Shelf Sag Calculator

Shelves almost never break. They bow. A board that could carry many times its load without cracking still sinks in the middle until the row of books above it traces a shallow smile, and the smile deepens with the years.

How much bow is too much is a matter of eyesight, not safety. The woodworker’s rule quoted by WoodBin’s Sagulator is that the eye will notice a deflection of 1/32 inch per foot of shelf. On a three foot bookshelf that is 3/32 inch, or the span divided by 384. It is a guideline, not a building code, and a strict one: the timber rules tabulated by Swedish Wood let a beam on two supports finish with a net sag of a 250th to a 350th of its span.

So the question before the saw comes out is not whether the shelf will hold. It is how far it will move, on the day it is loaded and ten years later. Both can be worked out in advance.

A shelf is a beam

A shelf resting on two sides is the simplest case in structural engineering, a beam on two supports. The theory that describes it carries the names of Leonhard Euler and the Bernoullis. It was first set out around 1750 and was not applied on a large scale until the Eiffel Tower and the Ferris wheel, late in the nineteenth century. It predicts deflection from the load, the supports and a property called flexural rigidity, and it is limited to small deflections in the elastic range. A bookshelf moves a fraction of an inch, well inside those limits.

For a load spread evenly along a shelf that simply rests on its supports, the sag at the center is five times the total load times the cube of the span, divided by 384 times the stiffness of the material times a geometric quantity, the moment of inertia. The American Wood Council prints it among its beam design formulas, and the Forest Products Laboratory’s Wood Handbook lists the same coefficient, 5/384, in its table of beam loadings. Put the same load in the middle and the coefficient becomes 1/48, which makes the sag 1.6 times larger. The fraction changes with the loading. The shape of the expression does not.

Nor does it care what the beam is made of. The Engineering ToolBox uses the center-load version to find 0.025 inch of deflection in a 100 inch steel beam under 10,000 pounds.

For a rectangular board the moment of inertia is the depth of the shelf times the cube of its thickness, divided by twelve. That cube is the most useful fact in shelf design. Double the depth and the sag halves; double the thickness and it falls to one eighth. Span works against you even harder. Under a fixed load, sag grows with the cube of the span, but books fill whatever length they are given, so the sag of a full shelf climbs with the fourth power. A shelf twice as long, loaded at the same weight per foot, sags sixteen times as much.

What the wood brings

The material enters through one number, the modulus of elasticity, and the way it is measured is the shelf formula run backward. A test piece is laid across two supports and loaded at the center, its deflection is read, and the modulus is whatever makes the equation balance. The Wood Handbook’s tables come from tests of that kind on clear, straight-grained pieces at 12 percent moisture content. They give northern red oak 1.82 million pounds per square inch, sugar maple 1.83, black walnut 1.68, black cherry 1.49 and eastern white pine 1.24. Among those five, the stiffest is about half again as stiff as the softest.

Those figures are averages. The same handbook puts the typical coefficient of variation of the modulus of clear wood at 22 percent, so a prediction for one particular board is an estimate, however many decimals it shows.

Sheet goods are a different class. In the handbook’s comparison of wood-based composites, plywood has a bending modulus of 6.96 to 8.55 gigapascals, particleboard 2.76 to 4.14 and medium-density fiberboard 3.59, against 12.27 for clear white oak. MDF is wood fiber and resin joined under heat and pressure; its trade association describes a surface that is flat, smooth and free of knots and grain. It is also about three and a half times less stiff than red oak.

Particleboard, which the same association lists for shelving among its common uses, is made of residual chips and shavings and lands in the same range. Plywood sits higher, though still well short of solid wood. It is built from cross-laminated veneers, and its makers’ association notes that its stiffness depends on the direction of stress relative to the panel’s strength axis, so which way a shelf is cut from the sheet matters.

Stiffness is not strength, and for shelves the gap between them is wide. Take a red oak shelf 36 inches long, 10 inches deep and 3/4 inch thick, carrying 90 pounds of books. That is 30 pounds per foot, in the middle of WoodBin’s 20 to 40 for a fully loaded bookshelf. The bending stress at the center works out to 432 pounds per square inch. The handbook’s modulus of rupture for clear northern red oak is 14,300. The board is at about 3 percent of its breaking stress and already close to sagging visibly. Appearance gives out long before the wood does.

