Вы находитесь на странице: 1из 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266466213

Seismic Force Demand on Brace Connections of Conventional Construction


Braced Steel Frames Designed According to Canadian Seismic Provisions

Article · May 2010


DOI: 10.1061/41130(369)124

CITATION READS

1 69

4 authors, including:

Robert Tremblay Sanda Koboevic


Polytechnique Montréal Polytechnique Montréal
226 PUBLICATIONS   2,967 CITATIONS    26 PUBLICATIONS   53 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NEESR: Reserve Capacity in New and Existing Low-Ductility Steel Braced Frames View project

Effect of P-delta on earthquakes respond of columns in steel moment-resisting frames. View project

All content following this page was uploaded by Sanda Koboevic on 21 October 2015.

The user has requested enhancement of the downloaded file.


2010 Structures Congress © 2010 ASCE 1361

Seismic force demand on brace connections


of conventional construction braced steel frames
designed according to Canadian seismic provisions

R. Tremblay1, P.X. Castonguay1, K. Guilini-Charette1, and S. Koboevic1


1
Group for Research in Structural Engineering, Department of Civil Geological and
Mining Engineering, Ecole Polytechnique, Montreal, QC, Canada, H3C 3A7.
Email: robert.tremblay@polymtl.ca

ABSTRACT

This paper presents the results of an ongoing research project on the seismic
performance of concentrically braced steel frames of the Conventional Construction
category designed according to NBCC 2005 and CSA-S16 seismic provisions. These
structures are expected to remain essentially elastic during strong ground motions.
Linear time history analyses were performed on typical braced steel frames of this
category to assess the force demand on brace connections. Different building heights
were considered. The structures were assumed to be located in both eastern and
western Canada to reflect two different seismic conditions. Two local soil conditions
were also examined: site classes C and E. The computed force demand varied with
the building height, the location and the site class. In several cases, it exceeded the
predictions from multi-modal response spectrum analysis, which is mainly attributed
to the assumptions made on damping and higher mode response. Recommendations
on design forces are given to prevent brittle failure of brace connections.

INTRODUCTION

The design of steel seismic force resisting systems in Canada must conform
to the seismic provisions of the 2005 National Building Code of Canada (NBCC)
(NRCC, 2005) and the design and detailing rules prescribed in the CSA-S16-01
standard for the design of steel structures (CSA 2005). In NBCC 2005, a ductility-
related force modification factor, Rd, is used to account for the capacity of the
various seismic force resisting systems to withstand severe earthquakes through
inelastic cyclic response. The value of Rd varies between 5.0 for the most ductile
systems to 1.0 for the less ductile ones (Mitchell et al., 2003). For steel structures of
the Conventional Construction category (Type CC structures), only the inherent
ductility of steel and other sources of energy dissipation present in ordinary steel
frames (e.g., friction) are relied upon to dissipate the seismic energy input. An Rd
factor of 1.5 is therefore specified in NBCC for these structures, and their response is
expected to be predominantly elastic with limited ductility demand. In regions of
moderate and high seismicity, these structures are also limited to 15 m in height,
which limits their use in many populated areas of Canada (Tremblay et al., 2009).
While structural steel members generally exhibit minimum ductility when
loaded beyond their factored resistance, this is not always the case for connections.
In concentrically braced steel frames, for instance, non ductile failure modes such as
2010 Structures Congress © 2010 ASCE 1362

