Professional Documents
Culture Documents
Overstrength and Ductility Factors For Steel Frames
Overstrength and Ductility Factors For Steel Frames
Abstract: Buildings sited outside the seismic zones such as those in Singapore are not designed for seismic loads. However, the
Downloaded from ascelibrary.org by Northern Arizona University on 07/25/13. Copyright ASCE. For personal use only; all rights reserved.
buildings in these regions vibrate due to tremors caused by faults in Sumatra, Indonesia. Therefore, in order to assess the vulnerability of
these buildings under low-seismic excitation, it is important to determine the overstrength and ductility of these buildings designed for
gravity loads and wind loads or notional horizontal load. In this study, steel frames of three-, six- and ten-story three-bay moment resisting
frames 共MRFs兲, concentrically braced frames 共CBFs兲, and semirigid frames are designed according to BS 5950, which does not have any
seismic provisions. A nonlinear push-over analysis involving P-⌬ effect has been performed on these steel frames, considering the effects
of various combination of dead load 共DL兲 and imposed load 共IL兲, viz 1.0DL⫹0.4IL 共most common state兲, 1.0DL⫹1.0IL 共serviceability
state兲, and 1.2DL⫹1.2IL 共ultimate state兲. By using the N2 method, the seismic performance of the steel frames are investigated when they
are subjected to the ‘‘design earthquake’’ for Singapore. The results show that MRF, CBF, and semirigid steel frame would be subjected
to a base shear up to 4.2% of seismic gravity load, while the corresponding value for infilled frames is up to 7.8%. These values are much
less than the strength capacity of the steel frames with considerable response modification factors which are the combination of the
overstrength and ductility.
DOI: 10.1061/共ASCE兲0733-9445共2003兲129:8共1019兲
CE Database subject headings: Ductility; Seismic design; Steel frames; Singapore.
1
Associate Professor, Dept. of Civil Engineering, National Univ. of
Singapore, NUS, 10 King Ridge Crescent, Singapore 119260, Singapore.
E-mail: cvebalen@nus.edu.sg
2
Graduate Student, Dept. of Civil Engineering, National Univ. of
Singapore, NUS, 10 King Ridge Crescent, Singapore 119260, Singapore.
Note. Associate Editor: Marc I. Hoit. Discussion open until January 1,
2004. Separate discussions must be submitted for individual papers. To
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted
for review and possible publication on December 19, 2001; approved on
May 23, 2002. This paper is part of the Journal of Structural Engineer-
ing, Vol. 129, No. 8, August 1, 2003. ©ASCE, ISSN 0733-9445/2003/8-
Fig. 1. Frame overstrength factor and ductility factor
1019–1035/$18.00.
the overstrength due to most of the above factors, however, the strength and ductility of the structure, as illustrated in Fig. 1 共Bal-
overstrength due to redistribution of internal forces in the inelastic endra et al. 1999兲. The design base shear V d is defined as
range, arising from simplification in design procedure, is depend- Ve
able and can be estimated reliably. V d⫽ (1)
⍀
A review of the seismic codes adopted in several countries
reveals that a response modification factor 共or the seismic force where V e ⫽elastic base shear; ⍀⫽overstrength factor; and
reduction factor兲 is being used to reduce the elastic strength de- ⫽ductility factor defined as
mand due to the ultimate earthquake or worst credible event. In ⌬u
these seismic codes, the overstrength and ductility are incorpo- ⫽ (2)
rated through a response modification factor. The widely accepted ⌬y
limit state philosophy for seismic design considers at least two where ⌬ u ⫽ultimate displacement; and ⌬ y ⫽displacement at sig-
limit states associated with different levels of earthquake excita- nificant yield. Thus, the response modification factor
tion: The serviceability limit state for moderate earthquake shak-
R⫽⍀⫽overstrength factor⫻ductility factor (3)
ing and the ultimate limit state for severe earthquake shaking. The
modification factor is used to reduce the elastic base shear in- Osteraas and Krawinkler 共1990兲 studied structural over-
duced in a structure under the worst credible event, which is once strength of steel framing systems: distributed moment frames, pe-
in 475 years. The amount of reduction depends on the over- rimeter moment frames, and concentric braced frames designed in
compliance with the Uniform Building Code working-stress- The purpose of this study is to determine the overstrength and
design provisions. They reported overstrength factors ranging ductility of steel frames designed according to BS 5950. The
from 1.8 to 6.5 for the three framing systems. Also they found frames considered in this investigation are three-, six- and ten-
that the perimeter moment resisting frames have a smaller struc- story three-bay MRFs, CBFs, and semirigid frames. The latter has
tural overstrength than do moment resisting space frames, be- not received much attention by researchers. The contribution of
cause the gravity loads do not substantially influence the design infill wall to overstrength of frame has also been included.
