A simulation–optimization algorithm for reservoir capacity calculation: the effect of inflow data set length

Abstract

A simultion–optimization algorithm for sizing potential reservoirs on a river basin has been used to demonstrate the effects of the streamflow record length on capacity requirements. The sizing procedure involves the generation of synthetic streamflows and a reservoir reliability analysis. Two levels of reliability are present: the reliability in meeting water supply requirements, and the probability level associated with the final capacity selection. The procedure was applied to synthetic sequences generated from an 81-year inflow record and subsets of this sequence to produce frequency distributions of required capacities. The capacities with a 5% probability of exceedence were compared. Relying on short inflow sequences can lead to substantial overestimations or underestimations of capacity. A key factor is the presence of the critical period in the analysis. An open question remains regarding the effect of the final probability level selection on the range of capacities observed. Key words: design reservoirs, water supply, simulation, optimization, reliability, record length, synthetic streamflows.