Page 346 - Mechanics of Asphalt Microstructure and Micromechanics

T
338 Ch a p t e r e n

make samples in batches. Mixtures were made by controlling the weight of each compo-
nent proportionally. The weight of each component was used to calculate the aggregate
volume fractions.
10.2.2.2 Sample Scanning and Testing
The SkyScan microtomographic system and testing stage (SKYSCAN 2006) were used
in this study to obtain the microstructure of small samples and to conduct the uniaxial
compressive test. The X-ray source is 20 − 50 kV with maximum power of 40 W. The
pixel size is from 6 to 30 μm. The material testing stage can perform compression, ten-
sion, and torsion tests. The testing stage applies displacement to the top and bottom of
a sample. Samples with aggregates were scanned before testing. For these small sam-
ples, the whole sample was scanned with one rotation using the cone beam mode. The
resolution of the reconstructed images was 512 512 pixels. All samples were tested at
room temperature under uniaxial compression. The testing stage is displacement con-
trolled and the speed of the displacement was set at 15 μm/s. When the resistant force
reached the maximum allowable value of the testing stage, the application of displace-
ment was stopped. For the small sample used in this study, the application of displace-
ment was manually stopped at a stress of 0.8 MPa, i.e., 22.64N force on the sample.
Prior to each test, the testing stage was calibrated.

10.2.2.3 Sample Reconstruction, Test Simulation, and Parameter Back Calculation
After scanning, sectional images of the small sample were reconstructed. The recon-
structed sectional images were used to conduct 3D analysis, which determines volume
fractions for aggregates, binder, and voids. They were used to obtain the threshold val-
ue that will be used in the generation of the FE model. 3D visualization was carried out
to generate the digital representations of these physical samples.
Reconstructed sectional images were read with programs developed in Interchange
Data Language (IDL) and transferred into a data file read by a FORTRAN program to
generate the required inputs for the FEM. In this study, adaptive meshes were used for
asphalt binder to account for large distortions that may occur within thin asphalt layers
between two aggregates.
The simulation of the uniaxial compression test was conducted as a small-scale dy-
namic modulus test. The two-layer viscoplastic material model was used for asphalt
binder and an elasticity model was used for aggregates. Loading plates were modeled
as rigid surfaces. The movement of the top rigid surface was controlled with a down-
ward displacement velocity of 15 μm/s, which is the same as that in the physical test.
Parameters in the material model were estimated with back calculation based on a
parameter sensitivity study. Material parameters were calibrated from one set of sam-
ples. Then, the same set of material parameters was used in the simulation of other
samples with different aggregate volume fractions.
Figure 10.12 shows samples being scanned within the testing stage and a recon-
structed 3D image. The height of the sample can be measured on the screen. In the 3D
image, the black phase is aggregates.
Through reconstruction of the 3D microstructure and inclusion of the microstruc-
ture into FEM simulation (the digital test, Figure 10.13), the compression test can be
modeled. In this model, the asphalt binder is modeled as viscoplastic (Kichenin et al.,
1996); aggregates are modeled as elastic. In this study, specimens of the pure binder, 5%
aggregates (by volume of aggregate binder mixture), 25% aggregates, and 50% aggre-
gates were tested. In each case the mastic microstructure is scanned and represented in