What is the good time resolution in transient LES solution to capture pressure fluctuation.FFT requires(Nyquist frequency) min. 1/2*deltaT, which is for 1kHz frequiency =5.e-05[sec]!

You need a rough idea what is the frequency of the periodic phenomena you wish to capture... or a limit, say human hearing <20kHz.

And don't forget that you'll need to sample at at least twice the rate of the highest frequency you're interested in.

Unfortunately, my problem is the 10 fans with diferent sweep angle to compare theirs noise.I´m traying with T=1.e-o4 sec time step, and 512 (some older FFT codes reqires 2**n samples) iterations.That is 1/2*0.0001=5kHz, which is abouth highest frequency of mesured fan noise.Total solution time is optained from lowest frequency of captured noise TotalTimeSolution=1/Lowes freq.=1/20Hz=0.05[sec] 0.05/0.0001=500 iterations.

What you say?

Make sure you start from a steady state solution (use MRF method and then export field results via SMAP) or the transient part of the fan's response will really waste your time.

For FFT (or are you using DFT?...) to be accurate, you'll have to wait till the flow conditions are really repeating periodically so you can 'cut' out one repeating segment for FFT analysis. Otherwise, even with windowing, the answer can be spurious at the lower frequencies. Observe the residuals or monitor certain cells.

You might want to use the aeroacoutic model (need to turn on some cheatcode) to perhaps do a steady-state MRF to determine the viscous-shear-induced noise level.

These I think can only give you only relative comparisons of the noise-level between fan designs but to get exact figures you need to couple STAR-CD with some dedicated noise-analysis software like Sysnoise and last I heard, it's still experimental and requires substantial user-subroutine coding. Sysnoise ain't cheap too.

Just to correct that last point, the coupling from STAR to SYSNOISE (and others) is trivial in terms of user set-up, and requires no user coding.

You can get the coupling by going on the aero acoustics training course which covers coupling to acoustic propogation codes and more importantly how to capture acoustics in your cfd simulation.

Actually I was told by my support that it does require user coding. As they did not elaborate and I don't have Sysnoise, I just left it knowing such.

I will tray to import transient pressure field from fan baldes (area averaged ) transfered to frequency domain, in Sysnoise by Direct Boundary Method(DBEM).I´m just confused abouth compresibility , this model doesn´t work fine on Multiple regference Model in LES(KN)...

Just to clarify the STAR to SYSNOISE coupling, the current implementation is done through "user coding", in that it is called through posdat, however all coding and libraries are supplied by CD adapco, no coding by the user is required.

I objective of using MRF in the first place is to get roughly-accurate flow field starting point so that you don't have to wait for 10 to 20 cycles of the fan revolution to be simulated in full transient before you can get steady fluctuations to capture for FFT. So to that effect maybe you can start with a a simpler turbulence model if LES is shaky with MRF.

I was given the impression that it was still experimental until STAR-CD v3.2 and still required a certain depth of user customisation on the usubs. Of course, that was well over a year ago for me and at that time Sysnoise was sporting and was willing to give a free demonstration on coupling their code with STAR-CD.

Some other CFD codes(FLUENT) does it by user libray and transfer the sound in LES calculation tu receiver points, without export to the Sysnoise. Have you user library for STAR-CD?