IEA 22MW_Questions about small angle assumption violated & hydrodyn & servodyn

[haoran-Wang-1023]

Dear Dr. Jonkman and the NREL team,

I am a master student studying the topic of the new jacket foundation based on the IEA 22MW wind turbine. I have faced some questions about OpenFAST. I would appreciate it sincerely if you could give me some advice.

Because of the design of the TP and jacket height, I shortened the height of the tower. So, I changed the DISTRIBUTED TOWER PROPERTIES. But I cannot ensure that it is correct. When I run the program, the warning "Small angle assumption violated in SUBROUTINE SmllRotTrans() due to a large blade deflection (ElastoDyn SetCoordSy). The solution may be inaccurate. Simulation continuing, but future warnings from SmllRotTrans() will be suppressed." and "tower strike" always emerged, but finally, OpenFAST terminated normally. I searched the NREL forum and found some similar questions. Then, I changed the initial pitch angle and rotor speed, or set dt=0.0025s. It is indeed useful for solving the "tower strike."
Thus, my questions are
(1) How to solve the warning about the small angle assumption violated.
(2) Is the reason for the warning related to the fact that I shortened the tower height or that the result I recalculated is incorrect?
(3) Can I ignore this warning because the OpenFAST terminated normally?

Hydrodyn questions:
Because OpenFAST v3.5.3 cannot use the function of wave stretching, I used openfast_dev.x64 sharing from Maciej Mroczek. However, when the program runs to 390s (total run time = 600s, dt = 0.006s), the error "For each surface-piercing member, at least two elements must remain fully submerged" emerges. I don't know what happened to hydrodyn. So, I tried to change the MDivSize from 1.0 to 0.5. But it cannot solve this problem. 
Thus, my questions about hydrodyn are
(1) What does "at least two elements" mean?
(2) how to solve this problem.

Servodyn questions:
(1) IEA 22MW shared us the libdiscon.dll and IEA-22-280-RWT-DISCON.IN about monopile. Can these files also be directly applied to the jacket? Do I need to recompile the .dll?
(2) When I set different wind conditions like steady wind/turbsim wind, do I need to change the servodyn or .in file or just set the match inflow wind file?

Thank you very much!

Best regards,
Haoran

[jjonkman]

Dear @haoran-Wang-1023,

Regarding the "small angle assumption violated" warnings and "tower strike" errors, often these are caused by a numerical instability due to an improper model definition (incorrect geometry, mass/stiffness, or time step) or poor choice of model initial conditions (rotor speed and/or blade pitch not matching their expected (mean) values condition on the mean hub-height wind speed being simulated). I'd have to know more about what you changed in your OpenFAST model to comment specifically on what is causing the issue in your model.

Regarding your "HydroDyn" questions, I'm not sure where you obtained the version of OpenFAST you are using, but wave stretching will be included in the official version of OpenFAST in the upcoming release of OpenFAST v4.0. I would suggest upgrading to OpenFAST v4.0 once it is released and report if you are still running into the same issue at that point.

Regarding your "ServoDyn" questions, you shouldn't need to recompile the ROSCO DISCON DLL unless you need to change its source code for some reason. I would think the ROSCO input file for the IEA Wind 22-MW RWT atop the monopile would apply equally for the same turbine atop a jacket, unless the lowest support structure natural frequencies changed a lot in the process. I wouldn't expect that you'd change ServoDyn or ROSCO when changing the wind conditions, but you should change the initial conditions in ElastoDyn for the rotor speed and blade-pitch angles as mentioned in my response to your first question.

Best regards,

[jjonkman]

Dear @haoran-Wang-1023,

If the thrust is similar in the original model, than the jacket is not the source of the issue.

ElastoDyn output RotThrust is the reaction force within the shaft, and so, includes a contribution of rotor weight (projected along the shaft as a result of tilt). If you want the pure aerodynamic applied thrust, you can use AeroDyn output RtAeroFxh.

