Spring, 2007

ISE 102

Introduction to Linear
Programming (LP)

Asst. Prof. Dr. Mahmut Ali GÖKÇE

Industrial Systems Engineering Dept.

İzmir University of Economics

1
www.izmirekonomi.edu.tr
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
Spring, 2007

Introduction to Linear
Programming
 Many managerial decisions involve trying to
make the most effective use of an organization’s
resources. Resources typically include:
 Machinery/equipment
 Labor
 Money
 Time
 Energy
 Raw materials
 These resources may be used to produce
 Products (machines, furniture, food, or clothing)
 Services (airline schedules, advertising policies, or
investment decisions)
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
2 of 52
www.izmirekonomi.edu.tr
Spring, 2007

What is Linear Programming?

 Linear Programming is a mathematical technique

designed to help managers plan and make
necessary decisions to allocate resources
 Linear Programming (LP) is one the most widely
used decision tools of Operations Research &
Management Science (ORMS)
 In a survey of Fortune 500 corporations, 85 % of
those responding said that they had used LP

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

3 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Brief History of LP

 LP was developed to solve military logistics

problems during World War II
 In 1947, George Dantzig developed a
solution procedure for solving linear
programming problems (Simplex Method)
 This method turned out to be so efficient for
solving large problems quickly
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
4 of 52
www.izmirekonomi.edu.tr
Spring, 2007

History of LP (contd)

 The simultaneous development of the

computer technology established LP as an
important tool in various fields
 Simplex Method is still the most important
solution method for LP problems
 In recent years, a more efficient method for
extremely large problems has been
developed (Karmarkar’s Algorithm)
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
5 of 52
www.izmirekonomi.edu.tr
Spring, 2007

LP Problems

 A large number of real problems can be

formulated and solved using LP. A partial list
includes:
 Scheduling of personnel
 Production planning and inventory control
 Assignment problems
 Several varieties of blending problems including
ice cream, steel making, crude oil processing
 Distribution and logistics problems
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
6 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Typical Applications of LP

 Aggregate Planning
Develop a production schedule which
 satisfies specified sales demands in future periods
 satisfies limitations on production capacity
 minimizes total production/inventory costs

 Scheduling Problem
Produce a workforce schedule which
 satisfies minimum staffing requirements
 utilizes reasonable shifts for the workers
 is least costly

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

7 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Typical Applications of LP
(contd)
 Product Mix (“Blending”) Problem
Develop the product mix which
 satisfies restrictions/requirements for customers
 does not exceed capacity and resource constraints
 results in highest profit

 Logistics
Determine a distribution system which
 meets customer demand
 minimizes transportation costs

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

8 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Typical Applications of LP
 Marketing
(contd)
Determine the media mix which
 meets a fixed budget
 maximizes advertising effectiveness

 Financial Planning
Establish an investment portfolio which
 meets the total investment amount
 meets any maximum/minimum restrictions of
investing in the available alternatives
 maximizes ROI
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
9 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Typical Applications of LP
(contd)
 What do these applications have in common?
All are concerned with maximizing or
minimizing some quantity, called the
objective of the problem
All have constraints which limit the degree
to which the objective function can be
pursued

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

10 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Typical Applications of LP
(contd)at Delta Air Lines
Fleet Assignment
 Delta Air Lines flies over 2500 domestic flight legs every
day, using about 450 aircrafts from 10 different fleets
that vary by speed, capacity, amount of noise
generated, etc.
 The fleet assignment problem is to match aircrafts (e.g.
Boeing 747, 757, DC-10, or MD80) to flight legs so that
seats are filled with paying passengers

 Delta is one the first airlines to solve to completion this

fleet assignment problem, one of the largest and most
difficult problems in airline industry
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
11 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Fleet Assignment at Delta

(contd)
 An airline seat is the most perishable
commodity in the world
 Each time an aircraft takes off with an empty
seat, a revenue opportunity is lost forever
 The flight schedule must be designed to
capture as much business as possible,
maximizing revenues with as little direct
operating cost as possible

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

12 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Fleet Assignment at Delta

