The 10-variable regression equation presented above has some of the
variables in Equation 4.21. The log (D D ) and log (D D ) terms
1 4 1 2
were found to make the most significant contribution to the R2 value of
0.953 (39). It must be noted that D is in Equation 4.22. This sensor
deflection measurement is currently not made with the conventional
sensor array utilized by the FDOT (see Section 3.2.1).
4.3.3.3 Prediction Equations for E The modulus of the sta-
bilized subgrade layer, E3 does not contribute any significant change
on FWD deflection basins. As in the case of the Dynaflect, this layer
was found to be difficult to develop moduli prediction equations. From
the sensitivity analysis, the maximum percent change in deflections
occurred at FWD deflections 02 and D3 when the original E3 value was
doubled or halved (Tables 4.2 and 4.4). Therefore initial analyses
involved the examination of the relationships between E3 and D2 or D3
using log-log plots. Simple power law equations with R2 values greater
than 0.998 were generally obtained for various levels of E E E4 and
t These parameters were found to have considerable influence on the
intercepts (K ) and slopes (K2) of the power law relationships. Because
of the complex interactions involved, it was difficult to combine some
of the power law equations.
Multiple linear regression analysis was performed on a subset of
the theoretical deflections. An 8-variable prediction equation was
obtained from the analysis. This is shown in Case 1.
Case 1. For 150.0 < E < 300.0 ksi, 1.5 < t < 6.0 in.,
1 1
42.5 < E < 170.0 ksi, 30.0 < E < 75.0 ksi,
2 3
and 10.0 < E < 40.0 ksi,
4