(App. B). ML5-R-5 experienced rapid increases
Schmidt model was used. This series also used the
in damage during the thaw periods, but also had
high density 1206 form for the subgrade modulus
an equivalent increase in damage over the sum-
calculations. Results for this series are shown in
mer months. The other sections experienced a
Table 15 in rows labeled "Ullidtz." In nearly all
majority of their damage during the thaw periods
cases, this series produced higher predicted appli-
and very little damage during the summer.
cations to failure than the corresponding Schmidt
The low damage predictions result from the
series.
fact that the model predicts damage at the center
The third series also used the Ullidtz model to
of a slab, whereas the majority of damage occurs
calculate the asphalt modulus. In addition, it ad-
at the slab edges. Chou (1989) recommends in-
dressed the problem from the Phase 1 simulations
creasing the stresses computed from the layered
that during the summer the predicted subgrade
elastic method by a factor of 1.33 to estimate the
moduli based on the high-density 1206, or "nor-
coverage levels for roadway pavements.
mal" form were higher than FWD-measured val-
ues. For this series, the unfrozen subgrade modu-
lus, when recovered, was set to be constant at
Phase 2A
103,000 kPa (15,000 lb/in.2). This value was cho-
To address problems discovered in the Phase 1
sen as being approximately equal to an average
modeling series, three additional simulation se-
value back-calculated from FWD measurements
ries were conducted on the eight flexible test sec-
tions studied in Phase 1 with the water table lo-
on subgrade during fall 1991.* As a result of this
cated at the shallowest position. A change that
modulus approximation, the linear model (model
applied to all three series was that the normalized
0 in Table 4) was used in NELAPAV for all
volumetric unfrozen water content form of the
subgrade calculations. The applications to failure
frozen modulus equations (Table 10a) was used,
from this series are listed in Table 15 as "Ull-
rather than the normalized gravimetric form used
15K." In the conventional sections including a
in Phase 1.
base/subbase, the applications to failure were
The first of the new series was essentially a
slightly lower than those resulting from the previ-
repeat of the initial series in that it used the Schmidt
ous case with the Ullidtz asphalt model and the
model (eq 8) for calculation of the asphalt modu-
higher-modulus subgrade; in the full-depth sec-
lus and the high density 1206 form of the subgrade
tions, they were quite a bit lower.
for modulus calculations. It differed from the ini-
A comparison of the Ullidtz/15K subgrade se-
tial series in that the normalized volumetric unfro-
ries results with the originally designed failure at
zen water content form of the frozen modulus
3,300,000 applications for the 5-year sections and
equations was utilized. A mistake in the CUMDAM
at 6,600,000 applications for the 10-year sections
program was also remedied--correcting the value
yields the following conclusions:
of the subgrade modulus used in the Corps of
1. In general, the Mechanistic Pavement De-
Engineers vertical strain damage calculation. Table
sign Procedure predicts the 5-year conventional
15 lists the results of this series in the rows la-
sections to fail due to asphalt cracking prior to the
beled "Schmidt." In general, predictions of appli-
end of the design life, while failure from subgrade
cations to failure in the conventional cross sec-
rutting is predicted beyond the design life.
tions were less than in the original series using the
2. Predictions for the 5-year full-depth section
horizontal strain criteria and greater using the ver-
indicate that it will not fail from asphalt cracking,
tical strain criteria. In the full-depth sections (F4
but two of the three criteria for subgrade rutting
and F14), applications to failure were increased in
indicate early failure.
both the horizontal and vertical strain criteria.
3. The two 10-year conventional sections with
The second series addressed the issue that the
class 5 and class 6 special bases (F21 and F22) are
Phase 1 simulations computed summer asphalt
predicted to fail from asphalt cracking at about
moduli that were lower than considered reason-
the end of the design applications, while subgrade
able. In this series, the Ullidtz model (eq 9) was
rutting criteria indicate longer life.
used to calculate the asphalt modulus at tempera-
tures greater than 1C; at colder temperatures, the
* D. Van Deusen, Mn/ROAD, pers. comm. 1992.
25
Previous Page
