by G. Fenves and R. Desroches
March 1995, 132 pp.
Click on the link below for the full text:
Abstract
The response of the Northwest Connector at the I-10/215 Interchange in the
1992 Landers and Big Bear earthquakes was recorded by a strong motion
instrumentation network. The Connector is a 2540 ft. curved bridge constructed
in 1973 and retrofitted for improved earthquake performance in 1991. Although
the peak ground acceleration was approximately 0.10 g at the site for both
earthquakes, much less than the level of ground motion expected in a major
earthquake, the strong motion data provides important information about the
earthquake response of a typical curved freeway bridge. The objectives of this
study are to: evaluate the importance of non-uniform support motion on the
response of the bridge; determine the vibration properties of the bridge;
determine efficacy of typical modeling and dynamic analysis techniques used in
the design of bridges to predict the response recorded in the earthquakes; and
examine the role of the intermediate hinges on the earthquake response of the
bridge.
The input motion was fairly uniform in the two earthquakes. Consequently,
the spatial variation of free-field input motion is not important for these cases.
The earthquake response of the Connector is predominantly transverse. Bent 8
near the center of the bridge is the most heavily instrumented bent. The
maximum transverse deformation of the column was 4.76 in. for Landers
(drift=0.86%), and 2.98 in. for Big Bear (drift=0.54%). Pile cap rotation produced
10 to 15 percent of the column displacement in the transverse direction.
The strong motion records show the effects of pounding as evidenced by
large acceleration spikes at the five intermediate hinges in the Connector. Using
the processed displacement records, Hinge 7 had the largest opening: 1.41 in. for
Landers and 1.70 in. for Big Bear. Several of the closing excursions have the high
frequency oscillation associated with pounding. Accumulated crushing of the
filler material in the hinge may have occurred in the second earthquake. In the
transverse direction, the relative hinge displacement is constrained by a shear
key. The shear keys were effective in limiting the relative displacement to a
maximum of 0.61 in. The maximum relative transverse displacements,
however, were greater than the nominal 1/4-inch gap, indicating that the shear
keys likely suffered local crushing of filler material.
The most significant finding is the difference in the fundamental mode
period of the Connector in the Landers and Big Bear earthquakes. The fundamental
period lengthened from 1.56 sec in Landers to 1.75 sec in Big Bear. The other noticeable
difference between the two earthquakes is that the damping ratio increased in the
first and third mode, although it decreased in the second mode. The lengthening of
the vibration periods and generally increased damping may indicate that the bridge
"softened" in the Landers earthquake. Changes in the pile foundations appear to be
the most likely cause, although reduction in stiffness of the steel jacketed columns
is a contributing factor in the lengthening of the vibration periods.
Standard three-dimensional "stick models" are used in addition to a
nonlinear model of the intermediate hinges. The gross flexural rigidity (EI) for
the columns, increased 10 to 15 percent for the steel jackets, are then modified
to match the vibration periods of the model with the identified periods of the
Connector in the two earthquakes. For the Landers model, the modification
factor is 1.05; for the Big Bear earthquake the modification factor is 0.85. The
reduction of the column stiffness between the two earthquakes, along with the
reduced rotational stiffness for the pile foundations, accounts for the lengthened
vibration periods identified from the recorded motion. With these assumptions,
the comparison between the model motion and recorded motion is good in
many aspects. Although there are some differences, the comparison provides
confidence that standard analysis techniques are adequate for design with proper
assumptions about column and foundation flexibility. However, the modeling
of hinges using gap elements does not provide accurate high frequency response
due to pounding of adjacent frames.
The analysis shows that the most heavily loaded column was Bent 9 in the
Landers earthquake, with a bending moment that was 73 percent of the ultimate
flexural capacity. The nonlinear model showed the restrainers developed a
maximum stress of 97 ksi, less than the yield stress. The hinge pounding
included in the nonlinear model caused a moderate increase in some column
forces compared with the linear tension and compression models which neglect
pounding.
|