IN THE CLASSROOM

■ RICK HESSE, Feature Editor, Pepperdine University

Transportation Model Reports

for Excel
Rick Hesse, Feature Editor

T
he standard transportation problem optimal solution, using the Solver included
formulated as a spreadsheet model with Excel from Frontline Systems is also
to be optimized by the Solver add-in shown.
has many advantages in terms of setup and A surprise is that the extra production
solution for teaching. However, I have al- is stored in New York, even though the
ways been frustrated trying to get some cost ($0.80) is over twice as much as keep-
form of report for the standard Transporta- ing the extra production in Denver ($0.30)
tion model that would be informative to a or Atlanta ($0.20). But because the cost of
manager instead of a mathematician. The shipping out of New York to all the other
standard assignment model is a very simple locations is so high, it is better to keep the
model to have the report written automati- extra production in New York than the
cally because there is only one assignment other two plants.
per row and the value of the assignments The key cell formulas and range names
are 0s and 1s. But the transportation model are given below.
will have more than one shipment for one Cell Formulas:
or more rows and the values aren’t 0 and 1. F10: =SUM(B10:E10) copy to F11:F12
B13: =SUM(B10:B12) copy to C13:E13
Excel Transportation Model F13: =SUMPRODUCT(COST,SHIP)
The data for the standard transportation Range Names:
model for three plants (Denver, Atlanta and FROM: A4:A6; TO: B3:E3
New York) shipping to three warehouses COST: B4:E6; SHIP: B10:E12
(Tucson, Miami and San Diego) as well as Figure 2 gives the Solver setup show-
the inventory (Storage) is given in rows 3- ing that all seven constraints are equalities,
Rick Hesse
is professor of quantitative 7 in Figure 1. It is a balanced problem (Sup- with 12 variables and the total cost to be
methods at Pepperdine ply = Demand) and a total of 300 units must minimized.
University in the Graziadia be left at one or more of the plants. The
Graduate School of Busi-
ness. He received his BS,
MS, and DSc at Washing-
ton University School of
Engineering in applied
math and computer science. Dr. Hesse is the
author of Managerial Spreadsheet Model-
ing & Analysis and Applied Management
Science: A Quick & Dirty Approach (with
Gene Woolsey), articles in numerous journals,
and software for personal computers. Rick was
the first professor to be awarded the Outstand-
ing Civilian Service Medal by the Department
of the Army at West Point in 1982, and was the
winner of the Decision Sciences Institute’s Inno-
vative Instructional Award in 1981.

Dr. Rick Hesse

Graziadia Graduate School of Business
Pepperdine University
Malibu, CA 90265
[email protected]
Figure 1: Transportation spreadsheet model.

Decision Line, March 2001 7

Transportation Report
For those of us who are used to matrices, it
is a fairly simple matter to read off the so-
lution—but it is not very user-friendly.
What is more desirable is a report like the
one shown in Figure 3. Such a report can
certainly be created “manually”—that is,
each entry needs to either be typed in or
just “point and click” to the labels and val-
ues. The last column is a simple formula
multiplying the number of cases times the
cost per case. The Total formulas are also
simple sums.
But the challenge has always been to Figure 2: Solver setup for transportation model.
find a way to at least set up a semi-auto-
matic report. To do this, I have used the
INDEX and MATCH functions in Excel in
conjunction with the defined range names.
C18: =INDEX(SHIP,MATCH(A18,
FROM,0),MATCH(B18,TO,0)) copy to
C19:C23
D18: =INDEX(COST,MATCH(A18,
FROM,0),MATCH(B18,TO,0)) copy to
D19:D23
E18: =C18*D18 copy to E19:E23
The INDEX(range,row,column) func-
tion returns the value in range found in the
specified row and column. Thus
INDEX(SHIP,1,1) returns the value 600. To
determine which row and column is de- Figure 3: Transportation model report.
sired, the MATCH(value,range,0) function
searches the specified range for an exact
(the 0 is the code for that) match of the
specified value. In this case, the value is the
name of the plant or wholesaler. The use of
the 0 code in the MATCH function allows
the list in the range not to have be ordered
alphabetically. By using these two functions,
then for any size model the report can be
written fairly quickly and painlessly. All that
is needed is to “point and click” to the
names for the supply and demand to put
into columns A and B of the report. Inci-
dentally, to show what range names have
been defined in your current workbook,
click on the arrow on the box with the cell
reference (upper left of the screen) and the
drop-down list will show all the range
names. Highlight any one of them and Ex-
cel will automatically highlight that range
(taking you to another worksheet if neces-
sary). Figure 4: Edit function in Excel.

