Database Reference
In-Depth Information
The error may be either in your STELLA program or your understanding of the
system that you wish to model—or both.
What do you really have here? How does STELLA determine the time path of the
state variable? At the beginning of each time period, starting with time = 0 days (the
initial period), STELLA looks at all the components for the required calculations.
The values of the state variables will form the basis for these calculations. Only
the variable GETTING SICK depends on the state variable SICK. To estimate the
value of GETTING SICK after the first time period, STELLA multiplies 0.05 by the
value SICK (at time = 0) or 10 (provided by the information arrows) to arrive at 0.5.
From time = 1 to time = 2, the next DT, STELLA repeats the process and continues
through the length of the model. When you plot your model results in a table, you
find that, for this simple model, STELLA calculates fractions of SICK from time =
1 onward. This problem is easy to solve; for example, by having STELLA round the
calculated number of SICK—there is a built-in function that can do that—or just by
reinterpreting the population size as “thousands of SICK.”
This process of calculating stocks from flows highlights the important role
played by the state variable. The computer carries that information—and only that
information—from one DT to the next, which is why it is defined as the variable
that represents the
condition
of the system.
You can drill down in the STELLA model to see the parameters and equa-
tions that you have specified and how STELLA makes use of them. Click on the
downward-pointing arrow at the left of your STELLA diagram.
The equations and parameters of your models are listed here. The model equations
are also listed at the end of each chapter of this topic so you can more easily recreate
the models. Note how the SICK population in time period
t
is calculated from the
population one small time step, DT, earlier and all the flows that occurred during
that DT.
The model of the SICK population dynamics is simple. So simple, in fact, that
it could be solved with pencil and paper, using analytic or symbolic techniques.
The model is also linear and unrealistic. Next, add a dimension of reality—and
explore some of STELLA's flexibility. This may be justified by the observation that,
as populations get large, mechanisms set in that influence the rate of GETTING
SICK.
To account for feedback between the size of the SICK population and its rate of
GETTING SICK, an information arrow is needed to connect SICK with CONTA-
GION RATE. The connection will cause a question mark to appear in the symbol
for CONTAGION RATE. The previous specification is no longer correct; it now
requires SICK as an input. We will make this input indirectly. First we construct
