Advanced Topics in System Dynamics Optimization Ken Carpenter Luis Luna Introduction System Dynamics has relied on the use of intuition and experience by system owners and analysts to help design policies for improving system behavior over time Most descriptions of the System Dynamics method emphasize the iterative process required to validate Dynamic Hypothesis. Introduction Desired Model Behavior (Wolstenholme, 1990) Manual changes to simulation Model Actual Model Behavior New parameter values and structures Simulation Introduction (Wolstenholme, 1990) The concept of optimization is based on the automation of the manual procedure. Desired value of objective function in Model Optimization Actual value of objective function in model New parameter values and structure Simulation Most common optimization problems Optimization for Model Calibration Curve fitting Optimization for Policy Analysis Save the World!! Two Approaches (Oliva, 2000) Full-information maximum likelihood via optimal filtering (FIMLOF) Engineering statistics Model Reference Optimization (MRO) Non-linear optimization techniques Model Reference Optimization Main steps Define an objective function (Pay-off) Policy Analysis Calibration Define the Constraints (Parameters) Apply one of the Optimization Methods (Use Vensim or other software).