Towards the algorithmic detection of archetypal structures in system dynamics



Traditionally, model analysis follows qualitative, heuristic, and trial-and-error-driven approaches for testing dynamic hypotheses. Only recently have other methods like loop dominance analysis or control theory been proposed for this purpose. We advocate complementing established qualitative heuristics with a quantitative method for model analysis. To that end, we propose two algorithms to detect Wolstenholme's four generic problem archetypes within models. We tested these algorithms using the Maintenance and World Dynamics models. The approach presented in this paper is a first important step towards the identification of system archetypes in system dynamics and contributes to improving model analysis and diagnosis. Furthermore, our approach goes beyond diagnosis to eliciting solution archetypes, which foster the design and implementation of effective policies. Copyright © 2015 System Dynamics Society