Abstract
A robust optimization (RO) model to manage water quality in rivers is presented. RO is a framework that is characterized by defining solution robustness and model robustness and controlling the effects of data uncertainties in the model on optimal solutions. Discharge, water depth and water temperature are assumed to vary stochastically in rivers and several scenarios are generated. To maximize expected total BOD loads from outfalls, solution robustness and model robustness are objectives in the RO model. Constraints consist of BOD and DO transport equations discretized by the finite element method, water quality standards and effluent limitation standards in rivers. Optimal allocations of effluent BOD loads can be determined by simulating the stream flow and operating the RO model with various values of multiobjective weights. Application to a hypothetical river system demonstrates that the model can be more effective tool for providing alternative solutions than a commonly used stochastic programming model.
