Mapping the wind resource over UK cities

Decentralised energy sources, such as small-scale-wind energy, have a number of well-known advantages. However, within urban areas, the potential for energy generation from the wind is not currently fully utilised. One of the most significant reasons for this is that the complexity of air flows within the urban boundary layer makes accurate predictions of the wind resource difficult to achieve. Without sufficiently accurate methods of predicting this resource, there is a danger that wind turbines will either be installed at unsuitable locations or that many viable sites will be overlooked. In this paper, we compare the accuracy of three different analytical methodologies for predicting above-roof mean wind speeds across a number of UK cities. The first is based upon a methodology developed by the UK Meteorological Office. We then implement two more complex methods which utilise maps of surface aerodynamic parameters derived from detailed building data. The predictions are compared with measured mean wind speeds from a wide variety of UK urban locations. The results show that the methodologies are generally more accurate when more complexity is used in the approach, particularly for the sites which are well exposed to the wind. The best agreement with measured data is achieved when the influence of wind direction is thoroughly considered and aerodynamic parameters are derived from detailed building data. However, some uncertainties in the building data add to the errors inherent within the methodologies. Consequently, it is suggested that a detailed description of both the shapes and heights of the local building roofs is required to maximise the accuracy of wind speed predictions.

► Three methods for predicting above-roof mean wind speeds in urban areas are described.
► Model predictions are compared to measured data from various urban locations.
► Inputting high resolution building data into the model increases predictive accuracy.
► Accounting for the influence of wind direction also improves model accuracy.
► By accounting for these complexities, model errors are reduced to less than 20%.