Optimization of Reservoir Operational Policy Using Simulated Annealing Algorithm (Case Study: Mahabad reservoir)

In this paper two models have been written and solved by two approaches: non-linear programming (GAMS/MINOS solver) and simulated annealing methods in order to optimize operational rules in Mahabad reservoir. Because of missing the required eco-system data, these required data were calculated using Tennant method. The objective function was defined to minimize the deference between the amounts of agriculture water demand and the released water from the reservoir during 23 years of flow data (1370-1393) and their performance were evaluated using the operational indexes. Obtained results of both optimization models showed that, the simulated annealing algorithm with objective function value of 22.01, reliability value of 25%, resiliency velocity value of 22.22%, vulnerability value of 41.04% and sustainability value of 0.032, gave relatively better results compared to the non-linear programming method with the objective function value of 88.92, reliability value of 48.18%, resiliency velocity value of 32.86%, vulnerability value of 84.27% and sustainability value of 0.024. Also, it was found that, simulated annealing algorithm took longer running time to achieve the global optimal point in comparison to non-linear programming. However, using simulated annealing algorithm estimated a lower water shortage of 1632.24 MCM compared to the non-linear programming which estimated water shortage of 3351.4 MCM. Simulated annealing algorithm decreased the shortage by distributing shortages in different months and estimated a better performance.