Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 10

THE UNIVERSITY OF AKRON

6500: 680 SUPPLY CHAIN LOGISTICS MANAGEMENT


MODULE 3 CASE ASSIGNMENT

FACILITY LOCATION FOR DINKO’S PRINTING SERVICES

Dinko’s Printing Services, Inc. is a small business located in Gotham


City. Dinko's offers its customers a full range of black-and-white, color,
and custom printing, copying and binding services. It also creates
custom mailings, posters, presentations, and print advertisements.

A typical customer transaction begins with a call to Dinko’s toll-free


number. If the job involves an electronic document created by the
customer, he or she sends the document to Dinko’s via the internet. If
the job involves physical documents, Dinko’s sends a bicycle messenger
to the customer’s location to pick up the job. After the job is finished,
another messenger is dispatched to the client with the completed work.

A competitive priority for Dinko’s is its speed of delivery. Most print


jobs are accepted with very short deadlines, and customers are sensitive to late deliveries. A significant
portion of a job’s completion time is the travel time to and from a customer’s location. Dinko’s operates
five facilities scattered across Gotham. When a customer dials the central toll-free number, the operator
routes the request to the facility nearest the customer.

Evelyn Abeles, Dinko’s owner, is considering expanding the company’s operations to nearby Metropolis.
She wants to make a gradual entry into the Metropolis market, and as a start she has identified ten
downtown office buildings as potential sources of demand for print jobs. Evelyn has assigned each a
weight between 1 and 10 based on her estimate of its demand. The buildings, their weights, and their
locations are listed in Table 1.

Table 1: Customer Locations in Metropolis


Building Raw
Number Building Name Weight Avenue Street
1 Marine Midland Building 8 39 12
2 Masius Benton & Bowles 9 26 42
3 Empire Building 5 22 41
4 The Gage Towers 8 4 11
5 Wang Plaza 1 37 25
6 Farmer’s Loan & Trust Bldg. 9 14 35
7 Clark Kent Tower 2 30 17
8 Federal Building 5 11 16
9 The Hancock Center 1 38 6
10 Oglivy & Mather Worldwide 4 23 7

The founders of Metropolis designed the city’s downtown as a grid. The downtown district consists of 50
numbered avenues running north-south, and 50 numbered streets running east-west. City blocks are
square, as shown in Figure 1.

Page 1 of 10
Figure 1: A Portion of Downtown Metropolis

Evelyn believes two facilities should be sufficient to satisfy the demand from the ten office buildings listed
in Table 1. Since she plans to lease the space required for the two facilities, she has already spent some
time looking at available commercial real estate in Metropolis. She has identified five locations that meet
her requirements for space, utilities, accessibility, and rent. These potential locations are listed in Table 2.

Table 2: Potential Facility Locations in Metropolis


Locatio
n Avenue Street
Number
1 20 35
2 15 17
3 30 47
4 22 25
5 5 18

Evelyn is uncertain about whether to establish both Dinko’s facilities immediately. Instead, she could
open a single facility and wait to see whether it is successful before opening the second. Therefore she
wants to address the following questions:
1) If she opens just a single facility, where should it be located? If the first Dinko’s is a success,
where should she place the second?
2) If she establishes two facilities simultaneously, where should they be located?

Our Excel model should be set up to solve both the one- and two-facility location problems.

Solution

The first step is to decide how to measure the distance between buildings. Since the city streets are
arranged as a grid, we can place imaginary coordinate axes along 1 st Street and 1st Avenue, with their
intersection as the origin. The position of each building can now be identified using its street and avenue

Page 2 of 10
numbers as coordinates. For example, the position of the Empire Building is (22,41), as shown in Figure
1. In the equations below, we will refer to the position of the ith building in Table 1 as (Xi, Yi).
Now consider the distance between two buildings numbered i and j. We might measure their separation
using the Euclidean distance formula:

 X i  X j  2  Yi  Y j  2 .

For example, the Euclidean distance between the two buildings shown in Figure 1 is:

 26  22  2   42  41 2  4.12 blocks.

