Model simplification and validation: Illustration with indirect structure validity tests

Abstract

Simplification of a large system dynamics model and validation of the simplified version is illustrated. The original model represents agricultural and environmental problems of irrigation development in Southeast Turkey and consists of 62 stock variables. Its simplified version with a narrow model boundary and higher level of aggregation is a general representation of its selected dynamics and consists of 11 stock variables only. Analysis of reference behaviours, indirect structure validity tests and scenario runs reveal simplified model as a valid and useful version of the original. Simplification helps distilling essential model structures that cause selected problems and increases the quality and understanding of models. It can also be a step towards building theory-like structures and general representation of case specific problems in various application domains.