Chapter 8 TTAT

Chapter 8
TRANSPORTATION, TRANSSHIPMENT
ASSIGNMENT, AND TRAVELLING
SALESPERSON PROBLEM
Mar B. Cruz
TRANSPORTATION, TRANSSHIPMENT, ASSIGNMENT, AND TRAVELING SALESPERSON PROBLEMS | 3
8. Transportation, Transshipment, Assignment, and Travelling

Salesperson Problems
So far, we have discussed the general way of formulating and solving linear programming problems. There are linear programming
problems that have unique or special features/structures on their models which paved the way for the development of algorithms
that particularly solve such problems. In this section, we cover four special LP problems: transportation, transshipment,
assignment, and traveling salesperson problems.
8.1. Transportation Problems
In a transportation problem, the task is to determine the number of units to

be transported/shipped from various sources to various destinations that will
minimize the total transportation costs. The adjacent figure generally
illustrates the transportation problem where we have m sources and n
destinations.
Let i (1, 2, 3, …, m) be a source and j (1, 2, 3, …, n) be a destination so that

xij is the number of units to be transported from Source i to Destination j
corresponding to a per unit cost of cij. Meanwhile, let ai be the maximum
number of units that can be supplied by or transported from Source i and bj
be the number of units demanded or required by Destination j. When the
transportation problem is balanced where total supply is equal to total demand,
the transportation model is stated as
m n
Min Z = ∑ ∑ cij xij

i = 1j = 1
Subject to:
m n
Supply: ∑ xij = ai ∀i Demand: ∑ xij = bj ∀j

j=1 i=1
NN: xij ≥ 0.
Most of the time, however, total supply and total demand are not equal. A more general form of the model to capture the situation
where total supply is greater than or equal to the total demand is provided below by the revised constraints:
m n
Supply: ∑ xij ≤ ai ∀i Demand: ∑ xij ≥ bj ∀j

j=1 i=1
Meanwhile, when the total demand exceeds the total supply, the problem may be deemed infeasible because the premise is to
satisfy the demand. If such shipments are allowed or acceptable despite the lack of supply, the demand of a destination may or
may not be completely filled; hence, we use “≤” in the demand constraints instead while the supply constraint is in the form of
equality. If this supply constraints remains to be ≤-constraints, we will end up with zero values for all our decision variables. This
is not what we want if the idea is to exhaust the supply of each source to fulfill some or all of the units demanded by the
destination points.
In summary1,
Let xij – number of units to be transported from Source i to Destination j; i = 1, 2, 3, …, m; j = 1, 2, 3, …, n

m n

i = 1j = 1
1One peculiar characteristic of the transportation model is that there is one redundant constraint in the model. A thorough discussion of this instance is
discussed elsewhere.
MANUAL ON MANAGEMENT SCIENCE (MGT 221)

Subject to:
Total supply ≥ Total demand Total supply < Total demand
m m
Supply: ∑ xij ≤ ai ∑ xij = ai ∀i

j=1 j=1
n n
Demand: ∑ xij ≥ bj ∑ xij ≤ bj ∀j

i=1 i=1
NN: xij ≥ 0 xij ≥ 0.
where cij is the per unit cost of shipping from Source i to Destination j, ai is the supply or capacity of Source i and bj is the
demand or requirement of Destination j.
8.1.1. Modeling
EXAMPLE 8.1 Modeling – Transportation Problems
MG Auto. MG Auto has three plants in Los Angeles, Detroit and New Orleans, and two major distribution centers in Denver
and Miami. The capacities of the three plants during the next quarter are 1,000, 1,500, and 1,200 cars, respectively. The quarterly
demands at the two distribution centers are 2,300 and 1,400 cars, respectively. The transportation costs ($) from sources to
destinations are given below:
From\To Denver (1) Miami (2)

Los Angeles (1) 80 215
Detroit (2) 100 108
New Orleans (3) 102 68
Formulate an LP model that will minimize the total transportation costs.
Modeling
In the problem, we have three sources (Los Angeles, Detroit, and New Orleans) and
two destinations (Denver and Miami). Since the total supply and total demand are
equal at 3,700 cars or units, then we have a balanced transportation problem. The
figures provided in the table are the transportation costs on a per unit basis.
Our decision variables are x11, x12, x21, x22, x31 and x32; representing the number of
units to be transported from Los Angeles (1) to Denver (1), Los Angeles (1) to Miami
(2), Detroit (2) to Denver (1), Detroit (2) to Miami (2), New Orleans (3) to Denver
(1) and New Orleans (3) to Miami (2), respectively. With these, the corresponding
objective is
Min Z = 80x11 + 215x12 + 100x21 + 108x22 + 102x31 + 68x32
For the supply constraints, we want the total number of units transported from a source to the destinations to be equal or less
than the capacity of that source. Consider the case of Source 1, i.e., Los Angeles. What are the variables associated with this
source? The variables x11 and x12. These variables represent those outflows from this source, and the total should not exceed the
capacity of this source, which is 1,000 units. Mathematically, the constraint is expressed as
Los Angeles: x11 + x12 ≤ 1,000.
It is easy to see now that the remaining supply constraints are
Detroit: x21 + x22 ≤ 1,500

New Orleans: x31 + x32 ≤ 1,200.
Let us proceed now to the demand constraints. These constraints are those associated with the destinations. Consider the case of
Destination 1, i.e., Denver. Here, we talk with the total number of units being transported to Denver from the different sources;
such inflows to Denver are represented by the variables x11, x21 and x31. The sum of these variables should be at least the number
of units demanded by Denver.
Denver: x11 + x21 + x31 ≥ 2,300

Meanwhile, the remaining demand constraint is
Miami: x12 + x22 + x32 ≥ 1,400.
Finally, we include the NN constraint.
NN: x11, x12, x21, x22, x31, x32 ≥ 0
One may wish to add the integer constraint since we do not want fractional number of cars to be shipped. However, this may no
longer be necessary. As we solve the transportation model through the Transportation Simplex Algorithm, one will realize that
for as long as all our supply and demand parameters are integer values, then the values of the decision variables will be integers
as well.
In summary,
Let x11 – number of units to be transported from Los Angeles to Denver

x21 – number of units to be transported from Los Angeles to Miami
x31 – number of units to be transported from Detroit to Denver
x12 – number of units to be transported from Detroit to Miami
x22 – number of units to be transported from New Orleans to Denver
x32 – number of units to be transported from New Orleans to Miami
Min Z = 80x11 + 215x12 + 100x21 + 108x22 + 102x31 + 68x32
Subject to:
Los Angeles: x11 + x12 ≤ 1,000
Detroit: x21 + x22 ≤ 1,500
New Orleans: x31 + x32 ≤ 1,200
Denver: x11 + x21 + x31 ≥ 2,300
Miami: x12 + x22 + x32 ≥ 1,400
NN: x11, x12, x21, x22, x31, x32 ≥ 0.
o An optimal solution is: Z = $313,200; x12 = x31 = 0; x11 = 1,000, x21 = 1,300, x22 = 200, x32 = 1,200.
End.
MG Auto. Suppose now the problem is revised in two ways. First, the total supply is greater than the total demand. Second, the
total demand is greater than the total supply. In what follows presents a visual representation and the corresponding models
(constraints part) of these two situations.
A. Total Supply > Total Demand B. Total Supply < Total Demand
Subject to: Subject to:

Los Angeles: x11 + x12 ≤ 1,000 Los Angeles: x11 + x12 = 1,000
Detroit: x21 + x22 ≤ 1,500 Detroit: x21 + x22 = 1,300
New Orleans: x31 + x32 ≤ 1,200 New Orleans: x31 + x32 = 1,200
Denver: x11 + x21 + x31 ≥ 1,900 Denver: x11 + x21 + x31 ≤ 2,300
Miami: x12 + x22 + x32 ≥ 1,400 Miami: x12 + x22 + x32 ≤ 1,400
NN: x11, x12, x21, x22, x31, x32 ≥ 0 NN: x11, x12, x21, x22, x31, x32 ≥ 0
o An optimal solution is: Z = $273,200; x12 = x31 = o An optimal solution is: Z = $291,600; x12 = x22 =
0; x11 = 1,000, x21 = 900, x22 = 200, x32 = 1,200. x31 = 0; x11 = 1,000, x21 = 1,300, x32 = 1,200.
End.

Wheat-Wiw. Wheat is harvested in the Midwest and stored in grain elevators in three different cities, Kansas, Omaha, and Des
Moines. These grain elevators supply three flour mills, located in Chicago, St. Louis and Cincinnati. Grain is shipped to the mills
in railroad cars, each car capable of holding one ton of wheat. Each grain elevator can supply the following number of tons (i.e.,
railroad cars) of wheat to the mills monthly. The following tables summarized the supply and demand for each source and
destination, respectively.
Grain Elevator Supply Mill Demand

Kansas (1) 150 Chicago (1) 200
Omaha (2) 175 St. Louis (2) 100
Des Moines (3) 275 Cincinnati (3) 300
Total 600 Total 600
The cost of transporting one ton of wheat from each grain elevator (source) to each mill (destination) differs, according to the
distance and rail system. For example, the cost of shipping one ton of wheat from the grain elevator at Omaha to the mill at
Chicago is $7. These costs are shown in the following table.
From\To Chicago (1) St. Louis (2) Cincinnati (3)

Kansas (1) $6 8 10
Omaha (2) 7 11 11
Des Moines (3) 4 5 12
Formulate the corresponding LP model that will minimize the total transportation costs.
Modeling
The illustration of the problem and the corresponding model are provided below. The decision variables are generalized with xij
to save space. When the problem gets complicated, i.e., the number of sources and/or destinations increases, it pays to model the
problem in general form but be sure to define the indices properly.
Let xij – tons of grains to be shipped from Source i to Destination j; i, j = 1, 2, 3

Min Z = 6x11 + 8x12 + 10x13 + 7x21 + 11x22 + 11x23 + 4x31 + 5x32 + 12x33
Subject to:
Kansas: x11 + x12 + x13 ≤ 150
Omaha: x21 + x22 + x23 ≤ 175
Des Moines: x31 + x32 + x33 ≤ 275
Chicago: x11 + x21 + x31 ≥ 200
St. Louis: x12 + x22 + x32 ≥ 100
Cincinnati: x13 + x23 + x33 ≥ 300
NN: xij ≥ 0
o An optimal solution Z = $4,525; x13 = 150, x21 = 25, x23 = 150, x31 = 175, x32 = 100; x11 = x12 = x22 = x33 = 0.
End.
There are also problems that do not literally involve shipping physical goods; they can be modeled as transportation problems as
illustrated in the succeeding examples.
Royal Printing Corp. Royal Printing Corp. (RSC) receives an order to print 100 pcs. of Math books, 250 pcs. of Science books,
and 150 pcs. of Reading books. RSC has two machines where these books can be printed. Machine A produces a book at a rate
of 6 pcs. per hour, while the new Machine B can print ten books per hour. Assume that there are only 3.5 available working days
to print all these books before its deadline. At ten working hours per day, this translates to 35 hours availability of each machine.
As a result of different book features (e.g., Science has a lot of colored pictures,
Machine\Book Math Science Reading
while Reading is usually black-and-white text), cost of printing these books
also depends on which machine they are printed. The cost per piece, in Php, Machine A 21 27 19
is provided in the adjacent table. Formulate the corresponding model to find Machine B 25 30 23
the best book printing setup.

Modeling
Here, we can consider the two machines as the supply points, and the three disciplines are the destination points; so that the
problem is how many books the machines should “supply” to the three disciplines (Math, Science and Reading) to minimize the
total cost of printing the different books. Hence, the decision variables are described based on the printing machines and disciplines.
xAM – number of Math books printed with Machine A

xAS – number of Science books printed with Machine A
xAR – number of Reading books printed with Machine A
xBM – number of Math books printed with Machine B
xBS – number of Science books printed with Machine B
xBR – number of Reading books printed with Machine B
The objective is expressed as
Min Z = 21xAM + 27xAS + 19xAR + 25xBM + 30xBS + 23xBR.
For the supply constraints, we start with Machine A. How many books this machine can print? It says in the problem that it can
print at a rate of 6 pcs. per hour. With 35 working hours, the machine can print 210 books. Following this, Machine B can then
print 350 books over the 35-working-hour-duration. The total “supply” therefore is 560 books. Is this equal to the total number of
books demanded? The order stipulates that 100 pcs., 250 pcs. and 150 pcs. of Math, Science and Reading books, respectively,
should be printed; for a total of 500 books. The problem is not balanced, and specifically, the total “supply” is greater than the
total “demand.” The constraints, as discussed, are now specified as
Machine A: xAM + xAS + xAR ≤ 210 Math: xAM + xBM ≥ 100

Machine B: xBM + xBS + xBR ≤ 350 Science: xAS + xBS ≥ 250
Reading: xAR + xBR ≥ 150.
In summary,
Let xAM – number of Math books printed with Machine A

xAS – number of Science books printed with Machine A
xAR – number of Reading books printed with Machine A
xBM – number of Math books printed with Machine B
xBS – number of Science books printed with Machine B
xBR – number of Reading books printed with Machine B
Min Z = 21xAM + 27xAS + 19xAR + 25xBM + 30xBS + 23xBR
Subject to:
Machine A: xAM + xAS + xAR ≤ 210 Math: xAM + xBM ≥ 100
Machine B: xBM + xBS + xBR ≤ 350 Science: xAS + xBS ≥ 250
NN: xBR ≥ 0 Reading: xAR + xBR ≥ 150.
o An optimal solution is: Z = Php 12,610; xAS = xBR = 0; xAM = 60, xAR = 150, xBM = 40, xBS = 250.
End.
Pinoy Premium Pasta. With the expected uptick of demand during December for pasta noodles, Pinoy Premium Pasta plans to
determine the least-cost aggregate plan for the next four months. Backlogs are strictly not allowed as competitors will surely take
any unfulfilled demand for the product offered by the company. Furthermore, the hiring of part-time workers is not allowed. The
demand and the relevant costs are given below:
Mode of Production Monthly Capacity Unit Cost (labor + material)

Regular 500 boxes per worker Php 2,000 per box
Overtime 100 boxes per worker Php 2,500 per box
Subcontracting Not applicable Php 3,200 per box
September October November December

Demand (in boxes) 35,000 45,000 60,000 70,000
Holding cost per month is Php 50 per box. Initial inventory at the start of September is 5,000 boxes. At the end of December, at
least 6,000 boxes should be available as buffer inventory. The company has 85 workers. Provide the corresponding LP model to
find the optimal production plan for the company.

Modeling
How do we supply the monthly demands? By regular production, overtime and subcontracting, right? Further, we can use or
distribute the beginning inventory over the three months; but most of the time, beginning inventory is used or consumed
immediately. Hence, we can view the three modes of production in each of the three months plus the beginning inventory as the
supply points. Meanwhile, since each month requires a number of boxes of the pasta, the months September, October, November,
and December are considered as the destination points. To aid us in completing the elements of the model, we work with the
following table:
Destination
Source Capacity (boxes)
Sept. (1) Oct. (2) Nov. (3) Dec. (4)
Beg. Inventory (1) 100,000 100,000 100,000 100,000 5,000
Regular (2) 100,000 100,000 100,000 100,000 42,500
Sept.
Overtime (3) 100,000 100,000 100,000 100,000 8,500

Subcontracting (4) 100,000 100,000 100,000 100,000 M
Regular (5) 100,000 100,000 100,000 100,000 42,500
Oct.
Overtime (6) 100,000 100,000 100,000 100,000 8,500

Regular (8) 100,000 100,000 100,000 100,000 42,500
Nov.
Overtime (9) 100,000 100,000 100,000 100,000 8,500

Regular (11) 100,000 100,000 100,000 100,000 42,500
Dec.
Overtime (12) 100,000 100,000 100,000 100,000 8,500

Demand (boxes) 35,000 45,000 60,000 76,000
Let us complete the table one step at a time. We start with the monthly demands. As specified in the problem, the demands for
September, October, and November are 35,000 boxes, 45,000 boxes, and 60,000 boxes, respectively. For December, aside from the
70,000 boxes needed, we want to have a buffer of at least 6,000 boxes; Hence, we want to satisfy a demand of at least 76,000
boxes in December. The following table reflects such information:
Destination
Sept. (1) Oct. (2) Nov. (3) Dec. (4)
Beg. Inventory (1) 100,000 100,000 100,000 100,000 5,000
Regular (2) 100,000 100,000 100,000 100,000 42,500
Sept.
Overtime (3) 100,000 100,000 100,000 100,000 8,500

Regular (5) 100,000 100,000 100,000 100,000 42,500
Oct.
Overtime (6) 100,000 100,000 100,000 100,000 8,500

Regular (8) 100,000 100,000 100,000 100,000 42,500
Nov.
Overtime (9) 100,000 100,000 100,000 100,000 8,500

Regular (11) 100,000 100,000 100,000 100,000 42,500
Dec.
Overtime (12) 100,000 100,000 100,000 100,000 8,500

Demand (boxes) 35,000 45,000 60,000 76,000
Next, the capacity of each “supplier.” The beginning inventory is 5,000 boxes. This value also serves as the capacity of the
beginning inventory being a “supplier.” Meanwhile, for regular production for either month, the capacity is 42,500 boxes (500
boxes per worker x 85 workers). With overtime production, the capacity is 8,500 boxes (100 boxes per worker x 85 workers).
Finally, there is no limit to subcontracting. However, for modeling purposes, we put a ceiling to the number of units subcontracted
in such a way that the ceiling is very large, and it will appear that as if there is no ceiling to subcontracting. Therefore, the
capacity for subcontracting can be represented by M.

Destination
Sept. (1) Oct. (2) Nov. (3) Dec. (4)
Beg. Inventory (1) 100,000 100,000 100,000 100,000 5,000
Regular (2) 100,000 100,000 100,000 100,000 42,500
Sept.
Overtime (3) 100,000 100,000 100,000 100,000 8,500
Regular (5) 100,000 100,000 100,000 100,000 42,500
Oct.
Overtime (6) 100,000 100,000 100,000 100,000 8,500

Regular (8) 100,000 100,000 100,000 100,000 42,500
Nov.
Overtime (9) 100,000 100,000 100,000 100,000 8,500

Regular (11) 100,000 100,000 100,000 100,000 42,500
Dec.
Overtime (12) 100,000 100,000 100,000 100,000 8,500

Demand (boxes) 35,000 45,000 60,000 76,000
We are now to fill the remaining cells with unit cost values. Let us first reflect with the statement, “Backlog is strictly not allowed
as competitors will surely take any unfulfilled demand for the product offered by the company.” The statement imposes, for
example, that October production (whether regular, overtime or subcontracting) can address the demands of October, November,
and December but not September because backlogs are not allowed. For the algorithm to avoid assigning any “shipments” between
October production and September demand, we make such “shipments” costly. How? We again use M to signify that such
“shipments” are costly.
Destination
Sept. (1) Oct. (2) Nov. (3) Dec. (4)
Beg. Inventory (1) 100,000 100,000 100,000 100,000 5,000
Regular (2) 100,000 100,000 100,000 100,000 42,500
Sept.
Overtime (3) 100,000 100,000 100,000 100,000 8,500

Regular (5) M 100,000 100,000 100,000 42,500
Oct.
Overtime (6) M 100,000 100,000 100,000 8,500

Subcontracting (7) M 100,000 100,000 100,000 M
Regular (8) M M 100,000 100,000 42,500
Nov.
Overtime (9) M M 100,000 100,000 8,500

Subcontracting (10) M M 100,000 100,000 M
Regular (11) M M M 100,000 42,500
Dec.
Overtime (12) M M M 100,000 8,500

Subcontracting (13) M M M 100,000 M
Demand (boxes) 35,000 45,000 60,000 76,000
Let us now go to the costs of production. When a monthly demand is satisfied within the same production month, then we only
incur the productions costs; thus,
Destination
Sept. (1) Oct. (2) Nov. (3) Dec. (4)
Beg. Inventory (1) 0 100,000 100,000 100,000 5,000
Regular (2) 2,000 100,000 100,000 100,000 42,500
Sept.
Overtime (3) 2,500 100,000 100,000 100,000 8,500

Regular (5) M 2,000 100,000 100,000 42,500
Oct.
Overtime (6) M 2,500 100,000 100,000 8,500

Regular (8) M M 2,000 100,000 42,500
Nov.
Overtime (9) M M 2,500 100,000 8,500

Regular (11) M M M 2,000 42,500
Dec.
Overtime (12) M M M 2,500 8,500

Demand (boxes) 35,000 45,000 60,000 76,000
However, when we use our production to satisfy a demand later on (for example, regular production of a unit held in September
but to be used in October), we incur an additional cost – holding cost. In this case, the holding cost is Php 50 per box per month.
Thus, one box regularly produced on September and kept for October costs Php 2,050 (Php 2,000 + Php 50); when it is kept for
November, the cost will be Php 2,100 (Php 2,000 + Php 50 + Php 50).

