PAD 724 - Lecture 2 - Concept models

I. Further thoughts on project proposals?

II. Revision to syllabus

A. Abbreviated review of building blocks of models
B. Substitute covering Vensim “molecules” of structure?
C. Add (tonight) talking about Concept Models

III. Software
A. Downloading Vensim DSS
1. Follow the instructions given in class or email.
B. Downloading the Molecules
1. Go to the Download page of www.Vensim.com
2. Select “Free downloads”
3. Select Molecules.
4. Download
5. Note that a PDF file called “Tutor” is contained among the molecules

IV. Concept Models

A. What are they?
1. Very small, vividly clear, simple models focused on a proposed problem or
project, capturing the “concept” of the study in both structure and dynamics
2. See “Teamwork in Group Model Building”

B. How have they been used?

1. Group model building
a) To introduce the system dynamics approach that would be used in a
group modeling session
b) To introduce stocks, flows, and feedback loops and the icons we use
for them
c) To introduce simulating a dynamic system
d) To illustrate that there are links between structure and behavior, that
“behavior is a consequence of structure”
e) To initiate of discussion of the structure and behavior of the real
system involved in the problem description
2. Professional modeling
a) To begin a process of conceptualization of a difficult problem, starting
with a whole-problem view at a very simplified level
b) To drop back from complicated modeling to clarify purpose and
directions for the more complete and complicated modeling, and
develop understandings

V. Concept Model Examples

A. Welfare reform (Cortland folder)

1. A sequence of three models (or one model in Vensim DSS that is unfolded
bit by bit and run using “partial simulation”)
 Job loss rate
Fraction losing job
per year
On At risk
assistance Job finding employed
rate

Avg length of stay

on assistance
2.
At Risk Populations
4,000

3,000

2,000

1,000

0
0 2 4 6 8
Time (Year)

On assistance : welf1
At risk employed : welf1
3.

Job loss rate Fraction losing job

per year
On At risk
assistance Job finding employed
rate
Employment
ratio
Avg length of stay
on assistance
Jobs
4.
At Risk Populations
4,000

3,000

2,000

1,000

0
0 2 4 6 8
Time (Year)

On assistance : welf2
At risk employed : welf2
5.
 Job loss rate
Fraction losing job
per year
On At risk
assistance Job finding employed
rate
Loss of Employment
assistance rate ratio
Avg length of stay
on assistance
Unemployed
and Jobs
unassisted Avg total time on
assistance
Fraction losing
assistance per year
6.
At Risk Populations
4,000

3,000

2,000

1,000

0
0 2 4 6 8
Time (Year)

On assistance : welf3
At risk employed : welf3
Unemployed and unassisted : welf3
7.

8. We note the behavior changes as the structure changes. The group also
realizes that the model is theirs to change, to better fit their senses of reality.
B. Office of Real Property Systems (ORPS)
1. A single simple model built beforehand and built up with the group piece by
piece.
fraction revaluing per year

Fraction Fraction
moving to becoming
continuous Revaluing compliant
reval each yr towns far from each yr
reval

Towns recently Towns not too Towns far from

revalued far from reval reval
Slipping from Slipping far
recent reval from reval
Moving towns Moving
to continuous recalcitrant
reval towns to reval
time to slip from time to slip to
Towns recent reval far from reval
Towns seldom
continuously
revaluing
revaluing

<Total cost> <Average time since reval>

2.
3. After experiments with fake costs, the group estimated real annual costs for
towns in the stocks and transition costs for moving a town from one category
to another, with the following dramatic results:
Graph for Average time since reval
6

4.5

1.5

0
1990 1995 2000 2005 2010
Time (Year)
4.
Graph for Total cost
200 M

175 M

150 M

125 M

100 M
1990 1995 2000 2005 2010
Time (Year)
5.
6. The last graph suggests that the only way to lower costs is to (somehow) get
to revaluing continuously, but getting there will cost more in the short run
(worse before better behavior).
C. ATKearney
Total
Target growth rate personnel Fraction experienced

Experienced
New personnel
personnel
Hiring rate Experienced
Gaining experience rate
leave rate

Time to become
Frac exp leaving p year
1. experienced

Total personnel Fraction experienced

Target growth
rate

Experienced
New personnel
personnel
Assimilation Experienced
Hiring rate rate leave rate

Time to Frac exp

become leaving p year
experienced
Mentoring time available
Mentoring time needed

Adequacy of
Mentoring time
mentoring
needed per New
2.
Total personnel Fraction experienced

Target growth
rate

Experienced
New personnel
personnel
Assimilation Experienced
Hiring rate rate leave rate

Time to Frac exp

become leaving p year
experienced
Mentoring time available
Mentoring time needed

Adequacy of
Mentoring time
mentoring Mentoring time
needed per New available per Exp

Perceived time
to become Exp
Pressure for mentoring

3.
 D. Basell
Delivery
Delivery delay
delay perceived by
the market

Phantom
Inventory
Producing Shipping ordering loops

Phantom
ordering
Production Ordering +
capacity
Inventory Demand Fraction of
Building orders shipped
coverage
capacity Availability loop
Capacity
acquisition
delay
Capacity in
construction Capacity
adjustment loop

Capacity
ordering Planned growth
in capacity
1.

VI. What do concept models mean to us?

A. Smart to use them to start a modeling project.

1. We will try to build into 724 an exercise that involves building a concept
model related to the project.
B. Smart to use them in the middle of a modeling project if it begins to get complicated
and off track
1. We may build that in too, or instead.
C. Very smart to use them with clients.
1. Takes less than half an hour to introduce our tools and perspectives and begin
to talk about the clients’ problem in their terms with our contributions.

VII. Coflows

A. The idea
1. Tracking an attribute of a stock and its flows, like energy efficiency of
housing, the cost (value) of production capacity, the quality of admits to a
degree program or hires in an agency, and so on.

B. Generic example
 C. Coflow tracking smokers in an aging chain

nonsmoking smoking nonsmoking

adults elderly
adults dying dying dying

Children Teens Adults Elderly

births p year children maturing teens aging adults aging smoking elderly
dying

fraction children fraction teens fraction adults fraction elderly

smoking smoking smoking smoking

<Fraction
Children Teens Adults Elderly
smoking>
smoking smoking smoking smoking
child smokers teen smokers adult smokers elderly smokers
maturing aging aging dying
new teens new adults Adult smokers new elderly
new children
smoking smoking smoking dying smoking

D.

VIII. Exercise 1 - Using a coflow in Roger Naill’s natural gas model