Bayesian Stochastic Genetic Algorithm for Operation of Reservoirs

Operation of reservoir systems using the Bayesian stochastic GA-based optimization model (BSGA) is investigated in this paper. This model considers the joint probability distribution of inflow and forecast flow to the reservoir. In this way, the intrinsic flow and forecast uncertainties are considered. In this study, a multi objective approach that considers the interest of different agencies, water users and stakeholders in water allocation from the reservoir is taken. The proposed model maximizes an objective function which is based on the expected value of the Nash product which includes different utility functions of different players as well as their relative authorities on the water allocation process. This work is an extension of a Bayesian Stochastic Dynamic Programming (BSDP) developed by Karamouz and Vasiliadis (1992). They utilize the Bayesian decision theory in the optimization algorithm for reservoir operation. The proposed GA-based optimization model does not have the dimensionality problem of the BSDP. In order to test the proposed methodology, the model is applied to the Satarkhan Reservoir system in north-western part of Iran. The results show the significant value of the proposed model in allocated water in a real world system, considering the forecast uncertainty and reducing the run time compared to an alternative optimization model such as BSDP for reservoir operation problems.

Keywords:

: Reservoir Operation , Bayesian Stochastic Genetic Algorithm , Water Allocation , Conflict Resolution