Relying on construction adhesive for structural capacity 2

[DTS419]

Especially in residential construction, it can be common in practice to use construction adhesive in addition to mechanical fasteners to make connections. Examples of this would include sill plates on foundation walls, wood sheathing to support members, sistering joists, etc. As a structural engineer, do you or would you ever account for the capacity of construction adhesive? If so, how do you do it an acceptable manner?

 

[JoshPlumSE]

As a structural engineer, do you or would you ever account for the capacity of construction adhesive?

I would require a published ICC report that backs up the capacity I'd be using.

The only caveat to this would be in helping with some "serviceability" checks that I do. If I'm want to get a little extra stiffness to pass a LL deflection check and we're using a really good construction adhesive on the joist, then I might use a bit of the sheathing as a "composite" joist for the deflection check.

Same sort of thing if I'm doing some kind of vibration check.

But, I wouldn't rely on it for strength / failure checks unless I had some hard data to justify it.

 

[Eng16080]

However, we do rely on construction adhesive for non-strength based checks, i.e. deflection, in that you develop partial composite action between the wood members and sheathing, which is reflected in deflection calculations.

This is interesting. If I'm guessing correctly that you're involved in the I-joist industry, then I assume your statement means something like: the maximum published I-joist span lengths (which are mostly deflection limited) are based on the assumption that the subfloor is correctly glued to the I-joists. I always assumed the glue was just something extra that wasn't contributing to strength or stiffness. So then, the span lengths would be non-conservative in the event that the field crew doesn't glue the subfloor or doesn't glue it per the specs of the I-joist supplier. I'm trying to gain an understanding in case I might be overlooking something in spec'ing I-joists on my plans.

 

[ChorasDen]

quote

You pretty much hit the nail on the head here on everything you wrote, but I would say some of your assumptions may be overlooking some effects. I'll try and summarize here best I can with limited space, but it's a fairly involved topic at the end of the day.

I assume your statement means something like: the maximum published I-joist span lengths (which are mostly deflection limited) are based on the assumption that the subfloor is correctly glued to the I-joists

This is correct within the U.S. Every i-joist manufacturer that I am aware of that publish span tables includes a footnote indicating that the spans are based upon composite action with 24" rated subfloor (ie 23/32" panel) that are glued and nailed. Some manufacturers have you subtract some amount of distance from total span if the connection is nailed only.

This is different in Canada, because most of the spans are vibration controlled rather than strictly deflection controlled. However, I do not work in Canada, and I am not familiar with CSA 086 enough to say whether composite action provides any benefit to vibration checks.

I always assumed the glue was just something extra that wasn't contributing to strength or stiffness.

The link I included above discusses the derivation and calculation for deflection checks including composite action, as I assume you saw. Adhesive is only considered in this type of check, which is a serviceability check, rather than a strength check.

So then, the span lengths would be non-conservative in the event that the field crew doesn't glue the subfloor or doesn't glue it per the specs of the I-joist supplier.

This is where things get a bit tricky. Not sure how familiar you are with structural wood products including sawn lumber, but wood is inherently an inconsistent structural material, there is significant variability from one piece to another in any binned product, whether that be visually graded, mechanically graded, or an engineered wood product. Because MOE is a serviceability related design value, it's design value is not controlled very tightly. Published MOE value for all wood products are based upon the median design value of a normally distributed population, without any adjustments. If you have a copy of AWC's National Design Specification, check out Appendix F.

Per Appendix F, you'll see that the coefficient of variation for MOE of visually graded sawn lumber is 0.25, this is a large COV value, and means there is significant variability on stiffness of members within the population. In addition, because MOE is based on the median value, each piece has a 50% chance of having a lower stiffness than what was designed. Appendix F gives methods of adjusting MOE for critical applications. This is similar to the adjustments used for Emin, which is a strength critical design value rather than serviceability based design value.

What does this mean? Well, if you're concerned that the span tables are non-conservative, be aware that 50% of the joists (whether that be sawn lumber, mechanically graded, or engineered wood) have lower stiffness than the published value used for design. This is true for the span tables in the IBC/IRC as well, but is found to be acceptable practice.

 

[Eng16080]

ChorasDen, I appreciate the excellent explanation.

I suppose it likely varies by manufacturer, but I imagine most I-joists use MEL or MSR lumber rather than visually graded, thus having a lower COV value and more predictable MOE. If correct, this would be a slight benefit of using I-joists over sawn lumber as you would be more likely to have a "bad batch" of lumber with a lower than average MOE. Whether this makes up for the glue issue (or perceived issue), I don't know. I've certainly never accounted for a stiffness benefit in gluing subfloor to sawn lumber.

I'll take a deeper look at the document you attached.

 

[ChorasDen]

LVL is the most common product for flange stock, with MSR probably the next most common.

