Repair Split Legs, how?

[Bill McDermott]

This bench is made of White Oak. It got the old steel wool and vinegar treatment in honor of our Black Lab. While Rocky was a very robust dog that gave us over 16 years of sturdy companionship, my memorial bench is crippled after only a year. Now about the repair... I could use some advise.

The leg blanks are glued up to give the thickness I wanted. Plenty of Titebond III, generous application and full clamping. I had some scraps in the cutoff bins and tested them just now. The bonds broke in the wood, not at the glueline. Of course, the bench has lived outside in northern Illinois for a full cycle of seasons. While the scraps have been in the shop.

FYI, I sealed up the bottom of the legs with epoxy. There is some small checking down there, but nothing one would not expect or easily tolerate. Both legs on one side split. Neither leg on the other side shows any sign of splitting.

Here's what I think went wrong. The legs that split face West. They get bombed with full sun for several hours a day. The east side legs are shaded by the bench itself. I think the dark color soaks in the solar heat and loosened up the glue. But I am curious if you have another idea - or a similar experience.

So, I have two questions for the creek.

1) What went wrong?

2) What's the right way to repair it?

Thanks very much.

Bill

[Jim Matthews]

A clear split, the length of the glue line, and no warp in either half?

They still look dead straight.

Is there a housed tenon at the top?
I'm guessing here, but I don't think it's the legs
that moved in the heat - I suppose it's the stretcher.

Is disassembly possible, or are you looking for a fix without taking things apart?

It's a handsome piece of work, particularly in the stretcher and lower rung details.

[Bill McDermott]

Thanks Jim. I don't mind taking it apart, although that will be an adventure. That's where my head is right now. Open it, plane it and glue it with something (TBD) that is not reactive to heat. ...and add some screws on the inside for safe keeping.
Yes, there is a stretcher tenon right in line with that opening.
The half of the leg the opened only moved on one side. So it did warp, or perhaps I should say it cupped. The outside of the leg is still glued tight. On both legs it was only the inside glue joint that opened. So your focus on the tenon is valid. Do you think that 3/8" thick tenon really had enough in it to open that whole glue line?

[Jim Matthews]

...and add some screws on the inside for safe keeping.
That was my first thought to remediate the problem. If you still have similar material, plugs could hide the repair.


Do you think that 3/8" thick tenon really had enough in it to open that whole glue line?

I make legs the same way, for outdoor furniture - crossgrain laminations.
I just used Titebond III and none have failed in four years.

We get big swings in temperature and humidity, although not so cold as Chicago land.
(Go Hawks, GO!)

I don't use fully housed mortises to attach the long rails, only on the shorter rungs.
The long rails are "dropped" into a bridle joint, because it's easier to align on the fly.

While it is possible that the cupping force was enough to separate the two halves,
it looks like it has opened more at the top than the bottom, like splitting a log with a wedge.

Kudos on your design, you built it better than it has to be.

[Patrick Harper]

It's hard to tell from the pictures, but did a glue joint fail or is it splitting with the grain?

[Bill McDermott]

Opened up right down the glue joints on the inside (mortise) of both of the legs that get cooked in the afternoon sun. No wood split at all. It's a clean line where the boards were glued.

[Karl Andersson]

Bill,
Just my observations from doing a lot of outdoor water damage repair on wood structures as well as furniture; I think Jim's theory about the tenons is the main source of the splits, but something made that Titebond give way when is wasn't supposed to. I think water might be a main factor in giving the tenons extra strength when swelling and also possibly weakening the "waterproof" glue - consider that this may have happened in the swampy summer, or it could have happened during a freeze after the joint was soaked (by ice melting during the day in a western exposure). Having the seat planking gap right above the M&T joint might be allowing the wood at the joint to get soaked in certain weather situations, swelling both the inside of the mortise, the leg and the tenon - then the sun orientation made whatever difference was needed for the glue to fail.

Either way, I'd take steps to protect the joint "opening" from water before you repair the legs; even if it wasn't the main cause of the problem, it will keep the bench rot-free longer. If possible, you could put a strip of flashing, maybe just an inch wide across the top of the joint, between the leg/ stretcher and the seat planks just under the seat planking gap, or at least put a waterproof epoxy or caulk layer in the gap down against the top of the joint, sloped so it doesn't just hold water there. This is similar to the practice of putting a 2-inch wide strip of tar paper on the top of all outdoor deck joists before nailing on the decking - it keeps water from sitting in the joint and soaking the wood.

Water, especially in a temperate climate, is a constant source of structural attacks -not to mention temperature and UV rays. Something you may want to try: put one of your scrap glue-ups in the freezer overnight, then try to split it when it's still frozen- see if the glue breaks then. Or maybe the cold theory had nothing to do with it if you took the bench in for the winter... but I think water did.
good luck,
Karl

[Bill McDermott]

Karl, Good thinking. So obvious now that I regret not putting a roof over the joinery from the get go. Thanks.

[ian maybury]

Hi Bill. It's easy for guys like me to be wise after the fact (this stuff is always a balancing act), but expansion as a result of moisture can by all accounts apply very high forces - especially when there's a fairly large area involved like possibly here.

Even without the weakening effect of the moisture on the adhesive - stressing a glue joint at one end in a way that tries to peel the pieces apart (like driving a wedge into one end of the joint) tends to concentrate most of the stress at the very end of the glue line. Which when it fails leads to the stress being applied to the next bit on, which… It zips the joint apart. This is a classic premature failure mode with glued joints, it's normal to design to avoid the situation arising.

Peeling happens when there is flexibility in the substrate (if the glued pieces were both thick and rigid the load would be spread over the full area of the joint) - it's likely that the heavy waisting of the leg just below the tenon played quite a large part in this.

It seems that Titebond doesn't stick well over itself, so disassembly and replaning the joint surfaces is probably a good way to go. It's not supposed to be highly waterproof either - maybe the matt finish left it more exposed than might a gloss. Perhaps use a polyurethane or resorcinol adhesive if it's staying outside?

I'm not sure about the tenon in the top of the leg - it seems likely to cause ongoing problems. It might seem a bit of a cop out, but might it be possible to get rid of the through tenons and link the stretchers behind the leg using a heavy duty version of a metal (stainless?) bracket like this instead: http://videos.osbornewood.com/table-construction-attaching-a-table-leg-using-a-metal-corner-bracket Or go a glued and screwed wooden variant of it along these lines? http://lumberjocks.com/projects/80522

[Pat Barry]

I'm not seeing the evidence that the stretcher caused this split so I think this is unlikely. It looks more like a simple glue joint failure. Maybe because not enough glue, maybe because the wood wasn't amenable to gluing due to either prep or moisture content, maybe too much clamp pressure, etc. I know its a lot of work but I would recommend trying to take those legs apart, re-surface the faces needing glue and then re glue with Titebond III.