Dear all,
Currently I’m working on a research to compare the NREL 5 MW wind turbine model with land based foundation and the 4 FOWT platforms under different conditions. I conducted some preliminary simulations with very simple environmental conditions to verify that everything was OK. This are the conditions :
-
SteadyWind of 8 m/s; conditions set on NRELOffshrBsline5MW_InflowWind_Steady8mps.dat
1 WindType
8 HWindSpeed
90 RefHt
0.2 PLExp
-
No waves; conditions set on every SeaState file of the 4 FOWTs
0 WaveMod
-
RotSpeed 8.5 m/s; initial condition set on evey ElastoDyn file to minimize the transient behaviour at the beging of the simulations.
8.5 RotSpeed
Value calculated from the expression:
TSR=V_tip/V_wind=80 (m/s)/11,4 (m/s) ≈7
RotSpeed (rpm)=(TSR·60·V_wind)/(2pi·R)=(7·60·8)/(2pi·63)≈8.5 rpm
-
I prolonged the simulations to 600 s at the .fst files
600 TMax
As you can see those are the only files and variables that I have modified. The rest of the files are the original ones that come with every model when you download it from GitHub.
I have attached the results. I was expecting something similar to the GenPwr curve from the land foundation in every FOWT, but only the semi submersible model satisfy my expectations. I can’t understand why the other 3 models (Barge, Spar and TLP) project that output.
I would be grateful if you could enlighten me about the behaviour of those models under such simple conditions.
Kind regards,
Ignacio López
Dear @Ignacio.Lopez,
Presumably the first and third plots you are showing are for the land-based and OC4-DeepCwind semisubmersible, respectively; is that correct?
The other models are showing clear signs of a strong start-up transient, resulting from not setting good initial conditions for platform surge and pitch. To avoid this transient, we generally recommend for floating wind systems to set the initial platform surge and pitch angles in ElastoDyn (PtfmSurge, PtfmPitch) to their expected (mean) values conditioned on the mean hub-height wind speed you are using. (The r-test for the OC4-DeepCwind semisubmersible sets good initial for surge and pitch in ElastoDyn based on a mean wind speed of 8 m/s).
Best regards,
Dear @Jason.Jonkman ,
Yes, the first and third plots are for the land-based and OC4-DeepCwind semisubmersible, respectively.
As you suggested, I changed the initial conditions of the variables PtfmSurge and PtfmPitch to their expected mean values at a mean wind speed of 8 m/s. Now the GenPwr is behaving normally.
I have some new doubts. The method I used to calculate those values is to check the results of the PtfmSurge and PtfmPitch in the firts simulations (without the correct initial conditions). Then I estimated the mean value and remade the simulations with the good initial conditions.
-
Is there another more “academic” way to calculate the expected values of PtfmSurge and PtfmPitch under different wind conditions?
-
If I understand your answer correctly, to calculate those initial conditions, in the case I use a turbulent wind, I should only consider the mean hub-height wind speed, right? But what if I also have irregular waves? Are they not taken into account?
-
Lastly, what’s the typical transient time in the FOWTs with turbulent wind and irregular waves?
Thanks a lot.
Best regards,
Dear @Jason.Jonkman,
I have simulated a simple scenario of uniform wind and no waves with the four types of floating platforms, and I have evaluated the GenPwr. The uniform wind condition I have simulated is a step wind of 8-14 m/s with a transition time of 5 s. Here are the results:
I think the responses from the Barge, Land, and TLP are something expected, but I am not sure about the Spar and Semi. I don’t think a peak of almost 6 MW is normal. Is it due to my wind conditions being too aggressive or unrealistic?
Best regards,
Dear @Ignacio.Lopez,
The NREL 5-MW baseline wind turbine atop the OC3-Hywind spar and OC4-DeepCwind semisubmersible share the same baseline controller, which I presume you are using. This controller responds more slowly (less gain) than the baseline controller of the other systems you are simulating due to the long natural periods of the platform-pitch modes in the spar and semi to avoid controller-induced negative damping. I suspect this controller change is the reason for the overshoot in power you are seeing.
Best regards,
Dear @Jason.Jonkman,
Thank you for your quick reply.
So, if the controller in the Spar and Semi responds more slowly and because of that I am obtaining a peak of almost 6 MW, then what is the capacity of the wind turbine to bear an overshoot in power?
I remember reading somewhere that the maximum overshoot that the turbine could withstand was around 10 % of its capacity. That’s why the results from the Spar and Semi didn’t add up for me.
Is this assumption correct or the limit for an overshoot in power is higher?
Best regards,
Dear @Ignacio.Lopez,
Well the NREL 5-MW baseline wind turbine is not real, and the amount of power “overshoot” that is permitted is not specified for this “paper” turbine. I agree that 10% sounds more reasonable in reality, but like you said, this specific step change in wind speed is likely unrealistic as well.
Best regards,
Dear @Jason.Jonkman ,
Thank you for your previous answer.
