MODELLING
Wind flow modelling can be split into two main categories; linearised and non-linearised. The former modelling technique uses simplified steady-state solutions of the Navier-Stokes equations. It assumes a uniform surface roughness and generally flat terrain. This technique is generally used on mesoscale models as it is less computationally intensive and is the backbone of well-established industry tools (discussed later). A mesoscale model can aid in identifying regions of the country with the most wind resource potential. The drawback of such modelling is the inaccuracies exhibited in mountainous or coastal terrain due to sharp changes in gradient.
To remedy this, non-linearised modelling techniques can be used in conjunction with the linearised models in order to produce a higher resolution and more accurate analysis of a particular region. This microscale model dictates the placement of a wind farm or even individual turbines within the region identified by the mesoscale model. The accuracy of this modelling system stems from its ability to simulate flow separation, boundary layer thermal properties as well as turbulence due to more complex surface types [1]. The drawback of this system is that it is much more resource intensive, requiring a balance to be maintained between precision and runtime. Additionally, the complex nature of the analysis results in more inconsistencies and can be difficult to validate.
To remedy this, non-linearised modelling techniques can be used in conjunction with the linearised models in order to produce a higher resolution and more accurate analysis of a particular region. This microscale model dictates the placement of a wind farm or even individual turbines within the region identified by the mesoscale model. The accuracy of this modelling system stems from its ability to simulate flow separation, boundary layer thermal properties as well as turbulence due to more complex surface types [1]. The drawback of this system is that it is much more resource intensive, requiring a balance to be maintained between precision and runtime. Additionally, the complex nature of the analysis results in more inconsistencies and can be difficult to validate.
The terrain elevation data can be obtained in many different spatial resolutions, sometimes as accurate as 50 cm. The Shuttle Radar Topography Mission (SRTM) is a 90 m resolution data available freely from many sources and is therefore chosen for initial development. The given text file is converted from geographic to Cartesian coordinates in meters using MATLAB, and can be stored in any desired format.
Arthur's Seat in Edinburgh was chosen as the test area. The extracted elevation data of an area of around 6 km squared is presented in graphical form on the left. It is immediately clear to anybody familiar with the area that the SRTM data fails to capture the full features of the terrain, especially that of steep gradients. This results in an overall smooth representation of the area. Despite this drawback, the data is suitable enough to be used in the initial stages of the project to build and test the models until higher quality data can be secured. |
Both the figures presented on this page are derived using MATLAB, the file necessary to duplicate these results is available for download from the link below.
![]()
|