weld failure or net-section fracture of braces can occur in brace connections, unless
the connections are specially designed and detailed to trigger ductile inelastic
response (e.g., Tremblay et al., 1996; Yang and Mahin, 2005; Martinez-Saucedo et
al., 2008; Fell and Kanvinde, 2009). In CSA-S16, capacity design procedure and
special detailing for ductile response are not prescribed for Type CC structures and
non ductile failure modes may therefore govern the performance of connections. In
order to prevent premature brittle brace connection failure and detrimental
consequences on structural integrity, CSA-S16 requires that the design seismic loads
for connections be amplified by 1.5 when the governing failure mode is not ductile.
These amplified design forces correspond to fully elastic seismic force level.
A research project has been initiated to understand better the seismic
behavior of Type CC structures in Canada (Tremblay et al., 2009). The study focuses
on concentrically braced steel frames as this system is among the most popular in
Canada due to its simplicity and relatively low cost. The main objectives of the
project are to assess the force and deformation demands on members and
connections in these structures and propose design recommendations to achieve
adequate seismic performance. Presently, very limited guidance is available to
engineers regarding the design of ductile connections for Type CC structures and the
use of amplified design connection loads has gained wide acceptance in practice.
This paper presents a study of the seismic force demand imposed on Type CC
braced steel frames used in building structures. Buildings with heights varying
between 2 and 10 stories and with split-X and chevron bracing configurations are
studied. The structures are located in two different seismic regions of Canada:
Montreal, QC, in eastern Canada, and Vancouver, BC, along the pacific west coast.
Both sites are representative of several major cities located in zones of moderate
seismicity in northeast and northwest areas of the United States. At both locations,
site classes C and E are investigated. The design of the structures is discussed first.
The analysis models and ground motions used in the analyses are then described.
Analysis results are presented and commented for each case. A discussion follows on
the high force demand that was observed in the buildings.
BUILDINGS STUDIED
Building Description. A total of 24 buildings were examined in the study: 11 and 13
for the Montreal and Vancouver locations, respectively. The building plan view
illustrated in Figure 1 was adopted for all structures. Other geometric parameters are
varied: number of stories (n = 2, 3, 4, 5, 6, 8, and 10), story height pattern, bracing
configurations (split-X and chevron bracing), and position of the bracing in the
building (on the perimeter, as shown in Figure 1, or along the first interior column
line). For each location, site classes C and E were examined. The braced frame
elevations for the 4-story split-X and chevron bracings are given in Figure 1 for the
normal story height pattern, together with the gravity loads and design spectra for
each location and site class. The normal height pattern is used for the 2-, 4-, 6-, 8-
and 10-story structures. For the 3- and 5-story structures, the story heights are
adjusted such that the total building height, hn, is the same as for the 4-story building
(15.6 m). This parameter was varied to examine whether building height or number
of stories affects most the structure seismic response.
2010 Structures Congress © 2010 ASCE 1363

N
T = 0.05 hn for n = 2 3,4,5 6 8 10
Studied
1.0
Frame Montreal - Site C
Montreal - Site E
Vancouver - Site C

S (g)
0.5 Vancouver - Site E
5 @ 9 m = 45 m

0.0
0.0 1.0 2.0 3.0
T (s)

3 @ 3.75 m
= 11.25 m
Gravity Loads (kPa)

Roof: Dead = 3.0


Snow = 2.48 (Montreal)
1.64 (Vancouver)
Floor: Dead = 3.5 kPa
4.35 m Partitions = 1.0 kPa
5 @ 9 m = 45 m Live = 3.8 kPa
Exterior walls = 1.2 kPa
Plan View Split-X Chevron

Figure 1. Building geometry and design loads.

Seismic Design. Automated design was performed according to NBCC 2005 and
CSA 2001 with the commercially available Graitec Advanced Design-America
program. The effects of earthquakes were determined using the response spectrum
dynamic analysis method. According to NBCC, the member forces from the analysis
are adjusted such that the base shear from the spectrum analysis is equal to Vd = VeIE
/ RoRd, where Ve is the elastic base shear from the response spectrum analysis, IE is
the importance factor (= 1.0 for the structures studied herein), and Ro and Rd are,
respectively, the overstrength and ductility force modification factors (Ro = 1.3, Rd =
1.5 for Type CC structures). For regular structures such as the ones studied herein,
the value of Vd in NBCC need not exceed 80% of the lateral earthquake design force,
V, given by V = S(Ta) Mv IE W / (RoRd). In this expression, S(Ta) is the design
spectral acceleration for the given site at the design fundamental period of vibration
of the building, Ta, MV is a factor accounting for higher mode effects, and W is the
seismic weight. The base shear V need not exceed 2/3 the value of V with Ta = 0.2 s
and must be greater than the value obtained with Ta = 2.0 s. The design period Ta can
be calculated using modal analysis but cannot exceed 0.05 hn.
In NBCC 2005, site specific 2% in 50 years uniform hazard spectral
ordinates, Sa, are given at periods of 0.2, 0.5, 1.0, and 2.0 s. Local soil effects are
accounted for by applying intensity- and frequency-dependent acceleration-based
and velocity-based site coefficients, Fa and Fv. The design spectrum, S(T), is equal to
FaSa(0.2) for T < 0.2 s, the smaller of FvSa(0.5) and FaSa(0.2) for T = 0.5 s, FvSa(1.0)
for T = 1.0 s, FvSa(2.0) for T = 2.0 s, and FvSa(2.0)/2 for T > 4.0 s. The S values for
intermediate period values are obtained by linear interpolation. For site class C, Fa =
Fv = 1.0. For site class E in Montreal, Fa = 1.17 and Fv = 2.06. For Vancouver, Fa =
0.95 and Fv = 1.84 for site class E. The resulting spectra are shown in Figure 1.
Seismic forces and deformations were obtained from 3D response spectrum
analysis, including accidental in-plane torsion. Inelastic P-Δ effects were also
considered. The design of the steel framing was then performed according to the
CSA-S16 standard. Detail of the design can be found in Castonguay (2009).
2010 Structures Congress © 2010 ASCE 1364

