Proceedings of the 9th U.S.

National and 10th Canadian Conference on Earthquake Engineering

Compte Rendu de la 9ième Conférence Nationale Américaine et
10ième Conférence Canadienne de Génie Parasismique
July 25-29, 2010, Toronto, Ontario, Canada • Paper No 1599

SEISMIC BEHAVIOUR OF RECTANGULAR R/C BRIDGE COLUMNS UNDER

BIDIRECTIONAL EARTHQUAKE COMPONENTS

A. Khaled1 , B. Massicotte2 and R. Tremblay2

ABSTRACT

Modern bridge design codes and regulations generally require that the seismic
demand on bridge columns under bidirectional earthquake ground motions be
determined using the 30%-Rule of the effects of each component computed
independently using response spectrum analysis. A research project consisting of
a comprehensive analytical and experimental study has been undertaken to
examine the adequacy of the 30%-Rule to predict the seismic demand on R/C
bridge columns subjected to bidirectional earthquake components for two North
American sites in areas of moderate seismic hazard: Montreal, in the East, and
Vancouver, along the West coast. Results of the analytical study show that the
combination rule is tributary of both ground motion and bridge characteristics and
suggest that a lower percentage for eastern sites and a higher percentage for
western sites would be more appropriate compared to the 30%-Rule currently
prescribed by codes and regulations. A new combination rule with varying
weighted percentage, 20% for eastern sites and 40% for western sites, is proposed
to account for the differences observed for the two sites. The objective of the
experimental program described in this paper is to validate the proposed
combination rule through the testing of large scale rectangular R/C bridge
columns under bidirectional ground motions.

Introduction

The prediction of the seismic demand on structural elements subjected to bidirectional

ground motion earthquakes can be achieved using the response spectrum dynamic analysis method.
A unique ground motion spectrum is usually adopted and input independently in each of the two
orthogonal directions and the structure is analyzed accordingly. The critical seismic demand is then
estimated by combining the two maximum demand values computed in each of the two orthogonal
directions using combination methods. These methods are the Percentage Rule (Newmark 1975,
Rosenblueth and Contreras 1977), the SRSS Rule (Rosenblueth 1951), or the CQC3 (Lopez and
Torres 1997, Menum and Der kiureguian 1998).
For bridge columns, whose responses depend on the interaction of several forces (e.g.
axial force and bending moments), rigorous advanced response-spectrum-based methods for
combining the effects of ground motion components have been developed (Menum and Der
Kiureghian 2000). These methods underlie complex theories and imply cumbersome calculations
1
Professor, Dept. of Construction, École de Technologie Supérieure, Montreal, Quebec, H3C 1K3
2
Professor, Dept. of Civil Geological and Mining Engineering, École Polytechnique, Montreal, Quebec, H3C 3A7
that are beyond the normal day-to-day practice for design engineering firms. As an alternative,
the Percentage Rule represents a simpler procedure for combining the effects of ground motion
components for bridge columns, especially for the common case of regular bridge structures.
Consequently, the Percentage Rule has been widely accepted and is now prescribed in most
seismic bridge design codes.
In North American bridge design codes (e.g., CSA 2006, ATC 2003, AASHTO 1996),
the critical seismic response R in bridge columns is obtained by combining the maximum
responses R1 and R2 associated to each component of the ground motion earthquake, using the
Percentage Rule:

R = R1 + α R2 (1)
R = α R1 + R2 (2)

where α is the weighted percentage defining the Rule. A unique value of α is typically
prescribed. However, there is no general acceptance on the value of α among codes and
regulations; while some codes prescribe α = 0.3 (e.g. CSA 2006, AASHTO 1996), as suggested
by Rosenblueth and Contreras (1977), others prescribe rather α = 0.4 (e.g., ATC 2003), as
suggested by Newmark (1975). Past studies related to the adequacy of the Percentage Rule
generally indicate that the Rule can either underestimate or overestimate the critical response by
as much 30% when compared to the exact response obtained from time-history analyses (e.g.,
Wilson and al. 1995, Menum and Der Kiureghian 1998). However, these studies were conducted
for single response quantities from which the adequacy of the Percentage Rule cannot be
generalized to the case of multiple response quantities that characterize structural elements such
as bridge columns. Furthermore, none of these studies addressed the issue of the effect of the
structure or the ground motion characteristics on the Percentage Rule.
A research project consisting of a comprehensive analytical and experimental study has
been undertaken by the authors to examine the adequacy of the 30%-Rule to predict the seismic
demand on regular R/C bridge columns subjected to bidirectional earthquake components for
two North American sites in areas of moderate seismic hazard: Montreal, in the East, and
Vancouver, along the West coast. This paper summarizes the key findings of this study. The
adequacy of the 30%-Rule is assessed throughout elastic and nonlinear dynamic analyses on a
series of common regular bridge models. A new Percentage Rule that takes explicitly into
account the ground motion and the bridge characteristics is proposed. The test set up and
procedure used for the multi-axis testing of rectangular bridge piers designed with the proposed
Percentage Rule and subjected to bidirectional seismic loadings is also briefly discussed.