(If we want to know the distance between these buildings in feet, we must multiply 4.12 by the length of a
block.) The Euclidean formula measures the straight-line distance between the two buildings, which
probably doesn’t reflect the actual route a messenger would take. If the messenger uses the city streets to
travel between buildings, we should use the formula for metropolitan distance:
Xi  X j  Yi  Y j .
The metropolitan distance between the two buildings shown in Figure 1 is:
26  22  42  41  5 blocks.
We can now formulate Evelyn’s location problem as a math program. When she wants to place a single
facility, the decision variables are the coordinates of its location: (XD1,YD1). The objective function
minimizes the average distance from the facility to the ten office buildings:
10
Minimize :  w  X
i 1
i i  X D1  Yi  YD1  ,

where wi is the normalized weight assigned to building i. (Normalizing the weights means rescaling the
raw weights to sum to 1.) The absolute value functions used in the objective function are nonlinear, so
this formulation is not a linear program and can be difficult to solve.

We also include a constraint to limit the allowable choice of location (i.e. the feasible region) to the five
possibilities listed in Table 2:

 X D1 ,YD1     20, 35 , 15, 17 ,  30, 47  ,  22, 25 ,  5, 18 .


This sort of constraint is difficult to express in a way that can be used with the Solver tool, so we will solve
Evelyn’s problem manually. [If we wished to use Solver, we could formulate the problem with five binary
decision variables (X1,…,X5), one for each of the potential locations. A decision variable with a value of
“1” indicates its corresponding location should contain a Dinko’s. The constraint X1 +…+ X5 = 1 ensures
that exactly one location is selected. Unfortunately Solver’s solution algorithm is not well suited for this
kind of nonlinear-integer-program. For a problem of this size, we are better off finding the solution
manually.]

When we wish to locate both facilities simultaneously, the decision variables are the coordinates of the
two locations: (XD1,YD1) and (XD2,YD2). The objective function minimizes the average distance between
each office building and the nearest Dinko’s location:
10
Minimize :  w  MIN  X
i 1
i i  X D1  Yi  YD1 ,  X i  X D 2  Yi  YD 2  ,
and the constraints limit the allowable choice of locations to the possibilities listed in Table 2:

 X D1 ,YD1  and  X D 2 , YD 2     20, 35 , 15, 17  ,  30, 47  ,  22, 25 ,  5, 18 .

Page 3 of 10
Developing the Dinko’s Facility Location Model

Open the file titled “Shell for Dinko’s.xls,” which is an incomplete version of the model shown in the
figures below. Go to the Model worksheet. The cells shaded green in Figure 2 are user inputs. Enter
these numbers directly; no equations are needed.

Figure 2: Inputs Section of the Excel Model

Figure 3 shows the work area used for locating a single facility. The decision variables, (XD1,YD1), are
contained in cells C19:D19 and are shaded yellow. The objective function is the weighted average distance
and is contained in cell E34. Notice that this is the sum of the weighted distances, not their average.

Figure 3: Work Area for Locating a Single Facility

Since we are not using the Solver tool for this problem, we must find the optimal facility location
manually. Try entering the five locations from Table 2, one at a time, into cells C19:D19. You should find
that the fourth location, the corner of 22 nd Avenue and 25th Street, minimizes the weighted average
distance: 22.788 blocks.
Next use Excel’s Chart Wizard tool to create a street map of the city showing the locations of the ten office

Page 4 of 10
buildings, the five potential Dinko’s locations, and the optimal location contained in cells C19:D19. On
the worksheet labeled One Facility Map, enter the data shown in Figure 4. You should use references to
the Model worksheet so that the map will update when you change the locations of the office buildings or
the Dinko’s facility.

Figure 4: The One Facility Map Worksheet

Notice that the avenues for all sixteen locations are in the same column, but the streets for the office
buildings, the potential locations, and the optimal Dinko’s location are in different columns. This setup
separates the office buildings, potential locations, and optimal location into different data series. The
result is that they will appear on the map with different colors and symbols.

Now select cells B2:E18 (this includes one of the label rows), and in the Chart Wizard window select the
XY (Scatter) chart with no lines, as shown in Figure 5.

