGEMENT SCIENCE TECHNIQUES

 Project scheduling: PERT (Program Evaluation and  Waiting line (or queuing) models help managers
Review Technique) and CPM (Critical Path Method) understand and make better decisions concerning the
help managers responsible for planning, scheduling, operation of systems involving waiting lines.
and controlling projects that consist of numerous  Simulation is a technique used to model the operation
separate jobs or tasks performed by a variety of of a system. This technique employs a computer
departments, individuals, and so forth. program to model the operation and perform
 Inventory models are used by managers faced with the simulation computations.
dual problems of maintaining sufficient inventories to  Decision analysis can be used to determine optimal
meet demand for goods and, at the same time, incurring strategies in situations involving several decision
the lowest possible inventory holding costs. alternatives and an uncertain pattern of future events.
 Forecasting methods are techniques that can be used
to predict future aspects of a business operation.
37 38

MANAGEMENT SCIENCE TECHNIQUES KEY TERMINOLOGIES

 Goal programming is a technique for solving multi- Feasible Solution:


criteria decision problems, usually within the
•Any solution LPP/NLP which does not violate constraints is called
framework of linear programming. feasible solution.
 Analytic hierarchy process is a multi-criteria decision-
• Feasible solution may be optimal (best) solution.
making technique that permits the inclusion of
subjective factors in arriving at a recommended •Any LPP/NLP can have more than one feasible solution.
decision.
•Optimal solution to LPP/NLP must be a feasible solution.
 Markov-process models are useful in studying the
evolution of certain systems over repeated trials (such Infeasible Solution:
as describing the probability that a machine, •Any solution that violates at least one constraint is called infeasible
functioning in one period, will function or break down solution.
in another period).
•Any LPP/NLP may have infinite number of infeasible solution.
39 40
•Infeasible solution lies outside the bounded region.

10
09/11/2020

KEY TERMINOLOGIES

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.
41

11
11/11/2020

INTRODUCTION TO LINEAR PROGRAMMING

 Linear Programming Problem


DECISION MODELLING  Problem Formulation
 A Simple Maximization Problem
TERM-II
 Graphical Solution Procedure
MBA (2020-21)
 Slack & Surplus Variables
 Extreme Points and the Optimal Solution

SESSION 2  Reduced Cost

LINEAR PROGRAMMING  Computer Solutions


[FORMULATION OF THE LP MODEL]  A Simple Minimization Problem
 Special Cases
Dr. Devendra Kumar Pathak
(Ph.D., IIT Delhi)
Assistant Professor, 2
Operations Management & Decision Sciences,
Indian Institute of Management (IIM) Kashipur

OR MODELLING SOME POPULAR MANAGEMENT PROBLEMS

OR : Solving management decision making


problems through mathematical modeling 1. Product Mix Problem (Production Management)
Applications 2. Media Selection Problem (Marketing
Accounting: Cash flow planning, Credit policy Management)
Finance: Capital budgeting, Portfolio analysis, Dividend policy 3. Portfolio Selection Problem (Finance Management)
Project management: Allocation of resources to projects, 4. Labor Planning / Staffing Problem (HR Management)
Determination of crew size, Project scheduling
5. Diet Problem (Food industry, Hospitals etc)
Facility planning: Facility size & location, Capacity Planning,
6. Blending Problem (Chemical Industry)
Production / Manufacturing: Product mix, Production planning /
scheduling, Lot size, Inventory planning & control, Procurement 7. Production scheduling Problem (Multiperiod planning /
mix Aggregate Planning)
Marketing: Sales promotional planning, Sales mix, Sales allocation
to markets
OB/ HR: Manpower planning, Scheduling of training programs
R&D: Product introduction timing, Appropriate product design

1
11/11/2020

LINEAR PROGRAMMING (LP) PROBLEM LINEAR PROGRAMMING (LP) PROBLEM


 Linear programming involves choosing a course of
action when the mathematical model of the problem  If both the objective function and the
contains only linear functions. constraints are linear, the problem is referred to
 The maximization or minimization of some quantity is as a linear programming problem.
the objective in all linear programming problems.
 Linear functions are functions in which each
 All LP problems have constraints that limit the degree to variable appears in a separate term raised to the
which the objective can be pursued. first power and is multiplied by a constant (which
 A feasible solution satisfies all the problem's could be 0).
constraints.
 An optimal solution is a feasible solution that results in  Linear constraints are linear functions that are
the largest possible objective function value when restricted to be "less than or equal to", "equal to",
maximizing (or smallest when minimizing). or "greater than or equal to" a constant.

5 6
 A graphical solution method can be used to solve a
linear program with two variables.

PROBLEM FORMULATION GUIDELINES FOR MODEL FORMULATION

 Problem formulation or modeling is the process of  Understand the problem thoroughly.


translating a verbal statement of a problem  Describe the objective.
into a mathematical statement.
 Describe each constraint (capacity, time)

 Define the decision variables (CI)

 Write the objective in terms of the decision


 Every LP problems has some unique features, but
most problems also have common features. variables.
 Write the constraints in terms of the decision

 General guidelines for LP model formulation are variables.


illustrated on the slides that follow.

7 8

2
11/11/2020

EXAMPLE 1: A SIMPLE MAXIMIZATION PROBLEM EXAMPLE 1: GRAPHICAL SOLUTION

 LP Formulation  First Constraint Graphed


Objective
x2
Max 5x1 + 7x2 Function
8
s.t. x1 < 6 “Regular”
7 x1 = 6

2x1 + 3x2 < 19 Constraints 6


Shaded region
x1 + x2 < 8 5 contains all
Non-negativity feasible points
4
x1 > 0 and x2 > 0 Constraints for this constraint
3
2 (6, 0)
1
x1
9 1 2 3 4 5 6 7 8 9 10 10

EXAMPLE 1: GRAPHICAL SOLUTION EXAMPLE 1: GRAPHICAL SOLUTION

 Second Constraint Graphed  Third Constraint Graphed


x2 x2 (0, 8)
8 (0, 6 1/3) 8
7 7
6 6 x1 + x2 = 8
2x1 + 3x2 = 19
5 5
4 4
Shaded Shaded
3 3
region contains region contains
2 all feasible points 2 all feasible points
(9 1/2, 0)
for this constraint for this constraint (8, 0)
1 1
x1 x1
1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 8 9 10 12

3
11/11/2020

EXAMPLE 1: GRAPHICAL SOLUTION EXAMPLE 1: GRAPHICAL SOLUTION

 Combined-Constraint Graph Showing Feasible Region


 Objective Function Line
x2 x2
x1 + x2 = 8
8 8 Arbitrary
value
7 7
(0, 5)
x1 = 6
6 6 Objective Function
5x1 + 7x2 = 35
5 5
4 4
3 3
Feasible Region
2x1 + 3x2 = 19
2 (set of all feasible solutions) 2
(7, 0)
1 1
x1 x1
1 2 3 4 5 6 7 8 9 10 13 1 2 3 4 5 6 7 8 9 10 14

EXAMPLE 1: GRAPHICAL SOLUTION EXAMPLE 1: GRAPHICAL SOLUTION

Selected Objective Function Lines  Optimal Solution


x2 x2 Maximum
Objective Function Line
8 8 5x1 + 7x2 = 46

7 7
5x1 + 7x2 = 35
6 6 Optimal Solution
5 5x1 + 7x2 = 39 5 (x1 = 5, x2 = 3)

4 4
3 5x1 + 7x2 = 42 3
2 2
1 1
x1 x1
1 2 3 4 5 6 7 8 9 10 15 1 2 3 4 5 6 7 8 9 10 16

4
11/11/2020

SUMMARY OF THE GRAPHICAL SOLUTION PROCEDURE


FOR MAXIMIZATION PROBLEMS SLACK AND SURPLUS VARIABLES

 Prepare a graph of the feasible solutions for each of  A linear program in which all the variables are non-
the constraints. negative and all the constraints are equalities is
 Determine the feasible region that satisfies all the said to be in standard form.
constraints simultaneously.
 Draw an objective function line.  Standard form is attained by adding slack variables
 Move parallel objective function lines toward larger to "less than or equal to" constraints, and by
objective function values without entirely leaving subtracting surplus variables from "greater than or
the feasible region. equal to" constraints.
 Any feasible solution on the objective function line
with the largest value is an optimal solution.  Slack and surplus variables represent the difference
between the left and right sides of the constraints.

 Slack and surplus variables have objective


17 18
function coefficients equal to 0.

SLACK VARIABLE (FOR ≤ CONSTRAINTS) SLACK VARIABLE

Example 1 in Standard Form Optimal Solution


x2 Third
Max 5x1 + 7x2 + 0s1 + 0s2 + 0s3 Constraint: First
x1 + x2 = 8 Constraint:
8 x1 = 6
s.t. x1 + s1 = 6 7
s3 = 0
2x1 + 3x2 + s2 = 19 6 s1 = 1
x1 + x2 + s3 = 8 5
Second
x1, x2 , s1 , s2 , s3 > 0 4 Constraint:
2x1 + 3x2 = 19
3
Optimal
s1 , s2 , and s3 2 Solution s2 = 0
Max 5x1 + 7x2 are slack variables (x1 = 5, x2 = 3)
1
s.t. x1 < 6
2x1 + 3x2 < 19 x1
19 20
x1 + x2 < 8 1 2 3 4 5 6 7 8 9 10
x1 > 0 and x2 > 0

5
11/11/2020

EXTREME POINTS AND THE OPTIMAL SOLUTION EXAMPLE 1: EXTREME POINTS

 The corners or vertices of the feasible region are x2


referred to as the extreme points.
8
 An optimal solution to an LP problem can be 7 5 (0, 6 1/3)
found at an extreme point of the feasible region. 6
5
 When looking for the optimal solution, you do not 4
have to evaluate all feasible solution points. Feasible
4 (5, 3)
3
Region 3 (6, 2)
2
 You have to consider only the extreme points
1 1 (0, 0) 2 (6, 0)
of the feasible region.
x1
21 1 2 3 4 5 6 7 8 9 10 22

COMPUTER SOLUTIONS INTERPRETATION OF COMPUTER OUTPUT

 LP problems involving 1000s of variables and 1000s of  Now we will discuss the following output:
constraints are now routinely solved with computer  objective function value
packages.  values of the decision variables
 Linear programming solvers are now part of many  reduced costs
spreadsheet packages, such as Microsoft Excel.  slack and surplus

 In the next session we will discuss how an optimal


 Leading commercial packages include CPLEX, solution is affected by a change in:
LINGO, MOSEK, Xpress-MP, and Premium Solver  a coefficient of the objective function
for Excel.  the right-hand side value of a constraint

23 24

6
11/11/2020

EXAMPLE 1: LINGO SOLUTION EXAMPLE 1: LINGO SOLUTION

Amount
unused

25 26

REDUCED COST EXAMPLE 1: LINGO SOLUTION


Reduced Costs
 The reduced cost for a decision variable whose
value is 0 in the optimal solution is:
the amount the variable's objective function coefficient
would have to improve (increase for maximization
problems, decrease for minimization problems) before
this variable could assume a positive value (i.e., to be
the part of optimal solution).

 The reduced cost for a decision variable whose Max 5x1 + 7x2 + 9x3
value is > 0 in the optimal solution is 0. s.t. x1 < 6
2x1 + 3x2 < 19
x1 + x2 < 8
x1 > 0 and x2 > 0

27 28

7
11/11/2020

EXAMPLE 2: A SIMPLE MINIMIZATION PROBLEM EXAMPLE 2: GRAPHICAL SOLUTION

 LP Formulation  Graph the Constraints


Constraint 1: When x1 = 0, then x2 = 2; when x2 = 0,
Min 5x1 + 2x2 then x1 = 5. Connect (5,0) and (0,2). The ">" side is
s.t. 2x1 + 5x2 > 10 above this line.
4x1 - x2 > 12 Constraint 2: When x2 = 0, then x1 = 3. But setting x1
x1 + x2 > 4 to 0 will yield x2 = -12, which is not on the graph. Thus,
to get a second point on this line, set x1 to any
x1, x2 > 0 number larger than 3 and solve for x2: when x1 = 5,
then x2 = 8. Connect (3,0) and (5,8). The ">” side is to
the right.

