I think you are being overly cautious. I'm leaning against a European beech bench that is 40 years old and has cycled between 30% and 90% RH annually. It was glued up with PVA and has spent most of its life without a finish.
Boats are not a good comparison to interior furniture. No heavy sun and water exposure. Boat construction also has to deal with extreme dynamic loads. Your glue joints will be straight edge grain, most likely having the ring orientation more or less the same to show quarter figure. This means that the wood will expand and contract in unison and not stress the glue joint.
As a bonus, you will be able to achieve a finer glue joint with PVA than with epoxy. PVA joints are stronger when there is a near zero joint thickness. Epoxy needs a bit of gap so as not to starve the joint.

My background Tom is engineering R&D - including with a number of years in adhesives development. It's probably this and my exposure to engineering adhesives that makes me think of using epoxy.
My experience of PVA glues on wood is much more recent...
You're probably correct in that most unthinkingly reach for the PVA - we'd probably (or maybe not?) have heard by now if it failed in such applications.
That said there are those using epoxy too for this kind of work for similar reasons that swear by it. Others that accuse it of unexpected failures - although there's a load of reasons why this could happen which are not fundamental to the adhesive.
I've encountered the thicker bond line issue using thickened epoxy adhesives when building fixtures in birch ply.
Epoxy is very strong, but that is potentially also its downside....

PS. This paper sets out pretty clearly the reality that epoxies don't don't do so well in wood bonds subject to movement resulting from changes in moisture content: https://www.fpl.fs.fed.us/documnts/pdf2006/fpl_2006_frihart003.pdf
It also sets up a scenario where there might be reason to think that a very flexible epoxy such as the West G/flex might (?) overcome the problem - although if so it's a bit unclear why the company is not more clearly pushing the product for bonding wood subject to changes in moisture content...

Most woodworking adhesive failures in my opinion are a cause of poor joint design or poor execution. G2 would be beneficial where you have large sections of wood coming together at right angles such as a rail and stile joint on an exterior door. Most often though a rigid glue line is sufficient and desirable. PVA is as strong as the surrounding wood, inexpensive and easy to work with. There are a variety of formulations tailored to just about any use. There are a few instances where PVAs pose problems and epoxy is more appropriate.
As long as you start with properly dried wood, machine true glue surfaces and clamp evenly with sufficient force under the right conditions your bench top glue ups will be solid for decades with a good quality PVA.

The difficulty Tom is that standard rigid epoxies are known to fail adjacent to the bond line if there is significant movement in the wood.
The problem seems to be that the adhesive is too strong and too stiff - it can prevent a layer of the wood adjacent to the bond line going with the movement in the body of the piece.
The failure in these cases can look as though it was in the epoxy since it occurs in cells made brittle by its filling them - but it's not. (see the linked paper)
As with a lot of these things it's to some degree about cycle numbers and joint and wood condition specifics - and dimensions.
Intererestingly enough West US technical have just come back tonight to say that they would expect the G/flex 650 to be fine in the application (the 655 is silica filled and likely too viscous for the application) - that the highly flexible formulation is both capable of enough extension and not so stiff that it causes failure in the wood.
The 650 is quite a bit more viscous than their stock 105/205 system - you can't use the can mounted plastic metering pumps with it.
They also say that in their experience the six week intensive moisture content cycling test mentioned above indicates this - perhaps because it while short is very extreme and would have caused damage and hence reduced the strength numbers obtained when the test pieces were pull tested afterwards.

That is an interesting article. I'm not sure that the test results pertain to your bench project because I don't see the wood movement you will experience will be sufficient to destroy the epoxy bond.
However to answer your original question, search with the woodweb search function "g flex epoxy" and you will get plenty of results that should help you.

