Chi-Kong Ng, SEEM2420, Dept.

of SEEM, CUHK 3:1

Chapter 3.

OR Modeling Approach and Linear

Programming Model
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:2

3.1. Overview of the OR Modeling

Approach
3.1.1. Six (Overlapping) Steps of Modeling
Step 1. Defining the Problem and Gathering Data
• Most practical problems encountered by OR teams are initially
described to them in a vague, imprecise way.

• Make sure the problem is the right problem. The right problem
often refers to the one from the management’s viewpoint.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:3

• Develop a well-defined statement of the problem: Appropriate

objective, constraints on what can be done, interrelationship
between the area to be studied and other areas in the organization,
possible alternative decisions, time limits for making a decision, etc.
• Remember: It is difficult to extract a “right” answer from the
“wrong” problem.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:4

Step 2. Formulating a Mathematical Model

Model:
An idealized and approximate representation of more complex sys-
tems in real world.

Mathematical model:
A model expressed in terms of mathematical symbols and expressions.

Elements of Most OR Problems:

• Decision Variables:
If there are n related quantifiable decisions to be made, they
are represented as decision variables (say, x1, x2, . . . , xn) whose
respective values are to be determined.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:5

• Constraints:
Any restrictions on the values that can be assigned to the decision
variables are called constraints.
• Objective Function:
In many OR problems, the decision maker wants to maximize
(usually revenue or profit) or minimize (usually costs) some function
of the decision variables. The function to be maximized or
minimized is called the objective function.
• Parameters:
The constants (namely, the coefficients and right-hand sides) in the
constraints and the objective function are called parameters.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:6

Goal of the mathematical model:

To choose the value of the decision variables so as to optimize (i.e.,
maximize/minimize) the objective function, subject to the specified
constraints, for each set of given parameters.

Advantages of mathematical model:

Concise, systematic and efficient.

Problems with mathematical model:

Assumptions.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:7

Step 3. Deriving Solutions from the Model

• Theoretically, a common theme in OR is the search for an optimal
solution. Practically, it should tend to be a good approximation
to the ideal decision for the real problem.
• When we study an OR problem, we should also consider the cost of
the study itself and then maximize the net benefits resulting from
the study. Occasionally, OR teams use heuristic procedures to find
a good solution.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:8

• Post-optimality analysis:
Analysis done after finding an optimal solution.
◦ What-if analysis:
What would happen to the optimal solution if different assump-
tions are made about future condition?
◦ Sensitivity analysis:
Which parameters of the model are most critical in determining
the solution?
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:9

Step 4. Testing the Model

• Given the difficulty of communicating and understanding all the
aspects and subtleties of a complex operational problem, it is very
likely that an OR team either has not been given all the true facts
of the situation or has not interpreted them properly. Thus model
validation and solution test are necessary and important.
• Methods include:
◦ checking for obvious errors/oversights in the model;
◦ reexamining formulation of the problem and comparing with
the model;
◦ varying the parameters/decision variables and checking the
outputs.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:10

Step 5. Preparing to Apply the Model

If a model is to be used repeatedly, it is necessary to install a well-
documented system for applying for the model, which includes:
• the model
• the solution procedure
• operating procedure for implementation
• a plan for detecting the changes in the values of parameters and
making necessary modifications
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:11

Step 6. Implementation
• Implementation must meet the requirement/expectation of the
management.
• Implementation involves several steps:
◦ Explaining to operating management the final solution and how
it relates to operating realities
◦ Developing the procedures to put this solution into operation
◦ Monitoring the initial experience with the solution and identi-
fying any necessary modifications
◦ Documenting the methodology clearly and accurately to make
it a reproducible work
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:12

3.1.2. Focus in this Course

Formulating and solving mathematical models (steps 2 and 3).
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:13

3.1.3. Prototype Example — Wyndor Glass Co.

Problem
Problem Description:
Top management of the Wyndor Glass Co. decided to revamp the
company’s product line. Unprofitable products are being discontinued,
releasing production capacity to launch two new products having large
sales potential.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:14