29 30

EXAMPLE 2: GRAPHICAL SOLUTION EXAMPLE 2: GRAPHICAL SOLUTION

Graph the Constraints (continued)  Constraints Graphed


Constraint 3: When x1 = 0, then x2 = 4; when x2 = 0, x2
6
then x1 = 4. Connect (4,0) and (0,4). The ">" side is Feasible Region
above this line. 5
4x1 - x2 > 12
4
x1 + x2 > 4
3

2 2x1 + 5x2 > 10

x1
31 1 2 3 4 5 6 32

8
11/11/2020

EXAMPLE 2: GRAPHICAL SOLUTION EXAMPLE 2: GRAPHICAL SOLUTION

 Graph the Objective Function Objective Function Graphed


Set the objective function equal to an arbitrary x2 Min 5x1 + 2x2
constant (say 20) and graph it. For 5x1 + 2x2 = 20, 6
when x1 = 0, then x2 = 10; when x2= 0, then x1 = 4. 5
Connect (4,0) and (0,10). 4x1 - x2 > 12
4
 Move the Objective Function Line Toward x1 + x2 > 4
Optimality 3
Move it in the direction which lowers its value 2x1 + 5x2 > 10
2
(down), since we are minimizing, until it touches the
last point of the feasible region, determined by the 1
last two constraints.
x1
33 1 2 3 4 5 6 34

EXAMPLE 2: GRAPHICAL SOLUTION EXAMPLE 2: GRAPHICAL SOLUTION

 Solve for the Extreme Point at the Intersection of the Optimal Solution
Two Binding Constraints (slack/surplus value 0) x2
4x1 - x2 = 12 6
x1+ x2 = 4 4x1 - x2 > 12
5
Adding these two equations gives: x1 + x2 > 4
4
5x1 = 16 or x1 = 16/5 Optimal Solution:
3 x1 = 16/5, x2 = 4/5,
Substituting this into x1 + x2 = 4 gives: x2 = 4/5 5x1 + 2x2 = 17.6
2
Solve for the Optimal Value of the Objective Function 2x1 + 5x2 > 10
1
5x1 + 2x2 = 5(16/5) + 2(4/5) = 88/5
x1
35 1 2 3 4 5 6 36

9
11/11/2020

SUMMARY OF THE GRAPHICAL SOLUTION PROCEDURE


FOR MINIMIZATION PROBLEMS SURPLUS VARIABLES (FOR ≥ CONSTRAINTS)

Prepare a graph of the feasible solutions for each of Example 2 in Standard Form
the constraints.
Determine the feasible region that satisfies all the Min 5x1 + 2x2 + 0s1 + 0s2 + 0s3
constraints simultaneously.
s.t. 2x1 + 5x2 - s1 = 10
Draw an objective function line.
4x1 - x2 - s2 = 12
Move parallel objective function lines toward smaller Surplus Variables
objective function values without entirely leaving the x1 + x2 - s3 = 4
feasible region. x1, x2, s1, s2, s3 > 0
Any feasible solution on the objective function line
with the smallest value is an optimal solution.
Min 5x1 + 2x2 s1 , s2 , and s3 are
s.t. 2x1 + 5x2 > 10 surplus variables
4x1 - x2 > 12
37 x1 + x2 > 4 38
x1, x2 > 0

EXAMPLE 2: LINGO SOLUTION EXAMPLE 2: LINGO SOLUTION

Excess
Amount

39 40

10
11/11/2020

SPECIAL CASES EXAMPLE: ALTERNATIVE OPTIMAL SOLUTIONS

 Alternative Optimal Solutions


In the graphical method, if the objective function Consider the following LP problem.
line is parallel to a boundary constraint in the
direction of optimization, there are alternate Max 4x1 + 6x2
optimal solutions, with all points on this line
segment being optimal. s.t. x1 < 6
2x1 + 3x2 < 18
x1 + x2 < 7

x1 > 0 and x2 > 0

42 43

EXAMPLE: ALTERNATIVE OPTIMAL SOLUTIONS SPECIAL CASES

Boundary constraint 2x1 + 3x2 < 18 and objective Infeasibility


function Max 4x1 + 6x2 are parallel. All points on • No solution to the LP problem satisfies all the
line segment A – B are optimal solutions. constraints, including the non-negativity conditions.
x2
7
x1 + x2 < 7
• Graphically, this means a feasible region does not
exist.
6
A • Causes include:
5
B x1 < 6 • A formulation error has been made.
4
• Management’s expectations are too high.
3 2x1 + 3x2 < 18
• Too many restrictions have been placed on the
2
problem (i.e. the problem is over-constrained).
1
x1
1 2 3 4 5 6 7 8 9 10 44 45

11
11/11/2020

EXAMPLE: INFEASIBLE PROBLEM EXAMPLE: INFEASIBLE PROBLEM

 Consider the following LP problem.  There are no points that satisfy both constraints, so
there is no feasible region (and no feasible solution).
Max 2x1 + 6x2 x2
10
s.t. 4x1 + 3x2 < 12
2x1 + x2 > 8 2x1 + x2 > 8
8

x1, x2 > 0 6
4x1 + 3x2 < 12
4

x1
46
2 4 6 8 10 47

SPECIAL CASES EXAMPLE: UNBOUNDED SOLUTION

Unbounded  Consider the following LP problem.


• The solution to a maximization LP problem is
unbounded if the value of the solution may be Max 4x1 + 5x2
made indefinitely large without violating any of the
constraints. s.t. x1 + x2 > 5
• For real problems, this is the result of improper 3x1 + x2 > 8
formulation. (Quite likely, a constraint has been x1, x2 > 0
accidentally omitted.)

48 49

12
11/11/2020

EXAMPLE: UNBOUNDED SOLUTION

 The feasible region is unbounded and the objective


function line can be moved outward from the origin
without bound, infinitely increasing the objective
function. x2
10
3x1 + x2 > 8
8

4
x1 + x2 > 5
2

x1 50
2 4 6 8 10

13
11/11/2020

MEDIA SELECTION

 One application of linear programming in marketing is


media selection.
DECISION MODELLING
TERM-II  LP can be used to help marketing managers allocate a
MBA (2020-21) fixed budget to various advertising media.

 The objective is to maximize reach, frequency, and


quality of exposure.
CASE 1
MEDIA SELECTION  Restrictions on the allowable allocation usually arise
during consideration of company policy, contract
requirements, and media availability.
Dr. Devendra Kumar Pathak
(Ph.D., IIT Delhi)
Assistant Professor, 2
Operations Management & Decision Sciences,
Indian Institute of Management (IIM) Kashipur

MEDIA SELECTION MEDIA SELECTION

SMM Company recently developed a new instant Estimated Audience


salad machine, has a maximum of $282,000 to spend Ad Type Reached With Each Ad Cost Per Ad
on advertising. The product is to be initially test Daytime (D) 3,000 $5,000
marketed in the Dallas Evening News (E) 4,000 $7,000
area. The money is to be spent on a TV advertising blitz Sunday Game (G) 75,000 $100,000
during one weekend (Friday, Saturday, and Sunday)
in November. Media Selection SMM wants to take out at least one ad of each
The three options available are: daytime type (daytime, evening-news, and game-time).
advertising, evening news advertising, and Sunday Further, there are a max. of two game-time ad spots
game-time advertising. A mixture of one-minute TV available. There are a max. of ten daytime spots and
spots is desired. six evening news spots available daily. SMM wants
to have at least 5 ads per day, but spend no more than
$50,000 on Friday and no more than $75,000 on
3 Saturday. 4

1
11/11/2020

MEDIA SELECTION

Define the Decision Variables

DFR = number of daytime ads on Friday

MEDIA SELECTION DSA = number of daytime ads on Saturday


DSU = number of daytime ads on Sunday
EFR = number of evening ads on Friday
ESA = number of evening ads on Saturday
ESU = number of evening ads on Sunday
GSU = number of game-time ads on Sunday

5 6

MEDIA SELECTION MEDIA SELECTION

Define the Objective Function Define the Constraints


Take out at least one ad of each type:
(1) DFR + DSA + DSU > 1
Maximize the total audience reached: (2) EFR + ESA + ESU > 1
(3) GSU > 1
Max (audience reached per ad of each type) Ten daytime spots available:
x (number of ads used of each type) (4) DFR < 10
Max 3000DFR +3000DSA +3000DSU +4000EFR (5) DSA < 10
+4000ESA +4000ESU +75000GSU (6) DSU < 10
Six evening news spots available:
(7) EFR < 6
(8) ESA < 6
(9) ESU < 6
7 8

2
11/11/2020

MEDIA SELECTION MEDIA SELECTION

Define the Constraints (continued) Define the Constraints (continued)


Only two Sunday game-time ad spots available: Spend no more than $75,000 on Saturday:
(10) GSU < 2 (15) 5000DSA + 7000ESA < 75000
At least 5 ads per day: Spend no more than $282,000 in total:
(11) DFR + EFR > 5 (16) 5000DFR + 5000DSA + 5000DSU + 7000EFR
(12) DSA + ESA > 5 + 7000ESA + 7000ESU + 100000GSU < 282000
(13) DSU + ESU + GSU > 5
Non-negativity:
Spend no more than $50,000 on Friday: DFR, DSA, DSU, EFR, ESA, ESU, GSU > 0
(14) 5000DFR + 7000EFR < 50000

9 10

MEDIA SELECTION MEDIA SELECTION

The Management Scientist Solution Solution Summary

Objective Function Value = 199000.000 Total new audience reached = 199,000


Number of daytime ads on Friday = 8
Variable Value Reduced Costs Number of daytime ads on Saturday = 5
DFR 8.000 0.000 Number of daytime ads on Sunday = 2
DSA 5.000 0.000 Number of evening ads on Friday = 0
DSU 2.000 0.000 Number of evening ads on Saturday = 0
EFR 0.000 0.000 Number of evening ads on Sunday = 1
ESA 0.000 0.000 Number of game-time ads on Sunday = 2
ESU 1.000 0.000
GSU 2.000 0.000

11 12

3
20/11/2020

INTRODUCTION TO SENSITIVITY ANALYSIS

 In the previous sessions we discussed:


 objective function value
DECISION MODELLING  values of the decision variables
TERM-II  reduced costs
MBA (2020-21)  slack/surplus

SESSION 3-4  In this session we will discuss:


LINEAR PROGRAMMING  changes in the coefficients of the objective function
[LP - SENSITIVITY ANALYSIS  changes in the right-hand side value of a constraint
AND INTERPRETATION OF SOLUTION]
 Shadow Price

Dr. Devendra Kumar Pathak


Assistant Professor,
Operations Management & Decision Sciences, 2
Indian Institute of Management (IIM) Kashipur

INTRODUCTION TO SENSITIVITY ANALYSIS EXAMPLE 1

 Sensitivity analysis (or post-optimality  LP Formulation


analysis) is used to determine how the optimal
solution is affected by changes, within specified
Max 5x1 + 7x2
ranges, in:
 the objective function coefficients s.t. x1 < 6
 the right-hand side (RHS) values in constraints 2x1 + 3x2 < 19
x1 + x2 < 8
 Sensitivity analysis is important to a manager who
must operate in a dynamic environment with x1, x2 > 0
imprecise estimates of the coefficients.

 Sensitivity analysis allows a manager to ask certain


what-if questions about the problem. 3 4

1
20/11/2020

EXAMPLE 1 OBJECTIVE FUNCTION COEFFICIENTS


Binding
 Graphical Solution Constraint  Let us consider how changes in the objective
x2 function coefficients might affect the optimal
x1 + x2 < 8 solution.
8 Max 5x1 + 7x2
 The range of optimality for each coefficient
7
provides the range of values over which the
6 x1 < 6 current solution will remain optimal.
5 Optimal Solution:  Managers should focus on those objective
4 x1 = 5, x2 = 3 coefficients that have a narrow range of optimality
3 and coefficients near the endpoints of the range.
2x1 + 3x2 < 19
2
1
x1
5 6
1 2 3 4 5 6 7 8 9 10

EXAMPLE 1 RANGE OF OPTIMALITY

 Changing Slope of Objective Function  Graphically, the limits of a range of optimality are
x2 found by changing the slope of the objective
Coincides with
8 x1 + x2 < 8 function line within the limits of the slopes of the
constraint line binding constraint lines.
7
6 Objective function

5 5 line for 5x1 + 7x2


4 Coincides with
2x1 + 3x2 < 19
3 Feasible
4 constraint line
2 Region
3
1
1 2
x1
1 2 3 4 5 6 7 8 9 10 7 8

2
20/11/2020

SENSITIVITY ANALYSIS: COMPUTER SOLUTION EXAMPLE 1


Software packages provide the following LP information:

Information about the objective function:


• its optimal value
• coefficient ranges (ranges of optimality)
Information about the decision variables:
Sensitivity Analysis: Computer Solution
• their optimal values
• their reduced costs
Information about the constraints:
• the amount of slack or surplus
• the dual prices
• right-hand side ranges (ranges of feasibility)
9 10

EXAMPLE 1 EXAMPLE 1

Range of Optimality for c1 and c2

Variable Cells
Model Final Reduced Objective Allowable Allowable
Variable Name Value Cost Coefficient Increase Decrease
X1 X1 5.000 0.000 5.000 2.000 0.333
X2 X2 3.000 0.000 7.000 0.500 2.000

Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 1 5.000 0.000 6.000 1E+30 1.000
2 2 19.000 2.000 19.000 5.000 1.000
3 3 8.000 1.000 8.000 0.333 1.667

11 12

3
20/11/2020

RIGHT-HAND SIDES SHADOW PRICE

 Let us consider how a change in the right-hand side  Graphically, a shadow price is determined by
for a constraint might affect the feasible region and adding +1 to the right hand side value in question
perhaps cause a change in the optimal solution. and then resolving for the optimal solution in
terms of the same two binding constraints.
 The improvement in the value of the optimal
solution per unit increase in the right-hand side  The shadow price for a nonbinding constraint
is called the shadow price. is 0.
 The range of feasibility is the range over which the
shadow price is applicable.  A negative shadow price indicates that the objective
function will not improve if the RHS is increased.
 As the RHS increases, other constraints will become
binding and limit the change in the value of the
objective function. 13 14

EXAMPLE 1 EXAMPLE 1

 Shadow Prices  Shadow Prices


Constraint 1: Since x1 < 6 is not a binding Constraint 3: Change the RHS value of the third
constraint, its shadow price is 0. constraint to 9 and resolve for the optimal point
determined by the last two constraints: 2x1 + 3x2 =
Constraint 2: Change the RHS value of the second 19
Constraint to 20 and resolve for the optimal point and x1 + x2 = 9.
determined by the last two constraints: The solution is: x1 = 8, x2 = 1, z = 47.
2x1 + 3x2 = 20 and x1 + x2 = 8. The shadow price is znew - zold = 47 - 46 = 1.
The solution is x1 = 4, x2 = 4, z = 48. Hence, the
shadow price = znew - zold = 48 - 46 = 2.
Max 5x1 + 7x2 Max 5x1 + 7x2
s.t. x1 < 6 s.t. x1 < 6
2x1 + 3x2 < 19 15 2x1 + 3x2 < 19 16
x1 + x2 < 8 x1 + x2 < 8
x1, x2 > 0 x1, x2 > 0

