McKee Box Compression Formula (Short Form)

In August 1963, three researchers at the Institute of Paper Chemistry published a formula in a trade magazine, and sixty years later it is still how the world's corrugated boxes get specified. Robert McKee, John Gander and John Wachuta had been working on why boxes fail in compression, and what they arrived at was not a curve fit but a mechanics result with a curve fit inside it: $BCT = 5.874 \, ECT \, h^{0.508} Z^{0.492}$.

Start with the exponents, because they are the tell. Almost everyone writes this formula as $5.87 \, ECT \sqrt{hZ}$, on the entirely reasonable ground that 0.508 and 0.492 are both very nearly a half. They are — but $h$ and $Z$ are not nearly anything: a C-flute box has a caliper of 0.16 inches and a perimeter of 56, four hundred times larger. The ratio raised to the difference of the exponents, $(h/Z)^{0.008}$, comes out at 0.954, so the square-root version returns about 4.7% MORE than the exponents McKee published. On a number this consequential that is not a rounding, and it always errs on the unsafe side. This site uses 0.508 and 0.492.

Now the thing almost nobody points out, which is that both McKee constants are DIMENSIONLESS. Look at what each side carries. Compression strength is a force. $ECT$ is a force per unit length. Caliper and perimeter are lengths. So the right-hand side has a force exponent of one — good — and a length exponent of $-1 + 0.508 + 0.492 = 0$. The lengths cancel exactly. That cancellation is not luck; the exponents were CHOSEN so that it would happen, which is why they are 0.508 and 0.492 rather than any other pair summing near one. The consequence is that 5.874 is a pure number and the equation works in any self-consistent unit system: pounds-force with pounds-force per inch and inches, or newtons with newtons per metre and metres. Same constant, same answer.

And that is why this site can do something a vendor calculator cannot. The industry's conventional metric unit set is not self-consistent — it quotes $ECT$ in kN/m, caliper in millimetres, perimeter in centimetres and the answer in kilograms-force. Mix those and the cancellation breaks, so the trade patches it with a separate metric constant near 1.855, and at least one major vendor's documentation never states out loud that its metric output is kgf rather than newtons. A units engine that converts everything to SI before it computes makes the entire problem vanish. You can verify it on this page: enter 32 lbf/in, 0.16 in, 56 in and you get 536.89 lbf; enter the same board as 5.604 kN/m, 4.064 mm, 1422.4 mm and you get 2388.21 N, which is 536.89 lbf to twelve figures. One constant, no patches.

What the equation is DOING, physically, is a plate-buckling calculation. The four side panels of the box are thin plates standing on edge, and under a top load they bow outward and fail — the same phenomenon as a slender column buckling under Euler's formula, one dimension up. That is why board bending stiffness belongs in the equation, and caliper is standing in for stiffness rather than mattering in its own right. The perimeter $Z$ is the total loaded edge length; $ECT$ is the strength of that edge per unit of it. Understand that and every term has a job.

Now the honesty, and it is substantial. Urbanik and Frank pooled box compression data spanning forty-six years and found the stock constant overestimated measured strength by 52% on average against one published data set. Refitting the constant to that data cut the average error to 25%. Neither number is a scandal: $k = 5.874$ was fitted in 1963 to single-wall regular slotted containers loaded top to bottom, and refitting a constant to your own board and your own boxes is normal, expected practice rather than an admission of failure. What it does mean is that this figure is a design starting point, not a measurement, and that a thin margin computed from it is thinner than it looks.

Three classic mistakes, in order of how much damage they do. The first is entering caliper in millimetres where mils were meant, or the reverse — C-flute is 0.16 in, 160 mils and 4.06 mm, all the same board, and the two wrong readings are off by factors of about 6.5 and 25 respectively. The units engine here removes that mistake entirely, because you say which unit you meant. The second is ignoring the shallow-box limit: below a depth of about $Z/7$ the panels are too short to buckle at all and this equation has the wrong form, not merely the wrong calibration. The third, and the expensive one, is treating the answer as a service capacity. It is a laboratory number on new, dry, undamaged board tested in about thirty seconds at 50% relative humidity, and the box in your warehouse is none of those things.