Professional Documents
Culture Documents
Uploaddocument 2235
Uploaddocument 2235
Uploaddocument 2235
CHAPTER 10
Integer Programming, Goal Programming, and
Nonlinear Programming
TEACHING SUGGESTIONS
Teaching Suggestion 10.1: Topics in This Chapter.
The overall purpose of this chapter is to provide a framework for the topics of integer program-
ming, nonlinear programming and goal programming. These are fairly advanced topics in a
mathematical sense, and the chapter’s intention is solely to introduce them through a series of
simple graphical problems. Some of the topics are on the cutting edge of QA. For example, in
integer and nonlinear programming, no one solution procedure exists to handle all problems.
Teaching Suggestion 10.2: Using the Computer to Solve Mixed-Integer Programming Problems.
Note that the Excel printout in Program 10.2 allows users to specify which variables are integers
and which, by default, can be fractional.
Teaching Suggestion 10.3: Multiple Goals.
Ask students what other goals a company might have beyond maximizing profit. Socially con-
scious firms need to state as their mission a whole series of objectives. Encourage students to
research an article showing a goal programming application. There is a wealth of research in
journals.
Teaching Suggestion 10.4: Deviational Variables Are the Key in Goal Programming.
The concept of deviational variables requires careful explanation to the class. Students are accus-
tomed to the decision variables of X1 and X2. Now they need to concentrate on goal achievement.
The minus and plus signs on deviational variables need a thoughtful classroom discussion.
Teaching Suggestion 10.5: Difficulty of Graphical Goal Programming.
Solving goal programming problems graphically can be a confusing concept relative to graphical
LP. Students often have difficulty with the direction of deviational variables.
10-1
Copyright ©2015 Pearson, Inc.
10-2
Copyright ©2015 Pearson, Inc.
10-5. Satisficing is a term used in goal programming because it is often not possible to “opti-
mize” a multi-goal problem. We come as close as possible to reaching goals.
10-6. Deviational variables in LP, are the difference between set goals and the current solution.
In LP problems, only “real” variables are used, representing physical quantities. This is discussed
in Section 10.4.
10-7. A college president’s goals might be to (1) increase enrollments by 1,000 students; (2) stay
within budget; (3) keep class sizes down to an average of 25 students; (4) increase faculty sala-
ries; (5) develop 10 new online courses; (6) reduce average teaching loads to three courses per
semester, and so on. There will be financial, space, tenure, and many other constraints.
10-8. Ranking goals just means more weight can be placed on one goal over another. The high-
er-ranked goals must be achieved completely before goal programming moves on to meet lower-
ranked goals.
10-9. a. Linear
b. Nonlinear because of 8X1X2 in objective
c. Goal programming
d. Nonlinear because X12 in first constraint
e. Nonlinear and quadratic objective function
10-10. a. Let X = number of prime time ads per week
Y = number of off-peak ads per week
Maximize audience exposure = 8200X + 5100Y
Subject to:
390X + 240Y 1800
X2
Y6
X, Y 0
Solution: X = 2; Y = 4.25; audience = 38,075
b. X = 2, Y = 4; audience = 36,800
There are other good solutions.
c. Optimal integer solution:
X = 4, Y = 1; audience = 37,900
10-3
Copyright ©2015 Pearson, Inc.
10-11. a. Let Xi = 1 if item i is selected and 0 otherwise, for i = 1 to 8.
Maximize utility 80X1 + 20X2 + 50X3 + 55X4 + 50X5 + 75X6 + 30X7 + 70X8
Subject to: 8X1 + X2 + 7X3 + 6X4 + 3X5 + 12X6 + 5X7 + 14X8 35
Xi = 0, 1
Solution using QM for Windows Integer and Mixed-Integer Programming Module:
X1 = X2 = X4 = X5 = X6 = X7 = 1
X3 = X8 = 0
Objective function = 310
b. Add constraint X5 – X3 0
X1 = X3 = X5 = X6 = X7 = 1
X2 = X4 = X8 = 0
Objective function = 285
10-12. X1 = number of larger posters
X2 = number of smaller posters
Maximize profit = 3X1 + 2X2
subject to X1 3
X2 5
2X1 + X2 10
X1, X2 0
See graph below.
