EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS, VOL.

20,535-549 (1991)

DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES UNDER

THE COMBINED EFFECT O F VERTICAL AND HORIZONTAL
EARTHQUAKE MOTIONS
M. ALA SAADEGHVAZIRI'
Departmenr of Civil and Environmenial Engineering, New Jersey Instiiute-of Technology, Newark, NJ 07102. U.S.A.

AND

D. A. FOUTCHt
Department of Civil Engineering, Universiiy of Illinois at Urbana-Champaign. Urbana. IL 61801-2397, V.S.A.

SUMMARY
Measurements of ground motions during past earthquakes indicate that the vertical acceleration can reach values
comparable to horizontal accelerations or may even exceed these accelerations. Furthermore,measurements of structural
response show the possibility of significant amplification in the response of bridges in the vertical direction that can be
attributed to the vertical component of ground motion. In this study, the relative importanceof the vertical component of
ground motion on the inelastic response of R/C highway bridges is investigated. Particular emphasis is placed on
modelling of the deck and piers to account for complex loading histories under combined vertical and horizontal
earthquake motions. Analyses of actual bridges indicate that, in general, the vertical motion will increase the level of
response and the amount of damage sustained by a highway bridge. Vertical motion generates fluctuatingaxial forces in
the columns, which cause unstability of the hysteresis loops and increase the ductility demand. Furthermore, vertical
motion can generate forces of high magnitude in the abutments and foundationsthat are not accounted for by the current
seismic design guidelines. Thus, it is important to consider this component of the ground motion in the design of highway
bridges, especially for those located in regions near seismic faults.

INTRODUCTION
The destruction of many highway bridges during the San Fernando earthquake of Feb. 9,1971 revealed the
seismic vulnerability of highway bridges and provided a stimulus for both theoretical and experimental
research related to seismic response of this class of structures. The lack of extensive research prior to this
earthquake is due to the nature of the damage caused by previous earthquakes, which was mostly due to
failure of the substructure or the surrounding soil. The vibrational effect is reported to be the cause of damage
to a relatively small number of bridges during past earthquakes such as the Kanto earthquake of 1923 in
Japan, the 1964 Alaskan earthquake, the Madang earthquake of 1970 and the Chilean earthquake of 1971.l.'
Unlike the pattern of damage during the previous earthquakes, structural vibration was the main cause of
damage to, or failure of, many bridges during the San Fernando earthquake of February 1971. During this
earthquake 62 bridges in the epicentral region of the earthquake were damaged, at a cost of over 15 million
d ~ l l a r s .Damage
~ assessment brought out some of the shortcomings in the design. Among the key
deficiencies were: lack of ductility, small seat width at expansion joints and lack of horizontal ties and
inadequacy of abutments and wing walls. Many of these problems have been studied by researchers both
analytically and experimentally. A compilation of literature resulting from some of this work can be found in

*Assistant Professor.
'Professor.

ooS8-8847/91/060535-15$07.50 Received 19 October 1989

0 1991 by John Wiley & Sons, Ltd. Revised 29 October 1990
536 M. A. SAADEGHVAZIRI A N D D. A. FOUTCH

a report published by the Applied Technology C ~ u n c i lAlthough

.~ the previous work has been extensive and
has contributed significantly to the enhancement of seismic design of highway bridges, it is not complete in a
sense that most of this work considered only the horizontal motions of an earthquake. As a result of this, the
current guidelines for seismic design of highway bridges, recommended by AASHTO,’ consider only the two
horizontal components of ground motion in the analysis and design of bridge structures. AASHTO
recognizes this limitation and states that the specifications ‘. . . will be subjected to refinement as the state-of-
the-art in the effects of vertical seismic excitation develop^'.^
It is the purpose of this paper to present the findings of an extensive study on the combined effects of
vertical and horizontal earthquake motions on the inelastic response of highway bridges.6

