Force and Pressure Coefficients Correct, but acoustic predictions are still miles off?

[niall-oneill]

Hi,
I am trying to predict the noise from a NACA 0012 airfiol, for which I have information on the pressure and force coefficients. I have set up a simulation and am getting almost identical values for these coefficients. This led me to be confident that the acoustic predictions would be reasonably similar.

My controlDict, and other acoustic files are available at https://github.com/niall-oneill/openFOAM/tree/main. I have more or less used the defaults that I saw in other tutorials aside from things that were unique to my case like the c0, dref (width of domain). Some setting that i left as default, but am unsure of are U0 and Ufft.

For curle's analogy, I am just getting a more or less constant value, and for Fwh, I am getting negative values, that do not resemble the experiment (experimental and simulated data is attached below as well)

My time step is quite small, but maybe not enough, and perhaps my mesh could be finer. My question really is what would the most important parameters to accurately describe the SPL with a sufficient resolution be? I cannot seem to find adequate information online about this.

For curle's analogy, I am just getting a more or less constant value, and for Fwh, I am getting negative values, that do not resemble the experiment.
Thanks,
Niall

[mkraposhin]

Hello, sir!

1. FWH and Curle analogies are based on the change of momentum and force across a given surface. Therefore, if your flow is steady state, then you are not to get any noise.
2. Both analogies are 3D . That means, if you want a reliable result, then you should setup 3D case. The library has some functions to run in 2D, but I'm still not sure whether the theory behind 2D version of the analogies implementation is correct.

[niall-oneill]

Hello, sir!

1. FWH and Curle analogies are based on the change of momentum and force across a given surface. Therefore, if your flow is steady state, then you are not to get any noise.
2. Both analogies are 3D . That means, if you want a reliable result, then you should setup 3D case. The library has some functions to run in 2D, but I'm still not sure whether the theory behind 2D version of the analogies implementation is correct.

Hello, thanks for the reply.

Previously, I set up a cylinder vortex shedding case, in which I was able to get closer to the experimental data, even with a 2D mesh. Based on this, I assumed that an airfoil would also be suitable. However fromwhat you said, it makes sense that the cylinder would be a more appropriate example, since the vortex shedding is unsteady and would produce more noise than an airfoil.

From reading this paper: https://acoustics.ippt.gov.pl/index.php/aa/article/viewFile/1765/pdf_303
who validated the noise on an Airfoil using ANSYS, I assumed that there would be no issue, however it seems from closer inspection on the paper that the SPL equations may be fundamentally different to those used in libAcoustics.

[mkraposhin]

Well, the flow in your case is turbulent and this phenomena is substantially 3D. Hence, some justifications are needed for usage of the 2D domain, because otherwise you can't obtain proper flow fields.

Secondly, as I told, FWH/Curle analogies are 3D, therefore you must derive an approach or approximation how to convert 2D results into 3D. For example, as a first step, you can say that your 2D profile contains results for thin layer with with dx and if total width of the 3D airfoil is W, N=W/dx, then perhaps you can assess SPL using assumption as several (N) non-coherent sources: p'_tot=sqrt(N)*p'_2D

Of course, p' and p'_2D are pressure fluctuations in Pa, not in dB.

[mkraposhin]

And before doing CAA simulations, I strongly recommend you to validate aeroacostics model. At least, values and fluctuations of Cd and Cl should match experimental results for this Re number.