|Home » Forums » Architectural Woodworking » Message||Login|
You are not logged in. Consider these WOODWEB Member advantages:
warped slab door9/9
We made four exterior veneered slab doors last year for a residential job in northern New York, two of which are problematic. All are constructed and finished the same, all are approximately 2 1/4"x 36"x 100". 1 1/4" foam core (Klegecell or Divinycell structural foam), 1/2" Extira skins, solid Honduras Mahogany edgebands and blocking, sliced Mahogany faces with vertical grain direction, 6 coats of Epifanes varnish, laid up with West System epoxy. All secured with 3 point hardware with Tectus hinges. The pair were prehung in the shop and the other two hung in the field due to builder's schedule requiring those jambs in place before the doors were ready. All were flat in the shop, and as installed in late winter. The main entry pair have remained flat. One singleton bowed slightly over its length towards the inside soon after installation enough that the upper bolt of the three point latch would not engage (5/32" proud of the jamb). We ground the tip of the bolt back from center to engage the strike sooner, and it does work now, though the weatherstripping is not fully contacting at that corner unless the top and bottom bolts are engaged. The fourth door stayed flat through the heating season, then bowed more drastically concave toward the interior(over 3/16" proud of the jamb at top and bottom). We are faced with replacing this door, but are not sure what course to take. None of the doors have cupped across their width. The doors are well protected from the weather with large overhangs. The main entry and the badly warped door are in the main house which is climate controlled, the fourth door is in the garage which is heated but probably not air conditioned. We have made several similarly constructed doors in the past without problems. We have considered using thicker veneer and thinner skins (Extira doesn't seem to be available in less than 1/2" thickness). We don't understand why two doors have remained stable and two not. Comments/ suggestions welcome.
Did you vac bag the whole layup?
One problem I might be the extira. It came to you non-flat like all sheet goods. Each sheet has memory. If you happen to put each sheet bowing in the same direction it might cause your problem. The stuff is also like a sponge. One side could easily soak up more resin than the other.
The marine coring products are designed to conform to non flat surfaces so they provide no useful flatness.
How much solid mahogany blocking did you use? I would be inclined to use at least 1 1/4" x 2 1/2" internal stiles.
They tell medical students, "When you hear hoofbeats, think horse not zebra." There might be some super-subtle difference between the warped and unwarped doors, like a sheet of Extira with a bad memory, but we should rule out big differences first.
The two doors that remained flat were prehung (presumably before finishing) in your shop. The two that warped were hung in the field. Did they get trimmed in the field? Have you inspected the warped doors to see whether all cuts, holes, etc... were filled and finished?
What else was different about these doors? Exposure? Overhangs? Are the jambs good (did the doors have to get tortured to fit in the jambs)?
Dr. Jenness to Forensics, please! ;8>)
The blanks were pressed in a vacuum press, the Extira was reasonably flat, the banding was approximately 1 1/2"x 2 1/4". The blanks came out of the press flat and were flat as installed. The doors were not trimmed in the field. The jamb of the worst door was set slightly in wind, but not severely so-perhaps 1/8" difference in plumb between the two legs. That door is not twisted, but really bowed along its latch edge. I believe all the penetrations were sealed, but I will check that. The conditions are essentially the same for all doors, with the exception of the garage door which may not have AC inside. I have speculated that the entry doors may have fared better because they are used infrequently and are typically latched at 3 points on the latch edge, while the other two are less often secured with the top and bottom bolts. What baffles me is the variation in behavior amongst the three doors with basically the same conditions, and the fact that the worst one stayed flat through the heating season and only started to bow in the spring. I wish we had data on the interior and exterior RH through the seasons. Thanks for the responses.
You have more to do on the forensic side to be sure, but my first thought is what the unfinished doors on site experienced before and during finishing.
A casual reading of any door warranty obviously puts a lot of responsibility on the site people to store and finish any wood door properly and promptly. In the real world, this almost never happens. One side gets finished half way and sits for a week, the other side gets done, then back to the first face. Different products in different places, applied different ways. The problem is that this will be difficult to reconstruct since anyone near the doors probably is aware of the problem, and will be defensive.