VERTICAL MOTIONS
Field measurements and reports
The assumption that the maximum vertical acceleration is small and always smaller than the maximum
acceleration in the horizontal directions is a major factor in the lack of provisions for considering the vertical
component of ground motion in the seismic design of buildings as well as bridges. However, contrary to this
assumption, reports on recent earthquakes indicate that not only is the maximum acceleration in the vertical
direction significant, but also there have been earthquakes where the maximum acceleration in this direction
has exceeded that in the horizontal directions.
The maximum vertical acceleration of the San Fernando earthquake was recorded at Pacoima Dam and
was 0.7 g . The maximum displacement observed across the fault trace is reported to have been about six feet
in the vertical direction and about five feet in the lateral direction.’ The El Asnam earthquake of Oct. 10,1980
is considered extraordinary because of the strong movement in the vertical direction.8 Unfortunately, no
acceleration records of this earthquake are available. However, there have been indications that the vertical
acceleration may have reached a maximum very close to 1.0 g, while the maximum horizontal acceleration is
estimated to have been about 025 9.’ Records of an earthquake of magnitude 5.9 in Southern California
show that the vertical acceleration was significant and much higher than the horizontal acceleration. In one
location the peak horizontal acceleration was recorded to be 0.33 g, while the peak vertical acceleration was
equal to 0.59 g . ’ O During the Loma Prieta earthquake of Oct. 17, 1989, in several locations the peak vertical
acceleration was greater than the peak horizontal acceleration and reached values close to 0.7 9.l’

Structural response
Disregarding the vertical component of earthquake motion in the analysis and design of structures under
seismic load has resulted in insufficient attention to recording the vibrational response of structures in this
direction. However, the available measurements indicate the possibility of vertical excitation of structural
systems caused by ground motion. For example, recordings of the Meloland Road Overpass show that,
during the Imperial Valley earthquake of Oct. 9, 1979, the road deck horizontal and vertical peak
accelerations were comparable, and they reached acceleration levels of 0.5 g and 0-51 g , respectively.”
Records from Rio Del Overpass (a two-span R/C box girder bridge) also support such a possibility and
peculation.'^ During the Cape Mendocino offshore earthquake of August 1983, the peak vertical accelera-
tion at the centre of a span was larger than that at the abutment by nearly a factor of ten. That is, the vertical
amplification was equal to 10. The same factor for the transverse motion was only 2.

ANALYTICAL MODEL FOR THE BRIDGE

Bridge deck model
To investigate the effect of vertical motion along with other components of an earthquake, 3-D modelling
of the bridge deck system is essential. Furthermore, to perform non-linear time history analyses the model
must be computationally effective. Therefore, it is very useful to develop a model that is simple and
economical and yet is accurate in representing the vibrational characteristics of the system in 3-D space.
 DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES 537

These objectives can be achieved by modelling the bridge deck with a set of beam elements forming a grid
(Figure 1). Properties of constituent elements in such a model (GRID MODEL) are obtained from the
geometry of the box cross-section and from a model using isoparametric shell elements by matching several
fundamental frequencies and mode shapes of the two models. The accuracy of the model was demonstrated
by comparing the results of a grid model of the Meloland Road Overpass subjected to the 1979 Imperial
Valley earthquake to the measured response.6

Column model
Vertical motion will generate a large axial force in the piers/columns of highway bridges. Furthermore, the
variations of this force will not be proportional to that of the lateral force or bending moment. Thus, as the
primary source of energy dissipation, it is important to represent these components of the bridge with as
realistic a non-linear model as is practical that can simulate the complex behaviour of R/C members under
arbitrary loading conditions and histories. Another objective in developing the column model was that it
must be capable of depicting the extent and type of damage to be expected.
Since simple and computationally effective discrete models are mostly based on a set of predefined
phenomenological rules, they can not simulate the complex interaction between the axial and flexural
deformations, especially when the variation of axial and lateral forces is un~oupled.'~ This is because most of
these hysteretic models are established under the assumption of a constant axial force. Furthermore, the
extent and type of damage can not be observed using these models. Therefore, it was decided to model the
R/C column using inelastic isoparametric plane stress elements to represent the concrete and elastic-plastic
bar elements to model the reinforcing steel. A typical column is modelled as shown in Figure 2: 3-node truss
elements are used for better compatibility between these and the 9-node plane stress elements.
This type of modelling, although expensive and time consuming computationally, is capable of simulating
the complex behaviour of R/C members under non-proportional fluctuations in axial and lateral loads. The
versatility is due to the fact that no predefined hysteresis rules are involved, and only the stress-strain
relationships of the constituent materials are needed. The non-linear concrete material model includes
cracking as well as crushing of the concrete. The degree of concrete confinement can also be accounted for
through the constitutive model for the concrete. A bilinear elasto-plastic material model with kinematic

Figure 1. Grid model of a typical bridge deck

538 M. A. SAADEGHVAZIRI AND D. A. FOUTCH

9-Node Isoparameteric
Plane Stress Element

d
Steel: 3-Node Truss Element

1: + Reinforcing Bars

Typical Cross-section

Figure 2. Finite element model of typical R/C column

hardening is used for the reinforcement. For a detailed description of the column model and practical
examples on the adequacy of the model see Saadeghvaziri and F ~ u t c h . ' ~ . ~

NUMERICAL EXAMPLES
The models developed for the deck and the column were employed to analyse several hypothetical bridges
under two different cases. These two cases are: (i) considering only transverse (horizontal) motion, referred to
hereafter as H-only; and (ii) considering both vertical and transverse motions simultaneously, referred to as
V&H.

