Re: When to model?

Stephen Wehrenberg (stephenw@gwis2.circ.gwu.edu)
Tue, 29 Nov 1994 07:03:08 -0500 (EST)

Well, most recently ...

I'm a human resource planner for the US Coast Guard. My last three SD
efforts were

1. A model of the gross flows of people through an HR system,
from initial entry to retirement, with a flight simulator interface that
gives decision makers (read "students") control over the recruiting,
promotion, and loss "valves." They quickly understand that only by
careful and thoughtful adjustment of all the valves in a coordinated
fashion can they "win the game."

2. A model of the "front end" of the Coast Guard's HR system,
focusing on recruiting, selection, basic training for enlisted personnel,
first assignment to operating units, entry level technician training, and
eventual assignment to an operating unit as an apprentice technician.
Time frame is about two years (although the effects of the decisions made
about flows live in the system for 20 or more years); numbers involved
are an average flow of 4-6000 people. Again, lumps are to be avoided.
Problem statement had to do with reducing training infrastructure (plant,
instructional staff, etc.) by reducing variance in flows ... reducing the
need for excess capacity. (In a firehouse model, the analog might be
that we need to better schedule and control the incidence of fires so we
can balance our workload and prevent the need for excess capacity ... the
firemen sitting around the station with nothing to do.)

3. A simple multiple bathtub model that examined various ways to
achieve a desired distribution of ethnic/racial groups and genders so as
to ensure that the workforce represents the labor market. Stocks,
losses, and gains are the obvious variables.

Three different looking applications of the SD hammer (iThink), but in
reality, the same nail in each case ... stocks and flows. I choose the
SD models over others when the primary objective has to do with
understanding the dynamics of the system, rather than just developing a
one-time "point" solution. Particularly true when a major facet of the
system in question is reciprocal interaction (but then I'm a psychologist
by trade, with a definite leaning toward social behaviorism, social
learning theory, and bi-directional causality).

Stephen Wehrenberg, Ph.D.
stephenw@gwis2.circ.gwu.edu
"Life is a jam session."

On Mon, 28 Nov 1994, Joe Kilbride wrote:

> Each of us has a number of different "tools" we can use when responding to
> a clients' request for help. For what types of "requests for help" are
> System Dynamics models (e.g., ithink and its cousins) most appropriate?
> What are the criteria you use for choosing SD models over other approaches?
>
> If no interest in this topic, I promise to drop it this time. It's just
> that I have this new hammer, and suddenly everything looks like a nail...
>