Abstract for: On parallel aggregation and disaggregation in system dynamics simulation models
We use two versions of the beer distribution game to test and extend existing guidelines and recommendations regarding parallel aggregation in system dynamics (SD) simulation models. First, we build an SD model of a five-stage supply chain that mimics the oscillations, amplification and phase-lag typically associated with a phenomena of supply chain instability referred to as the bullwhip effect. In a second step, we perform parallel disaggregation by duplicating stage one to four. According to the guidelines provided in literature both model versions should exhibit similar dynamics, because the decision rules of the stages are identical and the time the items spend in the individual stocks is similar. We can show, however, that during the process of disaggregation additional decision rules need to be added to the model, which might potentially alter model dynamics.