Abstract

This chapter introduces a tentative procedure for estimating the minimum level of complexity of a computational model for designing a cost-effective combined sewer overflow (CSO) system. In this approach, model complexity is related to a measure of model uncertainty, and to the design costs. Ultimately the goal is to determine the optimal (least cost) complexity for a model (version 4.3 of the U.S. EPA Storm Water Management Model, SWMM) which is itself uncertain.

Differences between modelers, and errors in datafiles could invalidate our proposed method. Thus we propose a procedure that first produces consistent models free of user-input error. Developed in the form of an expert system shell, our procedure uses a front-end rule-based (FERB) decision support system (DSS). This FERB-DSS comprises two functions, model development, and sensitivity analysis: first the user is led through the development of consistent input files; second a heuristic sensitivity analysis procedure is applied to selected model parameters. In a following step, calibrated parameters that consistently reproduce observations are determined. Finally, minimum uncertainty and corresponding cost production functions are derived, and from these parameters an optimal level of model complexity is defined. This chapter describes the final step, as well as the problem of determining a globally optimum set of calibration parameters for SWMM.

Using 1994 U.S. EPA policy as a guideline for design, our proposed method is applied to a complex and costly problem: CSO controls in Columbus OH. We conclude that analysis of uncertainty is essential when using models to design CSO controls.

This paper is only available in PDF Format:

  View full text PDF