This Lessons Learnt Report outlines the key learnings Fulcrum3D has learnt in the period between October 2019 – April 2020.
Report extract
- Our forecasting models continue to be more accurate than ASEFS such that they pass AEMO’s accuracy requirements with respect to RMSE and MAE on 5-minute power production.
- The MP5F API is increasingly robust. Further streamlining of the MP5F API registration and authentication processes is an area for improvement. The suggestion of some worked examples is an excellent idea.
- Accurate power modelling is non-trivial but provides bonus insights. It confirms our original hypothesis that a detailed physical based models are essential for accurate forecasting.
- There is currently minimal value in using forecasts for FCAS participation or export control but further increases in FCAS costs may strengthen the case for using forecasting for these services.