Page - 172 - in Book of Full Papers - Symposium Hydro Engineering

COMMISSION INTERNATIONALE DES GRANDS BARRAGES ------- VINGT-SIXIÈME CONGRÈS DES GRANDS BARRAGES Autriche, juillet 2018 ------- MODELLING THE UNCERTAINTY OF INFLOW TO SHORT-TERM RESERVOIR OPERATION USING FUNCTIONAL EXPANSION-BASED METHOD Duan CHEN Vice Dean of Trans-boundary River Research Department, CHANGJIANG RIVER SCIENTIFIC RESEARCH INSTITITUE Bo HU Vice Dean of Water and Soil Conservation Department, CHANGJIANG RIVER SCIENTIFIC RESEARCH INSTITITUE CHINA 1. INTRODUCTION Single trace of inflow forecast has been commonly used for guiding short- term reservoir operation and provided a better way of managing water resources, compared to traditional rule-based-only operations. However, uncertainty of the inflow information is a major challenge to the operator as the single trace of inflow forecast may deviate far from the reality due to forecast errors. To better represent the uncertainty associated with the inflow forecast, ensemble stream flow prediction [1], also called ESP is gradually used to replace the single trace inflow forecast. The ESP is essentially a collection of multiple inflow forecasts that are resulted from multiple runs of weather models. The advantage of the ESP is that the inflow uncertainty under largely variable weather condition e.g., climate change are accounted through different realizations of inflow. However, the usage of the ESP may lead to a high computational cost for optimization of reservoir operation due to uncertainty modeling. This study use a functional expansion-based method namely Karhunen-Loeve 172