Linear Programming:

Formulation and Applications

Outline:
1. Example: Personnel Scheduling Problem
2. Example: Transportation Problem
3. Example: Assignment Problem
4. Model Formulation from a Broader Perspective
5. Summary

Learning objectives:
• Recognize various kinds of managerial problems to which linear programming can
be applied.
• Formulate a linear programming model from a description of a problem in any of
these categories.
• Identify the key components of any linear programming model and the kind of
spreadsheet cells used for each component.
• Describe different type of constraints and benefit constraints, and the difference in
how they arise.
• Understand the flexibility that managers have in prescribing key considerations that
can be incorporated into a linear programming model.

Linear programming problems come in many guises. And their models take various forms.
This diversity can be confusing to both students and managers, making it difficult to
recognize when linear programming can be applied to address a managerial problem. Since
managers instigate management science studies, the ability to recognize the applicability
of linear programming is an important managerial skill. This chapter focuses largely on
developing this skill.

Our approach will be to emphasize these common threads—the identifying features —

that tie together linear programming problems even when they arise in very different
contexts. We will describe some broad categories of linear programming problems and the
identifying features that characterize them. Then we will use diverse examples, but with
the purpose of illustrating and emphasizing the common threads among them.

We will focus on five key categories of linear programming problems via examples:
resource-allocation problems (e.g., Wyndor problem), cost–benefit–trade-off problems
(e.g., Profit & Gambit Problem, Union Airways Personnel Scheduling), mixed problems
(e.g., in an assignment problem), transportation problems (e.g., Big M Company), and
assignment problems (e.g., Sellmore Company). In each case, an important identifying
feature is the nature of the restrictions on what decisions can be made, and thus the nature
of the resulting functional constraints in the linear programming model. For each
category/example, you will see how the basic data for a problem lead directly to a linear
programming model with a certain distinctive form. Thus, model formulation becomes a
by-product of proper problem formulation.
1. Example: Personnel Scheduling Problem
The personnel scheduling for a company that provides some kind of service involves cost–
benefit–trade-off analysis. The objective is to schedule the work times of the company’s
employees so as to minimize the cost of providing the level of service specified by
management. The following example illustrates how this can be done.

The Problem

Union Airways is adding more flights to and from its hub airport and so needs to hire
additional customer service agents. However, it is not clear just how many more should be
hired. Management recognizes the need for cost control while also consistently providing a
satisfactory level of service to the company’s customers, so a desirable trade-off between
these two factors is being sought. Therefore, a management science team is studying how
to schedule the agents to provide satisfactory service with the smallest personnel cost.

Based on the new schedule of flights, an analysis has been made of the minimum number
of customer service agents that need to be on duty at different times of the day to provide a
satisfactory level of service. (The queueing models, another topic of this course, can be
used to determine the minimum numbers of agents needed to keep customer waiting times
reasonable.) These numbers are shown in the last column of Table 5 for the time periods
given in the first column. The other entries in this table reflect one of the provisions in the
company’s current contract with the union that represents the customer service agents. The
provision is that each agent works an eight-hour shift. The authorized shifts are
Shift 1: 6:00 am to 2:00 pm.
Shift 2: 8:00 am to 4:00 pm.
Shift 3: Noon to 8:00 pm.
Shift 4: 4:00 pm to midnight.
Shift 5: 10:00 pm to 6:00 am.
Check marks in the main body of Table 5 show the time periods covered by the respective
shifts. Because some shifts are less desirable than others, the wages specified in the
contract differ by shift. For each shift, the daily compensation (including benefits) for each
agent is shown in the bottom row. The problem is to determine how many agents should be
assigned to the respective shifts each day to minimize the total personnel cost for agents,
based on this bottom row, while meeting (or surpassing) the service requirements given in
the last column.

Formulation

This problem is, in fact, a pure cost–benefit–trade-off problem (another example is the
Profit & Gambit Problem). To formulate the problem, we need to identify the activities
and benefits involved.
Activities correspond to shifts.
The level of each activity is the number of agents assigned to that shift.
A unit of each activity is one agent assigned to that shift.
Thus, the general description of a linear programming problem as finding the best mix of
activity levels can be expressed for this specific application as finding the best mix of shift
sizes.
Benefits correspond to time periods.
For each time period, the benefit provided by the activities is the service that agents
provide customers during that period.
 The level of a benefit is measured by the number of agents on duty during that time
period.

