Characterization and implementation of stress dependent resilient modulus of asphalt treated base for flexible pavement design

Abstract

Asphalt treated base (ATB) is the most commonly used type of stabilized material in pavements because of material availability and relatively low cost in Alaska. The treatment enhances the material's properties to overcome deficiencies in some marginal materials. Resilient modulus (MR) of these materials is an essential pavement design input. Currently, in the Alaska Flexible Pavement Design (AKFPD) Manual, MRS of ATBs were back calculated using testing results of falling weight deflectometer (FWD). There is a need for an accurate laboratory characterization of these materials. In this study, the MRS of hot asphalt treated base (HATB), emulsifed asphalt treated base (EATB), foamed asphalt treated base (FATB), and a mixture of reclaimed asphalt pavement (RAP) and D-1 aggregate at a 50: 50 ratio (RAP 50:50) were measured using repeated triaxial tests. D-1 granular materials used for base course construction were collected from three regions in Alaska. HATB specimens were compacted using Superpave gyratory compactor and three binder contents were used: 2.5%, 3.5% and 4.5%. EATB and FATB specimens were compacted according to ASTM D1557 and three residual binder contents were used: 1.5%, 2.5% and 3.5%. RAP 50:50 was also compacted according to ASTM D1557 and no additional additives were added. MR was measured at three temperatures (i.e. -10°C, 0°C, 20°C for HATB, EATB and FATB; -10°C, -2°C, 20°C for RAP 50:50). The stress-dependent property of MR was successfully characterized by the modified universal soil model, in which the MR was expressed as a function of bulk stress (θ) and octahedral shear stress (τoct). Generally, MR increased with an increase of θ and decreased with an increase of τoct. Stress-dependent patterns of each type of ATB were analyzed and discussed. Predictive equations for MR were developed for all types of ATBs investigated in this study. The equations were based on the modified universal soil model. The material properties (i.e. binder content and percentage fracture surface), temperature and the interactions among them were incorporated into equations. The developed predictive equations had very high coefficient of determination (R²). The R² s of equations HATB_10, EATB_10, FATB_10 and RAP_9, in which the influencing factors and second order interactions among factors were included, were all greater than 99%. These equations can be also used to estimate nonlinear elastic constants of ATBs in the modified universal soil model (i.e. k₁, k₂ and k₃). The stress dependent property of MR was incorporated into pavement structural analysis using the finite element method (FEM) program Abaqus through user defined material that was programmed in the user subroutine. Comparisons were made between pavement responses obtained from nonlinear FEM and traditional linear elastic layered system. The representative MR of ATBs were determined and recommended based on the equivalent critical pavement response of the typical Alaska flexible pavement structure. Predictive equations were developed to estimate the critical pavement responses. The equations were developed through regression analyses using a database generated from 16,848 nonlinear pavement FEM analyses, which covered a variety of pavement structure combinations. These nonlinear pavement analyses were implemented through the function of a parametric study provided in Abaqus FEM package. In total 9 independent variables were included, which were the thickness of the surface course, base course, and subbase, moduli of HMA, subbase and subgrade, and nonlinear elastic constants of ATB (i.e. k₁ k₂ k₃) in the MR model. The interactions among these variables were also included. The R²s of predictive equations were at least 0.9725. The predictive equations can be used for routine pavement analysis and design purposes.