Continuing with the theme of wood engineering topics - the last post was about how to avoid sagging shelves - here I’ll touch on the basics of how moisture effects wood, and what to do to minimize that impact.
No, I’m not talking about wood decay and rot that the homeowners among us suffer in our siding and patios. That’s “simply” avoided by keeping that fine furniture dry. Rather, wood expands and contracts with changes in the atmospheric humidity. In Understanding Wood, Bruce Hoadley estimates that the seasonal change in humidity in the average U.S. climate for a two-story wood-framed house can change the height of a house by 3/4”!
But for many smaller “products”, the dimensional changes might be imperceptible either in absolute size or even in relative size. Nonetheless, even imperceptible changes can tear a beautiful piece of furniture apart if it’s not well designed to account for that movement; this earlier post on veneering showed one such picture. Or you may have noticed the impact with the drawer or door that sticks in summer but moves freely in winter.
In short, wood moisture content depends on the humidity. And wood size depends on wood moisture content. This can be a long and complicated topic, but I’ll boil it down to a few key points, and a handy online calculator!
Background in Layman’s Terms
Trees are collections of very long cells. When the tree is growing, those cells have water in them, and the cell walls also have water. The weight of water in a species in comparison to the completely dry version is called the moisture content (MC) - for instance, a board weighing 20# when freshly cut may only weigh 12# after it is fully dry, which yields a moisture content of 8# / 12# = 67%. Hardwoods have moisture content from 60% - 100%, but can range from 40% to 160% depending on species and where in the tree is being measured.
Wood bought from the lumberyard is typically dried - whether in a fast process in a kiln or more gradually by exposure to atmospheric air over years - to a moisture content level of 8-14%. That removes all of the water inside the cells, but leaves some water between the cells. However, the wood is gradually reacting with the environment, absorbing or releasing water. Though it differs slightly by species and a few other attributes, there is a relationship between relative humidity (how much “free” water is in the air) and the “equilibrium” moisture content (EMC) of the wood - that is, the moisture content at which the wood will not absorb or release more moisture.
But let’s back up for a moment - where is this moisture stored? Well, it’s stored where there already is some remaining moisture, in the cell walls. And because those cells are really long along the direction the tree grows, the effect of adding or removing water is somewhat akin to raking up autumn leaves with a steel-tined rake: it may get a little wider as leaves push the tines apart, but it doesn’t really get any longer.
Similarly, wood expands in the radial and tangential directions, but not really much in the longitudinal direction. In fact, that “3/4” inch” that a wood-framed house might grow or shrink in height isn’t at all because of the height of the vertical framing members, but due almost solely to roughly 38” thick of joists, top plates, sole plates, and subflooring in a two-story structure; a typical 8’ tall 2x4 installed might shrink only 1/16th of an inch even from a “green” fully saturated state to average 8% MC - so this direction of movement is insignificant and typically ignored in movement calculations and construction considerations.
So there’s a difference in shrinkage between the horizontal axes of the tree (which we need to be concerned about) and the vertical axis of the tree (which we can ignore). In fact, its a bit more nuanced: the expansion is larger in the tangential direction than in the radial direction. This is thought to be because of internal structures in the wood, but regardless this difference can be a relatively modest 1.4x in black walnut, or as much as 3.0x in California laurel.
How do you know if you have radial or tangential axis on the face of a board - or both? Though sometimes you may know from how the board was initially cut from the log, there is apparently some confusion among the terms quartersawn and riftsawn. So the best way to tell is to look at the board itself, observing how the growth rings cross the surface or the end.
So How Much Does it Expand & Contract?
How much wood moves depends on the seasonal fluctuation in relative humidity of the environment the wood is in, the wood type, and which axis of wood (tangential or radial) we are focusing on. You can find tables of tangential shrinkage (St) and radial shrinkage (Sr) percentages, such as here or in the Wood Database; these are the % shrinkage in that dimension from fully saturated to oven dry (i.e.: 0% MC). However, as shrinkage does not occur as the cell cavities dry out - shrinkage only occurs as the cell walls dry - and the cell cavities are empty only at approximately 28% moisture content, this is essentially the percent shrinkage in that direction as the wood changes from roughly 28% MC to 0% MC. And just to introduce a little more terminology, that 28% is called the “fiber saturation point”, or fsp.