Once again, a careful formulation of the problem, including gathering all the relevant data,
leads rather directly to a spreadsheet model. This model is shown in Figure 5, and we
outline its formulation below.

The Data As indicated in this figure, all the data in Table 5 have been entered directly into
the data cells CostPerShift (C5:G5), ShiftWorksTimePeriod (C8:G17), and
MinimumNeeded (J8:J17). For the ShiftWorksTimePeriod (C8:G17) data, an entry of 1
indicates that the corresponding shift includes that time period whereas 0 indicates not.
Like any cost–benefit–tradeoff problem, these numbers indicate the contribution of each
activity to each benefit. Each agent working a shift contributes either 0 or 1 toward the
minimum number of agents needed in a time period.

The Decisions Since the activities in this case correspond to the five shifts, the decisions to
be made are
S1 = Number of agents to assign to Shift 1 (starts at 6 am )
S2 = Number of agents to assign to Shift 2 (starts at 8 am )
S3 = Number of agents to assign to Shift 3 (starts at noon)
S4 = Number of agents to assign to Shift 4 (starts at 4 pm )
S5 = Number of agents to assign to Shift 5 (starts at 10 pm )
The changing cells to hold these numbers have been placed in the activity columns in row
21, so
S1 → cell C21, S2 → cell D21, ……, S5 → cell G21
where these cells are collectively referred to by the range name NumberWorking
(C21:G21).

The Constraints These changing cells need to be nonnegative. In addition, we need 10

benefit constraints, where each one specifies that the total number of agents serving in the
corresponding time period listed in column B must be no less than the minimum
acceptable number given in column J. Thus, these constraints are
Total number of agents serving 6--8 am  48 (min. acceptable)
Total number of agents serving 8--10 am  79 (min. acceptable)
……
Total number of agents serving midnight--6 am  15 (min. acceptable)
Since columns C to G indicate which of the shifts serve each of the time periods, these
totals are
Total number of agents serving 628 am = S1
Total number of agents serving 8210 am = S1 + S2
……
Total number of agents serving midnight 26 am = S5
These totals are calculated in the output cells TotalWorking (H8:H17) using the
SUMPRODUCT functions shown below the spreadsheet in Figure 5.

Note: One other type of constraint is that the number of agents assigned to each shift must
have an integer value. With Excel’s Solver they can be added in the Add Constraint dialog
box by entering NumberWorking on the left-hand side and then choosing “int” from the
pop-up menu between the left-hand side and the right-hand side. The set of constraints,
NumberWorking = integer, then appears in the Solver Parameters, as shown in Figure 5.
Fortunately, in this example, researchers have shown that we will be able to obtain the
correct (integer) solution even if we omit the integer requirement.
The Measure of Performance The objective is to
Minimize Cost = Total daily personnel cost for all agents
Since CostPerShift (C5:G5) gives the daily cost per agent on each shift and
NumberWorking (C21:G21) gives the number of agents working each shift,
Cost = 170 S1 + 160 S2 + 175 S3 + 180 S4 + 195 S5 (in dollars)
= SUMPRODUCT(CostPerShift, NumberWorking)
→ cell J21

Summary of the Formulation The above steps provide the complete formulation of the
linear programming model on a spreadsheet, as summarized below (in algebraic form).

Minimize Cost = 170 S1 + 160 S2 + 175 S3 + 180 S4 + 195 S5 (in dollars)

subject to
Total agents 6--8 am: S1  48
Total agents 8--10 am: S1 + S2  79
……
Total agents midnight--6 am: S5  15
and
S1  0, S2  0, S3  0, S4  0, S5  0

TABLE 5 Data for the Union Airways Personnel Scheduling Problem

Solving the Model The lower left-hand corner of Figure 5 shows the entries needed in
Solver, along with the selection of the usual two options. After solving, NumberWorking
(C21:G21) in the spreadsheet shows the resulting optimal solution for the number of
agents that should be assigned to each shift. TotalCost (J21) indicates that this plan would
cost $30,610 per day.
FIGURE 5 The spreadsheet model for the Union Airways problem, including the formulas
for the objective cell TotalCost (J21) and the other output cells TotalWorking (H8:H17), as
well as the specifications needed to set up Solver. The changing cells NumberWorking
(C21:G21) show the optimal solution obtained by Solver.

