Stacking Safety Factor

Capacity divided by demand. The arithmetic takes a second; the question of what belongs in the numerator is where the whole subject lives.

The warning first, because it is the reason this page is written at all. The safety factor and the derating multiplier chain are COMPETING METHODS, not complementary ones, and applying both in full double-counts the same losses. The trade rule "divide the laboratory BCT by three to seven" was written for people who were NOT going to derate factor by factor — the three-to-seven band is a lumped stand-in for precisely the humidity, creep, overhang and pattern effects the derating page models explicitly. A realistic explicit chain lands near 0.12 by itself, which is already a divisor of about 8.5. Multiply a safety factor of five on top and you have divided by more than forty. The result is a board grade nobody makes, a quotation you will lose, and a great deal of fibre doing no work.

So decide which method you are using and say so on the drawing. If you have derated factor by factor, the remaining safety factor covers only what the factors did not model — handling damage, board-to-board variability, the pallet that got dropped — and 1.2 to 1.5 is an honest range. If you have applied a lumped divisor to the laboratory figure, do not also apply the factors. What is never defensible is a safety factor whose value nobody can trace, applied to a capacity nobody can trace, resting on a McKee prediction that Urbanik and Frank found overestimated one published data set by 52%.

What a safety factor is not. It is a ratio of central estimates, and it is not a probability of survival. The quantity you actually want is the overlap between the distribution of box strengths coming off the converting line and the distribution of loads in the warehouse, and those distributions have widths. A box population with a wide spread can fail at a nominal safety factor of two, while a tightly controlled one survives at 1.3. When two suppliers quote the same ECT and one of them has half the standard deviation, they are not selling the same box, and a safety factor cannot see the difference.

Which capacity goes on top. Judging a box on its laboratory BCT is the single most expensive mistake in packaging, and it is expensive precisely because the answer looks so reassuring. This site's worked example is a 32 ECT C-flute shipper that tests at 537 lbf and carries 100 lbf in a six-high stack: an apparent margin of 5.4 to one. Derate it honestly for moisture, storage time, overhang and pattern and the service capacity is 110 lbf, so the real margin is 1.10. That box passed the laboratory by more than five and is running on almost nothing in the warehouse — and nothing about the certificate would tell you.