Now I am checking the step response of the NREL 5-MW baseline wind turbine atop OC4-DeepCwind semisubmersible and the land-based foundation. The wind profile I’m using and the blade pitch of both cases are shown under
The step of 12 m/s caught my attention. For the OC4-DeepCwind case, having in mind your words:
This controller responds more slowly (less gain) than the baseline controller of the other systems you are simulating due to the long natural periods of the platform-pitch modes in the spar and semi to avoid controller-induced negative damping.
and having read the specifications of the floating system (https://docs.nrel.gov/docs/fy14osti/60601.pdf) I wanted to know if what is happening there is the beging of a large resonant motion. Because of that, I simulated a simple case of the semisubmersible with steady wind of 12 m/s and no waves (having set good inial conditions) and it is imposible to obtain any reasonable outputs because of the resonant motion of variables such PtfmSurge and BldPitch.
For the land-based foundation also, I don’t understand the behavior of the blade pitch in the 12 m/s step.
Best regards,
Dear @Ignacio.Lopez,
Did you change the OpenFAST models of the NREL 5-MW baseline wind turbine on land and offshore atop the OC4-DeepCwind semisubmersible in any way from how they provided in the OpenFAST r-test (other than the wind speed and initial conditions)? I would not expect either model to have strong oscillations in blade pitch after a wind speed step around rated.
Best regards,
Dear @Jason.Jonkman,
I did not change, at least consciusly, any of the OpenFAST models and their variables apart from the aforementioned. I just repeted the simulations from zero and obtained the same results.
If you need to check any of the files I’m using, just let me know a way to send them to you, since in the forum I can only attach pictures.
Best regards,
Dear @Ignacio.Lopez,
Could you can share the files through Google Drive?
Best regards,
Dear @Jason.Jonkman,
Here it is the link to files. These are the exact files I used to obtain the graphs of the previous post.
If you change the InflowWind file in the semisubmersible case and set the steady wind of 12 m/s, with initial conditions RotSpeed = 12.1 rpm, BldPitch(1,2,3)=1.5 deg, PtfmSurge=8 m and PtfmPitch=3.5 deg you will see the strong oscillation in blade pitch I referred to.
Best regards,
Hi Ignacio,
Jason asked me to weigh in on your results.
I think this is normal behavior for the NREL-5MW reference controller. Steady wind near rated is difficult to dampen fully without additional control loops (e.g., floating feedback) or peak shaving. It also looks like you are using a constant power torque controller, which also is more difficult to dampen, compared to constant torque control.
I hope this helps. Please let us know if you have any questions.
Best, Dan
Dear @Daniel.Zalkind,
Thank you so much for your response.
Given that you brought it up, the controllers I’m using are the standard ones: DISCON.dll for the land-based turbine and the DISCON_OC3Hywind.dll for the semisubmersible. I’m planning to do a complete study with turbulent wind (created with TurbSim) and irregular waves (set in SeaState).
Could I have any problem using those controllers, or was this just an isolated phenomenon related to the wind condition I simulated?
Best regards,
Hi Ignacio,
I would have to do a simulation and loads analysis, like you did, to be sure about those controllers. I’m not familiar enough with their parameters or behaviors to fully assure you, but I do not think they have floating feedback or peak shaving to dampen platform motion.
At risk of self-promoting (twice), we have the ROSCO controller, which can be tuned for a variety of turbines and platforms using only a few tuning parameters. In this article, we tuned/optimized ROSCO for 4 different floating substructures: https://www.nrel.gov/docs/fy22osti/82038.pdf. You can see in Fig. 2 that the controller parameters will affect your loading.
I hope this helps.
Best, Dan
Dear @Daniel.Zalkind,
Thank you once again for your answer. It has been very helpful.
As I need to continue my work, I have a new doubt I hope you or @Jason.Jonkman can solve.
During a simulation with the semisumersible model, using multidirectional waves, I found this error:
C:\Users\ignacio\Desktop\cwd\UniWind_Waves\5MW_OC4Semi\../5MW_Baseline/HydroData/marin_semi.12d:
only contains one first wave direction at 0 degreesIt cannot be used with multidirectional waves.
Set WaveDirMod to 0 to use this file.
CheckWAMIT2WvHdgDiffData: WAMIT output file
C:\Users\ignacio\Desktop\cwd\UniWind_Waves\5MW_OC4Semi\../5MW_Baseline/HydroData/marin_semi.12d:
only contains one second wave direction at 0 degreesIt cannot be used with multidirect
The solution to this error is clear (set WaveDirMod to 0) but my question is, how much less realistic would it be to make the simulations without multidirectional waves?
I have already read in the forum some posts like this one: How to use SESAM \ WADAM to generate.12d file - #2 by Jason.Jonkman. I think I should follow the steps described there to create a new marin_semi.12d file with the second-order potential-flow solution needed to simulate multidirectional waves.
The problem with that is that I am not familiar at all with hydrodynamic calculations at this scale.
I hope you can help me with this issue.
Best regards,
Dear @Ignacio.Lopez,
It is quite common to assume waves are uni-directional (long crested, without directional spreading), but the validity of this assumption is likely site-dependent.
It is not to difficult to compute a first-order potential-flow solution for different wave directions, but it would be very computationally expensive to compute full second-order QTFs for different wave directions, and this is rarely done in practice.
Best regards,
Dear @Jason.Jonkman,
Thank you for your comments.
I have checked the wave rose of Hs of my site, attached bellow:
Under those conditions, it is fair to assume unidirectional waves, right?
Best regards,
Dear @Ignacio.Lopez,
I would guess your figure is illustrating the wave height as a function of primary wave propagation direction, but does not directly address the topic of long- (uni-directional) versus short-crested waves (wave directional spreading).
Best regards