The sag that arrives later

The formula gives the sag on the day the books go on. Wood then keeps moving. The Wood Handbook calls it creep: deformation that continues under a steady load, even a small one, for years. At typical loads and conditions, it says, the extra deformation after several years may roughly equal the initial one. The handbook records the custom of holding the first-day deflection of dry timber to about two thirds of the long-run limit, and the US National Design Specification multiplies the deflection from a long-term load by 1.5 for seasoned lumber and 2.0 for wood structural panels in dry service.

European timber design is more specific, and harder on sheet goods. Eurocode 5 assigns each material a deformation factor, and under a permanent load the final sag is the first-day sag times one plus that factor. In a heated room the factor is 0.60 for solid timber and 0.80 for plywood, as tabulated by Swedish Wood, but 2.25 for load-bearing particleboard and for MDF. A solid shelf therefore ends at about 1.6 times its first-day sag. An MDF shelf ends at 3.25 times. That is the case against the long MDF shelf: it starts three and a half times more flexible and is then multiplied by twice as much.

Moisture makes it worse. In a Forest Products Laboratory study, plywood held under load at 50 percent relative humidity for six months had deflected to 1.4 times its initial value on average, and oriented strandboard to 1.7 times; at 85 percent humidity the figures were 1.9 and 5.2. A later test of exterior-grade MDF found creep deflections under cycling humidity two to three times those at a steady 65 percent. The heated-room factors do not apply in a garage or a basement.

One piece of folk practice has a basis. On unloading, the handbook says, the elastic deformation returns at once and about half of the creep follows in time. An emptied, flipped shelf starts again from a smaller bow.

Four ways to buy stiffness

The calculator above opens on that same red oak shelf: 36 inches of span, 10 inches deep, 3/4 inch thick, 90 pounds spread evenly, resting on two supports. In metric that is 914.4 millimeters, 254, 19.05 and 40.8 kilograms. The moment of inertia is 10 times 0.75 cubed over 12, or 0.3516 inches to the fourth power, and the sag comes to 0.085 inch, the span divided by 421. That passes the eye test of 0.094 inch. Multiply by 1.6 for the years ahead and it reaches 0.137 inch, the span divided by 263. The shelf looks flat when it is built and bowed when it is old.

Thickness is the cheapest cure. At 7/8 inch the same shelf sags 0.054 inch on the first day and 0.086 after creep, inside the guideline. An eighth of an inch of extra wood does what swapping red oak for any other species in the handbook’s table of American woods cannot. Shortening works as well. Held to 3/4 inch and the same 30 pounds per foot, the longest red oak shelf that stays under the guideline for years is 31 5/8 inches. The same board in MDF sags 0.299 inch at once and 0.97 inch eventually, and its longest flat span is 16 1/2 inches.

The third cure is the lip, a strip glued along the front edge. It works because material counts toward the moment of inertia by the square of its distance from the axis the section bends about. That is the parallel-axis theorem taught in mechanics courses. A strip 3/4 inch thick and 1 1/2 inches tall, flush with the top, raises the moment of inertia of the example shelf from 0.3516 to 0.7001 and halves the sag to 0.043 inch. The MDF version needs a lip 2 7/8 inches tall to stay under the guideline. The tool treats the lip as the shelf’s own material, so it understates what a hardwood strip does for fiberboard.

The fourth is to stop the ends from tilting. For a beam fixed at both ends the coefficient for a spread load falls from 5/384 to 1/384, so the sag is one-fifth of the resting figure: 0.017 inch for the red oak shelf. That is the ideal of a perfectly rigid joint. A shelf on pins or cleats is free to tilt and gets none of it, and a shelf glued into a housing or screwed through the sides gets some unknown share. The Forest Products Laboratory’s older advice on particleboard shelving is plainer: keep the supports close together and, where possible, support the rear edge.

What the arithmetic leaves out

The calculation is a model with plain limits. It is elastic bending and nothing else. It does not check whether the shelf, the pins or the screws are strong enough. It ignores the shelf’s own weight and the extra deflection from shear, which the handbook says is partly folded into its moduli and is worth about 10 percent. It treats a lipped shelf as one rigid section, which a wide, thin panel is not. Its creep factors are code values for load-bearing boards in a heated room, not measurements of the panel in your hands.

The Forest Products Laboratory warned decades ago that much particleboard shelving is sold with no indication of its grade, and the 22 percent scatter in solid wood applies to every plank. The honest reading of any result is a range. A shelf predicted to finish at the span divided by 400 may or may not show its bow. One predicted at the span divided by 800 will not. Margin is cheap when it is bought in thickness.

The bookcase that sags was not overloaded in any sense a structural engineer would recognize. It was too thin by an eighth of an inch, or too long by four inches, or made of the wrong board. All three could have been known before the first cut.