A Stochastic Optimization Model for Managing Water Supply Operations

This paper presents a reliability-based water resources management framework that utilizes stochastic optimization techniques to account for uncertainties associated with of prediction of climatic condition, water demand, surface water availability, baseline groundwater levels, non-anthropogenic reservoir water budget, and hydrologic/hydrogeologic properties. The framework was developed to manage water resources from over 180 groundwater production wells, stream flow withdrawal, regional reservoir, and desalination plant in the Tampa Bay region in Florida, USA. The developed method maximizes the reliability of achieving the goals that all protected wetlands in the area are healthy and sea water intrusion is prevented. The framework involves (1) a system simulation model to represent the water resources routing under the OROP and (2) a Monte Carlo simulation model to generate realizations of climatic events, water demand, available surface water quantity, and (3) a unit response matrix (URM) that relates groundwater level response to groundwater extraction. A simulator response model simulates how water supply operators adjust the optimized rates of groundwater extraction, surface water withdrawal, and reservoir inflow/outflow according to meeting the water demand in all circumstances. The reliability optimization problem is solved using a differential evolutionary algorithm.