Similarly, we arrived at the following updated table:
Destination
Sept. (1) Oct. (2) Nov. (3) Dec. (4)
Beg. Inventory (1) 0 50 100 150 5,000
Sept.Regular (2) 2,000 2,050 2,100 2,150 42,500
Overtime (3) 2,500 2,550 2,600 2,650 8,500
Regular (5) M 2,000 2,050 2,100 42,500
Oct.
Overtime (6) M 2,500 2,550 2,600 8,500

Regular (8) M M 2,000 2,050 42,500
Nov.
Overtime (9) M M 2,500 2,550 8,500

Regular (11) M M M 2,000 42,500
Dec.
Overtime (12) M M M 2,500 8,500

Demand (boxes) 35,000 45,000 60,000 76,000
Note that one of the assumptions in the model is that the beginning inventory in September can be distributed throughout the
four months. It is not necessary that all will be used for September and that it can be further stored at a cost to satisfy demand
in later months. With the preceding cost-table now, we can easily formulate now the LP model. In general, the model is
Let xi-j – number of boxes to be produced with Mode i and to be used to satisfy demand in Month j;
i = 1, 2, 3, …, 13; j = 1, 2, 3, 4
13 4
Min Z = ∑ ∑ ci-j xi-j

i = 1j = 1
Subject to:
13
Supply: ∑ xi-j ≤ ai ∀i = 1, 2, 3, …, 13
j=1
4
Demand: ∑ xi-j ≥ bj ∀j = 1, 2, 3, 4
i=1
NN: xi-j ≥ 0.
o An optimal solution is: Z = $450,700,000; x1-2 = x1-4 = x2-2 = x2-3 = x3-1 = x3-2 = x3-3 = x4-1 = x4-2 = x4-3 = x4-4 = x5-1 =
x5-4 = x6-1 = x6-3 = x6-4 = x7-1 = x7-2 = x7-3 = x7-4 = x8-1 = x8-2 = x8-4 = x9-1 = x9-2 = x9-4 = x10-1 = x10-2 = x10-3 = x10-4 =
x11-1 = x11-2 = x11-3 = x12-1 = x12-2 = x12-3 = x13-1 = x13-2 = x13-3 = 0; x1-1=2,000, x1-3 = 3,000, x2-1 = 33,000, x2-4 = 9,500,
x3-4 = 8,500, x5-2 = 36,500, x5-3 = 6,000, x6-2 = 8,500, x8-3 = 42,500, x9-3 = 8,500, x11-4 = 42,500, x12-4 = 8,500, x13-4 =
7,000.
o In summary:
i\j Sept. (1) Oct. (2) Nov. (3) Dec. (4) Supply
Beg. Inventory (1) x11 = 2,000 0 3,000 0 5,000
Regular (2) 33,000 0 0 9,500 42,500
Sep.
Overtime (3) 0 0 0 8,500 8,500

Subcontracting (4) 0 0 0 0 0
Regular (5) 0 36,500 6,000 0 42,500
Oct.
Overtime (6) 0 8,500 0 0 8,500

Regular (8) 0 0 4,2500 0 42,500
Nov.
Overtime (9) 0 0 8,500 0 8,500

Regular (11) 0 0 0 42,500 42,500
Dec.
Overtime (12) 0 0 0 8,500 8,500

Subcontracting (13) 0 0 0 7,000 7,000
Demand (boxes) 35,000 45,000 60,000 76,000 216000\216000
o From the summary table, it is recommended that we use 2,000 of the 5,000 boxes of our beginning inventory, and
produce 33,000 boxes to meet the demand in September. Meanwhile, for October demand, we resort to regular
production (36,500 boxes) and overtime production (8,500 boxes). The remaining figures are interpreted similarly.

Notice that the results indeed assigned 0 values for decision variables pertaining to backlogs; as backlogs were made
costly with the use of M.
End.
8.1.2. Transportation Simplex Algorithm
The Simplex Algorithm can practically solve any transportation model. Although, because of its special property or simplex
configuration, the transportation model can be solved differently through the Transportation Simplex Algorithm.
The algorithm requires the following: balanced transportation tableau and an initial basic feasible solution. After satisfying these
two requirements, the entering and basic variables are determined, and the necessary iterations follow. Just like the Simplex
Algorithm, the optimal solution is obtained when there is no longer entering variable. A thorough illustration of the Transportation
Simplex Algorithm is provided in what follows.
EXAMPLE 8.6 Transportation Simplex Algorithm – Transportation Problems
Wheat-Wiw. Recall the LP model from EXAMPLE 8.3.
Let xij – volume (tons) of grains to be shipped from Source i to Destination j; i, j = 1, 2, 3

Min Z = 6x11 + 8x12 + 10x13 + 7x21 + 11x22 + 11x23 + 4x31 + 5x32 + 12x33
Subject to:
Kansas: x11 + x12 + x13 ≤ 150
Omaha: x21 + x22 + x23 ≤ 175
Des Moines: x31 + x32 + x33 ≤ 275
Chicago: x11 + x21 + x31 ≥ 200
St. Louis: x12 + x22 + x32 ≥ 100
Cincinnati: x13 + x23 + x33 ≥ 300
NN: xij ≥ 0
Constructing the balanced transportation tableau
The transportation tableau is similar to the table we have constructed in Example 8.3. The following is the transportation tableau
for this example:
From\To Chicago (1) St. Louis (2) Cincinnati (3) Supply

6 8 10
Kansas (1) 150
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 275
Demand 200 100 300 600
Because the problem is already balanced (total supply is equal to total demand), then the transportation tableau is already
balanced.
The blank cells are associated with the decision variables. For example, the cell described as the intersection of Kansas row and
Chicago column is associated with the variable x11 and each ton transported from Kansas and Chicago costs $6. The total number
of units shipped from Kansas to Chicago, St. Louis and Cincinnati should be equal to 150 tons. Meanwhile, the 200 units demanded
by Chicago should be met by the sources.

EXAMPLE 8.3 showcases a balanced transportation problem. How do we construct now a balanced transportation tableau when
the transportation problem is not balanced? For illustration purposes, let us consider EXAMPLE 8.2.
A. Total Supply > Total Demand B. Total Supply < Total Demand
Subject to: Subject to:

Los Angeles: x11 + x12 ≤ 1,000 Los Angeles: x11 + x12 = 1,000
Detroit: x21 + x22 ≤ 1,500 Detroit: x21 + x22 = 1,300
New Orleans: x31 + x32 ≤ 1,200 New Orleans: x31 + x32 = 1,200
Denver: x11 + x21 + x31 ≥ 1,900 Denver: x11 + x21 + x31 ≤ 2,300
Miami: x12 + x22 + x32 ≥ 1,400 Miami: x12 + x22 + x32 ≤ 1,400
NN: x11, x12, x21, x22, x31, x32 ≥ 0 NN: x11, x12, x21, x22, x31, x32 ≥ 0
o An optimal solution is: Z = $273,200; x12 = x31 = o An optimal solution is: Z = $291,600; x12 = x22
0; x11 = 1,000, x21 = 900, x22 = 200, x32 = 1,200. = x31 = 0; x11 = 1,000, x21 = 1,300, x32 = 1,200.
How do we arrive with a balanced transportation tableau when total supply is greater than total demand? In Situation A,
there is an excess supply of 400 units; we want these units to be “demanded” as well so that the total supply will now be equal
to total demand. To do this, we add an artificial or dummy destination that will absorb these excess units. Note that there is
no cost involved in shipments between any legitimate source and dummy destination.
The corresponding balanced transportation tableau is
From\To Denver (1) Miami (2) Dummy (3) Supply

80 215 0
Los Angeles (1) 1,000
100 108 0
Detroit (2) 1,500
102 68 0
New Orleans (3) 1,200
Demand 1,900 1,400 400 3,700
We follow the same technique when total demand is greater than total supply; except that we have a dummy source to supply
the unmet demand of 200 units in the case of Situation B. The corresponding balanced transportation tableau is
From\To Denver (1) Miami (2) Supply

80 215
Los Angeles (1) 1,000
100 108
Detroit (2) 1,300
102 68
New Orleans (3) 1,200
0 0
Dummy (4) 0200
Demand 2,300 1,400 3,700
Likewise, there is no cost involved in shipments between a dummy source and any legitimate destination.

Determining an initial basic feasible solution
After securing a balanced transportation tableau, we are now ready to determine the initial basic feasible solution. There are
several ways to do it. For this part, we cover only two methods: Northwest Corner Method and Least-Cost Method. Other
methods are, to name a few, the Vogel’s Approximation Method and Russell’s Approximation Method.
Recall the balanced transportation tableau from the previous step.

6 8 10
Kansas (1) 150
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 275
Demand 200 100 300 600
Northwest Corner Method
As the term implies, we identify the northwest-most cell in every iteration. The northwest-most cell is the cell associated with
Kansas-Chicago; for brevity, we label this as Cell 1-1. In this cell, what is the maximum number of units that can be shipped
from Kansas to Chicago? It is 150 units. It cannot be greater than 150 because the capacity of Kansas is 150 units. Thus,

6 8 10
Kansas (1) 150 150, 0
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 275
Demand 200, 50 100 300 600
With 150 units, the supply of Kansas is exhausted (hence, we make the necessary shading). Meanwhile, the demand of Chicago
is down to 50 tons because Kansas source has fulfilled 150 of the 200 tons required.
Afterward, we repeat the same process – identify the new northwest-most cell among the remaining empty cells. This time, the
northwest-most cell is Cell 2-1. This cell is associated with Omaha-Chicago. What is the maximum number now that we can
transport from Omaha to Chicago? Since the demand in Chicago is only 50 tons now against the supply of 175 tons, then it is
worthwhile to transport only 50 units from Omaha to Chicago. Thus, during this process, we always select the minimum.

6 8 10
Kansas (1) 150 150, 0
7 11 11
Omaha (2) 50 175, 125
4 5 12
Des Moines (3) 275
Demand 200, 50, 0 100 300 600
The new northwest-most cell now is Cell 2-2 (Omaha-St. Louis). Between 125 tons and 100 tons, we select 100 tons to fill the
said cell. As a consequence,

6 8 10
Kansas (1) 150 150, 0
7 11 11
Omaha (2) 50 100 175, 125, 25
4 5 12
Des Moines (3) 275
Demand 200, 50, 0 100, 0 300 600
Now, we are left with Cell 2-3 and Cell 3-3. Between the two, it is Cell 2-3 that is the northwest most. The corresponding value
for the latter cell is 25 tons.

Continuing,

6 8 10
Kansas (1) 150 150, 0
7 11 11
Omaha (2) 50 100 25 175, 125, 25, 0
4 5 12
Des Moines (3) 275
Demand 200, 50, 0 100, 0 300, 275 600
Finally, it is clear to see that

6 8 10
Kansas (1) 150 150, 0
7 11 11
Omaha (2) 50 100 25 175, 125, 25, 0
4 5 12
Des Moines (3) 275 275, 0
Demand 200, 50, 0 100, 0 300, 275, 0 600
The tableau now with an initial basic feasible solution is

6 8 10
Kansas (1) 150 150
7 11 11
Omaha (2) 50 100 25 175
4 5 12
Des Moines (3) 275 275
Demand 200 100 300 600
The basic variables and corresponding values are x11 = 150, x21 = 50, x22 = 100, x23 = 25 and x33 = 275. The rest of the variables,
x12, x13, x31 and x32, are non-basic; hence, their values are 0. The total transportation costs associated with this solution is $5,925.00
(6 x 150 + $7 x 50 + $11 x 100 + $11 x 25 + $12 x 275).
When faced with a choice between 0 and a positive number, we of course select 0 and proceed as is. Such situations point to
the case of degeneracy. Degeneracy tends to exist because again, in transportation models, there is always one redundant
constraint, as mentioned earlier. Consider the situation below, illustrating the Northwest Corner Method and degeneracy.
From\To D1 (1) D2 (2) D3 (3) D4 (4) D5 (5) Supply

5 8 6 6 3
S1 (1) 8
4 7 7 6 5
S2 (2) 5
8 4 6 6 4
S3 (3) 9
1 5 2 2 6
S4 (4) 3
Demand 4 4 5 4 8 25

5 8 6 6 3
S1 (1) 4 84
4 7 7 6 5
S2 (2) 5
8 4 6 6 4
S3 (3) 9
1 5 2 2 6
S4 (4) 3
Demand 40 4 5 4 8 25


5 8 6 6 3
S1 (1) 4 4 840
4 7 7 6 5
S2 (2) 5
8 4 6 6 4
S3 (3) 9
1 5 2 2 6
S4 (4) 3
Demand 40 40 5 4 8 70
In case of ties, we simply write the maximum number of units in the cell and arbitrarily crossing (shading) out either the row
corresponding to 0-capacity or the column corresponding to 0-demand. Although, for our common agreement, we choose to
cross out the column. Thus, we shaded the D2 column in the tableau above upon realizing that both the supply and demand
were exhausted.
Continuing, the northwest-most cell now is the S1 – D3 cell or the cell associated with x13. Here, we ponder on what is the
maximum amount we can ship to D3 with reference to S1? The answer is 0. Hence,

5 8 6 6 3
S1 (1) 4 4 0 8400
4 7 7 6 5
S2 (2) 5
8 4 6 6 4
S3 (3) 9
1 5 2 2 6
S4 (4) 3
Demand 40 40 55 4 8 70
At this point, the value of x13 = 0. Because x13 is basic and equal to 0, degeneracy is observed. Recall in the LP chapter that
one of the special cases is degeneracy where a basic variable has a value of 0. We should not be worried about degeneracy in
this case. We can proceed with the Northwest Corner Method as is.
We should achieve with the following initial basic feasible solution:

5 8 6 6 3
S1 (1) 4 4 0 8400
4 7 7 6 5
S2 (2) 5 0 500
8 4 6 6 4
S3 (3) 4 5 950
1 5 2 2 6
S4 (4) 3 30
Demand 40 40 550 440 830 70
The initial solution for our transportation problem is defined by the following:
BVs NBVs
x11 = 4 x13 = 0 x24 = 0 x35 = 5 x14 = x15 = x21 = x22 = x25 = x31 =
x12 = 4 x23 = 5 x34 = 4 x45 = 3 x32 = x33 = x41 = x42 = x43 = x44 = 0

Least-Cost Method
In this method, instead of finding the northwest-most cell, we locate the cell containing the least cost. Let us work with the same
balanced transportation tableau.

6 8 10
Kansas (1) 150
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 275
Demand 200 100 300 600
What cell contains the least cost? It is Cell 3-1 (Des Moines-Chicago) with $4. How many units can we ship from Des Moines to
Chicago? It is 200 tons. Hence,

6 8 10
Kansas (1) 150
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 200 275, 75
Demand 200, 0 100 300 600
Which among the remaining empty cells contains now the least cost? It is Cell 3-2. The maximum number of tons that we can
ship from Des Moines to St. Louis is 75 tons. Thus,

6 8 10
Kansas (1) 150
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 200 75 275, 75, 0
Demand 200, 0 100, 25 300 600
We repeat the same process. With Cell 1-2,

6 8 10
Kansas (1) 25 150, 125
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 200 75 275, 75, 0
Demand 200, 0 100, 25, 0 300 600
Next, with Cell 1-3,

6 8 10
Kansas (1) 25 125 150, 125, 0
7 11 11
Omaha (2) 175
4 5 12
Des Moines (3) 200 75 275, 75, 0
Demand 200, 0 100, 25, 0 300, 175 600

Finally, with Cell 2-3,

6 8 10
Kansas (1) 25 125 150, 125, 0
7 11 11
Omaha (2) 175 175, 0
4 5 12
Des Moines (3) 200 75 275, 75, 0
Demand 200, 0 100, 25, 0 300, 175 600
The tableau now with an initial basic feasible solution is

6 8 10
Kansas (1) 25 125 150
7 11 11
Omaha (2) 175 175
4 5 12
Des Moines (3) 200 75 275
Demand 200 100 300 600
The basic variables and corresponding values are x12 = 25, x13 = 125, x23 = 175, x31 = 200 and x32 = 75. The non-basic variables
are x11, x21, x22 and x31. The total transportation cost (Z) is $4,550.00 ($8 x 25 + $10 x 125 + $11 x 175 + $4 x 200 + $5 x 75).
Here, we have five basic variables. Remember that the number of sources plus destinations less 1, m + n – 1, is equal to the
number of basic variables in transportation models. In this example, we have three sources and three destinations; so, 3 + 3 – 1
= 5. Further, there are two other things to remember particular to the transportation tableau. One, for each row or column, there
should be at least one filled cell. Second, there should be no cycles – the filled cells should not form any (closed) loop.
Determining the entering variable
There are two ways to determine the entering variable. The more convenient and intuitive way is the Stepping Stone Method.
Another approach is the use of the Method of Multipliers. This method is based on the duality property of the transportation
model. Further discussion on this is left elsewhere.
Recall the initial basic feasible solution obtained through the Northwest Corner Method from the previous step.

6 8 10
Kansas (1) 150 150
7 11 11
Omaha (2) 50 100 25 175
4 5 12
Demand 200 100 300 600
Stepping Stone Method
Referring to the tableau above, the NBV’s are x12, x13, x31 and x32. They have zero values – no unit-flow between the associated
source and destination. Consider x12 (Kansas-St. Louis). What if we allow one unit of flow from Kansas to St. Louis? What will
happen?

6 ◉ 8 10
Kansas (1) 150 +1 150
7 11 11
Omaha (2) 50 100 25 175
4 5 12
Demand 200 100 300 600

If a unit flow happens, then we exceed the capacity of Kansas. Hence, we make the necessary adjustments2 to maintain the supply
level of 150 units from Kansas.

