Author(s): B. Nirmala; P. P. Mujumdar

Linked Author(s):

Keywords: Bayesian stochastic; Dynamic programming; ANN; Reservoir operation; Hydropower

Abstract: Reservoir operation problems may be characterized by two sources of uncertainties: first one is the natural uncertainty with respect to the actual realization of stream flow and the second one is the uncertainty associated with the accuracy of forecasts. This paper presents the development of an operating policy model for a multi-reservoir system for hydropower generation by addressing forecast uncertainty along with inflow uncertainty. The stochastic optimization tool adopted is the Bayesian Stochastic Dynamic Programming (BSDP). The BSDP model proposed by Karamouz and Vasiliadis, [1992] is essentially a dynamic programming model, which incorporates a Bayesian approach within the classical Stochastic Dynamic Programming (SDP) formulation [Loucks et al., 1981]. BSDP differs from the classical SDP in the selection of state variables and the way in which transition probabilities are derived. The model solution results in a steady state release policy for individual reservoirs. The BSDP policy is simulated with historical inflows and different types of forecasts viz., inflow forecasts resulting from an ANN rainfall forecast, forecasts as mean values of monthly inflows, and perfect forecasts of inflows i. e., forecasts equal to historical inflows. The simulation results evaluate the sensitivity of the operating model to the accuracy of forecasts.

DOI:

Year: 2005