[scoop_them_mids]

Hi,


To check the resolution of turbulence in an LES simulation, I would like to plot the turbulent power spectrum density against frequency (the famous one a -5/3 slope), but never attempted this before. All I have is the time-averaged resolved TKE. Is there a way to do it in Paraview or some other script?



Thanks

[scoop_them_mids]

Ok, found some clues here:
Energy spectrum calculation

[LuckyTran]

You can't do it with just a time-averaged tke. The time-averaged tke is just one number, e.g. 1. It has no spectrum.

You need the time-accurate tke, or the time-accurate velocity.


In paraview, i'm not sure. But in general, you just take the velocity into Fourier space and then square it in Fourier space, or take the kinetic energy into Fourier space directly. Basically, you need to do an DFT/FFT and then plot it.

[scoop_them_mids]

You are right, also I think Paraview has an FFT function but it is probably best to probe the transient data at one particular location during runtime, and then post-treat a text file directly with an external script.

[FMDenaro]

Originally Posted by scoop_them_mids

You are right, also I think Paraview has an FFT function but it is probably best to probe the transient data at one particular location during runtime, and then post-treat a text file directly with an external script.

The spectrum can be evaluated in time at some location but you have to enforce a statistical averaging over several period.
If your problem has some directions of homogeneity you can do the spectrum along the space.

[scoop_them_mids]

Originally Posted by FMDenaro

The spectrum can be evaluated in time at some location but you have to enforce a statistical averaging over several period.
If your problem has some directions of homogeneity you can do the spectrum along the space.

Someone mentioned that in the other conversation (plot k against wavenumber instead of frequency). One aspect of my problem is checking the shear layer axis in a wind tunnel, definitely worth trying...