6.73 therefore compares the relationship- between the results from that
study and those obtained from the test pavement sections used in this
study. From a practical standpoint, there is very little difference
between the Eq prediction equations. This difference is not significant
enough to warrant the use of one equation in preference to the other,
except when Do1 is less than 0.06 mils or much greater than 1.0 mil. It
was shown in Section 4.4.1.4 that the use of the simplified E4 predic-
tion equation result in the overprediction of weak subgrades (E4 < 10.0
ksi) and underprediction of high or stiff subgrades (E4 > 100.0 ksi).
The results obtained with the use of the modified Dynaflect system
indicate that separation of loaded areas produces double bending which
allows for the optimal placement of sensors to separate the response of
the different pavement layers. This somewhat unique load-sensor config-
uration makes it possible to develop simplified (power law) equations
for prediction of layer moduli. If desired, the predicted layer moduli
can be used as "seed moduli" in iterative elastic multi-layer computer
programs (e.g., BISDEF). This would help to ensure that unique
solutions are obtained. It has been demonstrated in Section 6.5 that
predicted E, and Eq values are reliable and seldom require much adjust-
ment or tuning to match the measured deflection basin. Therefore, it
appears that the most desirable approach in computer simulation is to
use E2 and E3 values as "seed moduli" for any iterative or judgment
modified analysis.
Appendix F describes a recommended testing and analysis procedures
using the modified Dynaflect testing system. The positions of the five
geophones in the conventional system were modified into the form shown
in Figure F.1. The computational algorithms of the recommended