Implementation in PCSWMM Using Genetic Algorithms for Auto Calibration and Design-Optimization

Abstract

This paper discusses the development, application and performance evaluation of a genetic algorithm-based software tool (PCSWMM) for calibration of the Storm Water Management Model (SWMM version 4.4h). Model calibration is a crucial step in developing a useful storm water model, especially when the model is used to evaluate one or more "what-if" scenarios in an existing storm water system. While a SWMM model can be applied to very simple modeling problems, it can also be quite complex, containing thousands of significant hydraulic and hydrologic entities. As each model entity may contain as many as a dozen sensitive, uncertain parameters, and as the volume of available observed time series data increases, rigorous manual calibration can be expensive and time-consuming. For this reason, model calibration is often not performed, or performed inadequately. An automated calibration tool is described that significantly reduces the effort required for calibration and design optimization. A sample application is provided. Such tools encourage the adoption of more thorough model development and verification protocols, and better design.