Presentations

Abstract

In Post Construction Yield Assessments (PCYAs), the industry best practice relies on estimating a turbines potential production (Gross) adjusted with a long-term correction to a reference period. A crucial part of undertaking a PCYA is the calculation of losses experienced by each individual turbine for the investigated historic period. The primary source of wind data for the PCYA is the nacelle anemometer, which is not reliable for accurately measuring a turbine's power curve. But often it's the only data signal available, thus PCYAs use a wind turbine's historic power curve "under normal operation" as a benchmark for evaluation of the lost production for periods of partial performance or production stops due to various reasons. We will present a method for quantifying the related uncertainty of individual turbine's calculated losses by means of bootstrapping applied on random subsets of valid normal operation data, where the actual kWh produced is known and describe the method of how to select and quantify the ratio between actual samples and the predicted values following the best practice for PCYA loss estimation methodology. Furthermore, for sites under icing with a "frozen" or zero wind speed reading or in Grid loss conditions, where data is missing, we will demonstrate how to identify the second best windspeed data signal available by using the main methodology of the FGW - TR10 guideline and by substitution of the second-best alternative windspeed signal how to quantify the uncertainties on predicted potential production during conditions where original nacelle windspeed signal are faulty or missing. The method will be described to an extent where the audience will be able to re-produce the methodology, with the hope of it becoming best practice in future for PCYA on how to quantify the related uncertainties on the calculated lost production.