Conjiguration and design

All of the hypothetical bridges are assumed to consist of two spans (140-ft each') with either a single- or
double-column bent at the centre. For the single-column bent the cross-section is a 4 ft x 8 ft rectangle, and it

1 ft = 0.3048m.
 DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES 539

is 3 ft x 5 ft rectangular sections spaced 20-ft apart for the double-column bent. The properties of the decks
are determined from a typical 2- or 4-lane R/C box girder bridge. The deck-to-abutment connection is
assumed pinned, except as noted. The length of the columns is assumed to be 30 ft, and they are assumed to
be fixed at the base. The model for bridge deck and pier is the same as that which was described earlier.
Bridges are designed based on AASHTOs’ single mode response spectrum method. The piers are designed
for the moments from the static analysis divided by the appropriate response modification factor, R, and the
strength reduction factor, 4. However, one bridge is designed strictly based on the required steel for moment
capacity and the minimum reinforcement requirement is ignored. Assumed properties for concrete and steel
materials are shown in Table I. These values are used for all bridges except for one bridge where perfect
confinement of concrete is not assumed. For that bridge, the slope of the stress--strain relationship for
concrete beyond the maximum compressive stress is determined based on the amount of lateral
reinforcement.6-
For a proper comparison of structural responses it is desirable to have an accelerogram that generates
approximately the same elastic structuraI response at all frequencies within a given range. Therefore, an
artificial accelerogram developed by Barenberg and Foutch16 is used in this study. The artificial accelero-
gram generates almost the same acceleration in structures with frequencies in the range of 2 to 10 Hz.
The artificial earthquake is scaled such that the effective peak acceleration is 0.7 g. Note that 0.4 g effective
peak acceleration as assumed in the design will cause no or very small yielding in the piers. This is due to the
fact that the strength reduction factor assumed in the design of the columns, the softening of the piers upon
cracking, and strain hardening of the reinforcing steel offset the effect of the response modification factor.
Another factor is that for most of the bridges the minimum reinforcement requirement controls the design.
Considering all these factors, no significant inelastic behaviour takes place under the design earthquake. It
should also be noted that peak accelerations in excess of design values are expected in the faulting region of
major earthquakes. In fact, CalTrans16incorporates this into its own provisions even though this is ignored
by AASHTO.

Bridges # 1 & # 2
Bridge # 1 has a single-column at the centre and, based on the codes, the minimum reinforcement of 1 per
cent governs the design of the column. The moment-curvature (M-4) hysteresis loops at the base of the
column for both cases (i.e. H-only and V&H) are shown in Figure 3. When considering only transverse
motion the hysteresis loops are very stable with no strength deterioration. There is not any sudden or
unpredicted change in the stiffness of the column. On the other hand, the M-4 loops for the V&H case are
very unstable and asymmetric. There are significant changes in the strength and stiffness of the column
because of the variation in axial load resulting from vertical excitation. As a result of this variation in
stiffness, the percentage of total lateral load carried by the column is significantly different for each case
(Figure4). For the V&H case, there are moments of abrupt change in the column’s participation in the lateral
load carrying capacity of the system. Note that both the maximum and minimum column shears occur for
the V&H case. For the presence of a very rigid deck and other reasons to be discussed later, the maximum

Table I. Properties of concrete and steel’

Maximum compressive strength of concrete 4.0 ksi

Strain at maximum compressive strength 0.002 infin
Ultimate compressive strength of concrete 40 ksi
Strain at ultimate compressive strength 0.050 infin
Initial modulus of elasticity for concrete 4OOO.O ksi
Yield stress of steel 60.0ksi
Modulus of elasticity for steel 30 O00.0 ksi
Modulus of elasticity for steel after yielding 600.0 ksi

‘1 ksi = 6.895 MN/mZ; 1 in = 0.0254 m.