150 6 ◉ 8 10
Kansas (1) +1 150
–1
7 11 11
Omaha (2) 50 100 25 175
4 5 12
Demand 200 100 300 600
However, we are short now of one unit for Chicago’s demand. Hence,

150 6 ◉ 8 10
Kansas (1) 150
–1 +1
50 7 11 11
Omaha (2) 100 25 175
+1
4 5 12
Demand 200 100 300 600
Again, we have a problem here. We have exceeded the capacity of Omaha by one unit. Thus, we have to subtract one unit from
the 100 units in Cell 2-2. As a result,

150 6 ◉ 8 10
Kansas (1) +1 150
–1
50 7 100 11 11
Omaha (2) 25 175
+1 –1
4 5 12
Demand 200 100 300 600
One may ask why we subtract 1 from 100 in Cell 2-2 and not from 25 in Cell 2-3? It is because when we do the latter and further
adjust that in Cell 3-3, then we find that there is no way now we can make further adjustment to maintain the 275-unit capacity
of the Des Moines. Thus, when we perform the simulation or adjustments, we should always total to the supply or demand. The
revised tableau is

6 8 10
Kansas (1) 149 1 150
7 11 11
Omaha (2) 51 99 25 175
4 5 12
Demand 200 100 300 600
The starting basic feasible solution from the Northwest Corner Method provides a Z = $5,925.00. With the preceding tableau
which resulted from the one-unit flow simulation in Cell 1-2, the Z-value is $5,923.00 ($6 x 149 + $215 x 1 + $8 x 1 + $7 x 51 +
$11 x 99 + $11 x 25 + $12 x 275). Causing a one-unit flow between Kansas and St. Louis results in a reduction of $2. This
reduction of $2 is the interpretation of –$2 computed from +$8 – $6 + $7 – $11. This sequence of positive and negative numbers
is a result of the unit-flow increase and unit-flow decrease in the concerned cells.3 The –2 is the rate of improvement associated
with Kansas-St. Louis.
Now that we know that there is a better solution than the starting basic feasible solution, we safely conclude that the starting
basic feasible solution or current solution, in general, is not yet optimal. There is at least one non-basic variable, say x12, that
when it becomes basic, the total transportation cost is reduced. The next question now is how about the other NBVs, x31, and
x32? When they become basic, do they reduce the Z? If so, do they provide a greater significant reduction in the Z than that of
x12? Let us see.
2 We can instead adjust Omaha-St. Louis first to maintain the 100-unit requirement of St. Louis. This is done by subtracting one-unit flow from the 100
units in Cell 2-2. This alternative approach implies that we can either work in a clockwise or counter-clockwise manner in the simulation of one-unit flow.
3 Note that the sequence of adding and subtracting a unit-flow in cells rendered the term “stepping stone”.

If we allow x13 to be basic, i.e., we allow, say, a unit-flow between Kansas and Cincinnati,

150 6 8 ◉ 10
Kansas (1) 150
–1 +1
50 7 11 25 11
Omaha (2) 100 175
+1 –1
4 5 12
Demand 200 100 300 600
The current Z of $5,925.00 is neither reduced nor improved because the associated rate of improvement for Kansas-Cincinnati is
0. How did we arrive at 0? It is from +$10 – $6 + $7 – $11. Meanwhile, with x31 or Des Moines-Chicago,

150 6 8 10
Kansas (1) 150
50 7 11 25 11
Omaha (2) 100 175
–1 +1
◉ 4 5 275 12
Des Moines (3) +1 275
–1
Demand 200 100 300 600
This time, the current Z is reduced by $4 with one 2unit of flow, i.e., $5,921.00. How did we know such? Because the rate of
improvement associated to Des Moines-Chicago is –4; computed from +$4 – $12 + $11 – $7.
Finally, with x32 or Omaha-St. Louis,

150 6 8 10
Kansas (1) 150
7 100 11 25 11
Omaha (2) 50 175
–1 +1
4 ◉ 5 275 12
Des Moines (3) 275
+1 –1
Demand 200 100 300 600
A further reduction in Z is achieved with a unit flow in Des Moines-St. Louis. The rate of improvement is +$5 – $12 + $11 – $11
= –$7. Thus, Z would be $5,918.00. In summary,
Rate of improvement
Associated variable
(ROI) ($)
Kansas-St. Louis x12 –2
Kansas-Cincinnati x13 0
Des Moines-Chicago x31 –4
Des Moines-St. Louis x32 –7
Since a unit flow in Des Moines-St. Louis provides the highest reduction in the current Z, the associated variable, x32, which is
currently non-basic, is chosen to become a basic variable. Thus, x32 is the entering variable.
When all rates of improvement are non-negative, then we say we have arrived at the optimal solution. Further, note that when
there is a rate of improvement of 0 with the optimal tableau or solution, then we have a case of alternative optima. The alternative
optimal solution is obtained by forcing the non-basic variable associated with the 0 rate of improvement to enter.

Before we end this section, few remarks are made. As you may have observed, the sequence of adding and subtracting a unit-
flow in the cells formed a rectangular or square loop. However, this may not always be the case.
For illustration, consider this new tableau:
From\To Mill 1 (1) Mill 2 (2) Mill 3 (3) Mill 4 (4) Supply
5 10 10 2 20 11
Silo 1 (1) 15
–1 +1
12 5 7 9 5 20
Silo 2 (2) 15 25
–1 +1
◉ 4 14 16 10 18
Silo 3 (3) 10
+1 –1
Demand 5 15 15 15 50
Another example of a loop is shown in the following tableau:
From\To Mill 1 (1) Mill 2 (2) Mill 3 (3) Supply

10 10 2 20 20
Silo 1 (1) 30
–1 +1
12 7 9
Silo 2 (2) 20 20
◉ 4 14 5 16
Silo 3 (3) +1 5
–1
5 3 10 10 2
Silo 4 (4) 15
–1 +1
Demand 25 20 25 70
m
Method of Multipliers
This time, let us work with the Method of Multipliers to identify the entering variable. Let us go back again to the initial basic
feasible solution derived from the Northwest Corner Method.

6 8 10
Kansas (1) 150 150
7 11 11
Omaha (2) 50 100 25 175
4 5 12
Demand 200 100 300 600
The first step in this method is to make the following associations:
v1 v2 v3
6 8 10
u1 Kansas (1) 150 150
7 11 11
u2 Omaha (2) 50 100 25 175
4 5 12
u3 Des Moines (3) 275 275
Demand 200 100 300 600
We associate multiplier ui with row i and multiplier vj with column j. Recall again that index i relates to a source while index j
relates to a destination. The next step now is the find the values of the multipliers that satisfy the equations: ui + vj = cij for
basic variables xij, with one multiplier (usually the first row’s multiplier) arbitrarily set to 0. The following table is then provided:
Basic variable Equation Solution and multiplier value

x11 u1 + v1 = c11 0 + v1 = 6; thus, v1 = 6
x21 u2 + v1 = c21 u2 + 6 = 7; thus, u2 = 1
x22 u2 + v2 = c22 1 + v2 = 11; thus, v2 = 10
x23 u2 + v3 = c23 1 + v3 = 11; thus, v3 = 10
x33 u3 + v3 = c33 u3 + 10 = 12; thus, u3 = 2

Thus,
v1 = 6 v2 = 10 v3 = 10
6 8 10
u1 = 0 Kansas (1) 150 150
7 11 11
u2 = 1 Omaha (2) 50 100 25 175
4 5 12
u3 = 2 Des Moines (3) 275 275
Demand 200 100 300 600
For each non-basic variable xij (empty cell), compute ui + vj – cij and indicate the value in the respective cells in parenthesis.
v1 = 6 v2 = 10 v3 = 10
6 8 10
u1 = 0 Kansas (1) 150 (2) (0) 150
7 11 11
u2 = 1 Omaha (2) 50 100 25 175
4 5 12
u3 = 2 Des Moines (3) (4) (7) 275 275
Demand 200 100 300 600
Those numbers in parenthesis can also be treated as the rates of improvement based on one-unit flow. However, they are
interpreted oppositely – the “more positive” the value, the higher the reduction in Z with a unit-flow. Consider these values as
the magnitude or degree of reduction in the Z with a unit flow. Therefore, in this case, the entering variable is x32. This is
consistent with that of the Stepping Stone Method.
Determine the leaving variable and iteration
Earlier in the Stepping Stone Method, our simulations were based on a one-unit flow; as well as how the ui + vj – cij values were
interpreted and used for selecting the entering variable, which is x32 (Des Moines-St. Louis). The question is why not two units
or more? The more the units can be shipped from Des Moines to St. Louis, the higher reduction in Z is achieved. With two units,
the reduction is $14 ($7 x 2). Meanwhile, with three units, the reduction is $21 ($7 x 3). We generalize now this maximum number
of flow units as 𝜃 (theta), so that

6 8 10
Kansas (1) 150 150
7 100 11 25 11
Omaha (2) 50 175
–𝜃 +𝜃
4 ◉ 5 275 12
Des Moines (3) 275
+𝜃 –𝜃
Demand 200 100 300 600
To find the value of 𝜃, we recall that all the decision variables are limited to non-negative values as demonstrated by the non-
negativity constraint. Hence,
x22 = 100 – 𝜃 ≥ 0 x22 = 100 – 𝜃 ≥ 0 x32 = 𝜃 ≥ 0 x33 = 275 – 𝜃 ≥ 0.
Now, think of the maximum value of 𝜃 that can simultaneously satisfy all the inequalities. It is clear to see that 𝜃 is equal to 100
units. As a result,

6 8 10
Kansas (1) 150 150
7 11 11
Omaha (2) 50 125 175
4 5 12
Des Moines (3) 100 175 275
Demand 200 100 300 600

Note that with 𝜃 = 100, x22 becomes 0 and thus becomes non-basic from being basic. Therefore, x22 is the leaving variable. Because
it becomes basic, the associated cell is left blank. Do not write 0 because writing zero signifies that x22 is basic equal and equal to
zero. At this point, the following describes the tableau:
BVs NBVs
x11 = 150 x32 = 100 x12 = 0 x22 = 0
x21 = 50 x33 = 175 x13 = 0 x31 = 0.
x23 = 125
Z = $6 x 150 + $7 x 50 + $11 x 125 + $5 x 100 + $12 x 175 = $5,225. Alternatively, Z = $5,925 + Rate of Improvement = –$7
x (𝜃 = 100) = $5,225. Is the solution optimal? Let us perform again simulations, i.e., compute for the new set of rates of
improvement to see whether the current solution is optimal. Those indicated in parenthesis are the rates of improvement (verify!).

6 8 10
Kansas (1) 150 (5) (0) 150
7 11 11
Omaha (2) 50 (7) 125 175
4 5 12
Des Moines (3) (–4) 100 175 275
Demand 200 100 300 600
We see from the preceding that the entering variable is x31. For the leaving variable, we have

6 8 10
Kansas (1) 150 150
50 7 11 125 11
Omaha (2) 175
–𝜃 +𝜃
◉ 4 5 175 12
+𝜃 –𝜃
Demand 200 100 300 600
so that
x21 = 50 – 𝜃 ≥ 0 x23 = 125 + 𝜃 ≥ 0 x31 = 𝜃 ≥ 0 x33 = 175 – 𝜃 ≥ 0.
Thus, 𝜃 = 50 and the leaving variable is x21. The iterated tableau is

6 8 10
Kansas (1) 150 150
7 11 11
Omaha (2) 175 175
4 5 12
Des Moines (3) 50 100 125 275
Demand 200 100 300 600
Does the tableau now provide the optimal solution? Let us compute the rates of improvement again. It is hoped that one sees
clearly how the following rates (those in parentheses) were computed:

6 8 10
Kansas (1) 150 (1) (–4) 150
7 11 11
Omaha (2) (4) (7) 175 175
4 5 12
Des Moines (3) 50 100 125 275
Demand 200 100 300 600
The entering variable now is x13 because it is the best rate of improvement; actually, the variable is the only non-basic variable
that brings a reduction in the current Z when it becomes basic.

For the leaving variable, we have

150 6 8 ◉ 10
Kansas (1) 150
–𝜃 +𝜃
7 11 11
Omaha (2) 175 175
50 4 5 125 12
+𝜃 –𝜃
Demand 200 100 300 600
so that
x11 = 150 – 𝜃 ≥ 0 x13 = 𝜃 ≥ 0 x31 = 50 + 𝜃 ≥ 0 x33 = 125 – 𝜃 ≥ 0.
Thus, 𝜃 = 125. With 𝜃 = 125, the leaving variable is x33. The revised tableau is

6 8 10
Kansas (1) 25 125 150
7 11 11
Omaha (2) 175 175
4 5 12
Des Moines (3) 175 100 275
Demand 200 100 300 600
Based on the tableau above, we see that the 125 units (𝜃) come from x33 making the variable as the leaving variable, i.e., it
has become non-basic. Since non-basic variables are associated with empty cells, then it is advised not to write any value
inside the cell. There is a difference between a cell containing 0 and an empty cell. The first cell described means that the
associated variable is basic and has a value of 0 (hence, a case of degeneracy) while an empty cell tells us that the associated
variable is non-basic and thus, has a value of 0. Consider the situation (from Taha) below where x31 is the entering variable.
5 10 10 2 20 11
Silo 1 (1) 15
–𝜃 +𝜃
12 5 7 9 5 20
Silo 2 (2) 15 25
–𝜃 +𝜃
◉ 4 14 16 10 18
Silo 3 (3) 10
+𝜃 –𝜃
Demand 5 15 15 15 50
Thus,
x31 = 𝜃 ≥ 0 x34 = 10 – 𝜃 ≥ 0 x24 = 5 + 𝜃 ≥ 0 x22 = 5 – 𝜃 ≥ 0 x12 = 10 + 𝜃 ≥ 0 x11 = 5 – 𝜃 ≥ 0
so that 𝜃 = 5. But the question is, where will we get this 5 units? We can get it from either x22 or x11. Suppose we want to get
it from x11 so that x11 will leave the solution or will become non-basic. Thus, we do not write 0 in the cell associated with x11
to signify that x11 has become non-basic. Meanwhile, by the iteration process, x22 becomes 0 but remains to be basic.
x11 has left the solution 10 2 20 11
(x11 became non-basic) Silo 1 (1) 15 15
12 7 9 20
x22 remains basic and equal Silo 2 (2) 0 15 10 25
to 0 (a case of degeneracy)
4 14 16 18
Silo 3 (3) 5 5 10
Demand 5 15 15 15 50
The story is consistent with that of the Simplex Algorithm. We know that degeneracy is about to happen when there are ties
for the leaving variable (as in the case of x11 and x22), and there is degeneracy where there is at least one basic variable that
is equal to 0 (as in the case of x11).

Let us check again whether the solution now is optimal. This again requires us to compute for the rates of improvement.

6 8 10
Kansas (1) 25 (1) 125 150
7 11 11
Omaha (2) (0) (3) 175 175
4 5 12
Des Moines (3) 175 100 (4) 275
Demand 200 100 300 600
All the rates now are non-negative. This means that we have arrived at the optimal solution. The optimal solution is
x11 = 25 x23 = 175 x32 = 100

x13 = 125 x31 = 1753 All remaining variables are equal to 0.
We ship 25 tons of grains from Kansas to Chicago; 125 tons from Kansas to Cincinnati; 175 tons from Omaha to Cincinnati, and
from Des Moines to Chicago; finally, 100 tons from Des Moines to St. Louis. The minimum total transportation costs, optimal Z,
is $6 x 25 + $10 x 125 + $11 x 175 + $4 x 175 + $5 x 100 = $4,525.
It is worthwhile to note also that we have an alternative solution as suggested by the rate of improvement associated with x21,
which is 0. If we force x21 to enter or to become basic,

25 6 8 125 10
Kansas (1) 150
–𝜃 +𝜃
◉ 7 11 175 11
Omaha (2) 175
+𝜃 –𝜃
4 5 12
Des Moines (3) 175 100 275
Demand 200 100 300 600
so that
x11 = 25 – 𝜃 ≥ 0 x13 = 125 + 𝜃 ≥ 0 x21 = 𝜃 ≥ 0 x23 = 175 – 𝜃 ≥ 0.
With 𝜃 = 25, the leaving variable is x11. The alternative optimal solution now is provided by the tableau below:

6 8 10
Kansas (1) 150 150
7 11 11
Omaha (2) 25 150 175
4 5 12
Des Moines (3) 175 100 275
Demand 200 100 300 600
Note that Z = $10 x 150 + $7 x 25 + $11 x 150 + $4 x 175 + $5 x 100 = $4,525. We have arrived at the same total transportation
cost provided by the earlier optimal solution.
End.
8.1.3. Problem Set
PROBLEM 8.1
Versatech Corporation has decided to produce three new products. Five branch plants now have excess production capacity. The
unit manufacturing cost of the first product would be $31, $29, $32, $28, and $29 in Plants 1, 2, 3, 4, and 5, respectively. The
unit manufacturing cost of the second product would be $45, $41, $46, $42, and $43 in Plants 1, 2, 3, 4, and 5, respectively. The
unit manufacturing cost of the third product would be $38, $35, and $40 in Plants 1, 2, and 3, respectively, whereas Plants 4 and
5 do not have the capability for producing this product. Sales forecasts indicate that 600, 1,000, and 800 units of products 1, 2,
and 3, respectively, should be produced per day. Plants 1, 2, 3, 4, and 5 can produce 400, 600, 400, 600, and 1,000 units daily,
respectively, regardless of the product or combination of products involved. Assume that any plant having the capability and
capacity to produce them can produce any combination of the products in any quantity. The management wishes to know how
to allocate the new products to the plants to minimize total manufacturing cost. Obtain an optimal solution.