4
20/11/2020

EXAMPLE 1 RANGE OF FEASIBILITY

 Shadow Prices  The range of feasibility for a change in the right


hand side value is the range of values for this
Variable Cells
coefficient in which the original dual price
Model Final Reduced Objective Allowable Allowable
Variable Name Value Cost Coefficient Increase Decrease
remains constant.
X1 X1 5.000 0.000 5.000 2.000 0.333
X2 X2 3.000 0.000 7.000 0.500 2.000

Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 1 5.000 0.000 6.000 1E+30 1.000
2 2 19.000 2.000 19.000 5.000 1.000
3 3 8.000 1.000 8.000 0.333 1.667

17 18

EXAMPLE 1 EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Feasibility Olympic Bike is introducing two new lightweight


bicycle frames, the Deluxe and the Professional, to be
Variable Cells
Model Final Reduced Objective Allowable Allowable made from special aluminum and steel alloys. The
Variable Name Value Cost Coefficient Increase Decrease anticipated unit profits are $10 for the Deluxe and
X1 X1 5.000 0.000 5.000 2.000 0.333
$15 for the Professional.
X2 X2 3.000 0.000 7.000 0.500 2.000

Constraints
The number of pounds of each alloy needed per
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease frame is summarized on the next slide.
1 1 5.000 0.000 6.000 1E+30 1.000
2 2 19.000 2.000 19.000 5.000 1.000
3 3 8.000 1.000 8.000 0.333 1.667

19 20

5
20/11/2020

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

A supplier delivers 100 pounds of the aluminum  Model Formulation


alloy and 80 pounds of the steel alloy weekly.  Verbal Statement of the Objective Function
Maximize total weekly profit.
Aluminum Alloy Steel Alloy  Verbal Statement of the Constraints
Deluxe 2 3
Professional 4 2 Total weekly usage of aluminum alloy < 100
pounds.
How many Deluxe and Professional frames should Total weekly usage of steel alloy < 80 pounds.
Olympic produce each week?  Definition of the Decision Variables
x1 = number of Deluxe frames produced weekly.
x2 = number of Professional frames produced
weekly.

21 22

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Model Formulation (continued)

Max 10x1 + 15x2 (Total Weekly Profit)

s.t. 2x1 + 4x2 <=100 (Aluminum Available)


3x1 + 2x2 <=80 (Steel Available)

x1 >= 0
x2 >=0

23 24

6
20/11/2020

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Optimal Solution

According to the output:

x1 (Deluxe frames) = 15
x2 (Professional frames) = 17.5
Objective function value = $412.50

25 26

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Optimality

Question
Suppose the profit on deluxe frames is
increased to $20. Is the above solution still
optimal? What is the value of the objective function
when this unit profit is increased to $20?

Max 10x1 + 15x2

s.t. 2x1 + 4x2 < 100


3x1 + 2x2 < 80 27 28
x 1, x 2 > 0

7
20/11/2020

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Sensitivity Report  Range of Optimality

Variable Cells
Model Final Reduced Objective Allowable Allowable Answer
Variable Name Value Cost Coefficient Increase Decrease The output states that the solution remains
X1 Deluxe 15.000 0.000 10.000 12.500 2.500
X2 Profes. 17.500 0.000 15.000 5.000 8.333 optimal as long as the objective function coefficient of
x1 is between 7.5 and 22.5. Because 20 is within this
Constraints
range, the optimal solution will not change. The
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease optimal profit will change: 20x1 + 15x2 = 20(15)
1 Alum. 100.000 3.125 100.000 60.000 46.667 + 15(17.5) = $562.50.
2 Steel 80.000 1.250 80.000 70.000 30.000

29 30

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Optimality  Range of Optimality

Variable Cells
Question Model Final Reduced Objective Allowable Allowable
If the unit profit on deluxe frames were $6 Variable Name Value Cost Coefficient Increase Decrease
X1 Deluxe 15.000 0.000 10.000 12.500 2.500
instead of $10, would the optimal solution change? X2 Profes. 17.500 0.000 15.000 5.000 8.333

Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 Alum. 100.000 3.125 100.000 60.000 46.667
2 Steel 80.000 1.250 80.000 70.000 30.000

31 32

8
20/11/2020

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Optimality  Range of Feasibility

Answer Question
The output states that the solution remains If the right-hand side of constraint 2 is
optimal as long as the objective function coefficient increased by 2, what will be the effect on the
of x1 is between 7.5 and 22.5. Because 6 is outside objective function value?
this range, the optimal solution would change.

Max 10x1 + 15x2

s.t. 2x1 + 4x2 < 100


33 3x1 + 2x2 < 80 34
x 1, x 2 > 0

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Feasibility  Range of Feasibility

Variable Cells
Model Final Reduced Objective Allowable Allowable Answer
Variable Name Value Cost Coefficient Increase Decrease A shadow price represents the improvement
X1 Deluxe 15.000 0.000 10.000 12.500 2.500
X2 Profes. 17.500 0.000 15.000 5.000 8.333 in the objective function value per unit increase
in the right-hand side.
Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease Since the RHS remains within the range of feasibility, we
1 Alum. 100.000 3.125 100.000 60.000 46.667
2 Steel 80.000 1.250 80.000 70.000 30.000
can calculate Z value by using SP. The objective function
value increases by $2.50.

35 36

9
20/11/2020

SIMULTANEOUS CHANGES EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Optimality and 100% Rule  Range of Optimality and 100% Rule
The 100% rule states that simultaneous changes in
objective function coefficients will not change the Question
optimal solution as long as the sum of the If simultaneously the profit on Deluxe frames
percentages of the change divided by the was raised to $16 and the profit on Professional
corresponding maximum allowable change in the frames was raised to $17, would the current
range of optimality for each coefficient [i.e., solution be optimal?
∑(Change*100)/(AI/ AD)DV] does not exceed 100%.

Max 10x1 + 15x2

s.t. 2x1 + 4x2 < 100


37 3x1 + 2x2 < 80 38
x 1, x 2 > 0

EXAMPLE 2: OLYMPIC BIKE CO. EXAMPLE 2: OLYMPIC BIKE CO.

 Range of Optimality Range of Optimality and 100% Rule

Variable Cells
Model Final Reduced Objective Allowable Allowable
Answer
Variable Name Value Cost Coefficient Increase Decrease If c1 = 16, the amount c1 changed is 16 - 10 = 6 .
X1 Deluxe 15.000 0.000 10.000 12.500 2.500 The maximum allowable increase is 22.5 - 10 = 12.5,
X2 Profes. 17.500 0.000 15.000 5.000 8.333
so this is a 6/12.5 = 48% change. If c2 = 17, the
Constraints amount that c2 changed is 17 - 15 = 2. The maximum
Constraint Final Shadow Constraint Allowable Allowable allowable increase is 20 - 15 = 5 so this is a 2/5 = 40%
Number Name Value Price R.H. Side Increase Decrease
1 Alum. 100.000 3.125 100.000 60.000 46.667 change. The sum of the change percentages is 88%.
2 Steel 80.000 1.250 80.000 70.000 30.000 Since this does not exceed 100%, the optimal solution
would not change.

39 40

10
20/11/2020

SIMULTANEOUS CHANGES EXAMPLE 3

 Range of Feasibility and 100% Rule  Consider the following linear program:
The 100% rule states that simultaneous changes in
right-hand sides will not change the dual prices as Min 6x1 + 9x2 ($ cost)
long as the sum of the percentages of the changes
divided by the corresponding maximum allowable s.t. x1 + 2x2 < 8
change in the range of feasibility for each right-hand 10x1 + 7.5x2 > 30
side [i.e., ∑(Change*100)/(AI/ AD)RHS] does not x2 > 2
exceed 100%.
x1, x2 > 0

41 42

EXAMPLE 3 EXAMPLE 3

Sensitivity Report  Optimal Solution

Variable Cells According to the output:


Model Final Reduced Objective Allowable Allowable
Variable Name Value Cost Coefficient Increase Decrease
X1 X1 1.500 0.000 6.000 6.000 6.000 x1 = 1.5
X2 X2 2.000 0.000 9.000 1E+30 4.500
x2 = 2.0
Constraints Objective function value = 27.00
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 1 5.500 2.500 8.000 1E+30 2.500
2 2 30.000 -0.600 30.000 25.000 15.000
3 3 2.000 -4.500 2.000 2.000 2.000

43 44

11
20/11/2020

EXAMPLE 3 EXAMPLE 3

 Range of Optimality Sensitivity Report

Variable Cells
Question Model Final Reduced Objective Allowable Allowable
Suppose the unit cost of x1 is decreased to $4. Variable Name Value Cost Coefficient Increase Decrease
X1 X1 1.500 0.000 6.000 6.000 6.000
Is the current solution still optimal? What is the X2 X2 2.000 0.000 9.000 1E+30 4.500
value of the objective function when this unit cost is
decreased to $4? Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 1 5.500 2.500 8.000 1E+30 2.500
2 2 30.000 -0.600 30.000 25.000 15.000
3 3 2.000 -4.500 2.000 2.000 2.000
Min 6x1 + 9x2
s.t. x1 + 2x2 < 8
10x1 + 7.5x2 > 30 45 46
x2 > 2
x1, x2 > 0

EXAMPLE 3 EXAMPLE 3

 Range of Optimality  Range of Optimality

Answer Question
The output states that the solution remains How much can the unit cost of x2 be decreased
optimal as long as the objective function coefficient without concern for the optimal solution changing?
of x1 is between 0 and 12. Because 4 is within
this range, the optimal solution will not change.
However, the optimal total cost will change:
4x1 + 9x2 = 4(1.5) + 9(2.0) = $24.00.

47 48

12
20/11/2020

EXAMPLE 3 EXAMPLE 3

Sensitivity Report  Range of Optimality

Variable Cells
Model Final Reduced Objective Allowable Allowable Answer
Variable Name Value Cost Coefficient Increase Decrease The output states that the solution remains
X1 X1 1.500 0.000 6.000 6.000 6.000
X2 X2 2.000 0.000 9.000 1E+30 4.500 optimal as long as the objective function coefficient
of x2 does not fall below 4.5.
Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 1 5.500 2.500 8.000 1E+30 2.500
2 2 30.000 -0.600 30.000 25.000 15.000
3 3 2.000 -4.500 2.000 2.000 2.000

49 50

EXAMPLE 3 EXAMPLE 3

 Range of Optimality and 100% Rule Sensitivity Report

Variable Cells
Question Model Final Reduced Objective Allowable Allowable
If simultaneously the cost of x1 was raised to $7.5 Variable Name Value Cost Coefficient Increase Decrease
X1 X1 1.500 0.000 6.000 6.000 6.000
and the cost of x2 was reduced to $6, would the X2 X2 2.000 0.000 9.000 1E+30 4.500
current solution remain optimal?
Constraints
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease
1 1 5.500 2.500 8.000 1E+30 2.500
2 2 30.000 -0.600 30.000 25.000 15.000
3 3 2.000 -4.500 2.000 2.000 2.000
Min 6x1 + 9x2
s.t. x1 + 2x2 < 8
10x1 + 7.5x2 > 30 51 52
x2 > 2
x1, x2 > 0

13
20/11/2020

EXAMPLE 3 EXAMPLE 3

 Range of Optimality and 100% Rule  Range of Feasibility

Answer Question
If c1 = 7.5, the amount c1 changed is 7.5 - 6 = If the right-hand side of constraint 3 is
1.5. The maximum allowable increase is 12 - 6 = 6, increased by 1, what will be the effect on the
so this is a 1.5/6 = 25% change. If c2 = 6, the optimal solution?
amount that c2 changed is 9 - 6 = 3. The maximum
allowable decrease is 9 - 4.5 = 4.5, so this is a 3/4.5
= 66.7% change. The sum of the change
percentages is 25% + 66.7% = 91.7%. Because this
does not exceed 100%, the optimal solution would Min 6x1 + 9x2
not change. s.t. x1 + 2x2 < 8
53 10x1 + 7.5x2 > 30 54
x2 > 2
x1, x2 > 0

EXAMPLE 3 EXAMPLE 3

Sensitivity Report  Range of Feasibility

Variable Cells
Model Final Reduced Objective Allowable Allowable Answer
Variable Name Value Cost Coefficient Increase Decrease A shadow price represents the improvement
X1 X1 1.500 0.000 6.000 6.000 6.000
X2 X2 2.000 0.000 9.000 1E+30 4.500 in the objective function value per unit increase
in the right-hand side. A negative shadow price
Constraints
indicates a negative improvement in the objective,
Constraint Final Shadow Constraint Allowable Allowable
Number Name Value Price R.H. Side Increase Decrease which in this problem means an increase in total cost
1 1 5.500 2.500 8.000 1E+30 2.500 because we're minimizing. Since the RHS remains
2 2 30.000 -0.600 30.000 25.000 15.000 within the range of feasibility, we can calculate Z value
3 3 2.000 -4.500 2.000 2.000 2.000
by using SP. The objective function value increases by
$4.50.
55 56