540 M. A. SAADEGHVAZIRI A N D D. A. FOUTCH

24000.

12000.

-12000.

-24000.
-3.8 -1.9 0.0 1.9 3.8
CURVATURE (1 08*-3 r a d / f t )
(a)

24000.

12OOO.

-12000.

-24000. I
-3.8 -1.0 0.0 1.0 8
CURVATURE (1O * * 4 rad/ft)
(b)
Figure 3. Moment-curvature hysteresis loops for Bridge # 1: (a) H-only excitation; (b) V&H excitations
 DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES 54 1

-.
1.5 2.5 3.5
TlME (eec)
Figure 4. Time histories of the percentage of lateral load carried by the column

1
(4 (a)
Figure 5. Extent of damage for Bridge # 1: (a) H-only excitation; (b) V&H excitations

lateral displacement is about the same for both cases and is 3.4 in at the centre of the deck. The difference in
hysteretic behaviour manifests itself also in the amount of damage sustained by the column, as shown in
Figure 5, where the sustained damage for the V&H case is twice as much as that for the H-only case (note that
the concrete is considered damaged when its maximum strength is reached).
With regard to other components of the bridge: the maximum moment to be camed by the column
foundation is 30 per cent higher for the V&H case (23OOO kip-ft vs. 17 600 kip-ftt). The axial load for the H-
only case is a constant force of 2000 kip, corresponding to the dead load, while for the VLH case the
maximum compressive axial load is equal to 5450 kip and a maximum net tensile axial load of 1400 kip
occurs. Considering vertical motion the maximum downward and upward vertical forces at the abutments
are 1520 kip and 420 kip, respectively, compared to 600 kip and zero for the H-only case.
The second bridge (Bridge # 2) is exactly the same as Bridge # 1 except for the degree of concrete
confinement. For this bridge, rather than assuming perfect confinement, the stress-strain curve for concrete

1' k i p 4 = 1.356 kN m; 1 kip = 4448 kN.

542 M. A. SAADEGHVAZIRI A N D D. A. FOUTCH

softens upon reaching the maximum compressive strength and its slope depends on the amount of transverse
reinforcement. For the H-only case no major difference was observed compared to Bridge # 1. However, for
the other case (V&H), although the maximum moment was the same as that for Bridge # 1, the maximum
curvature was 50 per cent higher (5.8 x vs. 3.8 x lop3rad/ft). This means not only much higher
damage to the concrete, but also indicates the demand for more local ductility under such loading. Thus,
proper detailing of the plastic zones must be warranted.

m.

4250.

?
cI
L
$
0.
!z
9a
-4250.

-8500.
-4.6 -2.3 0.0 23 4.8
CURYA’I’VRE(10**-3 rad/ft)
(a)
8500.

4250.

-
r
4
3
3 0.

-4250.

-8500.

CURVATURE (1 0**-3 rad/ft)

(b)
Figure 6. Moment-curvature hysteresis loops for both columns of Bridge # 3 under H-only excitation
 DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES 543

-4.6 -2.3 0.0 2.3

CURVATURE (10.O-3 rad/ft)
(4

-4250.

-6500. ' I

CURYATUEE (1 0-3 rad/ft)

(b)
Figure 7. Moment4urvature hysteresis loops for both columns of Bridge # 3 under V&H excitations

Bridges # 3 & # 4
These two bridges are assumed to have double columns at the centre bent. For the first bridge the
minimum reinforcement requirement, which controls the design, is considered. For the second bridge (Bridge
# 4) this requirement is ignored and the columns are designed to carry the required moment and axial force.
544 M. A. SAADEGHVAZIRI A N D D. A. FOUTCH

8-only

1.7 9.2 3.7

TlME (sec)

Figure 8. Time histories of variations in ultimate moment

Shown in Figures 6 and 7 are the M-4 curves for Bridge # 3 for both columns. Similar to previous
examples, for the H-only case the hysteresis loops, although asymmetric, are stable with no strength decay for
each half cycle. The asymmetry of the loops is due to variation in the axial force from framing action in the
lateral direction. Note that, unlike non-proportional axial force, under proportional variations in axial force
there is a compensating effect between the two column^.'^ Under vertical motion the loops are again very
unstable and demonstrate significant variations in strength and stiffness (Figure 7). The compensating effect
of columns for the H-only case results in a symmetric relationship, while this is not true for the V&H case.
Diagrams of the variations in ultimate moment of the columns for both cases are plotted in Figure 8. The
damage for the V&H case was 70 per cent more than that for the H-only case.
The effect of vertical motion on Bridge # 4, which is designed strictly based on the required moment
capacity, was even more pronounced. For both cases, compared to Bridge # 3, the damage sustained by the
columns was more than 50 per cent higher. The maximum displacement for this bridge was 5.7 in, which is 55
per cent more than that for Bridge # 3 (with 3.7 in of displacement). Note that ignoring the minimum
reinforcement requirement is similar to increasing the response modification factor, R, for design, which has
been suggested by previous investigators.

