Cumulative ESALs with Traffic Growth

This page exists mostly to stop one mistake, and the mistake is so common that it deserves the top of the page rather than a footnote.

The growth factor is already a sum. $\mathrm{GF} = ((1+g)^{Y} - 1)/g$ is the closed form of a geometric series: it has already added up year one, plus year two at $(1+g)$, plus year three at $(1+g)^2$, all the way to year $Y$. The chain is nevertheless written, over and over, in coursework and in spreadsheets and occasionally in reports that get built from, as "annual ESALs × growth factor × design life". That multiplies by the analysis period twice. At twenty years it inflates the answer by a factor of twenty, and a pavement designed on it would be absurd — except that ESAL totals are large abstract numbers that nobody has an intuition for, so a design traffic of two hundred million where ten million was meant does not look obviously wrong on a page.

The check takes two seconds and it never fails you: at $g = 0$, the growth factor must equal the number of years. With no growth, the total is simply $Y$ identical years, so $\mathrm{GF} = Y$. Put zero growth into whatever you are using and see what comes out. If it gives $Y$, the factor is behaving; if the chain then multiplies by $Y$ again, you have found the bug. (Mathematically, $\mathrm{GF}$ at $g = 0$ is $0/0$, and its limit is $Y$. This site's brain supplies that limit rather than returning a division error, because a reader who genuinely expects flat traffic deserves an answer.)

Now the harder truth about the number itself. The growth rate is the least defensible input in the whole design chain. Nobody can forecast truck traffic twenty years out. Growth is not smooth — it is a step function punctuated by a new distribution centre opening, a mill closing, a competing route being twinned, a border crossing being rebuilt, a recession. Compounding a smooth annual percentage across two decades produces a number with far more apparent precision than the underlying knowledge supports, and the compounding is not gentle: at 3% a year, traffic in the final year is 1.75 times the first year's; at 5% it is 2.53 times.

The honest treatment is therefore not to find the right growth rate. It is to run the design at two or three rates and look at how much the pavement thickness actually moves. Very often the answer is: not much, because the structural number depends on the logarithm of cumulative traffic, and a factor of two in ESALs is a fraction of an inch of asphalt. That is a genuinely reassuring result and it is worth knowing, because it tells you where the argument is worth having and where it is not. When the thickness does swing — usually on a lightly trafficked road where a new industrial user might double the truck count — that is your signal to design for the possibility rather than the forecast.

Two smaller cautions. First, the compound form assumes a constant percentage; where a specific known development will step the traffic up in year six, model it as a step, not as a rate. Second, the equation counts whole years from a first-year base, so be clear whether your $w_1$ is the traffic in the year the road opens or the traffic today — a design that spends three years in procurement and construction has already lost three years of growth from its base.