Books 3337 0 0 (1361-1400)

The RISKTRIGEN Function
See Figure 13. Sometimes we want to use a triangular random variable, but we are not
sure of the absolute best and worst possibilities. We may believe that there is a 10%
chance that market share will be less than or equal to 30%, that the most likely share is
40%, and that there is a 10% chance that share will exceed 75%. The RISKTRIGEN func-
tion is used in this situation. The formula
RISKTRIGEN(.3,.4,.75,10,90)
would be appropriate for this situation. Then @Risk draws a triangle that yields
■ A 10% chance that market share is less than or equal to 30%. This requires a
worst possible market share of around 20%.
■ A most likely market share of 40%.
■ A 10% chance that market share is greater than or equal to 75%. This requires a
best possible market share of around 95%.
Again, the probability of a market share between 20% and 50% is just the area under the
triangle between 20% and 50%.
The RISKUNIFORM Function

See Figure 14. Suppose a competitor’s bid is equally likely to be anywhere between 10 and
30 thousand dollars. This can be modeled by a uniform random variable with the formula
RISKUNIFORM(10,30)
Again, this function makes any bid between 10 and 30 thousand dollars equally likely.
The probability of a bid between 15 and 28 thousand would be the area of the rectangle
bounded by x 15 and x 28. This would equal (28 15)(.05) .65.
The RISKGENERAL Function

What if a continuous random variable does not appear to follow a normal or a triangular
distribution? We can model it with the RISKGENERAL function.
FIGURE 13
APPENDIX 1 @Risk Crib Sheet 1345

FIGURE 14
Suppose that a market share of between 0 and 60% is possible, and a 45% share is
most likely. There are five market-share levels for which we feel comfortable about com-
paring relative likelihood. (See Table 1.) Thus, a market share of 45% is 8 times as likely
as 10%; 20% and 55% are equally likely; etc. Note that this distribution cannot be trian-
gular, because then 20% would be (20/45) as likely as peak of 45%, and 20% would be
.75 as likely as 45%. To model this, enter the formula
RISKGENERAL(0,60,{10,20,45,50,55},{1,6,8,7,6})
The syntax of RISKGENERAL is as follows:
■ Begin with the smallest and largest possible values.
■ Then enclose in {} the numbers for which you feel you can compare relative
likelihoods.
■ Finally, enclose in {} the relative likelihoods of the numbers you have previously
listed.
Running this in @Risk yields the output shown in Figure 15. Note that 20 is 6/8 likely as
45; 10 is 1/8 as likely as 45; 50 is 7/8 as likely as 45; 55 is 6/8 as likely as 45; etc. In be-
TA B L E 1
Market Share Relative Likelihood
10% 1
20% 6
45% 8
50% 7
55% 6
1346 APPENDIX 1 @Risk Crib Sheet

A B C D E F G
1 EXAMPLE OF
2 RISKGENERAL
3 DISTRIBUTION
4
5 Minimum 0
6 Maximum 60
7 Specified Points
8 10 1
9 20 6
10 45 8
11 50 7
12 55 6
13 45.28889 =RISKGENERAL(0,60,{10,20,45,50,55},{1,6,8,7,6} )
14
15 Distribution for DISTRIBUTION
16 0.08
17
18 0.06
PROBABILITY
19
20 0.05
21
0.03
22
23
0.02
24
25 0.00
26 1.5 11.0 20.5 30.0 39.5 49.0 58.5
27
FIGURE 15 28
tween the given points, the density function changes at a linear rate. Thus, 30 would have
a likelihood of
(30 20)*(8 -6)
6 6.8
(45 20)
Modeling Correlations
Suppose we have three normal random variables, each having mean 0 and standard devi-
ation 1, correlated as follows:
■ Variable 1 and variable 2 have .7 correlation.
■ Variable 1 and variable 3 have a .8 correlation.
■ Variable 2 and variable 3 have a .75 correlation.
To model this correlation structure, we use the RISKCORRMAT command. Simply en-
ter your correlation matrix somewhere in the worksheet. In Figure 16, we chose C27:E29.
B C D E F G H
25
26
27 1 0.7 0.8
28 0.7 1 0.75
29 0.8 0.75 1
30
31 1 Variable 1 1.793028 risknormal(0,1,riskcorrmat(c27:e29,1))
32 2 Variable 2 -0.449129 risknormal(0,1,riskcorrmat(c27:e29,2))
33 3 Variable 3 -0.521328 risknormal(0,1,riskcorrmat(c27:e29,3))
FIGURE 16

For each variable, type in front of the variable’s actual distribution the syntax
Actual Risk Function, RISKCORRMAT(Matrix, i)
Here, Matrix (C27:E29 in this case) indicates where the correlation matrix resides, and i
is the column of the correlation matrix that contains the correlations for variable i. Thus,
for variable 1, the correlations come from the first column of the correlation matrix.
If you run a simulation and extract the data for cells D31:D33, you will find that
■ Each cell has a mean of around 0 and a standard deviation around 1.
■ Each cell follows a normal distribution.
■ D31 has around a .7 correlation with D32.
Truncating Random Variables

Suppose you believe that market share for a product is approximately normally distrib-
uted, with mean .6 and standard deviation .1. This random variable could exceed 1 or be
negative, which would be inconsistent with the fact that market share must be between 0
and 1. To resolve this, you may enter the random variable from the Define Distribution
icon as shown in Figure 17.
You could also type in formula
RISKNORMAL(.6,.1,RISKTRUNCATE(0,1))
Then @Risk generates a normal random variable with mean .6 and standard deviation .1.
If the random variable assumes a value between 0 (the lower truncation value) and 1 (the
upper truncation value), that value is retained. Otherwise, another value is generated. The
truncation values must be within 5 standard deviations of the mean.
FIGURE 17
1348 APPENDIX 1 @Risk Crib Sheet

The RISKPERT Function
This function is similar to the RISKTRIANG function. The RISKPERT function is used
to model the duration of projects. For example,
RISKPERT(5,10,20)
would be used to model the duration of an activity that always takes at least 5 days, never
takes more than 20 days, and is most likely to take 10 days. Whereas RISKTRIANG has
a piecewise linear density function, the RISKPERT density has no linear segments. It is
a special case of a Beta random variable.
Common Error Message

The error message “Invalid number of arguments” means that an incorrect syntax has
been used with an @Risk function. For example, RISKDUNIFORM({A1:A7}) may
have been used instead of RISKDUNIFORM(A1:A7).

APPENDIX 2
Cases
Jeffrey B. Goldberg
UNIVERSITY OF ARIZONA
CASE 1 Help, I’m Not Getting Any Younger! 1351
CASE 2 Solar Energy for Your Home 1351
CASE 3 Golf-Sport: Managing Operations 1352
CASE 4 Vision Corporation: Production Planning and Shipping 1355
CASE 5 Material Handling in a General Mail-Handling Facility 1356
CASE 6 Selecting Corporate Training Programs 1359
CASE 7 BestChip: Expansion Strategy 1362
CASE 8 Emergency Vehicle Location in Springfield 1364
CASE 9 System Design: Project Management 1365
CASE 10 Modular Design for the Help-You Company 1366
CASE 11 Brite Power: Capacity Expansion 1368

■ I might consider taking vitamin pills to get nutri-
CASE 1 tional requirements, but I would rather eat food.
Help, I’m Not Getting Any Younger! Key Questions

Profile of a university professor: ■ What should I eat at each meal?
■ 45-year-old formerly athletic male ■ If I allowed less variety, would your recommenda-
■ 215 pounds tion change?
■ 71 inches tall ■ If I allowed more than $100 per week, would your
recommendation change? How?
■ Exercises no more than once a week. Walks 0.5
miles daily to and from the car while carrying a ■ What key minerals and vitamins constrain the
10-pound briefcase. solution?
■ Family history of adult diabetes.
I need help! My diet is terrible, and I have been
gaining weight and feeling more tired. CASE 2
I heard that Professor George Dantzig of Stanford
once used linear programming to construct a diet. It Solar Energy for Your Home
would be great if you could tell me what to eat dur-
ing each day. So, because I’m a firm believer in math- As our ability to extract and process fossil fuels de-
ematical models, I want you to use linear program- creases, many people are looking to renewable re-
ming to determine a reasonable diet for me to eat sources to meet their energy needs. In particular, solar
during a week. It is your job to collect data for use in energy is becoming an advanced technology that has
the model. economic promise. In areas with large solar insola-
I have the following requirements for the diet: tions, there can be enough energy to power an entire
home. The amount of solar energy reaching the earth
■ I like variety. You cannot prescribe a diet in which each year is many times greater than worldwide energy
I eat just one food during the entire week (like 10 demand; it varies, of course, with location, time of day,
boxes of Total cereal). I would like to eat at least and season. Sunlight is also a widespread resource and
15 different foods during the week. can be captured from virtually anywhere on earth.
■ You have to give me something from each of the There are two categories of home solar systems:
four basic food groups (dairy, fruit and vegetable, passive and active. In a passive system, the solar en-
meat, and grains)—not Mcfood, frozen food, pizza ergy heats a material that is used in a productive man-
food, or food on a stick. ner. For example, in Arizona it is common to use a
■ I like nutrition. You cannot prescribe a diet that passive solar system to heat swimming pools and the
does not meet minimum daily requirements for es- water used in the home. Every building has some of
sential minerals and vitamins. You cannot prescribe its heating requirements met by solar energy. Sunlight
a diet in which I gain a lot of weight. I could stand passing through windows is a source of heat, and the
to lose a few pounds. value of passive solar heating is enhanced by proper
building insulation. A well-insulated building requires
■ I hate Brussels sprouts, sweet potatoes, pears, and
less energy for heating; thus, much of the heating load
organ meats such as liver and kidney.
can be met by passive solar features. Optimum pas-
■ Forget about any canned fruits or vegetables. Yuck. sive solar design begins with the layout of a building
■ I do not eat any pork or pork products. lot; a house must be oriented so that it can take full
■ I am not a big fan of frozen dinners, no matter how advantage of available solar energy.
nutritious or convenient they are. Active systems are more complex and generally in-
volve converting the solar energy to electrical energy.
■ I don’t drink milk with any meal except breakfast.
Photovoltaic (PV) cells use the energy of the sun to
■ I work for the university, so I have a limited bud- produce electricity. They produce none of the green-
get for food. Try to keep costs less than $100 per house or acid gas emissions that are commonly asso-
week (the lower the better). ciated with the use of fossil fuels to generate electric-
CASE 2 Solar Energy for Your Home 1351