In this example, we have applied a useful concept in modeling, the incidence matrix,
which is a 0-1 matrix (“table”) that shows the relationship between two classes of objects
(or, two sets of indexes). In such a matrix/table, the entry in row-x and column-y is 1 if x
and y are related (called incident in this context) and 0 if they are not. For example,
“period-3 row (10am-12pm)” is related to (served by) “shift-2 column (agent)”.

2. Example: Transportation Problem

One of the most common applications of linear programming involves optimizing a
shipping plan for transporting goods. In a typical application, a company has several plants
producing a certain product that needs to be shipped to the company’s customers (or
perhaps to distribution centers). How much should each plant ship to each customer in
order to minimize the total cost? Linear programming can provide the answer. This type of
linear programming problem is called a transportation problem.

This kind of application normally needs two kinds of functional constraints. One kind
specifies that the amount of the product produced at each plant must equal the total amount
shipped to customers. The other kind specifies that the total amount received from the
plants by each customer must equal the amount ordered. These are fixed-requirement
constraints, which makes the problem a fixed-requirements problem.

The Big M Company Transportation Problem

The Big M Company produces a variety of heavy duty machines at two factories. One of
its products is a large turret lathe. Orders have been received from three customers to
purchase some of these turret lathes next month. These lathes will be shipped individually,
and Table 9 shows what the cost will be for shipping each lathe from each factory to each
customer. This table also shows how many lathes have been ordered by each customer and
how many will be produced by each factory. The company’s distribution manager now
wants to determine how many machines to ship from each factory to each customer to
minimize the total shipping cost.

Figure 9 depicts (logically) the distribution network for this problem. This network ignores
the geographical layout of the factories and customers and instead lines up the two
factories in one column on the left and the three customers in one column on the right.
Each arrow shows one of the shipping lanes through this distribution network.

Formulation of the Problem in Linear Programming Terms

We need to identify the activities and requirements of this transportation problem to

formulate it as a linear programming problem. In this case, two kinds of activities have
been mentioned—the production of the turret lathes at the two factories and the shipping
of these lathes along the various shipping lanes. However, we know the specific amounts
to be produced at each factory, so no decisions need to be made about the production
activities. The decisions to be made concern the levels of the shipping activities —how
many lathes to ship through each shipping lane. Therefore, we need to focus on the
shipping activities for the linear programming formulation.
The activities correspond to shipping lanes, depicted by arrows in Figure 9.
The level of each activity is the number of lathes shipped through the
corresponding shipping lane.

Just as any linear programming problem can be described as finding the best mix of
activity levels, this one involves finding the best mix of shipping amounts for the various
shipping lanes. The decisions to be made are
SF1-C1 = Number of lathes shipped from Factory 1 to Customer 1
SF1-C2 = Number of lathes shipped from Factory 1 to Customer 2
SF1-C3 = Number of lathes shipped from Factory 1 to Customer 3
SF2-C1 = Number of lathes shipped from Factory 2 to Customer 1
SF2-C2 = Number of lathes shipped from Factory 2 to Customer 2
SF2-C3 = Number of lathes shipped from Factory 2 to Customer 3
so six changing cells will be needed in the spreadsheet.
The objective is to
Minimize Cost = Total cost for shipping the lathes
Using the shipping costs given in Table 9,
Cost = 700SF1-C1 + 900SF1-C2 + 800SF1-C3 + 800SF1-C1 + 900SF2-C2 + 700SF2-C3
is the quantity in dollars to be entered into the objective cell. (We will use a
SUMPRODUCT function to do this a little later.)

The spreadsheet model also will need five constraints involving fixed requirements. Both
Table 9 and Figure 9 show these requirements.
Requirement 1: Factory 1 must ship 12 lathes.
Requirement 2: Factory 2 must ship 15 lathes.
Requirement 3: Customer 1 must receive 10 lathes.
Requirement 4: Customer 2 must receive 8 lathes.
Requirement 5: Customer 3 must receive 9 lathes.
Thus, there is a specific requirement associated with each of the five locations in the
distribution network shown in Figure 9.