10-4
Copyright ©2015 Pearson, Inc.
Step 1. Optimal LP solution at a is (X1 = 21/2, X2 = 5, profit = $17.50). Step 2. Integer solution at
b is (X1 = 3, X2 = 4, profit = $17). Integer solution at c is (X1 = 2, X2 = 5, profit = $16). Hence the
optimal integer solution is X1 = 3 large posters and X2 = 4 small posters (seen at point b).
10-13. X1 = number of Boeing 757s purchased
X2 = number of Boeing 767s purchased
Maximize passenger carrying capability = 125,000X1 + 81,000X2
subject to
80X1 + 110X2 1,600 ($ million available)
[X1 1
(X1 + X2)] or
3
2
X1 – 1 X2 0 (one-third 757s)
3 3
10-5
Copyright ©2015 Pearson, Inc.
10-15. Let Xi = 1 if location i is selected and 0 otherwise, for i = 1 to 6.
Minimize X1 + X2 + X3 + X4 + X5 + X6
Subject to:
X1 + X6 1
X1 + X2 1
X2 + X3 1
X1 + X3 + X6 1
X3 + X4 + X5 1
X3 + X4 + X6 1
X2 + X5 1
X4 + X5 1
Xi = 0, 1 for i = 1 to 6.
Solution using QM for Windows Mixed-Integer Programming Module:
X1 = X2 = X4 = 1; all other variables = 0. Objective function value = 3
This means only locations 1, 2, and 4 will be used.
10-16.
1 if location i is selected
a. Let: X i =
0 if location i is not selected
Maximize profit = $5,000X1 + 6,000X2 + 10,000X3
+ 12,000X4 + 8,000X5 + 3,000X6
+ 9,000X7 + 10,000X8
subject to $60,000X1 + 50,000X2 + 82,000X3
+ 103,000X4 + 50,000X5
+ 41,000X6 + 80,000X7
+ 69,000X8 $300,000
b. X1 = 0, X2 = 1, X3 = 1, X4 = 0, X5 = 1, X6 = 1, X7 = 0, X8 = 1, Profit = $37,000
10-6
Copyright ©2015 Pearson, Inc.
10-17. a. Let X1 = 1 if apartment project is undertaken; 0 otherwise
Let X2 = 1 if shopping center project is undertaken; 0 otherwise
Let X3 = 1 if mini-warehouse project is undertaken; 0 otherwise
Maximize NPV = 18X1 + 15X2 + 14X3
Subject to:
40X1 + 30X2 + 20X3 80
30X1 + 20X2 + 20X3 50
X1, X2, X3 = 1 or 0
b. The optimal solution is X1 = 1, X2 = 1, X3 = 0. NPV = 33. This means that both the apartment
project and the shopping center project will be undertaken. The amount of money spent in year 1
would be $70 (thousand) and in year 2 would be $50 (thousand).
10-18. a. X1 X2 This means that if the apartment is not built (X1 = 0), the shopping center can-
not be built (X2 must equal 0).
b. X1 + X2 + X3 = 2
10-19. a. Let Xij = 1 if generator i is functioning during time period j, and 0 otherwise; where i =
1, 2, 3 and j = 1 for 6–2 time period; j = 2 for 2–10 time period; j = 3 for 6–10 time period.
Let Yij = megawatts produced by generator i in time period j, where i = 1, 2, 3 and j = 1 for 6–2
time period; j = 2 for 2–10 time period.
