Seismic Design and Behavior of Ductile Knee-Braced Moment Frames

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Seismic Design and Behavior of Ductile

Knee-Braced Moment Frames


Sutat Leelataviwat1; Bunyarit Suksan2; Jarun Srechai3; and Pennung Warnitchai4

Abstract: The design and behavior of a ductile structural system called a knee-braced moment frame (KBMF) are presented in this paper.
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

The design of this structural system is based on a capacity-design concept that results in ductile behavior. For this system, the frames are
designed so that the knee braces will yield and buckle under seismic loads; this is followed by plastic hinging of beams at the ends of the beam
segments outside the knee portions. Through this concept, inelastic activities are confined to the designated elements. The knee braces also
provide much less obstruction than the braces of conventional systems, making this structural system architecturally attractive. The design
concept of this new structural system is addressed first. Results from an experimental study into the seismic behavior of the proposed system
are then presented. Two approximately half-scale KBMF specimens were tested. The load-deformation characteristics obtained from the test
indicate that the newly developed system can be a viable alternative to conventional structural systems. Finally, results from dynamic analyses
of an example KBMF structure designed by the proposed method are provided. DOI: 10.1061/(ASCE)ST.1943-541X.0000301. © 2011
American Society of Civil Engineers.
CE Database subject headings: Frames; Bracing; Structural steel; Seismic design; Cyclic tests.
Author keywords: Moment frames; Knee braces; Structural steel system; Seismic design; Cyclic test.

Introduction resulting in a stable hysteretic response (Engelhardt and Popov


1989, 1992).
Conventional structural steel framing systems, such as moment A hybrid structural system that combines the key characteristics
resisting frames (MRFs), concentrically braced frames (CBFs), of MRFs and CBFs is presented in this paper. This system, called a
and eccentrically braced frames (EBFs), are widely used in seismi- knee-braced moment frame (KBMF), relies on moment-resisting
cally active areas. Each of these structural systems offers different frames, together with stocky knee braces, as means to resist seismic
advantages and disadvantages. For example, MRFs provide a very forces. Knee braces were used in the past for wind-resistant design
stable hysteretic behavior and create no architectural obstructions. and have been recently explored in various forms for seismic
However, the performance of these frames is strongly dependent applications (Aristizabal-Ochoa 1986; Balendra 1991; Sarraf and
on the quality of materials and workmanship, particularly at the Bruneau 1996; Balendra et al. 2001; Kim and Seo 2003; Inouel
beam-to-column connections (FEMA 2000a, b; AISC 2005a). et al. 2006). The design of the proposed KBMF structural system
CBFs provide excellent stiffness but they rely essentially on the is based on a capacity-design concept that results in ductile behav-
postbuckling behavior of the braces. Unfortunately, postbuckling ior. For this system, the frames are designed so that the knee braces
behavior is complex and not as well understood. The braces will yield and buckle under seismic loads; this is followed by plas-
may exhibit a considerable degradation in strength after buckling,
tic hinging of beams at the ends of the beam segments outside the
leading to asymmetrical cyclic behaviors with respect to tension
knee portions. All inelastic activities are directed away from the
and compression (Sabelli 2001). Research has suggested that
critical areas, decreasing the dependence of the performance on
the performance of CBFs may vary significantly (Uriz and
the material and quality of the welded joints.
Mahin 2008). Recent advances have included the use of buckling
Because the moment connections are expected to remain
restrained braces (BRBs) to enhance hysteretic response (Uang and
elastic, relatively simple details can be used. Consequently,
Nakashima 2004). EBFs also rely on the yielding of ductile links,
demanding quality assurance procedures can be avoided. In addi-
tion, after a moderate earthquake, the knee braces can be repaired,
1
Assistant Professor, Dept. of Civil Engineering, King Mongkut’s Univ. provided that the columns are not distressed. More importantly, the
of Technology, Bangkok, Thailand (corresponding author). E-mail: sutat knee braces provide much less obstruction than the braces of con-
.lee@kmutt.ac.th
2 ventional systems, making this system architecturally attractive.
Structural Engineer, Worley Parsons (Thailand) Co., Ltd., Bangkok,
Thailand. They can also be utilized in the seismic strengthening of existing
3
Doctoral Student, Dept. of Civil Engineering, Chulalongkorn Univ., MRFs.
Bangkok, Thailand. An experimental study into the seismic behavior of KBMFs was
4
Associate Professor, School of Engineering and Technology, Asian carried out to investigate the ability of the system to dissipate
Institute of Technology, Patumthani, Thailand. earthquake energy. The results of the study are reported herein.
Note. This manuscript was submitted on July 12, 2009; approved on In this paper, the design concept and key parameters that govern
August 5, 2010; published online on August 25, 2010. Discussion period
the seismic response of the frame are addressed first. The test
open until October 1, 2011; separate discussions must be submitted for in-
dividual papers. This paper is part of the Journal of Structural Engineer- program and the details of the test specimens are then provided.
ing, Vol. 137, No. 5, May 1, 2011. ©ASCE, ISSN 0733-9445/2011/5-579– Finally, results from the quasi-static, cyclic tests of the specimens
588/$25.00. are discussed.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011 / 579

