Use of PU for a bench build?

[ian maybury]

The long planned Schwarz-ish Roubo in German beech is finally under way (just finishing the drawings), and the provisional plan is to use a liquid polyurethane structural adhesive. (generically similar to e.g. Gorilla Glue) PVA is of course a possibility, and has successfully built many benches - but I'm wary of the the possibility of creep and of its reluctance to bond to itself.

Plus I've had good luck with a specific PU from a local source, and have found it easy to scrape off what foams out of the joint.

Gap filling, no creep/structural bonding capability, longevity, colour (it's a little dark), repairability (bonds to itself) are other considerations.

Trouble is that I've never used it for large scale work like this.

I'd appreciate your thoughts on suitability and other considerations.

Failing that epoxy is an option, although it's in my experience quite a lot less convenient (mixing and clean up), and I'd have to order from the UK.


Thanks

ian

[Chris Fournier]

I don't understand your concern with PVA creep in this application Ian. PVA would be my choice because it has a proven successful track record and it's the least expensive option. I have built and helped build with others about 8 benches - all PVA, no failures spanning over 20 years. Now if you want troubles, use epoxy...

I have used PU for larger end grain butcher blocks because I felt that the application merited the additional cost. Quite frankly I could have used plastic resin glue at a fraction of the cost with equal measure of success.

[Jamie Buxton]

My experience with Gorilla glue was not terrific. The stuff foams up as it cures, and and pushes the workpieces apart unless they are clamped extremely well. Worse than that, the joint isn't as strong as PVA. When I made test butt joints (like I was making a table top), and stressed them to breaking, the break was in the glue. In contrast, PVA will break in the wood. That is, PVA is stronger than wood, while Gorilla glue is weaker than wood.

[ian maybury]

My view is partially technical, and partially gut feeling guys. That said I've done a fair amount of digging on PU in woodworking, and was exposed to polyurethanes in an adhesives R&D context back in the late 80s and 90s too.

I'm used to viewing PU as a structural adhesive (one that delivers a high strength and high adhesion semi-rigid polymer) much like an epoxy, and likewise to seeing PVA as a moderate strength water based commodity adhesive - although in the end an adhesive need only be as strong as the substrate, and modern Titebond and similar PVAs are quite heavily modified and clearly are as strong as most woods.

The major negatives about PVA that come up when you do some reading are the fact that it won't bond to itself for repair purposes, plus some questions about long term stability under load and longevity, especially in less than perfect (e.g. damp) conditions. It wins for convenience - PU is potentially messy without good technique.

There's bum stories about regarding PU too, although many of them are hard to reconcile with the known properties of the polymer. There seem to be those that love them, and those that hate them. I was relatively neutral, but having been eased into using one recently by the need to face bond phenolic ply was presently surprised at the ease of spreading, decent open time, ability to bond awkward surfaces (strong enough to cause wood failures on face to face bonded pieces of sanded and solvent washed phenolic coated ply), toughness, and easy removal of the squeeze/foam out. Water resistance is another positive.

I'd need a lot of convincing to buy the perception of some that PU can't do an excellent job of structurally bonding wood. There's some issues in play though it seems that can result in problems:

1. If gaps caused by poor fits and/or clamping allow it to foam in the bond line it's going to be weak. Good fits and plenty of clamping force are essential. i.e. whatever gap filling capability it has is best regarded as solely cosmetic.
2. Wiping of water on one of the bond surfaces seems to be essential - it promotes both penetration and more complete polymerisation.
3. It's actually better than most at wetting and hence bonding a wide variety of surfaces, but without solvent washing etc its like any other adhesive going to struggle on resinous or oily woods and polyolefin plastics. (polyethylene, polypropylene etc) There could be tight pored woods it doesn't do so well on too.

My guess is that the above plus the fact that it's relatively new in the US account for most of the negative stories - but thoughts appreciated.

My question actually presumed it could more than do the business, but that use over the large surfaces involved in laminating a bench top might present some practical issues.

It seems it's at least do-able, a quick dig just now before posting brought up this piece on the use of Gorilla Glue on a Benchcrafted Roubo bench and Moxon vise build by a blogger named Brownell who does demos etc for GG. (one of the US liquid PUs) http://brownellfurniture.com/tag/benchcrafted-roubo/ He doesn't look to be doing anything in terms of clamping or spreading that's not pretty typical....

ian

[Bill White]

Tried to use the stuff twice. No more. I'm stickin' with PVA glues unless I'm gonna stick two rocks together.
Saying that, I use Gorilla PVA because of the "grab" time.
I've never (yet) had a PVA joint fail.
Bill

[Peter Quinn]

About 6 years ago I made a set of carraige house doors for my garage as a temporary fix. The garage is old and wrecked, the over head door company would not/ could not put a door on it, so I made one. My main goal? Cheap. And quick. I used spruce 2x 4 decking from the local lumber yard, t-111 DF plywood for the panels. Glue? I did one with PL and one with a PVA type III . Loose tenon construction, glued the plywood panels into deep grooves. Only hitch was in my haste I forgot to note which door used which glue. I figured when the first one fell apart I'd see what type of glue was used, a sort of laymans science expiriment. This spring the garage comes down, no failures in the doors to date. So which glue is stronger? Both. Are stronger than they need to be. Take your pick.

I prefer PVA for ease of use , Pl has a longer open time and scrapes easier but sucks on open grain species. Think grain filler. I don't think any of the consumer based pva's or pl adhesives are considered structural for engineering purposes.

[Alan Schwabacher]

There was a test some time ago of various glues where the interesting variation seemed to me to be clamping pressure. With high clamping pressure they all gave strong joints, but with lower clamping pressure the strength of the bond decreased much more for polyurethane glues than it did for the other glues.

[Jamie Buxton]

..I'd need a lot of convincing to buy the perception of some that PU can't do an excellent job of structurally bonding wood...

Geez, if your mind's already made up, why'd you pose the question? :)

[Bob Falk]

Are you going to leave this bench out in the rain? If not, use Titebond and make life easy....Gorilla glue is a sticky, ugly, mess and is impossible to clean up. My $0.02

[ian maybury]

Sorry Jamie, that wasn't what I meant - no offence intended, and pardon it coming across that way.

From where I sat it seemed pretty clear that PU bonds wood very effectively indeed. That if there are issues they are more likely to be to do with handling of the adhesive, and the nature of the job it's used for. i.e. issues to with the specific application, rather than the fundamental limits or capability of the adhesive.

These pros and cons are still capable of determining whether it might be a good choice for a given job - and either way there's nothing that says that there are not reasons to avoid using it at all.

My question was posed in this context. I'd not seen it used for larger work like bench building. My sense was that the inconveniences (e.g. more difficult clean up when wet) could be worked around (you can't bath in it like you might with a PVA :-)), but thinking of the large joint areas involved in laminating the planks for the bench top and legs I didn't want to run into a real show stopper of issue like e.g. joint weakness as a result of insufficient clamping allowing foaming to open up the joints.

I guess I was hoping that a few people could chime in to say they had experience of using PU on larger work of a similar nature, and advise on what to look out for. The initially negative replies caused me to do some more googling which turned up the Gorilla Glue guy's blog reporting his build of a Benchcrafted Roubo using PU - which while not exactly a neutral view seems at least to confirm how he did it and that it works.

On PVA and the more general comments. It's clearly the default choice for most here (me too actually), most are familiar with and comfortable with it, and it's convenient to use.

