266
6.7.2 Analysis of Dynaflect Tuned Data
6.7.2.1 Comparison of Measured and Predicted Deflections. It was
mentioned in Section 6.5.2 that the measured and predicted Dynaflect
deflections (Figures 6.15 to 6.31) were generally in good agreement for
the test pavement sections. The measured and predicted deflections
listed in Tables 6.1 and 6.8, respectively, were regressed to evaluate
the reliability of the BISAR predicted Dynaflect deflections. The
regression analyses utilized only sensors 1, 4, 7, and 10 deflections.
These were the sensor deflections selected from the theoretical analysis
(Chapter 4) to be related to the moduli of specific layers.
Figures 6,66, 6.67, 6.68 and 6.69 present the relationship between
BISAR predicted and field measured Dynaflect deflections at modified
sensor locations 1, 4, 7, and 10, respectively. In all cases, the high
R2 value (R2 > 0.96) indicated an expectionally good correlation between
predicted and measured deflections. The regression equations for D^ and
Dy (Figures 6.67 and 6.68) provided an almost perfect correlation with
the intercept and slope being within 0.015 mils of zero and 0.018 mils
of unity, respectively.
The Dx values (Figure 6.66) tended to yield a slightly higher
intercept (0.065) and slope (1.107) which results in the predicted
deflections being slightly greater than those measured. There were four
test sites where predicted Dx values were about 0.2 to 0.3 mils greater
than the measured D1 values. This difference may be due to sensor
placement variation, the use of single Dx measurement in the earlier
tests, variation in measured Dx response according to wheel positioning,
or where complete tuning was not achieved (e.g., SR 24 and SR 12).