14
20/11/2020

NON-INTUITIVE SHADOW PRICES EXAMPLE 2: OLYMPIC BIKE CO. (REVISED)

Constraints with variables naturally on both the Recall that Olympic Bike is introducing two new
left-hand and right-hand sides often lead to lightweight bicycle frames, the Deluxe and the
shadow prices that have a non-intuitive explanation Professional, to be made from special aluminum and
(appropriate reasoning is required).
steel alloys. The objective is to maximize total profit,
subject to limits on the availability of aluminum and
steel.
Let us now introduce an additional constraint. The
number of Deluxe frames produced (x1) must be greater
than or equal to the number of Professional frames
Max 10x1 + 15x2 produced (x2) .
s.t. 2x1 + 4x2 < 100 x1 > x2
3x1 + 2x2 < 80 57 58
x 1, x 2 > 0

EXAMPLE 2: OLYMPIC BIKE CO. (REVISED) EXAMPLE 2: OLYMPIC BIKE CO. (REVISED)

Model Formulation (continued) Sensitivity Report (Revised)

Variable Cells
Max 10x1 + 15x2 (Total Weekly Profit) Model Final Reduced Objective Allowable Allowable
Variable Name Value Cost Coefficient Increase Decrease
s.t. 2x1 + 4x2 < 100 (Aluminum Available) X1 Deluxe 16.000 0.000 10.000 12.500 10.000
3x1 + 2x2 < 80 (Steel Available) X2 Profes. 16.000 0.000 15.000 1E+30 8.333

x1 - x2 > 0 (Product Ratio) Constraints


Constraint Final Shadow Constraint Allowable Allowable
x1, x2 > 0 Number Name Value Price R.H. Side Increase Decrease
1 Alum. 96.000 0.000 100.000 1E+30 4.000
2 Steel 80.000 5.000 80.000 3.333 80.000
3 Ratio 0.000 -5.000 0.000 26.667 2.500

59 60

15
20/11/2020

EXAMPLE 2: OLYMPIC BIKE CO. (REVISED)

Shadow Price for Constraint #3


The interpretation of the shadow price -5.00 is
correctly stated as follows:

“If we are forced to produce 1 more Deluxe frame


over and above the number of Professional frames
produced, total profits will decrease by $5.00.”

61

16
23/11/2020

PRODUCT MIX
INTRODUCTION
TJ Inc.’s makes three nut mixes for sale to grocery chains located in the
Southeast. The three mixes, referred to as the Regular Mix, the Deluxe
DECISION MODELLING Mix, and the Holiday Mix, are made by mixing different percentages of
TERM-II five types of nuts. In preparation for the fall season, TJ Inc.’s just
purchased the following shipments of nuts at the prices shown:
MBA (2020-21)

CASE 2
LINEAR PROGRAMMING
[PRODUCT MIX]

Dr. Devendra Kumar Pathak The Regular Mix consists of 15% almonds, 25% Brazil nuts, 25% filberts,
Assistant Professor, 10% pecans, and 25% walnuts. The Deluxe Mix consists of 20% of each
Operations Management & Decision Sciences, type of nut, and the Holiday Mix consists of 25% almonds, 15% Brazil nuts,
Indian Institute of Management (IIM) Kashipur 15% filberts, 25% pecans, and 20% walnuts.

PRODUCT MIX PRODUCT MIX


INTRODUCTION REPORT REQUIRED
An accountant at TJ, Inc. analyzed the cost of packaging materials, sales Perform an analysis of the TJ Inc.’s product-mix problem, and prepare a
price per pound, and so forth, and determined that the profit contribution report for the president of TJ, Inc. that summarizes your findings.
per pound is $1.65 for the Regular Mix, $2.00 for the Deluxe Mix, and
$2.25 for the Holiday Mix. These figures do not include the cost of specific 1. The cost per pound of the nuts included in the Regular, Deluxe,
types of nuts in the different mixes because that cost can vary greatly in the and Holiday mixes
commodity markets. Customer orders already received are summarized
here: 2. The optimal product mix and the total profit contribution

3. Recommendations regarding how the total profit contribution can be


increased if additional quantities of nuts can be purchased

4. A recommendation as to whether TJ, Inc. should purchase an additional


1000 pounds of almonds for $1000 from a supplier who overbought
Because demand is running high, it is expected that TJ, Inc. will receive
many more orders than can be satisfied. TJ, Inc. is committed to using the
5. Recommendations on how profit contribution could be increased (if at all)
available nuts to maximize profit over the fall season; nuts not used will be
if TJ, Inc. does not satisfy all existing orders
given to a local charity. Even if it is not profitable to do so, TJ, Inc.’s president
3 4
indicated that the orders already received must be satisfied.

1
23/11/2020

PRODUCT MIX
SOLUTION 1

The difference between relevant and sunk costs is critical. The cost of the shipment
of nuts is a sunk cost.

PRODUCT MIX Cost per pound of ingredients

Almonds $7500/6000 = $1.25


Brazil $7125/7500 = $.95
Filberts $6750/7500 = $.90
Pecans $7200/6000 = $1.20
Walnuts $7875/7500 = $1.05

Cost of nuts in three mixes:


Regular mix: .15($1.25) + .25($.95) + .25($90) + .10($1.20) + .25($1.05) = $1.0325
Deluxe mix .20($1.25) + .20($.95) + .20($.90) + .20($1.20) + .20($1.05) = $1.07
Holiday mix: .25($1.25) + .15($.95) + .15($.90) + .25($1.20) + .20($1.05) = $1.10
5 6

PRODUCT MIX PRODUCT MIX


SOLUTION 2 SOLUTION 2
Let R = pounds of Regular Mix produced
D = pounds of Deluxe Mix produced
H = pounds of Holiday Mix produced
Note that the cost of the five shipments of nuts is a sunk (not a relevant) cost and should
not affect the decision.
The following linear programming model can be solved to maximize profit contribution
for the nuts already purchased.

Max 1.65R + 2.00D + 2.25H


s.t.
0.15R + 0.20D + 0.25H  6000 Almonds
0.25R + 0.20D + 0.15H  7500 Brazil
0.25R + 0.20D + 0.15H  7500 Filberts
0.10R + 0.20D + 0.25H  6000 Pecans
0.25R + 0.20D + 0.20H  7500 Walnuts
R  10000 Regular
D  3000 Deluxe
7 8
H  5000 Holiday
R, D, H  0

2
23/11/2020

PRODUCT MIX PRODUCT MIX


SOLUTION 2 SOLUTION 3

Q3) Recommendations regarding how the total profit


contribution can be increased if additional quantities of
nuts can be purchased

9 10

PRODUCT MIX PRODUCT MIX


SOLUTION 3 SOLUTION 3

11 12

3
23/11/2020

PRODUCT MIX PRODUCT MIX


SOLUTION 3 SOLUTION 4

From the dual values it can be seen that additional almonds are Q4) A recommendation as to whether TJ, Inc. should purchase
worth $8.50 per pound to TJ. an additional 1000 pounds of almonds for $1000 from a
supplier who overbought.
Additional walnuts are worth $1.50 per pound.

From the slack variables, we see that additional Brazil nut,


Filberts, and Pecans are of no value since they are already in
excess supply.

13 14

PRODUCT MIX PRODUCT MIX


SOLUTION 4 SOLUTION 4

Yes, purchase the almonds.


The dual value shows that each pound is worth $8.50; the
dual value is applicable for increases up to 583.33
pounds.

Resolving the problem by changing the right-hand side of


constraint 1 from 6000 to 7000 yields the following optimal
solution.

The optimal solution has increased in value by $4958.34.


Note that only 583.33 pounds of the additional almonds
were used, but that the increase in profit contribution more
15 than justifies the $1000 cost of the shipment. 16

4
23/11/2020

PRODUCT MIX PRODUCT MIX


SOLUTION 4 SOLUTION 4

17 18

PRODUCT MIX PRODUCT MIX


SOLUTION 5 SOLUTION 5

Q5) Recommendations on how profit contribution


could be increased (if at all) if TJ, Inc. does not
satisfy all existing orders

7000

19 20

5
23/11/2020

PRODUCT MIX
SOLUTION 5

From the dual values it is clear that there is no advantage


to not satisfying the orders for the Regular and Deluxe
mixes.

However, it would be advantageous to negotiate a decrease


in the Holiday mix requirement.

21

6
23/11/2020

DECISION MODELLING
TERM-II
MBA (2020-21)

CASE 1
LINEAR PROGRAMMING
[WORKLOAD BALANCING]

Dr. Devendra Kumar Pathak


Assistant Professor,
Operations Management & Decision Sciences, 2
Indian Institute of Management (IIM) Kashipur

WORKLOAD BALANCING WORKLOAD BALANCING


INTRODUCTION QUESTIONS
Digital Imaging (DI) produces color printers for both the Perform an analysis for Digital Imaging in order to determine how many
professional and consumer markets. The DI consumer division units of each printer to produce. Prepare a report to DI’s president
recently introduced two new color printers. The DI-910 model can presenting your findings and recommendations. Include (but do not limit
produce a 4” 3 6” borderless color print in approximately 37 seconds. your discussion to) a consideration of the following:
The more sophisticated and faster DI-950 can even produce a 13” 3
19” borderless color print. Financial projections show profit 1. The recommended number of units of each printer to produce to
maximize the total contribution to profit for an 8-hour shift.
contributions of $42 for each DI-910 and $87 for each DI-950.
What reasons might management have for not implementing
your recommendation?
The printers are assembled, tested, and packaged at DI’s plant located
in New Bern, North Carolina. This plant is highly automated and 2. Suppose that management also states that the number of DI-910
uses two manufacturing lines to produce the printers. Line 1 printers produced must be at least as great as the number of DI-950
performs the assembly operation with times of 3 minutes per DI-910 units produced. Assuming that the objective is to maximize the total
printer and 6 minutes per DI-950 printer. Line 2 performs both contribution to profit for an 8-hour shift, how many units of each printer
the testing and packaging operations. Times are 4 minutes per DI- should be produced?
910 printer and 2 minutes per DI-950 printer. The shorter time
for the DI-950 printer is a result of its faster print speed. Both 4
manufacturing lines are in operation one 8-hour shift per day.
3

1
23/11/2020

WORKLOAD BALANCING
QUESTIONS
3. Does the solution you developed in part (2) balance the total time spent
on line 1 and the total time spent on line 2? Why might this balance or lack
of it be a concern to management?

4. Management requested an expansion of the model in part (2) that would


provide a better balance between the total time on line 1 and the total time
on line 2. Management wants to limit the difference between the total time
on line 1 and the total time on line 2 to 30 minutes or less. If the objective is
WORKLOAD BALANCING
still to maximize the total contribution to profit, how many units of each
printer should be produced? What effect does this workload balancing have
on total profit in part (2)?

5. Suppose that in part (1) management specified the objective of


maximizing the total number of printers produced each shift rather than
total profit contribution. With this objective, how many units of each printer
should be produced per shift? What effect does this objective have on total
profit and workload balancing? For each solution that you develop, include
a copy of your linear programming model and graphical solution in the 5 6

appendix to your report.

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 1 SOLUTION 1
Production Rate
(minutes per printer)
Model Line 1 Line 2 Profit Contribution ($)
DI-910 3 4 42
DI-950 6 2 87
Capacity: 8 hours*60 minutes/hour = 480 minutes per day

Let D1 = number of units of the DI-910 produced


D2 = number of units of the DI-950 produced
Max 42D1 + 87D2
s.t.
3D1 + 6D2  480 Line 1 Capacity
4D1 + 2D2  480 Line 2 Capacity
D1 , D2  0

7 8

2
23/11/2020

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 1 SOLUTION 2

Q2) Suppose that management also states that the


The optimal solution is D1 = 0, D2 = 80. The value of the optimal solution is
$6960.
number of DI-910 printers produced must be at least as
great as the number of DI-950 units produced.
Assuming that the objective is to maximize the total
Q1) The recommended number of units of each printer to produce
to maximize the total contribution to profit for an 8-hour shift. contribution to profit for an 8-hour shift, how many
What reasons might management have for not implementing your units of each printer should be produced?
recommendation?

Management would not implement this


solution because no units of the DI-910 would
be produced.
9 10

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 2 SOLUTION 2

Adding the constraint D1  D2 and resolving the linear


program results in the optimal solution D1 = 53.333, D2
= 53.333. The value of the optimal solution is $6880.

11 12

3
23/11/2020

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 3 SOLUTION 3

Q3) Does the solution you developed in part (2) balance the
total time spent on line 1 and the total time spent on line 2?
Why might this balance or lack of it be a concern to
management?

13 14

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 3 SOLUTION 4

Time spent on Line 1: 3(53.333) + 6(53.333) = 480 minutes Q4) Management requested an expansion of the model in part
(2) that would provide a better balance between the total time
Time spent on Line 2: 4(53.333) + 2(53.333) = 320 minutes on line 1 and the total time on line 2. Management wants to
limit the difference between the total time on line 1 and
Thus, the solution does not balance the total time spent on Line 1 the total time on line 2 to 30 minutes or less. If the
and the total time spent on Line 2. This might be a concern to objective is still to maximize the total contribution to profit,
management if no other work assignments were available for the how many units of each printer should be produced? What
employees on Line 2. effect does this workload balancing have on total profit in part
(2)?

