Instability associated with platform DOFs

Wind & Water Computer-Aided Engineering Software Tools
1

I have a question on the platform DOFs that I’m hoping to get some help on.

It’s discussed in the SubDyn User Manual that to avoid numerical issue, additional damping in the platform heave DOF may be needed in the HydroDyn input file by specifying the AddBLin(3,3) term which can be calculated by Eq.1.

I’m modeling a monopile-supported fixed bottom offshore wind turbine. The platform is defined in ElastoDyn as follows: PtfmMass=0, PtfmRIner=0, PtfmPIner=0, PtfmYIner=total rotational inertia of the undeflected tower about its centerline. The AddF0, AddCLin, AddBLin, and AddBQuad terms are all zero in the HydroDyn input file except for AddBLin(3,3) term representing zeta=0.01 damping for platform heave. The Guyan reduction is used in the SubDyn module.

Running OpenFAST v2.3.0 with a time step of 0.01s (fixed by the controller DLL), I observe the following:

  1. Platform heave still grows unbounded unless zeta is increased to around 0.05.
  2. Platform yaw also experiences instability. I do not know how to properly add damping for platform yaw, so as an attempt, I used Eq.1 and simply substituted M_SD_3,3 and K_SD_3,3 by M_SD_6,6 and K_SD_6,6. Using AddBLin(6,6) that corresponds to zeta=0.01, the platform yaw instability goes away.

My questions are:

  • Can Eq.1 be extended to platform DOFs other than platform heave, or should different formulas be used?
  • Do the diagonal terms of AddBLin, while useful for removing instability, impact the dynamics of the support structure? Is there a recommended threshold? Is the highest damping I’ve gone to i.e. 5% too big so as to significantly alter the behavior of the model away from the actual model?
  • Following the above, if I’m limited both by the time step size and by a AddBLin threshold, are there additional ways to address platform instability?

Thank you in advance,
Jing

2

Dear Jing,

I’m a bit surprised by your obersvations. If the heave and yaw are growing without bound for the model configuration you describe, I would expect numerical instability to be the cause, which I would not expect the addition of physical damping to improve. Instead, I would have expected a lower time step to help.

Regarding your direct questions:

  • Eq. (1) from the SubDyn User’s Guide and Theory Manual can be used for other DOFs, but there are likely other terms that are important for some DOF, e.g., pitch and roll, because the platform reference point is likely not the center of mass. For yaw, I would use K_66^(SD), M_66^(SD), and ensure M^(ED) is the rotational inertia of the undleflected RNA + tower + platform about the tower axis.
  • 5% damping seems rather high for structural damping, which is what is being modeled here; 1% is more common.
  • Even if the controller DLL requires a fixed time step, you can use DLL_DT in ServoDyn to specify a controller time step (DT = 0.01 s in your case) that is bigger than the glue code time step, but you should ensure that the glue code time step is an integer divisor of DLL_DT.

I hope that helps.

Best regards,

3

Dear Jason,

Thank you very much for your time and response!

Following your advice, I lowered the glue-code time step while maintaining the controller DLL time step (we didn’t know this is possible), lowered the platform-heave damping to 1% critical damping ratio, and removed the platform-yaw damping entirely. The simulation stability is maintained this way. So it does appear that the previous instability is primarily due to the time step not being low enough to resolve high-frequency dynamics, although I also cannot speak to why it was suppressed by increased physical damping.

Thanks again and best regards,
Jing