Why Wind Turbines Underperform
Many wind operators are still grappling with the imprecision of traditional power curve-based methods.
Underperformance of turbines remains one of the biggest challenges to operators of wind fleets. Most operators are still relying on legacy power curve-based techniques to address underperformance and have no automated way to drive revenue growth through improved turbine performance.
Reliance on traditional underperformance identification costs many operators up to 2 percent of their annual energy production (AEP). The Electric Power Research Institute estimates that just a 1 percent boost in productivity at a typical wind farm with 100 two-megawatt turbines results in revenue increases of $250,000 - $500,000.
Knowing that operators lose significant revenue due to underperformance, we've set out to explain why power curve-based techniques are leading to performance management strategies that fall short.
3 Problems with Power Curves
Power performance approaches are primarily based on the manufacturer’s design curve. If a turbine falls below an OEM threshold, reliability teams can only sometimes filter out likely reasons for underperformance including curtailment, icing, and anemometer degradation. This approach presents several problems:
1. Power curves can’t capture the full operating context
The design curve is based on the performance of wind turbines located in an area with minimal turbulence and average head-on winds. Factors that make up the total operating context — like confounding variables of air density, wake effects from turbine arrangement, or hilly terrain — are missing from the manufacturer’s curve that operators use to track underperformance.
By not accounting for these variables, operators get a partial and misleading picture of performance.
2. Power curves are turbine-specific, not operator-friendly
The OEM curve is a forecast of performance for a specific set of turbines. Over time, turbine models under management change. Site to site, and with regular redesigns and repowering, performance benchmarks that are tailored to specific power curves become a site’s model for power performance management.
With repowering and development initiatives, rendering analytics from one make or model of turbines to another becomes another important way to sustain productivity and build internal best practices around power performance management. Operators shouldn’t have to fall behind or re-train reliability teams when they buy more durable, better-performing turbines. Power performance management should have the flexibility to adapt with procurement and personnel decisions.
3. Power curves aren’t linked to value impact
The design curve is a technical reflection of what’s going on — it abstracts turbine productivity from business goals around preventing downtime, driving productivity, and performing cost-effective maintenance. For power performance management to earn enterprise-wide buy-in, operations and maintenance teams must have a shared understanding about the impact of specific underperformance issues on revenue.
Current power performance management is inefficient
These problems stick operators with alerts that make power performance a burden — on average, design curves deliver true positive alerts only 4 percent of the time. Because of this imprecision, performance and reliability engineering teams pour more time and money trying to isolate and understand true cases of underperformance.
Often, they cannot provide enough evidence to enable action or manage service providers, because they’re preoccupied with availability issues and already facing torrents of data at their sites. Power curve-based alerts aren’t reliable to maintenance teams, either. False positives often just turn up skepticism about future service requests.
In the event engineering teams can pull together evidence from an alert to enable service providers to take action, it becomes challenging to prove the importance of underperformance in dollar values. Performing service on underperformance at the right time demands that it’s prioritized among other competing maintenance issues.
Traditional underperformance techniques aren’t powerful enough to optimize maintenance and help wind operators get the most out of their turbines. The solution? AI for power performance gives engineering and operations teams a shared and precise view on how to optimize maintenance for underperformance issues based on root cause. That way, there is clarity about the bottom-line impact of addressing underperformance at any given time, and supportive evidence that enables smarter decisions and faster action.