(contd)
 The airline industry combines
 the capital-intensive quality of the manufacturing sector
 low profit margin quality of the retail sector
 Airlines are capital, fuel, and labor intensive
 Survival and success depend on the ability to
operate flights along the schedule as efficiently as
possible

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

13 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Fleet Assignment at Delta

(contd)
 Both the size of the fleet and the number of different
types of aircrafts have significant impact on schedule
planning
 If the airline assigns too small a plane to a particular
market:
it will lose potential passengers
 If it assigns too large a plane:
it will suffer the expense of the larger plane
transporting empty seats

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

14 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Stating the LP Model

 Delta implemented a large scale linear

program to assign fleet types to flight
legs so as to minimize a combination of
operating and passenger “spill” costs,
subject to a variety of operation
constraints

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

15 of 52
www.izmirekonomi.edu.tr
Spring, 2007

What are the constraints?

 Some of the complicating factors include:
number of aircrafts available in each fleet
planning for scheduled maintenance (which city is
the best to do the maintenance?)
matching which crews have the skills to fly which
aircrafts
providing sufficient opportunity for crew rest time
range and speed capability of the aircraft
airport restrictions (noise levels!)

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

16 of 52
www.izmirekonomi.edu.tr
Spring, 2007

The result?!

 The typical size of the LP model that Delta

has to optimize daily is:
 40,000 constraints
 60,000 decision variables

 The use of the LP model was expected to

save Delta $300 million over the 3 years
(1997)

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

17 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Formulating LP Models
 An LP model is a model that seeks to maximize or
minimize a linear objective function subject to a
set of constraints

 An LP model consists of three parts:

a well-defined set of decision variables
an overall objective to be maximized or minimized
a set of constraints

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

18 of 52
www.izmirekonomi.edu.tr
Spring, 2007

PetCare Problem
PetCare specializes in high quality care for large
dogs. Part of this care includes the assurance that
each dog receives a minimum recommended
amount of protein and fat on a daily basis. Two
different ingredients, Mix 1 and Mix 2, are
combined to create the proper diet for a dog. Each
kg of Mix 1 provides 300 gr of protein, 200 gr of
fat, and costs $.80, while each kg of Mix 2
provides 200 gr of protein, 400 gr of fat, and costs
$.60. If PetCare has a dog that requires at least
1100 gr of protein and 1000 gr of fat, how many
kgs of each mix should be combined to meet the
nutritional requirements at a minimum cost?
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
19 of 52
www.izmirekonomi.edu.tr
Spring, 2007

LP Formulation Steps

 STEP 1: Understand the Problem

 STEP 2: Identify the decision variables

 STEP 3: State the objective function

 STEP 4: State the constraints

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

20 of 52
www.izmirekonomi.edu.tr
Spring, 2007

PetCare Problem
PetCare specializes in high quality care for large
dogs. Part of this care includes the assurance that
each dog receives a minimum recommended
amount of protein and fat on a daily basis. Two
different ingredients, Mix 1 and Mix 2, are
combined to create the proper diet for a dog. Each
kg of Mix 1 provides 300 gr of protein, 200 gr of
fat, and costs $.80, while each kg of Mix 2
provides 200 gr of protein, 400 gr of fat, and
costs $.60. If PetCare has a dog that requires at
least 1100 gr of protein and 1000 gr of fat, how
many kgs of each mix should be combined to meet
the nutritional requirements at a minimum cost?
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
21 of 52
www.izmirekonomi.edu.tr
Spring, 2007

PetCare: LP Formulation
 STEP 1: Understand the Problem
 STEP 2: Identify the decision variables
x1 : kgs of mix 1 to be used to feed the dog
x2 : kgs of mix 2 to be used to feed the dog
 STEP 3: State the objective function
minimize 0.8 x1 + 0.6 x2 (total cost)

 STEP 4: State the constraints

subject to 300 x1 + 200 x2  1100 (protein constraint)
200 x1 + 400 x2  1000 (fat constraint)
x1  0 (sign restriction)
x2  0 (sign restriction)

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

22 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Furnco Company Problem

Furnco manufactures desks and chairs. Each

desk uses 4 units of wood, and each chair uses 3
units of wood. A desk contributes $40 to profit,
and a chair contributes $25. Marketing restrictions
require that the number of chairs produced must
be at least twice the number of desks produced.
There are 20 units of wood available. Formulate
the Linear Programming model to maximize
Furnco’s profit.

