Wind Noise through Windshield 3

[saurabhsuresh]

Hi Guys,

Im trying to model a part of the windshield in an impedance tube to check the noise reduction across it. When I model it in ANSYS, the noise reduction seems to change with the area of cross section. But in reality, no matter what the area of cross section is, the noise reduction should be the same at one particular frequency right?
Can anyone tell me what I am doing fundamentally wrong?

Any help would be appreciated
Thanks!

 

[IRstuff]

Why do you think there's no area impact? The resonant frequency of an object is related to its dimensions, isn't it?

TTFN

FAQ731-376

 

[saurabhsuresh]

Yes that is exactly what I was thinking as well. But if I want to to the same test at higher frequencies (say 5000-8000Hz), the tube inlet size has to be reduced to accommodate this frequency which in turn reduces the sample size. But then the resonant frequency of the smaller sample reaches almost 44,000Hz on calculation and ANSYS shows that at lower frequencies of interest it is so stiff that it does not allow any noise to pass through. Which is not the case if we do it in an experimental setup. Even at high frequencies, there are noise transmission through.

So is there anything fundamentally wrong when I am doing an acoustic analysis of the sample?

 

[IRstuff]

That may have little to do with the screen and more to do with how you might have actually installed it in the system, or a number of other semi-related things.

TTFN

FAQ731-376

 

[saurabhsuresh]

Right..so all i want to do is check the noise reduction across the glass sample...say Im not considering a windscreen. Jst to know how the sample would react at different frequencies....
Would you know how to test it acoustically?

 

[GregLocock]

Easiest way is in a sound transmission suite, essentially two reverberant rooms separated by an aperture. You put the test item in the aperture, make a noise in one room, record the noise in the other room, and you can work out the transmission loss. That is how many architectural acoustic treatements are measured. Smaller self contained machines like APAMATs do the same thing, albeit with less authority.

I must confess I struggle with the idea of measuring the acoustic properties of a thing like a windscreen using a small disc in an impedance tube, since in practice its transmissibility will be controlled by its mounting, and bending modes etc, rather than very much that is innate in the material.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[saurabhsuresh]

Thanks Greg.... Yea i know that would work, bu the thing is, I am tryin to simulate this condition in ANSYS....So if you would happen to know how I could go about that in ANSYS that would be really helpful!..Thanks again

 

[hacksaw]

a model of propagation/insertion loss of a windscreen consists of propagation of longitudinal waves through the thickness, while your description of the ANSYS model suggests that it is set up to describe is flex modes mainly in transverse plane.

What meshing are you using relative to the windscreen thickness.

 

[saurabhsuresh]

What you said about longitudinal waves makes a lot of sense..and yes a i realized that the sound being transmitted based on the vibration modes. When you say what meshing am I using..I don't quite understand what you mean. I mesh it regularly using quad mesh with one element along the thickness (which is 5mm thick compared to 0.5m x 0.5m area). Is that ok?

Would you be able to guide me as to how to model the longitudinal waves passing through? because i am facing problems when I try to go to higher frequencies by reducing the sample size. And the reason I am guessing is the flex modes in the transverse plane like u mentioned. Any ideas would really help.... Thanks hacksaw

 

[MikeyP]

If the transmission loss is dominated by the non-resonant (mass law) transmission path (as it will be below the critical frequency), then the dynamic stiffness properties will not be important.

What is the critical frequency for your sample?

M

--
Dr Michael F Platten

 

[saurabhsuresh]

Thanks for the response Dr Platten....the critical frequency is around 3000Hz. So I was un der the impression that no matter what the size of the sample, the Transmission loss would remain the same and I would see a dip around that frequency. But my ANSYS model seems to be governed by the resonant frequencies which change according to the size of the sample.
Any ideas?

 

[hacksaw]

with propagation you need more than one element in the thickness direction, even static deformations require 4-5

 

[saurabhsuresh]

ok...i think i tried meshing it with more elemnts along the thickness and that did not seem to change the result, because ultimately it is still vibrating the panel and the nodes are still transmitting the pressure from one side to the other through flex modes of the plate

 

[hacksaw]

true, that is what your simulation describes. your experiment, however, invites a more complex model.

propagation requires more that one element in the thickness direction, otherwise how do you handle the boundary conditions at the surfaces.