Hi,
I am going to simulate a 2D aeroacoustics simulation in Fluent using a DES model on a stator. Actually, the simulation should be performed on a complete 3D rotor-stator setup but due to some limitations, I don’t have access to the rotor geometry and computing power. A simplified model of my setup is shown in the attached picture.
I have been searching for parameters that I can extract from the DES simulations. These are the following parameters that I can think of:
1)A common parameter is the FFT of pressure data recorded in the wake region of the blade. Using FFT, I want to see if I can see tonal noise peaks or broadband peaks. Based on these observations I can characterize the sound sources such as vortex shedding or just random turbulence.
2) Another parameter that I have come across is the sound pressure which can be calculated by subtracting the instantaneous pressure from the average pressure. How this can be done in Fluent, I am not sure.
3) The other is " divergence of velocity" or dilatation contours. I found it in this article
https://www.researchgate.net/publication/268583775_Affordable_Compressible_LES_of_Airfoil-Turbulence_Interaction_in_a_Free_Jet
I am not able to understand what's the significance of this parameter.
4) Another quantity that can be visualized is the normalized Q-criterion, but this will be good to see the turbulence sources and not aeroacoustics as such.
5) Also, I was wondering if one can actually extract the noise from a pressure signal at a point so that you can hear it. I came across this in the STAR-CCM+‘s user guide some time ago and they say that it can be done. The guide mentioned that that one needs to perform inverse FFT of the signal.
Perhaps, people with experience in aeroacoustics can suggest me any additional parameters that can be used. Although the near-field region sound is of interest and not far-field, I am open to using the FW-H model if it provides additional insights.
Dear Ishan Nande,
Here are my suggestions:
1) a 2d simulation of turbulence might be questioned by many people
2) If you extract the pressure signals from the wake this will not be purely noise, but noise+flow. To talk about pure acoustics you need to study it out of the flow, in the far-field.
3) a calculation of the Overall Sound Pressure Level (OASPL) is based on a value of the Root Mean Square (RMS) of pressure fluctuations p'rms [Pa] (referenced to the treshhold of human hearing 2*10^-5 Pa): OASPL dB = 10 log (p'rms^2/p'ref^2) dB. The Sound Pressure Level is a frequency dependent parameter. To get it you need to make a DFFT of p'(t) [Pa] first and transfer the result to the logarithmic scale and get power spectrum of the signal (exact formulas to do that you can get from Matlab examples). As a result you get SPL(f). If you integrate SPL(f) you should get OASPL dB (valuable check).
4) FW-H model is designed for MOVING (e.g. rotating) surfaces in open space...
I hope it was usefull.
Oskar S.
Dear Ishan Nande,
If you are interested in separating the acoustic and turbulent pressure fluctuations, you could use wavenumber filtering. Depending on the mean flow velocity they appear in two distinct wavenumber domains. However, in order to get a good wavenumber resolution you need to detect the Fourier-transformed pressure fluctuation over certain axial extend, which might be difficult if the numerical grid is not large enough. Also the mean flow field should not vary strongly in the region, where the wavenumber filtering is applied.
I hope it is useful.
M. Behn
Ishan,
Following up on M. Behn's answer, you might want to look at "INFINITE BEAMFORMING: WAVENUMBER DECOMPOSITION OF SURFACE PRESSURE FLUCTUATIONS" by Stefan Haxter and Carsten Spehr (http://www.bebec.eu/Downloads/BeBeC2014/Papers/BeBeC-2014-04.pdf). I think Stefan's method would be easy for you to implement.
Bob
- Badges
- Science topic