Optimal reservoir sizing with yield reliability.

A hybrid simulation-optimization model is presented for selecting amongst and sizing potential reservoirs on a river basin. The model considers monthly reservoir simulation to meet target system water supply annual yields with a specified reliability structure. Storage requirements at each site are functionally related to reservoir yields, the reliability structure, and a historical or synthetic flow record through a modified sequent peak algorithm. Optimal reservoir sizes are determined through nonlinear programming. The optimization component of the model explicitly solves for reservoir yields, and for at site annual supply reliabilities that are constrained to meet a contractually specified reliability structure for the system. The use of simulation to relate yields, storage capacities and streamflow leads to a significant reduction on the numertical size of the optimization model, and also allows for greater flexibility in modelling the complexities (e.g. in stream diversions, water rights structure, reach gains/losses) of the stream network. This paper extends the model of Lall and Miller (1988) by explicitly solving for annual at site yield reliabilities, rather than pre-specifying them. An example application to sites in the Jordan River basin in Utah is presented. (A)