[Olympical]

Hi Guys,

I am a newcomer to the world of computational fluid dynamics so please bear with me. I have been playing with a channel flow similar with periodic boundary conditions in the x and z direction. Now I am trying to compute the 1D and 2D energy spectrum of the channel. As far as I understand it simply a fourier transform for the time series of the time series of the directional velocities for the 1D case. Is this the case ? I have also outputted the results of my simulation every 2000 steps, would I need to have more frequent output for the purposes of the energy spectrum?

Thanks so much for your help
David

[FMDenaro]

Have a look to similar questions, for example

Energy spectrum-wrong spectrum shape

[Olympical]

Thanks for the reply, sorry I missed the previous post which I have now gone through and it has been an immense help.

One question that pops up when I am trying to perform these calculations myself. If I am trying to compute the spectra along Kx, I am taking a constant-y plane then performing fft over the time series of each z value correct? What I am confused about is that each z value also has a corresponding x values so am I performing the fft at each grid point in the y-plane? If it's only for each z values then what do I do with the points along a certain z.

Thanks very much for your patience, I am still quite new to CFD in general so these questions might seem a bit basic.

[FMDenaro]

Originally Posted by Olympical

Thanks for the reply, sorry I missed the previous post which I have now gone through and it has been an immense help.

One question that pops up when I am trying to perform these calculations myself. If I am trying to compute the spectra along Kx, I am taking a constant-y plane then performing fft over the time series of each z value correct? What I am confused about is that each z value also has a corresponding x values so am I performing the fft at each grid point in the y-plane? If it's only for each z values then what do I do with the points along a certain z.

Thanks very much for your patience, I am still quite new to CFD in general so these questions might seem a bit basic.

The order can be the following:


1) fix a time and use the velocity field at this time
2) fix a position y (distance from the wall)
3) Compute the spectra in the plane xz, that is the spectra along kx (kz) for all the discrete positions z (x) of the grid.
4) Perform the ensemble averaging in z (x) of the spectra along kx (kz)
5) Repeat from 1) at a different time


After you have a sufficient sample of spectra in time, perform the ensemble averaging in time.


As final results you can show the spectra Euu,Evv,Eww both in kx and in kz.