Earthquake ground motion records

Montreal and Vancouver were chosen as representative cities for eastern and western
Canada, respectively. The selection of earthquakes representative of these regions was made
based on the deaggregation of the seismic risk at each site in terms of most likely magnitude (M)
and epicentral distance (R) scenarios for a probability of exceedance of 2% in 50%. The
predominant M-R scenarios are M6.0 at 30 km, M6.5 at 50 km, and M7.0 at 70 km for Montreal,
and M6.5 at 30 km and M7.2 at 70 km for Vancouver. Seven and nine pairs of orthogonal
ground motion records were selected for Montreal and Vancouver, respectively. The earthquakes
consist of recorded historical events as well as simulated earthquakes (SIM) provided by Prof. G.
Atkinson (Personal communication 2005).
The site condition for the recorded and the simulated earthquakes are representative of
site Class C, corresponding to an average shear-wave velocity of about 560 m/s, except for the
Saguenay and Nahanni earthquakes which were recorded on hard rock. Site response analyses
were first performed to modify the Saguenay and Nahanni recordings and make them appropriate
for Class C site conditions. Finally, the horizontal components of all the selected earthquakes
were transformed into their principal directions according to the Penzien and Watabe approach
(Penzien and Watabe 1975).

Effectiveness of the 30%-Rule

The adequacy of the 30%-Rule to predict the seismic demand on bridge columns
subjected to bidirectional earthquake components was examined (Khaled and al. 2009). Nine
(09) two-span generic bridge models (Fig.1) were considered. The prototype bridges used to
develop the generic bridge models are regular bridges that are representative, in terms of
characteristics, of a large number of bridge structures encountered in Montreal and Vancouver
highways and consist of: (i) straight bridges, (ii) skewed bridges with 22o skew angle, and (iii)
curved bridges with 30 m radius of curvature. The bent systems of these models are single
column bents and multi-columns bents with circular, rectangular, and wall type columns. The
columns are 8.0 m tall. They are fixed at the base and pin-connected at the top. The geometric
characteristics as well as the fundamental periods of vibration along the longitudinal and
transverse directions of the bridge models considered in this study are summarized in Table1.

1
Skew angle θ
X

Column cross-section

Figure 1. Typical elevation view of the generic bridge models.

Two series of dynamic analyses were performed on the selected bridge models using the
SAP2000 program (CSI 2006). In the first series, time-history analyses were performed to
determine the critical seismic response, RTH, of the bridge columns under the simultaneous action
of orthogonal earthquake principal components. The input angle of the two orthogonal
earthquakes components was varied from 0o to 180o in 30o increments. In the second series,
response spectrum analyses of the bridge models were performed using the 5% damped
acceleration spectra of the selected earthquakes to determine the maximum response in each
direction. The probable response, R30%, is then estimated using the 30%-Rule.

Table 1. Characteristics of the generic bridge models.

