ps8
Cartes 13. Nonlinear Models: Dynamic, Goal, and Nonlinear Programming
“Thus any such storage arrangement that can be soured for less thon $18,000
cover the heaton of the project provides a cost benefit to CJM. These options
inolide the followin;
1, Find a larger storage facility that can be leased for less than $3600 per
‘month (=$18,000/5).
‘Arpaia toca gs ua per uit bach ar Gos of lem than $2000.
($18,000) per vehicle pet month
3. Utlize the storage option at the Grand Palace ‘Theater (dating the
‘months of July aad August, tain workers and lease the secure three-
vehicle garage
"The monthly eapenditares incurred under this third option are summarized
in Table TV.
‘Tapie IV. Monthly Cos Analyses (Leasing Secure Storage Facility and
Unliang the Grand Palace Theater)
| ‘Conte
Mediation Somer Ga
‘Month’ Quantity Garage Polace Produccon Fed Tinining Stonge Tou!
May. f 0 © $63,000 2/000 $10,000 $1,600 *$: 75,600,
foe Fa 0 sas $300 $0. S300 5 ae00
iy 4 re er
(oe $5 FT: Scop Sim so $410 § of000
Se o Cbs 08 0 $0 stem s 10
Foe 12 Soup _si2000 som sip00_ S000
Besides being cost effeesve; by eliminating che September production this
‘option gives CIM a one-month advsnee on its next project. We would be
pleased to provide avitional analyses to GJM should there be any changes * |
the data provided ot shoold edditional options become available.
13.5 Goal Programming
In real life, victully all problems have several objectives, not just one. For exam-
ple, ifa mamvfacturing company concentrates solely on cost containment, i might
neglect the environmental or social concerns of the community it serves? If
health care company concentrates solely on top-quality health care, its owners!
profit could be slighted. If publishing company pots too much emphasis on de
veloping an error-free textbook, the amount of time spent reviewing a potential
text might push the poblishing date beyond that required for adoptions. In shoct,
finding optimal solutions to a model formulated with a single objective can seri
cusly affer other aspirations and goals that are important to an organization.
‘A goal programming model seeks to simultaneously take into account sev-
eral objectives or goals that are of concern toa decision maker. While a linear pro-
gramming model consists of constraints and a single objective fanction to be
‘maximized oF minimized, a goal programming model consists of constraints and 3
set of goals that are prioritized in some sense.
Tn both linear snd goo! programming problems, ifthe comirains are inconsis
tent, there are no feasible solutions for the model. In goal programming, however,
‘one can expect that although there is a set of feasible solutions satisfying the co
straints, none of them may simultaneously satisfy all the conflicting gees of the13.5 Goal Programming CD159
‘organization, The objective of goal programming is to find a solution that satisfies
the true constraints and cames clesest to meeting the stated goals.
Goal programming approaches analyze how much a proposed solution devi-
ates from each stated goal. Accordingly, for each goal a pair of deviation variables
are defined (one equaling the amount by which the solution overachieves the goal;
the other equaling the amount by which it fails to meet the goal). For example,
suppose a company is considering three forms of advertising: television (X,), radio
‘%,), and newspaper (X.) Each television spot costs $3000 to ran and reaches 1000
new potential customers; each radio spot costs $800 to run and reaches 500 new
‘customers; and each newspaper ad costs $250 to run and reaches 200 new cus-
‘tomers, Suppose the company has three goals:
Goal 1: Spend no more than $25,000 on advertising
Goal 2: Reach at least 30,000 new potential customers
Goal 3: Run at least 10 television spots
JE these were constraints rather than goals, the functional constraints wonld be
‘written as
3000X, + 800%, + 250X, = 25,000
1O00X, + S00; + 200X, = 30,000
x = 10
If the third constraint, X; = 10, is satisfied, then the minimum value for the lef
side of the first constraint must be (3000)(10) = 30,040. Since this exceeds the
‘maximam limit of 25,000, no feasible sofution (with the X's = 0) exists that satis-
fies all three of these conditions simultaneously.
Tr these conditions were simply goals or targets instead of constraints, then for
each goal, i, define
U, = the amount by which the left side falls short of (under) its
right-hand side value
he amount by which the left side exceeds its right-hand side value
‘The goals can now be writen as equations:
3000X, + 800X; + 250%, + Uy ~ Ey = 25,000,
1000X, + S00X; + 200X; +U:-E = 30,000
X +U-E= 10
Here, E; (the amount spent over $25,000), Uz (the number of potential new eus-
tomers under 30,000 reached), and Us (the number of television ads below 10
sired) represent the detrimental deviecions for this problem. However, corhing
in under budget (U, > 0), reaching more chan 30,000 povential new customers
, > 0), or running more than 10 television ads (F > 0) are all indications that
the respective goals have been met. The objective of a goal programning
problem somehow involves minimizing the detrimental variables. Just how this is
done differentiates two approaches 10 solving goal programming probleins—the
nonpreemptive and the preemptive approach
NONPREEMPTIVE GOAL PROGRAMMING
In the nonpreemptive approach to goal programming, relative weights are as-
signed to the detrimental deviations. These act as a per unit penalty for failure to
‘meet a stated goal, Using these weights one can convert the goal programming
0, Ey must be 0;
and since E, > 0, Us most be 0.)FIGURE 13.5 No Feasible
Points Meet the Priority
FIGURE 13.6 Minimizing
the Priority 3 Objective
15,5 Goal Programming CD-165
Seung E, = Oand U, = 0, the goal equations become:
4X, + 10K, + Uy = 100 or Uy = 100 ~.04X, ~ 10x,
ond
2K, + 3X, — Ey = 1600 or y= 2X, + 3X; 1600
‘The priority 3 objective function (MINIMIZE 30U, + E,) can then be expressed
as: MINIMIZE 30(100 ~ .08X, ~ .10X,) + @X; + 3X; ~ 1600). Combining
the terms gives:
MINIMIZE 1400 + 8X,
Since 1400 is a constant, this function is minimized by minimizing $X,. Imposing
this objective Fonction on the feasible region obtained by satisfying the priority 2
{gives che situation shown in Figure 13.6. Here, all pints on the line X; = 400 be-
tween X; = 600 and X, = 800 are optimal. This gives V; = 8400) + 1400 = 1720.
