[a_habib]

Thanks.
Actually my problem seems to be in installing the fftw-3.3.3
the permission for righting files to /user/local ... is denied
But I'll try it again using your tips.

[a_habib]

Thanks Alot, It worked

[a_habib]

I'm Wondering if some one used this library.It's calculate the acoustic pressure and the fft to calculate the SPL sv. frequency, my question is at any point does it calculate the fft? at the end time? or during the runtime solution????

[a_habib]

Hello FOAMERS
I need a validation case to validate libAcoustic (something simple unlike the Tandem Cylinder case)
Any Help

[skeptik]

Here is a work about cavity aeroacoustic measurements, for instance

http://www.lr.tudelft.nl/fileadmin/F...ant_Parkhi.pdf

[maHein]

Hello,

I was wondering why the libAcoustics does calculate an average distance between surface and observer? Why is the distance not evaluated for each face separately?

Kind regards,

Martin

[skeptik]

Originally Posted by maHein

Hello,

I was wondering why the libAcoustics does calculate an average distance between surface and observer? Why is the distance not evaluated for each face separately?

Kind regards,

Martin

True. Sadly i've not developed this version of the Curle analogy, i've only adopted for 2.3 to work with it. May be we will make a revision and develop a new version.

It was found at least one error in the old realization of this library.

[maHein]

I am currently working on reimplementing Curle for OpenFOAM 2.4.0 to get a better understanding of acoustic analogies on the way to Ffowcs-Williams-Hawkings. That's why I came up with the question...

[skeptik]

Originally Posted by maHein

I am currently working on reimplementing Curle for OpenFOAM 2.4.0 to get a better understanding of acoustic analogies on the way to Ffowcs-Williams-Hawkings. That's why I came up with the question...

It's nice question and nice challenge. We're developing FWH in addition to Curle now and if you find some errors in libAcoustics, please, report. I hope we will have a time to review old versions and to make a test-cases and validate our results more.

[a_habib]

What is the ((dRef)) in the lib code??
Is it the the long in the z direction if the simulation is 2D [x,y]
If so, what I should use for it in the 3D simulations

thanks in advance

[maHein]

If you set dRef to zero, it is not considered in the calculations. If you set the value, the pressure fluctuations are divided by dRef.

I think this is only used for 2D simulations.

[skeptik]

Originally Posted by a_habib

What is the ((dRef)) in the lib code??
Is it the the long in the z direction if the simulation is 2D [x,y]
If so, what I should use for it in the 3D simulations

thanks in advance

If you solve 2D problem dRef is your mesh dZ.
If you solve 3D problem assign to dRef any negative value.

[a_habib]

Hello.
I tried to validate this analogy using the paper ((Two and Three-Dimensional Simulation of Sound Generated by Flow Around a Circular Cylinder - 30th AIAA Aeroacoustics Conference)).
every thing was good as you can see my results except when I calculated the SPL & OASPL for different cylinder span length. The result where on the opposite, the higher SPL in my calculation where related to the shortest cylinder (in the paper where related to the longest cylinder).
any one can tell me why this happened.
you can check my controlDit


SPL.jpg

controlDict.txt

Results1.jpg

Results2.jpg

[maHein]

I simulated the NASA tandem cylinder validation case using a Spalart Allmaras DDES turbulence model with a span length of 3 times the diameter. The results are quite satisfactory and similar to other publications on this case.

[skeptik]

Originally Posted by maHein

I simulated the NASA tandem cylinder validation case using a Spalart Allmaras DDES turbulence model with a span length of 3 times the diameter. The results are quite satisfactory and similar to other publications on this case.

Very nice results.

I have few questions to clarify DDES-related problems:
1. Did you measure y+? How much is it?
2. What's about wallFunctions? Did you use it? (connected to 1. question)
3. Can you show your fvSchemes file? Or at least say what was your grad and div schemes?

Thank you in advance.

[maHein]

Hello,

the mesh contains contains 8 mio cells with an average y+ value of about 1.9 for the first and 1.4 for the second cylinder. I also prepared two finer meshes but did not test them yet. For nuSgs I used the nutLowReWallFunction and fixedValue 0.0 for nuTilda.

My fvSchemes looks as following:

ddtSchemes
{
    default         backward;
}

gradSchemes
{
    default         Gauss linear;
}

divSchemes
{
    default          Gauss linear;
}

laplacianSchemes
{
    div(phi,U)       Gauss LUST grad(U);
    div(phi,nuTilda) Gauss linearUpwind grad(nuTilda);

    div((nuEff*dev(T(grad(U))))) Gauss linear;
}

interpolationSchemes
{
    default         linear;
}

snGradSchemes
{
    default         corrected;
}

fluxRequired
{
    default         no;
    p;
}

I tried Gauss linear for div(phi,U), but it caused massive checkerboarding. Maybe a less diffusive scheme for div(phi,nuTilda) could also be tested.

Regards,

Martin

[skeptik]

Originally Posted by maHein

Hello,

the mesh contains contains 8 mio cells with an average y+ value of about 1.9 for the first and 1.4 for the second cylinder. I also prepared two finer meshes but did not test them yet. For nuSgs I used the nutLowReWallFunction and fixedValue 0.0 for nuTilda.

My fvSchemes looks as following:

ddtSchemes
{
    default         backward;
}

gradSchemes
{
    default         Gauss linear;
}

divSchemes
{
    default          Gauss linear;
}

laplacianSchemes
{
    div(phi,U)       Gauss LUST grad(U);
    div(phi,nuTilda) Gauss linearUpwind grad(nuTilda);

    div((nuEff*dev(T(grad(U))))) Gauss linear;
}

interpolationSchemes
{
    default         linear;
}

snGradSchemes
{
    default         corrected;
}

fluxRequired
{
    default         no;
    p;
}

I tried Gauss linear for div(phi,U), but it caused massive checkerboarding. Maybe a less diffusive scheme for div(phi,nuTilda) could also be tested.

Regards,

Martin

Why laplacian

laplacianSchemes
{
    div(phi,U)       Gauss LUST grad(U);
    div(phi,nuTilda) Gauss linearUpwind grad(nuTilda);

    div((nuEff*dev(T(grad(U))))) Gauss linear;
}

?

May be

divSchemes
{
    div(phi,U)       Gauss LUST grad(U);
    div(phi,nuTilda) Gauss linearUpwind grad(nuTilda);

    div((nuEff*dev(T(grad(U))))) Gauss linear;
}

?

[maHein]

ups, that's my mistake! I mixed up the settings... Now I have to rerun everything ...

[skeptik]

Originally Posted by maHein

ups, that's my mistake! I mixed up the settings... Now I have to rerun everything ...

Wow.
We spent a huge time and resources investigating different numerical schemes, may be we should try something like yours. Despite such errors you get good results, something in your fvSchemes was ignored but something not) will see.
Which version you use?

[maHein]

I'm using OpenFOAM 2.4.0.

I'm currently rerunning the simulation with the correct settings. I can show you the results in a few days.

Regards,

Martin