Bridge Model Bridge Type Bent System Column Cross-Section (m2) Tx (s) Ty (s)
DPCU Straight Single Column Circular (Ø 2.2 m) 0.764 0.338
DPRU Straight Single Column Rectangular (1.4×2.8) 1.005 0.284
DPM Straight Wall-Type Column Rectangular (0.8×10) 1.250 0.109
DPCM Straight Multi-Columns Circular (Ø 1.2 m) 0.747 0.322
DPRM Straight Multi-Columns Rectangular (0.8×1.6) 1.024 0.257
BPRU Skewed Single Column Rectangular (1.5×3.0) 0.886 0.265
BPRM Skewed Multi-Columns Rectangular (0.9×1.8) 0.714 0.233
CPRU Curved Single Column Rectangular (1.6×3.2) 0.770 0.190
CPRM Curved Multi-Columns Rectangular (1.0×2.0) 0.575 0.227

The critical responses RTH and R30% were identified, through a design process, as the ones
producing the largest failure surface of the columns. The adequacy of the 30%-Rule at predicting
the seismic response is investigated by comparing R30% to RTH, in terms of interacting moments
(Mx, My) at the base of the columns. For the nine bridge models and the sixteen ground motions
considered, a total of 144 graphs, representing the seismic response of the bridge columns to
bidirectional earthquakes at both sites were obtained and analyzed. Typical results are presented
in Fig. 2 for Montreal (east site) and Vancouver (west site).
Analysis of the results indicate that the critical seismic demand on bridge models located
in Vancouver (west site) tend to be underestimated by the 30% - Rule, whereas for bridge
models
(a) BPRU model under Saguenay EQ (b) DPCU model under Whittier Narrow EQ
20000 15000
0 0
12000
15000 30 30
60 9000 60
10000 90 90
6000
120
Moment X-X (kNm)

120
Moment X-X (kNm)

5000 150 3000 150

180 180
0 0
30% 30%
-5000 -4000 -3000 -2000 -1000 0 1000 2000 3000 4000 5000 -30000 -20000 -10000 0 10000 20000 30000
-3000
-5000
-6000
-10000
-9000 RTH
RTH
R30%
-15000 R30% -12000

-15000
-20000
Moment Y-Y (kNm) Moment Y-Y (kNm)

Figure 2. Comparison of time-history (RTH) and response spectrum (R30%) results for:
(a) Montreal (east site), (b) Vancouver (west site).
located in Montreal (east site) the 30%-Rule tends to overestimate on average the seismic
response. The use of the 30%-Rule introduced a mean absolute error of the order of 7% when
compared with time-history results. In some cases, the 30%-Rule was found to underestimate or
overestimate the exact response by more than 21% and 19%, respectively. These results show a
difference in the demand expected from earthquakes expected in eastern and western sites and
suggest that a lower value of α (in Eqs. 1 and 2) for eastern Canada and a higher value for
western Canada could yield more accurate results compared to the unique 0.3 value currently
prescribed by the majority of bridge design codes.

Improved Percentage Rule

An improved combination rule for predicting the seismic demand on regular bridge
columns is developed. The improved combination rule implicitly accounts for both the bridge
and ground motion’s characteristics by specifying a variable weighted percentage. The required
column reinforcement ratios of nine regular bridge models with varying characteristics (Table 2 )
were determined using the combination rule with different values of the weighted percentage α
(0, 0.3, and 1.0) and compared to those obtained from multiple elastic time-history analyses to
identify the bridge characteristics that influence the most the combination rule.

Scaling of the earthquake ground motions

The two orthogonal principal components of the selected ground motion earthquakes
listed in Table 1 were scaled to 2%/50 year uniform hazard spectra (UHS) using a loose spectral
technique. The two resultant spectra were then separated by scaling up and down the response
spectra to have a spectral ratio of the minor to the major component of 0.75 (Lopez et al. 2006).
The spectral accelerations at T = 0.1, 0.15, 0.2, 0.3, 0.4, 0.5, 1.0, 2.0, and 4.0 s are equal to,
respectively, 0.65, 0.71, 0.69, 0.5, 0.39, 0.34, 0.14, 0.048 and 0.024 for Montreal and to 0.80,
0.95, 0.96, 0.84, 0.74, 0.66, 0.34, 0.18 and 0.09 for Vancouver.

Table 2. Characteristics of the bridge models.

