Dynamic modulus is the key property used to characterize stiffness of asphaltic mixtures in pavement performance evaluation programs such as the Guide for Mechanistic–Empirical Design of New and Rehabilitated Pavement Structures. This paper investigates various models for predicting the dynamic modulus of asphalt mixtures and compares model predictions with experimental test results. The predictions of two semi-empirical models (Witczak's model, modified Hirsch model), an analytical micromechanics model (Hashin's model), and the computational micromechanics model are compared with the dynamic modulus test results obtained from cylindrical asphalt concrete specimens. For the computational micromechanics approach, the finite element method was incorporated with laboratory tests that characterize the properties of individual mixture constituents and with a digital image analysis technique to represent detailed microstructure characteristics of asphalt concrete mixtures. All predicting models investigated in this paper are in fair agreement with the test results. Witczak's equation simulates dynamic moduli somewhat greater than laboratory test results, whereas the modified Hirsch model generally underpredicts moduli. The computational micromechanics model presents a relatively higher deviation at lower loading frequencies, but it shows better predictions because the loading frequency is higher. Hashin's analytical micromechanics model is limited to accurately predicting the dynamic modulus of the asphalt mixtures because of geometric simplifications and assumptions. With further improvements, the computational micromechanics method incorporated with the testing protocol seems attractive, because it can directly account for geometric complexity due to aggregates and inelastic mixture component properties with fewer of the required laboratory tests.

No abstract

This study presents a multiscale computational model for predicting the mechanical behavior of asphalt mixtures. The model can account for mixture heterogeneities by considering individual mixture constituents through the scale-linking technique: a local scale in a form of the heterogeneous representative volume element and a global scale that has been homogenized from local scale responses. The model is implemented with a finite element formulation, so that geometric complexities, material inelasticity, and the growth of time-dependent damage can be properly handled. Damage is in the form of cracks modeled with nonlinear viscoelastic cohesive zones. The primary purpose of this paper is to present the multiscale modeling framework developed and to evaluate the applicability of the multiscale modeling technique to determine the performance of asphalt mixtures and structures when damaged. This is accomplished by employing only material properties at the constituent level (local scale) as model inputs. The indirect tensile test of fine-aggregate matrix mixture is simulated as an example, and the simulation results are compared with experimental results to evaluate the applicability of the model. Predictive power of the model and the benefits related to the reduction of computational efforts and laboratory tests are further discussed.

No abstract

This paper presents an experimental verification of geometrically defined representative volume elements (RVEs) of heterogeneous asphalt concrete mixtures before any significant damage is initiated. A typical dense-graded Superpave mixture (12.5 mm nominal maximum aggregate size) is selected as a representative roadway paving mixture and used in this study to accomplish two parallel approaches: Geometrical analysis of mixture heterogeneity using two-dimensional actual images of asphalt concrete inner structures and experimental evaluation through uniaxial tensile tests of asphalt concrete mixtures incorporated with digital image correlation (DIC) technique. To properly address the significant heterogeneity of asphalt concrete mixtures in defining the RVE, several geometrical factors such as area fraction, gradation, orientation, and the distribution of aggregate particles in asphalt concrete mixtures are considered altogether. For the uniaxial tensile test with the DIC, the mean strains and their standard deviations captured by DIC are analyzed to confirm statistical homogeneity of RVEs evaluated from the geometrical analyses. The two approaches present similar results, indicating that typical dense-graded asphalt mixtures can be characterized for their material properties with an approximate RVE size of 60 mm. Findings from this study further imply that the simple geometrical analysis can be an efficient tool to reasonably determine the RVE of asphalt mixtures and other granular composites where significant heterogeneity is involved.