As for multipoints holding the doors flat and preventing warping, I'd debate that one. If wood wants to move (remember, always in response to changes in MC most often due to changes in RH), it will move. With the latches disengaged, the doors will exhibit their natural shape.
What is troubling is that your construction sounds robust and not marginal or shaky. I don't like the Extira, but that is not a death dealing blow, either. I don't know what the structural foam is. I have always made torsion box type flush doors and laid in the foam with no glue, no force, allowing it to float, so to speak. The torsion boxes, if made flat, stay flat.
The foam is typically used for coring the decks & hulls of fiberglass boats. The increase in hull thickness increases the stiffness quickly. Essentially like deeper beams. The composite panels are dimensionally stable and strong.
I agree that the door will move regardless of latches, etc if it wants to move.
What thickness veneers were used?
The doors were finished in the shop and hung by us' although two of jambs were set by site carpenters. If the penetrations for hinges, etc., were not sealed, that is on us.
The veneers are thin (.023") sliced mahogany. Not what I would have chosen, but I don't run the show.
The question is what to do now. Any recommendations for different type or thickness of components?
Sorry, Kevin, I don't have your answer, but a few things jump out at me. First, you have too much thickness in your skins for a proper build with rigid structural foam: use 1/4" max. Stiffness in this type of construction comes from the tensile strength of the skins, and the shear strength of the bond between the skins and the core: overkill is not always stronger/better. Secondly, vac bagging with epoxy is overly difficult in terms of preventing glue starvation. You'd be better off with resorcinol.
What to do now? I don't have an answer except the one that is easy to say and hard to do - replace the warped doors.
Then do you build the same way or a new improved way? I vote for thinner skins - 1/4" max, and torsion box with long verticals and a few horizontals, with PUR in a vacuum bag with less pressure.
I will say that after 40 years of watching others and trying it myself many times (we can get desperate), over bending, or wetting one side or forcing a twist will not fix the problem for more than a few minutes. Think of the warped door that is forced to close for 20 years - it is forced into flat, but still retains the warp when released. Even after all the years.
Replacement is the plan, we are trying to decide on the rebuild specs. David and Mark, what sheet material in 1/4" thickness would you recommend? We have considered marine ply, though it will probably mean adding a crossband under the vertical face veneer unless we could source ply in an 8'x4' configuration. Extira is not available in less than 1/2" thickness.
I will say that we have laminated many items with epoxy in a vacuum press over the years with no problems, and this project does not seem to be suffering from delamination or glue starvation. I will do a sample with the materials used just to test this possibility, but I will be very surprised if there is a deficient bond between the foam and Extira.
Mark and David, what is the principle behind advocating for thinner skins? And David, why do you think PUR glue would be superior, and what specific product would you recommend?
We use a product called Truespec (essentially an chipboard bonded with a waterproof adhesive). We use it at 1 1/2" thickness. We CNC a ladder core with rigid foam in the openings. 1/8" marine ply on both faces. Solid edging. 1/4" solid front and back faces. Makes a nice 2 1/4" door.
Still I have had one problem with a door that was skinned with western red cedar running across the width. That was a big mistake on my part as the moisture content was too high, resulting in a bowed door. Frankly, I don't think I will ever run skins across the width again. Even a small change in moisture content can lead to problems.
Otherwise, we have no problems on these doors when they are finished properly. A rigid core is key.
What thickness are the face veneers?
I am getting ready to build 2 such slab doors so I'm interested in the outcome.
Thinking the hotpress platens will be flatter than I can ever get my vacume bag to be.
The face veneers were thin(.024") sliced veneers. Again, not my choice for durability, but that's what we used. I would not use sawn veneers thicker than 1/8" on a plywood or similar substrate unless the parts were very narrow.
I like the Powdered Urea Resin for its workability and open time. While it is not pristine, it is less of a mess than epoxy. It must have 70 degrees or better is one drawback. We use the Custom Pak product.
Thin skins help keep the weight down. We have pressed the skins only (1/8" 'door skins' or 1/4" 3 ply) back to back with the face veneers and then once out of the bag, pressed them to the core - a latticework of sticks with foam, lock blocks and edgebands. Nerve-wracking as all hell, but it worked well.
Looked for pictures, but all I found is attached. Similar principles, but all much heavier - 3/4" faces each side and 1-3/4" core frames.