mum density condition, the model predicts
lifespans of the sections when judged by the hori-
28,585,000 applications to failure. In case f4w6ld,
zontal strain criteria (Table 15).
the same test section with a low density value for
Results from the Phase 2A series that used the
the 1206 subgrade, the model predicts 1,905,000
Ullidtz model for asphalt modulus calculations
applications to failure. And finally, in case f4w6ss,
and a recovered summer subgrade modulus of
103,00 kPa (15,000 lb/in.2), a close approxima-
with the 1232 subgrade that produced lower resil-
tion of the value backcalculated from FWD mea-
ient modulus values in lab tests, failure is pre-
surements on subgrade during fall 1991, yield per-
dicted after only 72,000 applications, or in less
formance predictions that are as "fine-tuned" as
than one year.
possible, so far (Table 15 "Ull-15K"). When these
After the Phase 1 simulations were complete,
Mechanistic Pavement Design Procedure predic-
the resilient modulus data from the 1206 subgrade
tions are compared with the originally designed
were reviewed. A calibration error of unknown
failure at 3,300,000 applications for the 5-year sec-
magnitude was discovered in the resilient modu-
tions and 6,600,000 applications for the 10-year
lus equipment used to measure the unfrozen modu-
sections, the following statements can be made:
lus on that material. Modulus testing of the 1206
1. Predictions for the full-depth sections, both
subgrade in the frozen condition and all testing on
5 and 10 year, indicate that they will not fail from
the other materials were conducted on a different
asphalt cracking, but two of the three criteria for
machine, which passed its calibration checks. Com-
subgrade rutting indicate early failure;
paring frozen and unfrozen resilient modulus test
2. Conventional sections are predicted not to
results from the 1206 subgrade with those from
fail due to subgrade rutting; however, sections
other materials tested, Figure 7, indicates that the
unfrozen Mr values for the 1206 subgrade may be
with more frost-susceptible bases are predicted to
about an order of magnitude too high. A lower Mr
fail because of asphalt cracking relatively early
for the 1206 subgrade would have resulted in ear-
in their design life, and sections with non-frost-
lier failures than computed in this report where
susceptible bases are predicted to fail towards the
the 1206 optimum density subgrade was used.
end of their design life.
We are currently investigating the unfrozen
By modeling 21 years of environmental condi-
1206 subgrade results and will include findings
tions in the Phase 3 series, we were hoping to find
and revised performance predictions in Berg (in
the characteristics of a "most severe" winter that
prep.)
could be used for design purposes. The results
It is obviously extremely important to use the
were complex and a relation between predicted
representative subgrade conditions in the design
failure and characteristics of freeze seasons was
simulations. Results are consistent with observa-
elusive, at best.
tions from in-service pavements; e.g., weak areas
Based on the above results, we recommend the
fail much more rapidly than strong ones and high
following studies:
water tables cause failures before similar pave-
1. The class 4 special and class 5 special base
ments constructed over lower water tables. Both
materials should receive the full complement of
the 1206 and 1232 subgrade samples were ob-
laboratory tests so that simulations may be run
tained from the test site and were located less than
using properties of actual materials from the site
one-half mile apart, yet their tested behavior was
rather those of substitute materials.
quite different.
2. Once performance data are received from
Another important aspect governing the results
Mn/ROAD, predicted and measured values of
of the Mechanistic Pavement Design Procedure is
moisture, temperature, and strains should be com-
the method used to calculate the asphalt concrete
pared.
resilient modulus. The use of the Schmidt model,
3. Use performance data from Mn/ROAD to
which produces extremely low summer season
revise, refine or develop new damage models for
moduli, partially accounts for the short lifespans
rutting and fatigue cracking.
predicted in the Phase 1 modeling series. Replace-
4. Use performance data from Mn/ROAD to
ment with the Ullidtz model in part of the Phase
refine or develop a new model for the change of
2A modeling series increased the predicted
the asphalt modulus with temperature.
34
Previous Page
