Page 25 - MODELING OF ASPHALT CONCRETE

Modeling of Asphalt Concr ete 3

Future of Asphalt Concrete Modeling
The modeling of asphalt concrete is an evolving subject. Continuing developments and
improvements in computational power and test techniques will allow asphalt materials
and pavement engineers to use more realistic, powerful models to predict the
performance of asphalt materials and pavements. The following subsections attempt to
shed some light on the possibilities for future models of asphalt concrete.

Pavement Response Model versus Performance Model
A traditional approach to asphalt pavement performance prediction is divided into two
steps: pavement response prediction and pavement performance prediction. In this
approach, responses of an undamaged pavement (e.g., tensile strain at the bottom of the
asphalt layer) are estimated from a structural model (e.g., the multilayered elastic
theory) using initial, undamaged properties of the layer materials. Asphalt concrete
performance models are developed using laboratory test results and relate the initial
response of the asphalt concrete specimens to the life of those specimens. The responses
estimated from the structural model are then input to the performance model to
determine the life of the pavement. This approach is the state-of-the-practice method
that is adopted in most recent mechanistic-empirical pavement design methods,
including the Mechanistic-Empirical Pavement Design Guide (MEPDG) developed under
the NCHRP project 1-37A (2004).
These models are simple to use because the only measured response of the
mixture is at the initial stage of fatigue testing. Such models deserve credit for the basic
foundation of current mechanistic-empirical pavement designs. However, there are
several weaknesses in this traditional approach. First, damage evolution in complex
structures and modified materials may not be captured accurately. For example, complex
combinations of layer material types and thicknesses in perpetual pavements make it
more difficult to accurately predict the failure mechanisms using conventional hot mix
asphalt (HMA) performance prediction models and pavement response models.
Secondly, most performance models used in the two-step approach are mode-of-
loading dependent. These models are developed using results obtained from laboratory
tests, which are conducted either in controlled stress mode or in controlled strain mode.
Currently available two-step approaches do not have the ability to discern the mode of
loading in a mechanistic manner and, therefore, could result in an unreliable performance
prediction.
Thirdly, the laboratory test methods used in the traditional two-step approach are
designed to simulate the boundary conditions of pavement structures rather than to
define the material’s constitutive behavior in the representative volume element (RVE).
Often these laboratory test methods predict the performance under only some selected
pavement conditions. So, because the test methods simulate the pavement boundary
conditions rather than capture the behavior of the RVE, the number of tests needed to
cover the wide range of pavement conditions that are expected in the field is undesirably
large.
The weaknesses of the two-step approach can be overcome using a mechanistic
approach that combines HMA material models and the pavement response model. In
this approach, the material model describes the stress-strain behavior of the material in
the RVE. The material model is then implemented into the pavement response model
where boundary conditions of the pavement structure in question are applied. This