An enhanced multi-objective evolutionary algorithm for the rehabilitation of urban drainage systems

Abstract

This study aimed to develop a hybrid simulation-optimization model to find optimal strategies for the rehabilitation of urban drainage systems. The HEC-HMS was used for rainfall-runoff analysis and the EPA-SWMM model to hydraulically route the floods in urban channels. A combination of EPA-SWMM with a new developed multi-objective evolutionary algorithm (MOEA), called non-dominated sorting enhanced differential evolution (NSEDE), was used to optimize the size of flood walls, cross-structures and detention ponds, with the objective of minimizing the rehabilitation costs and runoff surcharge. NSEDE exhibited better performance in terms of convergence and solution diversity compared to three well-known MOEAs. Comparison of the model outputs with the rehabilitation plan of Tehran municipality demonstrated the superiority of the optimum designs. At the same level of cost and flooding with respect to the Tehran municipality plan, the optimum designs reduced the cost of rehabilitation by 61.7% and network flooding by 37.5%.