Bridges # 5 & # 6
To see the relative importance of vertical motion on 4-lane bridges which have stiffer and more massive
decks, two bridges with decks having properties similar to those of a typical 4-lane bridge are studied. Bridge
# 5 has a single column at the central bent and the abutments are assumed pinned. The second bridge
(Bridge # 6 ) is exactly the same as the first one except for the deck-abutment connection which is assumed to
be non-linear. Note that for these examples the minimum reinforcement ratio controls the design. Owing to
high stiffness of the deck in the transverse direction, this amount of reinforcement is even much higher than
that required for an earthquake with 0.7 g peak acceleration.
Hysteresis loops for Bridge # 5 are shown in Figure 9. When considering only transverse motion, the
column remains almost elastic. For the V&H case, however, this is not the case, as can be seen from this
figure. As a result of tensile axial force, the cross-section yields at a much lower bending moment and the
shear capacity decreases. Also, owing to the compressive axial load the column becomes much stiffer.
Consequently, more shear and moment are generated in the column. Quantitatively the amount of damage
for the second case is four times that for the first case. Time histories of column axial force are shown in
Figure 10. Note that a similar trend is observed for vertical forces on the abutments.
Another possible mode of failure is shear failure. This is due to the fact that the shear capacity of R/C
members is affected by the level of axial force. Compressive axial force increases and tensile axial force
decreases the shear capacity. Note that, although compressive axial force increases the shear capacity, it also
 DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES 545

CURVATURE (1 0-3 rad/ft)

(4
Figure 9. Moment-curvature hysteresis loops for Bridge # 5 (a) H-only excitation; (b) V&H excitations

increases the demand owing to the increase in the stiffness of the column. Thus, it is possible to have shear
failure even when the axial force is compressive.
Blume et al.” recommend the following formulae for the amount of shear, V,, carried by the concrete:
546 M. A. SAADEGHVAZIRI AND D. A. FOUTCH

"""".
0.0 2.5 5.0
TIME (see)

Figure 10. Time histories of column axial force for Bridge # 5

if P is compressive, and

for tensile axial load P , . A,, is the transformed area of the uncracked section; d is the effective depth of the
section; b is the minimum width of the section; and F i is the maximum compressive strength of concrete.
Using these equations and assuming zero capacity when the entire section is cracked, time histories of the
provided shear capacity along with applied shear are plotted in Figure 11. For the H-only case, the shear
capacity is almost constant and is always higher than the applied shear. However, owing to varying axial
load, the shear capacity for the V&H case is changing. In several instances the applied shear exceeds the
provided capacity, indicating the possibility of shear failure. In many other instances the factor of safety
against shear failure is very close to unity, which is not desirable for structures that are designed to resist
seismic loads.
Note that similar behavioural characteristics and effects are observed for a 4-lane bridge with a double-
column bent. Comparing the results to those of other examples, it appears that the effect of vertical motion is
more pronounced on the 4-lane bridges. This is due to the fact that higher forces develop in this type of bridge
since the decks are so massive. Most of the shear in the transverse direction is carried by the abutments;
therefore, considering only horizontal motion, the columns remain almost elastic. However, in conjunction
with vertical motion due to the changes in the stiffness of the column and as a result of coupling between axial
force and bending moment, much higher yielding takes place and the sustained damage is significantly
higher.
In practice it is not always possible or practical to accommodate such high forces at the abutments.
Therefore, abutments might be designed to yield. To see the relative importance of vertical motion on this
type of boundary condition, Bridge # 6 is analysed with non-linear elasto-plastic springs at the ends. The
yield force is equal to 1/3 of the design shear at the abutments. Compared to Bridge # 5 the sustained
damage is higher for both cases although it is still insignificant for the H-only case. As expected, the
maximum displacement is higher and is equal to 1.8 in (that is about 80 per cent increase compared to Bridge
# 5). For the V&H case, compared to the H-only case, not only is the extent of damage significantly higher.
but it is accumulated on only one side of the cross-section. This is due to phasing of the axial force with
respect to the lateral load. Accumulation of damage at certain regions can result in structural collapse that
will not occur if damage is uniformly distributed. Or, it may necessitate in replacement of an entire pier owing
to deep penetration of damage beyond a repairable level.
The maximum displacements for both cases are equal. One would expect the displacement to be higher for
the V&H case, since the deck does not resist any lateral load upon yielding of the abutments. The reason that
this does not happen is that, when the abutments yield and the column is under tensile axial force, the stiffness
 DYNAMIC BEHAVIOUR OF R/C HIGHWAY BRIDGES 547

