A simulation-optimization approach to retrieve reservoir releasing strategies under the trade-off objectives considering flooding, sedimentation, turbidity and water supply during typhoons

Tuesday, 16 December 2014
Chien-Lin Huang1, Nien-Sheng Hsu1, William W-G Yeh2 and Gene J-Y You3, (1)NTU National Taiwan University, Taipei, Taiwan, (2)University of California Los Angeles, Los Angeles, CA, United States, (3)National Taiwan University, Taipei, Taiwan
This study develops a simulation-optimization approach for retrieving optimal multi-layer reservoir conjunctive release strategies considering the natural hazards of sedimentation, turbidity and flooding during typhoon invasion. The purposes of the developed approach are: (1) to apply WASP-based fluid dynamic sediment concentration simulation model and the developed extracting method of ideal releasing practice to search the optimal initial solution for optimization; and (2) to construct the replacing sediment concentration simulation model which embedded in the optimization model. In this study, the optimization model is solved by tabu search, and the optimized releasing hydrograph is then used for construction of the decision model. This study applies Adaptive Network-based Fuzzy Inference System (ANFIS) and Real-time Recurrent Learning Neural Network (RTRLNN) as construction tool of the concentration simulation model for total suspended solids. This developed approach is applied to the Shihmen Reservoir basin, Taiwan. The assessment index of operational outcome of multi-purpose multi-layer conjunctive releasing are maximum sediment concentration at Yuan-Shan weir, sediment removed ratio, highest water level at Shan-Yin Bridge, and final water level in Shihmen reservoir. The analyzed and optimizing results shows the following: (1) The multi-layer releasing during the stages before flood coming and before peak flow possess high potential for flood detention and sedimentation control; and during the stages after peak flow, for turbidity control and storage; (2) The ability of error toleration and adaption of ANFIS is superior, so ANFIS-based sediment concentration simulation model surpass RTRLNN-based model on simulating the mechanism and characteristics of sediment transport; and (3) The developed approach can effectively and automatically retrieve the optimal multi-layer releasing strategies under the trade-off control between flooding, sedimentation, turbidity and water supply.