Minimize cost = 6,000(X11 + X12 + X13) + 5,000(X21 + X22 + X23) + 4,000(X31 + X32 + X33) + 8(Y11
+ Y12) + 9(Y21 + Y22) + 7(Y31 + Y32)
Subject to:
Y11 + Y21 + Y31 3,200 megawatts requirements from 6–2
Y12 + Y22 + Y32 5,700 megawatts requirements from 2–10
Y11 2,400(X11 + X13) maximum megawatts from #1 from 6–2
Y12 2,400(X12 + X13) maximum megawatts from #1 from 2–10
Y21 2,100(X21 + X23) maximum megawatts from #2 from 6–2
Y22 2,100(X22 + X23) maximum megawatts from #2 from 2–10
Y31 3,300(X31 + X33) maximum megawatts from #3 from 6–2
Y32 3,300(X32 + X33) maximum megawatts from #3 from 2–10
X11 + X12 + X13 1 generator #1 starts up at most once
X21 + X22 + X23 1 generator #2 starts up at most once
X31 + X32 + X33 1 generator #3 starts up at most once
Xij = 0 or 1 for all i, j
Yij 0 for all i, j
10-7
Copyright ©2015 Pearson, Inc.
b. The solution is: X12 = 1, X33 = 1, Y12 = 2,400, Y31 = 3,200, Y32 = 3,300, total cost = $74,700.
Thus, generator #1 will be utilized in the period 2–10 and will generate 2,400 megawatts of elec-
tricity. Generator #3 will be started at 6 and utilized for the entire 16 hours. It will generate 3,200
megawatts during the 6–2 time period, and 3,300 megawatts during the 2–10 time period.
10-20. Let T = number of TV ads, R = number of newspaper ads, B = number of billboard ads,
and N = number of newspaper ads.
Minimize P1d1– + P2d2+ + P3d3– + P4d4+ + P4d5+ + P4d6+ + P4d7+
Subject to:
(1) number of people reached
40,000T + 32,000R + 34,000B + 17,000N – d1+ + d1– = 1,500,000
(2) budget
900T + 500R + 600B + 180N – d2+ + d2– = 16,000 budget constraint
(3) number of TV or radio ads
T + R – d3+ + d3– = 6
(4) restriction on number of each individual type of ad
T – d4+ + d4– = 10
R – d5+ + d5– = 10
B – d6+ + d6– = 10
N – d7+ + d7– = 10
All variables 0
b. T = 0, R = 0.73, B = 0, N = 86.86 (These are rounded to two decimals)
c. Goal 1 (number of people reached) and goal 2 (budget) are met completely. The number of
TV, radio, and billboard ads are each less than 10. The other goals are not met.
10-8
Copyright ©2015 Pearson, Inc.
10-21. Let: X1 = number of two-drawer cabinets produced each week
X2 = number of three-drawer cabinets produced each week
d1– = underachievement of profit goal
d1+ = overachievement of profit goal
d2– = idle time in production capacity
d3– = underachievement of sales goal for two-drawer files
d4– = underachievement of sales goal for three-drawer files
Minimize deviations
= P1d1– + P1d1+ + P2d2– + P3d3– + P3d4–
subject to
10X1 + 15X2 + d1– – d1+ = $11,000 (profit target)
1X1 + 2X2 + d2 – = 1,300 hours (production limit)
1X1 + d3 – = 600 (two-drawer sales limit)
X2 + d4 – = 400 (three-drawer sales limit)
All Xi, di variables 0
10-22. Because we want to achieve the profit goal as closely as possible (minimize both d1– and
d1+), the line ABC becomes the feasible region. When the P2 priority is included, the feasible re-
gion is reduced to the segment AB. P3 priority applies to both d3– and d4–. The three-drawer goal
(d4–) is fully attained at point B and the two-drawer goal (d3–) is almost reached.
Using QM for Windows to solve this, we get the best solution: X1 = 500, X2 = 400. The value
of d3– = 100, meaning the two-drawer sales goal is underachieved by 100 cabinets. See the graph
below.
10-9
Copyright ©2015 Pearson, Inc.
