Task 19 proposes a method that is robust, easily adaptable, filters outliers automatically and does not assume a normal distribution of the SCADA data for individual turbines and wind farms. The proposed method uses percentiles of the reference, non-iced power curve in combination with temperature measurements. Ice build-up on turbine blades gradually deteriorates the power output (or results to overproduction to iced anemometer) so for increased accuracy the method uses three consecutive 10-minute data points for defining start-stop timestamps for icing events. In other words, the turbine rotor is used as an ice detector. Iced turbine power losses are defined by comparing the performance to the calculated power curve using heated anemometers from nacelle and the measured reference, expected power curve. Production losses are separated into 2 categories: operation and standstill losses due to icing.
On a general level, the method can be divided into 3 main steps:
- Calculate reference, non-iced power curve
- Calculate start-stop timestamps for different icing event classes
- Calculate production losses due to icing
Below is the list of required SCADA data signals used as input for the production loss calculation method.