AFMG Network Forum

andrzejhd · Member

Hi again. I'm having a problem with accurate RT60 predictions in a modeled room (big factory hall). I did some physical measurements of this space and RT60 (T20) times range from 2.7s (125Hz) to 3.6s (1.6kHz).

I managed to calibrate the model with 15% accuracy using Local Decay Time, but Ray Tracing (Probe module) gives me values that are much lower and the 'ratios' between frequencies change as well.

What RT60 model should I use? Ray Tracing with high order and high density is the recommended method, right?

Thanks for help,
A. Wink

andrzejhd · Member

Anyone?

Bruce · Member

The recommended method is to use AURA.

andrzejhd · Member

Thanks for your reply! Unfortunately my budget doesn't allow for the AURA module. Sad

thomas · Member

Hi Andrzej,
possible reasons for too short rev. time (simulated) are:
- simulated impulse response is too short
(IR has to be at least 40 dB long!
-> >=2/3*expectet RT + "flighttime" of the direct sound
(important in realy large rooms))

- wrong distribution of absorbing faces in the model
(the effectiveness of an absorbing face depends on radiation strength!
A fact, that is'nt taken into account in statistical methods (Sabine, Eyring)).
And: in a large factory hall is it possible, that no absorbing material is on
the ceiling and on the walls, all is "on the floor", so sound that travels
only between the walls never sees any absorbing face and the measured
RT is higher than with Sabine predicted.
So make sure, that your absorbing materials are in the right position.

Another possible problem occurs, when scattering and diffraction are
really important but simulation only takes specular reflections into
account. But then, the simulated RT should be too long.

Best regards
thomas

andrzejhd · Member

Thanks Thomas! I'll give it a go...

Btw. Could you guys recommend any books that are addressing simulation issues? I can find a lot of books on general acoustics, but there is not so much 'simulation specific' stuff.

Thanks,
Andrzej

thomas · Member

Hi Andrzej,
there is an EASE-tutorial I would recommend (look for "ease afmg tutorial").
And then simulate rooms, you always know.
Start with very simple rooms. Always think about
the results ("What happens in reality, what does the
simulation, what is the difference (if there is one)?").
Good luck
thomas

AFMG Pedro Lima · Forum Moderator

Hi Andrzej,

The most related books I know are probably:

Vorlaender, Michael. Auralization: Fundamentals of Acoustics, Modelling, Simulation, Algorithms and Acoustic Virtual Reality.

Feistel, Stefan. Modeling the Radiation of Modern Sound Reinforcement Systems in High Resolution.
http://www.logos-verlag.de/cgi-bin/engbuchmid?isbn=3710&lng=eng&id=

The second is more specifically related to modelling of sound sources (usually of interest for manufacturers who would like to insert their speakers in EASE or Focus).

Kind Regards,
Pedro Lima

Agustín Arias · Member

Did you add a random tail to your reflectogram? can you post a picture of your reflectogram?

andrzejhd · Member

Thank you Pedro!

I'm still working on my model guys. As soon as I finish, I'll try to post the reflectogram.

PS. Do you know how to deal with coefficients bigger than 1? I've been trying to find some info on that and I get the physics of it, but don't know how to normalize it in a right way.

Thanks,
A.

Agustín Arias · Member

If you have values greater than 1, use 1, which means total absorption. Values greater than 1 appears in measurements of absorption coefficients in laboratory when, for instance, the sample under test has rigid boundaries (edge effect). So it has to be set to 1 for EASE

andrzejhd · Member

Yeah, but should I scale all the coefficients in a "smooth way" or just "clip" the ones that are bigger than 1?

I'm testing the AFMG SoundFlow and when I specify the sample size - some coefficients are as big as 3.6... Sad

Agustín Arias · Member

Yes, "clip" the values greater than one to 1. Do not modify the others coefficients

andrzejhd · Member

Cool. Thanks for that Agustin! Very Happy

Agustín Arias · Member

andrzejhd, sorry but I've just read your previous comment!:

"I'm testing the AFMG SoundFlow and when I specify the sample size - some coefficients are as big as 3.6".

That result is very weird. I though that you were using an absorption coefficients database where some materials could show values such as 1.2. But 3.6 is too much!!! My Soundflow demo license has expired so I can not check that, but I'm pretty sure that the values calculated with Soundflow are always between 0 and 1.

Maybe that value (3.6) is the surface absorption coeficient = absorption coefficient*surface