I am still working on my FAST/OrcaFlex model for a floating wind turbine, and I am currently running into some difficulties. In order to identify the source of my problems, a colleague advised me to study the OC4 project. Therefore I downloaded the OC4 FAST/OrcaFlex model available on the NREL website, and I compared the results of the different OC4 Load Cases with public data available on your drive “Task 30 OC4”.
The LC 1.x match well. However, for the LC 2.1 (Airy waves only), the pitch and heave amplitudes are greater than expected (see attached figures, the FAST/OrcaFlex model is the large red line). I don’t understand these differences: I did not change the OC4 model. I am using FAST v8.16.00a-bjj, and OrcaFlex 10.2b. I’ve read the Minutes of Meeting of the OC4: I have not seen anything about this kind of issue.
To those working with FAST/OrcaFlex or those involved in the OC4 project, could you give me some clue about this issue?
As usual, thank you for your impressive responsiveness!
In order to be totally sure, I did the test once more: when I download FAST, FASTLink and FAST OrcaFlex models from your link, and then I run the LC2.1 model with no modification except for the Total run time (1060 s instead of 60 s to avoid transient effect), I have a max pitch amplitude of 3.2° (instead of 2° in the OC4 results) and a max heave of 2m (instead of 1.2m in the OC4 results), as previously.
I am probably missing something because I know that it shoul be working… however it is a really annoying issue for me, since it needs to be solved and justified before I can keep working on my subject.
Here are some news about my previous problem, in case someone needs it in the future.
I change the hydrodynamic model in the OrcaFlex file: I am now using the load RAOs and I have set the Ca value for the 6D buoys at 0. The results are now matching well.
However, there is still a point which particularly worries me. I found out that in the OC4 NREL model, the pitch and roll hydrostatic stiffness is negative. I found that quite odd so I did several tests with my FAST OrcaFlex model and a standalone OrcaFlex model and they gave me the following conclusions:
In the FAST/OrcaFlex model, the pitch is larger when the hydrostatic stiffness is positive,
In the standalone OrcaFlex model, the simulation aborts when the hydrostatic stiffness is negative,
The standalone OrcaFlex model with positive hydrostatic stiffness seems to match better with the FAST/OrcaFlex model which has negative hydrostatic stiffness!
These observations are really annoying because it seems that there is a convention issue between FAST and OrcaFlex. Let me know if someone has a clue about these strange observations.
Regarding your earlier post, I’m not too familiar with these sample FAST-OrcaFlex models, but I forwarded you question to Morten Andersen of Aalborg University, who original developed these models during a stay at NREL. Hopefully he can find time to respond.
Regarding the negative values of the pitch and roll hydrostatic stiffness when coupled to FAST, this is because the rotor+nacelle+tower+platform mass are accounted for in FAST, not OrcaFlex. So, the hydrostatic stiffness specified in OrcaFlex should only account for restoring provided by buoyancy, but not the restoring provided by body mass or moorings. This is the same requirement when specifying the hydrostatic restoring in FAST’s HydroDyn module–see the description of the *.hst file in section 6.8.4 of the draft HydroDyn User’s Guide and Theory Manual for more information: wind.nrel.gov/nwtc/docs/HydroDyn_Manual.pdf.
Thank you very much, it would be really interesting to have some details about this model.
I didn’t realize the impact of the weight in the hydrostatic stiffness matrix. Thank you for these explanations, I now understand my strange observations and I’ll be able to correct my model.
Is possible to have the input files for the Turbulent wind (Mann model) used in LC 3.2, 3.3, 3.5and 3.9b for the OC4? I am trying to compute them with the mann turbulence generator of the DUT but the parameter definition is a bit obscure to me.
Sorry if this last question is slightly off topic because it is related to the LC 3.x and not to the LC 2.1, I was not sure that it deserved a new topic.
I asked Amy Robertson who led OC4 Phase II and she said that the Mann wind files are not stored on any server due to the large size (abour 400 MB each). The wind files for OC4 Phase II were generated by Torben Larsen of DTU Wind Energy using DTU’s IEC Turbulence Simulator. I suggest reaching out to Torben to ask him for the associated input files he used.
I finally managed to set up an account so I could comment on your issue.
It’s been some years since I worked on these models, but I hope that I’ll be able to help you out.
From your results it looks to me like the axial damping in the heave plates are too low. Try to add both Drag Area and Cd in Z for some of the elements in the base of the three columns. Area is trivial, but Cd you’ll probably need to tune a little until you obtain the results you desire.
Looking at the OrcaFlex model, there’s some things that I would have done differently If I had to build the models from scratch today.
Instead of using 67 ‘lumped buoys’ I would remodel it by using 10 ‘spar buoys’ (type of 6D buoy). These can be modeled to consist of several cylindrical subelements with varying properties. Here you can have geometrically calculated drag area, and then adjust you Cds manually.
NOTE: The ‘spar buoy’ displaces water and hence provides non-linear which is desirable, but it means that you’ll have to remove the hydrostatic stiffness matrix from WAMIT under ‘vessel type’.