However, the parameters which are incorporated in a simulation model
and the conditions under which a model is operating, span huge
multidimensional spaces. Therefore, it is not a trivial task to
achieve an acceptable fit for each set of the model's operating
conditions. Some iterative procedure is necessary to identify
acceptable settings of the model parameters. The difficulty arises
from the coupling between model parameters. This means that these
settings cannot be determined individually while at the same time all
remaining parameters are kept constant. Instead of searching for the
value of each parameter individually one must rather search for an
optimal set of *all* parameters as long as they are not
independent from each other, which will rarely be the case.

Figure 2.9 depicts the basic procedure which is
used to find an optimal set of model parameters.

Relying on *measurements*, an *optimizer* is tuning the
parameters of a model in order to obtain that set of parameters which
results in a minimal overall error. This procedure includes several
degrees of freedom. At first, a measure has to be defined which
characterizes the error between the model's results and the
corresponding measurements. Secondly, one needs to combine the results
for particular operating conditions into an overall error.

1999-05-27