122
Ex (Predicted) = 75.612 + 0.848E1 (Actual) Eqn. 4.31
(N = 97, R2 = 0.848)
Deleting the cases with E^ equal to 0.35 and 200 ksi resulted in
Ex (Predicted) = 28.707 + 0.989E1 (Actual) Eqn. 4.32
(N = 90, R2 = 0.934)
The improvement in R2 value suggests that the Ex prediction equa
tions for Case 2 should be used with caution when E, values are
extremely low or high.
Case 3: 10.0 < E < 85.0 ksi; and 7.0 < t < 10.0 in.
2 l
For this case, Equations 4.1, 4.6, and 4.7 were derived to predict E .
As mentioned previously, these equations appear to be simple compared
with those for Cases 1 and 2. The relatively simple Ex prediction
equations for Case 3 were developed using data for tx = 8.0 in. only,
but it was found to be applicable for thicknesses between 7 and 10 in.
This also suggests that for thicker pavements, the effect of tx on Ej,
E2, and becomes negligible, as shown in Figures 4.17 and 4.18.
The percent difference between actual and predicted Ex values for
Case 3 were within +6 percent. Only three pavements exceeded this,
having less than +10 percent error. When the predicted and true moduli
values were regressed, the following equation was obtained.
E1 (Predicted) = 2.896 + 0.991E1 (Actual) Eqn. 4.33
(N = 22, R2 = 0.998)