We can roughly approximate* that shrinkage as linear as the MC of the wood changes. So we now need to just know the seasonal variation in relative humidity, use that to get to the seasonal variation in EMC. The equation is: width * shrinkage % * (delta EMC / fsp). Just to aid in intuition, the term (delta EMC / fsp) is essentially what fraction of the total possible range over which the loss in moisture content causes shrinkage is actually experienced by the board.
Putting it together in an example, the shrinkage of a 9” wide plain sawn yellow birch board from summer to winter is:
Shrinkage percentage: plain sawn implies we should use tangential shrinkage percent; for yellow birch St is 9.2%
Delta EMC (equilibrium moisture content): this depends on the change in relative humidity in the region; fortunately, the USDA Forestry Products Lab has put together a map to aid here, reproduced below. For the San Francisco Bay Area, we should use 8% in January to 13% in July, for a delta of 5%. Some parts of the country have a range of as small as 1%; others experience a range as great as 6%. And if a piece might be taken around the US (or the world), it should be able to handle a range of 9%.
fsp: varies slightly between species, but 28% is generally a good-enough value to use for hardwoods
Putting it all together, that 9” poplar board in summer will get narrower by 9” * 9.2% * 5% / 28% = 0.148” in winter. That’s 1.6%, or nearly 5/32”.
* This approximation has an error of about 5%, which is small enough to be good for almost all applications, but when real precision is needed, refer to a more complex formula in Chapter 6 of Understanding Wood. And I’ve skipped over a few other details, such as the assumption that the board you’re working with is already at the EMC, and the EMC of your environment falls within the EMC of the environment where the piece will live. But hey, in a short blog post, some concessions must be made.
Oy vey. That’s a lot to think of. Well, fortunately here there’s a handy online calculator that automatically pulls up the right data for different wood species and does the calculation. You still need to figure out where on the map you are (or the finished piece will be), and which dimension is radial or tangential. But the rest is done for you. Check out the Shrinkulator here.
Wood moves - so what? Well, a few examples are given at the top: sticking door, or the split coffee table. But there are two general potential problems from this movement:
The actual dimensional change
The uneven change in surfaces
Anyone who has done a wood floor - whether solid hardwood or engineered - has heard that the materials must be delivered and left onsite to “breathe” for a few days before installation; this is to acclimatize the planks to the on-site humidity so that there is less sideways movement that would open up gaps or cause buckling. This movement in floors is also less problematic in narrower planks as the more frequent gaps allow for a little more expansion or allow gaps to be spread amongst more lines, making them less visible. This flooring contractor highlights six common hardwood floor problems, and five of them are caused by movement-related moisture before, during, or after installation.
Or the uneven movement in a joint can cause problems as well. While just about every piece of furniture mates different faces of wood together, two common joint and construction techniques highlight the potential issues here:
Mortise & tenon joinery: Common in chairs, doors, and lots of furniture, a rectangular or round tenon fits snugly into a mating slot or hole cut for the mortise. If this is very snug when the wood is drier, say in winter, it’s attempt to expand in summer will be constrained by the lack of movement of the mortise (as that mortise will not grow in height with a change in humidity). This constrained expansion may deform the tenon so that, in winter when it shrinks again, it breaks the glue line and draws away from the earlier tight fit.
Rail & stile panel doors: Doors - for your kitchen cabinet or for your front door - may be of a rail and stile construction technique with solid wood panels. In short, the rail and stile form a channelled frame into which the panel can be inserted; a beginner woodworker may cut the panel to fit snugly in the channel. If the panel fits snugly in winter, as it expands in summer, it will burst or crack the frame. Or if the panel fits snugly in summer (and is then painted or finished after assembly), as it shrinks in winter, unfinished edges of the panel will start to show along the border of the panel.
As an interesting aside, wood movement is the reason plywood always has odd numbers of plies (or layers). Each ply is placed perpendicular to the previous one and would exert force on the face along the direction of the larger movement. An odd number allows those forces to cancel each other out.
There are countless other ways that movement can be a problem. For a piece to last for generations, a good woodworker will need to take movement into account in their craft.
How to Manage Movement?
