Room Modeling Workflow
by Pat Brown
An alternative title for this blog could be “How a Balloon Pop Can Save the Day” for a sound system designer. When building room models for acoustics simulation, the order of workflow is important. The intuitive order for room modeling workflow is:
- 1. Build wireframe
- 2. Assign absorption (ABS)
- 3. Calculate reverb time (RT)
Logical, but it can lead to some huge errors. If a surface material has a low alpha, such as concrete, and it covers a large area, a small error in the coefficient will lead to a huge error in the RT.
On a recent project, a parking garage, the calculated reverb time of the space was over 9 seconds, using the “default” concrete coefficients and the Sabine equation. Measurements in the space revealed the actual RT to be just over 2 seconds. This illustrates some of the problems with statistical reverberation equations, which is that they are fickle and conditional. The parking garage, with its low ceiling, does not have a “mixing” geometry. The ABS coefficients for concrete are not correct for this scenario, and since there is so much of it a small discrepancy produces a very large error.
A Better Workflow Order
- 1. Build wireframe
- 2. Measure the reverb time
- 3. Assign absorption coefficients
Working toward a target RT based on measured data increases the accuracy dramatically, because you are starting the process with the right answer, and working toward the conditions that produce it.
I know what you are thinking. “What if the room doesn’t exist?” Not a problem. There are rooms like it out there, and they can be measured. Evaluate a similar space, ideally several, and get a feel for what the RT is likely to be, and then engineer to that. It’s much easier to spot a bogus RT than a bogus absorption coefficient.
RT Measurements, Let Me Count the Ways
On the parking garage project, the class attendees measured the RT using three different methods.
- 1. Noise bursts with a stopwatch
- 2. Balloon pops with an RT timer (NTI XL2)
- 3. Measured IRs using a log sweep
All three methods produced very similar results for this “2 second” space with an ambient noise level of LAS = 65 dB! Any of these methods would have flagged the bogus 9 second RT of the initial calculation. The more spaces you measure, the more intuition you will develop on the likely reverb time of the unbuilt space.
For practical reasons, sound system designers are stuck with statistical reverberation time equations and Geometric Acoustics – GA. It is vital to the design process to understand that both of these methods have conditions and limits. Acoustic predictions is far from an exact science. It’s not about being “accurate.” It’s about being “less wrong” and “right enough.”
We must always apply sanity checks to keep things matched up with the real world. pb