WindSim | Technical Basics

WindSim is CFD-based wind farm design tool (WFDT) that includes 3D visualization.

WindSim | How does it work?

Local wind fields are highly influenced by local topography. The input basis for WindSim consists of a digital terrain model on a proper length scale, according to the phenomenon under consideration. WindSim can be used in a variety of length scales ranging from detailed micro siting up to larger meso scale wind resource assessments. WindSim uses so called body fitted co-ordinates (BFC) with refinement towards the ground.

In addition to the digital terrain model, a similar model with terrain roughness must be supplied. The terrain roughness has a particular impact in a zone towards the ground.

Finally, WindSim needs meteorological data from at least one point within the modelled area. With these primary inputs the wind resources for the whole area can be calculated, the energy production from any number of wind turbines can be obtained, and the area with infrastructure can be visualized in the 3D interactive visualization module.


Can a numerical model reproduce a wind field scenario as it is observed in nature? Use of numerical models introduces inaccuracies. It is important to be aware of these inaccuracies and their origin. In the following the primary sources of inaccuracies will be highlighted.

Grid resolution

The numerical model uses height and roughness information from a specified grid. The accuracy of the numerical simulation depends of the resolution of this grid. Due to restricted computational resources this grid cannot always be constructed with the desired resolution. Typical a resolution in the order of 100×100 meter is used for meso scale modelling within larger areas in the order of 1000 kmxkm, while a finer resolution in the order of 10×10 meter is necessary for micro scale modelling.

These conditions are illustrated with an example. In the below figure a mountain is shown where half of the points used in the discrete representation have successively been removed. The mountain at the left has 25 meter point spacing, while the same mountain with a 200 meter point spacing is found at the far right.

Figure 1. Discrete terrain with successive removal of points.

In a similar way the estimated wind resources will depend on the grid resolution. A wind field above an island is shown on the two figures below. The extension of the model is 3500×3800 meter and the highest top reaches 68 meter above the sea level. Even for this island with rather modest topography the wind field in the finest model reveals much more details than the model on a coarser grid.

Figure 2. Wind fields at 50 meters height for models with different grid resolutions, wind direction 30 °.

Boundary conditions

The numerical model simulates the wind field in a 3D computational domain. Along the border of the computational domain information about the flow field must be supplied. These so called boundary conditions are given as fully developed flow profiles taking into account the given roughness at the border. However, the model will not have any information about the wind field outside the computational domain, any abrupt changes in topography or roughness along the border will contaminate the flow field. A border zone, where the flow field is allowed to adapt to the surroundings is introduced in order to reduce this problem. No results are presented in the border zone. In some situations there might be impossible to avoid areas with abrupt changes in topography and roughness along the border, these areas should be treated with care.

If terrain and roughness information is available from larger areas than those of particular interest, an efficient nesting technique can be used. Nesting means that the results from a larger outer model are used as boundary conditions in a refined inner model. In this way the problem with inaccurate boundary conditions are eliminated in the refined model.


The quality of the meteorological input data is crucial for the quality of the numerical results. If the meteorological data has been collected during a short time interval these data must be correlated to long-term nearby statistical data in order to represent the long-term wind climate in the area. The meteorological input data must also be representative for the whole computational area. Finally, it must be compatible with the scale of the numerical model, i.e. it must not contain effects from smaller or larger scales than those resolved in the numerical model.

Meteorological input must be supplied for at least one point within the modelled area. WindSim has several means of assimilation if wind data from additional positions are available. Increasing the amount of wind data will of course improve the accuracy of the numerical results.