Theme 5-11-Optimal operation of irrigation reservoirs using stochastic dynamic programming.

Abstract

Stochastic Dynamic Programming (SDP) is widely used in deriving optimal operating policies for reservoir. Usually inflow is represented in SDP as markovian first order in every time period or independent for every time period. One state variable is more in markovian first order inflow compared to independent inflow. In this work SDP models with three types of inflow assumptions namely inependent inflow in every time period ,first order markovian inflow in every time period and a mix of both first order markovian inflow assumption and independent inflow assumption based on the statistical significance of serial correlation values has been developed and applied to a irrigation reservoir. The mixed inflow assumption has been found to have less dimensionality problem. The results of using all the three types of inflow assumption in SDP model are compared and discussed.