15 16

4
23/11/2020

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 4 SOLUTION 4
Let T1 = total time spent on Line 1
T2 = total time spent on Line 2
Whatever the value of T2 is,
T1 - T2 <= 30
with T1 = 3D1 + 6D2 and T2 = 4D1 + 2D2

Adding this constraint to the linear program formulated in part


(2) and resolving we obtain the optimal solution D1 = 96.667, D2 =
31.667. The value of the optimal solution is $6815.

Line 1 is scheduled for 480 minutes and Line 2 for 450 minutes.

The effect of workload balancing is to reduce the total contribution to


17 18
profit by $6880 - $6815 = $65 per shift.

WORKLOAD BALANCING WORKLOAD BALANCING


SOLUTION 5 SOLUTION 5

Q5) Suppose that in part (1) management specified the


objective of maximizing the total number of printers
produced each shift rather than total profit contribution. With
this objective, how many units of each printer should be
produced per shift? What effect does this objective have on
total profit and workload balancing?

19 20

5
23/11/2020

WORKLOAD BALANCING
SOLUTION 5

The optimal solution is D1 = 106.667, D2 = 26.667. The total


profit contribution is

42(106.667) + 87(26.667) = $6800

Comparing the solutions to part (4) and part (5), maximizing


the number of printers produced (106.667 + 26.667 = 133.33)
has increased the production by 133.33 - (96.667 + 31.667) =
5 printers but has reduced profit contribution by $6815 -
$6800 = $15.

But, this solution results in perfect workload balancing


because the total time spent on each line is 480 minutes. 21

6
11/25/2020

FINANCE APPLICATIONS
Portfolio Selection Problem

 International City Trust investment decisions


Decision Modelling

Term-II ICT has $5 million available & intends to maximize the interest earned

MBA (2020-21)
INVESTMENT INTEREST EARNED RISK SCORE
Trade credits 7% 1.7
Corporate bonds 10% 1.2
SESSION 6 Gold stocks 19% 3.7
LINEAR PROGRAMMING APPLICATIONS Platinum stocks 12% 2.4
(MODEL FORMULATION) Mortgage securities 8% 2.0
Construction loans 14% 2.9

Dr. Devendra Kumar Pathak


(Ph.D., IIT Delhi)
Assistant Professor,
Operations Management & Decision Sciences,
Indian Institute of Management (IIM) Kashipur

FINANCE APPLICATIONS FINANCE APPLICATIONS


Portfolio Selection Problem Portfolio Selection Problem

 Decision variables
Conditions:
 No more than 25% of total investment may be in any single type of
investment T = dollars invested in trade credit
 At least 30% of total investment must be in precious metals B = dollars invested in corporate bonds
 At least 45% of total investment must be in trade credits and G = dollars invested in gold stocks
corporate bond P = dollars invested in platinum stocks
 Average risk score cannot be more than 2 M = dollars invested in mortgage securities
C = dollars invested in construction loans

1
11/25/2020

FINANCE APPLICATIONS EMPLOYEE STAFFING


Portfolio Selection Problem Worker Planning Problem

 ABC Bank staffing requirements


Objective function
Maximize dollars of interest earned = 0.07T + 0.10B + 0.19G
+ 0.12P + 0.08M + 0.14C
subject to TIME PERIOD NUMBER REQUIRED
T + B + G + P + M + C ≤ 5,000,000 9 a.m.–10 a.m. 10 (Min.)
T ≤ 0.25(T + B + G + P + M + C)
B ≤ 0.25(T + B + G + P + M + C) 10 a.m.–11 a.m. 12
G ≤ 0.25(T + B + G + P + M + C) 11 a.m.–Noon 14
P ≤ 0.25(T + B + G + P + M + C)
M ≤ 0.25(T + B + G + P + M + C) Noon–1 p.m. 16
C ≤ 0.25(T + B + G + P + M + C) 1 p.m.–2 p.m. 18
G+P ≥ 0.30(T + B + G + P + M + C)
T+B ≥ 0.45(T + B + G + P + M + C) 2 p.m.–3 p.m. 17 (Max.)
1.7T + 1.2B + 3.7G + 2.4P + 2.0M + 2.9C 3 p.m.–4 p.m. 15
≤2
T+B+G+P+M+C
4 p.m.–5 p.m. 10
T, B, G, P, M, C ≥0

EMPLOYEE STAFFING EMPLOYEE STAFFING


Worker Planning Problem Worker Planning Problem

 Decision variables
Conditions & relevant data:
 Bank now employs 12 full-time tellers, but it has several people available
on its roster of part-time employees
 A full-timer works from 9 am to 5 pm with 1 hour lunch break. Half of F = number of full-time tellers (all starting at 9 a.m.)
full-timer takes the lunch between 11 am to noon and other half from P1 = number of part-timers, starting at 9 a.m. (leaving at 1 p.m.)
noon to 1 pm. P2 = number of part-timers, starting at 10 a.m. (leaving at 2 p.m.)
 A part-timer must work exactly for 4 hours a day, but can start anytime P3 = number of part-timers, starting at 11 a.m. (leaving at 3 p.m.)
between 9 am and 1 pm.
P4 = number of part-timers, starting at noon (leaving at 4 p.m.)
 The bank’s corporate policy limits part-time hours to a maximum of 50%
P5 = number of part-timers, starting at 1 p.m. (leaving at 5 p.m.)
of the day’s total staff requirements
 Part-timers earn $7 per hour ($28 per day) and full-timers earn $90 per
day in salary and other benefits on average

2
11/25/2020

EMPLOYEE STAFFING MULTI-PERIOD APPLICATIONS


Worker Planning Problem Production Scheduling Problem

Objective Function
 Greenberg Motors order schedule
Minimize total daily personnel cost = $90F + $28(P1 + P2 + P3 + P4 + P5)
subject to
F + P1 ≥ 10 (9 a.m. – 10 a.m. requirement)
F + P1 + P2 ≥ 12 (10 a.m. – 11 a.m. requirement) MODEL JANUARY FEBRUARY MARCH APRIL
0.5F + P1 + P2 + P3 ≥ 14 (11 a.m. – 12 noon requirement)
0.5F + P1 + P2 + P3 + P4 ≥ 16 (12 noon – 1 p.m. requirement) GM3A 800 700 1,000 1,100
F + P2 + P3 + P4 + P5 ≥ 18 (1 p.m. – 2 p.m. requirement) GM3B 1,000 1,200 1,400 1,400
F + P3 + P4 + P5 ≤ 17 (2 p.m. – 3 p.m. requirement)
F + P4 + P5 ≤ 15 (3 p.m. – 4 p.m. requirement)
F + P5 ≤ 10 (4 p.m. – 5 p.m. requirement)
F ≤ 12 (full time tellers available)
4P1 + 4P2 + 4P3 + 4P4 + 4P5 ≤ 56 (part time worker hours not exceeding
50% of total requirements)
F, P1, P2, P3, P4, P5 ≥ 0 (non-negativity)

MULTIPERIOD APPLICATIONS MULTIPERIOD APPLICATIONS


Production Scheduling Problem Production Scheduling Problem

 Greenberg Motors data Decision variables


PAt = number of model GM3A motors produced in month t
(t = 1, 2, 3, 4 for January–April)
UNIT COST UNIT COST HOLDING ENDING LABOUR PBt = number of model GM3B motors produced in month t
MODEL (Jan – Feb) (Mar - Apr) COST INVENTORY HOURS IAt = Level of on-hand inventory for GM3A at the end of
GM3A $10 $11 $0.18 450 1.3 month t
IBt = Level of on-hand inventory for GM3B at the end of
GM3B $6 $6.60 $0.13 300 0.9
month t
Objective function
Minimize total costs = cost of production + cost of carrying inventory
Maximum inventory = 3,300 units = 10PA1 + 10PA2 + 11PA3 + 11PA4 + 6PB1
No layoff policy + 6PB2 + 6.60PB3 + 6.60PB4 + 0.18IA1
2,240 labour regular hours available each month + 0.18IA2 + 0.18IA3 + 0.18IA4 + 0.13IB1
320 maximum additional labour hours available each month + 0.13IB2 + 0.13IB3 + 0.13IB4

3
11/25/2020

MULTIPERIOD APPLICATIONS MULTIPERIOD APPLICATIONS


Production Scheduling Problem Production Scheduling Problem

Balance equation subject to


0 + PA1 – 800 = IA1 (GM3A motors in Jan)
0 + PB1 – 1,000 = IB1 (GM3B motors in Jan)
IA1 + PA2 – 700 = IA2 (GM3A motors in Feb)
IB1 + PB2 – 1,200 = IB2 (GM3B motors in Feb)
IA2 + PA3 – 1,000 = IA3 (GM3A motors in Mar)
IB2 + PB3 – 1,400 = IB3 (GM3B motors in Mar)
IA3 + PA4 – 1,100 = IA4 (GM3A motors in Apr)
IB3 + PB4 – 1,400 = IB4 (GM3B motors in Apr)
IA4 = 450 (ending inventory GM3A)
IB4 = 300 (ending inventory GM3B)
IA1 + IB1 ≤ 3,300 (max inventory Jan)
IA2 + IB2 ≤ 3,300 (max inventory Feb)
IA3 + IB3 ≤ 3,300 (max inventory Mar)
IA4 + IB4 ≤ 3,300 (max inventory Apr)

MULTIPERIOD APPLICATIONS OM APPLICATIONS


Production Scheduling Problem Make or Buy Decision

subject to Janders is preparing to introduce two new calculators: one for


the business market called the Financial Manager and one for
1.3PA1 + 0.9PB1 ≥ 2,240 (Jan labour minimum) the engineering market called the Technician.
1.3PA1 + 0.9PB1 ≤ 2,560 (Jan labour maximum)
1.3PA2 + 0.9PB2 ≥ 2,240 (Feb labour minimum)
Each calculator has three components: a base, an electronic
1.3PA2 + 0.9PB2 ≤ 2,560 (Feb labour maximum)
1.3PA3 + 0.9PB3 ≥ 2,240 (Mar labour minimum)
cartridge, and a faceplate or top. The same base is used for
1.3PA3 + 0.9PB3 ≤ 2,560 (Mar labour maximum) both calculators, but the cartridges and tops are different.
1.3PA4 + 0.9PB4 ≥ 2,240 (Apr labour minimum)
1.3PA4 + 0.9PB4 ≤ 2,560 (Apr labour maximum) All components can be manufactured by the company or
All variables ≥ 0 (nonnegativity)
purchased from outside suppliers. The manufacturing costs and
purchase prices for the components are summarized in Table.

4
11/25/2020

OM APPLICATIONS OM APPLICATIONS
Make or Buy Decision Make or Buy Decision

Company forecasters indicate that 3000 Financial Manager


calculators and 2000 Technician calculators will be needed.

However, manufacturing capacity is limited. The company


has 200 hours of regular manufacturing time and 50 hours
of overtime that can be scheduled for the calculators.

Overtime involves a premium at the additional cost of $9 per


Minutes hour.

The problem for Janders is to determine how many


units of each component to manufacture and how
many units of each component to purchase.

OM APPLICATIONS OM APPLICATIONS
Make or Buy Decision Make or Buy Decision

5
11/25/2020

OM APPLICATIONS OM APPLICATIONS
Make or Buy Decision Make or Buy Decision

MARKETING APPLICATIONS MARKETING APPLICATIONS


Marketing Research Marketing Research
MSI agreed to conduct door-to-door personal interviews to obtain
The marketing research firm’s objective is to conduct the responses from households with children and households without
survey so as to meet the client’s needs at a minimum cost. children. In addition, MSI agreed to conduct both day and evening
interviews. Specifically, the client’s contract called for MSI to conduct
1000 interviews under the following quota guidelines.
Market Survey Inc. (MSI), specializes in evaluating consumer
1. Interview at least 400 households with children.
reaction to new products, services, and advertising campaigns. 2. Interview at least 400 households without children.
3. The total number of households interviewed during the evening must
A client firm requested MSI’s assistance in ascertaining be at least as great as the number of households interviewed during the
consumer reaction to a recently marketed household day.
product. 4. At least 40% of the interviews for households with children must be
conducted during the evening.
5. At least 60% of the interviews for households without children must be
conducted during the evening.

6
11/25/2020

MARKETING APPLICATIONS MARKETING APPLICATIONS


Marketing Research Marketing Research

What is the household, time-of-day interview plan


that will satisfy the contract requirements at a
minimum total interviewing cost?

MARKETING APPLICATIONS
Marketing Research

7
12/2/2020

REQUIRED DATA

Dr. Devendra Kumar Pathak


(Ph.D., IIT Delhi)
Assistant Professor,
Operations Management & Decision Sciences,
Indian Institute of Management (IIM) Kashipur
2

REQUIRED DATA REQUIRED DATA

Selling Price $39000 $38000


3 4

1
12/2/2020

ANALYSIS OBJECTIVE FUNCTION

 Unit contribution for Model 101 trucks


= selling price – variable cost
Problem Formulation—Analytical
= selling price – (direct material+ direct labor+ variable overhead )
=$39,000-$(24,000+4,000+8,000) = $3,000
Decision variables
 Similarly, unit contribution for Model 102 trucks
=$38,000-$(20,000+4,500+8,500) =$5,000
Let N101 and N102 be the monthly production rates of Model 101 trucks and
Model 102 trucks.  Merton,s objective function is:
Max (3,000*N101)+(5,000*N102)

CONSTRAINTS SOLUTION
ONE MODEL CASE
 For engine assembly we require  Suppose Merton wants to produce Model 101 or Model 102, but not both.
 1.0*N101+2.0*N102≤4,000  Which model should it manufacturer?
For metal stamping we require  Model 102 has a higher unit contribution ($5,000 vs. $3,000)
 2.0*N101+2.0*N102≤ 6,000
Model 101 assembly,
 But there are constraints. These restrict Model 101 production to Min [4000/1, 6000/2, 5000/2]
 2.0*N101 ≤5,000 = 2500 and 102 production to Min [4000/2, 6000/2, 4500/3] = 1500.
Model 102 assembly,
 3.0*N102 ≤ 4,500  In our case, the unit contributions for model 101 and Model 102 are $3000 and $5000,
Non-negativity constraint respectively.
 N101 ≥ 0; N102 ≥ 0
 If Merton were to produce only Model 101 trucks, the maximum contribution would also be
$7.5 million. Since the maximum contribution is the same in either case, the company should
be indifferent to producing Model 101 trucks of Model 101 trucks
 At this level of production, Model 101 assembly capacity is fully used up, but there is excess
engine assembly, metal stamping, and Model 102 assembly (1500, 1000, and 4500
machines hours).

