Keywords
water resources, uncertainty, dss, scenario analysis, pro-active approach
Start Date
1-7-2006 12:00 AM
Abstract
In water resources management problems, uncertainty is mainly associated with the value of hydrological exogenous inflows and demand patterns. Deterministic models are inadequate to represent these problems and traditional stochastic optimization models cannot be used if there is insufficient statistical information to support the model. In this paper the uncertainty is modelled by a scenario approach in a multistage environment which includes different possible system configurations in a wide time horizon. A robust chance optimization model is used in order to obtain a so-called barycentric value with respect to decision variables. The successive reoptimization step, based on this barycentric solution, allows reducing the consequences deriving from a wrong decision. The improved version of WARGI DSS performs scenario analysis by identifying trends and essential features on which to base a robust decision policy. The current version of WARGI can be linked to commercial solvers as well as to some free solvers such as IdrScen. IdrScen is a new package for large dimension problems based on open source philosophy, that exploits the speed of network simplex methods in order to obtain very efficient solutions to the scenario problems. Moreover, the application to a real water resource system in Sardinia, Italy, shows the usefulness of the scenario analysis in water resources problems affected by a high level of uncertainty in data input. It appears that IdrScen can be a promising alternative tool to commercial codes for large size optimization problems coming for complex real resource systems.
Scenario Analysis in Water Resources Management Under Data Uncertainty
In water resources management problems, uncertainty is mainly associated with the value of hydrological exogenous inflows and demand patterns. Deterministic models are inadequate to represent these problems and traditional stochastic optimization models cannot be used if there is insufficient statistical information to support the model. In this paper the uncertainty is modelled by a scenario approach in a multistage environment which includes different possible system configurations in a wide time horizon. A robust chance optimization model is used in order to obtain a so-called barycentric value with respect to decision variables. The successive reoptimization step, based on this barycentric solution, allows reducing the consequences deriving from a wrong decision. The improved version of WARGI DSS performs scenario analysis by identifying trends and essential features on which to base a robust decision policy. The current version of WARGI can be linked to commercial solvers as well as to some free solvers such as IdrScen. IdrScen is a new package for large dimension problems based on open source philosophy, that exploits the speed of network simplex methods in order to obtain very efficient solutions to the scenario problems. Moreover, the application to a real water resource system in Sardinia, Italy, shows the usefulness of the scenario analysis in water resources problems affected by a high level of uncertainty in data input. It appears that IdrScen can be a promising alternative tool to commercial codes for large size optimization problems coming for complex real resource systems.
