Technical information

4-8
For performing the backcalculation, the n1 pavement system was represented
using four layers. The three HMA lifts were combined into one (top) layer, 5 in. (127
mm) thick with an assumed Poisson’s ratio of
30.0
1
=
ν
. The second layer from the top
represented the crushed aggregate base course, with a thickness of 6 in. (152.4 mm) and
35.0
2
=
ν
(assumed). Because no instrumentation was embedded in the subgrade (only
on top), there was no available data to support its sub-layering. Hence, the upper and
lower subgrade layers were combined into one layer (third layer from the top) having a
total thickness of 61 in. (1.55 m) and 40.0
3
=
ν
(assumed). The fourth and final layer,
with semi-infinite thickness, represented the concrete floor of the test pit. The elastic
properties of this layer were fixed to the following values:
000,000,4
4
=
E psi (27,580
MPa) and
20.0
4
=
ν
. The dual-wheel loading was represented by two circular areas,
each 8 in. (203 mm) in diameter, transferring uniform vertical stresses of 150 psi (1.03
MPa) to the pavement surface. The spacing between the centers of the loads was taken
as 13.5 in. (343 mm). For simulating the moving APT carriage, the quasi-static
approach was applied in which dynamic (inertial) effects are disregarded. This
assumption seemed reasonable because of the relatively slow loading speeds in the
APT.
Generated model responses were compared to measured responses and a
nonlinear optimization algorithm (Fylstra et al., 1998) was applied to manipulate the
material properties until a best fit was achieved. This process was repeated twice to
separately analyze the structure during pass #5,000 and during pass #80,000. Due to the
non-symmetric strain response of the pavement, only data from the approaching branch
were used for the comparison. Subsequently, 25 data points were pre-selected from each
time history, corresponding to 25 different APT carriage positions relative to the gauge
location with denser spacing closer to the gauge. These ‘offset’ distances ranged
between 70 in. (1.78 m), for which readings were negligible, and zero, in which the
APT carriage was exactly in line with the gauge along the Y-axis (see Figure 3.5.1).
Regardless of the number of data points used for the comparison between model
and experiment there were only three moduli that needed to be backcalculated (for a
given pass level), namely the HMA modulus (
1
E ), the aggregate base modulus (
2
E ),