Reliability programing in reservoir management: 3. System of multipurpose reservoirs

Abstract

A reliability programing technique, which includes the concept of ‘reliability or risk’ in an optimization, is applied to multiple multipurpose reservoir systems. The procedure can be applied to any multipurpose multiunit reservoir system with two general types of linkage: normal channel flow for reservoir releases and pipelines or pumping canals. Thus each reservoir could be connected to every other reservoir, and each could receive releases from any or all other reservoirs as dictated by a particular system configuration. A two‐level solution algorithm is proposed. A solution can be obtained for a reservoir system with few purposes (flood control, power production, irrigation, water supply, and water quality enhancement) and random inflows and demands. The random inflows and demands are represented by conditional distribution functions. The objective function of economic efficiency, representing the tradeoff between benefits and risks embodied by a risk‐loss function, is included in the present approach. The reliability programing model is nonlinear and can be split into two models: search model and special linear programing model. The procedure is illustrated using a portion of the Red River system in Oklahoma and Texas, a system of three multipurpose reservoirs. The three reservoirs individually satisfy purposes (flood protection, hydroelectric power generation, water supply, and water quality enhancement), and two of the reservoirs work together to satisfy additional water requirements (flood control and water quality enhancement downstream). Input data necessary for solving the optimization model are presented. Results of the operation of the system, including optimal operating policies for the reservoirs and reliabilities of the operation, illustrate the major advantages of the reliability programing approach compared with other stochastic optimization techniques.