Natural frequency of NREL DS monotower in OC3.

Dear Jason,

I am looking at using FAST to model monopile support structure for offshore wind turbines. Before doing the compilation of FAST I would like to reproduce the 1st natural frequency of the support structure (0.25) from NREL report “Offshore Code Comparison Collaboration (OC3) for IEA Task 23 Offshore Wind Technology and Deployment”.

I construct a NREL_DS FAST model based on discussion in Compile FAST to model distributed spring model for monopile. to include the soil-pile interaction. In this process, I have not updated the mode shapes of NREL_DS model.

I may have some points need to be clarified with you:

  1. If I use NREL originally developed FAST.exe (can be considered not including soil effect) to run NREL_DS model, the 1st natural frequency of support structure is 0.2124.
  2. If I use the FAST.exe compiled by me (maybe incorporate the soil effect correctly) and set TwrLdMod = 2 (Do I have to?), I will get 1st natural frequency as 0.2842. If I do not set TwrLdMod =2, I will get 1st natural frequency as 0.2124.
  3. But since either way could not give me a result around 0.25. I have also tried BModes NREL_DS model by wind.nrel.gov/public/gbir/BModes-for-Erica/ with a small modification for node location. I can get the 1st natural frequency at around 0.25.

So, could you please help me to check which one is the number presented in OC3 report? And whether there is some mistakes I have made through the process? I have attached the platform and tower data.

Really appreciate!

Best regards,
NRELOffshrBsline5MW_Tower_Monopile_DS.doc (4.04 KB)
NRELOffshrBsline5MW_Platform_Monopile_DS.doc (6.14 KB)

Dear Linling,

I have looked briefly at your attachments and noticed a couple problems:

  1. You have not enabled any platform DOFs, which means that the monopile will be fixed 56 m below the MSL. To properly “feel” the soil stiffness, you should enable platform surge, sway, roll, and pitch DOFs (PtfmSgDOF, PtfmSwDOF, PtfmRDOF, and PtfmPDOF) when using the DS foundation model.
  2. You have not used mode shapes of the support structure appropriate to the DS foundation, but are using the mode shapes for the monopile with rigid foundation. You must use proper mode shapes if you expect to get the correct natural frequency. You referred to your forum topic where you were using BModes to derive frequencies and mode shapes for the DS model. You should use these mode shapes in your FAST model.

Regarding your question on TwrLdMod, “yes”, you must set TwrLdMod to 2 if you wish to use the UserTwrLd_DS.f90 source file you’ve compiled in.

Best regards,

Dear Jason,

Really appreciate!

Do you still have the DS_tower.dat as well as the DS_platform.dat files available? Do you mind if you can send me a copy of them?

For the discussion above, I still have two questions in my mind:

  1. Why would not incorporate the Heave and Yaw motions of the platform?

  2. If we have set TwrLdMod = 2, does that mean that we cannot see the hydrodynamic effect (by Morison’s equation) and soil effect at the same time if we run time-marching simulation with FAST? But we can see both of them simultaneously by either a CS or an AF model.

Best regards,

Dear Lingling,

Yes, please find attached my FAST v7 tower and platform input files for the OC3-monopile for the NREL 5-MW turbine with distributed-springs (DS) foundation model.

Here are my answers to your questions:

  1. The distributed springs are transverse to the pile, so, only impact transverse displacements – i.e., from surge, sway, roll, and pitch. Enabling yaw or heave DOFs will require that you include axial or torsional springs.

  2. Setting TwrLdMod to 2 in FAST v7 causes FAST to bypass the hydrodynamic load calculation, however, this is remedied in the UserTwrLd_DS.f90 source file because within routine UserTwrLd() is a CALL to routine MorisonTwrLd() – the routine normally called when TwrLdMod is set to 1.

Best regards,
DSFiles.zip (3.5 KB)

Dear Jason,

Really appreciate!