That said some casual reading suggests there's equally some genuine questions that get asked about its longevity (a bench hopefully needs to last generations rather than for a few years), resistance to adverse conditions and stability under load. I've already said that it seems to have been improved in the interval, but in the Euro adhesive R&D environment I worked in it wasn't (unlike polyurethanes which have been used in engineering and structural applications for many years - probably for longer than the current generation of modified PVAs have been around in woodworking) regarded as a structural adhesive either.

Which isn't to say it can't do a long term 'permanent' job in a bench that hopefully will survive to be passed on to family or whatever, but does explain the questions. I'm hearing the positive comments about PVA too....

ian

[ian maybury]

PS to the above. I'd been finding generalised hearsay references on the PVA and creep issue, but more digging just now brought up this highly definitive post by Hugh Evans (a tech guy at Titebond) in response to the question of whether or not PVA is suitable for load bearing structural applications on Wood Central: http://tiny.cc/6qscew There's a whole series of linked posts around the topic.

He very specifically says that none of the Titebonds (1, 2 or 3) are suitable for load bearing applications - that they are all subject to creep. (he links a piece defining this too) Creep testing is standardised - there are various BS, DIN and other standards with which a glue has to comply to be OK for use in structural applications. (this is properly why PU, epoxies and the like are known generically as 'structural adhesives'.

Coincidentally the guys (an industrial manufacturer) I've been buying my liquid PU from came back on a specific enquiry this morning too:

'We make large volumes of high quality PVA wood adhesive but they are not recommended for load bearing applications.'



Here is the full message with names removed:

'PU adhesive joints are permanent. Correct joint design and adhesive application are important. Timber moisture should be in the region of 7% to 10%.

Ideally, timber should be planed within 24 hours of adhesive been applied. Surfaces to be bonded must be smooth, flat and parallel.

The bonds should be clamped (100 pounds per sq inch for low density timbers and 250 pounds per sq inch for high density timbers). Clamping has several functions, it forces entrapped air from the joint, it brings adhesive into molecular contact with the wood surfaces, it forces adhesive to penetrate into the wood structure for more effective mechanical interlocking, it squeezes the adhesive into a thin continuous film and it holds the assembly in position while the adhesive cures. If the pressure is too high, the adhesive can over penetrate porous woods and cause starved joints that are inferior in bond strength. The strongest joints result when the consistency of the adhesive permits the use of pressures that are consistent with the recommended pressures for the density of the wood.

Like all adhesives, wood adhesives in particular can vary in quality. It is only recently that polyurethanes are found in the DIY woodworkers cabinet and the quality of these adhesives may not be the same as its equivalent industrial product. PU's are reactive and moisture sensitive. They typically have a shelf life of three months from date of manufacture and for the occasional home woodworker, they may offer less then optimum bond strength if the product as the adhesive in the container ages. With our 'xyz' PU, squeeze the bottle to expel air and replace the cap. This will extend the shelf life of the product from open.

We make large volumes of high quality PVA wood adhesive but they are not recommended for load bearing applications.

Re your enquiry on 'xyz' PVA adhesives. These are DIY adhesives and like DIY PU (for example 'xyz' PU), do not have the same performance characteristics as correctly applied industrial adhesives.'




The point on the poor quality of some consumer grade PUs might (along with the need for moisture, good fits and clamping) explain some more of the issue seen at times with them.

PVA is successfully used for lots of load bearing jobs. The distinction it seems is short versus long term continuous loading. It can creep (flow like plastic) in the case of the latter, and for this reason is not recommended for tasks in continuously loaded applications like bent laminations, and beams where it can lead to structural issues and/or the telegraphing of the adhesive line through the finish over time.

It does OK on stuff like furniture because while it can be heavily loaded it's usually only for short periods.

There are a few PVAs about that are claimed to be OK for load bearing applications, but it's basically shading it by all accounts. As ever the performance of most adhesives is inherent in their chemistry, and the properties of the resulting polymer. There's lots of claims made about wonder additives, but in the end it's rare for them to make that much difference.

ian

[joe milana]

How about Resourcinal resin? It seems to address all your concerns. Long pot life, no creep, structural bond, and if your ever in a flood....
Oh, and those dark glue joints speak quality in my book!

[John Coloccia]

"Load bearing" in this case means for something like laminating beams for a house....continuous loads, not lightly loaded furniture and workbenches. Guitar builders worry about creep for neck joints because guitars can have significant loads on them for many years in poor conditions (like, in a case in the trunk of a car). That said, most still use Titebond though some use hide glue. Just about no one uses PU except for some notoriously hard to glue woods in laminations...and even with those Titebond works just fine with proper preparation.

Hugh Evans has also said that Titebond Extend has the least creep, and that's what I've started using on my loaded joints, such when I build my Drop Top model and have to bend a thick piece of maple and stick it down to the back. That joint gets Extend because I don't want it slowly pulling apart over the next 20 or 30 years.

If you really want no creep, use hide glue.

[Peter Quinn]

How about Resourcinal resin? It seems to address all your concerns. Long pot life, no creep, structural bond, and if your ever in a flood....
Oh, and those dark glue joints speak quality in my book!

Then he has to worry about the sanding dust being toxic, and there is a lot of leveling to do on a bench top. And there are minimum application temperature issues. Does it reach 70 degrees in Ireland?

[ian maybury]

Resorcinol would do the job all right I'm sure Joe based on what's about. It seems like we only get over 70 deg here for about a week a year mind you, and I know nothing about the stuff.

Don't get me wrong - there's clearly at the practical level no major issue with building benches from PVA - certainly nothing that surfaces in any reasonable period of time. I have to confess though to being a bit of a glue snob (those working/that have worked in the structural and engineering adhesives world like to dump on PVA as commodity gunk), and it does seem from what I'm picking up that some of good PVAs do indeed come close to being able to pass the creep testing required of a structural adhesive. (i found a German one claiming to do so) Intermittent loading no doubt gets the others the rest of the way.

The thought of using PU came from some recent good experience with it, having some left, and the feeling that a big bench is not the lightest of duties.

What's surprising is the amount of negative comment regarding PUs. It seems to suggest that between the difficulty of cleaning up and the need for tight gaps, proper wetting and good clamping that while it's basically a very capable adhesive that it is possible to get into trouble with it if it's not handled right.

ian

[joe milana]

I heard from an instrument maker that original titebond was the best PVA for anti creep, but not repairable. I'll admit that I've steered clear of the PU's because of comments I've read on this forum. Based on the info you've provided, that with tight fitting joints, and adequate clamping pressure, it's the way to go, and foaming is not much of a concern with a bench build. Tight gaps, a little moisture, and lots of clamps is not too much to ask of a woodworker. Good luck with the build, I'm getting ready to start one as well.

[Chris Fournier]

Two aspects of this thread are bordering on absurd. One is the negative spin on PU. It is simple to use and very effective - if you have done your homework and understand how it works. If you're grabbing your bottle of PU and using it like PVA well of course it sucks. The second absurdity is the use of the terms "load bearing and structural" when referring to a work bench, glue creep and PVA failure.

You'd think we were talking about gluing heat shielding tiles on a space craft here. Sorry, load bearing heat shielding tiles.

No one gets brownie points for using the most sophisticated, cutting edge or expensive glue on their woodworking projects. Use a product that does the job and represents the best value and concentrate on good design and quality craftsmanship. And by craftsmanship I'd be prepared to limit the reference to tight joints

[ian maybury]

The information Joe is just what I tripped over when I went looking, so go carefully and make of it what you will. I've as above still not made a final decision on which glue to use for my bench, although I'm leaning towards the PU. The plan is to run some test pieces as soon as I have some of the wood milled up.

