A real-world use of icing forecasts in short-term power trading

Wind turbine icing is one of the most significant weather-related risks for wind power producers and traders, frequently resulting in production losses, turbine downtime, and increased balancing costs. We spoke with Maarten Hofhuis, Portfolio Manager at Pure Energie, to learn how the company manages icing risks and how Dexter Energy’s icing forecasts and alerts have benefited their short-term trading operation.

How did you monitor and manage the risk of icing before implementing icing alerts?

Before using icing alerts, we mainly relied on general weather forecasts. If cold temperatures were expected, we’d consult with our asset management team to discuss potential downtime. Their assessments were typically based on past experience rather than specialized icing forecasts.

What kind of impact does wind turbine icing have on your operations and trading activities?

The financial impact of icing can be substantial. While not every event results in high balancing costs, a few notable ones each year can have a major effect. Including icing forecasts in our planning is essential to mitigate these financial risks.

Estimating the share of yearly balancing costs attributable to icing is challenging, but our initial indication suggests it could be between 5% and 10%.

Apart from financial losses, icing can also affect turbine safety and maintenance. Some wind farms are shut down preventively based on weather conditions to reduce safety risks.

What led Pure Energie to start using Dexter’s icing forecasts?

Several factors influenced our decision. For starters, we have a long-standing, positive collaboration with Dexter. Secondly, we had a clear need for accurate icing forecasts and for reliable volume calculations to anticipate downtime.

We also had the chance to provide feedback during the pilot phase, which helped shape the product to fit our operational needs.

We began using the forecast in the 2023–2024 winter season, around September or October. It’s now fully integrated into our operational workflow, and we feel well prepared for winter 2025.

How do you currently use the icing alerts in your day-to-day operations?

When an icing alert is received, our department discusses the situation internally and with asset management to determine the appropriate response.

To streamline this process, we are considering integrating these alerts into a Power BI dashboard for improved visualization and automation. Exploring API-based delivery instead of emails could further enhance efficiency.

What do you typically do when an icing alert is triggered?

The most common action is adjusting our bidding strategy on EPEX Spot. We feed Dexter’s adjusted forecast data into our bidding system to account for potential downtime. In most cases, we trust Dexter’s forecast and act accordingly.

How do you verify the accuracy of the icing alerts?

We cross-check the alerts with turbine downtime data from our asset management team. They track when turbines stop due to icing, allowing us to compare actual incidents against forecasts.

What have you learned about the trade-off between accuracy and risk?

We’ve found that it’s often better to forecast slightly more icing events than strictly necessary. Missing even one severe icing event can lead to large financial losses, so a cautious approach tends to yield better overall results.

We’re conducting a broader evaluation over a longer period to determine the overall impact on balancing costs. It’s likely that the outcome will be positive.

Would you recommend Dexter’s icing forecast?

Yes, definitely. Missing an icing event can cause significant losses. Having an accurate forecast allows us to anticipate these events and mitigate the risks effectively.