Predictive maintenance is a hot topic as the natural evolution of wind farming technology will heavily support the preventive and corrective procedures already in place.
This project aimed to give our client the ability to:
● Be able to take proactive decisions regarding turbine maintenance.
● Provide a Real-time overview of the health of turbine components
and help better predict potential downtime thereby increasing
turbine uptime and production.
● Provide analytics to support optimal turbine placement and angles.
● Provide emergency alerts.
The detection of underperformance, for its part, would make it possible to considerably improve the efficiency of the turbines by focusing on operational parameters that are easy to intervene on and by detecting in real-time production anomalies at the level of the machines, which included:
● Degradation of the power curves.
● Improvement of the alignment of the turbines concerning the wind.
● Optimal exploitation of the park’s deposit and more.
The power curve is a graph that represents the output power of a wind turbine at different wind speeds. It is developed from measurements made on-site by attaching an anemometer to the wind turbine itself or by using a measuring mast independent of the turbine for wind measurements.
We can distinguish 4 parts on this curve:
- From 0 to the starting speed (here 5 m / s): the output power is zero, the wind is not strong enough to cause the rotation of the rotor from the starting speed to the nominal speed (here 15 m / s): the output power increases until it reaches the nominal power (here 750 kW)
- From nominal speed to breaking speed (here 25 m / s): the output power is maintained at the nominal power almost constant thanks to the regulation device.
Degradation of the power curve
Turbines over time experience a deterioration in efficiency (for various reasons), which is reflected in a degradation of the power curve over time. The need here was to set up a methodology for predicting this degradation, which would serve as a reference to assess the condition of the turbines and to identify the causes linked to abnormal degradation. This model will consider different aspects of the operation of the turbine but also those related to maintenance:
● Hours of operation since commissioning.
● Impact of singular events such as a major fix (change of a majorpart).
● Impact of cyclical events (annual preventive maintenance).
Detection of clamps without alarms
Power curve analysis reveals “unexpected” and unlisted underperformance points via alarms. As illustrated in the chart below, this turbine experienced a period of “clamping” (the dots framed in red) where it should have produced nominal energy. These periods of clamping should normally be notified by the monitoring system, but in some of them we have what are called clamps without alarm. The solution can predict those clamps and report them back in real-time even if the monitoring system doesn’t.
Detection of underperformance (improvement of operational parameters of pitch, yaw)
The nacelle rotation mechanism, “yaw mechanism,” uses electric motors to turn the nacelle with the rotor and present the propeller to the wind. The rotation mechanism is actuated by an electronic controller which detects the wind direction using the vane placed at the level of the turbine.
In the case of strong winds, wind turbines are equipped with aerodynamic braking systems to reduce the forces exerted on the propeller blades. Several techniques are used. The pitch control blade tilt system, through a hydraulic control and a control station, rotates the blades to orient them to the wind.
● Detection of performance anomalies linked to an incorrect pitch / yaw angle configuration (outlier positions)
● Detection of underperformance linked to a less than optimal configuration of the pitch / yaw systems or degradation of the initial configuration
● Identification of KPIs such as the shortfall and recommendations for parameter adjustments for certain speed ranges and wind directions