2
12/2/2020

1 TWO MODEL CASE 1 TWO MODEL CASE

11000000

9 10

1 TWO MODEL CASE 1 TWO MODEL CASE

11 12

3
12/2/2020

1 TWO MODEL CASE 1 TWO MODEL CASE

13 14

1 TWO MODEL CASE 1 TWO MODEL CASE

15 16

4
12/2/2020

2 SOURCING OUT ENGINE ASSEMBLY 2 SOURCING OUT ENGINE ASSEMBLY

11000000

17 18

2 SOURCING OUT ENGINE ASSEMBLY 3 MODEL 103?

19 20

5
12/2/2020

3 MODEL 103? 3 MODEL 103 CONTRIBUTION INCREASED BY 350

N103 >= 0

N103 >= 0

21 22

3 MODEL 103 CONTRIBUTION INCREASED BY 350 4. REQUIRED DATA

N103 >= 0

Selling Price $39000 $38000


23 24

6
12/2/2020

4. REQUIRED DATA 4 ENGINE ASSEMBLY OVERTIME PRODUCTION?

25 26

4 ENGINE ASSEMBLY OVERTIME PRODUCTION? 5 MARKETING CONSTRAINT?

27 28

7
12/2/2020

5 MARKETING CONSTRAINT?

29

8
04/12/2020

THE ESSENCE OF DUALITY THEORY


 Every linear programming problem has an
associated problem called the dual
DECISION MODELLING
TERM-II  Original problem is known as the primal
MBA (2020-21)

 Consider a maximization primal problem in


SESSION 8 canonical form
LINEAR PROGRAMMING
[DUALITY IN LINEAR PROGRAMMING]
 Dual is a minimization problem
 Dual uses same parameters but at different
Dr. Devendra Kumar Pathak locations
Assistant Professor,
Operations Management & Decision Sciences,
Indian Institute of Management (IIM) Kashipur
2

THE ESSENCE OF DUALITY THEORY CANONICAL FORM

 A maximization linear program is said to be in


canonical form if all constraints are "less than or
equal to" constraints and the variables are non-
negative.

 A minimization linear program is said to be in


canonical form if all constraints are "greater than
or equal to" constraints and the variables are non-
negative.

1
04/12/2020

CANONICAL FORM PRIMAL AND DUAL PROBLEMS

 Convert any linear program to a maximization  The optimal value of the objective function of the
problem in canonical form as follows: primal problem equals the optimal value of the
 minimization objective function: objective function of the dual problem.
multiply it by -1
 "less than or equal to" constraint:
leave it alone  Solving the dual might be computationally more
 "greater than or equal to" constraint: efficient when the primal has numerous
multiply it by -1 constraints and few variables.
 "equal to" constraint:
form two constraints, one "less than or equal to",
the other "greater or equal to"; then multiply this
"greater than or equal to" constraint by -1.

The dual of a maximization problem 5 6


in canonical form is a minimization
problem in canonical form.

THE ESSENCE OF DUALITY THEORY THE ESSENCE OF DUALITY THEORY

2
04/12/2020

THE ESSENCE OF DUALITY THEORY EXAMPLE: PRIMAL

 Coefficients in the objective function of the primal  LP Formulation


problem:
 Are right-hand sides of the functional constraints in the Max 2x1 + x2 + 3x3
dual problem
 Right-hand sides of the functional constraints in the s.t. x1 + 2x2 + 3x3 < 15
primal problem: 3x1 + 4x2 + 6x3 > 24
 Are the coefficients in the objective function of the dual x1 + x2 + x3 = 10
problem
 Coefficients of a variable in the functional constraints x1, x2, x3 > 0
of the primal problem:
 Are the coefficients in a functional constraint of the dual
problem

10

EXAMPLE: PRIMAL EXAMPLE: PRIMAL

 Primal in Canonical Form  Primal in Canonical Form (continued)


 Constraint (1) is a "<" constraint. Leave it alone.
 Constraint (2) is a ">" constraint. Multiply it by -1. Max 2x1 + x2 + 3x3
 Constraint (3) is an "=" constraint. Rewrite this as two
constraints, one a "<", the other a ">" constraint. Then s.t. x1 + 2x2 + 3x3 < 15
multiply the ">" constraint by -1. -3x1 - 4x2 - 6x3 < -24
(result on next slide) x1 + x2 + x3 < 10
-x1 - x2 - x3 < -10
x1, x2, x3 > 0

11 12

3
04/12/2020

EXAMPLE: DUAL EXAMPLE: DUAL

 Dual of the Canonical Primal  Dual of the Canonical Primal (continued)


 There are four dual variables, U1, U2, U3', U3".
 The objective function coefficients of the dual are the RHS Min 15U1 - 24U2 + 10U3' - 10U3"
of the primal.
 The RHS of the dual is the objective function coefficients of s.t. U1 - 3U2 + U3' - U3" > 2
the primal. 2U1 - 4U2 + U3' - U3" > 1
 The rows of the dual are the columns of the primal. 3U1 - 6U2 + U3' - U3" > 3
(result on next slide)
U1, U2, U3', U3" > 0

13 14

EXAMPLE: DUAL ECONOMIC INTERPRETATION OF THE DUAL VARIABLES


General statements about the dual of a maximization problem in
The objective function values for the primal and dual problems must be
canonical form. equal.
• The dual is a minimization problem in canonical form.

• When the primal has n decision variables, the dual will have n HighTech Industries problem (Display Monitors). The original
constraints. formulation—the primal problem—is as follows:

• When the primal has m constraints, the dual will have m decision
variables.

• The right-hand sides of the primal constraints become the objective function
coefficients in the dual.

• The objective function coefficients of the primal become the right-hand sides of
the dual constraints.

• The constraint coefficients of the ith primal variable become the coefficients in
the ith constraint of the dual. 15 16

4
04/12/2020

ECONOMIC INTERPRETATION OF THE DUAL VARIABLES ECONOMIC INTERPRETATION OF THE DUAL VARIABLES

The HighTech dual problem is as follows:

Each dual variable is associated with one of the constraints in the


primal.

Specifically, U1 is associated with the assembly time constraint, U2


17 with the Portable display constraint, and U3 with the warehouse space 18
constraint.

ECONOMIC INTERPRETATION OF THE DUAL VARIABLES ECONOMIC INTERPRETATION OF THE DUAL VARIABLES
The objective function values for the primal and dual problems must be equal.

Dual variables must carry the interpretations of being the value per unit of
resource.

19 20

5
04/12/2020

ECONOMIC INTERPRETATION OF THE DUAL VARIABLES EXAMPLE: WYNDOR GLASS CO.


For a minimization problem, the dual prices are the same in absolute value
but have opposite signs.
 Wyndor Glass Co.
 Produces windows and glass doors
Following interpretation of the primal and dual problems can be made when the
 Plant 1 makes aluminum frames and hardware
primal is a product-mix problem.
 Plant 2 makes wood frames
Primal Problem: Given a per-unit value of each product, determine how much  Plant 3 produces glass and assembles products
of each should be produced to maximize the value of the total production.
Constraints require the amount of each resource used to be less than or equal  Company introducing two new products
to the amount available.  Product 1: 8 ft. glass door with aluminum frame
Dual Problem: Given the availability of each resource, determine the per-unit  Product 2: 4 x 6 ft. double-hung, wood-framed window
value such that the total value of the resources used is minimized. Constraints  Problem: What mix of products would be most
require the resource value per unit be greater than or equal to the value of each
unit of output.
profitable?
 Assuming company could sell as much of either product
Dual Variables: The variable in a dual linear as could be produced
programming problem. Its optimal value provides the 21 22
dual price for the associated primal resource.

EXAMPLE: WYNDOR GLASS CO. EXAMPLE: WYNDOR GLASS CO.


 Products produced in batches of 20  Formulating the model
 Data needed x1 = number of batches of product 1 produced per week
 Number of hours of production time available per week in each x2 = number of batches of product 2 produced per week
plant for new products Z = total profit per week (thousands of dollars) from producing these
two products
 Production time used in each plant for each batch of each new
product  From bottom row of Table
 Profit per batch of each new product 𝑍 = 3 𝑥1 + 5𝑥2
 Constraints
𝑥1 ≤ 4
2𝑥2 ≤ 12
3𝑥1 + 2𝑥2 ≤ 18
𝑥1 ≥ 0
𝑥2 ≥ 0
 Classic example of resource-allocation problem
 Most common type of linear programming problem

23 24

6
04/12/2020

EXAMPLE: WYNDOR GLASS CO. EXAMPLE: WYNDOR GLASS CO. PRIMAL

25 26

EXAMPLE: WYNDOR GLASS CO. PRIMAL EXAMPLE: WYNDOR GLASS CO. DUAL

27 28

7
04/12/2020

EXAMPLE: WYNDOR GLASS CO. DUAL PRIMAL AND DUAL VARIABLES

 The dual variables are the "value per unit" of the


corresponding primal resource, i.e. the shadow prices.

 If the dual is solved, OFV will remain the same

29 30

THE ESSENCE OF DUALITY THEORY THE ESSENCE OF DUALITY THEORY

 Weak duality property


 Summary of primary-dual relationships
 If x is a feasible solution for the primal problem and y is
 Weak duality property a feasible solution for the dual problem, then cx ≤ yb.
 Strong duality property
 Complementary solutions property
 Complementary optimal solutions property
 Symmetry property

• Strong duality property

If x* is an optimal solution for the primal problem and


y* is an optimal solution for the dual problem, then cx*
31 = y*b. 32

8
04/12/2020

THE ESSENCE OF DUALITY THEORY THE ESSENCE OF DUALITY THEORY

 Complementary solutions property (cont’d.)


 If x is not optimal for the primal problem, then y is not
 Complementary solutions property
feasible for the dual problem
 At each iteration, the simplex method simultaneously
identifies a Corner - Point Feasible (CPF) solution x for
the primal problem and a complementary solution y for  Complementary optimal solutions property
the dual problem  The simplex method identifies (at its final iteration) an
 Where cx = yb optimal solution x* for the primal problem and a
complementary optimal solution y* for the dual problem
 Where cx* = y*b

33 34

THE ESSENCE OF DUALITY THEORY THE ESSENCE OF DUALITY THEORY

 Symmetry property
 For any primal problem and its dual problem
 Duality theorem (cont’d.)
 All relationships between them must be symmetric
 If one problem has feasible solutions and an unbounded
objective function, then the other problem has no
feasible solutions
 Duality theorem  If one problem has no feasible solutions, then the other
 Identifies the only possible relationships between the problem either has no feasible solutions or an
primal and dual problems unbounded objective function
 If one problem has feasible solutions and a bounded
objective function, then so does the other problem
 Both weak and strong duality properties apply

35 36

9
04/12/2020

EXAMPLE: PRIMAL EXAMPLE: PRIMAL

 Primal in Canonical Form


Max 2x1 + x2 + 3x3

s.t. x1 + 2x2 + 3x3 < 15


-3x1 - 4x2 - 6x3 < -24
x1 + x2 + x3 < 10
-x1 - x2 - x3 < -10

x1, x2, x3 > 0

37 38

EXAMPLE: DUAL EXAMPLE: PRIMAL

 Dual of the Canonical Primal (continued)

Min 15U1 - 24U2 + 10U3' - 10U3"

s.t. U1 - 3U2 + U3' - U3" > 2


2U1 - 4U2 + U3' - U3" > 1
3U1 - 6U2 + U3' - U3" > 3

U1, U2, U3', U3" > 0

39 40

10
07/12/2020

OUTLINE

 Transportation Modeling
DECISION MODELLING
TERM-II  Developing an Initial Solution
MBA (2020-21)  The Northwest-Corner Rule
 The Lowest-Cost Method
SESSION 9  Vogel's Approximation
TRANSPORTATION PROBLEM
 Special Issues in Modeling
 Demand Not Equal to Supply
Dr. Devendra Kumar Pathak
Assistant Professor,  Degeneracy
Operations Management & Decision Sciences, 2
Indian Institute of Management (IIM) Kashipur

DISTRIBUTION AND NETWORK MODELS NETWORK PROBLEMS

 Supply Chain Models  A network model is one which can be represented by


 Transportation Problem a set of nodes, a set of arcs, and functions (e.g.
 Transshipment Problem costs, supplies, demands, etc.) associated with the
 Assignment Problem arcs and/or nodes.

 Transportation, transshipment, assignment, shortest-


route, and maximal flow problems as well as the
PERT/CPM problems are all examples of network
problems.

