Professional Documents
Culture Documents
A Performance-Based Design Approach For Rectangular Concrete-Filled Steel Tube (RCFT) Frames Under Seismic Loading
A Performance-Based Design Approach For Rectangular Concrete-Filled Steel Tube (RCFT) Frames Under Seismic Loading
Authors:
INTRODUCTION
Performance of buildings under seismic loading is often decided based on damage states
observed in the structural and non-structural members. The type and extent of damage
states exhibit a wide range depending on the structural system resisting the earthquake
loading. While developing performance-based design methodologies, it is critical to
provide computational tools to assess the available capacity of structural members
corresponding to various performance-levels (e.g., Immediate Occupancy, Life Safety).
In this paper the progression of local damage states (e.g., concrete crushing, steel
yielding) of rectangular concrete-filled steel tube (RCFT) members and the effect of local
damage states on the structural performance of composite RCFT frames was investigated
using a new distributed plasticity finite element formulation. The computational findings
of the detailed structural performance, particularly the interaction of the local damage
states, were compared with experiments, as well as to damage indices of RCFT members
and frames proposed by [Tort and Hajjar, 2003]. Adopting the local damage states as the
main design parameters, a performance-based design methodology is introduced and
illustrated for the RCFT beam-columns of a three-story composite RCFT frame.
1
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
axial loading applied at a constant eccentricity. In this type of test, pinned-roller support
conditions at the member ends were usually used. The analysis results are presented for
Specimen 2 by [Cederwall et al., 1990] and Specimen SHC-4 by [Bridge, 1976], where
the latter specimen experienced biaxial bending plus axial compression, since the
member was turned about its longitudinal axis by 45º. In the monotonically-loaded non-
proportionally-loaded beam-column tests, constant axial load was introduced first, and
the specimen was then subjected to uniform bending. Pinned-roller support conditions
were imposed at the element ends. The analyses were conducted for Specimen III-3 by
[Tomii and Sakino, 1979] and Specimen BC32-80-40 by [Varma et al., 2002].
Cyclically-loaded portal frames with RCFT columns and steel girders were tested under
constant axial load and cyclically varying shear loading. The columns were subjected to
double curvature with end supports restrained against rotation. Specimen 21C30C by
[Kawaguchi, 2000] was analyzed to investigate local damage characteristics under
repeated load cycles.
Based on the analysis results and experimental observations, the typical local damage
states of RCFT members were found to include concrete cracking ( CR ),concrete
crushing ( CC ), yielding of compression flange of the steel tube ( CY ),yielding of the
tension flange of the steel tube ( TY ), local buckling of the compression flange of the
steel tube ( LBF ), and local buckling of the web of the steel tube ( LBW ). However, in
the case of axially loaded columns, yielding of the steel tube ( CY ) and local buckling of
the steel tube ( LB ) can not be differentiated between flanges and webs, due to the
uniform distribution of axial stresses. The progression of local damage of the specimens
is presented in Figures 1 through 7 for the seven specimens listed above. In each figure,
experimental and computational load vs. deformation response of the specimens are
illustrated. In addition, the occurrence of local damage states are shown on the load-
deformation plots based on three sets of data. In the first data set, the initiation of local
damage states is shown as they were documented in the experimental reports ([Tort and
Hajjar, 2004] discuss how this data was determined). The second data set was
constructed using the damage function equations by [Tort and Hajjar, 2004] to detect the
initiation of local damage states. In the third data set, the computational results from the
fiber element analysis of RCFT members by [Tort and Hajjar, 2006] were utilized. The
individual stress vs. strain response of steel and concrete fibers over the critical RCFT
cross-section along the member length was monitored during the analysis and the
initiation of local damage states were determined with respect to the stress and strain
levels attained by the material fibers. In Figures 1 through 7, numerical information
about the loading, material, and geometric properties of the specimens are also provided
including yield strength of steel tube ( f y ), compressive strength of concrete ( f c ), depth
2
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
over thickness ratio ( D / t ), length over depth ratio ( L / D ), eccentricity over depth ratio
( e / D ), and axial load over nominal cross-section strength ( P / Po ), where appropriate.
Error analyses were conducted for the fiber element analysis and damage function
data presented in Figures 1 through 7 if there experimental data existed for the initiation
of specific local damage states. The discrepancies between the local damage state
estimations and the experimental results were quantified by calculating the percentage
error of the deformation-based damage indices. The deformation-based damage index for
the experiment was calculated as the deformation value at the initiation of the local
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
damage state divided by the deformation at the peak load level from the experiment. The
deformation-based damage index for the fiber element computation was calculated as as
the deformation value at the initiation of local damage state divided by the deformation at
the peak load level from the fiber element analysis. The deformation-based damage
index from [Tort and Hajjar, 2003;2004] was calculated directly from the predictive
equations outlined in that work. The percentage errors were then calculated as the
experimental damage index value minus the computational or analytical damage index
value, divided by the experimental damage index value. This normalization by response
parameters at peak load level helped identify the error in the damage progression itself,
eliminating the other sources of errors such as assumptions in the analysis and variations
in the material and geometric properties.
Figures 1 through 7 illustrate the complex and differing progression of damage that
can occur in composite members due to the interplay between the two materials. In some
cases, such as for axially-loaded RCFT columns, the occurrence of the local damage
states often exhibits a significant effect on the load-deformation response in the form of
either strength or stiffness degradation as can be seen in Figures 1 and 2. For specimen
SC32-80, in Figure 1, the fiber analysis results and the damage assessment study by Tort
and Hajjar, 2003; 2004] shows that yielding of the steel tube takes place earlier than
concrete crushing compared to the experimental results. However, the difference
between the analysis and experimental results is not significant since the steel tube
yielding and concrete crushing were found to occur approximately simultaneously. The
damage function equation to estimate yielding of the steel tube [Tort and Hajjar, 2003;
2004] was derived from experimental data points assuming that yielding of the steel tube
initiates when the axial strain of the RCFT column is equal to the yielding strain of the
steel tube. The early detection of yielding of the steel tube in the loading history can be
attributed to inaccurate calculation of the yielding strain due for example to not
accounting directly for residual stresses in the steel tube. In all sets of data presented in
Figure 1, local buckling of the steel tube is determined to take place following the
attainment of peak axial load.
For specimen CR4-D-2 in Figure 2, since no experimental results documenting local
damage states were available, the damage states were studied based on finite element
analysis results and the damage assessment study by [Tort and Hajjar, 2003; 2004].
Similar to specimen SC32-80, the damage function equations estimates the occurrence of
the yielding of the steel tube earlier compared to the computational results, but the other
damage states correlate well between the different approaches. Concrete crushing and
local buckling of the steel tube take place subsequently when the peak axial load level is
reached. The RCFT column experiences a significant reduction in strength following
local buckling.
3
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
Table 1 illustrates the error analysis for the axially-loaded RCFT columns. Since no
experimental local damage data existed for specimen CR4-D-2, error analysis is
conducted only for specimen SC32-80. Reasonable errors were found throughout except
for yielding of the steel tube for the reasons discussed above. It was found that the fiber
element analysis produced smaller error values for all local damage states compared to
the damage function equations, which in turn are based on regression analysis from
numerous experiments [Tort and Hajjar, 2003; 2004].
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
TABLE 1
ERROR ANALYSIS FOR AXIALLY-LOADRD RCFT COLUMNS
16000.0
D/t = 34.3
14000.0 fy = 600.0 MPa
fc = 110.6 MPa
12000.0
L/D = 4
Axial Force (kN)
10000.0
Experiment
8000.0 CY - Experiment
CC - Experiment
LB - Experiment
6000.0 Analysis
CY - Analysis
4000.0 CC - Analysis
RCFT LB - Analysis
CY - Tort and Hajjar (2003)
2000.0 CC - Tort and Hajjar (2003)
LB - Tort and Hajjar (2003)
0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Axial Deformation (mm)
FIGURE 1
LOCAL DAMAGE STATES FOR RCFT COLUMN UNDER PURE COMPRESSION FOR SPECIMEN SC32-80 [VARMA,
2000]
4
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
As can be seen in Figure 3, the fiber element analysis results and the damage function
equations detect concrete cracking to be the earliest damage state for specimen 2.
Compared to the damage function equations, however, the fiber element results estimates
the initiation of concrete cracking at a load level closer to the attainment of peak axial
load, which can be attributed to the fact that the damage function equations for concrete
cracking were governed by specimens having larger e / D ratios than specimen 2 [Tort
and Hajjar, 2003].
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
4500.0
D/t = 73.7
4000.0
fy = 262.0 MPa
3500.0 f’c = 25.4 MPa
L/D = 3
3000.0
Axial Force (kN)
2500.0 Experiment
Analysis
2000.0
CY - Analysis
1500.0 CC - Analysis
LB - Analysis
1000.0
RCFT CY - Tort and Hajjar (2003)
CC - Tort and Hajjar (2003)
500.0
LB - Tort and Hajjar (2003)
0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Axial Deformation (mm)
FIGURE 2
LOCAL DAMAGE STATES FOR RCFT COLUMN UNDER PURE COMPRESSION FOR SPECIMEN CR4-D-2
[NAKAHARA AND SAKINO, 1998]
The fiber element results for specimen 2 indicates that the peak axial load is reached
following concrete crushing and yielding of the compression flange. On the other hand,
yielding of the tension flange is consistently determined to take place at large
deformation levels following the attainment of the peak axial load. Thus, the basic
progression of damge is being predicted well by the analysis and the damage indices.
Local buckling of the compression flange is the last damage level in the fiber element
results. It is important to note that, unlike for the examples shown for axially-loaded
members, detection of the damage states based on observation of the load-deformation
results alone (a common approach to damage assessment) is often insufficient for the
more complex response seen in beam-columns, as the effect of initiation of individual
local damage states on the load-deformation response is often not clear both in the
experimental and the fiber element analysis results. The only significant change seen in
Figure 3 is the reduction in stiffness and strength when the peak axial load is breached.
For Specimen SHC-4, a specimen for which limited experimental damage
observations are available, Figure 4 shows that concrete cracking takes place at an axial
load level approximately half of the peak axial load. Both damage function equations
and fiber element analysis results illustrate that a noticeable reduction in the stiffness
takes place after yielding of the compressive flange initiates. The fiber element results
also show that the degradation in stiffness accelerates following the occurrence of
5
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
concrete crushing. The peak axial load is reached when yielding of the tension flange
occurs. However, local buckling of the compression flange does not cause a noticeable
change on the behavior of the RCFT column.
The error analysis of proportionally-loaded specimens is illustrated in Table 2. Due
to the lack of experimental data, the error values are calculated only for yielding of the
compression flange and yielding of the tension flange local damage states.
800.0
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
700.0
600.0
Experiment
Axial Force (kN)
500.0 CY - Experiment
TY - Experiment
400.0 Analysis
D/t = 24 CR - Analysis
300.0
fy = 420 MPa CY - Analysis
f’c = 46.2 MPa CC - Analysis
100.0
RCFT CR - Tort and Hajjar (2003)
CY - Tort and Hajjar (2003)
TY - Tort and Hajjar (2003)
0.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
Lateral Mid-Height Displacement (mm)
FIGURE 3
LOCAL DAMAGE STATES FOR RCFT COLUMN UNDER ECCENTRIC AXIAL LOADING FOR SPECIMEN 2
[CEDERWALL ET AL., 1990]
In contrast to the axially loaded specimens, the displacement-based error values show
that damage function equations exhibit better correlation with experimental results than
fiber element analysis. In the case of specimen SHC4, yielding of compression flange is
estimated with better accuracy compared to specimen 2. However, these discrepancies
will vary some with each specimen, sometimes overestimating and sometimes
underestimating the experimental results based upon the procedures used to target mean
response when developing the damage indices in [Tort and Hajjar, 2003; 2004] and to
calibrate the constitutive models used in [Tort and Hajjar, 2006].
For non-portionally-loaded specimens, based on the fiber element results, specimen
III-3 undergoes concrete cracking early in the loading history. Concrete cracking is
followed by yielding of the compression flange and concrete crushing. The peak axial
load level is reached through yielding of the tension flange of the steel tube. This
progression of damage is consistent with all sets of data in Figure 5. The fiber element
analysis and experimental results indicates that local buckling of the steel tube initiates
soon after the peak moment value is attained. The damage function equations, on the
other hand, detected the occurrence local buckling of the compression flange after
significant inelastic deformation takes place after peak moment value is breached which
is due to the fact the effect of axial load on local buckling was underestimated.since the
majority of the empirical data was from specimens having axial load ratios less than 0.3.
Nevertheless, the load vs. deformation response does not exhibit a significant change due
6
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
to local buckling of the compression flange. This speaks to the complexity of estimating
limit states within these composite members. The deformation level at local buckling of
the steel tube web is the largest among the other local damage states.
TABLE 2
ERROR ANALYSIS OF PROPORTIONALLY-LOADED RCFT BEAM-COLUMNS
3000.0
D/t = 20 fy = 317 MPa
2500.0 fc = 39.2 MPa L/D = 10
e/D = 0.188
2000.0
Axial Load (kN)
Experiment
1500.0
CY - Experiment
Analysis
1000.0 CR - Analysis
CY - Analysis
RCFT CC - Analysis
500.0
TY - Analysis
LBF - Analysis
0.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
Mid-Height Lateral Displacement (mm)
FIGURE 4
LOCAL DAMAGE STATES OF RCFT COLUMN UNDER ECCENTRIC AXIAL LOADING FOR SPECIMEN SHC-4
[BRIDGE, 1976]
7
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
18.0
16.0
14.0
Experiment
12.0 D/t = 33 CY - Experiment
Moment (kN.m)
TY - Experiment
fy = 288 MPa LBF - Experiment
10.0 fc = 20.6 MPa LBW - Experiment
8.0
L/D = 3.0 Analysis
CR - Analysis
P/Po = 0.3
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
CY - Analysis
6.0 TY - Analysis
CC - Analysis
4.0 LBF - Analysis
LBW - Analysis
2.0 RCFT CY - Tort and Hajjar (2003)
TY - Tort and Hajjar (2003)
LBF - Tort and Hajjar (2003)
0.0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Curvature x Depth of the Steel Tube
FIGURE 5
LOCAL DAMAGE STATES FOR RCFT COLUMN UNDER CONSTANT AXIAL LOADING AND UNIFORM BENDING
MOMENT FOR SPECIMEN III-3 [TOMII AND SAKINO, 1979]
As can be seen from Table 3, the damage function equations have the largest error
values when detecting concrete crushing and local buckling of the steel tube flange. A
similar trend can also be observed for the fiber element analysis results.
TABLE 3
ERROR ANALYSIS FOR NON-PROPORTIONALLY-LOADED RCFT BEAM-COLUMNS
The progression of local damage states for cyclically loaded RCFT beam-columns is
presented in Figure 7. A similar sequence of local damage states with the non-
proportionally loaded RCFT beam-columns can be observed as in Figures 5 and 6. The
fiber element analysis results indicate that the portal frame experiences a degradation in
stiffness once concrete crushing takes place. The fiber element analysis and damage
function equations produced consistent results with respect initiation of local damage
states. For example, concrete cracking, yielding of the compression flange, and concrete
crushing occurred approximately at similar deformation levels with respect to the both
methods of estimating local damage states.
In this section, two methods were presented to identify the progression of local
damage states of RCFT members during the loading history. The damage function
8
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
equations were found to serve as an efficient method to detect local damage states. The
efficiency is mainly attributed to the ease of calculating deformation levels corresponding
to the local damage states and also to the satisfactory correlations with experimental
results. For example, the initiation of concrete cracking, yielding of the tension flange,
local buckling of the steel tube web were detected with good accuracy using the damage
function equations. The fiber element analysis provides a detailed investigation of the
local damage states of RCFT members. Using fiber element analysis, not only the
initiation but also the evolution the local damage states along the RCFT members can be
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
evaluated. For example, in Table 4, from fiber element analysis results, the damage
states of concrete and steel fibers at the critical RCFT cross-section is documented when
the peak moment attained for specimen BC32-80-40. In addition, compared to the
damage function equations, fiber element analysis can be applied to RCFT specimens
with a broader range of material and geometric properties since damage function
equations were derived from a limited set of experimental data [Tort and Hajjar, 2003;
2004]. The accuracy of the fiber element method while detecting the local damage states
was often found to be satisfactory, particularly for the limit states associated with the
steel tube.
900.0
D/t = 34 fy = 600 MPa
800.0 L/D = 4.9 f’c = 110 MPa
P/Po = 0.4
700.0
RCFT
600.0
Moment (kN.m)
500.0
FIGURE 6
LOCAL DAMAGE STATES FOR RCFT COLUMN UNDER CONSTANT AXIAL LOADING AND UNIFORM BENDING
MOMENT FOR SPECIMEN BC32-80-40 [VARMA ET AL., 2002]
9
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
350.0
Experiment TY - Tort and LBF - Tort and LBF - Analysis
Analysis Hajjar (2003) Hajjar (2003)
CC - Tort and 250.0
CR - Analysis Hajjar (2003)
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
CY - Analysis
CY - Tort and
150.0
Hajjar (2003)
CC - Analysis CR - Tort and
Hajjar (2003) CC - Analysis
TY - Analysis 50.0
TY - Analysis
LBF - Analysis
-50.0 -30.0 -10.0 10.0 30.0 50.0
-50.0
CR - Tort and Drift
Hajjar (2003)
CY - Tort and
(mm)
-150.0 CY - Analysis
Hajjar (2003)
TY - Tort and CR - Analysis
Hajjar (2003)
CC - Tort and -250.0
Hajjar (2003)
D/t = 21.6 fc = 17.9 MPa
LBF - Tort and fy = 404 MPa L/D = 8
Hajjar (2003) Shear (kN) P/Po = 0.30
-350.0
FIGURE 7
LOCAL DAMAGE STATES FOR A PORTAL FRAME WITH RCFT COLUMNS AND STEEL GIRDER FOR SPECIMEN
21C30C [KAWAGUCHI, 2000]
TABLE 4
DAMAGE DISTRIBUTION AT THE CRITICAL RCFT CROSS-SECTION FOR SPECIMEN BC32-80-40 [VARMA ET
AL., 2002]
A total 7 earthquake records were selected from the data set by [Medina and
Krawinkler, 2005]. The basic characteristics of the ground motions are documented in
Table 5 including magnitude (M), distance to the fault (R), and peak ground acceleration
(PGA). The acceleration time histories were scaled with respect to the first mode period
of the frame (0.657 seconds) at three hazard levels representing seismic loading with
2%/50year, 10%/50year, and 50%/50 probabilities of exceedence [FEMA 356, 2001].
The results of the time history analysis are presented in terms of maximum IDR (%)
for each story along with the statistical properties of mean (µ) and standard deviation (σ)
in Table 6. The damage experienced by the RCFT columns is defined through a
deformation-based damage function which utilizes the monotonic load-deformation
10
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
response under constant gravity load and linearly increasing lateral force, putting the
members into double curvature. The ratio of the maximum mean IDR to the IDR when
the member attains its peak lateral load (IDRo) on the monotonic load-deformation
response is defined as the damage ratio ( D$ ) that the RCFT beam-column is subjected to.
The calculation of the damage value ( D$ ) for the column in the first story between the
nodes A’D and A’C is documented in Figure 9. The load-deformation curve of Figure 9
was obtained by analyzing the RCFT column under constant axial load and
monotonically increasing shear loading. The rest of the columns in the first story will
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
have the same monotonic load-deformation curve if the axial load and RCFT cross-
section properties are similar. Otherwise, a separate analysis needs to be conducted for
the other RCFT columns in the story while calculating their damage ratios. The IDRo
value to normalize the maximum mean IDR was taken as the point when the stiffness of
the member reduces to 10% of its initial value. The maximum mean D$ values of the
member at each hazard level are given in Table 7. Based on the performance level scale
proposed by [Tort and Hajjar , 2004], the RCFT column may be associated with a
specific performance level at each hazard level.
Girders: A’ B’ C’ D’ E’
fy = 345 MPa m3 m4 m4 m4 m4 m4 m4 m4 m4 m4 m4 m4 m3
RCFT Columns: m1 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m1
B
fy = 317 MPa 1 1 1 1
2
f’c = 27.6 MPa 1 2 2 1
1 – HSS482x483x9.53 m1 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m1
2 – HS 559x559x15.9 C 1 1 1 1
1 2 2 2 1
m1 = 12328 kg D
m2 = 13623 kg
m3 = 8790 kg
m4 = 14639 kg
4 bays @ 9.1 m = 36.4 m
FIGURE 8
RCFT COMPOSITE FRAME FOR PERFORMANCE EVALUATION
11
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
where:
γ - demand uncertainty factor, γ a - analysis uncertainty factor, φ - resistance factor
TABLE 5
SUMMARY OF GROUND MOTION CHARACTERISTICS FOR PARAMETRIC STUDY OF THREE-STORY RCFT FRAME
TABLE 6
SUMMARY OF SUMMARY OF IDR (%) VALUES
The value of φ accounts for the uncertainty and randomness in determining the
capacity of structural members. [Tort and Hajjar, 2003] defined the capacity of RCFT
members at different performance levels based on damage function values. The φ factors
for these damage function values as capacity of RCFT members, as in Table 7, can be
quantified using Equation (2) by [Galambos and Ravindra, 1977]:
12
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
and Hajjar, 2003], an RCFT member with no sign of yielding in tension can be
considered to have mild damage. Table 8 summarizes resistance factor calculations for
initiation of yielding of the tension flange as an example. Since yielding of the tension
flange is considered here to define the boundary between Immediate Occupancy and Life
Safety performance levels, its associated data may be used to characterize the uncertainty
in performance level correlations. Rm and V R are thus determined from the
corresponding experimental data reported in [Tort and Hajjar, 2003] for RCFT beam-
column specimens tested under double curvature and constant axial compression. The
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
damage function equations by [Tort and Hajjar, 2003] were utilized for calculating Rn .
The target reliability was set to be 3.0. Rm exhibited a significant variation because the
capacity of RCFT members for the structural limit states highly depend on the geometric
and material properties of the RCFT columns. Therefore, the variation of resistance for
the available specimens was calculated after the resistance is normalized by its nominal
value. For example, the measured damage function values at tension flange yielding can
be normalized by the calculated values from damage function equations or fiber element
analysis. The resulting value of φ of 0.64 would be appropriate to round to 0.65.
800.0
700.0
600.0
Lateral Load (kN)
500.0
IDRo
400.0 max. mean IDR (2%50)
max. mean IDR (10%50)
300.0 max mean IDR (50%50)
200.0
max. mean IDR
D$ =
100.0 IDRo
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
InterStory Drift Ratio (%)
FIGURE 9
DEFORMATION-BASED DAMAGE FUCNTION CALCULATION FOR THE RCFT COLUMN
TABLE 7
PERFORMANCE LEVEL DETERMINATION FOR RCFTS
The dispersion in the capacity of the structural members has two main sources,
including randomness and uncertainty [Yun et al., 2002]. The uncertainty factors of
βRC and βUC established by [Yun et al., 2002] were defined to quantify the variation in
capacity due to randomness and uncertainty, respectively. βRC was taken as 0.20 as
13
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
recommended by [Yun et al., 2002] for steel moment frame structures; it represents the
variation due to random factors in the structural response, such as the characteristics of
the ground motion input. βUC is calculated as the standard deviation of the natural
logarithm of Rm / Rn based on the experimental data and damage function equations at
yielding of the tension flange established by [Tort and Hajjar, 2003]. βUC was determined
to be equal to 0.29, which quantifies the variation due to the factors with more
predictable nature (e.g., testing method, material properties). The square of the total
uncertainty in capacity ( βCU ) can be quantified as sum of the squares of βRC and βUC
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
( βCU 2 = βRC 2 + βUC 2 ). The variation in structural response with respect to the
acceleration time histories is accounted for through a demand uncertainty factor given in
Equation (3):
γ = ek ×β RD / 2b
2
(3)
where:
k - slope of the hazard curve (set to 3.0, for California [Yun et al., 2002]),
βRD - uncertainty factor of demand due to randomness,
b - correlation coefficient between spectral acceleration and structural demand
(1.0 as recommended by [Cornell et al., 2002])
βRD can be calculated as the standard deviation of the natural log of the max IDR of the
structural members. For the composite frame, βRD was quantified as 0.35, 0.31, and
0.38 for the 2%/50, 10%/50, and 50%/50 hazard levels, respectively.
Rm / Rn VR β φ
1.07 0.31 3 0.64
TABLE 8
RESISTANCE FACTOR CALCULATION FOR YIELDING OF THE STEEL TUBE IN TENSION
The analysis uncertainty factor given in Equation (4) represents the uncertainty
resulting from approximations and assumptions in the computational procedures (e.g.,
time history analysis, damping, live load, etc.). In Equation (4), it was assumed that the
uncertainty due to damping and other factors is insignificant compared to the uncertainty
in the nonlinear time history analysis.
γ a = ek ×β
2
/ 2b
NTH
(4)
where:
β NTH - uncertainty factor due to inaccuracies in nonlinear time history analysis
(0.15 for a 3-story structure, as per [Yun et al., 2002]).
The square of total uncertainty factor in drift demand ( βDU ) can be calculated as the
sum of squares of βRD and β NTH . Similarly, the square of total uncertainty in demand
and capacity ( βUT ) can be expressed as the sum of the squares of βDU and βCU . The
confidence level for the performance evaluation of the 1st story column of the composite
14
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
frame can be quantified using standard Gaussian tables. The confidence level (C.L.)
calculations corresponding to each hazard level are summarized in Table 9.
TABLE 9
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
ACKNOWLEDGEMENTS
This research was funded by the National Science Foundation (Grant No. CMS-0084848)
and by the University of Minnesota. Supercomputing resources were provided by the
University of Minnesota Supercomputing Institute. The authors would also like to thank
Susan LaFore and Steven Gartner for their assistance with this research. Any opinions,
findings, and conclusions or recommendations expressed in this material are those of the
authors and do not necessarily reflect the views of the National Science Foundation.
15
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers
Structural Engineering Research Frontiers © 2007 ASCE
References
[1] Bridge, R. Q. “Concrete Filled Steel Tubular Columns,” Report No. R283, School of Civil Engineering, University of Sydney,
Sydney, Australia, 1976.
[2] Cederwall, K, Engstrom, B, and Grauers, M, “High Strength Concrete Used in Composite Columns ,” 2nd Intl. Symp. on Util. of
High-Str. Conc., ACI, Detroit, MI, 1990, pp. 195-214.
[3] Cornell, C, A, Jalayer, F, Hamburger, R, O, and Douglas, A F, “Probabilistic Basis for 2000 SAC Federal Emergency
Management Agency Steel Moment Frame Guidelines,” Journal of Structural Engineering, ASCE, Vol. 128, No. 4, 2002, pp.
526-533.
Downloaded from ascelibrary.org by New York University on 03/04/16. Copyright ASCE. For personal use only; all rights reserved.
[4] Federal Emergency Management Agency, FEMA 356, “Prestandard and Commentary for the Seismic Rehabilitation of
Buildings,” Washington, DC, 2001.
[5] Galambos, T, V and Ravindra, M, K, “The Basis for Load and Resistance Factor Design Criteria of Steel Building,” Canadian
Journal of Civil Engineering, Vol. 4, No. 2, 1977, pp. 178-189.
[6] International Conference of Building Officials, “International Building Code,” International Code Council, Falls Church, VA,
2003.
[7] Kawaguchi, J, “Experimental Study on The Elastic-Plastic Behavior of Concrete-Filled Steel Tube Structures,” Internal Report,
Mie University, Tsu-shi, Japan, 2000.
[8] LaFore, S and Hajjar, J, F, “Design of Concrete-Filled Steel Tube Frames for Assessment Under Seismic Loading ,” ST-05-01,
Dept. of Civil Eng., Univ. of Minnesota, Minneapolis, MN, 2005.
[9] Medina, R, A, and Krawinkler, H, “Strength Demand Issues Relevant for the Seismic Design of Moment-Resisting Frames,”
Earthquake Spectra, Vol 21, No. 2, 2005, 415-439.
[10] Nakahara, H and Sakino, K, “Axial Compressive and Uniform Bending Tests of High Strength Concrete Filled Square Steel
Tubular Columns ,” Proceedings of the Fifth Pacific Structural Steel Conference, October 13-16, Seoul, Korea, 1998, pp. 943-
948.
[11] Nishiyama, Morino, S., Sakino, K., Nakahara, H., and Fujimoto, T. et al. “Summary of Research on Concrete-filled Structural
Steel Tube Column System Carried Out Under The US-JAPAN Cooperative Research Program on Composite and Hybrid
Structures,” BRI Research Paper No. 147, Building Research Institute, January, Tsukuba, Japan, 2002.
[12] Shakir-Khalil, H and Mouli, M, “Further Tests on Concrete-Filled Rectangular Hollow-Section Columns ,” The Structural
Engineer, Vol. 68, No. 20, 1990, pp. 405-413.
[13] Tomii, M and Sakino, K, “Experimental Studies on the Ultimate Moment of Concrete Filled Square Steel Tubular Beam-
Columns,” Transactions of the Architectural Institute of Japan, No 275, 1979, pp. 55-63.
[14] Tort, C and Hajjar, J, F, “Damage Assessment of Concrete-Filled Steel Tube Members and Frames,” Report No. ST-03-1, Dept.
of Civil Eng., Univ. of Minnesota, Minneapolis, MN, 2003.
[15] Tort, C and Hajjar, J, F, “Damage Assessment of Rectangular Concrete-Filled Steel Tubes for Performance-Based Design,”
Earthquake Spectra, Vol. 20, No. 4, 2004, pp. 1317-1348.
[16] Tort, C and Hajjar, J, F, “Seismic Demand and Capacity Evaluation of Rectangular Concrete-Filled Steel Tube (RCFT)
Members and Frames,” STESSA 2006 Behavior of Steel Structures in Seismic Areas, August 14-27, Yokohama, Japan, 2006, pp.
651-656.
[17] Varma, A, H, “Seismic Behavior, Analysis, and Design of High Strength Square Concrete Filled Steel Tube (CFT) Columns,”
Ph.D. Diss., Dept. of Civil Eng., Lehigh Univ., Bethlehem, PA, 2000.
[18]Varma, A. H., Ricles, J, M, Sause, R, Lu, L, ”Seismic behavior and modeling of high-strength composite concrete-filled steel tube
(CFT) beam-columns,” Journal of Constructional Steel Research, Vol. 58, No. 5-8, 2002, pp. 725-758.
[19] Yun, S, Hamburger, R O, Cornell, C A, and Foutch, D A, “Seismic Performance Evaluation for Steel Moment Frames,” Journal
of Structural Engineering, ASCE, Vol. 128, No. 4, 2002, pp. 534-545.
16
Copyright ASCE 2007 Structures Congress 2007
Structural Engineering Research Frontiers