gypsum/drywall: glued vs screwed

[Bob]

To answer a thread in another forum I threw together this page
http://www.bobgolds.com/GlueOrScrew/home.htm

The change in the coincidence dip was right where I was expecting.
But the drop in STC was unexpected.

The page is with insul48SA.exe, although I'm using IRC761's naming conventions for parts.

[Eric.Desart]

Bob did you use a poetic license here, by just substituting 2 panels by 1 thicker?
Or do you use an official version of the program?

Kind regards
Eric

[Bob]

Bob did you use a poetic license here, by just substituting 2 panels by 1 thicker?

Yes. The first wall has two layers of 16mm gypsum on each side. The second wall has one layer of 32mm gypsum on each side.

Or do you use an official version of the program?

the evaluation version.

[Eric.Desart]

Hello Bob,

I thought so.....
People who read your message do believe that you really calculated the difference.
Well it isn't..

That's good for a private approach to see what happens for yourself, with this coincidence.

And what happens here with the STC is very logical, you just plant this dip to lower frequencies, where it adjust to the lower level were you connect it.
As such you see much more spare on the lower frequencies.
But therefore first check how STC is calculated.

2) for music this has almost no are minor effect.

And if you do such experiments. Don't put them on a page telling that this is the difference between glued and screwed.
Just tell on the page that you were experimenting around, not that this represent the difference between 2 systems.

Best regards
Eric

[Rod Gervais]

One of the things that drives me crazy about these programs is how unpredictable the model is (in the end) in comparison with real life........

In this particular case you end up even a bit better with screwed (vrs lami) than the model predicts........ but a big difference @ 4k (an additional 13.4db of reduction) in the real world vrs the model - and STC69 vrs the 67 the model indicates (with the screwed drywall)

HOWEVER - it can easily go the other way as well.

I imagine as a tool they (modeling packages) are usefull to predict a trend (perhaps) to get one closer to reality prior to testing.

But I see some pretty wide swings betwee the model predictions and what has actually been tested.......... and this (in and of itself) makes me tense if people consider these software packages to be anything other than what I "paint them to be" above.

That being a tool to get me close enough so that perhaps I do not throw away my testing money - but not for anything as serious as actually using this for a design tool for an assembly that is never tested prior to construction.

Sincerely,

Rod

Rod

[Eric.Desart]

Hello Rod,

Those software packages are OK in this sense they they combine a lot of know-how.
But using a tryout version with limited capabilities, and substituting things by others, is really OK to learn and to get the feel, NOT to tell others how things are.

What you can conclude from Bob's page is a confirmation from the known, and already often discussed fact that the coincidence lowers when making a panel thicker, and secondly that Bob still hasn't got the feel how STC is calculated and how a double leaf system behaves.
Again: NOT knowing is no disgrace, LEARNING AND BEING CURIOUS IS HONORABLE.
But the first thing is to recognize the questions, not using this for others knowing less than Bob to tell the difference between two systems.

A Hewlett Packard calculator which can do about anything, does not make an engineer, but is a tool for an engineer.
So there is no problem with the software, but one should understand what they do, in order to understand the limits.

I don't blame my calculator because I'm unable to design the Eifel Tower in Paris. :):)

And I can assure you that ALL algorithms used in such software ARE tested AND studied and published by CLEVER guys with better measurement facilities than most of us can dream about. This is not just a result from a guy behind his desk.

Kind regards
Eric

[Rod Gervais]

Eric,

Thanks, I have a better picture now..............

Rod

[Bob]

secondly that Bob still hasn't got the feel how STC is calculated and how a double leaf system behaves.

The keyword in there is 'feel'. I've read the STC algorithm, but I can't apply it instinctively yet.

I've read in text by Eric several times that the coincidence dip would move, but in the insul48SA.exe predictor did more than just move it over (down 1khz), the entire curve above 250hz went down (TL db) with it more than I was expecting. I thought there might be an error in insul48SA.exe, due to my strange 32mm material. (an error beyond the usual measurement vs prediction errors)

This is the STC algorithm, as far as I know:
http://www.auralex.com/auralex_acoustic ... s.asp?Q=13

I seem to recall Eric or Scott writing once at yahoo that 5 layers of 10mm (3/8") gypsum were better than 3 layers of 16mm (5/8"). i.e. more layers of the same mass were better than fewer layers. But in all the related posts Eric always said two things:
a) didn't make much of a difference in music (<250hz)
b) moved the coincidence dip
and, as far as I recall, never really mentioned the drop on STC. Obviously, it wasn't obvious to me.

The tradeoff comes of course when you have to install the stuff.
Is 5/8 + 3/8 + 5/8 that much better than 5/8 + 5/8 + 5/8 on both sides of a double stud wall? Is it worth the possible mixup or runing out of 3/8 at the job site?

