Actual reverberation time and error from Pachyderm simulation

Hi all,
I'm currently working on calculating reverberation time using Pachyderm Acoustics.
I have a reverberation chamber close to me, and I have an environment where I can check the actual reverberation time.

The reverberation chamber of the attached GH file was given a condition of low sound absorption coefficient, and the reverberation time was simulated. (For example, the sound absorption coefficient of the plaster in the material library was set to 1% and analyzed.) )
Then, the reverberation time of Pachyderm's simulation showed a reverberation time of about 2 seconds in each frequency band, but this reverberation time was very different from the actual measured time. (In actual measurements, each microphone was approximately 125 Hz for 17.5 seconds, 250 Hz for about 14 seconds, 500 Hz for about 9.5 seconds, 1000 Hz for about 7.5 seconds, 2000 Hz for about 6.s, and 4000 Hz for about 3.5 seconds.) )
What are the possible causes?
I will attach the gh file so that you can check it and get an answer.

If the results of the respondent's analysis of the reverberation time in Pachyderm are similar to my measurements,
I would like to know the settings at that time.
Also, please let me know if there is any information required for verification

Thank you

Attachments:

question.gh, 21 KB

Replies to This Discussion

Permalink Reply by FUYU on June 21, 2024 at 3:17am

This is the result of my analysis of the reverberation time. Try to match it.

Attachments:

Results of my reverberation time analysis.csv, 1 KB

Permalink Reply by Arthur van der Harten on June 22, 2024 at 12:19am

FUYU,

The grasshopper file doesn't have any geometry in it, and so is only a simulation method. This doesn't show me what your model looks like, whether it is using the correct units, or how you set up your absorption and scattering.

Please send me your Rhino model for evaluation.

Another suggestion - maybe walk before you run. There was nothing in that grasshopper file that couldn't have been done in the Rhino interface, and probably with less chance of user error. Maybe try using the Pachyderm for Rhino interface. You might get some insight into what you might not have set correctly from that.

Arthur

Permalink Reply by FUYU on June 23, 2024 at 5:58pm

Hello Arthur,

Thank you for your quick reply.
I'm attaching the Rhino model to this comment.

Also, thank you for your suggestions regarding the settings.
Can you tell us more about your suggestions?

I've never changed the settings of Pachyderm for Rhino except for the material.

In the example of a project that Arthur has done in the past, where he simulated a long reverberation time, could you tell me what kind of settings were used?
I would like to use that as a reference to modify the settings and ask questions here.
Also, please let me know if there is any information required for verification

room.3dm

Permalink Reply by Arthur van der Harten on June 23, 2024 at 6:22pm

Hi,

I'm running it myself right now. Two important comments:

1. You must set scattering. There are aids on the scattering page in the main rhino interface. Please use them.

2. Your cutofftime is very low for such a reverberant room - 1000 ms. This means that your simulation is not running nearly so long. 1000 ms is appropriate for many of the rooms that you will model, but a reverberation chamber is a very special case, and requires much longer calculation times, and much more rigor to simulate.

Arthur

Permalink Reply by FUYU on June 23, 2024 at 6:36pm

I'm surprised at how quick the reply was.
Thank you so much

1. Thanks for telling me about the scattering coefficient. I'll check and set it up. Do I need to set transparency?

2. Does that mean that if you want to mimic a reverberation chamber, you should have a longer cut-off time? Do you have any recommended cut-off times?

Permalink Reply by Arthur van der Harten on June 23, 2024 at 7:17pm

Hi,

My pleasure.

2: You need to have a cutofftime at least 75% of the expected reverberation time, longer if at all possible (in case there are non-diffuse effects that Sabine would not predict.).

Arthur

Permalink Reply by FUYU on June 23, 2024 at 7:27pm

Dear Arthur,

Oh my God.
Then it seems that it should be about 15,000.
I'll try it with this setting once.
(It's going to take a lot of time, but...)

If there is something you don't understand after trying, let me ask you.

Thank you

Permalink Reply by Arthur van der Harten on June 23, 2024 at 7:58pm

It will. Like I said, a reverberation chamber is a very special case that most people will not need to run. The last one I did took about 4 hours, with a 7 second reverb time.

You should be aware - geometric acoustics algorithms are only really accurate for a particular part of the spectrum, the part above the schroeder frequency for your room, and the part for which the geometry is accurate by wavelength.

You have very long RTs at low frequencies, but you might consider selecting your cutofftime for another band, such as 500 Hz.

Suffice it to say that 1000 ms is too short, but 15,000 sounds excessive. I selected 7000 ms. It is already about 10 to 20 percent complete.

I would also like to point out that there isn't really a free lunch with these methods, not if you want a real scientific tool. Pachyderm's algorithm isn't the fastest, but it is one of the most reliable when it comes to non-sabinian rooms with a tendency to generate flutter echoes and other anomalies.

Arthur

Permalink Reply by Arthur van der Harten on June 23, 2024 at 10:02pm

No. You only get one cut off time, but my point is that it probably isn't worth choosing it for a band that isn't that accurate when simulated using a geometric algorithm.

Yes, it is running on my personal machine now. I'll post the results sometime tomorrow.

Arthur

Permalink Reply by FUYU on June 23, 2024 at 10:17pm

Dear Arthur
Thank you, I understood.

I'm very happy about that. I'm really looking forward to tomorrow!

thank you

Permalink Reply by Arthur van der Harten on June 24, 2024 at 11:31pm

Hi Fuyu,

The simulation completed. My results were about 3 seconds, all octaves. It may be possible that your room really is as reverberant as you say, but for this to be the case in pachyderm, much lower absorption and scattering would be needed. Such conditions are only possible with very heavy and polished surfaces, but they aren't unusual in a reverb chamber.

Also, I get a room volume of 264 cubic meters, and a sabine calc gets a similar 3 second reverb time.

So how do you know it has those reverberation times? Did you do a measurement? If so, would you send the impulse response to me? Photos of the room would also be useful.

Arthur

Permalink Reply by FUYU on June 25, 2024 at 12:01am

Hello, Arthur

Thanks for the analysis.
Is it 3 seconds? It's quite different from my actual measurements.

By the way, I also did an analysis yesterday.
When analyzed with a sound absorption rate of 5%, scattering of 25%, and a cutoff time of 10000, the reverberation time was approximately 2.5 seconds.
(Scattering was provisionally chosen only because it has a light computational load)
We are currently analyzing it with a sound absorption coefficient of 1.

The reverberation time was measured with the arrangement stored in the data.
There is no equipment to measure the impulse response.
Due to confidentiality obligations, it is difficult to send photos.
I am sorry

----------------
Question.
Can you tell us the conditions (sound absorption coefficient, scattering, cut-off time) of Arthur's analysis?
If you don't mind, I'd like you to upload Rhino and GH data.

We would like to ignore the calculation of the low frequency band because it reduces the accuracy of the geometric analysis and causes the calculation time to increase.
In this case, if I reduce the computational load by setting the sound absorption coefficient of the material I want to ignore (let's say 250 Hz or less) to 100%, for example, will it affect the calculation result of 500 Hz or more?
Is there a mechanism in Pachyderm to store wav files and analyze them using them?

I'm sorry for all the questions and requests.
Please help me.