Nonlinear optimization model for screening multipurpose reservoir systems

An optimization model for selecting and sizing potential reservoir and hydr opower plant sites on river basins is presented. The model improves on the work of Lall and Miller by replacing the modified sequent peak algorithm fo r sizing reservoirs with a behavior analysis algorithm that allows operatio n of the reservoir system with realistic operating policies. A heuristic si mulation procedure is used to assess the feasibility of developing self-sus tainable hydropower plants at each site. A gradient (derivative) based opti mization algorithm is employed to integrate these simulations into an optim al screening model. To enhance the efficiency of the algorithm, the approxi mate and expensive approach of evaluating derivatives by divided difference s is replaced by automatic differentiation. A comparison of the computation al performance shows that the optimal solution with automatic differentiati on is obtained four times faster than with the divided difference method. T he model is developed in the context of river basins in India.