of perimeter moment resisting frames. Uang 共1991兲 reported
overstrength of 2 to 3 for 4- to 12-story special steel moment
resisting frames, and for steel frames located in high-seismic re-
Modeling of Steel Frames
gion, overstrength of a four-story steel frame is about 40% higher
than that of a 12-story steel frame. Rahgozar and Humar 共1998兲
obtained the overstrength factor ranging from 1.5 to 3.5 for dif- In this study, the finite element system ABAQUS 共Hibbitt, Karls-
ferent types of ten-story braced frame. son and Sorensen, Inc., Pawtucket, R.I., 1996兲 has been used to
Also, Osman et al. 共1995兲 did a nonlinear push-over analysis model the various steel frames. Each beam and column member is
on six-story moment resisting frames 共MRFs兲 designed according discretized into several three-node beam elements. The cross sec-
to the National Building Code of Canada 共NBCC兲 1990. The tion of each element is subdivided into fibers, as illustrated in Fig.
predicted ductility factors are around 1.5 to 2.7. Corresponding to 2. There are 21 strips in the web and 15 strips in the flange of I
the ductility factor defined in this study, the National Earth- section. The finite elements used to model column and beam
quake Hazards Reduction Program 共NEHRP 1994兲 gives the same members are based on Timoshenko beam theory.
definition to a displacement amplification factor C d (C d In ABAQUS, the beam element is a one-dimensional line ele-
⫽⌬ u /⌬ y ). Values of C d for different steel framing systems are: ment in three-dimensional space that has stiffness associated with
5.5 for special MRF, 4 for ordinary MRF, and 4.5 for concentri- deformation of the axis of beam. The deformations consist of
cally braced frame 共CBF兲. axial stretch, curvature change 共bending兲, and torsion. The active
⫽ 冑
4 E i t sin共 2 兲
4E f I c h ⬘
(12)
冉 冊 冉 冊
G gi ⫽total of gravity load located at storey i; h x ⫽height above l⫽0.3997⫺ ⫺ (15)
E E
the base of the structure to story x; h i ⫽height above the base of 480 5.695
the structure to story i; n⫽total number of stories in structure;
and V b ⫽total horizontal earthquake base force. Three different where
combinations of dead and imposed vertical load are defined, that
is, 1.0DL⫹0.4IL 共most common state兲, 1.0DL⫹1.0IL 共service- ␣f⫽
b
tf
, ␣ w⫽
h
t w
, and ⫽ 冑 300
y
ability state兲 and 1.2DL⫹1.2IL 共ultimate state兲.
b⫽half flange width; h⫽web depth; t f ⫽flange thickness; and
t w ⫽web thickness.
Failure Modes
Collapse Mechanism
The static push-over analysis is used in which the gravity loads When adequate number of plastic hinges form to cause a sway
are held constant while the magnitude of earthquake forces of mechanism.
assumed distribution pattern along the height of the building, are
gradually increased until the failure mode occurred. The failure Shear Capacity
mode is taken as the one of the following which ever occurs first When shear force in beam or column exceed the shear capacity
when P-⌬ effect is included. evaluated as
Fig. 9. Base shear-top floor lateral displacement curve under various vertical load combinations for MRFs and CBFs.
Fig. 10. Sequence of developing hinge and local buckling location of ten-story frames: 共a兲 MRF and 共b兲 split-X-braced frame
tical load condition, as a ratio of notional horizontal load is 8.00 combination, the ductility factor for three-, six- and ten-story
and 4.66, respectively. For the ten-story frame, the base shear MRFs are 1.95, 1.94, and 1.67, respectively.
