An improved dynamic programming algorithm by integrating discrete differential and successive approximation for real...

Urban drainage systems (UDSs) are facing increasing demands for storage and discharge capacity in urban flood control and pollution prevention, where real-time orifice operation plays a crucial role. However, little research has recognized the importance of improving UDS storage-drainage capacity by enhancing orifice control precision and reducing optimization dimensionality. To fully exploit the storage capacity of UDS, a new optimization algorithm is developed in this study by coupling Discrete Differential Dynamic Programming (DDDP) with Successive Approximation (DDDP-SA). The new method leverages successive approximation to reduce the dimensionality of the optimization problem and subsequently uses DDDP to search optimal solution within each feasible corridor. The capability of the new method is demonstrated in Yueyang, China, and is compared with three existing control strategies including passive, rule-based control (RBC), and the Dynamic Programming with Successive Approximation considering the time lag of flow routing (DPSA-TL). Results indicate that DDDP-SA can further reduce combined sewer overflow (CSO) volume by 1.84 % to 11.03 % across all rainfall events comparing with the passive strategy. However, the CSO mitigation effect diminishes as rainfall intensity increases. For real-time control (RTC), DDDP-SA maintains superior effectiveness, achieving 1.83 % to 10.64 % reduction of CSO. Compared with DPSA-TL, DDDP-SA can improve the precision and adaptability of UDS control in terms of increased orifice usage duration and opening standard deviation by 4 % to 8 %. This paper presents a novel algorithm designed for the efficient real-time optimization and fine-grained orifice operation of UDS. The proposed methodology facilitates the strategic utilization of the pipe network's inherent storage capacity, thereby significantly improving the overall resilience of UDS.