Wind Speed Plotting

hi All,
I have been using the Simulink examples that come with FAST. I was wondering how I plot the wind velocity after the simulation is complete in MATLAB?


Dear Amit,

If you’ve added the wind speed (e.g., WindVxi) to the FAST OutList before running the simulation in Simulink, you can access the wind speed output from the MATLAB workspace just like with the other FAST outputs.

Best regards,

Thanks, Jason.


Hi Jason

If I need to use the Wind output from FAST to an external CFD program to calculate the time varying pressure on 5 MW blade, then should I use below output values exported by FAST?


If these are correct parameters then can you describe the xi-y-zi coordinate system orientation so that these values can be correctly applied in CFD tool?
Is there any diagram / paper where the FAST-CFD linkage is used?

-Thanks and Best Regards
Pravin Kulkarni

Hi Jason

  1. Basically I want to know if I should be using FAST output values WindVxi, WindVyi, WindVzi OR TurbSim output wind speed U,V, W to calculate the pressure loads on the blade using CFD tool. My aim is to generate 10 min average wind speed, calculate pressure in CFD using these speed components as input and then find fatigue life of blade.

  2. Also I found that neither of these sets of values (WindVxi, WindVyi, WindVzi or U,V,W) follow the Weibull distribution. As per TurbSim and FAST theory, should they be following this distribution?

Eagerly awaiting your reply.

-Best Regards

Dear Pravin,

I’m not really sure I understand what you are trying to do. A few comments:

  • FAST outputs WindVxi/WindVyi/WindVzi only provide the wind speed at a single point, which is fine for uniform inflow, but not much use for load prediction when considering spatial-temporal turbulence variations across the rotor disk.
  • In FAST outputs WindVxi/WindVyi/WindVzi, Xi/Yi/Zi is the FAST inertial frame coordinate system, with Xi directed along the zero-degree wind direction, Zi directed vertically upward, and Yi directed to the left when looking downwind.
  • The Weibull distribution applies to the long-term wind speed probability, but most time-domain simulations apply to short-term conditions (e.g., 10-minutes), where the wind is stationary.
  • Fatigue is driven strongly by turbulence, so, you would need to consider turbulent inflow to properly predict fatigue.
  • FAST has been coupled to the OpenFOAM CFD tool for wind-plant aerodynamic analysis in a tool called SOWFA:
  • You mention pressure loads, so, presumably you want to consider blade-surface-resolved CFD, but this is rarely done in the wind community due to the heavy computational expense. Not even our SOWFA tool uses a blade-surface-resolved solution (instead, SOWFA relies on an actuator-line representation of the blade).

I hope that helps.

Best regards,

Thanks Jason as always for quick and clear answer !.

My purpose is to analyze a single blade and not go for wind farm or spatio-temporal scope. I want to calculate time varying loads or pressure on the 17 sections of NREL 5 MW land based blade and do fatigue analysis in Ansys. Based on the discussion so far with you, here is the summary of options

  1. Use of ForcN, ForcT - Ruled out since they do not consider inertial loads
  2. Use of SpniFLxbi…SpniMLxbi - These are possible candidates as they include applied load + interia effect

For option 2 I could calculate the equivalent loads for each section as per Yahya’s suggestion However I was thinking of having a way to calculate pressure on airfoils instead of point loads. If there is no other option then should I go with option 2 above?

Secondly, if I want to use time-domain method (weibul wind—>rainflow counting—>stress range–>Fatigue life analysis process) instead of spectral density method then is there any way to calculate wind velocity components?. I believe TubSim is a spectral analysis tool

-Best Regards

Dear Pravin,

Option 2 is the approach commonly employed in industry, although as I’ve said in prior posts, I don’t know all of the details.

I’m not sure I understand your second question, but the time-domain approach is the one commonly employed in industry.

Best regards,