capacity under common vertical load condition is about 3.8 times
the ultimate wind load 共1.2WL兲. This is equivalent to 8.5% of
seismic gravity load W g , taken to be 100% of dead load. The Concentrically Braced Frame
corresponding values for three- and six-story frames are 11.2 and
The base shear versus top displacement curves of three-, six- and
6.5%, respectively. Since P-⌬ effects are greater for the service-
ten-story CBFs are depicted in Fig. 9 including inverted-V-braced
ability state and ultimate state, the ratio of base shear to seismic
gravity load declines to 5.5, 4.2, and 4.4% for three-, six- and frame and split-X-braced frame. It is observed that the failures of
ten-story MRFs, respectively, under ultimate vertical load condi- all these CBFs are caused by the formation of a mechanism in the
tion. One reason for a large overstrength factor is because the first story of these CBFs. As for the inverted-V-braced frame, the
choices of members were based on ultimate limit state design compression brace elastically buckles early because of initial
philosophy with the partial safety factor for loads, and the re- compression stress and large slenderness ratio which is around
quirement of stability check for sway frames. Because the loading 160. Then, as soon as the tension brace yields, the collapse
combination governing the design of three- and six-story frames mechanism forms in the first story of the frame. In contrast, for
共COMB 3兲 is different from that of ten-story frames 共COMB 2兲, the split-X-braced frame, the tension braces of a lower story such
there is no monotonic trend in the base shear capacity. as the second and the fourth story might yield earlier than the first
The ductile performance of MRF depends on the rotation ca- story because these braces are pulled initially by gravity load.
pacity of the plastic hinges at the beam and column ends. The And since the braces in the second story carry a part of the gravity
potential locations of the plastic hinges for ten-story have been load imposed on the first floor, the initial compression stress is
depicted in Fig. 10. For three- and six-story frames, the rotation small for the first-story brace 共Fig. 10兲. Thus, the collapse mecha-
capacity exceeded in the end bay beams of the second floor; for nism forms much later than the maximum lateral load resistance
the ten-story frame, the rotation capacity exceeded at the seventh point which is determined by the modified Riks method used in
story column. This is largely because the flange of the I section ABAQUS 共Hibbitt, Karlsson and Sorensen, Inc., Pawtucket, R.I.,
member is slender to meet the rotation demand under large lateral 1996兲, as shown in the load versus deflection curves of split-X-
sway, which has exceeded the 2% drift limit. The local buckling braced frames 共Fig. 9兲.
occurs before failures such as shear failure and collapse mecha- The overstrength and ductility factors are listed in Table 2 for
nism. For ten-story MRF, under the ultimate vertical load combi- CBFs. Under the most common vertical load combination
nation, it is found that the local buckling occurs almost at the (1.0DL⫹0.4IL), the overstrength factors are 5.57, 3.33, and 2.48
same time as the excessive interstory drift. for three-, six- and ten-story inverted-V-braced frames, respec-
As a consequence of the above observation, it can be con- tively. It is found that the vertical load combination imposed on
cluded that the rotation capacities of all the beam and column the CBFs have relatively small influences on overstrength factors
members meet the demand of code allowable interstory drift of of these CBFs. For instance, the overstrength factor of a ten-story
2% story height. However, it should be highlighted that the local inverted-V-braced frame in ultimate loading decreases by 15.7%
buckling occurs close to the excessive drift point, and the lateral from that of the most common state. The base shear capacities
resistances at the local buckling points are similar to those at the under the most common vertical load combination, is equivalent
excessive interstory drift points, especially for ten-story MRF. For to 5.6% of seismic gravity load. The corresponding values for
relatively tall MRFs, the columns located in higher-story levels three- and six-story frames are 7.8 and 4.6%, respectively. For
have slender flanges of width to thickness ratio b/t f of about nine, split-X-braced frames, the overstrength factors are similar to the
and thus they have the tendency to buckle due to earthquake load. inverted-V-braced frames, which are 3.42 and 2.60 for six- and
It is seen that the ductility factor varies little with different ten-story frames, respectively, under the most common vertical
vertical load combinations. Under the most common vertical load load combination. Like the MRFs, under the most common, ser-
Fig. 11. Base shear-top floor lateral displacement curve under vari- Fig. 12. Comparison of load-displacement curves for between semi-
ous vertical load combinations for semirigid steel frames 共Type A兲 rigid frames 共Type B兲 and MRFs
Table 3. Overstrength, Ductility, and Response Modification Factors of Semirigid Frames 共Type A兲
Number of Vertical load Overstrength Ductility Response modification
stories combination ⍀ (V b /V d ) V b /0.01W g factor, R Failure mode
Semirigid Frame
Semirigid Frame „Type B… and infilled frames under the most common vertical load combi-
In order to investigate the effects of semirigid connections on the nations. It is shown that the three- and six-story semirigid frames
seismic behavior of the steel frame, the rigid connections of the 共Type A兲 have greater response modification factors than the cor-
MRFs are replaced with a semirigid counterpart. Also, the non- responding CBFs, and the values are about 94 and 80% of the
linear behavior of the panel zone is introduced in the push-over corresponding values of three- and six-story MRFs, respectively.
