Goal Programming

ƒ Firms usually have more than one goal. For example,

¾ maximizing total profit,
profit

¾ maximizing market share,

¾ maintaining full employment,
¾ providing quality ecological management,
management

¾ minimizing noise level in the neighborhood, and

¾ meeting numerous other non-economic goals.

To accompany Quantitative Analysis for 11-1 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Goal Programming

ƒ It is not possible for LP to have multiple goals unless they are all

measured in the same units (such as dollars),

¾ a highly unusual situation.

ƒ An
A iimportant technique
i that has been developed to supplement

LP is called ggoal pprogramming.

g g

To accompany Quantitative Analysis for 11-2 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Goal Programming (continued)

ƒ Goal programming “satisfices,”

¾as opposed to LP, which tries to “optimize.”
¾Satisfice means coming as close as possible to reaching
goals.

ƒ The objective function is the main difference between Goal

Programming and Linear Programming.

ƒ In Goal programming, the purpose is to minimize deviational

variables,

¾which are the only terms in11-3the objective function.

To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
Classical Linear Programming Formulation
Graphical Solution
 Example of Goal Programming
Harrison Electric Revisited

Goals Harrison’s management wants to

achieve each equal in priority:
achieve,
ƒ Goal 1: to produce as much profit above $30 as
possible during the production period.
period
ƒ Goal 2: to fully utilize the available wiring department
hours.
ƒ Goal 3: to avoid overtime in the assembly department.
ƒ Goal 4: to meet a contract requirement to produce at
least seven ceiling fans.

To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Ranking Goals with Priority Levels

A key idea in goal programming is that one goal is

more important than another.
another Priorities are assigned
to each deviational variable.

Priority 1 is infinitely
To accompany Quantitative Analysis for more important
11-8
than Priority
© 2006 2, which is
by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
infinitely
by Render/Stair/Hannamore important than the next goal, and so on.
 Profit Goal is converted to constraint

7 X1 + 6 X2 >= 30

Let’s define d1 as deviation of LHS from $30

d1 = ((7 X1 + 6 X2)) – 30

The deviation d1 can be either +ve or –ve denoted by d1+ or d1-

d1+ - d1- = 7 X1 + 6 X2 – 30

Management, 9e
The region specified by
To accompany Quantitative Analysis for first goal is shown
11-9 © 2006 by on graph
Prentice Hall, Inc.
Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Deviation below $30 (d1-) should
be minimized, because Goal is to
increase profit above $30 as much
as possible

To accompany Quantitative Analysis for 11-10 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
Wiring department has 12 hours. According to Goal 2 , all these hours
should be fully utilized.
utilized Hence the implied constraint is

2 X1 + 3 X2 >= 12

Let’s define d2 as deviation of LHS from 12 hours

d2 = (2 X1 + 3 X2) – 12

ve denoted by d2+ or d2-

The deviation d2 can be either +ve or –ve

d2+ - d2- = 2 X1 + 3 X2 – 12
To accompany Quantitative Analysis for 11-11 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
The region specified by 2nd goal is shown on graph
 Analysis of First and Second Goals

Deviation below 12 hours (d2-)

should be minimized
minimized, because
Goal is to use all the hours in
wiring department (no under-
utilization of hours in wiring)

To accompany Quantitative Analysis for 11-12 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Assembly department has 30 hours scheduled
According to Goal 3 , the consumed hours should not exceed 30.
H
Hence the
h iimplied
li d constraint
i isi

6 X1 + 5 X2 <= 30

Let’s define d3 as deviation of LHS from 30 hours

d3 = (6 X1 + 5 X2) – 30

Th deviation
The d i ti d3 can be
b either
ith +ve
+ or –ve denoted
d t d byb d3+ or d3-

d3+ - d3- = 6 X1 + 5 X2 – 30
To accompany Quantitative Analysis for 11-13 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna The region specified by 3rd goal is shown on graph
 7 X1 + 6 X2 >= 30
2 X1 + 3 X2 >= 12
6 X1 + 5 X2 <= 30

