The wind industry is being buffeted on all sides by shrinking subsidies, increasing competition, and soaring maintenance costs. High wholesale energy prices also mean that even incremental uplifts in productivity for older turbines could have a big effect.
Operational data from wind farms holds the secrets to dramatically boosting asset performance and power output, improving operating margins and lowering the levelised cost of electricity across the wind industry.
Bitbloom’s ML-driven approach achieves a step-change in the precision with which available power and performance changes can be measured that could transform wind industry productivity. The model was able to detect subtle shifts as small as 0.75% with 90% confidence over a 90-day test period. When compared to a purely statistical method without the use of ML, this could allow operators to identify a 1% dip in turbine performance within 50 days instead of 170 days, potentially delivering 4 months of increased productivity.