 

[cam_b]

I have specified many epoxy anchors in the past. When I'm attending site, if the epoxy is still wet then i randomly pull out one of the epoxy anchors. It has made me reluctant to specify them now, because in my experience the holes are rarely cleaned and the embedment depth is often not even close to what has been specified.

 

[BridgeSmith]

...in my experience the holes are rarely cleaned and the embedment depth is often not even close to what has been specified.

Our experience has been similar - holes not getting cleaned and the epoxy not getting mixed well are the common problems that have caused us to not use them in any tension applications. We still use them for dowels in shear and traffic railing anchorages.

 

[DTS419]

I agree with most of what has been said in this thread, especially the parts about adhesive anchors being very prone to improper installation. I have as many stories about that as anyone. However, one perspective I will offer is that it's easy for engineers to hunker in a protective and safe position that construction adhesive should never be counted. But, try explaining that to a contractor who has seen construction adhesive outperform mechanical fasteners that you won't let him take credit for it even though he is using it anyway, and that instead he needs to use twice as many bolts on an project with tight margins. Sometimes as engineers we need to come out of our protective shells and be a little more practical.

 

[SE2607]

Off the original topic just a bit, but it's disappointing to read about the poor quality control issues relative to epoxy anchors.

What can be done to improve this situation?

 

[BridgeSmith]

Sometimes as engineers we need to come out of our protective shells and be a little more practical.

I think taking into account the performance (or lack thereof) of a product in the field when considering it for design, is very practical. We had several consequential failures of epoxy anchors in tension applications. No injuries, thankfully, but it's certainly potentially dangerous and a headache for everyone, when the anchors that are supposed to hold down the ends of bridge girders to their abutment, don't hold and a allow a 10 ton bridge girder to pop up in the air. Granted, it was during construction, but construction workers matter, too.

If there is adequate oversight to ensure the proper process in the application of the adhesive, then sure, it can be accounted for in the design.

It seems we've strayed quite a bit from the original question, though, which appeared to be about making wood-to-concrete/masonry and wood-to-wood connections with construction adhesive. Wood-to-wood connections seem to be far less prone to the problems associated with wood to concrete or masonry, so we may need to make a distinction between the applications. There's also established design procedures for the use of construction adhesive in the connections of subfloors to joists, or at least I-joists, so there's that. It seems reasonable to me to account for the construction adhesive in that specific circumstance, but to be cautious in counting on it for design in other circumstances, where the QC/QA may not be adequate, and where the strength or stiffness gains are not readily quantified.

 

[Eng16080]

DTS419, I don't think anybody here disagrees that glue can provide incredible strength. Take glulams, for example.

However, as an engineer, stamping plans and needing to make decisions that I may be held accountable for later, if my past experience has been that:
[ol 1]
[li]glue/adhesives are prone to improper installation, and[/li]
[li]contractors have told me that they've "seen construction adhesive outperform mechanical fasteners"[/li]
[/ol]
my general conclusion will likely be to not allow this type of connection for strength determinations. Perhaps I might account for some increase in stiffness (as discussed above), like in a floor, although this is debatable.

I would personally take what the contractor told me with a big grain of salt. Not because I think they're trying to be deceptive, but because their experience with the glued connection is likely only for a very short duration over the structure's entire life, presumably prior to the wood initially shrinking/drying and later moving seasonally, and presumably without the connection being loaded to anywhere near the design load.

In terms of having a connection which is a hybrid glued and mechanically fastened connection, I'd be concerned that the glue might provide a stiffer load path within the connection than the mechanical fasteners and that all the load would initially go to the glue, until it failed, and then to the mechanical fasteners (which are presumably under-designed to take the full load).

I would be most concerned with a wood to wood glued connection where:
[ol A]
[li]the wood species and grade differ,[/li]
[li]the moisture content differs, or[/li]
[li]most importantly, the wood grain directions are not aligned.[/li]
[/ol]

Basically, if there is any chance of the two pieces wanting to move independently of one another, the glued connection will be susceptible to failure if the glue cannot accommodate that movement. I suspect that the failure then might not even be in the glue itself but rather in the wood fibers tearing/shearing off.

While most of my thoughts here are only that, the following statement mentioned above and substantiated by testing carries more weight:

In the wood industry, we do not rely on construction adhesive for any strength based calculation...

 

[cam_b]

try explaining that to a contractor who has seen construction adhesive outperform mechanical fasteners

How do they know it outperforms the mechanical fasteners? Neither one should be failing, unless it's designed or installed incorrectly.

 

[Tomfh]

Our experience has been similar - holes not getting cleaned and the epoxy not getting mixed well are the common problems that have caused us to not use them in any tension applications

I came across a new one recently. I inspected a few anchor bolts, and found I could pull them out easily by hand. They'd been installed days before, so it should have been cured. Turns out the installer didn't have any mixer nozzles, and had resorted to squirting it straight from the cartridge.