Bending Stair Rail in Three Dimensions

Experienced stair rail professionals weigh in on a deep and complex subject multi-dimensional wood bending in this long, detailed thread. February 25, 2009

I'm building a meandering staircase that scales along an angled wall, around a fireplace and up to the second floor. The handrails are proving to be a fabrication problem. I need to bend them in left/right as well as up/down directions. I've been thinking my best bet is to steam bend a laminated railing out of thin oak strips (say 10 strips approx 3/16" x 3" x 16 feet each) and glue it up.

I'm finding the grain orientation limits how I am able to bend the wood, even with steam. In other words, the laminations will bend just fine to the left-right along the grain, but it bends very little (hardly at all) along the edge grain in the up/down direction. I've been steaming test boards of similar dimensions with little success bending along the edge grain. My steam oven is approx. 180-190 deg F consistently inside, and I leave the boards to cook for about half an hour. It would seem this cannot be done without a little brain picking out there. Mostly, when reading about lamination steam bending or when making bent laminations, there's little mention out there about the grain orientation. I'd be very interested in some ideas and thoughts.

Forum Responses
(Architectural Woodworking Forum)
From contributor H:
Railing as you wish to bend it is not done by bending. Rather it is done in carved sections then joined with carved fittings. You can laminate and use carved fittings to join the pieces, which is the easiest way. Not that it is ever easy. There is far more to it than it would first appear. But you have found that out. You do not need to steam 3/16 strips. Steam bending something a solid that large is problematic .

From contributor H:
I agree with contributor H that you shouldn't need to steam 3/16" strips. Even 1/4" strips should bend at a reasonable radius (down to approx. 12" to 18"). However, if you are going to steam bend then 180 to 190 degrees is not hot enough. I find 195 as the minimum required temp for my steam box in order to make the wood pliable. A few degrees does make a difference.

From the original questioner:
Remember: the problem occurs when trying to bend along the edge grain. Therein lies my problem and the reason for picking your brains. In other words (for example): resaw a 1x4 so you have a board about 3/16" x 3.5" x however" long. Bend this board along the 3.5" face, no problem. Bend it along the 3/16" edge, there's no way you're going to get a 12-18" radius. Carving the sections is of course the standard way. If I can pull it off, I'd love to try and make out of one piece first. "Is not done by" is not yet in my vocabulary.

From contributor H:
I've never heard of anyone bending a board edgewise. If we could steam bend a 3/4" x 3 1/2" board into a 1/2-round casing that is 3 1/2" wide I would be changing our entire construction process.

I don't know if you can do that even with the steamed/compressed wood now available from a few sources (big time megabucks investment for the equipment). I'm not sure why you want to bend in that orientation. Won't it work for you with standard strip laminations? With oak the seams pretty much disappear into the grain and the final glued up product looks like 1/4-sawn wood.

From contributor W:
You could try 1/4 x 1/4 this should bend in any direction no steam.

From contributor S:
How about this new-fangled "compressed wood" that can be bent in every direction? I haven't used it, but I have read about it here on WOODWEB and also seen before and after in a customer's shop.

From contributor K:
I don't know how abrupt your change in pitch occurs, but here is an approach that I used on an elliptical curve that I did about 15 years ago. The inside rail had the greatest change in pitch, and the tightest bend, so I will describe that.

After picking my pattern off of the form, for the vertical axis, and sighting it from the end, there was a gentle sweep up to a pretty hard hump in the middle about one foot higher than a straight line from end to end.

I joined up two panels of solid 4/4 lumber about 2' wide, and over half the needed length. I then cut a long scarf along opposite ends of the two panels to be joined, and then ran a Freud finger joint router bit along those edges, and glued them together with epoxy. I then used my pattern to cut blanks out of that, which I then re-sawed for 3 laminates each. By shuffling the individual plies, they were gaining some support from each one outside of any individual joint. I used epoxy for that glue-up in a vacuum bag.

By sawing the vertical curve out of nested parts from this special glued up blank, there was not a lot of waste, but there was a lot of preparation, and obviously you wouldn't want to rush the bend before the total cure of the glue in the finger joints. I hope you can follow this brief description.

Oh, one last thing. There is a strong tendency for laminates with much sawn in curve to try to roll over and splay like a deck of cards while the bend is being made, if you have not made provisions to counter it.

From contributor D:
The two basic methods of making railings are as mentioned above - "wreathing" as in large blocks of wood shaped to fit the 3-d curves as needed, and bent lamination, typically cold bent. The compressed wood will certainly become a third method.

Wreathing is described in excruciating mathematical detail in Linden Press' "A Treatise on Stairbuilding and Handrailing" by Mowat. Bent lamination is as done in most curved stair shops. All curved, raked rail is actually twisted, or helical, and therefore even as things change pitch or rake, and radius, and rise, the amount of twist varies also.

One way to accommodate strong changes in pitch in either a consistent or non consistent radius is to form up a cold bent lamination of sufficient width to the plan radius, but do the lamination on a helical pitch. This will allow you to mark out the rail's location, and then saw it out from that. This enables a fluid form with tight changes, and no joints to worry about.

