The No.1 Website for Pro Audio
Porous Absorber Calculator.. Accuracy issues?
Old 10th April 2014
  #1
Lives for gear
 
nms's Avatar
Porous Absorber Calculator.. Accuracy issues?

I have to say, I haven't been that impressed with the real life results vs modeled predictions when it comes to which material combinations should outperform others. How do you guys feel about its accuracy?

http://www.acousticmodelling.com/multi.php

I've been experimenting with different layer combinations of Roxul and fluffy for rear wall traps, and it's done a poor job of predicting what configurations will perform best.

3 configs I've modeled and tried are as follows:

Roxul 3" followed by 16" fluffy fiberglass (JM R-28)
Roxul 9" followed by 8" fluffy
Roxul 2" - 8" air - 8" fluffy

I attached the modeled comparison which clearly shows the one with 16" of fluffy to be the better performer, but particularly at 55hz which is the location of my lowest resonance. However, in reality the config with 9" of Roxul consistently does a (slightly) better job of reducing the amplitude of that resonance. I tested it at 4 different height locations on that wall and in all cases the one with more Roxul came out front. The one with more fluffy performed just slightly better around 40hz, but I have no problems there.

The model also predicts that the best performer at 55hz will be 3" roxul - 8" air gap - 8" fluffy. In reality this is the worst measured combination of the 3.

Not impressed!

p.s. This was using the default Allard and Champoux model. 0 incidence angle.
Attached Thumbnails
Porous Absorber Calculator.. Accuracy issues?-roxul-vs-fluffy-accuracy.gif  
Old 10th April 2014
  #2
Lives for gear
 
Jens Eklund's Avatar
 

The calculator in question does not take into account the mass of the porous material (that has some effect on low frequency performance due to the membrane action it might cause), only flow resistivity, and although the most important factor; the mass can sometimes affect the result quite a lot, even if the FR value is the same. More advanced tools include this, such as AFMG SoundFlow.
Old 10th April 2014
  #3
Lives for gear
 

Thanks for the interesting post nms, and thanks for taking the time and trouble to do this comparison. Do you have the "actual" room results to go with the prediction, REW mdat? That would also be very useful to see?

+1 Jens.

Maybe Demetris could comment if he is out there......
Old 10th April 2014
  #4
Lives for gear
 
John White's Avatar
 

Thanks nms for posting your findings. Would you do the same for the Komatsu model? It may answer some of your questions (and mine).


Thanks Jens.
Old 11th April 2014
  #5
Lives for gear
 
nms's Avatar
All I have saved is the measurements for 16" of fluffy with and without 3" Roxul panel in front (the Roxul layer reduces the 55hz mode by 1dB and also the decay time).

I tried a few different configs and found the best results by alternating the materials from floor to ceiling in this way (each line is 2' high):

[Wall] 8" fluffy - 9" Roxul
[Wall] 16" fluffy - 3" Roxul
[Wall] 8" fluffy - 9" Roxul
[Wall] 16" fluffy - 3" Roxul

The config with 9" of Roxul and 8" of fluffy was best at reducing the amplitude of the 55hz mode, while the one with more fluffy was better at reducing the decay time in the 40hz region. A combination of both seemed to be the way to go. I don't think I'll be using the absorber models for anything after this. The real answers come from testing it out yourself.
Old 11th April 2014
  #6
Gear Head
 

Hi nms

Thanks for your posts. Your results are interesting though not altogether surprising.

Firstly I would like to point out that it is not the accuracy of the calculator which is in question (this can be easily verified by comparing the calculator's results with laboratory test results published in various papers). The issue is whether it provides useful results in real-world situations.

I would agree with Jens that the membrane action of the more dense material is likely to be a factor, but there are several other issues which should be taken into account when comparing test results with predictions from this type of calculator.

Can you please tell us more about the test configuration. In particular, the size and shape of the absorber(s) you tested, and also whether the edges were covered or left open.

