Troubleshooting Glue Line Swelling in Furniture

In this long and complicated thread, woodworkers and experts try to figure out what's causing a slight ridge to form at the glue line in a wood chair lamination. July 29, 2014

Question
I'm having a small issue with a glue line on some chairs I make. The chairs are solid walnut and I make the backs from two pieces of 8/4 stock rather than cut the backs out of 16/4 stock. Where the two pieces meet along the top edge, a slight ridge sometimes forms long after sanding and finishing where one piece seems to have shrunk or expanded. How can I control this?

I've been using Titebond III. I like it because it's non-toxic and I can use it in my cold shop (it gets into the mid 40’s at night occasionally). I thought of using plastic resin glue, but I do not like the long clamp time, temperature requirements, mess and toxicity. Is there another rigid glue I might want to try? Would a more rigid glue even help? The other option I've considered was to make the backs, set them aside for a few months, then sand them, assuming the problem was just the moisture from the glue and time would resolve this.


Click here for higher quality, full size image

Forum Responses
(Adhesive Forum)
From Contributor G:
Unless you let it dry for around three days you will get the shrinkage. What happens is when you apply a water-based glue the wood swells. Then the glue dries and you cut and shape the piece, but the wood near the area of the glue joint is still swelled from the water in the glue. After all the moisture comes out of the wood the cells shrink back and you get that little sunken area. You need to wait a few days to let the water come out naturally before you cut into the wood. One nice thing about the other glues is no water, so no swelling. The longer glue times compared to waiting the three days can be worth it in the long run.



From the original questioner:
The chairs have several days or weeks to dry before final sanding, that's why I'm entertaining letting them dry for a couple months. The Dap Plastic Resin Glue that's readily available uses water anyway, so I'm not sure that would solve my problems. I'm open to trying a two-part resin glue or an epoxy if the clamp times aren't too long. Would a polyurethane glue minimize shrinkage? I'll do some tests, of course, but the tests take months and I want to know what to try. I'd love to hear some brands and sources for any slightly hard-to find products.


Click here for higher quality, full size image



From Contributor G:
Have you looked into UniBond 800? Not sure what the clamp times are like.


From Contributor O:
I have found that the yellow glues and their cousins are prone to doing this. In my case, I first saw it 20 years ago on stave glued turnings that were nicely sanded and then the next day the glue lines were all raised. I sanded them down- easy enough. The next day they were back. Sand, return - like a 3 Stooges skit. Using a rigid glue-line glue like plastic resin or Unibond will prevent the problem. As I understand it, the yellow glue is always moving (never rigid) and creeps out of the joint. I have heard the wood swelling and shrinking reasoning, but my experience is a little different.


From the original questioner
I'm going to try Unibond and polyurethane glue and see if that helps. I suspect the issue is that the glue line is cut at an angle, so any swelling or shrinking is magnified. I originally used 16/4 walnut, but figured out each back would be around $45 plus versus $11 in 8/4 (due party to cost, but mostly to waste). Hopefully, one of these solutions works and I won't need to raise the price of my chairs.


From Contributor G:
Can you incorporate a design aspect at the glue line to conceal it? Maybe rout a small groove at the two glue lines and keep the line at the edge of the groove. That could eliminate a bunch of testing and headaches.


From the original questioner
Thanks for the idea Contributor G but it will not work for the look I want.


From Contributor G:
I figured as much. The thought popped in my head and it is an easy thing to do. Good luck with the UB 800. I wouldn't trust the poly glue.


From Contributor O:
I agree with Contributor G, I would not trust the poly glue. Uni-bond is a bit more expensive than plastic resin glue (Weldwood), harder to get, and a little harder to mix up. As I understand it, they are the same basic glue, just two ways to get there.


From Jeff Pitcher, forum technical advisor:
Just to clarify a few things: Unibond 800, Dap Weldwood and all PVA glues are all waterbased adhesives. Reactive Polyurethanes do not contain water. Both the Unibond and the Weldwood are thermosetting resins which do not creep under load. PVA will allow creep if the joint is under stress as it might be if holding together bent laminations. It will not creep if not under load. I would suggest checking the moisture content of the wood you are using. It should be 7-10 %. My guess is that it is higher and that is likely causing your problems. Changing to a polyurethane might help your problem, changing to a UF resin will not fix it.