Bridge Number. Column Column cross-section

Type Bent system
model of spans height (m) (m2)
DPCU Straight Single column 2 6.0 Circular: Ø 1.2 m
DPRU Straight Single column 2 6.0 Rectangular : 1.2×2.4
DPM Straight Wall-type 2 6.0 Rectangular: 0.8×6.0
BPRU Skewed Single column 2 6.0 Rectangular : 1.2×2.4
BPM Skewed Wall-type 2 6.0 Rectangular: 0.8×6.0
CPRU Curved Single column 2 6.0 Rectangular: 1.6×2.8
DPRI Straight Multi-columns 4 6.0 / 10.0 Rectangular: 1.2×2.4
IRR-C Skewed/Curved Multi-columns 5 6.0 / 10.0 Circular: Ø 1.8 m
IRR-R Skewed/Curved Multi-columns 5 6.0 / 10.0 Rectangular: 1.2×2.2
Analysis and design procedures
Elastic response spectrum analyses of the bridge models were performed independently
in each of the longitudinal and transverse directions using the NBCC 2005 design spectrum
(Table 3) and the critical combined biaxial moments (Mx, My) were estimated using the
combination rule given by Eqs. (1) and (2). Three values of α were considered: (a) α = 0, (b) α =
0.3, and α = 1.0. The critical combined biaxial moments were also computed using multiple
linear modal time-history analyses with the pair of the scaled components of the earthquake
ground motion earthquakes as input.
The bridge columns were designed for both sites based on the seismic requirements of
the CSA-S6-06 standards (CSA 2006). The specified strength f’c and the modulus of elasticity
Ec of the concrete were set equal to 35 MPa and 26 000 MPa, respectively, whereas the yield
stress fy = 400 MPa and the modulus of elasticity Es = 200 000 MPa, were assumed for the steel
reinforcement. The critical seismic induced biaxial moments were reduced by the appropriate
reduction factor R, as prescribed by the CSA-S6: R = 3.0 for Mx and My, for single or multiple
ductile columns and 2.0 for My for wall-type piers. The required reinforcement ratio ρ was
determined for a combination of (P, Mx, My), where P is the axial load in the column. For each
bridge model and site, the optimum required longitudinal reinforcement ratios ρ0%, ρ30%, ρ100%,
and ρTH were computed adopting α = 0 for ρ0% , α = 0.3 for ρ30% , α = 1.0 for ρ100% , and based on
time-history analysis results for ρTH.
The results show that the required reinforcement ratios for the bridge columns designed
at the Montreal site, with α = 0 and α = 0.3, varies between 0.35% and 0.73%. These values are
below the lower limit of 0.8% specified in CSA-S6. That lower limit requirement is met for all
bridges located in Vancouver. The results also show that for both sites, the computed value of
ρ0% and ρ30% are nearly equal for the bridge models without skewed columns (DPCU, DPRU,
DPM, CPRU, and DPRI), the increase in steel being less than 4% among all cases when applying
the 30%-Rule (average increase = 1.7%). This difference is more pronounced in the case of the
bridge models with skewed columns (BPRU, BPM, IRR-C, and IRR-R) with an average increase
of 21% and 19% for Montreal and Vancouver, respectively. This suggests that the skew angle θ
of the bridge columns may have a significant influence on the weighted percentage α used in the
combination rule.

Effect of the column skew angle on the combination rule

To assess the significance of the effect of the column skew angle on the weighted
percentage α, a second series of analyses and designs were performed on a two spans simple
column skewed bridge (BPRU type bridge) by varying the column skew angle θ from 0˚ to 45˚
and following the same procedure used previously. The results are summarized in Table 3.
The required reinforcement ratios ρ0%, ρ30%, and ρ100% are compared to ρTH for both sites
in terms of the relative error given by:

⎛ ρ ⎞
Re lative Error = ⎜ % − 1⎟ ×100 (%) (3)
⎝ ρTH ⎠
Table 3. Required reinforcement ratio of the bridge column for different values of θ.

Bridge Montreal (east site) Vancouver (west site)

ρ0% ρ30% ρ100% ρTH ρ0% ρ30% ρ100% ρTH
o
BPRU_0 0.54 0.56 0.67 0.56 1.56 1.60 1.75 1.75
BPRU_10o 0.54 0.60 0.79 0.56 1.56 1.70 2.15 1.75
BPRU_20o 0.52 0.63 0.90 0.56 1.51 1.75 2.43 1.80
BPRU_30o 0.48 0.63 0.96 0.56 1.36 1.70 2.60 1.80
BPRU_35o 0.38 0.60 0.98 0.56 1.26 1.65 2.71 1.80
BPRU_45o 0.40 0.56 0.98 0.56 1.07 1.51 2.71 1.80

At both site, the relative error generally increases when increasing the skew angle to
reach approximately 30% for Montreal site and 40% for Vancouver site. For Montreal, the 30%-
Rule slightly overestimates the required amount of steel with a maximum relative error of 12.5%
for θ = 30o, whereas for Vancouver the 30%-Rule underestimates the required amount of steel
with a minimum relative error of -16.1% for θ = 45o.

