286
E = 39.427(D )0.023(D )-1.026 Eqn. 6.16
4 6 7
(R2 = 0.999 and N = 22)
The percent prediction errors of Equations 6.14 to 6.16 were
generally below 10 percent. The maximum prediction error of Equa-
tion 6.14 was 7.0 percent while the other two had 9.0 percent pre-
diction errors. However, Equations 6.15 and 6.16 had greater pre-
diction accuracy than Equation 6.14, with prediction errors gener-
ally less than 4 percent.
The relationship between E4 and D6 or D7 is illustrated in
Figure 6.75. The corresponding equations (Equations 6.14 and 6.15)
apply to a wider range of E4 values than those obtained from the
theoretical analysis. Also, the slopes of Equations 6.14 and 6.15
are close to unity, approaching the format of the Dynaflect simpli-
fied E4 prediction equation.
It was initially believed (in Section 4.4.2.4) that the use of
Equations 4.28 or 6.16, which incorporate two sensor deflections,
should generally minimize the potential for prediction error due to
measurement variablity. However, it was found that variation in D
as high as 100 percent would have about one percent change in the
predicted E4 value with the use of Equation 6.16. Thus, Equation
6.16 is not very sensitive to changes in D6 as compared to that of
D7. Therefore, Equations 6.14 and 6.15 (Figure 6.75) should be
used for E4 predictions, and whenever possible, an average value
should be used. Where surface cracks exist, one of the equations
might be preferable to the other depending on the ability of the
pavement to transfer loads to the geophone locations.