In a large scale cyclic storage system ,as the number of rule parameters and/or number of operating period increase, general purpose gradient-based NLP solvers and/or genetic algorithms may loose their merits in finding optimally feasible solution to the problem. In these cases hybrid GA which decomposes the main problem into two manageable sub-problems with an iterative scheme between GA and LP solvers may be considered as a sound alternative This research develops a hybrid GA-LP algorithm to optimally design and operate a nonlinear, non-convex, and large scale lumped cyclic storage system. For optimal operation of the system a set of operating rules are derived for joint utilization of surface and groundwater storage capacities to meet a predefined demand with minimal construction and operation cost over a 20 seasonal planning period. Performance of the proposed model is compared with a non-cyclic storage system. The management model minimizes the present value of the design and operation cost of the cyclic and non-cyclic systems under specified and governing constraints, employing the developed GA-LP hybrid model. Results show that cyclic storage dominates non-cyclic storage system both in cost and operation flexibility.

Keywords: conjunctive use, cyclic storage, hybrid GA, decomposition method

Full-Text [PDF 320 kb]   (3610 Downloads)