To accompany Quantitative Analysis for 11-14 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Deviation above 30 hours (d3+)
should be minimized, because
Goal is to avoid using more than
30 hours in assembly department
( over -utilization
(no tili ti off hours
h in
i
assembly)

To accompany Quantitative Analysis for 11-15 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Assembly department has 30 hours scheduled
According to Goal 4 , the ceiling fans minimum quantity
Sh ld bbe 77. H
Should Hence the
h iimplied
li d constraint
i isi

X2 >= 7

Let’s define d4 as deviation of LHS 7 units of X2

d4 = X2 – 7

Th deviation
The d i ti d4 can be
b either
ith +ve
+ or –ve denoted
d t d byb d4+ or d4-

d4+ - d4- = X2 – 7
To accompany Quantitative Analysis for 11-16 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna The region specified by 4th goal is shown on graph
 Analysis of All Four Priority Goals

Deviation below 7 (d4-) should be

minimized, because Goal is to
produce at least 7 units

To accompany Quantitative Analysis for 11-17 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Analysis of All Four Priority Goals

The constraint X2>=7 lies outside the feasible region. Excluding this constraint,
points A , B, C & D are extreme points in the feasible region.

To accompany Quantitative Analysis for 11-18 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Analysis of All Four Priority Goals

P4

Point Priorities
A (0
(0,6)
6) P3
B (3.75,1.5) P2, P3
C ((2,2.66)) P1,P2
D ( 0, 5) P1

P1 P3 P2
To accompany Quantitative Analysis for 11-19 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Solution at point A ( 0,6); priority P3

• Goal 1 (Priority 1) : Profit should exceed $30

•
• 7 X1 + 6 X2 = 7 (0) + 6 ( 6) = 36
36, Profit exceeds by $6

• Goal 2 : (Priority 2) Fully utilize wiring department’s hours of 12

•
• 2 X1 + 3 X2 = 2 (0) + 3 ( 6) = 18,

• wiring dept’s utilization exceeds by 6 hours of minimum value of 12

• Goal 3 (Priority 3) : Avoid using more than 30 hours in assembly dept

• 6 X1 + 5 X2 = 6 (0) + 5 ( 6) = 30,
30 (zero overtime)

• Goal 4 (Priority 4) : Make at least 7 ceiling fans; X2 >= 7

• X2 = 6 (does not achieve this goal)

To accompany Quantitative Analysis for 11-20 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
Solution at point B (3.75, 1.5); priority P2,P3

• Goal 1 (Priority 1) : Profit should exceed $30

•
• 7 X1 + 6 X2 = 7 (3.75)
(3 75) + 6 ( 11.5
5 ) = 35
35.3,
3 Profit exceeds by $5.3
$5 3

• Goal 2 : (Priority 2) Fully utilize wiring department’s hours of 12

•
• 2 X1 + 3 X2 = 2 (3.75) + 3 ( 1.5) = 12,

• wiring dept’s utilization exceeds by 0 hours of minimum value of 12

• Goal 3 (Priority 3) : Avoid using more than 30 hours in assembly dept

• 6 X1 + 5 X2 = 6 (3.75)
(3 75) + 5 ( 1.5)
1 5) = 30,
30 (zero overtime)

• Goal 4 (Priority 4) : Make at least 7 ceiling fans; X2 >= 7

• X2 = 1.5 (Goal falls short by 5.5 units)

To accompany Quantitative Analysis for 11-21 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Solution at point C (2, 2 2/3); priority P1,P2

