[MardyOwens]

Hi,
I am running a case in a CFD LES software with implicit filter.
Have you ever experienced an aliasing due to the grid? The spectra in frequency with a coarser grid seems to have bigger value of energy then the same case with a finer grid. Can this be due to an aliasing caused by the grid?

[FMDenaro]

Originally Posted by MardyOwens

Hi,
I am running a case in a CFD LES software with implicit filter.
Have you ever experienced an aliasing due to the grid? The spectra in frequency with a coarser grid seems to have bigger value of energy then the same case with a finer grid. Can this be due to an aliasing caused by the grid?

It could be an effect of aliasing (but what about your discretization of the non-linear terms?) but that could be one among other possible causes.
Could you provide details of your simulation and post the figures of the two spectra?

[MardyOwens]

If I have an aliasing, is it possible that some terms that should have been counted as sub-grid scales, are instead counted in the lower frequency?

[FMDenaro]

Originally Posted by MardyOwens

If I have an aliasing, is it possible that some terms that should have been counted as sub-grid scales, are instead counted in the lower frequency?

Aliasing means that the non-linear quadratic terms produce components behind the cut-off grid frequency, that is in the range of the unresolved terms.

But do not forget that the SGS model by itself acts adding a contribution only on the lower resolved frequencies. Thus, also some problems in the SGS model could generate an increasing in the resolved energy level.
Without details I cannot say more.

[gnwt4a]

u r probably using finite differences which means aliasing from high wavenumbers is not an issue. also, the unpredictability of low resolution simulations based on truncation error for stabilization means that u must increase resolution 'sufficiently' before start considering fanciful alternatives.


also, yr domain may be too short but u have given shod-all info on yr simulation.
--

[FMDenaro]

Originally Posted by gnwt4a

u r probably using finite differences which means aliasing from high wavenumbers is not an issue. also, the unpredictability of low resolution simulations based on truncation error for stabilization means that u must increase resolution 'sufficiently' before start considering fanciful alternatives.


also, yr domain may be too short but u have given shod-all info on yr simulation.
--

"using finite differences which means aliasing from high wavenumbers is not an issue"


why do you wrote that?