Optimization of Multireservoir Water Resources Systems Operation Using Genetic Algorithm

Since the river flow regime is not always in harmony with the downstream water requirements, reservoir systems are constructed to regulate the natural river flow. Because of the spatial distribution of the water requirement sites, the storage system on a river may consist of several reservoirs. Due to the variable rainfall and river regime, the management policies play an important role for operation of the reservoir system. In this study, a deterministic Genetic Algoritem model is developed for optimal operation of the multireservoir water resource system in the north of Khorasan, northeastern Iran. The reservoirs are single purpose and regulate water for an irrigation project. The system is intended to maximize the total farm income. The system is made up of two reservoirs in series on Zangelanloo and Shoorkal rivers. Objective downstream farming fields are cultivated with a pre-determined multiple cropping pattern of wheat (27% and 18% in field 1 and 2, respectively), barley (30% and 26% in field 1 and 2, respectively), and sorghum (43% and 56% in field 1 and 2, respectively). The model developed in this study is used to obtain the optimal pattern of reservoir operation and water allocation among different crops for a definite combinations of state variables (reservoir storage classes at the beginning of the season and rainfall and inflow regimes). Total farm income were maximized. Running the model for 12 combinations of the state variables (4 reservoir storage classes and 3 regimes of dry, wet and average for rainfall and river inflow) showed that the results corresponding to the dry regime were sensitive to the reservoir storage class at the beginning of the season. In other regimes this sensitivity decreased. Also relative crop yield of field 2 decreased more in the dry regime, which may be due to the smaller reservoir in Shoorkal dam.