A Study Investigating the Inter-Correlation of Wind Speed and Turbulence on the Accuracy of SCADA Based Wind Turbine Blade Load Reconstructions

Abstract

The structural health of wind turbines is of great importance. Wind turbines are often sited at remote locations and repair jobs are only possible under certain weather conditions. If major parts are failing as for instance the gearbox or wind turbine blades not only weather and spare part availably needs to be considered, in addition are in general cranes or heavy lifting vessel needed. Wind turbines are equipped with a vast amount of sensors controlling the performance and power production of the turbine. Those sensors are not only driving the internal controller they are as well an essential part of the wind turbine reporting. The sensors provide the input parameters for the Supervisory Control and Data Acquisition (SCADA) System, a system transmitting in 10-minute intervals about 200 instances to the turbine operator. The SCADA system provides performance related attributes such as the generated power, the state of the turbine, the availability of the grid, a description of the wind field as well as control related attributes summarizing the activities of the yaw or pitch system [1]. A connection between the available operational data and the experienced loads, allows us to monitor the loads on the base of the SCADA data, data that is available for each and every commercial wind turbine. Load information can be used for advanced turbine control, to forecast incipient faults and to re-evaluate the remaining wind turbine lifetime. The relationship has a complex nature; wind turbine loads are driven by the inflow conditions such as turbulent, non-uniform or skewed inflow. In addition is the relation between wind and loading biased by the turbine control. Pervious research showed that it is possible to reconstruct the experienced loading if SCADA data is available [2]. The reconstruction process is based on statistics describing a time interval of 10-minutes. As for instance the wind field, which is here described by the average, max, min and standard deviation, calculated over a 10-minute interval. For load simulations is according to [3] is the assumption of a constant level of turbulence (per 10-minute bin) reasonable, at least for a wind speed ranging between 5 and 20 m/s. But how does the constant wind speed and turbulence intensity effect the reconstruction process, in particular for those wind regimes where the simplification is not reasonable? Here the data has been binned according to wind speed and turbulence intensity to test the impact of wind speed and turbulence intensity on the precision of the reconstruction.