An eigenvalue analysis for the evaluation of policy parameters in a system dynamics model

This paper utilizes the eigenvalue analysis in identifying model parameters and selecting their appropriate values to obtain a stable system response in the production-stock problem of Coyle. The model is first assumed to be in steady-state equilibrium. This obviates the need for the delays in physical flows and the smoothing levels in the information flows, and helps in converting an eight-order model into a second-order model. Applying the eigenvalue analysis to select appropriate parameter values, it is shown that the initial policy set, which was rejected by Coyle because it produced undesirable, explosive oscillations, can result in extremely good model behaviour. The model behaviour so obtained has been found to be superior even to that obtained with the final policy of Coyle.