All five of these requirements can be expressed in constraint form as

Amount provided = Required amount
For example, Requirement 1 can be expressed algebraically as
SF1-C1 + SF1-C2 + SF1-C3 = 12
where the left-hand side gives the total number of lathes shipped from Factory 1, and 12 is
the required amount to be shipped from Factory 1. Therefore, this constraint restricts SF1-C1,
SF1-C2, and SF1-C3 to values that sum to the required amount of 12. In contrast to the  form
for resource constraints and the  form for benefit constraints, the constraints express
fixed requirements that must hold with equality, so this transportation problem falls into
the category of fixed-requirements problems introduced in the preceding section.
However, variants of transportation problems can have resource constraints or benefit
constraints as well. For example, if 12 lathes represent the manufacturing capacity of
Factory 1 (the maximum number that can be shipped) rather than a requirement for how
many must be shipped, the constraint just given for Requirement 1 would become a 
resource constraint instead. Such variations can be incorporated readily into the
spreadsheet model.

Formulation of the Spreadsheet Model

In preparation for formulating the model, the problem has been formulated above by
identifying the decisions to be made, the constraints on these decisions, and the overall
measure of performance, as well as gathering all the important data displayed in Table 9.
All this information leads to the spreadsheet model shown in Figure 10. The data cells
include ShippingCost (C5:E6), Output (H11:H12), and OrderSize (C15:E15),
incorporating all the data from Table 9. The changing cells are UnitsShipped (C11:E12),
which give the decisions on the amounts to be shipped through the respective shipping
lanes. The output cells are TotalShippedOut (F11:F12) and TotalToCustomer (C13:E13),
where the SUM functions entered into these cells are shown below the spreadsheet in
Figure 10. The constraints are that TotalShippedOut is required to equal Output and
TotalToCustomer is required to equal OrderSize. These constraints have been specified on
the spreadsheet and entered into Solver. The objective cell is TotalCost (H15), where its
SUMPRODUCT function gives the total shipping cost. The lower left-hand corner of
Figure 10 shows the entries needed in Solver, along with the selection of the usual two
options.
The layout of the spreadsheet is different than for all the prior linear programming
examples. Rather than a separate column for each activity and a separate row for each
constraint, the cost data and changing cells are laid out in a table format. This format
provides a more natural and compact way of displaying the constraints and results.
UnitsShipped (C11:E12) in the spreadsheet in Figure 10 shows the result of applying
Solver to obtain an optimal solution for the number of lathes to ship through each shipping
lane. TotalCost (H15) indicates that the total shipping cost for this shipping plan is
$20,500.

Since any transportation problem is a special type of linear programming problem, it

makes the standard assumption that fractional solutions are allowed. However, we actually
don’t want this assumption for this particular application since only integer numbers of
lathes can be shipped from a factory to a customer. Fortunately, even while making the
standard assumption (allowing fractional solutions), the numbers in the optimal solution
shown in UnitsShipped (C11:E12) only have integer values. This is no coincidence.
Because of the form of its model, almost any transportation problem (including this one) is
guaranteed in advance to have an optimal solution that has only integer values despite the
fact that fractional solutions also are allowed. In particular, as long as the data for the
problem includes only integer values for all the supplies and demands (which are the
outputs and order sizes in the Big M Company problem), any transportation problem with
feasible solutions is guaranteed to have an optimal solution with integer values for all its
decision variables. Therefore, it is not necessary to add constraints to the model that
require these variables to have only integer values.

To summarize, here is the algebraic form of the linear programming model that has been
formulated in the spreadsheet:
Minimize Cost = 700SF1-C1 + 900SF1-C2 + 800SF1-C3 + 800SF1-C1 + 900SF2-C2 + 700SF2-C3
subject to the following constraints:
1. Fixed-requirement constraints:
SF1-C1 + SF1-C2 + SF1-C3 = 12 (Factory 1)
SF2-C1 + SF2-C2 + SF2-C3 = 15 (Factory 2)
SF1-C1 + SF2-C1 = 10 (Customer 1)
SF1-C2 + SF2-C2 = 8 (Customer 2)
SF1-C3 + SF2-C3 = 9 (Customer 3)
2. Nonnegativity constraints:
SF1-C1  0, SF1-C2  0, SF1-C3  0, SF2-C1  0, SF2-C2  0, SF2-C3  0

TABLE 9 Some Data for the Big M Company Distribution-Network Problem

FIGURE 9 The distribution network for the Big M Company problem.
FIGURE 10 The spreadsheet model for the Big M Company problem, including the
formulas for the objective cell TotalCost (H15) and the other output cells TotalShippedOut
(F11:F12) and TotalToCustomer (C13:E13), as well as the specifications needed to set up
Solver. The changing cells UnitsShipped (C11:E12) show the optimal solution obtained by
Solver.

