Asset Publisher

TransAlta Inflow Ensemble Forecasting System (IEFS)

TransAlta is a power generation and wholesale energy marketing company with more than 100 years of industry experience. TransAlta has 27 hydropower plants, 13 of which are located in Alberta under a Power Purchase Arrangement (PPA) obligation. The 13 PPA plants primarily serve to provide electricity during periods of peak electrical demand. Water management is also a key concern from flood mitigation for stabilizing flow rates through towns such as Canmore, Drayton Valley, Rocky Mountain House and cities like Calgary and Edmonton.

TransAlta had an inflow forecast system based on deterministic forecasts run through the Raven modelling framework to run the University of British Columbia Watershed Model (UBCWM) model. TransAlta had the desire to add ensemble forecasting capabilities to improve their inflow forecasting and knowledge of uncertainty for use in short- and long-term generation forecast. In this project the existing operational functionality of the inflow modelling forecast process was migrated to Delft-FEWS along with process improvements. 

Key process improvements for the new Inflow Ensemble Forecasting System (IEFS) system are:

  • Raven integration with Delft FEWS, including state and parameters updating as well as model state visualization via the FEWS GUI.
  • Ensemble forecasting using several meteorological ensembles and Ensemble Streamflow Prediction (ESP) for a longer term outlook
  • Use of FEWS Web Service to share forecast data internally.
  • RTC Tools 2 reservoir modelling and pilot flow optimization.

TransAlta is now running a system (Delft-FEWS + Raven) which is used by an increasing user community, including hydropower industry peers and governmental forecasting agencies. As part of this process, TransAlta has contributed to the shared development of Raven model adaptors to Delft-FEWS. The introduction of RTC Tools is expected to allow better use of forecast data to provide operational guidance and allow TransAlta to more fully utilize additional information from ensemble forecasting.