Lateral Overstrength. As of result of several factors such as code limits on


maximum design periods, minimum seismic loads, provisions for P-Δ effects and
minimum requirements for members, the actual lateral resistance of many of the
designed structures exceeded the anticipated seismic demand. Such lateral
overstrength may influence the performance of the buildings and should be taken
into account in the analysis of the results. When assessing the seismic force demand,
only the overstrength due to differences between design and expected seismic effects
is relevant. This contribution to overstrength can be represented by the ratio Vde/Ve,
where Vde is the design base shear Vd multiplied by the factors RoRd and Ve is the
elastic base shear from response spectrum analysis.
Table 1 gives the building height, hn, the upper limit on design period (= 0.05
hn), the first mode period obtained from modal analysis, T1, and the Vde/Ve ratio for
the various building cases studied. Note that only the structures with the braced
frames located along the building perimeter are considered in Table 1 (22 cases out
of 24); the two cases with braced frames located along interior column lines are
discussed later. As shown, with the exception of the 4-story split-X braced frame on
site class E, all structures in Montreal possess overstrength varying from 14% to
67%. In all cases, the upper limit on the period governed the calculation of Vde.
Lateral overstrength is typically larger for the more flexible structures exhibiting
longer fundamental periods T1. This is the case for the chevron bracing (generally
more flexible than the split-X geometry) and structures on site class C (seismic loads
lower compared to site class E). Similar trends are observed for Vancouver, but the
overstrength is generally less due to the higher seismic loads and, thereby, stiffer
structures. The Montreal buildings generally have dynamic base shears below the
80% of the equivalent static base shear, while the Vancouver buildings have dynamic
base shears between 80% and 100% of the equivalent static base shear.

Table 1. Fundamental period, T1 (in s), and overstrength ratio Vde/Ve (T1 - Vde/Ve).

Site n hn 0.05hn Montreal Vancouver


class (m) (s) Split-X Chevron Split-X Chevron
C 2 8.1 0.41 0.54 - 1.15 - 0.43 - 1.00 -
3 15.6 0.78 - 1.03 - 1.37 - 0.79 - 1.00
4 15.6 0.78 1.00 - 1.23 1.12 - 1.49 0.73 - 1.00 0.88 - 1.18
5 15.6 0.78 - 1.22 - 1.67 - 0.91 - 1.05
6 23.1 1.16 1.70 - 1.20 - 1.23 - 1.00 -
8 30.6 1.53 2.40 - 1.45 2.52 - 1.75 1.72 - 1.00 1.84 - 1.12
10 38.1 1.91 3.24 - 1.33 - 2.33 - 1.00 -
E 4 15.6 0.78 0.73 - 1.00 - 0.57 - 1.00 -
8 30.6 1.53 2.00 - 1.14 - 1.17 - 1.03 -

