Load Equivalency Factor (the Fourth Power Law)

Start with the thing everybody knows, because it is not true. "The fourth power law" does not appear in the AASHO Road Test. The phrase is nowhere in Special Report 61E's three hundred and seventy-one pages, and neither is the exponent 4. What the Road Test actually published was a serviceability model — a logarithmic relation in which pavement condition falls from an initial value toward a terminal one as load applications accumulate — with a load term inside it. The exponents fitted to that load term were 4.79 for flexible pavement and 4.62 for rigid. Not four, and not even the same number for the two kinds of road.

So where does four come from? It is a later curve fit. If you take the Road Test's own equations, generate equivalency factors across the design space they cover, and then ask what simple power law $\mathrm{LEF} \propto L^{n}$ best reproduces them, you get an exponent that lands between roughly 3.2 and 4.5 — and which value you get depends on the pavement type, on the structural number, and on the terminal serviceability you decided to call failure. The flexible-pavement curve is not even monotonic in structural number: push the thickness up and the effective exponent goes down and then back up again. Four is a round number that sits inside that spread and is easy to say. That is its entire claim.

This site therefore makes $n$ a field you type, defaulting to 4. That is the same decision the catalog makes about Richter's local magnitude, about the $b$ in Gutenberg-Richter, and about $R_0$ in the Kermack-McKendrick model: where a celebrated number is a fitted parameter rather than a constant of nature, the reader gets to move it and see what happens. Here it is not an academic exercise. $n$ is the number underneath every overweight-permit argument, every spring load restriction, and every road-user-charge schedule in the world. Change it from 4 to 3.5 and a 20% overload goes from costing 107% extra damage to costing 89% extra. Change it to 4.79 and the same overload costs 155% extra. Fortunes and statutes turn on the choice, and the choice is a fit.

The second thing to get right is that the power form describes a SINGLE axle. The commonest error in the subject is to apply it axle by axle down a truck and add the answers up. Do that to a 34,000 lb tandem — two axles at 17,000 lb — and you get $2 \times (17/18)^4 = 1.591$. The tabulated equivalency factor for that tandem is about 1.10. The power form is 45% high, and the reason is mechanical: the two axles of a tandem are close enough together that they share one deflection basin. The second axle rolls into ground the first has already pushed down, so it does not deliver a full independent load cycle. Nothing on this page corrects for that, because the correction is the tabulated data, and that data is not free to reproduce. What the page can do is tell you the error exists and which way it points.

Third: the calibration. The AASHO Road Test ran from 1958 to 1960 on six loops near Ottawa, Illinois. One subgrade, one climate, one construction season, two years of trafficking, and bias-ply tyres at 75 to 80 psi. Modern highway trucks run radials at 85 to 145 psi, on wider single tyres the test never saw, at speeds and axle configurations that did not exist. Every equivalency factor in daily use is an extrapolation past its calibration, and the source reports say so themselves. The Road Test was a magnificent piece of work and it is still the largest full-scale pavement experiment ever run; it is also sixty-five years old and it measured one road.

What survives all of that is the shape of the relation, and the shape is the useful part. Damage rises far faster than load — steeply enough that light vehicles vanish from the arithmetic entirely and heavy ones dominate it completely. At $n = 4$, one axle of a loaded five-axle semi does about five thousand times the damage of one axle of a car. A whole loaded semi, using honest tandem factors, is worth something like twenty to twenty-five thousand cars; use the naive per-axle sum instead and the same truck comes out at over thirty thousand. That spread — the same physics, the same exponent, one modelling choice — is the honest uncertainty in the most-quoted statistic in the field.