PROBLEM 8.2
The B. Hall Real Estate Investment Corporation has identified four small apartment buildings in which it would like to invest.
Mrs. Hall has approached three savings and loan companies regarding financing. Because Hall has been a good client in the past
and has maintained a high credit rating in the community, each savings and loan company is willing to consider providing all or
part of the mortgage loan needed on each property. Each loan officer has set differing interest rates on each property (rates are
affected by the neighborhood of the apartment building, condition of the property, and desire by the individual savings and loan
to finance various-size buildings), and each loan company has placed a maximum credit ceiling on how much it will lend Hall in
total. This information is summarized in the accompanying table.
Property (Interest rates) (%) Maximum

Savings and loan company
Hill St. Banks St. Park Ave. Dury Lane credit line ($)
First Homestead 8 8 10 11 80,000
Commonwealth 9 10 12 10 100,000
Washington Federal 9 11 10 9 120,000
Loan required to purchase building $60,000 $40,000 $130,000 $70,000
Each apartment building is equally attractive as an investment to Hall, so she has decided to purchase all buildings possible at
the lowest total payment of interest. From which savings and loan companies should she borrow to purchase which buildings?
More than one savings and loan can finance the same property.
o Let xij be the amount to be borrowed from Company i to purchase a building in Location j; i = 1 (First Homestead),
2 (Commonwealth), 3 (Washington Federal); j = 1 (Hill St.), 2 (Banks St.), 3 (Park Ave.), 4 (Dury Lane).
o An optimal solution is: Z = $28,300; x11 = 30,000, x12 = 40,000, x13 = 10,000, x21 = 30,000, x24 = 70,000, x33 = 120,000;
all remaining decision variables are equal to 0.
PROBLEM 8.3
Cars are shipped from three distribution centers to five dealers. The shipping cost is based on the mileage between the sources
and the destinations and is independent of whether the truck makes the trip with partial or full loads. The following table
summarizes the mileage between the distribution centers and the dealers together with the monthly supply and demand figures
given in number of cars. A full truck truckload includes 18 cars. The transportation cost per truck mile is $25. Determine the
optimal shipping schedule.
Center\Dealer 1 2 3 4 5 Supply
1 100 150 200 140 35 400
2 50 70 60 65 80 200
3 40 90 100 150 130 150
Demand 100 200 150 160 140
o Let xij be the number of cars to be shipped from Center i to Dealer j; i = 1, 2, 3; j = 1, 2, 3, 4, 5.

o An optimal solution is: Z = $1,582,500; x11 = 100, x14 = 160, x15 = 140, x22 = 50, x23 = x32 = 150; all remaining decision
variables are equal to 0.
PROBLEM 8.4
The J. Mehta Company’s production manager is planning for a series of 1-month production periods for stainless steel sinks. The
demand for the next four months is presented in the accompanying figure. The Mehta firm can normally produce 100 stainless
steel sinks in a month. This is done during regular production hours at $100 per sink. If demand in any 1 month cannot be
satisfied by regular production, the production manager has three other choices: (1) He can produce up to 50 more sinks per
month in overtime but at the cost of $130 per sink; (2) he can purchase a limited number of sinks from a friendly competitor for
resale (the maximum number of outside purchases over the 4-month period is 450 sinks, at $150 each); or (3) he can fill the
demand from his on-hand inventory. The inventory carrying cost is $10 per sink per month. Backorders are not permitted.
Inventory on hand at the beginning of month 1 is 40 sinks. Set up this “production smoothing” problem as a transportation
problem to minimize cost. Use the northwest corner rule to find an initial level for production and outside purchases over the 4
months.
Month Demand for stainless steel sinks

1 120
2 160
3 240
4 100

PROBLEM 8.5
The National Western Railroad’s rail network covers most of the U.S. West and Midwest. On a daily basis, it sends empty freight
cars from various locations in its rail network to its customers for their use. Sometimes there are not enough freight cars to meet
customer demand. The transportation costs for shipping empty freight cars, shown in the table provided, are directly related to
distance traveled and the number of rail centers that must handle the car movement. Determine the number of empty freight
cars that should be sent from each rail network location to customers to meet demand at the minimum total cost.
Freight car Customer location

Supply
location Milwaukee Omaha Topeka Tucson Denver Wichita Minneapolis Memphis Kansas
Portland 27 23 23 26 21 29 40 45 23 1,100
Fresno 31 26 25 22 20 34 47 43 26 720
Long Beach 38 31 32 18 24 27 51 48 34 1,450
Salt Lake 28 18 17 24 9 20 32 35 19 980
El Paso 41 27 24 11 18 22 46 30 25 650
Houston 38 24 22 16 27 25 41 28 23 1,025
St. Louis 15 14 10 27 23 12 19 10 9 1,330
Chicago 12 13 15 31 26 17 14 15 14 1,275
Demand 974 1,225 1,690 710 1,261 663 301 479 1,227
o Let xij be the number of empty freight cars that should be sent from each rail network location to customers; i = 1
(Portland), 2 (Fresno), 3 (Long Beach), …, 8 (Chicago); j = 1 (Milwaukee), 2 (Omaha), 3 (Topeka), …, 9 (Kansas).
o An optimal solution is: Z = 149,777; x12 = 723, x19 = 376, x22 = 55, x23 = 665, x32 = 446, x34 = 60, x35 = 281, x36 = 663,
x45 = 980, x54 = 650, x63 = 1,025, x78 = 479, x79 = 851, x81 = 974, x87 = 301; all remaining decision variables are equal
to 0.
PROBLEM 8.6
Binford Tools manufactures garden tools. It uses inventory, overtime, and subcontracting to absorb demand fluctuations. Expected
demand, regular and overtime production capacity, and subcontracting capacity are provided in the following table for the next
four quarters for its basic line of steel garden tools. The regular production cost per unit is $20, the overtime cost per unit is $25,
the cost to subcontract a unit is $27, and the inventory carrying cost is $2 per unit. The company has 300 units in inventory at
the beginning of the year. Determine the optimal production schedule for the four quarters to minimize total costs.
Quarter Demand Regular capacity Overtime capacity Subcontracting capacity

1 9,000 9,000 1,000 3,000
2 12,000 10,000 1,500 3,000
3 16,000 12,000 2,000 3,000
4 19,000 12,000 2,000 3,000
o Let xi-j be the number of units sourced through Mode i to be used/consumed in Quarter j; i = 1 (beginning inventory),
2 (Qtr. 1 regular production), 3 (Qtr. 1 overtime production), 4 (Qtr. 1 subcontracting), …, 11 (Qtr. 4 regular
production), 12 (Qtr. 4 overtime production), 13 (Qtr. 4 subcontracting); j = 1 (Qtr. 1), 2 (Qtr. 2), 3 (Qtr. 3), 4 (Qtr.
4).
o Assumption: Backlogs or backorders are not allowed.
o An optimal solution is: Z = $1,198,500; x1-2 = 300, x2-1 = 9,000, x3-2 = x5-3 = x10-3 = 1,000, x5-2 = 9,000, x6-2 = 1,500,
x7-2 = 200, x8-3 = x11-4 = 12,000, x93 = x12-4 = x10-4 = 2,000, x13-4 = 3,000; all remaining decision variables are equal to
0.
PROBLEM 8.7
The J. Mehta Company’s production manager is planning for a series of 1-month production periods for stainless steel sinks. The
demand for the next four months is presented in the accompanying figure. The Mehta firm can normally produce 100 stainless
steel sinks in a month. This is done during regular production hours at $100 per sink. If demand in any 1 month cannot be
satisfied by regular production, the production manager has three other choices: (1) He can produce up to 50 more sinks per
month in overtime but at a cost of $130 per sink; (2) he can purchase a limited number of sinks from a friendly competitor for
resale (the maximum number of outside purchases over the 4-month period is 450 sinks, at a cost of $150 each); or (3) he can fill
the demand from his on-hand inventory. The inventory carrying cost is $10 per sink per month. Back orders are not permitted.
Inventory on hand at the beginning of month 1 is 40 sinks. Set up this “production smoothing” problem as a transportation
problem to minimize cost. Use the northwest corner rule to find an initial level for production and outside purchases over the 4-
month period.
Month Demand for stainless steel sinks

1 120
2 160
3 240
4 100

PROBLEM 8.8
Three electric power plants with capacities of 25, 40, and 30 million kWh supply electricity to three cities. The maximum demands
at the three cities are estimated at 30, 35, and 25 million kWh. The price per million kWh in the three cities is given in the
following table. During August, there is a 20% increase in demand at each of the three cities, which can be met by purchasing
electricity from another network at a premium rate of $1,000 per million kWh. The network is not linked to city 3, however. The
utility company wishes to determine the most economical plan for the distribution and purchase of additional energy. Determine
an optimal distribution plan for the utility company.
Plant\City 1 2 3
1 $600 $700 $400
2 $320 $300 $350
3 $500 $480 $450
o Let xij be the amount of power (million kWh) supplied to City j from Source i; i = 1 (Plant 1), 2 (Plant 2), 3 (Plant
3), 4 (Network); j = 1, 2, 3.
o An optimal solution is: Z = $49,710; x13 = 25, x22 = 40, x31 = 23, x32 = 2, x33 = 5, x41 = 13; all remaining decision
variables are equal to 0.
PROBLEM 8.9
The MJK Manufacturing Company must produce two products in sufficient quantity to meet contracted sales in each of the next
three months. The two products share the same production facilities, and each unit of both products requires the same amount
of production capacity. The available production and storage facilities are changing month by month, so the production capacities,
unit production costs, and unit storage costs vary by month. Therefore, it may be worthwhile to overproduce one or both products
in some months and store them until needed.
For each of the three months, the second column of the following table gives the maximum number of units of the two products
combined that can be produced on Regular Time (RT) and Overtime (O). For each of the two products, the subsequent columns
give (1) the number of units needed for the contracted sales, (2) the cost (in thousands of dollars) per unit produced on Regular
Time, (3) the cost (in thousands of dollars) per unit produced on Overtime, and (4) the cost (in thousands of dollars) of storing
each extra unit that is held over into the next month. In each case, the numbers for the two products are separated by a slash /,
with the number for Product 1 on the left and the number for Product 2 on the right.
Product 1/Product 2
Maximum combined production
Month Unit cost of production ($1,000’s) Unit cost of
Sales
RT OT RT OT storage
1 10 3 5/3 15/16 18/20 1/2
2 8 2 3/5 17/15 20/18 2/1
3 10 3 4/4 19/17 22/22
The production manager wants a schedule developed for the number of units of each of the two products to be produced on
Regular Time and (if Regular Time production capacity is used up) on Overtime in each of the three months. The objective is to
minimize the total of the production and storage costs while meeting the contracted sales for each month. There is no initial
inventory, and no final inventory is desired after the three months. Obtain an optimal solution to this problem.
PROBLEM 8.10
MG Auto produce four car models: M1, M2, M3, and M4. The Detroit plant produces models M1, M2, and M4. Models M1 and
M2 are also produced in New Orleans. The Los Angeles plant manufactures models M3 and M4. The capacities of the various
plants and the demands at the distribution centers and the transportation cost per car are provided in the accompanying tables.
Additionally, it is possible to satisfy a percentage of the demand for some models from the supply of others according to the
specifications in the following table. Determine the optimum shipping schedule. Hint: Add four new destinations corresponding
to the new combinations [M1, M2], [M3, M4], [M1, M2] and [M2, M4]. The demand at the new destinations is determined from
the given percentages.
Capacities and Demand

Model
Total
M1 M2 M3 M4
Plant
Los Angeles – – 700 300 1,000
Detroit 500 600 – 400 1,500
New Orleans 800 400 – – 1,200
Distribution center
Denver 700 500 500 600 2,300
Miami 600 500 200 100 1,400

Interchangeable Models
Distribution center Percentage of demand Interchangeable models
Denver 10 M1, M2
20 M3, M4
Miami 10 M1, M2
5 M2, M4
8.2. Transshipment Problems
In the transportation problem, we either have pure supply or pure demand points. Meaning, those supply points (for example,
that in Kansas) are only capable of supplying while the demand points are only capable of demanding. In some cases, some points
are capable of both supplying and demanding. These points are called transient or transshipment points (nodes). Such nodes
appear to be intermediary nodes or nodes between these pure and demand nodes. Transshipment nodes are usually in the form
of warehouses, distributors or dealers and the likes; so that for example, from a source or producer, it may be more convenient or
cheaper for the products to pass through first a distributor before it goes to end-users. Let us go directly now to our concrete
examples.
8.2.1. Modeling
EXAMPLE 8.7 Modeling – Transshipment Problems
DeMont Chemical Company. DeMont Chemical Company manufactures fertilizer in three plants, referred to as P1, P2, and P3.
The company ships its products from plants to two central distribution centers (DCs), D1 and D2, and then from the DCs to five
regional warehouses, W1–W5. At the DCs, no demand occurs, and no capacity limits exist. Demand is associated with the
warehouses, and capacities exist at the plants. The system is described in the following two tables, one for each stage. The units
for capacity and demand are pounds of fertilizer, and the unit costs are given per pound.
(To) DC
(From) Plant Capacity
D1 (4) D2 (5)
P1 (1) $1.36 $1.28 2,400
P2 (2) $1.28 $1.35 2,750
P3 (3) $1.68 $1.55 2,500
(To) Warehouse
(From) DC
W1 (6) W2 (7) W3 (8) W4 (9) W5 (10)
D1 (4) $0.60 $0.36 $0.32 $0.44 $0.72
D2 (5) $0.80 $0.56 $0.42 $0.40 $0.55
Requirement 1,250 1,000 1,600 1,750 1,500
Modeling
To aid us in modeling the transshipment problem, we

visually illustrate the problem.
The transshipment problem involves adequate supply

(7,650 units) to meet the demands or requirements
(7,100 units). We see that there are three pure supply
nodes (P1, P2, and P3), five pure demand nodes (W1,
W2, W3, W4, and W5) and two transshipment nodes
(D1 and D2). There are ten nodes all in all. A
combination of these nodes defines a decision variable.
Suppose we denoted P1, P2, P3, D1, D2, W1, W2, W3,
W4, and W5 with indices 1 to 10, respectively, a decision
variable is defined as
Let xi-j – pounds of fertilizers to be shipped from

Node i to Node j; i, j = 1, 2, 3, …, 10 such that route
i-j exists or is feasible.
Notice that instead of using the terms “source” and “destination” in the definition, we used the term “node” to recognize that a
node can be not only a source node but also a demand node, as in the case of a transshipment node.

Meanwhile, the objective is to minimize the total transshipment cost.
Min Z = 1.36x1-4 + 1.28x1-5 + 1.28x2-4 + 1.35x2-5 + 1.68x3-4 + 1.55x3-5 + 0.60x4-6 + 0.36x4-7 + 0.32x4-8
+ 0.44x4-9 + 0.72x4-10 + 0.80x5-6 + 0.56x5-7 + 0.42x5-8 + 0.40x5-9 + 0.55x5-10
For the constraints, there are three sets: one for the pure supply nodes, another for the pure demand nodes and finally, for the
transshipment nodes. For the pure supply constraints, the idea is that the outflow from a pure supply node should not exceed its
capacity. Hence,
P1 (1): x1-4 + x1-5 ≤ 2, 400

P2 (2): x2-4 + x2-5 ≤ 2,750
P3 (3): x2-4 + x2-5 ≤ 2,750.
It is obvious also that the pure demand constraints are
W1 (6): x4-6 + x5-6 ≥ 1,250

W2 (7): x4-7 + x5-7 ≥ 1,000
W3 (8): x4-8 + x5-8 ≥ 1,600
W4 (9): x4-9 + x5-9 ≥ 1,750
W5 (10): x4-10 + x5-10 ≥ 1,500.
Meanwhile, for the constraints related to the transshipment nodes, we follow basically this rule: inflow=outflow, i.e., whatever
comes in equals whatever comes out. It means then that
D1 (4): x1-4 + x2-4 + x3-4 = x4-6 + x4-7 + x4-8 + x4-9 + x4-10

D2 (5): x1-5 + x2-5 + x3-5 = x5-6 + x5-7 + x5-8 + x5-9 + x5-10.
We restate these constraints in the following manner:
D1 (4): x1-4 + x2-4 + x3-4 – x4-6 – x4-7 – x4-8 – x4-9 – x4-10 = 0

D2 (5): x1-5 + x2-5 + x3-5 – x5-6 – x5-7 – x5-8 – x5-9 – x5-10 = 0.
Finally, we include the NN constraint.
NN: xi-j ≥ 0
In summary,
Let xi-j – pounds of fertilizers to be shipped from Node i to Node j; i, j = 1, 2, 3, …, 10

such that Route i-j exists or is feasible.
Min Z = 1.36x1-4 + 1.28x1-5 + 1.28x2-4 + 1.35x2-5 + 1.68x3-4 + 1.55x3-5 + 0.60x4-6 + 0.36x4-7 + 0.32x4-8 + 0.44x4-9
+ 0.72x4-10 + 0.80x5-6 + 0.56x5-7 + 0.42x5-8 + 0.40x5-9 + 0.55x5-10
Subject to:
P1 (1): x1-4 + x1-5 ≤ 2, 400 D1 (4): x1-4 + x2-4 + x3-4 – x4-6 – x4-7 – x4-8 – x4-9 – x4-10 = 0
P2 (2): x2-4 + x2-5 ≤ 2,750 D2 (5): x1-5 + x2-5 + x3-5 – x5-6 – x5-7 – x5-8 – x5-9 – x5-10 = 0
P3 (3): x2-4 + x2-5 ≤ 2,750
W1 (6): x4-6 + x5-6 ≥ 1,250 W4 (9): x4-9 + x5-9 ≥ 1,750
W2 (7): x4-7 + x5-7 ≥ 1,000 W5 (10): x4-10 + x5-10 ≥ 1,500
W3 (8): x4-8 + x5-8 ≥ 1,600 NN: xi-j ≥ 0.
o An optimal solution is: Z = $12,849.50; x1-4 = 1,100, x1-5 = 1,300, x2-4 = 2,750, x3-5 = 1,950, x4-6 = 1,250, x4-7 = 1,000,
x4-8 = 1,600, x5-9 = 1,750, x5-10 = 1,500; x2-5 = x3-4 = x4-9 = x4-10 = x5-6 = x5-7 = x5-8 = 0.
End.
Note that when we have a sufficient supply, the model format showcased in EXAMPLE 8.7 applies. How about when the problem
does not have sufficient supply to meet all the requirements? Consider EXAMPLE 8.8.

Consider the transshipment problem illustrated by the adjacent figure.

Suppose that the management deems it is necessary that Node 7’s
demand of 300 should be completely fulfilled. Since the supply cannot
completely fulfill the demand, every unit of unmet demand for Node
6 and 8 is given a penalty of 4 and 6, respectively. Provide the
corresponding LP model of the transshipment problem.
Modeling
To reflect the penalties for unmet demand, we can create a dummy

pure supply node dedicated to supplying only Node 6 and Node 7 as
shown in the following figure:
Let xij – number of units to be shipped from Node i to Node j;
i, j = 1, 2, 3, 4, 5, 6, 7, 8, 9 such that Route i-j is feasible
Min Z = 6x13 + 5x14 + 4x23 + 6x24 + 7x25 + 4x36 + 8x37 + 7x47 +
9x48 + 2x45 + 3x57 + 6x58 + 4x96 + 6x98
Subject to:
Node 1: x13 + x14 ≤ 400
Node 2: x23 + x24 + x25 ≤ 300
Node 3: x13 + x23 – x36 – x37 = 0
Node 4: x14 + x24 – x47 – x48 – x45 = 0
Node 5: x25 + x45 – x57 – x58 = 0
Node 6: x36 + x96 ≥ 250
Node 7: x37 + x47 + x57 ≥ 300
Node 8: x48 + x58 + x98 ≥ 200
Node 9: x96 + x98 ≤ 50
NN: xij ≥ 0
Notice in the figure that there is no route between the dummy supply Node 9 and Node 7 because any unit shipped between these
two points are hypothetical considering that the latter node requires all its demand be met. With the inclusion of the dummy
supply, the transshipment problem becomes balanced. The corresponding LP is shown above.
o An optimal solution is: Z = 7,250; x14 = 400, x23 = 250, x25 = 50, x36 = 250, x45 = 400, x57 = 300, x58 = 150, x98 = 50;
x13 = x24 = x37 = x47 = x48 = x96 = 0.
What if the problem does not involve any penalty system and we simply model it as
Let xij – number of units to be shipped from Node i to Node j; i, j = 1, 2, 3, 4, 5, 6, 7, 8;

such that Route i-j is feasible
Min Z = 6x13 + 5x14 + 4x23 + 6x24 + 7x25 + 4x36 + 8x37 + 7x47 + 9x48 + 2x45 + 3x57 + 6x58
Subject to:
Node 1: x13 + x14 = 400 Node 5: x25 + x45 – x57 – x58 = 0
Node 2: x23 + x24 + x25 = 300 Node 6: x36 ≤ 250
Node 3: x13 + x23 – x36 – x37 = 0 Node 7: x37 + x47 + x57 ≤ 300
Node 4: x14 + x24 – x45 – x47 – x48 = 0 Node 8: x48 + x58 ≤ 200
NN: xij ≥ 0.
Here, we followed the idea that when supply is less than demand, the supply of each pure supply node will be exhausted. Hence,
we have Node 1 and Node 2 constraints as equalities. Further, we also expect that the demand for each pure demand node may
be completely or partly satisfied. Hence, we have Node 6, Node 7 and Node 8 constraints as less-than-or-equal-to constraints.
After running the model, we obtained the following solution: x14 = 400, x24 = 300, x45 = 700 and all the remaining variables are
equal to 0. The solution does not allow any shipment from the transshipment nodes to the pure demand nodes. Though this
solution is valid, this is senseless when the premise is to at least partly satisfy the demands of the pure demand nodes at the most
efficient way. We cannot automatically use the technique in the transportation problem when the transshipment problem is
unbalanced. The presence of transshipment nodes brings this complexity. The technique is to balance first the problem by adding
the necessary dummy supply points and implementing a penalty system for unmet demand if deemed necessary.
End.

