Stochastic Multiobjective Optimization Model for Urban Drainage Network Rehabilitation

Abstract

Flooding in urban areas has become increasingly common in recent decades, as a result of increased urbanization, decreased infiltration rates, and climate change. Hydraulic rehabilitation plans can be developed and implemented to maintain suitable urban drainage system performance. Determining the effective plans, however, requires the involvement of rainfall uncertainties in the modeling and using special tools. In this study, statistical copula functions are established to determine joint probability distribution of rainfall variables considering their dependence structure. The most credible distribution is then used through the Monte Carlo simulation (MCS) to investigate rainfall uncertainties. A multiobjective optimization model is also developed and, after validation, is linked to the EPA-SWMM model for evaluating urban drainage rehabilitation scenarios. The copula-based multiobjective optimization model represents a range of cost-effective rehabilitation plans in terms of overflow improvements and their confidence intervals. This provides decision makers with valuable information about the robustness of the solutions and the possibility of selecting a trade-off region given constrained rehabilitation resources.