AFMG Network Forum
AFMG Network Forum Index -> EASE 4 -> Rev. Time vs. C80, C50 and other parameters
Post new topic  Reply to topic View previous topic :: View next topic 
Rev. Time vs. C80, C50 and other parameters
PostPosted: Wed Jun 06, 2007 10:49 am Reply with quote
Michal Kaminski
Member
 
Joined: 06 Jun 2007
Posts: 2
Location: Wroclaw, POLAND




I have a large project to do. I need to simulate acoustics in Operahouse in Wroclaw (Breslau). In 2006 the Operahouse was modernized, so acoustics has also changed. Measurements has been made few times during modernization. I know the results of measurements at some stages of workings:
1) Auditorium without listeners seats (without scene and with closed orchestra pit),
2) Only scene,
3) Auditorium with listeners seats (without scene and with closed orchestra pit),
4) Auditorium with listeners seats (without scene and with open orchestra pit),
5) Auditorium with the scene (and open orchestra pit)
So pretty much.

All measuring and simulating emitters/receivers are in the same place - at the auditorium (excluding 5) of course).

This kind of comparing is very helpful and can teach of what is wrong with the model and how to make models that fits the reality, but...

Everything is ok at the first point 1) and 2), almost every coeficient (T30, T20, EDT, C80, C50, G and others) are quite correct. Problems start during simulating listeners seats (point 3)) and everything is getting worse at next stages.
Most important is in this case T30 coefficient and when I choose proper geometry (in my opinion) and absorption coefficients, T30 is ok, but other parameters gets crazy (especially C80, C50).
For example - differences in C80 (in every 1/1 oct. band) between measurments and simulations at points are:
1) no differences,
3) about 2,5 dB more in simulations than measurements,
4) about 3 dB more in simulations than measurements,
5) about 6 dB more in simulations than measurements.

Only T30, T20, EDT are quite correct.
Can anyone tell me how are these parameters calculating in EASE (4.1)? What kind of algorithms or anything what could help me explain these differences. Maybe something strange about calculating reverb-tail or any other strange things connected with impulse response.
Thanks.
View user's profile Send private message Send e-mail
Rev. Time vs. C80, C50 and other parameters
PostPosted: Mon Jun 11, 2007 10:45 pm Reply with quote
Bruce
Member
 
Joined: 19 Apr 2005
Posts: 459
Location: Minneapolis, MN, USA




Hello Michal,

Welcome to the forum.

I am not quite sure what you mean for 3 and 4. Have you added the seats to the model as faces? When the seats were added for the measurements, was the microphone close to the seats? Then the scenery, how has it been modeled? As faces, as well?

Often it is better to add some of the details by applying appropriate scattering coefficients to the surfaces. In other cases, you will want to add the detail near your measurement points for the seats and backs of the chairs only, then use scattering for other surfaces.

In general, you should keep increasing resolution in AURA until you do not see significant changes in the calculations. At that point you can be confident that there is sufficient data to produce reliable results.

The same happens if you have some of the source locations located in the orchestra pit. Now the surfaces can be close enough to the source that you will maybe need to add a bit of detail for them.

I hope this helps.

_________________
Best Regards,
Bruce C. Olson
View user's profile Send private message Send e-mail Visit poster's website
PostPosted: Tue Jun 12, 2007 8:00 am Reply with quote
Michal Kaminski
Member
 
Joined: 06 Jun 2007
Posts: 2
Location: Wroclaw, POLAND




Thanks for your reply.
... to be more precise:

Without scene means that there is a curtain made of steel down (so the scene is isolated). Closed orchestra pit means that movable orchestra pit is set up to the level of the stage (so the situation is like there is no orchestra pit and proscenium is larger), open orchestra pit means that it was moved about 2,5 meters down)...

The resolution of simulations is quite big: 200000 particles, 3000 ms. Scattering coefficients does not change the situation as much as it is a result of measurements. I also modeled listeners seats (always as faces) in different ways - as planes and as 3D (more seats looking like) objects.
There might be a problem with the distance between listeners seats and receivers (there are situated near the seats - about 2 meters above it - problem doesn't exist in case of orchestra pit, there is no sources/receivers situated inside of it), but I can't move it to much (because the receivers were at these places during measurements, and some of them would be than to close to the ceiling (quite complicated room).

It must be a problem with listeners seats, because without them (when most of the surfaces are strongly reflecting) there is no problem.

My task is also to find the differences between simulations and measurements and explain why it is happening, thats why I asked about the algorithms. I can see nothing special in impulse responces and energy time curves (to calculate C80, C50 manually).

Other thing I can't explain is why there is such a big difference between T30 by AURA Mappings and this counted by EASE with Eyring/Sabine formula (even about 1s differences) (View/Room Info/Data).

I have no more ideas :/

Anyway thanks for trying to help me.
View user's profile Send private message Send e-mail
PostPosted: Tue Jun 12, 2007 2:29 pm Reply with quote
Bruce
Member
 
Joined: 19 Apr 2005
Posts: 459
Location: Minneapolis, MN, USA




Michal Kaminski wrote:
It must be a problem with listeners seats, because without them (when most of the surfaces are strongly reflecting) there is no problem.

Generally I model these as 1m high boxes.
Michal Kaminski wrote:
Other thing I can't explain is why there is such a big difference between T30 by AURA Mappings and this counted by EASE with Eyring/Sabine formula (even about 1s differences) (View/Room Info/Data).


This will usually happen. Eyring/Sabine are statistical measures that depend on roughly equal absorption at all surfaces and a strongly mixing environment. These conditions are seldom both met in real rooms. Therefore they are only a very rough approximation of the RT. Impulse Responses are far better indicators.

For T30, you also need to be careful that the Schroeder integral is not including noise and/or late discrete reflections. If it is quite different from T10 and T20, look at the IR and Schroeder to determine why.

_________________
Best Regards,
Bruce C. Olson
View user's profile Send private message Send e-mail Visit poster's website
PostPosted: Tue Jun 12, 2007 2:36 pm Reply with quote
Emad El-Saghir
Member
 
Joined: 27 Jun 2005
Posts: 8
Location: Cairo, Egypt




Hello,
I have the following comments:
1. A resolution of 200'000 particles is not really what we consider "quite big", especially in such a case.
2. Scattering coefficients shall not change the measurement results of course, but they still influence the simulation results, and specially, again, in such a case.
3. Sabine's and Eyring's formulae are based on a certain set of assumptions, which are not met in all cases, ray tracing is always more accurate. One-second difference is not a big difference, if you know how far your model is from fulfilling the statistical formulae assumptions.
Regards,

_________________
Emad El-Saghir
Acoustic Design Ahnert Limited
View user's profile Send private message Send e-mail Visit poster's website
AFMG Network Forum Index -> EASE 4
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
All times are GMT  
Page 1 of 1  

  
  
 Post new topic  Reply to topic  


Powered by phpBB © 2001-2003 phpBB Group
Theme created by Vjacheslav Trushkin
Variation by CodeWeavers and AFMG