Module 12 Practice Problems

1) The Tulsa Symphony Orchestra Dilemma

Each year, the Smith, Jones and Sharda families join together to buy 6 season tickets for
the TSO and use a lottery style process to decide how they will split the tickets across the
performances.

This year, there are 6 performances spread out between September and May, all ‘themed’
on different countries – Germany (September), Italy (October), America (November),
England (January), France (March) and Latin America (May).

Given six performances, each family will then get to attend exactly 2 separate concerts.

The families have not been satisfied with the manner in which they split up the tickets in
the past, and so this year, they decide upon the following process. Each family is
allocated 100 points to use to “weight” their desire to attend each individual performance.
Thus, a family can ‘allocate’ 15 units to 5 of the concerts, and 25 units to another in
expressing their preferences (higher is more preferred). Once the families have done this
(secretly), then assignments of performances to families are done such that the overall
sum of ‘assigned’ preference points is maximized. So, for instance, in the example
below, if the Smith’s got the first two concerts, the Joneses’ the next two, and the Sharda
family the last two, that would result in a score of 33+13+9+21+13+12 = 101.

Many factors can be considered by the families – timing of the concerts, the country
being highlighted, other activities, etc.

Below is the preferences allocated by the three families.

Germany Italy America England France L.

America
Smith 33 13 11 17 16 10
Jones 26 14 9 21 17 13
Sharda 34 9 12 20 13 12

Using the objective stated above, find how this process would allocate performances to
families.

2) Christopher Cross’s Crew (adapted from John Toczek’s “Choose your crew” from
April 2011 OR/MS Today, pg. 14).

The table below shows an assessment of 9 possible crew members on 3 critical skills (and
their cost) for a long sea voyage. A captain’s most important decision for a long sea
voyage is selecting the right crew, and the different mix of skills necessary to be
successful (and survive). Such skills include fishing (have to eat), sailing (have to move
forward) and navigation (can’t get lost).

History has shown that to be successful, your crew must have a cumulative skill value of
at least15 units on each of the three critical skills. You can have as large as crew as you
need. What is the minimal cost crew that will lead to a successful voyage?

Name Fishing Sailing Navigation Cost (K)

Amy 3 5 1 46
Bill 1 2 5 43
Carl 3 4 2 47
Dan 4 3 1 36
Eva 4 2 2 43
Fred 1 3 4 55
Greg 3 1 5 68
Henry 5 4 2 64
Ida 3 3 3 60

3) Project Selection at US EAST Communcations

The Project Steering committee at US EAST Communications is sorting through its

information system project selection process for the next quarter. Twelve projects are
under consideration. Their attributes are shown below:
Project A B C D E F G H I J k L
TYPE CS CS E E OG OG EXP EXP EXP NEW NEW NEW
NPV 40 50 70 100 60 110 -20 50 100 70 280 220
CAPbud 200 175 80 100 100 125 100 200 175 125 240 160
Staff MM 100 80 40 50 75 100 50 175 100 85 150 220
Last Year? 100 75
Next Year? 1 1 0 0 0 0 0 1 1 0 1 1

TYPE = Critical (CS), Enhancement (E), On-Going (OG), Experimental (EXP) and New
process (NEW).
NPV = Assessed payoff in 000 – the objective to be maximized in the selection of our
projects.
CAPBud – capital budget for each project.
Staff MM – estimated staff expense, in Man-months.
Last Year: Budget expense for last year. 0 implied the project is not on-going.
Next Year? – A ‘1’ indicates it is a multi-year project

Create an IP model to select the best set of projects that maximize NPV. Items to
consider for selection:
a) There are 1000 units for CAPBud and 450 units for Staff MM.
b) EXP projects are very risky but might be very rewarding – select exactly one of these
projects.
c) Select at least 1 critical project.
d) No more than ½ of the projects selected should be multi-year projects (“Y” for Next
Year).
e) If Project B is chosen, Project J must be as well. The reverse is not true.
f) Run the model two ways – one forcing the on-going projects to be selected, another
model without forcing them (letting the model decide if it should select them ‘naturally’).

4) The China Syndrome’s Holiday Staffing- II

Take the model you built for Jane Turner in Module 11 and add the following restrictions
to a revised Integer programming model: Everyone must work at least once. People
cannot work both days at Thanksgiving or at New Year’s.

5) GBR Systems

You manage a small software development company called GBR Systems. You have 10
employees available for assignment to 2 projects. Each employee has an assessed rating
(higher means better performer) and you’ve noted which employees have earned
certification A and certification B (see table below).

