Kosovic Model - Multiphase Support

[hgopalan]

Summary

The existing LES models overpredict turbulent viscosity near the interface for multiphase flow simulations. This arises due to the inconsistency between the ABL profile and the profile specified in the ocean. To enable turbulence in the ocean, there has to be a consistent initial profile between atmosphere and ocean to avoid over-prediction of the turbulence diffusion terms due to initial condition generated shear and buoyancy production.

The multi-phase model handles this incompatibility and ensures that the diffusion term do not cause a drop in the time step. The following changes are made

1. A new density production term is added and added to the calculation of the turbulent viscosity. This term ensures that the turbulent viscosity is driven to zero at the interface where there is a vertical density gradient.
2. A horizontal damping term to turn off turbulent viscosity within the strong wave region. Wave motion is close to solid body motion and we cannot have the turbulence diffusion term inside it.
3. A water density term to turn off the turbulence diffusion in the water region. Currently a uniform current speed and temperature value is specified in the water region. This causes inconsistency in the diffusion term at the interface. This term ensures that the inconsistency does not cause solver divergence or reduction in time step.

The multiphase solver runs at 1/3rd of the time step when compared to the single phase immersed forcing wave model. The maximum possible Aspect Ratio is ~ 2-3. Solver fails at AR=4 with JONSWAP spectrum.

Please check the type of change introduced:

• [ X] Bugfix
• [X ] Feature
• Code style update (formatting, renaming)
• Refactoring (no functional changes, no api changes)
• Build related changes
• Documentation content changes
• Other (please describe):

Checklist

The following is included:

• new unit-test(s)
• new regression test(s)
• documentation for new capability

This PR was tested by running:

• the unit tests
  • on GPU
  • on CPU
• the regression tests
  • on GPU
  • on CPU

Additional background

Linear Waves

Streamwise Windspeed X-Y plane

Turbulent Viscosity

Vertical Velocity at z=16 showing wave-aligned streaks

Vertical Velocity at z=48 showing ABL streaks

Issue Number: #1462

[hgopalan]

@rybchuk Thank you for the test cases. It was very helpful.

[hgopalan]

Adding the changes to the Kosovic model would make the code too cluttered. This was the reason for a separate model.

[hgopalan]

Added test case

[hgopalan]

documentation done