But the combination of onsite inspections and gathering data remotely through a Scada system (see below) is essential for maximising a turbine's performance.
To make sense of data collected by Scada, a warranted power curve is usually used as a basepoint. This is a data curve, supplied by the turbine maker, that predicts the hypothetical performance of its machine during the initial warranty period. The power curve is based on the turbine's predicted performance at an ideal site, watered-down to make it more realistic.
As each wind turbine may depict variations due to differences in its components or surroundings, the warranted power curve alone is insufficient for understanding performance and a historical reference for the specific wind turbine is required. This reference can be defined using an earlier period when the wind turbine was operating as expected after filtering out unwanted data. It is important to ensure that probable software or hardware changes have not affected the operation of the wind turbine in relation to the historical reference.
Historical references can also be used for aspects that may affect a turbine's performance, such as the temperature of the components and control parameters. In addition, the references can be applied between identical wind turbines, to define potential performance increase or a case of underperformance.
As a wind turbine is a complex assembly of electric, hydraulic and mechanical systems, the amount of information that can be gathered is huge, requiring a structured database and standardised analysis methods. Performance data is either continuous when it comes from Scada or periodical when gathered from inspections.
Continuous information can provide indications on the performance of the turbine's control system itself as well as the status of its components. For example, a wind turbine whose pitch parameters deviate from the optimal settings would result in high revenue loss.
Periodical information is gathered from inspections that can vary from general wind-turbine checks to specific gearbox endoscopic inspections, oil analysis and vibration measurements. Like Scada, this information must be collected in a structured manner to be available for reviewing when needed. It is valuable for identifying possible root causes when issues appear in the Scada system. A high gearbox temperature that leads to sequential stoppages during high winds can be directly attributed to a gear oil pump that is found to malfunction during inspection.
The combination of this information is also essential to identify issues that may either not be visible in the Scada system or be misinterpreted by sensors. For example, a misplaced ambient air temperature sensor will lead to wrong measurements and, consequently, an error in the normalisation of the power output in relation to wind speed.
HOW DOES IT WORK? SCADA AND CONDITION MONITORING SYSTEMS TO ASSESS TURBINE PERFORMANCE
A supervisory control and data acquisition (Scada) system collects electronic data from sensors fixed to a wind turbine and fed to a central computer. It is usually supplied when the turbines are installed to ensure owners and operators can access statistical data on turbine performance.
The sensors monitor results such as power output and blade pitch changes, averaging the data over ten-minute periods. It can also show signals from voltage and temperature sensors, or vibration data. This low-frequency data may also contain maximum, minimum and standard deviations, and the Scada information can also contain alarms triggered by predefined limits set by the maker.
The periodical Scada system can be complemented by high-frequency data from condition monitoring systems (CMS). These assess the status of turbine systems and sub-systems using sensors such as accelerometers, strain gauges, and oil particle counters. The signals from these instruments provide data in very short time frames — less than a second. Because each CMS instrument is standalone, the temperature and alarm data, for example, provided from the Scada system can be used to correlate possible deviations in the CMS readings and indicate root causes.
Filippos Amoiralis is a project engineer at Mecal Wind Farm Services