Calibration of an objective function for the optimization of real‐time reservoir operations

Optimization models have been used as simulation tools to mimic complex decision making processes such as real‐time reservoir system operations. A recurring difficulty in the development of optimization models for use as simulation tools is the proper specification of the objective function. A new method for calibrating the objective function of an optimization model is presented. The method is developed specifically for the quadratic program class of models, but is applicable to general nonlinear programs with polynomial objective functions. A simplified case study is used to exercise the method in calibrating a quadratic objective function of a math program used to simulate operation of the Green River Basin hydrosystem in Kentucky. Results of the case study indicate that the method may be successfully used in certain cases to determine objective function parameter values to approximate decision making processes.