From contributor J:
Slice it up, steam bend it, compress it, twist and torture it and finally clamp the snot out of it. Don't forget to try some (deadly) anhydrous ammonia gas in a vacuum chamber (that works too). There are better methods. Some of them are more than a hundred years old and some are just coming off the computer.

From contributor P:
I'm tickled by all your responses. Thank you so much. I had looked into the "bendywood" type of pre-steamed, pre-compressed products. It is about $2000 for a 16' stick, plus freight, plus hydraulic pipe bending equipment so thats a no go. To answer the "why along the edge grain?" question, the railing swerves left and right as well as makes vertical pitch changes.

Good thought on the 1/4" strips. Im wondering if the assembled piece would look funny with all those horizontal and vertical laminations. I believe that would work, though.

Contributor K - you're way beyond me. I don't think I quite follow the method you described, though it sounds most interesting. Contributor J - I like your style.

From contributor K:
The idea was to lay out and join the plys end to end so the grain follows your rail path. You can't just cut end to end miters, because they would break. The finger joint will work, but only if it is slanted to lengthen it, rather than the short miter, and then doing it out of a couple of wide panels is faster and easier than doing one plank at a time.

From contributor O:

I think all the answers are good ones but need to be combined for your situation. I suggest bending your wood for the horizontal parts of the rail seeing whereas it will bend easily even without steam, and then joining the wood and shaping it for your vertical pitch changes; making sure you pick wood that has similar grain so when you get your final shape it is a functional good looking railing. Put down your books and go with the basics. It's just wood.

From contributor J:
Old fashion tangential geometry produces the patterns that are used to cut and square helical handrails from flat boards. The boards or planks themselves are usually not much thicker then the molded rail.

Handrail easings due to pitch or direction changes within the floor-plan or elevation, are usually regular and uniform because they are derived from a single plane. These parts are also first cut-out square from flat boards.

By employing tangent methods, you will be able to use the least amount of material possible and end up with the most graceful possible transitions. You will also have more control over handrail height etc. This is what I've found to be true and you can forget about bending, twisting and building forms. By using both laminated and solid-cut rail (as required) you can have the best of both methods and produce some really good work.

From Gene Wengert, forum technical advisor:
A technical explanation for the difficulty in side bending is that a piece twice as thick requires four times more effort to bend. If it a narrow piece (1/4" by 1", for example) and you try to bend it the hard way, it will tend to buckle. As noted, steam bending must be close to 212 F and oftentimes is done at a slight pressure to increase the temperature while still at 100% RH. For severe bends, wood should be 22% to 26% MC. For gentle bends, you do not need to steam usually.

From contributor N:
As a long time stairbuilder and author of a book on bending, I can't resist offering a couple of comments. First, bending a piece of wood 1/4" by 2 1/2" on edge is very close to impossible. If you're still determined to produce these pieces, you could bend a piece 2 1/2" square, (not easy but not impossible) then band-saw the piece into 1/4" strips.

You could also mill 1/4" pieces as wide and long as you need them, band-saw the contour and you'd have the same curved 1/4" strips in your hand at the end of the day. What you'll discover, though, is that even when you get the blank glued-up, running the handrail profile on a twisty piece is really hard and can be quite dangerous.

From the original questioner:
Those of you who noted a few deg F made a difference were spot-on, of course, among most of the other suggestions collected here. I've successfully bent some more test laminations (1/2"x1/2" and 3/4"x3/4", no bigger) along my tortuous path, and I've also taken note of the helical way a piece bends on its own. I will use that feature to my advantage. If I gang enough small bendable pieces together this might work. Once bent up and glued up I'll attack it with scrapers and sandpaper and guns and knives. I'm thinking of using Liquid Hide Glue and 4000 clamps for this one. Epoxy is a little more difficult to work with/clean up so I'm shying away from it.

From contributor Z:
I have a little to add. It was about 1850 when Thonet figured out that if he used square profiles (rods) instead of flat strips, he could bend in three dimensions instead of two. He patented that about 160 years ago, and it still works just as well. As you said above, maybe a little clean up with knives and guns (like that image - tempted myself - but my angle grinder gets picked up first). Something I find that works for clamping up all those "rods" is surgical tubing. If you bend with glue in the joints the rods slide as they bend. The glue doesn't work well combined with steaming though, so the rods have to be pretty thin and near your target moisture content to glue up that way.

The compressed wood mentioned above is my product. I'm in the exciting position of having the only Compwood press in the Americas. Bendywood was mentioned above, but dismissed as too expensive. Bendywood comes from Italy through a US distributor. They have lots of hand rail experience, and I defer to their experience in this area. I use my compressed wood for architectural fabrications in bent wood and make it available for others to use in their projects. One day, one of you will try it in a hand rail and let me know how it worked out. Actually, I have one customer in the Seattle area building one now and am waiting for pictures.