ANALYSIS
Numerical Models. Linear time history dynamic analysis was performed using the
OpenSees analysis platform (McKenna and Fenves, 2004). The analysis was
performed in the direction parallel to the braced frame studied (E-W in Figure 1)
2010 Structures Congress © 2010 ASCE 1365

using a two-dimensional model. The model included one of the two bracing bents
acting in the E-W direction, together with the leaning gravity columns laterally
supported by the braced frame studied. The columns were continuous over two
stories with semi-rigid splice connections. Rayleigh damping was specified with 3%
of critical damping in the first two modes of vibration. P-Δ effects were considered
in the analyses, with gravity loads consisting of the dead load plus 50% of the live
load and 25% of the roof snow load.
Selection and Scaling of Ground Motions. Each structure was subjected to an
ensemble of 10 synthetic ground motion time histories generated by Atkinson
(2009). For each ensemble, a total of 120 time histories were first chosen for five
different magnitude-distance scenarios dominating the hazard at the site. From those
large ensembles, the 10 records that best matched the design response spectrum
without scaling in the period range of 0.2-2.0 s for site class C, and 0.5-2.0 s for site
Class E, were then selected. The structures located in Vancouver were also subjected
to historical ground motions selected from the PEER database to match the dominant
magnitude-distance scenarios and local site conditions: 10 records for Class C and 12
records for Class E. These records were linearly scaled to fit the design spectra.
Additional information on ground motion selection and scaling can be found in
Guilini-Charette (2009). The resulting ground motion spectra are compared to the
design spectra in Figure 2. In the analyses, all synthetic and historical ground
motions were amplified by a factor corresponding to the effect of accidental torsion
considered in design on the first story shear in the bracing bent studied.

Montreal-Simulated Vancouver-Simulated Vancouver-Historical


a) 2.2 3 3
2
1.8 2.5 2.5
1.6
1.4 2 2
Sa (g)

Sa (g)

Sa (g)

1.2
1.5 1.5
1
0.8 1 1
0.6
0.4 0.5 0.5
0.2
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
T (s) T (s) T (s)

b) 2.2 3 3
2
1.8 2.5 2.5

1.6
2 2
1.4
Sa (g)

Sa (g)
Sa (g)

1.2
1.5 1.5
1
0.8 1 1
0.6
0.4 0.5 0.5
0.2
0 0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
T (s) T (s) T (s)

Figure 2. Design spectra (in black) and 5% damped absolute acceleration response
spectra of the selected ground motions (individual records in grey; median ground
motion response spectra in red): a) site class C; b) site class E.
2010 Structures Congress © 2010 ASCE 1366

ANALYSIS RESULTS

Reference Buildings. The structures with the normal story height pattern and built
with split-X bracing on site class C are examined first. These structures are used
thereafter as a reference to assess the influence of the other parameters studied.
Median (50th percentile), 84th percentile and maximum (100th percentile) values of
the computed brace axial loads, P, are presented at every level of each building for
each ground motion ensemble. The median values are used herein to assess the force
demand. The other percentile values are used to evaluate the variability in the
demand. The computed brace axial loads are normalized with respect to the axial
loads Pde corresponding to the design base shears Vde defined earlier. In CSA-S16,
brace connection loads for Type CC structures must be amplified by 1.5 when non
ductile failure is expected. This factor corresponds to the Rd value. Hence, a ratio
P/Pde equal to 1.0 means the force demand from the time history analysis is identical
to the expected resistance of a non-ductile connection, assuming that this resistance
is equal to 1.3 times the amplified seismic force used in design. Following the
approach propsed by Mitchell et al. (2003), it can be shown that this 1.3 factor
reflects dependable overstrength likely to be present in most steel connections.
In the figures, the buildings are designated using 13-digit labels. The first two
letters of the label correspond to the location of the building (Mo = Montreal, Va =
Vancouver). The third letter corresponds to the site class (C or E). The following
four digits correspond to the total building height in meters (8.2, 15.6, 23.1, 30.6, or
38.1). The following letter corresponds to the storey height pattern (N = normal, L =
low, when 5 stories are fit in hn = 15.6 m, and T = Tall, when 3 stories used in hn =
15.6 m). The next two numbers, followed by the letter “s”, correspond to the number
of stories (02s, 03s, 04s, 05s, 06s, 08s, or 10s). The following two letters refer to the
bracing system (SX = split-X or CV = chevron bracings). The final letter refers to the
position of the bracing bays (E = Exterior or I = Interior of the building).
Figures 3 and 4 give P/Pde ratios for the reference buildings in Montreal and
Vancouver (synthetic and historical records), respectively. In Montreal, the median
P/Pde values are generally below 1.0. However, in the 6- and 8-story structures, two
of the stories have median connection force demands greater than 1.0, indicating that
the current 1.5 amplification factor is adequate, provided that the structure meets the
current 15 m height limit prescribed for Type CC structures. The amplification factor
would need to be increased by 1.25 if structures of this type were to be used up to 10
stories. In Vancouver, the median P/Pde results vary between 1.06 and 1.47 for
synthetic ground motions and 0.96 and 1.50 for historical ground motions. In
general, the force demand tends to decrease with the building height. However,
structures in Vancouver are clearly subjected to greater connection force demands
for all building heights. The results suggest that the seismic design forces for non-
ductile connections in Type CC structures in Vancouver will have to be amplified by
an additional 1.5 factor. The generally lower demands in Montreal are due to the
inherent overstrength of the buildings at that location, as indicated by the Vde/Ve
ratios in Table 1 being larger than 1.0, versus being equal to 1.0 for most of the
buildings in Vancouver.
Influence of the Bracing Configuration. In Figure 5 the brace axial load demands
2010 Structures Congress © 2010 ASCE 1367

