Limits Of PVA In Structural Bonding?


From original questioner:

Hi all. Because modified PVAs like TB seem to have limitations in structural applications i've been investigating the possibility of needing to use a liquid polyurethane on a heavy work bench in European beech. Failures in the field would be serious.

PVA has handling advantages, but it's unclear at what point e.g. creep may become an issue. Plus there seems to be little info about on it's longevity.

Liquid PU seems to offer excellent structural performance (based on some experience so far), but needs rather more careful handling to avoid inconvenience and deliver consistent results.

Can anybody say (preferably in fairly quantitative technical/engineering as well as general terms) where the limits beyond which PVA is likely to start running into trouble are likely to lie. e.g. stress levels, environment (heat/moisture/UV etc), degradation and so on?

There presumably is a threshold beyond which a structural adhesive like the PU becomes necessary, and below which it's overkill - but where does this boundary lie?

Thank you

From contributor Je

It's difficult to give you precise numbers because it depends on the species being glued. In general though, PVA is not intended for structural use. By structural I mean either a joint that's under stress or under load. Normal PVAs don't really have much moisture resistance either. You would need to look to a crosslinking PVA for any real water resistance. PUR is probably an excellent choice for the bench you're gluing. It has more thermosetting characteristics and very high moisture resistance. You might also consider resorcinol resin if you want it to be bullet proof.

From contributor Ia

Thank you Jeff. I've a little background in adhesives development, but not specifically with PU or PVAs and am perhaps a little more wary of the differences than is absolutely necessary - while heavy there is at least plenty of joint area in a big bench.

As often there's lots of generic information about, but it's hard to get to any real data on topic, even for the more recent modified PVAs.

Sounds though like PU used correctly (tight joints, adequate moisture and proper clamping) is at least a safe choice.

I'm treading cautiously as (a) i'd rather not have my own bench come apart (it should be something to be passed on for long term use), and (b) there's a likelihood of doing some commercially.


From contributor Le

Why don't you use Unibond 800 or West System Epoxy or Resorcinol. All are pretty proven glues.

Pretty sure if you are building a workbench then Titebond II would work fine.

From contributor Ia

I'm pretty sure it would be OK too Leo, but it bugs me not to be able to verify where the limits of PVA lie in broadly quantitative/data based terms as i'm an engineer. The advertising meanwhile does everything it can to walk you into using the stuff....

I'm in Ireland so West system is an option from the UK, I thought of the PU only because i've used it before with good results, and liked the easy removal of squeeze out.

Plus while I've lots of finishing and small joint experience with epoxy I've never used it on a big glue up. For sure it would eliminate any uncertainty though.

Think I'll check it out..


From contributor Le

Sometimes engineers just over think too much LOL

With PVA you have to really rush the glue up. If it is a multi lamination glue up, unless you have a glue roller then you will be pressed for time or need to do the bench in section. glue 3 1' footers together then glue the 3 together.

Epoxy and Resorcinol have a good open time, maybe 20-30 minutes. Not sure about the Unibond 800

From contributor La

I'd like to see Gene comment on this.

All 3 glues are strong enough. Unless Titebond is being used under water or this is for an out door bench it's water resistance is good. I've got a laminated cutting broad made 30 years ago with it that shows no sign of glue line failure. Washed off in the sink after each use. Resorcinol is a great glue, no creep, but a PIA to use.
Urethanes are even a bigger PIA.

Assuming the stock is well prepared, I don't see creep as becoming an issue here.

My German made work bench, very old, looks like it was made with urea formaldehyde glue. It just had a failure due to me over stressing the tail vice. The fits we good but UF seems to get kind of brittle. With fluctuations in humidity, stresses are put on joints and the brittle nature of UF seems not to cope with it well. I've had this happen on other things glued with UF also.

From contributor Ia

:-) I guess Leo that one of the things an engineer learns early on that gets ingrained is that unlike in some fields (where it's possible to claim the sun, moon and stars if carefully worded and get away with it) - the basic process in engineering design (after the creative bit) is the identification and successive knocking down of the unknowns. i.e. 'it' can't be presumed it's going to work until this is done.

