Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Frequency-Domain Hydrodynamic Analysis about OC4 #2578

Open
axxyyyyyyyy378 opened this issue Dec 21, 2024 · 16 comments
Open

Frequency-Domain Hydrodynamic Analysis about OC4 #2578

axxyyyyyyyy378 opened this issue Dec 21, 2024 · 16 comments

Comments

@axxyyyyyyyy378
Copy link

axxyyyyyyyy378 commented Dec 21, 2024

Dear @jjonkman ,

Today, I attempted to use ANSYS AQWA as an alternative to WAMIT (since I do not have access to the latter) to calculate the frequency-domain hydrodynamic parameters for the OC4 wind turbine. Below, I have attached some of the geometric parameters and wave conditions used during the calculations. I have modeled the geometry of the floating platform and defined the baseline parameters following the reference (Robertson, A., Jonkman, J., Masciola, M., Song, H., Goupee, A., Coulling, A., Luan, C., 2014. Definition of the Semisubmersible Floating System for Phase II of OC4 (No. NREL/TP-5000-60601, 1155123).However, it seems there were some errors in my settings.

Beacuse when I imported the hydrodynamic parameter files generated by AQWA (.1, .3, .hst, .12s, .12d) into OpenFAST, the simulation produced incorrect results. Under a steady unidirectional wind of U = 9 m/s and unidirectional irregular waves with Hs = 1.2646 m and Tp = 10 s, the wind turbine exhibited unreasonable platform displacements.

I have also included the erroneous simulation results below. Could you kindly help identify where my settings might be incorrect?
**AQWA settings below:
1734797892162
1734797909065
1734797930264
1734799099813
1734797946937
1734797959558
**Unreasonable results:
1734798810081
1734798844928

Thanks in advance.

@axxyyyyyyyy378 axxyyyyyyyy378 changed the title Feature request Frequency-Domain Hydrodynamic Analysis about OC4 Dec 21, 2024
@jjonkman
Copy link
Collaborator

Dear @axxyyyyyyyy378,

It looks like your OpenFAST model is physically or numerically unstable.

To isolate the problem, I would first disable second-order hydrodynamics. Is your solution stable then?

If not, I would check that your ANSYS AQWA-developed first-order solution (.1, .3, .hst) is consistent with the first-order solution provided in the OpenFAST r-test, which was derived from WAMIT.

FYI: the body mass and inertia should not be needed to compute the first-order solution (.1, .3, .hst). But the values you say you are using for the second-order solution are not correct because they only represent the platform itself, not the mass/inertia of the full floating offshore wind turbine.

Best regards,

@axxyyyyyyyy378
Copy link
Author

axxyyyyyyyy378 commented Dec 25, 2024

Dear @jjonkman,

Merry Christmas! (maybe we have a time difference T.T) Thank you for your patient response. After reading your reply, I first disabled the second-order hydrodynamic calculations, but the results were still incorrect. I then manually compared the first-order solution generated by ANSYS AQWA with the reference solution provided by OpenFAST, and I found some mismatches in the hydrodynamic parameters. For instance, the hydrostatic stiffness coefficients for roll and pitch in the .hst file showed significant differences. I discovered that this issue was caused by an incorrect center of gravity setting in the model.

After correcting this, the .hst file (hydrostatic stiffness coefficients) results became consistent with the reference solution, and the results stabilized when the second-order hydrodynamic calculations were disabled. However, it is worth noting that there are still significant differences in the results of the .1 and .3 files (as shown in the attached figure). I am unsure if this indicates that the AQWA calculations are still incorrect. Could you kindly provide some guidance on this?

**significant differences in the results of the .1 files:
1735126061486

Additionally, based on your comments, I did not account for the total mass/inertia of the full OC4 floating wind turbine when calculating the second-order hydrodynamic parameters (.12s, .12d). However, I am unsure where to find accurate values for these parameters. In the reference I consulted 'Robertson, A., Jonkman, J., Masciola, M., Song, H., Goupee, A., Coulling, A., Luan, C., 2014. Definition of the Semisubmersible Floating System for Phase II of OC4 (No. NREL/TP-5000-60601, 1155123)', it seems that data regarding the overall center of mass and inertia of the full OC4 floating wind turbine is not provided.

Thank you in advance!

@jjonkman
Copy link
Collaborator

Dear @axxyyyyyyyy378,

I'm not sure why your first-order AQWA solution differs from the WAMIT solution provided by NREL. If it helps, the dimensional NREL solution is plotted in Figure 4-3 of the NREL/TP-5000-60601 report you reference. How do your results compare when plotted on top of each other?

The full-system mass, center of mass, and inertias of the NREL 5-MW baseline wind turbine atop the OC4-DeepCwind semisubmersible are documented in my post dated Feb 13, 2020 in the following topic on our forum: https://forums.nrel.gov/t/oc4-deepcwind-semisubmersible-wamit-files/2203.

