Steam bending and bending jigs

Everything below is workshop experience — the kind of judgement that gets passed along a bench, not measured in a laboratory. Where a rule of thumb appears it is labelled as one, and where a number would normally be quoted you will find a test bend on scrap instead. Species behave differently, boards within a species behave differently, and your steam box is not the same as anyone else's. Treat this page as a map of the decisions, not a table of settings.

§1 The choiceSteam-bend or laminate?

The first decision is not how to bend, but which method the part wants. Both produce a curve; they fail in different ways and look different when they succeed.

Steam bending keeps the section solid and the grain continuous around the curve, which is why chair backs, boat frames and walking sticks are traditionally bent rather than sawn: a sawn curve severs the grain at the shoulders and leaves short, weak fibres where the load runs. Steam bending needs straight-grained stock, a box, a strap and a former, and it asks you to accept some variability — the wood has a say in the final shape, and a proportion of blanks will fail. It suits parts where strength along a curve matters and where a little movement after bending can be accommodated.

Laminating — gluing a stack of thin, easily flexed strips over a former — trades that grain continuity for predictability. The cured part holds its shape with far less springback than a steamed bend, repeated parts come out near-identical, and there is no steam kit to build. The costs are glue lines on show at every edge, the work of sawing or resawing accurate strips, and a former that must be built to take clamping pressure over its whole face. It suits tight or compound curves, parts that must match a drawing closely, and any curve you need to make several of.

• Grain continuity and a solid section matter → steam bend.
• The curve must match a drawing, repeatedly → laminate.
• Visible glue lines would spoil the part → steam bend.
• The curve is tight, compound, or both → laminate.
• No steam kit and no wish to build one → laminate.
• The stock available is thin, clear and plentiful → laminate; thick, straight and clear → either.

§2 StockStock that wants to bend

Bends fail at defects. Choose blanks with the straightest grain you can find, and reject anything where the grain runs off the edge of the blank along the curve — grain runout is where splinters lift and breaks start. Knots, even tight ones, are grounds for rejection in a bending blank; so are checks and shakes. Riven or rift-sawn stock, where the fibres follow the length of the blank, is the traditional preference for exactly this reason.

Moisture history matters as much as grain. In workshop experience, air-dried stock bends more willingly than kiln-dried: kiln schedules leave the fibres less able to soften and slide, and a bend that air-dried ash takes calmly can fracture in kiln-dried stock of the same species. That is bench experience, not published data — but it is consistent enough that if you have a choice of boards, the air-dried one is the bending blank. Species choice is covered properly in thetimber selection guide; for bending, the workshop rule of thumb is that the ring-porous temperate hardwoods — ash, oak, elm and their relatives — are the willing ones, and that brittle or resinous species earn their reputation the hard way.

§3 SteamThe steam box and the strap

A steam box is a lidded chamber a little longer than the blank, fed with steam from a wallpaper stripper or purpose-made generator, with a drain at the low end and dowel racks inside so steam reaches every face of the stock. It does not need to be pressure-tight — it should leak a little, so fresh steam keeps moving through. Build it from softwood or exterior ply, slope it towards the drain, and drill a hole for a probe thermometer so you know it is properly hot rather than merely warm.

Timing folklore abounds, usually quoted per inch of thickness. This site does not repeat those numbers: schedules vary with species, moisture content and how well your box holds heat, and an over-steamed blank fails as readily as an under-steamed one. Steam a test piece from the same board, bend it, and let the result set your schedule. The one certainty is that once a blank leaves the box you have well under a minute of useful pliability — everything must be laid out, clamps open, gloves on, before the lid lifts.

The strap is what makes thick bends survive. Wood tolerates being compressed along the grain far better than being stretched; an unsupported bend stretches its outside face until the fibres tear. A steel or aluminium strap clamped to the outside of the curve, with solid end stops bearing on the ends of the blank, forces most of the movement into compression on the inside face instead. The end stops carry the real load — make them stout, and make the strap fit the blank's length closely, because slack is stretch.

§4 JigsFormers and clamping jigs

The former gives the bend its shape, for steamed and laminated work alike. Band-saw it from thick material or stack and screw layers of ply or MDF; fair the curve with a spokeshave and check it with a batten, because every lump in the former prints into the part. Give it clamping ears or a foot that fixes to the bench, and cut clamp access — notches or through-holes — before you need them, with a dry run of the full clamping sequence while everything is cold and calm.

• Work from the middle of the curve outwards, one clamp at a time, so nothing bunches.
• Use cauls between clamp heads and the work; a steamed face is soft and marks easily.
• For laminating, a two-part former that closes on the stack spreads pressure best.
• Wax or tape the former faces for glue work, or the part becomes part of the jig.
• Leave the part clamped until it has fully cooled and dried — impatience undoes bends.

§5 SpringbackAllowing for springback

Every bend relaxes a little when the clamps come off. Steamed parts move more, laminated parts less, but neither holds the former's exact curve. You will find percentage allowances quoted elsewhere; this site does not invent one, because springback depends on species, thickness, moisture and the curve itself. The workshop answer is to over-bend against a former whose curve is tighter than the drawing, by an amount you establish from a test bend in the same stock — bend, release, measure against the drawing, and adjust the former. One sacrificial blank tells you more than any table would.

Bent stock also keeps responding to its environment, which is one more argument for planning parts so that a small drift in curvature will not strand the whole assembly — and for cutting bending blanks with generous length so the sprung ends can be trimmed to fit. Offcut discipline and blank planning belong to the same habit of mind asreducing waste in the workshop: the material you save on a failed bend is the blank you test the next one with.

§6 ChecklistBefore the first real bend

• Blank straight-grained, defect-free, no runout along the curve; air-dried if you have it.
• Former faired, fixed down, clamp access cut, sequence rehearsed dry.
• Strap fitted with solid end stops; gloves and clamps laid out in reach.
• Box up to temperature on a probe, not on faith; drain clear.
• A test blank from the same board steamed and bent first — schedule and springback set from it.
• Time cleared: from lid to last clamp is one unhurried, uninterrupted sequence.