3 4

1
07/12/2020

TRANSPORTATION PROBLEM TRANSPORTATION PROBLEM


 Network Representation
 The transportation problem seeks to minimize the
total shipping costs of transporting goods from m
origins (each with a supply si) to n destinations (each 1 d1
with a demand dj), when the unit shipping cost from an c11
origin, i, to a destination, j, is cij. 1 c12
s1
c13
 The network representation for a transportation c21
2 d2
problem with two sources and three destinations is
given on the next slide. s2 2 c22
c23
3 d3

Sources Destinations
5 6

TRANSPORTATION PROBLEM TRANSPORTATION PROBLEM

Linear Programming Formulation  Linear Programming Formulation (continued)

Using the notation: 𝑚 𝑛

Min ෍ ෍ 𝑐𝑖𝑗 𝑥𝑖𝑗


xij = number of units shipped from 𝑖=1 𝑗=1
origin i to destination j 𝑚

cij = cost per unit of shipping from ෍ 𝑥𝑖𝑗 ≤ 𝑠𝑖 𝑓𝑜𝑟 𝑖 = 1, 2, … , 𝑚 Supply


𝑖=1
origin i to destination j
𝑛
si = supply or capacity in units at origin i
෍ 𝑥𝑖𝑗 ≤ 𝑑𝑗 𝑓𝑜𝑟 𝑗 = 1, 2, … , 𝑛 Demand
dj = demand in units at destination j 𝑗=1

continued
xij > 0 for all i and j

7 8

2
07/12/2020

TRANSPORTATION PROBLEM TRANSPORTATION PROBLEM: EXAMPLE #1


 A special class of linear programming problem
(LPP)

To Warehouse
 Need to know
From (Factory) Mumbai Patna Lucknow
1. The origin points and the capacity or supply
per period at each Jaipur $5 $4 $3

2. The destination points and the demand per Nagpur $8 $4 $3


period at each Bhubaneswar $9 $7 $5
3. The cost of shipping one unit from each
origin to each destination
9 10

4 5

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


TRANSPORTATION MATRIX

Jaipur To Jaipur
(100 units Factory
Mumbai Patna Lucknow capacity capacity
capacity) Patna From constraint
(200 units
$5 $4 $3
required) Jaipur 100
Lucknow Cell
(200 units representing
required) $8 $4 $3 a possible
Nagpur 300 source-to-
destination
$9 $7 $5 shipping
Bhubaneswar 300 assignment
Nagpur (Nagpur to
Mumbai (300 units Lucknow)
(300 units capacity) Warehouse
required) requirement 300 200 200 700

Bhubaneswar
(300 units 11 Cost of shipping 1 unit from Bhubaneswar Lucknow Total demand 12
capacity) factory to Patna warehouse warehouse demand and total supply
6 7

3
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


TRANSPORTATION SIMPLEX METHOD NORTHWEST-CORNER RULE

 The transportation problem is solved in two phases:


 Phase I -- Finding an initial feasible solution  Start in the upper left-hand cell (or northwest
 Phase II – Iterating to the optimal solution corner) of the table and allocate units to shipping
routes as follows:
 In Phase I, Northwest rule or Lowest cost or VAM 1. Exhaust the supply (factory capacity) of each row before moving
can be used to establish an initial basic feasible down to the next row
solution without doing numerous iterations of the 2. Exhaust the warehouse requirements of each column before
simplex method. moving to the next column
3. Check to ensure that all supplies and demands are met

 In Phase II, the Stepping Stone Method, using the


MODI method for evaluating the reduced costs may
be used to move from the initial feasible solution to the
optimal one. 14

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


NORTHWEST-CORNER RULE NORTHWEST-CORNER RULE

1. Assign 100 units from Jaipur to Mumbai (exhausting Jaipur’s supply)


To (A) (B) (C) Factory
From Mumbai Patna Lucknow capacity

$5 $4 $3 2. Assign 200 units from Nagpur to Mumbai (exhausting Mumbai’s


(D) Jaipur 100 100 demand)
3. Assign 100 units from Nagpur to Patna (exhausting Nagpur’s supply)
$8 $4 $3
(E) Nagpur 200 100 300 4. Assign 100 units from Bhubaneswar to Patna (exhausting Patna’s
demand)
$9 $7 $5 5. Assign 200 units from Bhubaneswar to Lucknow (exhausting
(F) Bhubaneswar 100 200 300 Lucknow’s demand and Bhubaneswar’s supply)

Warehouse
requirement 300 200 200 700

Means that the firm is shipping 100 units from 15 16


Bhubaneswar to Patna
10 9

4
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


NORTHWEST-CORNER RULE LOWEST-COST METHOD

Computed Shipping Cost 1. Identify the cell with the lowest cost
Route 2. Allocate as many units as possible to that cell
From To Tubs Shipped Cost per Unit Total Cost without exceeding supply or demand; then cross out
D A 100 $5 $500 the row or column (or both) that is exhausted by this
E A 200 8 1,600 assignment
E B 100 4 400
F B 100 7 700 3. Find the cell with the lowest cost from the remaining
F C 200 5 $1,000 cells
Total: $4,200
4. Repeat steps 2 and 3 until all units have been
This is a feasible solution allocated
but not necessarily the
17 18
lowest cost alternative.
11 12

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


LOWEST-COST METHOD LOWEST-COST METHOD

To (A) (B) (C) Factory To (A) (B) (C) Factory


From Mumbai Patna Lucknow capacity From Mumbai Patna Lucknow capacity

$5 $4 $3 $5 $4 $3
(D) Jaipur 100 100 (D) Jaipur 100 100

$8 $4 $3 $8 $4 $3
(E) Nagpur 300 (E) Nagpur 100 300

$9 $7 $5 $9 $7 $5
(F) Bhubaneswar 300 (F) Bhubaneswar 300

Warehouse Warehouse
requirement 300 200 200 700 requirement 300 200 200 700

First, $3 is the lowest cost cell so ship 100 units from Jaipur Second, $3 is again the lowest cost cell so ship 100 units
to Lucknow and cross off the first row as Jaipur is satisfied 19 from Nagpur to Lucknow and cross off column C as Lucknow 20
is satisfied
13 14

5
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


LOWEST-COST METHOD LOWEST-COST METHOD

To (A) (B) (C) Factory To (A) (B) (C) Factory


From Mumbai Patna Lucknow capacity From Mumbai Patna Lucknow capacity

$5 $4 $3 $5 $4 $3
(D) Jaipur 100 100 (D) Jaipur 100 100

$8 $4 $3 $8 $4 $3
(E) Nagpur 200 100 300 (E) Nagpur 200 100 300

$9 $7 $5 $9 $7 $5
(F) Bhubaneswar 300 (F) Bhubaneswar 300 300

Warehouse Warehouse
requirement 300 200 200 700 requirement 300 200 200 700

Third, $4 is the lowest cost cell so ship 200 units from Finally, ship 300 units from F to A as this is the only
Nagpur to Patna and cross off column B and row E as 21 remaining cell to complete the allocations 22
Nagpur and Patna are satisfied
15 16

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #1


LOWEST-COST METHOD LOWEST-COST METHOD
To (A) (B) (C) Factory
From Mumbai Patna Lucknow capacity
To (A) (B) (C) Factory $5 $4 $3
From Mumbai Patna Lucknow capacity (D) Jaipur 100 100
$5 $4 $3
(D) Jaipur 100 100 $8 $4 $3
(E) Nagpur 200 100 300
$8 $4 $3
(E) Nagpur 200 100 300 $9 $7 $5
(F) Bhubaneswar 300 300
$9 $7 $5
(F) Bhubaneswar 300 300 Warehouse
requirement 300 200 200 700
Warehouse
requirement 300 200 200 700 Total Cost = $3(100) + $3(100) + $4(200) + $9(300) = $ 4100

Total Cost = $3(100) + $3(100) + $4(200) + $9(300)


This is a feasible solution, and an improvement over
= $4,100 23 the previous solution, but not necessarily the lowest24
17
cost alternative

6
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM


STEPPING-STONE METHOD (MODI) SPECIAL ISSUES IN MODELING
To (A) (B) (C) Factory
From Mumbai Patna Lucknow capacity

(D) Jaipur 100


$5 $4 $3
100
 Demand not equal to supply
 Called an unbalanced problem
$8 $4 $3
(E) Nagpur 100 200 300
 Common situation in the real world
$9 $7 $5
(F) Bhubaneswar 100 200 300  Resolved by introducing dummy
Warehouse
sources or dummy destinations as
requirement 300 200 200 700 necessary with cost coefficients of zero

Total Cost = $5(100) + $8(100) + $4(200) + $9(100) + $5(200)


= $4,000
25 26

26 27

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM


SPECIAL ISSUES IN MODELING SPECIAL ISSUES IN MODELING

Total Cost = 250($5) + 50($8) + 200($4) + 50($3) + 150($5) + 150(0)


= $3,350  Degeneracy
 To use the stepping-stone methodology,
To
From
(A)
Mumbai
(B)
Patna
(C)
Lucknow
Dummy
Factory
capacity the number of occupied squares in any
$5 $4 $3 0
solution must be equal to the number of
(D) Jaipur 250 250 rows in the table plus the number of
$8 $4 $3 0
columns minus 1 (m+n-1)
(E) Nagpur 50 200 50 300
 If a solution does not satisfy this rule it is
(F) Bhubaneswar
$9 $7
150
$5
150
0
300 called degenerate
Warehouse
requirement 300 200 200 150 850
27 28
New
Jaipur 28 29
capacity

7
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #1 TRANSPORTATION PROBLEM: EXAMPLE #2


SPECIAL ISSUES IN MODELING TRANSPORTATION MODEL (VAM)
To Customer Customer Customer Warehouse
From 1 2 3 supply • Transportation problems are solved manually within a tableau format.

Warehouse 1 100
$8 $2 $6
100 • Each cell in a transportation tableau is analogous to a decision variable that
indicates the amount allocated from a source to a destination.
$10 $9 $9
Warehouse 2 0 100 20 120

$7 $10 $7
Warehouse 3 80 80

Customer
demand 100 100 100 300

Initial solution is degenerate


Place a zero quantity in an unused square and
29 30
proceed computing improvement indices
30 31

TRANSPORTATION PROBLEM: EXAMPLE #2 TRANSPORTATION PROBLEM: EXAMPLE #2


VOGEL’S APPROXIMATION METHOD (VAM) (1 OF 5) VOGEL’S APPROXIMATION METHOD (VAM) (2 OF 5)
- Method is based on the concept of penalty cost or regret. VAM allocates as much as possible to the minimum cost cell in the row or column
- In VAM the first step is to develop a penalty cost for supply and demand. with the largest penalty cost.

- Penalty cost is calculated by subtracting the minimum cell cost from the next higher cell
cost in each row and column.

31 32

32 33

8
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #2 TRANSPORTATION PROBLEM: EXAMPLE #2


VOGEL’S APPROXIMATION METHOD (VAM) (3 OF 5) VOGEL’S APPROXIMATION METHOD (VAM) (4 OF 5)
- After each VAM cell allocation, all row and column penalty costs are recomputed. - Recomputed penalty costs after the third allocation.

33 34

34 35

TRANSPORTATION PROBLEM: EXAMPLE #2 TRANSPORTATION PROBLEM


VOGEL’S APPROXIMATION METHOD (VAM) (5 OF 5) VOGEL’S APPROXIMATION METHOD (VAM)
SUMMARY OF STEPS
- The initial VAM solution; total cost = $5,125

1. Determine the penalty cost for each row and column.


2. Select the row or column with the highest penalty cost.
3. Allocate as much as possible to the feasible cell with the lowest transportation cost in the
row or column with the highest penalty cost.
4. Repeat steps 1, 2, and 3 until all requirements have been met.

35 36

36 37

9
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #3 TRANSPORTATION PROBLEM: EXAMPLE #3

Acme Block Company has orders for 80 tons of Delivery Cost Per Ton
concrete blocks at three suburban locations as follows:
Northwood -- 25 tons, Westwood -- 45 tons, and Northwood Westwood Eastwood
Eastwood -- 10 tons. Acme has two plants, each of Plant 1 24 30 40
which can produce 50 tons per week. Delivery cost per Plant 2 30 40 42
ton from each plant to each suburban location is shown
on the next slide.
How should end of week shipments be made to fill
the above orders?