Two automobile plants, P1 and P2, are linked to three dealers,

D1, D2, and D3, by way of two transport centers, T1 and T2,
according to the network shown below. The supply amounts at
plants P1 and P2 are 1,000 and 1,200 cars, and the demand
amounts at dealers D1, D2, and D3, are 800, 900, and 500 cars.
The shipping costs per car (in hundreds of dollars) between
pairs of nodes are shown on the connecting links (or arc) of the
network. Formulate the LP model.
Modeling
By inspecting the network, we see that the transshipment nodes

are not only T1 and T2 but also D1 and D2. Further, D1 and
D2 require 800 and 900 units, respectively. These requirements
can be viewed as the net flows in these transshipment nodes.
If again we let xij be the number of units to be transported from Node i to Node j where i, j = 1, 2, 3, 4, 5, 6 such that Route i-j
is feasible, the associated constraints with D1 and D2 are
D1 (5): x35 – x56 ≥ 800

D2 (6): x36 + x46 + x56 – x67 ≥ 900.
Notice that we can have these two constraints as equalities because of their transshipment nature. However, but to be consistent
with the idea that demand is fulfilled exactly or exceeded, we instead use “≥” in the two constraints.
The complete LP model is
Let xij – number of units to be transported from Node i to Node j where i, j = 1, 2, 3, 4, 5, 6, 7

such that Route i-j is feasible
Min Z = 3x13 + 4x14 + 2x23 + 5x24 + 7x34 + 8x35 + 6x36 + 4x46 + 9x47 + 5x56 + 3x67
Subject to:
Node 1: x13 + x14 ≤ 1,000 Node 5: x35 – x56 ≥ 800
Node 2: x23 + x24 ≤ 1,200 Node 6: x36 + x46 + x56 – x67 ≥ 900
Node 3: x13 + x23 – x34 – x35 – x36 = 0 Node 7: x47 + x67 ≥ 500
Node 4: x14 + x24 + x34 – x46 – x47 = 0 NN: xij ≥ 0.
o An optimal solution is: Z = 20,700; x14 = 1,000, x23 = 1,200, x35 = 800, x36 = 400, x46 = 1,000, x67 = 500; x13 = x24 =
x34 = x47 = x56 = 0.
End.
8.2.2. Transportation Simplex Algorithm
Manually, we can solve a transshipment model by working with its transportation equivalent. From the transshipment model, we
construct the corresponding transportation tableau and perform the Transportation Simplex Algorithm. In this section, for each
example, we end with the constructed initial transportation tableau. It is up for the students to proceed using the algorithm to
derive the values of the decision variables.
EXAMPLE 8.10 Transportation Simplex Algorithm – Transshipment Problems
Recall EXAMPLE 8.8.
Let xij – number of units to be shipped from Node i to Node j; i, j = 1, 2, 3, 4, 5, 6, 7, 8, 9 such that Route ij is feasible
Min Z = 6x13 + 5x14 + 4x23 + 6x24 + 7x25 + 4x36 + 8x37 + 7x47 + 9x48 + 2x45 + 3x57 + 6x58 + 4x96 + 6x98
Subject to:
Node 1: x13 + x14 ≤ 400 Node 6: x36 + x96 ≥ 250
Node 2: x23 + x24 + x25 ≤ 300 Node 7: x37 + x47 + x57 ≥ 300
Node 3: x13 + x23 – x36 – x37 = 0 Node 8: x48 + x58 + x98 ≥ 200
Node 4: x14 + x24 – x47 – x48 – x45 = 0 Node 9: x96 + x98 ≤ 50
Node 5: x25 + x45 – x57 – x58 = 0 NN: xij ≥ 0

Solving
As in the transportation tableau, we need to identify the sources and

destinations. The sources are nodes 1, 2, 3, 4, 5, and 9, while the
destinations are nodes 3, 4, 5, 6, and 8. Recall that the transshipment
nodes 3, 4, and 5 are both sources and destinations.
The starting transportation tableau of the transshipment problem is

provided in the succeeding tableau. The costs associated with
infeasible routes are M so that the algorithm will avoid assigning any
shipment. Note that the cost associated with variable xij where i = j
(such as the case of Source 1 – Destination 1, Source 3 – Destination
3, and so on) is 0.
From\To Node 3 Node 4 Node 5 Node 6 Node 7 Node 8 Supply

6 5 M M M M
Node 1
4 6 7 M M M
Node 2
0 M M 4 8 M
Node 3
M 0 2 M 7 9
Node 4
M M 0 M 3 6
Node 5
M M M 4 M 6
Node 9
Demand
The supply of any transshipment node is equal to the original/initial supply at that node plus a buffer. Meanwhile, for the demand
of any transshipment node, it is equal to the original/initial demand at the node plus a buffer. The buffer is equal to the total
supply or total demand provided that the transshipment problem is balanced. Hence,

6 5 M M M M
Node 1 400
4 6 7 M M M
Node 2 300
0 M M 4 8 M 0 + 750 =
Node 3
750
M 0 2 M 7 9 0 + 750 =
Node 4
750
M M 0 M 3 6 0 + 750 =
Node 5
750
M M M 4 M 6
Node 9 50
0 + 750 = 0 + 750 = 0 + 750 =
Demand 250 300 300 3,000
750 750 750
This is now the equivalent transportation tableau. To solve now for the decision variables, we work with an initial set of decision
variables using Northwest Corner Method or Least Cost Method and proceed with the transportation algorithm. Note that M
must be substituted with a very large number relative to the given cost parameters. Iteration 0 shows the Northwest Corner Rule
solution (M = 100) and the computed ROIs (those in parenthesis) associated with non-basic cells.

Iteration 0
6 5 M M M M
Node 1 400 (–101) (–8) (–108) (–108) (–14) 400
4 6 7 M M M
Node 2 300 (–98) (–99) (–106) (–106) (–12) 300
0 M M 4 8 M
Node 3 50 700 (–2) (–198) (–198) (–8) 750
M 0 2 M 7 9
Node 4 (200) 50 700 (–2) (–2) (1) 750
M M 0 M 3 6
Node 5 (202) (102) 50 250 250 150 750
M M M 4 M 6
Node 9 (202) (102) (100) (–96) (–96) 50 50
Demand 750 750 750 250 300 200 3,000
The final iteration is

6 5 M M M M
Node 1 (2) 400 (93) (92) (90) (87) 400
4 6 7 M M M
Node 2 250 (1) 50 (92) (90) (87) 300
0 M M 4 8 M
Node 3 500 (99) (97) 250 (2) (91) 750
M 0 2 M 7 9
Node 4 (101) 350 400 (97) (2) (1) 750
M M 0 M 3 6
Node 5 (103) (102) 300 (99) 300 150 750
M M M 4 M 6
Node 9 (103) (102) (100) (3) (97) 50 50
Demand 750 750 750 250 300 200 3,000
The total transshipment cost is minimized to $7,250. The solution generated is an alternative solution to what we had earlier: Z
= 7,250; x14 = 400, x23 = 250, x25 = 50, x36 = 250, x45 = 400, x57 = 300, x58 = 150, x98 = 50; x13 = x24 = x37 = x47 = x48 = x96 =
0. How do we interpret results such as shipping 500 units from node 3 to node 3?
End.
EXAMPLE 8.11 Transportation Simplex Algorithm – Transshipment Problems
Recall EXAMPLE 8.9.

Let xij – number of units to be transported from Node i
to Node j where i, j = 1, 2, 3, 4, 5, 6, 7 such that
Route i-j is feasible
Min Z = 3x13 + 4x14 + 2x23 + 5x24 + 7x34 + 8x35 + 6x36
+ 4x46 + 9x47 + 5x56 + 3x67
Subject to:
Node 1: x13 + x14 ≤ 1,000
Node 2: x23 + x24 ≤ 1,200
Node 3: x13 + x23 – x34 – x35 – x36 = 0
Node 4: x14 + x24 + x34 – x46 – x47 = 0
Node 5: x35 – x56 ≥ 800
Node 6: x36 + x46 + x56 – x67 ≥ 900
Node 7: x47 + x67 ≥ 500
NN: xij ≥ 0

The equivalent transportation tableau is
From\To Node 3 Node 4 Node 5 Node 6 Node 7 Supply

3 4 M M M
Node 1 1,000
2 5 M M M
Node 2 1,200
0 7 8 6 M 0 + 2,200
Node 3
= 2,200
M 0 M 4 9 0 + 2,200
Node 4
= 2,200
M M 0 5 M 0 + 2,200
Node 5
= 2,200
M M M 0 3 0 + 2,200
Node 6
= 2,200
0 + 2,200 0 + 2,200 800 + 2,200 900 + 2,200
Demand 500 11,000
= 2,200 = 2,200 = 3,000 = 3,100
Meanwhile, using the Least Cost Method with M = 100, the starting basic feasible solution (Iteration 0) and the rates of
improvement associated with the non-basic variables are provided in the tableau below:
Iteration 0
3 4 100 100 100
Node 1 (1) (–92) (800) 200 (0) 1,000
2 5 100 100 100
Node 2 (0) (–91) (0) 700 500 1,200
0 7 8 6 100
Node 3 2,200 (–87) (–90) (–92) (2) 2,200
100 0 100 4 9
Node 4 (194) 2,200 (96) (0) (5) 2,200
100 100 0 5 100
Node 5 (198) (104) 2,200 (5) (100) 2,200
100 100 100 0 3

Node 6 (198) (104) (100) 2,200 (3) 2,200
Demand 2,200 2,200 3,000 3,100 500 11,000
The optimal tableau leads to Z = $20,700.

3 4 100 100 100
Node 1 (1) 1,000 (90) (92) (89) 1,000
2 5 100 100 100
Node 2 1,200 (1) (90) (92) (89) 1,200
0 7 8 6 100
Node 3 1,000 (5) 800 400 (91) 2,200
100 0 100 4 9
Node 4 (102) 1,200 (94) 1,000 (2) 2,200
100 100 0 5 100
Node 5 (108) (106) 2,200 (7) (99) 2,200
100 100 100 0 3
Node 6 (106) (104) (98) 1,700 500 2,200
Demand 2,200 2,200 3,000 3,100 500 11,000
End.
8.2.3. Problem Set
PROBLEM 8.11
Sadeghian Company is in the business of buying and selling grain. An important aspect of the company’s business is arranging
for the purchased grain to be shipped to customers. If the company can keep freight costs low, its profitability will be improved.
Currently, the company has purchased three rail cars of grain at Peoria, seven rail cars at Iowa City, and six rail cars at Lawrence.
Fourteen carloads of grain have been sold. The locations and the amount sold at each location are given in Table A.

Table A Table B
Location Rail carloads Louisville Dayton
Augusta 2 Peoria 1,800 1,500
Gainesville 4 Iowa City 1,400 1,900
Oxford 3 Lawrence 1,200 1,600
Columbia 5
Table C
Augusta Gainesville Oxford Columbia
Louisville 4,400 3,600 3,300 3,200
Dayton 4,200 3,500 3,100 2,700
All shipments must be routed through either Louisville or Dayton. Shown in Table B are the shipping costs per rail car from the
origins to Louisville and Dayton. Meanwhile, Table C shows the costs per rail car to ship from Louisville and Dayton to the
destinations. Determine a shipping schedule that will minimize the freight costs necessary to satisfy demand.
o Let xij be the number of rail carloads to be shipped from Node i to Node j; i, j = 1 (Peoria), 2 (Iowa City), 3 (Lawrence),
4 (Louisville), 5 (Dayton), 6 (Augusta), 7 (Gainesville), 8 (Oxford), 9 (Columbia) such that Route i-j exists.
o An optimal solution is: Z = $6,520; x15 = 8, x34 = 6; x46 = 2; x47 = 4; x58 = 3, x59 = 5; all remaining decision variables
are equal to 0.
PROBLEM 8.12
World Foods, Inc., imports food products such as meats, cheese, and pastries to the United States from warehouses at ports in
Hamburg, Marseilles, and Liverpool. Ships from these ports deliver the products to Norfolk, New York, and Savannah, where
they are stored in company warehouses before being shipped to distribution centers in Dallas, St. Louis, and Chicago. The products
are then distributed to specialty food stores and sold through catalogs. The shipping costs ($/1,000 lb.) from the European ports
to the U.S. cities and the available supplies (1,000 lb.) at the European ports are provided in Table A. Meanwhile, the
transportation costs ($/1,000 lb.) from each U.S. city of the three distribution centers and the demands (1,000 lb.) at the
distribution centers are shown in Table B. Determine the optimal shipments between the European ports and the warehouses and
the distribution centers to minimize total transportation costs.
Table A Table B
U.S. City Distribution Center
European Warehouse
New Supply Dallas St. Louis Chicago
Port Norfolk Savannah
York Norfolk 75 63 81
Hamburg 420 390 610 55 New York 125 110 95
Marseilles 510 590 470 78 Savannah 68 82 95
Liverpool 450 360 480 37 Demand 60 45 50
o Let xij be the amount of food products (1,000 lbs.) to be shipped Node i to Node j; i, j = 1 (Hamburg), 2 (Marseilles),
3 (Liverpool), 4 (Norfolk), 5 (New York), 6 (Savannah), 7 (Dallas), 8 (St. Louis), 9 (Chicago) such that Route i-j exists.
o An optimal solution is: Z = $77,362; x14 = 42, x15 = 13, x26 = 63, x35 = 37, x48 = 42, x59 = 50, x67 = 60; all remaining
decision variables are equal to 0.
PROBLEM 8.13
Walsh’s Fruit Company contracts with growers in Ohio, Pennsylvania, and New York to purchase grapes. The grapes are processed
into juice at the farms and stored in refrigerated vats. Then the juice is shipped to two plants, where it is processed into bottled
grape juice and frozen concentrate. The juice and concentrate are then transported to three food warehouses/distribution centers.
The transportation costs ($) per ton from the farms to the plants and from the plants to the distributors and the supply (1,000
tons) at the farms and demand (1,000 tons) at the distribution centers are summarized in the accompanying tables. Determine
the optimal shipments from farms to plants to distribution centers to minimize total transportation costs. What would be the
effect on the solution if the capacity at each plant were 140,000 tons?
Plant Supply Distributor

Farm Plant
Indiana Georgia (1,000 tons) Virginia Kentucky Louisiana
Ohio 16 21 72 Indiana 23 15 29
Pennsylvania 18 16 105 Georgia 20 17 24
New York 22 25 83 Demand 90 80 120
PROBLEM 8.14
KanTech Corporation is a global distributor of electrical parts and components. Its customers are electronics companies in the
United States, including computer manufacturers and audio/visual product manufacturers. The company contracts to purchase
components and parts from manufacturers in Russia, Eastern and Western Europe, and the Mediterranean and it has them

delivered to warehouses in three European ports, Gdansk, Hamburg, and Lisbon. The various components and parts are loaded
into containers based on demand from U.S. customers. Each port has a limited fixed number of containers available each month.
The containers are then shipped overseas by container ships to the ports of Norfolk, Jacksonville, New Orleans, and Galveston.
From these seaports, the containers are typically coupled with trucks and hauled to inland ports in Front Royal (Virginia), Kansas
City, and Dallas. There are a fixed number of freight haulers available at each port each month. These inland ports are sometimes
called “freight villages,” or intermodal junctions, where the containers are collected and transferred from one transport mode to
another (i.e., from the truck to rail or vice versa). From the inland ports, the containers are transported to KanTech’s distribution
centers in Tucson, Pittsburgh, Denver, Nashville, and Cleveland. Following are the handling and shipping costs ($/container)
between each of the embarkation and destination points along this overseas supply chain and the available containers at each
port.
U.S. Port Available

European Port
Norfolk Jacksonville New Orleans Galveston Containers
Gdansk 1,725 1,800 2,345 2,700 125
Hamburg 1,825 1,750 1,945 2,320 210
Lisbon 2,060 2,175 2,050 2,475 160
Inland Port Intermodal Capacity

U.S. Port
Dallas Kansas City Front Royal (containers)
Norfolk 825 545 320 125
Jacksonville 750 675 450 210
New Orleans 325 605 690 160
Galveston 270 510 1,050
Intermodal Capacity 170 240 140
(containers)
Distribution Center
Inland Port
Tucson Denver Pittsburgh Nashville Cleveland
Dallas 450 830 565 420 960
Kansas City 880 520 450 380 660
Front Royal 1,350 390 1,200 450 310
Demand 85 60 105 50 120
Determine the optimal shipments from each point of embarkation to each destination along this supply chain that will result in
the minimum total shipping cost.
PROBLEM 8.15
The Pinnacle Company is a U.S.-based manufacturer of furniture and appliances that offshored all of its actual manufacturing
operations to Asia about a decade ago. It then set up distribution centers at various locations on the East Coast, near ports where
its items were imported on container ships. In many cases, Pinnacle’s appliances and furniture arrive partially assembled, and the
company completes the assembly at its distribution centers before sending the finished products to retailers. For example,
appliance motors, electric controls, housings, and furniture pieces might arrive from different Asian manufacturers in separate
containers. Recently Pinnacle began exporting its products to various locations in Europe and demand steadily increased. As a
result, the company determined that shipping items to the United States, assembling the products, and then turning around and
shipping them to Europe was inefficient and not cost effective. The company now plans to open three new distribution centers
near ports in Europe, and it will ship its items from Asian ports to distribution centers at the European ports, offload some of
the items for final product assembly, and then ship the partially filled containers on to the U.S. distribution centers.
The following table shows the seven possible locations near container ports in Europe, and their container capacity (units) that
Pinnacle has identified to construct its proposed three distribution centers; the container shipments (units) from each of its Asian
ports; and the container shipping cost from each of the Asian ports to each possible distribution center location:
Proposed Distribution Center

Asian Port Cont.
Rotterdam Hamburg Antwerp Bremen Valencia Lisbon Le Havre
Ship.
Center Cost 16,725,000 19,351,000 13,766,000 15,463,000 12,542,000 13,811,000 22,365,000
Hongkong 3,466 3,560 3,125 3,345 3,060 3,120 3,658 235
Shanghai 3,190 3,020 3,278 3,269 2,987 2,864 3,725 170
Busan 2,815 2,700 2,890 3,005 2,465 2,321 3,145 165
Mumbai 2,412 2,560 2,515 2,875 2,325 2,133 2,758 325
Kaoshiung 2,600 2,800 2,735 2,755 2,473 2,410 2,925 405
Capacity 565 485 520 490 310 410 605

The following table shows the demand (units) from each of the U.S. ports and the cost ($) for container shipments from each of
the possible distribution center locations to each of the U.S. ports:
Proposed U.S. Port

