Professional Documents
Culture Documents
Seismic Performance of Highway Bridges
Seismic Performance of Highway Bridges
highway bridges
A. Ghobarah and H. M. Ali
The 1971 San Fernando earthquake was a major turning damping, which are the two basic elements required for
point in North America in the development of seismic effective base-isolation.
design criteria for bridges, in much the same way as the The objective of this study is to investigate the seismic
1933 Long Beach earthquake was for the earthquake behaviour of the isolated and non-isolated bridges when
response of buildings. Similarly, the 1929 Murchison subjected to earthquake ground motions with different
earthquake in New Zealand altered the view prevailing intensity and frequency content. In the non-isolated
then that earthquakes were of scientific interest only. bridges, the energy is dissipated by allowing the piers to
Although significant advances have been achieved since behave inelastically. The impact of various design para-
that time in the design and construction of an earthquake- meters, such as the lead yield force and the location of
resistant bridge, numerous gaps still remain in the under- the lead plugs on the performance of base-isolated
standing of the seismic behaviour of bridges. highway bridges, is evaluated.
It has been generally recognized that it is uneconomical
to design a bridge to resist a severe earthquake elastically. Bridge model
Piers represent the clearly identifiable members in which
potential yield zones can be allowed to occur provided Consider a typical three-span highway bridge such as the
that the stability of the bridge is not compromised. one shown schematically in Figure 1. The bridge is a
Another approach is to restrict the energy absorption reinforced concrete box type that is continuous over three
capacity to a non-structural region and to allow the rest spans and is supported by two single column piers. The
of the bridge to maintain its integrity during the earth- structural system may be idealized by the analytical
quake. Use of special isolation devices is an alternative model shown in Figure 2. The structure is assumed to
to the dissipation of energy through the plastic deforma- consist of a series of line beam-column elements and
tion of piers. A lead-rubber bearing is such an isolation translational massless springs (truss elements). The bridge
system, which allows differential displacement between structure is assumed to be symmetric. This assumption
the superstructure and the substructure and provides an is not necessary but makes it possible to restrain many
energy dissipation mechanism. The laminated elastomeric of the possible degrees of freedom in the model, especially
part of the bearing carries the weight of the superstructure those referring to the torsional behaviour.
and supplies a horizontally restoring force. The plastic In the analysis, it is assumed that the ground motion
deformation of the lead produces hysteretic damping. So, is applied in either the lateral or longitudinal directions.
the lead-rubber system provides both the flexibility and The vertical bridge response due to the vertical ground
0141-0296/88/03157-! 0/$03.00
© 1988 Butterworth & Co (Publishers) Ltd Eng. Struct. 1 988, Vol. 10, J u l y 157
Seismic performance of highway bridges. A. Ghobarah and H. M. Aft
Structural nodes are assumed at thc intersection of
Deck ~ A
pier-bearings and deck-bearing elements. No inter-
mediate nodes are assigned for the deck elements. This
c assumption has been used in girders and columns of
Abutment ' Abutment
wall Pier Pier walt building frames with sufficient accuracy. In bridge
structures, span lengths are relatively large and inter-
a
mediate deck nodal points may seem necessary. However,
the absense of intermediate nodes did not lead in previous
studies ~ to any pronounced inaccuracies in free vibration
results and the earthquake response of bridges.
p~~/i///////
E Effect of rigid deck assumption
The bridge model can be significantly simplified if the
bridge deck is assumed to be rigid in its own plane. The
Pier top ~ simplified model shown in Figure 3 could be regarded as
simulating half the bridge structure. The model repre-
sents either of the longitudinal or the transverse responses.
The same idealizations and assumptions of the first model
hold here except that the deck is assumed to act as a
rigid diaphragm in plan. This assumption allows the deck
to displace only in the direction of loading while restrain-
ing its rotational degree of freedom,
The two models as discussed allow for the variability
Figure I Typical three span highway bridge: (a) elevation; (b) of stiffness of the bearings at the abutments and piers.
section A-A Both of the models could be used to study energy
dissipating devices located at either or both abutment
and pier, by allocating suitable force-displacement char-
acteristics to the bearings (truss elements). The main
Deck
advantage of the first model is that it enables the actual
y ~ earings representation of the deck stiffness. No guidelines are
available as to when diaphragm action ma3/be assumed
in highway bridge design. In some cases, such as that of
a box girder superstructure, it is clear that the deck is
stiff in its own plane over several spans. However, it is
not clear how other types of superstructure, such as
precast girders supporting a cast in-place concrete deck,
will behave. Owing to the difference in rotation between
Deck the piers and abutment tops, some rotation of the deck
"y~ Bearings,~.~,~ may occur. It is believed that the in-plane rigidity is still
likely to be high in highway bridges up to three spans z.
