A Cvar-Based Nonlinear Stochastic Model For Water Resources Management

Shao Liguo,Luan Shengji

2010-01-01

Abstract:A CVaR-based nonlinear two-stage stochastic programming (CNTSP) model was developed in this study for supporting water resources allocation under uncertainty. CNTSP model was formulated based on an integration of Conditional Value-at-Risk (CVaR) constraint into a two-stage stochastic programming (TSP) framework, and could be used to deal with uncertainties expressed as probability distributions. Most importantly, the measure of extreme expected loss on the second-stage penalty cost was incorporated into the model constraints, such that the trade-off between system economy and extreme expected loss could be conducted And the introduction of a nonlinear penalty function avoid the situation that single water user undertook large mount of expected risk loss. The developed model was applied to a water resources allocation problem. The result demonstrated that the CNTSP model could help decision makers obtain viable water allocation alternatives, and conduct trade-off analysis between economic benefit and risk control.