Optimization of System Dynamics Models: a Novel Methodology

A novel methodology for the optimization of system dynamics models is summarized and illustrated via the application of this methodology to a case study of coal-transportation management. The objective of this case study is to minimize total cost while satisfying a given constraint for the efficiency of the simulated system. This methodology combines the “Karush-Kuhn-Tucker” conditions (which are well known in mathematical optimization) with “efficient global optimization”, which is closely related to “Bayesian optimization” and “machine learning” as they all use Gaussian processes or Kriging to approximate black-box models. The case study numerically illustrates the methodology’s effectiveness and efficiency, compared with the optimizer in the “Insight Maker” software for system dynamics models.