Best regards,

[haoran-Wang-1023]

Dear @jjonkman ,

Thanks for your reply.

I output the RtAeroFxh. It was 3400 kN, which is smaller than RotThrust (3800 kN). However, it still has a near 600 kN gap with 2793 kN.

I also try to run the IEA-22-280-RWT-Semi model. The RtAeroFxh was also about 3400 kN, which made me confused.

I really don't know what's wrong with the model. Could you give me more advice, or do you need me to provide some setting files?

Best regards,
haoran

[jjonkman]

Dear @haoran-Wang-1023,

Are you getting the rotor speed and blade-pitch angles that you expect for the rated conditions you are simulating, or are these different from what you expect as well?

Best regards,

[haoran-Wang-1023]

Dear @jjonkman ,

I set the initial pitch angle and rotor speed in ElastoDyn. The rotor speed is a constant value. Pitch angle will change with time. But it did not have a big gap with the initial pitch angle.

Best regards,

[haoran-Wang-1023]

Dear @jjonkman ,

I have solved this question by using the ElastoDyn+BeamDyn.

Thanks for your help.

Best regards,

[haoran-Wang-1023]

Dear @jjonkman,

Recently, I compared the tower top displacement between OpenFAST and the FE software, ANSYS. I simplify the RNA to a mass point and consider the interias in ANSYS. The two models have nearly equal frequencies. However, when I compare the tower top displacements, they do not match each other totally.

The load case is very simple; just at steady wind speed, wave and SSI are not considered. I output the loads from AeroDyn, including RtAeroFxh,RtAeroFyh,RtAeroFzh,RtAeroMxh,RtAeroMyh,RtAeroMzh (see Figure 1).

Figure 1: load series

Then, I input the load series into ANSYS and output the tower top displacement to compare with OpenFAST (TwrTpTDxi, TwrTpTDyi). The results are very different (see Figure 2).

Figure 2: Comparison of tower top displacements

Thus, I hope you can give me some advice to solve this problem.

Sincerely,
Haoran

[jjonkman]

Dear @haoran-Wang-1023,

How do the reaction loads at the tower top compare between OpenFAST and ANSYS, e.g., ElastoDyn outputs YawBrFxp, ..., YawBrMzp?

Best regards,

[haoran-Wang-1023]

Dear @jjonkman,

Honestly, at the beginning of the comparison between OpenFAST and ANSYS, I indeed output the ElastoDyn parameters YawBrFxp, ..., YawBrMzp. Then, I input these parameters into tower top in ANSYS. But the results are not good. So, I try to learn other methods mentioned in other papers, such as outputting the AeroDyn parameters.

Come back to your question. I can use APDL to output the reaction loads at the tower top node in ANSYS and compare them with ElastoDyn outputs YawBrFxp,... YawBrMzp.

Best regards,
haoran

[haoran-Wang-1023]

Dear @jjonkman ,

Recently, I ran the regular wave & no wind case condition based on the new jacket foundation.
According to your reply in forum, I set the .fst and the elastodyn file as below:

• CompInflow = 0
• CompAero = 0
• CompServo = 0
• FlapDOF = False
• EdgeDOF = False
• DrTrDOF = False
• GenDOF = False
• YawDOF = False
  At hydrodyn, I set Hs = 8m, Tp = 10s, and chose the WaveMod = 1(regular wave). The Wave1Elev are shown as below.
  

However, when I output the tower top deflection or React-forces, the results were very weird.

• TwrTpTDxi (0-40s)

• TwrTpTDxi (400-1200s)

• ReactFXss (0-100s)

• ReactFXss (400-1200s)

Thus, I sincerely hope you could give me some advice.

Best regards,

[jjonkman]

Dear @haoran-Wang-1023,

I'm not sure I really understand your question, but if you are artificially increasing the damping to reduce the time for the start-up transients to die out, this artificially high damping will still be present in the periodic response that you want, which could bias that response.