From Contributor B:
Are you using a meter to verify that the moisture level of your walnut stock is low enough for working? You also mentioned a cool shop; is the humidity in the shop elevated? I would look carefully at these things as the source of your difficulties.


From the original questioner
I had the impression that Unibond didn't have water, but Weldwood does. What's the big issue with polyurethane that I might have for interior furniture? The wood is dry and from the same boards, so I'm not concerned about that.


From Contributor G:
There’s no water in UniBond 800.


From Jeff Pitcher, forum technical advisor:

Contributor G I assure you Unibond 800 contains about 40% water. All urea formaldehyde resins are water based. I'm sure you could contact the manufacturer directly for verification.


From Contributor G:
I was looking at a lot of different veneering sites about PVA vs UB 800. It seems that they like to use the UB 800 because it doesn't have water issues. I also found this: “Unibond 800 is a two part urea resin glue that offers excellent adhesion and stability. The real benefit is that you are not introducing water to the veneer and substrate which can cause seams to separate, thin substrates to warp, and some veneers to develop moldy areas. Cleanup can be difficult and a gallon of this adhesive is a bit expensive, but for that ultra-special veneer, this might be your best option. This adhesive is also good for adhering oily veneers such as rosewood. Unibond requires four to six hours in the vacuum press to bond.”


From Contributor O:
Jeff - I'll go out on a limb here and respectfully disagree with your take on moisture content. I have quit using yellow glues in some applications for the reason as stated by the original poster. Creep may or may not be the correct term. I have asked the question repeatedly, and get a lot of tech talk back, but the glue still expands out of the joint over time - usually overnight or a few days. I can slice it off with a sharp chisel - not every joint, not every time as it is not predictable - at least in my attempts to reproduce the problem. It is only yellow glue, and it only expands, never shrinks. The wood is well within the proper MC.

I first noticed this and still see it most frequently on turnings. I then theorized that the heat of sanding caused the glue to expand - the thermo in thermo-plastic. After all sanding the line will return and then can be sliced off or hand sanded (no heat) and then it comes back. This can go one for four-five cycles before it stops, or stops long enough to get the stuff out of the shop. I don't want to be argumentative, but this has been my experience, my observations.

To the original questioner: Have you removed the line and had it return?



From Contributor G:
I actually have the opposite happen with TB II (yellow PVA). I gut shrinkage of the glue line. I will try to make a bunch of stuff and then do the rough sanding. Then I’ll go back several days later to do the final sanding to prevent the small indent at the glue line.


From the original questioner
My issue isn't the glue expanding or shrinking, but the wood expanding or shrinking. The ridge is because one piece has shrunk or expanded differently than another. I'd like the glue to lock them at the same glue point. I think it has returned before. I noticed it on some chairs that I had finished, sanded them smooth, and then refinished that part. I recently looked at them all again maybe a month later, and noticed one had a ridge again, but I don't recall if I had sanded that one down or not. I ought to sand that one down, mark it, and see what it looks like in the summer. Titebond III does seem a little more rigid than yellow glues, but that's based only on the dried bits on glue bottles. The other issue I had was the oil finish I was using wasn't having the same sheen as the rest of the chair around the glue joint, like the glue had spread while sanding about 1/4" away from the glue line, so hopefully another glue will solve this issue too (the top gets trimmed off after the gluing, so it's not an issue of a messy glue up).


From contributor D:
I suggest you try Structan, a polyurethane from Excel adhesives. It contains no water and thus you introduce no moisture to your gluing surfaces. It cures to an extremely tough consistency, unlike the honey colored gorilla glues which cure like styrofoam.