ity. The main barrier to increased use of this technol- FIGURE 1
ogy is cost. A common semiconducting material used Daily Energy Flow
in PV cells is single crystal silicon. Single crystal sil- Electrical energy from sun in day t
icon cells are generally the most efficient type of PV
cells, converting as much as 23% of incoming solar
energy into electricity. The main problem with them
Day t
is their production cost. Polycrystalline silicon cells Battery storage Battery storage
end of day t –1 end of day t
are less expensive to manufacture but less efficient
than single crystal cells (15% to 17%). Thin films
(0.001–0.002 mm thick) of amorphous or uncrystal-
lized silicon are another PV alternative. These thin Electrical usage in day t
films are inexpensive and may be easily deposited on
materials such as glass and metal, thus lending them-
be met by this system (no natural gas will be used for
selves to mass production. Amorphous silicon thin-
heating or cooking, for example).
film PV cells are widely used in commercial elec-
Your design should include the following:
tronics, powering watches and calculators. These cells,
however, are not especially efficient—12% in the lab, ■ the area of the PV collectors and the amount of
7% for commercial cells—and they degrade with time, battery storage that you need,
losing as much as 50% of their efficiency with expo- ■ an estimate of the cost of the system (you may in-
sure to sunlight. clude any tax advantages that accrue from the pur-
Solar power is an intermittent source of electricity. chase of solar energy systems),
If PV cells are your only source, then the storage of
■ a profile of the battery storage levels at the end of
electricity may be necessary. Electricity for a home
each day for a six-month period,
can be stored in batteries, which can be expensive.
Also, to generate sufficient electricity, you need a ■ an estimate of cost savings (or loss) over buying
large area of collectors on your roof or somewhere on your electrical power from the local utility company.
your property. The amount of solar energy captured
depends on the surface area of the collectors and their
conversion efficiency.
A solar energy system can often be looked at as a CASE 3
conservation system. Figure 1 depicts one way to look
at the daily flow of energy. Golf-Sport: Managing Operations
Your job is to design an active solar system for a
home in your area. For the analysis, you will have to Golf-Sport is a small-sized company that produces
collect data on: high-quality components for people who build their
own golf clubs and prebuilt sets of clubs. There are
■ system cost and efficiency, five components—steel shafts, graphite shafts, forged
■ daily solar insolation in your area (usually mea- iron heads, metal wood heads, and metal wood heads
sured in watts/meter2; this information can be found with titanium inserts—made in three plants—Chan-
locally where weather data are stored and col- dler, Glendale, and Tucson—in the Golf-Sport sys-
lected), and tem. Each plant can produce any of the components,
■ typical daily power requirements for a home in although each plant has a different set of individual
your area. constraints and unit costs. These constraints cover la-
bor and packaging machine time (the machine is used
The costs of the system generally include a fixed com- by all components); the specific values for each com-
ponent and variable components that depend on the ponent–plant combination are given in Tables 1–3.
total area of the PV collectors, the type of material Note that even though the components are identical in
used in the collector (usually only material is chosen), the three plants, different production processes are
and the amount of battery storage needed. Your analy- used, and therefore the products use different amounts
sis should cover at least 6 months of data (12 months of resources in different plants.
would be better, because you would like your design Besides component sales, the company takes the
to be appropriate for the entire year). You should as- components and manufactures sets of golf clubs. Each
sume that all energy requirements for the home will set requires 13 shafts, 10 iron heads, and 3 wood
1352 APPENDIX 2 Cases

TA B L E 1
Product-Resource Constraints: Chandler
Resources
Labor Packing Advertising
Products (Minutes/Unit) (Minutes/Unit) ($/Unit)
Steel shafts 1 4 1.0

Graphite shafts 1.5 4 1.5
Forged iron heads 1.5 5 1.1
Metal wood heads 3 6 1.5
Titanium insert heads 4 6 1.9
Monthly availability 12,000 20,000 —
(minutes)
TA B L E 2
Product-Resource Constraints: Glendale
Resources
Steel shafts 3.5 7 1.1

Graphite shafts 3.5 7 1.1
Forged iron heads 4.5 8 1.1
Metal wood heads 4.5 9 1.2
Titanium insert heads 5.0 7 1.9
(minutes)
TA B L E 3
Product-Resource Constraints: Tucson
Resources
Steel shafts 3 7.5 1.3

Graphite shafts 3.5 7.5 1.3
Forged iron heads 4 8.5 1.3
Metal wood heads 4.5 9.5 1.3
Titanium insert heads 5.5 8.0 1.9
(minutes)
CASE 3 Golf-Sport: Managing Operations 1353

heads. All of the shafts in a set must be the same type Your job is to determine a recommendation for the
(steel or graphite), and all of the wood heads must be company. A recommendation must include a plan for
the same type (metal or metal with inserts). Assembly production and sales. In addition, you should also ad-
times for the sets at each plant are shown in Table 4. dress the following sensitivity-analysis issues in your
Each plant of Golf-Sport has a retail outlet to sell recommendation:
components and sets, and the specific plant is the only
■ If you could get more graphite or advertising cash,
supplier for its retail outlet. The minimum and maxi-
how much would you like, how would you use it,
mum amounts of demand for each plant–product pair
and what would you be willing to pay?
are given in Table 5. Note that, although the mini-
mums must be satisfied, you do not need to satisfy de- ■ At what site(s) would you like to add extra pack-
mand up to the maximum amount. ing machine hours, assembly hours, and/or extra
This planning problem is for two months. The costs labor hours? How much would you be willing to
in Table 6 increase by 12% for the second month, and pay per hour and how many extra hours would you
production times are stationary. Inventory costs are like?
based on end-of-period inventory for each product set ■ Marketing is trying to get Golf-Sport to consider
and cost out at 8% of the cost values in Table 6. Table an advertising program that promises a 50% in-
7 lists the revenue generated by each product. Ini- crease in their maximum demand. Can we handle
tially, there is no inventory. this with the current system or do we need more
The corporation controls the capital available for resources? How much more is the production go-
expenses; the cash requirements for each product are ing to cost if we take on the additional demand?
given in the last column of Tables 1–3. There is a to-
tal of $20,000 available for advertising for the entire
system during each month, and any money not spent TA B L E 6
in a month is not available the next month. The cor- Material, Production, and Assembly Costs ($) per Part or Set
poration also controls graphite. Each shaft requires 4 Plants
ounces of graphite; a total of 1,000 pounds is avail- Products Chandler Glendale Tucson
able for each of the two months.
Steel shafts 6 5 7
Graphite shafts 19 18 20
TA B L E 4 Forged iron heads 4 5 5
Metal wood heads 10 11 12
Time Total Time Available
Plant (Minutes per set) (Minutes) Titanium insert heads 26 24 27
Set: Steel, metal 178 175 180
Chandler 65 5,500
Set: Steel, insert 228 220 240
Glendale 60 5,000
Set: Graphite, metal 350 360 370
Tucson 65 6,000
Set: Graphite, insert 420 435 450
TA B L E 5
Minimum and Maximum Product Demand per Month
Store (or Plant)
Products Chandler Glendale Tucson
Steel shafts [0, 2,000] [0, 2,000] [0, 2,000]

Graphite shafts [100, 2,000] [100, 2,000] [50, 2,000]
Forged iron heads [200, 2,000] [200, 2,000] [100, 2,000]
Metal wood heads [30, 2,000] [30, 2,000] [15, 2,000]
Titanium insert heads [100, 2,000] [100, 2,000] [100, 2,000]
Set: Steel, metal [0, 200] [0, 200] [0, 200]
Set: Steel, insert [0, 100] [0, 100] [0, 100]
Set: Graphite, metal [0, 300] [0, 300] [0, 300]
Set: Graphite, insert [0, 400] [0, 400] [0, 400]

TA B L E 7 TA B L E 9
Revenue per Part or Set ($) Product-Resource Constraints: Plant 2
Plants Resources
Products Chandler Glendale Tucson Labor Machine Material
Products (Hours/Unit) (Hours/Unit) (Lb./Unit)
Steel shafts 10 10 12
Graphite shafts 25 25 30 Small 3.5 7 1.1
Forged iron heads 8 8 10 Medium 3.5 7 1.0
Metal wood heads 18 18 22 Large 4.5 8 1.1
Titanium insert heads 40 40 45 Precision 4.5 9 1.4
Set: Steel, metal 290 290 310 Total available 5,000 12,500 —
Set: Steel, insert 380 380 420
Set: Graphite, metal 560 560 640
Set: Graphite, insert 650 650 720 TA B L E 10
Product-Resource Constraints: Plant 3
Resources
Labor Machine Material
CASE 4
Small 3 7.5 1.1
Vision Corporation: Production Medium 3.5 7.5 1.1
Large 4 8.5 1.3
Planning and Shipping Precision 4.5 8.5 1.3
Vision is a large company that produces video- Total available 3,000 6,000 —
capturing devices for military applications such as
missiles, long-range cameras, and aerial drones. Four
different types of cameras (differing mainly by lens pounds of material is available for the entire system
type) are made in the three plants in the system. Each during the planning period.
plant can produce any of the four camera types, Transport has 3 major customers (RAYco, HONco,
although each plant has its own individual constraints and MMco) for its products. The maximum sales for
and unit costs. These constraints cover labor and ma- each customer–product pair is given in Table 11. Prod-
chining restrictions, and the specific values are given uct sales prices are given in Table 12, and the ship-
in Tables 8–10. Note that even though the products ping costs from each plant to each customer are de-
are identical in the three plants, different production tailed in Table 13. Table 14 contains the production
processes are used and thus the products use different costs for each product–plant pair.
amounts of resources in different plants. The corpo- All shipping from plants 1 and 2 that goes to
ration controls the material that goes into the lenses; RAYco or HONco must go through a special inspec-
the material requirements for each product are given tion. These units are sent to a central site, inspected,
in the last column of Tables 8–10. A total of 3,500 and then sent to their destination. The capacity of this
special inspection site is 1,500 pieces.
Your job is to determine a recommendation for the
TA B L E 8
company. A recommendation must include a plan for
Product-Resource Constraints: Plant 1 production and shipping as well as the cost and rev-
Resources enue generated from each plant. In addition, you
Labor Machine Material should address the following potential issues in your
recommendation:
Small 3 8 1.0
■ If you could get more material, how much would
Medium 3 8.5 1.1
you like? How would you use it? What would you
Large 4 9 1.2
be willing to pay?
Precision 4 9 1.3
■ If you could get more inspection capacity, how
Total available 6,000 10,000 —
much would you like? How would you use it? What
would you be willing to pay?
CASE 4 Vision Corporation: Production Planning and Shipping 1355

TA B L E 11
Maximum Product Sales ($) per Unit CASE 5
Customers
Products RAYco HONco MMco
Material Handling in a General
Small 200 400 200 Mail-Handling Facility†
Medium 300 300 400
Large 500 200 300 For more than 200 years, the United States Postal Ser-
Precision 200 400 300 vice (USPS) has delivered mail across the country.
Daily delivery goes to some 137 million households;
in 2001, the USPS processed and delivered more than
207 billion pieces of mail to a delivery network that
TA B L E 12
grew by 1.7 million new addresses. Clearly, the USPS
Product Sales Price ($) per Unit
is the largest material handler (in terms of pieces) in
Customers the world. Statistics recorded by Pricewaterhouse
Products RAYco HONco MMco Coopers show that 94% of first-class mail destined for
Small 17 16 16 next-day delivery received overnight service—and
Medium 18 18 17 this was a record performance for a second straight
Large 22 22 23 year. Despite the high volume, the USPS managed to
Precision 29 26 27 cut costs by $900 million in 2001 while maintaining
record service performance and high levels of cus-
tomer satisfaction.
To process mail quickly, one must use advanced
TA B L E 13
mechanization. Mail-sorting machines can process 10
Shipping Costs ($) per Unit
letters per second (we are long past the days of hand
Customers sorting in front of a large set of post boxes). Sorting
Plant RAYco HONco MMco using the zip4 standard can result in a mail sort
1 1.0 1.6 1.1 down to an individual carrier’s walk sequence, which
2 1.2 1.5 1.0 saves significant carrier time.
3 1.4 1.5 1.3 Five major operations can be performed on each
letter, and each operation has its own machine:
Automatic facer and canceller (AFC) This machine
TA B L E 14 cancels the stamp and orients all of the letters so
Production Costs ($) per Unit that the stamp is in the upper-right corner. This
Plant
machine also separates mail into one of three
streams—automation, mechanization, or manual.
Products 1 2 3
Letter sorting machine (LSM) This machine is semi-
Small 14 13 14
automated and helps human operators sort mail.
Medium 16 17 15
The operator reads the address and then types in a
Large 18 20 19
destination code. The machine then routes the let-
Precision 26 24 23
ter to the appropriate bin.
Optical character reader (OCR) This machine reads
handwritten or typed addresses and then prints a
■ At what plant(s) would you like to add extra ma- machine-readable barcode on the envelope.
chine hours? How much would you be willing to
Barcode sorter (BCS) This machine reads the bar-
pay per hour? How many extra hours would you
like? code on the letter (either printed by the OCR or by
the sender’s equipment) and then sorts it to a bin.
■ Marketing is trying to get RAYco to consider a
50% increase in its demand. Can we handle this †
Based on work done jointly with Ron Askin and Sanjay Jagdale,
with the current system or do we need more re- 1994.
sources? How much more money can we make if
we take on the additional demand?

