The data in Figure 1 show a significant difference between the cumulative
distributions for the base layer and subgrade layer pressure cells. The two cumulative
distributions do not overlap, and the separation in absolute difference between the two
distributions increases as percentile increases. At the 95th percentile, the difference
between the base and subgrade distributions is approximately 0.5 psi. The 95th
percentile base pressure difference is 0.78 psi while the 95th percentile subgrade pressure
difference is 0.29 psi. To statistically validate the difference between the two
distributions, a Kolmogorov-Smirnoff (K-S) test (α = 0.05) was performed on the two
cumulative distributions (St. Johns, 2008). The K-S test yielded a p-value of 0.022,validating the visual interpretation of Figure 3 by providing statistical evidence of a
difference between the two data sets.
Figure 1. Cumulative distributions for base and subgrade pressure cells.
To assess the relative significance of these absolute difference values, a quick
analysis was performed on the dynamic pressure response data from the 2006 Test
Track. These data were current through mid-2008 at the time of analysis. For this
analysis, a database of single axles containing the 95th percentile pressure registered
from each gauge on the various testing dates was compiled. The 95th percentile pressure
was utilized as a representative pressure from each testing date so as to remove positive
outliers from the dataset. From these data, an average base pressure and subgrade
pressure that was registered from the instruments under traffic was determined. For the
base pressure cells under single axle load, the average pressure response was
approximately 9.6 psi. For the subgrade cells under single axle load, the average
pressure response was approximately 6 psi. For both the base and subgrade pressure
cells, the measure of variability (95th percentile absolute pressure difference shown in
Figure 1) is less than 10% of a typical field pressure reading.
Load Level Comparisons. Given the differences between the cumulative distributions
of the base and subgrade pressure cells, the question now became whether or not these
differences were a function of the overlying material (HMA versus unbound granular
material), or a function of other factors (e.g. load level, pressure magnitude, or
pavement condition). To answer this question, the absolute difference data were further
analyzed according to the magnitude of the load placed on the pressure cells.
To determine the effect of load level on pressure cell measurement repeatability,
the absolute difference values for the base and subgrade pressure plates were analyzed
by load level. A combined data set including all base and subgrade pressure data was
analyzed as well, with load level being the only subdivision. For this analysis, the
average and standard deviation of the absolute difference for each of the three data sets