There is an advantage to compressed wood for this type of thing. You can bend it solid, in any direction without backing straps, to a much smaller radii than with steam, and without using steam - you bend it cold. This allows for bends in multiple directions along the same plank. Attempting to steam and bend a complex bend from solid hardwood presents complications with setting up the backing straps.

There was mention above of using specialized bending equipment with compressed hardwoods. I use them, but for handrails? No way. The radius is not very small, and if you have enough leverage on the piece, it will bend quite easily. If it gets tough, I section it into two pieces to make the bend "four times" as easy to use Gene's formula. Then glue it up later. Not much clean up from one glue line (if you even need to do that sectioning).

The compressed wood is so interesting, it can help with these "challenges" that customers present. I'm not happy either if I'm not scratching my head and reinventing the wheel. My products tend to lead to more head scratching - but the good kind. I look forward to sharing what I have with those creative head scratches out there.

From contributor B:
Can you bend a 1x4 into a half circle? That is, you would see the 3 1/2" face up if the half circle was set on the bench after bending. This is the orientation that would be required for bending a half circle casing.

From contributor Z:
The question is - can 1 x 4 Compwood be bent into a half circle the hard way? The short answer is yes. But it isn't what I would do for a number of reasons. The first answer exists in a post above - the wood will want to twist and lay flat. The second is that the radius will easily exceed what can be accomplished with steam - but still doesn't rule out Compressed Hardwoods. Wood bends best in square profiles or flat rectangles.

Rectangles on edge would need some special support for the sides and bending something 4" thick will take a lot of leverage. This would necessitate a long overhang, made longer by attaching it to something else (begin to picture the difficulty of adding a 10' long plank secured to each end to increase your leverage and it still might not be enough).

Here is the long answer. Lets say this is a face molding for a 36" half round window. Inside diameter = 36", Outside diameter = 44". The plank will be Pi*radius = 3.14 x 20" (I'm taking the midpoint of the plank) ~ 63" long. To bend it, the plank must stretch to 22*Pi ~ 69" on the outside of the curve, and compress on the inside of the curve to 18*Pi ~ 57". So the plank after bending is 57" long on the inside and 69" long on the outside. That is a 6" stretch on the outside of the curve and a 6" compression on the inside of the curve. Or the plank on the outside is 12" longer than the plank on the inside. This is easier to picture if you were laminating 3/16" veneers and the outside veneer is 12" longer than the inside veneer. Yikes! Never-the-less, this does not exceed the maximum bend that Compressed Hardwoods can achieve (a 10% stretch on the outside and a 10% compression on the inside). So you could do it, but, it can be done so much easier if you just saw it into 2 or 3 pieces, bend them, and glue them back together after fixing the shape by drying the wood. In this way you can cold bend it by hand without thinking about leverage.

Additionally, there is an interesting thing that can be accomplished in moldings by bending squares and gentle rectangles rather than a 1x4 (or 3/4 x 3 1/2) on edge. Take a look at the image included. I've built up a profile with several pieces of ~1x1 +/- to create a stepped profile. It looks like custom crown molding, but it is bent from solid cherry, and is certainly not paint grade.

By the way, I mention above that Compressed Hardwoods can stretch by up to about 10%. This is the "secret" of why it works for "Extreme Wood Bending". Compwood is pre-compressed by about 20%. It settles back while cooling to a net ~10% length loss. And when you bend it, it'll stretch that 10% on the outside of the curve and compress a further 10% (back to the point where it was fully compressed in my press) on the inside of the curve through the bellows action that was created in pre-compression. No backing strap or additional steam is required to do this. And because a backing strap isn't needed, you can bend it any which way, all along the length of the plank. This is where millwork is not quite so interesting for Compressed Hardwoods. They are usually two dimension bends. When you get into three dimension bends, you are able to take advantage of a very unique capability with Compressed Hardwoods.

Click here for full size image

From contributor K:
Contributor Z, you did an excellent job of describing your process. I still have a few questions though. When you are bending the individual strips, did you make forms for each radii, or is each one just laid over the previous? Could you use a glue which works on moist wood, and just brad them for alignment, then clamp for the final drying?

From the original questioner:
Any thoughts out there on the adhesive question? I'm thinking about using Titebond Liquid Hide Glue due to its open time, cleanup ease, and stainability. Shying away from epoxy mostly due to the cleanup factor and I'm not going to dye the glue ahead of time. Plastic resin glue leaves behind brown/amber glue lines and I'd like the assembled lamination to be as invisible as possible.

The comments below were added after this Forum discussion was archived as a Knowledge Base article (add your comment).

Comment from contributor A:
You should be able to obtain the change in elevation pitch and still use something like a 3/16 to 3/8 x 3" board by doing a twist and bend after steaming. The only steaming I've done so far has been experimenting using a microwave, but have been able to bend a 3/8" x 6" piece of ponderosa pine to a 10" diameter by wrapping the piece in a wet cloth and putting it in the microwave it be steamed.