283


from the 8.0 in. used in the theoretical study to a high value of 24.0

in. (SR 715). Since these values represent actual test pavements, the

resultant prediction equations would be more reliable than those of

Section 4.3.3, unless of course their prediction accuracies are lower.

 6.7.3.2.1 Asphalt concrete modulus, El. Multiple linear

 regression analysis of the test data resulted in the following

 equation:


 log E, = 3.229 1.0683 log (tl) 2.8217 log (Di D2)


 + 1.008 log (Di D3) + 0.8835 log (D1 05)
 Eqn. 6.10

 (R2 = 0.885 and N = 22)


 Error analysis indicated that 5 out of 22 pavements had pre-

 dictions with errors greater than 20 percent with one pavement

 having as high as 44 percent prediction error. This pavement was

 SR 12 which had an asphalt concrete thickness of 1.5 in. As

 explained in Section 4.4.3, the SR 12 pavement was deleted from the

 data base and the remaining data analyzed to obtain Equation 6.11.


 log E1 = 2.215 0.2481 log (tl) 12.445 log (D1 D2)


 + 17.205 log (Di D3) 5.871 log (Di D4)
 Eqn. 6.11


(R2 = 0.959 and N = 21)