A ROBUST HYBRID APPROACH FOR EFFICIENT OPTIMAL DESIGN OF SEWER SYSTEMS

摘要

For the optimal cost-effective design of sewer system improvements, one of the state-of-the-art approaches is to incorporate a Genetic Algorithm with a mathematical sewer simulator, which simulates detailed hydraulic performance of the system. However, intensive calculations in both the system evaluation and the optimization processes always result in a very high computation cost for this approach to achieve a sound solution, especially when designing large sewer networks. In this paper the authors introduce an innovative design approach, which firstly combines a simplified fast sewer simulator (Fastnett [4]) with a single-objective GA. Due to the vast computation saving achieved by Fastnett, a design solution can be obtained in an extraordinarily short computation time. However, as compensation for its simplification, Fastnett cannot precisely capture the full hydraulic behavior in sewers. Therefore, the solution may be defective, I.e infeasible or too expensive. So the system performance is re-evaluated by using dynamic wave routing in EPA SWMM [10], and a secondary optimization method based on cellular automata (CA) principles is applied for fine adjustments to the preliminary design. CA is adopted here because of its key features: locality, homogeneity and parallelism, which enable it to make design improvements locally and efficiently. A small artificial and a large real sewer system were employed for case studies. The results indicate that at the second stage CA optimization highly reduces or eliminates the flooding problems remaining in the preliminary GA solution and eventually decreases the total cost still further. In the first case, its solution was also compared to that found by the conventional approach, namely directly incorporating the GA and SWMM simulation. The proposed approach clearly demonstrates its capability to be a highly efficient and effective tool for sewer system design.