Velocity field jumps at AMR levels interface

[euzenatf]

We aim at reproducing the stable SWIFT benchmark using AMR-Wind.
We follow a precursor > turbine simulation workflow to account for ABL conditions. The wind turbine is simulated using an ALM approach with FLLC. We are using AMR-Wind 3.1.4.

Once the inflow conditions are matched, the Vestas V27 (hub height: 32.1m, diameter: 27m) is included in the domain. Two additional mesh refinement zones surrouding the wind turbine are included to capture the dynamics close to the turbine. Some unphysical "high speed" zones are developing at the finest AMR level interfaces (i.e. the finest level being the one containing the turbine, see below) propagating downstream in the domain (increasing the domain size in X does not seem to solve the issue).

More details about what is happening at the interfaces is shown below. The 2 finest refinement zones are highlighted in the top-right figure for an instantaneous flow snapshot (the "high speed" zones are also visible for instantaneous velocity snapshots). A plane through the domain is then drawn and displayed on the bottom-right part. Some changes in the "mean_velocityY" and "mean_velocityZ" fields are clearly seen right at the AMR-level interfaces.

[hgopalan]

Can you share the input files for us to look into it? Did you observe this for neutral and stable ABLs too? @mchurchf @rthedin

[michaelasprague]

@moprak-nrel @mbkuhn -- wanted to get your eyes on this too.

[euzenatf]

Please find below the input files

abl_stable.txt

alm_stable.txt

[moprak-nrel]

Hi @euzenatf , this is really strange behavior. Could you also share the output logs from the ABL/ALM runs? Want to make sure there isn't some type of convergence issue.

[euzenatf]

Hello,

Thanks for checking our case. Here are the logs.
Due to file size limit constraints, I only kept the first 50s and last 70s used for the simulation (job was cancelled at the end because we had enough data).
Please let me know if more details are required.

logs_short.txt

[moprak-nrel]

Hi @euzenatf, I have been looking through ALM input file and there are a couple things that could be causing the issue. Is there any way for you to share the ABL checkpoint file and boundary planes with me so I can test these out?

In particular I want to see if the level 4 and level 5 refinement zones being so close might be the issue:

tagging.g4.g4.origin = 110. 25. 0.
tagging.g5.g5.origin = 120. 35. 0.

[github-actions]

This issue is stale because it has been open 30 days with no activity.

[github-actions]

This issue is stale because it has been open 30 days with no activity.