From facility selection to real-time scheduling: An SWMM-MPC framework for urban drainage control under extreme...

Increasingly frequent extreme rainfall poses severe challenges to conventional urban drainage systems (UDSs), often rendering them inadequate for effective flood control. Mitigating flood risk necessitates an integrated approach that leverages the conveyance capacity of urban roads and the storage capacity of green infrastructures (i.e. community parks). However, existing studies often fail to optimize the configuration and operation of emergency storage and drainage facilities (ESDFs) under varying rainfall scenarios. Moreover, the conventional strategy of minimizing total system overflow volume fails to adequately protect critical roads and facilities. To address this gap, this study proposes a mechanism for optimized, tiered utilization of ESDFs based on road risk levels. An optimal facility allocation strategy is derived using dynamic indicator weights that adapt to rainfall intensity. The results indicate that risk assessment is dominated by the density of critical infrastructure (weighting 0.44∼0.57). As rainfall intensity increases, the system's reliance shifts from the pipe network during low-to-moderate rainfall to community parks; their dispersed nature becomes crucial for enhancing resilience in extreme scenarios. A multi-scenario real-time optimization framework integrating SWMM with Model Predictive Control (MPC) was developed to minimize road inundation depth through the coordinated regulation of grey-green-blue infrastructure. Results show that the model reduces road inundation depths across the entire network by 0.13∼0.62 m (3.92%∼8.42%) for 10- to 50-year return period events. Even under extreme 100- to 200-year events, flooding depths on high-risk roads were reduced by 0.11∼0.25 m (2.93%∼6.96%), demonstrating its effectiveness as the system approaches saturation.