Prestressed/post-tensioned concrete bridges - Computational Analysis and Design of Bridge Structures - page 163

model (1982) is adopted in the AASHTO LRFD specifications (2013).
SFRAME (Ketchum and Scordelis 1986), developed in the University of
California at Berkeley in the 1980s as a comparison with VBDS, however,
adopts the original ACI 209 as its creep and shrinkage model. Although
the creep models are different between two numerical models, when some
parameters are taken as standards, the two creep models are very similar in
nature. One great shortcoming of applying the LRFD creep model is that the
maximum volume-surface area ratio used in the evaluation is limited to six,
while some of the structures may require a ratio over six.
The construction sequence is modeled in 41 stages to simulate the erec-
tion and tendon prestressing of each section. It takes one week for each
launching and prestressing. At day 100, the 18.3-m long girder at the side
span starts to be cast, and the side spans and the center span close at day
168 and 182, respectively. All prestressing tendons are jacked at a unique
stress of 1393 MPa (202 ksi), and the losses are taken as 15% of the jack
stress. Unlike SFRAME, all losses are simply treated to be a constant along
their path in this analysis by VBDS.
The time-dependent analysis for the 27 years following construction is
performed by a smart step adjustment. The basic step is one week. It will be
increased by one week whenever the differences of two adjacent analyses are
less than a designated threshold or will be decreased by one week if they are
above the threshold. Usually it varies between 1 and 12 weeks.
Table 5.5 shows the results and their comparison between VBDS and
SFRAME. The differences between two numerical solutions are checked.
Stresses of cases for maximum dual cantilever, ready to serve and 27 years
later are shown. Figures 5.40 through 5.43 show some screens captured
from VBDS; they show only the stress distribution on the top flange of the
box girder at the maximum dual cantilever stage, after secondary dead load
imposed, 27 years later, and on the stress envelop of HS-20, respectively.
The jagged stress plots shown in Figures 5.40 through 5.42 are caused by the
axial forces induced by the cantilever or local tendons. Jagged locations are
where tendons terminate. The live load stresses show the smoothness across
the whole girder. The live load analysis indicates that the live load stress
along the girder may be incorrect if it is calculated by using simple girder
principles based on its moment and axial force envelope. Unlike the dead
load, which is already distributed over a statically determined structure
before closure, the live load will cause significant axial force over the girder
(−6300 kN/4000 kN at the center of the main span) because the bridge
is fixed with two piers and the centroid of the girder shapes a flat arch.
Therefore, the main span behaves like an arch bridge. In this case, it may
not be sufficiently accurate to take the extreme moment and its correspon-
dent axial force or the extreme axial force and its correspondent moment to
calculate the stress over the girder in the main span. In VBDS, however, the