I searched the forum but didn't really find definitive information Tom. I even posted the same query quite a while ago.
You're correct I suspect regarding movement in the bench - it may not be enough to cause a problem one way or the other.
The day could come too though where it is left in the rain while being transported or something.
I'm like yourself unsure whether or not the G/flex test described above proves durability in the very long term in more moderate conditions and with less movement.
West probably do have a fair idea though, and that was the precise question I posed to them this time around.
The guy I spoke to gave it to me in writing and was a senior type with experience - not the usual junior that gets handed the job in many places. They often can't even understand the problem you are trying to get an answer to.
What seems definitely to be the case is that modified PVAs do on occasion creep in these applications - I've had it from a guy manufacturing heaven benches in the UK who switched to epoxy for this reason.
The G/flex at this point feels like the best show in town unless somebody has definitive experience.
Stock epoxies don't do well where there is wood movement - that's one of the reasons why West always specify that wood in boat applications is sealed with a coat of epoxy.
Also why epoxies are not FAA approved for use on wood in aircraft construction. They are widely used on homebuilts though...

..." used widely in home builts..." not a glowing endorsement with the regularity that they seem to fall out of the sky around here :)

PVA. If in doubt use a glue line joint.
Done

I agree with Dave. You are overthinking this. It's a bench. Titebond has a long history of success holding wood together. Use TB II if you are concerned about short term rain exposure.

You can't have it both ways, so rigid that there's no possibility of creep but not so rigid that it will accommodate wood movement. The glue that solves one concern will fail at the other. TB Original or Extend is a good compromise.

John

"heaven benches"? I guess they have to be perfect?

Here you go.

West System (Gougeon Brothers) pioneered the use of typical West 105 resin system for direct bonding of milled 1/4" thick teak to composite(fibreglass/carbon fibre/foam) decks of sail & powerboats.

The deck gets soaking wet, dries completely in the baking sun. Structurally the deck goes thru all kinds of cyclic twisting. The wood in theory is incompatible with the composite deck.

The typical West 105 restrains the thickish veneer of an oily wood. There are no other fasteners. This has gotten to point of them glueing up large sections on a shop table. Cutting it to a design or matching a pattern by hand or CNC. Then vacuum bagging it to the deck at a later date. You rough up all of the surfaces with 80 grit, acetone wash, bonded with thickened epoxy

The teak is not sealed in any way on the deck side. Fully exposed to the elements. Teak is used on decks of boats because it has nonskid properties.

Thanks Adam. That's good information - and good news since I've gone ahead and bought some West G/flex 650.
The plan is to get stuck into a first build in a couple of weeks.
Bonding 1/4in teak decking to fibreglass is certainly is a demanding outdoor application for the epoxy.
A quick skim just now suggests that its been around for a while and that a variety of adhesives including purpose formulated flexible epoxies are used.
There's at least two pieces on the topic in West's EpoxyWorks mag - they in one place (maybe a later piece?) suggest using the more flexible G/flex product which make sense given the difference in movement between the wood and the substrate.
One factor which possibly helps these joints to stay put/makes the application not entirely generalisable is perhaps the use of 1/4in teak - it's as in the case of veneers presumably not thick/strong enough to overstress the joint.

Ian,

The Gougeon Brothers started doing bond down decks in the mid 1980’s(40 years). The vast majority of them would have been done with the West 105. It’s described throughly in their Boat construction manual. It’s now a free pdf online.

G/Flex is just over 10 years old now. It’s a great product. I’ve used a few of the small bottle kits over the years. It does stick pretty good to anything and remains flexible enough that it doesn’t crack.

The theory’s we have of things moving & breaking are sometimes valid. That said epoxy by its superior adhesive & the surface area can overcome those movements. Carbon fibre(zero movement) can be used with wood if you know what is possible.

Keep in mind that typical wood flooring in a house is defying the wood rules.

I’m an architectural woodworker/boatbuilder.

The devil for sure (and also the solution) is often in the detail.
: )
I spent some years early in my career in adhesives R&D, and saw many times how on the one hand it can very difficult to predict theoretically what will work.
Sometimes the physics subtly align in what looks a bit unlikely.
Sometimes the opposite.
How on the other hand that making proper arrangements for the accomodation of especially differing materials movements (typically thermal in engineering adhesives, but wood movement is a similar deal) is essential if bonds are to be long lived.
There are various ways of getting around the movement issue - a thick bond line (spacers of one sort or another are sometimes used), matching of material properties etc - but what is clear is that when there is movement it often takes more than just a physically strong, environmentally resistant and very sticky adhesive to get the job done.
Adhesives can also sometimes 'come unstuck' (joke) for very obscure reasons - when for example something subtle and not very obvious changes in the processing of the substrates, or in the post bond environment.
Accelerated aging/testing doesn't always accurately predict long term durability either - although commercial pressures mean that manufacturers tend often to talk as though it does.
Many engineers have learned the hard way that mechanical fasteners are genrally more predictable, but there are of course applications where adhesives offer unique advantages...