Delivery barcode sorter (DBCS) This machine does One of the keys to faster processing and increasing
a two-pass sort that uses barcodes with the zip4 utilization is an effective material-handling and data
code and sorts down almost to a walk sequence. system. Each tray has a barcode that describes the
When using the DBCS, the result is such that a salient characteristics of its mail. When a sorting ma-
carrier requires little to no processing at the carrier chine is ready for operation (say, for example, that we
station to deliver the mail. are going to sort down to a group of 10 zip codes), a
call goes out to bring all trays with appropriate mail
Items known as flats (for example, magazines and
to the appropriate BCS machine. The data system
8.5 11–inch envelopes) also are processed through
must (1) know where those trays are located, (2) go
the system. A flat-sorting machine (FSM) is used in a
and get them, (3) bring the trays to the machine-input
semiautomated process. An operator loads a piece
area, and (4) exit the area. The faster this can be done,
onto the machine and keys in a code based on the
the better.
flat’s address; the machine then routes the piece to an
The network for our GMF is given in Figure 3.
appropriate bin.
A letter that enters a general mail facility (GMF) Each machine has an input and output point. For ex-
follows a routing that depends on the machine read- ample, nodes 1 and 28 are the inputs and output,
ability of the address and the presence of an existing respectively, for AFC 1. For this application, the
material-handling system is an overhead monorail.
barcode. Although many routes are possible, the ma-
Carriers that hold one tray circulate around the sys-
jor ones are given in Figure 2. Once letters go through
the AFC, they are stored in cardboard or plastic trays tem to pick up and deliver trays; they rest in the park-
that hold approximately 400 letters. The letters are ing lot when not in use. The arcs in the diagram are
moved in these trays throughout the facility, and each the links of the monorail; all links are one-way. The
dotted lines on Figure 3 represent links that are above
machine sorts the mail into different trays.
the machine level and offer shortcuts across the facil-
As part of quality improvement, the post office is
ity. The facility also contains switches (nodes 29, 18,
always looking for ways to cut costs while expediting
and 32) that allow carriers to change directions. Node
mail processing. To meet goals for overnight delivery
34 is the link to the shipping dock, and all trays enter
and three-day cross-country delivery, letters that ar-
and exit the system at this point.
rive by 6 p.m. from box pickup must be processed that
The carriers travel at approximately 1 mile per
night and be on planes or trucks for the next destina-
tion. Because the sorting and character-reading ma- hour, and there must be 15 feet between carriers on
chines cost millions of dollars each, increasing ma- the same link. For the purpose of this study, assume
chine use and saving the purchase of even one machine that there is a bypass at each node so a carrier can
pass other carriers that are stopped for loading and
in a GMF is a significant achievement.
unloading operations. Also assume that the switches
operate quickly relative to the speed of the vehicles so
that collisions do not occur and the switch capacity is
FIGURE 2 not constraining. Figure 3 is drawn approximately to
Processing Routes for Letters scale. The facility is approximately 220 feet long by
160 feet wide. At 1 mile per hour, it takes a carrier
AFC OCR BCS DOCK
approximately 2.5 minutes to run the length of the fa-
cility (from node 14 to node 1, for example).
AFC BCS DBCS DOCK Table 15 contains the tray movement loads for the
peak hour. Each load has an origin node, a destination
AFC OCR LSM node, and the number of trays that must be moved.
Each load is a leg in the route for a particular tray of
AFC LSM DOCK mail. The system must have capacity to move an empty
carrier from the parking lot to the origin, load the tray,
AFC OCR DOCK move the tray to the destination, unload the tray, and
then return to the parking lot. All carriers are dis-
AFC BCS DOCK
patched from the parking lot because this simplifies
the logic of the scheduler. By capacity, there must be
a sufficient number of vehicles and the capacity on
BCS DBCS DOCK
each link between nodes cannot be overloaded. As-
CASE 5 Material Handling in a General Mail-Handling Facility 1357

FIGURE 3
General Mail Facility: Track Layout
29
1 AFC 1 28 FSM 30
2 AFC 2 27 FSM 31
3 LSM 1 26 32
4 LSM 2 25
5 OCR 1 24 Carrier
parking lot
OCR 2 Node 33
6 23
7 OCR 3 22
8 BCS 1 21 Staging
interface
shipping to
9 10 11 12 13 20 other GMFs
and carrier
stations
Node 34
BCS DBCS DBCS DBCS
2 1 2 3
16
14 15 16 17 18
TA B L E 15 sume that it takes one minute to perform a loading or

Load Data for the Peak Hour unloading operation.
Load Number Origin Node Destination Node Number of Trays
Your jobs are to:
1 33 1 15 ■ Determine if this system has enough material-

2 28 4 20 handling capacity for moving the trays in the peak
3 22 14 30 hour (we generally design for peak hour so that we
4 10 33 30 are sure that the system will not get bogged down
5 24 8 15 when demand is high).
6 21 33 30 ■ Suggest where we might add extra track to relieve
7 24 4 5 capacity congestion. This should be minimized be-
8 25 33 15 cause track cost is high.
9 27 17 15 ■ Determine the flows of trays through the network
10 13 33 40 during the peak hour. Which routes are chosen for
each load?

■ Determine places in the network that are risky— Increased training costs have occurred for many
that is, if a link goes down, machines can be cut reasons. Employees view training in the form of for-
off from the material-handling system. mal degrees and documented technical skills as im-
■ Estimate the total carrier travel distance during the portant for job security. Technology is changing at a
peak hour. rapid pace. It has been claimed that high schools and
universities are not producing the skills needed by in-
■ Investigate the effect of reducing the intercarrier
spacing. You need space between carriers to pre- dustry, so industry must train and reeducate recent
vent carrier collisions. If we put better sensors on graduates. For high school graduates, this may in-
the front of the carriers, they can stop more quickly clude training in technology-based skills; for college
and we can have less spacing. graduates, this may include developing nontechnical
skills such as leadership, communications, interper-
Some tips: sonal relations, and ethics.
■ This is a difficult problem. Be patient and try not
to become discouraged. Problem Environment
■ Do not forget the empty carrier movements to the For a corporation, the primary purpose of training is
origin and from the destination. to ensure that employees have the key skills needed to
■ Compute the capacity on each arc. Initially, as- effectively manage and operate the business. There
sume a single lane. To increase capacity, consider are many options for providing training. For example,
multiple lanes between nodes or consider adding to train staff members in computer skills, a corpora-
arcs to give more paths between origin nodes and tion may use any of the following strategies:
destination nodes.
■ hiring an outside consultant to develop and present
■ A precise formulation of this problem can be larger an on-site training course,
than most problems you have seen. You should not
■ using corporate personnel to develop and present
undertake to solve a large-sized formulation unless
an on-site training course,
you have software that can handle large problems.
Some approximate formulations are more manage- ■ purchasing a training course and having employees
able, but they still can require hundreds of vari- use it for self-study,
ables and constraints. ■ contracting with a local college or university to
provide training, or
■ sending employees to an off-site training seminar.
CASE 6 The above possibilities are for a single skill. The
† purpose of many training programs, however, is to
Selecting Corporate Training Programs give employees a broad set of skills. Often the skill
sets of two or more programs partially overlap. When
Introduction
this happens, the corporation must choose the set of
Training has become a major cost of doing business. A programs that give employees the required skills for
1995 survey of all U.S. businesses with 100 or more their jobs and the appropriate employees for each
employees revealed that approximately $52 billion was training program. In any case, training decisions made
being spent on training; it has been estimated that $90 in an ad-hoc “pay-as-you-go” manner will be ineffi-
to $100 billion is being spent for training overall. cient and generally result in additional expense.
Developing strategies to implement cross-training is a To give the decision problem structure, the follow-
current topic in the operations research (OR) literature. ing assumptions are appropriate:
Management consultants advocate aggressive educa-
■ We have a known study period—for example, the
tion and professional development to remain competi-
next 3 or the next 5 years—over which we need to
tive in the global and local markets. Employees now
plan training. The study period should fit with the
expect job and skill growth to be a major component
overall business strategy and enable accurate esti-
of their duties.
mates for training needs and available resources.
†
■ There is a known set of skills that employees need.
Based on work done jointly with John V. Farr, and David A. Thomas
at USMA, 1995.
Among others, these may include technical, inter-
personal, communication, and management skills.

TA B L E 16
Skills List
No. Skill No. Skill
1 New employee orientation 22 Stress management

2 Performance appraisals 23 Computer programming
3 Personal computer apps 24 Diversity
4 Leadership 25 Data processing/MIS
5 Sexual harassment 26 Planning
6 Team building 27 Public speaking and presentation
7 Safety 28 Strategic planning
8 Hiring and selection process 29 Writing skills
9 New equipment operation 30 Negotiating skills
10 Training the trainer 31 Finance
11 Product knowledge 32 Marketing
12 Decision making 33 Substance abuse
13 Listening skills 34 Ethics
14 Time management 35 Outplacement and retirement
15 Conducting meetings 36 Creativity
16 Quality imiprovements 37 Purchasing
17 Delegation skills 38 Smoking cessation
18 Problem solving 39 Financial and business literacy
19 Goal setting 40 Reengineering
20 Managing change 41 Foreign language
21 Motivation
TA B L E 17
Salary and Skills Required for Each Job Classification
Person Salary ($) Skills 1–20
Senior Manager 250,000 0 1 0 1 0 0 0 0 0 0 0 1 1 1 0 1 0 1 1 0

Project Manager 200,000 0 1 0 0 1 0 0 0 0 0 1 1 1 1 1 0 0 1 0 0
Professional 150,000 1 1 0 0 1 1 1 0 1 0 0 0 1 0 1 0 0 0 0 1
Sales 150,000 1 1 0 0 0 0 0 0 1 1 0 0 1 0 0 0 1 0 0 1
Technician 100,000 1 1 1 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1
Administrative 80,000 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Assistant
Person Skills 21–41
Senior Manager 1 1 1 1 0 1 1 0 1 0 1 1 0 1 0 1 1 1 0 0 0
Project Manager 0 1 1 1 0 1 1 1 1 0 1 0 1 0 0 1 1 1 1 0 1
Professional 0 0 0 0 1 1 0 0 0 1 0 0 1 0 1 0 0 1 1 0 1
Sales 0 0 0 0 1 1 0 0 0 1 0 0 1 0 1 0 0 1 1 1 1
Technician 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1
Administrative 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 1 0
Assistant

