Analytics for the wind industry have long leveraged data to generate insights on downtime. With advances in AI, wind operators now have an opportunity to more effectively address underperformance, yielding production increases not possible with reliability models alone.
Availability was a Necessary First Step
Traditional turbine monitoring providers have focused analytics on downtime for several reasons:
- Downtime is identifiable: Downtime is easily detectable and understandable, thanks to improved classification systems developed in anticipation of NERC-GADS. Underperformance, however, is more difficult to spot — especially with traditional power curve techniques that have a high signal-to-noise ratio.
- Downtime is measurable: Most warranties set targets for downtime rather than productivity, partly because it’s inarguable whether or not turbines are running. Given those incentives, service providers are more likely to take action on downtime, and the software they build and buy reflects that reality.
- Downtime is addressable: Major component failures eat into availability, are expensive, and jeopardize planned maintenance. As many owners move out of warranty and self-perform their operations and maintenance, they’re taking reliability even more seriously than they were before.
Reliability analytics have accelerated availability improvements across much of North America and Europe, with a typical wind owner now operating at 96-98 percent availability. In addition, with the Production Tax Credit-driven development and repowering of the last few years, availability has become a lower priority.
The Opportunity for Power Performance
As a result, many operators have turned to performance optimization as a high-priority area for operational improvement and production gains. In its engagements with customers, Uptake has found that many operators have room for as much as 2 percent in increased annual energy production (AEP). Since lost AEP issues are often trends across several turbines or event sites, gaining insight into root causes of underperformance like turbine derating or yaw misalignment can provide outsized production increases.
Challenges with Underperformance Analytics
Simple analytics approaches identifying underperformance monitoring often fail to provide ROI. That’s because most rely on the manufacturer’s design curve, not taking into account confounding variables like icing and constrained operations, or even anemometer degradation. These approaches generate true positives about 4 percent of the time; busy performance engineering teams don’t have time to sift through hundreds of supposed underperformance alerts. Even with accurate alerts, engineers can struggle to prioritize underperformance issues among individual turbine issues and system improvements — especially based on value.
Advanced analytics holds significant promise for addressing these challenges by benchmarking a turbine’s current production against historic performance and peer turbines, as well as the power curve. AI-powered automation and access to deep supporting evidence on underperformance issues can also help identify root causes and develop a plan of action, whether it's with internal operations teams or by stronger service provider management.
For most wind owners, performance and availability are two key pieces of their wind engineering and operations decisions. Unless they have high-quality analytics insights to inform both, they’ll continue to leave megawatt-hours — and, consequently, revenue — on the table.
By targeting underperformance and downtime using AI, our wind customers have increased AEP by up to 2%.