Best regards,

[whr1023]

Dear @jjonkman ,

I compared the NREL-5MW jacket wind turbine and IEA-22MW jacket wind turbine under the regular wave and irregular wave.

I set

• CompInflow = 0

• CompAero = 0

• CompServo = 0

• FlapDOF = False

• EdgeDOF = False

• DrTrDOF = False

• GenDOF = False

• YawDOF = False

and I set the rotorspeed = 0, but the trends of results were different.

Here are the results of NREL-5MW.

NREL TOP DISP UNDER REGULAR WAVE

In my view, the result of the IEA-22MW jacket wind turbine will have the same trend as the NREL-5MW.
However, the results I got all have decay trends, as below shown.

IEA22-MW TOP DISP

SO, I am so confused about that. And I check once again and again and run the static case.

I think I didn't set the right initial ptfmheave or something in ElastoDyn.dat file. How can I set the right initial value of ElastoDyn.dat file?

I sincerely need your help.

Best regards,
Haoran

[jjonkman]

Dear @haoran-Wang-1023,

What particularly concerns you bout the differences between these two tower responses? Are you concerned with the final mean value, the damping level, or the oscillations? I would say these will all depend on the system properties (mass, stiffness, geometry).

I'm not familiar with a jacket model for the IEA Wind 22-MW RWT. Can you clarify where you obtained this model? The IEA Wind 22-MW repository currently only has monopile and semi OpenFAST models: https://github.com/IEAWindTask37/IEA-22-280-RWT/tree/main/OpenFAST.

Best regards,

[whr1023]

Dear @jjonkman ,

Thanks for your reply. I'm sorry I haven't explained the details of my concern. What I concern concludes the final mean value of the tower top displacement, tower top acceleration, and the mudline forces.

Regarding the jacket model, it is designed by myself. Thus, it cannot be found in the IEA Wind 22-MW repository.

Actually, when I set the total time = 5400s under the regular wave cases, the tower top displacement will be in a steady state at 3000s. The tower-top acceleration still decays and is not in a steady state.

Thus, the damping level is also concerned. Because I noticed that in the ElastoDyn.dat file, IEA-22MW monopile set the RayleighDamp = 0.0 0.00183. I'm not sure if it's because I didn't set the damping coefficients correctly that the regular wave takes 3000s to reach steady state.

Besides, the mean values of tower top displacement under different regular wave cases are equal. For this result, I cannot confirm it is right.

Best regards,

[jjonkman]

Dear @whr1023,

Thanks for clarifying, but I'm still not sure I really understand your concern.

Depending on the importance of nonlinearities in the system, regular waves are more likely to impact oscillations in the response rather than mean response. In offshore wind, the mean tower-top displacement is likely more driven by gravitational and aerodynamic loads than wave loads.

I'm not sure if your stiffness-proportional damping constant is appropriate for your model or not, but the lower the damping is, the longer the start-up transient will last.

Best regards,

[haoran-Wang-1023]

Dear @jjonkman ,

Recently, I am validating the frequency of IEA 22MW based on the monopile. I have a question about it. I would appreicate it very much if you could solve it.

Actually, in Table 13 at IEA22 report, the turbine clamped at monopile base is 0.16 Hz as shown. At first, I thought it must be clamped at -79m.

However, when I run the .fst from github and check the SubDyn.dat. I found it clamped at -34m and the frequecny of wind turbine based on the -34m is 0.16 Hz. Thus, my question is if I run it wrong or the 0.16 Hz in report indeed means the monopile at mudline(-34m) not at -79m.

Best regards,
Haoran

[jjonkman]

Dear @haoran-Wang-1023,

I was not the one to write the IEA Wind 22-MW RWT specifications report, but when a monopile is referred to as "clamped", it typically means that it is clamped at the mudline (not the free end of the monopile), which would be -34 m for this system.

Best regards,