54
mm mixes with optimum and optimum minus 0.5% AC, respectively. The figures show a
fairly consistent trend, i. e., the lower the air void content, the higher the axial dynamic
stiffness is.
5.4 Analysis of axial stiffness
This section deals with analysis of axial stiffness data. The sensitivity of axial
stiffness to various mix and test parameters is investigated using statistical analysis, and
surrogate models are developed for the prediction of axial stiffness and axial loss
stiffness. The models presented herein, as well as in the following sections were utilized
for the development of fatigue models and pavement analysis and design for the fatigue
distress.
5.4.1 Surrogate models for axial stiffness
The axial stiffness model development procedure followed was similar to that
used for fatigue characterization. The models presented in this section are the general
models for axial stiffness | E*|, axial loss stiffness E", and axial stiffness
Hz
E
10
* at 10 Hz
frequency.
Table 5- 2 through Table 5- 4 provides summary of regression analysis results for
the various models. The axial stiffness models based on GLM are as follows:
At 10 Hz frequency:
17.5153 105 exp( 0.03956 0.01256 0.31472 0.11671 ) 2 0.94
10
E* = ´ AC + GR - Temp - Va R =
Hz
( 5.2)
For variable frequency:
E* = 9.9535 ´ 105 exp( 0.08946AC + 0.0368GR - 0.46242Temp - 0.15345Va) × ( Freq) 0.35152 R2 = 0.96 ( 5.3)
E" = 5.7278 ´ 105 exp( 0.09032AC + 0.20616GR - 0.41168Temp - 0.13566Va ) × ( Freq) 0.35152 R2 = 0.91 ( 5.4)