Comparison of Supply Chain Modeling Techniques

[SangDon Lee]

Im not sure whether you are interested in specific comparison of
supply chain modeling techniques or general comparison among OR,
Discrete-event simulation and SD. If you are interested in general
comparison, Dr. John Stermans paper would be good.

John D. Sterman (1998), A skeptics guide to computer models in
Gerald O. Barney, W. Brian Kreutzer and Martha J. Garrett (eds),
Managing a Nation: the microcomputer software catalog, Boulder:
Westview Press, pp.209-29. Also appeared in Modeling for management by
G. P. Richardson, Dartmouth Publishing Company :This paper compares
the strength and weakness of optimization, simulation and econometrics
approach.


Here is some of MY thoughts about appicability of SD OR and
discrete-event simulation.
+---------++-----------------+------------------------| |
||static problem | dynamic problem |
+---------++-----------------+------------------------+
|aggregate|| | SD |
+---------++-----------------+------------------------+
|detail ||Operations | Discrete-event |
| ||Spread Sheet | Simulation |
+---------++-----------------+------------------------+



Comparison of SD & Discrete-Event Simulation(DES)

** Objectives(problems to deal with) **
SD: Guiding policy/strategy analysis
DES: Daily/operational decision analysis
(e.g., No. of machines, buffer size and location)

** Focus **
SD: Information feedback structures
Understanding of interdependencies and time delays
(e.g., interaction between production orders and supply
leadtime)
DES: Uncertainty
andomness of events
(e.g., material or customer arrivals)

** Theory **
SD: Information feedback and control theory
DES: Probability and Queueing theory

** Model **
SD: inflow and outflow model with feedbacks
nonlinear ordinary difference equations
DES: input and output model with usually no emphasis on
feedback.

** Solution **
SD: Aggregate & generic
(e.g., SD says we have to be very careful when we update
supplier leadtime because there is a vicious circle
involved in supplier leadtime and production orders. This
creates some problems to customers because they have to go
one more step to extract specific decisions after
understanding the interactions, which is quite difficult,
I believe, considering time pressure that managers are under.
I think many system dynamist may have this kind of responses
from customers: O.K., SD is great, but just tell me what
to do (*@!#$%), Do I order now or not, and how much ?"
Also, SD is a continuous-type simulation, but production
orders do occur in discrete way, at least thats
how most customers think.)
DES: Detail and specific
(e.g., discrete-event simulation will show the input buffer
size of stamping machine has to be increased because of the
increasing supplier leadtime in order to prevent the stamping
machine from starving.)

** Sensitivity of a model **
SD: The outcome of a model is generally insensitive to changes
in parameter values because of the effect of multiple
feedbacks.
DES: The outcome is more sensitive to the selection of
distribution of random variables/parameters (e.g., use of
normal or exponential distribution for material processing
time may provide quite different answers)

** Time span **
SD: long-term (weeks, months or years) This is I believe why SD
has not been widely used in companies since most people are busy
for fire-fighting.
DES: short-term (minutes, hours or days)

** Quantitative /qualitative factors **
SD: usually SD involves qualitative factors such as social and
psychological factors.
(e.g., the impact of schedule pressure on quality and
productivity on project management)
DES: usually DES doesnt involve qualitative factors and quite
difficult to consider.

** How to improve a system **
SD: usually modify information feedback structure
DES usually change parameter values


I hope this would give some clue to compare supply chain modeling
techniques.

Sangdon Lee
sangdonlee@yahoo.com



==
Be happy

Sangdon Lee

_________________________________________________________
DO YOU YAHOO!?
Get your free @yahoo.com address at http://mail.yahoo.com

[Bruce Campbell]

> Sterman, John D. 1988. "A Skeptics Guide to Computer Models." pp. 133-169.
> In: Lindsey Grant, ed. Foresight and National Decisions. Lanham, MD:
> University Press of America.

