- Introduction To STELLA -
![]() |
STELLA is a graphical simulation program which operates in both the Apply Macintosh environment and PC environments and is commercially available. It offers a highly user-friendly and intuitive interface which allows for the easy capture of individual expertise of scientists who may have no previous computer programming skills. |
STELLA provides four basic graphical "tools" which are linked to create the
basic structure of a model. These are referred to as stocks, flows, converters,
and connecting arrows.
![]()
![]()
In the example above, a sample population is represented by a stock. Stock
variables are the rectangular boxes which represent a 'reservoir', or state
quantity which is conserved. Flows are control variables which are able to
alter the state of the stock variables and can be both uni and bi-directional.
In this case, there are two flows representing both the births and deaths
within a population. Converters modify the behaviour of all three variable
classes, and connecting arrows serve as the internal communication 'wiring'
for the model. In this example, the properties of a population's birth rate
are defined by a converter, which is linked to the model via connection arrows.
To build a STELLA model, the user first defines the model's stock variables
and positions them roughly on the computer screen. Depending on the nature
of modeled phenomenon and the modeler's preferences, flows, converters, and
connection arrows are then positioned on the screen to create the general
schematic of the model. Individual variables are then defined by 'double-clicking'
on their screen icons which brings up a definition menu in which the modeler
enters the appropriate variable descriptors. Text may also be added within
a variable definition menu. Variable descriptors include a broad assortment
of mathematical, statistical, or logical "built-in" functions, as well as
user-defined constants or graphical relationships. A modeling 'time-step'
must also be selected which represents the amount of time required to complete
one model cycle using Runge-Kutta simulation techniques.