0.0 5.0
TIME (see)

0.0 2.5 6.0

TIME (sec)
(4
Figure 11. Time histories of applied shear vs. shear capacity for Bridge #5: (a) V&H excitations; (b) H-only excitation

of the system goes toward zero. This in turn means a significant reduction in input energy; therefore, there is
not any lateral force to cause any additional displacement.

CONCLUSIONS
From these examples, it can be said that, in general, vertical motion does affect the response of highway
bridges. In conjunction with transverse motion, it will aggravate the sustained damage or may even cause
damage and/or failure which would not have occurred otherwise. Of course, each component of a bridge is
affected differently by this component of ground motion, and, depending on the intensity of motion, the
extent of damage varies. The following conclusions with regard to the relative importance of the vertical
component of ground motion on various components of bridges can be made.

Columns and piers

Owing to uncoupled variations in the axial and lateral forces, the hysteresis loops are very unstable and
asymmetric. They demonstrate significant fluctuations in the stiffness and strength of the column. As a result
of compressive axial load the column stiffness increases. Consequently, the amount of lateral shear and
moment that is carried by the column increases. This, in turn, increases the possibility of failure in the
column, its foundation and abutments. On the other hand, tensile axial force reduces the shear and moment
capacity of the cross-section. This may lead to shear failure or yielding under bending moments that are
548 M.A. SAADEGHVAZIRI AND D. A. FOUTCH

much lower than the anticipated design capacity. Note that unexpected yielding of columns has been
reported for many bridges during past earthquakes. Other undesirable effects are: substantial increase in
ductility demand, higher damage at the plastic zones, accumulation of damage on only one side of the
column cross-section, cracking of the entire column cross-section, the possibility of pullout and buckling of
the reinforcing bars and increased forces to be carried by the abutments.
For a double-column bent, the effect of the vertical component of ground motion is even more
pronounced. This is because of the higher response modification factor used in the design of multiple-column
bents. Compared to 2-lane bridges, the relative effect of this motion on the columns of 4-lane bridges is more
important. For 4-lane bridges, as a result of the very stiff deck in the lateral direction, a significant portion of
the total lateral load is carried by the abutments. Therefore, considering only the transverse motion, the
columns remain almost elastic. However, when the combined effect of vertical and transverse motion is
considered,much greater4nelasticbehaviour takes place, the sustained damage is significantlyhigher and the
possibility of shear failure prevails.

Abutments and connections

For all of the bridges discussed here, the varying forces induced by the vertical motion on the abutments
would not be accounted for in the current guidelines for seismic design of highway bridges. For major
earthquakes, the intensity of these forces can be so high that compressive forces as large as three times the
dead load are exerted. Net tensile axial loads of great magnitude are also generated. To summarize, the effects
of this component of ground motion are expected to be: (i) significant variation in the mechanism of load
transfer through the bearing. For example, if normal stresses are zero or tensile the shear capacity of these
elements reduces substantially; (ii) possible failure of the hold-down devices, since these components are not
designed for any tensile vertical force if the abutments are pinned in the longitudinal direction;
(iii) modification, of soil-abutment interaction, increasing the complexity of this phenomenon. For example
the codes recommend that abutments of the free-standing type should be proportioned to slide rather than
tilt. To comply with this recommendation, consideration of the changes in the frictional forces due to
variations in normal stresses is essential; and (iv) higher settlement and greater forces on the wingwalls.

Foundation
Effects of the vertical component of a major earthquake on the pier foundation are: (i) higher compressive
and tensile axial forces: (ii) a significant increase in the bending moment and lateral shear due to the increase
in compressive axial force acting on the columns; (iii) possibility of pullout of the reinforcement and
inadequacy of longitudinal steel in the piles; and (iv) possibility of foundation uplift and stability problems in
the piles.