It's kind of tough to weasel your way out when there's a pile of broken parts on the floor in front of you...

Think you have to be right Larry that PVA will do an excellent job. Even if the open time is a little short for comfort. There's been lots of lovely benches built with it.

You don't see mention of adhesive choice by those doing them commercially - but i'm not sure how to read that.

My guess is that there are issues they seem most likely to be longer term...

Ta guys


From contributor Le

I've been using Titebond I & II for 20 years now and only had a few failures. All of them evident immediately. Most of the time the wood will break before the glue will let go. Just don't starve the joint. I have never been able to clamp it too tight and squeeze the glue out as they claim can happen.

Titbond I will have less creep than II, but TB II is water resistant to the point of water proof on furniture. I wipe the glue on my pants and then do a hot wash in the machine with detergent and it NEVER comes off. I suspect it will stick to the wood pretty good too LOL

From contributor Ia

Good to know Leo. Titebond is actually only recently available over here.

I meant by the way 'if' there are issues in the last sentence.

I've used a lot of specially bought yellow aliphatic while model aircraft building because it sanded well, but it's not generally available here either.

We do get lots of white glues/PVAs, and some are 'modified' types with better water resistance, but it's actually next to impossible to know how they differ from the bottle unless you buy a professional product. I've used quite a bit, but generally not on what you might call 'stressed' joints.


From contributor Ri

My concern with epoxy would be overtightening to show no gluelines and starving the joint. Titebond I on my European style bench in walnut and red oak is over 30 years old. Made my living on it for 8 years, then moved it back to my basement when I moved into a corporate woodworking job. No failures with the top, or the wood tail vise. Only steel in it are the screws.

From contributor Ia

Thank you Rich, there's been several post similarly of long lived bench builds on S/C too.

Which all bodes well for the TB route.


From contributor Le

Pretty much the only time I don't use TB II is when I want a waterless glue.

From contributor Ad

This whole idea of the structural integrity of a woodworking glue is not valid.

The glues mentioned in this thread are all stronger than the wood itself. During stress tests the grain seperates near the glue line.

The only reason that epoxy resin is not structurally rated for things like LVL and the likes is the low temp problem. Epoxy starts to loose some mechanical properties above 200F. I would assume that PVA has a similar issue.

I would contend that epoxy resin laminations in wooden boat construction especiallay masts have been tested far beyond any other adhesive. Keep in mind that the exact same epoxy resins are used to fabricate all carbon composite in boat construction including masts.

The longevity of PVA wood glues is in my opinion indefinite. Unless they are subjected to alot of water.

I'm curious as to "liquid poly" adhesive. If this is Gorilla glue you are way off base on all of your previous assumptions.

From contributor La

I don't have a lot of experience with urethane glues. What little I do have has been mostly negative. It has a very short shelf life once opened, it expands as it absorbs moisture for curing making a mess that is difficult to clean up. I don't have any experience to know about creep.

"I'm curious as to "liquid poly" adhesive. If this is Gorilla glue you are way off base on all of your previous assumptions." Please explain.

From contributor Le

I think he is referring to the PUR hot poly glues

From contributor La

From Ian's first post "Liquid PU seems to----" I didn't take this as the hot melt PU.

From contributor Ad


I was asking which PU.

Gorilla type glues are poor wood adhesives. I really do not consider them as wood glue. They fall into that large group of adhesives at the hardware store that includes GOOP. The ones that you use for art projects, dissimilar materials, and as a last hope.

I have no experience with the hot melt PU what so ever.

From contributor La

We've got the Franklin hot melt PU gun, rarely used. The adhesive is expensive and once the cartridge seal is broken a short life. The stuff seems to stick OK for what we were using it for. Just used to put small decorative trims on.

From contributor Ia

By liquid poly i meant single part low viscoity moisture curing polyurethanes guys.

My original thought was that while I knew that quality PVAs clearly do an excellent job on most wood applications that there might be longer term issues with creep, moisture resistance or other problems to do with the cured polymer degrading over time.