J. Struct. Eng. 2011.137:579-588.


Design Concept of Ductile Knee-Braced For a frame where the length of the clear span, Lc , is relatively
Moment Frames large, the beam yields primarily in bending. The fully yielded and
strain-hardened plastic moment of the beam can be estimated by
Yield Mechanism and Required Strength multiplying the plastic moment by an overstrength factor
of Key Elements M max ¼ ξM p ð2Þ
The design of a KBMF is based on a preselected yield mechanism
that limits inelastic activities to ductile segments of the frame. For where ξ = overstrength factor; and M p = plastic moment of the
this structural system, seismic energy is dissipated by means of the beam based on expected yield stress. The axial force in the beam
yielding and buckling of the knee braces and flexural yielding of can be taken into account when calculating the plastic moment of
the beams outside the knee regions with a selected yield mecha- the beam, leading to a smaller value of M p . However, for design
nism, as illustrated in Fig. 1. The strength of the knee braces is purposes, this effect is disregarded as a first approximation. The
selected to achieve the desired mechanism. In addition, the braces maximum shear force that can be developed is related to the maxi-
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

are designed to limit the moments at the beam-to-column connec- mum moment and the clear span
tions to a selected value below the yield moment. The other mem-
V max ¼ 2M max =Lc ¼ 2ξM p =Lc ð3Þ
bers in the frame are designed to remain elastic under the largest
forces generated by fully yielded and strain-hardened plastic hinges The moment at the beam-to-column connection given by Eq. (1) is
and knee braces, except at the column bases where plastic hinges to be limited in such a way that it is always less than a fraction of
are required to complete the mechanism. the plastic moment, i.e.,
The first step in designing a KBMF is the design of the base-
moment frame. The sizing of the moment frame members can be M C ≤ γM p ð4Þ
carried out using any accepted design procedure. After the base-
moment frame has been designed, the sizes of the knee braces where γ = numerical factor with a value less than 1.0. This factor
are selected to ensure the formation of the desired mechanism. can be chosen depending on the allowable moment at the connec-
After the selection and incorporation of knee braces, the columns tions. By using Eqs. (2)–(4), and taking L ¼ Lc þ 2Lk , Eq. (1) can
can be evaluated against additional forces. The size of the knee be cast in a dimensionless form as follows:
braces at each story level can be determined using the following
αPcr ðξ  γÞ 2ξ
assumptions and procedure. ≥ þ ð5Þ
As a first approximation, the gravity loads are considered small M p =L ðLk =LÞ sinðθÞ ð1  2Lk =LÞ sinðθÞ
and are disregarded in the design. This assumption is generally true
because lateral load-resisting frames are normally placed around The above equation provides the required postbuckling strength of
the perimeter of the buildings, and are designed to resist much the knee brace for a given beam size and knee brace configuration.
larger seismic forces than forces from gravity loads. For a frame The goal is to keep the moment at the connection less than the
with key parameters, as illustrated in Fig. 1, it can be shown that specified level and, at the same time, to achieve the selected
the maximum moment at the beam-to-column connection, M c , at a mechanism.
given story can be expressed as For a given story, once the size of the beam, the length of the
knee portion, and the angle of the knee brace have been selected,
M C ¼ M max þ V max Lk  αPcr sinðθÞLk ð1Þ the size of the knee brace can be determined. The columns can then
be evaluated against the forces that occur because of the knee
where M max and V max = maximum moment and the maximum braces. A capacity-design approach that considers the equilibrium
shear force, respectively, that can be developed at the plastic hinges; of the entire column (Goel and Chao 2008) subjected to forces
α = postbuckling strength reduction factor (Remennikov and generated by the beams and the knee braces could be used.
Walpole 1998); Pcr = buckling strength of the knee brace; θ = angle Alternatively, a pushover analysis can be carried out up to an
that the knee braces make with the beam; and Lk = length of the expected displacement demand level, assuming that elastic col-
knee portion. Eq. (1) indicates that the maximum moment at umns and the forces obtained from the analysis can be used in
the connection occurs when the beam reaches the fully yielded the design of the columns.
and strain-hardened plastic moment and the knee brace reaches The preceding analysis is based on the assumption that the
its final postbuckling strength, αPcr . gravity loads are small. In actual design, all different loading