It's hard not to agree 100% Chris that the negativity about PUs is probably heavily influenced by spin or something similar - perhaps as a consequence of liquid PUs having been fairly positioned in the US as the 'new' (not in Europe) wonder wood working adhesive. Hence threatening the commercial position and longstanding user familiarity represented by PVA.

It's clear too that while an excellent adhesive that PU has it's own handling and use requirements, but none so far seem to be show stoppers.

The thread was intended to dig out some experience of the practicalities of using PU from those having used it on a bench or other large and heavy project. Right now it feels more like it walked into somebody else's argument - one where the battle lines are well and truly drawn. Few it seems have experience, or are prepared to even discuss the topic on merit.

What I don't agree with is that PU is somehow some sort of exotica. I'm buying top quality industrial liquid PU in Ireland for roughly $15 per litre locally. It's coming from a well known manufacturing supplier. Titebond is basically positioned as a consumer adhesive, and as such costs around $18.50 per litre in a gallon bottle. Bought retail in the US Titebond 3 for example is quite a bit cheaper than say Gorilla Glue, but the maybe $100 difference is hardly going to force the decision either way given the total cost of a high end hardwood bench with good quality hardware. The chances are anyway that an industrial liquid PU can be bought rather more cheaply from an industrial supplier.

Whatever you make of the competing technical claims it's clear that Titebond is one hell of a marketing operation. The reality though is that both a good PVA and a liquid PU correctly used should be able to deliver bonds causing failure of the wood in pretty much all cases.

Longevity and long term stability are really the topic here - whether or not there is any possibility that PVAs in a heavy structure like a big bench can run into creep issues and whether PUs might be a better choice in this sort of situation.

The term 'load bearing' is quite normal usage in the topic - it's used by the manufacturers themselves. 'Structural' is equally normal usage for adhesives that form stable and consistently long chain polymers (elastomeric as opposed to semi-plastic) when they cure. The information on creep resistance is what it is - step over whatever the PSI stress level is on a sustained basis and you're going to see creep whether it's in a 1/2in sq bond in a musical instrument that's stressed long term as a result of a curve in the wood, or a 20 sq ft bond in a laminated beam holding up part of a roof. (which stress threshold is significantly lowered by elevated temperatures)

There's as before absolutely no question but that perfectly successful work benches are built every day with PVA.

The implicit view in a lot of the comment that I can't fathom though is why I wouldn't (if the claims stand up, and there are no other reasons not to) use a longer lived and properly repairable adhesive on a very expensive bench project that I have no intention of repeating again, and which I want to do my best to ensure will still be fully functional for somebody else to use in 30 or 40 years time.....

ian

[John Coloccia]

Ian, what are you talking about? You're making it sound like us unenlightened Americans are just brainwashed into using Titebond. You asked a direct question about poly and we said that it requires a lot of clamping and it's messy. We also said that the advantages to using it don't seem to be of any benefit building a workbench. Since most of the glue joints are long grain to long grain, repairability isn't even a factor because those never come apart. It's cross grain joints that could potentially be a concern later on, like M&T for example. A bench should either probably be bolted or draw bored anyhow, so even that's not a concern.

You're hung up on the creep thing, so go ahead and use polyurethane, but I don't see what you're talking about with battle lines, walking into other people's arguments or anything else. It just seems like you started this thread expecting controversy and it never showed up, but you wouldn't know it from reading through your posts. I would hate for someone to start reading this and to come away with the impression that this is some sort of never-ending debate or that there's some controversy here because there's none that I can tell. Many people just prefer other glues to poly, that's all.

I'm sure your bench will be fine with polyurethane. I do suggest stocking up on clamps to get nice, tight glue lines and strong joints. Poly does not tolerate inadequate clamping pressure well at all...probably the worst of any glue I've ever used in this respect, and a work bench will have lots of loooonnnggg glue joints.

[ian maybury]

I'm not here to argue John, and have no axe to grind - it'd suit me very well to use PVA.

The causes of problems with PU do seem to have emerged (clamping force/moisture availability/gaps), but that's an argument in its favour. When a question about using it for a particular application sees it roundly dissed by almost all despite widespread success over many years - while the alternative gets nothing but praise - it however seems pretty clear that somehow there has been some heavy negative conditioning going on.

Especially when even a cursory dig on the web suggests that the choice is more nuanced than that.

None of which means that PVA can't do a decent job, but frustratingly despite the questions about creep, repairability and longevity that have been posed all over the place it's notable that none of the makers (so far as I can tell) has rushed into the space to say that if you use it in a piece that your grandchildren won't (on balance) experience problems with the joints in 30 years.

Which doesn't necessarily prove that PU is any better, but its makers do at least seem to have no difficulty in saying that the polymer is basically permanent.

Which with reality being the complex deal it is still leaves a judgement call to be made.

Anyway. Thank you all for your input, time to let it stew for a while....

ian

[George Gyulatyan]

Actually, it might be a good idea to use PVA for the top at least because of it's ability to creep. Presumably your benchtop is going to be a glued lamination, face-grain to face-grain. The wood in your top is likely to move more in the direction of it's thickness (up/down). If you use rigid glue such as PU/Epoxy/UF, then you're asking for trouble down the line, where it's likely your joints will remain sound, but you'll end up with a cracked bench top.

You're likely to need to flatten your benchtop from time to time anyway, so any issues along the joint lines due to creep with PVA glue will be taken care of along the process.

By using rigid glue, you'll be stressing your benchtop unnecessarily by limiting it's ability to contract/expand with seasonal changes and invite more trouble than you're attempting to solve.

[ian maybury]

That's the point I think George - the best solution to these issues isn't (barring some wonder adhesive) likely to be a one size fits all deal.

Exchanging PMs with John last night resulted in his raising what is a similar point - that a bench contains a lot of joint configurations anyway. That e.g. glue in the likes of one of the very large mortice and tenon joints in say the legs (which involves crossed grain) is probably going to struggle to survive long term anyway because of cyclical stresses caused by seasonal wood movement. He made the point that maybe a mechanical fixing like draw boring is actually the best long term bet in a case like that.

Which all to my mind underlies the point that for sure every single joint is best considered on it's own merits - in terms of both the adhesive selection and the situation.

What tees me off a bit (and is the reason I dug in a bit in this thread) is that it's very hard to get past the marketing whitewash (which tends to imply that 'ours is best for everything', with a bit of fine print here and there to provide a fall back in event of problems) put out by most of the consumer glue manufacturers - to get to the point where a more fact/merit (actually applications engineering as in industry) based approach becomes possible.

The different adhesive types truly do have pros and cons to them, meaning that good information is essential.

Woodweb with its fairly commercial tone and membership has some very ood discussion on the pros and cons of adhesive systems. I tripped over the mention of an apparently very comprehensive manual on adhesive types and methods there in the past few day while digging around, but forgot to make a note, and can't find it now. Think it's published by the US forestry service or somebody like that.

Does anybody have details, or know of anything like this?

Thanks

ian

[ian maybury]

Just for info i found the USDA Forest products lab paper on their testing of liquid PUs here: http://tiny.cc/p24hew

It seems to confirm that PU is way ahead of PVA in purely structural terms. Glues though are about horses for courses. The relatively undemanding nature of most wood applications means that PVA can handle many of them, with the result that its easy handling comes to the fore.

