DTU 10-MW Reference Wind Turbine

Dear Abhinary,

It sounds like you are able to use a DLL controller in the version of FAST you are using, so, if the DTU DLL is not loading properly, I would guess the problem is in the path or compilation of the DLL. I’m not sure I can give you more specific advice.

I’m not sure I understand your comment about the unsteady airfoil aerodynamics model of AeroDyn v15. The Beddoes-Leishman dynamic stall model is implemented in AeroDyn v15. Perhaps you were running an old version?

Best regards,

Dear Jason,

When I tried to run the model operation DLCs provided in the repository there are no problems with the DTU controller. I till cant figure out the error in my setup. As you mentioned could be a mistake in the directory.

The dtu_we_controller_bladed is running with the feedback loops from generator_servo_input and controller_input. Is there an open-sorce master controller for the DTU 10 MW RWT?

Thanks and regards
Abhinay Goga

Dear Abhinay,

I have not run simulations with the DTU 10-MW turbine myself. Hopefully someone else on this forum can answer your questions regarding the controller.

Best regards,

Dear Jason and all,

Can anyone explain why the electrical output is facing with such numerical instabilities?
The frequency of the fluctuations seems not resulted from wind shear, tower shadow (3P-effects) and does not match with the wave frequency as well.
Any suggestions are appreciated.

Thanks and regards
Abhinay Goga

Dear Jason and all,

LIFES50+ D1.2 states that, AeroDynv14 calculates the unsteady aerodynamics at fixed initial position rather with the moving tower. For a floating wind turbine the tower moves along with the floater and initial position for aerodynamics calculation is actually a frame of reference positions. The report later states that, “at present the moving tower potential flow model is not implemented with UA, hence decided to use AeroDynv14”. Is it still the case?

Thanks and regards
Abhinay Goga

Dear Abhinay,

I’m not sure I really understand what the text you are quoting is saying.

Regardless, the tower influence model in AeroDyn v14 is limited to a fixed undeflected/undisplaced position of the tower, so, I would not enable AeroDyn v14’s tower influence model for floating wind turbines, which have the potential for substantial platform displacement or tower deflection. AeroDyn v14 is no longer being worked on and will be deprecated soon.

That said, AeroDyn v15 has more functionality than AeroDyn v14 and AeroDyn v15 does consider the impact of floating platform motions and tower deflection on the tower influence model.

Best regards,

Dear Dr.Jason
Thanks for reading my email.
I’m using openfast v2.3 to simulate the IEA 10 MW reference wind turbine. And I used the DISCON_IEA_offshore.dll controller. However,I faced some problems when I simulated the rated wind speed condition(turbulent wind). The simulation can run but the blade pitch is zero during the simulation under the rated wind speed, which casue large rotor thrust force and genrator power (over 12 MW). Aslo, when i tried to simulate the over-rated wind speed condition, the simulation aborted. I have uploaded some input files and results.Can you give me some advice?I’m not familiar with controller, maybe the problem of the controller?inputfile.rar (242 KB)

Thanks a lot in advance!
Best regards
Lixian Zhang

Dear Lixian,

I’m not familiar enough with the DTU 10-MW reference wind turbine OpenFAST model that you are simulating to comment much, but I don’t see that you are referencing the DISCON_IEA_offshore.dll in your ServoDyn input file (instead, you set DLL_FileName = “controller\DTU_10MW_Braceless_Controller_Lib.dll”); could that be a problem? Also, are the initial conditions you’ve set (RotSpeed = 9 rpm, BlPitch = 0.0) the expected values for the mean wind speed you are simulating?

Best regards,

Dear Dr.Jason
Thanks for your quick response.
I’m so sorry that I uploaded the wrong input files of servodyn. I think that I use the right path of dll file into servodyn file after I doubled checked the input files. I also found some simliar questiones in the fourm, and try to use “dtu_we_controller” into openfast. However, simliar problem happened. The controller can not be loaded into either openfast 32-bit and 64-bit. I found that no matter how I changed the intial pitch angle, the output pithc angle will become zero and the simualation will be aborted in seconds under over rated wind speed(18 m/s:constant wind).


I downloaded the input files of IEA 10 MW wind turbine including the controller in github afew mothons ago. I’m wondering if I use the wrong dll file for 10MW controller. Is there any newest version for the dll of 10 MW?
IEA-10.0-198-RWT_ServoDyn.rar (3.03 KB)
Thank you very much!

Best regards
Lixian Zhang

Dear Lixian,

I have been using the IEA-10.0-198-RWT (based on the offshore DTU 10-MW wind turbine design) within the OC6 project.

At the beginning I was also having problems with the blades pitch. The IEA-10.0-198-RWT uses the ROSCO controller. You can find a 64-bit version of this controller already compiled in the next GitHub repo: https://github.com/IEAWindTask37/IEA-10.0-198-RWT/tree/master/controller/dll_windows

In order to run this controller, you will need the controller parameters file (*_DISCON.IN). You can find this file in the same GitHub repository that I was commenting: https://github.com/IEAWindTask37/IEA-10.0-198-RWT/tree/master/openfast

Note that the wind turbine is a direct-drive wind turbine. Accordingly, you can see that line 76 in the parameters file defines the gearbox ratio as 1:

1.0                 ! WE_GearboxRatio	- Gearbox ratio [>=1],  [-]

At this point, there is something very important: line 79 of the parameters file defines the location of the text file IEA-10.0-198-RWT_Cp_Ct_Cq.txt that contains the rotor performance tables (Cp, Ct, Cq). You must edit the parameters file and point to the proper location in your computer. I recommend using an absolute path rather than a relative path when pointing to this file. For example, in my case, line 79 looks like this:

"C:\OC6_PhaseII\OpenFAST_model\IEA-10.0-198_Wind_Turbine\ServoData\IEA-10.0-198-RWT_Cp_Ct_Cq.txt"      ! PerfFileName      - File containing rotor performance tables (Cp,Ct,Cq)

Note that this IEA-10.0-198-RWT_Cp_Ct_Cq.txt file is also available at: https://github.com/IEAWindTask37/IEA-10.0-198-RWT/tree/master/openfast

I guess this should solve your problems with the controller. Please, let me know if it’s not the case.