TA B L E 18
Enrollment Cost and Skills of Each Program
Program Enrollment Cost ($) Skills 1–20
Program 1 500 1 1 0 1 0 0 0 0 0 0 1 1 1 1 1 0 1 0 0 0
Program 2 300 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0
Program 3 500 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 1 0 0 0 1
Program 4 575 0 1 0 1 0 0 0 0 1 1 0 0 1 0 0 0 1 0 0 0
Program 5 800 0 1 1 1 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 0
Program 6 400 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 1
Program 7 200 0 0 0 0 1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1
Program 8 1,000 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 1 0
Program 9 200 1 1 1 1 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0
Program 10 500 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0
Program 11 700 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0
Program 12 600 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0
Program 13 400 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 1 1 1 0
Program 14 900 1 0 0 0 1 1 0 0 0 0 0 0 0 0 0 1 0 1 0 0
Program 15 700 1 0 1 0 1 0 1 1 1 0 0 0 0 0 0 0 0 0 1 0
Program Skills 21–41
Program 1 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0
Program 2 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0
Program 3 1 1 1 1 1 1 1 0 1 0 1 0 0 0 0 1 0 0 0 0 0
Program 4 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 1
Program 5 0 0 0 1 0 1 0 0 0 0 0 0 1 0 0 0 0 0 1 1 0
Program 6 0 0 0 0 0 0 0 0 1 0 0 1 1 0 0 1 1 0 0 0 0
Program 7 0 1 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0
Program 8 1 1 1 1 0 0 1 0 1 1 0 0 0 0 0 0 1 0 0 0 0
Program 9 0 0 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0
Program 10 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0
Program 11 0 1 1 0 0 0 1 0 1 1 0 1 0 0 1 1 0 0 0 0 0
Program 12 0 0 0 0 0 0 0 1 0 1 1 0 0 0 0 0 0 1 1 1 0
Program 13 1 0 1 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0
Program 14 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 1
Program 15 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 1
■ Employees are divided into classes. In each class, ■ Training is equally effective for all people, thus we
we have estimates for (1) the number of employ- are concerned with which programs to offer and
ees, (2) the employee hourly wage, (3) the number which employees in each class to assign to each
of employees that require each particular skill, and program. If we know the quality of the training for
(4) the maximum time available for training em- individual skills for individual classes, then we can
ployees in each class during the study period. relax this assumption.
■ There is a list of training programs. For each pro-
gram, we assume we have (1) the set of skills Potential Corporate Setting
taught, (2) the cost, (3) the development time,
(4) the completion time for an employee, and Your job is to develop models to aid businesses and
(5) the maximum number of employees who can corporations in determining the appropriate training
participate per decision cycle. programs to use. The type of model and issues often
depend on the size of the corporation and the potential

TA B L E 19 are 15 programs available for use; Table 18 contains
Interfering Programs the cost per person and the skills covered for each
Program Number (Days Long) Programs that Interfere program. In Table 18, a 1 in the row for program p
and the column for skill s implies that program p con-
Program 1 (2) 3 5 8
tains skill s. Table 19 lists the programs that conflict
Program 2 (2) 3 7 10
in time with other programs (for example, programs
Program 3 (4) 1 2 12
3, 5, and 8 conflict with program 1). An employee
Program 4 (3) 6 7 14
cannot take two programs simultaneously. It is com-
Program 5 (2) 1 9 12
pany policy that each employee is limited to 15 days
Program 6 (3) 4 7 11 14
for training per year.
Program 7 (5) 2 4 6
Program 8 (2) 1 10 13
Program 9 (3) 5 15 Key Questions
Program 10 (3) 2 8
Your job is to develop a recommendation for the com-
Program 11 (2) 6 12 15
pany for addressing its training needs. In particular,
Program 12 (4) 3 5 11
you should address the following key questions:
Program 13 (3) 8 14
Program 14 (4) 4 6 13 ■ Which training programs should we be using? What
Program 15 (3) 9 11 is the assignment of personnel to those programs?
■ Identify programs with heavy use that may justify
the development of an in-house course. How much
uses of the models. For large corporations, there are would you be willing to pay for that development
many employees in each class, so it is not necessary to if you could use the program for the next three
model and schedule down to the individual employee. years?
Concentrate instead on the assignment of classifica- ■ We have the opportunity to negotiate prices for
tions to programs and ignore the assignment of spe- programs. Which programs would you suggest are
cific individuals to programs. Also, sufficient resources candidates for negotiation?
exist to develop internal training programs, hence you ■ What skills are especially expensive for us to cover?
should consider program development costs as well as If we were to develop our own programs, what
employee costs (lost work time, travel, lodging, meals, skills should be covered?
course materials) in the objective. A large corporation
can use the model to plan the development of courses. ■ Would your recommendation change if we allowed
This will help determine (1) program-development more days of training per year?
costs so that in-house programs are cost-effective and
(2) appropriate programs for each employee classifica-
tion so that, on average, there is sufficient time to com-
plete the assigned programs within the available time.
CASE 7
For small businesses, the focus is often different.
Typically, these companies do not develop in-house
BestChip: Expansion Strategy
programs because they do not train enough employ- BestChip (BC) is a large nationwide corporation that
ees to justify development costs. Because the number produces low-fat snack products for an expanding
of employees is small, it is important to model down market (pun intended). Basically, BC takes materials
to the employee level and schedule employees so that (corn, wheat, and potatoes) and turns them into two
both training and job tasks can be completed. types of snacks: chips (regular and green onion) and
Your OR consulting firm has been hired to design party mix (one variety). BC is expanding into the
the training program for a small company. There are western United States and is considering sites for lo-
no in-house classes, and vendors provide all training. cating production facilities.
The company has determined 41 skills that are im- BC currently has eight candidate sites. Table 20
portant for its employees; these are listed in Table 16. shows the sites’ purchase prices and the purchase and
There are six employees; the salary level and skills re- shipping cost per ton of each material to each site.
quired for each person are given in Table 17. You can
The purchase cost represents the yearly amortized
assume that there are 250 working days per year. There
cost of opening and operating the site (exclusive of

TA B L E 20
Site Information and Material Shipping Cost
Material Shipping Cost ($/Ton)
Site Location Purchase Cost ($/Year) Corn Wheat Potato
Yuma, AZ 125,000 10 5 16
Fresno, CA 130,000 12 8 11
Tucson, AZ 140,000 9 10 15
Pomona, CA 160,000 11 7 14
Santa Fe, NM 150,000 8 14 10
Flagstaff, AZ 170,000 10 12 11
Las Vegas, NV 155,000 13 12 9
St. George, UT 115,000 14 15 8
TA B L E 21
Demand Information
Demand
Company Location Regular Green Onion Party Mix
Jones Salt Lake City 1,300 900 1,700

YZCO Albuquerque 1,400 1,100 1,700
Square Q Phoenix 1,200 800 1,800
AJ Stores San Diego 1,900 1,200 2,200
Sun Quest Los Angeles 1,900 1,400 2,300
Harm’s Path Tucson 1,500 1,000 1,400
shipping costs). Each site may produce as many as TA B L E 22

20,000 tons of product per year. Product-Ingredient Mix
BC has six major customers, and all demand is Ingredient
shipped by truck from the plant to the customer ware- Product Corn Wheat Potato
house. The shipping cost depends on the tonnage and
distance and comes to $0.15 per ton-mile. The cus- Regular chips 70 20 10
tomers, their location, and their yearly demand in tons Green onion chips 30 15 55
for each product are listed in Table 21. You must meet Party mix 20 50 30
demand.
The makeup of the products does not depend on
the production plant. Table 22 gives the product-
ingredient mix data. The company requires that we ■ What sites should be selected? How should the
consolidate our business, so we cannot locate plants customers be served?
in more than two states. ■ If gasoline gets more expensive and our trucking
For this analysis, ignore the differences in property costs change, then how is the recommendation
and income tax rates between the states (this is usually affected?
critical, but it gets us far afield of the key issue of math
■ If rail freight costs for material shipping increase,
programming). In addition, many critical factors actu-
then how is the recommendation affected?
ally determine locations; for example, the method of
financing the site purchase will also be a major factor Please consider other sensitivity-analysis issues
in the decision—but we will ignore that also. that you feel might be important for management’s
Your job is to determine how we should expand decision-making process.
into the west and develop alternatives. Questions you
should answer include:
CASE 7 BestChip: Expansion Strategy 1363

department has estimated the number of fire calls, the
CASE 8 number of false alarms, and the number of medical
calls per 12-hour day. These data are listed in Table
Emergency Vehicle Location in Springfield 23. Currently, there are five fire stations in the city;
these are listed in Figure 4. Each existing station costs
You are the logistics manager for the Springfield Fire $20,000 per year to operate. The yearly costs for each
Department. You are to develop a recommendation for potential station (including the amortized cost of con-
providing emergency service to Springfield. The de- struction) are also listed in Table 23. Each station can
partment’s resources include engine trucks, ladder hold two vehicles at most.
trucks, and paramedic vehicles. The budget suggests For fire calls, an engine and a ladder truck must re-
a total of 15 vehicles are fundable in the coming year. spond. For medical calls, a paramedic vehicle always
Currently, seven engines, three ladder trucks, and five responds and an engine also goes when one is avail-
paramedic vehicles are in operation. This system runs able and closer than the nearest paramedic vehicle.
24 hours per day. On average, fire calls take 2.5 hours, false alarms 10
The city has been divided into 10 zones (see Fig- minutes, and medical calls 45 minutes.
ure 4). The map is drawn to scale. For each zone, the
FIGURE 4
Map of the City (17 miles by 11 miles)
X
2 X
X
1
3 4
X
5
X
X
6
X X
8 10
7 X X 9
X marks the spot of the existing sites and the possible sites in each zone
TA B L E 23
Demand Information per Year
Zone Fire Calls False Alarms Medical Calls Base Cost ($/Year)
1 100 200 1000 40,000

2 50 100 450 Existing
3 75 100 600 35,000
4 120 75 1300 50,000
5 150 100 1400 Existing
6 300 150 1000 50,000
7 200 100 800 Existing
8 250 175 1000 Existing
9 100 25 900 Existing
10 75 50 650 35,000

When vehicles are dispatched on a call, the closest hire subcontractors, ensure each task is completed
idle vehicle is dispatched first. If no vehicles are idle, within specification, determine how much labor to as-
then the call must be sent to a private provider; these sign to each task, and generally ensure the project’s
responses cost the city $5,000 per medical call, success.
$15,000 per fire call, and $200 per false alarm. There SD is really a subcontractor within the larger proj-
is no queueing. The street network is largely rectan- ect of building the power plant. Table 24 details SD’s
gular, and the fire department estimates that the cost plant-construction and data-system-design tasks. SD
per mile for travel is $1.50 per mile for engines and is directly in charge of tasks 2, 6, 7, 10, 14, 15, and
ladders and $0.75 per mile for paramedic units. 19. The remaining tasks in Table 24 interact with those
Your job is to design a system for the fire depart- in SD’s charge. Assume that the remaining tasks
ment. The questions that should be considered are (1) will start whenever their predecessor tasks are com-
these: plete and (2) will finish exactly after their duration.
To shorten the seven SD tasks, you must pay addi-
■ What sites should be selected and how should the
tional labor and overhead costs. Table 25 lists the
vehicles be distributed?
functions you can use to compute the cost of chang-
■ If travel gets more expensive, how is the recom- ing each task duration to a new value. (Note: tj is the
mendation affected? original duration of task j; dj is the minimum duration
■ If the cost of using the private provider increases, of task j.) The table also lists the lowest possible task-
then how should the system be changed? duration value. You may not increase the duration of
■ Is all of this equipment needed to serve the public? any task.
■ How much more demand can be handled with the The revenue that SD obtains from the project de-
full complement of vehicles? pends not only on its tasks but also on when the total
project is completed. The project is due at day 900,
Your write-up should include a description of your and SD receives the contract price of $600,000 if the
models and any assumptions made in model formula- project is done then. If the project is finished x days
tion. You will have to make simplifying assumptions, early, then SD receives a total of $600,000
because this problem has details that may be difficult $15,000x0.7 in revenue. If SD finishes x days late, then
to model. There are many ways to model parts of this it receives $600,000 $20,000x1.4 in revenue.
system, and you can use different approaches to an- Expediting tasks can be profitable and necessary to
swer different questions. You may use Excel or meet deadlines, although employees do not really like
LINDO, or you may use heuristics. Your call will de- it. Task completion quality is a function of the task
pend on your modeling approaches. completion time, and we would like to have a high
Hint: This case is less specific and has vague com- quality. This may conflict with our objective of max-
ponents; simplify as a first approach and then get imizing profit. Because quality affects future revenues,
more complex. If you try to include everything, you it is difficult to estimate the dollar impact of poor
will become frustrated because this does not fit any quality. If tj is the original duration of task j and xj is
standard modeling paradigm. the expedited duration time for task j, then quality,
measured on a scale of 0–100 (with 100 being the
best), can be represented by the function
CASE 9 100 min [100, (tj xj)2.2]
System Design: Project Management† This is only the quality for a task. It is unclear how
one might quantify the quality of the project.
Your job is to determine how we should proceed
System Design (SD) is a small corporation that con- with our tasks. Your analysis should answer some of
tracts to manage systems and industrial engineering the following questions:
projects. In this case it must manage the design and
construction of a power plant’s data-processing and ■ What tasks are critical to project completion? What
data-collection system. SD’s role in the project is to tasks will you expedite?
■ How are you measuring system quality, and how
†
does your recommendation measure up relative to
This material is expanded from a homework problem in Applied
that objective?
Mathematical Programming by Wayne Winston.
CASE 9 System Design: Project Management 1365