analysis of the frames. The vertical loads are the same as what However, the response modification factor of the ten-story semi-
were used for MRFs, and the most common vertical load combi- rigid frame 共Type A兲 is similar to that of CBF, and it is almost
nation (1.0DL⫹0.4IL) is considered for the three-, six-, and ten- 60% of the corresponding MRF. Although the infilled frames have
story frames to make comparison. the greatest overstrength among the different steel frames, which
Fig. 12 illustrates the base shear versus top displacement is 2–3 times the values of MRFs, the response modification fac-
curves obtained for these two types of frames. Table 4 shows the tors are only 1.16 –1.45 times the corresponding values of MRFs
overstrength and ductility of the frames with a semirigid connec- due to the lack of ductility of infilled frames. It is interesting to
tion obtained using Eq. 共1兲, where design base shear V d is taken note that the response modification factors are almost the same
to be same values as the corresponding MRF. The overstrength for both types of semirigid ten-story frames and ten-story CBF.
factors are 4.37, 1.83, and 1.27 for three-, six-, and ten-story The semirigid frames possess the greatest ductility factors, while
frames, and the ductility factors are 2.50, 2.75, and 2.90, respec- the infilled frames have no apparent ductility.
tively.
Although the overstrength factors of three-, six-, and ten-story
semirigid frames 共Type B兲 are less by 45.4, 60.7, and 66.3% from Performance Evaluation
the corresponding values of MRFs the ductility factors are higher
by 28.2, 41.8, and 73.7%, respectively. A nonlinear method, N2 method 共Fajfar 2000兲, is chosen herein to
evaluate the performance of MRFs, CBFs, semirigid frames, and
infilled frames when they are subjected to the ‘‘design earth-
Effect of Infill Walls
quake’’ for Singapore proposed by Balendra et al. 共2002兲. The
The MRFs were investigated with infill walls which were mod- base shear versus top displacement curves obtained in the push-
eled as previously described. The infill walls were added to each over analysis for the steel frames are transformed to the capacity
story of the steel frame. The nondimensional curves shown in Fig. diagrams of a single-degree-of-freedom 共SDOF兲 system, and the
13 depict the base shear versus top displacement of these three capacity diagrams are compared with the seismic response spectra
infilled frames. All the infilled frames failed due to local shear 共demand兲. The steps of the simple version of the N2 method are
failure at the bottom of first-story leeward column. When the described below:
failure occurred the maximum compression stress in the wall was 1. Seismic demand for SDOF system: The seismic demand is
still within the permissible limits. traditionally defined in the form of an elastic 共pseudo兲 accel-
Table 5 lists the overstrength factors of three-, six-, and ten- eration spectrum S ae 共‘‘pseudo’’ will be omitted henceforth兲,
story infilled frames under different vertical load combinations. in which spectral accelerations are given as a function of the
The values decrease only slightly as the vertical load increases. In natural period of the structure T and specified damping coef-
the most common state, the corresponding values are 18.11, ficient. To obtain the design response spectrum for Singapore,
12.37, and 9.16 for three-, six-, and ten-story infilled frames, re- Balendra et al. 共2002兲 used a hazard prediction model com-
spectively, which are equivalent to 25.4, 17.3, and 20.5% of seis- ponent attenuation model to estimate the bedrock motion in
mic gravity load per frame. Singapore after the model has been validated by the historical
Since the yielding occurs after the shear failure point under all attenuation data. Since soft soil can significantly amplify the
the vertical load combinations, the ductility factor of these three bedrock motion, three sites in the Kallang formation on the
infilled frames is 1 for various loading states. eastern part of Singapore are selected in the predominant pe-
riod range of the rock motion. The sites are located at Marine
Parade 共MP兲, Katong Park, and Katong area. An ensemble of
Response Modification Factor
18 synthetic accelerograms which have been stochastically
Fig. 14 and Table 6 depict the overstrength, ductility, and re- simulated for the critical M-R 共where M stands for magnitude
sponse modification factors of MRFs, CBFs, semirigid frames, of earthquake and R stands for source to site distance兲 com-
Fig. 14. Comparison of overstrength factors, ductility factors, and response odification factors of various frames under most common vertical
load combination
for ten-story frames whose values are around 0.07%. The demand
6 Moment 4.66 1.94 9.04 on strength V e /W g where V e is the base shear corresponding to
resisting displacement demand and W g is the seismic gravity load, are also
frames presented for various steel frames in Table 7. The highest demand
Concentrically 3.33 1.57 5.23 of 7.8% of seismic gravity load W g is obtained for a ten-story
braced infilled frame. As shown in Fig. 15, the intersection of the capac-
frames ity diagram with the seismic spectrum for a ten-story infilled
Semirigid 2.80 2.61 7.31 frame is very close to the peak of spectrum of the MP site, and as
Infilled 12.37 1.00 12.37 such the demand is large. Three-story frames are subjected to
10 Moment 3.77 1.67 6.30 relatively higher-strength demand than six- and ten-story MRF,
resisting CBF, and semirigid frames, which vary from 3.2 to 4.2% of W g .
