5MW FOWT linearization issue

Dear Rana,

A few more clarifying questions:

  • What do you mean when you say you use a step function for torque and pitch?
  • Is the time-series solution in equilibrium (periodic steady state) when you linearize?
  • How did you post-process the results to extract the damping ratio–did you apply MBC followed by azimuth averaging and eigenanalysis?
  • Which mode is negative damped?

Because of the wrapping of the text in your pasted *.lin file, I can’t post-process your results to verify your eigensolution. Please upload an attachment next time.

Best regards,

Dear Jason,

Getting back to the previous topic, yes I did as FAST v7 User’s Guide explains for linearization, TrimCase to 3, VSContrl to 1, VS_RtTq to the desired constant generator torque (43093.55Nm in this case), and VS_RtGnSp, VS_Rgn2K, and VS_SlPc to 9999.9E-9 (very small don’t cares > 0.0). But doing this, the system does not converge.
To solve this convergence issue I increase the TMax value as it is explained in the Guide up to 9999s. Also I try changing DispTol and VelTol, but the pitch angle varies between 8 and 4 degrees aprox., which is a large variation I think.
Last option would be to increase the system platform pitch damping, but if I do this, I would be changing the system behaviour and I want to obtain as much accurate linear model as possible. There is no other way?

Thank you in advance,
JOE

Dear JOE,

I haven’t played around much with TrimCase = 3 for a floating system. Perhaps the gain on the proportional feedback control law for the trim solution is set too high, exciting the platform-pitch motion? You could change this gain, but this will require a recompile of FAST v7.

Best regards,

Dear Jason,

Thanks for your instantaneous reply.

That was my thought. Ok, I will try it. I will tell you if I get it.

Best regards,
JOE

Dear Jason

Really i am thankful for you, the answers are as following

1- desired torque entered system as in step input function and the same thing for blade pitch angle
2- yes the time-series solution was in equilibrium
3- i execute GetMats-f8 and after that i execute mbc3, one of the results is MBC_ eigen vals. i don’t execute cce. m
4- no one is negative damped but number 13 equals 0 .

i attached the MBC-eigien- vals
eigen-vals.txt (633 Bytes)

Dear Rana,

Ah, I see now. The negative damping is showing up for a zero-frequency i.e. rigid-body mode. The damping of the rigid-body mode does not necessarily indicate that the system is unstable. As described in this forum topic: http://forums.nrel.gov/t/learizing-baseline-5mw-wind-turbine-with-fast/494/1, rigid-body modes show up in MBC3 as a pair of zero-valued (or near-zero-valued) frequencies with +/- inf damping (i.e., eigenvalues with real values only). That is, each rigid-body mode will introduce an additional mode beyond the number of enabled DOFs and the damping is unphysical.

Best regards,

Dear Jason,

I linearized the OC3-Hywind Model in FAST 7.02 for different Windspeeds. All Eigenfrequencies match the one I expected but the PtfmYaw Sensor (0.08 Hz instead of 0.12 Hz).
I have not used the trim option because of the not converging simulation and instead used the average for Windspeed, RotSpeed and RtTq that I found in time series calculations to define a point of operation.
I don’t think I misunderstood the values from CampbellDiagramm.xls (with GetMats.m and MBC.m) because there is no mode around 0.12 Hz found at all.
In time series calculation the peaks in the ROA´s (with Wave excitation) and powerspectra (with turbulent wnd excitation) match the 0.12Hz acceptable well. So I don’t think the model itself has errors.
I attached the Main file as well as the linearization setting file.

Do you have any idea where this error could evolve from?

Best regards,

Simon
NRELOffshrBsline5MW_Linear.txt (2.05 KB)
FAST.7.02_v11.00 _Main.txt (26.3 KB)

Dear Simon,

I’m not sure I understand your problem. Are you saying the time series shows excitation of the platform-yaw mode at 0.12 Hz, but linearization shows the natural frequency of the platform-yaw mode to be 0.08 Hz for the same conditions?

It’s a bit troubling that the trim option is not finding a converged solution. I did see that you have set VSContrl to 2 instead of 1 in your FAST primary input file; could this be causing a problem?

I would also suggest simplifying the model e.g. by eliminating degrees of freedom (DOFs) to better isolate the problem.

Best regards,

Dear Jason,

Yes, that was my Problem.

by switching VBSContrl to 1 and CalcStdy to True with TrimCase =3 linearization converged and i got 0.12 Hz for PtfmYawDOF.
I set
12.74235216 VS_RtGnSp - Rated generator speed for simple variable-speed generator control (HSS side) (rpm)
43090 VS_RtTq - Rated generator torque/constant generator torque in Region 3 for simple variable-speed generator control (HSS side) (N-m)
212.6790715 VS_Rgn2K - Generator torque constant in Region 2 for simple variable-speed generator control (HSS side) (N-m/rpm^2)
10 VS_SlPc - Rated generator slip percentage in Region 2 1/2 for simple variable-speed generator control (%)

For 0mps Windspeed that works well but for the others Igot small angel assumption violated warning due to large tower deflection (I tried different start settings for Initial Platform positions withoutsuccess) followed by the error, that due to smllRottrans() Catenary cannot solve quasi-static mooring

Do you have a idea to solve this?
Best regards,

Simon

Dear Simon,

Are you saying that your time-domain solution is now not working? You should be using VSContrl = 2 in time-domain simulations within FAST v7.02 in order to use the torque controller from the Bladed-style DLL.

