You are on page 1of 24

Goal Programming Models

• LP/IP -> single objective


• Multiple objectives
• Real-life problems

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-1
Goal Programming Models
• Goal programming “satisfices” as opposed
to LP models which “optimize”

• This means finding a point which satisfies all


the objectives as best as possible

• There may be a hierarchical order among


objectives

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-2
Goal Programming Models
• In goal programming
• The problem specific constraints (the
ordinary LP constraints) are known as
hard constraints.

• The goals(objectives) are expressed as


constraints and these are called as soft
constraints.

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-3
Goal Programming Models
• Deviation variables are used to express
under- and over-achivement from the goals.

• There are 2 nonnegative deviation variables


for every goal: one for underachivement and
one for overachivement.

di- = amount of underachievement for goal i


di+ = amount of overachievement for goal i

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-4
Goal Programming Models
• Example:
x1: amount of units to be produced from product 1
x2: amount of units to be produced from product 2

One unit of x1 and x2 will bring $3 and $2, respectively.

• The decision maker states his goal 1 as “achieve profit


of at least $3,000”

• Not hard constraint overachievement

• 3x1 + 2x2 + d1- - d1+ = 3000 (soft constraint)


based on underachievement
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-5
Goal Programming Models
• 3x1 + 2x2 + d1- - d1+ = 3000 (soft constraint)

• d1- and d1+ are nonnegative variables, both cannot have


positive values at the same time
overachievement is
• If profit is 5,000, then d1- =0, d1+ =2,000 captured by d1+

underachievement
• If profit is 2,000, then d1- =1,000, d1+ =0 is captured by d1-

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-6
Goal Programming Models
• The objective function of goal programming is formed by
deviation variables only and we want to minimize these
variables

• Goal 1: achieve profit of at least $3,000

• If profit > 3,000, no problem

• Minimize underachievement

Min d1- + …
for other goals
for goal 1
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-7
Baynet Computer
Baynet is a computer company that produces the CP4X and the
CP5Z computers. The computers use different mother boards
produced in abundant supply by the company, but use the same
cases and disk drives. The CP4X models use two DVD-RW disk
drives and no Blu-Ray disk drive whereas the CP5Z models use
one DVD-RW disk drive and one Blu-Ray disk drive.
The DVD drives and cases are bought from vendors. There are
1000 DVD-RW disk drives, 500 Blu-Ray disk drives, and 600
cases available to Baynet on a weekly basis. It takes one man-
hour to manufacture a CP4X and its profit is 0.2 bitcoin and it
takes one and one-half man-hours to manufacture a CP5Z and its
profit is 0.5 bitcoin.
For simplicity, it is assumed that the amount of computers can
have fractional values.

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-8
Baynet Computer
The company has three goals which are given below:

Goal 1: Make at least 500 total computers weekly.

Goal 2: Make a profit of at least 250 bitcoins weekly.

Goal 3: Use no more than 400 man-hours per week.

© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-9


Baynet Computer
• Decision variables
x1 = amount of CP4X computers produced weekly
x2 = amount of CP5Z computers produced weekly

di- = amount of underachievement for goal i


di+ = amount of overachievement for goal i
i=1,2,3

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-10
Baynet Computer soft
constraints

• Goals expressed as soft constraints


Goal 1: Make at least 500 total computers weekly.
x1 + x2 + d1- - d1+ = 500 min underachievement
(in the objective function, min d1-)

Goal 2: Make a profit of at least 250 bitcoins weekly.


0.2x1 + 0.5x2 + d2- - d2+ = 250 min underachievement
(in the objective function, min d2-)

Goal 3: Use no more than 400 man-hours per week.


x1 + 1.5x2 + d3- - d3+ = 400 min overachievement
(in the objective function, min d3+)
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-11
Baynet Computer
• Hard constraints
2x1 + x2 < 1000 (Availability of DVD-RW drives)
x2 < 500 (Availability of Blu-Ray drives)
x1 + x2 < 600 (Availability of cases)

• Non-negativity:
x1, x2, di-, di+ > 0 for i=1,2,3

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-12
underachivement underachivement overachivement
for goal 1 for goal 2 for goal 3

Min d1- + d2- + d3+

s.t.
2x1 + x2 < 1000 (Availability of DVD-RW
hard
drives)
constraints
x2 < 500 (Availability of Blu-Ray drives)
x1 + x2 < 600 (Availability of cases)
soft
constraints x1 + x2 +d1- -d1+ = 500 (Goal 1)
0.2x1 + 0.5x2 +d2- -d2+ = 250 (Goal 2)
x1+1.5x2 +d3- -d3+ = 400 (Goal 3)
x1, x2, d1-, d1+, d2-, d2+, d3-, d3+ > 0
based on (nonnegativity)
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-13
Equal importance
second goal is third goal is second goal is third goal is
underachieved overachieved total
decision
by 150 by 100 deviation
variable

all goals are expressed


equal as = constraints, using
importance deviation variables
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-14
Priorities among goals

• 3 ways of defining priority among goals


• Equal importance

• Weighted importance

• Prioritized or ranked goals

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-15
Weighted importance
• Goal 2 is 2 times more important than Goal 1
• Goal 3 is 5 times more important than Goal 1

• Min d1- + 2d2- + 5d3+

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-16
Weighted importance
first goal is second goal is
underachieved underachieved
by 100 by 170

weighted
importance
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-17
Ranked goals the most important,
highest-ranked

• Use priorities when it is difficult to assign weights for


deviation variables

• Rank the goals: R , R ,… 1 2

• Lower-ranked goals are considered only after higher-


ranked goals are met

• Requires the solution of a series of LP problems:


First solve for highest-ranked goal and then solve for
lower ranks one by one

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-18
Ranked goals
Baynet ranks the goals as

• Rank R : Goal 1 1

• Rank R : Goal 2 2

• Rank R : Goal 3 3

• Min R (d 1 1
-
)+ R2(d2-) + R3(d3+)
Goal 1 is the Goal 2 is the Goal 3 is the
most important second least important
goal important goal goal
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-19
Ranked goals
Min d1- Minimize rank R1 deviations

Solve and set the value of “d1- = optimal value” as


a new constraint.
Min d2- Minimize rank R 2

deviations

Solve and set the value of “d2- = optimal


value” as new constraint
Min d3+ Minimize rank R deviations
3

Solve

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-20
Ranked goals Optimal value
for d1-

consider only the


most important goal
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-21
Ranked goals Set d2- = 0 in
the next
model.

using previous consider only the


model, set d1- = 0. second important goal
based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-22
Ranked goals third goal is
overachieved by 350

using previous models,


set d1- = 0 and d2- = 0. consider only the
based on least important goal
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-23
Priorities among goals
• Equal importance

• Weighted importance

• Ranked goals

based on
© 2013 Pearson Education, Inc. publishing as Prentice Hall 6-24

You might also like