Rating Cert A Cert B

Tagge 80 1
Brownson 76 1 1
Humm 69
Ferragamo 65 1
Luck 74 1
Garcia 90
Sorley 88
Quinn 62 1
Hager 71 1
Burns 74 1 1

Project Pelini requires at least 3 team members. There needs to be at least 2 people
assigned that have Certification A and at least 1 person assigned with Certification B.
They can overlap, obviously. Because Pelini is an important project, the weighted
average of the employees assigned to it must average at least 82.

Project Devaney also requires at least 3 team members, and 2 with Certification A. The
weighted average of the employees assigned to this project must average at least 75.
More than 3 employees may be assigned to each project. Find the optimal assignment –
your objective is to maximize the total employee rating of all assigned staff members
(aggregately for both projects).

Perhaps it should be mentioned that each employee can be assigned at most to one of the
two projects.

6) I’m A Pod Man!

You are a consultant to the selection committee for the Division III NAIA Basketball
tournament. The Championship is determined by a 16-team, single elimination
tournament. Four teams are assigned to each of four regions (KC, FARGO, ST. JOE,
WICHITA). Each region must have a #1 seed, a #2 seed, a #3 seed and a #4 seed. In each
region, first round games see the #1 seed play the #4 seed, and the #2 seed play the #3
seed. The winners play, and that winner advances to the Final Four.

The final 16 teams have already been determined, as have their seeds (there are 4 #1
seeds, 4 #2 seeds, etc.). Your task is to assign teams to regions while minimizing some
distance metric. The distance from the school to each of the four regional sites is shown
in the attached table. The schools conference affiliation and their seed are also shown in
the table. Note that “None” implies that the school is not in a conference. The only other
rule in assigning teams is that teams from the same conference must be assigned to
different regionals (this is an absolute rule and includes considering both games played at
the regional sites).

Create a model that is feasible and minimizes overall distance traveled by all teams
aggregately.
7) NCAA March Madness Scheduling – 2010 Revisited

Each year, there are many complaints about how the NCAA Men’s Basketball
Committee seeds and schedules teams in the 65-team single elimination tournament to
determine the National Champion. It is inevitable that somebody is unhappy.

You have been asked to look at some different scenarios in assigning teams to regions
and sub-regions using 2010 data so the committee can reevaluate their processes for
subsequent years. You will be using only the 8 teams that are #1 and #2 seeds, but all 4
regions and 8 sub-regions. The criterion used will be minimizing the sum of distance
traveled by all the teams involved, using their assigned regions and sub-regions as the
specific measure. Distance data is provided on a separate spreadsheet.

Some of the complaints and other aspects important in assigning teams to regions and
sub-regions that the committee has asked to be examined:

a) Each of the four regions (Midwest-St. Louis, West-Salt Lake City, East-Syracuse,
South – Houston) will have exactly one #1 seed assigned (from Kansas, Duke, Kentucky
and Syracuse) and exactly one #2 seed assigned (Kansas State, West Virginia, Villanova,
Ohio State).

b) NCAA scheduling policy requires that a team not play on its home court during the
tournament – so Syracuse (the team) cannot be assigned to the East region (in Syracuse,
NY).

c) Due to complaints from the host sites of the sub-regionals, each sub-region MUST
have either a #1 seed or a #2 seed assigned to it. Normally, this is NOT required (each
sub-region typically must have TWO total #1,2,3 or 4 seeds assigned – we are not
considering #3 and #4 seeds). The sub-regionals are held at sites in Buffalo, Providence,
Jacksonville, Milwaukee, New Orleans, Oklahoma City, San Jose and Spokane.

d) Teams from the same conference cannot be assigned to the same region. Of the 8
teams we are considering, Kansas and Kansas State are from the Big 12 Conference,
while Syracuse, West Virginia and Villanova are from the Big East Conference. Duke
(Atlantic Coast Conference), Kentucky (Southeast Conference), and Ohio State (Big
Eleven Conference) are the other affiliations. Note these were the conference affiliations
as of 2010.

e) A team cannot be assigned to a region and sub-region such that the selections choose
BOTH the “closest” region AND the “closest” sub-region. So if Team Y’s closest region
is Salt Lake City and its closest sub-region is Providence , Team Y can be assigned either
to Salt Lake City, OR Providence, OR neither, but NOT BOTH.
Scenario #1 - Implement an appropriate linear programming model that assigns the 8
teams to Regions and Sub-regions, minimizing the sum of overall distances subject to the
items listed above.

Scenario #2 – Augment the model above by removing restriction 3 – this means that a
sub-region can have 0,1, or 2 #1 and #2 seeds assigned to it.