in split-X and chevron bracings are compared for the same 4- and 8-story building
applications in Vancouver. The results show that the demand is reduced when
chevron bracing is used. Similar reduction is observed for the same buildings in
Montreal. This reduction is likely due to the higher overstrength exhibited by the
more flexible chevron bracing structures. In spite of this reduction, connection
design forces would still need to be amplified by an additional 1.5 factor for Type
CC structures up to 8 stories in Vancouver.

10 MoC08.1N02sSXE
9 MoC15.6N04sSXE
8 MoC23.1N06sSXE
7 MoC30.6N08sSXE
Storey

6 MoC38.1N10sSXE
5
4
3
2
1
0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.6 0.8 1 1.2 1.4 1.6 1.8 2
P / Pde P / Pde P / Pde
(50th) (84th) (100th)

Figure 3. P/Pde ratios for split-X bracing on class C site in Montreal.


10
9
8
7
6
5 Synthetic Earthquakes
4
3
VaC08.1N02sSXE
2
VaC15.6N04sSXE
Storey

1
VaC23.1N06sSXE
10
9 VaC30.6N08sSXE
8 VaC38.1N10sSXE
7
6
5
Historical Earthquakes
4
3
2
1
0.8 1.2 1.6 2 0.8 1.2 1.6 2 0.8 1.2 1.6 2
P / Pde P / Pde P / Pde
(50th) (84th) (100th)

Figure 4. P/Pde ratios for split-X bracing on class C site in Vancouver.

Influence of the Site Class. The median axial load demands on brace connections in
split-X braced frames built on class C and Class E sites are compared in Figure 6. As
illustrated, higher demand is generally expected on site Class E and the differences
between the two site classes are as noticeable in Montreal as they are in Vancouver.
2010 Structures Congress © 2010 ASCE 1368

8
7
6
5
4
Synthetic Earthquakes

3
2 VaC15.6N04sSXE
Storey

1 VaC15.6N04sCVE
8 VaC30.6N08sSXE
7 VaC30.6N08sCVE
6
5
Historical Earthquakes
4
3
2
1
0.9 1.2 1.5 1.8 2.1 0.9 1.2 1.5 1.8 2.1 0.9 1.2 1.5 1.8 2.1
P / Pde P / Pde P / Pde
(50th) (84th) (100th)

Figure 5. P/Pde ratios for split-X and chevron bracings on class C site in Vancouver.

a) b) 8
7
6

MoC15.6N04sSXE 5
MoE15.6N04sSXE 4
Synthetic Earthquakes
MoC30.6N08sSXE 3
MoE30.6N08sSXE
2 VaC15.6N04sSXE
Storey

