The digitalization of wind energy production is generating extraordinary levels of data from wind energy assets. Modern wind turbines can generate upwards of 50,000 or even 80,000 data tags every 10 minutes.
Making use of the resulting visibility is, however, beyond the capacity of most small or mid-sized wind owners and operators. Busy maintaining expensive assets and meeting customer demands, these owners and operators often plan maintenance on scheduled programs and rely on their turbines to perform according to manufacturers’ specifications.
Advanced data analytics can help when deriving value from these large data sets, using machine learning algorithms to make sense of the data and generating actionable insights for operators in a cost-effective way. These insights can identify potential issues of concern, identify when minor equipment adjustments are warranted, and reduce unnecessary downtime through better maintenance planning and execution. The value of the resulting improvements may not be game changers individually. Yet collectively and over the lifetime of assets, they add up to significant savings and increased output.
The first step in analytics is the creation of an accurate model of turbine performance for each turbine. These performance models, also known as performance digital twins, are trained using subsets of data points from periods of normal operating conditions. They allow for the detection of small changes in performance, as compared to the baseline. Further, this baseline also provides the basis for any loss estimates during faults and service outages.
Information about anomalous operational behavior can help operators identify a problem that, had it not been caught, could have impacted efficiency until annual or semi-annual scheduled inspections. For example, environmental erosion of blade leading edges will reduce the aerodynamic efficiency of the blade. This damage can result in as much as a 2% drop in energy capture. Analytics can identify compensatory control behaviours connected to the operation of the turbine that signal this sort of problem, which could prompt an operator to inspect the blades sooner than scheduled.
Other insights derived through analysis involve the exposure of changes in operational behaviour that, if not caught early, can cascade in their impacts and related costs. Modern wind turbines are all equipped with thermal management systems to control and eliminate waste heat. As the efficacy of these systems begin to degrade, early intervention can return them to full performance before degradation results in a loss in performance. Similarly, small changes to the thermal behaviour of a subsystem can also indicate increasing inefficiencies in that system. Early intervention in these cases can guide early repairs that can reduce or even avoid downtime.
Analytics also provides owners with ongoing visibility into maintenance contractor performance. The continuous flow of information can reveal hidden and costly asset downtime. Meteorological models of the wind farm serve as an independent source of wind speed and direction data for those turbines whose monitoring systems have been shut off. These models, when linked to a performance digital twin, can shed light on the quantity of energy that wasn’t captured during the monitoring outage. Highlighting the frequency and magnitude of these service-related production gaps can be the first step to closing the “no data available” loophole.
Like any analysis, the insights produced through analytics and machine learning can vary significantly in quality depending on the inputs. The involvement of technical experts in wind energy who understand and can filter relevant data are just as important as skilled data analysts. As many facilities don’t have the necessary human, technical, or financial resources in house, it is worthwhile to call on the help of some outside experts. When assets include dozens or even hundreds of turbines, the accumulated gains in efficiency and output will offset costs of adoption and add up to some serious money that is no longer being left on the table, or left to simply blow downwind.