Compare how the solution changes between Scenario #1 and Scenario #2. Use ‘words’
or ‘pictures’ or ‘tables’ or some kind of non-telepathic communication.

STL SLC H SYR

B12 KU 271 1092 758 1170
BE SYR 890 2061 1703 0
SEC KENT 346 1665 1070 667
ACC DUKE 803 2123 1174 636

B12 KSU 367 1010 731 1250

BE NOVA 889 2151 1543 254
B10 OSU 428 1711 1160 472
BE WVU 625 1921 1344 429

SPOKANE SJOSE OKC NO MILW JACK PROV BUFF

KU 1649 1860 319 875 612 1199 1440 1033
SYR 2462 2842 1384 1382 772 1076 306 150
KENT 2190 2444 839 748 470 717 895 519
DUKE 2599 2771 1188 852 879 479 664 691

KSU 1570 1778 290 963 687 1275 1515 1109

NOVA 2539 2920 1383 1219 850 847 271 404
OSU 2168 2479 914 911 448 807 737 330
WVU 2310 2690 1119 1021 620 758 569 287

8) University of South Oklahoma (USO) – Final Exams.

Due to global warming, a freak snow storm has hit the Dallas, TX campus of the
University of South Oklahoma during the first day of Finals. Six (6) business related
classes had their final exam postponed.

Now that the streets have been cleared, and order has been restored, the 6 classes need to
be ‘paired-up’ and scheduled in 3 different time blocks. (The actual times are irrelevant
since the exam will take place on Saturday). Administration wishes to pair-up the classes
(and thus give the students their finals) in such a way as to minimize the number of
conflicts – students who are in both classes assigned to one Final time block. The table
below shows the number of simultaneously enrolled students in each class-pair. Students
with conflicts will be taken care of outside the 3 time blocks – thus, their numbers should
be minimized. For instance, the table shows that the FIN and ACCT class share 15
students (there are 15 conflicts, 15 students who are taking both classes).

Create an appropriate IP model that determines how best to pair up classes, minimizing
the number of conflicts.

FIN ACCT ECON MGMT MSIS STAT

FIN 15 8 6 2 11
ACCT 5 4 9 12
ECON 10 13 7
MGMT 14 7
MSIS 3
STAT

9) Headline: Oklahoma tourism officials closing state parks

Associated Press

Oklahoma City — The Oklahoma Tourism and Recreation Department says a number of
state parks will close for good this year because of budget cuts.

A sign of the times – budget cuts forcing state tourism officials to close state parks. Was
this process does rationally?

Assume you are a Tourism and Recreation Department Director for a local region facing
similar issues. You manage 18 state parks (P1 thru P18) that are spread across 5 different
sub-regions (A thru E in the picture below).
Each park is in one of the 5 sub-regions A, B, C, D or E.

Each park contains one or more of the following facilities: CAMPING, CABINS,
LODGE.

Each park also has an ANNUAL budget and has a ‘VISITation’ parameter that represents
a combined measure of number of annual visitors and their ‘enjoyment’ level as
measured through surveys. This visitation parameter is on a scale of 1-100, with 100
being ‘mega-millions of visitors all perfectly satisfied with their experience’.

Data is shown below.

As a first attempt at determining which parks to keep open (therefore, which ones to
close), you establish the following requirements.

Keep enough parks open such that their combined (summed) measure of
visitation/enjoyment is at least 702 units. Obviously, you don’t want to close popular and
well-visited parks if you don’t have to.

In each sub-region, keeps parks open such that there is at least 1 park that has
CAMPING, 1 park that has CABINS and 1 park that has a LODGE. Note that 1 park
could conceivably fill all three requirements. In this case, you want to provide a variety
of park characteristics in each sub-region.

For each contiguous sub-region pair (i.e., sub-regions that border each other, such as Sub-
region A and Sub-region B), there must be at least a total of 4 parks across both sub-
regions combined. Similarly, you want to make sure that state parks are accessible to
everyone across the different sub-regions of the state. Again, this is for sub-region
PAIRS.

Your initial objective is to see how by much you can reduce the annual budget while
meeting these ‘park open’ requirements in deciding whether each park should remain
open or be closed.