In order to assess the effect of neglecting the stiffness
of the deck on the dynamic response of the bridge, the
qUg error in the undamped fundamental period of the
Figure2 Modellingofthe bridgestructure:(a) transversedirection; structure that results from the rigid deck assumption is
(b) longitudinaldirection shown in Figure 4. The error percentage is plotted against
the total stiffness of the elastomeric bearings, K=, normal-
ized to the weight of the superstructure, IV. The error in
period is defined as: [(7"1 - T2)/T1] x 100, where TLis the
motion component is not taken into account. Accord- period of free vibration in the transverse direction using
ingly, the corresponding degrees of freedom can be
restrained without significantly affecting the expected Deck displacement
response. Out of the six degrees of freedom in the space
frame, only two are allowed for the nodes on the deck Dec
kM-~ass Pier
top mass Pier top
and at the pier. For the tranverse loading, the nodes at
Abutment ~ ~ ~ " ~ displacement&
rotation
the piers are allowed to displace in the tranverse direction
(Z axis), and rotate about the longitudinal direction of
the bridge (X axis). For the longitudinal loading, the Y
nodes at the piers are allowed to displace in the longi-
tudinal direction (X axis) and rotate about the transverse Pie
direction (Z axis). For the piers and abutments, the
vertical displacement is restrained since no significantly
axial shortening is expected. The pier is assumed to be Z [/[//7111fll
fixed to the ground where no soil-structure interaction Ground
is considered. Figure 3 Simplifiedmodelof the bridgestructure
o.o9 /--
Curvature ~'~urvature
Displacement ductility
Earthquake Strength demand
(elastic) Elastoplastic Clough's degrading
ship, respectively, were adopted. The response is intended earthquake record, the response is dramatically affected
to provide the strength and ductility demand required by many cycles that increase the shift in period, reducing
for the bridge to survive severe ground motions. The four the displacement ductility to five compared with the 13.66
actual earthquake time histories described above are used given by the elastoplastic hysteretic model.
as input ground motion. Since the vertical earthquake When limitation on the design forces is provided by
effects are not included in the analysis, the axial load in the yielding of selected structural members, such as the
the pier, or the gravity load effect is constant. The yield pier for highway bridges, it must be designed to deform
surface can therefore be represented by a two-dimensional in a ductile manner. The associated dissipation of energy
elastoplastic moment interaction surface. in the yielding members reduces the seismic response of
The bridge response was predicted using the Newmark the structure but entails some form of damage. This
integration method with the constants ~, and fl taken to procedure may be useful for the El Centro and the
be 0.5 and 0.25, respectively, for the three cases; elastic, Parkfield earthquakes where the ductility requirements
elastoplastic, and stiffness degrading response analysis. were found to be around 1.7 to 1.86. For the Pacoima
The integration time step is taken to be 0.02 s, which is Dam record of the San Fernando earthquake, where the
approximately 1/20th of the fundamental free vibration ductility requirement is much higher, it is not practical
period of 0.412 s. The integration time step is also equal to provide such ductility in the design.
to the time interval of the ground motion records. For Although the structure in the cases of the El Centro
the plastic analysis, the response is determined for a more and the Parkfield earthquakes may survive, the expected
refined time step of 0.01 s; however, a comparison with damage may mean costly repairs and difficulty in the
the results of 0.02 s indicates a negligible difference of restoration of permanent deflections. The resulting
ca. 0.5%. The results of the elastic and inelastic analyses structural damage in the form of permanent deflection
are summarized in Table 2. Considering the pier as a are shown in Figures 7 and 8 for the elastoplastic and
simple cantilever, the displacement of the top relative to Clough degrading stiffness models for pier behaviour,
the base at the first yield, uy, can be determined. By using respectively. Other disadvantages of the conventional
this displacement as the reference value, the pier 'displace- ductile approach are that design procedures may be
ment ductility' is defined as the maximum observed complicated and the required reinforcing steel details are
displacement at the top of the pier, divided by the expensive and difficult to place.
reference relative displacement at first yield, Uy. The
'strength demand' is defined as the maximum base
moment of the pier, divided by the yield moment My.
