Automated Calibration using Optimization Techniques with SWMM RUNOFF

Abstract

The usefulness of a hydrologic model is directly related to its application and how well it is calibrated. Calibration is a subjective exercise where model parameters are adjusted to reduce discrepancies between measured data and modeled predictions. Automated calibration can be used to accelerate the model calibration process, minimize modeler bias, and increase the goodness of fit between measured and modeled hydrographs. During calibration of a complex hydrologic model, it may be difficult to simultaneously adjust predicted output hydrographs to correspondingly match multiple objectives (peak flows, total volume and shape of the hydrograph). Custom programming was used to link SWMM Runoff version 4.4h with Palisade’s Evolver software to improve model goodness of fit. A small sanitary sewer basin was simulated as part of a collection system rehabilitation pilot program to judge the effectiveness of infiltration and inflow (I/I) removal. A one-month time series of hourly flow measurements were used and calibration was performed with an automated calibration method that applied a genetic algorithm solution technique. Several goodness-of-fit metrics revealed an improved calibration for both pre- and post-rehabilitation flow hydrograph, as well as for projected hydrographs to a design event. This study demonstrates an accurate and cost-effective automated method for model calibration that is not only valuable for repeated model analyses performed throughout a collection system rehabilitation program, but can also be applied to other watershed models.