Application of Multi-Objective Genetic Algorithm for Optimization of Detention Pond Selection

Chiao-Wen Tai

The study aims at proposing a simulation-optimization model for deciding the optimal combination of detention ponds, which comprises a physiographic drainage-inundation model and a non-dominated sorting genetic algorithm II (NSGA-II). The Dian-Bao Creek Basin in southern Taiwan is chosen as the study area. Five detention pond candidates with different sizes and lacations are adopted for optimizing their combination by minimizing the investment cost and the cost of inundation damage. One-day design rainfalls for different return periods (i.e., 2, 5, 10, 25-years) are used as the model input. During the optimization process, the two conflicting objectives, the investment cost and the cost of inundation damage, are minimized to obtain the Pareto-optimal solutions by using NSGA-II. Based on the post optimization approach, the compromise solutions for different return periods are obtained. An easy cost-benefit analysis is used to evaluate the compromise solutions for different return periods. The optimal combination of detention ponds for the return period (e.g., 10 years in thestudy case) with the hightest direct benefic-coat ratio can be suggested for decision making.

Keywords:  multi-objective genetic algorithm、inundation model、detention pond、optimization