OpenFAST
WindCrete Simulation Issue
Hello everyone,
Background: I am currently running simulations on the WindCrete Spar platform developed under the COREWIND project. Notably, I am primarily using OpenFAST v4.1.2 for my simulations, but the most recent WindCrete files I could find use the older OpenFAST v3.5.3 version. Therefore, I migrated the source input files onto the v4.1.2 template and performed some debugging. I have now encountered an issue.
The Issue: I am attempting to conduct control validation, specifically as shown in Figure 5 of the article, "Response of the International Energy Agency (IEA) Wind 15 MW WindCrete and Activefloat floating wind turbines to wind and second-order waves" (available at: https://wes.copernicus.org/articles/6/867/2021/):
IEA-15-240-RWT-WindCrete_MoorDyn.txt
Thank you in advance for your time and guidance. Sincerely
Dear @marisa16-daze,
I haven't reviewed your MoorDyn files, but it sounds like you are concerned with the floater heave and mooring tensions. But I'm not sure which is the cause and which is the effect. I would suggest running a simple case without wind inflow and incident waves, with (a) all platform DOFs disabled, and (b) with all platform DOFs disabled except heave. Do the mooring tensions match between versions for (a)? Do you obtain the same heave response between versions in (b)? Hopefully that simple test will help isolate the issue.
Best regards,
Dear.jonkman Thank you for your quick response and guidance. I performed simulations for both Case (a) and Case (b).
In Case (a) (all DOFs disabled), I found that the sum of the vertical connector forces (Con*fz total) in the migrated file is approximately 2500 N greater than in the original WindCrete source file (which used OpenFAST v3.5.3). The overall force magnitude is on the order of 10^6 N.
In Case (b) (Heave enabled only), the Heave response showed a difference, as expected (as shown below). This difference is likely one of the direct causes of the subsequent dynamic Heave discrepancy. However, I am unsure if the 2500 N discrepancy in Con*fz is the actual cause of the difference in Heave response, as I have checked the input files repeatedly, and they are the same(maybe i ignore some para? ).
Dar @marisa16-daze,
What is the heave restoring associated with water plane area and is the difference of 2500 N in mooring tension enough to cause the shift in heave?
Regardless, in your original post, I saw that heave was off by just less than 1 m, but in your simplified analysis of (b), heave is only off by just less than 0.1 m, so, perhaps something else is off.
Best regards,
Dear Dr.jonkman Yes, in my original post, the Heave response under step wind showed a large deviation, around 1 m, which I wonder if is due to the inclusion of wind loads. However, in Case (b), it's more like that simulates the static equilibrium position of the WindCrete, the deviation is less than 0.1 m, but the oscillation/fluctuation in the migrated file is several times larger than that of the source file.
Please excuse my lack of knowledge regarding moorings; I am not entirely sure of the exact name for the "heave restoring associated with water plane area" in the OpenFAST output channels.However, I recently compared the HydroFz (Total Hydrodynamic Vertical Force) and the total vertical force from the mooring lines (LINE*FZ sum, lines 1–9) between the migrated and source WindCrete files under Case (b).
For the LINEFZ forces, I summed them directly, taking into account their positive and negative signs.For the HydroFz (Total Hydrodynamic Vertical Force), the value from the migrated file (4.07859810^8 N） is greater than the source file (4.077206 10^8 N), differ is 1.39210^5N.
For the total vertical mooring line forces (LINE*FZ), the migrated file (5.45 *10^6 N) is greater than the source file (4.48 10^6 N), differ is 9.710^5N. Could this be the direct cause of both the static and dynamic discrepancies mentioned above?
I have checked the input file differences again and noticed that the new version of the HydroDyn module includes several parameters that were absent in the old version. When editing the new input file, I set some of these new parameters to default values like 1 or 0. I wonder if any of these potentially contribute to the differences in mooring forces and, subsequently, the Heave response. Some examples are shown below (the input file on the left is the old version, and the one on the right is the new version):
Additionally, the new version of the blade model appears to be more detailed/refined than the old version, but this should not lead to numerical differences in the platform displacement.
Thank you for your modeling guidance and patient answers. SinCerely
Dear @marisa16-daze,
Regarding the "heave restoring associated with water plane area", this is an input to HydroDyn. I gather from what you shared that hydrostatic restoring is accounted for in your model through the potential-flow solution (rather than through the strip-theory solution), so, the heave restoring is provided in the (3,3) element of the stiffness matrix specified in your .hst file (although these data are nondimensional, and so, they must be dimensionalized).
Regarding your values of HydroFzi and summed LINE*FZ, again, I recommend comparing these for the various cases (with/without heave), to understand which is cause and which is an effect of the heave motion.
FYI: The hydrostatic load from the potential flow solution in the vertical direction for body 1 can be isolated via output B1HdsFzi, whereas HydroFzi includes the total integration of all hydrodynamic loads in the vertical direction.