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

23 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Furnco Company (contd)

x1 : number of desks produced
x2 : number of chairs produced

maximize 40 x1 + 25 x2 (objective
function)

subject to 4 x1 + 3 x2  20 (wood constraint)

2 x1 - x2  0 (marketing constraint)
x1 , x2  0 (sign restrictions)
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
24 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Farmer Jane Problem

Farmer Jane owns 45 acres of land. She is going
to plant each acre with wheat or corn. Each acre
planted with wheat yields $200 profit; each with
corn yields $300 profit. The labor and fertilizer
used for each acre are as follows:
Wheat Corn
Labor 3 workers 2 workers
Fertilizer 2 tons 4 tons
100 workers and 120 tons of fertilizer are
available. Formulate the Linear Programming
model to maximize the farmer’s profit.

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

25 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Farmer Jane (contd)

x1 : acres of land planted with wheat

x2 : acres of land planted with corn

maximize 200 x1 + 300 x2 (objective function)

subject to x1 + x2  45 (land constraint)

3 x1 + 2 x2  100 (labor constraint)
2 x1 + 4 x2  120 (fertilizer constraint )
x1 , x2  0 (sign restrictions)

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

26 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Truck-Co Company Problem

Truck-co manufactures two types of trucks: 1 and
2. Each truck must go through the painting shop
and the assembly shop. If the painting shop were
completely devoted to painting type 1 trucks, 800
per day could be painted, whereas if it were
completely devoted to painting type 2 trucks, 700
per day could be painted. Is the assembly shop
were completely devoted to assembling truck 1
engines, 1500 per day could be assembled, and if
it were completely devoted to assembling truck 2
engines, 1200 per day could be assembled. Each
type 1 truck contributes $300 to profit; each type
2 truck contributes $500. Formulate the LP
problem to maximize Truckco’s profit.

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

27 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Truckco Company (contd)

x1 : number of type 1 trucks manufactured

x2 : number of type 2 trucks manufactured

maximize 300 x1 + 500 x2 (objective function)

subject to 7 x1 + 8 x2  5600 (painting constraint)

12 x1 + 15 x2  18000 (assembly constraint)
x1 , x2  0 (sign restrictions)

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

28 of 52
www.izmirekonomi.edu.tr
Spring, 2007

McDamat’s Fast Food

Restaurant
McDamat's fast food restaurant requires different number of full
time employees on different days of the week. The table below
shows the minimum requirements per day of a typical week:
Day of week Empl Reqd Day of week Empl Reqd
Monday 7 Friday 4
Tuesday 3 Saturday 6
Wednesday 5 Sunday 4
Thursday 9
Union rules state that each full-time employee must work 5
consecutive days and then receive 2 days off. The restaurant wants
to meet its daily requirements using only full time personnel.
Formulate the LP model to minimize the number of full time
employees required.

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

29 of 52
www.izmirekonomi.edu.tr
Spring, 2007

McDamat’s Fast Food

Restaurant (contd)
 Defining Decision Variables
xi : number of employees beginning work on
day i where i = Monday, …. ,
Sunday

 Defining the Objective Function

min Z = xmon + xtue + xwed + xthu + xfri + xsat
+ xsun
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
30 of 52
www.izmirekonomi.edu.tr
Spring, 2007

McDamat’s Fast Food

Restaurant
 (contd)
Defining the Constraint Set
xmon + xthu + xfri + xsat + xsun  7 (Mon Reqts)
xmon + xtue + xfri + xsat + xsun  3 (Tue Reqts)
xmon + xtue + xwed + xsat + xsun  5(Wed Reqts)
xmon + xtue + xwed + xthu + xsun  9(Thu Reqts)
xmon + xtue + xwed + xthu + xfri  4 (Fri Reqts)
xtue + xwed + xthu + xfri + xsat  6 (Sat Reqts)
xwed + xthu + xfri + xsat + xsun  4 (Sun Reqts)
 Non-Negativity Condition (Sign Restriction)
xmon , …. , xsun  0
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
31 of 52
www.izmirekonomi.edu.tr
Spring, 2007

