communicating SD models

[bonnell@Kodak.COM]

Date: 03/11/95 15:47:56
To: NYSERNET--KODAKR INTERNET

>From: William Bonnell
Mgt.Services Div., Eastman Kodak Company
Bldg. 56-4, KP, Rochester, NY 14652-4530
Subject: communicating SD models
(I apologize if this turns out to be a duplicate. I sent it a few days
ago but it didnt appear on the list.)

RE:
>I guess Im looking for a protocol for presenting this kind of work to senior
>people who are coming into the field "cold." I can never have them for
>enough time to give them an academic training session in SD, modeling, and
>systems simulation. We have discussed various ways of going about it: MFS
>first to pique interest, followed by stock/flow model, followed by Causal
>Loop diagrams... or, start with the causal loop diagram first, followed by
>model, followed by MFS. Any ideas....???

>LtCol Robert Glitz
================================ response =================================
One of the things Ive done to prepare a group of SD unaware people is to
build a small model in front of their eyes of something that they intuitively
understand. I usually use population, births, deaths, fertility and avg. age,
for a starter. Using Ithink, I go through all the features and tools in the
process of building it. I also ask them questions such as "what increases
population". This always brings a couple funny answers before I say, Oh, you
mean fertility. Then I place that on the diagram. As soon as the diagram
is done, I put a value of .04 in fertility and tell them unit analysis
is important. This .04 is people per person per year. I use 70 years for an
average age and 5.5 billion as initial population. Before I run the model for
100 years while they watch a graph of births, deaths and population, I ask
them what the plot should look like. They usually get this right.

After this I ask about limits to growth. After a little discussion I put a
variable on the diagram call MAX POPULATION, and I explain that I really dont
need to model all the detail right now but that there is a limit based on the
discussion we just had. I feed POPULATION and MAX POPULATION into a new
variable called FRACTION OF MAX and ask them what the equation should be. Most
understand that it should be POPULATION/MAX POPULATION. Then I feed this
FRACTION OF MAX into AVERAGE AGE. I ask what the equation for this should be.
Trouble occurs here, but I explain you could use something like 35/FRACTION OF
MAX. Then I show the graphical input function. I open AVERAGE AGE and set it
equal to FRACTION OF MAX. Then click BECOME GRAPH. When I sketch the AVERAGE
AGE going from 70 to about 20 as FRACTION OF MAX goes from .5 to 1, there are
usually a few "ahs". Finally, I ask what effect this will have on population
by sketching an XY chart of POPULATION versus TIME and having them give me
some scenarios. Usually people hesitate and I ask will population go above
the MAX POPULATION. Will it decline? What will the steady state value be.
The shape, pattern and final value will depend on what you sketch and what the
final value for AVERAGE AGE is when FRACTION MAX =1.

======================= EQUATIONS ===============================

POPULATION(t) = POPULATION(t - dt) + (BIRTHS - DEATHS) * dt
INIT POPULATION = 5.5E9

BIRTHS = POPULATION*FERTILITY
DEATHS = POPULATION/AVG_AGE
FERTILITY = .04 {PEOPLE PER PERSON PER YEAR}
FRACTION_MAX = POPULATION/MAX_POP
MAX_POP = 10E9
AVG_AGE = GRAPH(FRACTION_MAX)
(0.00, 70.0), (0.1, 70.0), (0.2, 70.0), (0.3, 70.0), (0.4, 70.0), (0.5, 70.0),
(0.6, 68.5), (0.7, 63.0), (0.8, 51.0), (0.9, 30.5), (1, 9.00)

IF AVG_AGE IS 70 FOR ALL VALUES OF FRACTION_MAX, POPULATION REACHES 6.7E10

UNDER THE ABOVE ASSUMPTIONS, POPULATION REACHES 9.26E9


This entire exercise takes about 20 to 30 minutes. Then we start work on
their problem.

=====================================================================
Bill Bonnell (716) 477-7874 | For the things we have to learn
bonnell@kodak.com | before we can do them,
73323.2112@compuserve.com | We learn by doing them -- Aristotle

[da5e@rain.org (Dale Kirby)]

In the March 11, 1995 edition of the L.A. Times Metro section there is a
short description of the work of UCLA gerontologist John F. Schnelle using
systems analysis and quality control techniques with nursing homes. If
there is interest I will synopsis the article.

Dale Kirby

______________________________________________________________________________
Dale Kirby Moving the agenda forward
Free-Lance Database Developer with focused creativity.
da5e@rain.org
Http://www.rain.org/~da5e
____________________________________________________________________________
__