Two typical pavement structures were selected in this dissertation: a thin asphalt pavement with a granular base layer (simulating a secondary road), and a thick asphalt pavement, such as that used for interstate highway systems. The two pavement structures
were exposed to the accelerated pavement testing (APT) using the Accelerated Testing Loading ASsembly (ATLAS) (Figure 4.10). The full-scale pavement testing provides an acceptable middle ground between real pavement loading in the field and laboratory tests. Various pavement responses were measured during APT using in-situ instrumentations, including strains, stresses, and deflections.
(a) (b)
Figure 4.10 Advanced Transportation Loading ASsembly (ATLAS) with a (a) dual-tire assembly; and (b) wide-base tire
The secondary road pavement structure resembled one of the existing test sections constructed in a previous project for evaluating the effectiveness of geogrid reinforcement (Al-Qadi et al. 2007). It is composed of a 76-mm asphalt layer and a 305- mm unbound aggregate base layer, as shown in Figure 4.11(a). PG 64-22 binder was used in the SM-9.5 mix (surface mix with maximum nominal aggregate size of 9.5mm) for the asphalt layer. Dense-graded crushed limestone aggregates were used for the granular base layer. The pavement section was constructed on subgrade with a low CBR of 4. During construction, strain gauges were embedded at the bottom of the asphalt layer, and pressure cells and linear variable deflection transformers (LVDTs) were embedded in the base layer and subgrade. Thermocouples were placed in the asphalt layer, base layer, and subgrade to monitor the pavement temperature.
The thick pavement structure is one of the test sections built as part of an extended- life pavement project (Carpenter 2008). It is composed of a 254-mm asphalt layer directly
over a lime-stabilized subgrade, as shown in Figure 4.11(b). Two asphalt binders were used in the asphalt layers: a PG 64-22 for base course, and an SBS PG 70-22 for polymer-modified binder course and wearing surface. The asphalt contents of the binder and base courses are 4.5%; while the asphalt content of wearing surface is 5.4%. No liquid anti-strips were used in any mixture. The aggregate used in all mixes is limestone. The subgrade is lime-stabilized to address the high water content existing in the natural soil. Longitudinal strain measurements were obtained at the stabilized subgrade–asphalt layer interface using an H-shape strain gauge. Temperature data was continuously collected using T-type copper–constantan thermocouples throughout the pavement depth.
(a) (b)
Figure 4.11 Cross sections of (a) thin asphalt pavement and (b) thick asphalt pavement
The FE solutions are compared with the field measurements to further check the accuracy of the model. Table 4.4 compares the measured and calculated longitudinal tensile strains at the bottom of the asphalt layer in the full-depth pavement section under the loading of two different tire configurations (690kPa, 8km/h, and 25ºC). Generally, the predicted tensile strains were found slightly greater than the measured strains. A good agreement of the response ratios caused by the two tire configurations is achieved between the predicted and measured results.
Table 4.5 compares the calculated pavement responses with the field measurements under the loading of a dual-tire assembly in the thin asphalt pavement section (35.5kN,
690kPa, 8km/h, and 30ºC). The pavement responses are calculated using various scenarios of contact stress distributions and material models. The use of 3-D contact stress distribution and the nonlinear model for the base layer and subgrade results in smaller differences between the measured and calculated stresses and strains, compared to the values obtained when the tire contact stresses are assumed uniform or the conventional linear isotropic model is used. It was found that, for the tensile strain at the bottom of the asphalt layer, the calculated strain is smaller than the measured strain, but the calculated responses in the base layer and subgrade are greater than the measured ones.
Table 4.4 Measured and Calculated Longitudinal Tensile Strains for Thick Asphalt Pavement Section (strain units: micro)
Tire configurations Field measurements FEM
35.5kN 44.4kN 53.3kN 35.5kN 44.4kN 53.3kN
Dual-tire assembly 89 105 119 102 115 131
Wide-base 455 tire 117 134 146 128 150 168
Ratios 1.31 1.28 1.23 1.25 1.30 1.28
Table 4.5 Measured and Calculated Pavement Responses for Thin Asphalt Pavement Section
Pavement responses FEM
Field
Tire contact stress Uniform 3-D
Base and subgrade model Nonlinear Linear Nonlinear Tensile strain at the bottom
of asphalt layer (micro) 370 324 397 529
Vertical stress at the bottom
of base (kPa) 68 66 63 44
Compressive strain on top of
The accuracy of FE analysis is affected by the material models used in the analysis. The discrepancies between the measured and the calculated pavement responses could be related to two reasons: 1) the nonlinear viscoelasticity of asphalt concrete was not considered in the analysis and these factors have more significant effects on the thin asphalt pavement responses with weak supports; 2) the measured tensile strains in the thin asphalt layer could be overstated because at the bottom of the thin asphalt layer the bending is more significant than tension. Considering the variability of as-built pavement thickness, environmental conditions, laboratory-measured material stiffness, and instrumentation responses, the discrepancies between the measured and the calculated responses were considered acceptable in this study for the purpose of comparing pavement responses at various tire loading conditions.
4.7 Summary
A 3-D FE model of flexible pavement is developed to analyze pavement responses under vehicular loading. This model utilizes implicit dynamic analysis and simulates the vehicular loading as a continuous moving load with three-dimensional contact stresses at the tire-pavement interface. In the model, the asphalt layer is modeled as a linear viscoelastic material and its relaxation modulus is converted from the laboratory- determined creep compliance data. The granular base layer is modeled as a nonlinear anisotropic material and its vertical, horizontal, and shear modulus is dependent on both the bulk stress and shear stress. A UMAT is developed to implement the constitutive model of the granular base layer in the FE model. The FE model results were compared with field measurements and the discrepancies were considered acceptable for the purpose of comparing pavement responses at various tire loading conditions.