Thanks
Demetris
Old 11th April 2014
  #7
Gear Guru
Real World

How fibre behaves near boundaries as opposed to the simple quarter wave notion.
Last two posts here.
http://www.johnlsayers.com/phpBB2/vi...7&hilit=galaxy
DD
Old 12th April 2014
  #8
Lives for gear
 
Jens Eklund's Avatar
 

And a discussion about the claim:
Need Knowledgable Acousticians for loaded questions
Old 15th April 2014
  #9
Lives for gear
 
John White's Avatar
 

nms,

Is your room solid? Quiet? Sealed? Does it have lossy walls, weak door seals?

These may be contributing factors to the disparity between measurements and predictions.

And as Jens wrote regarding placement and capacity due to boundary interference and specific modal patterns. I think there may be too many variables for which to account with such limited testing and questionable conditions.

Eager to hear more.
Old 15th April 2014
  #10
Lives for gear
 
nms's Avatar
Quote:
Originally Posted by John White View Post
I think there may be too many variables
I summed up my feelings in another thread as such..

I have no doubt there are lab conditions which would have yielded otherwise, but in the real world.. with speakers in a studio and treatments placed near walls.. it seems to be a misleading waste of time.

I think there are too many variables and differences involved when you compare modelled predictions with the real world implementation of room treatments.

When doing thick treatments often the best results will come from a combination of materials, and the inability to account for the mechanical behavior of these combinations puts it at a further disadvantage.
Old 15th April 2014
  #11
Lives for gear
 
John White's Avatar
 

Thanks, and yes it's difficult to account for all the variables and differences. That's so true.

Still, I'd be interested to hear if your studio is sealed and solid or more closely resembling that of a more domestic type fabrication.

Thanks again for a good thread.
Old 15th April 2014
  #12
Lives for gear
 
nms's Avatar
It's a smallish room with flush mounted monitors and 8° angled side walls in the front half as well as a 16° sloped ceiling in the front half. 5/8 drywall, sealed. The rear of the room has an open doorway on one side which leads to another room. The rear wall has the highest measured concentration of LF, specifically 55hz. Certainly a crucial location to treat in rooms of this design.
Old 18th April 2014
  #13
Gear Guru
 
Glenn Kuras's Avatar
nms,

Well done! I myself have never liked using a simulator. The only way to really know is to test in a lab or just test it in your room to see which is best.
Old 12th May 2014
  #14
Lives for gear
 
nms's Avatar
Moving this here to reply:
Quote:
Originally Posted by prairiedog View Post
nms, I agree that testing can provide better results than modelling but do not agree that the calculators are not useful. The most important issue is understanding what the calculators actually do. That will give you more realistic expectations, and more importantly help you to predict when the results will be useful and when they are likely to be unreliable. I will provide more information in your own thread when I have time.

Regarding the differences between Soundflow and my own online calculator, Soundflow is a commercial product and has some excellent features which will certainly justify the expense for many users. However, for basic calculations relating to multi-layer porous absorbers, both calculators should give almost identical results (provided the extra options in Soundflow are not used). Most importantly, these results should match those in the papers written by the people who developed the models. Users with access to the relevant academic papers can check this themselves.

Hope that's useful,
Demetris
Demetris, first off I want to say that I thought you did a fantastic job with the user interface of your absorber calculator. 10/10. For someone who had never used them it was easier to figure out than any other I've seen. I really appreciate your intentions & contribution there.

My problem is not with your work, but the models they rely on and their frequent inability to transfer over into the real world treatment of our listening environments. Their inability to account for the membrane effect seems to be a large contributor to that, which is something I believe soundflow can account for if I'm not mistaken. How well it works I do not know. Certainly, any multilayer calculations that can't model this behavior will be misleading as it is a significant part.

It's a bad state of affairs when people who haven't tested and confirmed the answers these calculators spit out start passing the info around as gospel and advising others based on it. This causes people who don't know any better to spend time & money building treatments that don't perform as well.