Proposed weighted percentage α for regular skew bridges

An ideal weighted percentage is derived by determining, by linear interpolation between

the steel quantities ρ0%, ρ30%, and ρ100%, the percentage α that will give the value of ρTH obtained
from time-history analysis. For Montreal, the optimum weighted percentage increases nearly
linearly from 0 to 30% with the skew angle of the bridge column, with an average value of 20%,
whereas for Vancouver it remains almost constant, at an approximate value of 40%.
Based on the results presented in the previous sections, the following values of the
weighted percentage are suggested for the combination rule for design of regular bridges located
in eastern and western Canada.
Eastern Canada:
For skew angles less or equal to 10˚, α = 0%
For skew angles larger than 10˚, α = 20%
Western Canada:
For skew angles less or equal to 10˚, α = 0%
For skew angles larger than 10˚, α = 40%

Validation of the proposed combination rule

Multi-directional cyclic testing has been performed on half-scale bridge columns to

validate the proposed combination rule and the seismic design requirements for bidirectional
seismic loading, with special interest in the difference in demand anticipated from earthquakes
expected in eastern and western regions in North America. Four R/C 3000 mm tall bridge
column specimens having a 600 × 1200 mm rectangular cross-section were tested in the Hydro-
Québec Structural Engineering Laboratory at Ecole Polytechnique of Montreal. The four
specimens were divided in two sets. The first set, S1 and S2, representative of skewed columns
(θ = 30o) located in Montreal (east site) were designed with α = 0 and α = 0.3, whereas the
second set, S3 and S4, representative of skewed columns (θ = 45o) located in Vancouver (west
site) were designed with α = 0 and α = 0.4. Table 4 summarizes the design results.

Table 4. Design of the column specimens.

Provided
Column Combination Axial Load Longitudinal Transverse
Reinforcement
Specimen Rule Index Rebar ∗ Reinforcement
Ratio (%)
S1 100%-0% 0.06 0.41 42 - #3 #3 @ 45 mm
S2 100%-30% 0.06 0.57 32 - #4 #3 @ 45 mm
S3 100%-0% 0.06 0.94 34 - #5 #3 @ 40 mm
S4 100%-40% 0.06 1.72 32 - #7 #3 @ 45 mm
∗ Imperial bar sizes.

Two general bidirectional displacement-prescribed cyclic loading were developed based

on results of nonlinear time-history analyses. The bidirectional displacement-prescribed cyclic
loadings follow a “butterfly-type” path, as illustrated in Fig. 3(b), and were intended to
reproduce the estimated average orbital displacement of the prototype bridges (Fig. 3(a)) under
earthquakes expected in eastern and western regions of North America.

ΔT
(a) (b)
0.03
e c
0.02
Transverse displ. (m)
Longitudinal displ. (m)
Transverse displ. (m)

0.01
b
0 ΔL
-0.03 -0.02 -0.01 0 0.01 0.02 0.03
-0.01
g

-0.02
f d
-0.03
Longitudinal Displ.
Longitudinal displ.(m)
(m)

Figure 3. Bidirectional cyclic loading: (a) Orbital displacement of the prototype column;
(b) Displacement-prescribed loading path used for cyclic testing.
(a) (b)
1.4
E-W 1.2
1
0.8

Normalized Lateral Force V/Vy

0.6
0.4
0.2
0
-4 -3 -2 -1 -0.2 0 1 2 3 4
-0.4
-0.6
-0.8
-1
-1.2
-1.4
Drift Ratio Δ/L (%)

Figure 4. Experimental program: (a) Test setup; (b) Typical response of the column.