3. Example: Assignment Problem

We now turn to another special type of linear programming problem called assignment
problems. As the name suggests, this kind of problem involves making assignments.
Frequently, these are assignments of people to jobs. Thus, many applications of the
assignment problem involve aiding managers in matching up their personnel with tasks to
be performed. Other applications might instead involve assigning machines, vehicles, or
plants to tasks. Here is a typical example.

An Example: The Sellmore Company Problem

The marketing manager of the Sellmore Company will be holding the company’s annual
sales conference soon for sales regional managers and personnel. To assist in the
administration of the conference, he is hiring four temporary employees (Ann, Ian, Joan,
and Sean), where each will handle one of the following four tasks:
 1. Word processing of written presentations.
2. Computer graphics for both oral and written presentations.
3. Preparation of conference packets, including copying and organizing written
materials.
4. Handling of advance and on-site registrations for the conference.
He now needs to decide which person to assign to each task.

Although each temporary employee has at least the minimal background necessary to
perform any of the four tasks, they differ considerably in how efficiently they can handle
the different types of work. Table 10 shows how many hours each would need for each
task. The rightmost column gives the hourly wage based on the background of each
employee.

Formulation of a Spreadsheet Model

Figure 11 shows a spreadsheet model for this problem. Table 10 is entered at the top.
Combining these required times and wages gives the cost (cells D15:G18) for each
possible assignment of a temporary employee to a task, using equations shown at the
bottom of Figure 11. This cost table is just the way that any assignment problem is
displayed. The objective is to determine which assignments should be made to minimize
the sum of the associated costs.

The values of 1 in Supply (J24:J27) indicate that each person (assignee) listed in column C
must perform exactly one task. The values of 1 in Demand (D30:G30) indicate that each
task must be performed by exactly one person. These requirements then are specified in
the constraints given in Solver.

Each of the changing cells Assignment (D24:G27) is given a value of 1 when the
corresponding assignment is being made, and a value of 0 otherwise. Therefore, the Excel
equation for the objective cell, TotalCost = SUMPRODUCT(Cost, Assignment), gives the
total cost for the assignments being made. The Solver Parameters box specifies that the
goal is to minimize this objective cell.

The changing cells in Figure 11 show the optimal solution obtained after running Solver.
This solution is
Assign Ann to prepare conference packets.
Assign Ian to do the computer graphics.
Assign Joan to handle registrations.
Assign Sean to do the word processing.
The total cost given in cell J30 is $1,957.

Characteristics of Assignment Problems

Note that all the functional constraints of the Sellmore Co. problem (as shown in cells
H24:J27 and D28:G30 of Figure 11) are fixed-requirement constraints which require each
person to perform exactly one task and require each task to be performed by exactly one
person. Thus, like the Big M Company transportation problem, the Sellmore Co. is a fixed-
requirements problem. This is a characteristic of all pure assignment problems; however,
there are variants of assignment problems where this is not the case.

Like the changing cells Assignment (D24:G27) in Figure 11, the changing cells in the
spreadsheet model for any pure assignment problem gives a value of 1 when the
corresponding assignment is being made, and a value of 0 otherwise. Since the fixed-
requirement constraints require only each row or column of changing cells to add up to 1
(which could happen, e.g., if two of the changing cells in the same row or column had a
value of 0.5 and the rest 0), this would seem to necessitate adding the constraints that each
of the changing cells must be integer. After choosing the Solver option to make the
changing cells nonnegative, this then would force each of the changing cells to be 0 or 1.
However, it turned out to be unnecessary to add the constraints that require the changing
cells to have values of 0 or 1 in Figure 11 because Solver gave an optimal solution that had
only values of 0 or 1 anyway. In fact, a general characteristic of pure assignment problems
is that Solver always provides such an optimal solution without needing to add these
additional constraints.