Best regards,

@axxyyyyyyyy378
Copy link
Author

axxyyyyyyyy378 commented Dec 25, 2024

Dear @jjonkman,

Thank you for your response. As you mentioned, I compared the AQWA results with the NREL solution in Figure 4-3 of the NREL/TP-5000-60601 report. From the comparison, most results match well; however, there are significant differences in the surge-pitch elements (A15, B15) and sway-roll elements (A24, B24). What could be the possible reasons for this? Could you kindly provide some suggestions?

1735144861355

Best regards,

@jjonkman
Copy link
Collaborator

Dear @axxyyyyyyyy378,

Good question. In addition to the sign flip on the 15 (surge-pitch) and 24 (sway-roll) coupling terms, I also see that the AQWA results are quite a bit smaller than WAMIT in magnitude for the 15, 24, 44, and 55 terms, while the results look nearly identical for the 11, 22, 33, and 66 terms. Are you using the same reference point in AQWA as used by the WAMIT results, which is the intersection of the vertical centerline of the platform and still water level?

Best regards,

@axxyyyyyyyy378
Copy link
Author

axxyyyyyyyy378 commented Dec 26, 2024

Dear @jjonkman,

Thank you for your patient response. You are absolutely correct—I mistakenly set the reference point at the actual center of gravity of the platform (13.46m below the waterline). When I adjusted the reference point to the intersection of the vertical centerline of the platform and the SWL, the results in .1 matched perfectly (see the figure below).

Subsequently, I recalculated the second-order hydrodynamic coefficients (.12s and .12d) with the reference point returned to the actual center of gravity of the platform (without considering the full wind turbine yet). Afterward, I conducted a comparison of the six-degree-of-freedom responses of the platform under a specific load condition (steady wind speed U = 8 m/s , irregular waves H_s = 1.2646 m, T_p = 8 s) using AQWA results (.hst, .1, .3, .12s, .12d) and WAMIT reference results (see the figure below). The comparison shows excellent agreement between the two.

However, as you previously pointed out, when calculating second-order hydrodynamic coefficients, I need to account for the mass and inertia parameters of full wind turbine , including tower, nacelle, and blades. Yet, based on the current results, it seems that considering only the floating platform has already yielded accurate outcomes, which leaves me somewhat puzzled.
Additionally, when calculating the second-order hydrodynamic coefficients, is the reference point still set at the intersection of the vertical centerline of the platform and the SWL?

**the results in .1 matched perfectly :
1735204080444

**comparison of the six-degree-of-freedom responses of the platform :
image

In subsequent tests, I encountered another issue. I attempted to include all wave directions (-180° to 180°, in 10° intervals) in the second-order hydrodynamic parameter files (.12s, .12d) to simulate varying incoming wave directions. However, when I imported the hydrodynamic parameters for all wave directions into the files, the calculation failed, displaying an error that the second-order WAMIT data was "too sparse" (see the figure below). Interestingly, if I only imported the parameters for a single wave direction, the calculation proceeded without any issues.

**error message:
1735222066275

Best regards,

@jjonkman
Copy link
Collaborator

Dear @axxyyyyyyyy378,

When calculating the second-order potential-flow solution for HydroDyn, the full system mass, center of mass, and inertia should be used and the reference point for the hydrodynamic outputs should be the same as the first-order solution. The reference point is user-specified in HydroDyn via PtfmRefxt/PtfmRefyt/PtfmRefzt and need not be the intersection of the vertical centerline of the platform and still water level in HydroDyn, but that point is the reference point used by the OpenFAST model of the DeepCwind semisumbersible used in the OpenFAST r-test.

If your results from AQWA match those from WAMIT without these settings, then perhaps the second-order effect is not playing a large role for the conditions you are simulating.

Regarding the "sparse" error, does the direction and frequency data in second-order solution include all combinations of direction and frequencies pairs?

Best regards,

@axxyyyyyyyy378
Copy link
Author

Dear @jjonkman,

Thank you for your response. Your comments are very professional, and I now fully understand the issue with setting the reference point. As you mentioned, since the environmental loads I simulated were relatively mild, the second-order effects did not play a significant role.

To address this, I made adjustments by using the full system mass, center of mass, and inertia in AQWA, and ensured that the hydrodynamic output reference point matched the first-order solution's reference point. I then "correctly" calculated the second-order hydrodynamic coefficients in HydroDyn. Subsequently, I compared the platform displacement responses obtained using AQWA and WAMIT results in OpenFAST under a set of extreme loads (turbulent wind U = 22.1 m/s, irregular waves Hs = 10.9 m, Tp = 16 s) to amplify the second-order effects. From the results (see the figure below), it is difficult to conclude that the two results align well, as there is a significant discrepancy in the pitch response. This seems to suggest that the second-order hydrodynamic coefficients from AQWA are still incorrect, but I cannot pinpoint the cause of the error. Do you have any other suggestions?