In terms of wood adhesives. PVA glues like tight glue lines. You have to clamp a joint very hard to squeeze it out. It doesn't matter if its straight off of the jointer or planer with sharp knives.

Epoxy likes thick glue joints. The reason the industry settled on 80 grit(which is not a commonly used grit) is it automatically produces a thick glue line regardless of clamping pressure. It's not simply the more surface area theory.

The teak boat deck is one of those things that you wouldn't consider until somebody else tried it.

Very few people know the Gougeon Brothers(West System Epoxy) were the largest builder of large wind turbine blades from the 1980 into the 1990's. Those were highly engineered from wood/plywood/fibreglass/epoxy.

They have an engineering/fabrication department whereas the other epoxy companies have sales/tech departments.

Thanks Adam.
The low clamping pressure requirement and gap tolerance as you say are reasons why epoxy in this specific regard (large area laminations) may bring advantages - and why PVA may have some downsides.
The high strength and stiffness of the former (especially standard more rigid epoxies) may however bring other longer term risks into play - hence this thread.
Clamping pressures of 200 - 250psi get quoted as being ideal when using PVA on hardwoods.
This subject to geometry issues (e.g. a mortise and tenon is largely dependent on an interference fit and swelling of the wood to generate clamping pressure in the important long grain parts of the joint) is not a problem in the case of typical e.g. furniture joints - it's even easy to overdo the pressure.
Face joints between wider boards (especially those that don't have the back up of mechanical/joint interlocking) are however a different matter - it's easy to underestimate the effects of the square law. (even 1psi for example generates 144lbs force when applied to a 12in square)
Take 4in wide boards clamped face to face at generous 12in centres using parallel clamps.
12x4x250psi = 12,000lbs force (9,600 at 200psi) Your average parallel clamp requires some heavy duty twisting to generate more than 1,000lbs force - yet that's in this case only 8-10% of the recommended clamping pressure.
Not only that - these are average /sq in. numbers - the pressure may be much higher right under the clamp pad, but will fall very rapidly while moving away from it.
Widen the boards to 5in and the average figures drop to 6-8%.
Now lots use PVA successfully in this sort of situation (and the large bond areas must help with this by reducing the stress on the adhesive) - but it may also be the case that it's at least part of the reason why you hear the odd report of creep and localised problems occurring. (e.g. splits at the ends of laminations)
Most will NOT report a problem in their pride and joy unless they really are under pressure to find a fix to the problem....
Gougeon Bros by the way do indeed to my mind have a refreshingly practical and hands on approach to technical matters - in an industry which in general has been known to oversell its' products.

I am reminded of the saying "Paralysis by analysis." and dealt with it myself many times during a long engineering career. You will never have perfect or enough data to cover all scenarios. You make the best decision you can with the data you have at the time. And you move forward.

John

There's always a balance John, and many error in the other direction.
That said I've as above committed and bought the G/flex having concluded that while the indications were positive there wasn't really solid information about...

I didn't say don't do your homework, and I'm glad you've done yours, gathered as much input as you could, made a decision and are moving forward. Best of luck.

John

I just don't see the likelihood of such extreme wood movement as to require using G-Flex over pva in this application assuming properly dried and machined lumber. If I were in doubt as you are I would do some short samples with different adhesives and expose them to moisture cycling over some months' time before going into production. The short length would allow for relatively rapid mc changes in a large cross section of wood.

I certainly wouldn’t lose any sleep over it. My example of the boat decks defy most of the principles of wood construction. It shouldn’t work. But, it does and it’s a brilliant solution to a very old boat building problem.

I agree with the others, that you're overthinking it.

But, if you want the ultimate peace of mind....what about no adhesive?

Drawbored mortise and tenons have stood the test of time.