Application to the Des Moines River of Multiple Reservoir Operating Strategies Incorporating Short and Long Term Information in Real Time

摘要

Optimal reservoir operation is a complex and important problem. The problem of optimizing the operation of two Des Moines River, Iowa reservoirs (Saylorville and Red Rock), was segregated into four subproblems for reservoir operation: during normal flow period, during minor and major floods, and under drought conditions. Each subproblem reflects a different reservoir operation mode based on the hydraulic situation. Two particular problems were solved. The first dealt with the means for determining the optimum operation of Saylorville Reservoir under drought conditions, and illustrates the use of stochastic dynamic programming. The second problem was that of conjunctive operation of both reservoirs during a period of minor floods, and illustrates the use of Linear Quadratic Gaussian control in the operation of multi-reservoir systems, leading to linear release rules.