None of which defines precisely where the threshold/criteria that justifies upgrading to a structural adhesive for e.g. something like a bench lies. Or if it even exists. In the spirit of wanting to know I've contacted Andy Brownell who demos Gorilla PU (re. polyurethane bench build blog linked a few posts back), he's planning to post some feedback.

The paper seems (subject to my getting it right) to have found as follows. They tested a variety of liquid PUs against resorcinol, getting similar results from them all except for some differences on water resistance which seems to be influenced by the additives used, see below. Resorcinol was used as their reference standard since it seems it's basically the gold in waterproof structural wood adhesives.

1. On dry wood (douglas fir and yellow birch) PU is much stronger, although it doesn't add to joint strength since the wood fails first.

2. On water saturated wood (the test method uses pressure and vacuum cycles to rapidly produce wet samples after bonding dry) PU produced similar strength joints, but failures when they eventually occurred were in the adhesive bond (by de-lamination) and not in the wood.

Which meant that while it performed very well it didn't technically satisfy the ASTM wet strength standard.

3. On de-lamination testing (probably undertaken to get a better understanding of the above - the method uses repetitive wetting/drying/steaming/wetting cycles) PU (while not always failing) produced higher levels of de-lamination. (that is spontaneous peeling apart of the bonded layers as in the way cheap plywood comes apart when left outside)

4. Use of a primer seemed to resolve these issues. (it's painted on to the wood before the adhesive - structural bonding PUs used in the automotive area routinely specify the use of primers too on some surfaces/applications. It seems likely there's some issue about the actual PU to surface bond possibly coming under attack in very wet conditions that it resolves)

The primer tested was HMR (hydroxymethylated resorcinol), it dramatically improved the de-lamination performance (and hence stops the bond failures in (2)). Based on more limited testing it seemed to more than enable the PU to easily beat the 8% de-lamination rate required by the ASTM test for wet structural wood adhesives.

5. Creep testing of PU (60 day sustained high temperature/humidity/stress (60 - 300psi)) showed no movement at all. It far exceeded the requirements of the relevant ASTM test for structural adhesives.

This suggests that PU used correctly (good fits, well clamped, moisture about) is on this basis going to easily handle the requirements of a bench build, and seems likely to offer excellent longevity. It's actually even without the primer pretty highly waterproof, it's only in fairly extreme saturation and de-lamination situations that it starts to show issues. While not mentioned here it's also very good for bonding to dissimilar substrates. (metals, plastics, wood, composites etc)


All this to my mind means that whatever questions may arise (about the use of PU on a specific wood application) they seem more likely (as already surmised) to relate to practicalities and situation specifics than to the fundamental capability or nature of PU.

These guys used surfaces planed within 24hrs, fine water mist, 0.098kg/m2 of adhesive rolled on, immediate assembly and rubbing, clamping after 15 min to 69N/cm2 (fir) and 103N/cm2 (birch - the makers of PU seem to suggest using higher pressures for more dense woods) for 24 hrs, then re-conditioning to around 8% EMC.

[Hugh Evans]

Hi Everyone,

I thought I felt my ears burning and figured I would chime in to lend my expertise. Since concerns have been voiced regarding corporate affiliations I will start by stating that I have none. Franklin parted ways with me at the end of March after some lovely politically motivated corporate BS, and it should say everything that I am considered eligible for re-employment... So I promise, I'm still very competent in the technical realm of adhesives. Unfortunately, the tech department is now supervised by a guy with a sales background and an MBA. If you ever talk to him on the Titebond support line you can look forward to being quoted all manner of questionable information from official marketing literature. For example: Titebond III shear strength of 4000 PSI from its data sheet was the result of 15 D905 specimens, one of which happened to be an abnormally strong piece of maple that failed at 4000 PSI. Maple, on average will fail between 3200 and 3400 PSI. When the bond results in 100% wood failure you are no longer testing the adhesive, you are testing the wood. Marketing makes use of what they call "typical" values, which rather than having any basis in statistical results, are convenient for the purposes of selling products. If you need a fun question to ask, try questioning why every other product on titebond.com has a strength value at 150°F aside from Titebond III (it loses over 2/3 of its bond strength.)

The article above references some of the lowest level structural tests on softwoods. From my experience, liquid polyurethanes are mediocre wood glues at best (and I have data to support this claim.) They typically fail to cause wood failure on maple during D-905 testing and lose considerable strength at elevated temperatures, which is a strong indicator of creep performance. If they could pass the appropriate ASTM specs, they would have started marketing PURs for structural applications long ago.

With that said, a bench will not see anything close to structural loads and is not an example of a situation where creep is a legitimate design concern. The best example of a situation where creep becomes a major concern are laminated load bearing beams that will be carrying many tons on an indefinite basis without the benefit of any mechanical fasteners. In these situations, resorcinol, epoxy, and some EPIs are the true viable options. Again, they will not be needed for a bench especially if good joinery is expected in the build process.

I've probably raised some new questions, so go ahead and shoot. I'm unemployed and aside from hunting for a new job I am waiting on paperwork to finish processing to start my own company as well. Aside from starting a guitar manufacturing operation I will also be offering my expertise as a consultant. This means in the meantime, I'm happy to answer some questions in my free time... It's fun and it keeps me sharp.

[ian maybury]

Hi Hugh, dead pleased to see you come on in that the problem all along has been to get past the marketing generalities to some data based information. Your ex-employer has reached the point of being the default choice for bonding everything wood - which as an engineer I tend to be slow to buy. (which resistance unintentionally - sorry guys - caused some offence here) It instantly sets my antennae twitching as I'm conditioned to seek out much harder data.

I'd appreciate running a few questions by you. I'd open by saying that (a) I've no particular attachment to liquid PU except that (a) it's locally available and not expensive, (b) i've had good if limited experience with it, and (c) having worked around engineering adhesives R&D tend to regard it as a decent structural adhesive. Also that it's clear that lots of beautiful benches are built with PVA, and that it does a perfectly acceptable job in the short if not the long term. Also that because of the large joint areas and (apart from its own weight) highly intermittent loads that it's probably not as you say a very tough application.

Doing some reading since my initial post (USDA Handbook - Wood as an Engineering Material) one issue that's emerged that i didn't know about is that wood itself is highly prone to creep - that even relatively low stress levels of maybe only 500psi will if sustained cause almost all to take a 'set' (which doesn't bode well for through bolts staying tight), and that a load of only 60% or above of that causing failure in a short term test is likely to lead to failure under a sustained load.

Sounds (a) like it's lucky that most furniture is only intermittently loaded, and (b) that wood adhesives don't actually need to be all that strong......

Pardon my asking so much (maybe you can point to a manual covering the ground), but from the perspective of viewing a bench or a piece of high quality furniture as an item that should last at least a couple of generations:

(1) Does PVA/modified PVA truly hold up long term?
(2) What adhesive would you recommend on technical grounds with cost as an issue (but secondary) for occasionally (but highly) stressed furniture joints as seen in e.g. a slim chair? i.e. is it necessary to step up to using a structural adhesive, and if so which type.
(3) Do you possibly have to hand some real world numbers for permissible design stresses, physical characteristics (impact resistance is not one that gets much consideration) and major strengths and weaknesses of the major wood adhesives? Feel free to apply whatever caveats are necessary.
(4) Is there a trustworthy wood adhesives 'bible' you can recommend - covering not just quantitative technical data but also more quantitative but highly practical real world handling and use considerations?

