Substructure implementation error


I am attempting to implement Test21, but with an alternate substructure, and have run into a few issues. When I run the test I am receiving the following error after changing out the HydroDyn and SubDyn input files.

“FAST_Solution: CalcOutputs_And_SolveForInputs:SolveOption2:Inflow_Wind_CalcOutput:CalcOutput:ifW_TSF
FWind_CalcOutput [position=(NaN,NaN,NaN)in wind-file coordinates]: Error:FF wind array was exhausted at 1.0000E-02 seconds (trying to access data at NaN seconds).”

In looking at the forums this has previously been attributed to model instability, but also appeared to have occurred much further into the simulation than in my case which leads me to believe that my substructure is not properly interface with the upper portion of the model. Is this likely? (particularly since the ElastoDyn/AeroDyn/ServoDyn files are unaltered from Test21) I used Paraview to try and determine whether the support structure files were being read (image Test6 Output.png)

The structure I am attempting to create would appear like this.

There seemed to be an issue the fact that my brace members were attaching to the edges of the monopile members and not the centers, so I added connecting members in a polygon at the top and bottom of the braced section so each member would be able to attach to the axis of another. The lines in the Test6 image are all in the correct places, but I am at a bit of a loss as to how I can achieve a better visual.

My two primary questions are:

  1. Can you recommend a strategy for how to achieve the support structure I am going for within the capabilities of FAST
  2. Is it likely that my wind array error is the result of this sub-structure not being properly recognized/modeled or is there another area of the model I should be looking to fix?

Any guidance you can provide is extremely appreciated. Thank you.

MC Anderson
HybridSupport_SubDyn3_COPY.txt (11.2 KB)

Dear M.C. Anderson,

I agree that the error you receiving means that the model is going numerically unstable.

I think that your substructure should be able to be modeled in FAST/SubDyn/HydroDyn. I took a brief look at your SubDyn file and I’m a bit confused. It looks like you’ve added a member between joint 5 and 14, which is unphysical for your substructure. Also, it doesn’t appear that joints 6-13 get connected to joint 5. Likewise, joints 15-22 don’t get connected to joint 14. There may be other problems, but I would start there.

Regarding your figure, it looks like you are using surface visualization for the turbine, but stick-figure visualization for the substructure. How did you do that? Alternatively, perhaps you’ve set the Morison members of HydroDyn to have a very small diameter?

Best regards,


Thank you for the rapid feedback. Per your suggestion, I added members from 6-13 to 5 and 15-22 to 14, removing the non-physical member between 5 and 14 and have successfully achieved the images below. Do you foresee any problems with me defining the newly added members as having extremely small diameter/thickness/density, as they are also “non-physical” for the actual design?

As for your question regarding how I generate the image with both surfaces and lines – purely by accident. The Morison surfaces were not properly initialized (due to the joint definition issues previously discussed) and the lines in that previous figure belonged to a “DebugError.HD_MorisonDistrib.t0.vtp” as opposed to the actual Morison surface.

The updated model now runs successfully for about 10 seconds before exceeding the .4 radian limit for small rotations.

In looking at the Paraview visualization for the last executed timestep - it appears that I haven’t adequately defined the connection between the brace members and monopile sections.

Is it advisable to treat the brace member ends as interface joints and lock some of the DOF, or is there another method that you would recommend to model the setup?

Kind regards,
MC Anderson

Dear M.C. Anderson,

While the additional members you’ve added between joints 6-13 and 5 and joints 15-22 and 14 are unphysical, there are needed to properly interconnect the beam assembly. I would expect negligible in-plane warping of the bottom pile and top cylinder, so, these new members should have sufficiently high stiffness (although there mass can be low) (but not so high stiffness and so low mass so as to cause numerical problems).

In SubDyn, “interface joints” are those that are rigidly connected to the tower base and “base reaction joints” are those that are rigidly connected to the seabed. So, these cannot be used to interconnect beam members that are offset from each other, unless that offset happens at the tower base or seabed. Instead, you should add unphysical members that lead to the required behavior, as I explained above.

Best regards,