1 VaE15.6N04sSXE
8 8 VaC30.6N08sSXE
7 7 VaE30.6N08sSXE
6 6
Storey

5 5
Historical Earthquakes
4 4
3 3
2 2
1 1
0.8 1.2 1.6 2 1 1.6 2.2 2.8 3.44
P / Pde P / Pde
(50th) (50th)

Figure 6. Influence of site class on P/Pde ratios for split-X bracing in:
a) Montreal; b) Vancouver
2010 Structures Congress © 2010 ASCE 1369

For Montreal, this can be explained in part by the lower overstrength present in the
structures constructed on site Class E. In Vancouver, a greater increase in force
demand is observed for the 8-story building compared to the 4-story one. From the
figure, it can also be seen that the synthetic earthquakes in Vancouver yielded
significantly larger P/Pde ratios for site class E compared to the historical records.
Based on these results, conventional construction with non-ductile connections is not
suitable for site classes E in either Montreal or Vancouver. Alternatively, the
structures could be used provided that the current 1.5 amplification factor be
increased by 30% in Montreal and 50% in Vancouver. These numbers would allow
structures of any height up to 30.1 m and structures up to 15 m in height to be built
in Montreal and Vancouver, respectively.
Influence of the Story Height Pattern and Braced Frame Location. The influence
of varying the story height pattern and number of stories for the 15.6 m tall chevron
bracing structures on class C site in Vancouver is illustrated in Figure 7. For the
median values, no obvious trend exists between the three different story height
patterns other than lower maximum P/Pde ratios over the structure height are obtained
for the 5-story structure under the historical earthquake records. Similar observations
were made from the results of the analyses for the same three 15.6 m tall structures
in Montreal. Hence, it can be concluded that the story pattern does not influence
significantly the elastic brace force demand for 15 m tall Type CC structures.
The influence of moving the bracing bent from an exterior column line to the
adjacent interior column line was studied for the 4- and 8-story split-X braced frames
on site C in Vancouver. The results showed that there is no significant difference, or
tendency, between interior and exterior braced bent locations.
5

3 Synthetic Earthquakes

2
VaC15.6T03sCVE
Storey

1 VaC15.6N04sCVE
5 VaC15.6L05sCVE

3 Historical Earthquakes

1
0.9 1.2 1.5 1.8 2.1 0.9 1.2 1.5 1.8 2.1 0.9 1.2 1.5 1.8 2.1
P / Pde P / Pde P / Pde
(50th) (84th) (100th)

Figure 7. Influence of story height pattern on P/Pde ratios for chevron bracing on
site class C in Vancouver.
2010 Structures Congress © 2010 ASCE 1370