Regarding the "sparse data" error, the second-order hydrodynamic coefficient file (.12d, .12s) that generated the error includes all directions from -180° to 180° at 10° intervals (37 directions in total) and all frequencies from 0.03333 Hz to 0.57716 Hz at 0.01133 Hz intervals (47 frequencies in total). The total number of rows in the file is 533,022, which covers all wave direction and frequency data. However, when I include only a single wave direction in the .12d, .12s file, the total number of rows becomes 533,022/37 = 14,406, and the calculation proceeds without any errors.

**platform displacement responses obtained using AQWA and WAMIT results in OpenFAST:
image

Best regards,

@jjonkman
Copy link
Collaborator

jjonkman commented Dec 28, 2024

Dear @axxyyyyyyyy378,

Regarding the second-order solution calculated by AQWA, have you plotted the QTFs against the equivalent results from WAMIT at 0 degrees wave direction (like you did with the added mass and damping matrices)?

Regarding the "sparse" error, presumably you mean 49 unique frequencies (rather than 47. Regardless, it seems like you are missing a dimension in your QTF. I would expect there to be 6 * nd^2 * nf^2 rows in the QTF, where nd = number of wave directions and nf = number of frequencies. With nd = 37 and nf = 49, this would be 19,721,814 rows; you are missing the combinations of wave direction pairs.

Best regards,

@axxyyyyyyyy378
Copy link
Author

Dear @jjonkman ,

Thank you for your response, but I’m not sure how to post-process the second-order hydrodynamic data or what kind of plots to create for comparison and validation. Do you have any suggestions?

Best regards,

@jjonkman
Copy link
Collaborator

Dear @axxyyyyyyyy378,

For the second-order data at a given wave direction (zero degrees), I would plot the QTF as a contour / surface plot as a function of the two frequencies. The QTF is complex valued; so, you could plot the magnitude and phase or the real and imaginary components. You could scale the data to physical units or keep the data nondimensional.

Best regards,

@axxyyyyyyyy378
Copy link
Author

Dear @jjonkman,

Thank you for your prompt response. I compared the QTF magnitudes for three degrees of freedom (surge, heave, and pitch). However, the results show a significant discrepancy between the reference values from WAMIT and the values I obtained using AQWA. Considering that the first-order hydrodynamic coefficients matched very well previously, what do you think could be the reason for this difference?

image
image

Best regards,

@jjonkman
Copy link
Collaborator

jjonkman commented Jan 1, 2025

Dear @axxyyyyyyyy378,

I'm not sure. The second-order difference-frequency QTF depends both on the quadratic interaction of first-order quantities as well the solving of the second-order potential. Perhaps you could isolate the contribution of each, although I don't believe I have access to a WAMIT solution for the OC4 DeepCwind semisubmersible that separates the two contributions. The diagonal of the difference-frequency QTF (where f1-f2 = 0, representing the mean drift term) depends only on the first-order solution, so, hopefully that compares well (it is hard to tell from your contour plots). If that differs, I would first identify what is causing that.

Best regards,

@axxyyyyyyyy378
Copy link
Author

Dear @jjonkman,

Happy New Year! Thanks for your professional advice. I compared the QTF magnitudes for the three degrees of freedom (surge, heave, and pitch) as quickly as possible when f1 = f2. From the comparison, the overall trend of the curves appears similar, but the results from AQWA exhibit more significant fluctuations and pronounced spiky noise. Could you provide some other suggestions?

image

Best regards,

@jjonkman
Copy link
Collaborator

jjonkman commented Jan 2, 2025

Dear @axxyyyyyyyy378,

I'm not sure why AQWA has more "noise" than WAMIT for the slow-drift terms.

FYI: You could generate similar plots at other difference frequencies by plotting the results for f1-f2 = constant, where the constant is the specific difference frequency of interest.

Best regards,

@axxyyyyyyyy378
Copy link
Author

Dear @jjonkman,

Thanks for your advice. As far as I know, the wave periods worth attention in certain sea areas are roughly between 3s and 16s, so the wave frequencies fall within the range of 1/16 Hz to 1/3 Hz. I have redrawn the QTF magnitudes for the three degrees of freedom (surge, heave, and pitch) for f1-f2 = constant. From the results, the surge shows good agreement, the heave matches less well in the high-frequency range, and the pitch exhibits the poorest agreement. However, it seems that this is the best we can achieve for now. It looks very challenging to completely resolve this issue.

1736225828117

Best regards,

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

2 participants