Distribution Centers New York Savannah Miami New Orleans
Rotterdam 2,045 1,875 1,675 2,320
Hamburg 2,875 2,130 1,856 2,415
Antwerp 2,415 2,056 1,956 2,228
Bremen 2,225 1,875 2,075 2,652
Valencia 1,865 1,725 1,548 1,815
Lisbon 1,750 1,555 1,420 1,475
Le Havre 3,056 2,280 2,065 2,425
Demand 440 305 190 365
Determine which three distribution center locations in Europe Pinnacle should be selected, and the shipments from each of the
Asian ports to these selected distribution centers and from the European distribution centers to the U.S. ports.
PROBLEM 8.16
The adjacent figure shows the routes for shipping cars from three
plants (Node 1, 2 and 3) to three dealers (Node 6 to 8) by way
of two distribution centers (Node 4 and 5). The shipping costs
per car (in $100) are shown on the arcs. Solve the problem as a
transshipment model. Further, find the new optimum solution if
distribution Center 4 can sell 240 cars directly to customers.
o Let xij be the number cars to be shipped from Node i

to Node j; i, j = 1, 2, 3, …, 8 such that Route i-j exists.
o An optimal solution is: Z = $8,640; x15 = 900, x24 =
1,400, x34 = 1,000, x45 = 1,300, x46 = 1,100, x57 =
1,000, x58 = 1,200; all remaining decision variables are
equal to 0.
PROBLEM 8.17
Consider the oil pipeline network shown in the adjacent figure. The
different nodes represent pumping and receiving stations. Distances in
miles between the stations are shown on the adjacent network. The
transportation cost per gallon between two nodes is directly proportional
to the length of the pipeline. Develop the associated transshipment
model and find the optimum solution.
o Let xij be volume (gallon) to flow from Node i to Node j; i, j

= 1, 2, 3, …, 7 such that Link i-j exists.
o An optimal solution is: Z = 3,380,000; x12 = 50,000, x37 =
60,000, x62 = 40,000, x75 = 60,000, x54 = 20,000, x56 = 40,000;
x17 = x34 = x72 = x57 = x65 =0.
PROBLEM 8.18
Consider the transportation problem in which two factories supply three stores with a commodity. The numbers of supply units
available at sources 1 and 2 are 200 and 300; those demanded at stores 1, 2 and 3 are 100, 200 and 50, respectively. Units may
be transshipped among the factories and the stores before reaching their final destination. Find the optimal shipping schedule
based on the unit costs ($) provided by the table below.
Factory 1 Factory 2 Store 1 Store 2 Store 3

Factory 1 0 6 7 8 9
Factory 2 6 0 5 4 3
Store 1 7 2 0 5 1
Store 2 1 5 1 0 4
Store 3 8 9 7 6 0

PROBLEM 8.19
Texxon Oil Distributors, Inc., has three active oil wells in a west Texas oil field. Well 1 has a capacity of 93 thousand barrels per
day (TBD), Well 2 can produce 88 TBD, and Well 3 can produce 95 TBD. The company has five refineries along the Gulf Coast,
all of which have been operating at stable demand levels. In addition, three pump stations have been built to move the oil along
the pipelines from the wells to the refineries. Oil can flow from any one of the wells to any of the pump stations, and from any
one of the pump stations to any of the refineries, and Texxon is looking for a minimum cost schedule. The refineries’ requirements
are as follows:
Refinery 1 2 3 4 5
Requirement (TBD) 30 57 48 91 48
The company’s cost accounting system recognizes charges by the segment of pipeline that is used. These daily costs are given in
the tables below, in dollars per thousand barrels.
From\To Pump A Pump B Pump C From\To Refinery 1 2 3 4 5

Well 1 1.52 1.60 1.40 Pump A 5.15 5.69 6.13 5.63 5.80
Well 2 1.70 1.63 1.55 Pump B 5.12 5.47 6.05 6.12 5.71
Well 3 1.45 1.57 1.30 Pump C 5.32 6.16 6.25 6.17 5.87
What is the minimum cost of providing oil to the refineries? Which wells are used to capacity in the optimal schedule?
o Let xi-j be the volume of oil (TBD) to be shipped from Node i to Node j; i, j = 1 (Well 1), 2 (Well 2), 3 (Well 3), 4
(Pump A), 5 (Pump B), 6 (Pump C), 7 (R1), 8 (R2), …, 11 (R5) such that Link i-j exists.
o An optimal solution is: Z = $1,963.82; x1-4 = 93, x2-5 = 86, x3-4 = 28, x3-6 = 67, x4-7 = 30, x4-10 = 91, x5-8 = 57, x5-9 =
29, x6-9 = 19, x6-11 = 48; x1-5 = x1-6 = x2-4 = x2-6 = x3-5 = x4-8 = x4-9 = x4-11 = x5-7 = x5-10 = x5-11 = x6-7 = x6-8 = x6-10 =
0.
PROBLEM 8.20
The Krotzer Company manufactures and distributes meters used to measure electric power consumption. The company started
with a small production plant in El Paso and gradually built a customer base throughout Texas. A distribution center was
established in Ft Worth, and later, as the business expanded, a second distribution center was established in Santa Fe. The El
Paso plant was expanded when the company began marketing its meters in Arizona, California, Nevada, and Utah. With the
growth of the West Coast business, the company opened a third distribution center in Las Vegas and just two years ago opened
a second production plant in Sacramento.
Manufacturing costs differ between the company’s production plants. The cost of each meter produced at the El Paso plant is
$10.50. The Sacramento plant uses newer and more efficient equipment, and as a result, its manufacturing costs come to only
$10.00 per unit. The quarterly production capacity is 30,000 meters at the older El Paso plant and 20,000 meters at the Sacramento
plant. No shipments are allowed from the Sacramento plant to the Ft. Worth distribution center.
Due to the firm’s rapid growth, little attention has been paid to the efficiency of the distribution system, but company management
has decided that it is time to address this issue. The cost of shipping a meter from each of the two plants to each of the three
distribution centers is shown in the following Table A.
Table A Table C
Distribution Center Demand
Plant Customer zone
Ft Worth Santa Fe Las Vegas (meters)
El Paso 3.20 2.20 4.20 Dallas 6,300
Sacrameto – 3.90 1.20 San Antonio 4,880
Wichita 2,130
Table B Kansas City 1,210
Customer zone Denver 6,120
DC
Dal SA Wich KC Den SLC Pho LA SD Salt Lake City 4,830
FW 0.30 2.10 3.10 4.40 6.00 – – – – Phoenix 2,750
SF 5.20 5.40 4.50 6.00 2.70 4.70 3.40 3.30 2.70 Los Angeles 8,580
LV – – – – 5.40 3.30 2.40 2.10 2.50 San Diego 4,460
The company serves nine customer zones from the three distribution centers. The forecast for the number of meters needed in
each customer zone for the next quarter is shown in Table B. The cost per unit of shipping from each distribution center to each
customer zone is given in Table C. Note that some distribution centers do not serve certain customer zones because the costs
would be prohibitive.
In the current distribution system, demand at the Dallas, San Antonio, Wichita, and Kansas City customer zones is satisfied by
shipments from the Ft Worth distribution center. In a similar manner, the Denver, Salt Lake City, and Phoenix customer zones
are served by the Santa Fe distribution center, and the Las Vegas distribution center serves the Los Angeles and San Diego

customer zones. The El Paso plant supplies the Ft. Worth and Santa Fe distribution centers, while the Sacramento plant supplies
the Las Vegas distribution center.
You have been called in to make recommendations for improving the distribution system, and, in particular, to address the
following issues.
• If the company does not change its current distribution strategy, what will the distribution system cost be for the
following quarter?
• Suppose that the company is willing to consider dropping the distribution center limitations. In other words, customer
zones would not necessarily be assigned to unique distribution centers, and distribution centers would not necessarily
be assigned to unique plants. With this added flexibility, by how much could cost be reduced?
• In the foreseeable future, the company anticipates moderate growth of about 20 percent in demand. Suppose this growth
is met using the current routes and expanding plant capacity as needed. What plant capacities would be required?
What would be the total system cost?
8.3. Assignment Problems
The assignment problem is a special type of transportation problem where the supply and demand at each “source” and
“destination”, respectively, is one unit. The goal in an assignment problem is to find the optimal assignment or pairing (e.g.,
assignment of people to projects, salespeople to territories, contracts to bidders, jobs to machines, and so on) where total
assignment costs is minimized. These people, projects, territories and machines among others are those that we treat as either
“source” or “destination” in the context of the transportation problem.
The figure below generally illustrates the assignment problem. The assignment model is binary in nature. The decision variables
are binary. A value of 1 for xij means that i is assigned to j and 0 means otherwise. Since the concern in assignment problem is
determining the most efficient pairings, we do not make any distinction between “sources” and “destinations.” Like the
transportation model, the model for a balanced assignment problem (there is a one-is-to-one correspondence between “sources”
and “destinations”) is also shown below.
1, assign i to j
Let xij = {
0, otherwise
i, j = 1, 2, 3, …, n
n n

i = 1j = 1
Subject to:
𝑛
“Supply”: ∑ xij = 1 ∀i
j=1
n
“Demand”: ∑ xij = 1 ∀j
i=1
NN: xij ≥ 0
If the assignment problem is unbalanced, we follow the same approach in the transportation problem. Refer to EXAMPLE 8.12 to
better appreciate this class of problem.
8.3.1. Modeling
EXAMPLE 8.12 Modeling – Assignment Problems
The Triplets. Triplets Anna, Bernice, and Carla are to find partners from an opposite set of triplets Denver, Earl and Francis.
From the personality preferences, the three girls have rated the three guys on favorability as a partner. This information is
summarized in the following table.
Denver (1) Earl (2) Francis (3)

Anna (1) 5 10 11
Bernice (2) 11 5 10
Carla (3) 10 8 12
Formulate the corresponding model to find the assignment of partners for the three girls that will maximize total favorability.

Modeling
We can consider Anna, Bernice, and Carla as the “sources” while Denver, Earl, and Francis as the “destinations” if we contextualize
the problem as a transportation problem. Meanwhile, the figures given are favorability ratings; hence, these are not costs in
general that need to be minimized. However, by default, the assignment problem is a minimization problem. These figures can be
negated and thus converted as cost figures to work with a minimization objective.
Denver (1) Earl (2) Francis (3)

Anna (1) –5 –10 –11
Bernice (2) –11 –5 –10
Carla (3) –10 –8 –12
Working with the binary variable xij where i = 1 (Anna), 2 (Bernice), 3 (Carla) and j = 1 (Denver), 2 (Earl), 3 (Francis), the
objective of our assignment model is
Min Z = –5x11 – 10x12 – 11x13 – 11x21 – 5x22 – 10x23 – 10x31 – 8x32 – 12x33.
The constraints are then expressed as
“Supply” “Demand”
Anna (1): x11 + x12 + x13 = 1 Denver (1): x11 + x21 + x31 = 1
Bernice (2): x21 + x22 + x23 = 1 Earl (2): x12 + x22 + x32 = 1
Carla (3): x31 + x32 + x33 = 1 Francis (3): x13 + x23 + x33 = 1.
Finally,
Bin: xij = (0, 1).
In summary,
1, Assign i to j
Let xij = {
0, Otherwise
i, j = 1, 2, 3
Min Z = –5x11 – 10x12 – 11x13 – 11x21 – 5x22 – 10x23 – 10x31 – 8x32 – 12x33
Subject to:
Anna (1): x11 + x12 + x13 = 1
Bernice (2): x21 + x22 + x23 = 1
Carla (3): x31 + x32 + x33 = 1
Denver (1): x11 + x21 + x31 = 1
Earl (2): x12 + x22 + x32 = 1
Francis (3): x13 + x23 + x33 = 1
Bin: xij = (0, 1).
o An optimal solution is: Z = –33; x12 = x21 = x33 = 1; x11 = x13 = x22 = x23 = x31 = x32 = 0. This means that Anna is
paired with Earl, Bernice is paired with Denver and Carla is paired with Francis.
End.
Fix-It-Shop. Fix-It-Shop has just received three new rush projects to repair: (1) a radio, (2) a toaster oven, and (3) a broken coffee
table. Three repair persons, each with different talents and abilities, are available to do the jobs. The Fix-It-Shop owner estimates
what it will cost in wages to assign each of the workers to each of the three projects. The costs ($), which are shown in the table,
differ because the owner believes that each worker will differ in speed and skill on these varied jobs. The owner’s objective is to
assign the three projects to the workers in a way that will result in the lowest total cost to the shop. Provide the appropriate
model for this problem.
Radio (1) Oven (2) Table (3)

Adam (1) 11 14 6
Ben (2) 8 10 11
Carlo (3) 9 12 7

Modeling
Similar to EXAMPLE 8.12, the LP model for this problem is as follows:
1, assign i to j
Let xij = {
0, otherwise
i, j = 1, 2, 3
Min Z = 11x11 + 14x12 + 6x13 + 8x21 + 10x22 + 11x23 + 9x31 + 12x32 + 7x33
Subject to:
Adam (1): x11 + x12 + x13 = 1
Ben (2): x21 + x22 + x23 = 1
Carlo (3): x31 + x32 + x33 = 1
Radio (1): x11 + x21 + x31 = 1
Oven (2): x12 + x22 + x32 = 1
Table (3): x13 + x23 + x33 = 1
Bin: xij = (0, 1).
o An optimal solution is: Z = $25; x13 = x22 = x31 = 1; x11 = x12 = x21 = x23 = x32 = x33 = 0. The optimal plan is Adam
to fix the table, Ben to fix the oven and Carlo to fix the radio.
End.
The Furniture Factory. In furniture factory floor, eight machines may be used to perform different jobs. Currently, there are
eight pending jobs. Formulate the model to find the optimal assignment of jobs (1-8) to machines (A-H) to minimize the machining
time to finish all the jobs. A machine can only handle a single job. Given below are the processing times, in hrs., of the jobs in
the machines.
1 2 3 4 5 6 7 8
A 1.4 1.6 1.0 2.4 2.1 0.8 1.4 1.7
B 1.3 0.9 1.0 1.7 1.1 1.3 2.3 2.3
C 1.2 1.9 2.2 2.0 2.0 – 1.3 2.0
D 1.8 1.4 – 1.6 2.2 – 1.0 2.4
E 0.9 2.0 2.4 0.8 0.6 1.6 0.7 2.5
F 0.5 1.4 1.2 2.1 1.0 – 1.3 1.1
G 1.6 0.8 1.4 0.9 1.6 0.6 1.3 1.0
H 2.0 – 1.0 2.0 – 1.3 1.2 2.3
Modeling
Those cells with a dash inside mean that such job-machine assignments are infeasible. How do we make sure that when we solve
the problem, such assignments are avoided? This is where M again comes in. We make such assignments very costly or very
inefficient (longer processing time) as represented by M. Hence,
1 2 3 4 5 6 7 8
A 1.4 1.6 1.0 2.4 2.1 0.8 1.4 1.7
B 1.3 0.9 1.0 1.7 1.1 1.3 2.3 2.3
C 1.2 1.9 2.2 2.0 2.0 M 1.3 2.0
D 1.8 1.4 M 1.6 2.2 M 1.0 2.4
E 0.9 2.0 2.4 0.8 0.6 1.6 0.7 2.5
F 0.5 1.4 1.2 2.1 1.0 M 1.3 1.1
G 1.6 0.8 1.4 0.9 1.6 0.6 1.3 1.0
H 2.0 M 1.0 2.0 M 1.3 1.2 2.3
With this development, we can now formulate the model just like any other assignment problems. Therefore,
1, Assign i to j
Let xij = {
0, Otherwise
i = A, B, C, …, H; j = 1, 2, 3, …, 8
Min Z = 1.4xA1 + 1.6xA2 + …. + MxC6 + … + MxH5 + 1.3xH6 + 1.2xH7 + 2.3xH8

Subject to:
A: xA1 + xA2 + xA3 + xA4 + xA5 + xA6 + xA7 + xA8 = 1
B: xB1 + xB2 + xB3 + xB4 + xB5 + xB6 + xB7 + xB8 = 1
C: xC1 + xC2 + xC3 + xC4 + xC5 + xC6 + xC7 + xC8 = 1
… …
H: xH1 + xH2 + xH3 + xH4 + xH5 + xH6 + xH7 + xH8 = 1
1: xA1 + xB1 + xC1 + xD1 + xE1 + xF1 + xG1 + xH1 = 1
… …
Bin: xij = (0, 1)
o An optimal solution is: Z = 7.5; xA6 = xB2 = xC1 = xD7 = xE5 = xF8 = xG4 = xH3 = 1; the remaining decision variables
are equal to 0.
End.
Supposed that we have only seven machines in EXAMPLE 8.14 against the eight pending jobs. How do we model this assignment
problem? Can we simply follow the approach in case of unbalanced problem in the Transportation Problem, as discussed earlier?
The student is left to provide the model then.
8.3.2. Hungarian Algorithm
The assignment problem can be manually solved through the Hungarian Algorithm. This section will demonstrate only the method
and will not explain the logic of the method. The following are the steps of the Hungarian Algorithm,
1. Ensure that the problem is a balanced minimization assignment model.

2. Identify each row’s minimum and subtract it from all other entries of the row.
3. Identify each column’s minimum and subtract it from all other entries of the column.
4. Cover all zeros using the minimum number of horizontal and vertical lines. An optimal assignment is now available if
the number of lines used equals the number of rows (or columns). If no optimal assignment is available, proceed to Step
5. Identify then the optimal assignment by performing the following steps:
A. Find a row or column that has only one cell that contains a 0 value. This 0 is recognized as unique 0, and the
pair associated with this 0 shall be included in the optimal solution. If no unique 0 is available, arbitrarily select
a 0.
B. Draw horizontal and vertical lines passing through this unique 0.
C. Among the uncovered cells with lines, repeat the same process of finding a unique 0 and drawing lines as
instructed. Do this until all required pairings are achieved.
5. Find the smallest number uncovered by the drawn lines in the previous step. Revise the table by (a) subtracting that
number from all uncovered elements of the table and (b) by adding the same number to all elements corresponding to
the intersections of two lines. Repeat Step 4 until an optimal solution is obtained.
The algorithm is demonstrated below:
EXAMPLE 8.15 Hungarian Algorithm
The Triplets. Recall EXAMPLE 8.12.
1. Ensure that the problem is a 2. Identify each row’s minimum and 3. Identify each column’s minimum
balanced minimization subtract it from all other entries and subtract it from all other
assignment model. of the row. entries of the column.
–5 –10 –11 –5 6 –10 1 –11 0 66 10 00

–11 –5 –10 –11 0 –5 6 –10 1 00 65 11
–10 –8 –12 –10 2 –8 4 –12 0 22 43 00

4. Cover all zeros using the minimum number of horizontal and vertical lines. An optimal assignment is available if the
minimum number of lines is equal to the number of rows (or columns). Otherwise, proceed to Step 5.
6 0 0 Since the minimum number of lines covering all zeroes is now

0 5 1 equal to the number of rows or columns, then an optimal
2 3 0 assignment is available.
How to identify the optimal assignment?
A. Find a row or column that has only one 6 0 0 This unique 0 is associated with
cell that contains a 0 value. This 0 is 0 5 1 Bernice-Denver. Hence, x21 = 1.
recognized as unique 0, and the pair 2 3 0
associated with this 0 shall be included in
the optimal solution.
6 0 0
B. Draw horizontal and vertical lines passing 0 5 1
through this unique 0. 2 3 0
C. Among the uncovered cells with lines, 6 0 0 This unique 0 is associated with
repeat the same process of finding a 0 5 1 Anna-Earl. Hence, x12 = 1.
unique 0 and drawing lines as instructed. 2 3 0
Do this until all required pairings are
achieved. With the last unique 0, Carla is
6 0 0
0 5 1 paired to Francis; x33 = 1.
2 3 0
Anna is assigned to Earl, Bernice to Denver, and Carla to Francis.

