On Cross-Anisotropy of Flexible Pavement Layer Materials for Improved Pavement Responses

Pavement layer materials are typically assumed to be isotropic in determining pavement responses, such as stress and strain. The main benefit of the isotropic assumption is that stiffness defined by modulus of elasticity (E-value) is equal in all directions. In reality, stiffness along vertical and horizontal (i.e., and ) may vary due to density gradients in compacted layer materials caused by vertical compaction during construction. Therefore, pavement layer materials with different vertical and horizontal modulus (i.e., ≠ ) are cross-anisotropic. Past studies have reported that degree of cross-anisotropy, ( ) ranges from 0.2 to 0.85 in Asphalt Concrete (AC) layers. Yet, the presence of cross-anisotropy is not considered in pavement layers for determining stress and strains, which are used for prediction of pavement distresses such as rutting and fatigue cracking. Ignoring the presence of AC cross-anisotropy may lead to an error in stress-strain calculations, and in pavement distress value predictions. To this end, this study examines the effects of cross-anisotropy on pavement stress-strain and thereby, evaluates the importance of cross-anisotropy in pavement design. In addition to cross-anisotropy, AC modulus is temperature and frequency dependent. Pavement temperature varies not only at different times over a day but also at different seasons over a year. Also, temperature is not constant over the depth of an AC layer which leads to a non-homogenous distribution of AC stiffness. AC modulus varies due to different loading frequencies caused by variable speed of moving vehicles. Unbound layers such as base and subbase are stress-dependent. The stiffness of an unbound layer differs at different months over a year due to varying moisture contents. This study considers all of these factors such as temperature, frequency, and moisture with material cross-anisotropy through laboratory testing and finite element modeling. In this study, a dynamic Finite Element Model (FEM) is developed based on the geometry of an instrumented pavement section at Milepost (MP) 141 on I-40, Rio Puerco, New Mexico and deflections, stress, and strains are predicted under truck tire pressure. Two different material models are employed: (i) generalized cross-anisotropic and temperature dependent viscoelastic model, and (ii) nonlinear elastic and stress-dependent model. The first one is for the AC layer whereas the second one is for unbound layers such as base and subbase courses. In this study, dynamic modulus tests were conducted in the laboratory on field-compacted vertical and horizontal AC cores to determine the parameters required for developing the generalized cross-anisotropic viscoelastic model. Resilient modulus tests were conducted on granular aggregates collected from the base and subbase layers to develop nonlinear elastic and stress-dependent material model. After integrating these material models, the dynamic FEM is simulated under a Falling Weight Deflectometer (FWD) test type load (79.6 psi over a circular area with 6 inch radius) for model validation. The simulated responses are compared to in-situ deflections, stress, and strains under a FWD test. It is observed that the simulated pavement responses are close to in-situ responses. Using the validated model under non-uniform vertical tire-pavement contact stress, FEM simulations are run at varying n-values and temperatures to determine tensile strain at the bottom of AC layer and vertical compressive strains in all layers. It is observed that strains are sensitive to material cross-anisotropy and strains with cross-anisotropy assumption are greater than those with isotropy assumption. In particular, both horizontal tensile and vertical compressive strains increased in case of AC cross-anisotropy, whereas only vertical strains increase in case of unbound layer cross-anisotropy. These strain variations due to cross-anisotropy increase at high temperature and in presence of stress-dependent unbound layer. In order to determine pavement distresses, simulated strains are used to determine damage due to fatigue and permanent deformation using Miner’s rule. In case of AC cross-anisotropy, damage decreases as n-value increases towards isotropy ( ). However, unbound layer cross-anisotropy leads to an increase in fatigue damage as n-value increases towards isotropy. It is also observed that damage due to AC cross-anisotropy is very high compared to unbound layer cross-anisotropy. It indicates that earlier pavement damage is mostly caused by the AC cross-anisotropy. That is, AC cross-anisotropy is more important than the unbound layer cross-anisotropy when considering rutting and fatigue damage only. This study recommends to incorporate the AC cross-anisotropy, at a minimum, in pavement analysis and design.PhD, Civil EngineeringDoctoralUniversity of New Mexico. Dept. of Civil EngineeringTarefder, RafiqulNg, Tang-TatMaji, ArupShen, Yu-Li