Short-Term Stochastic Optimization of Hydropower Production and Marketing under Meteorological Uncertainty

Abstract:

Hydroelectric power systems are characterized by variability and uncertainty in yield and water resources obligations. Market volatility and the growing number of operational constraints for flood control, navigation, environmental obligations and ancillary services (including load balancing requirements for renewable resources) further the need to quantify sources of uncertainty.

This research presents an integrated framework to handle several sources of uncertainty. Main focus is on the meteorological forecast uncertainty represented by a probabilistic Numerical Weather Predictions (NWP), its consistent propagation through load and streamflow forecasts, and the generation of scenario trees with multi-dimensional distance metrics. The short-term management approach adapts to this forecast uncertainty and its resolution by the use of multi-stage stochastic optimization.

The Federal Columbia River Power System (FCRPS), managed by the Bonneville Power Administration, the US Army Corps of Engineers and the Bureau of Reclamation, serves as a large-scale test case for the application of the new framework. It is used to quantify the operational flexibility of the hydropower system and to propagate the forecast uncertainty through the system. Objectives and constraints enable the operating staff to shape and direct the system such that variability remains in the hydropower system or gets absorbs by the energy market.