I have the following problems to discuss with you:

  1. I have used the tower.dat file run with my compiled FAST.exe. If I only turned on DoFs of tower (8 DoFs in total), I have obtained the first two natural frequencies as 0.2260, 0.2380. If I turn on all the DoFs in FAST (which is 19 DoFs in this case), I got the first two natural frequencies as 0.2225, 0.2482. Do you think the results in the correct range?

  2. When I used BModes results provided by wind.nrel.gov/public/gbir/BModes-for-Erica/ and the ModeShapePolyFitting.xls in FAST archive, I could not get the mode shapes in your tower.dat file. I just have copied the first two columns of the first S-S mode in BModes output to the input sheet of the ModeShapePolyFitting.xls. I have attached the ModeShapePolyFitting.xls file I used to reproduce the mode shapes and the output from Gunjit S. Bir. Could you please take a look whether I have made some mistakes in the procedure?

Best regards,
monopile_tower_with_added_m.xls (74.1 KB)
ModeShapePolyFitting_tower_with_added.xls (3.52 MB)

Dear Lingling,

  1. The frequencies you are reporting are a bit low–I would expect the 1st two modes to be closer to 0.25 Hz. Are you using a heavily discretized tower in FAST? I used 99 TwrNodes.

  2. Your approach to running ModeShapePolyFitting.xlsx looks sound. Please note that you can obtain the “Slope” of cell F5 in worksheet “Input” from the BModes output. For the mode in question, this would be 0.001304, the s-s slope at the base of the first mode. The mode looks qualitatively close to the one I derived, but I would not expect a perfect match because I derived my mode shapes using ADAMS in place of BModes, as discussed in the following forum topics: http://forums.nrel.gov/t/tower-eigenfrequencies-of-nrel-5mw-turbine/517/1 (see my posted dated Tue Aug 28, 2012).

Best regards,

Dear Jason,

Really appreciate!

Now the questions are clear, but the results still deviate some. I have DT = 0.00125 and I don’t know whether this would yeild some difference. I have attached the simulation file, could you please take a look of it?

Thanks and best regards,
4NRELOffshrBsline5MW_Monopile_DS_FAST original - Copy.zip (1.43 MB)

Dear Lingling,

Your model and results look correct. Actually, the frequencies you reported earlier (0.2225 and 0.2482) match what I’ve produced with this model as well. Sorry for the confusion earlier–I think I was looking at the results from a different model.

Best regards,

Dear Jason,

Really appreciate!

Best regards,

Dear Jason,
I am using FAST to model monopile flexible support structure for offshore wind turbines.
I use Distributed Springs method to compile my FAST.exe and the modal frequency is almost the same as oc3 Phase II result.
After that I have some question, how can I output
(1) the fore-aft and side-to-side displacements of the monopile at 7m below the mudline
(2) the fore-aft shear, side-to-side shear, side-to-side bending moment, and fore-aft bending moment of the
monopile at 7m below the mudline
I try to read codes form FAST_IO.f90 but I can’t find out some parameter about the monopile at 7m below the mudline.
DO u give me some suggestions or information about this, please?
Thank you very much.

Best Regards
Jason. Lai

Dear Jason,

I’m assuming you are using FAST v7 for your analysis.

In FAST v7, you can place virtual strain gages at up to nine locations along the tower via inputs NTwGages and TwrGagNd. You can then output tower motions (e.g. absolute positions relative to the inertial frame origin, deflections relative to the tower base) and tower loads (e.g. shear forces and bending moments) at each of these gages via outputs starting with TwHt#, where # is the gage number (between 1 and 9). For example, If TwrGagNd(4) = 12, where node 12 was located at 7 m below the mudline, then TwHt4TPxi would give the translational position along the xi axis of the node 7-m below the mudline and TwHt4FLxt would give the fore-aft shear force along the Lxt axis at the node 7-m below the mudline. See the FAST User’s Guide for more information.

Best regards,

Dear Jason,
Really appreciate!
After that,I want to exert seismic loading to monopile flexible support structure for offshore wind turbines.
According to UserPtfmLd_Seismic.f90, I try to revise subroutine UerPtmLd in UserSubs.f90 and then recompile my FAST.
Is the method feasible?

