Request for system boundary setting guidance

Roy Greenhalgh
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Request for system boundary setting guidance

Post by Roy Greenhalgh »

I am currently struggling with how to create boundaries around
processes/systems. I have read a couple of texts about Computer
Simulation, and find “Introduction to Computer Simulation” by Nancy
Roberts, David Anderson, Ralph Deal et al very useful in this area.

Could anyone recommend other texts or sources to help me?


Roy Greenhalgh
From: Roy Greenhalgh <>

>From Sun Nov 26 14:47:18 2000 -0500
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id OAA26184
for <>; Sun, 26 Nov 2000 14:46:07 -0500 (EST)
Received: (from daemon@localhost)
by (8.9.3/8.9.3) id OAA16336;
Sun, 26 Nov 2000 14:46:06 -0500 (EST)
Date: Sun, 26 Nov 2000 14:46:06 -0500 (EST)
Message-Id: <>
Subject: BOUNCE Approval required:

>From Sun Nov 26 14:46:04 2000
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id OAA15729
for <>; Sun, 26 Nov 2000 14:41:18 -0500 (EST)
Received: from
by (mail_out_v28.33.) id 1.3e.40812fc (4339)
for <>; Sun, 26 Nov 2000 14:41:10 -0500 (EST)
Message-ID: <>
Date: Sun, 26 Nov 2000 14:41:09 EST
Subject: REPLY SD & Electronic Commerce (SD2997)
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
X-Mailer: Windows AOL sub 124


Several colleagues and I developed an SD model-based simulator to help
newspapers evaluate different strategies for making the transition from their
traditional print business to online delivery of news and e-commerce. An
elaborate set of interfaces and "canned" runs enable users without a
model-building background to use this strategy simulator to look at a variety
of strategies for a hypothetical newspaper/media company as well as
developing their own variants. The simulator can be downloaded from the
website While not totally about e-commerce, this
application has a significant e-business component. Hope this is helpful.

Gary Hirsch

Bruce Campbell
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Request for system boundary setting guidance

Post by Bruce Campbell »

There are a number of classic texts in this area. However, the book of
choice at the moment seems to be:
Sterman, John, 2000, Business Dynamics: Systems Thinking and Modeling
for a Complex World. McGraw Hill.

I havent seen the book (the exchange rate in Australia makes it
horrendously expensive) but it has had some good reports on the list.

Another source of information is the SD "Road Maps" series from MIT that
are available on the web (
This is a self learning tutorial series. It is still not complete, but
if you work your way through what is there, you will be pretty
knowledgable. They also give details of other text books.

To answer your specific question, the rule of thumb is: Model a problem,
not a system. Then, anything that affects the problem behaviour is
inside the boundary, and anything else should be ignored.

The manuals that come with iThink caution against modelling a system.
Believe me, once you have been involved in modelling a system rather
than a problem you will never wish to repeat the experience. Two issues
immediately arise: where is the boundary and; many investigators want to
model things to infinite detail. Either is guaranteed to lead to the
failure of the modelling effort. If you model a problem the boundary is
more obvious, and you have an argument to allow aggregation of similarly
behaving items. The latter allows the construction of simpler models
that avoid the detail that makes validation and understanding almost

As Jay Forrester has repeatedly pointed out on this list, most insights
into problem behaviour come from very simple models - usually no more
than 50 objects, and often less than 20. These models are then often
expanded to include many more objects to make them acceptable to
management and for policy decision making. However, the underlying
behaviour of these expanded models is the same as the simple model. All
the insights I have gained as a result of SD modelling has been from
extremely simple models. The reaction to the models is normally "Well,
thats pretty obvious." So obvious, in fact, that nobody has considered
that explanation before!

Hope this helps,

Bruce Campbell

Bruce Campbell
Joint Research Centre for Advanced Systems Engineering
Division of Information and Communication Sciences
Macquarie University 2109

Ph: +61 2 9850 9107
Fax: +61 2 9850 9102

>From Mon Nov 27 22:43:43 2000 -0500
Status: R
Return-Path: <system-dynamics-approval@WORLD.STD.COM>
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id WAA15624
for <vensim@WORLD.STD.COM>; Mon, 27 Nov 2000 22:42:54 -0500 (EST)
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id WAA24026
for <>; Mon, 27 Nov 2000 22:42:52 -0500 (EST)
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id WAA11675
for <>; Mon, 27 Nov 2000 22:42:07 -0500 (EST)
Received: (from daemon@localhost)
by (8.9.3/8.9.3) id WAA23787;
Mon, 27 Nov 2000 22:42:02 -0500 (EST)
Date: Mon, 27 Nov 2000 22:42:02 -0500 (EST)
Message-Id: <>
Subject: BOUNCE Approval required:
MIME-Version: 1.0
Content-Type: text/plain; charset=unknown-8bit
Content-Transfer-Encoding: 8bit

