Predicting Wind Farm Production and Icing Losses with AFRY DAPS

Based on AFRY’s Winterwind 2026 contribution by Helmi Uusitalo

Icing: the most underestimated challenge in cold climate wind energy

For wind farms operating in cold climates, icing is one of the most disruptive factors affecting production. When ice forms on turbine blades, energy output drops sharply and in some cases stops completely. The technical challenge is well known. But the real problem is not icing itself; it is the inability to predict when icing will occur, how severe it will be, and when it will end. This unpredictability leads to:

• high imbalance and regulating costs
• inaccurate day ahead bidding
• unnecessary downtime
• reduced grid stability

AFRY has developed a solution: AFRY DAPS, an advanced forecasting method that integrates physics-based icing and normal operation production models, SCADA-driven parameterization, and weather calibration models. Together, these capabilities dramatically improve next-day production forecasts for wind farms in cold climates.

A new approach to cold climate wind forecasting

AFRY DAPS combines four modelling pillars, each supporting high-accuracy predictions.

1. Physics-based icing modeling

AFRY uses ice accretion physics to model how ice forms on blades under real atmospheric conditions. The model incorporates:

• in-cloud and precipitation icing
• glaze, rime, and wet snow formation
• turbine-specific model optimization and parameterization
• local weather inputs and microclimate variation

This enables detailed simulation of icing behavior, a critical foundation for predicting production losses.

2. Data-driven parameterization

Each turbine behaves differently under icing. AFRY DAPS calibrates parameters turbine-by-turbine using:

• historical SCADA data
• real-time turbine measurements
• optimization models

This approach accounts for site-specific differences that pure physics or pure weather models cannot capture.

3. High accuracy normal- operation production forecast

Reliable icing-loss forecasting requires a precise baseline. AFRY DAPS uses:

• regression models
• power curve optimization
• weather calibration models

The result is accurate next-day normal-production predictions, a prerequisite for quantifying icing impact.

4. Weather calibration to correct forecast errors

Even the best model collapses without accurate weather data. AFRY’s calibration tool adjusts forecasts using observed deviations between predicted and actual weather conditions, significantly reducing model error.

The breakthrough: predicting ice release moments

One of the most important modeling advances is the ability to forecast the ice release point, the exact moment when ice begins to detach from the blade.

Predicting this moment accurately has an immediate financial impact:

Scenario - result
Ice release predicted too late - lost production and missed revenue.
Ice release predicted too early - high imbalance costs.
Ice release predicted correctly - optimized production, minimized economic losses.

AFRY’s model accounts for multiple interacting weather conditions and turbine behavior to locate this point with high precision.

Validation: up to 80% of icing events predicted

Model testing shows strong results across different wind farms:

• ~80% of icing events can be identified and predicted
• Normal-production forecasts match actual production with high accuracy
• Severe icing events can generate tens of thousands of euros in avoided regulatory costs

Years of development integrating physics, data, and weather intelligence have enabled AFRY to deliver one of the most accurate cold-climate wind-power forecasting methods available today.

From research to real-world implementation

AFRY’s modeling methods are led by Helmi Uusitalo (M.Sc., Engineering Physics), who has developed icing loss modeling both academically and in operational consultancy. Her combined background in atmospheric physics, turbine modeling, and SCADA analytics underpins the scientific rigour behind AFRY DAPS.

AFRY’s cross-disciplinary expertise, spanning meteorology, data science, engineering, and energy markets, ensures the solution supports not only technical forecasting, but also bidding, asset management, and grid integration decisions.

Conclusion: Icing will remain complex, but unpredictability doesn’t have to

AFRY DAPS offers a practical and scientifically robust way to reduce forecasting uncertainty in cold climate wind power. With improved accuracy in both normal production and icing loss prediction, operators can:

• stabilize their production forecasts
• reduce financial exposure
• optimize market participation
• improve grid reliability
• plan operations more efficiently

Icing may be a natural part of northern wind power, but with AFRY DAPS, its consequences no longer have to be unpredictable.