ABSTRACT
OPT~ING THE RULE FOR ANNOUNCED ALLOCATION - POLICY ITERATION MODEL A computer program was developed to optimise the announced allocation rule for the system. The optimisation technique used was discrete differential dynamic programming, with a policy iteration method. A policy relating the announced allocation to the amount in storage at the start of the year is specified initially. A corridor of releases for those specified storage levels is searched to find a more optimal release policy. The corridor around the new optimal policy is then searched in an attempt to find a better policy. This process continues until a specified level of convergence in operation policy or annual return is achieved. The objective function of this model consists of the benefit obtained for supplying water coupled with penalties for violating the following constraints.
