Modification of Dynamic Modulus Predictive Models for Asphalt ...

LABORATORY TESTING DESIGN
Dynamic Modulus Test
The dynamic modulus tests were performed with a universal testing machine (UTM) in a
environmental chamber shown in Figure 28. The UTM has a hydraulically driven load frame
which can provide a maximum load of 25kN (5620lbs) at various frequencies. The magnitude
and frequency of applied load can be precisely controlled. The applied sinusoidal load was
carefully selected to maintain a sample strain level of 85 to 110με to ensure a measurable strain
and prevent excessive damage to sample. Excessive unrecoverable deformation can be achieved
if the strain is too high. The testing was conducted following ASTM D3496. The standards
requires
(height by diameter) asphalt cylinder cored from a
cylinder compacted by the Superpave Gyratory Compactor (SGC). Extra time and costs
are needed to core an asphalt cylinder. According to Robinette and Williams’ study (2006), the
dynamic modulus test results of cored and directly compacted samples are not significantly
different. More than 70 samples were made for this research including trial samples to determine
the required mass for target air void and samples for testing dynamic modulus. In order to lower
the cost and make the laboratory work more practical, the 150mm by 100mm cylinders in this
research were compacted by the SGC with a special mold designed specifically for samples in
this size. The sample strain was tested with three evenly spaced LVDTs attached on the side of
the sample.
The statistical design of dynamic modulus experiment is split plot. Dynamic moduli were
tested at 4°C , 21°C , and 37°C and nine frequencies (0.1, 0.2, 0.5, 1, 2, 5, 10, 20, and 25Hz) at
each temperature for each sample. Therefore, each sample was tested with 27 treatment
combinations. The 13 types of mixtures form the whole plots. Mixture type is the whole plot
factor. Five samples were made from each mixture to form the sub plots. The sub plot factor is
the individual cylinder. Thus, the variability caused by mixture type and treatments can be
determined. The variability between the five replications for each mixture can be also calculated
and separated from the total error, increasing the chance to identify the significant differences
between different types of mixtures.
53