Cost–Resilience Trade-Offs in Urban Stormwater Pipe Network Upgrading Under Increasing Rainfall Severity

Increasing rainfall severity is placing growing pressure on urban stormwater drainage systems, yet the cost–resilience efficiency and attainable performance of conventional pipe upgrading under severe rainfall remain insufficiently understood. This study developed a coupled Storm Water Management Model–Non-dominated Sorting Genetic Algorithm II (SWMM–NSGA-II) framework to optimize stormwater pipe networks under 5-, 20-, 50-, and 100-year design rainfall events in a highly urbanized catchment of Beijing, China. Pipe diameter was optimized to maximize flood resilience while minimizing life-cycle cost, and Pareto-optimal solutions were evaluated to identify representative knee-point configurations. Total node flooding volume increased from 13,325 to 46,544 m3 between the 5- and 100-year events, while the Gini coefficient declined from 0.838 to 0.742, indicating a broader redistribution of flooding burden across the network. Mean optimized resilience decreased from 97.35 to 83.00, although relative improvements over baseline conditions increased from 16.17% to 21.33%. In contrast, mean life-cycle cost varied by only 4.39% across rainfall scenarios. Pareto fronts exhibited progressively smaller resilience gains toward the high-cost range, and knee-point solutions showed selective rather than uniform pipe enlargement. These findings highlight a growing mismatch between infrastructure investment and attainable resilience under severe rainfall, supporting sustainable, performance-based stormwater rehabilitation with potential applications to drainage retrofitting in industrial parks and aging residential communities undergoing urban renewal.