Expansion joints
Although not considered in this study, it is anticipated that vertical motions will have major effects on the
behaviour of these elements. Note that the load-deformation relationships of the bearing pads are affected
significantly by the variations in the applied normal stresses. Thus, by affecting the performance of the
bearings and exerting high forces on the vertical ties of an expansionjoint, the vertical motion may cause loss
of span due to excessive displacement of the girders.

REFERENCES
effects of highway bridges’, Report No. EERC 72-11, Earthquake
1 . T. Iwasaki. J. Penzien and R. Clough, ‘Literature survey-Seismic
Engineering Research Center, University of California, Berkeley, CA. 1972.
2. G . C. Sturman, ‘The Alaska highway system’, The Great Alaska Earthquake of 1964-Engineering, National Academy of Sciences,
Washington, D.C., 1973, pp. 987-1009.
3. A. L. Elliot and 1. Nagai, ‘Earthquake damage to freeway bridges’, San Fernando, Cal$ornia, Earthquake of February Y, 15271, U.S
Department of Commerce, National Oceanic and Atmospheric Administration, Washington, D.C., 1973, pp. 201-233.
4. Applied Technology Council, ‘Earthquake resistance of highway bridges’, Proc. workshop on earthquake resistance highway bridges
Palo Alto, CA (1979).
 DYNAMIC BEHAVIOUR O F R/C HIGHWAY BRIDGES 549

5. AASHTO (American Association of State Highway and Transportation Oflicials), Guide Specijcationrr for Seismic Design of
Highway Bridges, 1983, AASHTO, Washington, D.C. 1983.
6. M. A. Saadeghvaziri and D. A. Foutch, ‘Inelastic behavior of R/C highway bridges under horizontal and vertical earthquake
motions’, Civil Engineering Studies, SRS No. 540, University of Illinois at Urbana-Champaign, Urabana, Illinois, 1988.
7. G. W. Housner, ‘General features of the San Fernando earthquake’, Engineering Features on the San Fernando Earthquake of
February 9, 1971, (Ed. P. C . Jennings), EERLTl-02, Earthquake Engineering Research Laboratory, California Institute of
Technology, Pasadena, CA, 1971.
8. D. V. Mallick, ‘Engineering studies of building response during El Asnam earthquake of October 10, 1980’, Proc., 8th world conf.
earthquake eng. San Francisco, CA IV, 783-790 (1984).
9. J. Despeyroux, ‘Some lessons to be drawn from the El Asnam earthquake of October 10,1980’, Proc., 8th worldconf.earrhquake eng.
San Francisco, California, V, 549-556 (1984).
10. EERI, Earthquake eng. res. inst. newsletter 2, No. 8, (1986).
11. EERC, ‘Preliminary report on the seismological and engineering aspects of the October 17, 1989 Santa Cruz (Loma Prieta)
earthquake’, Report No. UCB/EERC-89/14,Earthquake Engineering Research Center, University of California, Berkeley, 1989.
12. S. D. Werner, M.B. Levin and J. L. Beck, ‘Seismic response characteristics of Meloland Road Overpass during 1979 Imperial Valley
earthquake’, Report No. R-8222-5603; Agbabian Associates, El Segundo, California, March 1985.
13. A. F. Shakal et al., ‘CSMIP strong-motion instrumentation and records from transportation structures-Bridges’, Reprinted from a
Symposium Sponsored by the Technical Council on Lifeline Earthquake Engineering/ASCE, National Convention, San Francisco,
CA, October, 1984.
14. M.A. Saadeghvaziri and D. A. Foutch, ‘Behavior of R/C columns under nonproportionally varying axial load’, J . stmct. eng. A X E
116, 1835-1856 (1990).
15. R. Park and T. Paulay, Reinforced Concrete Structures, Wiley New York, 1975.
16. M. E.Barenberg and D. A. Foutch, ‘Inelastic response and damage assessment of reinforced concrete highway bridges subjected to
seismic loadings’, Civil Engineering Studies, SRS No. 528, University of Illinois at Urbana-Champaign, Urbana, Illinois, 1986.
17. J. A. Blume, N. M. Newmark and L. H. Corning, ‘Design of multistory reinforced concrete buildings for earthquake motions’,
Portland Cement Association, Chicago, Illinois, 1961.