Sei sulla pagina 1di 19

Goal Programming

Multiple Objectives and Minimization of Goal Deviational Variables

Goal Programming
In LP, only one objective was maximized or minimized. However, there are instances which involve multiple objectives like Productivity, Quality and Profit. These may not be reduced to a common unit of measure. Goal Programming (GP) is a variation of LP, used for problems with multiple objectives

Goal Programming (GP)


Relates to multiple objectives Goals are expressed as constraints. Goal constraints specify desirable level of performance. Constraints include hard (non goal) and soft (goal) ones.. Solution must satisfy the hard (non goal) constraints. Non goal constraints may not be satisfied. Deviations from goals are minimized.

GP deals with Satisficing


When one or more goals are not achieved by a solution, it is because there are conflicts either between goals or between goals and hard constraints. One constraint puts X1 = 10, while another puts X1 = 20. In LP, there would not be any feasible solution. In GP, deviations from the goals are permitted to obtain a solution. In GP, the objective is to satisfy the hard constraints and achieve reasonably acceptable levels for goal constraints. This is SATISFICING.

Deviation Variables
Deviation variables represent the differences between actual performance and target performance. Two kinds of deviations under achievement (actual < target) and over achievement (actual > target). Both under achievement (u) and over achievement (v) variables are added to each goal constraint. 4X1 + 2X2 + u1 v1 = 100 labor hours If 4X1 + 2X2 = 80, then u1 = 20 (under achievement) If 4X1 + 2X2 = 110, then v1 = 10 (over achievement) If 4X1 + 2X2 = 100, then u1 = 0, v1 = 0. One of the deviation variables has to be zero.

Model Formulation
Identify the decision variables Identify the constraints and determine which ones are goal constraints. Formulate the non goal (hard) constraints. Formulate the goal (soft) constraints. Formulate the objective function Add the non negativity constraints Set the priority of goals

Goal Programming Example


Minimize P1u1, P2v1, P3U2 Subject to : 4X1 + 2X2 <= 40 (non goal) 2X1 + 6X2 <= 60 (non goal) 3X1 + 3X2 + u1 v1 = 75 (goal) X1 + 2X2 + u2 v2 = 50 (goal) X1, X2, u1, v1, u2, v2 >= 0 P1 = first priority, P2 = second priority, P3 = third priority.

Goal Programming Example


A company manufactures 3 products X1, X2, X3. Material availability is 600 pounds and X1, X2, X3 require 2, 4 and 3 units respectively. Assembly time for X1,X2,X3 are 9, 8, 7 mins respectively with total availability of 900 mins. The packing time for X1,X2,X3 are 1, 2 and 3 min respectively with total availability of 300 minutes. The manager has the following objectives in order of priority : 1. Minimize overtime in the assembly department 2. Minimize undertime in the assembly department 3. Minimize both overtime and undertime in the packaging dept. Material : 2X1 + 4X2 + 3X3 <= 600 Assembly : 9X1 + 8X2 + 7X3 + u1 v1 = 900 mins. Packaging : 1X1 + 2X2 + 3X3 + u2 v2 = 300 mins Minimize P1v1, P2u1, P3(u2+v2) X1,X2,X3,u1,v1,u2,v2 >= 0

GP Graphical
Minimize P1u1, P2u2, P3u3 Subject to : 5X1 + 3X2 <= 150 (A) 2X1 + 5X2 + u1 v1 = 100 (1) 3X1 + 3X2 + u2 v2 = 180 (2) X1 + u3 v3 = 40 (3) X1, X2, u1, v1, u2,v2, u3, v3 >= 0

X2

50

Hard A 5X1 + 3X2 <= 150

Feasible Solution space

30

X1

X2

50

A(0,50) , u1 = 0 B(0,20), u1 = 0 C(22,10), u1 = 6 Hard A 5X1 + 3X2 <= 150

Acceptable 20 B Region

P1 C
Soft 1 2X1 + 5X2 + u1 v1 = 100

30

50

X1

X2 60 50 A P2

A(0, 50) , u1 = 0, u2 = 30 B(0, 20), u1 = 0, u2 = 120 Priority 2 goal does not Overlap the feasible Solution space Priority goal achievement to remain unaltered

Hard A

Acceptable 20 B Region

P1 C Soft 1

Soft 2 3X1 + 3X2 + u2 v2 = 180

30

50

60

X1

X2 60 50 A P2

A(0, 50) , u1 = 0, u2 = 30, u3 = 40 Optimum B(0,20), u1 = 0, u2 = 120, u3 = 40

Hard A

Soft 3 X1 + u3 v3 = 40
P3 Priority goal 3 constraint does not overlap the feasible Solution space Priority 1 & 2 goals achievement unaltered

Acceptable 20 B Region

P1 C
Soft 2

Soft 1

30

40

50

60

X1

Goal Programming Problem


X1,X2,X3 = Qty of Product 1, 2 and 3 Max profit Z = 4X1 + 2X2 + 3X3 Subject to : Labor : 2X1 + x2 + 4X3 <= 160 hours Storage : X1 + 2X2 + 3X3 <= 150 sqft Product 1 : X1 >= 10 units X1,X2,X3 >= 0 Convert this LP into a GP with the following priorities order : Minimize the underutilization of labor Achieve a satisfactory profit level of $300. Try to avoid making less than 10 units of product 1.

Goal Programming Problem Formulation


X1,X2,X3 = Qty of Product 1, 2 and 3 u1, u2, u3 = underachievement of Goals 1, 2 and 3 v1, v2, v3 = overachievement of Goals 1, 2 and 3 P1, P2, P3 = First, second and third priority Minimize P1u1, P2u2, P3u3 Subject to : Storage : X1 + 2X2 + 3X3 <= 150 sqft Labor : 2X1 + X2 + 4X3 + u1 v1 = 160 hours (First Goal) Profit : 4X1 + 2X2 + 3X3 + u2 v2 = 300 (Second Goal) Product 1 : X1 + u3 v3 = 10 units (Third Goal) X1,X2,X3, u1,u2,u3,v1,v2,v3 >= 0 Solve for min of u1, then for min of u2, followed by for min of u3

Goal Programming (Solver)


Minimize P1u1,P2u2,P3u3,P4u4,P5v1 2X1 + 5X2 + u1 v1 = 100 (Goal, Labor hours) 3X1 + 3.6X2 + u2 v2 = 180 (Goal, M/C hours) X1 + u3 v3 = 40 (Goal, Demand for Product 1) X2 + u4 v4 = 50 (Goal, Demand for Product 2) X1,X2, u1,u2,u3,u4,v1,v2,v3,v4 >= 0 Goal 1 : Fully utilize Labor hours Goal 2 : Fully utilize M/C hours Goal 3 : Production of X1 = 40 Goal 4 : Production of X2 = 50 Goal 5 : Minimize labor overtime

Other Topics
Multiple Objective Linear Programming (Step Method)

Multiple Objective Linear Programming (Step Method)


Basically a LP Model with Multiple Objectives. The constraint set is solved for each objective. A pay-off Matrix is constructed. The Decision Maker relaxes one of the satisfied objective. There may be an improvement in another objectives. The solutions are shown to the DM. This process goes on till the DM is satisfied. A preferred solution is obtained

MOLP Example
Media Selection Problem One objective Minimize Cost with a minimum Target audience to reach Another Objective Maximize Target Audience with a maximum budget (Cost) available. Solve the same problem with one objective at a time. Check the audience per unit cost outlay. Arrive at a preferred solution

Potrebbero piacerti anche