Simulation going unstable in extreme conditions

Hi everyone,

I am trying to simulate a spar-type FOWT (my own model) in extreme wind and waves, with the rotor idle. I have been able to run the model in milder conditions but when I try to run it in wind speeds >20m/s and Hs >5m the simulation crashes within the first 2 minutes of simulation time, and I get the following error message:

FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:HydroDyn_CalcOutp
ut: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:HydroDyn_CalcOutp
ut: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.

FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption2:InflowWind_CalcOutput:CalcOutput:IfW_TSF
FWind_CalcOutput [position=(29.048, -44.23, 5.4457) in wind-file coordinates]: FF wind array
boundaries violated: Grid too small in Y direction. Y=-44.23; Y boundaries = [-40, 40]
CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in
GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.

FAST encountered an error at simulation time 29.345 of 3660 seconds.
Simulation error level: FATAL ERROR

Aborting FAST.

I have turned off BEMT and unsteady aerodynamics, I have added a correction step, I have tried reducing DT and DT_UJac and I have tried turning off AeroDyn altogether but they have not helped. After playing around with the ElastoDyn settings I have found that the problematic degrees of freedom appear to be PtfmPitch and PtfmRoll, so I’m guessing that the problem is related to hydrostatics, but I am not sure how to solve the problem without altering the platform model significantly. Does anyone have any suggestions?

Thank you, any help will be greatly appreciated.

Best regards,
Rachael

Dear Rachael,

I agree with your assessment and with the steps you’ve taken to solve the problem so far. Indeed, it sounds like the system spar-type FOWT is hydrostatically unstable, or at least susceptible to excessive motion in large waves. Have you calculated the natural frequencies of the platform in FAST (e.g., via free-decay test simulations) and are the natural frequencies what you expect? Are the platform-pitch and roll natural frequencies sufficiently far away from the first-order wave frequencies / periods (0.05 - 0.5 Hz; 2-20 s)?

Best regards,

Dear Jason,

Thank you for your reply. Yes, I have done free decay tests in FAST and the pitch and roll natural frequencies are around 24 s, although I did have to increase C44 and C55 in the AddCLin matrix to achieve this desired value. I am using a Jonswap spectrum with a wave period Tp = 12.5 s.

Best regards,
Rachael

Dear Rachel,

Do you have second-order hydrodynamic terms enabled in HydroDyn? Perhaps the platform-pitch mode is being excited by second-order effects?

Best regards,

Dear Jason,

No, I have not enabled second order hydrodynamics. However, when I ran the model in BModes to obtain the tower mode shapes for FAST, the roll and pitch natural frequencies were 0.135 Hz, which doesn’t correspond to the natural periods from the free decay tests in FAST but might correspond with first order wave frequencies. I’ve checked my inputs in BModes and am fairly confident that they are correct, but do you know of any reason why there might be this difference?

Best regards,
Rachael

Dear Rachael,

I would expect that BModes and FAST would predict very similar natural frequencies. Have you properly accounted for the gravitational restoring in the BModes solution (because gravitational restoring is not intrinsically accounted for in BModes, you must manually add it through a stiffness matrix)?

Best regards,

Dear Jason

Yes, I input the gravitational restoring values into the stiffness matrix (hydro_K) in BModes. I calculated values of -77559000 for pitch and roll restoring based on a displaced volume of 389.89 m3 and a centre of buoyancy 19.79 m below the mean sea level, and 44531 for heave restoring based on a waterplane area of 4.43 m2.

Best regards,
Rachael

Dear Rachael,

I would expect that you’d specify a positive value for the pitch and roll stiffness. I agree that the buoyancy term you’ve calculated is negative-valued, but I would expect an equally large (or larger) positive value from the gravitational restoring associated with the full system weight and center of mass (-massgz_CM).

Best regards,

Dear Jason,

Thank you for the advice, I had not realised that gravitational restoring needed to be included in BModes. The natural frequencies from BModes now agree with those from FAST, around 0.04 Hz. I have corrected the tower mode shapes, but the FAST model is still going unstable. Sorry for the trouble, but do you have any more suggestions?

Best regards,
Rachael

Dear Rachael,

I guess I would start by computing the response amplitude operators (RAOs) via FAST. Perhaps large excitation is expected, even though the natural frequencies are outside the wave-excitation range? We’ve published a paper on how to compute RAOs via FAST–see: nrel.gov/docs/fy13osti/58098.pdf.