I'm currently nearing completion of my new production studio, and I figured this would be a good time to learn the calculator. After running countless simulations I thought "Damn, I definitely need to go pick up a lot more fluffy fiberglass! What will I do with all this extra Roxul?". However, after a night of testing different configurations that turned into "Good thing I have all of this Roxul!" If I didn't know any better or test and measure before proceeding I would have done some things to my room that I'm very glad I didn't.

One trend I've seen is the great embellishment of predicted results involving air gaps. If you visit Bob Gold's absorption coefficients and compare to the OC703 figures for instance there's significant discrepancy, but it's all the more evident in cases like the one in the first post. There you can see it predicts better results with air in place of fluffy fiberglass. This was not remotely accurate and in real life I've yet to see air outperform fluffy fiberglass in any porous absorber. It definitely goes against what we consider common knowledge around here.
Old 12th May 2014
  #15
Lives for gear
 
ciro's Avatar
 

I don´t know if I´m off topic , but the only experience I had with a perforated panel based on PAC, I did it with the slotted calculator (because I read its perforated formula was not accurate). But after the panel done it didn´t touch the intended frequency (50Hz), even covering some holes.

I put it in the center of my rear wall floor , where I had a modal peak (of course) and longer decay at this frequency. It worked for 100hz though(which I didn´t want to kill).

Chances are, I did something wrong - who knows.I ended with lots of fluffy stuff in the corner and it decreased the 50hz decay the way I expected,

Ciro
Old 12th May 2014
  #16
Gear Guru
3

A suspended ceiling would often fully cover one boundary.
I don't see how this can be represented by say a 10 m^2 patch in a lab.
Then if we break it up into small patches we get closer to the reality of 2'x4' panels as commonly used. Does the increased area of edge enhance absorption by edge effect while at the same time allowing sound waves to simply pass by the now relatively small surface areas? What is the net result? Does the air gap notion apply at all to the typical 2 side reflection traps or even the 4 cloud traps?

And we haven't even touched the corner issue or the damped membrane effect....
Seems to me that is a lot of validity of Ethan's empirical testing in actual rooms.
Rigid fiberglass density tests
Plus Glenn's current and frequent tests in his dedicated room.

DD
Old 14th May 2014
  #17
Lives for gear
 
Rod Gervais's Avatar
 

Quote:
Originally Posted by DanDan View Post
A suspended ceiling would often fully cover one boundary.
I don't see how this can be represented by say a 10 m^2 patch in a lab.
Then if we break it up into small patches we get closer to the reality of 2'x4' panels as commonly used. Does the increased area of edge enhance absorption by edge effect while at the same time allowing sound waves to simply pass by the now relatively small surface areas? What is the net result? Does the air gap notion apply at all to the typical 2 side reflection traps or even the 4 cloud traps?

And we haven't even touched the corner issue or the damped membrane effect....
Seems to me that is a lot of validity of Ethan's empirical testing in actual rooms.
Rigid fiberglass density tests
Plus Glenn's current and frequent tests in his dedicated room.

DD
Lab tests are of HUGE value when used for exactly what they are intended to be used for....

They allow one the ability to view different products from different manufacturer's when tested in exactly the same manner........

This is the starting point for making a determination as to where might want to head from the point of view of product - but not the end all as to the expectations of what one might expect with "in place results" outside of that lab.

That's a horse of an entirely different color - one which is more difficult (hence expensive) to model when it comes to real world conditions.......

For example - conditions in which the results from the exact same treatments sitting inside of 2 rooms that are exactly the same finished dimensions - with the same door locations - lighting - speakers/speaker locations - furniture and placement - etc., etc., etc., could be completely different

Those rooms can have completely different results because of the construction outside of those room finishes........