8 Decision Line, March 2001

 One of the problems with using range
names is that the formulas will not rede-
fine themselves for another model
worksheet in the same workbook with a
transportation model of different size
(number of rows and/or columns). And if
the ranges are redefined for one worksheet,
they are redeifined for all the other
worksheets in that workbook.
The cell formulas can be written using
just the cell references (shown below):
C18: =INDEX($B$10:$E$12,
MATCH(A18,$A$4:$A$6,0),
MATCH(B18,$B$3:$E$3,0))
D18: =INDEX($B$4:$E$6,
MATCH(A18,$A$4:$A$6,0),
MATCH(B18,$B$3:$E$3,0)) Figure 5: Column graph with data table.
Now when these formulas are copied
over to another worksheet for a model with
a different number of plants or warehouses,
it is a simple matter to use the edit function
in Excel for cell C18 to resize the areas for
the INDEX and MATCH functions, as
shown in Figure 4. Just click on cell C18,
then click inside the edit bar at the top of
the screen and magically all the cell refer-
ences in the formula are outlined.
Although you cannot see the colors in
this article, Excel color codes each cell ref-
erence, so that the shipping range (B10:E12)
is blue and the outline around that range is
blue. Instead of retyping the cell reference,
you can grab the lower right corner of the Figure 6: Black and white graph for larger transportation problem.
outline with the mouse and pull it up, down,
left or right and resize the area that you
want and then press <Enter>. If the rect-
angle is not in the proper upper left-hand
corner, you can grab any side of the rect-
angle with the mouse and place the rect-
angle where desired and then resize it. This
is an extremely useful way of editing for-
mulas and a great time-saving device for
spreadsheet users. In similar fashion we can
resize the cell reference area in B3:E3 and
the reference to A4:A6 needs no changing.
The same process is repeated for the cost
per case in cell D18. Once those two for-
mulas are “resized” we can copy them
down as needed for any new model. Armed
with just one example template, students
can quickly set up this semi-automatic re-
port for any similar transportation model.
Figure 7: 3D graph of transportation model results.
See Classroom, page 12

Decision Line, March 2001 9

 To avoid misunderstandings, students • Give first authorship to the person who Should my publication strategy in
should discuss co-authorship arrangements needs a publication the most. This crite- the years following my studies
with their advisors well in advance of the rion introduces a selfish motive and a logi- emphasize quality or quantity?
completion of the dissertation. A well-cho- cal flaw. We could all argue that we need
Quality.
sen advisor should have little difficulty re- first authorship because it does more for
lating to the student’s need to clarify us to have our names first than to have a You can fulfill all of your professional ambi-
ownership and responsibility for manu- co-author’s name first. First authorship tions with a series of premier journal articles,
scripts emanating from the dissertation. should not be a charitable contribution by and your colleagues will recognize you for
any author to another, so find another it. A portfolio filled with large numbers of
In general, how should the order of criterion that makes more sense. articles in obscure journals is valued less than
authorship in co-authored work be a portfolio with three, four or five outstand-
determined? What should my research priority be ing articles in recognized top-tier journals.
Put the project initiator and leader first. in the first year after getting my Clearly, not everything you write will end
Ph.D.? up in premier journals. But premier journals
This criterion acknowledges the importance should be your main target.
You should prepare two manuscripts from
of vision and direction on a project. On pa-
your dissertation for submission to top
pers initiated by students (e.g., course pa- I am sure doctoral students have many more
journals. At least one of these should be solo- questions than the eight that I have posed
pers or dissertations), I believe that first
authorship should go to the student. How-
authored. on their behalf. My answers to these ques-
ever, there are several additional criteria to The importance of “mining” your disser- tions should not be taken as the only correct
consider: tation cannot be overemphasized. Al- answers, and I do not assume that all doc-
• Alternate the order of authorship on mul- though you may be tired of the dissertation toral students are the same and need the
tiple papers. This criterion works for a after spending two to three years with it, same advice. One’s individual advisor may
team that will share work equally across you must shape it into the form of journal be able to provide deeper insights into a
multiple articles. articles soon after your defense is com- particular student’s situation. Nonetheless,
• Place the authors in alphabetical or ran- pleted. The dissertation is probably the having pledged not to waffle on my advice,
dom order. These criteria usually are se- most carefully supervised work you will the answers given above are direct and re-
lected when contributions are equal and ever do, and it would be a shame not to flect the collective concerns of many students
where the authors cannot determine other take advantage of the extensive feedback in many different academic settings. ■
criteria to apply. you have received.
• Put senior faculty first, followed by junior Many students seek new projects with
faculty and students. This criterion be- new colleagues when they take their first
stows first authorship on people who en- academic positions. I recommend against
Julie E. Kendall
joy a higher social position, independent making binding commitments to new
School of Business-Camden
of their contributions to the work. I per- work before the old work is complete. You Rutgers University
sonally think that first authorship should can discuss possible future work with your Camden, NJ 08102
be earned, not bestowed, so I do not fa- new faculty colleagues, but you need most (609) 225-6585
vor this criterion. of your research time in your first year in a fax: (609) 424-6157
new position to convert your dissertation [email protected]
into published articles. http://www.thekendalls.org

Classroom, from page 9

Graphical Report longer, or there are too many rows or col- A final option is to choose the 3D col-
The shipping pattern and labels found in umns (more than 5). When the model has umn graph (option #7) for problems with
A9:E12 in Figure 1 makes a good candidate a few rows and many columns, then the a moderate number of rows and columns,
column graph without the data table is a as shown in Figure 7 for a third Transpor-
for a graphical report, and there are sev-
good option. At that point, the colors may tation model example. The only problem
eral good options. The first works well
be a bit confusing, and so if black is chosen with this option is that some columns can
when there are only a few rows and col-
umns and includes a table of the shipments along with some patterns, it is not a bad get hidden behind taller ones in front. The
and is shown in Figure 5. This is a color choice. Figure 6 shows such a graph for a 100 units of Lost Sales at Tucson are hidden
column graph (option #4) with a data table. problem with 3 Plants by 6 Warehouses. in the back corner unless the series are re-
This could be accompanied by the semi- versed under the graph options. ■
The inclusion of a Data Table with the
automatic report instead of a data table.
graph may not be as attractive when the
names of the supply and demand cities are

12 Decision Line, March 2001