Regarding your settings of newer HyroDyn inputs like WaveDisp, AMMod, AxFDMod, MSecGeom, etc., I agree that your settings disable the newer associated functionality.
Best regards,
Dear.Dr.jonkman I compared the results with and without the Heave DOF enabled (i.e., your suggested Cases (a) and (b)), and the results are as follows:
Mooring force comparison with/without Heave DOF, under no wind and no wave conditions: Original file:
File after Migration：
I also compared the calculated results for the WindCrete Step Wind test before and after migration. Referencing Figure 1 from the first conversation (Figure 5 from the paper: https://wes.copernicus.org/articles/6/867/2021/), the results are as follows:
The difference between the mooring force ($\text{LINE*FZ}$) and the hydrostatic force (HydroFzi) seems to reflect the Heave deviation, exhibiting a similar trend and magnitude of offset.Furthermore, to rule out potential issues caused by version differences, I downloaded the $\text{IEA-15-240-RWT-UMaineSemi}$ files applicable to v4.1.2 and above and the original files for v3.5.5, and performed the same Step Wind calculations. The results were quite consistent across the two official versions.
Therefore, the limited conclusion I can draw now is that the issue is almost certainly due to an error in the input file migration, rather than originating from the different OpenFAST versions. However, I have repeatedly and meticulously compared the input files, and found almost no differences except for the blades. This makes the problem quite tricky.
I am unsure if you can pinpoint the source of the issue based on the results provided. If it's truly unresolvable, I might have no choice but to use v3.5.5 to carry out the remaining work.Thank you for your guidance on modeling. SinCerely
Dear @marisa16-daze,
I'm seeing a very small change in the vertical mooring tensions between the original and migrated versions for the no-heave case. But the 3rd plot in each group is too blurry to interpret. Can you re-upload those?
Best regards,
Dear.Dr.jonkman My apologies, Jason. I have retaken the screenshots; the previous ones were indeed quite blurry.
HydroFz from the pre-migration file:
HydroFzi from migrated file:
Green line for only heave,blue line for noDOF
calculated results for the WindCrete Step Wind test:
In the stepwind test figures, the blue lines represent the migrated file, and the green lines represent the original file. Best regards,
Dear @marisa16-daze,
Thanks for cleaning up these plots. From these, I'm seeing a change of 140,000 N in vertical hydrodynamic force between the original and migrated versions for the no-heave case. This is quite a bit more than difference in vertical mooring tensions you shared yesterday, so, the change in vertical hydrodynamic force in the undisplaced condition seems to be the source of the issue. You haven't shared what the hydrostatic restoring from waterplane area in heave is, but does this offset explain the 1-m change in heave between the original and migrated versions?
Best regards,
Dear.Dr.jonkman I consulted the official technical documentation (https://corewind.eu/wp-content/uploads/files/publications/COREWIND-public-design-and-FAST-models-of-the-two-15mw-floater-turbine-concepts.pdf). On page 13, it states that the tower base has a diameter of 13.2 m at mean sea level (MSL). Since WindCrete has a circular cross-section at this location, the waterplane area can be approximately estimated asA_wp = π × r² ≈ 3.1416 × 43.56 ≈ 136.85 m²；Alternatively, it can be estimated based on the hydrostatic restoring stiffness in the heave direction shown on page 15, the C33=1.3746 × 10⁶ N/m，A_wp = C33 / (ρ × g) ≈ 1.3746 × 10⁶ / (1025 × 9.81) ≈ 1.3746 × 10⁶ / 10055.25 ≈ 136.7 m²,which is in good agreement with the geometric calculation.
Therefore, can the difference in heave displacement be estimated as Δz=ΔF/ρgA_wp=140000/（136.710259.81)=0.1m，Can this explain the 0.08 m difference in heave displacement observed in my tests, the difference in the heave equilibrium position in the static equilibrium test, and, more importantly, the larger discrepancy in heave response observed in the step wind test?
In any case, the root of the issue seems to point to HydroDyn, and I will continue to focus my investigation there. Thank you very much for your modeling guidance and prompt response. Sincerely
Dear @marisa16-daze,
Thanks for clarifying. I noticed that your migrated files have 140000 N of additional buoyancy in the no-heave case than your original files, but in your heave free-decay actually shows your migrated model heaves downward a bit compared to the original, which is backwards from what I would expect. Given that the vertical mooring tensions are nearly identical between the old and new files, there must a difference in body weight (rotor, nacelle, tower, or floater) as well that would explain that.
Also, your step-wind test has a greater change in the heave displacement between the old and new models than the heave-only test, which suggests that there is other coupling at play. Are the mooring lines taut such that the surge offset you are showing between the original and modified models could be related to the heave offset (set-down effect)?
Best regards,