The optimal total assignment cost, Z, is –33 (–10*x12 = 1 – 11*x21 =
1 – 12*x33 = 1).
the intersections of two lines. Repeat Step 4 until an optimal solution is obtained.
Because we have reached the optimal solution in Step 4, then this step is no longer applicable.
End.
Fix-It-Shop. Recall EXAMPLE 8.13.
After performing Step 1, 2, and 3, we have reached the following revised tableau:
5 6 0
0 0 3
2 3 0
Because the minimum number of lines required to cover all zeroes is not equal to the number of rows or columns, i.e., 3,
5 6 0
0 0 3
2 3 0
we then proceed to the fifth step.

the intersections of two lines. Repeat Step 4 until an optimal solution is obtained
53 64 00 3 4 0 Since the minimum number of lines covering all

00 00 35 0 0 5 zeroes is now equal to the number of rows or
20 31 00 0 1 0 columns, then an optimal assignment is available.
How to identify the optimal assignment?
A. Find a row or column that has only one 3 4 0 This unique 0 is associated with
cell that contains a 0 value. This 0 is 0 0 5 Ben-Oven. Hence, x22 = 1.
recognized as unique 0 and the pair 0 1 0
associated with this 0 shall be included in
the optimal solution.
3 4 0
B. Draw horizontal and vertical lines passing 0 0 5
through this unique 0. 0 1 0
C. Among the uncovered cells with lines, 3 4 0 This unique 0 is associated with
repeat the same process of finding a 0 0 5 Adam-Table. Hence, x13 = 1.
unique 0 and drawing lines as instructed. 0 1 0
Do this until all required pairings are
achieved. (With the last unique 0, Carlo
3 4 0
0 0 5 shall repair the radio; x31 = 1.)
0 1 0
Adam will repair the table, Ben will do the oven, and Carlo will be
in-charge of the radio. Meanwhile, Z = $25.
End.
The Furniture Factory. Recall EXAMPLE 8.14.
1 2 3 4 5 6 7 8
A 1.4 1.6 1.0 2.4 2.1 0.8 1.4 1.7
B 1.3 0.9 1.0 1.7 1.1 1.3 2.3 2.3
C 1.2 1.9 2.2 2.0 2.0 M 1.3 2.0
D 1.8 1.4 M 1.6 2.2 M 1.0 2.4
E 0.9 2.0 2.4 0.8 0.6 1.6 0.7 2.5
F 0.5 1.4 1.2 2.1 1.0 M 1.3 1.1
G 1.6 0.8 1.4 0.9 1.6 0.6 1.3 1.0
H 2.0 M 1.0 2.0 M 1.3 1.2 2.3
Let us perform the first three steps of the algorithm. The problem is already balanced, so the first step is satisfied. The second
step is to identify each row’s minimum and subtract it from all other entries of the row. For this step as well as the third step,
we do not need to assign any value to M. Regardless of what is subtracted from M, the answer will still be a big number and can
still be generalized as M as exhibited in the following tables.
1 2 3 4 5 6 7 8
A 1.4 0.6 1.6 0.8 1.0 0.2 2.4 1.6 2.1 1.3 0.8 0.0 1.4 0.6 1.7 0.9
B 1.3 0.4 0.9 0.0 1.0 0.1 1.7 0.8 1.1 0.2 1.3 0.4 2.3 1.4 2.3 1.4
C 1.2 0.0 1.9 0.7 2.2 1.0 2.0 0.8 2.0 0.8 MM 1.3 0.1 2.0 0.8
D 1.8 0.8 1.4 0.4 MM 1.6 0.6 2.2 1.2 MM 1.0 0.0 2.4 1.4
E 0.9 0.3 2.0 1.4 2.4 1.8 0.8 0.2 0.6 0.0 1.6 1.0 0.7 0.1 2.5 1.9
F 0.5 0.0 1.4 0.9 1.2 0.7 2.1 1.6 1.0 0.5 MM 1.3 0.8 1.1 0.6
G 1.6 1.0 0.8 0.2 1.4 0.8 0.9 0.3 1.6 1.0 0.6 0.0 1.3 0.7 1.0 0.4
H 2.0 1.0 MM 1.0 0.0 2.0 1.0 MM 1.3 0.3 1.2 0.2 2.3 1.3

Next, identify each column’s minimum and subtract it from all other entries of the column.
1 2 3 4 5 6 7 8
A 0.6 0.6 0.8 0.8 0.2 0.2 1.6 1.4 1.3 1.3 0.0 0.0 0.6 0.6 0.9 0.5
B 0.4 0.4 0.0 0.0 0.1 0.1 0.8 0.6 0.2 0.2 0.4 0.4 1.4 1.4 1.4 1.0
C 0.0 0.0 0.7 0.7 1.0 1.0 0.8 0.6 0.8 0.8 MM 0.1 0.1 0.8 0.4
D 0.8 0.8 0.4 0.4 MM 0.6 0.4 1.2 1.2 MM 0.0 0.0 1.4 1.0
E 0.3 0.3 1.4 1.4 1.8 1.8 0.2 0.0 0.0 0.0 1.0 1.0 0.1 0.1 1.9 1.5
F 0.0 0.0 0.9 0.9 0.7 0.7 1.6 1.4 0.5 0.5 MM 0.8 0.8 0.6 0.2
G 1.0 1.0 0.2 0.2 0.8 0.8 0.3 0.1 1.0 1.0 0.0 0.0 0.7 0.7 0.4 0.0
H 1.0 1.0 MM 0.0 0.0 1.0 0.8 MM 0.3 0.3 0.2 0.2 1.3 0.9
We are now ready to proceed with the next step with the “reduced” matrix below:
1 2 3 4 5 6 7 8
A 0.6 0.8 0.2 1.4 1.3 0 0.6 0.5
B 0.4 0 0.1 0.6 0.2 0.4 1.4 1
C 0 0.7 1 0.6 0.8 M 0.1 0.4
D 0.8 0.4 M 0.4 1.2 M 0 1
E 0.3 1.4 1.8 0 0 1 0.1 1.5
F 0 0.9 0.7 1.4 0.5 M 0.8 0.2
G 1 0.2 0.8 0.1 1 0 0.7 0
H 1 M 0 0.8 M 0.3 0.2 0.9
End.
8.3.3. Problem Set
PROBLEM 8.21. The coach of an age group swim team needs to assign swimmers to a 200-yard medley relay team to send to the
Junior Olympics. Since most of his best swimmers are very fast in more than one stroke, it is not clear which swimmer should be
assigned to each of the four strokes. The five fastest swimmers and the best times (in seconds) they have achieved in each of the
strokes (for 50 yards) are given in the table provided. The coach wishes to determine how to assign four swimmers to the four
different strokes to minimize the sum of the corresponding best times. Obtain an optimal solution.
Carl Chris Davide Tony Ken

Backstroke 37.7 32.9 33.8 37.0 35.4
Breaststroke 43.4 33.1 42.2 34.7 41.8
Butterfly 33.3 28.5 38.9 30.4 33.6
Freestyle 29.2 26.4 29.6 28.5 31.1
o Let xij be equal to 1 if swim Style i is assigned to Swimmer j and 0 otherwise; i = 1 (Backstroke), 2 (Breaststroke), 3
(Butterfly), 4 (Freestyle); j = 1 (Carl), 2 (Chris), 3 (Davide), 4 (Tony), 5 (Ken).
o An optimal assignment is: Z = 126. 2 seconds; x13 = x24 = x32 = x41 = 1; all the remaining variables are equal to 0.
PROBLEM 8.22.
Four cargo ships will be used for shipping goods from one port to four other ports (labeled 1, 2, 3, 4). Any ship can be used for
making any one of these four trips. However, because of differences in the ships and cargoes, the total cost ($) of loading,
transporting, and unloading the goods for the different ship-port combinations varies considerably, as shown in the table provided.
The objective is to assign the four ships to four different ports in such a way as to minimize the total cost for all four shipments.
Ship\Port 1 2 3 4
1 500 400 600 700
2 600 600 700 500
3 700 500 700 600
4 500 400 600 600
PROBLEM 8.23.
The Patricia Garcia Company is producing seven new medical products. Each of Garcia’s eight plants can add one more product
to its current line of medical devices. The unit manufacturing costs ($) for producing the different parts at the eight plants are
shown in the accompanying table. How should Garcia assign the new products to the plants to minimize manufacturing costs?

Plant
Product
1 2 3 4 5 6 7 8
C53 0.10 0.12 0.13 0.11 0.10 0.06 0.16 0.12
C81 0.05 0.06 0.04 0.08 0.04 0.09 0.06 0.06
D5 0.32 0.40 0.31 0.30 0.42 0.35 0.36 0.49
D44 0.17 0.14 0.19 0.15 0.10 0.16 0.19 0.12
E2 0.06 0.07 0.10 0.05 0.08 0.10 0.11 0.05
E35 0.08 0.10 0.12 0.08 0.09 0.10 0.09 0.06
G99 0.55 0.62 0.61 0.70 0.62 0.63 0.65 0.59
o Let xij be equal to 1 if Product i is assigned to Plant j and 0 otherwise; i = 1 (C53), 2 (C81), 3 (D5), …, 7 (G99); j =
1 (1), 2 (2), 3 (3), …, 8 (8).
o An optimal assignment is: Z = $1.18; x16 = x23 = x34 = x45 = x52 = x68 = x71 = 1; all remaining decision variables are
equal to 0.
PROBLEM 8.24.
NASA’s astronaut crew currently includes 10 mission specialists who hold a doctoral degree in either astrophysics or astromedicine.
One of these specialists will be assigned to each of the 10 flights scheduled for the upcoming nine months. Mission specialists are
responsible for carrying out scientific and medical experiments in space or for launching, retrieving, or repairing satellites. The
chief of astronaut personnel, himself a former crew member with three missions under his belt, must decide who should be assigned
and trained for each of the very different missions. Clearly, astronauts with medical educations are more suited to missions
involving biological or medical experiments, whereas those with engineering- or physics-oriented degrees are best suited to other
types of missions. The chief assigns each astronaut a rating on a scale of 1 to 10 for each possible mission, with a 10 being a
perfect match for the task at hand and a 1 being a mismatch. Only one specialist is assigned to each flight, and none is reassigned
until all others have flown at least once.
Mission
Astronaut
Jan. 12 Jan. 27 Feb. 5 Feb. 26 Mar. 26 Apr. 12 May 1 Jun. 9 Aug. 20 Sep. 19
Vincze 9 7 2 1 10 9 8 9 2 6
Veit 8 8 3 4 7 9 7 7 4 4
Anderson 2 1 10 10 1 4 7 6 6 7
Herbert 4 4 10 9 9 9 1 2 3 4
Schatz 10 10 9 9 8 9 1 1 1 1
Plane 1 3 5 7 9 7 10 10 9 2
Certo 9 9 8 8 9 1 1 2 2 9
Moses 3 2 7 6 4 3 9 7 7 9
Brandon 5 4 5 9 10 10 5 4 9 8
Drtina 10 10 9 7 6 7 5 4 8 8
Who should be assigned to which flight? If NASA has just been notified that Anderson is getting married in February and has
been granted a highly sought publicity tour in Europe that month, how does this change the final schedule?
PROBLEM 8.25
Suppose a data processing department wishes to assign five programmers to five programming tasks (one programmer to each
task). Management has estimated the total number of days each programmer would take if assigned to the different jobs, and
these estimates are summarized in the table below. Determine the assignment that minimized the total programmer days required
to complete all five jobs.
Task
Programmer
o Let xij be equal to 1 if Programmer i is assigned to Task j and 0 1 2 3 4 5
otherwise; i = 1 (1), 2 (2), 3 (3), …, 5 (5); j = 1 (1), 2 (2), 3 (3), 1 50 25 78 64 60
…, 5 (5). 2 43 30 70 56 72
o An optimal assignment is: Z = 261 days; x15 = x21 = x32 = x44 = 3 60 28 80 66 68
x53 = 1; all remaining decision variables are equal to 0. 4 54 29 75 60 70
5 45 32 70 62 75
PROBLEM 8.26
Kathleen Taylor is a freshman at Roanoke College, and she wants to develop her schedule for the spring semester. Courses are
offered with class periods either on Monday and Wednesday or Tuesday and Thursday for 1 hour and 15 minutes duration, with
15 minutes between class periods. For example, a course designated as 8M meets on Monday and Wednesday from 8:00 A.M. to
9:15 A.M.; the next class on Monday and Wednesday (9M) meets from 9:30 to 10:45; the next class (11M) is from 11:00 A.M. to
12:15 P.M.; etc.

Kathleen wants to take the following six freshman courses, with the available sections shown in order of her preference, based on
the professor who is teaching the course and the time.
Course Sections Available

Math 11T, 12T, 9T, 11M, 12M, 9M, 8T, 8M
History 11T, 11M, 14T, 14M, 8T, 8M
English 9T, 11T, 14T, 11M, 12T, 14M, 12M, 9M
Biology 14T, 11M, 12M, 14M, 9M, 8T, 8M
Spanish 9T, 11M, 12M, 8T
Psychology 14T, 11T, 12T, 9T, 14M, 8M
First, determine a class schedule for Kathleen that most closely meets her preferences. Second, determine a class schedule for
Kathleen if she wants to leave 11:00 A.M. to noon open for lunch every day. Finally, suppose Kathleen wants all her classes on
two days, either Monday and Wednesday or Tuesday and Thursday. Determine schedules for each and indicate which most
closely matches her preferences.
PROBLEM 8.27
Biggio’s Department Store has six employees available to assign to four departments in the store – home furnishings, china,
appliances, and jewelry. Most of the six employees have worked in each of the four departments on several occasions in the past
and have demonstrated that they perform better in some departments than in others. The average daily sales ($) for each of the
six employees in each of the four departments are shown in the following table. Employee 3 has not worked in the China
department before, so the manager does not want to assign this employee to China. Determine which employee to assign to each
department and indicate the total expected daily sales.
Department Sales
Employee
Home Furnishings China Appliances Jewelry
1 340 160 610 290
2 560 370 520 450
3 270 – 350 420
4 360 220 630 150
5 450 190 570 310
6 280 320 490 360
o Let xij be equal to 1 if Employee i is assigned to Department j and 0 otherwise; i = 1 (1), 2 (2), 3 (3), …, 6 (6); j = 1
(home furnishings), 2 (china), 3 (appliances), 4 (jewelry).
o An optimal assignment is: Z (max) = $1,930; x21 = x34 = x43 = x62 = 1; all remaining decision variables are equal to 0.
PROBLEM 8.28
The accompanying figure gives a schematic layout of a machine shop

with its existing work centers designated by Square 1, 2, 3, and 4. Four
new work centers, I, II, III and IV, are to be added to the shop at the
locations designated by Circle a, b, c, and d. The objective is to assign
the new centers to the proposed locations to minimize the total materials
handling traffic between the existing centers and the proposed ones.
Meanwhile, the following table summarizes the frequency of trips
between the new centers and the old ones. Materials handling equipment
travels along the rectangular aisles intersecting at the locations of the
centers. For example, the one-way travel distance (in meters) between
Center 1 and Location b is 30 + 20 = 50 m.
Existing New center

center I II III IV
1 10 2 4 3
2 7 1 9 5
3 0 8 6 2
4 11 4 0 7

PROBLEM 8.29
The Southeastern Conference has nine basketball officials who must be Game
assigned to three conference games, three to each game. The conference office Official
Athens Columbia Nashville
wants to assign the officials so that the total distance they travel will be 1 165 90 130
minimized. The distance (in miles) each official would travel to each game is 2 75 210 320
given in the table provided. Note, however, Official 2 and 8 recently had a 3 180 170 140
confrontation with one of the coaches in the game in Athens. They were forced 4 220 80 60
to eject the coach after several technical fouls. The conference office decided 5 410 140 80
that it would not be a good idea to have these two officials work the Athens 6 150 170 190
game so soon after this confrontation, so they decided that Official 2 and 8 7 170 110 150
will not be assigned to the Athens game. Obtain an optimal assignment. 8 105 125 160
9 240 200 155
PROBLEM 8.30
A business executive must make the four round trips listed
Departure date from Dallas Return date to Dallas
in the adjusted table between the head office in Dallas and a
Monday, June 3 Friday, June 7
branch office in Atlanta. The price of a round-trip ticket from
Monday, June 10 Wednesday, June 12
Dallas is $400. A discount of 25% is granted if the dates of
Monday, June 17 Friday, June 21
arrival and departure of a ticket span a weekend (Saturday
Tuesday, June 25 Friday, June 28
and Sunday).
If the stay in Atlanta lasts more than 21 days, the discount is increased to 30%. A one-way ticket between Dallas and Atlanta
(either direction) costs $250. How should the executive purchase the tickets?
o Let xij be equal to 1 if Departure Date i is paired with Return Date j and 0 otherwise; i = 1 (June 3), 2 (June 10), 3
(June 17), 4 (June 25); j = 1 (June 7), 2 (June 12), 3 (June 21), 4 (June 28).
o An optimal solution is: Z = $1,180; x14 = x21 = x32 = x43 = 1; all remaining decision variables are equal to 0.
8.4. Traveling Salesperson Problems
A special application of the assignment problem is the traveling salesperson problem (TSP). The latter problem can be framed as
an assignment problem with an additional, peculiar constraint – no subtours! What is a subtour? To define subtour, let us define
first what a tour means. Consider a salesman that needs to visit five cities. The challenge is to start in one city, visit the other
cities exactly once, and return to the starting city most efficiently, i.e., the shortest total distance or time traveled. If the cities
are 1, 2, 3, 4 and 5, then a tour, for example, is 1 – 4 – 2 – 5 – 3 – 1. This means that the salesman’s route starts in City 1 then
heads to City 4, 2, 5 and 3 then goes back to City 1. A subtour is observed when for example the salesman starts in City 1 goes
to City 4 and 2 then heads back to City 1 (1 – 4 – 2 – 1) without having visited City 5 and 3. Another example of a subtour is 3
– 5 – 3. None between these two latter routes can visited all the cities in one “straight” travel. Given a tour then, any subset of
it is recognized as a subtour.
Going back to 1 – 4 – 2 – 3 – 1 as a sample tour. An important question is if there is a better way, if not the most efficient, to
visit all the cities starting from and ending at City 1? Why not take the 1 – 4 – 2 – 5 – 3 – 1 tour instead? What could be the
best tour then starting in any city and ending in that same city? This is what we want to answer in TSP. Let us now consider
specific examples to illustrate TSP.
8.4.1. Modeling and Solving
EXAMPLE 8.17 Traveling Salesperson Problem
5-City TSP. Consider the following 5-city TSP:
City 1 City 2 City 3 City 4 City 5

City 1 – 10 3 6 9
City 2 5 – 5 4 2
City 3 4 9 – 7 8
City 4 7 1 3 – 4
City 5 3 2 6 5 –
Modeling
Before we proceed to the corresponding model, let us try first to interpret figures. As per the table, the distance going from City
1 to City 2 to 10 distance units. However, going from City 2 to City 1 is only 5 distance units. It is common for distance

matrices/tables to exhibit symmetry, i.e., the distance from A to B is just equal to the distance from B to A. It must be noted
though that there are situations where distance information is asymmetric for various reasons which the readers are encouraged
to ponder on.
As mentioned, this TSP can be represented by an assignment model. For the decision variables, we declare the following:
1, City i and j are on the tour

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5.
Alternatively,
1, Go from City i to City j

xij = {
0, Otherwise.
The additional constraint is stated merely as
Subtour: No subtour.
The remaining elements of the model are constructed similarly with that of an assignment problem. The complete model is