Response of the isolated bridge
By using the elastic analysis, the moments in the pier The lead-rubber base-isolation pads are incorporated in
remain under the yield level for the Mexico City earth- the same bridge design at the piers and abutments. The
quake since the strength demand of 0.62 is less than unity, fundamental period of free vibration of the bridge, based
while for other earthquakes, moments exceed the yield on the elastomeric stiffness, was found to shift to 1.44 s.
moment. The strength demand at 7.56 is very high for The bridge structure is idealized by using the rigid-deck
Pacoima Dam, but it is moderate for the El Centro (1.545) model shown in Figure 3 with one beam element repre-
and the Parkfield (1.864) earthquakes. The response of senting the pier. The total of the elastomeric stiffness for
the pier using the elastoplastic model shows that the pier bearings located at the pier, which is the post-yield
displacement ductility is 13.66, which is very high for the stiffness as shown in Figure 6, is taken to be twice that
Pacoima Dam record. Displacement ductility of 1.86 and of bearings at the abutment. This is based on the fact
1.74 are required for the first bridge to survive the El that abutments carry about one half the load that the
Centro and the Parkfield earthquakes, respectively. For pier carries, considering all different combinations of dead
the Clough degrading stiffness model case, the ductility and live loads. The number of bearings at the pier are
requirements are lower than those arrived at by the then taken to be approximately twice the number at the
elastoplastic approach. This difference is because the abutment. Also, the number of lead plugs at the abutment
structure becomes more flexible as a result of stiffness is taken to be one half the number of the lead plugs at
degradation associated with a shift in the structure's the pier. Since all plugs are usually designed to have the
period to a region of lower energy, which reduces the same diameter for easier and systematic construction, the
forces acting on the structure. For E1 Centro and Park- yield level, which is the shear force at lead yield of the
field, only a slight difference in ductility requirements is lead-rubber bearings at the pier, is taken to be two thirds
noticed because the structure is subjected to a few of the characteristic yield strength of all lead plugs Qd.
response cycles that do not lead to significant impact on Consequently, the yield at abutment is taken as QJ3.
the pier's stiffness. For Pacoima Dam, which is a severe Based on experimental evidence, the elastic stiffness of
E
(J 2L:: I 11. .,I
E
i.
Centro
AA/VAA^A^ o AI
Ii
E 0
E
~J
03 _, V vV V
c~ Uy
ul
c~ -2
-2
I I
! I 0 10 15
0 5 I0 15
Parkfield 2F Parkfield
-~ Uy
<u 0 "-=~=
^^^A ~ 1f_ AA~I
o.
.~
-1
t"-'"v V'
Uy
-2 ° -2 F vv,, I I
I I I 0 5 10 15
0 5 10 15
San F e r n a n d o
o 4
~ Uy C
E 0
E I.. Uy v- ~1~ .-
(D Uy lv~ | ~ ~
~- -10 -4
£3
-20 I ,, I I -8 I I I
0 5 10 15 5 I0 15
* Mt and Dis refer to ratio of moment at base and ratio of displacement at top of piers
typical highway bridges will dramatically reduce the maximum displacement for the inelastic system. For a
acceleration response as a result of the period shift. given period, the response of different cases was normal-
However, soft soils may alter the frequency response of ized to the corresponding response at (Qdp/Qd= 0) when
an earthquake so that more of the energy is transmitted all the lead plugs are inserted in the bearings located at
at low frequencies. An example of an earthquake record the abutment. The mean values of the response for the
with predominant low frequency is the Mexico City, 1985, five period cases and f o r the normalized El Centro,
record. For such an earthquake, introducing the lead- Parkfield and San Fernando earthquake records are
rubber mountings tends to increase rather than to plotted in Figure 9.