[Eric.Desart]

Bob,

Do not misunderstand me.
I DOOOOOOOOO respect you and I'm jealous about your energy.

This must be a yought matter or something.

You simulate 2 things.
Just describe what you did, honestly. and exactly
You can state a relationship to gluing in function of this decreasing coincidence.

But the heck, why do you in fact FALSIFY those results by identifying them for what that are NOT, just to make a point..
Just explain on your page EXACTLY what you did, NOT what you assume it COULD be, because it's handy to prove a point.
And then you help others, really help them. You really show the phenomenon of the coincidence, and leave the questions open, where you don't have a clue yourself.
People knowing less than you are impressed by this software picture and do believe BLINDLY that those are exact things.

It's really because I respect you, that I really want to tell you something here. Something more important than why the STC lowers.

Warm regards
Eric

[Bob]

Hi Eric:

I think I get where you're coming from. To me the context of the http://www.bobgolds.com/GlueOrScrew/home.htm is constrained the the original post in http://www.avsforum.com/avs-vb/showthre ... ost3960391 .

I remember your advice about the
http://www.bobgolds.com/AbsorptionCoefficients.htm
page was to keep the text off the page, and put all the text in the posts. So I did that.

In my ETF5 tests pages, I didn't have a conclusions section. Just the method/aparatus/observations. The last time I ran the full bunch of them there were no conclusions to be drawn anyway due to the bad data, but I left them out primarily because I was sure I'd get any 'conculsions' wrong, missing something seemingly insignificant that turns out to be important.

I think the concern is with the title of http://www.bobgolds.com/GlueOrScrew/home.htm
so I'll clean it up a bit, and add some text to the end.

I interpret your meaning to be a continuation of and an expression of your desire to rid the world of more disinformation on the www. Obviously I'm willing to participate in that goal on my own website.

[Bob]

OK, now that that's cleaned up a bit, any thoughts on if this is a similar graph to what would could expect with real world results with glued vs screwed? Or has that already been answered?

I presume that the answer is yes these are similar to reality of glued vs screwed (little LF change, lower STC, coincidence dip move), it's just that the model I used is flaky because it's predictive software and it's software that doesn't have an option for glued vs screwed so what I'm actually modeling is a thick sheet rather than one that's glued together. So in reality with glued sheets there would be a little coincidence change (bending, change of speed, something like that) as the soundwave traverses the additional layers of paper and glue in the middle, and probably some other mass spring effects within that 1" of material that are more complex than I'm able to guess at right now.

But the exact numbers are less important to me right now, than if this is an enegineering approximatation (+- 5db) of the trends.
Or is this so far out of the ball park that it's not even worth mentioning these graphs with the word laminated anywhere near them.

[Eric.Desart]

Thanks Bob,

Haven't checked it now, do it tomorrow.
But again my respect for how you respond.

Warm regards
Eric

[Paul Woodlock]

....I don't blame my calculator because I'm unable to design the Eifel Tower in Paris. :):)
......Kind regards
Eric

I bet you could design some cool springs to float it on though :) :)

paul

[Eric.Desart]

LOL,

Don't think about it !!!!
Most people expect to see the Eifel tower in a vertical position.

If I put it on springs, it's well possible that ALL postcards must be substituted by new ones (landscape orientation).

Eric

[Bob]

Rod

In this particular case you end up even a bit better with screwed (vrs lami) than the model predicts........ but a big difference @ 4k (an additional 13.4db of reduction) in the real world vrs the model - and STC69 vrs the 67 the model indicates (with the screwed drywall)

FYI, I've charted the results from IRC761 and the insul prediction on the bottom of http://www.bobgolds.com/GlueOrScrew/home.htm

[Rod Gervais]

Bob,

Thanks for pointing that out for me - I must have missed it.

It helps when people can see the comparisons between the analysis and the real world tests.

Rod

[Bob]

Rod

I must have missed it.

You didn't miss it. It wasn't there before today. Your post, the quote, prompted me to do it.

[Bob]

FYI, the below thread's expert Dennis Erskine, recommends a particular wall system for Home Theater which includes a glued layers of gypsum. The thread also talks a little bit about various types of damping with glued layers of drywall, and their effects on the movement of the coincidence dip. (i.e. my graphs above are too exagurated for what happens)

http://www.avsforum.com/avs-vb/showthre ... ost3960391

I posted this:

My opinion is, at least for my own construction in my own home:
a) glue is optional/recommended on the joists/studs to reduce nail popping and rattles.
b) no glue between laminated layers. I note your statement about assisting installation with glue, but I'm fairly sure it'll reduce STC.
c) with two layers of drywall, a minimum number of screws on the first layer, and a full course of screws on the last layer. (as recommended in the USG manual)

and the reply came back from Dennis Erskine

First layer screwed and glued to framing.
Second layer laminated with glue and screwed through to the underlying framing.
...
5/8" drywall, R19 in the walls, ... Laminated and glued gypsum over RSIC clips.
No lab data that would be publically available. My own room, plus models I've run.

