Optimal monthly operation of interconnected hydroelectric power storages

Abstract

Many optimizing studies of water resource systems have given emphasis to the conceptual problem of formulating the design of a system as a set of mathematical functions to which optimizing techniques may be applied. By considering a real-world case of two interconnected reservoirs in the South Island of New Zealand where decisions must be made each month about the quantities of water to release for hydroelectric power generation, the practical problems of applying an optimizing technique are illustrated. The technique used is deterministic dynamic programming coupled with a multivariate stream-flow simulation model. The hydro power generating plants are operated in conjunction with thermal generating plants to provide a specified energy load for each month; the object in operation is to minimize the expected present value of the thermal generating costs; a real-time operating policy which specifies a planned release for each month as a function of the quantities of water stored at the beginning of each month is developed. The study illustrates important advantages of this particular dynamic programming approach over other optimizing techniques; in particular good account is taken of the stochastic nature of the inflows. These occur at four places in the river basin, show considerable seasonal variability, and substantial spatial and temporal intercorrelations. Excessive computational demands for more than two storages are a major restriction of the technique.