decrease the response. While the non-isolated bridge The introduction of the energy dissipators at the
under study can survive the Mexico City ground motion, abutment only reduces the shear in the pier dramatically.
it will be damaged if isolated. The displacement of the On average, the shear is about 43 % compared with the
deck is quite large, almost fifty times that of the non- shear in the case of introducing all lead plugs at the pier
isolated bridge. The large deformation of the bearings only, as shown in Figure 9. The percentage increases for
cannot be accommodated with reasonable design dimen- the other locations of the energy dissipators. On the other
sions. hand, when all the lead plugs are located on the abutment,
it suffers more forces, which are 2.8 times the force when
Location of lead plugs all lead plugs are located at the pier. The displacement
of the deck varies slightly with plug locations. The
The seismic energy dissipation mechanism in the lead- displacement has minimum values when half the lead
rubber bearing is by yielding of the lead plugs. Given the plugs are placed at the piers. However, the improvement
total shear force required to yield all the lead plugs, Qd, in the displacement of the deck is less sensitive than the
and the lead yield strength of 10 MPa, the total area of shear in the pier to variation in the location of the lead
the lead plugs can be determined. The shear force required plugs. When all the lead plugs are inserted in bearings
to yield the lead plugs located in bearings at the pier is located at the pier, the increase in displacement of the deck
termed Qdp- The ratio of the area of the lead plugs located
at the pier to the total lead area, Qdp/Qdmay vary between
zero and 1. The ratio of zero represents the case when 2.5
all the lead plugs are located at the abutments only, while
a ratio of 1 represents the case when all the lead plugs
are located at the pier only. By varying Qdp/Qd between
zero and one, the impact of the location of the energy Shear in pier
~
o
0.06
Lead shear resistance
,? J
No lead p l u g s
The shear force level at which yielding of the lead 0.04
plugs takes place is an important parameter in the design
of the base-isolator. Given the shear force level and the
lead yield strength of 10-10.5 MPa, the number and cross 0.02
sectional area of the lead plugs can be designed. The level
of yield force depends on the mechanism by which all I I I I
1.0 1 .2 1 .4 1 .6 1 .8 2.0
lateral loads, including those arising from sources other
than earthquakes, are resisted I ~. The level of yield force Bridge period, Tp(Sl
should be selected to achieve two objectives. First, the
Figure 11 Effect of the lead yield shear force o n t h e f o r c e o n
response of the bridge structure is required to remain abutment
elastic under the action of small earthquakes, wind and
braking forces. Second, during severe seismic events, the
level of forces and displacements are required to remain All lead plugs are inserted in the laminated bearings
within acceptable limits to ensure that the bridge con- located at the abutments. The elastomeric stiffness of the
tinues to function. bearings (i.e. post-yield stiffness) at a pier, Kpr, is assumed
The bridge response for different cases with varying to be twice the elastomeric stiffness at the abutment, K,,.
yield shear force of the lead dissipators were investigated For lead-rubber bearings, the ratio of the elastic stiffness
for the normalized E1 Centro, Parkfield and San to the post-yield stiffness (i.e. unloading stiffness ratio) is
Fernando earthquakes. Various ratios of the lead yield taken to be equal to ten. The relative stiffness of the pier
force, Qd, to the weight of superstructure, W,, up to 7 % to the stiffness of the deck was chosen to lie within the
were considered in the numerical analysis. The results ratio range of 20 to 40. The shear response in the pier,
are compared with the elastic response of the bridge force on abutment, and displacement of deck are plotted
without lead dissipators, Qd/W= 0. The response of the against the fundamental period of the bridge, Tp, in
bridge structure incorporating lead plugs as energy Figures 10, 11 and 12, respectively.
dissipating devices, over the practical period range of 1
to 2 seconds, are shown in Figures 10 to 12.
0.20 0.03
_m
0.15
c 0.02
"5_
._c
0.10
co
0.01
0.05
I I I I
I .0 I .2 I .4 I .6 I .8 .0
I I I I
1.0 I .2 I .4 I .6 I .8 .0 Bridge period, Tp(S)
Bridge period, Tp fs)
F i g u r e 12 E f f e c t o f t h e lead y i e l a shear force on the bridge d e c k
Figure I 0 Effect of the lead yield shear force o n t h e s h e a r in pier displacement