To glue or not to glue

[avare]

This started in another forum, and I have had no success posting back.

In another topic, Steve and I ended up discussing whether or not the layers in a leaf should be glued together. He wrote an excellent post detailing the theory of why leaves should not be glued together. I replied that my understanding was that gluing is better, but neither of us could refer to any studies or papers on this subject. I wrote that I would do some calculations to try and clarify the situation.

When I woke up this morning, I was prepared to spend several hours with Aaron (my computer) collecting data setting up equations using formulas for single panel isolation, double panel isolation, panel critical frequencies, other ones as I found them required, and trying to put them all together in my spreadsheet. One of the pieces of information I needed was Young's modulus of elasticity for drywall. I hopped onto the net and did a google search for it. The third search result was for a program called INSUL from Marshall Day Acoustics. Despite its limitations, after looking at briefly I decided to use it for this analysis. The program is very versatile within its limitations. I recommend it hugely for anyone working on sound isolation. The demo is available at:

insul

But that is not the purpose of this post.


The question is "Is it better to glue or not glue the layers in a leaf together?"

All the models have the common construction of: 1" leaf, X, fiberglass in the cavity, 1 1/4" leaf. The INS UL program does not have an option for glued or unglued leaves, however it does have capability of defining the thickness of leaves. To represent glued leaves, I used a single leaf of the same thickness as the sum of the individual leaves making up the leaf. I also did the models with resilient channel to see IT there were any interactions. The results are shown in the table below.

Model RC? STC Dip(db) Freq(Hz)Iso@125 Hz
Double 48 46 2.K 33
Single 41 37 1.25k 32
Double rc 56 55 2.5k 35
Single rc 50 46 1.25k 35

The first column is the type of construction where "double" means that two layers of half the total thickness drywall was used (eg the 1" leaf being 2 1/2" thick drywall layers), and "single" meaning that the model used a single drywall panel of the leaf thickness. The second indicates when rc was used. The third column is the calculated STC. The fourth column is the isolation at the maximum depth of the coincidence frequency resonance. The fifth column is the frequency of the 1/3 octave band where the dip occurred. The sixth column is the calculated isolation in the 125 Hz 1/3 octave band.

The single leaf model has an STC rating 6-7 points lower, the coincidence frequency lowered by an octave, and the low frequency isolation relatively unnaffected, It did not have any significant interaction with the rc.

I have not figured out how to paste the graphics from INSUL yet, but one point deserves comment The reduction in isolation was centred around the critical frequency dip. Because of the deficiency method in determining which STC value to use, the reduction in isolation is not as significant as would be implied by the reduction in the STC rating. Regardless of how significant it is, it is significant.

Gluing the layers in a leaf decreases the sound isolation of a partition made with drywall.


As I was working with the program I realized I could do another (well, sort of) A/B analysis on another common belief. That being it is good to not use the same thickness of layers in a partition. INSUL does not have the ability specify different thicknesses of layers in a leaf, but the amount of specified thickness and number of can be.


Again, the model construction is: first leaf, 2x4, fiberglass in cavity, second leaf. In trying to figure some model with a semblance to real life, I chose two leaves 1.5" thick.
In the first result "same", both leaves were three layers of half inch drywall. In the second result "Diff", one leaf was composed of two layers of 3/4" drywall.

Model STC Dip(db) Freq(Hz)Iso@125 Hz
same rc 60 49 3.15k 40
Diff rc 58 56 2.5k 39

What surprises me is that the one with different layers has an STC rating two points lower. Not surprisingly the single thickness model has a significantly larger dip.

More playing, uh, modeling called for.

I am certain that there are several weaknesses in this exercise, but it is more than I hace ever seen publsished on the subject.

[Paul Woodlock]

Greetings Avare

a good thread you'e started. very interesting stuff.

Eric Desart did explain to me last year as to why gluing the drywall together wasn't recommended, and better results could be had by not doing so. it seems you have confirmed that.

Eric also sugested that making up the layers with differing thickness AND/OR differing densities of drywall was also beneficial.

I cannot remember the exact reasoning behind eric's advice, but i'll to find them dep in my hard drive somewhere.


however I think it's got something to do with the resonances within the drywall itself. Think about it. Multiple layers of the same thickenss and same density drywall will resonate at the same freqs. ( a bit like the room modes reinforcing when two or more room dimensions are equal )

Whereas, with differing thickenesses and or different densities the resonant frequency of one layer will hopefully be dampened by another layer that doesn'twant to be excited at the same frequency. And in fact vice versa between layers.

This explanation I alos believe was part fo the NOT gluing theory. If you gluethem together they act as one big resonator. Not gluing keeps things, erm, seperate.

This is a layman's viewpoint ( i.e. mine ) :) , hopefully if Eric is still about on this group he can explain this for us.

But certainly from as laymans POV, it would seem to make sense.

PAul

[ February 25, 2004, 06:11 PM: Message edited by: Paul Woodlock ]

[knightfly]

Avare, I've posted the link to the insul calculator here before, and have played with it a bit - if you look at the tabs, you'll find that you CAN specify different thicknesses of layers per leaf - you are allowed to specify two different layers per leaf, each with its own specs. For heavier materials, you can even change the material properties - changing the density from 690 kG/m^3 to 2133 allows you to treat each 1/2" of "gypsum" as 1/2" of concrete. I've used this and come within 1 dB of claimed STC on 8" concrete walls, so it's close enough for guesswork.

Be careful to change the panel size (the double arrow at the top) it defaults to 2.7 x 4 meters.

Also, there is a trick to getting the two layers per leaf to work - you can't set the first leaf to 0 layers, only the second. But you CAN set one layer to 5/8" and the other to 1/2, and use two 5/8" layers and one 1/2" layer.

And, if you do that you'll see the coincidence dip flatten out in the graph. I'm not sure how much difference you would ACTUALLY get by placing the thinner layer in the middle (more changes in the sound path that way) because the free version doesn't let you do that.

Anyway, it's a cool program for a freebie, I'm considering buying the real one eventually...

Oh, BTW Eric was a substantial part of my info base on the glue/don't glue question many months ago - Steve

[avare]

Thanks to both or you . I have learned from b
oth of you,