TA B L E 24
Task Information
Task Duration Immediate
Task No. Task Name (Days) Predecessor Tasks
1 Preliminary system description 40 —

2 Develop specifications 100 1
3 Client approval 50 2
4 Develop input-output summary 60 2
5 Develop alarm list 40 4
6 Develop log formats 40 3, 5
7 Software definition 35 3
8 Hardware requirements 35 3
9 Finalize input-output summary 60 5, 6
10 Analysis performance calculations 70 9
11 Automatic turbine startup analysis 65 9
12 Boiler guides analysis 30 9
13 Fabricate and ship 200 10, 11, 12
14 Software preparation 80 7, 10, 11
15 Install and check 130 13, 14
16 Termination and wiring lists 30 9
17 Schematic wiring lists 60 16
18 Pulling terminals and cables 60 15, 17
19 Operational test 125 18
20 First firing 1 19
TA B L E 25
Expediting Costs ($1,000) and Limits
Duration (Days)
Task No. Current tj Minimum dj Cost to Decrease tj by x Days
2 100 70 1.5x1.8
6 40 20 2x2.0
7 35 20 1x2.0
10 70 40 1.8x1.9
14 80 60 1.9x1.6
15 130 120 0.95x2.7
19 125 80 0.9x2.9
■ How much money do you make on the project?

■ What will you give the decision maker to help with CASE 10
the decision?
■ If we can move minimum duration days to lower
Modular Design for the Help-You Company
values, then which values would you like to reduce?
The Help-You Company is in the business of manu-
■ If you could control additional tasks by paying facturing first-aid kits for cars, hikers, campers, sports
more money, which ones would you like to take, teams, and scouting groups. The company is located
and how much would you be willing to pay for in Tucson, Arizona, and all materials must be sent to
control? Tucson and then shipped to customers’ warehouses.
The company has done extensive market surveys of

its customers; Table 26 shows its estimates for the de- minimum requirements for each item are met. A
mand for its kits in the coming year. graphic of the approach is detailed in Figure 5.
Each kit contains the individual items shown in
Table 27, which are listed with their base sizes in If you design a module, for example, that has two
pounds. Help-You, for example, can buy packs of units of Band-Aids (as well as the other items) and
Acetaminophen extra-strength caplets. Each pack con- place three of these modules in a scouting kit, then the
tains 12 tablets and costs Help-You $1.50. kit will have 3*2 6 units of Band-Aids; this will
Help-You buys the individual items and then as- meet the requirement of four units of Band-Aids for
sembles kits based on the requirements for each part scouting kits. In this example, there is an “overage”
in each kit as shown in Table 28. of two extra units of Band-Aids that costs
These are minimum requirements in that the cus- 2 * $1 per unit $2
tomer expects at least the listed quantity of each item
in each specific kit. For example, in the kit for per each scout kit demanded. Also, the total unit con-
campers, there must be at least four blankets and at tent in a module cannot weigh more than 15 pounds
least three cold units (six cold packs) as well as the (an assembly requirement).
other items. Direct assembly meets requirements exactly, al-
There are two strategies available for assembly of though it usually has higher labor costs than modular
the kits: assembly. For storing inventory, modules are easier to
use because they tend to be smaller than kits.
■ In direct assembly, the exact requirements are put Develop a strategy for modular assembly. The key
into each kit. costs of the modular system are the overages that oc-
■ In modular assembly, one or more standard mod- cur. Your strategy must include the following:
ules are developed that can be assembled and com- ■ the number of modules you are designing (the more
bined into a kit with enough modules so that the modules you have, the closer you can match re-
quirements exactly, although the higher the costs
for assembly and inventory);
TA B L E 26
■ the unit content of each module designed; and
Kit Demand
■ an estimate for the total number of each module
Type Number of Kits Sold
required.
Cars 1,000
Your analysis should consider issues such as:
Hikers 800
Campers 100 ■ the trade-off between the number of different mod-
Sports teams 200 ules designed and the total overage cost (you do
Scouting groups 300 not need to try more than five different modules—
why?);
TA B L E 27
Item Cost and Base Size
Item Cost ($) Base Unit Base Unit Weight (Lb.)
Adhesive Band-Aids 1 10 per pack 0.20

Ace bandages 2 1 bandage 0.20
Flares 4 3 per pack 1.00
Blankets 15 1 blanket 2.00
Adhesive tape 2.50 1 roll 0.40
Cold packs 4 2 per pack 0.80
Sunburn cream 3.50 1 tube 0.40
Antiseptic cream 2 1 tube 0.50
Acetaminophen extra-strength 1.50 12 tablets 0.30
caplets
Rubber gloves 1.50 3 pairs 0.20
CASE 1 0 Modular Design for the Help-You Company 1367

TA B L E 28
Item to Kit Requirements (Base Units)
Kit Item Cars Hikers Campers Sports Teams Scouting Groups
Ace bandages 1 2 4 12 6
Band-Aids 0 2 4 4 4
Flares 2 1 1 0 2
Blankets 1 1 4 2 3
Adhesive tape 2 2 3 6 4
Cold packs 2 2 3 6 3
Sunburn cream 1 2 4 4 5
Antiseptic cream 1 2 3 2 4
Acetaminophen caplets 1 2 4 6 6
Rubber gloves 1 1 2 10 5
FIGURE 5
Modular Assembly
Car Hiker Camping Sports Scouts
Modules
Band-Aids Gloves
Ace bandage Tape
Flares Blankets
■ the sensitivity of your solution to the kit demand

estimates (for example, what happens to your rec- CASE 11
ommendations if the number of scouting kits sold
changes by 20%?); Brite Power: Capacity Expansion
■ the sensitivity of your recommendations to the unit
cost values; Brite Power is a small power provider in the Finger
Lakes region of New York state. Because of Califor-
■ the sensitivity of your recommendations to the
nia’s power shortage in summer 2001, Brite Power’s
weight limit on the size of each module; and
board of directors has decided to commission a study
■ discussion about your confidence that you have the to ensure that the company has sufficient power until
optimal solution in light of what you have covered 2020. The study requires a time horizon of 16 years;
concerning convex functions and sets. the first decisions can be implemented in January
2005. Even though operations plans for power com-
panies are important, this study is at a higher level.

Our concern is with power capacity during the year plant in 2007, for example, then 65% of the costs oc-
and not with day-to-day or hour-to-hour power-usage cur in 2007 and 35% in 2008; the plant then comes
fluctuations. online and can be used to satisfy demand in 2009.
By the start of 2005, Brite will have plants with 60 With this lead time, it is clear that Brite Power needs
megawatts’ worth of production capacity. Estimates of to do advanced planning.
demand for power from the company for the years 2005 Brite Power can build plants with 5-, 10-, 15-, and
to 2020 have been made and are listed in Table 29. 20-megawatt capacities. If the company invests now
The economics of a coal-fired plant run according in research for new technologies ($3 million per year
to a power law—that is, the cost of a new plant in for 5 years), then it can reduce the exponent in the
constant dollars (sometimes called “year 0 dollars”) power model from 0.8 to 0.65.
follows the following estimation rule: Besides building new capacity, Brite Power must
operate plants; operations costs are based on the
cost of plant with capacity K
0.8
amount of capacity used. If demand in year t is Dt

capacity K
*cost of the base size plant megawatts and total capacity in year t is Ct megawatts,
capacity of base size plant then the operations cost in constant dollars for the
For this analysis, the base plant’s production capac- year is:
ity is 5 megawatts; its cost is $18 million. The company
DC
0.5
t
estimates inflation at 4% per year for the duration of * Dt * $400,000
t
the time horizon. The company uses a discount rate of
12% per year; this assumes that actual dollars are used By and large, the company must satisfy all demand
in the analysis [1 actual dollar in year 1 is equivalent to in the year, although there are opportunities to buy
1/(1.12) .89 dollars now in year 0]. 3 megawatts per year from neighboring power com-
The time required for constructing a new plant is panies at $600,000 per megawatt (in constant
two years. The project requires 65% of the cost at the dollars).
start of the first year; the remaining 35% is spent at
the start of the second year. If Brite Power starts a new
Key Questions
■ The Brite Board would like to know when to aug-
TA B L E 29 ment capacity. How big should the expansions be?
Demand Demand When should they start?
Year (Megawatts) Year (Megawatts) ■ What are the actual dollars spent over the time
2005 54 2013 87 horizon to acquire and satisfy demand? What is the
2006 58 2014 87 discounted value of these expenditures?
2007 63 2015 90 ■ Should the company invest in research to lower the
2008 63 2016 90 power law coefficient?
2009 69 2017 100
■ If demand estimates are increased or decreased,
2010 75 2018 110
then how would the plan change?
2011 77 2019 110
2012 77 2020 120 ■ If the $400,000 value in the operations cost
changes, then how would the plan change?
CASE 1 1 Brite Power: Capacity Expansion 1369

APPENDIX 3
Answers to Selected Problems

Chapter 2 SECTION 2.4
1 Linearly dependent.
SECTION 2.1
2 Linearly independent.

1 2 3
1 a A 4 5 6 SECTION 2.5
7 8 9

12 12 12
1
1 2 3

3 6 9 2 A
12 12 12
b 3A 12 15 18
21 24 27 3 A1 does not exist.
1
8 a 1B .
c A 2B is undefined. 100

1 4 7
SECTION 2.6
d AT 2 5 8
3 6 9 2 30.
0 1
2
1
e BT REVIEW PROBLEMS
1 2
1 Infinite number of solutions of the form x1 k 1,

4 6 x2 3 k, x3 k.
f AB 10 15
T 0.20 T
Ut1 0.75 0 Ut
3
16 24 t1 0.90 t
g BA is undefined. 4 x1 0, x2 1.

y1 0.50 0 0.10 x1 13 Linearly independent.
2 y2 0.30 0.70 0.30 x2 14 Linearly dependent.
y3 0.20 0.30 0.60 x3
15 a Only if a, b, c, and d are all nonzero will rank A
4. Thus, A1 exists if and only if all of a, b, c, and d
SECTION 2.2 are nonzero.
b Applying the Gauss-Jordan method, we find if a, b,