Figure 5: The Chart Wizard

Continue through the Chart Wizard, and place the chart as an object in the One Facility Map worksheet.

Page 5 of 10
After some additional formatting (including showing both major and minor gridlines for each axis), you
can create a street map like the one shown in Figure 6.

Figure 6: A Street Map of Metropolis Showing the Optimal Location of the Single Facility

Returning again to the Model worksheet, Figure 7 shows the work area used for locating two facilities
simultaneously. The decision variables for this problem, (XD1,YD1) and (XD2,YD2), are contained in cells
C40:D41, which are shaded yellow. The objective function is the weighted average distance to the nearest
facility and is contained in cell G56. Again, the objective function is the sum of cells G45:G54, not their
average.

To find the optimal locations for the two facilities, enter pairs of locations from Table 2 into cells C40:D41
and look at the objective function. For example, the pair (20,35) and (5,18) results in a weighted average
distance of 16.827 blocks. There are ten distinct pairs of locations. You should find that the optimal pair
is (20, 35) and (15, 17), with a weighted average distance of 14.750 blocks.

Page 6 of 10
Figure 7: Work Area Used for Locating Two Facilities Simultaneously

On the Two Facilities Map worksheet, create another street map of the city, this time showing the office
buildings, the potential locations, and the optimal locations of the two Dinko’s facilities contained in cells
C40:D41. Refer to Figure 8.

Figure 8: A Street Map of Metropolis Showing the Locations of Both Facilities

Page 7 of 10
Validating the Dinko’s Facility Location Model

Save your workbook. Then change the avenues, streets, and raw weights for the ten office buildings as
shown in Figure 9.

Figure 9: Validation of the Model

Change the decision variables for the two problems as shown in Figure 10, and check the resulting work
areas to see that they match the figure.

Figure 10: Validation of the Model

Page 8 of 10
Using the Dinko’s Facility Location Model

1. Suppose Evelyn chooses not to open both facilities immediately. Instead she will open a single
facility and wait to see whether it is successful before opening the second.
a) If you followed the directions above, you found that the best location for a single facility is the
corner of 22nd Avenue and 25th Street. If this location is a success, where should Evelyn place the
second facility? (Use your work area for two facilities. Enter the location (22,25) into cells
C40:D40, and evaluate potential locations for the second Dinko’s by changing cells C41:D41.)
b) The instructions above also directed you to find the best locations for two facilities when they are
positioned simultaneously. These locations were the corner of 20 th Avenue and 35th Street, and
the corner of 15th Avenue and 17th Street. Which of these facilities should be opened first?
Explain.
c) How and why does the pair of locations identified in part (a) differ from the pair of locations in
part (b)? If Evelyn thinks it is very likely that she will open both facilities, which pair of locations
should she choose?

2. a) We used the metropolitan distance metric rather than the Euclidean distance. Is the Euclidean
distance between two buildings longer or shorter than the metropolitan distance?
b) When placing a single facility, our objective function minimized the weighted
average distance between the facility and the ten office buildings. What is the optimal location if
the objective function minimizes the maximum distance between the facility and the ten office
buildings? What is the value of the objective function (i.e. the maximum distance) at this optimal
location?
c) Under what circumstances does it make sense to choose an objective function
that minimizes the maximum distance between a facility and its customers? Give an example.

3. Suppose Evelyn is not restricted to the locations listed in Table 2; instead she can place her
facilities anywhere in the city. Try using Solver to identify the best location for a single facility. Use
your work area for one facility, and set up the Solver Parameters window as shown below.

This is not a linear program, so choose “GRG Nonlinear” as the Solving Method in the Solver Options
window.
a) Enter the initial location (1,1) into cells C19:D19, and run Solver. Then enter the initial location
(40,40), and re-run Solver. Solver will most likely return different solutions. Why does this
happen?
b) Since we are not limited to the five locations listed in Table 2, you might argue that we should
include the constraints: 1  X D1  50, 1  YD1  50, to keep the solution in the downtown
district. Why are these constraints unnecessary?

Page 9 of 10
Page 10 of 10

You might also like