Best Regards
Jason. Lai

Dear Jason,

I have not tried to use Seismic for a simulation with an offshore monopile model, but I think it should work without recompiling FAST, at least for the case of a rigid foundation (without distributed or coupled springs). I would try to use the FAST v7.02 model of the NREL 5-MW turbine atop the OC3-monopile (available here: nwtc.nrel.gov/Seismic). You’ll have to change PtfmSgDOF and PtfmSwDOF to True, PtfmMass to 7.0e5, and PtfmLdMod to 1 in NRELOffshrBsline5MW_Platform_Monopile_RF.dat. for use with Seismic, and of course use the new Seismic input file (*_Seismic.dat). Let me know if that works.

Best regards,

Dear Jason,
Really appreciate!
I will try it lately.
In OC3 phase II, which one is monopile rotational displacement, TwHt1RPxi or TwHt1RPyi ?
“1” is at z=0 m

Best Regards
Jason. Lai

Dear Jason,

TwHt1RPxi is the roll rotation about the xi axis and TwHt1RPyi is the pitch rotation about the yi axis.

Best regards,

Dear Jason,
It can work after I use new FAST 7.02 compiled with distributed springs and Seismic effect.
I will use Ansys software to compare with the result. (RNA set to be a point mass and tower input boundary condition such as distributed springs and Seismic acceleration.)
On the other hand, I want to exert seismic loading to jacket-type flexible support structure for offshore wind turbines.
I will try to use new FAST 8.09 compiled with distributed springs and Seismic effect.
Is the method feasible? If yes, can you give me some suggestion?
Thanks for your help.

Best Regards,
Jason.Lai

Dear Jason,

At this time, the reaction nodes (normally placed on the piles) in SubDyn within FAST v8 are assumed to be connected to the inertial frame. This limitation will have to be resolved before you can add seismic analysis to fixed-bottom offshore wind turbines within FAST v8. My guess is this would take a bit of work.

Best regards,

Dear Jason,
I compile a new FAST v7.02 version included wave + seismic effect and then I compare seismic results with this paper (see the attachment).
I can get similar results under idling and operation condition but I have a different result about Y-dir Base Moment in the emergency shutdown condition.

According to the paper, the emergency shutdown is initiated at 409.416 seconds, and it takes about 14 seconds for the rotor speed to reach zero RPM.
Why does Y-dir Base Moment decrease before the emergency shutdown at 409.416 seconds triggered?

In my result (blue curve), rotor speed to reach zero RPM exceed 423.416 seconds, can you give me some suggestion?

Thanks for your help.

Best Regards,
Jason. Lai
SR-5000-53872.pdf (647 KB)

Dear Jason Lai,

It is difficult for me to answer your question because I don’t have the exact input files used to generate the results from Prowell et al 2010.

To ensure that your executable is functioning as you expect, the first thing I would do is run the FAST 7.02 CertTest and ensure that you can reproduce the previous results (should be nearly identical, because from what I can tell, your changes will not impact the CertTest cases). The CertTest does check situations such as shutdown events.

If that checks out, my guess is something is slightly different in how you initiate the emergency shutdown from what is done in Prowell et al 2010.

Best regards,

Dear Jason,
I use FAST v7.02 version (A module for simulating seismic loads nwtc.nrel.gov/Seismic ) to compare El-Centro seismic results with this paper. (NREL/SR-5000-53872 “Seismic Loading for FAST”)
In the simulation, the earthquake initiated after 400 seconds is the same with above paper.
I get good agreement results in My (Y-dir Base Moment) under idling and operation condition. (as shown in first figure)

When I replace another seismic acceleration time history in this model, I get a confusion about the result. (as shown in second figure)
The seismic acceleration time history(duration: 90sec) is larger than El-Centro seismic acceleration time history(duration: 45sec).

The TwrBsMyt magnitude in idling condition is larger than normal operation condition after seismic occur and when the ending of earthquake, the TwrBsMyt loading convergence in idling condition is slower than normal operation condition to restore the original state.
Is it resonable??

Best Regards,
Jason. Lai