Best regards,

Dear Jason,

Thank you for your advice. I have attached the RAOs computed for roll and pitch. It looks like there is some excitation around 0.05 Hz, though I am not sure why.

Best regards,
Rachael


Dear Rachel,

Have you compared the FAST-generated RAOs with the RAOs generated via WAMIT? Do they both predict the response at 0.05 Hz?

Best regards,

Dears Dr. Jason Jonkman and Rachael Smith,

I see that this thread is quite old but I currently facing the same issue and perhaps your feedback can help me. I am also simulating an own spar-type FOWT, based on the Zefyros (ex Hywind-demo) structure. As Mrs Rachael Smith, the OpenFAST simulation is running in milder conditions but when I try to run it in extreme conditions (i.e., wave with Jonswap spectrum with Hs = 10 and Tp = 14s) a crash occurs in few iterations. Moreover, to identify the issue, I have decided to model aerodynamics in still air by enabling AeroDyn while disabling InfloWind and the wake calculation in AeroDynV15. The error message is closely like the one described in the first post of this thread.

I have found that the problematic degree of freedom appears to be PtfmPitch. In my research work the second-order hydrodynamic terms are enabled but even without them the simulation seems to crash. Otherwise, the natural frequencies from BModes in Pitch and Roll agree with those from FAST, around 0.04Hz.

As Mrs Rachael Smith, it looks like there is some excitation around 0.05Hz, as illustrated in the RAO pitch from OpenFAST and WAMIT.

Rao from OpenFAST
pitch_RAO_OF

RAO from WAMIT
pitch_RAO_WAMIT

Do you have any idea of the source of possible error?

Thank you in advance,

Best regards,

Adrien

Dear @Adrien.Hirvoas,

Rachel had received warnings about large platform rotations, which eventually triggered in error because an aerodynamic analysis node passed the boundary of the wind domain. In your case, you’ve disabled InflowWind, so, you should not receive the same error (although you can still receive a warning about large platform rotation). So, what do you mean when you say that the simulation crashes?

Best regards,

Dear @Jason.Jonkman,

Thank you for your reply. Indeed, in the described configuration setup, the error is not a FF wind array boundaries violation. Nevertheless, when I was enabling the InflowWind module, I had also an aerodynamic analysis node that was above the grid. That’s why, I decided to model the aerodynamics in still air in order to check if the system spar-type FOWT was hydrostatically unstable as mentioned in this thread. In that configuration set-up, the error message is the following:

FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:HydroDyn_CalcOutp
ut: Angles in GetSmllRotAngs() are larger than 0.4 radians.
HydroDyn_CalcOutput:HDOut_MapOutputs: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
HydroDyn_CalcOutput:HDOut_MapOutputs: Angles in GetSmllRotAngs() are larger than 0.4 radians.

FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:HydroDyn_CalcOutp
ut: Angles in GetSmllRotAngs() are larger than 0.4 radians.
HydroDyn_CalcOutput:HDOut_MapOutputs: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
HydroDyn_CalcOutput:HDOut_MapOutputs: Angles in GetSmllRotAngs() are larger than 0.4 radians.

FAST_Solution:FAST_AdvanceStates:AD_UpdateStates:SetInputs:TwrInfl:CheckTwrInfl:Division by
zero:Elements 2 and 1 are colocated.

To conclude, the warning is about large platform displacements not allowing with the OpenFAST theory. Then, I was thinking that the platform-pitch frequency was too close from the first-order wave frequencies. However, if you have any advice, I am more than interested.

Thank you in advance,

Best regards,

Adrien

Dear @Adrien.Hirvoas,

OK, thanks for clarifying. You could bypass this error by disabling the tower influence model in AeroDyn v15, i.e., TwrPotent = TwrShadow = 0.

I’m not sure what else to say. If your model functions as expected in less severe conditions, this implies that your model is not numerically unstable, rather your model is physically susceptible to excessive pitching in severe sea states.

Best regards,

Dear @Jason.Jonkman,

Thank you for your prompt response. I will follow your advice concerning AeroDyn v15.

Thank you again for your help and your work with OpenFAST.

Best regards,