1 1 4 1 1 4 c, and d are all nonzero,
x
1 2 1 1 6 or 2 1 6

x 1 0 0 0
1 3 2 8 1 3 8 a
1
0 0 0
A1 b
0 0 1 0
SECTION 2.3 c
0 0 0 1d
1 No solution. 18 4.
2 Infinite number of solutions of the form x1 2 2k,
x2 2 k, x3 k.
3 x1 2, x2 1.
Chapter 3 s.t. 8(x1 x4 x5 x6 x7) 4(x8 x11 x12
x13 x14) 88 (Sunday)
SECTION 3.1
s.t. 8(x1 x2 x5 x6 x7) 4(x8 x9 x12
1 max z 30x1 100x2 x13 x14) 136 (Monday)
1 s.t. x1 x2 7 (Land constraint) s.t. 8(x1 x2 x3 x6 x7) 4(x8 x9 x10
4x1 10x2 40 (Labor constraint) x13 x14) 104 (Tuesday)
1 s.t. 10x1 + 10 x2 30 (Government constraint) s.t. 8(x1 x2 x3 x4 x7) 4(x8 x9 x10
x1, x2 0 x11 x14) 120 (Wednesday)
2 No, the government constraint is not satisfied. s.t. 8(x1 x2 x3 x4 x5) 4(x8 x9 x10
b No, the labor constraint is not satisfied.
x11 x12) 152 (Thursday)
c No, x2 0 is not satisfied.
s.t. 8(x2 x3 x4 x5 x6) 4(x9 x10 x11
SECTION 3.2 x12 x13) 112 (Friday)
1 z $370, x1 3, x2 2.8. s.t. 8(x3 x4 x5 x6 x7) 4(x10 x11 x12
3 z $14, x1 3, x2 2. x13 x14) 128 (Saturday)
4 a We want to make x1 larger and x2 smaller, so we 20(x8 x9 x10 x11 x12 x13 x14)
move down and to the right. 0.25(136 104 120 152 112 128
b We want to make x1 smaller and x2 larger, so we
88)
move up and to the left.
b We want to make both x1 and x2 smaller, so we move (The last constraint ensures that part-time labor
down and to the left. will fulfill at most 25% of all labor requirements)
All variables 0
SECTION 3.3 3 Let x1 number of employees who start work on Sunday
1 No feasible solution. and work five days, x2 number of employees who start
work on Monday and work five days, . . . , x7 number of
2 Alternative optimal solutions. employees who start work on Saturday and work five days.
3 Unbounded LP. Also let o1 number of employees who start work on
Sunday and work six days, . . . , o7 number of employees
who start work on Saturday and work six days. Then the
SECTION 3.4 appropriate LP is
1 For i 1, 2, 3, let xi tons of processed factory i min z 250(x1 x2 x7)
waste. Then the appropriate LP is min z + 312(o1 o2 o7)
min z 15x1 10x2 20x3 s.t. x1 x4 x5 x6 x7 o1 o3 o4
s.t. 0.10x1 0.20x2 0.40x3 30 (Pollutant 1) o5 o6 o7 11 (Sunday)
s.t. 0.45x1 0.25x2 0.30x3 40 (Pollutant 2) s.t. x1 x2 x5 x6 x7 o1 o2 o4
x1, x2, x3 0 o5 o6 o7 17 (Monday)
It is doubtful that the processing cost is proportional to the s.t. x1 x2 x3 x6 x7 o1 o2 o3
amount of waste processed. For example, processing 10
tons of waste is probably not 10 times as costly as processing o5 o6 o7 13 (Tuesday)
1 ton of waste. The Divisibility and Certainty Assumptions s.t. x1 x2 x3 x4 x7 o1 o2 o3
seem reasonable.
o4 o6 o7 15 (Wednesday)
SECTION 3.5 s.t. x1 x2 x3 x4 x5 o1 o2 o3
1 Let x1 number of full-time employees (FTE) who o4 o5 o7 19 (Thursday)

start work on Sunday, x2 number of FTE who start work s.t. x2 x3 x4 x5 x6 o1 o2 o3
on Monday, . . . , x7 number of FTE who start work on
Saturday; x8 number of part-time employees (PTE) who o4 o5 o6 14 (Friday)
start work on Sunday, . . . , x14 number of PTE who start s.t. x3 x4 x5 x6 x7 o2 o3 o4
work on Saturday. Then the appropriate LP is
o5 o6 o7 16 (Saturday)
min z 15(8)(5)(x1 x2 x7)
All variables 0
min z 10(4)(5)(x8 x9 x14)
APPENDIX 3 Answers to Selected Problems 1371

SECTION 3.6 1 max z x3 2o1 3o2
2 NPV of investment 1 1 max z s.t. o1 2x1 x2
5 7 9 1 max z s.t. o2 3x1 3x2 2x3

= 6 2 3 $2.00.
1.1 (1.1) (1.1) 1 max z s.t. o1 200
NPV of investment 2 1 max z s.t. o2 300
3 9 7 g2 3x3 x1 3x2 (Gas 2 production)
= 8 2 3 $1.97.
1.1 (1.1) (1.1) x1 x2 x3 100 (100 hours per week
Let x1 fraction of investment 1 that is undertaken and of cracker time)
x2 fraction of investment 2 that is undertaken. If we All variables 0
measure NPV in thousands of dollars, we want to solve the
following LP: 5 Let A total number of units of A produced
max z 2x1 1.97x2 B total number of units of B produced
s.t. 6x1 8x2 10 CS total number of units of C produced (and

sold)
5x1 3x2 7
AS units of A sold
x1 3x2 1
BS units of B sold
x2 1
max z 10AS 56BS 100CS
All variables 0
s.t. A 2B 3C 40
The optimal solution to this LP is x1 1, x2 0.5, z
$2,985. A AS 2B
B BS CS
SECTION 3.7 All variables 0
1 z $2,500, x1 50, x2 100.
SECTION 3.10
SECTION 3.8 1 Let xt production during month t and it inventory
at end of month t.
1 Let ingredient 1 sugar, ingredient 2 nuts, ingredient
3 chocolate, candy 1 Slugger, and candy 2 Easy 1 min z 5x1 8x2 4x3 7x4
Out. Let xij ounces of ingredient i used to make candy j.
1 min z = 2i1 2i2 2i3 2i4 6i4
(All variables are in ounces.) The appropriate LP is
1 s.t. i1 x1 50
max z 25(x12 x22 x32) 20(x11 x21 x31)
1 s.t. i2 i1 x2 65
s.t. x11 x12 100 (Sugar constraint)
1 s.t. i3 i2 x3 100
x21 x22 20 (Nuts constraint)
1 s.t. i4 i3 x4 70
x31 x32 30 (Chocolate constraint)
All variables 0
x22 0.2(x12 x22 x32)
x21 0.1(x11 x21 x31) SECTION 3.11
x31 0.1(x11 x21 x31)
3 Let A dollars invested in A, B dollars invested in
All variables 0 B, c0 leftover cash at time 0, c1 leftover cash at time
1, and c2 leftover cash at time 2. Then a correct
SECTION 3.9 formulation is
max z c2 1.9B
1 Let x1 hours of process 1 run per week
s.t. A c0 10,000
(Time 0 available time 0 invested)
0.2A c0 B c1
1 Let g2 barrels of gas 2 sold per week
1 Let o1 barrels of oil 1 purchased per week
1.5A c1 c2
1 Let o2 barrels of oil 2 purchased per week
1 max z 9(2x1) 10g2 24(2x3) 5x1 4x2
All variables 0
1372 APPENDIX 3 Answers to Selected Problems

The optimal solution to this LP is B c0 $10,000, A min z 100x1 90x2 120x3 80x4
c1 c2 0, and z $19,000. Notice that it is optimal to
s.t. x1 x2 100
wait for the “good” investment (B) even though leftover
cash earns no interest. (Farm 1 land)
x3 x4 100
SECTION 3.12
(Farm 2 land)
2 Let JAN1 number of computers rented at beginning 500x1 650x3 7,000
of January for one month, and so on. Also define IJAN
number of computers available to meet January demand, (Corn requirement)
and so on. The appropriate LP is 400x2 350x4 11,000
min z 200(JAN1 FEB1 MAR1 APR1 (Wheat requirement)
min z + MAY1 JUN1) 350(JAN2 FEB2 x1, x2, x3, x4 0
min z + MAR2 APR2 MAY2 JUN2) 9 Let x1 units of process 1, x2 units of process 2, and
min z + 450(JAN3 FEB3 MAR3 APR3) x3 modeling hours hired. Then a correct formulation is
min z + MAY3 JUN3) 150MAY3 max z 5(3x1 5 x2) 3(x1 2x2)
min z 300 JUN3 175JUN2 2(2x1 3x2) 100x3
s.t. IJAN JAN1 JAN2 JAN3 s.t. x1 2x2 20,000 (Limited labor)
IFEB IJAN JAN1 FEB1 FEB2 FEB3 2x1 3x2 35,000 (Limited chemicals)
s.t. IMAR IFEB JAN2 FEB1 MAR1 3x1 5x2 1,000 200x3
MAR2 MAR3 (Perfume production perfume demands)
IAPR IMAR FEB2 MAR1 JAN3 x1, x2, x3 0
APR1 APR2 APR3 17 Let OT number of tables made of oak, OC number
of chairs made of oak, PT number of tables made of pine,
IMAY IAPR FEB3 MAR2 APR1 and PC number of chairs made of pine. Then the correct
s.t. IMAY MAY1 MAY2 MAY3 formulation is
IJUN IMAY MAR3 APR2 MAY1 max z 40(OT PT) 15(OC PC)
JUN1 JUN2 JUN3 s.t. 17(OT) 5(OC) 150
IJAN 9 (Use at most 150 board ft of oak)
IFEB 5 30PT 13PC 210
IMAR 7 (Use at most 210 board ft of pine)
IAPR 9 OT, OC, PT, PC 0
IMAY 10 18 Let school 1 Cooley High, and school 2 Walt
Whitman High. Let Mij number of minority students who
IJUN 5 live in district i who will attend school j, and let NMij
All variables 0 number of nonminority students who live in district i who
will attend school j. Then the correct LP is
REVIEW PROBLEMS min z (M11 NM11) 2(M12 NM12)
2 Let x1 number of chocolate cakes baked and x2 2(M21 NM21) (M22 NM22)
number of vanilla cakes baked. Then we must solve (M31 NM31) (M32 NM32)
max z x1 12 x2 s.t. M11 M12 50
1 2
s.t. x1
3
x2
3
8 M21 M22 50
4x1 x2 30 M31 M32 100
x1, x2 0 NM11 NM12 200
$69 36 66
The optimal solution is z ,
7
x1 ,
7
x2 .
7 NM21 NM22 250
8 Let x1 acres of farm 1 devoted to corn, x2 acres of NM31 NM32 150
farm 1 devoted to wheat, x3 acres of farm 2 devoted to
corn, x4 acres of farm 2 devoted to wheat. Then a correct
formulation is

For school 1, we obtain the following blending constraints: s.t. X16 X26 X36 X46 X56 X67 9
M11 M21 M31 (Friday constraint)
0.20
M11 M21 M31 NM11 NM21 NM31 All variables 0
0.30 49 Let Xij money invested at beginning of month i for
This yields the following two LP constraints: a period of j months. After noting that for each month
(money invested) (bills paid) (money available), we
0.8M11 0.8M21 0.8M31 0.2NM11 obtain the following formulation:
0.2NM21 0.2NM31> 0 max z 1.08X14 1.03X23 1.01X32 1.001X41
0.7M11 0.7M21 0.7M32 0.3NM11 s.t. X11 X12 X13 X14 600 400 400
0.3NM21 0.3NM31 0 (Month 1)
For school 2, we obtain the following blending constraints: s.t. X21 X22 X23 500 1.001X11 800
M12 M22 M32 (Month 2)
0.20
M12 M22 M32 NM12 NM22 NM32 s.t. X31 X32 500 1.01X12 1.001X21 300
0.30 (Month 3)
This yields the following two LP constraints: s.t. X41 250 1.03X13 1.01X22 1.001X31 300
0.8M12 0.8M22 0.8M32 0.20NM12 (Month 4)
0.20NM22 0.20NM32 0 All variables 0
0.7M12 0.7M22 0.7M32 0.30NM12 53 Let T1 number of type 1 turkeys purchased
0.30NM22 0.30NM32 0 53 Let T2 number of type 2 turkeys purchased
We must also ensure that each school has between 300 and D1 pounds of dark meat used in cutlet 1
500 students. Thus, we also need the following constraints:
W1 pounds of white meat used in cutlet 1
300 M11 NM11 M21 NM21 M31 NM31
D2 pounds of dark meat used in cutlet 2
500
W2 pounds of white meat used in cutlet 2
300 M12 NM12 M22 NM22 M32 NM32
Then the appropriate formulation is
500
max z 4(W1 D1) 3(W2 D2) 10T1 8T2
To complete the formulation, add the sign restrictions that
all variables are 0. s.t. W1 D1 50 (Cutlet 1 demand)
47 For i j, let Xij number of workers who get off days s.t. W2 D2 30 (Cutlet 2 demand)
i and j of week (day 1 Sunday, day 2 Monday, . . . , day s.t. W1 W2 5T1 3T2 (Don’t use more white
7 Saturday). meat than you have)
max z X12 X17 X23 X34 X45 X56 X67 D1 D2 2T1 3T2 (Don’t use more dark meat
s.t. X17 X27 X37 X47 X57 X67 2 than you have)
(Saturday constraint) W1/(W1 D1) 0.7 or 0.3W1 0.7D1
s.t. X12 X13 X14 X15 X16 X17 12 W2/(W2 D2) 0.6 or 0.4W2 0.6D2
(Sunday constraint) T1, T2, D1, W1, D2, W2 0
s.t. X12 X23 X24 X25 X26 X27 12
(Monday constraint) Chapter 4
s.t. X13 X23 X34 X35 X36 X37 6
SECTION 4.1
(Tuesday constraint)
1 max z 3x1 2x2
s.t. X14 X24 X34 X45 X46 X47 5
1 s.t. 2x1 x2 s1 + s2 s3 100
(Wednesday constraint)
x1 x2 + s1 + s2 80
s.t. X15 X25 X35 X45 X56 X57 14
x1 s3 40
(Thursday constraint)

3 min z 3x1 x2 Set 1 {x1, x2, s1}
1 s.t. x1 + x2 e1 + s2 3 Set 2 {x1, x2, s2}
1 s.t. x1 x2 s2 4 Set 3 {x1, x2, s3}
1 s.t. 2x1 x2 3
SECTIONS 4.12 AND 4.13
SECTION 4.4 1 z 1, x1 x2 0, x3 1.
1 From Figure 2 of Chapter 3, we find the extreme points 4 Infeasible LP.
of the feasible region.
Point Basic Variables SECTION 4.14
H (0, 0) s1 100, s2 80, s3 40 1 Let it i

t it be the inventory position at the end of
E (40, 0) x1 40, s1 month t. For each constraint in the original problem, replace
20, s2 40
it by i
t it. Also add the sign restrictions i
t 0 and it 0.
F (40, 20) x1 40, x2 20, s2 20 To ensure that demand is met by the end of quarter 4, add
G (20, 60) x1 20, x2 60, s3 20 constraint i 4 0. Replace the terms involving it in the
D (0, 80) x2 80, s1 20, s3 40 objective function by
(100i
1 100i1 100i
2 100i2
SECTION 4.5
100i
3 110i3 100i
4 110i4)
1 z 180, x1 20, x2 60. 2 z 5, x1 1, x2 3.
32 10
2 z ,
3
x1 ,
3
x2 43.
SECTION 4.16
SECTION 4.6
2 Let xi number of lots purchased from supplier i. The
1 z 5, x1 0, x2 5. appropriate LP is
min z 10s
1 6s 2 4s 3 s 4
4.7 60x1 50x2 40x3 s1 s
1 5,000
SECTION s.t.
2 Solution 1: z 6, x1 0, x2 1; solution 2: z 6, (Excellent chips)
x1 5167, x2 4157. By averaging these two solutions, we 20x1 35x2 20x3 s2 s
s.t. 2 3,000
obtain solution 3: z 6, x1 2187, x2 3117 .
(Good chips)
SECTION 4.8 s.t. 20x1 15x2 40x3 s3 s
3 1,000
(Mediocre chips)
1 x1 4,999, x2 5,000 has z 10,000.
s.t. 400x1 300x2 250x3 s4 s4 28,000
SECTION 4.10 (Budget constraint)
1 a Both very small numbers (for example, 0.000003) All variables 0
and large numbers (for example, 3,000,000) appear in
the problem. REVIEW PROBLEMS
b Let x1 units of product i produced (in millions).
If we measure our profit in millions of dollars, the LP 4 Unbounded LP.
becomes 5 z 6, x1 0, x2 3.
1 max 6x1 4x2 3x3 6 Infeasible LP.
1 s.t. 4x1 3x2 2x3 3 (Million labor hours) 8 z 12, x1 x2 2.
1 s.t. 3x1 2x2 x3 2 (lb of pollution) 10 Four types of furniture.
1 s.t. 3x1, x2, x3 0 15 z 17
, x2 32, x4 12.
2
17 a c 0 and b 0.
SECTION 4.11
b b 0 and c 0. Also need a2 > 0 and/or a3 > 0
1 z 16, x1 x2 2. The point where all three constraints to ensure that when x1 is pivoted in, a feasible solution
are binding (x1 x2 2) corresponds to the following results. If only a3 > 0, then we need b to be strictly
three sets of basic variables: positive.

c c 0, a2 0, a3 0 ensures that x1 can be s.t. C3 200R3
made arbitrarily large and z will become arbitrarily
large. s.t. C4 200R4, I4 0
20 Let ct net number of drivers hired at the beginning B1, B2, B3, B4 2
of the year t. Then ct ht ft, where ht number of drivers All variables 0
hired at beginning of year t, and ft number of drivers fired
at beginning of year t. Also let dt number of drivers after
drivers have been hired or fired at beginning of year t. Then Chapter 5
a correct formulation is (cost in thousands of dollars)
min z 10(d1 d2 d3 d4 d5) SECTION 5.1
2(f1 f2 f3 f4 f5) 1 Decision variables remain the same. New z-value is
4(h1 h2 h3 h4 h5) $210.
s.t. d1 50 h1 f1 4 a 530 c1 350.
c 4,000,000 HIW 84,000,000; x1 3.6 0.15,
s.t. d2 d1 h2 f2
x2 1.4 0.025.
s.t. d3 d2 h3 f3 f $310,000.
s.t. d4 d3 h4 f4
s.t. d5 d4 h5 f5 SECTION 5.2
s.t. d1 60, d2 70, d3 50, d4 65, 1 a $3,875.

s.t. d5 75 b Decision variables remain the same. New z-value is
$3,750.
All variables 0 c Solution remains the same.
26 Let 3 a Still 90¢.
Rt robots available during quarter t
b 95¢.
(after robots are bought or sold for the quarter)
c 95¢.
Bt robots bought during quarter t d Still 90¢.
St robots sold during quarter t e 82.5¢.
I
t cars in inventory at end of quarter t f 30¢ or less.
g 22.5¢ or less.
Ct cars produced during quarter t
It backlogged demand for cars at end of quarter t SECTION 5.3
Then a correct formulation is
3 2.5¢.
min z 500(R1 R2 R3 R4)
4 $2.
200(I
1 I
2 I
3 I
4)
5 Buy raw material, because it will reduce cost by $6.67.
5,000(B1 B2 B3 B4)
3,000(S1 S2 S3 S4) SECTION 5.4
300(I 1 I2 I 3 I4) 3 See Figures 1–4.
s.t. R1 2 B1 S1
s.t. R2 R1 B2 S2 FIGURE 1
s.t. R3 R2 B3 S3
s.t. R4 R3 B4 S4
I
1 I 1 C1 600
I
2 I2 I
1 I 1 C2 800
I
3 I 3 I
2 I2 C3 500
I
4 I4 I
3 I 3 C4 400
s.t. R4 2
s.t. C1 200R1
s.t. C2 200R2

FIGURE 2 REVIEW PROBLEMS
1 a $1,046,667.
b Yes.
c $33.33.
b $333.33.
7 a Decision variables remain unchanged. New z-value
is $1,815,000.
b Pay $0 for an additional 100 board ft of lumber. Pay
$1,350 for an additional 100 hours of labor.
c $1,310,000.
d $1,665,000.
Chapter 6
SECTION 6.1
1 Decision variables remain the same. New z-value is

$210.
FIGURE 3
SECTION 6.2
1 z 4s1 5s2 28
x1 s1 s2 6
x2 s1 2s2 10
SECTION 6.3
3 x1 2, x2 0, x3 8, z 280 (same as original

solution).
5 Home computer tables should not be produced.
6 a Profit for candy bar 1 6¢. If type 1 candy bar
earns 7¢ profit, new optimal solution is z $3.50,
x1 50, x2 x3 0.
b 5¢ candy bar 2 profit 15¢. If candy bar 2 profit
is 13¢, decision variables remain the same, but profit is
now $4.50.
FIGURE 4 100
c sugar 100.
3
d z $3.40, x1 0, x2 20, x3 40. If 30 oz of
sugar is available, current basis is no longer optimal,
and problem must be solved again.
e Make type 1 candy bars.
f Make type 4 candy bars.
8 a $16,667.67 comedy cost $350,000.
b 4 million HIW 84 million. For 40 million HIW
exposures, new optimal solution is x1 5.4,
x2 1.1, z $350,000.
c Advertise on news program.
SECTION 6.4
1 Yes.
2 No.
4 Yes.

SECTION 6.5 SECTION 6.8
1 min w y1 3y2 4y3 2 b New z-value $3.40.
s.t. y1 y2 y3 2 c New z-value $2.60.
d Since current basis is no longer optimal, the current
y1 y2 2y3 1
shadow prices cannot be used to determine the new
y1, y2, y3 0 z-value.
2 max z 4x1 x2 3x3 5 b Skilled labor shadow price 0, unskilled labor
shadow price 0, raw material shadow price 15, and
s.t. 2x1 x2 x3 1
product 2 constraint shadow price 5.
x1 x2 2x3 1 We would be willing to pay $0 for an additional hour
of either type of labor. We would pay up to $15 for an extra
x1, x2, x3 0
unit of raw material. Reducing the product 2 marketing
3 min w 5y1 7y2 6y3 4y4 requirement by 1 unit will save the company $5.
s.t. y1 2y2 y4 4 c b3 5, so new z-value 435 5(15) $510.
d Since shadow price of each labor constraint is zero,
1 s.t. y1 y2 2y3 1 the optimal z-value remains unchanged.
1 s.t. y1 y2 2y3 y4 2 e For a 5-unit requirement, b4 2. Thus, new
1 s.t. y1 y1, y2 0; y3 0; y4 urs z-value 435 2(5) $425. For a 2-unit requirement,
b4 1. Thus, new z-value 435 (1)(5)
4 max z 6x1 8x2 $440.
s.t. x1 x2 4 6 a If purchased at the given price of $1, an extra unit
1 s.t. 2x1 x2 2 of raw material increases profits by $2.50. Thus, the firm
would be willing to pay up to 1 2.5 $3.50 for an
1 s.t. 2x1 2x2 1 extra unit of raw material.
1 s.t. x1 0; x2 urs b Both labor constraints are nonbinding. All we can
say is that if an additional hour of skilled labor were
available at $3/hour, we would not buy it, and if an
SECTION 6.7 additional hour of unskilled labor were available at
1 a min w 100y1 80y2 40y3 $2/hour, we would not buy it.
s.t. 2y1 y2 y3 3 7 a New z-value $380,000.