• Goal 1 (Priority 1) : Profit should exceed $30

•
• 7 X1 + 6 X2 = 7 (2) + 6 (2 2/3 ) = 0,
0 Profit exceeds by $ 0

• Goal 2 : (Priority 2) Fully utilize wiring department’s hours of 12

•
• 2 X1 + 3 X2 = 2 (2) + 3 (2 2/3 ) = 12,

• wiring dept’s utilization exceeds by 0 hours of minimum value of 12

• Goal 3 (Priority 3) : Avoid using more than 30 hours in assembly dept

• 6 X1 + 5 X2 = 6 (2) + 5 (2 2/3 ) = 25 1/3, (4 2/3 hours

h are lleft
ft unutilized)
tili d)

• Goal 4 (Priority 4) : Make at least 7 ceiling fans; X2 >= 7

• X2 = 2 2/3 (Goal falls short by 4 1/3 units)

To accompany Quantitative Analysis for 11-22 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Solution at point D (0, 5); priority P1

• Goal 1 (Priority 1) : Profit should exceed $30

•
• 7 X1 + 6 X2 = 7 (0) + 6 (53 ) = 30,
30 Profit exceeds by $ 0

• Goal 2 : (Priority 2) Fully utilize wiring department’s hours of 12

•
• 2 X1 + 3 X2 = 2 (0) + 3 (5 ) = 15,

• wiring dept’s utilization exceeds by 3 hours of minimum value of 12

• Goal 3 (Priority 3) : Avoid using more than 30 hours in assembly dept

• 6 X1 + 5 X2 = 6 (0) + 5 (5 ) = 25 , (5 hours
h are lleft
ft unutilized)
tili d)

• Goal 4 (Priority 4) : Make at least 7 ceiling fans; X2 >= 7

• X2 = 5 (Goal falls short by 2 units)

To accompany Quantitative Analysis for 11-23 © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Linear Programming Formulation of
Goal Program
Step 1 :
Define objective function in terms of
Priority values of each Goal and its
corresponding deviations from the Goal
Step
p2:
Define constraint of the respective goal
with upper and lower deviation values

To accompany Quantitative Analysis for 11-24 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Linear Programming Formulation of
Goal Program
Goal 1:

Priority: P1

Negative deviation d1- should be

minimum
i i

Obj function:

P1 d1-

Related constraint:
d1+ - d1- = 7 X1 + 6 X2 – 30

To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Linear Programming Formulation of
Goal Program
Goal 2:

Priority: P2

Negative deviation d2- should be

minimum
i i

Obj function:

P2 d2-

Related constraint:
d2+ - d2- = 2 X1 + 3 X2 – 12

To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Linear Programming Formulation of
Goal Program
Goal 3:

Priority: P3

Positive deviation d3+ should be

minimum
i i

Objective function:

P3 d3+

Related constraint:
d3+ - d3- = 6 X1 + 5 X2 – 30

To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Linear Programming Formulation of
Goal Program
Goal 4:

Priority: P4

Negative deviation d4- should be

minimum
i i

Objective function:

P4 d4-

Related constraint:
d4+ - d4- = X2 – 7

To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Linear Programming Formulation of
Goal Program
Objective function:
{ Minimize prioritized deviations of all Goals }
Min P1 d1- + P2 d2- + P3 d3+ + P4 d4-

subject
j to
{ profit should exceed $30 }

7 X1 + 6 X2 – 30 = d1+ - d1-
{ wiring
g hours should be fullyy utilized}}

2 X1 + 3 X2 – 12 = d2+ - d2-
{ assembly hours should not exceed 30 hours }

6 X1 + 5 X2 – 30 = d3+ - d3-
{ x2 should be minimum 7}

X2 – 7 = d4+ - d4-

All variables are non-negative

To accompany Quantitative Analysis for
Management, 9e
© 2006 by Prentice Hall, Inc.
Upper Saddle River, NJ 07458
by Render/Stair/Hanna
Solution Methodology ( Ist Obj Function)

Solve the LP model according to Priority values in the

objective function;