__________________
Evil Bob

[Eric.Desart]

Bob,

Just simple:

Gluing a wall will decrease coincidence.
As a general rule you should avoid that (as you noticed yourself)

The very simple stylized basis is:
Keep mass-spring as low as possible and coincidence as high as possible (for double leave systems).
And this concept mass-spring covers as well mass, as cavity width as also symetry in this mass.

Gluing will or can add damping to panel. (That's why I immediately recognized that you used a poetic license).
However this is extremely complicated to calculate since that's entirely depending of how this is glued.

Basically you can compare that with laminated glass.
Laminated glass will be damped in the coincidence. In fact laminating glass is nothing else than gluing 2 or more window panes together (thermal process) via an extreme thin PvB foil (in fact originally designed as safety measure, this foil is incredibly strong).

Now you can say if it's good for laminated glass, then why not for drywall?
1) Mounted monolitic glass in itself has MUCH lower internal damping than mounted drywall.
2) You can not combine glass panes together without such process as you can do with drywall layers were you don't need this glue. So for thick glass panes it becomes much more important to damp this coincidence.
3) Also thick laminated glass panes will clearly show this lowered coincidence meaning has the same but slightly damped negative effect on insulation.
That's why Saint-Gobain developped this special foil and gluing procedure to combine this laminating advantages without the disadvantage of this lowering coincidence.
4) This whole coincidence effect has only sense in as far that the related coincidence dip, which becomes more a plateau when higher damped, is defining in the overall insulation versus the relative frequency distribution of the source you want to insulate.

Standard mounting methods of drywall has proven over and over again, that there is relative little you can do to improve those methods.
All other discussions go in the margins.

I once told about stiffness increasing low frequent behavior. Such things seem to get their own live (out of context), because I ALSO said that was valid for lightweight SINGLE leaf systems (often applied with steel). But there stiffness must be interpreted in function of real acoustic stiffness (strongly lowering the coincidence).
With double leaf systems this mass-spring remains the defining factor.
A brick wall is extremely stiff compared to drywall. Still a resilient drywall skin in front of that will still show the mas-spring dip.
A double brick wall (you hardly can go stiffer than that) will still be defined by this mass-spring behavior. In the latter case the mass-spring will even increase versus the calculation based on flexible walls. So while gaining a lot by the mass of a double brick wall, there is a drawback in the mass-spring calculation.

Still: If you should make a perfect decoupled drywall (boards, floating in the air) you should notice that TL in the low frequencies should drasticallty decrease.
It is the stiffness of the studs, which give standard grade office drywalls their high STC. This stiffness damps the mass-spring resonance.
A theoretical perfect decoupled drywall should drop at the mass-spring resonance to an about 0 dB insulation.
Since Ron always is rightfully reluctant to accept things without seeing measurements (in fact you're right about that Ron), I once made such a wall with a maximized decoupling based on nothing more than 2 times single 1/2 inch gypsumboards, glued as sandwich panel on a aglomerated (mainly ether) foam core with low density (which acts as a poorly damped spring).
It was meant as a separation in a rather huge building between the factory part and newly to be build commercial rooms (showroom, offices, etc.). The wall really splitted the whole buiding in 2 parts.
It was more an experiment of me to maximize high frequent TL .
The measurements are still somewhere in my archives.
The result was that I got EXTREME high high-frequent TL values (which was meant as such, since the factory did stainless steel activities as turnery, which caused very high frequent disturbance = sounds as a knife in your ears)
But at the low frequencies this wall acted as a filter. It really followed very closely a theoretical poorly damped double leaf system.
While I don't remember the exact measurements, I don't forget the resulting joke of my (back then) foreign colleages telling: don't ask Eric, he makes things that perfect that you can't use them anymore. Here they referred to the extreme poor Rw/STC this wall had, defined by this low frequent filter property (it becomes as a panel resonator with very sharp Q).
In fact studs on drywall acts as dampers. But this is frequency dependent, and does not mean that changing OC is that obvious. So never read more than what I'm saying, in the context I'm saying it.

I have hundreds of measurements more than the ones in this IRC761 report.
It's very difficult to state simple logic, for all those "in the margin" things.

I still plan to bring this all together in one huge database, made as such that one can make easier statistical conclusions in function of frequency weighted insulation.

I can not imagine that gluing a wall is a necessity to prevent a good construction from undesirable vibrations.
What you do know for sure is that this coincidence lowers, which is normally not a good thing.
And as said, if you want to damp the coincidence somewhat, there are easier ways to accomplish that.

Eric

[Eric.Desart]

Added a picture (total 2 now)

Eric