Jerry

The paper above is also available at John Stermans web site, which is
where I got it from.
http://web.mit.edu/jsterman/www/

Another more general paper is David F. Andersen, 1980, "How Differences
in Analytic Paradigms Can Lead to Differences in Policy Conclusions" in
Elements of the System Dynamics Method, J. Randers, ed., MIT Press (now
available thru Productivity Press), pp 61-75. This doesnt look at
supply chains in particular but how different methodologies can arrive
at different conclusions
ecommendations to the same problem.

Bruce Campbell

--
Bruce Campbell
Joint Research Centre for Advanced Systems Engineering
Macquarie University 2109
Australia

E-mail: Bruce.Campbell@mq.edu.au
Ph: +61 2 9850 9107
Fax: +61 2 9850 9102

[jerry_zygmuntowicz@om.cv.hp.com]

We have been involved in using SD for quantitative and predictive
applications on supply chain, marketing, and business strategy issues.
For supply chain in particular, there are multiple accepted
methodologies in use, and I have developed some sense, but not a clear
criteria for when each is best applied. SD is not a
one-model-fits-all solution, and it is as important to recognize when
to choose another approach as it is to identify a good problem for SD
analysis.

Within Hewlett Packard, we have internal consultants and pockets of
expertise around the company, some using advanced spreadsheet models
to configure supply chain networks, others using large LP models to
optimize shipments, still others employing large unix-based
simulations to tune inventory and capacity strategies, and finally
Greg Jacobus and others using SD models for a range of distribution
and manufacturing issues. As a business consultant within a large,
high-growth division, it is difficult to advise management teams on
what is the best tool to pull from the kit.

Does anyone know of some references that survey analytic approaches
such as linear programming, discrete event simulation, or other OR
approaches and SD and compares the strengths and weaknesses, or
appropriate applications, of each?

Jerry Zygmuntowicz
Senior Business Analyst
Hewlett Packard
1000 NE Circle Blvd.
Corvallis, OR 97330
541 715-1759
jerryz@cv.hp.com

[jerry_zygmuntowicz@om.cv.hp.com]

We have been involved in using SD for quantitative and predictive
applications on supply chain, marketing, and business strategy issues.
For supply chain in particular, there are multiple accepted
methodologies in use, and I have developed some sense, but not a clear
criteria for when each is best applied. SD is not a
one-model-fits-all solution, and it is as important to recognize when
to choose another approach as it is to identify a good problem for SD
analysis.

Within Hewlett Packard, we have internal consultants and pockets of
expertise around the company, some using advanced spreadsheet models
to configure supply chain networks, others using large LP models to
optimize shipments, still others employing large unix-based
simulations to tune inventory and capacity strategies, and finally
Greg Jacobus and others using SD models for a range of distribution
and manufacturing issues. As a business consultant within a large,
high-growth division, it is difficult to advise management teams on
what is the best tool to pull from the kit.

Does anyone know of some references that survey analytic approaches
such as linear programming, discrete event simulation, or other OR
approaches and SD and compares the strengths and weaknesses, or
appropriate applications, of each?

Jerry Zygmuntowicz
Senior Business Analyst
Hewlett Packard
1000 NE Circle Blvd.
Corvallis, OR 97330
541 715-1759
jerryz@cv.hp.com

["Michael J. Radzicki, Ph.D."]

Jerry:

> Does anyone know of some references that survey analytic approaches
> such as linear programming, discrete event simulation, or other OR
> approaches and SD and compares the strengths and weaknesses, or
> appropriate applications, of each?

I recommend reading the following:

Sterman, John D. 1988. "A Skeptics Guide to Computer Models." pp. 133-169.
In: Lindsey Grant, ed. Foresight and National Decisions. Lanham, MD:
University Press of America.

Mike Radzicki
Worcester Polytechnic Institute


Michael J. Radzicki, Ph.D.

Associate Professor of Economics
Department of Social Science & Policy Studies
Worcester Polytechnic Institute
Worcester, Massachusetts 01609-2280
United States of America
(508) 831-5767 (Voice)
(508) 831-5896 (Fax)
mjradz@wpi.edu (Email)

and

President
Sustainable Solutions, Inc.
210 Park Avenue
Box 225
Worcester, Massachusetts 01609
United States of America
sustsol@sunspot.tiac.net (Email)
http://www.tiac.net/users/sustsol (Web)