I'm sticking my neck out here, so feel free not to engage on all topics if it's more than you had in mind.

Thank you

ian

[George Gyulatyan]

When the bond results in 100% wood failure you are no longer testing the adhesive, you are testing the wood.
Given that there are many tests conducted on joint strength comparisons (for example: http://woodgears.ca/dovetail/index.html) show the primary reasons for joint failure being wood rather than glue failure, regardless of the glue being used, example above uses PVA I think, other environmental considerations would then have to be taken into consideration, I suppose, for choosing one type of glue over another, ex. water resistance, heat resistance, application type (veneering for example usually recommends UF), etc.

For me personally, one major consideration NOT to choose PVA is the relatively short open time, which can result in some hectic glueup sessions, depending on the project at hand.

[Stew Hagerty]

Hey Ian,

I'm just curious enough that I have to ask...

Are you an Engineer?

I mean, you're looking at this as some sort of "Super Project".


You'd think we were talking about gluing heat shielding tiles on a space craft here. Sorry, load bearing heat shielding tiles.

+1 for this statement

"Regular" (PVA) wood glue has been around for almost exactly 100 years (Polyvinyl acetate was discovered in Germany in 1912 by Dr. Fritz Klatte). Before that Hide glue has been used for centuries.

Care to make a guess as to how many woodworking benches have been built in that time?

A staggeringly large number!

For cryin out loud, your building a bench that dates back 240ish years.

Now, I certainly am a fan of technology, and have a fondness (sometimes too much so... $$$) for the newest/hottest thing out there. But we're talking about GLUE.

PVA works! And it works very very well.

Now there certainly is a place for PU Glue. I have some in my workshop right now. I use it on bent laminations mostly as I actually prefer Titebond III for exterior use.

Come on, think about how much you could have had done in your shop instead of doing all that research and writing a book on here.

By the way, there is certainly no offense intended in any of this.

[ian maybury]

I'm a mechanical engineer Stew, and am looking at this not just in the context of the bench, but as an opportunity to bring myself up to speed on the bonding of wood in general. I've used PVAs for years in an unthinking way, but not on stuff that put it under significant pressure.

That said I also find it highly unsatisfactory that even basic applications data for these PVAs is not published, at least not where I've been looking anyway. There's lots of generalised advertising claims about, but that's far from the same thing.

I know very well what you have said about benches, and that PVA is the default woodworking adhesive, and am given the tone of some of the replies by now more than a little conscious that many around here seem to regard as the universal wood bonding adhesive too.

The problem is that (even setting aside applications like lamination that definitely require a rigid glue line) reading a bit more widely shows that there's many out there that advocate other adhesives - including some very high profile names. Especially among professional woodworkers and furniture makers.

The problem disappears if the data and other considerations show that this is all about personal preferences, and simple issues like open time - if in fact there are no genuine engineering/structural requirements in woodworking requiring more capability than is delivered by PVA. The above people presumably have reasons for their choices too though.

To put it a bit more directly - if there's data about then it'd be great to have it. If not then let's just let the thread die.

ian

[John Coloccia]

Nice to see you here, Hugh. I was wondering when you'd show up. :)

[David Kumm]

I've never thought much about or cared about types of glue but have been happy Ian has done the heavy lifting here and I've learned something. I've spent many late nights learning about stuff no one else cares about and almost always I've found the issues more complex and the general consensus of opinion to be fairly incorrect or worse. I'm fine with a discussion getting deeper than the normal " I've tried it and it works fine" responses. Marketing info for woodworking stuff is pretty poor and generally dumbed down to be almost worthless. Some techinical info may be over my head but better than being uninformed or misled. Dave

[joe milana]

Couldn't agree with you more Dave, but I can't find anything in this thread to take with me. Ian suggests that the USDA data reports PU as a superior adhesive, yet Hugh Evans labels liquid PU glue as "mediocre at best", and "has data to support it". So which is it? Maybe if Ian would have left the "bench build" out of this thread, and made it simply about PU's vs. PVA's and the data to support them, the emotional responses would have been avoided.

[David Kumm]

Couldn't agree with you more Dave, but I can't find anything in this thread to take with me. Ian suggests that the USDA data reports PU as a superior adhesive, yet Hugh Evans labels liquid PU glue as "mediocre at best", and "has data to support it". So which is it? Maybe if Ian would have left the "bench build" out of this thread, and made it simply about PU's vs. PVA's and the data to support them, the emotional responses would have been avoided.

Joe, hopefully Hugh will respond. I agree that no conclusions yet but I'm still more informed than before and feel even better about using epoxy for entry and passage doors although its use was due more to open time than anything else. Dave

[ian maybury]

:) Thanks for seeing the point Dave.

It's evolved Joe. I started out by asking what I thought was a fairly straightforward question - based on the bench build and on the little I'd picked up on PVAs vs structural adhesives from working in industry, and on the fact that there's as you say multiple views about on choice of adhesive.

As the situation has progressed however it's started to feel more and more like this is well worn ground, with well established positions and people bought in all round.

I don't know about you guys, but when I find myself on the receiving end of 'just take the medicine and get on with it' type comment on technical matters my antennae start to twitch - by undermining what little information is around it progresses the whole deal into a more generalised matter. (my original question has under this stress developed some 'creep' :D )

One thing is absolutely for sure. There's a range of structural and less so adhesives out there. Each has it supporters, and each has carved out a market for itself. Which taken at face value surely suggests that unless the claims are mostly hot air (i.e. it doesn't matter a damn which you use - and the relative weakness of wood and some of the experience based comment suggests that this may be closer to the reality than many care to admit) each has it strengths and weaknesses, and there are right and wrong places to use each of them.

To repeat. If there's good data out there then lets have it. Otherwise let's let the thread drop.

ian

[John Piwaron]

I realize almost 2 weeks have passed since your original post. Here's my 2 cents:

I built my workbench from laminating 2" thick x 1.5" wide hard maple strips for the main portion and joining a built up dog block to it, and so on. All with Titebond. That was about 25 years ago. in the few cases where I felt a joint was not what it should be in terms of fit, I used epoxy in that specific joint. In all that time nothing on the bench has come apart requiring regluing or repair. Nothing.

About 10 or 12 years ago I built the outdoor planter and bench system from New Yankee Workshop. I used red cedar for it. I couldn't get the right thickness of stock, so in addition to making parts of the correct shape, I first planed and laminated red cedar construction materials to build up my stock. That'd be 2x4s, 2x6s and so on. I experimented. About 1/2 of the stock was laminated with gorilla glue, the other 1/2 with titebond2. As I write this now, after all that time outdoors nothing has come apart. Nothing. No delamination, no joints separating. The finish on the bench portion has been a Sikkens Cetol product, the planters are finished inside and out with ordinary exterior latex paint.

The point is that a PVA like Titebond and Titebond2 is entirely up to just about any woodworking task you can give it.