VALIDATION OF HIGH FORCE DEMAND RESULTS


Peak P/Pde values greater than 1.0 were routinely obtained from the linear
time history analyses performed on the various braced frames studied. This means
that force demands in the brace connections can exceed the forces corresponding to
fully elastic seismic response. Such high force demand was observed at the median
level, which, from a design standpoint, is an unexpected result as elastic response is
expected to represent an upper bound when the selected ground motion ensemble
matches the design level. This condition was generally met in this study as the
median acceleration spectra of all ground motion ensembles generally agreed well
with their associated design spectra (see Figure 2). Other reasons that can explain
this apparent discrepancy include: 1) the difference between the damping levels
assumed in design and analysis, and 2) the differences between the dynamic analysis
methods used for design and for seismic response assessment. Inspection of the time
history analysis results also reveals that the highest P/Pde values are often obtained in
the upper part of the structures, suggesting that consideration of the higher vibration
mode contribution may also have contributed to the relatively high force demand.
The response of a 4-storey split-X braced frame was carefully examined to
validate these assumptions. The structure was assumed to be located on a class C site
in Vancouver and was subjected to the 1989 Loma Prieta, Stanford University 360˚
ground motion record. That record is referred to herein as Record V17. The structure
periods in its first four lateral modes of vibration are 0.74 s, 0.32 s, 0.20 s, and 0.17
s, and the associated modal participation mass ratios are 0.75, 0.17, 0.072 and 0.002.
Hence, 92% of the total horizontal mass is included when considering the first two
vibration modes, which is sufficient to meet the 90% minimum mass participation
generally prescribed by codes when using response spectrum analysis.
A total of 6 different analyses of the structure were performed using the
SAP2000 computer program (CSI, 2009): 1) RSA-NBCC: Response spectrum
analysis (RSA) using the NBCC design spectrum; 2) RSA-5%: RSA using the 5%
damped spectrum of the V17 record; 3) RSA-3%: RSA using the 3% damped
spectrum of the V17 record; 4) MSTHA-3%: modal superposition time history
analysis assuming 3% damping in all modes; 5) MSTHA-R3%: same as 5) with 3%
damping in the first two vibration modes; and 6) DITHA-R3%: direct integration
(Newmark-Beta) time history analysis with 3% Rayleigh damping in the first two
vibration modes. For the last two analyses, the damping in the 3rd mode and 4th mode
are 3.9% and 4.3% of critical, respectively. The first method was used in design
whereas the last method was used in the analyses described in the previous section.
In Figure 8a, the NBCC design spectrum is compared to the 5% and 3%
damped ground motion acceleration spectra. The periods of the structure are also
shown in the graph. Differences exist between the NBCC and 5% ground spectra, but
those are expected to vanish when using an ensemble of properly selected and scaled
ground motions. As expected, reducing the damping from 5% to 3% increases the
acceleration and force demand from the ground motion. The brace forces obtained
from all 6 analysis methods at all 4 levels are presented in Figure 8b. The differences
between the RSA-NBCC and RSA-5% values are small and reflect the differences
between the design and ground motion spectra. A larger difference exists between
the RSA-5% and RSA-3% methods. The use of the lower damping increases by 20%
2010 Structures Congress © 2010 ASCE 1371

to 21% the forces in the four braces. This represents a very consistent increase over
the building height. However, this increase is larger than what would be predicted
using expressions found in design documents to account for the influence of
damping on spectral ordinates. For instance, the equation proposed in EC8 (CEN,
2004) predicts a 13% increase when reducing the damping from 5 to 3%.
In Figure 8b, the three time history analysis methods produced very similar
results for the four braces, confirming that the differences in damping modeling had
no significant impact on the brace force demand. At all levels but Level 4, the time
history results agree very well with the RSA-3% predictions, the differences being
less than 3%. At Level 4, time history analyses give forces 18 to 20% larger than the
RSA-3% method. In Figure 8c, it is shown that this is due to the fact that the peak
brace force occurs when first and second mode brace forces reach large values of the
same sign at the same time. For this situation, statistical modal combination rules
such as the SRSS or CQC methods generally used in response spectrum analysis will
underestimate the peak response. This study shows that the effects of actual damping
being less than 5% critical and, for components subjected to higher mode response,
the likelihood of having simultaneous peak modal responses must be properly
addressed in order to accurately predict the member or connection force demand. In
the design of ductile structural components, force demand in excess of the elastic
force level can be accommodated by a reduction of the seismic force modification
factor used for seismic design. Such an approach is not appropriate, however, for the
design of connections governed by a non ductile failure mode.

a) c) MSTHA-3%
T4 T3 T2 T1 2000
DITHA-R3%
P (kN)

1.2 V17-3% Damping


V17-5% Damping 0
NBCC
Sa (g)

0.8
-2000

0.4 0.3
ag (g)

0.0
0
0.0 0.5 1.0 1.5 V17 Record
-0.3
T (s)
b) 8000 0 5 10 15 20 25 30
RSA-NBCC
Time (s)
RSA-5% 3000
RSA-3%
6000 MSTHA-3% 2000 Mode 4 Mode 2
MSTHA-R3%
1000
P (kN)

DITHA-R3%
P (kN)

4000 0

-1000
2000 Mode 1
-2000 MSTHA-3%
DITHA-R3%
-3000 Max
0
1 2 3 4 10.0 10.5 11.0 11.5 12.0
Level Time (s)

Figure 8. Brace force response from RSA and THA analysis methods: a) Design
and ground motion acceleration spectra; b) Brace axial load; c) Brace force at
Level 4 from MSTHA-3% and DITHA-R3% methods.