Bruce Hoadley outlines five basic approaches in Understanding Wood: preshrinking, control moisture, mechanical restraint, chemical stabilization, and good design. While each of these is a book (or several) in itself, I’ll touch on them briefly here:
Preshrinking: Don’t work with green (freshly cut undried) wood! Green wood is useful in some circumstances - bending for instance. Typically, however, if you’re buying lumber from Lowes or from a specialty lumber store, it comes pre-dried, whether by kiln or air. But also - as the examples of hardwood floor issues above highlight - the wood may need to further acclimatize to your specific environment, particularly if it has been stored outside, unsheltered from the elements.
Control moisture: Unless the piece is museum grade, you’re probably not going to keep it in a hermetically sealed environmentally controlled box. But you can do a lot to retard moisture exchange with the environment, reducing the extremes in moisture content the piece would achieve at the shorter-duration extremes of humidity we might experience environmentally, by the application of a good finish. Different types of finishes vary in their ability here, so do a little research. And you can be mindful of the humidity you’re exposing the piece to, exerting some environmental controls (humidifier, dehumidifier) at the extremes which aren’t just good for your furniture, but also your comfort.
Mechanical restraint: Steel or plastics bolted to the wood can help restrain the wood - but depending on how much the wood needs to move, this could actually bow the wood between the bolts, bend plastics, or rip the bolts out! This technique is more commonly associated with plywood, where plies with perpendicular grain orientation are glued to each other to form a thicker panel. Douglas fir, for example, has tangential shrinkage of around 7.7% and longitudinal shrinkage of 0.1%; a plywood panel constructed of perpendicular plies of this has shrinkage (along both dimensions) of 0.5%, practically eliminating movement considerations.
Though plywood is often thought of as an ugly building material, there are beautiful architectural grade hardwood-veneered plywoods out there, far less expensive (and stabler and easier to work with and better for the environment) than the corresponding amount of hardwood. Just for rough numbers, in 2019, you can get a 4x8 3/4” sheet of a) construction-grade plywood for $30; b) casework-grade oak or maple plywood for $60; or c) architectural-grade walnut plywood for $250. With architectural grade, the veneers are bookmatched, consecutive, and defect free.
Chemical stabilization: Polyethylene glycol (PEG) or wood plastic composite (WPC) are two chemical approaches to stabilizing wood, though only the first is somewhat accessible to the small-shop woodworker. PEG essentially replaces the moisture in a green or fully-saturated board with a polymer of automobile antifreeze which does not evaporate, keeping the wood at its saturated dimension. I’ll confess I have no experience with either approach so will direct you to a google search to learn more!
Good design: This is where we often have the most control, between material selection, joint design and grain direction in the mating pieces, dimensional adjustments you can make (which may or may not be visible in the finished product), etc.
For a rail and stile cabinet door, for instance, a panel made of plywood won’t cause problems. Or if the design and profiles demand solid wood, a) sizing the panel small enough to allow for expansion without cracking the frame, b) finishing the panel before final assembly, and c) gluing the panel only in the center of the top & bottom rails can eliminate the problem; that’s how it’s been done for hundreds of years before the invention of plywood!
For a mortise & tenon joint, splitting a round tenon allows for a fracture line to occur inside the wood rather than at the glue line; or redesigning the joint to allow for two smaller-width tenons rather than one wider tenon makes the joint stronger in multiple ways, including halving the amount of expansion a single mortise has to accommodate.
Also, as the tangential face expands faster than the radial face, which expands faster than the longitudinal direction - flipping the boards, where possible, so that the rates of expansion are most closely aligned helps. In this diagram, we would ideally want the mortises cut in the tangential face and the horizontal board with the tenons to have the tangential face on the side. That way, the movement left-to-right is identical (if the boards are of the same material). And additionally, the vertical movement of the tenons is the smaller tangential expansion, to more closely align with the non-movement in the longitudinal axis of the vertical direction of the mortises.
For something like a solid wood countertop or tabletop anchored to a steel or composite frame - such as a hanging vanity in a bathroom - you would want to screw in the back solidly, but allow for horizontal movement by creating oval slots for the holes along the side. A 24” deep maple countertop may shift as much as 3/8” along its tangential axis. Or - if you don’t have an easy way to create oval slots in steel, just drill larger holes and use washers for the screws.
This is just a scratching the surface of good design for wood movement, and I’m only a novice here. But hopefully this has gotten you thinking about some of the considerations, either as a woodworker yourself, or in better knowing what to ask about or the language to use in working with a fine craftsmen in the field.