Development of probabilistic operating rules for Hluhluwe Dam, South Africa

Abstract

Hluhluwe Dam, with a 30 million m3 reservoir that supplies water for irrigation and Hluhluwe municipality in Kwa-Zulu Natal Province, South Africa, was consistently experiencing low storage levels over several non-drought years since 2001. The dam was operated by rules of thumb and there were no records of water releases for irrigation - the main user of the dam. This paper describes an assessment of the historic behaviour of the reservoir since its completion in 1964 and the development of operating rules that accounted for: i) the multiple and different levels of reliability at which municipal and irrigation demands need to be supplied, and ii) inter-annual and inter-decadal variability of climate and inflows into the dam. The assessment of the behaviour of the reservoir was done by simulation assuming trigonometric rule curves that were optimized to maximize both yield and storage state using the SCE-UA method. The resulting reservoir behaviour matched the observed historic trajectory reasonably well and indicated that the dam has mainly been operated at a demand of 10 million m3/year until 2000 when the demand suddenly rose to 25 million m3/year. Operating rules were developed from a statistical analysis of the base yields from 500 simulations of the reservoir each using 5 year-long stochastically generated sequences of inflows, rainfall and evaporation. After the implementation of the operating rules in 2009, the storage state of the dam improved and matched those of other reservoirs in the region that had established operating rules.