Step 1. Defining the Problem and Gathering Data

• Plant 1: Aluminum frames and hardware
• Plant 2: Wood frames.
• Plant 3: Glass and product assembly.
• New products:
◦ Product 1: An 8-foot glass door with aluminum frame.
It requires some of the production capacity in Plants 1 and 3.
◦ Product 2: A 4 × 6 foot double-hung wood-framed window.
It needs only Plants 2 and 3.
• In order to maximize the total profit, what are the production rates
for the two products subject to the restrictions imposed by the
production capacity.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:15

• Data:
◦ Each product is produced in batches of 20.
◦ The production rate = number of batches per week.
◦ The other details are shown below:
Production TimeProduction Time
Plant per Batch (Hours)
Available per Week
Product 1 Product 2 (Hours)
1 1 0 4
2 0 2 12
3 3 2 18
Profit/Batch ($) 3000 5000
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:16

Step 2. Formulating a Mathematical Model

Let xj = no. of batches of product j produced per week (for j =
1, 2);
Z = total profit per week (in thousands of dollars) from
producing these two products.
The Linear Programming Model:
Maximize
Z = 3x1 + 5x2 (Total Profit)
subject to
x1 ≤ 4 (Plant 1)
2x2 ≤ 12 (Plant 2)
3x1 + 2x2 ≤ 18 (Plant 3)
x1 ≥ 0
x2 ≥ 0
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:17

Step 3a. Deriving the Solution

x2

3x1+2x2=18
Z=36=3x1+5x2 x1=4

(2,6)
2x2=12

Z=20=3x1+5x2

Feasible
Z=10=3x1+5x2 Region

x1
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:18

3.1.4. Example: Vitamin Problem

Problem Description
John has recently seen a medical doctor and the medical doctor
recommends him to take some tablets as supplement for several
vitamins.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:19

Step 1. Defining the Problem and Gathering Data

• There are two types of tablets available: Tablets A and B.
• Every piece of Tablet A contains 10 units of Vitamin A, 4 units of
Vitamin B1 and 2 units of Vitamin C.
• Every piece of Tablet B contains 4 units of Vitamin A, 6 units of
Vitamin B1 and 6 units of Vitamin C.
• Tablet A costs $0.5 and Tablet B costs $1.0.
• According to the doctor’s advice, the minimum daily vitamin
requirement for John is 20 units, 18 units and 13 units for Vitamins
A, B1 and C, respectively.
• In order to minimize the total cost of buying the tablets and fulfilling
the daily requirement of vitamins, how many pieces of Tablets A
and B should John take everyday and buy every month (30 days)?
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:20

Step 2. Formulating a Mathematical Model

Please try to formulate it as a linear programming problem.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:21

Step 3a. Deriving the Solution

Please try to solve it by using the graphical method.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:22

Step 4. Testing the Model

• Does the solution make sense?
• Can we divide the tablets into a lot of small pieces?

Step 3b. Post-optimality Analysis:

• What will happen to the solution if the costs of the tablets change?

• What will happen if the daily requirements of the vitamins change?

Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:23

3.1.5. More on the Example of Vitamin Problem

• Now, if there are 3 more different tablets available, Tablets C, D
and E, formulate the problem again in a linear programming model.
• The components and prices of Tablets C, D and E (per piece) are
shown below:
Tablets
C D E
A 1 5 6
Vitamins B1 1 15 3
C 8 1 9
Unit Cost ($) 0.3 1.2 0.8
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:24

• Formulate the problem as a LP problem.

• How many pieces of Tablets A, B, C, D and E should John take

everyday and buy every month?
• Can we solve the problem using the graphical method?
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:25

3.2. Linear Programming

3.2.1. Introduction
• The development of LP has been ranked among the most important
scientific advances of the mid-20th century.
• Impact of Linear Programming (LP) since 1950 has been extraor-
dinary (after the birth of the simplex method by George Dantzig
in 1947).
• Linear means all the mathematical functions are linear.
• Programming means planning.
• LP involves the planning of activities to obtain an optimal result,
i.e., a result that reaches the specified goal best (according to the
mathematical model) among all feasible alternatives.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:26

3.2.2. The LP Model

Our Standard Form for the LP Problem
Maximize
Z = c1x1 + c2x2 + · · · + cn xn (Objective function)
subject to
a11x1 + a12x2 + · · · + a1nxn ≤ b1 (Functional constraint)
a21x1 + a22x2 + · · · + a2nxn ≤ b2 (Functional constraint)
...
am1x1 + am2x2 + · · · + amn xn ≤ bm (Functional constraint)
x1 ≥ 0, x2 ≥ 0, . . . , xn ≥ 0 (Nonnegativity constraints)
where
b1 ≥ 0, b2 ≥ 0, . . . , bm ≥ 0
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:27

Other Forms
• Minimization instead of maximization
• Some constraints take different sign restrictions:
All kinds of ≤, = and ≥.
• Some decision variables are not restricted by the non-negativity
constraints:
e.g., xj ≤ 0 (for some values of j),
e.g., xj unrestricted in sign (for some values of j).
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:28

Terminologies
A solution:
Any specification of values for the decision variables (x1, x2, . . . , xn).

A feasible solution:
A solution for which all the constraints are satisfied.

An infeasible solution:
A solution for which at least one constraint is violated.

The feasible region:

The collection of all feasible solutions. Note that it is possible for a
problem to have no feasible solutions.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:29

An optimal solution:
A feasible solution that has the most favorable value of the objective
function, i.e., the largest value if the objective function is to be
maximized. Note that it is possible for a problem to have multiple
optimal solutions.

A corner-point solution:
A solution that lies at the intersection of n constraint boundaries.

A corner-point feasible (CPF) solution:

A solution that lies at a corner of the feasible region.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:30

3.2.3. Assumptions of LP
Proportionality
• The contribution of each activity to the value of the objective
function Z is proportional to the level of the activity xj , as
represented by the cj xj term in the objective function.
• The contribution of each activity to the left-hand side of each
functional constraint is proportional to the level of the activity xj ,
as represented by the aij xj term in the constraint.
• Please give a counter example.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:31

Additivity
• Every function in a LP model is the sum of the individual
contributions of the respective activities.
• Please give a counter example.

Divisibility
• No constraint for being integer values.

Certainty
• The value of each parameter is known for certain.
• Please give a counter example.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:32

3.2.4. More Examples

• Design of Radiation Therapy (Hillier and Lieberman, 10th Edition,
p.45)
• Regional Planning (Hillier and Lieberman, 10th Edition, p.47)
• Controlling Air Pollution (Hillier and Lieberman, 10th Edition,
p.51)
• Reclaiming Solid Wastes (Hillier and Lieberman, 10th Edition,
p.53)
• Personnel Scheduling (Hillier and Lieberman, 10th Edition, p.57)
• Distributing Goods through a Distribution Network (Hillier and
Lieberman, 10th Edition, p.60)
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:33

3.2.5. Some Case Studies

• A “taste” of the realistic problem.
• Each of these is a classic application, initiated in the early 1980’s,
that has come to be regarded as a standard of excellence for future
applications of LP.

Choosing the Product Mix at Ponderosa Industrial

• 1980.
• 90 decision variables and 45 functional constraints
• The overall profitability was increased by 20%.
Chi-Kong Ng, SEEM2420, Dept. of SEEM, CUHK 3:34

Personnel Scheduling at United Airlines

• 1982.
• The core of the planning system was LP. (The details about the LP
model have not been published.)
• Over 20,000 decision variables.
• Saves US$ 6 million annually in just direct salary and benefit cost.

Planning Supply, Distribution and Marketing at Citgo

Petroleum Corporation
• 1984–1985.
• 15,000 decision variables and 3,000 equality constraints.
• The annual profit was increased by US$ 16.5 million.