Now - when you add to this an open doorway on one side of a room you're modeling - it becomes an even more complex equation - this because now the model cannot be simply the room (along with the construction of the surfaces surrounding that room) - but has to include the adjoining space - as well as the spaces to which that space connects......... and in that case - even the closing/opening of doors in rooms adjacent to that space can change the results to some degree in the room of interest......

Very complex modeling indeed.......

There is (of course) value in any data gathered that was gathered correctly - this whether in labs or real world applications - however neither of those really give anyone useful data from the perspective of what they can truly expect for results in their particular space.......

Rod
Old 26th July 2014
  #18
Lives for gear
Might the mla calculator be expected more predictable if only considering a single layer of porous absorber and a single layer of air?

Also, assuming a big enough surface area panel, would the calculations only be expected to have validity if the absorber layer and air layer are both fully enclosed in a box?

Using komatsu algorithm, the calculator estimates pretty good bass absorption with only 3" safe'n'sound + 18" air. And it estimates surprisingly good bass absorption with 6" safe'n'sound + 18" air.

I have some bulky toolchests and shelves in the rear of the home office. A movable absorber panel in front of each unit would probably be more pleasing to the eye than the utilitarian furniture which would be obscured, and this would space the movable panels about 18" from the walls without "wasting space".

I shall build 3 large panels for those locations. Built one last weekend and hopefully build the second absorber this weekend.

Each panel 3" thick safe'nsound, 4 feet wide and 6 feet tall. I tested with REW using the first panel, placed in each back corner and on the back wall, in each case about 18" from the wall in front of shelves or toolchest. Could note marginal improvement in all three spots, and left the first panel parked in the corner exhibiting the most improvement.

The first two traps I made, for straddling the front corners, were open on all sides, 9" deep, 22.5" wide, 6 ft tall safe'n'sound. These traps reduced the reverb time but only had minimal effect on bass peaks and nulls.

I got the impression that the addition of the 4 ft x 6 ft 3" panel, spaced well away from the wall, had a more noticeable effect on freq response than the first big front corner traps, though am sure that the front traps are contributing benefit in reducing bass reverb time and a dB or two on some peaks and nulls.

So these big thin rear traps touch the floor and one wall, but are open on one side and the top. OTOH, at four ft x six ft, they might be big enough that bass couldn't completely flow around the obstruction.

That is why I'm wondering, if that calculator might be somewhat relevant even if the absorber layers are not "sealed in a box", assuming a trap is "big enough"?

If that makes any sense at all.
Old 26th July 2014
  #19
Lives for gear
 
nms's Avatar
The big issue is that it doesn't model the mechanical behavior or "membrane effect". The Komatsu model is the best, but as soon as you step outside of a basic assembly with one type of material and no air gap, the results are really a crap shoot and not something I would recommend anyone act on. It's fine if you want to play around for ideas as a precursor to actually configuring different options and testing them in reality, but I don't recommend any use past that. If you want to learn what works best then build a frame and start taking measurements with the different materials or configurations you're considering. You'll end up with far more valuable knowledge than you would doing artificial modelling simulations.
Old 26th July 2014
  #20
Lives for gear
 
boggy's Avatar
Great job nms!

Quote:
Originally Posted by nms View Post
............ This was not remotely accurate and in real life I've yet to see air outperform fluffy fiberglass in any porous absorber. It definitely goes against what we consider common knowledge around here.
Yes, in short: "Air doesn't absorb sound". And yes, I know the theory or "common knowledge", but I always respect more my own experience. It is always better to check things in the real application (real problem solutions), even with all additional costs, it may be even cheaper than mistakes.



Old 27th July 2014
  #21
Lives for gear
Thanks nms for the good ideas.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 17 views: 1099
Avatar for Deleted 56021e5
Deleted 56021e5 4 weeks ago
replies: 101 views: 4520
Avatar for golgoth21
golgoth21 2nd April 2019
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
🖨️ Show Printable Version
✉️ Email this Page
🔍 Search thread
🎙️ View mentioned gear
Forum Jump
Forum Jump