Fig. 1. Yield mechanism of knee-braced moment frame

580 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011

J. Struct. Eng. 2011.137:579-588.


combinations must be considered. Because actual KBMFs may Recently, Lee and Bruneau (2005) conducted a review of exper-
vary in size and complexity, more rigorous analyses, such as imental studies on the cyclic testing of steel braces carried out
nonlinear time history and multiple-mode pushover analyses that between 1978 and 2006. The review included data from seven
consider realistic loading conditions, structural irregularity, specific independent studies covering 66 test specimens with a variety of
site effects, and multiple seismic hazard levels, may be necessary section types, lengths, and end restraints. The slenderness ratios
for a final design check and an evaluation of the structure. (kL=r) of the reviewed data ranged from 25 to 174. However,
out of 66 specimens reported, only three were stocky members with
Frame Geometry slenderness ratio values of approximately 30 or less, which is
Based on Eq. (5), the influence of the length of the knee portion, Lk , within the range of slenderness ratio values expected to be used
and the brace angle, θ, on the required strength of the knee brace in KBMFs. Nevertheless, the results clearly indicate, as expected,
relative to the beam plastic strength can be investigated. As an that stocky members provide superior energy dissipation at low
example, for a frame with ξ ¼ 1:25 and γ ¼ 0:80, the required displacement ductility levels, especially for tubular members.
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

brace strengths [as determined by Eq. (5)] for different values of Remennikov and Walpole (1998), on the basis of limited test results
Lk =L and θ are shown in Fig. 2. performed by various researchers, suggested an expression that can
As expected, the values of the required brace strength are very be used to compute the postbuckling strength reduction factor:
large when the values of Lk =L are small. The minimum required α ¼ 1:0 for 0 ≤ λe < 1 ð7a Þ
strength occurs when Lk =L is in the range of 0.13 to 0.25 for
all values of the brace angle. An Lk =L ratio of 0.2 appears to be α ¼ 1:038λ1:1 for 1 ≤ λe < 4 ð7b Þ
e
very practical for providing a large bay opening with relatively
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
low required brace strength. This ratio was used for the subsequent where λe ¼ ðkLb =rÞ ðF y =EÞ = effective slenderness ratio; F y =
testing program of this research. yield stress; E = modulus of elasticity; Lb = length of the brace;
The length of the knee portion can also be related to the plastic k = effective length factor; and r = radius of gyration. Trial designs
rotation of the beams. Using Fig. 1, it can be shown that of KBMFs with realistic member sizes indicate that the effective
  slenderness ratios of the knee braces should be approximately
2L 1.25 or less, leading to the values of the postbuckling strength
θp ¼ δ p 1 þ k ð6Þ
Lc reduction factor between 0.80 and 1.0. Using Eqs. (7a) and (7b),
the size of the knee brace that satisfies Eq. (5) can be determined.
where θp is the plastic rotation of the beam and δp is the plastic story The fracture of braces attributable to buckling and low-cycle
drift of the frame. Eq. (6) can be used to assess the ductility fatigue is also an important issue that must be addressed. The frac-
capacities of the beams given the story drift demands of the frame ture of the braces will lead to an undesirable increase in the stress at
under earthquake ground motions. the connections and might lead to fracture at the beam-to-column
connections, which would eventually lead to an unanticipated loss
Hysteretic Response and Postbuckling Strength of lateral strength.
of Knee Braces Past research results (Black et al. 1980; Fell et al. 2009),
although limited mostly to slender braces, have indicated that frac-
One of the key parameters in the design of KBMFs is the postbuck- ture initiation is primarily caused by the local strain induced by
ling strength reduction factor, α. KBMFs rely mainly on short and local and global buckling. However, this fracture could be delayed
stocky knee braces to dissipate seismic energy. The seismic behav- by using a small width-to-thickness ratio for the brace members.
ior of steel braces has been a subject of extensive research (Lee and Cross-sectional shapes also affect brace fracture. Test results (Fell
Bruneau 2005; Uriz et al. 2008; Fell et al. 2009). It is well known et al. 2009) have shown that the HSS shapes are prone to local frac-
that the hysteretic response of steel braces is influenced most by the ture because of buckling at the corners that significantly increases
brace slenderness ratio. The buckling strength of a brace is reduced the local strain. The pipe and W-shaped sections are less prone to
with each cycle of loading, leading to an unstable hysteretic fracture in this regard.
response. However, research conducted in the past mostly focused By selecting an appropriate shape with a low width-to-thickness
on the seismic response of slender braces commonly employed in ratio, the adverse effect attributable to early fracture could be
conventionally braced frames. avoided for the range of ductility expected for KBMFs. Never-
theless, because the past test results for braces have been made
mostly for slender braces tests, cyclic tests of stocky braces should
be carried out in the future to investigate this problem.