b New z-value $290,000.
y1 y2 2
1 a s.t. y1 , y2, y3 0 SECTION 6.9
b and c y1 1, y2 1, y3 0, w 180. Observe
that this solution has a w-value that equals the optimal 1 The current basis is no longer optimal. We should make
primal z-value. Since this solution is dual feasible, it computer tables, because they sell for $35 each and use only
must be optimal (by Lemma 2) for the dual. $30 worth of resources.
2 a min w 3y1 2y2 y3 2 a Current basis remains optimal if type 1 profit 6¢.
s.t. y1 y3 2
SECTION 6.10
y1 y2 1
1 a min w 600y1 400y2 500y3
y1 y2 y3 1
b y1 coefficient of s1 in optimal row 0 0 s.t. 4y1 y2 3y3 6
b y2 (coefficient of e2 in optimal row 0) 1 a s.t. 9y1 y2 4y3 10
b y2 1 1 a s.t. 7y1 3y2 2y3 9
b y3 coefficient of a3 in optimal row 0 M 1 a s.t. 10y1 40y2 y3 20
b y2 2 1 a s.t. 10y1 y1, y2, y3, y4 0
2,800 22 2
b w , y1 , y2 , y3 0.
b y2 coeffiOptimal w-value 0. 3 5 15
9 Dual is max w 28y1 24y2 SECTION 6.11

s.t. 7y1 2y2 50
1 z 9, x1 0, x2 14, x3 9.
2y1 12y2 100
2 a The current solution is still optimal.
y1, y2 0 b The LP is now infeasible.
Optimal dual solution is w $320,000, y1 5, y2 7.5.

c The new optimal solution is z 10, x1 1, 4 a New z-value 32,540 10(88) $33,420.
x2 4. Decision variables remain the same.
b Can’t tell, since allowable increase is 1.
SECTION 6.12 c $0.
d 32,540 (2)(20) $32,580.
4 Only HPER is inefficient.
e Produce jeeps.
REVIEW PROBLEMS 8 a New z-value $266.20.

b New z-value $270.70. Decision variables remain
1 a min w 6y1 3y2 10y3 the same.
1 a s.t. y1 y2 2y3 4 c $12.60.
d 20¢.
1 a s.t. 2y1 y2 y3 1
e Produce product 3.
1 a s.y1 urs; y2 0; y3 0
17 z 16, x1 8, x2 0.
Optimal dual solution is w 538, y1 23, y2 0, y3 73.
20 Optimal primal solution: z 13, x1 1, x2 x3 0,
b 9 b3 12. If b3 11, the new optimal solution
x4 2. Optimal dual solution: w 13, y1 1, y2 1.
is z 635, x1 136, x2 13.
21 a c1 3.
2 c1 12.
b c2 43.
3 a min w 6y1 8y2 2y3 c 0 b1 9.
3 a s.t. y1 6y2 + y3 5 d b2 10.
3 a s.t. y1 y3 1 28 LP 2 optimal solution: z 550, x1 0.5, x2 5.
100
3 a s.t. y1 y2 y3 2 Optimal solution to dual of LP 2: w 550, y1 y2 3
.
3 a s.t. y1, y2, y3 0 36 b2 3.

5 7
Optimal dual solution is w 9, y1 0, y2 ,
6
y3 .
6
b 0 c1 6.
c c2 76.
Chapter 7
SECTION 7.1

3

7 This is a maximization problem, so number in each cell
is a revenue, not a cost.
12 a Replace the M’s by incorporating a backlogging Laroach in month 1.

cost. For example, month 3 regular production can In Problem 7 of Section 7.1, Cliff gets 20,000 acres at
be used to meet month 1 demand at a cost of 400 site 1 and 20,000 acres at site 2. Blake gets 80,000 acres at
2(30) $460. site 2. Alexis gets 80,000 acres at site 1.
b Add a supply point called “lost sales,” with cost of
shipping a unit to any month’s demand being $450. SECTION 7.4
Supply of “lost sales” supply point should equal total
demand. Then adjust dummy demand point’s demand 2 Current basis remains optimal if c34 7.
to rebalance the problem.
4 New optimal solution is x12 12, x13 23, x21 45,
c A shipment from month 1 production to month 4 x23 5, x32 8, x34 30, and z 1,020 2(3) 2(10)
demand should have a cost of M. 994.
d For each month, add a month i subcontracting
supply point, with a supply of 10 and a cost that is $40
SECTION 7.5
more than the cost for the corresponding month i
regular supply point. Then adjust the demand at the 1 Person 1 does job 2, person 2 does job 1, person 3 does
dummy demand point so that the problem is balanced. no job, person 4 does job 4, and person 5 does job 3.
8 a Company 1 does route 1, company 2 does route 2,
SECTIONS 7.2 AND 7.3 company 3 does route 3, and company 4 does route 4.
The optimal solution to Problem 1 of Section 7.1 is to ship 10 b Company 3 does routes 3 and 1, company 2 does
units from warehouse 1 to customer 2, 30 units from warehouse routes 2 and 4.
1 to customer 3, and 30 units from warehouse 2 to customer 1.
The optimal solution to Problem 5 of Section 7.1 is to
buy 4 gallons from Daisy in month 2 and 3 gallons from
SECTION 7.6
1 a

b
REVIEW PROBLEMS
3 Meet January demand with 30 units of January from Bi to Bj means one black student from district i goes
production. Meet February demand with 5 units of January to school in district j. The costs of M ensure that shipments
production, 10 units of February production, and 15 units of from Wi to Bj or Bi to Wj cannot occur (table on next page).
March production. Meet March demand with 20 units of
8 Optimal solution is z 1,580, x11 40, x12 10, x13
March production.
10, x22 50, x32 10, x34 30.
4 Maid 1 does the bathroom, maid 2 straightens up, maid
13 Optimal solution is z 98, x13 5, x21 3, x24 7,
3 does the kitchen, maid 4 gets the day off, and maid 5
x32 3, x33 7, x34 5.
vacuums.
25 Sell painting 1 to customer 1, painting 2 to customer
7 Shipping 1 unit from Wi to Wj means one white student
2, painting 3 to customer 3, and painting 4 to customer 4.
from district i goes to school in district j. Shipping 1 unit

Chapter 8
SECTION 8.2
2 1–2–5 (length 14).

3
M large number to prevent shipping a unit through a 1 s.t. x0 xso,1 xso,2 (Node so)
nonexistent arc.
1 s.t. xso,1 x13 x12 (Node 1)
5 Replace the car at times 2, 4, and 6. Total cost $14,400.
1 s.t. x12 x32 xso,2 x24 (Node 2)
SECTION 8.3 1 s.t. x13 x32 x3,si (Node 3)

1 s.t. x24 x4,si (Node 4)
1 max z x0
1 s.t. x3,si x4,si x0 (Node si )
1 s.t. xso,1 6, xso,2 2, x12 1, x32 3,
1 s.t. All variables 0
1 s.t. x13 3, x3,si 2, x24 7, x4,si 7

Maximum flow 6. Cut associated with V
{2, 3, 4, si} SECTION 8.4
has capacity 6.
4 a See Figure 7.
2 max z x0
1 s.t. xso,1 2, x12 4, x1,si 3, x2,si 2,
FIGURE 7
1 s.t. x23 1, x3,si 2, xso,3 1
1 s.t. x0 xso,1 xso,3 (Node so)
1 s.t. xso,1 x1,si x12 (Node 1)
1 s.t. x12 x23 x2,si (Node 2)
1 s.t. x23 xso,3 x3,si (Node 3)
1 s.t. x1,si x2,si x3,si x0 (Node si )
1 s.t. All variables 0
Maximum flow 3. Cut associated with V
{1, 2, 3, si}
has capacity 3.
6 See Figure 5. An arc of capacity 1 goes from each
package type node to each truck node. If maximum flow b Critical path is A–C–G (project duration is 14 days).
21, all packages can be delivered. c Start project by June 13.
d min z x8 x1
FIGURE 5 d s.t. x2 x1 3
d s.t. x3 x2 6
d s.t. x4 x2 2
d s.t. x5 x4 3
d s.t. x5 x3 1
d s.t. x6 x5 1.5
d s.t. x8 x6 2
d s.t. x8 x7 3(x7 x3 2)
d s.t. x8 x2 3
d s.t. x8 x3 5
All variables urs
5 a See Figure 8. A–B–E–F–G and A–B–C–G are critical
paths. Duration of project is 26 days.
FIGURE 8
7 See Figure 6. If maximum flow 4, then all jobs can
be completed.
FIGURE 6
Activity Total Float Free Float
A 0 0
B 0 0
C 0 0
D 8 8
E 0 0
F 0 0
G 0 0

Books 3337 0 0 (1361-1400)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Books 3337 0 0 (1361-1400)

Uploaded by

Copyright:

Available Formats

The RISKTRIGEN Function

The RISKUNIFORM Function

The RISKGENERAL Function

APPENDIX 1 @Risk Crib Sheet 1345

1346 APPENDIX 1 @Risk Crib Sheet

APPENDIX 1 @Risk Crib Sheet 1347

Truncating Random Variables

1348 APPENDIX 1 @Risk Crib Sheet

Common Error Message

APPENDIX 1 @Risk Crib Sheet 1349

CASE 1 Help, I’m Not Getting Any Younger! 1351

CASE 2 Solar Energy for Your Home 1351

CASE 3 Golf-Sport: Managing Operations 1352

CASE 4 Vision Corporation: Production Planning and Shipping 1355

CASE 5 Material Handling in a General Mail-Handling Facility 1356

CASE 6 Selecting Corporate Training Programs 1359

CASE 7 BestChip: Expansion Strategy 1362

CASE 8 Emergency Vehicle Location in Springfield 1364

CASE 9 System Design: Project Management 1365

CASE 10 Modular Design for the Help-You Company 1366

CASE 11 Brite Power: Capacity Expansion 1368

Help, I’m Not Getting Any Younger! Key Questions

CASE 2 Solar Energy for Your Home 1351

1352 APPENDIX 2 Cases

Steel shafts 1 4 1.0

Steel shafts 3.5 7 1.1

Steel shafts 3 7.5 1.3

CASE 3 Golf-Sport: Managing Operations 1353

Steel shafts [0, 2,000] [0, 2,000] [0, 2,000]

1354 APPENDIX 2 Cases

CASE 4 Vision Corporation: Production Planning and Shipping 1355

1356 APPENDIX 2 Cases

CASE 5 Material Handling in a General Mail-Handling Facility 1357

TA B L E 15 sume that it takes one minute to perform a loading or

1 33 1 15 ■ Determine if this system has enough material-

1358 APPENDIX 2 Cases

CASE 6 Selecting Corporate Training Programs 1359

1 New employee orientation 22 Stress management

Senior Manager 250,000 0 1 0 1 0 0 0 0 0 0 0 1 1 1 0 1 0 1 1 0

1360 APPENDIX 2 Cases

CASE 6 Selecting Corporate Training Programs 1361

1362 APPENDIX 2 Cases

Jones Salt Lake City 1,300 900 1,700

shipping costs). Each site may produce as many as TA B L E 22

CASE 7 BestChip: Expansion Strategy 1363

1 100 200 1000 40,000

1364 APPENDIX 2 Cases

CASE 9 100 min [100, (tj xj)2.2]

CASE 9 System Design: Project Management 1365

1 Preliminary system description 40 —

■ How much money do you make on the project?

1366 APPENDIX 2 Cases

Adhesive Band-Aids 1 10 per pack 0.20

CASE 1 0 Modular Design for the Help-You Company 1367

Car Hiker Camping Sports Scouts

Ace bandage Tape

■ the sensitivity of your solution to the kit demand

1368 APPENDIX 2 Cases

CASE 1 1 Brite Power: Capacity Expansion 1369

Answers to Selected Problems

1 z $370, x1 3, x2 2.8. s.t. 8(x3 x4 x5 x6 x7) 4(x10 x11 x12

3 z $14, x1 3, x2 2. x13 x14)  128 (Saturday)

3 z $14, x1 3, x2 2. x13 x14) 128 (Saturday)

1 Let x1 number of full-time employees (FTE) who o4 o5 o7 19 (Thursday)

s.t. 6x1 8x2 10 CS total number of units of C produced (and

s.t. d1 60, d2 70, d3 50, d4 65, 1 a $3,875.

s.t. 2y1 y2 y3 3 7 a New z-value $380,000.

3 a s.t. y1, y2, y3 0 36 b2 3.