Question 7 of 7: Short answers on traffic growth, LEF and environmental factors

Question 7: Short answers on traffic growth, LEF and environmental factors (20 marks — (a) 7, (b) 7, (c) 6)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

Given. Three short discussion items on the AASHTO flexible design method: the choice of traffic growth rate, the meaning of a load equivalency factor that varies with $SN$, and the treatment of environmental effects.

Find. A reasoned answer to each, supported where possible by the numbers the method itself produces.

Part (a) — is the conservative growth rate justifiable?

Verdict: defensible on these numbers, but for the right reason and not as a habit. The first thing to do is price the conservatism rather than argue about it. Over a 20-year design period the two growth rates give

so choosing the upper bound raises the cumulative ESALs by 10.8 %. Feeding that into the design equation with the Question 5 parameters moves the required structural number from 3.977 to 4.038, i.e.

Fourteen hundredths of an inch is less than the placing tolerance of a paver, so on this project the "conservative" decision costs essentially nothing and buys real protection against a traffic forecast that turns out low. On that basis it is justifiable, and the engineer should say so explicitly in the design brief.

Two qualifications keep the answer honest. First, the AASHTO method already carries an explicit uncertainty mechanism: the reliability level $R$ and the overall standard deviation $S_0$, and AASHTO states that $S_0$ for flexible pavements (0.40 to 0.50) includes the error in the traffic prediction. Biasing the traffic input upward and then applying a high reliability double-counts the same uncertainty, and the compounded conservatism is no longer transparent to the client or to a reviewer. The defensible practice is to design on the best estimate — here 3.5 % — and to carry the uncertainty in $R$ and $S_0$ where it is visible and auditable.

Second, the logarithmic form of the design equation means pavement thickness is very insensitive to traffic but pavement life is very sensitive to thickness. A 10.8 % error in traffic is worth 0.06 in $SN$, but a 0.5 error in $SN$ is worth roughly a factor of two in life. The engineer's effort is therefore far better spent on the subgrade modulus and the drainage coefficients, which move $SN$ much more, than on shading the growth rate. The right deliverable is a short sensitivity table over 3 % to 4 % plus a life-cycle cost comparison, not a single conservative number.

Part (b) — does a higher LEF at a higher SN mean more damage?

Verdict: the inference is wrong. It mistakes a ratio for an absolute quantity. The load equivalency factor is defined as

with both numbers measured on the same pavement. It is therefore a relative index, and its numerator moves with $SN$ just as its denominator does. A larger $LEF$ at $SN=5$ says only that, as the pavement gets thicker, the standard 18-kip axle gains allowable repetitions slightly faster than the 40-kip tandem does. It says nothing whatever about absolute damage.

The absolute question is settled by the design equation itself. Taking $M_R=4000$ psi and $\Delta PSI=2.0$ at $R=50\ \%$, the allowable standard-axle loading is $W_{18}=2.698\times10^{6}$ at $SN=4$ and $1.287\times10^{7}$ at $SN=5$. Dividing each by the tabulated factor for the 40-kip tandem gives the allowable number of passes of that very axle:

The thicker pavement carries the same 40-kip tandem 4.7 times as many times before reaching the same terminal serviceability. Thicker is emphatically better, exactly as physical intuition demands: a stiffer structure spreads the wheel load over a larger area of subgrade and reduces the tensile strain at the bottom of the asphalt.

There is a further reason not to build an argument on that particular comparison. Reading down the $SN$ row for the 40-kip tandem gives 2.21, 2.16, 2.06, 2.03, 2.08, 2.14 — the factor falls, reaches a shallow minimum near $SN=4$, then rises again. This non-monotonic shape is an artefact of the regression fitted to the AASHO Road Test data, over which the heaviest axles were represented on a limited range of sections. A 2.5 % wobble in a fitted coefficient is well inside the scatter of the underlying experiment and cannot support a physical conclusion at all.

Part (c) — environmental factors in the AASHTO method

The AASHTO 1993 flexible method handles the environment in four distinct places rather than through a single factor, and a complete answer should name all four.

• Seasonal variation of subgrade support, through the effective resilient modulus. The roadbed soil is far stiffer when frozen and far weaker during spring thaw. The method divides the year into periods, assigns $M_R$ to each, converts each to a relative damage $u_f=1.18\times10^{8}M_R^{-2.32}$, averages the damage (not the modulus) over the year, and converts back. Because the exponent is $-2.32$, the weak thaw period dominates, and the resulting effective $M_R$ — the 4000 psi and 3800 psi values used in Questions 5 and 6 — is much closer to the spring value than to the annual mean.
• Moisture in the unbound layers, through the drainage coefficients $m_2$ and $m_3$. These multiply the base and subbase contributions in $SN=a_1D_1+a_2D_2m_2+a_3D_3m_3$. They are selected from a two-way table of drainage quality (time for the layer to drain to 50 % saturation) against the percentage of the year the layer is near saturation, and they range from 1.40 for excellent drainage that is rarely saturated down to 0.40 for very poor drainage saturated more than a quarter of the time. A pavement with poor drainage therefore needs a substantially thicker granular structure for the same $SN$.
• Frost heave and swelling, through an additional serviceability loss. The traffic-related loss $\Delta PSI_{TR}$ is reduced by the environmental losses $\Delta PSI_{SW}$ (swelling of expansive clay) and $\Delta PSI_{FH}$ (differential frost heave), which are estimated from the swell rate or frost-heave rate, the potential loss and the fraction of the project affected, and plotted against time. Only the remainder, $\Delta PSI_{TR}=p_0-p_t-\Delta PSI_{SW}-\Delta PSI_{FH}$, is available to be consumed by traffic, so environmental damage shortens the design life directly. Where heave or swell is severe the design is normally handled by subgrade treatment — excavation and replacement, non-frost-susceptible subbase to the frost depth, or moisture barriers — rather than by adding thickness.
• Temperature and aging of the bound layers, through material selection and the layer coefficient. The equation contains no explicit temperature term; low-temperature transverse cracking, thermal fatigue and binder aging are controlled outside the equation by choosing the binder grade (in Canada, the Superpave performance grade selected for the site’s climate percentiles), and by taking $a_1$ from a mix stiffness measured at a representative temperature.

In Canadian practice the TAC Pavement Asset Design and Management Guide keeps the same four mechanisms but places much heavier emphasis on frost: over most of the country the depth of frost-susceptible material that must be replaced, and the spring load restrictions applied to the finished road, are governed by frost penetration rather than by the structural number.