10-23. X1 = number of 64MB chips produced
X2 = number of 256MB chips produced
X3 = number of 512MB chips produced
d1– = underfilling customer’s order of 64MB chips
d2– = underfilling customer’s order of 256MB chips
d3– = underachievement of sales quotas for 64MB chips
d4– = underachievement of sales quotas for 256MB chips
d5– = underachievement of sales quotas for 512MB chips
d6– = underutilization of plant capacity
Minimize deviations
= P1d1– + P1d2– + P2d3– + P2d4– + P2d5– + P3d6–
subject to
X1 + d1– – d1+ = 30 (64MB chips order)
X2 + d2– – d2+ = 35 (256MB chips order)
–
X1 + d3 – d3+ = 40 (64MB sales goal)
–
X2 + d4 – d4+ = 50 (256MB sales goal)
–
X3 + d5 – d5+ = 60 (512MB sales goal)
8X1 + 13X2 + 16X3 + d6– = 1,200 (hours capacity)
All variables 0
10-24 The best solution found using the computer is
X1 = 15
X2 = 20
d1+ = 30
10-10
Copyright ©2015 Pearson, Inc.
10-25. a.
d1– = underachievement of class and study goal
d1+ = overachievement of class and study goal
d2+ = overachievement of sleeping goal
d3– = underachievement of social time goal
Major Bligh’s objective function becomes
minimize = d1– + d1+ + d2+ + d3–
subject to constraints (per week)
1X1 + 1X2 + 1X3 + 1X4 168
1X3 + d1– – d1+ = 30
1X1 – d2+ = 49
1X4 + d3– = 20
All variables 0
Since the goals have priority, they can be rewritten in this order, yielding to the absolute comple-
tion of each goal before attempting to achieve the next goal. The objective function would be-
come
Minimize = P1d1– + P1d1+ + P2d2+ + P3d3–
where P1 = meet class and study goal
P2 = meet sleeping goal
P3 = meet socializing goal
b. X1 = 49
X2 = 69
X3 = 30
X4 = 20
All goals are fully met.
10-11
Copyright ©2015 Pearson, Inc.
10-26. a. Let S = dollars invested in stocks; B = dollars invested in bonds;
R = dollars invested in real estate
Minimize d1– + d2– + d3+
Subject to
0.13S + 0.08B + 0.10R + Return is at least 10%
d1– – d1+ = 25,000
B + d2– – d2+ = 75,000 Amount in bonds is at least 30%
R + d3– – d3+ = 0.50(S + B) Real estate should not exceed half of stocks and bonds
S + B + R = 250,000
S 150,000
150,000
P 150,000
S, , P 0
b. S = $50,000 invested in stocks,
B = 75,000 invested in bonds;
R = $125,000 invested in real estate.
The total return is $25,000 (10%). The amount invested in real estate ($125,000) is more than
half of the amount invested in stocks and bonds ($125,000). This is the only goal that is not met.
10-12
Copyright ©2015 Pearson, Inc.
10-27. Maximize profit = X1(1,800 – 50X1)
+ X2(2,400 – 70X2)
= 1,800X1 – 50X21
+ 2,400X2 – 70X22
X1, X2 0
subject to 100X1 + 130X2 5,000 hours
X1, X2 0
10-28. Let X1 = no. of XJ6s and X2 = no. of XJ8s
subject to X1 + 2X2 = 40
X1, X2 0
b. X1 = 18.3; X2 = 10.8; revenue = $70,420
10-29. The optimal solution found using Solver in Excel is X = 62.73, Y = 8.64, Profit =
$720.41.
10-30. The optimal solution found using Solver in Excel is X = 0.333, Y = 0.667, with a variance
of 0.102 and a return of 0.09.
10-31. a. Total profit = (P1–6)X1 + (P2–8)X2
b. The optimal solution found using Solver in Excel is X1 = 260, X2 = 140, P1 = 20, P2 =
17.33, profit = $4,946.67.
10-32. a. Z = $665,000
Variable Value
X1 0
X2 0
X3 0
X4 0
X5 (South Orlando) 1
X6 0
X7 0
X8 (Apopka) 1
X9 (Lake Mary) 1
X10 (Cocoa Beach) 1
b. The expected return drops to $625,000. Osceola opens and Cocoa Beach closes.
10-13
Copyright ©2015 Pearson, Inc.
c. As seen below, with Apopka corrected, the new solution has a return of $635,000 but the
same locations as part a.