[ian maybury]

Thank you very much John. Regardless of my quest for numbers those have got to be a good practical examples of PVA doing the business in the real world - both in terms of the structural demands of a bench, and with some weather exposure.

ian

[Hugh Evans]

Hi Ian- I'm always ready to take on questions, and to be totally honest I'm used to getting a lot more than what you've started with. Amazingly enough, within the construction technical services group at Franklin I was the only employee with extensive woodworking experience and a background in science (Mechanical Engineering and Biochemistry, along with some advanced courses in applied statistics for fun.) I did a seminar in Cincinnati Ohio late last year to a group of local woodworking clubs, and as it turned out roughly 2/3rds of them were retired engineers. It was exceptionally technical, and a lot of fun for me. Now, on to your questions:

1.) Data on the the longevity of PVA adhesives only goes back to the mid 1940s, from which time there are still joints holding together. Hide glue generally seems to require repair or maintenance every 80 to 100 years, and I suspect part of the reason behind this is due to the fact that they are composed of naturally occurring proteins. Proteins degrade rather predictably over time and on top of that, microbes will happily digest them for energy if presented the opportunity. PVA on the other hand is a synthetic polymer that has been around for almost 100 years. Rather than having the better part of a billion years for life to develop means of digestive proteins, synthetic polymers do not degrade for the simple reason that nothing has evolved to break them down yet. In well designed joints, I have no reason to doubt that PVA adhesives will hold up for centuries if not longer.

2.) Good quality PVA, hands down. When I say good quality I mean PVA based adhesives that at the very least contain tackifying resins and are generally sold as yellow glues. Unless the pieces will be outdoors and exposed to the elements, non water resistant varieties generally have superior performance properties compared to their ANSI/HPVA type II counterparts. The reason for this is simple- getting water resistance from a PVA comes as a tradeoff. Let's take for example Titebond Original and Titebond II (or Gorilla Wood Glue which in testing I conducted a few months ago is every bit as capable of a product but without the dye.) When evaluated for shear strength at 150°F Titebond Original will lose around half of its strength while Titebond II will lose about 10% more than that, with strength returning as temperatures return to room temperature. This response to elevated temperatures is known as thermal plasticity, and if you're wondering if this has anything in common with the fact that all materials have a direct correlation between higher temperatures and increased creep you're absolutely correct. As a point of interest here, Titebond Extend followed by Titebond II Extend have the best creep resistance of any PVAs in Franklin's product line.

People often erroneously assume that joints rely entirely on adhesive for their strength. In reality, good joint design carries most of the stresses in a piece of furniture while the adhesive keeps it from slipping apart. Even if a piece of furniture was designed without any joinery and directly relied on adhesive alone, relatively enormous loads would be needed to induce creep over a long period of time. Franklin has two MTS universal test machines (most people refer to as Instron's, although this is much like referring to a photocopy as a Xerox) worth about $85k each. All but perhaps two or three people, when I was included, ever used these for anything aside from determining stress at failure. One project I was never able to complete involved the determination of the true proportional limit of all of the wood glues. My reason for this was simple: the equipment was available along with the necessary materials, and this knowledge would finally allow thermoplastic wood glues to be treated as true engineering materials. Engineers and architects alike would be able to use such products with standard practices such as design safety factors. Simply stating an adhesive will creep is useless information unless the required stresses and accompanying thermal characteristics are known and understood.

3.) I covered quite a bit of this question above. In the next few days I expect to receive a large cache of my old files from projects which will allow me to, in some cases, provide you with some actual hard numbers.

4.) I wish such a guide existed, but most of the knowledge regarding wood adhesives is known only within the industry. With that said, it has been suggested to me in the past to consider writing such a book myself. Now, I am in a particularly good position write such a book and am strongly considering doing so. One approach I have considered is that of affiliating with a major woodworking website and writing a comprehensive series of articles eventually leading up to a refined version making its way to print. It makes no sense in my mind to make such information so inaccessible to the public. I have personally spent hours with individual woodworkers on the phone answering questions they have sought the answers to for decades. Perhaps a good question for me to ask is if there are any sites people would recommend who might be interested in taking on such an endeavor? Maximizing audience size would be my main interest, and picking up a paycheck might be nice as well.

It would be abnormal if my answers failed to lead you to other questions, and I'm having fun, so if anyone wants to keep going: I'm ready for it.

[Van Huskey]

I love a thread that reaches into my (OK Dick Chaney voice) known unknowns and unknown unknowns...

I have never really had issues with glue, just followed conventional wisdom and liked the results, but I always knew there was much more to know but always had other intellectual stimulus queue ahead of "glue science". Based on this thread it is now in my mental woodworking queue, behind dust collection (I am actually dreading that one) and motor theory (which I am stoked about).

[ian maybury]

Just a quick PS to Joe, then onwards. What I said about PU was quite a bit more qualified. It's an excellent adhesive in terms of it's own properties (strong polymer, adjustable flexibility, great wetting/bonds lots of surfaces well, permanent, structural/minimal creep - all of which i've seen demonstrated at different times), but what I've been trying to get to is hard data on whether or not it (and now other adhesives as well) performs any better or even as well as seemingly inferior adhesive materials on wood. If only because the results in that USDA paper say that it's extra strength didn't deliver any more bond strength on the dry joints they tested because the wood failed first.

It's actually emerged since that there's a gazillion more reasons about too to explain why this is the case - another USDA paper turned up last night which addresses some of these issues. Which are mostly to do with subtler aspects of wood as a material. Especially that it's cellular, highly directional, contains many types of surfaces, variations in structure and compounds depending on type and processing, and that it moves a lot with moisture and temperature cycles etc etc.

This second paper still discusses the general situation and does not present repeatable data/performance/numbers that could be used in the engineering of joints - but it sheds a lot more light on the subtleties of adhesive bonds on wood. (see below)

ian

[ian maybury]

Thank you very much Hugh, that sheds more light on the situation from the inside perspective. It's re-assuring to think that PVA should be long term stable, and that handled correctly it's capable of even the more demanding variety of furniture joints. Even if as seems to be the case the work has not been done to generate all of the data I'm chasing.

I'm guessing that it's got decent fatigue resistance too, that it doesn't come apart if flexed over a long period of time?

It'd be great if you can continue as suits in due course with whatever you can manage.

What you said about wanting to generate data on the true proportional/elastic limits (how heavily it can be stressed before creep sets in) of quality modified PVAs over the temperature and wood moisture content ranges our stuff is likely to be exposed to is exactly the sort of subject I've been trying to turn up data on.

It's an engineering basic (which of course experience can get around - but that sort of knowledge is hard to transmit) to know what stress levels (in the usual shear, compression and peel modes) adhesive bonds in joints can be designed to safely accept.

I'm always of the view that it's better to try to progress these issues from the territory of the black art (with some with the 'knowledge' protecting their position like crazy) to one where it's a bit more science and engineering based.

It's never (thank goodness) going to be wholly predictable, and the judgement/experience/inutition of the craftsperson is always going to matter a lot - but even some sort of basic/ballpark set of working numbers fixing safe stress levels for adhesives in wood joints over the required temperature and %RH ranges would be so helpful.

I'm going to post again after this on what the second USDA paper i mentioned above http://www.treesearch.fs.fed.us/pubs/22048 seems to suggest to me, and try to outline some of the implications i seem to be drawing from it (please shout if sound to be talking bull) - but one thing it seems to make clear is that these limits depend heavily on not just the nature of the adhesive, but also on it's interaction with the properties of the wood in the zone immediately adjacent to the bond line.

That as George G rather perceptively pointed out way back 'Actually, it might be a good idea to use PVA for the top at least because of it's ability to creep.' One clear insight coming out of all of this discussion is that using a very rigid engineering adhesive may in some circumstances cause failures in the above zone by over stressing the timber in this boundary layer - which explains why a relatively weak polymer like cured PVA can make very effective bonds in wood.

