Last night I set about building the steam-bending rig for the crest rail for my Windsor chair. Steam-bending is one of my favorite processes in all of woodworking. There is something that feels heroic about taking a piece of 3/4″-thick white oak and bending it as though it were a popsicle stick. The simplicity and integrity of creating graceful curves by steam bending fall very much in line with my philosophy of woodcraft. The alternatives, for lack of a better word, suck.
I could have sawed out the curve for my crest rail from straight wood, but that would require a huge chuck of oak – 3 1/2″ thick – and a comparably huge saw. And worse, it would result in a piece with inferior strength due to short grain weakness near the ends.
I could have laminated the bend from multiple thin plies, but that would require a lot of glue, a lot of clamps, and (the worst part) a lot of machining to make those thin plies. After all that work, piece would be ugly because of all the glue lines. On the plus side, it would result in a piece with comparable strength to a steam-bent piece.
Or I could have split the crest out from a piece of wood with the appropriate bend already in it (like I did with my ladle) but then I’d have the trouble of finding the right piece of wood. And good luck finding the right tree if you decide to build a balloon-back.
With steam-bending, I can procure my stock from a straight-grained log with a wedge and a froe, shape it with a drawknife, and bend it with a steaming rig that I built using a few simple items from the hardware store (look for a post on my steam-bending rig before too long, though there’s plenty of information on the web. Nothing special about mine). Essentially, I can re-write the tree’s history and make it believe that it did, in fact, grow in the shape of a ballon-back (or almost any other shape I choose).
If you weren’t convinced of (or aware of) the simple beauty of steam-bending curved pieces before, I hope that you are now. But just how does steam-bending work? Glad you asked.
For this discussion, it will be critical to understand that wood is not a homogenous material. It is a composite material. Wikipedia defines a composite material as “a material made from two or more constituent materials with significantly different physical or chemical properties that, when combined, produce a material with characteristics different from the individual components.”
Composite materials most often comprise fibers to impart strength and stiffness with a matrix that binds the fibers together and provides form and additional strength to the material. Some familiar examples of composite materials are adobe (grass/straw fibers in an earthen matrix), fiberglass (glass fibers in a plastic resin matrix), and steel-reinforced concrete.
The primary constituents of wood are cellulose (40-50%) and lignin (25-30%), and hemicellulose (25-35%). Cellulose is a strong linear polymer that forms the majority of the cell walls of wood and all other plant parts. It forms the “fiber” of the composite material. Lignin is an amorphous polymer that fills the spaces between the cells walls, binding them together, which strengthens the wood. It forms the “matrix” of the composite.
Going back to an earlier example of a composite material, let’s visualize wood as steel-reinforced concrete. (Not because it’s the most accurate analogy, but because I assume almost everyone is familiar with it). The steel (like the cellulose) has more tensile strength than the concrete (like lignin). Suppose you have a straight steel-reinforced concrete post, but you need a bent one. What will happen when you bend it? The concrete will fracture, causing the post to fail. I’m assuming you have a method of actually bending steel-reinforced concrete, of course.
Same thing with wood. Most cross-grain failure are the result of ruptures in the lignin that then propagate through the wood. But what if we could plasticize our concrete, bend it while it’s soft, and then allow it to re-harden? The steel, with its high tensile strength, would be happy to comply. Unfortunately, concrete can’t be plasticized. But lignin can.
At normal temperatures, lignin locks the cellulose in place, preventing the strands from sliding past one another. This gives wood some of its characteristic stiffness. However, if you heat it to about 200°F, a very useful thing start to happen: it gets very soft and pliable, just like a plastic. If you cool the lignin back down to room temperature and dry it out, it will tend to retain the bent shape.
This, at its heart, is the basic gist of steam-bending: Heat wood to about 200°F to plasticize the lignin, bend the wood to a desired shape, and let it cool and dry out. When it is completely dry, the wood can be removed from the form and it will hold its shape.