Roger

Dear Dr. Roger
Thanks for your reply!
Now, It’s working after I used the ROSCO controller according to your comments. Thanks again!
I have another questions when using ROSCO controller.
When I simulated the constant wind with wind speed (11.4 m/s) for monopile wind turbine, the simulation results looked pretty good. However, when I simulated turbulent wind(11.4 m/s), it seams that some numerical instabilities occurred in generator power(Figure 1).And for overrated wind speed (18 m/s:constant wind ), the simulation results are also quite strange(Figure 2).


Figure 1 simulation results of turbulent wind(11.4 m/s)


Figure 2 simulation results of constant wind (18 m/s)

I would be appreciate if you can give me some suggestions.
Anyway, thank you very much!
Best regards
Lixian Zhang

Dear Lixian,

I’m glad that the controller is finally working at your side.

Regarding the issues observed, I guess that a developer of the ROSCO controller would be the most suitable person to explain the specific behavior that you are observing. Note that there is a GitHub repository for the ROSCO controller: https://github.com/NREL/ROSCO. You may contact them to better understand what could be the reason for your results.

Hopefully they can address your doubts with the controller performance.

Roger

Dear Dr. Roger

Thank you very much!

best regards
Lixian Zhang

Hi Lixian,

Glad to see that you got the ROSCO controller working. A few quick questions: Is the x-axis here time-steps? Or time?

Also, I am not sure I see what you mean when you say that there are “some numerical instabilities in the generator power”. Are you referring to the drops in generator power (e.g. around 1600 time-steps)? If so, these are likely attributed to the turbine dipping into region two - perhaps tuning the controller or the set-point smoother could improve this behavior a bit, but it does not look abnormal.

As far as the 18m/s simulation (Figure 2), this is certainly strange and something that I have not seen before. I just ran a 10 minute simulation at 18 m/s and am not seeing similar results (shown below). I also just pulled the most recent IEA 10MW model before running. Have you changed anything in the model, other than file paths and wind speeds?


Thanks,
Nikhar

Dear Dr. Nikhar

Thanks for your reply!
Sorry about the vague, the X-axis is referred to the simulation time, not time step. The simulation time is 5000s.

  1. Yes, "some numerical instabilities in the generator power’’ means the drops in the generator power(1600 s) ,in the turbulent wind simulation.
  2. For 18 m/s simulation, the simulation results seems quite good in the first 1800s. However, the results are quite strange after 1800s. I didn’t change anything for IEA 10 MW wind turbines. I’m not familiar with the controller before. Maybe not the controller issues? I will download the recent model of IEA wind 10MW in GitHub and to see if the same problems happened.

I would be appreciate it if I can receive you reply.
Thanks again!

Best regards
Lixian Zhang

Dear Jason and Matt,

For the regular wave simulation on OO-star semi floater there is a strange behavior. A sudden jump in the mooring line tensions (after 512secs) but no effect on platform displacements. LIFES50+ project also mentioned such behavior in their report. Could you explain the cause for such instability?

Thank and regards
Abhinay Goga

Hi Lixian,
I am not currently able to re-construct your results. I am running OpenFAST in single precision with a 64 bit architecture and the most recent version of ROSCO. My results are below. Let me know if if your problems persist after updating your IEA 10MW model.

Also, just to note, the power is slightly offset from 10MW in these results - this is due to a bug in how the rated torque was calculated for the DISCON.IN file in this model. I’ve talked to the person managing this and it will be remedied soon.

Nikhar

Dear Abhinay,

I appreciate you bringing this up and sharing the illustrative plots. This has come up before, and it’s tricky because I can’t think of anything in MoorDyn that would cause a change in behavior partway through a simulation. Other instances of this are documented in these threads:
http://forums.nrel.gov/t/feedback-on-moordyn-v1-01-02f-numerical-instability/1816/1
http://forums.nrel.gov/t/response-in-long-term-simulations-and-different-moordyn-time-step/2111/1
When the change occurs seems to depend on the time step size, and it only seems to be a problem in some versions of FAST/OpenFAST.

Could you let us know what version of FAST/OpenFAST you’re using, and what MoorDyn time step size (dtM) you’ve set?

If this is occurring with the current version, we will try to replicate the problem and investigate further.

Best,
Matt

Dear Matt,

The previous simulations were performed with FAST v8.16 and dtM 0.001. Following your comments in attached thread, I ran a new simulation with OpenFAST v2.3.0 and increased MoorDyn time step (dtM) to 0.0001. Yet the behavior remains same after 512secs and surprisingly the rise in magnitude seems a bit higher compared to the one with FAST v8.16.

Kind regards
Abhinay Goga

Dear Abhinay,

We traced this problem with the tensions back to how the time variable is stored in MoorDyn. It seems that the existing single-precision float causes problems with the interpolation/extrapolation of fairlead kinematics when the simulation time reaches a certain point (in this case 512 s). I was able to solve the problem by changing the variable to double precision. (I wish I had fixed this years ago, but unfortunately I did very little maintenance of the Fortran version of MoorDyn prior to joining NREL.)

If I remember right, you have been compiling OpenFAST yourself. In that case, you can access the updated code here: github.com/mattEhall/openfast/tree/MDmods

We will get this fix included in the next release of OpenFAST.

Thanks again for bringing this issue to our attention.
Matt