That there's very little hard data out there on the subject when you look (structural limits and longevity of bonds made using the various types of woodworking adhesives) didn't help matters.

Doing a lot of reading (there's some useful USDA papers out there) since the original post has brought out quite a lot of information.

PU is well able to deliver excellent bonds in dry wood, but the formulation, gaps, moisture availability, clamping and so on if not correct can all cause it to perform badly. Longevity is in terms of hard data so far an open question too.

As Adam implies there's USDA papers that suggest that most adhesives can do a decent job of bonding dry wood - but it's important that they are formulated to be flexible enough not to cause eventual failures in the boundary layer of adhesive impregnated wood. i.e. Maximising the strength of the polymer is far from the primary consideration.

PU and other structural adhesives it seems may on occasion end up too rigid, leading to longer term failures. One reason PVA often does very well seems to be because its relative flexibility doesn't over stress the interface.

Just what the precise limits and capability of either is is hard to find written down in engineering terms though.

It seems according to one USDA paper that because of the enormous complexity of the topic there's not been enough work done to fully understand what goes on in this boundary layer, the properties of the various adhesives on wood, and how failures occur - with the result that unexplained failures do arise with all wood glues.

Which means that glue use is a lot about experience, skill and gut feel - there isn't the data available to make it a fully engineering or science based topic.

I've not for example been able to raise even a simple number for the typical shear strength of a good PVA bond on a typical hardwood, and how it changes with temperature and moisture content.

It's no doubt out there somewhere, but it's not widely published by the glue makers.

My limited experience with a locally available (European) industrial wood bonding PU has been good, but like the PVA there's again no useful bond performance data included in the data sheet.

Anyway. My view has shifted, in that it seem that PVAs are a better long term prospect than I has suspected might be the case. It's equally become clear that the relatively high strength of PUs and other structural adhesives isn't walways an advantage, that they can be a bit of a double edged sword. They need to be correctly formulated for use on wood.

I've pretty much given up on obtaining hard engineering data for bonds on wood because the definitive version seems not to be out there - it's down to making a judgement call based on the experience of people like yourselves.


From contributor Ge

There are two issues in this thread: first is the strength and durability of the adhesive when under high load. The second is the failure of some adhesive joints.

It is well known that some adhesives do not become hard when cured and therefore WHEN UNDER HIGH CONTINUOUS LOADING will creep. That is, the two pieces of wood will slip or move past each other without breaking the joint. (There are some folks that use the word "creep" to explain short term movement of the adhesive when there is not any large load applied; this is different. Also, the creep here would not be an issue with a chair leg--it is loaded only once in a while and probably not at a high enough load level to result in creep that affects the product.) Most (maybe all) pva adhesives do have this long term creep and so cannot be used structurally. So, a so-called structural adhesive is needed. The definition of "too much creep" will vary from product to product, so it is hard to define limits.

The amount of creep is affected by heat and moisture. However, different pva formulations have different responses. For example, TB is very sensitive to the temperature in use and will creep when warm much faster. TB II or TB III are formulated so that they are not quite as sensitive to heat and moisture...a chemical reaction occurs that is not fully reversible.

Finally, when dealing with the acceptability of certain adhesives, there is the question of what damage is done or is likely to occur when creep happens. If life is threatened, such as with a laminated beam in a human-occupied building, there is a high risk and so we must be extra careful. If the question concerns a chicken house, we might not be as concerned as the damage done with failure is not too expensive or noteworthy....who cares if the roof sags a bit? Bottom line is that we tend to over-design, just in case.

The second issue is a failed joint. In my 50 years of experience, I very seldom see a failed joint that is caused by the adhesive itself. Rather, almost all failures are a result of the man-made issues, such as having too large or too small of a gap due to incorrect pressure, from changing pressure once the pieces have been joined; having non-flat surfaces, having non-fresh surfaces; and so on. The woodworker always seems to blame the adhesive rather than the way he is doing things. (Examples: How many times have you heard that epoxy joints are weak? When the joint is examined, most of the time, the joint is way too thin due to way too munch pressure for epoxy. Another example: When a table manufacturer surfaced the wood within ten minutes before gluing, rather than hours, their defect rate dropped from 8 out of 50 to 1 out of 50. how many times have you seen or heard about open glue joint on the ends of panels in the wintertime...caused by wood that is a bit too high in MC for its environment and not the adhesive.)

