IceLossForecast – wind farm icing loss forecasts
Minimize forecasting imbalance costs due to icing
Wind farm in cold climates
Winter is coming – and there’s money to save
In cold climates, icing on wind turbines is a major cause of production losses and thus loss of revenue. Generic weather services are not built for forecasting icing events. That’s why experts in the operational wind energy service sector such as balancing responsible parties (BRPs), wind farm operators, and transmission service operators (TSOs) rely on IceLossForecast – a state-of-the-art tool tailored for forecasting icing losses for wind farms.
With our high-resolution icing forecasts, you gain a competitive edge in both intraday and day-ahead trading, while staying ahead of turbine downtime and reduced power output.
Trusted by leading operators and BRPs across the Nordics
- Operational forecasts are used by +15 clients and nearly 50 wind farms across the Nordics
- Trusted by Fingrid, the Finnish TSO, since 2023
Business case
For a typical wind farm in Finland with average long-term icing losses on 4 % of annual energy production (AEP), using IceLossForecast reduced imbalance costs by an average of €21,000/year over 8 years – a total of €168,000.
- Payback time: 8–10 weeks (conservative estimate) and 3–4 weeks for winters 2022/2023 and 2023/2024
Your Questions Answered
How are the forecasts generated?
We use a specialized in-house setup where we combine different global models (GFS, ECMWF) and downscaling the global forecasts to local conditions. Optimized for cold climates, it estimates the icing mass for each turbine blade, and the wind farm icing loss is calculated using both physical (using our well known and extensively validated IceLoss methodology) and machine learning based models.
Is it difficult to integrated the forecasts into existing production forecasting systems?
The forecasts can easily be integrated to any existing production forecast system as the icing loss is provided as a icing loss factor (a value between zero and one). For example, if the forecast predicts icing losses of 15 % for tomorrow at 12:00, then the existing production forecast value is multiplied by an icing loss factor of 0.85.
How often are forecasts updated?
Every 6 hours by default. With real-time SCADA tuning, updates are available hourly (24 times/day).
What’s the forecast horizon?
Forecasts cover every hour for at least 48 hours ahead. Extended 7-day forecasts are also available.
How is uncertainty handled?
We use multiple weather models and a unique ensemble method to capture uncertainties in forecasted temperature, cloud placement and more. Our forecasts include deterministic values, ensemble means, min/max ranges and the probability of icing losses.
Can forecasts be tailored to my wind farm?
Absolutely. We offer both historical and real-time SCADA tuning, continuously updating forecasts with the latest data from the turbines.
How accurate are the forecasts?
The accuracy and chosen key performance indicator (KPI) depends on your goals – whether it’s minimizing imbalance costs, minimizing production forecast mean absolute error (MAE) or pinpointing high-loss icing events. We offer historical forecasts using our latest model setup, so you can evaluate performance before committing. We are disappointed if the correct hit-rate of icing events falls below 70 %.
How are forecasts delivered?
Via API or SFTP (CSV format). We support seamless integration and offer assistance with automation.
Want to see how IceLossForecast can work for you?
Please do not hesitate to contact us for more questions!
Mona Kurppa
Product Leader & Senior Adviser
Christoffer Hallgren
Senior Adviser