Turbulence intensity in TurbSim with TIMESR

Hi everyone,

I’m using Turbsim (v2, included in the release 3.4.1 of OpenFAST) to create a turbulent wind field starting from a time series of wind velocity (three components) that is recorded in correspondence of the turbine hub.

The turbulence intensity of the X component of the wind velocity, which is computed in MATLAB before creating the .timesr file, is 8.5%. After creating the wind field and running an OpenFAST simulation with it, the turbulence intensity of the wind speed at the turbine hub (which I get from the Wind1VelX variable of InflowWind) is 7.8%.

Trying to understand the issue, I went through the summary file generated by TurbSim where I’ve seen:

Hub-Height Simulated Turbulence Statistical Summary:

Type of Wind Min (m/s) Mean (m/s) Max (m/s) Sigma (m/s) TI (%)

U component 3.18 4.60 5.96 0.393 8.537

That is correct. However, at the end of the summary file:

U-component (X) statistics from the interpolated hub point:

TI = 7.7558 %

In the TurbSim user guide, when dealing with the ScaleIEC option, it is explained that: “For numerical reasons, the turbulence intensity (TI) of the IEC spectral models—without this scaling—usually is slightly less than the specified value. Increasing the time series length and/or decreasing the size of the time step results in values closer to the specified TI.

Does this also apply when creating the wind field from external storm series?

Hi @Alessandro.Fontantel,

Interesting topic. From your message, I understand that you are using the wind speed recorded at the hub location to create a spatially coherent wind in TurbSim.

When working with the hub location, I define an odd number of points in the horizontal and vertical directions of the grid in TurbSim. This will result in a grid point located at the hub location. Accordingly, you should be able to compare the recorded wind and the wind created by TurbSim at the hub location without interpolating different grid points. See below for reference:

By doing so, I would expect that your input wind and the wind at the hub location created by TurbSim to be the the same.

You can actually look at the wind generated by TurbSim by reading for example the *.wnd file. To do so, you can download the Matlab toolbox (GitHub - OpenFAST/matlab-toolbox: Collection of Matlab tools developed for use with OpenFAST) and take advantage of the function: readfile_WND.m (under the Utilities folder).

The below MATLAB code should help:

% Reading the wind file:
[velocity, y, z, nz, ny, dz, dy, dt, zHub, z1, SummVars] = readfile_WND('Alessandro.wnd'); % You will need the *.wnd and *.sum files

% Longitudinal wind speed at the hub location (if an odd number of points for the grid was defined):
Vx = velocity(:,1,(ny+1)/2,(nz+1)/2);
Time = 0:dt:(length(Vx)-1)*dt;

Can you plot the recorded and the wind created by TurbSim at the hub location? I would expect that following this approach, both time series are the same. Fingers crossed!

I hope that helps!

Hey Roger!

It turned out that you were right and the problem was the number of points in the grid. Indeed, I used an even number of points in the Z direction.

After changing the grid definition, the turbulence intensity at hub is correct and the time series of wind speed generated by TurbSim at that location matches the input. You can see it in the figure below, where I also reported the time series obtained at the point closest to the hub using the old grid. It is interesting, at least for me, to see that the difference in the output of TurbSim is quite large even if the change in the grid is small.

Thanks for your help Roger.

Kind regards,


1 Like