‘Thus the feasible region for the priority 4 problem is the line segment on the line
X, = 400 from (400, 600) to (400, $00) Points on thisline have V; = 1720.CD-AG6 —Charrex 12. Nonlin:
FIGURE 13.7 Minimizing
the Priority 4 Objective
®
New England Cysts
+ Model
+ Dynamic, Goal, and Nonlinear Programming
PRIORITY 4 PROGRAM [MINIMIZE (E, + 2E,)]
[In Figure 13.1, the lines fo goal 5: 2X, + Ky = 2200, and for goal 6: X, = 900 are
aude. All solutions on the line X;'= 400 from (400, 600) to (400, 800) have
X_ < 900; hence goa 6 is satisfied, = 0) by all remaining feasible solutions.
x
2000
seo
Goal 5 (having atleast 200 tres leftover) is met by all values such that 2X, +
2X, 2200, Since all remaining feasible solutions have X; = 400, this goal is
equivalent t9 2Xp = 1400, or Xp = 700. The points on the line X; = 400 between
(400, 600) and (400, 700) satisfy this constraint.
“Thus the oprimal solution i
X, = 400
X; = any value between 600 and 700
“That is, during next month's production run the New England Cycle Company
should produce 400 B2 models and between 600 and 700 $10 models,
USING EXCEL SOLVER TO SOLVE GOAL
PROGRAMMING MODELS
“The above is a graphical illustration of preemptive goal programming. One way
Excel Solver can be used sequentially to solve the preemptive goa) programming
‘model isto manually add the new constraints at each priority level and readjust the
objective function. Although this isa bit clumsy, it avoids the need to develop and
‘use macros in Excel, which is beyond the scope ofthis text, Figures 13.80-e ilustrate
the process, This could be done on a single worksheet; however, here (and in file
[New England Cycleais) separate worksheets are used for each priority level to pre
serve the Solver dialogue box that generate the results forthe corresponding level,
‘Asilstated in Figures 13.80-r, the steps are as follows:
+ Include sections for the priority level objective functions, the functional
constraints, and the goal constrains as shown in Figure 13.84. In Solver, for
Priority 1:
© Target Cell (Min) ~The priority! objective function value (Cell Q6)
© Constraints—The funetional constraints and the priority 1 (goal 1)
constraintFIGURE 13.80 New
England Cycle Priority 1
13.5 Goal Programming CD-167
+ The result is the worksheet in Figure 13.86. This shows thatthe minimum
value for the priority 1 objective in (cell Q6) is 0. For priority 2:
© Target Cell (Min)—The priority 2 objective fanction value (Cell Q7)
© Constraints—Add to the previous constraints the priority 2 (goal 2
constraint), and a constraint requiring the objective fanction valve of the
priority 1 (cell Q6) be 0.
+ The result is the worksheet in Figure 13.8c. This shows that the minimom
value forthe priority 2 objective in (cell Q7) is also 0, For priority 3:
© Terget Cell (Min)—The priority 3 objective fanction vaiue (Cell Q8)
© Constraints—Add to the previous constraints the priority 3 (goals 3 and 4
constrains), and a constraint requiring the objective function value of the
priority 2 (cell Q7) be 0.
+ The result isthe worksheet in Figure 13.84, This shows thatthe mininmam
‘value for priority 3 objective in (cell Q8) is 1720. For priority 4
© Target Cell (Min)—The priority 4 objective function value (Cell Q9)
© Constraints—Add to the previous constraints the priority 4 (goals § and 6
constraint), and a constraint requiring the objective function value of the
priority 3 (cell Q8) be 1720,
+ ‘The result is shown in Figure 13.8. This gives one of the possible solutions—
‘manufaeworing 400 Model B2 bicycles and 700 Model C10 bicycles.
Gall CasaleFuel Cais
‘Gea? Coarse
FIGURE 13.85 New
England Cycle Priority 2
ince mains
Costa
FIGURE 13.8¢ New England
Cycle Priority 3FIGURE 13.8/ New
England Cycle Priority 4
FIGURE 12.8¢_New
England Cycle Optimal
Solution
Eo?
euede 200