MoorDyn Response in Pitch/Roll

[Francisco.Pimenta]

Hi everyone,

I am currently trying to compare the influence of different sea states and wind conditions in the fatigue loads of the mooring lines of the 5MW Semisubmersible OC4 DeepWind Model. Before proceeding to further analysis, I've made a few tests in both ANSYS Aqwa and FAST in which I've provided the platform with some initial conditions and monitored the response. For the sway or surge displacement both the response and the cables forces where well predicted by both models:

T1f.png (28.58 KiB) Viewed 2020 times

However, when it comes to the picth or roll test the same does not happen. In this situation, the cable forces experience some high frequency variation that I was unable to explain:

T22f.png (40.02 KiB) Viewed 2020 times

They don't seem to affect largely the response however:

T22.png (24.8 KiB) Viewed 2020 times

I have already set the Moordyn damping to avoid numerically instability but i don't believe this to be the cause, as the problem only appears in the pitch/roll tests.

Does anyone have an explanation to this?

Thanks in advance

Francisco Pimenta

[Matt.Hall]

Hi Francisco,

The frequency of those tension oscillations could be affected by hydrodynamic drag as well as the internal damping that you mentioned, so I'd double check drag coefficients just to be safe. That said, because the oscillations seem to decay more rapidly than the pitch motion, I wonder if there could be an issue in how the initial 10 degree roll offset is handled by MoorDyn at initialization.

Could you share your MoorDyn input file? And, if you could you let me know what version of FAST and MoorDyn you are using, I can check if there is any initial condition oversight in the code.

Best,
Matt

[Francisco.Pimenta]

Thank you very much. I am currently using FAST v8.16 and the MoorDyn file is in the link below.

https://we.tl/t-PG7gByeyM1

I tried other combinations and even the response to JONSWAP spectrum generated waves seems to have this behaviour when compared to the response obtained from ANSYS. I tried decreasing the threshold for convergence and it seems to help, but I was wondering why this does not happen in the translational tests.

Thank you for your time