AFMG Network Forum

logandodson · Member

I am modeling a large gymnasium consisting of 4 basketball courts and a track with concrete walls, a metal deck ceiling, and a hard rubber floor. My expectation for the RT60 of a space this size with these reflective materials would be much higher than my result of 2.89s using Impact Ray Tracing.

My sabine time is 9.11s and I made sure to use the mean free path to determine my delay cutoff, and used enough rays per loudspeaker to attain a 90% impact chance.

Has anyone had a similar experience when modelling a space this large? Or is there something I am missing to get a more accurate RT60 result? Confused

Agustín Arias · Member

Hi Logan. What time did you use for ray tracing? Did you add a "statistical tail" to the reflectogram in the "Probe" module?

logandodson · Member

Agustín Arias · Member

That would be weird... can you share an image of your reflectogram at 500 or 1000 hz?

logandodson · Member

https://ibb.co/48CTtRs
https://ibb.co/TYvQzrd

So this is with and without using the random tail. My random tail this time did add a significant amount of time to the result. However I feel like the parameters I change when creating this random tail change the result so much and I am not sure what these values should be such as the pulse density and maximum density.

I also do no think the slope of the tail is an accurate continuation of the source. You can see clearly where the tail starts because the slope changes drastically.

Agustín Arias · Member

Hello Logan. My opinion (without having the model) is that the statistical tail must be adjusted correctly and the regression interval must be increased for the calculation of the RT Schroeder.

For this I recommend that you do the following:

1) open the project
2) open the Probe module
3) open the .imp or .rsp file
4) remove any statistical tail (only if you put one in and saved as .rsp file)
5) open the options panel (F9)
6) in the "Rev. time" tab, in the center is the "Replace by" button group. Within them, click on the one that says "Sabine" (Not on the checkbox that is inside the "Display" group at the right!!!). Finally, click OK
7. Add the statistical tail with "add random tail". In "Tail lenght" put a higher time, for example 4 (seconds)
8. Adjust the level and density of impacts until visually verifying that the statistical tail is correct.
9. Once you have added the statistical tail, open Options again and in the "Schroeder" tab verify that the integration time is greater than 4000 ms. Click on "Compute" and check results.

In any case, if the results are still very different, you can send me the model to my email below for a more precise inspection.

logandodson · Member

https://ibb.co/QJjX1xp

This is my statistical tail at 1000 Hz. The added tail's slope I would imagine is what is causing these high RTs but does this seem like an appropriate tail for the data?? At lower frequencies, the tail seems to fit the data better and is giving ~4.5s.

Bruce · Member

EASE Ray-Tracing is most useful for investigating the Early reflections. Investigation of the full IR is best done with the AURA 4 module. https://www.afmg.eu/en/aura-4-module

ervinjason · Member

I used the mean free path and found that I should be using above 500ms as my time. I think it was 560ms when I ran it. And yes after adding a statistical tail it has a very similar result (2.89 without and 2.82 with). I tried adding a random tail as well as adding missing impacts and attained similar results.

I am thinking a source of my error is that I selected a speaker near my listener seat. So the initial dB reading will be quite large due to direct impacts, and all impacts after will be attenuated by the shear size of the space before reflecting back. This will create a large spike at the beginning that could be a source of the low RT60 time since it is measured by the dB decrease over time.

Would cutting off the initial direct impacts be detrimental or helpful in determining the true RT60?