The test columns were subjected to a constant gravity load of 1566 kN and a
simultaneous displacement-prescribed loading in the two orthogonal directions using five high
performance structural actuators as illustrated in Fig. 4(a). In the test setup, constant gravity load
was applied using 2-1000 kN vertical actuators. Lateral displacement protocols were imposed
simultaneously along the X and Y axes (see Fig. 1) at the column top using three horizontal
actuators reacting against a 10 m tall L-shaped reaction wall. In-plane torsional rotation of the
column top was constrained to zero during the tests., such that response under the combination
of the bending moments about both X and Y directions, as determined from previous time-
history analyses could be examined. Typical response is shown in Fig. 4(b). The tests confirmed
the adequate behavior of the column specimens under the effects of combined seismic actions in
both orthogonal directions.

Conclusions

A research project consisting of a comprehensive analytical and experimental study has

been undertaken to examine the adequacy of the 30% combination rule to predict the seismic
demand on R/C bridge columns subjected to bidirectional earthquake ground motion. This study
is limited to regular bridge structures for which simple combination rules such as the Percentage
Rule are likely to be used in conjunction with response spectrum analysis in design. Based on the
results of this study, the following main conclusions were reached:
(i) The combination rule for predicting the seismic demand on regular bridge columns should
account for the effect of the skew angle and ground motion characteristics.
(ii) The value of the weighted percentage varies linearly with the skew angle for a bridge located
in eastern Canada (Montreal), whereas it remains nearly constant for a bridge located in western
Canada (Vancouver).
(iii) For an adequate prediction of the seismic demand on regular bridge columns under multi-
directional ground motion components, it is suggested to use a 20%-Rule for bridges located in
eastern site and a 40%-Rule for bridges located in western site, when the column skew angle
exceeds 10o.
(iv) Bi-directional testing was carried out that confirmed the adequacy of the proposed
combination rule. The columns exhibited satisfactory inelastic cyclic bi-directional response up
to high ductility levels way beyond the design level demand and were able to achieve their
predicted nominal flexural capacities up to the last cycle before failure.

References

American Association of State Highway and Transportation Officials (AASTHO), 1996. Standard
specifications for highway bridges, Division 1-A, Seismic Design., Washington, D.C.
Applied Technology Council (ATC), 2003. Recommended LRFD Guidelines for the Seismic Design of
Highway Bridges, Report No. ATC-49, Redwood City, CA.
Canadian Standard Association (CSA), 2006. CSA-S6 Canadian Highway Bridge Design Code, Rexdale,
ON.
Khaled, A., Tremblay, R., and Massicotte, B. 2009. Effectiveness of the 30%-Rule at predicting the
seismic demand on bridge columns subjected to bi-directional earthquake components. Earthquake
Spectra (submitted November 2009).
Lopez, O.A., Hernandez, J.J., Bonilla R., and Fernandez, A. 2006. Response spectra for multicomponent
structural analysis. Earthquake Spectra, 22(1): 85-113.
Lopez, O. A., and Torres, R., 1997. The Critical Angle of Seismic Incidence and the Maximum Structural
Response, Earthquake Eng. and Struct. Dyn., 26 (9), 881-894.
Menum, C., and Der Kiureghian, A., 1998. A Replacement for the 30%, 40%, and SRSS Rules for
Multicomponent Seismic Analysis, Earthquake Spectra, 14 (1), 153-163.
Menum, C., and Der Kiureghian, A., 2000a. Envelopes for seismic response vectors. I: Theory, J. Struct.
Engrg., ASCE, 126 (4), 467-473.
Newmark, N. M., 1975. Seismic Design Criteria for Structures and Facilities, Trans-Alaska Pipeline
System, Proceedings of First U.S. National Conference on Earthquake Engineering, Ann Arbor, MI,
94-103.
Penzien, J., and Watabe, M. 1975. Characteristics of 3-dimensional earthquake ground motions.
Earthquake Eng. and Struct. Dyn., 3 : 365-373.
Rosenblueth, E., and Contreras, H., 1977. Approximate Design for Multicomponent Earthquakes, J. Eng.
Mech. Div., ASCE, EM5 , 103 (5), 895-911.
Rosenblueth, E., 1951. A Basis for Aseismic Design, Doctoral Thesis, University of Illinois, Urbana, IL.
Wilson, E. L., Suharwardy, I., and Habibullah, A. (1995). A clarification of the orthogonal effects in a
three-dimensional seismic analysis, Earthquake Spectra, 11 (4), 659-666.