Bowing Problems with RF Glued Door Stiles
I do not believe this is the cause of our problems, but it keeps getting pointing out to me that while our problems did not start when we switched, it's the only common factor amongst all the jobs. We are located in the Midwest and go through huge swings in our shop related to humidity between summer and fall, which as I pointed out is when the problem doors were made. We strive to measure moisture content when the lumber comes into our shop and reject any loads that are not acceptable. Once the lumber comes in, however, we do struggle to keep the humidity in the building at acceptable levels, at least to me.
So my question is this. I know doors bow because the moisture content of the wood changes, basic woodworking 101. Could there be any correlation between the glue we are now using and the bowed doors? Around half of these doors were finished in house and left without being bowed. They are all finished all six sides. The vast majority of the doors since we changed glues are still flat. Some are still in our plant. The jobs are a variety of different woods, so I know we did not get in one bad load of lumber. Is there any possible chance the glue we are using is allowing the door stiles to bow? I hope some of this makes sense, as I am completely stumped and am looking for any insight as to what may be our problem.
From contributor D:
Check your rf operator too. Often if you glue different lengths of stiles, a lazy or uninformed employee may use the same setup, which may cause bowing. For example, if the pressure blocks point of pressure is directly on the ends of one length of stile, but on another is 10" offset from the edge, even with full length backers, the point of pressure will be different, and the stile will be glued up with stresses.
Another example is a stack of ten 98" stiles being glued up with two 82" stiles in the mix. If the shorts are on the outside, there will just be a reduction in pressure on the ends of the longs stiles, but if they are mixed in, the ends of the long stiles will have been glued up with stresses. Your backer blocks need to be calibrated often too. Don't just mic each one, stack a bunch of full length backers together and measure how thick they are on one end versus the other. A 1/32" difference in thickness can translate into introducing a 1/4" bow if you stack eight together. They get subjected to tremendous clamping pressure all the time and will eventually compress. If any single stile comes off the press not dead straight there is something wrong somewhere, and stiles are likely glued up with stresses. You can verify this by pulling a few from a suspect batch and leaving them out in the weather. They will bow.
I can only imagine your problems, as we had a few bowed stiles and they were even engineered stave core. The problem was in gluing operations, not the glue. We use Titebond 2 from 55 gal drums and do not add catalyst, having switched from various brands of two part glues with success, mainly I think because you can eliminate improper mix ratio or how thoroughly mixed the glue is as a variable.
From the original questioner:
Gene - hopefully this isn't ignorance on my part, but how would I go about measuring the water in the adhesive? If this is the problem, would we see this within a short time frame show up in bowing stiles? We also glue up panel stock in the RF and do not see any raised glue lines after we plane down our panel stock.
Contributor D - good info, as I'm just starting to really examine our RF operations. We typically glue up seven stiles at a time, sandwiched between two (I'm at home so I'm guessing at this one) 16" glued up blocks, think as if they're laminated beams. As far as I can tell, we shouldn't have any build up of tolerances. I randomly check our stiles as they come off the RF and they are straight as can be.
To add a little additional info I didn't think of yesterday. We stock some of our better selling doors. If, and it's a big one, we have a problem with our glue, shouldn't we see bowed doors before they leave our facility? These are whole jobs we are seeing warp, not something random, and definitely not all from the same batch of glue.
One final thing - it just occurred to me, and I'll look at it tomorrow. What would happen if we were not to look at growth rings before we glued up the stiles? I've read various arguments for and against this. How long would it take for bowing to show up if we're not putting opposing faces together? Some of these doors sat for a week before being finished, some five weeks, and one job sat in our shop for almost three months. The doors were all straight when they left. The more I look into this, the more stumped I become.
From contributor E:
I think any time you use a two ply lamination you're setting yourself up for trouble. With a three ply lamination you're at least keeping the stile balanced and that will help with the moisture swings after they leave your shop. I would be very interested to hear Gene’s comment on that. I do understand that it may not be practical for you to switch to a three ply lamination, especially if you've been using a two ply all these years, but I thought I would just throw it out there for consideration.
From Gene Wengert, forum technical advisor:
It does take a moisture change for pieces to warp, one-ply, two-ply or three-ply. I do think, like Contributor E, that thee-ply is more likely to stay flat. But you need to take all efforts to get the correct MC in the lumber and the same EMC in your plant. Both of these must match the EMC in use (or final MC). With very little or no moisture change, you will not have any problems. So, from this perspective, if you are having problems now, I suspect the wood is wetter and the source of water could be the new adhesive. Ok? The provider of the adhesive or a label on the product should provide the solid content and the liquid content. It is my impression that most (or maybe all) two part have less moisture.
From contributor D:
You probably don't have the same problem we had then, if every stile is dead straight. It may be the adhesive is the culprit. On many of the barrels we get there are acceptable temperature ranges they can be stored at, and during the winter we refused some barrels that arrived on trucks with no temperature control and it was freezing cold outside. A little detective work can reveal whether or not grain orientation is the culprit. A bowed stile with faces not opposed is almost the same as one solid piece. A telltale sign I've seen with PVA glue of all types is if it's been frozen or it's too cold, it will cure to a caulkier, duller consistency, instead of a harder, glossier, plastic consistency.
From the original questioner:
Thanks to all for the responses so far. It's definitely given me some things to look at. The solids content of our adhesive is 50.5%. Our old glue is listed at 50-55%. They are both crosslinking PVA’s. I'm still stumped as to why I can look at a fairly large amount of doors we stock in our facility and they are dead flat, then we finish some of them and they are still flat, then after we ship them they bow.
I fully understand a two piece stile is not the best construction method from a performance standpoint. For better or worse, we sell, and our market demands, a "solid" wood door. I'm slowly working towards an engineered stile that our sales staff and customers will accept, but were stuck with a two piece stile for now. A three piece stile for us would have a glue line running down the sticking, which I think, as do our customers, is completely unacceptable. Once again, thanks to all who have contributed.
From contributor F:
I love these challenges. Maybe there was a swing in the humidity during the twelve hours after the glue up or a type of humidity shift that happens during the evening and night hours. See what I'm looking at? If this glue has a secondary slight shift taking place during the final 50% of curing say a 24 hour period. Daily humidity changes could explain the bowing or warping. Also we have to be picky about things sitting on a flat surface when only one side is open to easy ambient conditions and changes. I've re ripped and re-glued several times over the years when someone did not maintain even conditions on items prior to sealing and finishing.
From contributor O:
I'm not sure if this is of any help but sometime ago, a customer who was using one of our machines to laminate softwood sections experienced a problem with distortion/bowing. It occurred really infrequently and after some head scratching, his conclusion was this: The RF cured the glue line and the laminated section was stacked on the floor. As the residual heat dissipated out through the timber, the concrete floor acted like a heat sink and cooled one side significantly enough for the moisture to pool on this cool side. This, together with the warmer side, made it stretch one side and contract the other - hence the bow. The reason as to why it was so infrequent was it was always the bottom one of the stack affected. His solution was to stack them with the glue line perpendicular to the floor - not parallel. I assumed it work as we never heard anything again.
Would you like to add information to this article?
Interested in writing or submitting an article?
Have a question about this article?
Have you reviewed the related Knowledge Base areas below?