SWMM Calibration Using Genetic Algorithms

Abstract

In order to improve the reliability of the Storm Water Management Model (SWMM), a parameter-optimization approach is required to determine the "best" input parameter sets. Within SWMM, the RUNOFF module is the best candidate module for uncertainty reduction by parameter optimization. The Genetic algorithm (GA) method is developed to optimize SWMM RUNOFF parameters. The basic principle of the GA is the same as that which controls the genetic reproduction process with crossover and mutation being the major operations. By applying the genetic algorithm to SWMM with the aid of the sensitivity wizard in PCSWMM, a sensitivity-based method for automating the calibration of the runoff model is developed. Overall, the average accuracy of the calibrated model was within 97% of the target dataset (TD) after approximately 58 cycles of the GA calibration program. The paper covers the genetic algorithm calibration method and its accuracy, efficiency, robustness and reliability.