Guidance on setting up directional inflow conditions

[Robson-Passos]

Hello,
I’m a new user of ARM-Wind and also new to CFD. I would like to combine CFD data with measurements to generate a Wind Resource File (WRG). The ARM-Wind would be used to calculate directional speedups (0°, 22.5°, 45°, etc.).

I would like to run the simulations under neutral atmospheric conditions so that the speedups are linearized ( independent of the Reynolds number). In this case, the speedups would depend only on the orography and surface roughness. Initially, I plan to use the RANS approach. However, I have some doubts about how to proceed (most likely due to my lack of experience in CFD).

I’m running a RANS simulation without a precursor. I used amr1D to generate an initial profile by forcing a wind speed of 10 m/s at 10 m height, and I set CoriolisForcing and GeostrophicForcing under ICNS.source_terms.

For the boundary conditions, I defined the type as "pressure_outflow".

I was able to successfully run the simulation with the geostrophic forcing. However, with this approach, the inflow is not aligned with the desired wind direction. I believe this is due to the Coriolis effect when running amr1D, which causes the wind to be deflected. If I use a precursor I will get the same effect.

I considered applying a logarithmic (Dirichlet) profile using PowerLawProfile
(https://exawind.github.io/amr-wind/user/inputs_Boundary_conditions.html#powerlawprofile), but in this case, the flow always has to be normal to the boundary plane (e.g., xlo), and I think I would need to rotate the topography to simulate different wind directions.

I’m opening this discussion to see if anyone has ideas on how I can impose a logarithmic inflow profile that is aligned with the desired wind direction (0°, 22.5°, 45°, etc.). Maybe there’s a way to define a pressure gradient to steer the flow?

Thanks in advance,
Robson Passos

[hgopalan]

There is an indirect way of doing it. Check the example here: test/test_files/abl_wallrans_neutral/abl_wallrans_neutral.inp

You modify the rans_1d.info file to provide a log law. The file requires following inputs

Z u v w k
...

So you compute your log
Law as

u = u* / kappa * log(z/z0) cos(dir)

v= u* / kappa * log(z/z0) sin(dir)

w= 0

k = u* ^2 / 0.556^2 ( this expression is to be consistent with RANS model)

And use python or matlab to write
The input file. Also when using log law you need to turn off geostrophic and coriolis forcing as they are inconsistent with log law profile which is valid only in surface layer.

[Robson-Passos]

Thank you very much for your response. I tested it and it worked properly. However, I have a question regarding the parameters RayleighDamping.reference_velocity and incflo.velocity
Even though I set ABL.initial_wind_profile = true and ABL.rans_1dprofile_file = "rans_1d.info", are these parameters still mandatory? What values should be assigned to them?

[hgopalan]

Rayleigh is hangover from the file I copied. Incflo.velocity is required during initialization and gets overwritten with profile.