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5
Min Z = Mx11 + 10x12 + 3x13 + 6x14 + 9x15 + 5x21 + Mx22 + 5x23 + 4x24 + 2x25 + 4x31 + 9x32
+ Mx33 + 7x34 + 8x35 + 7x41 + x42 + 3x43 + Mx44 + 4x45 + 3x51 + 2x52 + 6x53 + 5x54 + Mx55
Subject to:
City 1: x11 + x12 + x13 + x14 + x15 = 1 x11 + x21 + x31 + x41 + x51 = 1
City 2: x21 + x22 + x23 + x24 + x25 = 1 x12 + x22 + x32 + x42 + x52 = 1
City 3: x31 + x32 + x33 + x34 + x35 = 1 x13 + x23 + x33 + x43 + x53 = 1
City 4: x41 + x42 + x43 + x44 + x45 = 1 x14 + x24 + x34 + x44 + x54 = 1
City 5: x51 + x52 + x53 + x54 + x55 = 1 x15 + x25 + x35 + x45 + x55 = 1
Bin: xij = (0, 1)
Solving
With the model that we have, let us now try to solve the TSP. We run the assignment model iteratively until we obtain an
optimal solution for the assignment model that has no subtour. Running the assignment model (note that the subtour constraint
is removed for now),

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5
Subject to:
City 1: x11 + x12 + x13 + x14 + x15 = 1 x11 + x21 + x31 + x41 + x51 = 1
City 2: x21 + x22 + x23 + x24 + x25 = 1 x12 + x22 + x32 + x42 + x52 = 1
City 3: x31 + x32 + x33 + x34 + x35 = 1 x13 + x23 + x33 + x43 + x53 = 1
City 4: x41 + x42 + x43 + x44 + x45 = 1 x14 + x24 + x34 + x44 + x54 = 1
City 5: x51 + x52 + x53 + x54 + x55 = 1 x15 + x25 + x35 + x45 + x55 = 1
Bin: xij = (0, 1).
an optimal solution obtained is: Z = 15, x13 = x25 = x31 = x42 = x54 = 1; all remaining decision variables are equal to 0. Therefore,
the suggested sequences are 1 – 3 – 1 and 2 – 5 – 4 – 2. At this point, we have two subtours which we do not like. We must break
then one subtour. Since 1 – 3 – 1 involves only two cities, we choose this subtour to break for convenience. If we can eliminate
one subtour, in general, this will eliminate at least one of the other subtours; hopefully, all other subtours are eliminated. To
break now 1 – 3 – 1, we add the constraint,
1 – 3 – 1: x13 + x31 ≤ 1.
to force to have only at most one link between City 1 and City 3. This constraint now specifically represents the subtour constraint.
Running then the updated model provides the following optimal solution: Z = 16; x13 = x25 = x34 = x42 = x51 = 0; all remaining

decision variables are equal to 0. Hence, 1 – 3 – 4 – 2 – 5 – 1. We have now a tour! Thus, it is recommended to visit the five cities
in the following order: 1, 3, 4, 2, 5 and 1 (if we are to start in and end at City 1) with a total distance of 16 distance units.
In summary, the model providing the tour 1 – 3 – 4 – 2 – 5 – 1 is

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5
Subject to:
City 1: x11 + x12 + x13 + x14 + x15 = 1 x11 + x21 + x31 + x41 + x51 = 1
City 2: x21 + x22 + x23 + x24 + x25 = 1 x12 + x22 + x32 + x42 + x52 = 1
City 3: x31 + x32 + x33 + x34 + x35 = 1 x13 + x23 + x33 + x43 + x53 = 1
City 4: x41 + x42 + x43 + x44 + x45 = 1 x14 + x24 + x34 + x44 + x54 = 1
City 5: x51 + x52 + x53 + x54 + x55 = 1 x15 + x25 + x35 + x45 + x55 = 1
Bin: xij = (0, 1)
1 – 3 – 1: x13 + x31 ≤ 1.
What if instead of breaking 1 – 3 – 1, we decide to break the longer subtour, 2 – 5 – 4 – 2? Then we have
2 – 5 – 4 – 2: x25 + x54 + x42 + x52 + x45 + x24 ≤ 2.
With this constraint, not only the subtour 2 – 5 – 4 – 2 that is prohibited but also the subtour 2 – 4 – 5 – 2; hence, we are
allowing at most two links from the routes involving the three cities (2, 5 and 4). This is the reason now why the RHS is 2. In
general, the LHS includes all the possible combinations or pairs of cities in this case while the RHS is one less than the number
of cities contained in the subtour. Running the preceding model coincidentally provides the same recommendation: 1 – 3 – 4 – 2
– 5 – 1.
In summary, the alternative model is

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5
Subject to:
City 1: x11 + x12 + x13 + x14 + x15 = 1 x11 + x21 + x31 + x41 + x51 = 1
City 2: x21 + x22 + x23 + x24 + x25 = 1 x12 + x22 + x32 + x42 + x52 = 1
City 3: x31 + x32 + x33 + x34 + x35 = 1 x13 + x23 + x33 + x43 + x53 = 1
City 4: x41 + x42 + x43 + x44 + x45 = 1 x14 + x24 + x34 + x44 + x54 = 1
City 5: x51 + x52 + x53 + x54 + x55 = 1 x15 + x25 + x35 + x45 + x55 = 1
Bin: xij = (0, 1)
2 – 5 – 4 – 2: x25 + x54 + x42 + x52 + x45 + x24 ≤ 2.
End.
EXAMPLE 8.18 Traveling Salesperson Problem
Painting Operation. A painting operation is scheduled in blocks, where each block involves painting products with a particular
color. Cleaning time is required between each pair of blocks so that the equipment can be prepared for the new color. In each
cycle, there is one block of each color, and the total painting time is determined by the volume of orders. However, the actual
schedule length is determined by the sequence in which the blocks are scheduled since the cleaning time depends on the color in
the previous block and the color in the next block. The table below gives the number of minutes required to clean the equipment,
according to the color pair. Find the block sequence that minimizes the amount of time spent on cleaning during a full cycle.
From\To Color 1 Color 2 Color 3 Color 4 Color 5 Color 6

Color 1 – 4 8 6 8 2
Color 2 5 – 7 11 13 4
Color 3 11 6 – 8 4 3
Color 4 5 7 2 – 2 5
Color 5 10 9 7 5 – 2
Color 6 8 4 3 6 5 –

Modeling
It is not also necessary in TSP that we deal literally with traveling and distances. A setup problem such as this example can also
be analyzed as a TSP where the colors are treated as the cities and the setup times (number of minutes required to clean the
equipment according to color pair) are treated as the distances. The challenge is to find a sequence (tour) of these colors that will
minimize the total setup or cleaning times required.
The complete model is
1, Color j follows Color i

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5, 6
Min Z = Mx11 + 4x12 + 8x13 + 6x14 + 8x15 + 2x16 + 5x21 + Mx22 + 7x23 + 11x24 + 13x25 + 4x26
+ 11x31 + 6x32 + Mx33 + 8x34 + 4x35 + 3x36 + 5x41 + 7x42 + 2x43 + Mx44 + 2x45 + 5x46 + 10x51
+ 9x52 + 7x53 + 5x54 + Mx55 + 2x56 + 8x61 + 4x62 + 3x63 + 6x64 + 5x65 + Mx66
Subject to:
Color 1: x11 + x12 + x13 + x14 + x15 + x16 = 1 x11 + x21 + x31 + x41 + x51 + x61 = 1
Bin: xij = (0, 1)
Solving
We run the following assignment model:

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5, 6
Min Z = Mx11 + 4x12 + 8x13 + 6x14 + 8x15 + 2x16 + 5x21 + Mx22 + 7x23 + 11x24 + 13x25 + 4x26
+ 11x31 + 6x32 + Mx33 + 8x34 + 4x35 + 3x36 + 5x41 + 7x42 + 2x43 + Mx44 + 2x45 + 5x46 + 10x51
+ 9x52 + 7x53 + 5x54 + Mx55 + 2x56 + 8x61 + 4x62 + 3x63 + 6x64 + 5x65 + Mx66
Subject to:
Bin: xij = (0, 1).
An optimal solution obtained is: Z = 22 minutes: x12 = x21 = x36 = x45 = x54 = x63 = 1; all remaining decision variables are equal
to zero. The subtours are 1 – 2 – 1, 3 – 6 – 3, and 4 – 5 – 4. Since all the subtours involve two cities, we resort to arbitrarily
selecting one to break. In this case, we work first with 1 – 2 – 1.
1 – 2 – 1: x12 + x21 ≤ 1
An optimal solution is Z = 22; x16 = x21 = x35 = x43 = x54 = x62 =1; all remaining decision variables are equal to 0. The subtours
then are 1 – 6 – 2 – 1 and 3 – 5 – 4 – 3. We arbitrarily select now the first subtour to break.
1 – 6 – 2 – 1: x16 + x62 + x21 + x12 + x26 + x61 ≤ 2
An optimal solution is: Z = 23 minutes; x16 = x21 = x32 = x45 = x54 = x63 = 1; all remaining decision variables are equal to 0. The
subtours are 1 – 6 – 3 – 2 – 1 and 4 – 5 – 4. For convenience, we break the latter subtour.
4 – 5 – 4: x45 + x54 ≤ 1
An optimal solution is: Z = 23 minutes; x14 = x21 = x35 = x43 = x56 = x62 = 1; all remaining decision variables are equal to 0.
Finally, we arrive with a tour solution, 1 – 4 – 3 – 5 – 6 – 2 – 1. This is then suggested as the block sequence to minimize the
total cleaning time.

In summary,

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5, 6
Min Z = Mx11 + 4x12 + 8x13 + 6x14 + 8x15 + 2x16 + 5x21 + Mx22 + 7x23 + 11x24 + 13x25 + 4x26 +
11x31 + 6x32 + Mx33 + 8x34 + 4x35 + 3x36 + 5x41 + 7x42 + 2x43 + Mx44 + 2x45 + 5x46 + 10x51 +
9x52 + 7x53 + 5x54 + Mx55 + 2x56 + 8x61 + 4x62 + 3x63 + 6x64 + 5x65 + Mx66
Subject to:
Bin: xij = (0, 1)
1 – 2 – 1: x12 + x21 ≤ 1
1 – 6 – 2 – 1: x16 + x62 + x21 + x12 + x26 + x61 ≤ 2
4 – 5 – 4: x45 + x54 ≤ 1.
The three last constraints reflect the subtour constraint in general.
Supposed again we take a different path in selecting the subtours to break. After running the original assignment model, we
obtained 1 – 2 – 1, 3 – 6 – 3 and 4 – 5 – 4 as subtours. We selected 1 – 2 – 1, remember? Now, what if we have selected 3 – 6 –
3 instead, so that
3 – 6 – 3: x36 + x63 ≤ 1.
The subtours then are 1 – 6 – 2 – 1 and 3 – 5 – 4 – 3 with Z = 22. Suppose we break 3 – 5 – 4 – 3.
3 – 5 – 4 – 3: x35 + x54 + x43 + x34 + x45 + x53 ≤ 2
We have now a tour solution, 1 – 4 – 3 – 5 – 6 – 2 – 1. This iteration path is more efficient in finding a tour solution. But of
course, we do not know this path is more efficient unless we try it! In summary,

Let xij = {
0, Otherwise
i, j = 1, 2, 3, 4, 5, 6
Min Z = Mx11 + 4x12 + 8x13 + 6x14 + 8x15 + 2x16 + 5x21 + Mx22 + 7x23 + 11x24 + 13x25 + 4x26 +
11x31 + 6x32 + Mx33 + 8x34 + 4x35 + 3x36 + 5x41 + 7x42 + 2x43 + Mx44 + 2x45 + 5x46 + 10x51 + 9x52
+ 7x53 + 5x54 + Mx55 + 2x56 + 8x61 + 4x62 + 3x63 + 6x64 + 5x65 + Mx66
Subject to:
Bin: xij = (0, 1)
3 – 6 – 3: x36 + x63 ≤ 1
3 – 5 – 4 – 3: x35 + x54 + x43 + x34 + x45 + x53 ≤ 2
As for final remarks, one, notice that in this example and the previous, Z either increases or remains the same as we go along
with our subtour elimination process. Second, the constraints that we add to break subtours are logical constraints in nature.
Finally, for further reading, TSP is classified as NP-hard problem in Operations Research.
End.

8.4.2. Problem Set
PROBLEM 8.31
Recent graduate and amateur world traveler Alastair Bor is planning a European trip. His preferences are influenced by his
curiosity about urban culture in Europe and by his extensive study of international relations while he was in school. Accordingly,
he has decided to make one stop in each of 12 European capitals in the time he has available. He wants to find a sequence of the
cities that involves the least total mileage. He has calculated inter-city distances using published data on latitude and longitude
and applying the geometry for arcs of great circles. These distances are shown below. Find a minimum distance tour for Alastair,
starting and ending in Brussels. What is the length of the tour?
Ams Ath Ber Brus Cope Dub Lis Lon Lux Mad Par Rom
Amsterdam – 2166 577 175 622 712 1889 339 319 1462 430 1297
Athens – 1806 2092 2132 2817 2899 2377 1905 2313 2100 1053
Berlin – 653 348 1273 2345 912 598 1836 878 1184
Brussels – 768 732 1738 300 190 1293 262 1173
Copenhagen – 1203 2505 942 797 2046 1027 1527
Dublin – 1656 440 914 1452 743 1849
Lisbon – 1616 1747 600 1482 1907
London – 475 1259 331 1419
Luxembourg – 1254 293 987
Madrid – 1033 1308
Paris – 1108
Rome –
Note: Distance between two destinations are symmetric.
PROBLEM 8.32
A book salesperson who lives in Basin must call once a month on four Miles between cities
customers located in Wald, Bon, Mena, and Kiln. The adjacent table Basin Wald Bon Mena Kiln
gives the distances in miles among the different cities. The objective is Basin 0 120 220 150 210
to minimize the total distance traveled by the salesperson. Wald 120 0 80 110 130
Bon 220 80 0 160 185
Mena 150 110 160 0 190
Kiln 210 130 185 190 0
PROBLEM 8.33
A manager has a total of 10 employees working on six
Employee\Project 1 2 3 4 5 6
projects. There are overlaps among the assignments as the
1 X X X
adjacent table shows.
2 X X X
3 X X X X
The manager meets with each employee individually once a
4 X X X
week for a progress report. Each meeting lasts about 20
5 X X X
minutes for a total of 3 hours and 20 minutes for all 10
6 X X X X X
employees. To reduce the total time, the manager wants to
7 X X X X
hold group meetings depending on shared projects. The
8 X X X
objective is to schedule the meetings in a way that will reduce
9 X X
the traffic (number of employees) in and out of the meeting
10 X X X X X
room.
PROBLEM 8.34
Circuit boards (such as those used with PCs) are fitted with holes for mounting different electronic components. The holes are
drilled with a movable drill. The following table provides distances (in centimeters) between pairs of 6 holes of a specific circuit
board. Solve this problem as TSP.
– 1.2 0.5 2.6 4.1 3.2

1.2 – 3.4 4.6 2.9 5.2
0.5 3.4 – 3.5 4.6 6.2
‖dij ‖=
2.6 4.6 3.5 – 3.8 0.9
4.1 2.9 4.6 3.8 – 1.9
(3.2 5.2 6.2 0.9 1.9 – )
✦ End of chapter ✦

Chapter 8 TTAT

Uploaded by

Document Information

Original Description:

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

Chapter 8 TTAT

Uploaded by

Copyright:

Available Formats

Chapter 8

8. Transportation, Transshipment, Assignment, and Travelling

8.1. Transportation Problems

In a transportation problem, the task is to determine the number of units to

Let i (1, 2, 3, …, m) be a source and j (1, 2, 3, …, n) be a destination so that

Min Z = ∑ ∑ cij xij

Supply: ∑ xij = ai ∀i Demand: ∑ xij = bj ∀j

Supply: ∑ xij ≤ ai ∀i Demand: ∑ xij ≥ bj ∀j

Let xij – number of units to be transported from Source i to Destination j; i = 1, 2, 3, …, m; j = 1, 2, 3, …, n

Min Z = ∑ ∑ cij xij

MANUAL ON MANAGEMENT SCIENCE (MGT 221)

Supply: ∑ xij ≤ ai ∑ xij = ai ∀i

Demand: ∑ xij ≥ bj ∑ xij ≤ bj ∀j

NN: xij ≥ 0 xij ≥ 0.

EXAMPLE 8.1 Modeling – Transportation Problems

From\To Denver (1) Miami (2)

Formulate an LP model that will minimize the total transportation costs.

Min Z = 80x11 + 215x12 + 100x21 + 108x22 + 102x31 + 68x32

Los Angeles: x11 + x12 ≤ 1,000.

It is easy to see now that the remaining supply constraints are

Detroit: x21 + x22 ≤ 1,500

Denver: x11 + x21 + x31 ≥ 2,300

MANUAL ON MANAGEMENT SCIENCE (MGT 221)

Meanwhile, the remaining demand constraint is

Miami: x12 + x22 + x32 ≥ 1,400.

Finally, we include the NN constraint.

NN: x11, x12, x21, x22, x31, x32 ≥ 0

Let x11 – number of units to be transported from Los Angeles to Denver

EXAMPLE 8.2 Modeling – Transportation Problems

Subject to: Subject to:

MANUAL ON MANAGEMENT SCIENCE (MGT 221)

EXAMPLE 8.3 Modeling – Transportation Problems

Grain Elevator Supply Mill Demand

From\To Chicago (1) St. Louis (2) Cincinnati (3)

Let xij – tons of grains to be shipped from Source i to Destination j; i, j = 1, 2, 3

EXAMPLE 8.4 Modeling – Transportation Problems

MANUAL ON MANAGEMENT SCIENCE (MGT 221)

xAM – number of Math books printed with Machine A

The objective is expressed as

Min Z = 21xAM + 27xAS + 19xAR + 25xBM + 30xBS + 23xBR.

Machine A: xAM + xAS + xAR ≤ 210 Math: xAM + xBM ≥ 100

Let xAM – number of Math books printed with Machine A

EXAMPLE 8.5 Modeling – Transportation Problems

Mode of Production Monthly Capacity Unit Cost (labor + material)

September October November December

MANUAL ON MANAGEMENT SCIENCE (MGT 221)

Overtime (3) 100,000 100,000 100,000 100,000 8,500

Overtime (6) 100,000 100,000 100,000 100,000 8,500

Overtime (9) 100,000 100,000 100,000 100,000 8,500

Overtime (12) 100,000 100,000 100,000 100,000 8,500

Overtime (3) 100,000 100,000 100,000 100,000 8,500

Overtime (6) 100,000 100,000 100,000 100,000 8,500

Overtime (9) 100,000 100,000 100,000 100,000 8,500

Overtime (12) 100,000 100,000 100,000 100,000 8,500

MANUAL ON MANAGEMENT SCIENCE (MGT 221)

Overtime (6) 100,000 100,000 100,000 100,000 8,500

Overtime (9) 100,000 100,000 100,000 100,000 8,500

Overtime (12) 100,000 100,000 100,000 100,000 8,500

Overtime (3) 100,000 100,000 100,000 100,000 8,500

Overtime (6) M 100,000 100,000 100,000 8,500

Overtime (9) M M 100,000 100,000 8,500

Overtime (12) M M M 100,000 8,500

Overtime (3) 2,500 100,000 100,000 100,000 8,500