Multiple tests for validation of System Dynamics type of simulation models

Validation of model behavior is an important part of simulation validation in general, System Dynamics model validation in particular. System Dynamics methodology has often been criticized for its lack of quantitative/formal behavior evaluation tools. Indeed, System Dynamics models have certain characteristics that render standard statistical tests inappropriate. But the paper argues that such inappropriateness should not lead to adopting entirely qualitative procedures. The challenge is to design novel quantitative tests that are appropriate for System Dynamics behavior evaluation. Such tests should focus on major time patterns rather than individual data points. The paper suggests a set of procedures that focus on major pattern components of a given composite behavior pattern. It then investigates the appropriateness and properties of those procedures through a series of simulation experiments. The experiments show that the suggested procedures are easy to apply and interpret and suitable for System Dynamics type of behavior evaluation.