Another interesting characteristic of any pure assignment problem is that it can be viewed
as a special type of pure transportation problem. In particular, every fixed-requirement
constraint in the corresponding transportation problem would require that either a row or
column of changing cells add up to 1. This would result in Solver giving an optimal
solution where every changing cell has a value of either 0 or 1, just as for the original
assignment problem.

TABLE 10 Data for the Sellmore Co. Problem

FIGURE 11 A spreadsheet formulation of the Sellmore Co. problem.
4. Model Formulation from a Broader Perspective
Formulating and analyzing a linear programming model provides information to help
managers make their decisions. That means the model must accurately reflect the
managerial view of the problem:
• The overall measure of performance must capture what management wants
accomplished.
• When management limits the amounts of resources that will be made available to
the activities under consideration, these limitations should be expressed as
resource constraints.
• When management establishes minimum acceptable levels for benefits to be gained
from the activities, these managerial goals should be incorporated into the model as
benefit constraints.
• If management has fixed requirements for certain quantities, then fixed-
requirement constraints are needed.

With the help of spreadsheets, some managers now are able to formulate and solve small
linear programming models themselves. However, larger linear programming models may
be formulated by management science teams, not managers. When this is done, the
management science team must thoroughly understand the managerial view of the
problem. This requires clear communication with management from the very beginning of
the study and maintaining effective communication as new issues requiring managerial
guidance are identified. Management needs to clearly convey its view of the problem and
the important issues involved. A manager cannot expect to obtain a helpful linear
programming study without making clear just what help is wanted.

As is necessary in any textbook, the examples in this chapter are far smaller, simpler, and
more clearly spelled out than is typical of real applications. Many real studies require
formulating complicated linear programming models involving hundreds or thousands of
decisions and constraints. In these cases, there usually are many ambiguities about just
what should be incorporated into the model. Strong managerial input and support are vital
to the success of a linear programming study for such complex problems.

When dealing with huge real problems, there is no such thing as “the” correct linear
programming model for the problem. The model continually evolves throughout the course
of the study. Early in the study, various techniques are used to test initial versions of the
model to identify the errors and omissions that inevitably occur when constructing such a
large model. This testing process is referred to as model validation.

Once the basic formulation has been validated, there are many reasonable variations of the
model that might be used. Which variation to use depends on such factors as the
assumptions about the problem that seem most reasonable, the estimates of the parameters
of the model that seem most reliable, and the degree of detail desired in the model.

In large linear programming studies, a good approach is to begin with a relatively simple
version of the model and then use the experience gained with this model to evolve toward
more elaborate models that more nearly reflect the complexity of the real problem. This
process of model enrichment continues only as long as the model remains reasonably
easy to solve. It must be curtailed when the study’s results are needed by management.
Managers often need to curb the natural instinct of management science teams to continue
adding “bells and whistles” to the model rather than winding up the study in a timely
fashion with a less elegant but adequate model.
When managers study the output of the current model, they often detect some undesirable
characteristics that point toward needed model enrichments. These enrichments frequently
take the form of new benefit constraints to satisfy some managerial goals not previously
articulated.

Even though many reasonable variations of the model could be used, an optimal solution
can be solved for only with respect to one specific version of the model at a time. This is
why what-if analysis is such an important part of a linear programming study. After
obtaining an optimal solution with respect to one specific model, management will have
many what-if questions:
• What if the estimates of the parameters in the model are incorrect?
• How do the conclusions change if different plausible assumptions are made about
the problem?
• What happens when certain managerial options are pursued that are not
incorporated into the current model?
The topic “What-If Analysis for Linear Programming” is devoted primarily to describing
how what-if analysis addresses these and related issues, as well as how managers use this
information.

Because managers instigate management science studies, they need to know enough about
linear programming models and their formulation to be able to recognize managerial
problems to which linear programming can be applied. Furthermore, since managerial
input is so important for linear programming studies, managers need to understand the
kinds of managerial concerns that can be incorporated into the model. Developing these
two skills have been the most important goals of this chapter.

Review questions:
1. A linear programming model needs to reflect accurately whose view of the problem?
2. What is meant by model validation?
3. What is meant by the process of model enrichment?
4. Why is what-if analysis an important part of a linear programming study?