From contributor D:
Water only speeds the cure, nothing else. I've never added it. We've built thousands of custom doors using this adhesive over the past ten years. The joints are mortise and tenon cope and stick, with .007" slop when machined. Some complex doors have over twenty or so parts, and moist yellow glues swelled the tenons making assembly a nightmare. Structan on the other hand actually lubricates the joints and with an increased working time helps everything clamp up easily. I've never looked at their website - we've had many telephone conversations with their reps over the years however, and I assure you if you call them they will tell you water is not needed.


From Gene Wengert, forum technical advisor:
If the two pieces are moving differently this is not a problem caused by the adhesive, but a problem caused by different moisture contents of the wood. The initial information provides the clue. The shop is in the 40s, which means that the relative humidity will be quite high (compared to what the home owner will have or a heated room of any sort will have). I suspect the RH is above 60% RH in the shop (11% EMC) while a typical home or heated space will be 6 to 7% EMC. So, we know that there will be substantial moisture change and wood movement. If the grain of the two pieces is different, there will be different movement. Further, if the initial MC of the pieces is different, there will again be different movement. There are two steps involved to correct the problem. First, check the incoming MC with a pin meter to assure the average and also the core MC are under 8% MC. Then, keep the shop heated to 70 F so that the RH in the shop and in subsequent life of the chair will not change much. Temperature is not an issue (except for the adhesive's curing); for size changes, it is all about humidity and resultant moisture changes.


From Gene Wengert, forum technical advisor:
Regarding the comment about a sunken glue joint, the moisture in the glue (including 800) will cause local swelling of the wood at the glue line. Sand the joint smooth soon after gluing and then as the wood dries, the joint will shrink. For this reason, we wait a few days (while the joint is exposed to warm temperatures) for the swelling to go down before sanding. Check the archives here for more discussion of this common event. Also, note that with wood that is over 8% MC when joined, that is then exposed to a drier environment (such as the interior of a heated room), the wood will shrink but the rigid joint will not and this creates a raised joint. It is more common with the very rigid adhesives. Again, correct initial MC of the wood (6 to 7% MC for wintertime) is the key in most cases.


From contributor K:

I would suggest that you are wasting over 30 dollars per part fighting the glue issue. Prepping the parts for glue takes time, glue costs money, and clamp time is wasted time which equals money. All of this time and expense needs to be added to the part. You can't just say the wood is lower priced. It may actually be less expensive to use the thicker blank.


From the original questioner
Contributor D - I'll try the Structan too.

Gene - each back is cut from a single piece of walnut that is 6"x18"x1.75". Basically, I cut a 24" radius on the band saw 6" high , then glue the flat parts together, so the pieces are the exact same moisture content (my pin meter is reading about 3%, but I suspect I'm not pushing it deep enough. I need to buy a pinless). This leads me to believe that it's the moisture in the glue causing the issue. Heating the space isn't an option. My shop is in San Francisco so it never gets too hot or too cold to work, so we never installed a heater.

Contributor K - I constantly wonder which is the best option: gluing or buying thicker wood. Thirty dollars difference is really a best case scenario and gluing up parts isn't very time consuming if this issue is resolvable. My supplier has 8/4 walnut for $6.75 pbf, while the 16/4 is $11.75. With the method I use, I get maybe 70% yield, while I would get about 30% yield on the 16/4. What if I cut open a 16/4 board and hit a pith the whole way or there's some other flaw (cutting curves is sort of like opening up a geode: you never really know what you're going to get until it's open and the pattern on the back is an important detail)? That gets really expensive really quick. If I can make it work, I would prefer to use the 8/4. I hate to create so much pricey firewood. I also like that the entire chair can be made from 8/4 so that I can pick through the stack and use the best, clear, heartwood pieces for the backs, the knotty parts for the legs (because I can cut around the knots), and the sapwood parts for the seat (because I can face the light parts inward). Perhaps If I found a use for the waste, like in another chair, I would change my mind.

One more detail on the story, it only seems to happen after the oil finish is applied. I've had some cut parts sitting around for a few weeks or more, and they haven't formed any ridges (I ran my fingers along it, so this is not just a visual distinction).



From Gene Wengert, forum technical advisor:
I had a client in SF that had 10% mc in his shop and the customer's nearby home. Pin meters of any brand cannot read under 6.5% mc, some even not that low.