Posters

Abstract

To assess the wind resources for a wind farm project, on-site wind measurements are extrapolated to the long-term period using a MCP method. The measured data often contains gaps generated by instrument or data logger failures, especially for offshore wind measurements. This increases the uncertainty of the long-term extrapolation, which leads to inaccurate energy yield estimations, increasing the financial risks of the project. To avoid this, costly measures are taken, such as robust measurement set-ups, device monitoring, and redundancies. Measurement gaps are usually compensated for by extending the measurement campaign until the total amount of wanted data is obtained. Doing this does not only increase the cost of the campaign, but also means that not all seasons are equally represented in the data, leading to a biased wind resource assessment. These are the reasons for a growing interest in filling the gaps as an alternative. Traditionally, measurement data gaps are evaluated based on their effect on the measured time series. Instead, the present study evaluates gaps based on how they impact the final product of the wind resource assessment: the long-term extrapolation. The goal is to provide guidance on what gap lengths, amounts of gaps and overall data availability losses are acceptable regarding their influence on the wind resource assessment outcome. Furthermore, the mitigation of the gap effect by filling the gaps with data from a nearby site corrected with an MCP method is investigated. Previous studies concluded that the effect of single 30-day gaps on long-term extrapolations is minor and that it is not reduced by filling the gaps. We generalize those findings by investigating multiple cases of gaped measurements and applying various MCP methods. Analyses are performed on 2 years of concurrent data from three offshore sites in Europe. The effects on 20-year extrapolations of multiple gaps of various lengths are analyzed. This is done by introducing artificial gaps into the measured data and comparing the long-term extrapolated statistical parameters of the gaped and the original measured data. Randomly distributed gaps are introduced and their effect on the long-term extrapolated wind statistics is analyzed. This procedure is repeated until the average gap effect can be recognized. Three methods with different performances are investigated for filling the gaps. The long-term extrapolation is done using a widespread linear regression based MCP method. It is found that the effect of gaps on long-term extrapolations can be significant depending on the season in which gaps occur, the overall availability loss and the site. Therefore, the minimum acceptable data availability is to be assessed for each specific case following a procedure such as the one demonstrated in the present study. Filling the gaps reduces their impact on long-term extrapolations in two cases: * The gap-filling reference data has a better correlation to the measurement than the long-term reference data. This happens when nearby short-term data is available. * The MCP method used for gap-filling has a better performance than the MCP method used for long-term extrapolating.