On the Modeling and Computational Aspects of Dynamic Programming with Applications in Reservoir Control

Abstract

A discrete stochastic multistage decision model is developed, in the framework of which the dynamic programming algorithm is defined. It is demonstrated that the algorithm may be used as a solution procedure for non-routine reservoir control problems. The study demonstrates that analytical considerations may be used to significantly reduce the amount of computation needed for the implementation of the algorithm. Two reservoir control problems are studied: The "reliability problem" in which one seeks a minimum-cost operation rule under a constraint which specifies the maximum probability of shortage allowed during the life-time of the project, and the "range problem" in which the objective is the minimization of the expected value of the range of fluctuation around a critical storage level.