Solution:
Z = $635,000
Variable Value Variable Value
X1 0 X6 0
X2 0 X7 0
X3 0 X8 1
X4 0 X9 1
X5 1 X10 1
10-33. Solving this with Excel after changing the objective to maximization, we get a smoothing
constant of 0. The maximum MAD is 28.17 (ignoring the error in time period 1).
10-34. Excel is used to solve this. The optimal values for w1 and w2 are 0.3478 and 0.6522 as
shown below. The MAD is 11.504.
10-35. Maximize return = 50X1 + 100X2 + 30X3 + 45X4 + 65X5 + 20X6 + 90X7 + 35X8
subject to 500X1 + 1,000X2 + 350X3 + 490X4
+ 700X5 + 270X6 + 800X7 + 400X8 3,000
X1 + X2 + X3 + X4 + X5 + X6 + X7 + X8 5
X1 + X2 1
X3+ X4 + X5 2
X6+ X7 + X8 2
All Xi = 0 or 1
The solution is X1 = 0; X2 = 1; X3 = 1; X4 = 1; X5 = 0; X6 = 1; X7 = 1; X8 = 0; return = 285.
10-14
Copyright ©2015 Pearson, Inc.
10-36. Define the variables Ai, Bi, Ci, Di, to be 1 if job A, B, C, or D respectively is assigned to
machine i, where i = 1, 2, 3, and 4. The integer program is:
Minimize cost = 85A1 + 70A2 + 60A3 + 10A4 + 6B1 + 15B2 + 90B3 + 76B4 + 50C1 + 80C2 + 5C3 +
75C4 + 75D1 + 84D2 + 82D3 + 25D4
Subject to:
A1 + A2 + A3 + A4 = 1
B1 + B2 + B3 + B4 = 1
C1 + C2 + C3 + C4 = 1
D1 + D2 + D3 + D4 = 1
A1 + B1 + C1 + D1 = 1
A2 + B2 + C2 + D2 = 1
A3 + B3 + C3 + D3 = 1
A4 + B4 + C4 + D4 = 1
All variables = 0 or 1
Using Excel, we get two optimal solutions:
Assignment Cost Assignment Cost
A4 $ 10 A4 $ 10
B1 6 B2 15
C3 5 C3 5
D2 84 D1 $175
Cost $105 Cost $105
10-15
Copyright ©2015 Pearson, Inc.
10-38. The first two priorities are fully satisfied. The best solution is
X1 = 10 TV spots
X2 = 35 newspaper ads
Total exposure here is 8,250,000 people, so d4– = 750,000 people. In other words, the exposure
goal was underachieved by 3 million people.
4
Subject to:
(1) use 240 hours per week
All variables 0
10-40. The constraints are the same as in Problem 10-39. The objective function changes to:
Minimize d1− + d1+ + 0.5 d1− + 0.5 d3− + 0.5 d 4− + 0.333 d5−
10-16
Copyright ©2015 Pearson, Inc.
SOLUTION TO SCHANK MARKETING RESEARCH CASE
1. The first part of this case is an assignment problem that can be formulated with LP. A dummy
project manager can be added to create a balanced 4 4 cost matrix.
n m
Minimize = C
j =1 i =1
ij X ij
10-17
Copyright ©2015 Pearson, Inc.
This constraint looks at the deviation of d8+ from 0. In other words, the closer d8+ is to 0 (not as-
signing Gardener to CBT), the closer it comes to meeting the restriction.
Ruth earns $3,000 or more:
2,700X21 + 3,200X22 + 3,000X23
+ 3,100X24 + d9– – d9+ = $3,000
Here d9– represents underachievement of the goal, while d9+ is overachievement. The coefficients
are the costs per assignment.
Total costs:
( C X ij ) − d10 + = 0
4 3
ij
j =1 i =1
This attempts to minimize total cost, bringing it as close to zero as possible; d10+ is the deviation
from the goal.