P1 P2 P3 P4 P5 p6 P7 P8 P9 P10 P11 P12 P13 P14 P15 P16 P17 P18

Camping x x x x x x x x x x x x
Cabins x x x x x x x x x x x x
Lodge x x x x x x x x
Region A A A A B B B C C C D D D D E E E E
Annual 325 135 275 75 225 83 315 111 178 425 125 253 333 243 284 220 305 270
Visit 78 32 81 54 65 16 61 22 39 72 20 60 92 55 82 38 70 43

10) CHALCO CARTON COMPANY (CCC) - Revisited - The Chalco Carton

Company produces egg cartons (among other items) at three eastern Nebraska locations
(Gretna, Louisville, and, of course, Chalco). The company has just received contract to
provide a onetime shipment of egg cartons to ‘mom and pop’ grocery stores in Ashland,
Greenwood, Waverly and Eagle. The contract will pay CCC a fixed amount for the
satisfaction of the demand specified in the contract for the egg cartons at the grocery
stores.

A quick analysis of shipping costs using historical records and basic CCC accounting
principles leads to the following table which shows the per carton cost of transporting a
crate of cartons from the production locations to the grocery stores.

Ashland Greenwood Waverly Eagle

Gretna 10 7 8 15
Louisville 16 13 9 14
Chalco 12 11 12 16

Gretna has 75 crates of cartons in stock, Louisville, 80 crates, and Chalco, 55 crates. The
contract states that CCC needs to ship Ashland 50 creates, Greenwood 60 crates, Waverly
70 crates and Eagle, 65 crates.

You recognized early that you could not meet total demand, and you negotiated a penalty
clause – you agree to meet the demand of 3 of the 4 grocery stores. For the grocery store
whose demand was not met, there will be a penalty – Ashland’s penalty was $750,
Greenwood’s $900, Waverly $800 and Eagle, $1200.

Set up an appropriate IP model (one model) that minimizes total costs (shipping plus
penalty) while exactly meeting the demand of 3 of the 4 grocery stores.

11) A&W Bottlers supplies multiple lines of soda to the 11 county region in North
Central Oklahoma shown below. A&W acquired its warehouses over time via buying up
smaller operations, so it never really ‘designed’ its soda distribution supply chain. It has
warehouses in Enid, Guthrie, Stillwater, Perry and Sapulpa, and has determined that all of
the warehouses are likely not necessary.

Circles are warehouse locations:

Enid (Garfield), Guthrie (Logan),
Stillwater (Payne), Perry (Noble),
Sapulpa (Creek)
In analyzing A&W’s supply chain, the following monthly parameters have been
estimated/calculated.

Warehouse capacity (cases): Enid (10,000), Guthrie (12,500), Perry (15,000), Stillwater
(8,750), Sapulpa (20,000).

Monthly operating costs (if open): Enid (20,000), Guthrie (35,000), Perry (47,500),
Stillwater (17,500), Sapulpa (56,500).

Monthly county demand (cases):

County Demand County Demand County Demand

Grant 1000 Kay 750 Osage 6750
Garfield 2500 Noble 1500 Pawnee 4000
Kingfisher 1750 Payne 2000 Creek 2250
Logan 3500 Lincoln 5250

Distribution cost per case from warehouse to county (estimated):

Payn Cree
Grant Garfield Kfisher Logan e Lincoln k Pawnee Noble Kay Osage
Enid 2.2 0.4 3.1 4.7 4.3 8.3 11.3 5.7 2.6 5 13.1
Guthrie 7 5.2 3.3 0.3 1.8 3.1 8.4 4.9 4.5 10.3 10.6
Perry 3.6 2.4 5.1 4.6 2 6.2 8.1 3.1 0.2 4.6 7.6
Stillwater 6 4.1 5.4 2.1 0.1 4.8 5.9 3 2 7.1 8.2
Sapulpa 15 13.4 9.6 7.8 6.3 5.2 0.8 5.8 8.4 8.6 6.9

Using total monthly cost as your metric, determine which warehouses should be open,
and the optimal flow of soda from warehouse to county.

12) Linking Lumber

You are the order supervisor for a make-to-order Lumber company. You make three
different kinds of lumber products (A,B,C) which require two different resources to make
(RES1, RES2). You receive inquiries from two different possible customers:

Customer 1: Will pay you $7000 if you can provide 150 units of A,B,C combined in any
quantity.

Customer 2: Will pay you $4000 if you can provide 100 units of B and C combined in
any quantity.

Lumber A needs 2 units of both RES1 and RES2.

Lumber B needs 3 units of RES1 and 1 unit of RES2.
Lumber C needs 1 unit of RES1 and 4 units of RES2.
Both customers want their orders completed tomorrow. You have 325 units of RES1 and
275 units of RES2 available.

Create a model that helps you decide what you should promise (or which orders to take).

HINT: Find the optimal solution to an LP model where you are maximizing revenue of
the two customers, subject to the two resource constraints and the two constraints that
LINK the requirements for being paid and the payment (yes/no decisions on customers).