SD-modeling tools vs. Spreadsheets

This forum contains all archives from the SD Mailing list (go to http://www.systemdynamics.org/forum/ for more information). This is here as a read-only resource, please post any SD related questions to the SD Discussion forum.
Locked
Andreas Fiechtner
Newbie
Posts: 1
Joined: Fri Mar 29, 2002 3:39 am

SD-modeling tools vs. Spreadsheets

Post by Andreas Fiechtner »

we have tried out both and I would say that spreadsheest are defenetly a good tool
to create small modells and try things out. When a model becomes more complex
(let say more then 100 variables) it is quite difficult to handle it with a spreadsheet.

Advantages of spreadsheets:

easy to handle (provided you know how to use a spreadsheet aniway)
you can show the effect of changing a variable immediately without of
running a new simulation as you would have to do with VenSim.

Advantages of tools like for example Vensim are:

you can visualize the model easily
you can easily check the model for bugs.
you can easily change delta t for the simulation.
you can use subscripts

Well, I hope this will answer at least part of your question.

With kind regards

Andreas Fiechtner

________________________________________________________________________

Andreas Fiechtner Tel.: (+49) 0 94 39 - 90 20 90
Rauberweiherhaus 19 Fax: (+49) 0 94 39 - 4 93
D-92442 Wackersdorf Mail: andreas.fiechtner@ok2.de
_______________________________________________________
George Richardson
Senior Member
Posts: 68
Joined: Fri Mar 29, 2002 3:39 am

SD-modeling tools vs. Spreadsheets

Post by George Richardson »

There is a world (view) of difference between modeling in spreadsheets and
system dynammics modeling. The essence of the difference is that
spreadsheets capture DISCRETE year-to-year (or time unit-to-time unit)
changes while system dynamics simulations attempt to capture a CONTINUOUS
view of change.

One can build an exact replica of a system dynamics model in a
spreadsheet, but then one would have to have (in the most likely format) a
column for each Time Step (DT) in the simulation. For a time step of,
say, .25 year, one would need four columns to move from one year to the
next. Usually, people dont do that, preferring to let the successive
columns be the successive years (or whatever the time unit). Spreadsheets
in which successive columns represent successive time units are capturing
difference equation models, not continuous time models.

It can make a big difference. In a spreadsheet model representing a
difference equation structure, one does not have to obey our presumption
of at least one stock in every feedback loop. A variable can be set equal
to some variable "in the last period," and the loops that result may not
have any stocks in them. In a system dynamics model, there is no "last
period" -- the perspective is of a continuous system, with time flowing
continuously. Forrester had quite a bit to say about this in Industrial
Dynamics.

In the haands of careful experts, I think either modeling perspective can
yield robust results. But the discrete approach that most spreadsheet
users would produce is frought with peril. In Feedback Thought in Social
Science and Systems Theory, there is a lot on discrete and continuous
views. If you can find the book, look up "continuity and discreteness,"
"continuous point of view," "discreteness," and "events." Its a sobering
and instructive set of stories...

...George

-----------------------------------------------------------------------
George P. Richardson
G.P.Richardson@Albany.edu
Rockefeller College of Public Affairs and Policy Phone: 518-442-5257
University at Albany - SUNY, Albany, NY 12222 Fax: 518-442-5298
http://cnsvax.albany.edu/~gr383/
-----------------------------------------------------------------------
John Sterman
Senior Member
Posts: 117
Joined: Fri Mar 29, 2002 3:39 am

SD-modeling tools vs. Spreadsheets

Post by John Sterman »

I fully endorse Georges comments on the difficulties of spreadsheets for
dynamic modeling.

But there are other pragmatic reasons not to do this as well. While it is
possible to represent a system dynamics model in excel, I strongly advise
against it.

Properly capturing feedbacks and stock-flow structure in a spreadsheet is
tedious, and it is easy for errors to creep in undetected. Further, since
the spreadsheet emphasizes the contents of a cell, it is hard to visualize
the relationships in and overall structure of the model. It is a tool
designed for detail, not dynamic complexity.

Consequently, models built in excel are hard to document, explain, and
modify. These properties make it difficult to conduct the modeling process
in an open, inquiring mode, a mode in which client feedback can lead to
revisions with a short cycle time. Therefore client understanding and
confidence fall, and chances of meaningful implementation plummet, even, as
george says, in the hands of careful experts.

If a spreadsheet interface is required, a much better option is to build
the model in e.g. vensim, powersim or ithink, then send the data to and
from the spreadsheet via DDE.

In short, dont use spreadsheets when dynamic models of feedback systems
are appropriate.

John Sterman

John Sterman
J. Spencer Standish Professor of Management
Director, MIT System Dynamics Group
MIT Sloan School of Management
E53-351
30 Wadsworth Street
Cambridge, MA 02142
617/253-1951 (voice); 617/258-7579 (fax), jsterman@mit.edu
http://web.mit.edu/jsterman/www
Locked