Page 234 - MODELING OF ASPHALT CONCRETE

212 Cha pte r Ei g h t

9. Although a material under repetitive loading after the resilient conditions may
experience mainly elastic strains, Fig. 8-2, the concomitant microcracking and
fracture before and after the resilient condition can be affected by the history of
plastic strains or work that accumulate during the preresilient states. Hence, a
mechanistic model should be able to incorporate the accumulated plastic strains
or work on the subsequent response involving permanent strains (rutting),
microcracking, and fracture.
10. In the pavement literature (e.g., Witczak and Uzan 1988), it is claimed that the
RM approach can characterize the behavior of geologic (unbound) materials. In
fact, it has been noted by many investigators that such nonlinear elastic models
cannot represent realistic behavior of unbound (geologic) materials because they
are affected by factors such as plastic deformations, stress path, volume change,
type of loading, and in situ conditions (Desai and Siriwardane 1984; Desai 2001).

Other Models
In addition to the RM approach, a number of other (semi) empirical approaches are
used in pavement analysis for the evaluation of important distresses such as rutting,
damage, and fracture. These approaches are based on the computed stresses and strains
at selected locations in the pavement, often obtained by using layered elastic analysis or
nonlinear elastic finite element procedures. In conjunction with empirical factors based
on (field) observations, such approaches can sometimes provide reasonable predictions.
However, they may not be considered mechanistic because they do not involve
calculation of distresses based on evolving stresses and strains in the multidimensional
pavements, as affected by nonlinear material response.
In addition to the linear and nonlinear elastic models, plasticity models have often
been used for pavement materials. Plasticity models can include classical (e.g., von
Mises, Mohr-Coulomb, and Drucker-Prager) and enhanced (e.g., continuous hardening
or yielding: critical state and Cap, Vermeer and hierarchical single surface—HISS)
models. Although these models can provide improvements, particularly regarding
prediction of permanent deformations, they are not directly capable of handling other
important factors such as microcracking and fracture.
The SHRP project (Lytton et al. 1993) employed the viscoelastic models used
commonly for pavement materials such as asphalt (Schapery 1965, 1999), and other
models for plasticity, fracture, and damage. Each of these models was essentially an
independent one, and the overall model for the combined (elastic, plastic, creep, damage,
and fracture) responses may be considered to represent a combination of models. As a
consequence, the overall model can be complex and involve a large number of parameters,
many of which did not have physical meanings. In other words, they were not related to
specific states of material behavior, and their determination involved mainly curve
fitting procedures. In short, the overall model used may be unrealistic for rational and
realistic modeling of the unified behavior of bound and unbound (geologic) materials. It
was suggested that the RM model can be used for the unbound materials. As explained
before, this may not be realistic because previous research for geologic materials showed
that nonlinear elastic (resilient modulus) models cannot represent the actual behavior to
allow for factors such as plastic strains, volume change, stress paths, and repetitive
loading (Desai 1998b, 2001). It is believed that the approaches based on the combination