>From Mon Nov 27 22:41:58 2000
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id WAA21538
for <>; Mon, 27 Nov 2000 22:35:27 -0500 (EST)
Received: from ( [])
by (8.9.3/8.9.3) with ESMTP id VAA26168
for <>; Mon, 27 Nov 2000 21:35:24 -0600 (CST)
Message-ID: <>
Date: Mon, 27 Nov 2000 22:36:50 -0600
From: Alexander Leus <>
X-Mailer: Mozilla 4.5 [en]C-{C-UDP; TDSNET} (Win98; I)
X-Accept-Language: en
MIME-Version: 1.0
Subject: REPLY Request for system boundary setting guidance (SD2999)
References: <>
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Hi Roy,
Perhaps this idea will help you. One way of defining business process
boundaries is to get a good idea of the work flow from the customer
request to the service etc. that meets that request. You would do this
on a high level. You would use a cross functional team to define the
boundaries of the analysis.

Look at it if you were doing a flow chart of the work flow. Step 1 is
the starting point, step 2 is is end point. Once the team defines these
two points you have a better chance of developing a fruitful
process/system analysis. If you are analyzing engineering systems the
typical boundaries are at the interfaces of the key systems. This
occurs because of the way the subsystems are linked together.

If you want to discuss this more detail please email me.

Alex Leus

"Jim Hines"
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Request for system boundary setting guidance

Post by "Jim Hines" »

Roy Greenhalgh asks for adivice on setting boundaries for SD models.

It depends on what modeling approach you happen to be using. When using an
approach that involves an early phase of causal loop diagram, setting the
boundary is a no-brainer. When using an approach that goes directly to
stock and flow and equations, deciding on a boundary can be a bit of a

A very rough list of steps in the loop-first approach is

1) List variables
2) Draw reference modes for some of those variables
3) Come up with a set of loops that you think can generate the reference
4) Create a computer simulation model of (some) of those loops
5) etc.

Items 1 and 3 (usually) should be comprehensive -- theyre cheap so they
might as well be comprehensive. This means that after item 3 you will have
all of the loops that you think can generate the reference modes. There is
no boundary to the **loops** except your ability to generate theory.

What of the computer simulation model? The process of creating a computer
simulation model involves modeling one or two of the loops; then adding a
couple more loops; etc. The boundary of the computer model will always be
determined by the loops you are modeling. The boundary is a no brainer.
Figuring out how to model the loops in an operational way is a big-brainer,
but the boundary issue per se is trivial.

An alternative approach to modeling moves directly to stock and flow
diagramming and equation writing, without a significant looping stage. This
approach may have some advantages, but ease of setting a boundary is not one
of them. Most advice in texts tends to be a bit abstract and blue-sky. I
dont know where you will find concrete, specific, example-rich suggestions
for how to set a boundary when taking this stock-first approach. But one
place not to look is in this email.

Jim Hines

"Bob Cavana"
Junior Member
Posts: 14
Joined: Fri Mar 29, 2002 3:39 am

Request for system boundary setting guidance

Post by "Bob Cavana" »

hi Roy

you may find our recent systems thinking/system dynamics book
quite useful here. we outline a 5 phase systems thinking and
modelling process including problem structuring, causal loop
modelling, dynamic modelling, scenario planning & modelling,
and implementation and organisational learning. each phase
contains a number of steps which are outlined and illustrated in
the book. The book also includes a CD-rom with the models
and learning labs.

the full reference for the book is:

Systems Thinking and Modelling: Understanding Change and
Complexity by Kambiz E.Maani & Robert Y. Cavana, 2000,
Prentice Hall, New Zealand. (produced by Pearson Education
NZ Ltd)

ISBN 1 877258 00 8

It is available from:

Pegasus Communications

(Pegasus Communications also supplies a full range of systems
thinking/system dynamics books)

and Pearson Education outlets: go to
<> then click on Contact Us (purple box on left hand side).
Then click on Offices Worldwide and that page will give you a list
of all the contact details for Pearson around the world.

i hope you find our book useful

Bob Cavana
From: "Bob Cavana" <>
Dr Bob Cavana
Senior Lecturer in Decision Sciences
School of Business and Public Management, Victoria University of Wellington,
PO Box 600, Wellington, NEW ZEALAND
Tel: +64 4 4635137; Fax: +64 4 4635253; E-mail: