Seismic Design
Seismic Design
Seismic Design
................................................................................... 90
4.2.4 Nonlinear Dynamic Analysis Results ................................................ 91
4.2.5 Probability of Brace Demand Exceeding Connection
Capacity ............................................................................................. 93
4.3 Monte Carlo Dynamic Analyses of Single-Story Special
Concentrically Braced Frame Including Brace Connection
Failure ........................................................................................................ 98
4.3.1 Description of Analysis ..................................................................... 98
4.3.2 Monte-Carlo Dynamic Analysis Results ......................................... 101
4.3.3 Probability of Collapse Including Connection Fractures ................ 102
4.4 Conclusions ............................................................................................. 103
Chapter 5 Capacity-Based Design in Multi-Story Special
Concentrically Braced Frames ............................................................................ 127
5.1 Introduction ............................................................................................. 127
5.2 Dynamic Analysis of a 6-Story and a 16-Story Special
Concentrically Braced Frames ................................................................ 128
5.2.1 Description of Analysis ................................................................... 128
5.2.2 Brace Behavior ................................................................................ 129
5.2.3 Calculating R
................................................................................. 130
5.2.4 Nonlinear Dynamic Analysis: Force Demands on Brace
Connections...................................................................................... 134
5.2.5 Nonlinear Dynamic Analysis: Required Design Strength
of Brace Connections ....................................................................... 136
xiii
5.2.6 Nonlinear Dynamic Analyses: Axial Force Demands on
Columns ........................................................................................... 138
5.3 Dynamic Analysis of an Alternative Design of a 6-Story
Special Concentrically Braced Frame ..................................................... 144
5.3.1 Description of Analysis ................................................................... 144
5.3.2 Nonlinear Dynamic Analysis: Force Demands on Brace
Connections ..................................................................................... 145
5.3.3 Nonlinear Dynamic Analysis: Required Design Strength
of Brace Connections ....................................................................... 146
5.3.4 Nonlinear Dynamic Analyses: Axial Force Demands on
Columns ........................................................................................... 146
5.4 Probability of Collapse Including Connection Failures .......................... 147
5.5 Conclusions ............................................................................................. 149
Chapter 6 Application of Capacity Design Factor Methodology ....................... 187
6.1 Introduction ............................................................................................. 187
6.2 Digest of Proposed Methodology ........................................................... 188
6.3 Required Design Strength of Brace Connections ................................... 193
6.4 Conclusions ............................................................................................. 196
Chapter 7 Summary, Conclusions, Limitations and Future Work .................... 207
7.1 Summary and Conclusions ..................................................................... 207
7.1.1 Expected Demand on Capacity-Designed Components .................. 208
7.1.2 System Design Factors and Member Overstrength ......................... 210
7.1.3 Seismic Hazard Curve ..................................................................... 211
7.1.4 Component and System Reliability ................................................. 212
7.1.5 Capacity-Based Design of Brace Connections in SCBFs .............. 214
7.1.6 Capacity-Based Design of Columns in SCBFs .............................. 215
7.1.7 Column Demand in Tall Building Initiative.................................... 215
7.2 Limitations and Future Work .................................................................. 216
7.3 Concluding Remarks ............................................................................... 220
Notation List .............................................................................................................. 221
References .................................................................................................................. 224
Appendix A Incremental Dynamic Analysis of Low - Redundancy
Single-Story SCBF ................................................................................................. 233
xiv
Appendix B Conditional Mean Spectrum Effects on Capacity-
Designed Components ........................................................................................... 247
Appendix C Component Reliability Probability Formulas ................................ 254
Appendix D Special Moment Frame Connections ............................................... 257
Appendix E Collected Statistical Data: Material Properties,
Connection Capacity in SCBF and SMRF .......................................................... 265
xv
List of Tables
Table 2-1: Summary of Capacity Design Requirements in the AISC (2010)
Seismic Provisions.................................................................................... 35
Table 2-2: Summary of Capacity Design Requirements in the ACI 318-08 (2008) .. 36
Table 2-3: Relationship between probability of failure and the reliability index, .. 36
Table 3-1: Probability in 50 years that frames (T
1
= 0.2s) located in San
Francisco or New Madrid, will experience yielding of members based
on 2008 USGS hazard maps and R
......................................................... 65
Table 3-2: Target
R,Ha
-values to use when establishing capacity design factors.
The target
R,Ha
-values depend on R
...................................................................... 73
xvi
Table 3-13: Steps to calibrate the capacity design factors following the proposed
methodology when component strength probability distributions are
lognormal ................................................................................................. 74
Table 4-1: Properties of the two single-story braced frames investigated in
Chapter 4 ................................................................................................ 106
Table 4-2: Far-field loading protocol used to analyze brace behavior .................... 106
Table 4-3: Near-field tension and compression loading protocols used to analyze
brace behavior ........................................................................................ 107
Table 4-4: Estimation of R
using the ground motion sets median response spectrum ... 157
Table 5-9: Median of the normalized maximum brace tensile force vs. Sa
T1
.for 6-
story SCBF Design 1 ........................................................................... 158
Table 5-10: Dispersion of the normalized maximum brace tensile force vs.
Sa
T1
.for 6-story SCBF Design 1 .......................................................... 158
Table 5-11: Median of the normalized maximum brace tensile force vs. Sa
T1
.for
16-story SCBF ........................................................................................ 159
Table 5-12: Dispersion of the normalized maximum brace tensile force vs.
Sa
T1
.for 16-story SCBF ........................................................................... 160
Table 5-13: |/ -ratios for 6-Story SCBF - Design 1 calculated by both full
integration and by simplified method proposed in the methodology ..... 161
Table 5-14: |/ -ratios for 16-Story SCBF calculated by both full integration and
by simplified method proposed in the methodology .............................. 161
Table 5-15: Member sizes for the 6-Story SCBF Design 2 .................................... 161
Table 5-16: Summary of calculations performed to calculate the 6-story Design 2
frames R
and P(Coll
D>C
) in 50 years of
0.1% for selected connection failure modes in SCBFs located in San
Francisco. | = 0.75 ................................................................................. 200
Table 6-4: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of
0.1% for selected connection failure modes in SCBFs located in New
Madrid. | = 0.75 .................................................................................... 201
Table 6-5: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of
0.1% for selected connection failure modes in SCBFs located in San
Francisco. | = 0.75 ................................................................................. 202
Table 6-6: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of
0.1% for selected connection failure modes in SCBFs located in New
Madrid. | = 0.75 .................................................................................... 203
Table A-1: Frame properties .................................................................................... 238
Table A-2: Median and COV of normalized maximum brace forces, P
max
/P
y,exp
,
from analysis .......................................................................................... 238
Table A-3: MAFs for Frame 1 and various connection strengths ........................... 239
Table A-4: MAFs for Frame 2 and various connection strengths ........................... 239
Table B-1: Sa
T1
associated with each Run in Figures B-1 to B-4 ......................... 249
Table B-2: Median and COV of normalized maximum brace forces, P
max
/P
y,exp
from both analyses ................................................................................. 249
Table D-1: Bolted Moment End-Plate Connection capacity data used in reliability
analysis ................................................................................................... 260
Table D-2: Bolted Moment End-Plate Connection demand data used in reliability
analysis ................................................................................................... 260
Table D-3: Recommended | -factors for selected failure modes in bolted moment
end-plate connections in SMRF based on collected statistical data on
demand and capacity, R
system of 4
located in San Francisco and New Madrid. The MAF of exceeding the
design spectral acceleration has increased with the new hazard curves
for New-Madrid but decreased for San Francisco. ................................... 82
Figure 3-13: /|-ratio sensitivity to R
and V
tot
for frames located in San Francisco.
C
m
/C
n
and D
m
/D
n
= 1.0 ............................................................................. 83
Figure 3-16: /|-ratio sensitivity to R
and V
tot
for frames located in San Francisco.
C
m
/C
n
and D
m
/D
n
= 1.0 ............................................................................. 84
Figure 3-17: Ratio of calculated /|-values calculated based on being located in
San Francisco to those based on being located in New Madrid. .............. 84
Figure 4-1: SCBF analyzed for this example a) Plan b) Elevation ........................... 111
xxii
Figure 4-2: OpenSees model of braces. .................................................................... 112
Figure 4-3: Earthquake response spectra for the 44 ground motions used for the
Incremental Dynamic Analysis. The ground motions records are all
scaled to have the same spectral acceleration at the first mode period
of the frames. ......................................................................................... 112
Figure 4-4: Response of the OpenSees model of a HSS6x6x5/16 brace section
when subjected to a far-field loading protocol. E0 and m are
parameters of the fatigue material used. Brace fracture occurs at
relatively low axial deformations ........................................................... 113
Figure 4-5: Far-field loading protocol developed by Fell et al (2006) and used to
analyze brace behavior running dynamic analysis of SCBF frames ...... 113
Figure 4-6: Response of the OpenSees model of a HSS6x6x5/16 brace section
when subjected to a near-field tension loading protocol. E0 and m are
parameters of the fatigue material used. ................................................ 114
Figure 4-7: Near-field tension loading protocol developed by Fell et al (2006) and
used to analyze brace behavior running dynamic analysis of SCBF
frames ..................................................................................................... 114
Figure 4-8: Response of the OpenSees model of a HSS6x6x5/16 brace section
when subjected to a near-field compression loading protocol. E0 and
m are parameters of the fatigue material used. ...................................... 115
Figure 4-9: Near-field compression loading protocol developed by Fell et al
(2006) and used to analyze brace behavior running dynamic analysis
of SCBF frames ...................................................................................... 115
Figure 4-10: Results from pushover analysis on Frame 1 showing normalized base
shear on the left y-axis, estimated R
= < =
0
dx x f x F Q R P P
Q R f
(2-2)
18
where F
R
is the cumulative probability distribution function for R and f
Q
the probability
density function for Q. As implemented in the US design specifications, the LRFD
reliability scheme does not explicitly calculate probabilities of failure. Instead it estimates
the so called reliability index, , through the first order second moment (FOSM) method.
If the probability distributions of Q and R are normal and if Q and R are
statistically independent, the expression for the reliability index is:
2 2
Q R
m m
Q R
o o
|
+
= (2-3)
where Q
m
and R
m
are the mean values of load and resistance, respectively, and
Q
o and
R
o
their standard deviations. The randomness in Q arises due to randomness in load
intensities, uncertainties in the transformation of loads into load effects, and uncertainties
in structural analysis. The randomness in R arises due to variations in mechanical
properties of materials, variations in dimensions and of uncertainties in the theory
underlying the design definition of nominal member strength. (Ravindra and Galambos
1978) Q
m
and R
m
are therefore the products of the following parameters:
) (
[
=
Qi
x n m
Q Q (2-4)
) (
[
=
Ri
x n m
R R (2-5)
where the s are the mean values of the statistical parameters which represent the
randomness of Q and R, relative to their nominal values, Q
n
and R
n
If Q and R both have lognormal distributions, the reliability index is:
2
ln
2
ln
ln
Q R
m
m
Q
R
o o
|
+
|
|
.
|
\
|
= (2-6)
19
where
m
Q
and
m
R
\
|
= | (2-7)
The reliability index, , is a relative measure of reliability. As demonstrated in
Figure 2-2, if the limit state function is Z = R Q where z < 0 represents failure and R
and Q are normally distributed variables, then is the number of standard deviations (o
z
)
that the mean of Z (
z
) is from zero (i.e.,the limit state failure criterion). If R and Q are
statistically independent and all possible uncertainties are included in the analysis, then ,
as calculated using Equation 2-3 for normally distributed variables and 2-6 for
lognormally distributed variables, can be related to the probability of failure. Table 2-3
and Figure 2-3 show the relationship between the probability of failure and the reliability
index, , calculated with the following equation, where
1
u is the inverse cumulative
standard normal distribution function:
( )
f
P
1
u = | (2-8)
When only a subset of uncertainties are considered or when the probability
distributions of R and Q are non-normal, Table 2-3 can still be used to approximate the
probability of failure. However, as the results are not exact in such cases, the approximate
probabilities are more appropriate for relative comparisons as opposed to absolute
measurement of the failure risks.
Once the statistical parameters have been established, typically through a combination of
experimental data and judgment, the reliability index (or the probability of failure) of
each limit state can be calculated for a specified combination of | and (or alternatively,
20
the ratio of |/). Thus, | and/or can then be varied to achieve the desired reliability
index. Alternatively, if a certain reliability index, , has been decided upon, the target
safety level is achieved if:
| |
m Q R m
Q V V R
2 2
exp + > | (2-9)
The factor | |
2 2
exp
Q R
V V + | can be considered as a total load and resistance factor.
In the AISC LRFD this total factor was approximated by the product of two factors,
| |
R
V o| exp and | |
Q
V o| exp , in which is a coefficient equal to about 0.7 called a
splitting factor. With this approximation, one can write the safety check as:
| | | |
m Q m R
Q V R V o| o| exp exp > (2-10)
Adaptation of LRFD to Demand and Capacity Terminology: The LRFD reliability
method described above can be adapted to calculate the conditional reliability of
capacity-designed components in structural systems where the reliability is conditioned
upon forces induced by seismic deformation demands on the structure. For consistentcy
with capacity-based design terminology, the terms R and Q are replaced with capacity, C,
and demand, D, respectively. The term C represents the capacity of the force-controlled
member, e.g. a connection in a braced frame, and D represents the force demands induced
by yielding of one or more deformation-controlled members, e.g. the brace force in a
braced frame. Similarly, | and will be referred to as capacity and demand factors.
Lognormal probability distributions are often chosen to describe the behavior of
structural components, e.g. structural steel behavior (Galambos et al., 1982). Following
lognormal probability distribution assumptions, the expression for the reliability index of
capacity-designed components, conditioned upon a specific system deformation demand,
is as follows (see Appendix C for a more detailed derivation of Equation 2-11 to 2-14):
D C D C
m
m
V V V V
D
C
|
2
ln
2 2
+
|
|
.
|
\
|
= (2-11)
21
|
|
|
.
|
\
|
|
|
.
|
\
|
=
|
|
.
|
\
|
n
m
m
n
m
m
C
C
D
D
D
C
(2-12)
=
2
Ci
x C
V V (2-13)
=
2
Di
x D
V V (2-14)
where
m
D
and
m
C
, and
n
D and
n
C are the median and nominal values of the demand
and capacity probability distributions, respectively,
D
V and
C
V their lognormal standard
deviations, and is the correlation coefficient between demand and capacity. Equations
2-13 and 2-14 assume that the random variables representing demand are uncorrelated
and likewise for the random variables representing capacity. In most cases, the
correlation coefficient between demand and capacity is zero, since the demand and
capacity effects are independent. However, in some cases is non-zero, where there is
some correlation between the random factors affecting demand and capacity. Examples
where the demand and capacity are correlated include rupture of net section in a yielding
brace connection of an SCBF and the plastic hinge capacity of a beam outside of a
yielded link in an EBF. These are both cases where the demand and capacity are
controlled by the same member and therefore variations in material strength and member
dimensions are correlated between demand and capacity. Similarly, in reinforced
concrete walls, there is a correlation between the wall shear capacity and the induced
shear forces that are limited, to some extent, by the flexural strength of the wall. Strong-
column weak beam requirements of concrete frames are another example where the force
demands imposed by the beams have some correlation to the capacity of the column.
However, since the strength of the concrete members depends on certain components that
are common to the demand and capacity (e.g., concrete and workmanship) and others that
are independent (different reinforcing bars in beams/columns and wall shear/flexure),
these examples show that there is not always a clear cut distinction with regards to degree
of correlation between demand and capacity. In the absence of other information, it is
proposed to assume equal to 0 for components that are clearly independent and
equal to 0.5 for components where some correlation is likely to exist.
22
For the reliability calculations of capacity-designed components, small variance
approximations are not taken advantage of when lognormal probability distributions
apply, as the LRFD methodolgoy does, and the median and lognormal standard deviation,
often called the dispersion, of statistical parameters used instead (
D
V and
C
V ). However,
since variance is all likelihood sufficiently small, if mean and coefficient of variation are
used instead of median and lognormal standard deviation, the results should be
approximately the same.
Assuming the demand and capacity probability distributions are normal (as opposed
to lognormal), the expression for the reliability index of capacity-designed components,
conditioned upon a specific deformation demand, is:
D C D C
m m
D C
o o o o
|
2
2 2
+
= (2-15)
|
|
|
.
|
\
|
|
|
.
|
\
|
=
|
|
.
|
\
|
n
m
m
n
m
m
C
C
D
D
D
C
(2-16)
=
2
Ci
x C
o o (2-17)
=
2
Di
x D
o o (2-18)
where D
m
and C
m
are the mean values of load and resistance, respectively, and
D
o and
C
o are their standard deviations.
2.3.2 SAC Demand and Capacity Factor Design
The SAC Demand and Capacity Factor Design (SAC-DCFD) methodology is a
reliability methodology that was developed as a part of the SAC Federal Emergency
Management Agency (SAC/FEMA) program (Cornell et al 2002, Yun et al, 2002). It
provides probabilistic framework for seismic design and assessment of structures,
recognizing the dynamic and nonlinear displacement basis of the problem, such that a
23
specified performance objective is achieved. The performance objective is expressed as
the probability of exceeding a specified performance level. Similar to LRFD, the SAC-
DCFD methodology begins with a probabilistic model of demand and capacity and ends
with numerical factors (called demand and capacity factors) that are applied to mean or
nominal values of the demand and capacity. The framework ensures that if the factored
capacity exceeds the factored demand then the specified performance objective is
achieved.
The SAC-DCFD design criterion is based on satisfying the following condition:
)
`
>
)
`
2
1
exp
2
1
exp
2 2 P
m S D m C
D
b
k
C
b
k
a
| | (2-19)
which can be expressed in terms of factored demand and capacity as:
0
P
m m
D C | > (2-20)
where
0
P
m
D is the median drift demand under a given ground motion intensity, which in
turn is defined as the spectral acceleration with annual probability of P
0
of being
exceeded,
m
C
is the median drift capacity, k and b are local slopes of the seismic
hazard curve (Sa versus mean annual frequency of recurrence) and of the median
deformation demand curve (peak drifts versus Sa), respectively. The s are the standard
deviations of the natural logs of demand, D, and capacity, C, which are numerically equal
to the coefficients of variation, COVs, for smaller values, e.g. V<0.3. The SAC/FEMA
use of follows the tradition of the nuclear industry where the early roots of some of
these developments lie. However, in order to avoid confusion with the reliability index
from the LRFD scheme and to be consistent with LRFD, the standard deviations of the
natural logs of D and C are called Vs here (assuming small variance) and the SAC-
DCFD equation becomes:
)
`
>
)
`
2
1
exp
2
1
exp
2 2 P
m S D m C
D V
b
k
C V
b
k
a
(2-21)
24
The similarities between the SAC-DCFD equation (2-21) and the LRFD one
described earlier (3.12) are clear when compared together:
| | | |
m Q m R
Q V R V o| o| exp exp > (2-22)
Structural analysis (e.g. Incremental Dynamic Analysis) is needed to determine the
median drift demand for a given spectral acceleration on a given frame. By following the
procedure, a frame is subjected to a ground motion with spectral acceleration, Sa, with
annual probability P
0
of being exceeded. The outcome is the median drift demand
0
P
m
D
under that ground motion intensity. Using the median drift demand calculated, the
capacity required to ensure probability as low as P
0
of the demand exceeding the capacity
can be estimated as well. Figure 2-4 illustrates the main points of the SAC-DCFD
reliability framework. On the right hand side, the spectral acceleration is plotted versus
the maximum inter-story drift ratio and on the left hand side versus it is plotted versus the
spectral acceleration hazard. Dynamic analyses are performed at different spectral
accelerations for several ground motion records. The maximum inter-story drift ratio is
recorded for each dynamic analysis, resulting in a distribution, with a median,
0
P
m
D , and
dispersion,
Sa D
V , of the maximum inter-story drift demand at specified ground motion
intensity. Equation 2-21 can then be used to either calculate the required capacity such
that the probability of demand exceeding capacity is tolerable or given capacity, to
calculate if the probability is tolerable. Although the demand
0
P
m
D is established by
using records of just one Sa level, the whole range of Sa levels was considered in the
development of SAC-DCFD framework, where Sa levels were weighted by their relative
likelihood of being felt at the site.
2.3.3 FEMA P695 Quantification of Building Seismic Performance Factors
The FEMA P695 Quantification of Building Seismic Performance Factors (FEMA,
2009) was developed to evaluate seismic design provisions through inelastic static and
dynamic analyses of structural systems under earthquake ground motions. The intention
25
of the FEMA P695 reliability methodology is to provide a rational basis for establishing
global seismic performance factors (SPFs), including the response modification
coefficient (R factor), the system overstrength factor (O
0
) and the deflection
amplification factor (C
d
), of seismic force resisting systems proposed for inclusion in
modern building codes. The underlying basis of the method is to assess structural
collapse performance by nonlinear analysis of so-called archetype designs and to
ensure a consistent risk across all structural system types and materials. The details of the
methodology can be found in FEMA P695 Quantification of Building Seismic Response
Parameters (FEMA, 2009).
The FEMA P695 framework assesses the reliability of structural systems by
subjecting median models of structural archetypes, which are considered to closely
represent the system considered, to multiple ground motion records of increasing
intensity, i.e. the Incremental Dynamic Analysis (IDA), until structural collapse is
detected. (Vamvatsikos and Cornell, 2002). Structural collapse occurs when IDA
response plot reaches a plateau. Figure 2-5 illustrates the IDA process for a single ground
motion. The circles on the figure are the maximum story drift ratios (DR) recorded during
dynamic analyses plotted versus the ground motion spectral acceleration. Generally, the
DR increases as the spectral acceleration increases until the frame collapses. The DR and
spectral acceleration at which collapse is imminent in the simulation are referred to as
DR
SC
and S
T(SC)
where the subscript SC stands for a simulated collapse mode. Due to
both modeling limitations and practicality, some deterioration mechanisms leading to
collapse may not be included in the structural models. However, provisions are included
in the FEMA P695 methodology to assess the effects of deterioration mechanisms that
are not explicitly simulated but could cause collapse. These collapse modes are referred
to as non-simulated collapse modes. Examples include fracture in connections or hinge
regions of steel moment frame components, axial failure of steel columns or shear and
axial failure in reinforced concrete columns. The non-simulated collapse modes are
estimated after the incremental dynamic analysis has been performed and for each time
history analysis, if the capacity of the non-simulated collapse mode is exceeded, prior to
the collapse of the structure according to the time history analysis, the collapse point is
26
adjusted. Figure 2-5 shows the DR and spectral acceleration of a non-simulated collapse
mode, DR
NSC
and S
T(NSC)
.
Given the inherent variability in earthquake ground motions and possible variability
in the nonlinear response of a structural system to different ground motions (see Figure 2-
6), the results from the IDA are formulated in a probabilistic framework to evaluate
whether the collapse safety is tolerable. Based on the maximum considered earthquake
(MCE) ground motions, the so-called collapse margin ratio, CMR, is calculated. The
CMR is the median collapse level spectral acceleration, S
CT
, at the period of the system,
T, divided by the MCE spectral acceleration, S
MT
, at the period of the system, T. To
account for the unique spectral shape of extreme ground motions and their effect on
behavior, the CMR is then adjusted (based on the seismic hazard intensity and the
nonlinear deformation characteristics of the building) to calculate an adjusted CMR, or
ACMR. Based on the ACMR, the system is either deemed acceptable or unacceptable.
A second evaluation on whether the collapse safety is tolerable is to calculate the
probability of collapse under the MCE ground motions. The IDA response histories are
used to create a collapse fragility curve through a cumulative distribution function, CDF
(see Figure 2-7). The collapse fragility curve relates the ground motion intensity to the
probability of collapse. The collapse fragility curve is assumed to be lognormal with a
median equal to the ACMR and a specified dispersion. The specified dispersion, whose
value depends on how well the dynamic analyses are believed to capture the major
sources of uncertainty that contribute to variability in collapse capacity, varies from 0.275
to 0.950. From the collapse fragility curve, the probability of collapse under the MCE
ground motions is calculated, where the tolerable value is 10%.
The uncertainties considered in the FEMA P695 methodology are the following: a)
Record-to-Record Uncertainty, due variability in frame response to different ground
motion records b) Modeling Uncertainty, which includes uncertainties involved in the
idealization of the structural behavior and the characterization of the modeling
parameters for analysis c) Design Requirements-Related Uncertainty, which is based on
the completeness and appropriateness of the design requirements and the extent which the
27
archetype systems represents the actual structural design of the systems d) Test Data-
Related Uncertainty, which is based on the robustness and comprehensiveness of the test
data used to support the development of nonlinear archetype analysis models. Some of
these uncertainties are modeled explicitly and some are incorporated through judgment
with the use of the specified collapse fragility dispersion It should be pointed out that
failure of capacity-designed components is not explicitly accounted for in the FEMA
P695 dynamic analysis procedure and collapse fragility development. However, it could
be argued that the fraction of the collapse fragility curve dispersion due to Modeling
Uncertainty and Design Requirements-Related Uncertainty account for the risk of failure
of capacity-designed components.
In a related effort, as part of Project 07, the MCE seismic design maps have been
revised to provide more consistent collapse risk safety throughout various regions of the
United States. (Luco et al., 2007) The basis of this effort is to achieve a uniform target
collapse risk of 1% in 50 years and is therefore a move from the previous uniform-hazard
ground motions to uniform-risk ground motions. Recognizing that collapse capacity of
structures is uncertain, the use of uniform-hazard ground motions for design results in
structures with non-uniform collapse probability due to site-to-site variability in the shape
of ground motion hazard curves. To calculate collapse probabilities, lognormal collapse
fragility curves with dispersion of 0.8 (partially based on FEMA P695 analyses results
and recommendations) are integrated with site ground motion hazard curves. With a fixed
dispersion, only a single value of the lognormal collapse capacity curve is required to
fully describe it, and based on FEMA P695 recommendations of 10% probability of
collapse at the MCE intensity, that is the value chosen. The MCE intensity is then
increased or decreased for a given site, depending on if the calculated collapse probability
exceeds the 1% target or not. These combined efforts are significant since, for the first
time, they codify a target collapse safety risk that can provide the basis for establishing
seismic design guidelines. The calculated collapse probability depends on the specified
dispersion of the collapse fragility curve. Whether the dispersion of 0.8, and therefore the
target 1% collapse probability, is assumed to include the possibility of failure of capacity-
designed components is unclear.
28
2.3.4 Component vs. System Reliability
The target reliability of a component should be related to the impact of its failure on
the systems performance, as the ultimate goal is to limit the annual probability of system
collapse to a tolerable level. However, achieving this ultimate goal is far from an easy
task considering all the variables and uncertainties involved in the process. Calibrating
the probability of failure of each component within a system, such that the annual
probability of system collapse is tolerable, can therefore prove to be infeasible. One is
thus often forced to focus on component structural reliability with the assumption that by
satisfying reliability requirements at the component level, the system reliability is
tolerable. However, with the component failure consequences unknown, it can be
challenging to decide upon the required component reliability.
Figure 2-8 is a schematic diagram of the full scope of a reliability analysis of a
seismic resistant system. The figure is separated into 6 different levels based on the
details and complexities of the reliability analysis, ranging from a component to a full
structure reliability analysis. At the top of the figure is the actual structure of interest. The
end goal is to ensure its reliability, i.e. to limit its annual probability of collapse to a
tolerable level. However, it is essentially impossible to analyze the reliability of the
actual structure so the best alternative is to create a detailed analytical model of the
complete idealized structure, a model that is believed to capture the structures behavior
accurately (Level 5). This includes modeling the whole seismic force-resistant system
(SFR system) and the lateral resistance of the gravity system, including all possible
failure modes, to capture the true capacity of the structure and the benefits of its
redundancy. Further, due to variability in components strengths and ductilities, including
the variability of all random variables (modeling uncertainty) is necessary when
calculating the structures reliability. Similarly, the correlation between the random
variables needs to be included as it counteracts the positive benefits of redundancy.
Currently, seismic design codes do not take advantage of the lateral resistance of the
gravity system and in return, no special detailing of the gravity system is required with
regards to seismic forces. Therefore, taking advantage of the possible reduction in
collapse probability due to the inclusion of the gravity system in the analysis might not be
29
warranted. The Level 4 analysis in Figure 2-8 is in all respects the same as the Level 5
analysis except the lateral resistance of the gravity system is neglected but redundancy
within the SFR system is included. Level 3 further simplifies the analysis by only
modeling a part of the SFR system, illustrated in Figure 2-8 by modeling only one of the
two bays comprising the SFR system. The Level 3 analysis is therefore one where the
benefits of redundancy are limited.
The reliability analyses described in Level 3, 4 and 5 are very detailed, and time and
computer intensive, largely due to the direct inclusion of modeling uncertainty. Including
modeling uncertainty requires multiple realizations of each structure with different
random variable input values. Currently, this renders them fairly impractical, especially if
the goal is to analyze the reliability of a whole suite of structures subjected to different
ground motion records at multiple intensities. To go around that obstacle, all the
aforementioned analyses levels can be performed with median models to capture the
median response of the structure. That causes Level 3 and Level 4 to merge into
effectively the same analysis, as redundancy and correlation within the SSF system is
now neglected, and both become similar to Level 5 analysis, except there the gravity
system is included. This is essentially the approach that FEMA P695 takes to assess the
collapse risk of seismic resistant systems. The analytical models are median models
where expected values of all model parameters are used. To account for the fact that
modeling uncertainty is not directly included in the analysis, the dispersion of the
resulting collapse fragility curve is specified directly, rather than being based only
analysis results, and its value depends largely on the ground motion variability and the
overall confidence in the analytical model.
Level 2 is a simplified SFR system where only a system subassembly is modeled. By
modeling only a system subassembly, the number of random variables is significantly
reduced. Reducing the number of random variables allows for possibility of performing a
full uncertainty analysis, which can assist in evaluating whether some of the assumptions
made in the previous levels are justified, e.g. the specified dispersion in collapse fragility.
Level 1 is the simplest but most commonly used reliability analysis. It is a component
30
reliability analysis with no system behavior considerations. Level 1 analysis is the basis
of the LRFD component reliability methodology.
The reliability methodologies discussed here, LRFD and FEMA P695, each address
the objectives they set out to meet but provide limited information on the relationship
between component and system reliability. LRFD is a component limit state methodology
where objective is to limit the probability of exceeding a specific limit state. It does not
explicitly account for system reliability considerations except that it requires lower
probability (higher ) of exceeding connection limit states than member yielding limit
states based on the believe that exceeding connection limit states has more severe
consequences on the system reliability than exceeding member yielding limit states.
(Galambos, 1990). The higher reliability index for connections reflects a long tradition of
designing connections stronger than the members connected by them and in the
development of LRFD the target probability of connection failure was set arbitrarily at
2.5 orders of magnitude lower than member failure, which is represented in the
recommended reliability indices of 4.5 and 3.0 for connections and members,
respectively. (Ravindra and Galambos 1978, Ellingwood et al. 1980) Acknowledging that
the LRFD limit state design does not explicitly account for system reliability, Galambos
(1990) suggested a further distinction between different limit states based on the
presumed consequences of exceeding them on the system reliability by introducing a new
resistance/capacity factor, a system factor (|
s
) that could be multiplied with the already
established |-factors. The suggested |
s
factors ranged from 0.7 if exceeding the limit
state results in complete damage to 1.1 if exceeding if it results in only slight damage.
On the other hand, the FEMA P695 methodology evaluates the reliability of systems
and it generally lacks the resolution to assess the impact of the reliability of individual
components on the system reliability. Redundancy within the structural system and
uncertainty around components strength, stiffness and ductility is not included in the
methodology. Also, the non-simulated collapses modes, which are based on exceeding
deterministic values in the analysis (drift, rotations etc.), generally assume system
collapse if those values are exceeded. Capacity-designed components are assumed not to
fail in the methodology.
31
The SAC-DCFD methodology can be applied either at the component level or at the
system level, and can therefore be applied at any of the six levels listed in Figure 2-8. It
was developed specifically to check component or system limit states in seismic design
with recognition of the dynamic and nonlinear displacement basis of the problem. It
results in an annual probability of the chosen limit state being exceeded while only
checking it at one ground motion intensity. At the system level, similar to the FEMA
P695 methodology, it lacks the resolution to assess the impact of the reliability of
individual components on the system reliability.
Still, there are ways to assess the relationship between component reliability and
system reliability that can at least assist with understanding the impact of certain design
criteria. Sensitivity analysis can provide very insightful information on the response of
structures when model input parameters are varied. Often the approach is taken to fix all
variables at median values and then vary variables individually above and below the
median. Although a deterministic analysis, it provides information on which random
variables have the most impact on the structural response and has been used in many
studies, e.g. in Porter et al (2002) to examine which sources of uncertainty most strongly
affect the repair cost of a high-rise non-ductile reinforced-concrete moment-frame
building in a future earthquake. For many capacity-designed components, e.g. brace
connections, sensitivity analysis can provide limited information if not properly
performed. Increasing connection strength is likely not to change the structural response
and unless the connection strength is decreased enough, such that it fractures in the
structural analysis, decreasing the strength might not change the structural response either.
If the distributions of all, or at least most major, variables and the correlations
between them are known, realizations of each modeling random variable can be
generated using Monte-Carlo simulation methods (Melchers, 1999; Rubinstein, 1981)
This process is then repeated multiple times and for each set of simulated random
variables, the system reliability is assessed through methods such as incremental dynamic
analysis. The end result is a probability of collapse (or any other performance measure
being analyzed) distribution associated with the distribution of the input random variables.
Vamvatsikos and Fragiadakis (2009) and Dolsek (2009) both used Monte Carlo methods
32
combined with an efficient Latin Hypercube sampling method, a variance reduction
technique, to propagate model parameter uncertainty to the actual system seismic
performance on a nine-story steel moment resisting frame and a four-story reinforced
concrete frame, respectively. Although straightforward to implement and perhaps the
most comprehensive method to assess system reliability, Monte Carlo methods are
computationally intensive as the number of required simulations to achieve a stable
solution can be very large, especially when assessing small probabilities of failure, and
the time to complete each analysis is non-negligible. In addition, as all random variables
are lumped together, the Monte Carlo simulation methods give limited insight on the
influence of individual random variables, rather the impact of the uncertainties of random
variables as a whole on the system behavior. To assess the influence of individual
variables, the above procedure needs to be performed again with different distributions
for the input random variables, i.e. sensitivity analysis, which could be taken to represent
a change in the design provisions, thus further increasing analysis the time.
Response surface (Melchers, 1999; Pinto et al., 2004) is another method that can be
employed to assess system reliability with modeling uncertainties included. It can be
thought of as a hybrid technique between sensitivity analyses and full Monte Carlo
simulations, or any other technique that requires a closed form limit state function. By
systematically varying the input parameters such that most of the sample space is covered
and then assessing the system response of interest for each case, a limit state function or a
response surface is created by fitting a functional form between the input parameters and
the system response. With the relationship between the input parameters and the system
response known, a full Monte Carlo simulation can be performed using that estimated
functional form with minimal computational effort. Liel (2008) used Monte Carlo
simulations coupled with a response surface method to evaluate the collapse uncertainty
of reinforced concrete buildings, showing that modeling uncertainties will in most cases
both increase the dispersion of the system response as well as decrease the predicted
median collapse point compared to median model results. Generally, the response surface
method has worked well in structural reliability assessments, but its accuracy depends
largely on the characteristics of the limit state function, with highly nonlinear functions
providing less reliable results. (Melchers, 1999; Rajashekhar and Ellingwood, 1993)
33
2.4 Summary
Capacity design provisions are included in most modern seismic design codes. The
goal of employing capacity design principles is to ensure ductile response and energy
dissipation capacity in seismic resisting systems. The method to achieve this goal is to
confine significant inelastic deformations to selected structural components while others
are designed with sufficient strength to remain essentially elastic. While the concept is
clear, its implementation is more challenging. In an attempt to achieve the goals of
capacity design, US structural design codes adjust the design demand and the design
capacity by applying various factors to the relevant limit state equations. ASCE-7-10
includes an overstrength factor, the
0
-factor, which is meant capture the expected
capacity of deformation-controlled components and therefore the increased loads on
force-controlled members above the seismic design loads. The AISC Seismic Provisions
have made progress in addressing some of the issues of capacity-based design by
introducing the R
y
-factor to provide a consistent basis to evaluate expected yield
strength of deformation-controlled components. Other factors, such as factors to account
for strain hardening, the C
pr
-factor in the AISC Prequalified Connection Requirements
and the -factor used to adjust for compression strengths of BRBs have also been
introduced to assist in confining significant inelastic deformations to selected components.
Similarly, to assist in achieving the goals of capacity design ACI 318-08 includes a factor
of 1.25 to account for strain hardening in reinforcing steel in shear strength design.
However, these factors, which are intended to create a margin between demand and
capacity to ensure the desired behavior, have largely been established in an ad-hoc
manner, resulting in inconsistencies in the assumed deformation demands of components
and therefore inconsistencies in the reliability of capacity-designed components.
Many methodologies have been established to assess structural reliability. Here,
three of them were presented. The Load and Resistance Factor Design (LRFD)
methodology is a component methodology that was developed in the 1970s for static,
strength-based problems under various load types. It ensures a tolerable component
reliability index for a specified set of load and resistance factors. The SAC Demand and
Capacity Factor Design (SAC-DCFD) methodology is a reliability methodology that was
34
developed as a part of the SAC Federal Emergency Management Agency (SAC/FEMA)
program (Cornell et al 2002, Yun et al, 2002). SAC-DCFD was developed specifically
for seismic design and it ensures a tolerable annual probability of demand exceeding
capacity. Demand and capacity can be defined at the component level or at the system
level. The FEMA P695 Quantification of Building Seismic Performance Factors is a
system reliability methodology, developed to evaluate seismic design provisions through
inelastic static and dynamic analyses of structural systems under earthquake ground
motions with the objective of ensuring consistent risk across all structural system types
and materials. In the FEMA P695 methodology, the tolerable probability of collapse for a
structural system is 10% at the MCE ground motion intensity. In the development of new
MCE seismic design maps as part of Project 07, the collapse fragility curves developed
by FEMA P695 were integrated site seismic curves and the collapse fragility curves
adjusted until a target collapse probability of 1% in 50 years was attained. These two
efforts, FEMA P695 and Project 07, are the first to explicitly state target system
reliability for seismic design.
To relate the reliability of a component to the system reliability, the consequences of
component failures need to be known, a task far from easy considering all the variables
and uncertainties involved in the process. Many methods are available to assess the
relationship between component and structural reliability, each with its pros and cons.
Sensitivity analysis is simple and a straightforward to implement and provides important
information on which variables have the largest impact on the structural response.
However, it is both deterministic in nature and may give limited insight into the response
sensitivity of capacity-designed components whose values may not change the response
unless a certain threshold is exceeded. Monte Carlo simulation method is easy to
implement and perhaps the most comprehensive method to assess structural response.
Still, for most structural reliability assessments where nonlinear analysis is required, it
may prove to be too computationally intensive and provide limited insight into the impact
of individual variables. The response surface method combines the two methods
mentioned above, and has been shown to work well in structural reliability. Its accuracy
however, is very dependent on the shape of the limit state function, with highly nonlinear
functions providing less accurate results.
35
Table 2-1: Summary of Capacity Design Requirements in the AISC (2010) Seismic Provisions
Component
or System
AISC
Section
Description of Capacity Demand Requirements
*
Material
Properties
A3
R
y
and R
t
factors for determining expected yield and ultimate strengths of steel
materials (Table A3.1)
Columns D1.4
Required axial strength provisions, based on (a) amplified seismic load, (b) load
from yielding members including the effects of material overstrength and strain
hardening, (c) limiting resistance of foundation
Column
Splices
D2.5
Required strength of welded splices in seismic resisting system based on (a) 200%
of calculated loading effect, or (b) at least 50% of the strength of the expected
yield strength of column flanges (R
y
F
y
b
f
t
f
).
Column Bases D2.6
Required strength of bases shall be lesser of a) M
pc
/H (shear) and 1.1R
y
F
y
Z
x
(moment), or (b) the shear/moment calculated using the amplified seismic load.
SMF
Connections
E3.6
Required shear strength of beam column connections shall be based on load
combinations with amplified seismic load, E = 2[1.1R
y
M
p
/L
h
]. Other design
aspects covered by pre-qualified connection provisions or demonstrated by tests.
SMF Columns E3.4
Minimum required column flexural strengths determined without | factors and
compare with beam flexural strengths calculated based on 1.1R
y
F
y
Z
x
SMF Column
Splices
E3.6 Minimum required flexural strength equal to R
y
F
y
Z
x
and shear equal to EM
pc
/H.
SMF
Prequalified
Connections
AISC
(2010b)
Connection welds, bolts and other components generally designed to resist
maximum probable moment, C
pr
R
y
F
y
Z, where C
pr
is a factor for strain hardening
(determined as average of F
y
and F
u
and limited to 1.2). Resistance factors used in
conjunction with the maximum probable moment are |
d
= 1.0 for ductile limit
states and |
n
=0.9 for non-ductile limit states.
SCBF Brace F2.5
For braces with net section issues, the minimum required tensile strength at the net
section is R
y
F
y
A
g
or the maximum force that can be delivered by the system.
SCBF Brace
Connection
F2.6
Minimum required (a) axial strength of R
y
F
y
A
g
, less than amplified load effect,
(b) flexural strength 1.1R
y
M
p
, (c) compressive strength 1.14F
cre
A
g
SCBF Chevron
Beam
F2.4
Min. required flexural strength based on unbalanced vertical force from R
y
F
y
A
g
in
tension brace and 0.3 times 1.14F
cre
A
g
in compression brace, plus gravity loads.
SCBF
Columns
F2.3
Required strength based on larger of a) load from all braces at their expected
strength b) tension braces at their expected strength and compression braces at
their expected post-buckling strength but never to exceed load from amplified
seismic load in which all compression braces have been removed.
EBF Link-
Column
Connection
F3.6
Required strength of link-to-column connection is the nominal shear strength of
the link V
n
at the maximum rotation angle. Testing is required to demonstrate
deformation capacity of connection for some cases.
EBF Braces F3.4
Minimum required strength equal to axial force associated with 1.25 or 1.4 (I-
shaped/box shaped) times expected shear strength of link, R
y
V
n
.
EBF Beams F3.4
Minimum required strength equal to moment and axial force associated with 1.25
or 1.4 (I-shaped/box shaped) times expected shear strength of link, R
y
V
n
. The
available beam strength may also be increased by R
y.
EBF Beam-
Column
Connections
F3.6
For EBFs with moment connections, the required strength of beam-column
connections is the same as for OMF, 1.1R
y
M
p
.
BRBF Beams
and Adjoining
Members.
F4.2
Required strength of adjacent members (e.g., Chevron beams) and their
connections shall be based on adjusted brace strengths, |eR
y
P
ysc
in compression
and eR
y
P
ysc
in tension.
BRBF Brace
Connections
F4.6
Required strength of brace connections shall be 1.1 times the adjusted brace
strengths in tension and compression
*For explanation of symbols, see Notations section
36
Table 2-2: Summary of Capacity Design Requirements in the ACI 318-08 (2008)
Component
or System
ACI
Section
Description of Capacity Demand Requirements
*
Material
Properties
21
A factor of 1.25 is applied to reinforcing steel stress, f
y
, for shear strength design
forces to account for strain hardening.
Reinforcement 21.1.5
a) The actual yield strength based on mill tests does not exceed f
y
by more than
18,000 psi; and b)The ratio of the actual yield strength is not less than 1.25
IMF Shear
Strength
Requirements
21.3.3
Minimum required shear strength of beams and columns determined with | factors
shall exceed: a) The shear due to nominal moment strength developing at each
restrained end b) The maximum shear from design where the earthquake demand is
twice that prescribed in building codes
SMF Columns 21.6.2
Minimum required nominal column flexural strengths determined without | factors
and compare with factored nominal beam flexural strengths: M
nc
(6/5)M
nb
SMF Shear
Strength
Requirements
21.6.5
The design shear force, V
e
, shall be determined from consideration of the maximum
probable moment strengths, M
pr
, at each and of the member. M
pr
is determined using
| = 1.0 and a reinforcing steel stress equal to at least 1.25f
y
SMF Joints 21.7.2
Joint shear force generated by the flexural reinforcement is calculated for a stress of
1.25f
y
in the reinforcement.
Precast SMF
Ductile
Connections
21.8.2
Minimum required nominal shear strength, V
n
, without | factors to exceed 2V
e
of
flexural members where V
e
is calculated in 21.6.5 using 1.25f
y
for reinforcement
stress.
Precast SMF
Strong
Connections
21.8.3
Minimum required design strength of connections, S
n
, with | factors to exceed the
probable strength at intended yield locations, S
e
, calculated using 1.25f
y
for
reinforcement stress and | = 1.0. For column-to-column connections: |S
n
1.4S
e
,
|M
n
0.4M
pr
and |V
n
V
e
*For explanation of symbols, see Notations section
Table 2-3: Relationship between probability
of failure and the reliability index,
Probability of
Failure
Reliability Index,
0.50 0.00
0.25 0.67
0.15 .99
0.10 1.28
0.0 1.64
0 01 2.33
10
-3
3.09
10
-4
3.72
10
-5
4.26
10
-6
4.75
37
`
V
V
DBE
V
V
u,exp
(C
d
-1)
V
(
R
V( -1
In
= C
d
= V
u,exp
/V
V = V
DBE
/R
in
Figure 2-1: Idealized inelastic base shear versus story drift curve
Figure 2-2: Illustration of the reliability index, |, and its relationship with the failure
criterionrepresented by z<0
38
0 1 2 3 4 5
P
r
o
b
a
b
i
l
i
t
y
o
f
F
a
i
l
u
r
e
|
Figure 2-3: Relationship between probability of failure and the reliability index,
Figure 2-4: Illustration of the SAC-DCFD reliability framework (Image from Cornell et al., 2002)
39
Figure 2-5: Collapse of a system with simulated and non-simulated collapse modes using IDA (Image
from FEMA, 2009).
Figure 2-6: IDA response plot and collapse margin ratio (Image from FEMA, 2009).
40
Figure 2-7: Collapse fragility curves (Image from FEMA, 2009).
41
Figure 2-8: Structural Reliability: A schematic diagram of different levels of complexity in structural
reliability models. While the goal is to evaluate the actual structures reliability, most reliability
analyses are performed at the component level, a system subassembly level or at best with simplified
models of the actual structure where multiple variables and uncertainties are excluded from the
analysis. SFR System = Seismic Force Resisting System
42
43
Chapter 3
3 A Reliability-Based Methodology for
Establishing the Required Design Strength of
Capacity-Designed Components
3.1 Introduction
In this chapter a reliability-based methodology is presented for establishing the
required design strength of capacity-designed components. The goal is to provide a
framework that developers of design standards can use when establishing the required
design strengths of capacity-designed components that result in consistent reliability
between different systems. The chapter begins with a detailed description of the
development of the methodology, from the examination of deformation-controlled
components behavior to the determination of how capacity-designed components
failure affects the system reliability. Special attention is given to the impact of the system
response modification factor, i.e. the R-factor, and system overstrength on the reliability
of capacity-designed components.
Although a component reliability methodology, the capacity design reliability
methodology considers both system design parameters that affect component reliability
as well as the relationship between component and system reliability. The methodology
incorporates relevant aspects of the LRFD component reliability methodology, the
FEMA P695 system reliability methodology and risk consistent MCE maps of Project 07.
Guidelines for implementation of the methodology are provided.
44
3.2 Development of the Methodology
The reliability-based methodology for establishing the required design strength of
capacity-designed components utilizes the well-established LRFD component reliability
methodology with appropriate adjustments to address issues specific to capacity-based
design. The design strength of capacity-designed components is set by adjusting the load
and resistance factors, and |, until desired reliability is achieved. To be consistent with
capacity-based design terminology, load and resistance is referred to as demand and
capacity, and similarly, the load and resistance factors as demand and capacity factors
and collectively as capacity design factors. Demand and capacity factors are then applied
to nominal values of demand and capacity to ensure the failure probability is less than
that specified as tolerable. When demand and capacity probability distributions are
lognormally distributed, the basic equation to calculate the ratio between capacity design
factors is:
( )
D C D C Ha R
m
n
n
m
V V V V
C
C
D
D
|
|
2 exp
2 2
,
+ = (3-1)
where
m
D
and
m
C
, and
n
D and
n
C are the median and nominal values of the demand
and capacity probability distributions, respectively,
D
V and
C
V their lognormal standard
deviations, and the correlation between demand and capacity. The capacity design
reliability index,
R,Ha
, provides a measure of probability of demand exceeding capacity
of capacity-designed components. Equation 3-1 is a re-arrangement of the LRFD
Equations 2-11 and 2-12 from before, except that the definition and meaning of certain
terms are different.
The key differences between the calculation of capacity design factors as compared
to the conventional LRFD formulation is in the way the capacity design reliability index,
R,Ha
, and the demand parameters are selected. The demand in capacity-based design is
based on the strength capacity of the deformation-controlled components, as opposed to
more conventional loading effects due to dead, live and wind loads. The induced forces
from the deformation-controlled components can also vary depending on the deformation
45
demands in the structure. The reliability index in the LRFD methodology was originally
calibrated to pre-LRFD design equations and then used as a comparative value for
different failure modes. The reliability index,
R,Ha
, in Equation 3-1 serves a similar
purpose but takes the system effects (i.e. the systems response modification coefficient,
or R-factor and member overstrength) and the systems site ground motion hazard curve
into consideration. The subscript of the reliability index refers to the influence of the R-
factor and the site ground motion hazard curve (Ha) on the required index. The goal is to
attain a consistent level of risk due to failure of capacity-designed components in various
structural systems. Equation 3-1 assumes lognormal probability distributions. If the
probability distributions are not lognormally distributed, Equation 3-1 would need to be
reformulated.
The demand in capacity-based design is unique relative to other design concepts in
the sense that the demand on force-controlled components originates mainly from other
components within the system, i.e. from the deformation-controlled components as they
undergo inelastic deformation during seismic events. The design of the deformation-
controlled components therefore needs to be included in the methodology. Figure 3-1
shows brace responses when braces are subjected to different cyclic loading protocols: a
far-field and a near-field tension protocol. A common characteristic of both tests is that
the braces yield in tension at relatively low deformations. This results in brace
connections experiencing demands close to their maximum tensile demands at relatively
low deformations: deformations, which are likely to occur under low to moderate
earthquake intensities. Beyond this, one can see that the loading protocol can affect the
maximum tension forces reached in the braces. Figure 3-2 demonstrates how brace forces
develop under random earthquake loading. It shows incremental dynamic analysis results
for a single story SCBF where both the maximum drift (Figure 3-2c) and the maximum
brace force (Figure 3-2b) are recorded. For a given probability distribution of the brace
connection capacity and using the brace demand distribution from the analyses, the
probability of connection failure can be calculated, e.g. using Equation 2-2 from Chapter
2. It is evident that the probability of connection failure is negligible before brace
yielding and then saturates as the braces reach their maximum demands (Figure 3-2d).
The probability of connection failure can be controlled through the /|-ratio, which
46
defines the connection capacity relative to the demand. Referring to Fig. 3-2d, the
spectral acceleration when yielding initiates is the only other parameter needed to
describe the relationship between the probability of connection failure to spectral
acceleration demand. This behavior is likely to apply equally to other deformation-
controlled components. Predicting the ground motion intensity causing initiation of
deformation-controlled member yielding, Sa
y,exp
, is therefore pivotal in the development
of the methodology. A more detailed discussion of the single story SCBF incremental
dynamic analysis results can be found in Chapter 4.
When designing deformation-controlled components, the required design strength is
based on the seismic design forces of the overall frame. Following the ASCE 7 design
standard, the seismic design forces are based on the spectral acceleration for the
maximum considered earthquake (MCE) at the fundamental period of the structure, Sa
T1
.
This spectral intensity is then (a) multiplied by two-thirds to get the design basis
earthquake (DBE) intensity and (b) divided by the frames response modification
coefficient, or R-factor, to get the design spectral acceleration, Sa
design
. Thus, the seismic
design forces are associated with earthquake intensities significantly lower than the
maximum considered earthquake. Still, due to the use of capacity (resistance) factors (|),
member over-design and the use of nominal material values and nominal strength
equations, yielding is not expected to initiate at Sa
design
, but rather at some spectral
acceleration larger than Sa
design
. However, unlike Sa
design
, Sa
y,exp
is not unique for the
whole structure as it is based on components overall over-design, which varies from one
component to the next and between stories. The component with the lowest Sa
y,exp
is
therefore most likely to yield first, i.e. it is the weakest link, and could therefore possibly
shield other components from yielding. In addition, when the demand on capacity-
designed components originates from multiple deformation-controlled components,
Sa
y,exp
is not as clearly defined since overstrength of the different deformation-controlled
components varies.
To illustrate these points, an example of a 3-story SCBF is used. Figure 3-3 shows an
elevation view of the frame. The braces are the deformation-controlled components,
while the brace connections, beams, columns and foundation are capacity-designed
47
components. The demand on a given brace connection on the 2
nd
story originates from
the brace it is connecting to the frame. If the tension and compression capacity of the
braces in that story are identical, Sa
y,exp
for the demand on that connection can be
calculated based on the brace overdesign multiplied by Sa
design
. The fact that braces in
concentrically braced frames work in pairs and have different tension and compression
capacities, complicates the calculations, but this is discussed later. The same calculations
can be done for the 1
st
and 3
rd
story, and unless optimally designed, the calculated Sa
y,exp
for those will differ from each other and from the 2
nd
story. Similarly, Sa
y,exp
for the
demand on a beam in a given story is the same as it is for the brace connections in that
story. The difference is that the demand on a beam originates from both braces in a given
story. As this example demonstrates, Sa
y,exp
can vary from story to story. Therefore, for
the demand on the 1
st
story columns, which is based on the braces above yielding and
buckling, there is no clear Sa
y,exp
. Sa
y,exp
becomes more indistinct as the number stories
and braces above increases since the likelihood of simultaneous yielding or buckling of
the braces decreases as the number of them increases. As will be demonstrated with
columns in multi-story braced frames in Chapter 5, capacity design principles do often
not explain the demand on columns well since the full capacity is rarely reached.
Figure 3-4 shows an idealized static nonlinear pushover curve of a story in a 1-bay
braced frame, either a single story frame as in Figure 3-2 or a story within a multi-story
frame. Of interest is to estimate the story shear, which can be related back to spectral
acceleration that causes yielding of the tensile brace. The design story or base shear is V.
Due to over-design for any number of reasons, the factored nominal story or base shear
strength, |V
n
, exceeds V. Excluding capacity factors, (| ) and using expected material
values and strength equations, as opposed to nominal ones, the expected story or base
shear yield strength, V
y,exp
, will exceed |V
n
and due to strain hardening, the expected
story or base shear ultimate strength, V
u,exp
exceeds V
y,exp
. In Figure 3-4, V
y,exp
corresponds to the second kink in the idealized static nonlinear pushover curve,
corresponding to the point where the tension brace yields after the compression brace has
buckled. The compression strength of a brace is always less than its tension strength.
Therefore buckling occurs prior to tensile yielding for a pair of two opposing braces. The
48
ratio between V
y,exp
and V is referred to as the story or base shear yield over-strength,
y,exp
, and the ratio between the elastic design demand, V
DBE
, and V
y,exp
as R
. (See
Equation 3-2 and 3-3) R
(3-3)
R
Sa
Sa
MCE
y
3
2
exp ,
= (3-4)
Knowing Sa
y,exp
, the site ground motion hazard curve can be used to predict the
probability that the force-controlled components will experience large forces and cause
the deformation-controlled components to yield. Figure 3-5 shows this schematically
where on the right hand side there are idealized static nonlinear response curves versus
story drift for two systems with different Sa
y,exp
and on the left hand side are
representative seismic hazard curves for the Eastern and Western US. Based on Sa
y,exp
of
the two systems, the seismic hazard curves can be used to calculate the probability that
Sa
y,exp
is exceeded, i.e. that deformation-controlled components yield. Figure 3-5 shows
that the system with higher Sa
y,exp
has a lower probability of experiencing forces large
enough to yield its deformation-controlled components compared to the system with
lower Sa
y,exp
. Table 3-1 has calculated values for the probability in 50 years that the
spectral acceleration of the design basis earthquake, Sa
DBE
, divided by R
(i.e. Sa
y,exp
) is
exceeded for a San Francisco site (Lat = 38.0, Long = -121.7) and a New Madrid site (Lat
49
= 35.2, Long = -89.9) when T
1
= 0.2s. The seismic design values (Sa
T1
) and the
probability of exceedance are from the 2008 USGS hazard maps for each site. (Peterson
et al., 2008)
Table 3-1 shows that for a frame with R
= 2 (Frame 2) in San
Francisco, that probability is down to 28.7%. Because of the low-R-factor, Frame 2 has
only 1/3 the probability of the force-controlled member yielding compared to Frame 1.
This suggests that the connection capacity factors could be less stringent for Frame 2
compared to Frame 1. These results are reasonable when thinking about the impact the R-
factor has on frame behavior. Frames with large R-factors depend more on inelastic force
redistribution and inelastic behavior then frames with smaller R-factors during seismic
events and therefore the reliability of the force-controlled components becomes more
critical.
Table 3-1 also demonstrates the difference between two different sites. A frame with
R
= = > = >
i
x all
i i
x Sa MAF x Sa C D P C D MAF ) ( * ) ( ) ( (3-5)
where MAF(Sa=x
i
) is the mean annual frequency of observing a spectral acceleration in
some narrow range around x
i
, defined by the following equation (which approximates the
derivative of the site ground motion hazard curve at a given ground motion intensity
times the narrow range around x
i
, i.e. dx).
( )
( ) ( ) ( )
2
1 1 +
> >
= =
i i
i
x Sa MAF x Sa MAF
x Sa MAF (3-6)
In continuous integral form, Equation 3-5 is:
}
= > = > ) ( ) ( ) ( x dMAF x Sa C D P C D MAF (3-7)
where the notation |dMAF(x)| means the absolute value of the derivative of the sites
ground motion hazard curve times dx.
The summation in Equation 3-5 can be avoided by simplifying the probability of
demand exceeding capacity function to a step function, as demonstrated in Figure 3-6,
51
where the probability is zero when Sa<Sa
y,exp
and then a constant when Sa>Sa
y,exp
. This
approximation allows for simply multiplying that constant with the mean annual
frequency of Sa
y,exp
being exceeded, MAF(Sa>Sa
y,exp
), to calculate the mean annual
frequency of demand exceeding capacity. MAF(Sa>Sa
y,exp
) can be read directly from the
ground motion seismic hazard curve.
( )
exp , exp ,
* ) ( ) (
y y
Sa Sa MAF Sa Sa C D P C D MAF > > > ~ > (3-8)
where
( )
exp ,
exp ,
0
y
y
Sa Sa
Sa Sa
if
if
A
Sa C D P
>
s
= > (3-9)
Re-arranging Equation 3-8, the probability of demand exceeding capacity post-yielding
can be calculated, given the mean annual frequency of demand exceeding capacity and
the mean annual frequency Sa>Sa
y,exp
, as shown in Equation 3-10.
( )
Ha R
y
y
Sa Sa MAF
C D MAF
Sa Sa C D P
,
exp ,
exp ,
) (
) (
) ( | u =
>
>
~ > > (3-10)
The reliability index,
R,Ha
, for Equation 3-1 can thus be calculated by using the inverse
standard normal cumulative distribution function, according to the following:
( )
|
|
.
|
\
|
>
>
u =
exp ,
1
,
) (
y
Ha R
Sa Sa MAF
C D MAF
| (3-11)
Now that a method to calculate
R,Ha
has been developed, the last two steps to complete
the reliability framework are to decide upon an appropriate mean annual frequency of
demand exceeding capacity, MAF(D>C), and to decide what deformation demand to use
to represent the post-yielding demand.
The tolerable mean annual frequency of demand exceeding capacity is subjective and
depends on the consequences of the demand exceeding capacity, which in turn depend on
factors such as redundancy in the structural system and the correlation between the
52
components within the system. As a practical matter, it is proposed to make certain
assumptions to come up with a baseline number, which can be changed later if desired.
Here the assumption is made that the probability of failure of capacity-designed
components at a given spectral acceleration multiplied by the probability that component
failure causes frame collapse can be added to the probability of frame collapse from
incremental dynamic analysis at a given spectral acceleration (see Equation 3-12).
Further, it is assumed that the probability that component failure causes frame collapse is
independent of the ground motion intensity, Sa. However, as will be demonstrated in the
examples in Chapter 4 and Chapter 5, this probability depends on the ground motion
intensity and for the specific cases analyzed it is low at the lower ground motion
intensities. These assumptions are made for several reasons. First, the failure of capacity-
designed components can be neglected during the initial structural analysis and instead
considered at the component level by checking member forces, greatly simplifying the
structural analysis. Second, by adjusting the probability that that a capacity-designed
components failure causes collapse, the methodology allows for differentiating between
components whose failure is believed to cause collapse directly (often due to low
redundancy within the system) from others whose failure is not believed to have severe
adverse effects. Lastly, they give a rational basis for establishing component reliability
and will promote discussion and steer decision making toward the question of: What is
the tolerable probability of collapse of a structure in 50 years due to failure of capacity-
designed components? Project 07 has set the current standards of tolerable collapse
probability to 1% in 50 years (Luco et al. 2007). However, current system reliability
methodologies (FEMA P695) exclude the possibility of failure of capacity-designed
components, at least explicitly, in the collapse probability calculations, often on the
premise that capacity design principles will ensure that they do not fail. Therefore the
question becomes, if failure of capacity-designed components was included in the
analysis, what would be the total tolerable probability of collapse? And if the system has
a lower probability of collapse then the tolerable threshold, can the design requirements
on capacity-designed components be relaxed?
53
The system reliability is calculated by the following equation:
( ) ( ) ( ) ( ) Sa Coll C D P C D Coll P Sa Coll P Sa Coll P Sys
C D Sys
, > > + =
>
(3-12)
where ) | ( Sa Coll P is the total probability of collapse given Sa and ) | ( Sa Coll P
Sys
is the
probability of frame collapse at a given spectral acceleration, as calculated from
incremental dynamic analysis procedures but excluding consideration of failures of
capacity-designed components. ) | ( C D Coll P
C D
>
>
is the probability of frame collapse
due to demand exceeding the capacity of capacity-designed components and
) , | ( Sa Coll C D P Sys > is the probability of demand exceeding the capacity of a capacity-
designed component. At low spectral accelerations, the probability of a system level
frame collapse excluding failure of capacity-designed components, ) | ( Sa Coll P
Sys
, is
close to zero, but as pointed out before, the capacity-designed components may fail at
these low spectral accelerations and thus can contribute to the overall probability of
frame collapse. And because of the much greater frequency of occurrence of ground
motions with low Sa levels, failures of capacity-designed components at low ground
motion intensities contribute significantly to the annual frequency of system failure.
Figure 3-7 demonstrates this point illustratively. Figure 3-7a shows both a frame collapse
fragility curve at a given spectral acceleration excluding considerations of capacity-
designed components failures as well as a frame collapse fragility curve when it is
assumed that the probability of component failure (Figure 3-7b) causes frame collapse.
When the two collapse fragility curve are integrated with the site ground motion hazard
curve in Figure 3-7c (Los Angeles, Lat 33.99, Long = -118.16, T
1
= 0.2s), the calculated
probabilities of frame collapse vary significantly, or from 1.0% to 4.0% in 50 years
despite the only a 5.0% probability of component failure at the MCE ground motion
intensity.
With ) | ( Sa Coll P
Sys
close to zero at the low Sa levels, conditioning the probability of
demand exceeding capacity on the non-collapse of the system in the second term of
Equation 3-12 can justly omitted as the probability of non-collapse is close to 100% in
the region of interest. Equation 3-12 can then be re-written as:
54
( ) ( ) ( ) ( ) Sa C D P C D Coll P Sa Coll P Sa Coll P
C D Sys
> > + =
>
(3-13)
To calculate the mean annual frequency of collapse, Equation 3-13 is integrated with a
site ground motion hazard curve. This integration can be performed numerically as:
( ) ( ) ( ) ( ) ( )
= > > + =
>
x all
i C D Sys
x Sa MAF Sa C D P C D Coll P Sa Coll P Coll MAF ) ( (3-14)
Following the assumption from Equations 3-8 and 3-9 that P(D>CSa>Sa
y,exp
) is a
constant, Equation, 3-14 can be re-written as:
( ) ( ) ( ) ) ( ) ( C D MAF C D Coll P x Sa MAF Sa Coll P Coll MAF
C D
x all
i Sys
> > + = =
>
(3-15)
The mean annual frequency of collapse can be converted into a probability of collapse in
50 years (making the typical assumptions that Sa occurrences are Poissonian) with
Equation 3-16:
( )
) ( 50 50
50
1 1
Coll MAF
years
e e Collapse P
= =
(3-16)
As these calculations are mainly dealing with small probabilities of collapse, the added
contribution of collapse due to demand exceeding the capacity of force-controlled
components can be calculated as follows:
( )
( ) ) ( 50 50
50
1 1
C D MAF C D Coll P
years C D
C D
e e Coll P
> >
>
>
= = A
(3-17)
With Project 07 having set the tolerable probability of collapse to 1.0% in 50 years, it
seems appropriate to specify the tolerable added probability of collapse due to failure of
capacity-designed components relative to 1% in 50 years. Table 3-2 provides target
R,Ha
-values calculated with Equation 3-11, for different R
R,Ha
in order for the capacity-designed components in those systems to be as reliable as
those in systems with low R
R,Ha
-values then a system with higher R
= (3-18)
For each of the ground motion hazard curves, the
R,Ha
is calculated using Equation
3-19 for different MAF(D>C) and for R
\
|
>
>
u =
exp ,
1
,
) (
y
Ha R
Sa Sa MAF
C D MAF
| (3-19)
57
R
Sa
Sa
MCE
y
3
2
exp ,
= (3-20)
( ) ( )
( ) C D Coll P
Coll P
C D MAF
C D
years C D
>
A
= >
>
>
50
1 ln
) (
50
(3-21)
Tables 3-4 to 3-6 along with Figures 3-7 to 3-9 show representative results of the
recommended
R,Ha
based on the different parameters analyzed.
For a given R
and ( )
years C D
Coll P
50 >
A can be included in a function to calculate
R,Ha
.
A linear regression analysis was used to find the best functional form to calculate
R,Ha
and it is presented in Equation 3-22. How well the suggested function calculates the
R,Ha
is presented in Figure 3-11 where the predicted
R,Ha
is the value calculated using
Equation 3-22 and the true
R,Ha
is calculated using Equation 3-19.
07 . 2 ) ( 36 . 1 13 . 0 ) ln( 75 . 0
50 ,
+ A + =
> years C D Ha R
Coll P R k
| (3-22)
94 . 0
2
= R
Since Equation 3-22 includes no system dependent variables and the coefficient of
determination, R
2
, is close to 1.0, it can be useful to determine general values of the
reliability index for any component within any systems following capacity design
principles.
58
3.2.2 Effect of Risk-Targeted MCE Target on Calculated
R,Ha
Seismic design forces are based on the spectral acceleration for the maximum
considered earthquake (MCE) at the fundamental period of the structure, Sa
T1
. This
spectral intensity is then (a) multiplied by two-thirds to the design basis earthquake
(DBE) intensity and (b) divided by the frames response modification coefficient, or R-
factor. Depending on the slope of the seismic hazard curve, this approach can result in
greatly varying return periods (frequencies of exceedence) for design level spectral
accelerations of buildings located at different geographic locations. For sites with steep
hazard curves, the likelihood of seeing the design forces is considerably larger than for
sites with a more gradually inclined hazard curves. This can be seen in Table 3-7 and
illustratively in Figure 3-12 where ground motion hazard curves for the San Francisco
and the New Madrid sites used previously are plotted. Table 3-7 demonstrates that a
frame with R
,
2 2
, tot Ha R D C D C Ha R
m
n
n
m
V B V V V V
C
C
D
D
= + = | |
|
(3-23)
where for this analysis, B is a constant set to 1.0,
R,Ha
is calculated according to Equation
3-11 and Sa
y,exp
according to Equation 3-3. The results are presented in Figures 3-13 to 3-
17.
Figure 3-13 and 3-14 show that the /|-ratio is more sensitive to R
in San Francisco
than in New Madrid. Therefore, the possible reduction in required design strengths due to
R
is more pronounced in areas where the seismic hazard curve is fairly steep, such as
San Francisco, than in areas where it is flatter, such as New Madrid. For both locations
analyzed, the sensitivity increases when R
decreases. This is a result of the difference in the frequency between the seismic design
forces reducing as R
decreases.
3.3 The Methodology and Guidelines for Future Use
The proposed methodology to establish the required design strength of capacity-
designed components, which will be referred to as the Capacity Design Factor
methodology as the required design strength is set by adjusting the capacity design
factors, and |, is to use Equation 3-1 to establish capacity design factors for
components in seismic resistant systems whose strength probability distribution is
lognormal.
61
( )
D C D C Ha R
m
n
n
m
V V V V
C
C
D
D
|
|
2 exp
2 2
,
+ = (3-1)
The methodology is based on the LRFD component reliability methodology but
adapted for capacity-based design concepts. Comparison between the LRFD
methodology and the Capacity Design Factor methodology is presented in Table 3-9.
m
D
and
m
C
, and
n
D and
n
C are the median and nominal values of the demand and capacity
probability distributions, respectively,
D
V and
C
V their lognormal standard deviations,
and the correlation between demand and capacity. The demand probability distribution
is based on maximum strength demands up to deformation demands corresponding to the
MCE ground motion intensity. The capacity design reliability index,
R,Ha
, provides a
measure of probability of demand exceeding capacity of capacity-designed components,
for a specified pair of and |. The reliability index,
R,Ha
, is based on the seismic
resisting systems R-factor and member overstrength, the site ground motion hazard
curve, of the influence of capacity-designed component failures on the system collapse
safety, and the tolerable added probability of frame collapse due capacity-designed
components failure. Table 3-10 has recommended target
R,Ha
for varying R
between 1
and 6 for two US sites, one in the Western US and one in the Central and Eastern US, and
for probabilities of demand exceeding capacity ranging from 0.05% to 1.00% in 50 years.
Table 3-10 shows that for R
is based on the code R-factor and the member overstrength in yielding capacity
due to the use of capacity design factors, nominal material values and strength equations,
discrete member sizing etc. As described previously, R
-ratio can become considerably higher than 1.5 when all factors and constraints are
accounted for. If that is the case, R
and P(Sa>Sa
y,exp
). The difference in
required design strengths due to R
becomes
more pronounced for very low R
, i.e. for R
, but
in different locations, depends very much on the component strength dispersions. For
component strengths, the total dispersion (both demand and capacity), V
tot
, can be
expected to be around 0.2 0.3 and for R
Spectral
Acceleration
San Francisco
(Lat = 38.0, Long = -121.7)
New Madrid
(Lat = 35.2, Long = -89.9)
Sa
T1
= x
P(Sa>x)
in 50 years
Sa
T1
= x
P(Sa>x)
in 50 years
Sa
MCE,2005
1.38g 2.0% 1.29g 2.0%
Sa
MCE,2010
1.44g 1.8% 1.04g 2.9%
Sa
DBE
=2/3Sa
MCE,2010
0.96g 6.1% 0.69g 5.2%
Sa
DBE
/(R
San Francisco
Lat = 38.0, Long = -121.7)
New Madrid
(Lat = 35.2, Long = -89.9)
P(D>C) in 50 years P(D>C) in 50 years
0.05% 0.10% 0.20% 0.50% 1.00% 0.05% 0.10% 0.20% 0.50% 1.00%
1 2.4 2.1 1.9 1.4 1.0 2.3 2.1 1.8 1.3 0.9
2 3.0 2.8 2.5 2.2 1.9 2.6 2.4 2.1 1.7 1.3
3 3.2 3.0 2.8 2.5 2.2 2.4 2.5 2.2 1.9 1.6
4 3.4 3.2 3.0 2.7 2.4 2.8 2.6 2.3 2.0 1.6
5 3.5 3.3 3.1 2.8 2.5 2.9 2.6 2.4 2.0 1.7
6 3.5 3.3 3.1 2.9 2.6 2.9 2.7 2.5 2.1 1.8
66
Table 3-3: Calculated local slopes, k, of the chosen hazard curves calculated between Sa
DBE
and Sa
DBE
/4.
Site
Period
0.1s 0.2s 0.3s 0.5s 1.0s 2.0s
Los Angeles
(Lat=34.0, Long=-118.2)
2.13 2.22 2.23 2.28 2.30 2.32
San Francisco
(Lat=38.0, Long=-121.7)
2.13 2.17 2.23 2.27 2.26 2.31
Eureka
(Lat=41.3, Long=-124.3)
1.40 1.37 1.32 1.23 1.29 1.43
Sacramento
(Lat=38.7, Long=-121.6)
2.36 2.44 2.50 2.50 2.42 2.34
Memphis
(Lat=35.2, Long=-89.9)
0.96 0.96 0.94 0.90 0.89 0.85
Seattle
(Lat=47.6, Long=-122.3)
1.90 1.91 1.90 1.86 1.81 1.78
Portland
(Lat=45.5, Long=-122.7)
1.60 1.62 1.61 1.54 1.50 1.49
Las Vegas
(Lat=36.2, Long=-115.1)
1.69 1.84 1.92 1.99 1.94 1.88
Stanford
(Lat=37.4, Long=-122.2)
1.80 1.78 1.71 1.59 1.51 1.53
New York
(Lat=40.7, Long= -74.0)
1.08 1.17 1.23 1.31 1.35 1.39
Boston
(Lat=42.4, Long= -71.1)
1.24 1.34 1.40 1.42 1.37 1.36
Charleston
(Lat=32.8, Long= -79.9)
0.85 0.86 0.83 0.80 0.79 0.79
67
Table 3-4: Calculated
R,Ha
according to Equations 3-19 and 3-20 for each ground
motion hazard curve when R
San Francisco
Lat = 38.0, Long = -121.7)
New Madrid
(Lat = 35.2, Long = -89.9)
P(D>C)
2005 2010 2005 2010
0.10% 1.00% 0.10% 1.00% 0.10% 1.00% 0.10% 1.00%
1 2.2 1.1 2.1 1.0 2.0 0.7 2.1 0.9
2 2.8 1.9 2.8 1.9 2.3 1.2 2.4 1.3
3 3.0 2.3 3.0 2.2 2.4 1.4 2.5 1.6
4 3.2 2.4 3.2 2.4 2.5 1.6 2.6 1.6
5 3.3 2.6 3.3 2.5 2.6 1.6 2.6 1.7
6 3.4 2.7 3.3 2.6 2.6 1.7 2.7 1.8
71
Table 3-9: Comparison between LRFD and proposed methodology for establishing
capacity design factors
Variables Load and Resistance Factor Design Capacity Design Factors
D
m
/D
n
Central values of the load effect
statistical parameters including:
1) Dead load
2) Live load
3) Wind load
4) Other loads
Central values of the following
demand statistical parameters:
1) Ratio of component strengths at
deformation demands equal to MCE
ground motion demands to strengths
predicted by nominal equations.
Demands determined from
a) Cyclic tests and/or
b) Nonlinear dynamic analysis
2) Material strengths
3) Geometrical properties (fabrication)
4) Record-to-Record
C
m
/C
n
Central values of the following
resistance statistical parameters
1) Ratio of expected resistance
strengths to strengths predicted by
nominal equations
2) Material strengths
3) Geometrical properties (fabrication)
Same as LRFD
V
D
Variability of load effect statistical
parameters
Variability of the demand statistical
parameters
V
C
Variability of resistance statistical
parameters
Same as LRFD
Correlation between statistical parameters
= 0.0 as it is generally not included in
LRFD
Correlation between statistical parameters
= 0.5 when demand and capacity
originate from the same member,
= 0 otherwise
was chosen based on average values
in structural design prior to LRFD.
Suggested values:
= 2.5 3.0 for members
= 4.0 4.5 for connections
R,Ha
will depend on:
1) The frames R-factor
2) The member overstrength
3) The sites seismic ground motion
hazard curve
4) Tolerable P(D>C) in 50 years
AP(CollD>C) in 50 years
P(CollD>CD>C)
72
Table 3-10: Minimum
R,Ha
for Western US and Central and Eastern US
System
R
San Francisco
Lat = 38.0, Long = -121.7)
New Madrid
(Lat = 35.2, Long = -89.9)
P(D>C) in 50 years P(D>C) in 50 years
0.05% 0.10% 0.20% 0.50% 1.00% 0.05% 0.10% 0.20% 0.50% 1.00%
1 2.4 2.1 1.9 1.4 1.0 2.3 2.1 1.8 1.3 0.9
2 3.0 2.8 2.5 2.2 1.9 2.6 2.4 2.1 1.7 1.3
3 3.2 3.0 2.8 2.5 2.2 2.4 2.5 2.2 1.9 1.6
4 3.4 3.2 3.0 2.7 2.4 2.8 2.6 2.3 2.0 1.6
5 3.5 3.3 3.1 2.8 2.5 2.9 2.6 2.4 2.0 1.7
6 3.5 3.3 3.1 2.9 2.6 2.9 2.7 2.5 2.1 1.8
Table 3-11: Recommended P(D>C) in 50 years for components.
Consequences of Failure
Tolerable P(Coll)
in 50 years
Description P(Coll
D>C
D>C) 0.05% 0.10% 0.20%
Severe
Failure of component causes frame
collapse. Structural redundancy is
very low and/or correlation
between components high
100% 0.05% 0.10% 0.20%
Considerable
Failure is likely to cause frame
collapse. Structural redundancy is
low and/or correlation between
components fairly high
50% 0.10% 0.20% 0.40%
Moderate
Failure is unlikely to cause frame
collapse. Structural redundancy is
high and/or correlation between
components fairly low
20% 0.25% 0.50% 1.00%
Small
Failure rarely causes frame
collapse. Structural redundancy is
very high and/or correlation
between components low
10% 0.50% 1.00% 2.00%
73
Table 3-12: A procedure to calculate R
Step Description
1 Design a frame according to code specification
2
Calculate the expected yield shear force of every story, V
y,exp
, using
expected material properties. Where member strengths is different
depending on if its in tension or compression, as is the case for braces
in SCBF, use the average between the two.
3
Perform a Response Spectrum Analysis using the design spectrum and
record the story shear forces, V
RSA
4 Calculate the minimum V
y,exp
/V
RSA
ratio
5 Sa
y,exp
= Sa
T1
* min(V
y,exp
/V
RSA)
6 R
= 2/3 Sa
MCE
/ Sa
y,exp
74
Table 3-13: Steps to calibrate the capacity design factors following the proposed methodology when
component strength probability distributions are lognormal
Step 1
Capacity Statistical Properties
How Comment
X
1
: Component
Model Variable
Median and dispersion of variable are
obtained through comparison of capacity
strengths from test data to predicted
strengths using nominal strength equations
Measured material and geometrical
properties are used in predicting strengths
through the nominal strength equations
X
2
: Material
Strength
Variable
Median and dispersion of variable are
obtained through comparison of material
strengths from test data to nominal material
strengths
Depending on the failure modes, R
y
and
R
t
are either included or excluded in the
design equations
X
3
: Fabrication
Variable
Median and dispersion of variable are
obtained through comparison of measured
geometrical properties to nominal ones
Recommended values: X
m,3
= 1.00, V
X3
=
0.05
C
m
/C
n
X
m,1
* X
m,2
* X
m,3
Multiplication of median values of the
statistical parameters
V
C
2
V
X1
2
+ V
X2
2
+ V
X3
2
When sufficient test data is not available,
the recommended V
c
= 0.15 for ductile
failure modes and V
c
= 0.25 for brittle
failure modes
Step 2
Demand Statistical Properties
How Comment
X
4
: Load Model
Parameter
Variable
Median and dispersion of variable are
obtained through comparison of strengths
from test data where members are subjected
to cyclic loading to predicted strengths
using nominal strength equations
Members will generally have reached
their maximum capacity before the MCE
ground motion demand so this variables
becomes the ratio between the maximum
values from test data to predicted
strengths
X
5
: Material
Strength
Variable
Median and dispersion of variable are
obtained through comparison of material
strengths from test data to nominal material
strengths
With the introduction of R
y
and R
t
in the
nominal strength, C
n
, the average value of
this factor is between 0.95 and 1.05
X
6
: Fabrication
Variable
Median and dispersion of variable are
obtained through comparison of measured
geometrical properties to nominal ones
Recommended values: X
m,6
= 1.00, V
X3
=
0.05
X
7
: Record-to-
Record Variable
Median and dispersion of variables are
obtained through comparison of strengths
from dynamic analysis at given intensities
to expected strengths
Recommended values: X
m,7
= 1.00, V
X3
=
0.05
D
m
/D
n
X
m,4
* X
m,5
* X
m,6
* X
m,7
Multiplication of the median values of the
statistical parameters
V
D
2
V
X4
2
+ V
X5
2
+ V
X6
2
+ V
X7
2
When sufficient test data is not available,
the recommended V
D
= 0.20
Step 3
Correlation Coefficient,
How Comment
Correlation between statistical parameters is
analyzed through published test data when
demand and capacity originate from the
same member
Correlation will generally be between the
material and fabrication variable.
14
and
25
can be individually estimated or a
suggested value of
DC
= 0.5 can be used
75
Step 4
Yield Response Modification Factor, R
How Comment
R
=
1.5 2.0 should be used. If due to various
reasons, members are considerably
overdesigned, a designer can calculate R
for his design, see Table 3-7
Step 5
Hazard Curve Effects
How Comment
Hazard Curve
The hazard curve effects are broken into 3
categories. Is the frame located in the
Western US, Nort-Western US or Central
and Eastern US?
A more site specific analysis can be
performed and the MAF(Sa>Sa
y,exp
)
calculated directly.
Step 6
Tolerable P(D>C) in 50 years
How Comment
P(D>C) in 50
years
It is up to code developers to determine the
tolerable probability but recommended
values are in Table 3-11 with 0.10% in 50
years being the default value
Higher probability can be justified if
consequences of component failure is
considered small, mainly due to system
being highly redundant an components
uncorrelated
Step 7
R,Ha
How Comment
R,Ha
Use R
R,Ha
= 0.51ln(k) + 2.33
1.5
2.0
2.5
3.0
3.5
0.5 1.0 1.5 2.0 2.5 3.0
|
R
,
H
a
k
Figure 3-8: The calculated
R,Ha
from Table 3-4 as a function of the local slope k when R
= 2 and
P(D>C) in 50 years is 0.10%.
R,Ha
= 0.79ln(k) + 2.31
1.5
2.0
2.5
3.0
3.5
0.5 1.0 1.5 2.0 2.5 3.0
|
R
,
H
a
k
Figure 3-9: The calculated
R,Ha
from Table 3-5 as a function of the local slope k when R
= 4 and
P(D>C) in 50 years is 0.20%.
81
R,Ha
= 0.96ln(k) + 2.10
1.5
2.0
2.5
3.0
3.5
0.5 1.0 1.5 2.0 2.5 3.0
|
R
,
H
a
k
Figure 3-10: The calculated
R,Ha
from Table 3-6 as a function of the local slope k when R
= 6 and
P(D>C) in 50 years is 0.50%
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0 1.5 2.0 2.5 3.0 3.5 4.0
T
r
u
e
|
R
,
H
a
Predicted |
R,Ha
Figure 3-11: True
R,Ha
vs. the predicted
R,Ha
(Equation 3-19 and 3-22).
82
0.00
0.01
0.02
0.03
0 0.1 0.2 0.3 0.4
SaT1=0.2s [g]
M
A
F
(
S
a
>
y
)
Figure 3-12: New and old design spectral accelerations for a R
/
|
)
S
F
R
/
|
)
N
M
R
/
|
)
S
F
R
and V
tot
for frames located in San Francisco. C
m
/C
n
and D
m
/D
n
= 1.0
84
1.0
2.0
3.0
1.0 2.0 3.0 4.0 5.0 6.0
(
/
|
)
N
M
R
and V
tot
for frames located in New Madrid. C
m
/C
n
and D
m
/D
n
=
1.0
1.0
1.1
1.2
1.3
1.4
1.0 2.0 3.0 4.0 5.0 6.0
(
/
|
)
S
F
/
(
/
|
)
N
M
R
Vtot = 0.2
Vtot = 0.3
Vtot = 0.4
Figure 3-17: Ratio of calculated /|-values calculated based on being located in San Francisco to
those based on being located in New Madrid.
85
Chapter 4
4 Capacity-Based Design in Single-Story Special
Concentrically Braced Frames
4.1 Introduction
In Special Concentrically Braced Frames (SCBF), the braces are the deformation-
controlled components, and the brace connections, beams and columns are the force-
controlled components, whose required design strength is based on the brace strengths.
Since the capacity-design principles dictate that connections should be stronger than the
braces connected to them, connection failure is usually not modeled in nonlinear
structural analysis. (NIST GCR 10-917-8, 2010). The objectives of this chapter are to
investigate the demand on brace connections, their reliability and the potential
consequences of brace connection failure on the seismic collapse safety of single-story
SCBFs. The study entails nonlinear dynamic analyses of two single-story SCBF designs,
results of which are interpreted using the reliability methodology developed in Chapter 3.
In the first portion of this study, the two single-story SCBF designs are analyzed
through incremental dynamic analysis (Vamvatsikos & Cornell, 2002). These analyses
are intended to represent the median model response, where the model parameters are
based on the expected values of material strengths and element model parameters. As
such, the analyses do not directly simulate connection failures. The two frames are
designed for the same mass and seismic hazard, but they have different brace sizes so as
to result in different values of overstrength and R
much less than the maximum specified R. Figure 4-1 shows the plan
and elevation view of the frame models, and the key frame properties and design values
are listed in Table 4-1. The braces selected for Frame 1 and HSS 6x6x5/16, which are the
smallest braces that satisfy the design code requirements. Regardless, there is a
significant overstrength in Frame 1 due to discrete member sizes, the resistance factor
and differences in brace compression and tension strengths. In Frame 2, the brace areas
are doubled with the use of HSS 6x6x5/8 members. The R
for Frame 2 is 1.4, both values significantly below R = 6.
The frames are idealized as two-dimensional plane frame models and analyzed using
OpenSees (OpenSees, 2011). The models are developed to represent median conditions,
where the expected (versus nominal) material strengths, section properties, and
component model parameters are used. With the median models, the failure of capacity-
87
designed components is not modeled in the analysis since their median strengths are
designed to exceed the brace median strengths. The braces are modeled using nonlinear
force-based fiber elements. The beams and columns are modeled with elastic elements
that have nonlinear rotational springs to model member hinging adjacent to the beam-
column connections and column bases. The rotational behavior of the flexural hinge
springs follows a bilinear hysteretic response based on the Modified Ibarra Krawinkler
Deterioration Model (Ibarra et al. 2005, Lignos and Krawinkler 2009, 2010). Braces are
assumed to have pinned end connections to the frame, and the two exterior columns are
fixed at the base. The center elastic pin-ended column is included to exclude beam
bending due to unbalanced brace forces. To capture the stiffening effect of the brace
gusset plates, rigid offsets are used at the brace-to-frame connections and the effective
length of the braces is 80% of the work-point-to-work-point length. The gravity system is
idealized as a leaning column to simulate P-Delta effects and provides no lateral
resistance. The gravity load (1080 kips) is applied to the leaning column as axial load.
Rayleigh damping is assigned equal to 2% of critical damping at the first two vibration
periods of the models (T
1
= 0.39s and T
2
= 0.17s).
The hysteretic characteristics of the braces are modeled using a method developed by
Uriz (2005). An illustration of the brace OpenSees model is shown in Figure 4-2. Each
brace is subdivided into twelve nonlinear fiber members, each of which has three
integration points along the length. Each fiber section is modeled with twenty fibers
through its depth. To induce global buckling, initial camber in the middle of the braces is
0.1% (1/1000), as recommended by Uriz (2005). For these two-dimensional models,
brace buckling is modeled to occur in-plane, but in concept the buckling response is
equally indicative of buckling in- or out-of-plane. A Menegotto-Pinto material model
(Steel02) with isotropic strain hardening is used for the brace fibers (Filippou et al., 1983).
To model low-cycle fatigue rupture of braces, the fatigue material developed by Uriz
(2005), and implemented in OpenSees, is applied to the braces (uniaxialMaterial Fatigue).
It tracks the strain history in each fiber and uses a modified rainflow cycle counting
algorithm to determine the strain amplitudes. The fatigue material uses a Manson-Coffin
relation calibrated to multiple brace tests characterizes low-cycle fatigue damage to each
88
fiber during each cycle, and Miners rule determines whether the fatigue life had been
exceeded. The fatigue material model in OpenSees wraps around the brace steel material
model, and once the fatigue material reaches a damage level of 1.0, the stress of the steel
material becomes zero. (Uriz, 2005)
To account for the non-simulated failure criteria of columns, 10% story drift limit is
used following recommendations from the SCBF analysis in NIST GCR 10-917-8. Those
recommendations are based on results from Newell and Uang (2006), which show that
columns begin to lose their capacity after 7% to 9% story drift when subjected to cyclic
axial and lateral loads.
The frames are subjected to the 44 ground motion records selected by and used in
FEMA P695 Quantification of Building Seismic Performance Factors (FEMA, 2009).
Unlike in the FEMA P695 study, where the ground motion set is scaled as a whole (based
on the median of the set) to the spectral acceleration at the first mode period of the
structure, in this study the ground motion records are scaled individually based on the
spectral acceleration at the first mode period of the structure. This is done to
systematically relate component demand to spectral acceleration, which can in return be
related to frequencies of exceedance via the ground motion hazard curve for the site at
that spectral period. The earthquake response spectra for the 44 ground motions as used
in this analysis are plotted in Figure 4-3. The frames are analyzed using incremental time
history analysis technique where each ground motion record is scaled up until frames
collapse (Vamvatsikos & Cornell, 2002). During each dynamic analysis, the maximum
axial forces developed in the braces, i.e. the maximum demand on the connections, are
recorded. Connection fractures are not simulated directly in the analysis. Instead, the
probability of brace connection failure is determined by comparing the maximum braced
demand from each analysis to the probability distribution of connection strength. The
probability of demand exceeding connection capacity at a given spectral acceleration is
then calculated, i.e. ( ) Sa C D P > . Based on the postulated consequences of connection
fractures, specifically the probability of collapse given that the brace strength is
89
exceeded, ( ) C D Coll P
C D
>
>
, the total probability of frame collapse including connection
fractures is calculated using Equation 3-12.
4.2.2 Brace Behavior
As described earlier, the braces are modeled using nonlinear fiber elements that can
capture brace yielding and global buckling. A fatigue material is utilized to capture low-
cycle fatigue fracture due to the combined effects of global and local buckling. Model
parameters for the steel material and fracture models are based on calibration studies by
Uriz for square HSS sections. (Uriz, 2005)
To check the cyclic response of the brace model, several braces were subject to far-
field and near-field cycling loading protocols. The loading protocols were developed by
Fell et al. (2006) by adjusting loading protocols from ATC-24 (ATC, 1992) to represent
SCBF behavior. Table 4-2 shows the far-field loading protocol. For near-field loading
(see Table 4-3_, both asymmetric compression and tension loading protocols were used.
The brace axial deformations are related to the story drift angle by the following, Fell et
al (2006):
B B a
L L + + = A o u o o u
2 2
sin sin cos 2 1 (4-1)
where A
a
is the axial deformation, u is the drift angle, o the brace angle and L
b
the brace
length. For o of 45 and L
b
of 203in (80% of the work-point-to-work-point length in the
two frames analyzed), the relationship between the axial deformation and the drift angle
is:
u 50 . 0 ~
A
b
a
L
(4-2)
Figures 4-4 to 4-9 show the brace hysteretic response as well as the loading history
for the far-field, near-field tension and near-field compression loading protocols. In all
cases the brace compression buckling, tensile yielding and eventually brace fracture due
90
to low-cycle fatigue can be observed. The peak tensile forces for the three loading
protocols, far-field, near-field tension and near-field compression are 1.10, 1.15 and 1.09
times the expected brace yield strength and the corresponding compression forces are
1.25, 1.06 and 1.18 times the expected brace buckling strength.
4.2.3 Calculating R
To estimate the spectral accelerations at which tensile yielding in the braces can be
expected, Sa
y,exp
, the procedure in Table 3-12 is used to calculate R
and Sa
y,exp
, as calculated from the pushover
analyses, also help confirm how well the procedure from Table 3-12 estimates R
and
Sa
y,exp
. The calculated values of R
and Sa
y,exp
, based on the Table 3-12 procedures, are
summarized in Table 4-4.
The calculated values of R
, a pushover analysis is also performed. Figures 4-10 and 4-11 show the
pushover analyses results. In the figures, both the compression brace and tension brace
response is shown along with the combined response. Based on the story shear forces
when the tension braces yield, which represents the peak values in these cases, R
s are
2.6 and 1.5 for Frames 1 and 2, respectively. A normalized story shear force of 1.0 equals
the design force and R
calculated by pushover
analysis and the Table 3-12 methodology arises because the tension and compression
braces, as calculated by the pushover analysis, do not reach maximum strengths at the
same story drifts. As shown in Figures 4-10 and 4-11, the compression brace typically
reaches its peak buckling strength before the tension brace is fully yielded. This is in
contrast to the behavior as assumed in the calculations in Table 4-4, where the two peaks
are simply added together.
91
4.2.4 Nonlinear Dynamic Analysis Results
During each incremental dynamic analysis, the frame story drift ratios and the brace
axial forces are recorded. Figures 4-12 to 4-15 show the recorded brace axial responses
for Frame 1 when subjected to ground motion record No. 1 at ground motion intensities
equaling, Sa
T1
= 0.2g, 0.4g, 0.8g and 1.0g. The figures range from the demonstrating the
elastic response of the braces in Figure 4-12, to the compression brace buckling and
initiation of brace tensile yielding in Figure 4-13, to brace fracture after multiple inelastic
cycles in Figures 4-14 and 4-15. The maximum story drift and the brace tensile forces for
Frame 1 and ground motion record No. 1 are shown in Figure 4-16. From Figure 4-16a, it
can be seen that the maximum story drift increases as the spectral acceleration increases,
up to the point that the frame collapses. The collapsed point is shown in the figure as a
cross with a dotted line connecting the last non-collapsed analysis point with the
collapsed analysis point. Similarly, as demonstrated in Figure 4-16b, the maximum brace
tensile force increases as the spectral acceleration increases. However, unlike the
maximum story drift, the maximum developed brace force is limited by the brace strength
capacity, which causes the maximum developed braces forces to saturate at the brace
strength capacity. The maximum brace forces just before frame collapse is the cross in
the figure.
The analysis process just described is then repeated for all other ground motion
records, resulting in a distribution of story drift ratios and maximum brace tensile forces
at Sa level. Figures 4-17 and 4-18 show the maximum story drift ratios and maximum
brace tensile forces versus spectral acceleration for both frames. Figures 4-17a and 4-18a
show that the increased elastic strength of Frame 2 causes large story drifts to occur at
considerably higher spectral accelerations compared to Frame 1. Similarly, Figures 4-17b
and 4-18b show that the brace tensile yielding occurs at higher spectral accelerations for
Frame 2 than for Frame 1. The maximum brace forces are only recorded for the non-
collapsed cases. Due to the brace strength capacity limiting the maximum forces that can
be developed, the maximum brace forces for collapsed cases and non-collapsed cases are
very similar. However, since later the brace forces are used to calculate the probability of
connection failure and the probability of collapse due to connection failure of non-
92
collapsed cases, only the maximum brace forces from the non-collapsed cases are
recorded. The variation in peak drifts and forces at each Sa level represents the record-to-
record randomness. Using statistics from the collapse points, collapse fragility curves for
each frame are then developed by fitting a lognormal distribution to the observed means
and standard deviations and plotted as the dashed lines in Figure 4-19.
The incremental dynamic analysis was performed using the FEMA P695 ground
motion records, each of which was linearly scaled up for each increment in the analysis.
Linear scaling of the ground motion set results in conservative estimates of response and
collapse safety at extreme ground motions as it does not account for the unique spectral
shape of extreme ground motions and their effect on behavior (Baker and Cornell, 2006).
The FEMA P695 methodology handles this issue by multiplying the median collapse
point by a so-called spectral shape factor. The spectral shape factor is based on the
buildings fundamental period, its period-based ductility, and the applicable Seismic
Design Category. For the frames analyzed here, the spectral shape factor is 1.3 based on
Table 7-1b from FEMA P695 Quantification of Building Seismic Performance Factors
(FEMA, 2009), ductility index of 6 and fundamental period below 0.5s. The solid
collapse fragility curves shown in Figure 4-19 are calculated by shifting the original
collapse fragility curves (shown dashed) by the FEMA P695 Spectral Shape Factor.
Since there are questions as to whether the FEMA P695 Spectral Shape Factor
applies to these analyses of brace force demands (versus median collapse capacity), the
results were checked by comparing collapse analysis results obtained using an alternative
ground motion set that has the appropriate spectral shape. Appendix A and Appendix B
contain results from this study on a similar single-story SCBF subjected to both the
FEMA P695 ground motion set as well as a ground motion set selected to match the
target response spectrum and variance, i.e. i.e. the conditional spectrum or CS. The CS
ground motions were selected using an algorithm created by Jayaram et al. (2010). The
analysis demonstrated that the demands on brace connections are not affected by which
ground motion set is used, a consequence of both ground motion sets being similar at the
lower spectral accelerations at which brace yielding initiates.
93
Figure 4-20 plots together the collapse fragility curve (top plot) as well as the
normalized median brace tensile force (bottom plot) versus spectral accelerations for both
frames. The figure demonstrates two key points. First, the median collapse capacity of
Frame 2 exceeds the median collapse capacity of Frame 1 by 34%. This result was
expected since Frame 2 is designed significantly stronger than Frame 1. Second, as
compared to Sa
MCE
of 1.10g, the braces in both frames reach their yield strength at
relatively low spectral accelerations (on the order of Sa of 0.5g), and show only a modest
increase in the maximum forces developed past that point (see Figure 4-20 and Table 4-5).
The spectral accelerations at which yielding is expected, Sa
y,exp
, are calculated in Table 4-
4 and related to the R
as follows:
Frame 1: g
R
Sa
Sa
MCE
y
40 . 0
3
2
,exp
= =
Frame 2: g
R
Sa
Sa
MCE
y
70 . 0
3
2
,exp
= =
These calculated values compare well with the analysis results summarized in Table 4-5.
For example, Table 4-5 shows that the median P
max
/P
y,exp
for Frame 1 at 0.40g is 0.96 and
1.04 at 0.60g and then only a small increase passed that point, while for Frame 2, the
median P
max
/P
y,exp
is 0.97 at 0.60g and 1.01 at 0.80g. Note that in the proposed
methodology, the brace force demand ratios are calculated at the MCE demands, which
result in ratios of P
max
/P
y,exp
equal to about 1.05 for these frames. While the braces
continue to strain harden up to a ratio of 1.1, this occurs at large spectral accelerations
with low frequencies of occurrence.
4.2.5 Probability of Brace Demand Exceeding Connection Capacity
The models used in this analysis are median models where the expected element
properties (e.g. material strengths) are used. By using median models, the failure of brace
connections is by default excluded in the analysis as their median strength capacities
94
exceed the braces median strength capacities. To calculate the probability of brace
demand exceeding connection capacity and its possible influence on the overall frame
collapse performance, the demand distribution developed by recording the maximum
brace tensile forces in the dynamic analysis is used, along with an assumed probability
distribution of connection capacity. With the demand and capacity distributions, the
reliability index, which can be related to a probability of connection failure, is calculated
by the following equation, assuming lognormal probability distributions:
D C D C
m
m
V V V V
D
C
|
2
ln
2 2
+
|
|
.
|
\
|
= (4-3)
|
|
|
.
|
\
|
|
|
.
|
\
|
=
|
|
.
|
\
|
n
m
m
n
m
m
C
C
D
D
D
C
(4-4)
where
m
D
and
m
C
, and
n
D and
n
C are the median and nominal values of the demand
and capacity probability distributions, respectively,
D
V and
C
V are their lognormal
standard deviations, and is the correlation coefficient between demand and capacity,
For this analysis, the correlation coefficient is assumed to be zero.
The dispersion around the brace connection demand recorded from the analysis is
only due to the record-to-record randomness and does not reflect other uncertainties, such
as material and fabrication variability. To include those additional uncertainties, an
additional dispersion of 0.15 is added to the recorded analysis demand dispersion using
the square root of the sum of the squares (SRSS), .e.g Equation 2-14. The initial
dispersion varies from 0.20 at the low spectral acceleration to 0.02 at the high spectral
acceleration, i.e. past brace yielding. The total dispersion at the high spectral
accelerations is therefore 0.15, which compares well with collected statistics on brace
forces from brace test results reported in Table 6-2. The analysis results are taken to
represent the
n m
D D /
-ratio, where
m
D
= = > = >
i
x all
i i
x Sa MAF x Sa C D P C D MAF ) ( * ) ( ) ( (4-7)
The mean annual frequency of collapse can be converted into a probability of collapse in
50 years (making the typical assumption that Sa occurrences are Poissonian) using the
following equation:
( )
) ( 50
50
1
Coll MAF
years
e Collapse P
= (4-8)
The probabilities of collapse in 50 years for the two frames and three |/-ratios are
presented in Table 4-6 and plotted in Figure 4-24. Note that the value of |/ equal to 0 in
Figure 4-24 corresponds to the case without connection failure. These data demonstrate
that the probabilities of collapse of Frame 2 are considerably lower than of Frame 1 as
well demonstrating that the impact of connection failures is considerably lower for Frame
2, a result of the braces developing the maximum demand at higher spectral accelerations
for Frame 2 than for Frame 1. The probability of collapse in 50 years for Frame 1
excluding connection fractures is 2.0% compared to 0.5% for Frame 2, representing a 4
times larger risk or a 1.5% absolute increase in risk for Frame 1. When the |/-ratio is
1.00, the collapse probabilities are 7.1% and 2.8%, for Frames 1 and 2, respectively. In
this case, the connection failures increase the risk of collapse in Frame 1 by 3.6 times (a
5.1% absolute increase in collapse probability) and in Frame 2 by 5.6 times (a 2.3%
97
absolute increase in probability). Comparing the two frames, the 4.3% difference in
absolute risk (between 7.1% and 2.8%) could be attributed to a difference of 1.5% due to
the frame behavior (2.0% versus 0.5%) with the remaining2.8% difference due to
connection fractures.
The probabilities of collapse in Table 4-6 are calculated based on collapse
fragility curves that have not been adjusted for the ground motion spectral shape factor.
Therefore, the collapse probabilities are larger than the actual values that would be
calculated following the FEMA P695 procedures. The main points of the data are to
demonstrate the relative effect of the connection failures on collapse and how these differ
depending on the R
for Frame 2 was 1.4 while 2.5 for Frame 1. The results demonstrated
that it takes a considerably larger ground motions to cause yielding in the over-designed
frames and that the spectral acceleration at which yielding occurs can be estimated fairly
well with the predictive equations for Sa
y,exp
from Chapter 3 using calculated R
based on
guidelines from Table 3-12. Consequently, when calculating the mean annual frequency
of brace demand exceeding connection capacity, the values for Frame 1 were 2.5 to 4.0
times higher than for Frame 2 if designed according to the same criteria, i.e. the same
/| - ratio. This suggests that if consistent mean annual frequency is the objective, the
/| - ratio can be relaxed for Frame 2 to 0.90 compared to the current ratio of 0.75.
Additionally, the overall probability of collapse for Frame 2 was considerably lower than
for Frame 1, signifying an even further allowance for relaxing the /| - ratio for Frame 2.
The probability of collapse in 50 years, excluding the probability of connection fractures,
was 2.0% and 0.5% for Frame 1 and Frame 2, respectively, and with /| - ratio of 1.00,
the probability was 7.1% and 2.8% for the two frames.
Connection fractures were not included in the first set of analyses so when relating
the probability of connection fractures to the probability of frame collapse, the
simplifying assumption was made that a connection fracture equals frame collapse, i.e.
( ) C D Coll P
C D
>
>
= 1.0. To assess the probability of collapse due to connection fractures,
as well as the influence of including modeling uncertainty in the analysis, the second set
of analyses was a full modeling uncertainty analysis where connection fractures were
105
simulated directly in the dynamic analysis. The results from the analysis demonstrated
that the probability of collapse due to connection fractures is not a constant, as assumed
in the methodology from Chapter 3, but rather a function of the ground motion intensity,
Sa
T1
. The probability of collapse due to connection fractures was close to zero at spectral
accelerations around Sa
y,exp
and then increased as Sa
T1
increased. These finding suggest
that connection fractures are likely to occur, even for relatively moderate ground motions,
but that connection fractures are unlikely to cause system collapse. The low probability of
collapse due to connection fractures at the lower spectral accelerations therefore greatly
reduces the impact of connection fractures on the overall frame reliability and allows for
the possibility to increase the tolerable probability of connection fractures, i.e. to reduce
the required design strength of connections. These absolute values for ( ) C D Coll P
C D
>
>
are likely to be very sensitive to how well frames can re-distribute the load once a
connection fractures, i.e. to redundancy and correlation within the system. The
probabilities may also depend on factors such as ground motion duration effects, which
are not explicitly considered in this study. However, the trend of low values at the low
intensities with a gradual increase until the peak impact is reached is likely to be similar
for other seismic resistant systems. Also, as the frame analyzed here only included a
single pair of braces and accounted for no lateral stiffness from the gravity system, the
results can be considered to give an upper bound on the likely influence of connection
fractures.
106
Table 4-1: Properties of the two single-story braced frames investigated
in Chapter 4
Properties Frame 1 Frame 2
Width 30 ft -
Height 15 ft -
Seismic Weight 1040 kips -
Code T
0
0.15 s -
Design Base Shear 173 kips -
Beam Section W18x35 -
Beam: F
y,exp
55 kips -
Columns Section W12x72 -
Column: F
y,exp
55 kips -
Brace Section HSS 6x6x5/16 HSS 6x6x5/8
Brace: F
y,exp
60.3 kips 60.3 kips -
|P
n,c
158 266
|P
n,t
266 474
L/r 88 94
F
y,exp
A
g
388 706
F
cr
A
g
196 325
V
RSA
173 -
V
y,exp
413 729
R
= V
DBE
/V
y,exp
2.5 1.4
Table 4-2: Far-field loading protocol used to analyze brace behavior
Load
Step
Number of
Cycles
Peak u
(rads)
Peak A
a
/L
b
(rad)
1 6 0.00075 0.000375
2 6 0.0010 0.00050
3 6 0.0015 0.00075
4 4 0.0020 0.0012
5 2 0.01025 0.005
6 2 0.0185 0.009
7 2 0.02675 0.013
8 2 0.04 0.020
9 2 0.05 0.025
107
Table 4-3: Near-field tension and compression loading protocols used to analyze brace behavior
Load
Step
Number of
Cycles
Near-Field Tension Near-Field Compression
Peak u
(rads)
Peak A
a
/L
b
(rad)
Peak u
(rads)
Peak A
a
/L
b
(rad)
1 6 0.00075 0.000375 0.00075 0.000375
2 0.0455 0.0228 0.0185 0.009
3 0.0415 0.0210 0.0605 0.030
4 1 0.0185 0.009 0.0185 0.009
5 2 0.0150 0.007 0.0150 0.007
6 4 0.01025 0.005 0.01025 0.005
7 6 0.0015 0.00075 0.0015 0.00075
8 4 0.0020 0.0012 0.0020 0.0012
9 2 0.01025 0.005 0.01025 0.005
10 2 0.0185 0.009 0.0185 0.009
11 2 0.02675 0.013 0.02675 0.013
12 2 0.04 0.020 0.04 0.020
13 2 0.05 0.025 0.05 0.025
Table 4-4: Estimation of R
for the two frames analyzed following guidelines from Table 3-12
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
1
[kips]
Compression
2
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
[g]
R
1 HSS 6x6x5/16 388 196 413 173 2.4 0.38 2.5
2 HSS 6x6x5/8 706 325 729 173 4.2 0.67 1.4
1) F
y,exp
* A
g
2) Column Strength Equation using expected material properties. The
effective brace length is 80% of the work-point-to-work-point length
108
Table 4-5: Median and COV of normalized maximum brace
forces, P
max
/P
y,exp
, from the Incremental Dynamic Analyses
Sa
T1
Frame 1 Frame 2
[g] Median COV Median COV
0.10 0.37 0.20 0.21 0.22
0.20 0.72 0.12 0.43 0.20
0.30 0.89 0.08 0.63 0.16
0.40 0.96 0.05 0.84 0.10
0.60 1.04 0.03 0.97 0.06
0.80 1.05 0.02 1.01 0.03
1.00 1.06 0.02 1.05 0.02
1.25 1.07 0.02 1.07 0.02
1.50 1.08 0.02 1.07 0.02
2.00 1.09 0.02 1.08 0.02
2.50 1.10 0.02 1.09 0.02
3.00 1.10 0.02 1.10 0.02
Table 4-6: Calculated probabilities of collapse in 50 years for
Frames 1 and 2 for variable |/-ratios
P(Coll) in 50 years Frame 1 Frame 2
Connection Failure Excluded 0.020 0.005
|/ = 0.75 0.022 0.006
|/ = 0.90 0.037 0.013
|/ = 1.00 0.071 0.028
109
Table 4-7: Table of random model parameters
Random Variable Central Value Dispersion Distribution Source
Ground Motions
GM # Min: 1 Max: 44 Uniform Discrete FEMA P695
Brace HSS6x6x5/16
F
y
60.3 0.08 Lognormal Liu (2003)
F
u
75.4 0.05 Lognormal Liu (2003)
A
g
6.43 0.10 Lognormal Ellingwood et al (1980)
/L 0.25% 0.7 Lognormal Uriz(2005),Galambos(2009)
m 0.500 0.20 Lognormal Uriz (2005)
E
0
0.095 0.30 Lognormal Uriz (2005)
Connection
Strength 1.35*P
y,brace
0.15 Lognormal Multiple (See Table 5-23)
Column W12x72
M
y
/M
y,p
1.17 0.21 Lognormal Lignos & Krawinkler
(2009)
M
c
/M
y
1.11 0.05 Lognormal
u
p
0.034 0.43 Lognormal
u
pc
0.160 0.41 Lognormal
A 2.0 0.43 Lognormal
u
u
Min: 0.05 Max: 0.06 Uniform
Beam W18x35
M
y
/M
y,p
1.17 0.21 Lognormal Lignos & Krawinkler
(2009)
M
c
/M
y
1.11 0.05 Lognormal
u
p
0.038 0.43 Lognormal
u
pc
0.130 0.41 Lognormal
A 1.5 0.43 Lognormal
u
u
Min: 0.05 Max: 0.06 Uniform
m: Slope of Coffin-Manson curve
in log-log space
E
0
: Value of strain at which one
cycle causes failure
M
y
: Effective yield moment
M
y,p
: Nominal yield moment
M
c
: Capping moment (w/strain hard.)
u
p
: Pre-capping rotation capacity
u
pc
: Post-capping rotation capacity
A: Cyclic deterioration parameter
u
u
: Ultimate rotation capacity
/L: Brace initial imperfection
110
Table 4-8: Correlation matrix between parameters in the
Modified Ibarra Krawinkler Deterioration Model
Random
Variable
u
p
u
pc
A
u
p
1.00 0.69 0.44
u
pc
0.69 1.00 0.67
A 0.44 0.67 1.00
Table 4-9: Probability of collapse of the median model of Frame 1 at the spectral
accelerations used in the modeling uncertainty analysis
Sa
T1
[g]
Collapse
Probability:
Simulation
Results
Collapse Probability:
Fitted Distribution
(Median = 1.81g,
Disp = 0.45)
Median
P
max
/P
y,exp
COV
P
max
/P
y,exp
0.40 0/44 = 0.00 0.00 0.96 0.05
0.80 1/44 = 0.02 0.03 1.05 0.02
1.25 9/44 = 0.20 0.21 1.07 0.02
1.50 16/44 = 0.36 0.34 1.08 0.02
2.00 26/44 = 0.59 0.59 1.09 0.02
Table 4-10: Probability of brace demand exceeding connection capacity
Sa
T1
[g]
D
m
V
D
C
m
V
C
P(D>C)
0.40 0.92 0.10 1.35 0.15 0.02
0.80 1.07 0.10 - - 0.11
1.25 1.09 0.10 - - 0.12
1.50 1.10 0.09 - - 0.12
2.00 1.11 0.09 - - 0.13
111
Table 4-11: Results for Frame 1 including modeling uncertainty and connection fracture
Sa
T1
[g]
No. of Collapses
(Connection
Failures
Excluded)
No. of Non-
Collapsed Models
w/Connection
Failures
No. of
Additional
Collapses
P(Coll
D>C
D>C,No
Coll
sys
)
P(Coll)
0.40 0 (0.0%) 17 0 0.0% 0.0%
0.80 35 (3.5%) 111 11 9.9% 4.6%
1.25 214 (21.4%) 114 36 31.6% 25.0%
1.50 402 (40.2%) 70 19 27.1% 42.1%
2.00 602 (60.2%) 65 14 21.5% 61.6%
Table 4-12: Calculated probabilities of collapse in 50 years for Frame 1 based on P(CollD>C)
P(Coll) in 50 years
P(CollD>C)
= 1.0
P(CollD>C)
= 0.1
P(CollD>C)
= 0.2
P(CollD>C)
= Results
from 4.3.2*
Connection Failure Excluded 0.020 0.020 0.020 0.020
|/ = 0.75 0.022 0.020 0.020 0.020
|/ = 0.90 0.037 0.021 0.024 0.021
|/ = 1.00 0.071 0.025 0.030 0.023
* P(Coll
D>C
D>C,No Coll
sys
) results from Section 4.3.2 (and listed in Table 4-11) are linearly
interpolated between Sa levels and used with Equations 4-5, 4-7 and 4-8 to calculate probabilities
of collapse in 50 years for Frame 1.
180 ft
120 ft
a) b)
Figure 4-1: SCBF analyzed for this example a) Plan b) Elevation
112
Fig
ure 4-2: OpenSees model of braces.
10
-2
10
-1
10
0
10
1
10
-3
10
-2
10
-1
10
0
10
1
Period (s)
S
p
e
c
t
r
a
l
A
c
c
e
l
e
r
a
t
i
o
n
(
g
)
Figure 4-3: Earthquake response spectra for the 44 ground motions used for the Incremental
Dynamic Analysis. The ground motions records are all scaled to have the same spectral acceleration
at the first mode period of the frames.
113
Figure 4-4: Response of the OpenSees model of a HSS6x6x5/16 brace section when subjected to a far-
field loading protocol. E0 and m are parameters of the fatigue material used. Brace fracture occurs
at relatively low axial deformations
Figure 4-5: Far-field loading protocol developed by Fell et al (2006) and used to analyze brace
behavior running dynamic analysis of SCBF frames
114
Figure 4-6: Response of the OpenSees model of a HSS6x6x5/16 brace section when subjected to a
near-field tension loading protocol. E0 and m are parameters of the fatigue material used.
Figure 4-7: Near-field tension loading protocol developed by Fell et al (2006) and used to analyze
brace behavior running dynamic analysis of SCBF frames
115
Figure 4-8: Response of the OpenSees model of a HSS6x6x5/16 brace section when subjected to a
near-field compression loading protocol. E0 and m are parameters of the fatigue material used.
Figure 4-9: Near-field compression loading protocol developed by Fell et al (2006) and used to
analyze brace behavior running dynamic analysis of SCBF frames
116
2
3
6
1
R
Figure 4-10: Results from pushover analysis on Frame 1 showing normalized base shear on the left y-
axis, estimated R
Figure 4-11: Results from pushover analysis on Frame 2 showing normalized base shear on the left y-
axis, estimated R
P
m
a
x
/
P
y
,
e
x
p
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
Frame1:CollapseFragilityCurve
Frame2:CollapseFragilityCurve
Frame1:NormalizedMedianBraceConnectionDemand
Frame2:NormalizedMedianBraceConnectionDemand
Figure 4-20: Frame 1 and 2 collapse fragility curves (above) and the median of the normalized
maximum brace tensile forces vs. Sa
T1
(below)
123
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.0 1.0 2.0 3.0
P
(
D
>
C
)
Sa
T1
(g)
|/ = 0.9
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.0 1.0 2.0 3.0
P
(
D
>
C
)
Sa
T1
(g)
|/ = 0.9
a) b)
Figure 4-21: The probability of brace demand exceeding connection capacity based on brace
demand distributions from the Incremental Dynamic Analyses when |/ = 0.9, C
m
/C
n
= 1.4 and V
c
=
0.15 for a) Frame 1 and b) Frame 2.
0.00
0.25
0.50
0.75
1.00
0.0 1.0 2.0 3.0 4.0 5.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
P(Collapse)fromIDA
P(Collapse)fromIDA+
ConnectionFailures
0.00
0.25
0.50
0.75
1.00
0.0 1.0 2.0 3.0 4.0 5.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
P(Collapse)fromIDA
P(Collapse)fromIDA+
ConnectionFailures
a) b)
Figure 4-22: The collapse fragility curves for a) Frame 1 and b) Frame 2 both including (|/ = 0.9)
and excluding connection failures.
10
-3
10
-2
10
-1
10
0
10
1
10
-6
10
-4
10
-2
10
0
Sa
T1=1.02
(g)
A
n
n
u
a
l
R
a
t
e
o
f
E
x
c
e
e
d
a
n
c
e
Figure 4-23: Site ground motion hazard curve used in this example to calculate mean annual
frequencies of collapse is a San Francisco hazard curve (Lat 38.0, Long -121.7)
124
0.00
0.02
0.04
0.06
0.08
0.00 0.25 0.50 0.75 1.00
P
(
C
o
l
l
)
i
n
5
0
y
e
a
r
s
|/
Frame1
Frame2
Figure 4-24: Calculated probabilities of collapse in 50 years for Frame 1 and Frame 2 versus |/-
ratio, i.e. the connection strength.
- M
y
and
y
: Effective yield
strength and rotation
- K
e
= M
y
/
y
: Effective
stiffness
- (M
c
and
c
): Capping
strength and associated
rotation for monotonic
loading
-
p
: Pre-capping rotation
capacity for monotonic
loading
-
pc
: Post-capping rotation
capacity
- M
r
= M
y
: Residual strength
-
u
: Ultimate rotation
capacity
Figure 4-25: Modified Ibarra Krawinkler Deterioration Model (Image from Lignos & Krawinkler,
2009)
125
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.20
0.40
0.60
0.80
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
Frame1:CollapseFragilityCurve
Frame1:NormalizedMedianBraceConnectionDemand
P(Coll)andmedianbraceconnectiondemandatspectral
accelerationsselectedforMonteCarlosimulations
Figure 4-26: The collapse fragility curve (above) based on the median model, the median of the
normalized maximum brace tensile forces vs. Sa
T1
(below) and the representative values of Sa
T1
where the full uncertainty analysis is performed
126
0.0
0.1
0.2
0.3
0.4
0.0 0.5 1.0 1.5 2.0
P
(
C
o
l
l
a
p
s
e
|
D
>
C
,
N
o
C
o
l
l
s
y
s
)
Sa
T1
(g)
Figure 4-27: Probability of collapse due to brace connection fracture for Frame 1
0.00
0.25
0.50
0.75
0.0 0.5 1.0 1.5 2.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
MedianModel
ModelingUncertainty
ModelingUncertainty+Conn.Fracture
ModelingUncertainty+Conn.Fracture
w/P(Coll|D>C)=1.0)
Figure 4-28: Probabilities of collapse for Frame 1 including the influence of modeling uncertainty
and connection failures
127
Chapter 5
5 Capacity-Based Design in Multi-Story Special
Concentrically Braced Frames
5.1 Introduction
In this chapter, dynamic analyses of multi-story Special Concentrically Braced
Frames are conducted with goals to (1) demonstrate the applicability of the methodology
developed in Chapter 3 for multi-story systems, (2) calculate the conditional probability
of collapse caused by connection failure, (3) investigate the demand distribution on
columns in braced frames, and (4) use the findings to recommend capacity design factors
for selected failure modes in brace connections.
Median models of 6- and 16-story SCBFs are studied through incremental dynamic
analysis to investigate the effects of building height and the quantity of deformation-
controlled components in a system on the force demands on brace connections and
columns. Of particular interest is to confirm whether the R
and Sa
y,exp
, following the
procedure outlined in Table 3-12. From the nonlinear dynamic analyses, the maximum
force demands on the capacity-designed connections and columns are recorded and used
to evaluate a) accuracy of Sa
y,exp
, as calculated by the simplified methods proposed in
Table 3-12, to the observed brace yielding b) the demand distributions on connections in
multi-story frames relative to those in single-story frames, c) required connection design
strengths, and d) required axial column design strengths in multi-story braced frames.
129
The frame models for this analysis are idealized two-dimensional plane frame
models, implemented in OpenSees (OpenSees, 2011). Figure 5-1 shows the plan and
elevation view of the frame models. The models are developed to represent median
conditions, where the expected (versus nominal) material strengths, section properties,
and component model parameters are used. The braces are modeled using nonlinear
force-based fiber elements which captures both global buckling and low-cycle fatigue
rupture of the braces. The effective length of the braces is 80% of the work-point-to-
work-point length. The beams and columns are modeled with elastic elements that have
nonlinear rotational springs to model member hinging adjacent to the beam-column
connections and column bases. The beam-column connections where there are no braces
connecting to the frames are assumed to have no rotational stiffness. The rotational
behavior of the flexural hinge springs follows a bilinear hysteretic response based on the
Modified Ibarra Krawinkler Deterioration Model (Ibarra et al. 2005, Lignos and
Krawinkler 2009, 2010). Braces are assumed to have pinned end connections to the frame,
and the columns are fixed at the base. The gravity system is idealized as leaning columns
to simulate P-Delta effects and provides no lateral resistance. The gravity load (1056 kips
per floor and 794 kips roof) is applied to the leaning column as axial load. Rayleigh
damping is assigned equal to 2% of critical damping at the first two vibration periods of
the models. The OpenSees modeling and dynamic analysis procedure for the two frames
follows the procedure for the single-story SCBF analyses described in Chapter 4.
5.2.2 Brace Behavior
The brace behavior is critical in the response of SCBFs. The braces are modeled
using nonlinear fiber elements where each brace is subdivided into 12 nonlinear fiber
sections with 3 integration points along the length of each section. 20 fibers are used
across the depth of the cross section and an initial imperfection of the braces is 0.1%.
Before running the dynamic analysis, the brace behavior is analyzed by subjecting an
OpenSees model of them to far-field and near-field cycling loading protocols. The
description of the loading protocols and examples of the brace responses when subjected
to cyclic loading protocols can be found in Chapter 4.
130
5.2.3 Calculating R
The following is a summary of the step-by-step procedure to calculate Sa
y,exp
and R
for the 6-story frame following the procedure in Table 3-12. Results for both the 6- and
16-story frames are then summarized in Tables 5-5 to 5-8. Values calculated by the Table
3-12 method are then verified by comparison to data from dynamic analyses to assess the
accuracy of the simplified method works for multi-story frames.
Step 1: Design a frame according to code specification
The frame used for these calculations was designed as a part of NIST GCR 10-917-8
Evaluation of the FEMA P-695 Methodology for Quantification of Building Seismic
Factors (2010). The frame was designed using the Equivalent Lateral Force procedure
in ASCE 7-05 for Seismic Design Category D
max
and Soil Site Class D. The resulting
brace sizes are listed in Table 5-1.
Step 2: Calculate the expected yield shear force for each story, V
y,exp
, using expected
material properties. V
y,exp
is calculated as the sum of the horizontal projection of the
expected tensile and compression strengths of each pair of braces. The braces in this
example are inclined at a 45 angle. The calculation results are listed in Table 5-2.
Step 3: Perform a Modal Response Spectrum Analysis using the design response
spectrum divided by the response modification factor, R, and record the story shear
forces, V
RSA
. The RSA is performed following the procedure in ASCE 7-10 using SAP
2000. Figure 5-2 shows the design response spectrum and Table 5-3 summarizes the
results.
131
Step 4: Calculate the minimum ratio of V
y,exp
/V
RSA
The calculation results are listed in Table 5-4. Here the minimum V
y,exp
/V
RSA
-ratio is
2.42. The V
y,exp
/V
RSA
-ratio is fairly constant between stories, suggesting a balanced and
optimal design.
Step 5: Calculate the spectral acceleration at which yielding is expected, Sa
y,exp
Sa
y,exp
= Sa
T1
* V
y,exp
/V
RSA
= 0.122g * 2.42 = 0.30g
Step 6: Calculate the frames R
.
Sa
MCE
= 1.10g
R
= 2/3 * Sa
MCE
/Sa
y,exp
= 2/3*1.10g/0.30g = 2.5
The effective R-factor of 2.5 is considerably smaller than the code R-factor of 6,
despite the frame being fairly optimally designed. However, assuming an average
overstrength of 2.0 for SCBFs, a value close to 3.0 (R/
0
= 6/2 = 3.0) could be expected,
even for optimal designs. The R/R
=
o
(5-3)
Figure 5-4 shows the ratio for the expected tensile yield stress over the nominal critical
stress as a function of the slenderness ratio. The figure shows clearly that the F
y,exp
/F
cr,nom
ratio increases sharply as the slenderness ratio increases. Figure 5-5 shows the ratio
between the expected critical stress and the nominal critical stress as a function of the
slenderness ratio. In this case the F
cr,exp
/F
cr,nom
ratio decreases as the slenderness ratio
increases until it reaches a value of 1.0. This is due to reduced impact of the yield stress
as the slenderness ratio increases. Figure 5-6 finally shows the combined results of
Figures 5-4 and 5-5. As demonstrated in Figure 5-6, the ratio between the expected yield
shear strength and the nominal shear strength is very dependent on the slenderness ratio.
For slenderness ratios of 60, 80 and 100 and of 1.0, the ratios between the expected
133
yield shear strength and the nominal shear strength are 1.3, 1.5 and 1.7 for HSS circular
sections.
In the previous calculations, the elastic design force demands are based on the design
spectrum, which may or may not match the elastic force demands from specific ground
motions. To investigate the significance of this, the R
for the 6-story frame is 2.7 using the median response spectra
compared to 2.5 for the design response spectra. For the 16-story frame, the calculated R
is 1.1 using the median response spectra and 1.0 using the design response spectra. The
difference between the design spectra and median response spectra is due to differences
in the spectra values at periods less than the fundamental periods, i.e. in the spectra
values used for higher mode effects calculations. Shown in Figure 5-7 is a comparison of
the two spectra, anchored at the respective periods (T
1
= 0.82s and 1.71s) for the two
frames. For both frames, large differences at periods less than the fundamental period is
evident, e.g. from Figure 5-7b it can be seen that the median response spectra is up to
70% higher than the design response spectra at periods between 0.2s and 0.6s for the 16-
story frame. The nonlinear analysis results will show that using the median ground
motion response spectra calculates R
and Sa
y,exp
more accurately, not surprising since
those best represent the analysis input.
134
5.2.4 Nonlinear Dynamic Analysis: Force Demands on Brace Connections
Results from the incremental dynamic analysis of the 6-story Design 1 and the 16-
story frames are presented in this section, including story drift ratios and force demands
on brace connections. Figure 5-8 shows the maximum store drift ratio versus Sa
T1
and
Figure 5-9 the collapse fragility curves for both frames. For the 6-story frame, the
calculated median collapse spectral acceleration is 1.67g. With Sa
MCE
= 1.10g, the
calculated collapse margin ratio, CMR, is 1.52. Based on FEMA P-695 (FEMA, 2009),
the spectral shape factor, SSF, is 1.21 and the adjusted collapse margin ratio, ACMR, is
1.84. For the 16-story frame, the calculated median collapse point is 1.02, the CMR is
1.93 and ACMR is 2.32. Based FEMA P695 minimum acceptance criteria, ACMRs
equal to 1.56 and 1.46 for the 6-story and 16-story frames, both frames pass. However,
the ACMR values calculated in this study are considerably lower those reported in NIST
GCR 10-917-8 Evaluation of the FEMA P-695 Methodology for Quantification of
Building Seismic Factors (2010) where the ACMR values for the two frames are 2.64
and 3.20. The difference between the results is likely due to the difference in modeling
the beam-column connections. The models in NIST GCR 10-917-8 did not capture all of
the deterioration models associated with beam plastic hinging and as result, the models
often had considerable lateral force capacity, even after all of the braces at a given story
had lost all their capacity in both tension and compression. Although proper modeling of
the beam-column hinges is extremely important when calculating collapse risk, the main
objective of this study is to investigate demand on and reliability of capacity-designed
components within multi-story systems, not calculating collapse risk. The results will
demonstrate that the area of interest for those factors is at lower earthquake intensities
where the probability of collapse is small and the details of beam-column hinge modeling
are not a major concern.
The maximum brace tensile forces per story, P
max
, normalized by the expected yield
strength, P
y,exp
, are plotted versus spectral acceleration for the 6-story frame in Figure 5-
11. Two points are worth mentioning from the plots. First, no story remains elastic
throughout all ground motions and all intensities, a result of a fairly balanced design. This
is evident by observing that the P
max
/P
y,exp
exceeds 1.0 for all stories at the higher spectral
135
accelerations, i.e. passed 0.3s. Second, the braces in story 5 and story 6 yield at the lowest
spectral accelerations, closely followed by other stories. These trends follow those
evident in Table 5-6 where stories 5 and 6 have the largest R
of 2.7 and a Sa
y,exp
of 0.27g and based on
the analysis results, those predictions are fairly accurate, i.e. the median P
max
/P
y,exp
for
story 6 is 0.97 at Sa
T1
= 0.30s with a limited increase at higher spectral accelerations
(Table 5-9).
The median of the maximum brace tensile forces for each story of the 16-story frame
are plotted in Figure 5-14. From 5-14a, corresponding to a low intensity shaking of Sa
T1
= 0.30g, it can be seen that brace yielding occurs first in stories 3 and 13. As reported in
Table 5-8, these two stories had the highest calculated R
ranged from 0.6 to 1.1, implying that some stories may remain elastic while others
yield. Figure 5-14c shows that the median of the normalized maximum brace forces for
stories 8 and 16 are well below 1.0, even at Sa
T1
= 1.00g which is the median collapse
point. Stories 8 and 16 also have the lowest R
-ratio and for the simplified method, the demand distribution corresponding to
the maximum considered earthquake ground motion intensity is used. For the connection
capacity distribution,
n m
C C /
2 exp
2 2
,
+ = (5-4)
> = >
Sa
Sa MAF Sa C D P C D MAF ) ( * ) ( ) ( (5-5)
The calculated |/-ratios are reported in Tables 5-13 and 5-14, and the ratios
calculated by the two different methods are plotted against each other in Figure 5-22. For
comparison, the current |/-ratio for brace connections in AISC 2010 (AISC, 2010a) is
0.75. The calculated |/-ratios range from 0.72 to 0.81 for the 6-story frame and from
1.25 to 1.77 for the 16-story frame. The absolute values of the calculated |/-ratios were
not of primary concern in this analysis, but rather to assess how well the simplified
method compares with the full integration method. The large difference between the
calculated |/-ratios for the 6-story frame and the 16-story frame can be explained by (a)
the large overstrength in the 16-story frame compared to the 6-story frame, i.e. lower R
,
and (b) the lower median demand per story in the 16-story frame compared to the 6-story
frame. Since large inelastic deformations tend to concentrate in different stories for
different ground motions, taking advantage of the lower median demand per story as the
number of stories increases can lead to unconservative results. It would therefore be more
appropriate to use the maximum of all stories statistics to calculate the |/-ratios.
Regardless, from Tables 5-13 and 5-14 as well as Figure 5-16, it can be observed that the
simplified method and the detailed integration method compare quite well, especially for
the 6-story frame where the overstrength between stories is similar. The results using the
simplified method are unconservative for most cases due to the step-function
simplification (see Figure 3-5), which ignores the possibility of demand exceeding
138
capacity at spectral accelerations below Sa
y,exp
. The difference between the simplified
method and the detailed integration method ranges from -3.0% to 31.0%. The
unconservatism using the simplified method is more pronounced in stories which are
significantly over-designed compared to others within the same frame. In those instances,
that over-designed story may stay elastic for all ground motions and at all earthquake
intensities. As a result, the median demand never stabilizes and dispersion stays high
causing the step-function simplification not to capture the true behavior. Figure 5-17
shows the strong relationship between R
results in
higher |/-ratio.
5.2.6 Nonlinear Dynamic Analyses: Axial Force Demands on Columns
Column axial force demands from the 6 and 16-story frames dynamic analyses are
presented in this section. Based on the 2010 AISC Seismic Provisions, the required
design strength of columns in SCBFs is the maximum of forces from either one of two
following analyses:
i) An analysis in which all braces are assumed to resist forces corresponding to their
expected strength in compression or in tension.
ii) An analysis in which all braces in tension are assumed to resist forces
corresponding to their expected strength and all braces in compression are
assumed to resist their expected post-buckling strength.
However, the required design strength of columns does not need to exceed any of the
following:
a) The forces determined using design loads, including amplified seismic load (O
0
),
applied to a frame model in which all compression braces have been removed.
b) The forces corresponding to the resistance of the foundation to overturning uplift.
c) Forces determined from nonlinear analysis.
139
The determination of the required design strength of columns in SCBFs in the 2010
Seismic Provisions has become significantly more elaborate than in the 2005 version,
especially with the inclusion and exclusion of the compression braces in the analysis.
Richards (2009) demonstrated that the column axial demands can easily exceed
elastic design demand by more than the
0
-factor of 2.0, especially for low rise frames
and in the upper stories in high-rise frames, due to force redistribution after brace
buckling. To account for the force redistribution, Richards (2009) suggested an
amplification factor, A
x
, for axial column loads at story x which accounts for the force
redistribution in story x when the compression brace at that story is removed:
+ = + =
=
+ =
n
c x j
j
n
c x i
i
n
x i
x i
x
h F
h F
A 1 (5-6)
where F
i
= lateral force at the floor above story i, h
x
, h
j
= story height of story x or j, n the
number of stories in the frame, and c = 1 if story x has chevron brace or c = 0 if story x
has V-bracing. The A
x
amplification factor is largest for low-rise frames and in the upper-
stories of high-rise frames and then slowly decreases as the number of stories above story
x increases. However, as pointed out by Richards (2009), Equation 5-6 might not capture
the maximum column demands as both the possible brace compression capacity
overdesign and the inevitable brace tension capacity overdesign can cause the maximum
column demands to far exceed the elastic design demands, even the when amplified by A
x
.
Therefore, capacity design provisions where these factors are accounted for are required
to capture the theoretical maximum column demands.
Capacity-design concepts for high-rise frames, which assume all braces reach their
maximum capacities simultaneously, can cause columns in the lower-stories to be
excessively overdesigned. Studies on Eccentrically Braced Frames taller than 9 stories
demonstrated that simultaneous yielding of all links never occurred and that the column
demands at the base were as low as 60% of the theoretical maximum force (Koboevic
and Redwood, 1997; Richards, 2004). On the other hand, the conclusions from nonlinear
dynamic analyses on 2-, 4-, 8- and 12-story chevron braced steel frames by Tremblay and
140
Robert (2001) are that column compression force must be determined assuming that all
compression braces carry their buckling strength simultaneously and that the square root
of the sum of the squares (SRSS method) for accumulating brace compression loads is
un-conservative. However, the dispersion in the column demand at the lower stories of
both the 8-story and the 12-story frames is considerable. For example, the column
demand for the bottom stories in the 12-story frame ranges from 50% of what capacity-
design provisions suggest up to close to 100% of what they suggest.
The column seismic axial demand for the 6-story and 16-story frames is plotted in
Figures 5-18 through 5-27, described below. In all cases, C
max
is the maximum seismic
axial demand in a column for a given ground motion record. The demand is normalized
by two different design rules. C
exp
is the expected column demand based on capacity
design principles, as given by the maximum of either design rule i) or ii) above.
C
elastic,AISC
is the elastic design demand using design rule a) above calculated without
0
.
Starting from the 1
st
level, C
exp
for the 6-story frame is [2466, 2466, 1151, 1168, 324,
274] kips and for the 16-story frame it is [9852, 9852, 8039 8039 6511, 6511, 4984, 4984,
3584, 3584, 2286, 2286, 1264, 1264, 424, 424] kips. Similarly, C
elastic,AISC
for the 6-story
frame is [1520, 1520, 833, 833, 274, 274] kips and for the 16-story frame it is [2498,
2498, 2074, 2074, 1655, 1655, 1250, 1250, 873, 873, 540, 540, 269, 269, 82, 82] kips.
The median and the dispersion of C
max
/C
exp
are plotted for both frames in Figures 5-
18, 5-21 and 5-24 to 5-27. The general trend for both frames is that the median demand
decreases as the number of stories above it increases and the dispersion increases. The
median demand for the 1
st
-story of the 6-story frame is approximately 90% of what
capacity design rules suggest, even at the MCE demand (Sa
T1
= 1.10g). For the 16-story
frame, the 1
st
-story demand is approximately 50% of what capacity design rules suggest
at the MCE demand. (Sa
T1
= 0.53g). On distinct difference between the column demand
and the brace connection demand is that the influence of R
and Sa
y,exp
, values, following the procedure outlined in Table 3-12 and
used in Section 5.2.3. The results are listed in Table 5-16 and 5-17. During each
nonlinear dynamic analysis, the maximum demands on the capacity-designed connections
and columns are recorded and used to a) compare the spectral acceleration at which
braces begin to yield in tension to Sa
y,exp
calculated through simplified methods proposed
in Table 3-12, b) compare the demand distributions on connections in over-designed
frames to the ones in optimally designed frames, c) calculate the required connection
design strengths, and d) investigate the demand distribution on columns in multi-story
braced frames
145
5.3.2 Nonlinear Dynamic Analyses: Force Demands on Brace Connections
The simulation results from the incremental dynamic analysis of Design 2 of the 6-
Story SCBF are presented in this section. During each dynamic analysis, the frames
story drift ratios and the brace and column axial forces are recorded. Figure 5-31a shows
the maximum story drift ratio versus Sa
T1
and Figure 5-31b the collapse fragility curve.
The median collapse point is 1.66g, which is almost identical to the median collapse point
for Design 1 of the 6-story SCBF. The dispersion of the collapse fragility curve for
Design 2 is however reduced compared to Design 1, or 0.52 compared to 0.58. Figure 5-
34 compares the collapse fragility curves for the two frames. The smaller dispersion for
Design 2 is a result of the inelastic deformations occurring only in the bottom two stories
and therefore the frames collapse capacity is solely based on the their capacities (see
Table 5-18 for location of collapses). This point is further illustrated in Figure 5-32,
which compares the maximum story drift results for the two designs. Figure 5-32 shows
that at low spectral accelerations, the story drift dispersion for Design 2 are lower than
that of Design 1. This results in lower probability of collapse at the low spectral
accelerations for Design 2 than for Design 1. However, with the median collapse capacity
equal, the end result is a lower dispersion in the collapse fragility curve. Figure 5-33
compares the maximum story drift ratio in story 1 versus the spectral accelerations for the
two 6-story frames and it clearly shows the increased deformation demands in story 1 of
Design 2 compared to story 1 of Design 1 due to the concentration of inelastic
deformations in the lower stories.
The maximum brace tensile forces per story normalized by expected yield strength
of the braces are plotted in Figure 5-36. The plot shows that brace tensile yielding occurs
almost exclusively in story 1 and story 2 while the braces in the other stories experience
limited tensile yielding throughout the incremental dynamic analysis. This was expected
based on the considerable member overstrength in the upper stories. However, as the
braces in the upper stories have generally not yielded in tension, the dispersion around
the maximum brace forces is high as small differences in story drift ratios result in large
differences in the brace forces developed. Tables 5-19 and 5-20 list the median and the
dispersion of the brace forces per story and the maximum for the whole frame and Figure
146
5-37 plots the results. The R
of 2.5 and a
Sa
y,exp
of 0.30g and based on the analysis results, those predictions are fairly accurate.
5.3.3 Nonlinear Dynamic Analysis: Required Design Strength of Brace
Connections
To assess how well simplified methods of establishing required connection design
strengths proposed by the methodology match with those when a full integration of the
probability of demand exceeding capacity and a seismic hazard curve is performed, the
|/-ratios are calculated following the same technique as described in Section 5.2.4. The
calculated |/-ratios are reported in Tables 5-21 and Figure 5-38 plots the ratios
calculated through the two different methods against each other, including the results
from the other 2 previously analyzed frames. Again, the simplified method and the
detailed integration method compare quite well, especially for stories which do
experience inelastic deformations. The calculated |/-ratio for the 6
th
story in Design 2
using the simplified method is noticeably unconservative compared to the integration
method, or 19% larger. The 6
th
story stays essentially elastic for all ground motions at all
earthquake intensities and as a result, the median demand never stabilizes and dispersion
stays high causing the step-function simplification not to capture the true behavior.
(Figure 5-40) Figure 5-39 shows the strong relationship between R
increases.
5.3.4 Nonlinear Dynamic Analyses: Axial Force Demands on Columns
The seismic axial demands from the dynamic analysis of 6-story Design 2 SCBF are
presented in this section. Similar to what was done in 5.2.5, the demand is normalized by
two different design rules. C
exp
and C
elastic,AISC
, which have been described previously.
Starting from the 1
st
level, C
exp
is [3561, 3561, 2147, 2147, 732, 732] kips and C
elastic,AISC
is [1520, 1520, 833, 833, 274, 274] kips.The median and the dispersion of C
max
/C
exp
are
plotted in Figures 5-41, 5-44 and 5-45. The general trend for the demand is the opposite
147
of the other 2 frames analyzed, i.e. it the normalized median demand increases as the
number of stories above it increases and the dispersion decreases, i.e. the dispersion is the
highest in the top stories. This is due to the overdesign of the upper stories which causes
the braces not to develop their full capacity and to remain essentially elastic. Therefore
small differences in story drift ratios result in large differences in the brace forces
developed. This also causes the demand to continue to increase past the frames Sa
y,exp
.
The median demand at the base of the frame is approximately 80% of what capacity
design rules suggest at the MCE demand (Sa
T1
= 1.10g) with a 10% dispersion but 67%
at Sa
y,exp
with a dispersion of 10%.
Based on the AISC 2010 Seismic Provisions, the required design strength of columns
does not need to exceed elastic design demand, C
elastic,AISC
. Figure 5-43 plots the column
axial demand normalized by C
elastic,AISC
(
0
excluded). It can be seen that this elastic
design rule causes un-conservative estimates of the demand. This could be expected as
the design rule does not take into consideration the true member overstrength. Even if the
first stories of Design 1 and Design 2 have similar member overstrength, the upper stories
in Design 2 are significantly over-designed (see Table 5-17), thus creating the capacity to
develop larger forces. Therefore, similar to the observations for the 16-story frame, using
a fixed over-strength factor,
0
, irrelevant of the design causes the demand to be greatly
under-estimated.
5.4 Probability of Collapse Including Connection Failures
To assess the probability of collapse due to connection failures, ( ) C D Coll P
C D
>
>
,
dynamic analyses of the 6-story SCBF Design 1 are performed again, now with the
possibility of brace fracture included. Unlike the analysis on the single-story frame in
Chapter 4 where brace fracture was modeled in a full Monte Carlo analysis that included
variability in all nonlinear model parameters, the only random variables in this analysis
are the brace connection strengths. A total of 880 models are created by generating
realization of the brace connection strengths using Monte-Carlo simulation methods
(Melchers, 1999; Rubinstein, 1981). The median connection strength is 1.4 times the
148
expected yield strength, P
y,exp
, of the connected brace connected and the brace strength
dispersion is 0.18. The brace connection strength is intentionally low (|/ = 1.00 instead
of 0.75) to increase the instances of connection fractures. The brace connection strengths
are assumed to be statistically independent from each other, except the brace connections
on either end of the same brace are assumed to be fully correlated. Each frame model is
then matched with a ground motion such, such that each ground motion is used 20 times.
To better quantify the added probability of collapse due to connection failures only, all
models are initially analyzed without connection fractures included. This requires only 44
dynamic analyses (the previously performed incremental dynamic analysis), since with
the connection fractures excluded, all of the model realizations are identical, and the only
difference between different simulations is the ground motion record. For the resulting
non-collapsed analysis cases, where the maximum demand from the dynamic analysis
exceeds the previously simulated connection strength, the dynamic analysis is re-run,
now including possible connection fractures and the number of additional collapses is
recorded.
The spectral accelerations of special interest are the relatively low ones where the
braces have already yielded in tension and the frequency of exceedance is still high, i.e.
Sa
T1
between Sa
y,exp
and Sa
MCE
. The selected spectral accelerations for the analyses are
from 0.40 to 2.50g. The trends observed in these analyses general agree with those from
the single-story analyses of Chapter 4. Where connection fractures occur at low spectral
accelerations, they have a low probability of causing collapse (practically zero probability
at an impose intensity of Sa
y,exp
)At higher intensities, the probability of collapse due to
connection failure increases to maximum of about 25.6%. Thus, as in the single-story
example, these analyses indicate that the consequence of connection failures on system
collapse is relatively low. Table 5-22 and Figures 5-47 to 5-49 show the results of the
analyses. The circles in Figure 5-47 represent the probability of collapse of the median
model at the specified spectral accelerations. The same values are shown in column 2 of
Table 5-22. The number of frame models where connection failures occurred, but the
systems did not collapse in the first set of dynamic analysis, are shown in column 3 of
Table 5-22. If a connection failure equaled frame collapse, those could be directly added
149
to the frame models which collapsed in the initial analysis. The diamonds on Figure 5-47
show the change in the collapse fragility curve if that was the case. However, when the
dynamic analyses are re-run with the possibility of connection failures included, only a
fraction of the models collapse during the dynamic analyses. The result of the second set
of dynamic analysis is shown in column 4 of Table 5-22. The probability of collapse
given connection failure and that the system has not otherwise collapsed is calculated in
column 5 of Table 5-22 and shown visually in Figure 5-48. Eventually, the total
probability of collapse is calculated and presented in column 6 of Table 5-22 and the
values represented by the squares in Figure 5-47. Fitted distributions of these three
different cases are plotted in Figure 5-55.
5.5 Conclusions
The main objectives of this chapter are (1) to demonstrate the applicability of the
methodology developed in Chapter 3 for multi-story systems, (2) to calculate the
probability of system collapse given connection failure, (3) to investigate the differences
between multi-story and single-story systems with respect to the proposed methodology,
and (4) to investigate the demand on columns in braced frames. To meet these objectives,
dynamic analyses of two 6-story and one 16-story SCBFs were conducted.
Modal Response Spectrum Analyses were performed for each of the frames to assess
the R
and Sa
y,exp
, values, following the procedure outlined in Table 3-12. The Modal
Response Spectrum Analyses were performed using both the design response spectrum as
well as using the median response spectrum of the 44 ground motions used in the
dynamic analyses. Similar to the single-story systems, the spectral acceleration at which
braces began to yield in tension for all frames compared well with the with the previously
calculated Sa
y,exp
, except for the 16-story frame when the design response spectrum was
used. This was due to higher mode effects playing a larger role in the 16-story frame and
a significant difference between the design response spectrum and the median ground
motion response spectrum at periods less than the fundamental period, suggesting that
care needs to be given to which response spectrum are used for taller buildings. For best
150
results, a ground motion response spectrum that best represents the given site conditions
should be used in the Modal Response Spectrum Analysis.
For balanced designs, such as the 6-story Design 1 SCBF, no story can be expected
to remain elastic throughout all ground motions and all intensities. Capacity-designed
components therefore need to be designed to resist the demand from deformation-
controlled components as they undergo inelastic deformations. The calculated |/-ratios
for the 6-story Design 1 frame demonstrated this where the difference between stories is
minimal. However, compared to single-story frames, the median brace connection
demand per story decreases and the variability increases with increased number of
deformation-controlled components. This is due to concentration of large inelastic
deformations in different stories for different ground motions, thus shielding the other
ones from large deformations. This effect becomes more pronounced as the number of
stories increases. Nevertheless, for all three multi-story frames analyzed, the maximum
normalized brace tensile force for the whole frame versus spectral acceleration curves
compare well with the same curve for the single-story frames, since inelastic
deformations will occur somewhere within a frame when subjected to large enough
earthquake ground motions. Therefore, for |/-ratio calculations in balanced designs
where inelastic deformations are as likely to occur in one story as the next, it is not
appropriate to take advantage of this decrease in the median demand as it will result in
weaker connections. Instead, the maximum normalized brace tensile force statistics
should be used.
When member overstrength varies significantly between stories, as was the case for
both the 6-story Design 2 frame and the 16-story frame, some stories may remain elastic,
or close to elastic, throughout all dynamic analyses. The 6-story Design 2 frame was
specifically designed to study this case. All braces up the height of the frame were
identical, and therefore the member overstrength in the upper stories was significant. The
dynamic analysis results demonstrated that inelastic deformation occurred almost
exclusively in story 1 and story 2. As a consequence, the calculated |/-ratio for each
story varies significantly, or by 72% between story 1 and story 6. For this case, requiring
the margin between the brace connection strengths and the brace yield strengths to be the
151
same for both stories is conservative. Caution should be taken though when some stories
are significantly over-designed compared to others within the frame as the demand
dispersion can be significant due the members still behaving elastically and therefore
small increase in deformation demands will cause a large increase in brace forces. Those
are cases where the capacity-design concept does not apply as well since the
deformation-controlled components do not necessarily reach their full capacity. As a
result, the simplified method proposed in Chapter 3 to establish the |/-ratio, resulted in
quite unconservative values as the step-function assumption it makes does not capture the
probability of demand exceeding capacity curve accurately.
The demand on columns in braced frames is a complex matter. As the results from
the three braced frame studies demonstrated, capacity design principles do not explain
well the expected demand on columns, rather the maximum theoretical demand that can
be delivered to them. For columns where there are only a few stories above them, the
expected demand is this theoretical maximum demand, at least for balanced designs.
Given the column capacity distribution and the consequences of exceeding column
capacities, the simplified method proposed by the methodology to establish the required
column strength can be used. However, for columns with multiple stories above them, the
maximum demand predicted by capacity design principles rarely develops and the
demand on the columns continues to increase passed Sa
y,exp
and even up to the median
collapse point. Therefore, if risk-consistent column strength guidelines are the goal, the
simplified method proposed by the methodology will not suffice but a full integration of
the probability of exceeding the column capacity and a seismic ground motion hazard
curve is required. However, full integration is not a practical solution. Fortunately, for
design guidelines such as those in TBI (2010) which condition the design checks at the
MCE ground motion intensity, this means that the column demand at the MCE ground
motion intensity is considerably larger than the column demand at lower intensities.
Unfortunately, it also means that the MCE demand might not capture the maximum
column demand that can be developed. However, as the frequency of experiencing those
large demands associated with the maximum column demand becomes very small, that is
not of a great concern. Conditioning the reliability calculations, or design checks, at the
152
MCE demand is therefore as valid as it is for the simplified method proposed by the
methodology, but the margin between demand and capacity at the MCE demand needs be
adjusted to account for the different shape of the probability of demand exceeding
capacity curve.
The AISC 2010 Seismic Provisions include a limit on the maximum required design
strength of columns. The required design strength does not need to exceed the forces
determined using design loads, including amplified seismic load (O
0
), applied to a frame
model in which all compression braces have been removed. However, this study showed
that the use of a fixed overstrength factor,
0
, irrelevant of the design causes the demand
to be greatly under-estimated.
The dynamic analyses of the 6-story SCBF Design 1 where the possibility of brace
connection fractures was included demonstrated, similar to a similar analysis in the
single-story frame, that the probability of collapse given connection failure is not a
constant as assumed in the methodology but that it rather gradually increases until it
reaches its peak values around 25% at the MCE demand. This greatly reduces the
influence of connection fractures on the system reliability as the main contribution of
connection fractures to the mean annual frequency of collapse would be at the low
intensities with high frequencies of occurrence. These results agree with similar
conclusions on the impact of connection fractures on system behavior in Luco and
Cornell (2000).
153
Table 5-1: Member sizes for the 6-Story and 16-Story SCBFs
Story
6-Story Frame
1
16-Story Frame
Beams
2
Braces
3
Columns
2
Beams
2
Braces
3
Columns
2
16 - - - W18x65 HSS9-5/8x3/8 W12x45
15 - - - W18x35 HSS9-5/8x3/8 W12x45
14 - - - W18x71 HSS8-5/8x1/2 W14x82
13 - - - W18x35 HSS8-5/8x1/2 W14x82
12 - - - W18x86 HSS11-1/4x1/2 W14x120
11 - - - W18x35 HSS11-1/4x1/2 W14x120
10 - - - W18x86 HSS10x5/8 W14x176
9 - - - W18x35 HSS10x5/8 W14x176
8 - - - W18x97 HSS11-1/4x5/8 W14x233
7 - - - W18x35 HSS11-1/4x5/8 W14x233
6 W18x97 HSS7-1/2x5/16 W14x68 W18x97 HSS11-1/4x5/8 W14x283
5 W24x104 HSS9-5/8x3/8 W14x68 W18x35 HSS11-1/4x5/8 W14x283
4 W24x131 HSS9-5/8x1/2 W14x176 W21x93 HSS11-1/4x5/8 W14x342
3 W18x76 HSS11-1/4x1/2 W14x176 W18x35 HSS11-1/4x5/8 W14x342
2 W24x146 HSS12-1/2x1/2 W14x342 W24x146 W12x96 W14x370
1 W21x62 HSS12-1/2x1/2 W14x342 W18x35 W12x96 W14x370
1) 6-Story Frame Design 2 has HSS12-1/2x1/2 at all floors (F
y,exp
= 60.3 kips)
2) ASTM A992 (F
y,exp
= 55 kips)
3) ASTM A500 Grade B (F
y,exp
= 55 kips)
154
Table 5-2: Expected story shear yielding force, V
y,exp
Story
Expected Brace Strength
V
y,exp
[kips]
Brace
Tension
1
[kips]
Compression
2
[kips]
6 HSS7-1/2x5/16 388 221 431
5 HSS9-5/8x3/8 600 428 727
4 HSS9-5/8x1/2 788 557 951
3 HSS11-1/4x1/2 929 724 1169
2 HSS12-1/2x1/2 1035 847 1331
1 HSS12-1/2x1/2 1035 847 1331
1) F
y,exp
* A
g
2) Column Strength Equation using expected material properties
The effective brace length is 80% of the work-point-to-work-point length
Table 5-3: Design story shear forces from
Modal Response Spectrum Analysis
Story
V
RSA
[kips]
6 171
5 301
4 383
3 469
2 519
1 526
Table 5-4: The V
y,exp
/V
RSA
-ratio for each story
Story
V
RSA
[kips]
V
y,exp
/V
RSA
6 171 2.53
5 301 2.42
4 383 2.49
3 469 2.49
2 519 2.56
1 526 2.53
155
Table 5-5: Summary of calculations performed to calculate the 6-Story Design 1 frames R
using the
design response spectrum
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
[kips]
Compression
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
[g]
R
6 HSS 7-1/2x5/16 388 221 431 171 2.53 0.31 2.4
5 HSS 9-5/8x3/8 600 428 727 301 2.42 0.30 2.5
4 HSS 9-5/8x1/2 788 557 951 383 2.49 0.30 2.4
3 HSS 11-1/4x1/2 929 724 1169 469 2.49 0.30 2.4
2 HSS 12-1/2x1/2 1035 847 1331 519 2.56 0.31 2.3
1 HSS 12-1/2x1/2 1035 847 1331 526 2.53 0.31 2.4
Table 5-6: Summary of the calculations performed to calculate the 6-story Design 1frames R
using the ground motion sets median response spectrum
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
[kips]
Compression
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
[g]
R
6 HSS 7-1/2x5/16 388 221 431 194 2.22 0.27 2.7
5 HSS 9-5/8x3/8 600 428 727 317 2.30 0.28 2.6
4 HSS 9-5/8x1/2 788 557 951 385 2.47 0.30 2.4
3 HSS 11-1/4x1/2 929 724 1169 468 2.50 0.30 2.4
2 HSS 12-1/2x1/2 1035 847 1331 526 2.53 0.31 2.4
1 HSS 12-1/2x1/2 1035 847 1331 539 2.47 0.30 2.4
156
Table 5-7: Summary of the calculations performed to calculate the 16-story frames R
using the
design response spectrum
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
[kips]
Compression
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
R
16 HSS9-5/8x3/8 600 428 727 49.7 14.6 0.86 0.4
15 HSS9-5/8x3/8 600 428 727 92.1 7.9 0.47 0.8
14 HSS8-5/8x1/2 700 452 815 111.9 7.3 0.43 0.8
13 HSS8-5/8x1/2 700 452 815 123.1 6.6 0.39 0.9
12 HSS11-1/4x1/2 929 724 1169 124.0 9.4 0.56 0.6
11 HSS11-1/4x1/2 929 724 1169 131.4 8.9 0.53 0.7
10 HSS10x5/8 1011 730 1231 131.4 9.4 0.55 0.6
9 HSS10x5/8 1011 730 1231 144.3 8.5 0.50 0.7
8 HSS11-1/4x5/8 1147 889 1439 150.3 9.6 0.56 0.6
7 HSS11-1/4x5/8 1147 889 1439 171.1 8.4 0.50 0.7
6 HSS11-1/4x5/8 1147 889 1439 181.7 7.9 0.47 0.8
5 HSS11-1/4x5/8 1147 889 1439 208.9 6.9 0.41 0.9
4 HSS11-1/4x5/8 1147 889 1439 220.2 6.5 0.39 0.9
3 HSS11-1/4x5/8 1147 889 1439 241.5 6.0 0.35 1.0
2 W12x96 1551 1086 1864 252.3 7.4 0.44 0.8
1 W12x96 1551 1086 1864 258.8 7.2 0.42 0.8
157
Table 5-8: Summary of the calculations performed to calculate the 16-story frames R
using the
ground motion sets median response spectrum
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
[kips]
Compression
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
R
16 HSS9-5/8x3/8 600 428 727 69 10.6 0.62 0.6
15 HSS9-5/8x3/8 600 428 727 118 6.2 0.36 1.0
14 HSS8-5/8x1/2 700 452 815 131 6.2 0.37 1.0
13 HSS8-5/8x1/2 700 452 815 134 6.1 0.36 1.0
12 HSS11-1/4x1/2 929 724 1169 129 9.1 0.54 0.7
11 HSS11-1/4x1/2 929 724 1169 135 8.7 0.51 0.7
10 HSS10x5/8 1011 730 1231 138 8.9 0.53 0.7
9 HSS10x5/8 1011 730 1231 149 8.3 0.49 0.7
8 HSS11-1/4x5/8 1147 889 1439 154 9.3 0.55 0.6
7 HSS11-1/4x5/8 1147 889 1439 168 8.6 0.51 0.7
6 HSS11-1/4x5/8 1147 889 1439 181 8.0 0.47 0.8
5 HSS11-1/4x5/8 1147 889 1439 209 6.9 0.41 0.9
4 HSS11-1/4x5/8 1147 889 1439 231 6.2 0.37 1.0
3 HSS11-1/4x5/8 1147 889 1439 259 5.6 0.33 1.1
2 W12x96 1551 1086 1864 286 6.5 0.38 0.9
1 W12x96 1551 1086 1864 292 6.4 0.38 0.9
158
Table 5-9: Median of the normalized maximum brace tensile force vs. Sa
T1
.for 6-story
SCBF Design 1
Sa
T1
[g]
Story 1 Story 2 Story 3 Story 4 Story 5 Story 6 Max All
0.10 0.52 0.47 0.44 0.44 0.55 0.59 0.64
0.20 0.82 0.77 0.76 0.78 0.88 0.92 0.94
0.30 0.91 0.86 0.86 0.86 0.95 0.97 0.98
0.40 0.93 0.94 0.92 0.92 0.97 0.98 0.99
0.60 0.95 0.98 0.94 0.97 0.98 0.99 1.01
0.80 0.97 0.99 0.98 0.98 0.99 1.00 1.02
1.00 0.98 1.00 0.98 0.98 0.99 0.99 1.03
1.25 0.99 1.01 0.98 0.99 0.99 0.99 1.03
1.50 0.99 1.01 0.99 0.99 0.99 1.00 1.03
2.00 0.99 1.02 0.99 1.00 0.98 1.00 1.04
Table 5-10: Dispersion of the normalized maximum brace tensile force vs. Sa
T1
.for 6-
story SCBF Design 1
Sa
T1
[g]
Story 1 Story 2 Story 3 Story 4 Story 5 Story 6 Max All
0.10 0.21 0.19 0.18 0.17 0.25 0.29 0.26
0.20 0.10 0.11 0.12 0.13 0.09 0.11 0.07
0.30 0.09 0.09 0.08 0.10 0.06 0.07 0.03
0.40 0.04 0.08 0.09 0.08 0.06 0.06 0.03
0.60 0.05 0.07 0.07 0.07 0.05 0.04 0.02
0.80 0.04 0.07 0.05 0.06 0.04 0.03 0.02
1.00 0.04 0.05 0.03 0.06 0.03 0.03 0.03
1.25 0.04 0.04 0.03 0.05 0.02 0.03 0.03
1.50 0.05 0.03 0.04 0.04 0.02 0.04 0.03
2.00 0.04 0.03 0.03 0.03 0.04 0.02 0.03
159
Table 5-11: Median of the normalized maximum brace tensile force vs. Sa
T1
.for 16-story SCBF
Story
Sa
T1
[g]
0.15 0.30 0.45 0.60 0.90 1.35 1.80
1 0.55 0.73 0.78 0.81 0.83 0.82 0.86
2 0.50 0.65 0.71 0.77 0.82 0.86 0.90
3 0.67 0.85 0.87 0.90 0.92 0.89 0.90
4 0.59 0.81 0.88 0.92 0.92 0.91 0.93
5 0.57 0.74 0.79 0.83 0.86 0.84 0.85
6 0.49 0.69 0.75 0.84 0.85 0.81 0.83
7 0.48 0.66 0.70 0.74 0.76 0.73 0.72
8 0.43 0.61 0.67 0.73 0.70 0.74 0.79
9 0.46 0.64 0.72 0.77 0.79 0.80 0.81
10 0.43 0.64 0.69 0.78 0.80 0.76 0.79
11 0.52 0.68 0.75 0.78 0.82 0.81 0.77
12 0.46 0.64 0.68 0.76 0.75 0.76 0.76
13 0.66 0.84 0.88 0.87 0.88 0.89 0.86
14 0.61 0.81 0.89 0.92 0.91 0.82 0.84
15 0.68 0.82 0.83 0.84 0.85 0.83 0.83
16 0.50 0.65 0.69 0.73 0.73 0.72 0.69
All 0.73 0.96 0.98 1.01 1.03 1.07 1.10
160
Table 5-12: Dispersion of the normalized maximum brace tensile force vs. Sa
T1
.for 16-story SCBF
Story
Sa
T1
[g]
0.15 0.30 0.45 0.60 0.90 1.35 1.80
1 0.25 0.16 0.14 0.13 0.15 0.20 0.12
2 0.26 0.19 0.20 0.18 0.18 0.13 0.16
3 0.24 0.10 0.11 0.09 0.12 0.16 0.13
4 0.26 0.16 0.13 0.10 0.15 0.16 0.13
5 0.22 0.12 0.13 0.15 0.18 0.23 0.19
6 0.23 0.17 0.19 0.16 0.16 0.22 0.19
7 0.24 0.17 0.15 0.15 0.18 0.22 0.23
8 0.28 0.22 0.20 0.17 0.19 0.23 0.17
9 0.33 0.21 0.20 0.19 0.15 0.16 0.15
10 0.31 0.25 0.23 0.18 0.16 0.17 0.19
11 0.27 0.20 0.18 0.18 0.15 0.18 0.21
12 0.26 0.19 0.20 0.16 0.18 0.21 0.21
13 0.26 0.14 0.11 0.17 0.15 0.17 0.21
14 0.32 0.17 0.13 0.08 0.14 0.20 0.19
15 0.28 0.16 0.14 0.13 0.13 0.19 0.20
16 0.40 0.27 0.24 0.19 0.22 0.23 0.23
All 0.27 0.06 0.03 0.04 0.04 0.05 0.06
161
Table 5-13: |/-ratios for 6-Story SCBF - Design 1 calculated by both full
integration and by simplified method proposed in the methodology
Story Brace V
y,exp
/V
RSA
Sa
y,exp
[g]
R
Detailed
|/
Simplified
|/
6 HSS 7-1/2x5/16 2.22 0.27 2.7 0.72 0.78
5 HSS 9-5/8x3/8 2.30 0.28 2.6 0.76 0.79
4 HSS 9-5/8x1/2 2.47 0.30 2.4 0.80 0.81
3 HSS 11-1/4x1/2 2.50 0.30 2.4 0.81 0.81
2 HSS 12-1/2x1/2 2.53 0.31 2.4 0.80 0.78
1 HSS 12-1/2x1/2 2.47 0.30 2.4 0.79 0.80
Table 5-14: |/-ratios for 16-Story SCBF calculated by both full
integration and by simplified method proposed in the methodology
Story Brace V
y,exp
/V
RSA
Sa
y,exp
[g]
R
Detailed
|/
Simplified
|/
16 HSS9-5/8x3/8 10.58 0.62 0.6 1.46 1.92
15 HSS9-5/8x3/8 6.17 0.36 1.0 1.25 1.55
14 HSS8-5/8x1/2 6.20 0.37 1.0 1.30 1.52
13 HSS8-5/8x1/2 6.07 0.36 1.0 1.29 1.50
12 HSS11-1/4x1/2 9.08 0.54 0.7 1.77 1.83
11 HSS11-1/4x1/2 8.65 0.51 0.7 1.59 1.70
10 HSS10x5/8 8.92 0.53 0.7 1.75 1.76
9 HSS10x5/8 8.27 0.49 0.7 1.67 1.74
8 HSS11-1/4x5/8 9.33 0.55 0.6 1.84 1.90
7 HSS11-1/4x5/8 8.56 0.51 0.7 1.74 1.85
6 HSS11-1/4x5/8 7.97 0.47 0.8 1.70 1.65
5 HSS11-1/4x5/8 6.89 0.41 0.9 1.54 1.63
4 HSS11-1/4x5/8 6.22 0.37 1.0 1.40 1.51
3 HSS11-1/4x5/8 5.56 0.33 1.1 1.30 1.29
2 W12x96 6.52 0.38 0.9 1.65 1.63
1 W12x96 6.39 0.38 0.9 1.53 1.69
162
Table 5-15: Member sizes for the 6-Story SCBF Design 2
Story Beams Braces Columns
6 W18x97 HSS12-1/2x1/2 W14x99
5 W24x104 HSS12-1/2x1/2 W14x99
4 W24x131 HSS12-1/2x1/2 W14x193
3 W18x76 HSS12-1/2x1/2 W14x193
2 W24x146 HSS12-1/2x1/2 W14x398
1 W21x62 HSS12-1/2x1/2 W14x398
Table 5-16: Summary of calculations performed to calculate the 6-story Design 2 frames R
using
the design response spectrum
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
[kips]
Compression
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
[g]
R
6 HSS 12-1/2x1/2 1035 847 1331 165 8.07 0.98 0.7
5 HSS 12-1/2x1/2 1035 847 1331 302 4.41 0.54 1.4
4 HSS 12-1/2x1/2 1035 847 1331 392 3.39 0.41 1.8
3 HSS 12-1/2x1/2 1035 847 1331 476 2.79 0.34 2.2
2 HSS 12-1/2x1/2 1035 847 1331 526 2.53 0.31 2.4
1 HSS 12-1/2x1/2 1035 847 1331 536 2.48 0.30 2.4
Table 5-17: Summary of the calculations performed to calculate the 6-story Design 2 frames R
using the ground motion sets median response spectrum
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6
Story Brace
Tension
[kips]
Compression
[kips]
V
y,exp
[kips]
V
RSA
[kips]
V
y,exp
/V
RSA
Sa
y,exp
[g]
R
6 HSS 12-1/2x1/2 1035 847 1331 185 7.18 0.88 0.8
5 HSS 12-1/2x1/2 1035 847 1331 318 4.18 0.51 1.4
4 HSS 12-1/2x1/2 1035 847 1331 397 3.35 0.41 1.8
3 HSS 12-1/2x1/2 1035 847 1331 475 2.80 0.34 2.1
2 HSS 12-1/2x1/2 1035 847 1331 531 2.51 0.31 2.4
1 HSS 12-1/2x1/2 1035 847 1331 547 2.43 0.30 2.5
163
Table 5-18: Story location of collapsed cases.
Comparison between the two 6-Story SCBFs
Story Design 1 Design 2
5 &6
11
3 & 4
5
1 & 2
22 36
Table 5-19: Design 2 - Median values of the normalized maximum brace tensile force vs. Sa
T1
Sa
T1
Median P
max
/P
y,exp
[g] Story 1 Story 2 Story 3 Story 4 Story 5 Story 6 Max All
0.10 0.54 0.47 0.38 0.33 0.31 0.18 0.55
0.20 0.90 0.82 0.68 0.61 0.55 0.35 0.88
0.30 0.96 0.94 0.81 0.74 0.60 0.38 0.96
0.40 0.97 0.97 0.86 0.79 0.69 0.43 0.98
0.60 0.98 0.99 0.90 0.85 0.76 0.50 1.00
0.80 0.99 1.00 0.93 0.88 0.81 0.49 1.01
1.00 1.00 1.01 0.93 0.91 0.84 0.54 1.03
1.25 1.00 1.02 0.96 0.94 0.81 0.57 1.04
1.50 0.98 1.00 0.97 0.95 0.86 0.57 1.03
2.00 1.01 1.02 0.97 0.97 0.89 0.62 1.05
Table 5-20: Design 2 - Dispersion of the normalized maximum brace tensile vs. Sa
T1
Sa
T1
COV of P
max
/P
y,exp
[g] Story 1 Story 2 Story 3 Story 4 Story 5 Story 6 Max All
0.10 0.28 0.23 0.23 0.19 0.29 0.34 0.26
0.20 0.11 0.13 0.15 0.15 0.21 0.26 0.09
0.30 0.05 0.08 0.10 0.13 0.19 0.24 0.03
0.40 0.05 0.06 0.09 0.13 0.19 0.27 0.02
0.60 0.04 0.03 0.12 0.13 0.19 0.31 0.03
0.80 0.04 0.03 0.07 0.11 0.14 0.28 0.02
1.00 0.04 0.05 0.12 0.09 0.17 0.25 0.04
1.25 0.03 0.07 0.07 0.07 0.18 0.25 0.04
1.50 0.09 0.11 0.08 0.11 0.15 0.29 0.05
2.00 0.04 0.10 0.06 0.06 0.14 0.22 0.04
164
Table 5-21: |/-ratios for 6-Story SCBF - Design 2 calculated by both full
integration and by simplified method proposed in the methodology
Story Brace V
y,exp
/V
RSA
Sa
y,exp
[g]
R
Detailed
|/
Simplified
|/
6 HSS 12-1/2x1/2 7.18 0.88 0.8 1.00 1.19
5 HSS 12-1/2x1/2 4.18 0.51 1.4 0.83 0.82
4 HSS 12-1/2x1/2 3.35 0.41 1.8 0.80 0.78
3 HSS 12-1/2x1/2 2.80 0.34 2.1 0.78 0.75
2 HSS 12-1/2x1/2 2.51 0.31 2.4 0.72 0.71
1 HSS 12-1/2x1/2 2.43 0.30 2.5 0.71 0.73
Table 5-22: Results for Design 1 including brace connection fracture
Sa
T1
[g]
No. of
Collapses
(Connection
Failures
Excluded)
No. of Models
w/Connection
Failures (and
didnt collapse)
No. of
Additional
Collapses
P(Coll
D>C
D>C,No
Coll
Sys
)
P(Coll)
0.40 20 (2.3%) 92 0 0.0% 2.3%
0.60 20 (2.3%) 99 5 5.1% 2.8%
0.80 80 (9.1%) 114 10 8.8% 10.2%
1.00 160 (18.2%) 129 24 18.6% 20.9%
1.25 260 (29.5%) 117 30 25.6% 33.0%
1.50 380 (43.2%) 78 19 24.4% 44.3%
2.00 560 (63.6%) 74 18 23.1% 65.7%
2.50 640 (72.7%) 39 5 12.8% 73.3%
165
180 ft
a) b)
Figure 5-1: Plan and elevation of 6- and 16-story frames analyzed.
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0.0 0.5 1.0 1.5 2.0
S
a
T
1
/
g
T
1
[s]
Figure 5-2: Design response spectrum used in Modal Response Spectrum Analysis
166
Figure 5-3: Pushover analysis results from Frame 2 used in IDA analysis described in Chapter 4.
1.0
2.0
3.0
4.0
5.0
0 20 40 60 80 100 120 140 160
F
y
,
e
x
p
F
c
r
,
n
o
m
L/r
Figure 5-4: The ratio of the expected tensile yield stress over the nominal critical stress vs. the
slenderness ratio in HSS circular section.
167
1.0
1.1
1.2
1.3
1.4
1.5
0 20 40 60 80 100 120 140 160
F
c
r
,
e
x
p
F
c
r
,
n
o
m
L/r
Figure 5-5: The ratio between the expected critical stress and the nominal critical stress vs. the
slenderness ratio in HSS circular section.
1.0
1.5
2.0
2.5
3.0
3.5
0 20 40 60 80 100 120 140 160
V
n
o
m
V
d
e
s
i
g
n
L/r
HSSCircularSection(alpha=1.0)
HSSCircularSection(alpha=0.8)
HSSCircularSection(alpha=0.6)
Figure 5-6: The ratio between the expected yield shear strength and the nominal shear strength vs.
the slenderness ratio in HSS circular section.
0.0
0.5
1.0
1.5
2.0
2.5
0.0 1.0 2.0 3.0 4.0 5.0
S
a
T
1
T
1
[s]
DesignResponseSpectrum
GroundMotionSetMedianResponseSpectrum
0.0
0.5
1.0
1.5
2.0
2.5
0.0 1.0 2.0 3.0 4.0 5.0
S
a
T
1
T
1
[s]
DesignResponseSpectrum
GroundMotionSetMedianResponseSpectrum
a) b)
Figure 5-7: Comparison of design response spectrum and ground motion median response spectrum
for a) 6-Story SCBF Design 1 b) 16-Story SCBF
168
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
0.0
0.3
0.6
0.9
1.2
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
a) b)
Figure 5-8: Maximum story drift ratio vs. Sa
T1
for a) 6-story SCBF - Design 1 b) 16-story SCBF
0.00
0.25
0.50
0.75
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurve
(Median=1.67g,Dispersion=0.58)
IncrementalDynamicAnalysis
Results
0.00
0.25
0.50
0.75
1.00
0.0 0.5 1.0 1.5 2.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurve
(Median=1.02g,Dispersion=0.46)
IncrementalDynamicAnalysis
Results
a) b)
Figure 5-9: The collapse fragility curve, developed directly from incremental dynamic analysis
results, for a) 6-story SCBF Design 1 b) 16-story SCBF
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.20
0.40
0.60
0.80
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,e
x
p
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurve
MaxofAllBraces:NormalizedMedian
BraceConnectionDemand
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.20
0.40
0.60
0.80
1.00
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,e
x
p
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurve
(Median=1.02g,Dispersion=0.46)
MaxofAllBraces:Normalized
MedianBraceConnectionDemand
SaMCE = 0.54g
a) b)
Figure 5-10: The collapse fragility curve (above) and the median of the normalized maximum brace
tensile forces vs. Sa
T1
(below) for a) 6-story SCBF Design 1 b) 16-story SCBF
169
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story1
Story2
Story3
Story4
Story5
Story6
0.0
0.1
0.2
0.3
0.4
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
(
P
m
a
x
/
P
y
,
e
x
p
)
Sa
T1
(g)
Story1
Story2
Story3
Story4
Story5
Story6
a) b)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
MaxofAllStories
0.00
0.10
0.20
0.30
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
(
P
m
a
x
/
P
y
,
e
x
p
)
Sa
T1
(g)
MaxofAllStories
c) d)
Figure 5-11: 6-Story Design 1 incremental dynamic analysis results. a) Median of the normalized
maximum brace tensile force vs. Sa
T1
. b) COV of the normalized maximum brace tensile force vs.
Sa
T1
. c) Median of the normalized maximum brace tensile force for entire frame vs. Sa
T1
. d) COV of
the normalized maximum brace tensile force for entire frame vs. Sa
T1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story3
Story16
0.0
0.1
0.2
0.3
0.4
0.5
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
(
P
m
a
x
/
P
y
,
e
x
p
)
Sa
T1
(g)
Story3
Story16
a) b)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
MaxofAllStories
0.0
0.1
0.2
0.3
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
(
P
m
a
x
/
P
y
,
e
x
p
)
Sa
T1
(g)
MaxofAllStories
c) d)
Figure 5-12: 16-story incremental dynamic analysis results. a) Median of the normalized maximum
brace tensile force vs. Sa
T1
. b) COV of the normalized maximum brace tensile force vs. Sa
T1
. c)
Median of the normalized maximum brace tensile force for entire frame vs. Sa
T1
. d) COV of the
normalized maximum brace tensile force for entire frame vs. Sa
T1
170
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story6
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story5
a)
b)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story4
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story3
c)
d)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story1
e) f)
Figure 5-13: 6-story SCBF Design 1 IDA results. Normalized maximum brace tensile force vs. Sa
T1
for non-collapsed cases.
171
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=0.30g
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=0.60g
a) b)
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=1.00g
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-14: Median of the maximum brace tensile demand normalized by the expected brace
strength for 16-Story SCBF
0.0
0.1
0.2
0.3
0.4
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=0.30g
0.0
0.1
0.2
0.3
0.4
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=0.60g
a) b)
0.0
0.1
0.2
0.3
0.4
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
P
m
a
x
/
P
y
,
e
x
p
Story#
SaT1=1.00g
0.0
0.1
0.2
0.3
0.4
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
P
m
a
x
/
P
y
,e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-15: Dispersion of the maximum brace tensile demand normalized by the expected brace
strength for 16-Story SCBF
172
0.0
0.5
1.0
1.5
2.0
0.0 0.5 1.0 1.5 2.0
A
p
p
r
o
x
i
m
a
t
e
|
/
r
a
t
i
o
True|/ ratio
16StoryFrame
6StoryFrameDesign1
Figure 5-16: |/-ratios calculated through simplified method vs. |/-ratios calculated through
integration of dynamic analysis results for the 6-story and the 16-story frames
0.0
0.5
1.0
1.5
2.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T
r
u
e
|
/
r
a
t
i
o
R
eff
16StoryFrame
6StoryFrameDesign1
CurrentAISC2010ratio
Figure 5-17: Calculated |/-ratios vs. R
o
f
C
m
a
x
/
C
e
x
p
Sa
T1
(g)
Story1 Story2
Story3 Story4
Story5 Story6
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Sa
T1
(g)
Story1
Story2
Story3
Story4
Story5
Story6
a) b)
Figure 5-18: 6-story SCBF Design 1 a) Median of column axial force demand normalized by
expected demand vs. Sa
T1
. b) Dispersion of column axial force demand normalized by expected
demand vs. Sa
T1
.
0
1
2
3
4
5
6
0.50 0.75 1.00 1.25 1.50
S
t
o
r
y
C
max
/C
exp
Sa
T1
=0.30g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
0.50 0.75 1.00 1.25 1.50
S
t
o
r
y
C
max
/C
exp
Sa
T1
=1.10g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
0.50 0.75 1.00 1.25 1.50
S
t
o
r
y
C
max
/C
exp
Sa
T1
=1.65g
Median
84thpercentile
16thpercentile
a) b) c)
Figure 5-19: 6-story SCBF Design 1. Seismic axial demand in columns normalized by expected
demand using capacity design principles
0
1
2
3
4
5
6
1.0 1.5 2.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=0.3g
Median
84th percentile
16th percentile
0
1
2
3
4
5
6
1.0 1.5 2.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=1.10g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
1.0 1.5 2.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=1.65g
Median
84th percentile
16th percentile
a) b) c)
Figure 5-20: 6-story SCBF Design 1. Seismic axial demand in columns normalized by maximum
design strength as per AISC 2010 Seismic Provisions
174
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Sa
T1
(g)
Story1 Story5
Story9 Story11
Story13 Story15
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.0 0.5 1.0 1.5
C
O
V
o
f
C
m
a
x
/
C
e
x
p
Sa
T1
(g)
Story1 Story5
Story9 Story11
Story13 Story15
a) b)
Figure 5-21: 16-story SCBF a) Median of column axial force demand normalized by expected
demand vs. Sa
T1
. b) Dispersion of column axial force demand normalized by expected demand vs.
Sa
T1
.
0
2
4
6
8
10
12
14
16
0.00 0.25 0.50 0.75 1.00 1.25
S
t
o
r
y
C
max
/C
exp
Sa
T1
=0.30g
Median
84th percentile
16th percentile
0
2
4
6
8
10
12
14
16
0.00 0.25 0.50 0.75 1.00 1.25
S
t
o
r
y
C
max
/C
exp
Sa
T1
=0.53g
Median
84th percentile
16th percentile
0
2
4
6
8
10
12
14
16
0.00 0.25 0.50 0.75 1.00 1.25
S
t
o
r
y
C
max
/C
exp
Sa
T1
=1.00g
Median
84thpercentile
16thpercentile
a) b) c)
Figure 5-22: 16-story SCBF. Seismic axial demand in columns normalized by expected demand using
capacity design principles
0
2
4
6
8
10
12
14
16
0.0 2.0 4.0 6.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=0.30g
Median
84th percentile
16th percentile
0
2
4
6
8
10
12
14
16
0.0 2.0 4.0 6.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=0.53g
Median
84thpercentile
16thpercentile
0
2
4
6
8
10
12
14
16
0.0 2.0 4.0 6.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=1.00g
Median
84thpercentile
16thpercentile
a) b) c)
Figure 5-23: 16-story SCBF. Seismic axial demand in columns normalized by maximum design
strength as per AISC 2010 Seismic Provisions
175
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.30g
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.60g
a) b)
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.10g
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-24: Median of the maximum column axial demand normalized by the expected column
demand for 6-Story SCBF Design 1
0.00
0.05
0.10
0.15
0.20
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.30g
0.00
0.05
0.10
0.15
0.20
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.60g
a) b)
0.00
0.05
0.10
0.15
0.20
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.10g
0.00
0.05
0.10
0.15
0.20
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-25: COV of the maximum column axial demand normalized by the expected column
demand for 6-Story SCBF Design 1
176
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.30g
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.60g
a) b)
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.00g
0.0
0.2
0.4
0.6
0.8
1.0
0 2 4 6 8 10 12 14 16
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-26: Median of the maximum column axial demand normalized by the expected column
demand for 16-Story SCBF
0.00
0.10
0.20
0.30
0.40
0.50
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.30g
0.00
0.10
0.20
0.30
0.40
0.50
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.60g
a) b)
0.00
0.10
0.20
0.30
0.40
0.50
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.00g
0.00
0.10
0.20
0.30
0.40
0.50
0 2 4 6 8 10 12 14 16
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-27: Dispersion of the maximum column axial demand normalized by the expected column
demand for 16-Story SCBF
177
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
NormalizedForceDemand
TrueForceDemandPDF
(Dispersion0.4)
TBIDesignForceDemandCDF
(Mean=1.62,COV=0.17)
Probability of Design Demand
Exceeding the True Demand
Figure 5-28: Overlap of the TBI design force demand and the true force demand
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
C
max
/C
exp
AnalysisResultsatSa=0.30g
FittedLognormalDistribution
(Median=0.37,Disp=0.18)
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
C
max
/C
exp
TBIDesignForceDemandCDF
(Mean=0.67,COV=0.09)
AnalysisForceDemandPDF
(Median=0.37,Disp=0.18)
a) b)
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
C
max
/C
exp
AnalysisResultsatMCEDemand
FittedLognormalDistribution
(Median=0.44,Disp=0.20)
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
C
max
/C
exp
TBIDesignForceDemandCDF
(Mean=0.67,COV=0.09)
AnalysisForceDemandPDF
(Median=0.44,Disp=0.23)
c) d)
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
C
max
/C
exp
AnalysisResultsatSa=1.00g
FittedLognormalDistribution
(Median=0.49,Disp=0.24)
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
a
n
c
e
C
max
/C
exp
TBIDesignForceDemandCDF
(Mean=0.67,COV=0.09)
AnalysisForceDemandPDF
(Median=0.49,Disp=0.24)
e) f)
Figure 5-29: 16-story SCBF. Axial force demand and the probability of exceeding the TBI design
demand in the 1
st
story columns at a-b) Sa
y,exp
, c-d) Sa
MCE
, e-f) Median collapse point
178
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.0 0.5 1.0 1.5
P
r
o
b
a
b
i
l
i
t
y
o
f
E
x
c
e
e
d
i
n
g
T
B
I
D
e
s
i
g
n
D
e
m
a
n
d
Sa
T1
[g]
1stStoryColumn
Figure 5-30: 16-story SCBF. Probability of exceeding the TBI design demand in the 1
st
story columns
vs. Sa
T1
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
0.00
0.25
0.50
0.75
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
Design1:CollapseFragilityCurve
(Median=1.67g,Dispersion=0.58)
Design2:CollapseFragilityCurve
(Median=1.66g,Dispersion=0.52)
a) b)
Figure 5-31: 6-story SCBF - Design 2 a) Maximum story drift ratio vs. Sa
T1
b) The collapse fragility
curve developed directly from Incremental Dynamic Analysis results
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
a) b)
Figure 5-32: Comparison of the maximum story drift ratio vs. Sa
T1
for the two 6-story SCBFs
analyzed a) Design 1 and b) Design 2.
179
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
0.0
0.5
1.0
1.5
2.0
0.00 0.02 0.04 0.06 0.08 0.10
S
a
T
1
(
g
)
StoryDriftRatio
a) b)
Figure 5-33: Comparison of the maximum story drift ratio in story 1 vs. Sa
T1
for the two 6-story
SCBFs analyzed a) Design 1 and b) Design 2
0.00
0.25
0.50
0.75
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
Design1:CollapseFragilityCurve
(Median=1.67g,Dispersion=0.58)
Design2:CollapseFragilityCurve
(Median=1.66g,Dispersion=0.52)
Figure 5-34: Collapse fragility curves for the two 6-Story SCBFs.
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.20
0.40
0.60
0.80
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurve
(Median=1.66g,Dispersion=0.52)
MaxofAllBraces:Normalized
MedianBraceConnectionDemand
Figure 5-35: The collapse fragility curve (above) and the median of the normalized maximum brace
tensile forces vs. Sa
T1
(below) for the 6-Story SCBF Design 2
180
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story6
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story5
a)
b)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story4
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story3
c)
d)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story1
e) f)
Figure 5-36: 6-story SCBF Design 2 IDA results. Normalized maximum brace tensile force vs. Sa
T1
for non-collapsed cases.
181
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
Story1
Story2
Story3
Story4
Story5
Story6
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
(
P
m
a
x
/
P
y
,
e
x
p
)
Sa
T1
(g)
Story1
Story2
Story3
Story4
Story5
Story6
a) b)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
(g)
MaxofAllStories
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
(
P
m
a
x
/
P
y
,
e
x
p
)
Sa
T1
(g)
MaxofAllStories
c) d)
Figure 5-37: 6-Story SCBF Design 2 Incremental Dynamic Analysis results. a) Median of the
normalized maximum brace tensile force vs. Sa
T1
. b) COV of the normalized maximum brace tensile
force vs. Sa
T1
. c) Median of the normalized maximum brace tensile force for entire frame vs. Sa
T1
. d)
COV of the normalized maximum brace tensile force for entire frame vs. Sa
T1
182
0.0
0.5
1.0
1.5
2.0
0.0 0.5 1.0 1.5 2.0
A
p
p
r
o
x
i
m
a
t
e
|
/
r
a
t
i
o
True|/ ratio
16StoryFrame
6StoryFrameDesign1
6StoryFrameDesign2
Figure 5-38: |/-ratios calculated through simplified method vs. |/-ratios calculated through
integration of dynamic analysis results for all 3 frames
0.0
0.5
1.0
1.5
2.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
T
r
u
e
|
/
r
a
t
i
o
R
eff
16StoryFrame
6StoryFrameDesign1
6StoryFrameDesign2
Figure 5-39: Calculated |/-ratios vs. R
o
f
C
m
a
x
/
C
e
x
p
Sa
T1
(g)
Story1 Story2
Story3 Story4
Story5 Story6
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.0 0.5 1.0 1.5 2.0
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Sa
T1
(g)
Story1 Story2
Story3 Story4
Story5 Story6
a) b)
Figure 5-41: 6-story SCBF Design 2 a) Median of column axial force demand normalized by
expected demand vs. Sa
T1
. b) COV of column axial force demand normalized by expected demand vs.
Sa
T1
.
0
1
2
3
4
5
6
0.00 0.50 1.00 1.50 2.00
S
t
o
r
y
C
max
/C
exp
Sa
T1
=0.30g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
0.00 0.50 1.00 1.50 2.00
S
t
o
r
y
C
max
/C
exp
Sa
T1
=1.10g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
0.00 0.50 1.00 1.50 2.00
S
t
o
r
y
C
max
/C
exp
Sa
T1
=1.65g
Median
84thpercentile
16thpercentile
a) b) c)
Figure 5-42: 6-story SCBF Design 2. Seismic axial demand in columns normalized by expected
demand using capacity design principles
0
1
2
3
4
5
6
1.0 2.0 3.0 4.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=0.30g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
1.0 2.0 3.0 4.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=1.10g
Median
84thpercentile
16thpercentile
0
1
2
3
4
5
6
1.0 2.0 3.0 4.0
S
t
o
r
y
C
max
/C
elastic,AISC
Sa
T1
=1.65g
Median
84thpercentile
16thpercentile
a) b) c)
Figure 5-43: 6-story SCBF Design 2. Seismic axial demand in columns normalized by maximum
design strength as per AISC 2010 Seismic Provisions
184
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.30g
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.60g
a) b)
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.10g
0.00
0.25
0.50
0.75
1.00
1.25
0 2 4 6
M
e
d
i
a
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-44: Median of the maximum column axial demand normalized by the expected column
demand for 6-Story SCBF Design 2
0.00
0.10
0.20
0.30
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.30g
0.00
0.10
0.20
0.30
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=0.60g
a) b)
0.00
0.10
0.20
0.30
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.10g
0.00
0.10
0.20
0.30
0 2 4 6
D
i
s
p
e
r
s
i
o
n
o
f
C
m
a
x
/
C
e
x
p
Story#
SaT1=1.65g
c) d)
Figure 5-45: Dispersion of the maximum column axial demand normalized by the expected column
demand for 6-Story SCBF Design 2
185
1.20
1.00
0.80
0.60
0.40
0.20
0.00
0.20
0.40
0.60
0.80
1.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0
M
e
d
i
a
n
P
m
a
x
/
P
y
,
e
x
p
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurve
MaxofAllBraces:NormalizedMedianBraceConnectionDemand
P(Coll)andmedianbraceconnetiondemandatspectral
accelerationsselectedforMonteCarlosimulations
Figure 5-46: The collapse fragility curve (above), the median of the normalized maximum brace
tensile forces vs. Sa
T1
(below) and the representative values at the Sa
T1
where the Monte-Carlo
simulations are performed.
0.0
0.2
0.4
0.6
0.8
0.0 0.5 1.0 1.5 2.0 2.5
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
MedianModel
MedianModel+Conn.Fracture
MedianModel+Conn.Fracture
w/P(Coll|D>C)=1.0)
Figure 5-47: Probabilities of collapse for the 3 different models of Design 1
186
0.0
0.1
0.2
0.3
0.4
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
a
p
s
e
|
D
>
C
,
N
o
C
o
l
l
s
y
s
)
Sa
T1
(g)
Figure 5-48: Collapse fragility curve after brace connection fracture for Design 1
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
MedianModel
(Median=1.67,Dispersion=0.58)
MedianModel+Conn.Fracture
MedianModel+Conn.Fracture
w/P(Coll|D>C)=1.0)
Figure 5-49: Collapse fragility curves for Design 1. Comparison between the model with and without
brace connection fractures included in the analysis
187
Chapter 6
6 Application of Capacity Design Factor
Methodology
6.1 Introduction
In this chapter the proposed Capacity Design Factor methodology is summarized and
then applied to selected brace connection failure modes to demonstrate its use and
provide guidance on its application. Capacity design factors are calculated for the
selected failure modes, based on collected statistical data on brace connection demand
and capacity as well as collapse behavior and other force-demand data from the SCBF
analyses in Chapters 4 and 5. In these examples, the medians and dispersions of the force
demands are combined with data to characterize the statistical variation of force demands
and capacities in the force-controlled connections.
The calculations of capacity design factors are carried out using two methods. First is
the simplified method proposed in Chapter 3 that allows for a closed form solution of the
target reliability index, |
R,Ha
, and capacity design factors. Second is an integration
method that allows for relaxing some of the assumptions required for the closed form
solution in the simplified method, most notably the fixed probability of frame collapse
given a component failure. The results of two methods are then compared.
188
6.2 Digest of Proposed Methodology
The proposed Capacity Design Factor methodology is a component reliability
methodology that incorporates system reliability effects through definition of the load
demands and the target reliability index. It calculates the required design strength of
capacity-designed components such that the reliability is consistent between different
components, systems and seismic regions. Design equations for capacity-designed
components generally take on the following form:
n n
D C | > (6-1)
where
n
D and
n
C are the respective nominal values of demand, D, and capacity, C, as
specified in design codes, and and | are the demand and capacity factors (load and
resistance factors) meant to account for the uncertainties inherent in determination of the
nominal values of demand and capacity. The required design strength of capacity-
designed components is set by adjusting the demand and capacity factors, and |, until
the desired reliability is achieved. When demand and capacity probability distributions
are lognormal, the basic equation to calculate the ratio between capacity design factors is:
( )
D C D C Ha R
m
n
n
m
V V V V
C
C
D
D
|
|
2 exp
2 2
,
+ = (6-2)
where
m
D
and
m
C
R,Ha
, provides a measure of probability of demand exceeding capacity of capacity-
designed components. The relationship between the reliability index,
R,Ha
, and the
probability of demand exceeding capacity is shown in Equation 6-3 where
1
u is the
inverse cumulative standard normal distribution function.
( ) ( ) C D P
Ha R
> u =
1
,
| (6-3)
189
To apply the methodology, quantification of the demand and capacity of components
is required together with a target reliability index,
R,Ha
.
Capacity: The capacity is generally the strength limit state of capacity-designed
components. To quantify the capacity, the median ( X
X X X
C
C
n
m
= (6-4)
=
=
3
1
2
i
x C
i
V V (6-5)
Demand: For capacity-based design, the demand is based on the strength capacity of
the deformation-controlled components and on the deformation demands in the structure.
The calculated reliability index therefore varies based on the earthquake-induced
deformation demands in the structure, which in turn can be related to the input ground
motion intensity. Here, the spectral acceleration at the fundamental period of a structure
is typically used to describe the ground motion intensity. The methodology establishes
demands on capacity-designed components for the MCE ground motion intensity, Sa
MCE
,
but with consideration of how the demand accumulates at lower intensities. The risk of
demands exceeding capacity at intensity lower than the MCE is controlled through the
target reliability index. The demand associated with the MCE ground motion intensity is
quantified through the median ( X
X X X X
D
D
n
m
= (6-6)
=
=
7
4
2
i
x D
i
V V (6-7)
When there is correlation between demand and capacity random variables, the total
dispersion is calculated using the following equation,
= =
= +
7
1
7
1
,
2 2
2
i j
x x j i j i D C D C
j i
V V a a V V V V (6-8)
where a
i
is +1 if x
i
is a capacity random variable and -1 if x
i
is a demand random variable
and
i,j
is the correlation coefficient between random variable x
i
and x
j
.
Target
R,Ha
: Determining a target reliability index,
R,Ha
, is somewhat subjective,
since judgment plays a factor in determining some of the factors involved. The
methodology takes into consideration the many factors that influence the reliability of
capacity-designed components and results in risk consistency between capacity-designed
components. This requires both understanding of the demand in capacity-based design
and approximations in assessing the impact of component failure on system reliability. A
common characteristic of the demand in capacity-based design is that it increases quickly
at low spectral accelerations, where the deformation-controlled components behave
elastically. However, once deformation-controlled components yield, the demand
increase is much more gradual with spectral acceleration, and demand is even
approximately constant in many cases. The result is that the probability of demand
exceeding capacity is small prior to yielding of deformation-controlled components but
then increases quickly until it saturates. The spectral acceleration at which components
are expected to begin yielding is referred to as Sa
y,exp
. The proposed methodology makes
the assumption that the probability of demand exceeding capacity versus spectral
191
acceleration can be represented by a step function that is zero at spectral accelerations
below Sa
y,exp
and a constant non-zero value beyond Sa
y,exp
. The probability of demand
exceeding capacity beyond Sa
y,exp
is based on the probability at the MCE demand. Figure
6-1 illustrates the step-function approximation.
The step-function approximation allows for the closed-form solution of the reliability
index shown in Equation 6-9 that results in a consistent mean annual frequency of
demand exceeding capacity, MAF(D>C). Sa
y,exp
can be calculated using Equation 6-10
and MAF(Sa>Sa
y,exp
) is read directly from a site ground motion hazard curve. R
is here
defined as the yield response modification factor as it relates the deformation-controlled
components elastic demand to its expected yield strength.
( )
|
|
.
|
\
|
>
>
u =
exp ,
1
,
) (
y
Ha R
Sa Sa MAF
C D MAF
| (6-9)
R
Sa
Sa
MCE
y
3
2
exp ,
= (6-10)
Equation 6-9 allows for choosing a reliability index that results in consistent mean
annual frequency of demand exceeding capacity. The challenge then becomes deciding
upon the tolerable mean annual frequency. For a consistent basis, the methodology
suggests relating the impact of component failure to the system reliability through
Equation 6-11:
( ) ( ) ( ) ( ) Sa Coll C D P C D Coll P Sa Coll P Sa Coll P Sys
C D Sys
, > > + =
>
(6-11)
where ) | ( Sa Coll P is the total probability of collapse given Sa and ) | ( Sa Coll P
Sys
is the
probability of frame collapse at a given spectral acceleration, as calculated from
incremental dynamic analysis procedures but excluding consideration of failures of
capacity-designed components. ) | ( C D Coll P
C D
>
>
is the probability of frame collapse
due to demand exceeding the capacity of capacity-designed components and
) , | ( Sa Coll C D P Sys > is the probability of demand exceeding the capacity of a
192
capacity-designed component. As a default value, ) | ( Sa Coll P
Sys
is assumed to be a
lognormal collapse fragility curve with a dispersion of 0.8, and when integrated with a
site ground motion hazard curve, the probability of collapse is 1.0% in 50 years. The
methodology then proposes that, when the total collapse fragility curve, ) | ( Sa Coll P , is
integrated with a site ground motion hazard curve, the probability of collapse does not
exceed 1.1% in 50 years, or in other words that the added probability of collapse due to
failure of capacity-designed components does not exceed 0.1% in 50 years.
To allow for a closed-form solution, the methodologys simplified method proposes
using a constant ) | ( C D Coll P
C D
>
>
at all spectral accelerations. However, based on the
SCBF studies from Chapter 4 and Chapter 5, ) | ( C D Coll P
C D
>
>
does vary with ground
motion intensity and is close to zero at low spectral accelerations. When integrated with a
site ground motion hazard curve, a small difference in the probability of collapse at low
spectral accelerations can cause a large difference in the calculated probability of collapse
in 50 years. Consequently, use of a constant ) | ( C D Coll P
C D
>
>
curve leads to
conservative results. It is therefore proposed that when a probability of collapse curve and
a site ground motion hazard curve are integrated, that the ) | ( C D Coll P
C D
>
>
curve be
allowed to be represented by a bi-linear curve as shown in Figure 6-2 that goes from zero
to a representative value (20% in the figure) at the Sa
MCE
and is then constant beyond
Sa
MCE
.
Figure 6-3 illustrates how sensitive the total probability of collapse in 50 years is to
the value of the collapse fragility curves at the low spectral accelerations. The bottom
curve is a collapse fragility curve excluding component failures while the other two
curves are based on different assumptions regarding the form of ) | ( C D Coll P
C D
>
>
curve, i.e. bi-linear (integration method) or constant (simplified method). For both curves,
) | ( C D Coll P
C D
>
>
= 20% at Sa
MCE
. At first glance, the collapse fragility curve
associated with the simplified method appears to result in lower annual probability of
collapse. However, at the low spectral accelerations it exceeds the probability of collapse
of the fragility curve associated with the integration method and the result is that the
193
collapse risk is the same for the two collapse fragility curves when integrated with a San
Francisco site (Lat = 38.0, Long = -121.7) ground motion hazard curve. As shown in
Figure 6-4, the tolerable probability of demand exceeding capacity at spectral
accelerations larger than Sa
y,exp
is 2.7 times larger for the integration method than for the
simplified method due to the introduction of the bi-linear form of ) | ( C D Coll P
C D
>
>
curve. This example is for illustration purposes only. The probability of collapse for the
default collapse fragility curve is 1.0% in 50 years and 3.0% in 50 years for the other two
collapse fragility curves, or well above the suggested maximum value of 1.1% in 50 years.
6.3 Required Design Strength of Brace Connections
To illustrate the proposed reliability framework for establishing the required design
strength of capacity-designed components, six failure modes in brace connections
subjected to tensile brace forces are used, and capacity design factors recommended for
those failure modes. The selected failure modes are: 1) Net section failure, 2) Block shear
of welded gusset plate 3) Weld failure (SMAW) 4) Weld failure (FCAW) 5) Shear bolt
failure (A325 and A490) and 6) Block shear of bolted gusset plate. Figure 6-5 shows
examples of typical brace connections, e.g. bolted channel or angle, welded hollow
rectangular section or pipe with net-section reinforcement and bolted and welded W-
shape connected through web. Figure 6-6 illustrates typical connection details for a
welded hollow rectangular section with net-section reinforcement and possible failure
modes for this detail.
According to the AISC Seismic Provisions (AISC, 2010), the required design
strength of a brace connection in SCBFs has to exceed the expected brace yield strength,
R
y
F
y
A
g
. The design equation for brace connections has the following form:
n n
D C | > (6-12)
194
where - and |-factors are the capacity design factors, currently equal to = 1.00 and | =
0.75, C
n
is the nominal strength (capacity) of the brace connection as specified by the
Seismic Provisions and D
n
is:
g y y n
A F R D = (6-13)
where R
y
is the ratio of expected yield stress to specified minimum yield stress, F
y
is the
specified minimum yield stress and A
g
the brace gross area.
From the methodologys application guidelines listed in Table 3-13 in Chapter 3, the
first steps are to collect statistical data on component demand and capacity. For the
selected failure modes, statistical data was collected from multiple experimental test
results. For capacity, the variables of interest are the material, fabrication and connection
modeling variables. The median and dispersion of the connection capacities are listed in
Table 6-1 along with the sources. Table 6-2 lists the median and dispersion of the brace
demand based on loading protocol and deformation collected from cyclic experimental
test results. The values for the record-to-record variable on the demand side are based on
observations from the SCBF analyses in Chapters 4 and 5. The only failure mode where
demand and capacity are assumed to be correlated is the Net-section failure of welded
connections. The correlation coefficient between the demand and capacity material
variables is 0.60 based on values from Liu (2003) and the correlation coefficient between
the demand and capacity fabrication variables is set to 0.70 to account for the brace and
brace connection in this example being the same member. With the collected statistical
data, Equation 6-14 is used to calculate the capacity design factors, the /|-ratios,
suggested by the methodology.
( )
D C D C Ha R
m
n
n
m
V V V V
C
C
D
D
|
|
2 exp
2 2
,
+ = (6-14)
The reliability index, |
R,Ha
, is based on R
, P(Coll
D>C
) in 50 years of 0.1% for each
failure mode, P(Coll
D>C
D>C), site ground motion hazard curve, and which method is
used. For this example, a San Francisco site (Lat = 38.0, Long = -121.7) and a New
Madrid site (Lat = 35.2, Long = -89.9) ground motion hazard curves at T = 0.2s are used.
195
Brace forces associated with 4% cyclic loading are considered to represent the demand
on connections at the MCE ground motion intensity and used in the reliability
calculations. The methodology is based on using demands associated with the MCE
demand and Table 6-2 shows that at 4% cyclic loading, the median brace forces are 8%
above the yield brace forces. These results compare well with the developed median
brace forces at the MCE demand from the SCBF dynamic analyses from Chapters 4 and
5 which ranged from 1% to 8% above the yield brace forces.
The results for the selected failure modes in bracing connections using the simplified
method are shown in Tables 6-3 and 6-4. The results are represented as the required
demand factor, , given a capacity factor, | = 0.75 with a 0.1% probability of collapse in
50 years due to connection failures. All the failure modes selected are tensile failure
modes but the same method can easily be applied to compression or flexural failure
modes. The computed factors are based on the reliability of a single connection and a
single failure mode and does not include the presence of multiple connections and failure
modes. The inclusion of multiple connections in the reliability calculations might
increase the calculated factors. However, given the observed tendency of deformation
demands to localize in a few stories in the SCBF analyses from Chapter 4 and Chapter 5,
this effect is mitigated as other stories will in turn experience lower deformation demands
and the brace connections lower force demands. Zooming in on the results when
P(Coll
D>C
|D>C) is set to 20% and R
R,Ha
s, are 2.7 and 2.2. For New Madrid (R
R,Ha
s of 2.2 and 1.6.
The results suggest that the requirements for the capacity design factors for brace
connections in SCBFs can be relaxed from the current values of 1.00. The recommended
-factors are 0.9 and 0.8 for the Western US and the Central and Eastern US respectively.
198
Table 6-1: Brace connection capacity data used in reliability analysis
Brace Connection Capacity
Material
X
1
Fabrication
X
2
Connection Model
X
3
X
m,1
*X
m,2
*X
m,3
sqrt(V
xi
2
)
Net Section Failure
Welded Connection
1
X
m,i
V
xi
1.02
0.09
1.00
0.07
1.29
0.06
1.32
0.13
Block Shear of
Welded Gusset
2
X
m,i
V
xi
1.14
0.08
1.00
0.05
1.26
0.04
1.44
0.10
Weld Failure
(SMAW)
3
X
m,i
V
xi
1.40
0.14
1.00
0.10
1.17
0.16
1.64
0.23
Weld Failure
(FCAW)
3
X
m,i
V
xi
1.66
0.14
1.00
0.10
1.17
0.17
1.94
0.24
Shear Bolt Failure
(A325)
4
X
m,i
V
xi
1.41
0.05
1.00
0.05
1.00
0.05
1.41
0.10
Shear Bolt Failure
(A490)
4
X
m,i
V
xi
1.25
0.02
1.00
0.05
1.00
0.05
1.25
0.08
Block Shear of Bolted
Gusset
5
X
m,i
V
xi
1.14
0.08
1.00
0.05
1.19
0.05
1.36
0.11
1) Material values are based on data from Liu (2003)
Fabrication values are based on recommendations from Ravindra and Galambos (1978)
Connection model values are based on test results from Yang and Mahin (2005), Willibald et
al. (2006), Fell et al. (2006), Cheng et al. (1998)
Correlation between X
1
and X
4
(
14
) is 0.60 and between X
2
and X
5
(
25
) is 0.70
2) Material values are based on data from Liu (2003)
Fabrication values are based on recommendations from Ravindra and Galambos (1978)
Connection model values are based on results from Topkaya (2006)
3) Values are based on recommendations from Fisher et al. (1978), Ng. et al. (2002), Ng, Driver
and Grondin (2004), Lesik and Kennedy (1990) and Deng (2003)
4) Values are based on recommendations from Kulak, Fisher and Struik (1987)
5) Material values are based on data from Liu (2003)
Fabrication values are based on recommendations from Ravindra and Galambos (1978
Connection model values are based on test results from Hardash and Bjorhovde (1984),
Rabinovitch and Cheng (1993), Udagawa and Yamada (1998), Nast et al. (1999), Aalberg and
Larsen (1999), Huns et al (2002) and Mullin (2005)
199
Table 6-2: Brace connection demand data used in reliability analysis
Brace Connection
Demand
1
Material
X
1
Fabrication
X
2
Load
Model
X
3
Record-
to-Record
X
7
X
m,4
*X
m,5
*X
m,6
*X
m,7
sqrt(V
xi
2
)
Brace Yield
X
m,i
V
xi
0.95
0.11
1.00
0.05
1.00
0.05
1.00
0.05
0.95
0.14
2% Cyclic
X
m,i
V
xi
-
-
-
-
1.07
0.05
1.00
0.05
1.02
0.14
4% Cyclic
X
m,i
V
xi
-
-
-
-
1.08
0.06
1.00
0.05
1.03
0.14
4% Monotonic
2
X
m,i
V
xi
-
-
-
-
1.17
0.10
1.00
0.05
1.11
0.16
1) Material values are based on data from Liu (2003)
Fabrication values are based on recommendations from Ravindra and Galambos (1978)
Cyclic values are based on cyclic brace tests by Fell et al. (2006), Yang and Mahin (2005), Black
et al. (1981), Shaback (2001), Lehman et. al (2008), Han et al. (2007) and Elchalakani et al.
(2003)
2) Monotonic load model parameter values are based on test results that show that at 4% story drift,
braces have developed about 90% of their ultimate strength
200
Table 6-3: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of 0.1% for selected
connection failure modes in SCBFs located in San Francisco. | = 0.75
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll
D>C
|D>C) P(Coll
D>C
|D>C) P(Coll
D>C
|D>C)
40% 20% 10% 40% 20% 10% 40% 20% 10%
Net Section
Failure - Welded
0.75 0.71 0.67 0.83 0.80 0.77 0.88 0.86 0.83
Block Shear of
Welded Gusset
0.73 0.69 0.64 0.82 0.79 0.75 0.89 0.86 0.82
Weld Failure
(SMAW)
0.76 0.69 0.62 0.92 0.85 0.79 1.04 0.98 0.91
Weld Failure
(FCAW)
0.65 0.59 0.52 0.79 0.73 0.67 0.89 0.84 0.78
Shear Bolt
Failure (A325)
0.73 0.69 0.64 0.82 0.79 0.75 0.89 0.85 0.82
Shear Bolt
Failure (A490)
0.82 0.77 0.72 0.92 0.88 0.84 0.98 0.95 0.91
Block Shear of
Bolted Gusset
0.78 0.73 0.68 0.88 0.84 0.79 0.95 0.91 0.87
201
Table 6-4: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of 0.1% for selected
connection failure modes in SCBFs located in New Madrid. | = 0.75
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll
D>C
|D>C) P(Coll
D>C
|D>C) P(Coll
D>C
|D>C)
40% 20% 10% 40% 20% 10% 40% 20% 10%
Net Section
Failure - Welded
0.74 0.71 0.66 0.78 0.74 0.71 0.81 0.77 0.74
Block Shear of
Welded Gusset
0.72 0.68 0.63 0.76 0.72 0.68 0.80 0.76 0.72
Weld Failure
(SMAW)
0.75 0.68 0.60 0.81 0.75 0.68 0.87 0.80 0.74
Weld Failure
(FCAW)
0.64 0.57 0.51 0.69 0.64 0.58 0.74 0.69 0.63
Shear Bolt
Failure (A325)
0.72 0.68 0.63 0.76 0.72 0.68 0.79 0.76 0.72
Shear Bolt
Failure (A490)
0.81 0.76 0.71 0.85 0.81 0.76 0.89 0.85 0.80
Block Shear of
Bolted Gusset
0.77 0.72 0.67 0.81 0.77 0.72 0.85 0.81 0.76
202
Table 6-5: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of 0.1% for selected
connection failure modes in SCBFs located in San Francisco. | = 0.75
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll
D>C
|D>C,MCE) P(Coll
D>C
|D>C,MCE) P(Coll
D>C
|D>C,MCE)
40% 20% 10% 40% 20% 10% 40% 20% 10%
Net Section
Failure - Welded
0.74 0.70 0.66 0.80 0.77 0.73 0.83 0.80 0.77
Block Shear of
Welded Gusset
0.71 0.67 0.62 0.79 0.75 0.71 0.83 0.79 0.75
Weld Failure
(SMAW)
0.73 0.66 0.59 0.85 0.79 0.72 0.93 0.87 0.80
Weld Failure
(FCAW)
0.62 0.56 0.50 0.73 0.67 0.61 0.80 0.74 0.68
Shear Bolt
Failure (A325)
0.72 0.67 0.62 0.79 0.75 0.71 0.83 0.79 0.76
Shear Bolt
Failure (A490)
0.80 0.75 0.70 0.88 0.84 0.79 0.92 0.88 0.84
Block Shear of
Bolted Gusset
0.76 0.71 0.65 0.84 0.79 0.75 0.88 0.84 0.80
203
Table 6-6: Recommended -factors based on R
and P(Coll
D>C
) in 50 years of 0.1% for selected
connection failure modes in SCBFs located in New Madrid. | = 0.75
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll
D>C
|D>C,MCE) P(Coll
D>C
|D>C,MCE) P(Coll
D>C
|D>C,MCE)
40% 20% 10% 40% 20% 10% 40% 20% 10%
Net Section
Failure - Welded
0.73 0.69 0.64 0.75 0.72 0.68 0.77 0.73 0.69
Block Shear of
Welded Gusset
0.70 0.66 0.60 0.73 0.69 0.64 0.75 0.71 0.66
Weld Failure
(SMAW)
0.72 0.64 0.56 0.76 0.70 0.62 0.79 0.72 0.65
Weld Failure
(FCAW)
0.61 0.55 0.48 0.65 0.59 0.53 0.67 0.62 0.55
Shear Bolt
Failure (A325)
0.71 0.66 0.61 0.73 0.69 0.65 0.75 0.71 0.67
Shear Bolt
Failure (A490)
0.79 0.74 0.69 0.82 0.78 0.73 0.84 0.79 0.75
Block Shear of
Bolted Gusset
0.75 0.70 0.64 0.78 0.73 0.68 0.80 0.75 0.70
204
0.00
0.01
0.02
0.03
0.0 0.5 1.0 1.5 2.0
P
(
D
>
C
)
Sa
T1
(g)
Figure 6-1: Probability of imposed demand on a component exceeding its capacity is approximated
by a step function whose height depends on the calculated probability at Sa
MCE
.
0.0
0.1
0.2
0.3
0.4
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
D
>
C
)
Sa
T1
(g)
IntegrationMethod
SimplifiedMethod
Figure 6-2: The difference between the probability of collapse given component failure between the
integration method and the simplified method
205
0.0
0.1
0.2
0.3
0.4
0.5
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
(g)
CollapseFragilityCurveExcluding
ComponentFailures
TotalCollapseFragilityCurve
IntegrationMethod
TotalCollapseFragilityCurve
SimplifiedMethod
Figure 6-3: Difference in constant risk total collapse fragility curves between the integration method
and the simplified method
0.0
0.1
0.2
0.3
0.4
0.0 0.5 1.0 1.5 2.0 2.5 3.0
P
(
D
>
C
)
Sa
T1
(g)
IntegrationMethod
SimplifiedMethod
Figure 6-4: The difference between the probability of demand exceeding capacity between the
integration method and the simplified method
206
Figure 6-5: Examples of common SCBF brace connections a) Bolted channel or angle b) Welded
hollow rectangular section or pipe with net-section reinforcement c) Bolted and welded W-shape
connected through web
Figure 6-6: Braced Frame Connection Detail
207
Chapter 7
7 Summary, Conclusions, Limitations and
Future Work
7.1 Summary and Conclusions
Capacity design provisions are a vital part of most modern seismic building codes.
For both economic reasons as well as to provide ductile response and energy dissipation
capacity, seismic resisting systems are designed for only a fraction of the earthquake
forces that can be delivered to them during large seismic events. In US structural design
codes this is achieved through ASCE 7-10s response modification factor, R, which
reduces the design seismic forces. (ASCE, 2010) As a consequence, in the event of an
earthquake, inelastic deformations can be expected. To control which components deform
inelastically, the design strengths of selected components (deformation-controlled
components), are deliberately lower than those of others (force-controlled or capacity-
designed components). To achieve this, capacity design provisions in building codes have
capacity-designed components specifically designed to be stronger than the strength
capacity of the deformation-controlled components. Given the inherent uncertainties in
both local component strengths and overall system response, capacity design factors, |
and , are used by building codes to ensure a high probability that the desired behavior is
achieved, effectively adjusting the margin between the expected capacities of the
capacity-designed components and the expected demands from the deformation-
controlled components such that the probability of demand exceeding capacity is small.
However, rigorous treatment of the uncertainties involved and understanding of the
overall system behavior and its relationship with the demand on capacity-designed
components has been generally neglected when capacity design factors have been
established. This has resulted in inconsistencies in the way they are established for
208
different seismic force resisting systems. The primary objectives of this research were to
explore which uncertainties and system behavior aspects are important for the reliability
of capacity-designed components in seismic resistant systems and to develop a systematic
way to treat them.
The reliability-based methodology developed incorporates the main factors believed
to influence the reliability of capacity-designed components and it allows for making
informed decisions on their required design strengths. The factors considered are, the
system response modification factor (R-factor), member overstrength, site ground motion
seismic hazard curve and postulated consequences of capacity-designed components
failure on system behavior. The end result is a framework for establishing the required
design strength of capacity-designed components such that the tolerable probability of
frame collapse due to component failure is consistent between different components. The
calculated |/-ratios for the 6-story and 16-story SCBFs in Chapter 5, using both the
methodologys approximate method and direct integration of the probability of demand
exceeding capacity curve and a ground motion seismic hazard curve showed that the
approximate method provided very comparable results. Despite addressing numerous
factors that influence the reliability of capacity-designed components, applying the
methodology is straightforward once necessary statistical data has been collected. Digest
of the methodology and example application for selected brace connection failure modes
were provided in Chapter 6. Detailed conclusions corresponding to the objectives
outlined in Chapter, 1 along with conclusions from the analyses of the two capacity-
designed components, are summarized below.
7.1.1 Expected Demand on Capacity-Designed Components
The expected demand on capacity-designed components in seismic resistant systems
is the demand from deformation-controlled components as they undergo inelastic
deformations. The demand therefore depends on the deformation demands in the
structure, which in return can be related to the ground motion intensity it is subjected to.
Here, the spectral acceleration at the fundamental period of a structure is used to describe
209
the ground motion intensity. Some strain hardening can be expected but it depends on
members, materials and the system involved and its impact on the reliability is often not
the most critical concern as the probability of experiencing large enough ground motions
to cause significant strain hardening decreases quickly compared to the resulting increase
in demand. The demand associated with components initial yielding is, however, critical
as even for moderate ground motions, the capacity-designed components are likely to be
subjected to that demand, and the higher the R
, or Sa
y,exp
, is not unique
for the whole structure as it is based on components over-design, which varies from one
component to the next and between stories. For systems with high R
's <2.0.
7.1.3 Seismic Hazard Curve
The influence of R
and Sa
y,exp
on the component reliability is a result of
deformation-controlled components yielding and reaching maximum capacities at ground
motion intensities with varying frequencies of occurrence. Therefore, the ground motion
seismic hazard curve is very influential in the reliability of capacity-designed components,
as it reports the frequency of occurrence of a given earthquake intensity at a given site.
R
= 2 that probability
is down to 27.8%. This large difference is due to the steepness of the seismic hazard
curve in San Francisco, i.e. the probability of experiencing the earthquake intensities
corresponding to R
= 2. For
Central and Eastern US geographic locations where the seismic hazard curves are not as
steep as the ones in the Western US, the difference is less or for the same example the
212
probabilities would be 24.8% and 10.5% if the systems are located in the New Madrid
area. These two examples also demonstrated that even for an identical system located in
different geographic locations, the difference can be significant, and suggesting that the
required design strengths of capacity-designed components can justifiably vary based on
location. How much the required design strength can vary depends heavily on the
dispersion around the median demand and capacity. The results from Section 3.2.3
demonstrated that for a total dispersion of components demand and capacity
distributions between 0.2 and 0.3, the difference in calculated required design strengths
between the two locations using the proposed methodology ranges from 10% to 25%.
The difference between different geographic locations was reduced somewhat with
the introduction of the risk-targeted MCE target, as now the spectral acceleration of the
maximum considered earthquake in the Central and Eastern US generally decreases, and
therefore its frequency increases, while the reverse is generally the case in the Western
US. All the results presented here were based on the new risk-targeted MCE target.
7.1.4 Component and System Reliability
Most methods used to analyze reliability of structural systems, e.g. FEMA P695, do
not explicitly account for the failure of capacity-designed components within the
reliability assessment. This is justified by stating that capacity-designed components are
specifically designed to be stronger than deformation-controlled components and will
thus not fail (NIST GCR 10-917-8, 2010) (which is a very deterministic thinking within a
probabilistic framework) and through the Quality Rating of Index Archetype Models in
FEMA P-695 which increases the variance of the collapse probability distribution to
account for failure modes that are not modeled in the analysis. (FEMA P695, 2009)
Fortunately, the probability of failure of capacity-designed components is generally low
so the above justification is likely to be reasonable in most cases. In addition, redundancy
within systems will also decrease the adverse consequences of components failing.
However, to determine the required design strengths of capacity-designed components, or
213
any other components, the consequences of component failure on the overall system
reliability needs to be quantified.
Non-simulated collapse modes in FEMA P695 incorporate component failures that
are difficult to simulate directly, e.g. shear and axial failure in reinforced concrete
columns, fracture in connections or hinge regions of steel moment frame components, or
failure of tie-downs in light-frame wood shear walls, through limit state checks in post-
processing of analyses results. If a limit state is exceeded, the assumption made in FEMA
P695 is that it leads to a system collapse, i.e. P(Coll
D>C
|D>C) = 1.0. This is a
conservative approach that tends to shift the collapse fragility curve to lower values.
Many of the non-simulated collapse modes checked for are related to exceeding a
specific deformation, drift or rotation limit, often related to a deterioration mechanism,
and are therefore likely to occur at somewhat high spectral accelerations. Therefore, even
if exceeding those limit states is assumed to lead directly to collapse, the overall
conservatism in the collapse risk might be reasonable compared to the reduction in the
overall computational effort by neglecting those failure modes in the dynamic analysis.
However, if the same procedure is applied to account for failure of capacity-designed
components, which are force-based limit states, the conservatism in the collapse risk
might be significant if exceeding the limit states is assumed to equal system collapse. The
SCBF analyses demonstrated this clearly. Due to braces approaching peak strength
capacities at relatively low spectral accelerations, the brace connections begin to fail at
those low spectral accelerations as well. If brace connection failures equal system
collapse, it can have an un-proportional effect on the collapse risk due to the high
frequency of occurrence of the spectral accelerations where brace connections begin to
fail. The single-story SCBF analysis in 4.3 and the 6-story SCBF analysis in 5.4, where
connection fractures were directly simulated in the analyses, demonstrated that
connection fracture does not necessarily equal collapse. Further, analyses results
demonstrated that the probability of collapse given connection fracture depends on the
ground motion intensity. At spectral accelerations close to the MCE demand, the
probability of collapse due to connection fractures was 25%-30% for the cases studied.
214
Regardless of impact on system reliability, conditional reliability calculations of
capacity-designed components need to be done with awareness that the reliability of
many capacity-designed components is fairly constant beyond the point of deformation-
controlled components yielding and reaching maximum strengths. Therefore, even if the
system does not collapse, failure of capacity-designed components can be expected, even
at low spectral accelerations.
7.1.5 Capacity-Based Design of Brace Connections in SCBFs
Many of the observations made on the nature of the reliability of capacity-designed
components were inferred from the single-story SCBF analyses where frames with
different member overstrengths, i.e. varying R
.
For systems with lower R
on the development
of maximum demands on capacity-designed components and therefore their
reliability is a central part of the methodology. It was demonstrated that as R
increases, the frequency of capacity-designed components experiencing large
demands increases and likewise does the mean annual frequency of their failure.
Ignoring the effect R
Median capacity
C
n
Nominal capacity
C
d
Deflection amplification factor of seismic force resisting systems
C
pr
Factor in the AISC Prequalified Connection Requirements to account
for peak connection strength, including strain hardening. The factor is
based on a stress demand equal to the average between F
y
and F
u
CDF Cumulative distribution function
CMR Collapse margin ratio defined in FEMA P695 as the median collapse
spectral acceleration level divided by the MCE spectral acceleration
D Demand variable
D
m
Mean demand
m
D
Median demand
0
P
D Median drift demand under ground motion of intensity P
0,Sa
. (SAC-
DCFD)
D
n
Nominal demand on a member
DBE Design basis earthquake
DR Story drift ratio
DR
NSC
Story drift ratio of a non-simulated collapse mode
DR
SC
Story drift ratio of a simulated collapse mode
F
cre
Specified minimum critical compressive stress
F
u
Specified minimum tensile stress
F
y
Specified minimum yield stress
f
y
Specified minimum yield stress of reinforcing steel
H Height of story
Ha Site ground motion hazard curve
IDA Incremental dynamic analysis
IDR Inter-story drift ratio
k Coefficient for linear regression of hazard H(Sa) on intensity Sa in
proximity of region of interest in logarithmic space (SAC-DCFD)
L
h
Distance between plastic hinge locations
M
n
Nominal flexural strength
M
nc
Nominal flexural strength of columns
M
nb
Nominal flexural strength of beams
222
M
p
Nominal plastic flexural strength
M
pc
Nominal plastic flexural strength of columns
M
pr
Maximum probable flexural strength
MAF Mean annual frequency
MCE Maximum considered earthquake
MAF(Sa) The derivative of the ground motion hazard curve at Sa
P
0
Specific value for annual probability of performance level not being
met (SAC-DCFD)
P
ysc
Axial yield strength of steel core
P(Coll
Sys
Sa) Probability of frame collapse at a given spectral acceleration
calculated through incremental dynamic analysis, including the
failure of non-simulated collapse modes
P(Coll
D>C
D>C) Probability of frame collapse due to demand exceeding the capacity
of capacity-designed components
P(Collapse)
50years
Probability of frame collapse in 50 years
Q Load random variable used in LRFD
Q
n
Nominal load in LRFD
R Resistance random variable used in LRFD
R Frame code response modification factor used in seismic design
R
n
Nominal
resistance in LRFD
R
y
Ratio of expected yield stress to specified minimum yield stress, F
y
,
as specified in the AISC Seismic Provisions
R
t
Ratio of expected tensile stress to specified minimum tensile stress,
F
u
, as specified in the AISC Seismic Provisions
R
C
Dispersion measure for drift capacity C (SAC-DCFD)
DSa
Dispersion measure for drift demand D at given Sa level (SAC-
DCFD)
R,Ha
Capacity design factor reliability index which incorporates the
frames R-factor and site ground motion hazard curve such that
different failure modes in different systems have a consistent
reliability
Load/Demand factor
| Resistance/Capacity factor
|
d
Resistance/Capacity factor for ductile limit states as specified in
AISC 358
|
n
Resistance/Capacity factor for non-ductile limit states as specified in
AISC 358
Correlation coefficient
O
0
System overstrength factor
O
y,exp
Story or base shear yield over-strength
u Standard normal cumulative distribution function
AP(Coll
D>C
)
50years
Added probability of frame collapse in 50 years due to failure of
capacity-designed components
A Story drift
A
in
Inelastic story drift
e Strain hardening adjustment factor
224
References
Aalbert, A. and Larse, P.K. (1999) Strength and Ductility of Bolted Connections in
Normal and High Strength Steels. Department of Structural Engineering,
Norwegian University of Science and Technology, Trondheim, Norway.
ACI 318 (2010) Building Code Requirements for Structural Concrete and Commentary.
American Concrete Institute, Farmington Hills, MI.
American Institute of Steel Construction Inc. (AISC). (2010a). Seismic Provisions for
Structural Steel Buildings. American Institute of Steel Construction.
American Institute of Steel Construction Inc. (AISC) (2010b) Prequalified Connections
for Special and Intermediate Steel Moment Frames for Seismic Applications,
ANSI/AISC 358-10. Draft.
American Institute of Steel Construction Inc. (AISC) (2010c) Specification for Structural
Steel Buildings.
American Society of Civil Engineers (ASCE). (2010). ASCE-7-10: Minimum Design
Loads for Buildings and Other Structures, Reston, VA.
ATC (1992). ATC-24, Guidelines for Cyclic Seismic Testing of Components of Steel
Structures, Applied Technology Council.
Baker, J.W. (2005). Vector-Valued Ground Motion Intensity Measures For Probabilistic
Seismic Demand Analysis. PhD Dissertation, Department of Civil Engineering,
Stanford University, Stanford, CA.
Baker, J.W. and C.A. Cornell (2006). Spectral shape, epsilon and record selection,
Earthquake Engineering & Structural Dynamics, Vol. 34, No. 10, pp. 1193-1217.
Bartels, P.A. (2000) Net Section Rupture in Tension Members with Connection
Eccentricity. PhD.Dissertation. Department of Civil Engineering, West Virginia
University, Morgantown, WV.
Benjamin, J. R., & Cornell, C. A. (1970). Probability, Statistics, and Decisions for Civil
Engineering: McGraw-Hill Book Co., New York.
Black, G.R., Wenger, B.A. and Popov, E.P. (1980) Inelastic Buckling of Steel Struts
Under Cyclic Load Reversals. UCB/EERC-80/40, Earthquake Engineering
Research Center, Berkeley, CA.
225
Cornell, C.A. Jalayer, F, Hamburger, R.O. Foutch, D.A. (2002), The Probabilistic Basis
for the 2000 SAC/FEMA Steel Moment Frame Guidelines. Journal of Structural
Engineering, ASCE, Vol. 128, No. 4, pp. 526-533.
Cheng, J.J. Roger, Kulak G.L. and Khoo, H-A.. (1998) Strength of slotted tubular
tension members. Canadian Journal of Civil Engineering. Vol. 25, pp. 982-991.
Choi, S.W. and Park, H.S. (2009) Evaluation of the Minimum Column-to-Beam
Strength Ratios to Prevent the Formation of Plastic Hinges on the Column Parts
Consisting of Joints in Special Steel Moment Frames International Conference
on Civil and Environmental Engineering, ICCEE-2009. Pukyong National
University.
Deng, K. (2003). The Effect of Loading Angle on the Behavior of Fillet Welds Masters
Thesis. Department of Civil and Environmental Engineering, University of
Alberta, Edmonton, Alta.
Dexter, R.J., Melendrez, M.I. (2000) Through thickness properties of columns flanges in
welded moment connections Journal of Structural Engineering. Vol. 126, No. 1,
pp. 24-31.
Dolsek, M. (2009) Incremental dynamic analysis with consideration of modeling
uncertainties Earthquake Engineering and Structural Dynamics. Vol. 38, No. 6
pp. 805 - 825
Elchalakani, M., Zhao, X.L., Grzebieta, R. (2003) Test of Cold-Formed Circular
Tubular Braces under Cyclic Axial Loading Journal of Structural Engineering.
Vol. 129. pp.507-514.
Ellingwood, B., Galambos, T.V., MacGregor, J.G., Cornell, C.A. (1980). Development
of a Probability Load Criterion for American National Standard A58. National
Bureau of Standards. Publication 577.
Ellingwood, B., MacGregor, J.G., Galambos, T.V. and Cornell, C.A. (1982) Probability
Based Load Criteria: Load Factors and Load Combinations. Journal of the
Structural Division. Vol 108, pp. 978-997.
Engelhardt M. D., Venti M. J. (2000) Behavior and Design of Radius Cut Reduced Beam
Section Connections. Report No. SAC/BD-00/17, SAC Joint Venture.
Fell, B. V. (2008). Large-Scale Testing and Simulation of Earthquake-Induced Ultra Low
Cycle Fatigue in Bracing Members Subjected to Cyclic Inelastic Buckling. Ph.D.
Dissertation, University of California at Davis, Davis, CA.
226
Fell B.V., Kanvinde A.M., Deierlein G.G., Myers A.T, Fu X. (2006) Buckling and
fracture of concentric braces under inelastic cyclic loading. SteelTips Series,
Structural Steel Education Council, Moraga, CA.
FEMA P695 (2009). Quantification of Building Seismic Performance Factors. Federal
Emergency Management Agency.
Filippou, F. C., Popov, E. P., Bertero, V. V. (1983). Effects of Bond Deterioration on
Hysteretic Behavior of Reinforced Concrete Joints. Report EERC 83-19,
Earthquake Engineering Research Center, University of California, Berkeley.
Fisher, J.W., Ravindra, M.K., Kulak, G.L., Galambos, T.V. (1978) Load and Resistance
Factor Design Criteria for Connectors Journal of Structural Division. Vol. 104,
pp. 1427-1441.
Galambos, T.V. (1990) System Reliability and Structural Design Structural Safety, Vol.
7, Issue 2-4, March 1990, pp. 101 108.
Galambos, T.V. (2004). Reliability of the Member Stability Criteria in the 2005 AISC
Specifications. International Journal of Steel Structures. Korean Society of Steel
Construction. Vol. 4, pp 223-230.
Galambos, T.V. (2009) Reliability of Steel Beam-Columns, Personal Communication
Galambos T.V., Ellingwood, B., MacGregor, J.G. and Cornell, C.A. (1982). Probability-
Based Load Criteria: Assessment of Current Practice. Journal of the Struct.
Division, ASCE 108 (ST5), pp. 959-977.
Galambos, T.V., Ravindra, M.K. (1978). Properties of Steel for Use in LRFD. Journal
of the Struct. Division. ASCE, 104(ST9), pp. 1459-1468.
Gilton C., Chi B., and Uang C. M.(2000) Cyclic Response of RBS Moment Connections:
Weak-Axis Configuration and Deep Column Effects. Report No. SAC/BD-00/23,
SAC Joint Venture.
Han, S.W., Kim, W.T. and Foutch, D.A. (2007) Seismic Behavior of HSS Bracing
Members according to Width-Thickness Ratio under Symmetric Cyclic Loading
Journal of Structural Engineering. Vol. 133, No.2, pp. 264-273.
Hardash, S.G. and Bjorhovde, R. (1984) Block Shear Behavior of Coped Steel Beams.
Structural Engineering Report No. 244. Department of Civil and Environmental
Engineering, University of Alberta, Edmonton, Alta.
Haselton, C. B. (2006) Assessing Seismic Collapse Safety of Modern Reinforced Concrete
Moment Frame Building. PhD Dissertation, Department of Civil Engineering,
Stanford University, Stanford, CA.
227
Huns, B.B., Grond, G.Y., Driver, R.G. (2002) Block Shear Behaviour of Bolted Gusset
Plates. Structural Engineering Report No. 240. Department of Civil Engineering,
University of Alberta, Edmonton, Alta.
Huns, B.B., Grondin, G.Y., Driver, R.G. (2006) Tension and shear block failure of
bolted gusset plates. Canadian Journal of Civil Engineering. Vol. 33. pp. 395-
408
Ibarra, L. (2003). Global Collapse of Frame Structures under Seismic Excitations PhD
Dissertation, Department of Civil Engineering, Stanford University, Stanford, CA.
Ibarra, L. F., and Krawinkler, H. (2005). Global collapse of frame structures under
seismic excitations. Technical Report 152, The John A. Blume Earthquake
Engineering Research Center, Department of Civil Engineering, Stanford
University, Stanford, CA.
Ibarra, L. F., Medina, R. A., and Krawinkler, H. (2005). Hysteretic models that
incorporate strength and stiffness deterioration, Earthquake Engineering and
Structural Dynamics, Vol. 34, 12, pp. 1489-1511.
Jayaram, N., Lin, T., and Baker, J. W. (2010). A computationally efficient ground-
motion selection algorithm for matching a target response spectrum mean and
variance Earthquake Spectra, (in press).
Koboevic, S. and Redwood, R. (1997). Design and seismic response of shear critical
eccentrically braced frames. Canadian Journal of Civil Engineering, Vol. 21, No.
1, pp. 761-771.
Kulak, G.L., Fisher, J.W., and Struik, J.H.A. (1987) Guide to Design Criteria for Bolted
and Riveted Joints. Second Edition. New York. John Wiley & Sons.
Kulak, G.L. Wu, E.Y. (1997) Shear lag in bolted angle tension members. Journal of
Structural Engineering. Vol. 123, No. 9. September 1997. pp.1144-1152
Lehman, D.E., Roeder, C.W., Herman, D., Johnson, S., Kotulka, B. (2008) Improved
Seismic Performance of Gusset Plate Connections Journal of Structural
Engineering. Vol. 134, pp. 840-901
Lesik, D.F. and Kennedy, D.J.L. (1990) Ultimate Strength of Fillet Welded Connections
Loaded in Plane. Canadian Journal of Civil Engineering. Vol. 17, No. 1. pp. 55-
67.
Liel, A.B. (2008). Assessing the Risk of Californias Existing Reinforced Concrete Frame
Structures: Metrics for Seismic Safety Decisions. PhD Dissertation, Department
of Civil Engineering, Stanford University, Stanford, CA.
228
Liel, Abbie B., Curt B. Haselton, Gregory G. Deierlein, and Jack W. Baker. (2009)
Incorporating modeling uncertainties in the assessment of seismic collapse risk
of buildings. Structural Safety, Vol. 31, pp 197-211.
Lignos, D.G. (2008). Sidesway Collapse of Deteriorating Structural Systems under
Seismic Excitations. PhD Dissertation, Department of Civil Engineering, Stanford
University, Stanford, CA.
Lignos, D. G., and Krawinkler, H. (2009). Sidesway Collapse of Deteriorating Structural
Systems under Seismic Excitations. Technical Report 172, The John A. Blume
Earthquake Engineering Research Center, Department of Civil Engineering,
Stanford University, Stanford, CA.
Lignos, D. G., and Krawinkler, H. (2011). Deterioration Modeling of Steel Beams and
Columns in Support to Collapse Prediction of Steel Moment Frames, ASCE,
Journal of Structural Engineering (in press).
Ling, T.W., Zhao, X.L., Al-Mahaidi, R., Packer, J.A. (2007). Investigation of Block
Shear Tear-Out Failure in Gusset-Plate Welded Connections in Structural Steel
Hollow Sections and Very High Strength Tubes. Engineering Structures. Vol. 29,
2007, p. 469-482.
Liu, J. (2003), Examination of Expected Yield and Tensile Strength Ratios, Draft
Addendum Report, AISC, Chicago, IL.
Luco, N. and Cornell, C.A. (1998) Effects of random connection fractures on demands
and reliability for a 3-story pre-Northridge SMRF structure. Proceedings of the 6
th
U.S. National Conference on Earthquake Engineering. Paper No. 244. EERI: El
Cerrito, CA. Seattle, WA, pp. 1-12
Luco, N. and Cornell, C.A. (2000) Effects of connection fractures on SMRF seismic
drift demands. ASCE Journal of Structural Engineering. Vol 126, pp. 127-136.
Luco, N., Ellingwood, B.R., Hamburger, R.O., Hooper, J.D., Kimball, J.K., Kircher, C.A.,
(2007). Risk-Targeted versus Current Seismic Design Maps for the
Conterminous United States Proc. SEAOC 2007 Annual Conference, 13 pgs.
Medina, R.A. and Krawinkler, H. (2005) Strength Demand Issues Relevant for the
Seismic Design of Moment-Resisting Frames Earthquake Spectra, Vol. 21, Issue
2, pp. 415-439.
Melchers, R.E. (1999) Structural Reliability Analysis and Prediction. Second Edition.
New York. John Wiley & Sons, Inc.
Mullin, D. (2005) Gusset Plates as Energy Dissipaters in Seismically Loaded Structures.
Ph.D. Dissertation, Department of Civil and Environmental Engineering,
University of Alberta, Edmonton, Alta.
229
Nakashima, M., and Sawaizumi, S. (2000) Column-to-beam strength ratio required for
ensuring beam-collapse mechanisms in earthquake responses of steel moment
frames Proceedings, 12th World Conference on Earthquake Engineering, New
Zealand Society for Earthquake Engineering, paper No. 1109, Auckland, New
Zealand, 8 pp
Nast, T.E., Grondin, G.Y. and Cheng, J.J.R. (1999) Cyclic Behavior of Stiffened Gusset
Plate-Brace Member Assemblies, Structural Engineering Report 229, Department
of Civil and Environmental Engineering, University of Alberta, Edmonton, Alta.
Newell, J. and Uang, C.M., (2006) Cyclic Behavior of Steel Columns with Combined
High Axial Load and Drift Demand, Report No. SSRP-06/22, Structural Systems
Research Project, Department of Structural Engineering, University of California,
San Diego, California.
NIST GCR 10-917-8 (2010) Evaluation of the FEMA P-695 Methodology for
Quantification of Building Seismic Performance Factors. National Institute of
Standards and Technology, Gaithersburg, MD 208999-8600
Ng, A.K:F., Driver, R.G. and Grondin, G.Y. (2002). Behavior of Transverse Fillet Welds.
Structural Engineering Report 245, Department of Civil and Environmental
Engineering, University of Alberta, Edmonton, Alta.
Ng, A.K:F., Driver, R.G. and Grondin, G.Y. (2004). Behavior of Transverse Fillet
Welds: Parametric and Reliability Analyses. Engineering Journal. Second
Quarter. pp 55-67.
OpenSees (2011). Open System for Earthquake Engineering Simulation. Pacific Earthquake
Engineering Research Center. University of California, Berkeley. Online at
http://opensees.berkeley.edu/
Paulay, T. (1986) A critique of the special provisions for seismic design of the building
code requirements for reinforced concrete (ACI 318-83), ACI Journal. Vol. 83,
Issue 2, pp. 274-283.
Petersen, M. D., Frankel, A. D., Harmsen, S. C., Mueller, C. S., Haller, K. M., Wheeler,
R. L., Wesson, R. L., Zeng, Y., Boyd, O. S., Perkins, D. M., Luco, N., Field, E. H.,
Wills, C. J., and Rukstales, K. S. (2008). Documentation for the 2008 Update of
the United States National Seismic Hazard Maps: U.S. Geological Survey Open-
File Report 20081128.
Pinto, P.E., Giannin, R., & Franchin, P. (2004). Seismic reliability analysis of structures.
Pavia, Italy, IUSS Press.
230
Porter, K.A., Kennedy, R., Bachman, R. (2007), Creating Fragility Functions for
Performance-Based Earthquake Engineering, Earthquake Spectra, Vol 23, Issue 2,
pp. 471-489
Porter, K.A., Beck, J.L., Shaikhutdinov, R.V. (2002) Sensitivity of building loss
estimates to major uncertain variables Earthquake Spectra. Vol. 18, Issue. 4, pp.
719-743.
Rabinovitch, J.S. and Cheng, J.J.R. (1993) Cyclic Behavior of Steel Gusset Plate
Connections. Structures Report No. 191, Department of Civil and Environmental
Engineering, University of Alberta, Edmonton, Alta.
Rajashekhar, M.R. and Ellingwood B.R. (1993) A new look at the response surface
approach for reliability analysis Structural Safety. Vol. 12, Issue 3, pp. 205-220.
Ravindra, M.K., Cornell, C.A., and Galambos, T.V. (1978), Wind and Snow Load
Factors for Use in LRFD. Journal of the Struct. Division. ASCE 104(ST9), Proc.
Paper 14006, 9/1978, pp. 1443-1457.
Ravindra, M.K., Galambos, T.V. (1978) Load and resistance factor design for steel.
Journal of the Structural Division, ASCE 1978; 104(STD9): pp. 1337-1353.
Richards, P.W. (2004) Cyclic stability and capacity design of steel eccentrically braced
frames. PhD Dissertation, Department of Civil Engineering, University of
California, San Diego, La Jolla, CA.
Richards, P.W. (2009) Seismic Column Demands in Ductile Braced Frames Journal of
Structural Engineering, ASCE 2009, Vol. 135. No. 1, pp. 33-41
Ricles, J. M., Zhang X., Lu, L,, Fisher, J. (2004) Development of Seismic Guidelines for
Deep-Column Steel Moment Connection. ATLSS Report No. 04-13. June, 2004.
Rubinstein, R.Y. (1981) Simulation and the Monte Carlo Method. New York, John Wiley
and Sons.
Sabelli, R., Mahin, S., Chang, C. (2003) Seismic demands on steel braced frame
buildings with buckling-restrained braces Engineering Structures. Vol. 25, Issue
5, pp. 655 - 666
SEAOC Seismology Committee (2008) A Brief Guide to Seismic Design Factors
Structure Magazine, September 2008, pp. 30 32.
Shaback, J.B. (2001) Behavior of Square HSS Braces with End Connections under
Reversed Cyclic Axial Loading. Masters Thesis. Department of Civil Engineering,
University of Calgary, Calgary, Alta.
231
Song Yu, Q., Gilton, C. & Uang, C.-M. (2000) Cyclic Response of RBS Moment
Connections: Loading Sequence and Lateral Bracing Effects. Report No.
SAC/BD-00/22, SAC Joint Venture.
TBI (2010). The Tall Building Initiative, Pacific Earthquake Engineering Research Center,
http://peer.berkeley.edu/tbi/
Topkaya, C. (2007). Block Shear Failure of Gusset Plates with Welded Connections.
Engineering Structures. Vol 29. pp. 11-20
Tremblay, R. and Robert, N. (2001) Seismic performance of low- and medium-rise
chevron braced steel frames Canadian Journal of Civil Engineering, Vol. 28, No.
4, pp. 699-714.
Udagawa, K. and Yamada, T. (1998) Failure Modes and Ultimate Tensile Strength of
Steel Plates Joined with High-Strength Bolts. Journal of Structural Construction
Engineering, AIJ, No. 505, pp. 115-122.
Uriz, P. (2005). Towards Earthquake Resistant Design of Concentrically Braced Steel
Structures. Ph.D. Dissertation, University of California, Berkeley.
Uriz, P. and Mahin, S.A. (2004) Seismic Performance Assessment of Concentrically
Braced Steel Frames. Proceedings of the 13
th
World Conference on Earthquake
Engineering.
Vamvatsikos, D. and Cornell, C.A. (2002) Incremental Dynamic Analysis. Earthquake
Engineering and Structural Dynamics, Vol. 31, Issue 3, pp. 491-514.
Vamvatsikos, D. and Cornell, C.A. (2004) Applied Incremental Dynamic Analysis.
Earthquake Spectra. Vol 30, Issue 2, pp. 523-553
Willibald, S., Packer, J.A and Martinez-Saucedo, G. (2006) Behavior of gusset plate
connections to ends of round and elliptical hollow structural section members.
Canadian Journal of Civil Engineering. Vol. 33, pp 373-383.
Yang, F., and Mahin, S.A. (2005) Limiting Net Section Fracture in Slotted Tube
Braces. SteelTips Series, Structural Steel Education Council, Moraga, CA.
Yang, T.Y., Moehle, J.P. (2008). Shear in Walls, unpublished presentation at the 2008
meeting of the Los Angeles Tall Building Structural Design Council, Los
Angeles.
Yun, S-Y., Hamburger, R. O., Cornell, C.A., and Foutch, D.A. (2002). Seismic
Performance for Steel Moment Frames. Journal of Structural Engineering. Vol.
128, Issue 4, pp. 534-545.
232
Zareian, F. (2006). Simplified Performance-Based Earthquake Engineering. PhD
Dissertation, Department of Civil Engineering, Stanford University, Stanford, CA.
Zareian, F., Krainkler, H. (2007) Assessment of probability of collapse and design for
collapse safety. Earthquake Engineering and Structural Dynamics, Vol. 36, Issue
13, pp. 1901 1914.
233
A. Appendix A
Incremental Dynamic Analysis of Low -
Redundancy Single-Story SCBF
This appendix contains dynamic analysis results on low-redundancy single-story
SCBFs. The objective of the analysis is to investigate the influence of connection failures
on the reliability of single story SCBFs. The analysis is similar to the one from Chapter 4
except brace fracture is not simulated in the analysis and the beam-column hinges
provide no lateral resistance. To investigate the influence of connection failures on frame
capacity, two models of single story SCBFs are created. The frames are subjected to 44
ground motions and analyzed through incremental dynamic analysis. Connection failure
is simulated through post-processing of the results.
A.1 Description of Analysis
The frame models for this analysis are idealized two-dimensional plane frame
models of single story SCBFs with pinned connections at beam-column joints. Therefore,
the braces provide all lateral resistance of the frame. Figure A-1 shows the plan and
elevation view of the frame models. The models are median models where the expected
strengths of members are used. By using median models, the failure of capacity-designed
components is by default excluded in the analysis as their median strength capacities
exceed the braces median strength capacities. Connection fractures are therefore not
simulated directly in the analysis. The gravity system is idealized as leaning columns and
provides no lateral resistance. The frames are modeled with elastic beams and columns
and the braces are fiber sections which capture global buckling, but not local buckling or
fracture. 2.0% Rayleigh damping is applied to the models. The key difference between
the two frames is that Frame 2 is very over-designed, .i.e. the braces are stronger than
234
necessary, and therefore the R
Frame 2: g
g
R
Sa
Sa
MCE
y
59 . 0
7 . 1
0 . 1
3
2
exp ,
= = =
Table A-2 and Figure A-4a show that the calculated R
increases.
238
Table A-1: Frame properties
Properties Frame 1 Frame 2
Brace Section HSS 6x6x5/16 HSS 6x6x1/2
|P
n,c
158 230
|P
n,t
266 403
L/r 88 91
F
y,exp
A
g
388 623
F
cr
A
g
196 286
V
RSA
180 180
V
y,exp
413 630
R
= V
DBE
/V
y,exp
2.5 1.7
Table A-2: Median and COV of normalized maximum brace
forces, P
max
/P
y,exp
, from analysis
Sa
T1
Frame 1 Frame 2
[g] Median COV Median COV
0.10 0.25 0.12 0.16 0.12
0.20 0.50 0.12 0.33 0.13
0.30 0.75 0.09 0.50 0.13
0.40 0.92 0.06 0.67 0.111
0.50 1.01 0.04 0.84 0.09
0.60 1.04 0.03 0.97 0.07
0.70 1.06 0.02 1.04 0.05
0.80 1.07 0.02 1.07 0.03
0.90 1.07 0.02 1.08 0.02
1.00 1.08 0.02 1.08 0.02
1.50 1.09 0.02 1.10 0.02
2.00 1.10 0.02 1.11 0.02
239
Table A-3: MAFs for Frame 1 and various connection strengths
Connections Frame
MAF
conn
P(Fail) in
50 years
P(Fail|MCE) MAF
f
P(Coll) in
50 years
MAF
f
/MAF
f,no CF
0 - - 0.00057 0.028 1.00
0.00001 0.001 0.002 0.00058 0.029 1.01
0.0001 0.005 0.015 0.00066 0.032 1.16
0.001 0.048 0.117 0.00157 0.075 2.76
0.002 0.095 0.212 0.00260 0.122 4.56
0.005 0.221 0.440 0.00554 0.242 9.74
0.01 0.393 0.704 0.01055 0.411 18.56
Table A-4: MAFs for Frame 2 and various connection strengths
Connections Frame
MAF
conn
P(Fail) in
50 years
P(Fail|MCE) MAF
f
P(Coll) in
50 years
MAF
f
/MAF
f,no CF
0 - - 0.00032 0.016 1.00
0.00001 0.001 0.002 0.00033 0.016 1.02
0.0001 0.005 0.022 0.00042 0.021 1.28
0.001 0.048 0.182 0.00128 0.062 3.94
0.002 0.095 0.331 0.00226 0.107 6.96
0.005 0.221 0.671 0.00527 0.232 16.22
0.01 0.393 0.948 0.01046 0.407 32.18
240
1-bay or 4-bays
2 Idealized Single Story Frames
Area = 180 ft x 120 ft
Weight = 100 psf x 21600
= 2160 kips
Story Height = 15 ft,
Bay Width = 30 ft
V
design
= 173 kips
Figure A-1: SCBF analyzed for this example
0
1
2
3
4
5
6
7
8
9
0.00 0.01 0.02 0.03 0.04 0.05
Story Drift
V
/
V
d
e
s
i
g
n
Frame
Tension Brace
Compression Brace
a)
0
1
2
3
4
5
6
7
8
9
0.00 0.01 0.02 0.03 0.04 0.05
Story Drift
V
/
V
d
e
s
i
g
n
Frame
Tension Brace
Compression Brace
b)
Figure A-2: Results from pushover analysis showing normalized base shear on the left y-axis,
estimated R
on the right y-axis and story drift on the x-axis. a) Frame 1 b) Frame 2
241
a) b)
c) d)
Figure A-3: Frame maximum story drift and brace axial forces vs. spectral acceleration for a) Frame
1: maximum story drift b) Frame 2: maximum story drift c) Frame 1: maximum brace axial forces d)
Frame 2: maximum brace axial forces
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0 0.5 1.0 1.5 2.0
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
[g]
Frame 1
Frame 2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 0.01 0.02 0.03 0.04
P
m
a
x
/
P
y
,
e
x
p
Story Drift
Frame 1
Frame 2
a) b)
Figure A-4: a) Median of brace maximum axial forces vs. spectral acceleration and b) Median of
brace maximum axial forces vs. story drift.
0 0.2 0.4 0.6 0.8 1 1.2 1.4
0
1
2
3
4
5
Tension Brace: Max Forces Developed
S
a
P/P
y,exp
0 0.2 0.4 0.6 0.8 1 1.2 1.4
0
1
2
3
4
5
Tension Brace: Max Forces Developed
S
a
P/P
y,exp
S
a
0 0.01 0.02 0.03 0.04 0.05 0.06
0
1
2
3
4
5
1-Story SCBF IDA Curves
S
a
Story Drift
S
a
0 0.01 0.02 0.03 0.04 0.05 0.06
0
1
2
3
4
5
1-Story SCBF IDA Curves
S
a
Story Drift
242
10
-3
10
-2
10
-1
10
0
10
1
10
-6
10
-4
10
-2
10
0
Sa
T1=1.02
(g)
A
n
n
u
a
l
R
a
t
e
o
f
E
x
c
e
e
d
a
n
c
e
Figure A-5: Site ground motion hazard curve used in this example to calculate mean annual
frequencies of collapse is a San Francisco hazard curve (Lat 38.0, Long -121.7)
0.00
0.25
0.50
0.75
1.00
0.0 1.0 2.0 3.0 4.0 5.0 6.0
P
r
o
b
a
b
i
l
i
t
y
o
f
C
o
l
l
a
p
s
e
Sa
T1
[g]
Cm = inf, Connection MAF = 0.0
Cm = 1.25Dm, Conn. MAF = 0.002,
P(Conn. Fail/MCE) = 0.21
Cm = 1.33Dm, Conn. MAF = 0.001,
P(Conn. Fail/MCE) = 0.12
Cm = 1.56Dm, Conn. MAF = 0.0001,
P(Conn. Fail/MCE) = 0.02
Figure A-6: Frame 1 collapse fragility curves with and without connection failures. 3 different
median connection strengths are used for the simulation of connection failures.
243
0
0.00001
0.00002
0.00003
0.00004
0.00005
0.00006
0.0 1.0 2.0 3.0 4.0
P
(
C
o
l
l
a
p
s
e
|
S
a
)
*
P
(
S
a
)
Sa
T1
[g]
Cm = inf, Connection MAF = 0.0
Cm = 1.56Dm, Conn. MAF = 0.0001,
P(Conn. Fail/MCE) = 0.02
Cm = 1.33Dm, Conn. MAF = 0.001,
P(Conn. Fail/MCE) = 0.12
Cm = 1.25Dm, Conn. MAF = 0.002,
P(Conn. Fail/MCE) = 0.21
Figure A-7: Frame 1: Low Sa levels contribute significantly more to a frame MAF of collapse when
connection failures are included.
0.00
0.25
0.50
0.75
1.00
0.0 1.0 2.0 3.0 4.0 5.0 6.0
P
r
o
b
a
b
i
l
i
t
y
o
f
C
o
l
l
a
p
s
e
Sa
T1
[g]
Cm = inf, Connection MAF = 0.0
Cm = 1.18Dm, Conn. MAF = 0.002,
P(Conn. Fail/MCE) = 0.33
Cm = 1.27Dm, Conn. MAF = 0.001,
P(Conn. Fail/MCE) = 0.18
Cm = 1.53Dm, Conn. MAF = 0.0001,
P(Conn. Fail/MCE) = 0.02
Figure A-8: Frame 2 collapse fragility curves with and without connection failures. 3 different
median connection strengths are used for the simulation of connection failures
244
0
0.00001
0.00002
0.00003
0.00004
0.00005
0.00006
0.00007
0.0 1.0 2.0 3.0 4.0
P
(
C
o
l
l
a
p
s
e
/
S
a
)
*
P
(
S
a
)
Sa
T1
[g]
Cm = inf, Connection MAF = 0.0
Cm = 1.53Dm, Conn. MAF = 0.0001,
P(Conn. Fail/MCE) = 0.02
Cm = 1.27Dm, Conn. MAF = 0.001,
P(Conn. Fail/MCE) = 0.18
Cm = 1.18Dm, Conn. MAF = 0.002,
P(Conn. Fail/MCE) = 0.33
Figure A-9: Frame 2: Low Sa levels contribute significantly more to a frame MAF of collapse when
connection failures are included.
0.00
0.05
0.10
0.15
0.20
0.25
0.0000 0.0005 0.0010 0.0015 0.0020
P
(
C
o
n
n
F
a
i
l
u
r
e
|
M
C
E
)
MAF of Connection Failure
Frame 1
Frame 2
Figure A-10: If brace reliability calculations are conditioned at a fixed intensity level, e.g. the MCE
spectral acceleration, the results will be inconsistent in terms of MAF of connection failures.
245
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
0.000 0.001 0.002 0.003 0.004 0.005
MAF of Connection Failure
C
o
n
n
e
c
t
i
o
n
/
|
-
r
a
t
i
o
Frame 1
Frame 2
Figure A-11: The /| ratio can be determined based on the allowable MAF of connection failure.
0.000
0.001
0.002
0.003
0.004
0.005
0.000 0.001 0.002 0.003 0.004 0.005
MAF of Frame Collapse
M
A
F
o
f
C
o
n
n
e
c
t
i
o
n
F
a
i
l
u
r
e
Frame 1
Frame 2
Figure A-12: Fixed MAF of frame collapse can justify higher MAF of connection failure for Frame 2
then for Frame 1.
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
0.000 0.002 0.004 0.006 0.008 0.010
MAF of Frame Collapse
C
o
n
n
e
c
t
i
o
n
/
|
-
r
a
t
i
o
Frame 1
Frame 2
Figure A-13: Connection /| ratio can be based on MAF of frame collapse.
246
247
B. Appendix B
Conditional Spectrum Effects on Capacity-Designed
Components
This appendix investigates the effects different ground motion selection methods
have on the demand on capacity-designed components. The IDA SCBF analysis was
performed using the ATC-63 ground motion set which is linearly scaled up for each
increment in the analysis. Linearly scaling the ground motion set results in conservative
estimates at extreme ground motions as it does not account for the unique spectral shape
of extreme ground motions and their effect on behavior. The FEMA P695 methodology
handles this issue by multiplying the median collapse point by a so-called spectral shape
factor, SSF. The SSF is based on the buildings fundamental period, the period-based
ductility and the applicable Seismic Design Category. For Frame 1 the spectral shape
factor is 1.3 based on Table 7-1b from FEMA P695 Quantification of Building Seismic
Performance Factors (FEMA, 2009), ductility index of 6 and fundamental period below
0.5s.
However, as the focus of this study was not to calculate the median collapse point,
but rather to investigate the demand on capacity-designed components and the effect of
their failure on the frame reliability, multiplying the median collapse point by a constant,
does not tell us much. To analyze if the demand on capacity-designed components is
sensitive to whether the spectral shape of extreme ground motions, a new set of ground
motions were selected. These ground motions were chosen selected to match the target
response spectrum mean and variance, i.e. the conditional spectrum or CS. The CS
ground motions were selected using an algorithm created by Jayaram et al. (2010).
Figures B-1 to B-4 provide comparison of the median earthquake response spectra of the
CS ground motion sets and the ATC ground motion set for Sa
T1
from 0.3g to 4.5g. Table
B-1 lists the spectral accelerations associated with each run specified in Figures B-1 to
248
B-4. The figures show that as the spectral acceleration increases, the difference between
the two ground motion sets increases as well.
The incremental dynamic analysis was re-run on Frame 1, now with the frame
subjected to first the ATC-63 ground motion set and then the CS ground motion sets. The
frame collapse fragility curves based on the two analyses are presented in Figure B-5. A
lognormal distribution has been fitted to both results. Using the CS ground motions
pushes the frame collapse fragility to the right, i.e. increases the frame collapse capacity,
as expected. The ratio between the median collapse points, i.e. the spectral shape factors
from the analysis is 1.39, which is reasonably close to the spectral shape factor of 1.3
suggested by FEMA P695.
Table B-2 and Figure B-6 present median and COV of the maximum braces forces at
each spectral acceleration intensity for both analyses. The results indicate that the
maximum braces forces are not sensitive to which ground motion set is used and that the
spectral acceleration at which yielding is expected does not need to be multiplied by a
spectral shape factor if linearly scaled ground motions are used. The results should not
come as big surprise. In the elastic range, the response between the two ground motion
sets should be very similar and therefore the demand on capacity-designed components as
well. Also, the median earthquake response spectra are almost identical at low spectral
accelerations, e.g. see Sa
T1
= 0.31g and 0.41g in Figure B-1, the spectral accelerations
where yielding is likely to begin. Past yielding, i.e. past Sa
y,exp
= 0.42g, the maximum
brace forces saturate quickly to their maximum value and therefore the difference
between the two analyzes is minimal at higher spectral accelerations, where the two
response spectra differ considerably as for Sa
T1
of 2.10g in Figure B-3 and higher spectral
accelerations.
249
Table B-1: Sa
T1
associated with each Run in
Figures B-1 to B-4
Run # Sa
T1
[g]
1 0.31
2 0.41
3 0.51
4 0.56
5 0.71
6 0.78
7 0.96
8 1.25
9 1.70
10 2.10
11 2.50
12 3.00
13 3.75
14 4.50
Table B-2: Median and COV of normalized maximum brace
forces, P
max
/P
y,exp
from both analyses
Sa
T1
ATC Ground
Motions
CS Ground Motions
[g] Median COV Median COV
0.00 0.00 0.00 0.00 0.00
0.31 0.78 0.09 0.80 0.12
0.41 0.98 0.05 0.96 0.07
0.51 1.04 0.02 1.03 0.03
0.56 1.04 0.02 1.04 0.02
0.71 1.06 0.02 1.05 0.01
0.78 1.07 0.02 1.06 0.01
0.96 1.07 0.02 1.07 0.02
1.25 1.09 0.02 1.08 0.02
1.70 1.10 0.02 1.10 0.02
2.10 1.11 0.02 1.11 0.02
250
0 1 2 3 4
0
0.1
0.2
0.3
0.4
0.5
Run 1
S
a
Period
CMS
ATC
0 1 2 3 4
0
0.1
0.2
0.3
0.4
0.5
Run 2
S
a
Period
CMS
ATC
0 1 2 3 4
0
0.1
0.2
0.3
0.4
0.5
Run 3
S
a
Period
CMS
ATC
CMS
ATC
0 1 2 3 4
0
0.1
0.2
0.3
0.4
0.5
Run 4
S
a
Period
Figure B-1: Comparison between CS and ATC median earthquake response spectra for Sa
T1
of
0.31g, 0.41g, 0.51g and 0.56g
0 1 2 3 4
0
0.2
0.4
0.6
0.8
1
1.2
Run 5
S
a
Period
CMS
ATC
0 1 2 3 4
0
0.2
0.4
0.6
0.8
1
1.2
Run 6
S
a
Period
CMS
ATC
0 1 2 3 4
0
0.2
0.4
0.6
0.8
1
1.2
Run 7
S
a
Period
CMS
ATC
CMS
ATC
0 1 2 3 4
0
0.2
0.4
0.6
0.8
1
1.2
Run 8
S
a
Period
Figure B-2: Comparison between CS and ATC median earthquake response spectra for Sa
T1
of
0.71g, 0.78g, 0.96g and 1.25g
251
0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
Run 9
S
a
Period
CMS
ATC
0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
Run 10
S
a
Period
CMS
ATC
0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
Run 11
S
a
Period
CMS
ATC
CMS
ATC
0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
Run 12
S
a
Period
Figure B-3: Comparison between CS and ATC median earthquake response spectra for Sa
T1
of 1.70g,
2.10g, 2.50g and 3.00g
0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Run 13
S
a
Period
CMS
ATC
CMS
ATC
0 1 2 3 4
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Run 14
S
a
Period
Figure B-4: Comparison between CS and ATC median earthquake response spectra for Sa
T1
of 3.75g
and 4.50g
252
0.0
0.2
0.4
0.6
0.8
1.0
0.0 1.0 2.0 3.0 4.0 5.0 6.0
P
(
C
o
l
l
a
p
s
e
)
Sa
T1
[g]
CMS Lognormal Fitted Distribution
(Median = 2.62, Beta = 0.50)
ATC Lognormal Fitted Distribution
(Median = 1.89, Beta = 0.45)
Figure B-5: Comparison of calculated collapse fragility curve based on CS ground motion set and the
ATC-63 ground motion set.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.25 0.50 0.75 1.00 1.25 1.50
P
m
a
x
/
P
y
,
e
x
p
Sa
T1
[g]
CMS Median Results
ATC GM Median Results
Figure B-6: Median of maximum brace forces from both analyses. Analyzing the frame with CS
ground motions does not alter the demand on capacity-designed components.
253
C. Appendix C
Component Reliability Probability Formulas
This appendix provides a detailed derivation of the probability formulas from Chapter 2
used to perform reliability calculations for capacity-designed components conditioned at a
specific deformation level or earthquake intensity level.
The design equation for capacity-designed components is generally of this form:
n n
D C | > (C-1)
where C
n
and D
n
are the nominal capacity and nominal demand, and | and are the capacity and
demand factors.
The limit state equation, which takes values less than zero when demand exceeds capacity, is
written in the following format to facilitate an analytical solution below:
1 =
D
C
g (C-2)
where C and D are the capacity and demand random variables.
At the boundary, equation C-1 can be written as:
n n
D C | = (C-3)
We can re-write equation C-3 as:
D
D
D
C
C
C
n n
| = (C-4)
|
=
n
n
D
D
C
C
D
C
(C-5)
254
Substituting (C-5) into (C-2) gives:
1 =
|
n
n
D
D
C
C
g (C-6)
We have chosen to express capacities and demands in terms of ratios to nominal values, to
facilitate standardized representation of uncertainties. We further decompose those capacity and
demand ratios as follows:
) .... (
2 1 n
n
x x x
C
C
= (C-7)
) .... (
2 1 m n n n
n
x x x
D
D
+ + +
= (C-8)
The x
i
s are the normalized random variables involved in each analysis, e.g. the material and
fabrication variability or the model uncertainty etc. These x
i
s will be defined as needed in the
following examples. We are interested in calculating the probability that the demand exceeds the
capacity, which can be evaluated using the limit state equation:
) 1 / ( ) 0 ( s = s = D C P g P P
f
(C-9)
Taking logarithms, this equation becomes:
( ) ( ) 0 / ln s = D C P P
f
(C-10)
Assuming lognormal distributions for C and D, it can be shown that the solution to (9-10) is the
following simple formula, which depends only on the random variable medians and dispersions,
Vs. (i.e. the first-order second-moment (FOSM) method reliability result is exact):
|
|
|
|
|
.
|
\
|
|
.
|
\
|
u =
+
=
+
=
m n
i
m n
j
x x j i j i
m
m
f
j i
V V a a
D
C
P
1 1
,
ln
(C-11)
( )
LRFD
| u = (C-12)
255
Where
m
C
and
m
D
are the median values of the capacities and demands, respectively, and their
ratio is:
|
=
n
m
n
m
m
m
D
D
C
C
D
C
(C-13)
Substituting from equations C-7 and C-8 we obtain
( ) |
=
+ + + m n n n
n
m
m
X X X
X X X
D
C
...
)
...
(
2 1
2 1
(C-14)
The X
s are the median values of the normalized random variables involved in each analysis and
the V
x
s their dispersionss.
i,j
is the correlation coefficient between x
i
and x
j
and a
i
is the
gradient of the limit state function with respect to x
i
. For the case of equations C-6 C-8, a
i
is +1
if x
i
is a capacity random variable and -1 if x
i
is a demand random variable.
256
257
D. Appendix D
Special Moment Frame Connections
This appendix assesses the reliability and the required design strength of bolted end
plate moment connections in Special Moment Resisting Frames (SMRFs), using cyclic
test data and the proposed Capacity Design Factor methodology. SMRFs are seismic load
resisting systems where inelastic deformations are expected to mainly take place in beam
plastic hinges (along with some column panel zone deformations). The beam plastic
hinges are the energy dissipating components in a SMRF and are expected to provide
ductility and hysteretic damping for the frame when subjected to severe seismic ground
motions. Connections in SMRFs should be able to sustain a story drift angle of at least
0.04 radians when tested using a specific loading protocol. To ensure this desired
behavior the AISC Seismic Provisions (2010a) have force-controlled components be
designed to be stronger than the expected demands from beams when undergoing
inelastic deformation. For example, connection strength should be larger than the beam
hinge maximum probable moment (AISC, 2010b), column axial strength should exceed
the nominal strength of all yielding members, including material overstrength and strain
hardening, and the summation of column plastic moments should exceed the summation
of beam plastic moments.
To illustrate the reliability framework for establishing the required design strength
capacity-designed components in seismic resistant steel buildings, two failure modes for
bolted moment-end plate connections are used. The selected failure modes are: 1)
Tension bolt failure and 2) Shear bolt failure. The required design strength of bolted
moment end-plate connections in SMRF has to be, according to the AISC Prequalified
Connections for Special and Intermediate Steel Moment Frames for Seismic Applications
(2010b) larger than the beam hinge maximum probable moment, C
pr
R
y
F
y
Z. The current
-factor is 1.15 (i.e. C
pr
= ) and the |-factor is divided into two categories, |
n
= 0.90 and
258
|
d
= 1.00, where subscript n and d refer to non-ductile and ductile failure modes. Both of
the selected failure modes are considered non-ductile according to the AISC Prequalified
Connection Requirements (2010b).
f n n
M D C = > | (D-1)
where
h u pe f
S V M M + = (D-2)
x y y pr pe
Z F R C M = (D-3)
gravity
pe
u
V
L
M
V + =
'
2
(D-4)
The following are the results for the selected failure modes in bolted end-plate
moment connections (Table D-3 and D-4). The results are represented as both the
required capacity factor, |, given a demand factor, , to achieve given probability of
demand exceeding capacity in 50 years and wise versa. Only two possible failure modes
are selected but the same method can easily be applied to other failure modes given
statistical data has been collected or through expert judgment when data is not available.
The results in tables D-3 through D-6 are based on the collected statistical data on
demand and capacity presented in tables D-1 and D-2 and the reliability framework
described in Chapter 3. For this example, a San Francisco site (Lat = 38.0, Long = -
121.7) and a New Madrid site (Lat = 35.2, Long = -89.9) ground motion hazard curves at
T = 0.2s are used and taken to represent the ground motion hazard in the Western US and
the Central and Eastern US, respectively. Demands associated with the maximum forces
developed in testing of end-plate moment connections subjected to cyclic loading
protocols are considered to represent the demand on the connections up to the MCE
ground motion demand and is used for the reliability calculations.
For R
= 4, P(Coll
D>C
) in 50 years of 0.1%, P(Coll|D>C) = 20% and a San Francisco
site ground motion hazard curve, the range of recommended -factors is 1.15 1.43. The
upper values represent a significant increase in the currently used -factor of 1.15 (C
pr
=
259
= 1.15). Based on these results, the -factor needs to be increased or the tolerable P(D>C)
= 0.50% in 50 years increased to justify the currently used - and |-factors. Alternatively,
the |-factor could be adjusted to 0.72-0.93 while maintaining the -factor at 1.15,
suggesting that the commonly used |-factor of 0.75 for connections is more appropriate
for the case of bolted moment-end plate connections.
Due to the shape of the hazard curves in the Central and Eastern US, the
recommended -factors are lower. For R
= 4, P(Coll
D>C
) in 50 years of 0.1%,
P(Coll|D>C) = 20% and a New Madrid site ground motion hazard curve, the range of -
factors is between 1.02 and 1.29. Alternatively, the |-factor could be adjusted to 0.80 -
1.01 and maintaining the -factor at 1.15.
260
Table D-1: Bolted Moment End-Plate Connection capacity
1
data used in reliability analysis
Failure Mode
Material Fabrication Connection Model X
m,1
*X
m,2
*X
m,3
X
1
X
2
X
3
V
xi
2
Tension Bolt Failure X
m,i
1.20 1.00 1.00 1.20
A325 Bolts V
xi
0.07 0.05 0.00 0.09
Tension Bolt Failure X
m,i
1.07 1.00 1.00 1.07
A490 Bolts V
xi
0.02 0.05 0.00 0.05
Shear Bolt Failure X
m,i
1.41 1.00 1.00 1.41
A325 Bolts V
xi
0.07 0.05 0.05 0.10
Shear Bolt Failure X
m,i
1.25 1.00 1.00 1.25
A490 Bolts V
xi
0.02 0.05 0.05 0.08
Table D-2: Bolted Moment End-Plate Connection demand2 data used in reliability analysis
W-Shape Beams
Gr. 50
Material Fabrication
Load Model
Parameter
Record-
to-Record
X
m,4
*X
m,5
*X
m,6
*X
m,7
X
4
X
5
X
6
X
7
V
xi
2
X
m,i
1.00 1.00 1.15 1.00 1.15
V
xi
0.05 0.05 0.11 0.03 0.14
1) Values are based on recommendations from Kulak, Fisher and Struik (1987)
2) Material values are based on data from Liu, J. (2003)
Fabrication values are based on recommendations from Ravindra and Galambos
(1978)
Load model parameter values are based on cyclic test results on bolted moment end-
plated connections reported in Sumner, Mays and Murray (2000)
Note: The current C
pr
factor is set to 1.15 and from table D-2 it does well in representing
the expected strain hardening of the beam as it undergoes cyclic loading. However, by
definition, you can expect that the maximum demand will be exceed the predicted demand
approximately 50% of the time.
261
Table D-3: Recommended |-factors for selected failure modes in bolted moment end-plate
connections in SMRF based on collected statistical data on demand and capacity, R
and P(D>C) in
50 years
= C
pr
= 1.15, Western US
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll|D>C) P(Coll|D>C) P(Coll|D>C)
100% 20% 10% 100% 20% 10% 100% 20% 10%
Tension Bolt Failure
(A325 Bolts)
0.85 0.96 1.02 0.77 0.85 0.89 0.72 0.78 0.81
Tension Bolt Failure
(A490 Bolts)
0.78 0.87 0.92 0.71 0.78 0.81 0.67 0.72 0.75
Shear Bolt Failure
(A325 Bolts)
0.98 1.11 1.19 0.88 0.97 1.02 0.82 0.90 0.94
Shear Bolt Failure
(A490 Bolts)
0.89 1.00 1.07 0.81 0.89 0.93 0.76 0.83 0.86
Table D-4: Recommended -factors for selected failure modes in bolted moment end-plate
connections in SMRF based on collected statistical data on demand and capacity, R
and P(D>C) in
50 years
| = 0.90, Western US
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll|D>C) P(Coll|D>C) P(Coll|D>C)
100% 20% 10% 100% 20% 10% 100% 20% 10%
Tension Bolt Failure
(A325 Bolts)
1.22 1.08 1.01 1.34 1.22 1.17 1.44 1.32 1.27
Tension Bolt Failure
(A490 Bolts)
1.33 1.19 1.12 1.45 1.33 1.28 1.54 1.43 1.38
Shear Bolt Failure
(A325 Bolts)
1.06 0.93 0.87 1.17 1.06 1.01 1.26 1.15 1.11
Shear Bolt Failure
(A490 Bolts)
1.16 1.03 0.97 1.27 1.16 1.11 1.36 1.25 1.21
262
Table D-5: Recommended |-factors for selected failure modes in bolted moment end-plate
connections in SMRF based on collected statistical data on demand and capacity, R
and P(D>C) in
50 years
= C
pr
= 1.15, Central & Eastern US
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll|D>C) P(Coll|D>C) P(Coll|D>C)
100% 20% 10% 100% 20% 10% 100% 20% 10%
Tension Bolt Failure
(A325 Bolts)
0.86 0.97 1.04 0.82 0.91 0.97 0.79 0.87 0.92
Tension Bolt Failure
(A490 Bolts)
0.79 0.88 0.94 0.76 0.83 0.88 0.73 0.80 0.84
Shear Bolt Failure
(A325 Bolts)
0.99 1.13 1.22 0.94 1.06 1.12 0.91 1.01 1.07
Shear Bolt Failure
(A490 Bolts)
0.90 1.02 1.09 0.86 0.96 1.02 0.84 0.92 0.97
Table D-6: Recommended -factors for selected failure modes in bolted moment end-plate
connections in SMRF based on collected statistical data on demand and capacity, R
and P(D>C) in
50 years
| = 0.90, Central & Eastern US
Failure Modes
R
= 1 R
= 2 R
= 4
P(Coll|D>C) P(Coll|D>C) P(Coll|D>C)
100% 20% 10% 100% 20% 10% 100% 20% 10%
Tension Bolt Failure
(A325 Bolts)
1.21 1.07 0.99 1.26 1.13 1.07 1.31 1.18 1.12
Tension Bolt Failure
(A490 Bolts)
1.31 1.17 1.10 1.37 1.24 1.18 1.41 1.29 1.23
Shear Bolt Failure
(A325 Bolts)
1.04 0.92 0.85 1.10 0.98 0.92 1.14 1.02 0.97
Shear Bolt Failure
(A490 Bolts)
1.14 1.02 0.95 1.20 1.08 1.02 1.24 1.12 1.07
263
Figure D-1: Reduced beam section is commonly used in SMRFs to protect beam-column connections
from excessive demands and ensure that plastic hinging forms in beams rather than columns.
(Image from AISC, 2010b)
Figure D-2: Extended end-plate configurations (Image from AISC (2010))
Figure D-3: Bolt force model for four-bolt connection (Image from AISC (2010))
264
265
E. Appendix E
Collected Statistical Data: Material Properties,
Connection Capacity in SCBF and SMRF
This appendix provides a database of collected statistical data required for
component reliability calculations. Tables E-1 to E-10 contain statistical data on material
yield stress, F
y
, tensile stress, F
u
, and calculated correlation coefficient between F
y
and F
u
for different steel shapes and material specifications. Table E-11 provides data from
brace cyclic tests. The normalized brace forces at 1% to 5% story drift ratios are reported.
Table E-12 provides 13 tests results from 4 different experiments on net section failure.
The force at which net section failure occurs is normalized by the expected force based
on AISC (2010a) limit state equations. Similarly, Table E-13 provides 11 test results
from 1 experiment on block shear failure, and Table E-14 the results from 133 tests from
8 experiments on block shear failure of bolted gusset plate. Table E-15 provides the data
used for the component reliability calculations of brace connections in Chapter 6. Table
E-16 provides data on the maximum moment developed at RBS sections versus story
drift when subjected to cyclic loading. The data was collected from 37 cyclic tests on
RBS sections from 4 different experiments.
266
Table E-1: Material properties of steel members - yield stress
Shape
Material
Specification
Source Type F
y,nom
R
y
Fy
/F
y,nom
COV
Fy
F
y
/R
y
F
y,nom
No. of
Samples
A36 Liu, J. (2003) Mill 36 1.5 1.57 0.05 56.5 1.047 56
- Lignos, D. (2008) Flange coupon - - 1.25 0.12 45.0 0.833 101
- Lignos, D. (2008) Web coupon - - 1.40 0.13 50.3 0.931 80
A572 Gr. 50 Liu, J. (2003) Mill 50 1.1 1.20 0.05 60.0 1.091 8
- Lignos, D. (2008) Flange coupon - - 1.05 0.09 52.7 0.960 190
- Lignos, D. (2008) Web coupon - - 1.10 0.11 54.8 1.004 190
- Jacques and Frank (1999) Flange coupon - - 1.09 0.07 54.4 0.989 59
- Frank and Read (1993) Flange coupon - - 1.10 0.09 54.9 0.998 13536
A992 Liu, J. (2003) Mill 50 1.1 1.10 0.05 55.0 1.000 112
- Lignos, D. (2008) Flange coupon - - 1.10 0.10 55.1 1.002 33
- Lignos, D. (2008) Web coupon - - 1.09 0.06 54.7 0.995 33
- Bartlett et al (2003) Flange coupon - - 1.10 0.06 55.0 1.000 131
- Dexter et al (2000) Flange coupon - - 1.12 0.06 55.8 1.015 20295
A36 Liu, J. (2003) Mill 36 1.5 1.34 0.07 48.2 0.893 1668
A572 Gr. 50 Liu, J. (2003) Mill 50 1.1 1.29 0.07 64.5 1.173 232
A588 Liu, J. (2003) Mill 50 1.1 1.29 0.05 64.5 1.173 75
Channel A36 Liu, J. (2003) Mill 36 1.5 1.36 0.06 49.0 0.907 22
A36 Liu, J. (2003) Mill 36 1.3 1.39 0.07 50.0 1.069 43
A572 Gr. 50 Liu, J. (2003) Mill 50 1.1 1.16 0.07 58.0 1.055 35
A529 Gr. 50 Liu, J. (2003) Mill 50 1.2 1.22 0.05 61.0 1.017 550
A529 Gr. 55 Liu, J. (2003) Mill 55 1.1 1.1 0.05 60.5 1.000 1328
A572 Gr. 55 Liu, J. (2003) Mill 55 1.1 1.13 0.08 62.2 1.027 1307
A1011 SS Gr. 55 Liu, J. (2003) Mill 55 1.1 1.12 0.06 61.6 1.018 102
A1011 HSLAS Gr. 55 Liu, J. (2003) Mill 55 1.1 1.15 0.08 63.3 1.045 301
HSS
(round) A500 Gr. B Liu, J. (2003) Mill 42 1.4 1.36 0.07 57.1 0.971 645
HSS A500 Gr. B Liu, J. (2003) Mill 46 1.4 1.31 0.08 60.3 0.936 309
Pipe A53 Gr. B Liu, J. (2003) Mill 35 1.6 1.59 0.11 55.7 0.994 228
W-shape
Angle
Plate, Bar
Table E-2: Material properties of steel members - tensile stress
Shape
Material
Specification
Source Type F
u,nom
R
t
Fu
/F
u,nom
COV
Fu
F
u
/R
t
F
u,nom
No. of
Samples
A36 Liu, J. (2003) Mill 58 1.2 1.29 0.03 74.8 1.075 56
- Lignos, D. (2008) Flange coupon - - 1.14 0.07 66.1 0.950 101
- Lignos, D. (2008) Web coupon - - 1.18 0.07 68.3 0.981 80
A572 Gr. 50 Liu, J. (2003) Mill 65 1.1 1.2 0.04 78.0 1.091 8
- Lignos, D. (2008) Flange coupon - - 1.1 0.07 71.4 0.999 190
- Lignos, D. (2008) Web coupon - - 1.12 0.08 72.9 1.020 190
- Jacques and Frank (1999) Flange coupon - - 1.11 0.04 72.3 1.011 61
- Frank and Read (1993) Flange coupon - - 1.16 0.08 75.6 1.057 13536
A992 Liu, J. (2003) Mill 65 1.1 1.12 0.04 72.8 1.018 112
- Lignos, D. (2008) Flange coupon - - 1.15 0.06 74.5 1.042 33
- Lignos, D. (2008) Web coupon - - 1.15 0.06 74.9 1.048 33
- Bartlett et al (2003) Flange coupon - - 1.1 0.05 71.6 1.001 131
- Dexter et al (2000) Flange coupon - - 1.13 0.04 73.5 1.028 20295
A36 Liu, J. (2003) Mill 58 1.2 1.22 0.04 70.8 1.017 1668
A572 Gr. 50 Liu, J. (2003) Mill 65 1.1 1.38 0.06 89.7 1.255 232
A588 Liu, J. (2003) Mill 70 1.1 1.27 0.05 88.9 1.155 75
Channel A36 Liu, J. (2003) Mill 58 1.2 1.18 0.04 68.4 0.983 22
A36 Liu, J. (2003) Mill 58 1.2 1.23 0.04 71.3 1.025 43
A572 Gr. 50 Liu, J. (2003) Mill 65 1.2 1.26 0.07 81.9 1.050 35
A529 Gr. 50 Liu, J. (2003) Mill 70 1.2 1.22 0.05 85.4 1.017 550
A529 Gr. 55 Liu, J. (2003) Mill 70 1.2 1.22 0.01 85.4 1.017 1328
A572 Gr. 55 Liu, J. (2003) Mill 70 1.2 1.15 0.01 80.5 0.958 1307
A1011 SS Gr. 55 Liu, J. (2003) Mill 70 1.1 1.08 0.01 75.6 0.982 102
A1011 HSLAS Gr. 55 Liu, J. (2003) Mill 70 1.1 1.1 0.01 77.0 1.000 301
HSS
(round) A500 Gr. B Liu, J. (2003) Mill 58 1.3 1.24 0.04 71.9 0.954 645
HSS A500 Gr. B Liu, J. (2003) Mill 58 1.3 1.27 0.04 73.7 0.977 309
Pipe A53 Gr. B Liu, J. (2003) Mill 60 1.2 1.16 0.06 69.6 0.967 228
Angle
Plate, Bar
W-shape
267
Table E-3: Combined mean and COV of F
y
for steel members
Shape Material Specification
Fy
/F
y,nom
F
y
/R
y
F
y,nom COV No. of Samples
A36 1.57 1.047 0.05 56
A572 Gr. 50 1.2 1.091 0.05 8
A992 1.1 1.000 0.05 112
A36 1.34 0.893 0.07 1668
A572 Gr. 50 1.29 1.173 0.07 232
A588 1.29 1.173 0.05 75
Channel
A36 1.36 0.907 0.06 22
A36 1.39 1.069 0.07 43
A572 Gr. 50 1.16 1.055 0.07 35
A529 Gr. 50 1.22 1.017 0.05 550
A529 Gr. 55 1.10 1.000 0.05 1328
A572 Gr. 55 1.13 1.027 0.08 1307
A1011 SS Gr. 55 1.12 1.018 0.06 102
A1011 HSLAS Gr. 55 1.15 1.045 0.08 301
HSS (round) A500 Gr. B 1.36 0.971 0.07 645
HSS
A500 Gr. B 1.31 0.936 0.08 309
Pipe
A53 Gr. B 1.59 0.994 0.11 228
x4
1.14 1.02 Sum: 3666
V
x4 0.08 0.07
Plates and bars only
All specimens
W-shape
Angle
Plate, Bar
*
Values are from report by Liu (2003) on R
y
and R
t
factors for steel members
Table E-4: Combined mean and COV of F
u
for steel members
Shape Material Specification
Fu
/F
u,nom
F
u
/R
t
F
u,nom COV No. of Samples
A36 1.29 1.08 0.03 56
A572 Gr. 50 1.2 1.09 0.04 8
A992 1.12 1.02 0.04 112
A36 1.22 1.02 0.04 1668
A572 Gr. 50 1.38 1.25 0.06 232
A588 1.27 1.15 0.05 75
Channel
A36 1.18 0.98 0.04 22
A36 1.23 1.03 0.04 43
A572 Gr. 50 1.26 1.05 0.07 35
A529 Gr. 50 1.22 1.02 0.05 550
A529 Gr. 55 1.22 1.02 0.01 1328
A572 Gr. 55 1.15 0.96 0.01 1307
A1011 SS Gr. 55 1.08 0.98 0.01 102
A1011 HSLAS Gr. 55 1.1 1.00 0.01 301
HSS (round) A500 Gr. B 1.24 0.95 0.04 645
HSS
A500 Gr. B 1.27 0.98 0.04 309
Pipe
A53 Gr. B 1.16 0.97 0.06 228
x1
1.21 1.01 Sum: 7021
V
x1 0.06 0.07
All excluding
x1
1.23 1.02 Sum: 3355
plates and bars
V
x1 0.06 0.09
Fy,Fu
W-Sections
Table E-7: Estimated correlation coefficient between F
y
and F
u
for steel members
*
Values are from report by Liu (2003) on R
y
and R
t
factors for steel members
269
Table E-8: Ratio between yield and tensile stress for steel members
Correlation Coefficients assuming:
Shape Material Specification Mean Y/T COV Lognormal Normal
A36 0.76 0.04 0.58 0.58
A572 Gr. 50 0.77 0.07 -0.19 -0.21
A992 0.76 0.04 0.61 0.61
A36 0.68 0.05 0.72 0.72
A572 Gr. 50 0.72 0.04 0.82 0.82
A588 0.73 0.03 0.81 0.81
Channel
A36 0.72 0.03 0.89 0.89
A36 0.70 0.05 0.72 0.72
A572 Gr. 50 0.71 0.08 0.35 0.34
HSS (round)
A500 Gr. B 0.80 0.09 -0.35 -0.37
HSS
A500 Gr. B 0.82 0.05 0.84 0.85
Pipe
A53 Gr. B 0.80 0.09 0.57 0.57
W-shape
Plate, Bar
Y/T - Ratio
Angle
*
Values are from report by Liu (2003) on R
y
and R
t
factors for steel members
Table E-9: Combined correlation coefficient between F
y
and F
u
for steel members
*
Values are from report by Liu (2003) on R
y
and R
t
factors for steel members
270
Table E-10: Brace test results used in reliability calculations
Authors Shape 1 2 3 4 5
Fell et al (2007) HSS4x4x1/4 FF 1.07 1.09 1.09 1.09 1.09
Fell et al (2007) HSS4x4x1/4 NF -C 1.07 1.10 1.10 1.10 1.10
Fell et al (2007) HSS4x4x1/4 FF (EQ) 1.11 1.13 1.13 1.13 1.13
Fell et al (2007) HSS4x4x3/8 FF 0.94 1.01 1.01 1.01 1.01
Fell et al (2007) HSS4x4x3/8 FF(EQ) 1.03 1.05 1.05 1.05 1.05
Fell et al (2007) Pipe3STD FF 1.05 1.07 1.08 1.09 1.10
Fell et al (2007) Pipe3STD# FF 1.01 1.06 1.06 1.08 1.08
Fell et al (2007) Pipe5STD# FF 1.13 1.17 1.18 1.20 1.20
Fell et al (2007) Pipe5STD FF 1.17 1.19 1.19 1.19 1.19
Fell et al (2007) W12x16 NF-C 1.02 1.02 1.02 1.02 1.02
Fell et al (2007) W12x16 FF 1.00 1.01 1.01 1.01 1.01
Yang and Mahin (2007) HSS 6x6x3/8 NF-C 1.09 1.09 1.09 1.09 1.09
Yang and Mahin (2007) HSS 6x6x3/8 FF 1.07 1.11 1.11 1.11 1.11
Yang and Mahin (2007) HSS 6x6x3/8 FF 1.01 1.01 1.01 1.01 1.01
Yang and Mahin (2007) HSS 6x6x3/8 NF-T 0.99 1.00 1.01 1.01 1.01
Shaback (2001) HSS5x5x5/16 1.08 1.08 1.08 1.08 1.08
Shaback (2001) HSS6x6x5/16 1.08 1.11 1.11 1.11 1.11
Shaback (2001) HSS6x6x3/8 1.12 1.14 1.14 1.14 1.14
Shaback (2001) HSS5x5x1/4 1.00 1.07 1.07 1.07 1.07
Shaback (2001) HSS5x5x5/16 1.05 1.07 1.07 1.07 1.07
Shaback (2001) HSS5x5x3/8 1.11 1.17 1.17 1.17 1.17
Shaback (2001) HSS6x6x5/16 1.09 1.09 1.09 1.09 1.09
Shaback (2001) HSS6x6x3/8 1.10 1.12 1.12 1.12 1.12
Lehman & Roeder (2008) HSS 5x5x3/8 1.07 1.07 1.07 1.07 1.07
Lehman & Roeder (2008) HSS 5x5x3/8 1.10 1.10 1.10 1.10 1.10
Lehman & Roeder (2008) HSS 5x5x3/8 1.12 1.12 1.12 1.12 1.12
Lehman & Roeder (2008) HSS 5x5x3/8 1.06 1.09 1.09 1.09 1.09
Lehman & Roeder (2008) HSS 5x5x3/8 1.28 1.28 1.28 1.28 1.28
Han & Foutch (2007) HSS4x4x1/8 1.01 1.01 1.01 1.01 1.01
Han & Foutch (2007) HSS4x4x1/4 1.02 1.02 1.02 1.02 1.02
Han & Foutch (2007) HSS4x4x1/4 1.03 1.05 1.05 1.05 1.05
Han & Foutch (2007) HSS4x4x1/4 1.03 1.05 1.05 1.05 1.05
Black and Popov (1980) W8x20 1.00 1.07 1.07 1.07 1.07
Black and Popov (1980) W6x25 1.02 1.02 1.03 1.03 1.03
Black and Popov (1980) W6x20 1.02 1.12 1.12 1.12 1.12
Black and Popov (1980) W6x20 1.05 1.09 1.09 1.09 1.09
Black and Popov (1980) W6x20 1.09 1.11 1.11 1.11 1.11
Black and Popov (1980) W6x16 0.94 0.98 0.98 0.98 0.98
Black and Popov (1980) W6x15.5 0.99 0.99 0.99 0.99 0.99
Black and Popov (1980) W6x20 1.04 1.04 1.04 1.04 1.04
Black and Popov (1980) W5x16 1.00 1.00 1.00 1.00 1.00
Black and Popov (1980) Pipe 4 Std 0.99 1.04 1.04 1.04 1.04
Black and Popov (1980) Pipe 4 Std 0.95 1.02 1.02 1.02 1.02
Black and Popov (1980) Pipe 4 X-Strong 1.05 1.10 1.10 1.10 1.10
Black and Popov (1980) Pipe 4 X-Strong 1.23 1.23 1.23 1.23 1.23
Black and Popov (1980) Pipe 3-1/2 Std 0.99 0.99 0.99 0.99 0.99
Elchalakani et al (2003) CHS 139.7x3.5 (HSS 5.563x0.134) 0.99 1.01 1.01 1.01 1.01
Elchalakani et al (2003) CHS 139.7x3.5 (HSS 5.563x0.134) 0.95 0.97 0.97 0.97 0.97
x6
: 1.05 1.07 1.07 1.08 1.08
V
x6
: 0.06 0.06 0.06 0.06 0.06
Brace Loading Paramter
Story Drift
x
6
Italic values: If tensile force at a lower story drift is higher than at the higher story drift, the higher tensile
force governs. Maximum value of failed braces is extrapolated passed its failure point until all braces have
failed.
271
Table E-11: Experimental test results for net section failure
Net Section Failure
Experiment
Loading
Protocol
A
e
L
w
X
hat U F
u
P
max
/ A
e
F
u
1 NF (T) HSS 6x6x3/8
*
(in) ASTM A500 Gr. B 6.23 15 2.25 0.85 65 (ksi) 1.200
1 FF HSS 6x6x3/8* (in) ASTM A500 Gr. B 6.23 15 2.25 0.85 65 (ksi) 1.326
2 NF (T) Pipe3STD
*
(in) A53 Gr. B 1.69 6.5 1.11 0.83 67 (ksi) 1.271
2 NF (T) Pipe5STD
*
(in) A53 Gr. B 3.45 12 1.77 0.85 61 (ksi) 1.327
1 NF (T) Pipe 6 XS
*
(in) A53 Gr. B 6.58 15 1.91 0.87 55 (ksi) 1.139
3 Monotonic HSS 168x4.8* (mm) Class C 350 MPa 1453 156 53.5 0.66 540 (MPa) 1.315
3 Monotonic HSS 168x4.8* (mm) Class C 350 MPa 1596 192 53.5 0.72 540 (MPa) 1.339
3 Monotonic HSS 110x220x6.3
*
(mm) S355J2H 1499 145 62.2 0.57 530 (MPa) 1.396
3 Monotonic HSS 110x220x6.3
*
(mm) S355J2H 1728 182 62.2 0.66 530 (MPa) 1.350
3 Monotonic HSS 168x4.8* (mm) Class C 350 MPa 1453 156 53.5 0.66 540 (MPa) 1.365
3 Monotonic HSS 110x6.3x220
*
(mm) S355J2H 2073 185 38.9 0.79 530 (MPa) 1.167
4 Monotonic HSS 102x6.4
*
(mm) Gr. 350W Class H 1451 170 32.5 0.81 449 (MPa) 1.274
4 Monotonic HSS 220x8
*
(mm) Gr. 350W Class H 3739 275 69.7 0.75 431 (MPa) 1.327
y
= 1.292
y
= 0.079
13 number of tests in COV = 0.061
4 different experiments
lny
= 0.061
lny
= 0.254
1: Yang and Mahin (2005) Y
0.5
= 1.290
2: Fell et al. (2006)
3: Willibald et al. (2006)
4: Cheng et al. (1998)
* No reinforcement at net section
FF = Far Field Loading Protocol. NF (T) = Near Field Tension Pulse. NF (C) = Near Field Compression Pulse
Section
Table E-12: Experimental test results for block shear failure
Block Shear Failure
Experiment Loading Protocol Gusset Plate L
w b A
nt
A
nv
F
y
F
u
P
AISC
P
max
P
max
/ P
AISC
1 Monotonic 500x750x4 50.0 60.3 241.2 200 309 402 171 216 1.26
1 Monotonic 500x750x4 95.3 57.5 230 381.2 309 402 234 314 1.34
1 Monotonic 500x750x4 98.3 77.5 310 393.2 309 402 270 347 1.29
1 Monotonic 500x750x4 150.3 78.6 314.4 601.2 309 402 349 430 1.23
1 Monotonic 500x750x4 49.3 98.8 395.2 197.2 309 402 232 295 1.27
1 Monotonic 500x750x4 96.0 97.8 391.2 384 309 402 300 395 1.32
1 Monotonic 500x750x4 150.0 99.9 399.6 600 309 402 383 475 1.24
1 Monotonic 500x750x4 47.3 148.8 595.2 189.2 309 402 309 386 1.25
1 Monotonic 500x750x4 96.0 148.2 592.8 384 309 402 381 467 1.23
1 Monotonic 500x750x4 101.3 56.4 225.6 405.2 309 402 241 306 1.27
1 Monotonic 500x750x4 149.3 98.2 392.8 597.2 309 402 379 433 1.14
y
= 1.26
y
= 0.05
11 number of tests in COV = V
y
0.04
1 different experiments
lny
= 0.04
lny
= 0.23
Y
0.5
= 1.26
1: Topkaya (2006)
272
Table E-13: Experimental test results for block shear failure of bolted gusset plates
Block Shear of Bolted Gusset
Authors Year No. Of Tests TEST/AISC
2005
COV
Hardash and Bjorhovde 1984 28 1.22 0.06
Rabinovitch and Cheng 1983 5 1.22 0.05
Udagawa and Yamada 1998 73 1.18 0.05
Nash et al. 1999 3 1.35 0.01
Aalberg and Larsen 1999 8 1.21 0.03
Swanson and Leon 2000 1 1.17
Huns et al. 2002 10 1.13 0.16
Mullin 2005 5 1.15 0.04
Total 133 1.19 0.07
Table E-14: Experimental test results used for SCBF reliability analysis
Failure Mode Source Mean COV No. Samples
Net Section Failure - Welded Connection Yang and Mahin (2005) 1.22 0.08 3
- Fell et al. (2006) 1.30 0.03 2
- Willibald et al. (2006) 1.32 0.06 6
- Cheng et al. (1998) 1.30 0.03 2
- Combined 1.29 0.06 13
Block Shear of Welded Gusset Topkaya (2006) 1.26 0.04 11
Block Shear of Bolted Gusset Hardash and Bjorhovde (1984) 1.22 0.06 28
Rabinovitch and Cheng (1983) 1.22 0.05 5
Udagawa and Yamada (1998) 1.18 0.05 73
Nash et al. (1999) 1.35 0.01 3
Aalberg and Larsen (1999) 1.21 0.03 8
Swanson and Leon (2000) 1.17 - 1
Huns et al. (2002) 1.13 0.16 10
Mullin (2005) 1.15 0.04 5
Combined 1.19 0.07 133
Bolt Failure Kulak, Fisher and Struik (1987) 1.35 0.05 -
Weld Failure (SMAW)
Ng et al (2004), Lesik & Kennedy
(1990), Deng (2003) 1.17 0.16 86
Weld Failure (FCAW)
Ng et al (2004), Lesik & Kennedy
(1990), Deng (2003)
1.17 0.17 54
Failure Mode Source Mean COV No. Samples
Net Section Failure - Welded Connection Ravindra & Galambos (1978) 1.00 0.05 -
Block Shear of Welded Gusset Ravindra & Galambos (1978) 1.00 0.05 -
Block Shear of Bolted Gusset Ravindra & Galambos (1978) 1.22 0.06 -
Bolt Failure Kulak, Fisher and Struik (1987) 1.00 0.05 -
Weld Failure (SMAW)
Ng et al (2004), Lesik & Kennedy
(1990), Deng (2003) 1.00 0.10 86
Weld Failure (FCAW)
Ng et al (2004), Lesik & Kennedy
(1990), Deng (2003) 1.00 0.10 54
Failure Mode Source Mean COV No. Samples
Net Section Failure - Welded Connection Liu, J. (2003) 1.02 0.09 3355
Block Shear of Welded Gusset Liu, J. (2003) 1.14 0.08 3666
Block Shear of Bolted Gusset Liu, J. (2003) 1.14 0.08 3666
Bolt Failure Kulak, Fisher and Struik (1987) 1.00 0.05 -
Weld Failure (SMAW)
Ng et al (2004), Lesik & Kennedy
(1990), Deng (2003) 1.40 0.14 86
Weld Failure (FCAW)
Ng et al (2004), Lesik & Kennedy
(1990), Deng (2003) 1.66 0.14 54
Connection Capacity Model
Connection Fabrication
Connection Material
273
Table E-15: Experimental test results on the maximum moment developed at RBS sections vs. story
drift when subjected to cyclic loading
M
max,rbs
/M
p,rbs,exp
Story
Drift
All No Concrete Slab Concrete Slab
Mean COV Mean COV Mean COV
0.001 0.12 0.26 0.11 0.26 0.13 0.27
0.002 0.22 0.20 0.20 0.20 0.23 0.25
0.003 0.31 0.18 0.28 0.18 0.34 0.24
0.004 0.41 0.18 0.37 0.18 0.45 0.24
0.005 0.50 0.18 0.46 0.18 0.55 0.23
0.006 0.60 0.18 0.54 0.18 0.66 0.23
0.007 0.69 0.18 0.63 0.18 0.75 0.22
0.008 0.77 0.17 0.71 0.17 0.84 0.21
0.009 0.85 0.17 0.78 0.17 0.92 0.21
0.010 0.90 0.18 0.84 0.18 0.96 0.20
0.015 1.05 0.17 0.98 0.17 1.11 0.20
0.020 1.11 0.16 1.05 0.16 1.17 0.20
0.030 1.15 0.14 1.08 0.14 1.23 0.19
0.040 1.16 0.15 1.09 0.15 1.23 0.18
0.050 1.18 0.13 1.10 0.13 1.25 0.17
0.060 1.18 0.13 1.10 0.13 1.26 0.16
0.070 1.18 0.13 1.10 0.13 1.27 0.15
*Values from cyclic test results reported in Engelhardt & Venti (2000), Gilton et al (2000), Ricles et al
(2004) & Song et al (2000).
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07
M
m
a
x
,
r
b
s
/
M
p
,
r
b
s
,
e
x
p
StoryDrift
All
NoConcreteSlab
ConcreteSlab
Figure E-1: Plot of the normalized mean maximum moments developed at RBS sections vs. story
drift