SubDyn problem with DTU 10MW RWT

Dear All,

I have been experimenting with the FAST model of the DTU 10MW reference wind turbine, using the provided Bladed DLL controller. I have built a simple monopile model to try and use the turbine in an offshore environment. I have put the tower base 20m above mean sea level and chose a similar water depth. I have re-calculated the tower mode shapes using Abaqus, and modified them using the Excel tool provided with FAST. Then I calculated the initial heave displacement and heave damping using the formulation of Section 5.4 of the SubDyn user manual.

The model seemed to run well for my first several trial runs around the rated wind speed, and the PSD of the responses showed the correct frequency content when compared to analytical formulae. In general, the responses seemed reasonable.

However, in certain conditions, especially for high wind speeds like 20m/s (but below the cut-out wind speed) I encounter the usual stability problems (i.e. blades go supersonic, small angle assumption gets violated, occasionally FA array gets exhausted which I assume only means that due to large displacements the rotor moves out of the wind grid). I tried reducing the time step, I went as low as 0.0001s, but that did not solve the issue.

It is really hard to debug the problem because when I look at the FAST output file, everything goes haywire after a few seconds, all displacements/rotations, forces/moments, accelerations, etc become unphysically large and there is a resonant response. As the model runs fine without the foundation (i.e. without SubDyn) for the same wind conditions, I think the issue must be with the SubDyn model, or SubDyn’s coupling to ElastoDyn. The pile is just a 6.5m diameter monopile with a wall thickness of 0.07m, and it’s completely straight. I do not see how this could be an issue. The SubDyn model I built uses a simple Guyan reduction, but including Craig-Bampton modes only makes things worse.

To investigate the problem further, and get a sense of total damping (excluding aerodynamic damping) in the system I tried to run a free vibration test. I did this by setting CompAero=CompHydro=CompServo=CompInflow=0, and only calculating ElastoDyn and SubDyn. I deactivated the drive train rotational flexibility and generator DoFs, then set the initial fore-aft tower top displacement to 1m (also tried 0.5m). To my surprise, the vibration amplitudes increase exponentially and values turn into NaNs in the output file. It looks as if there was negative damping in the system somehow. The tower top responses in fore-aft (F-A) and side-to-side (S-S) directions, as well as the accelerations are shown below.

I do not know where this high frequency (~8.333Hz, 0.12s period) vibration comes from? Does anyone have any ideas on what could cause this or how to solve this problem? If that helps, I am happy to provide plots of any other responses.

Thanks and kind regards,
Laszlo Arany


Dear All,

To further elaborate on this, I have discovered that this issue happens even if there is no excitation whatsoever. No wind, no waves, no initial displacements (beside heave displacement from SubDyn Section 5.4).

It seems like a numerical instability but I can’t figure out what causes it. I tried using the double precision version of FAST, but as expected it did not help the issue.

Any help would be greatly appreciated.

Dear Laszlo,

I agree with all your comments so far…the problem sounds like some sort of physical or numerical instability. But it is hard for me to guess that is causing the problem. I suggest that you disable all DOFs in ElastoDyn and enable them one-by-one until you isolate the DOF (or DOFs) that is(are) causing the problem. Perhaps that will give you hint as to the problem.

Best regards,

Dear Jason and future readers,

The source of the problem was threefold:

  1. The approximated value of the initial displacement was off: I observed oscillations of increasing amplitude in the z direction around a certain value, so I modified the initial displacement to this value. With this I got rid of the heave oscillations.

  2. I increased the heave damping.

  3. I changed the tower mode shapes: I estimated the slope at the bottom by simply taking the tangent of the mode shape output from Abaqus manually (i.e. calculating the slope based on the first data points). When I modified the initial slope to be the rotation value output for the tower bottom node by Abaqus, the mode shape coefficients output by the provide Excel tool changed considerably. This got rid of all other oscillations.

Kind regards,

Laszlo Arany

Dear Laszlo,

Thanks for sharing how you solved the problems you were facing with SubDyn! Hopefully your comments will help other users who are building new FAST models with SubDyn.

Best regards,

Dear Laszlo,

Currently I’m also trying to get the FAST model of the DTU 10MW RWT running.
However, I am struggeling a bit with the controller dll. I have similar issues as mentioned in this forum topic: [url]]

From your posts above it sounds like you were able to get the Bladed dll controller running. How did you do this?
Would you mind sharing your dll?

Best regards,
Paul Schünemann