37 38

TRANSPORTATION PROBLEM: EXAMPLE #3 TRANSPORTATION PROBLEM: EXAMPLE #4

VAM Solution The Navy has 9,000 pounds of material in Albany,


Georgia that it wishes to ship to three destinations:
Variable From To Amount Cost San Diego, Norfolk, and Pensacola. They require 4,000,
x11 Plant 1 Northwood 5 120 2,500, and 2,500 pounds, respectively. Government
x12 Plant 1 Westwood 45 1,350 regulations require equal distribution of shipping
x13 Plant 1 Eastwood 0 0 among the three carriers.
x21 Plant 2 Northwood 20 600
x22 Plant 2 Westwood 0 0
x23 Plant 2 Eastwood 10 420
Total Cost = $2,490

39 40

10
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #4 TRANSPORTATION PROBLEM: EXAMPLE #4

The shipping costs per pound for truck, railroad,


and airplane transit are shown on the next slide. Destination
Formulate and solve a linear program to determine the Mode San Diego Norfolk Pensacola
shipping arrangements (mode, destination, and
Truck $12 $6 $5
quantity) that will minimize the total shipping cost.
Railroad 20 11 9
Airplane 30 26 28

41 42

TRANSPORTATION PROBLEM: EXAMPLE #4 TRANSPORTATION PROBLEM: EXAMPLE #4

Define the Decision Variables Define the Objective Function

We want to determine the pounds of material, xij , Minimize the total shipping cost.
to be shipped by mode i to destination j. The Min: (shipping cost per pound for each mode per
following table summarizes the decision variables:
destination pairing) x (number of pounds shipped
San Diego Norfolk Pensacola by mode per destination pairing).
Truck x11 x12 x13
Min: 12x11 + 6x12 + 5x13 + 20x21 + 11x22 + 9x23
Railroad x21 x22 x23
+ 30x31 + 26x32 + 28x33
Airplane x31 x32 x33

43 44

11
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #4 TRANSPORTATION PROBLEM: EXAMPLE #4

Define the Constraints


Equal use of transportation modes:
(1) x11 + x12 + x13 = 3000
(2) x21 + x22 + x23 = 3000
(3) x31 + x32 + x33 = 3000
Destination material requirements:
(4) x11 + x21 + x31 <= 4000
(5) x12 + x22 + x32 <= 2500
(6) x13 + x23 + x33 <= 2500
Non-negativity of variables:
xij > 0, i = 1, 2, 3 and j = 1, 2, 3
45 46

TRANSPORTATION PROBLEM: EXAMPLE #4 TRANSPORTATION PROBLEM: EXAMPLE #4

Computer Output

Objective Function Value = 142000.000


Variable Value Reduced Cost
x11 1000.000 0.000
x12 2000.000 0.000
x13 0.000 1.000
x21 0.000 3.000
x22 500.000 0.000
x23 2500.000 0.000
x31 3000.000 0.000
x32 0.000 2.000
x33 0.000 6.000
47 48

12
07/12/2020

TRANSPORTATION PROBLEM: EXAMPLE #4

Solution Summary
• San Diego will receive 1000 lbs. by truck
and 3000 lbs. by airplane.
• Norfolk will receive 2000 lbs. by truck
and 500 lbs. by railroad.
• Pensacola will receive 2500 lbs. by railroad.
• The total shipping cost will be $142,000.

49

13
09/12/2020

TRANSSHIPMENT PROBLEM
 Transshipment problems are transportation
problems in which a shipment may move through
DECISION MODELLING intermediate nodes (transshipment nodes) before
reaching a particular destination node.
TERM-II
MBA (2020-21)
 Transshipment problems can also be solved by general
purpose linear programming codes.
SESSION 10
TRANSSHIPMENT & ASSIGNMENT PROBLEM  The network representation for a transshipment
problem with two sources, three intermediate nodes,
and two destinations is shown on the next slide.

Dr. Devendra Kumar Pathak


Assistant Professor,
Operations Management & Decision Sciences,
Indian Institute of Management (IIM) Kashipur
2

TRANSSHIPMENT PROBLEM TRANSSHIPMENT PROBLEM


 Network Representation
Linear Programming Formulation

c36 Using the notation:


3
c13 xij = number of units shipped from node i to node j
s1 1 c37 6 d1
c14 cij = cost per unit of shipping from node i to node j
c15 c46 si = supply at origin node i
Supply 4 c47 Demand dj = demand at destination node j
c23
continued
c24 c56
s2 2 7 d2
c25
5 c57
Sources Destinations
Intermediate Nodes
3 4

1
09/12/2020

TRANSSHIPMENT PROBLEM TRANSSHIPMENT PROBLEM

Linear Programming Formulation (continued) LP Formulation Special Cases


Min ෍ 𝑐𝑖𝑗 𝑥𝑖𝑗
• Total supply not equal to total demand
all arcs • Maximization objective function
• Route capacities or route minimums
s.t. ෍ 𝑥𝑖𝑗 ≤ 𝑠𝑖
arcs out
Origin nodes i
• Unacceptable routes
෍ 𝑥𝑖𝑗 − ෍ 𝑥𝑖𝑗 = 0 Transshipment nodes
arcs out arcs in

෍ 𝑥𝑖𝑗 = 𝑑𝑗 Destination nodes j


arcs in

xij > 0 for all i and j continued


5 6

Destination demands should be completely satisfied.

SPECIAL CASES SPECIAL CASES


Total Supply Not Equal to Total Demand Maximization Objective Function: In some problems, the objective is to
find a solution that maximizes profit or revenue. Using the values for
If total supply exceeds total demand, no modification in the LP formulation. profit or revenue per unit as coefficients in the objective function, we
Excess supply will appear as slack (unused supply or amount not shipped).
simply solve a maximization rather than a minimization linear program.
If total supply is less than total demand, the linear programming model of a This change does not affect the constraints.
transportation problem will not have a feasible solution.

Modify the network representation by adding a dummy origin with a supply equal to the Route Capacities or Route Minimums: The LP formulation of the
difference between the total demand and the total supply. transportation problem also can accommodate capacities or minimum
quantities for one or more of the routes.
With the addition of the dummy origin and an arc from the dummy origin to each we can add constraints of the form Xij <= Lij if the route from origin i to
destination, the linear programming model will have a feasible solution. destination j has capacity Lij.
A zero cost per unit is assigned to each arc leaving the dummy origin so
that the value of the optimal solution for the revised problem will represent
the shipping cost for the units actually shipped (no shipments actually will Unacceptable Routes: Establishing a route from every origin to every
be made from the dummy origin). destination may not be possible. To handle this situation, we simply drop
When the optimal solution is implemented, the destinations showing the corresponding arc from the network and remove the
7 8
shipments being received from the dummy origin will be the corresponding variable from the linear programming formulation.
destinations experiencing a shortfall or unsatisfied demand.

2
09/12/2020

TRANSSHIPMENT PROBLEM: EXAMPLE TRANSSHIPMENT PROBLEM: EXAMPLE

The Northside and Southside facilities of Because of long standing contracts based on past
Zeron Industries supply three firms (Zrox, Hewes, orders, unit costs from the manufacturers to the
Rockrite) with customized shelving for its offices. suppliers are:
They both order shelving from the same two
Zeron N Zeron S
manufacturers, Arnold Manufacturers and
Arnold 5 8
Supershelf, Inc. Supershelf 7 4
Currently weekly demands by the users are 50 for
Zrox, 60 for Hewes, and 40 for Rockrite. Both Arnold The costs to install the shelving at the various
and Supershelf can supply at most 75 units to its locations are:
customers.
Zrox Hewes Rockrite
Additional data is shown on the next slide. Thomas 1 5 8
Washburn 3 4 4
9 10

TRANSSHIPMENT PROBLEM: EXAMPLE TRANSSHIPMENT PROBLEM: EXAMPLE

 Network Representation  Linear Programming Formulation


 Decision Variables Defined
Zrox 50 xij = amount shipped from manufacturer i to
5 Zeron
1 supplier j
75 Arnold
N 5 xjk = amount shipped from supplier j to customer k
8 8 where i = 1 (Arnold), 2 (Supershelf)
Hewes 60
j = 3 (Zeron N), 4 (Zeron S)
7 3 k = 5 (Zrox), 6 (Hewes), 7 (Rockrite)
Super Zeron 4  Objective Function Defined
75 Shelf
4 S
4 Minimize Overall Shipping Costs:
Rock-
Rite 40 Min 5x13 + 8x14 + 7x23 + 4x24 + 1x35 + 5x36 + 8x37
11
+ 3x45 + 4x46 + 4x47 12

3
09/12/2020

TRANSSHIPMENT PROBLEM: EXAMPLE TRANSSHIPMENT PROBLEM: EXAMPLE

 Constraints Defined
Amount Out of Arnold: x13 + x14 < 75
Amount Out of Supershelf: x23 + x24 < 75
Amount Through Zeron N: x13 + x23 - x35 - x36 - x37 = 0
Amount Through Zeron S: x14 + x24 - x45 - x46 - x47 = 0
Amount Into Zrox: x35 + x45 = 50
Amount Into Hewes: x36 + x46 = 60
Amount Into Rockrite: x37 + x47 = 40

Non-negativity of Variables: xij > 0, for all i and j.

13 14

TRANSSHIPMENT PROBLEM: EXAMPLE TRANSSHIPMENT PROBLEM: EXAMPLE

15 16
If you need to change the limiting capacity of any of the given
constraint by one unit, which one would you recommend?

4
09/12/2020

TRANSSHIPMENT PROBLEM: EXAMPLE ASSIGNMENT PROBLEM

Solution  An assignment problem seeks to minimize the total


cost assignment of m workers to n jobs, given that the
Zrox 50 cost of worker i performing job j is cij. [For
assigning jobs to Machine, agents to tasks]
5
75 1
Zeron
75 Arnold 5
N  It assumes all workers are assigned and each job is
8 8 performed.
Hewes 60

3 4  An assignment problem is a special case of a


7 transportation problem in which all supplies and
75
Super
Shelf
Zeron
S
all demands are equal to 1; hence assignment
4 4 problems may be solved as LPP.
Rock-
Rite 40
17 18

ASSIGNMENT PROBLEM ASSIGNMENT PROBLEM

 Network Representation
• In a balanced model supply equals demand.
c11
• In an unbalanced model supply does not equal 1
c12
1

demand. c13
Agents Tasks
• The network representation of an assignment c21
c22
problem with three workers and three jobs is 2 2
shown on the next slide. c23
c31
c32
3 3
c33
19 20

5
09/12/2020

ASSIGNMENT PROBLEM ASSIGNMENT PROBLEM

Linear Programming Formulation  Linear Programming Formulation (continued)

Using the notation: 𝑚 𝑛

Min ෍ ෍ 𝑐𝑖𝑗 𝑥𝑖𝑗


𝑖=1 𝑗=1
xij = 1 if agent i is assigned to task j
𝑚
0 otherwise
෍ 𝑥𝑖𝑗 = 1 𝑓𝑜𝑟 𝑗 = 1, 2, … , 𝑛 Tasks
𝑖=1
cij = cost of assigning agent i to task j
𝑛
continued ෍ 𝑥𝑖𝑗 ≤ 1 𝑓𝑜𝑟 𝑖 = 1, 2, … , 𝑚 Agents
𝑗=1

xij > 0 for all i and j

21 22
Each agent can be assigned to at most one task and each task must
have one assigned agent.

ASSIGNMENT PROBLEM ASSIGNMENT PROBLEM: EXAMPLE


An electrical contractor pays his subcontractors a
LP Formulation Special Cases fixed fee plus mileage for work performed. On a given
• Number of agents exceeds the number of tasks: day the contractor is faced with three electrical jobs
associated with various projects. Given below are the
Extra agents simply remain unassigned. distances between the subcontractors and the
• Number of tasks exceeds the number of agents: projects.
Projects
Add enough dummy agents to equalize the
number of agents and the number of tasks. Subcontractor A B C
The objective function coefficients for these Westside 50 36 16
new variable would be zero. Federated 28 30 18
Goliath 35 32 20
Universal 25 25 14
How should the contractors be assigned so that total
23 24
mileage is minimized?

6
09/12/2020

ASSIGNMENT PROBLEM: EXAMPLE ASSIGNMENT PROBLEM: EXAMPLE


Network Representation
50 Linear Programming Formulation
West. A
36
16 Min 50x11+36x12+16x13+28x21+30x22+18x23
28 +35x31+32x32+20x33+25x41+25x42+14x43
Fed.
30
B s.t. x11+x12+x13 < 1
18 x21+x22+x23 < 1
Agents
Subcontractors 35 32 Projects x31+x32+x33 < 1
x41+x42+x43 < 1
Gol.
20
C x11+x21+x31+x41 = 1
25
25 x12+x22+x32+x42 = 1 Tasks
x13+x23+x33+x43 = 1
Univ. 14
xij = 0 or 1 for all i and j
25 26

ASSIGNMENT PROBLEM: EXAMPLE ASSIGNMENT PROBLEM: EXAMPLE

27 28

7
09/12/2020

ASSIGNMENT PROBLEM: EXAMPLE ASSIGNMENT PROBLEM: EXAMPLE

 The optimal assignment is:

Subcontractor Project Distance


Westside C 16
Federated A 28
Goliath (unassigned)
Universal B 25
Total Distance = 69 miles

29 30

ASSIGNMENT PROBLEM: EXAMPLE 2 ASSIGNMENT PROBLEM: EXAMPLE 2 FORMULATION

Minimize Z = 210xAR + 90xAA + 180xAD + 160xAC + 100xBR +


Problem: Assign four teams of officials to four games in a
70xBA + 130xBD + 200xBC + 175xCR + 105xCA +
way that will minimize total distance traveled by the
140xCD + 170xCC + 80xDR + 65xDA + 105xDD +
officials. Supply is always one team of officials, demand is
120xDC
for only one team of officials at each game.
subject to:
Data:
xAR + xAA + xAD + xAC <= 1 xij  0
xBR + xBA + xBD + xBC <= 1
xCR + xCA + xCD + xCC <= 1
xDR + xDA + xDD + xDC <= 1
xAR + xBR + xCR + xDR = 1
xAA + xBA + xCA + xDA = 1
xAD + xBD + xCD + xDD = 1
xAC + xBC + xCC + xDC = 1
TRAVEL DISTANCES
7

8
09/12/2020

ASSIGNMENT PROBLEM: EXAMPLE 2 FORMULATION ASSIGNMENT PROBLEM: EXAMPLE 2 FORMULATION

33 34

ASSIGNMENT PROBLEM: EXAMPLE 2 FORMULATION ASSIGNMENT PROBLEM: EXAMPLE 2 SOLUTION


R

35 C

11

You might also like