ABSTRACT

A prolic author in the modeling eld Bran Selic, at the time this was written an employee of IBM Rational, has offered a list of characteristics for a useful model paraphrased as follows:

1. The use of abstraction to emphasize important aspects while removing irrelevant ones

2. Expressed in a form that is really understandable by observers 3. Fully and accurately represents the modeled system 4. Predictive such that it can be used to derive correct conclusions

about the modeled system 5. Inexpensive meaning it is much cheaper to construct and study

than simply building and observing the modeled system

17.2 How are models used in the synthesis work? As we move from problem space denition during the system development process into solution space design we need a set of design models to support our efforts because it is cheaper and easier to manipulate a good model to observe the effects of different solution possibilities than it is to build a prototype. While doing some work for Caterpillar several years ago the author was struck by a comment from several engineers and managers that they were tying to move from the use of as many as four sequential prototype versions of a new product, referred to as iron verication, to the use of simulation to evaluate different design possibilities.