The interesting angle that seems to be emerging is that there's a very strong convergence developing between your points Hugh, what the guys have been saying from experience (that PVA works most of the time, and that PUs and the like can work very well too, but that both can fail unexpectedly), and what this paper says. (see the post below for more detail on the paper)

You mention some possibilities like offering courses, writing for a website or writing a wood adhesives applications manual. That's music to my ears, but you'd have to rely on local feedback as to what the level of interest truly is.

One thing that strikes me (presuming the above is true) is that the adhesives manufacturers are by adopting what feels to me like a policy in their marketing of always claiming that ours is the strongest, the greatest, the most xyz and the best are doing both us and themselves a dis-service. That they may actually fear the publishing of fact based data - even if their product works just fine.

Who cares if cured PVA as a polymer is much weaker than an engineering/structural adhesive if in fact that property far from causing problems is just what's needed to deliver robust joints?

Educating people as to realities, and a subtler sell which doesn't treat us as morons (hung up on having the strongest glue - bragging rights for woodworkers :)!! Adhesive envy even...) would surely be preferable?

The magazines and their websites should surely be the natural route by which this sort of information gets transmitted (a series of articles could eventually be published as a manual), but there are perhaps barriers to this if their advertisers are not ready to run with a more fact based message of this sort.

Thanks again for the input, and my apologies if i'm boring the tail off anybody.

ian

[ian maybury]

On the latest USDA paper mentioned above: http://www.treesearch.fs.fed.us/pubs/22048 This is long, so it's for interested parties only.

It's doesn't present all that much by way of quantitative (numbers based) data, it spends a lot of time discussing the adhesive chemistry (rather than use), and it likewise spends quite a bit of time on urea formaldehyde and related adhesives given their widespread use in the production of lumber products - but it does seem to shed a lot of light on how adhesives work on wood, and on what the considerations in achieving decent performance are.

Which it seems are rather more subtle than just having the 'strongest' glue.

All of this is just me as an amateur trying to figure out what it seems to point towards - I'm not a professional. So take what I have to say with a heavy dose of caution.

I've been grouching about the lack of applications engineering data about on wood adhesives. They seem to suggest that this lack is genuine because the work simply hasn't been done to develop it - because the bonding of wood (compared to say engineering materials like metals) is fiendishly complicated, and is influenced by many often unpredictable variables.

I guess too that craft woodworking is pretty low rent territory for the adhesives players, and that people are getting by fairly well (albeit with some degree of unpredictability) based on experience with what's available.

The scenario described (as I read it - i'm being more explicit than the paper, and stating what I think they are pointing to) builds from the suggestion that the regular cycling of moisture levels and temperatures in use conditions and the nature of wood mean that most adhesive failures originate not in the adhesive itself, nor in the body of the parent wood - but in the material in the stressed boundary zone between the two which is comprised of variable mixtures of glue saturated, glue coated and uncoated wood.

The resulting stresses cycle repeatedly under the influences of environmental changes and during the curing of the adhesive, and may even reverse - creating the risk that the material in the affected zone may become over stressed and fail - if not immediately then eventually through fatigue. Joints may even fail spontaneously/de-laminate without any external loading as a result of this - which is perhaps another reason why a lot of wood adhesive testing involves temperature and humidity cycling.

What happens at the level of the internal structure of the wood is quite hard to predict, it seems there are lots of different failure mechanisms that can kick in. Bear in mind that the cellular structure of wood is highly directional and varied. Movement is different in different directions too depending on the grain orientation as we all know. The adhesive may coat or may penetrate the structure and alter its physical properties, as may moisture. There's potentially quite a wide variety of compounds present (or not) that can influence the physical behaviour of the wood itself (including moisture) and the adhesive (and it's cure) too.

They describe the various adhesive chemistries and broadly how they relate in general outline to this complexity, and suggest that a better knowledge of what goes on in this zone (where most failures occur) is the most fundamental requirement in improving to improving the performance of wood adhesives.

That improving their performance requires progressively tuning the properties of the various types to stop or delay the onset of as yet unknown modes of failure.

The adhesive chemistries and how they determine the physical properties of the cured material are it seems actually pretty well known. So they can be tweaked, but the million $ question is to know what is required.

It seems that because of these (and no doubt other) unknowns that there is a random factor in play - that adhesive joints on wood are subject to occasional and unpredictable failures. Albeit rarely, or else over a long enough time. Stuff can crop up as a result of accidental happenings/our not knowing what we need to control. (this happens in engineering adhesives too on surfaces like metals, but the complexity and lack of homogeneity of wood makes it more difficult)

The thrust of what they seem to have to say is that the balance of flexibility and other subtler properties are often more important in wood adhesives than ultimate strength. That for example a very rigid mass of cured material will sooner or later trigger failures of the sort described above in the boundary layer. Too much flexibility or plasticity can alternatively lead to movement and possibly creep.

HIgh quality PVA (while not immune to failures) does very well it seems precisely because it's flexibility (while presumably sailing close to the wind on creep and structural performance) means that it avoids issues like that above.

Despite the polymer being potentially so much stronger in the case of structural adhesives like PU and epoxy they can on occasion (even presuming the required flexibility has been built into the formulation) be subject to failures caused (presumably) by subtle issues in the boundary zone too. Maybe a little foaming in the bond line with a liquid PU isn't always a bad thing... :)

The bottom line seems to be that it's possible to bond wood very successfully with a wide variety of adhesive types. It sounds actually that getting a specific formulation that's been fine tuned to the needs of wood as above as well as taking accounts of the external boundaries of the application may possibly be more important than the specific type/chemistry. (i.e. PVA/epoxy/PU/plastic resin)

That's of course presuming the application has been correctly designed within the external boundaries that we are all familiar with. For example:

An adhesive has to cope with the expected environment, so the need for basics like water/temperature resistance, and ability to bond the wood in question is a given.

Joint layout and design are important, in that not only will the joint take quite a large part of the load in many situations, it also determines where and how the bond is stressed. A stress concentration in only a very local area can easily cause a highly local failure that will with the load cycling that's inevitable in wood bonds and time propagate and cause failure.

The processing of the wood is important, as is the possibility of contamination. Freshly cut surfaces seem to soak adhesive into the structure more easily, while those that have been hammered or heated by blunt tools can block absorbtion. Solvent washing just before bonding may be necessary to shift resin and or oils.

Clamping forces and gaps are important variables that play a part in determining both the properties of the cured adhesive and its penetration too.

The random variable suggests that testing is always going to advisable. While a quick whack it with a hammer test will tell a lot, the above complexity suggests that working the situation up to one where long term performance is predictable is going to be rather more complex, time consuming and equipment heavy than most of us can hope to support.

Which brings us back to the gap in availability of good applications engineering data for the wood adhesives we use.

Please pardon my hogging the thread - there no doubt angles in wood bonding that are not mentioned above too.

Thoughts?

ian

[John Coloccia]

My thoughts are simple. I rarely ever see a glue failure along edge joints. For example, I don't commonly see a table top that's just fallen apart, or a chair seat that just let go. I'm sure it happens, but I don't see it very often. I also don't commonly see laminations coming apart, multi-piece legs falling apart, laminated guitar necks falling apart, etc. Long grain joints, so long as they are well made, just don't ever seem to fail not matter what glue is used.

Cross grain joints, on the other hand, all seem to eventually fail no matter what glue is used, i.e. chair legs and mortise&tenons. IMHO, the longevity of a piece has everything to do with choosing appropriate joinery and executing it well, and practically nothing to do with the choice of adhesive. If the cross grain joinery depends upon glue for strength, then I believe the battle has been lost before it's started because that glue will eventually fail. If you really want to build for longevity then mechanical fasteners, draw bored M&T, and dovetails seem to outlast everything in my opinion.

