Optimization of real time operation of a multiple‐reservoir system

This paper addresses a typical requirement of real time multiple‐reservoir operation. A methodology is evolved utilizing a form of dynamic programing (DP) for the selection of an optimal reservoir storage policy path through a specified number of policy periods, and a linear programing (LP) routine is used for period by period optimization. The policy periods, in case of the real time operation, correspond to the number of time periods for which a streamflow prediction model is reasonably reliable. The method results in relatively easy computer implementation and high flexibility in that changes in constraints are simply made and additional reservoirs easily added. The method is of considerable value for the determination of release policies in real time operation of an existing system. As an example, application is made to a real system of reservoirs and hydroelectric facilities associated with the Shasta and Trinity sub‐systems of the California Central Valley Project.