5. Summary
Functional constraints with a  sign are called resource constraints, because they require
that the amount used of some resource must be less than or equal to the amount available
of that resource. The identifying feature of resource-allocation problems is that all their
functional constraints are resource constraints.

Functional constraints with a  sign are called benefit constraints, since their form is that
the level achieved for some benefit must be greater than or equal to the minimum
acceptable level for that benefit. Frequently, benefit constraints express goals prescribed
by management. If every functional constraint is a benefit constraint, then the problem is a
cost–benefit–trade-off problem.

Functional constraints with an = sign are called fixed-requirement constraints, because

they express the fixed requirement that, for some quantity, the amount provided must be
equal to the required amount. The identifying feature of fixed-requirements problems is
that their functional constraints are fixed-requirement constraints. One prominent type of
fixed-requirements problem is transportation problems, which typically involve finding a
shipping plan that minimizes the total cost of transporting a product from a number of
plants to a number of customers. Another prominent type is assignment problems, which
typically involves assigning people to tasks so as to minimize the total cost of performing
these tasks.

Linear programming problems that do not fit into any of these three categories are called
mixed problems.

In many real applications, management science teams formulate and analyze large linear
programming models to help guide managerial decision making. Such teams need strong
managerial input and support to help ensure that their work really meets management’s
needs.

References:
This notes is adapted from:
• F.S. Hillier and M.S. Hillier, Introduction to Management Science (5ed/6ed),
McGraw Hill, 2019. (Chapter 3)
• Personal notes; Internet.

Problem.
1 (Web Mercantile)
Web Mercantile sells many household products through an online catalog. The company
needs substantial warehouse space for storing its goods. Plans now are being made for
leasing warehouse storage space over the next five months. Just how much space will be
required in each of these months is known. However, since these space requirements are
quite different, it may be most economical to lease only the amount needed each month on
a month-by-month basis. On the other hand, the additional cost for leasing space for
additional month is much less than for the first month, so it may be less expensive to lease
the maximum amount needed for the entire five months. Another option is the intermediate
approach of changing the total amount of space leased (by adding a new lease and/or
having an old lease expire) at least once but not every month.

The space requirement and the leasing costs for the various leasing periods are as follows.

Required space Leasing period Cost per Sq.Ft. Leased

Month (Square Feet) (Months)
Month 1 30,000 Month 1 $65
Month 2 20,000 Month 2 100
Month 3 40,000 Month 3 135
Month 4 10,000 Month 4 160
Month 5 50,000 Month 5 190

The objective is to minimize the total leasing cost for meeting the space requirements.

a. Identify both the activities and the benefits being sought from these activities. (This
indicate why this is a "cost-benefit-trade-off problem".)

b. Identify verbally the decisions to be made, the constraints on these decisions, and the
overall measure of performance for the decisions.
c. Convert these verbal descriptions of the constraints and the measure of performance into
quantitative expressions in terms of the data and decisions.

d. Formulate the model in algebraic form.

e. Formulate a spreadsheet model for this problem. Identify the data cells, the changing
cells, the objective cell, and the other output cells. Also show the Excel equation for each
output cell expressed as a SUMPRODUCT function. Then use Solver to solve the model.
(Use Excel)

2. The Fagersta Steelworks currently is working two mines to obtain its iron ore. This
iron ore is shipped to either of two storage facilities. When needed, it then is shipped on to
the company’s steel plant. The diagram below depicts this distribution network, where M1
and M2 are the two mines, S1 and S2 are the two storage facilities, and P is the steel plant.
The diagram also shows the monthly amounts produced at the mines and needed at the
plant, as well as the shipping cost and the maximum amount that can be shipped per month
through each shipping lane.

Management now wants to determine the most economical plan for shipping the iron ore
from the mines through the distribution network to the steel plant.

a. Identify all the requirements that will need to be expressed in fixed-requirement

constraints.

b. Express this model in algebraic form.

c. Formulate and solve a linear programming model for this problem on a spreadsheet.
(Use Excel)

3. 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 following table.
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.

a. Describe how this problem fits into the format for an assignment problem.

b. Formulate and solve this problem on a spreadsheet. (Use Excel. Note, you may
formulate the algebraic model in draft paper first.)