The proposed model for the wind speed deficit in wind farms is analytical and encompasses both small wind farms and wind farms extending over large areas. As is often the need for offshore wind farms, the model handles a regular array geometry with straight rows of wind turbines and equidistant spacing between units in each row and equidistant spacing between rows. Firstly, the case with the flow direction being parallel to rows in a rectangular geometry is considered by defining three flow regimes. Secondly, when the flow is not in line with the main rows, solutions are suggested for the patterns of wind turbine units corresponding to each wind direction. The presentation is an outline of a model complex that will be adjusted and calibrated with measurements in the near future. Copyright © 2006 John Wiley & Sons, Ltd.

No abstract

No abstract

The economic feasibility of offshore wind power utilisation depends on the favourable wind conditions offshore as compared to sites on land. The higher wind speeds have to compensate the additional cost of offshore developments. However, not only the mean wind speed is different, but the whole flow regime, as can e.g. be seen in the vertical wind speed profile. The commonly used models to describe this profile have been developed mainly for land sites. Their applicability for wind power prediction at offshore sites is investigated using data from the measurement program Rødsand, located in the Danish Baltic Sea.Monin-Obukhov theory is often used for the description of the wind speed profile.From a given wind speed at one height, the profile is predicted using two parameters, Obukhov length and sea surface roughness. Different methods to estimate these parameters are discussed and compared. Significant deviations to Monin-Obukhov theory are found for near-neutral and stable conditions when warmer air is advected from land with a fetch of more than 30 km. The measured wind shear is larger than predicted.As a test application, the wind speed measured at 10 m height is extrapolated to 50 m height and the power production of a wind turbine at this height is predicted with the different models. The predicted wind speed is compared to the measured one and the predicted power output to the one using the measured wind speed. To be able to quantify the importance of the deviations from Monin-Obukhov theory, a simple correction method to account for this effect has been developed and is tested in the same way. 2IIVKRUH ZLQG UHVRXUFH /DQJH HW DO page 3 of 64The models for the estimation of the sea surface roughness were found to lead only to small differences. For the purpose of wind resource assessment even the assumption of a constant roughness was found to be sufficient. The different methods used to derive the Obukhov length L were found to differ significantly for near-neutral and stable atmospheric stratification. Here again the simplest method using only bulk measurements was found to be sufficient.For situations with near-neutral and stable atmospheric stratification and long (>30 km) fetch, the wind speed increase with height is larger than what is predicted from Monin-Obukhov theory for all methods to estimate L and z 0 . It is also found that this deviation occurs at wind speeds important for wind power utilisation, mainly at 5-9 ms -1 .The power output estimation has also been compared with the method of the resource estimation program WAsP. For the Rødsand data set the prediction error of WAsP is about 4%. For the extrapolation with Monin-Obukhov theory with different L and z 0 estimations it is 5-9%. The simple wind profile correction method, which has been developed, leads to a clear improvement of the wind speed and power output predictions. When the correction is applied, the error reduces to 2-5%.

A statistical method for automatic quality control of measurement data with distributions close to Gaussian is presented. For each data point a prediction is made, based on the mean, variance and point-to-point correlation of the time series. The predicted value is compared with the actual value and if the difference between the two is 'large' then that data point is either marked as an outlier or replaced by the forecast value. Four different prediction methods are tested and, using the best one of these, the method is tested using artificial and real turbulence data.