Hence first
fi LP model
d l is;
i

Min d1-

subject to
7 X1 + 6 X2 – 30 = d1+ - d1-
2 X1 + 3 X2 – 12 = d2+ - d2-
6 X1 + 5 X2 – 30 = d3+ - d3-
X2 –7 = d4+ - d4-
To accompany Quantitative Analysis for © 2006 by Prentice Hall, Inc.
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
Solution Methodology ( 2nd Obj Function)

Solution of Ist LP model will yield a value for variable d1-

say ; d1- = K1

Now add the value of this variable in constraint set and solve for second priority
variable as follows;

Min d2-

subject to
7 X1 + 6 X2 – 30 = d1+ - d1-
2 X1 + 3 X2 – 12 = d2+ - d2-
6 X1 + 5 X2 – 30 = d3+ - d3-
X2 –7 = d4 + - d4-
d1- = K1
To accompany Quantitative Analysis for
Management, 9e
by Render/Stair/Hanna
Solution Methodology ( 3rd Obj Function)

Solution of 2nd LP model will yield a value for variable d2-

say ; d2- = K2

Now add the value of this variable in constraint set and solve for third priority
variable as follows;

Min d3+

subject to
7 X1 + 6 X2 – 30 = d1+ - d1-
2 X1 + 3 X2 – 12 = d2+ - d2-
6 X1 + 5 X2 – 30 = d3+ - d3-
X2 –7 = d4 + - d4-
d1- = K1
d2- = K2
Solution Methodology ( 4th Obj Function)
Solution of 3rd LP model will yield a value for variable d3+
say ; d3+ = K3

Now add
N dd th
the value
l off this
thi variable
i bl in
i constraint
t i t sett andd solve
l for
f third
thi d priority
i it
variable as follows;

Min d4-

subject to
7 X1 + 6 X2 – 30 = d1+ - d1-
2 X1 + 3 X2 – 12 = d2+ - d2-
6 X1 + 5 X2 – 30 = d3+ - d3-
X2 – 7 = d4+ - d4-
d1- = K1
d2- = K2
d3+ = K3
APPLYING METHODOLOGY USING
LINGO
APPLYING METHODOLOGY USING
LINGO
APPLYING METHODOLOGY USING
LINGO
APPLYING METHODOLOGY USING LINGO
APPLYING METHODOLOGY USING
LINGO
APPLYING METHODOLOGY USING
LINGO
APPLYING METHODOLOGY USING
LINGO
APPLYING METHODOLOGY USING
LINGO
 Interpretation of Solution

X1 = 0 , X2 = 6

OK Profit Deviation ; D1 = D1P – D1N = 6 – 0 = 6

((Profit is $6
$ more than minimum of $30)
$ )

OK Wiring Hours Deviation ; D2 = D2P – D2N = 6 – 0 = 6

(Wiring hours is 6 hours more than minimum of 12)

OK Max Assembly Hours Deviation ; D3 = D3P – D3N = 0 – 0 = 0

(Assembly hours deviation is 0; exactly 30 hours are consumed)

✮ Minimum X2 >= 7 Deviation; D4 = D4P – D4N = 0 – 1 = -1

(One lessfor unit is produced11-42
than a minimum© 2006
requirement
To accompany Quantitative Analysis by Prentice Hall, of
Inc. 7 X2 units)
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Interpretation of Solution
What is ideal Solution? Objj fn == 0 why??
y

X1 = 0 , X2 = 6

0
OK Profit Deviation ; D1 = D1P – D1N = 6 – 0 = 6
((Profit is $6
$ more than minimum of $30)
$ )

OK Wiring Hours Deviation ; D2 = D2P – D2N = 6 – 0 = 6

(Wiring hours is 6 hours more than minimum of 12)

OK Max Assembly Hours Deviation ; D3 = D3P – D3N = 0 – 0 = 0

(Assembly hours deviation is 0; exactly 30 hours are consumed)

