Development of an uncertainty model for wind resource assessment techniques in complex terrain based on short scanning lidar time series

Background and motivation

The profitability of a wind farm project is determined by the expected annual energy production (AEP) of the wind turbine(s) at the potential site. In order to estimate the expected AEP, the wind resource must be known. An estimate of the wind resource can be obtained through a so-called wind resource assessment (WRA). The result is the long-term mean wind speed at the turbine(s) hub height, the wind direction probability distribution and sectorwise wind speed probability distribution. These values can then be combined with the turbine(s) power curve(s) and expected energy losses to calculate the expected AEP. It is therefore essential that thorough wind resource assessments are conducted prior to developing a new wind farm project.

Many different techniques for estimating the wind resource exist. Traditional WRA methods, such as the Wind atlas methodology used in WAsP, are typically based on time-series of wind measurements collected by cup-anemometers mounted on metrological masts (met masts). Ideally, the time-series should be representative of the wind climate over the turbine lifetime (typically 20 years), but as it is very expensive to carry out measurement campaigns of such durations, the wind is measured over a much shorter period, usually over one year. This data can then be long-term corrected (LTC) to represent the wind climate over a much longer period using measure-correlate-predict (MCP) techniques. Furthermore, in areas with complex terrain, traditional WRA lead to large uncertainties as a single met mast cannot capture the complexity and spatial variability of the flow. To achieve the desired accuracy, several met masts would have to be erected, which is quite costly.

A new project called RECAST started in February 2018 with the aim of making wind farm projects in complex terrain bankable by reducing the assessment time and uncertainty of the WRA. In the RECAST approach, the wind is measured using a WindScanner system, consisting of two or three synchronized lidars. The WindScanner system can thereby measure the wind at many positions and heights around a potential site without being moved. This is expected to lead to lower uncertainties in the horizontal extrapolation of the wind measurements to the turbine site(s). As it is
expected that the WindScanner system will reduce the uncertainty of the estimated wind resource in complex terrain, the measurement time could be reduced to make the wind farm project even more bankable.

Research objectives

This PhD project is carried out in collaboration with RECAST. The aim of the PhD project is to develop an uncertainty model that combines the uncertainty terms that differ between the traditional and RECAST WRA techniques. Using this model, it is then possible to compare and quantify the uncertainties in the gross AEP resulting from the differences in the two WRA techniques. The uncertainty sources that are expected to differ between the two WRA techniques include the uncertainty due to long-term correction, the uncertainty due to horizontal extrapolation, the uncertainty due to instrument type and the uncertainty due to data availability.

Contact

Elin Svensson
PhD student
DTU Wind Energy
+45 93 51 19 88