An Integrated Approach to Evaluating Simulation Credibility

Discussions about simulation credibility tend to focus on the issue of "validation", i.e., how well simulation predictions match real world observations. While validation is certainly a most direct and intuitively accessible measure of simulation credibility, the validation process has numerous well-known limitations. Validation is by no means the only measure of simulation credibility, however. This paper identifies and categorizes a spectrum of information that can be used to evaluate simulation credibility more robustly than reliance on validation results alone. It also describes a method by which the nature, scope and depth of information necessary to establish simulation credibility for a particular application can be determined from an analysis of the risks associated with that application.