Large-Scale Nonlinear Conjunctive Use Optimization Problem: Decomposition Algorithm

Abstract

A cyclic storage (CS) system is defined as physically-integrated and operationally interconnected surface water and groundwater subsystems with full direct interactions between the subsystems. The proposed definition treats surface and subsurface impoundment subsystems as competing and potentially interconnected parallel storage facilities that minimize most of the problems associated with large-scale surface impoundments for water supply purposes. This paper emphasizes on the development and implementation of a hybrid two-stage genetic algorithm (GA)-linear programming (LP) algorithm to optimize the design and operation of a nonlinear, nonconvex, and large-scale semidistributed, CS system in an irrigable area. Performance of the proposed model is tested with a 240-period problem where other approaches failed to locate a feasible solution. For optimal operation of the system, a set of operating rules are developed for the joint utilization of surface and subsurface storage capacities to meet a predefined demand minimizing construction and operation cost over a 20-seasonal planning period. Results show that CS dominates a noncyclic storage system both in cost and operation flexibility.