[Archived phorum post from squ1.com v2 by Olivier]

Nancy,

I am not too familiar with the acoustical analysis features, but it looks like the volume calculated in the statistical reverberation is the actual volume of the zone the analysis is being done, whereas the effective volume as computed under the acoustic response section would be the optimum volume (prefered) of the room for best acoustical performance under the same conditions.

I hope Andrew or Caroline can verify and confirm this to you, as well as inform you on the other issues you have.

Sorry I could not be of more help.

Olivier

[Archived phorum post from squ1.com v2 by Andrew]

Hi Nancy and Olivier,

The statistical and estimated RT values represent the two extremes of the calculation spectrum. The statistical RT assumes that the sound field is completely homogeneous and diffuse, with sound energy evenly distributed everywhere within the space and therefore weighting the absorbtion of each surface by its surface area.

The estimated RT is based in geometric ray-tracing in which the sound field is assumed purely specular (with a small level of randomness to simulate diffusion) and the absorption of each surface is weighted by the number of sound rays impacting on its surface.

Thus, the volume displayed in the statistcal analysis is the actual calculated geomteric volume of the space. However, sound rays may not be able to access all areas of that space - the effective volume displayed in the estimated analysis reflects this. The estimated volume is based on the mean free path length of each ray segment which, if you sample enough of them, has a direct linear relationship to the spatial volume.

Thus, the fact that the two volumes are substantially different indicates that the sound field in the space isn't fully homogeneous as there are some areas that the sound rays cannot access. Thus, as the real situation in any space is always some combination of specular and diffuse effects, the true acoustic response will lie somewhere between the two different values. Unfortunately, there is no real way of unequivocably calculating exactly where this is, so it is up to the user to look at the surfaces within the space and estimate the likely degree of specularity. Thus, spaces with large areas of exposed, flat hard surfaces will be dominated much more by specular effects than spaces with many many small rough or highly absorbent surfaces. Therefore, in such spaces, the real values will lie closer to the estimated values than the statistical.

Note also that the additional absoprtion from seating and occupancy is NOT included in the estimated RT as it considers solely the decay rate of specular sound rays as they bounce around the space.

Hope that helps...

Regards,
Andrew

Dr. Andrew Marsh
Square One research Pty Ltd

[Archived phorum post from squ1.com v2 by Valerio]

Hi,

I'm designing an auditorium hall for my degree and I have some questions about the Acoustic response section:

1) how can I calculate the absorbtion effect of an audience using the estimated RT in the acoustic response section? Do I need to model all the seats?

2) I noticed that spraying rays around creates an elapsed time of 1.720,8 ms in the animation section, while calculating the estimated RT in the acoustic response section with the same cut off and n? of bounces gives a RT varying from 1.50 to 3.50 approx. What is the relation between the two values (elapsed time and RT)?

3) calculating the estimated RT in the Acoustic response section we need to press esc to view the result. Does the duration of the calculation influences the RT? How can we know when to stop the calculation?

Thank you,
Valerio

[Archived phorum post from squ1.com v2 by Andrew]

Hi Valerio,

Sorry for the massive delay replying...

1) To model an audience plane, simply use the seat or occupant absoprtion. These are usually measured using a bank of seats fixed to a floor anyway so should give good agreement with the actual in-situ response. Included below is an image showing an example of the kinds of values you should use for fixed seating.

2) The estimated RT is very different to the analysis of existing rays. When you spray rays around, you set the actual number of bounces. Once the geometric paths of the rays have been generated, ECOTECT works out the time delay of the longest ray and uses that as its reference for the time period through which the rays can be animated.

The estimated RT calculation sprays a single ray around the enclosure until the selected decibel reduction is reached. In fact, when you select -30dB or -60dB, ECOTECT uses the first 50 rays to estimate the number of bounces required to ensure that most rays actually reach this level (that is why you may see the max bounces number change a bit when you start/restart a calc). It does not store this ray, it just keeps spraying it around and storing its decay rate at each frequency.

3) Basically, the longer you leave the estimated RT calculation to run, the more sample rays it will spray and the more accurate the resulting RT will be - hence you have to cancel it manually when you think its done enough.

If your decay lines are significantly curved or become horizontal before intersecting the bottom axis, chances are you do not have sufficient bounces set. Try doubling the current value (there is no limit). What you are after are stable decay rates that all intersect the bottom axis within the graph extents, then you can derive the RT curve.

Hope that helps...

Regards,
Andrew

Dr. Andrew Marsh
Square One research Pty Ltd

[Archived phorum post from squ1.com v2 by Valerio]

Thank you Andrew!

Now everything is clear except for point 1).

Do you mean seat or occupant absorption can be measured using only one single plane having the total seats surface, and assigning it the absorption coefficients?

Thank you,
Valerio

[Archived phorum post from squ1.com v2 by Andrew]

Hi Valerio,

Sorry for the delay in responding...

Basically yes. If you have the absorption coefficients for fixed seating, then you simply have to model the surface area of the seating as a single flat plane.

Obviously if the seats are arranged in an arena shape or sit on a series of angled floor facets you may have to model them as several flat planes, but you certainly don't need to model actual seats or even the seat surfaces.

Regards,
Andrew

Dr. Andrew Marsh
Square One research Pty Ltd

[Archived phorum post from squ1.com v2 by Valerio]

Thank you Andrew.