frames The corresponding values for the six- and ten-story MRFs, CBFs,
Concentrically 2.48 1.51 3.74 and semirigid frames vary from 0.6 to 1.9% of the seismic gravity
braced load. The ultimate lateral resistance of these frames are obtained
frames according to Eq. 共1兲, and listed in Table 7 in terms of RV d /W g ,
Semirigid 1.50 2.48 3.72 where R is the response modification factor and V d is the design
Infilled 9.16 1.00 9.16 lateral loads. The corresponding values for these frames are much
greater than the demands, which are 17.3 to 25.4% for infilled
frames, and 7.3 to 21.8% for MRFs, CBFs, and semirigid frames.
mined for a SDOF system. Thus, the multiple-degrees-of-
freedom 共MDOF兲 system needs to be converted to an equiva-
lent SODF system. The required equations that make the Conclusion
transformation are
兺mi⌽i m* The push-over analysis of MRFs reveals that the frames possess
⌫⫽ ⫽ (18)
兺mi⌽i2 兺mi⌽i2 significant overstrength and considerable ductility. The over-
strength factors of MRFs decrease from 8 to 3.5 as the number of
Dt storys increase from 3 to 10. The ductility factors of correspond-
Sd⫽ (19)
⌫ ing MRFs are all around 2. When the rigid connections are re-
V placed with semirigid connections, the overstrength factors of
F*⫽ (20) these three frames decrease around 50%, while the ductility fac-
⌫
tors increase more than 25%. The low-rise semirigid frames are
F* found to have considerable overstrength and ductility because the
Sa⫽ (21) semirigid connections are able to resist some moment and show
m*
where m i ⫽mass assigned to story i; ⌽ i ⫽component of dis- ductile behavior. The overstrength factors for three-, six-, and
placement shape 兵⌽其 for story i; D t ⫽time-dependent top dis- ten-story frames are 6, 3, and 1.5, respectively, while the ductility
placement; V⫽base shear of the MDOF model; m * factor is about 2.5. The infill walls are found to increase the
⫽equivalent mass of the SDOF system, and F * ⫽force of the overstrength of MRFs significantly, at the expense of ductility.
equivalent SDOF system. The steel frames with rigid connections and infill wall show sig-
nificant stiffness, which give rise to the small lateral drift and
significant shear force for the first-story columns. The over-
Capacity Diagram and Demand Curves for Various strength factors for three-, six-, and ten-story infilled frames are
Steel Frames found to be 18, 12, and 9, respectively. The overstrength and
ductility factors are almost the same for two different configura-
The base shear versus top displacement curves of the steel frames, tions of CBFs considered, viz inverted-V-braced and split-X-
such as MRFs, CBFs, 共inverted-V-braced兲, semirigid 共Type A兲 braced frames. The overstrength factors are 5.5, 3, and 2.5 for
frames 共the terms ‘‘inverted-V-braced’’ and ‘‘Type A’’ will be three-, six- and ten-story frames, respectively, and the ductility
omitted hereafter兲 and infilled frames, are converted to AD format factors are about 1.5. For all the frames, the response modification
as described above, and they are bilinearlized as in Fig. 1. Then factor decreases when the number of storys increases from 3 to
these capacity diagrams are compared with the seismic demand 10. The response modification factor is found to vary from 16 to
curves, which are shown in Fig. 15 for corresponding steel 6 for MRF, from 8.5 to 3.5 for CBF, and from 15 to 3.5 for a
Fig. 15. Seismic demand spectra versus capacity diagrams for 共a兲 MRFs, 共b兲 CBFs, 共c兲 semirigid frames, and 共d兲 infilled frames