[David Kumm]

While not on point regarding PU vs PVA, it seems I've read that epoxy that has been thickened with dust or silica performed much better than unthinned in certain applications. I've used epoxy a lot due mainly to open time in assembling passage and entry doors and any info regarding it would be appreciated as well. Maybe Hugh or others have some insight on proper use of epoxy as well. Dave

[Hugh Evans]

I have not yet had a chance to wade through the paper. However, John has raised an excellent point I would like to address. Edge and face gluing are ideal scenarios for wood glues (PVA specifically in this case, PUR is sufficiently brittle that I've seen 100% adhesive failure in shear at 1500 PSI.) If end grain must be glued one simple additional step will increase bond strength by 25% to 40% and that is diluting the adhesive with water in a 1:1 or 2:1 water:adhesive ratio to create sizing. brush a bit of this over the end grain and give it a few minutes to soak in before applying full strength adhesive... Your joints will be much better. This is also where 100% solids adhesives tend to shine because they behave the same way in both joint types. You are absolutely correct that proper design is one of the most important factors in the life of a piece. This is a great opportunity to mention the importance of tolerance within a joint. Glue lines in excess of about 0.010" and thicker will not perform like bond lines thinner than 0.005" and this is an extremely common cause of failure. The easiest rule of thumb for a closed joint such as M&T or dowels is to allow total play in the joint not to exceed 1/64" and if it's a friction fit, you're even better off.

[ian maybury]

Hi Hugh. Thanks for the info on bonding end grain, and thickness of bond lines.

John C has mentioned the face to face grain at 90 deg situation before (as in say a bridle or mortice and tenon joint) where the heavy wood movement is in directions at 90 deg to each other. This clearly is going to work an adhesive really hard in shear, and as he said there's an argument that it's then possibly better to rely on mechanical connections like draw boring. Except that it's done all the time. (on panel door frames etc)

Can PVA or any other adhesive realistically survive long term in this sort of high shear situation on a hardwood? Especially given what the second USDA paper above seems to be suggesting about the risk of failures over time as a result of cycling by the wood movement. Is a structural adhesive like the epoxy David has asked about likely to do any better in this sort of high fatigue scenario?

On the brittle failure of PUs you mentioned above. Do you know was there a reason in that particular case? It doesn't sound typical of a PU unless the product was badly formulated, the wood was weak, or the gap was large and it foamed into it or something. The first USDA paper i linked above had the product examples they tested all going to around 3,500psi in shear on yellow birch, and 1,700psi on fir - but the point was that they were all stronger than resorcinol on the same test, and the failures in all cases were the result of the wood failing first.

From what I've read PUs in general (as structural adhesives) give away a bit in terms of peak strength and heat resistance to epoxies and acrylics, but gain in toughness/flexibility and ability to bond a very wide range of surfaces.

The problem though is that it seems from the USDA material that these high strength adhesives may offer no gain on wood if in fact too much rigidity in the cured material leads to failures of the wood/adhesive mix in the boundary, or if there are other unpredictable factors that may kick in to cause problems.

It does seem to happen. I can remember in the 70s spending days scarf jointing (over about 12in x2) a break in a kayak paddle shaft to splice in a piece of ash to bridge it. The adhesive was a urea formaldehyde named cascamite that we were using to build kayaks. It snapped instantly on the first pull on the paddle. Not across the wood, but along the full length of one of the bonds leaving the adhesive film 100% intact. (a huge joint area) It didn't seem to be the adhesive that failed because it pulled out a thin layer of wood fibres, in retrospect it seems like it was the boundary layer that failed.

ian

[John Coloccia]

I don't want to speak for Hugh, but I've thought about this some and I believe that the tendency for wood to fail right at a glue line is that the very stiff glue creates a stress riser.

[Hugh Evans]

Actually, 90 degree grain orientation can be fine... If and only if the construction is balanced. This is the same reason you should never see plywood with an even number of layers. I don't feel comfortable completely ruling out unbalanced cross grain construction as a potential contributing factor to a failure, but I would be very surprised to see it as the primary causative mode. The construction method is sufficiently common that I have never encountered such a failure and aside from true load bearing applications would anticipate no appreciable advantage from using structural adhesives.

With respect to the polyurethanes I've tested, the brittleness is not a formulation problem so much as an inherent property of the polymer. Also, 3500 PSI on birch and 1700 PSI on fir. While wood failure is possible with PUR, it's not always guaranteed. My father, who has worked at Franklin for over 27 years now as a Technical Specialist in the industrial products group, has a saying that if PVA adhesives didn't exist they would be what everyone would be trying to invent. I fully agree with this statement, and even now that I am not affiliated with any adhesives manufacturer my viewpoint is that for the vast majority of applications there will be no advantage to using another chemistry than PVA. PUR, Epoxy, EPI, reactive hot melts, etc. have their places but by and large they are for highly specialized applications. A great classic demonstration of the brittle nature of PUR is to take two identical edge glued boards having assembled one with PVA and the other with PUR. Allow them to fully cure and then try dropping each one onto a hard surface such as concrete or tile. Sometimes it can take a few tries, but the PUR will fail on the bondline itself due solely to poor shock resistance. If anything breaks on the PVA board it will be wood.

I ran a test once out of curiosity on wood cutting boards. One was assembled with Titebond III and the other with Gorilla Glue. Both were placed in the break room dishwasher for a few months and the board assembled with Titebond III experienced no delaminations or failures of any kind. The Gorilla Glue board split at the ends on over half of the joints. This once again demonstrates the highly brittle nature of polyurethanes and in this case was induced by cyclic stress.

As structural adhesives go I'm a big fan of epoxy. Acrylics tend to be fairly brittle. Epoxies as a class have extraordinary range of formulations. Off the top of my head I cannot think of a single application where either epoxy or resorcinol would not be significantly more capable than polyurethane. One important factor for epoxies of which most woodworkers are unaware is that even a room temperature curing 5 minute epoxy can be made to perform much better by adding one ingredient: heat. The glass transition temperature (there's not a great way to explains what this means without delving into chemistry and materials engineering) of most epoxies is around 180°F give or take. By allowing them to cure for a few hours in this temperature range the epoxy will more completely react and result in a much more resilient product. In fact, if you look into how epoxies are used in the composites and aerospace industries you will find that an initial room temperature cure of an hour or so followed by another 2 to 4 in an oven is standard practice.

As far as your story about the kayak paddle goes, there are simply too many variables at play for me to even guess unless I could analyze it in a lab. UF can do a lot of strange things and especially back in the day before formaldehyde became associated with Satan, much more variation was possible. Modern UF, even in a thick sample, will rarely start to craze until several weeks have passed and indeed are nowhere near as capable as they once were. I read a preliminary internal report about 6 months ago from a veneer checking project testing a wide spectrum of adhesives. Shockingly, EPI was thoroughly trouncing the thermal plasticity of UF which is the last thing anyone expected to see.

[ian maybury]

Thanks for the thoughts Hugh. The kayak paddle was years ago, so heaven knows what happened.

TBH I have to throw in the towel so far as the attempt to raise comparative engineering data on wood bonding adhesives is concerned - at least so far as obtaining information that in any usable way sets out the strengths, weaknesses and likely performance of the various woodworking adhesives in a manner that's likely to be a reliable basis for design.

What's out there beyond the basics is often utterly contradictory and confusing....

ian