Best regards,

No, only the linearization for 0 mps is working. The others, with higher wind speed’s, got the mentioned error. Sorry for being to clear enough.

Time domain simulation’s are completed already.

Dear Simon,

The FAST User’s Guide describes how to use VSContrl = 1 with TrimCase = 3. Normally, I’d expect TrimCase = 3 to be used only above rated wind speed, whereby for the NREL 5-MW turbine I would set VS_RtTq = 43093.55 N-m (and VS_RtGnSp = VS_Rgn2K = VS_SlPc = small numbers). Below rated, I’d use TrimCase = 2.

Best regards,

Dear Jason,

thanks for clarifying. I ran the linearization with Trimcase =2 for Rotspeed <12rpm (rated). As the steady state solution was not found I tried it with compaero=false.

Now it works and the PtfmYaw DOF is found at 0.12 Hz. But I think that without air the results are not really useful, am I’m right?

Also there are some modes missing. Shouldn´t the Modes from Blade 1-3 the same? So 1st flw Bld1 = 1st flw Bld2 …

Is there a way to check the linearized model for errors?

Best regards,
Simon

Dear Simon,

The failure of TrimCase = 2 to converge is discussed in the following forum topic: http://forums.nrel.gov/t/linearisation-trim/628/1.

I’m not sure I understand your other questions.

Best regards,

Dear Jason,

I lowered the wind speed to get a higher TSR (now for sure higher than TSR at cp_max) but with CompAero = True no steady state solution is found.

With CompAero= false there is no problem for FAST finding a steady state solution and linearize, apart from that I think the model is different then and results are not reasonable anymore, right?

When I set GenDOF = false and CompAero = true the steady state solution is also found. Is it reasonable to set the VS_RtTq to an appropriate value for the windspeed / rotor speed and then assume a fixed speed turbine (GenDOF = false) to get the linearization of this variable speed turbine?

Best regards,

Simon

Dear Simon,

Does the steady-state solution with TrimCase = 2 seem to be converging such that increasing TMax would help, oscillating periodically such that increasing damping would help, or diverging?

I’m not sure what you mean by “reasonable”, but the linearization solution should be valid for small deviations around the operating point.

If you fix the generator DOF during linearization, I suspect you’ll get differences relative to a linearization with the generator DOF enabled in modes that involve coupling to the rotor rotation e.g. tower side-to-side bending, drivetrain torsion, and blade-edgewise bending.

Best regards,

Dear Jason,

Rotor speed, GenTq and the residues are oscillating periodically. I tried to increase TwrFADmp(1) and BldFlDmp(1) and the other tower and blade damping factors as well as the hydrodynamic damping matrix B in spar.1 by more than one order of magnitude but apart from rated condition the same problem is still showing up.
Tmax is already = 2000s so I don’t think I need to increase it further.
Even with very low wind speeds (5 rpm, 0.1 mps) the simulation is not converging.

Is anything else I can do to get a converged steady state condition without fixing the Gen DOF which you say would change the coupling between rotational modes.

Best regards,

Simon

Dear Simon,

I suggest that you debug by disabling structural DOFs. Does the rotor speed converge with TrimCase = 2 when only the generator DOF is enabled? If so, which DOF when enabled is causing the solution to not converge?

Best regards,

Dear all,

I was able to solve the error by switching IndModel in AeroDyn from NONE to SWIRL.

I disabled all structural DOFs but the error remained for larger wind speeds. So I tried changing aerodynamic settings. Now the steady state solution is found with all DOFs turned on again and with appropriate wind speed (still lower than for optimal TSR).

I hope someone find this helpful.

When looking for the eigenfrequencies with Getmats.m, MBC.m and the excel-sheet the blade modes for the individual Blades are not identical.
I assume that these are asymmetric modes ? For 1st edw bld 1 there is no mode fitting. Can someone help with identifying the blade modes?
1st flapwise I expect at 0.7 Hz 1st edgewise at 1Hz and 2nd flapwise at 2Hz.
Is there a way to get collective Blade modes?
Best regards,
Simon
FAST.7.02_rpm07.00_lin.txt (3.43 MB)

Dear Simon,

I’m glad you identified what was causing the lack of convergence. Just FYI – Normally, IndModel = NONE in AeroDyn is meant for simulations involving a parked/idling rotor. IndModel should be set to WAKE or SWIRL for operational rotors.

Collective modes have all blades deflecting in-phase. Asymmetric modes have one blade deflecting out-of-phase with the other(s). I have not reviewed your results, but you can find several examples on this forum where you can find interpretations of the eigensolution e.g.:

http://forums.nrel.gov/t/eigenanalysis-fast/362/11
http://forums.nrel.gov/t/linearization-in-fast-with-unbalanced-rotor/721/2
http://forums.nrel.gov/t/scaling-input-in-campbelldiagram-for-drivetrain-and-yaw-dof/683/4

These examples should help you as you interpret your eigensolution.

Best regards,