Hi,

I want to calculate the turbulent kinetic energy spectrum from my LES data to do one step in the validation process, but I am little bit confused about it.

There are two choices, one is to create a energy spectrum in space, where I have to find a homogenous direction, and the other to create a spectrum in time, where I have to choose a certain point in domain.

The problem is, which variant should I use and esp. which plane for the spectrum in space and which point in a surface-mounted cube simulation?

To calculate the spectrum I have to caculate the autocorrelation or cross-correlation, depending on the choice space/time. There exist an octave/matlab function 'xcorr' which can do that, but I would like to do that without using that function and without the use of FFT.

Would be nice if anybody has an idea!

Fabian

For space: invoke Taylor's frozen field hypothesis, For time: simply take FFT of the velocity signal

Hi,

thanks, do you have an idea which point (time) or plane (space) I should choose for a surface mounted cube calculation. I mean, I could try to find the best point which gives me Kolmogorovs -5/3 slope...

Fabian

In order to have a chance to find a -5/3 slope generally related to a wall-free turbulence at equilibrium, I suppose that the measurement point should be located in the wake and a few cube length away from the aft wall.

Hi Lionel,

thanks for the information. I see how this works.

Greetings! Fabian