CONCLUSIONS
A parametric study was performed to assess, through linear time history
analysis, the force demand in brace connections of Type CC braced steel frames up
2010 Structures Congress © 2010 ASCE 1372

to 10 stories in height designed according with Canadian seismic provisions. The


objective was to propose seismic loads for the design of brace connections that are
characterized by non ductile failure modes, i.e., connections that must remain
essentially elastic in order to prevent premature fracture.
The analysis showed that non-ductile brace connections in 2- to 10-story
braced frames located on class C sites in Montreal and Vancouver should be
designed for forces equal to 1.25 and 1.5 times, respectively, the amplified
connection design forces currently specified in CSA-S16. Larger additional
amplification is required for structures built on a site class E. In Montreal, the
additional amplification can be waived for structures less than 15 m in height on
class C sites. Smaller forces were obtained in more flexible structures and/or
structures exhibiting higher lateral overstrength. The study also revealed that further
study is needed on the influence of damping and higher mode response on the elastic
force demand imposed on members and connections of multi-story braced frames.
REFERENCES
Atkinson, G. M. (2009). “Earthquake Time Histories Compatible with the 2005 NBCC Uniform
Hazard Spectrum.” Can. J. of Civ. Eng., 36(6), 991-1000.
Castonguay, P.X. 2009. Seismic Performance of concentrically braced steel frames of the
conventional construction category. M.Sc. Thesis, Dept. of Civil, Geological and Mining Eng.,
Ecole Polytechnique, Montreal, QC, Canada.
CEN. (2004). EN 1998-1:2004:E, Eurocode 8: Design of structures for earthquake resistance - Part
1: General rules, seismic actions and rules for buildings. Comité Européen de Normalisation,
Brussels, Belgium.
CSA. (2005). CAN/CSA-S16.1 Limits States Design of Steel Construction, Including CSA-S16S1-05
Supplement No. 1, Canadian Standards Association, Willowdale, ON.
CSI. (2009). SAP2000 - Integrated Software for Structural Analysis and Design. Computer &
Structures, Berkeley, CA.
Fell, B.V., and Kanvinde, A.M. 2009. “Tensile forces for seismic design of braced frame connections
_ Experimental results.” J. Constr. Steel Res., doi:10.1016/j.jcsr.2009.10.002
Guilini-Charette, K. 2009. Effets des mouvements sismiques sur les structures en acier de la catégorie
des constructions conventionnelles. M.Sc. Thesis, Dept. of Civil, Geological and Mining Eng.,
Ecole Polytechnique, Montreal, QC, Canada. (in French)
Mitchell, D., Tremblay, R., Karacabeyli, E., Paultre, P., Saatcioglu, M., and Anderson, D. L. (2003).
“Seismic force modification factors for the proposed 2005 edition of the National Building Code
of Canada.” Can. J. of Civ. Eng., 30(2), 308-327.
McKenna, F. and Fenves, G.L. 2004. Open System for Earthquake Engineering Simulation
(OpenSees). Pacific Earthquake Engineering Research Center (PEER), University of California,
Berkeley, CA. (http://opensees.berkeley.edu/index.html)
Martinez-Saucedo, G., Packer, J.A., and Christopoulos, C. 2008. “Gusset Plate Connections to
Circular Hollow Section Braces under Inelastic Cyclic Loading.” J. Struct. Eng., ASCE, 134(7),
1252-1258.
NRCC. (2005). National Building Code of Canada, 12th ed., National Research Council of Canada,
Ottawa, ON.
Tremblay, R., Castonguay, P.X., Guilini-Charette, K., and Koboevic, S. (2009). “Seismic
Performance of Conventional Construction Braced Steel Frames Designed According to
Canadian Seismic Provisions.” Proc. 2009 ASCE Structures Congress, Austin, TX, 341, 87.
Tremblay, R., Bruneau, M., Nakashima, M., Prion, H.G.L., Filiatrault, A., and DeVall, R. 1996.
“Seismic Design of Steel Buildings: Lessons From the 1995 Hyogoken-Nanbu Earthquake.”
Can. J. of Civ. Eng., 23(3), 727-756.
Yang, F., and Mahin, S. 2005. Limiting net section fracture in slotted tube braces. Steel Tips
Technical Information and Product Service, Structural Steel Educational Council, Moraga, CA.

View publication stats

Вам также может понравиться