The Hybrid Energy Forecasting and Trading Competition challenged participants to forecast and trade the electricity generation from a 3.6GW portfolio of wind and solar farms in Great Britain for three months in 2024. The competition mimicked operational practice with participants required to submit genuine forecasts and market bids for the day-ahead on a daily basis. Prizes were awarded for forecasting performance measured by Pinball Score, trading performance measured by total revenue, and combined performance based on rank in the other two tracks. Here we present an analysis of the participants' performance and the learnings from the competition. The forecasting track reaffirms the competitiveness of popular gradient boosted tree algorithms for day-ahead wind and solar power forecasting, though other methods also yielded strong results, with performance in all cases highly dependent on implementation. The trading track offers insight into the relationship between forecast skill and value, with trading strategy and underlying forecasts influencing performance. All competition data, including power production, weather forecasts, electricity market data, and participants' submissions are shared for further analysis and benchmarking.
Abstract Wind resource measurement campaigns seldom span a period of sufficient length to characterize the long-term wind climate. Therefore, short-term measurements are correlated with long-term reference wind data to provide an unbiased prediction of the historical wind conditions. We present a regression neural network method that maps predictors from a numerical weather model to site-specific wind speed and wind direction, thereby producing representative long-term time series. Using offshore measurements at multiple, globally distributed sites we evaluate rolling training windows (6 months, 1 year, and 2 years) and validate against the unseen measurements. Comparison of the results is also done against uncorrected reference data and a baseline measure-correlate-predict implementation. The uncertainty is quantified through a representativeness-aware noise scheme. The time series corrected with the regression neural network exhibit improved fluctuation characteristics, reduced bias on the annual energy production, and realistic uncertainty estimates, indicating that machine-learning-based long-term correction is suitable for direct time series modelling of wind farms.
The geotechnical site investigation strategy for the development of an offshore wind farm - which often comprise more than 100 wind turbine genera-tors (WTGs) - employs a combination of both in situ testing and sampling bore-holes for classification and advanced laboratory testing. The piezocone cone pen-etration test (CPTu) is the main investigation tool and is conducted at each WTG foundation location to facilitate location-specific foundation designs. As such, a robust correlation between CPTu parameters and engineering properties needs to be established based on results of laboratory testing performed on high-quality samples. The effects of sampling on the triaxial behaviour of transitional soils have not been reported in the literature to the same extent as for clays. This paper explores the use of the gel-push (GP) sampler to retrieve intact specimens of tran-sitional soils to guide and inform the widely accepted practice of specimen re-constitution in the offshore industry – due to the difficulty in sampling in these types of soils. An onshore test site was selected because it provides more control on the operational procedures during sampling and in situ testing. Sample dis-turbance was assessed by comparing shear wave velocity measurements from seismic CPTu with values measured in laboratory tests. Triaxial tests were per-formed on intact and reconstituted specimens to evaluate the effects of fabric and structure on the constitutive response. The analyses show that the shear wave velocity of GP samples agree well with seismic CPTu measurements, indicating that the retrieved specimens are of good quality. In contrast, the intact and recon-stituted specimens exhibited dilatant and contractive behaviour, respectively, during triaxial testing indicating the importance of testing specimens mimicking the in-situ fabric and structure.