A Multi-Period Production
Planning
Sailco CorporationPr.
must determine how many sailboats to produce
during each of the next four quarters. The demand during each of
the next four quarters is as follows:
Quarters 1 2 3 4 .
Demand 40 60 75 25
At the beginning of the first quarter Sailco has an inventory of 10
sailboats.
At the beginning of each quarter Sailco must decide how many
sailboats to produce that quarter. Sailboats produced during a
quarter can be used to meet demand for that quarter.
Capacity Cost .
Regular Time 40 (sailboats) $400/sailboat
Overtime $450/sailboat
Inventory Holding Cost: $20/sailboat
Determine a production schedule to minimize the sum of production
and inventory holding costs during the next four quarters.
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
32 of 52
www.izmirekonomi.edu.tr
Spring, 2007

A Multiperiod PP Problem
(contd)
 Defining Decision Variables
R1 : regular time production at quarter 1
R2 : regular time production at quarter 2
…
…
Rt : regular time production at quarter t
Ot : overtime production at quarter t
It : inventory at the end of quarter t

 Defining the Objective Function

min 400 R1 + 400 R2 + 400 R3 + 400 R4 + 450 O1 + 450 O2 +
450 O3 + 450 O4 + 20 I1 + 20 I2 + 20 I3 + 20 I4

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

33 of 52
www.izmirekonomi.edu.tr
Spring, 2007

A Multiperiod PP Problem
(contd)
 Defining the Constraint Set
10 + R1 + O1 - I1 = 40
I1 + R2 + O2 -1I2 = 60
I2 + R3 + O3 - I3 = 75
I3 + R4 + O4 - I4 = 25
R1  40
R2  40
R3  40
R4  40

 Non-Negativity Condition (Sign Restriction)

R1, R2, R3, R4, O1, O2, O3, O4, I1, I2, I3, I4  0

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

34 of 52
www.izmirekonomi.edu.tr
Spring, 2007

LP Summary

 An LP problem is an optimization problem for

which we do the following:
 We attempt to maximize (or minimize) a linear function
of the decision variables. The function that is to be
maximized (or minimized) is called the objective
function
 The values of the decision variables must satisfy a set
of constraints. Each constraint must be a linear
equation or linear inequality
 A sign restriction is associated with each variable

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

35 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Graphical Solution Method

X2
Chair
s Furnco Company
7
[166.7 6.67
(2) Max 40 x1 + 25 x2
5] s.t. 4 x1 + 3 x2  20
6

2 x1 - x2  0
5 x1 , x2  0

4
(2,4)[180]

0 2.5 3.754
2 5 6 7 X1
Desk
(1) s
Z=15
Z=10 0
0
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
36 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Graphical Solution Method

X2 (4)

(contd)
Corn

60
Farner Jane (modified)
max 200 x1 + 300 x2
50 s.t x1 + x2  45
45 3 x1 + 2 x2  100
40
2 x1 + 4 x2  120
x1 ≥ 10
(20,20) x1 , x 2  0
30

(10,25)

20 (30,15)

10
(3)
(2) (1)
0
20 33.3 40 45 X1
10

30

50 60
[2000] Whea
[6667] t
Z=7080
Z=6000
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
37 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Special Cases of the Feasible

Region

Infeasible Redundant
Constraint
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
38 of 52
www.izmirekonomi.edu.tr
 Spring, 2007

More Special Cases of the

Feasible Region

Unbounded Feasible RegionUnbounded Feasible Region

Unbounded Solution Bounded Optimal Solution
Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics
39 of 52
www.izmirekonomi.edu.tr
Spring, 2007

Special Cases of the Optimal

Solution

Multiple Optima Unbounded Solution

Asst. Prof. Dr. Mahmut Ali GÖKÇE, Izmir University of Economics

40 of 52
www.izmirekonomi.edu.tr