Minimum X2 >= 7 Deviation; D4 = D4P – D4N = 0 – 1 = -1

(One
To accompany Quantitative lessfor unit is produced11-43
Analysis than a minimum© 2006
requirement
by Prentice Hall, of
Inc. 7 X2 units)
Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Goal Programming Versus Linear Programming

ƒ Multiple goals (instead of one goal)

ƒ Deviational variables minimized (instead of
maximizing profit or minimizing cost of LP)
ƒ “Satisficing” (instead of optimizing)
ƒ Deviational
D i i l variables
i bl are reall (and
( d replace
l slack
l k
variables)

To accompany Quantitative Analysis for 11-44 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming

ƒ Nonlinear objective function, linear constraints

ƒ Nonlinear objective function and nonlinear
constraints
ƒ Linear objective function and nonlinear
constraints

To accompany Quantitative Analysis for 11-45 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming

ƒ Nonlinear objective function, linear constraints

Max: 28X1 + 21X2 + 0.25X22

Subject to:
X1 + X2 ≤ 1000
0.5X1 + 0.4X2 ≤ 500

To accompany Quantitative Analysis for 11-46 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming

ƒ Nonlinear objective function and nonlinear

constraints.
constraints
Max: 13X1 + 6X1X2 + 5X2 + X2–1
S bj t to:
Subject t
2X12+ 4X22 ≤ 90
X1 + X23 ≤ 75
8X1 – 2X2 ≤ 61

To accompany Quantitative Analysis for 11-47 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming

The Problem has both Nonlinear Objective Function and

Nonlinear Constraints.
The solution to Great Western Appliance’s NLP Problem using
Excel Solver:

MAX= 28 * X1 + 21 * X2 + 00.25
25 * X2
X2^2;
2;
!subject to;
X1 + X2 <= 1000;
0 5* X1 + 0.4
0.5* 0 4 * X2 <= 500
500;

To accompany Quantitative Analysis for 11-48 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming

The problem has both Nonlinear Objective Function and

Nonlinear Constraints.
Constraints
Excel Solution to the Hospicare Corp.’s NLP Problem
using Solver:

Objective value: 271000.0

Variable Value Reduced Cost
X1 0.000000 493.0001
X2 1000.000 0.000000

To accompany Quantitative Analysis for 11-49 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming
ƒ Linear objective function and nonlinear
constraints

Max: 5X1 + 7X2

Subject to:
3X1+ 0.25X12 + 4X2 + 0.3X22 ≥ 125
13X1 + X13 ≥ 80
0.7X1 + X2 ≥ 17

To accompany Quantitative Analysis for 11-50 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming
Min = 5 * X1 + 7 * X2;
!S bj to;
!Subject
3 * X1 + 0.25 * X1 ^ 2 + 4 * X2 + 0.3 * X2 ^ 2 >= 125;
13 * X1 + X1 ^ 3 >= 80;
0.7 * X1 + X2 >= 17;

To accompany Quantitative Analysis for 11-51 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Nonlinear Programming

Objective value: 119.3325

Variable Value Reduced Cost

X1 3.325326 0.000000
X2 14.67227 0.000000

Row Slack or Surplus Dual Price

1 119.3325 -1.000000
2 11.01218 0.000000
3 0.4546953E-05 -0.2165749E-02
4 0
0.000000
000000 -7
7.000000
000000

To accompany Quantitative Analysis for 11-52 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna
 Computational Procedures -
Nonlinear Programming
ƒ Gradient method (steepest descent)
ƒ Separable programming - linear representation of
nonlinear problem
ƒ Separable programming deals with a class of
problems in which the objective and constraints are
approximated by linear functions. In this way, the
powerful simplex algorithm may again be applied.
ƒ In general, work in the area of NLP is the least charted
and most difficult of all the quantitative analysis
models.

To accompany Quantitative Analysis for 11-53 © 2006 by Prentice Hall, Inc.

Management, 9e Upper Saddle River, NJ 07458
by Render/Stair/Hanna