Page 133 - Mechanics of Asphalt Microstructure and Micromechanics

Experimental Methods to Characterize the Heterogeneous Strain F ield 125

Vertical displacements by DIC for the middle and bottom sections of a specimen sub-
jected to monotonic tension were compared to those measured by LVDT. A series of DIC
images captured during the monotonic and cyclic tests were used to study the evolu-
tion of the fracture process zone at the crack tip. The DIC examined deformation to a
random speckle pattern placed on the area of interest of the specimen, which yielded an
easier and faster specimen preparation method. Through careful manipulation of the
spray nozzle with proper pressure from a certain distance, a uniform-sized speckle pat-
tern was applied onto the surface randomly. The black paint was lightly sprayed sev-
eral times until the desired speckle pattern density was achieved. The optimum speckle
pattern density was an even mix of dark and white areas, producing a pattern that was
biased toward neither the white nor the black end of the gray scale. DIC demonstrated
a more accurate determination of the stress-strain behavior and the fracture process
zone. The applicability of DIC to a cylindrical specimen with a curved surface was
found to be more complicated than on prismatic specimens with a planar surface. The
DIC method was found to provide a more accurate constitutive relationship of the frac-
ture process zone than conventional LVDTs because of its full-field measurement and
post-processing nature.
Chehab et al. (2007) adopted DIC in developing a viscoelastoplasticity continuum
damage model (VEPCD) of asphalt mixtures. The formation of strains in the asphalt
mixtures was found to be highly localized as microcracks got densified, coalesced, and
further grew into macrocracks. Conventional LVDTs were found unable to capture the
localized process zone strain in the fracture process zone and ceased to accurately pre-
dict the performance of asphalt-aggregate mixtures after strain localization. DIC was
found to perform well in measuring the fracture process zone strains and could extend
the validity of the VEPCD model beyond localization. Methodologies that required
transfer from LVDT strains to DIC strains after strain localization for model calibration
and validation were presented for a range of loading and temperature conditions.
Romeo et al. (2006) developed a digital image correlation device to more accurately
capture localized or non-uniform stress distributions in asphalt mixtures and as a tool
to detect first fracture. The experimental analysis of asphalt mixture cracking behavior
was based on the HMA fracture mechanics visco-elastic crack growth law. Asphalt mix-
ture cracking mechanism and fundamental tensile failure limits were investigated us-
ing multiple laboratory test configurations, namely the Superpave Indirect Tensile
(IDT) test, the Semi-Circular Bending (SCB) test and the Three-Point Bending Beam
(3PB) test. Montepara et al. (2006) and Martin et al. (2006) extended Romeo’s work by
measuring more generally the elastic and plastic strains using the same device.
Birgisson et al. (2006, 2007, and 2008) used the DIC to verify fracture energy density
as a fundamental fracture threshold in HMA. Fracture energy density was evaluated
with the SCB test. The DIC method was based on the careful application of a sophisti-
cated image-matching technique (the least squares matching algorithm) with a proper
laboratory test configuration (specimen surface treatment, illumination, etc.). An image
sequence of the loaded specimen was acquired with a digital camera; a set of black pat-
terns artificially generated on the specimen surface was accurately tracked by the algo-
rithm along the loading sequence. The essence of the method was to compare the gray-
scale values of images of the deformed specimen with those of an initially un-deformed
specimen. The DIC was able to provide a dense and accurate displacement-strain field
of asphalt mixtures at the microstructural level. The resulting fracture behavior in the
SCB was predicted with a displacement discontinuity method to explicitly model the