So, some of the discussions prior to this have at times confused the issue by talking about one or the other items, or both.

Regarding PU adhesives, this is a family of adhesives with properties that range all over and that require different application techniques as well. Some adhesives are cheap and poor. Others are designed to flow easily when hot, others are not prone to creep but require time to cure. So, any discussion about PU adhesives should be very specific about the product it is being used on and how it is used (pressure, temperature, etc.). To condemn or to support all PU adhesives for any application is likely incorrect...there are good PU adhesives for many applications, but not the same one will work for all. Plus, PU adhesives require properly prepared wood, just like pva adhesives.

Incidentally, gluing end grain with any adhesive is really tough, if not impossible...assuming high strength is needed.

I hope these words provide some clarification and help with understanding some of the previous responses.

Regarding longevity (other than creep), I am not aware of longevity issues with commonly used adhesives for wood that is kept dry and not too hot.

From contributor Je

Thank you Gene, I agree completely with everything you wrote!

From contributor La

Thanks Gene

From contributor Ia

Hi Gene. Pardon my not noticing your post for a while.

Thank you for those clarifications which seem in essence to say that provided mainstream wood adhesives are selected and used as recommended that creep, unexplained failures and longer term degradation are unlikely to arise as issues in furniture and similar applications.

That seems reasonable to me. (given that each has carved a niche for itself, each has it supporters, and also because I've used pretty much all to some degree or another on wood before with universally good results)

My mistake was to undertake a short search to check out whether there was a stand out adhesive to use for the bench project - all sorts of contradictory claims, criticisms and feedback turned up. Which were not easily disentangled with the information and experience I could draw on.

Which I guess begged the question as to whether or not there might be some less well known issues lurking about that needed watching out for.

Thanks again for setting it out the big picture.


From contributor Gl

Fine Woodworking did a test of glue strengths and characteristics about a year ago (or 2 years). It might be helpful to you.

From contributor Ia

Ta Glenn. Think i saw that - they tested 90 deg bridle joints to destruction on an Instron (push/pull tester) ?

Trouble was it didn't really bring out the why's and the wherefores. e.g. they found PU to be weak, but 'surprise surprise' given that they used it in lightly clamped and possibly loose fitting joints.

The tests (which were purposely pitched at a fairly basic level) likewise made no attempt to address subtler issues like the ability of PU or PVA to handle long term sustained loads, cycling stresses caused by expansion and contraction, heat and moisture.


From contributor Ji

Here is some technical info on the Adhesive Bonding of Wood Materials from CED that may be helpfull.

From contributor Ia

Thanks for that Jim - a good find. It took a while to work through, but is a more comprehensive document than what I had seen so far.

Summarising what I seemed to retain it's of course process/sheet goods manufacturing oriented (the big user of wood adhesives), but it provides a rather more detailed snapshot of the underlying science - especially the mechanisms of adhesion, and some of the stuff that goes on in the wood/adhesive interface.

It more or less supports the view that the mainstream wood adhesive types work well if formulated and used correctly - but that each has it's own requirements. (e.g. tight gaps, decent clamping in the case of PU)

There's some informative stuff on timber types (bonding gets tougher with increasing density and/or oiliness), surface preparation (especially why freshly cut/planed with a sharp blade is best), and the effect that sanding, aging, contamination by various treatments (in the case of mostly sheet goods) and the like can have on bond strength.

Longevity of PVA (so far as is known) seems broadly OK, but it has lower limits in terms of humidity, sustained stress levels and heat - and of course does not bond properly to itself for repairs.

As before predicting actual numbers for use in joint design requires a bit more than is here, but it also shows why it's by no means a straightforward deal..

Thanks again.