Unfortunately, wood is a complex material and it doesn’t necessarily make things easy on us. Lignin tends to lose its plasticity when it has been dried beyond a certain point. There is no specific point at which wood is no longer useable for steam-bending, but if the moisture content drops below about 12%, your odds of bending it successfully will be severely diminished.
Kiln-dried wood – wood that has been heated to a temperature of 140-160°F and dried to 6-8% moisture content – is next to useless for steam-bending. Air-dried wood is ideal, and the bend is most likely to be successful if the wood is near the fiber saturation point (25-30%).
In addition to loss of plasticity, another reason for unsuccesful bending in drier wood is its lack of conductivity. You probably remember from middle school science class that wood is a poor conductor (i.e., a carrier of heat and/or electricity) while water is a good conductor. When wood loses its water, it also loses the primary medium by which it transfers heat from the outside of the piece to the inside. Green wood is far more conductive than dry wood. If the center of the workpiece is not heated to the appropriate temperature, the bend is likely to either splinter while bending or fail to hold the desired shape.
Not surprisingly, then, thinner pieces of wood are easier to bend successfully than thick pieces. It’s simply much easier to thoroughly heat a 1/4″-thick board than a 2″-thick board. In fact, if your pieces are thin enough, you don’t need steam at all. I think this is a rather common misconception. It is the heat, not the moisture, that plasticizes the lignin. The steam is simply a very efficient carrier of heat, and it has the added benefit of being the right temperature.
Guitar sides are traditionally heated and bent around a hot iron pipe, and crooked chair spindles may be heated and straightened with a heat gun. The downside to these methods is that it can be much easier to scorch your pieces if you aren’t careful.
Not all woods are created equally when it comes to steam-bending. Far from it, in fact. Generally speaking, softwoods are rated poorly for steam-bending. That’s not to say that they won’t bend – simply that you must take more steps to ensure success – thinner pieces, straighter grain, longer steaming times, and/or milder bends will improve your chances of success. Most steam-bending is done with hardwoods.
I wish I could offer you some scientific explanation as to why softwoods don’t bend as well as hardwoods (since that’s kind of my thing) but unfortunately I simply don’t know. Just a wild guess: It might have something to do with tensile strength, since the outer circumference of a bend will be under severe tensile stress until the piece sets.
In the 1962 classic “Machining and Related Characteristics of United States Hardwoods” (E.M. Davis, USFS Forest Products Laboratory), the author tested 25 species of hardwoods for steam-bending characteristics and concluded: “Specific gravity influenced bending, in that the heavy woods bent better than the light woods. In table 21, for instance, all the heavy woods (those with a specific gravity of 0.50 or over) except hard maple are in the upper half, whereas all the light woods (those with a specific gravity of less than 0.40) except willow are in the lower or the poor half. No consistent differences were noted in breakage be tween light and heavy pieces of the same wood.”
As we’ve previously discussed, there is a strong correlation between modulus of rupture and density, so perhaps the correlation that Davis noted is partially related to tensile strength? Feel free to send me any resources on this topic if you know of any.
Davis also noted “Ring-porous woods as a class gave better results than did diffuse-porous. The best 4 woods are all ring-porous, and 8 of the 10 ring-porous woods were among the best 12 woods.” Again, this may be an example of correlation and not causation, since the strongest/densest woods are mostly ring-porous.
But who knows? Perhaps there is a logical link between porosity and bending that I am unaware of. I’m including the entire publication on the Wood Properties Resources Page for perusal at your leisure, for anyone interested.
I also made this handy-dandy table. The left two columns (species and successful bend %) are from Davis’ report. The additional data (pore structure, specific gravity, and modulus of rupture) are from various other sources. Mostly the USFS Forest Products Laboratory. Where a range is given, it is because the wood type covers multiple species.
There is plenty more to write on this topic, but that’s all for now. Time for me to soak my aching typing fingers. Thanks for joining me for another steamy edition of “Wednesday, Woody Wednesday”.