Objective function:
minimize Z = P1d2– + P2d8+ + P3(2d1– + d3–) + P4d9– + P5d10+
For a given set of requirements, the smallest number of toll collectors that will meet them can be
obtained from the following integer linear programming problem:
minimize Z = X1 + X2 + X3 + X4 + X5 + X6 + X7
subject to
X1 + X2 + X3 + X4 + X5 + X6 + X7 R5
X1 + X2 + X3 + X4 + X5 R5
X2 + X3 + X4 + X5 + X6 R6
X3 + X4 + X5 + X6 + X7 R7
X1 + X4 + X5 + X6 + X7 R1
X1 + X2 + X5 + X6 + X7 R2
X1 + X2 + X3 + X6 + X7 R3
X1 + X2 + X3 + X4 + X7 R4
All variables 0
where Xj is the number of collectors starting on day j (j = 1 is Sunday) and Rj is the number re-
quired on day j.
10-18
Copyright ©2015 Pearson, Inc.
Solution Manual for Quantitative Analysis for Management, 12/E Barry Render, Ralph M. Stair,
1. The following table summarizes the requirements for shifts A, B, and C for each of the seven
days of the week along with the allocations that yield the minimum numbers of collectors start-
ing each shift: 18 for shift A, 16 for shift B, and 18 for shift C.
Toll Collector Requirements for Oakton River Case
Shift A B C Mix
DAY Req. Start Req. Start Req. Start Req. Start
Sun. 8 0 10 0 15 5 33 3
Mon. 13 3 10 1 13 2 36 9
Tue. 12 5 10 5 13 1 35 8
Wed. 12 0 10 1 12 4 34 6
Thu. 13 5 10 5 12 1 35 9
Fri. 13 1 13 1 13 5 39 7
Sat. 15 4 15 13 8 0 38 8
Total 18 16 18 50
Note: Alternative optimal solutions for each shift may be possible.
2. If mixing of shifts is allowed, the daily requirements become the sum of the daily shift re-
quirements, as shown in the second part of the table. The minimum number of collectors starting
each day is shown in the last Start column. The total 50 is a reduction of two from the total re-
quired without allowing for the mixing of shifts.
10-19
Copyright ©2015 Pearson, Inc.
KANSAN POIKA.
»Entä Rosenstein»?
»Hyvän nimenikö»?
»Miksi sitten istutte siinä ja raautte kuin joku pahanilman lintu? Jos
teillä on hyviä uutisia, niin miksi kujeilette kanssani pilkkaamalla
onnettomuuttani?»
»Vaan?»
»Herra kreivi!»
Miten András kulutti lopun päivästä, ei hän olisi voinut sanoa. Tiet
olivat niin ravalla, että Csillag voi vain vaikeasti löytää tien mudasta.
Kumminkin ymmärsi se isäntänsä surun, sillä se harhaili hänen
kanssaan pustalla kauan vielä senkin jälkeen kuin illan varjot olivat
kietoneet tasangon vaippaansa. Kaukaa kuuluva Tarnan kohina
lisäsi seudun luonteenomaista surullisuutta ja autiutta. Oli jo
myöhäinen, kun András vihdoin saapui rauhalliseen kotiinsa, missä
Etelka kehruuksiensa ääressä odotti levottomasti poikaansa. Kun
hän kuuli Csillagin kavioiden kapseen kartanolta, meni hän ovelle ja
katsoi pimeässä, miten András hoiti lemmikkiään ja vei sen sitten
mukavaan pilttuuseen yöksi. András ei nähtävästi ollut huomannut
äitiään, sillä silloin olisi hän ensimmäiseksi tervehtinyt häntä. Kun
Etelka huomasi, miten raskaasti ja hitaasti András lähestyi
rakennusta, tunsi hän kuvaamatonta surua. András pysähtyi
puutarhaan tuon ihanan ruusupuun juurelle, jossa oli paljon pieniä,
ihanaa kesäkuuta ennustavia nuppuja, ja Etelka ihmetteli, miksi
hänen poikansa ensin koski kädellään jokaiseen nuppuun ja
kumartui sitten suutelemaan niitä.
XX
VASTAUS.