Experimental Program

To validate the KBMF concept, cyclic testing of two approximately


half-scale, one-story, and one-bay KBMF specimens were per-
formed. The two frames were designed using the procedure previ-
ously described with two major different details as will be described
in the following sections.
The test specimens were laterally braced at every one-fourth of
the span length as well as at the knee locations. The brace spacing
is made smaller than the maximum spacing required for a special
moment resisting frame (SMRF), according to AISC (2005b), to
make certain that the test can be carried out well into the inelastic
Fig. 2. Required brace strength for different values of Lk=L and θ
range. The overview of the test setup is shown in Fig. 3. The

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011 / 581

J. Struct. Eng. 2011.137:579-588.


Table 1. Summary of Average Material Properties from Coupon Tests
Specimen Member Location Yield strength (MPa)
KBMF-1 Beam Flange 331
Web 337
Column Flange 326
Web 326
Knee brace 352
KBMF-2 Beam Flange 326
Web 338
Column Flange 315
Web 296
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

Knee brace Knee brace 326

based on a probable strength of 340 MPa, which is commonly


Fig. 3. Overview of KBMF test frame (photo by the authors) found in locally available steel. The actual values of yield strength
from subsequent coupon tests are given in Table 1.
The final design resulted in a KBMF consisting of a 4 m-long
beam H250 × 125 × 6 × 9–29:6 kg=m, 1.95 m-high H250 × 250×
9 × 14–72:4 kg=m columns, and ϕ76:2 × 3:9 knee braces (dimen-
sions in mm). The hollow circular section was used for the braces
because it provided the closest cross-sectional area to the value
required by the calculations. Fully welded beam-column connec-
tions were used in specimen 1. The knee braces and the gusset
plates were also fully welded to the beam and the columns. Visual
inspection alone was performed for the welds. No additional
precautions were taken.
The gusset plates were designed to avoid any out-of-plane buck-
ling to minimize out-of-plane forces that would act on the columns.
The gusset plates were first designed to resist the full tension and
compression from the knee brace based on the Whitmore width.
A stiffener was then added to prevent out-of-plane folding of
the plates, which resulted in a 10-mm thick gusset plate with a
125-mm wide transverse stiffener. The test specimen KBMF-1 is
Fig. 4. Loading history of KBMF test frame shown schematically in Fig. 5.
Fig. 6 shows the hysteretic loops obtained from the test. During
the test, localized yielding was first detected at the drift of approx-
imately 1.0% in one of the knee braces. The overall response of the
specimens were instrumented by strain gauges and displacement
frame started to deviate from linear behavior at the story drift of
transducers at key locations. The two specimens were subjected
approximately 1.5%. Between 2% and 3% story drifts, the beam
to quasi-static, cyclic loading, and the loading history is shown
started to plastify, and plastic hinges started to become visible at
in Fig. 4. The details and the observed behavior of the two spec-
the end of the 3% drift cycle.
imens are as follows.
The maximum load at the story drift of 3% was approximately
Test Specimen 1 (KBMF-1) 280 kN, slightly larger than expected. At the 3% story drift, the
mechanism, consisting of yielding and buckling of knee braces
The first specimen was designed by plastic-mechanism analysis and flexural hinging of the beam, formed as intended in the design.
such that the estimated yield strength of the frame was 250 kN. The average strain values, normalized by the uniaxial yield
This lateral strength was chosen primarily because of the capacity strain (εy ¼ F y =200 GPa) from coupon tests, at the beam-to-
of the actuator. The base-moment frame was first designed to meet column connection (strain gauges Set A in Fig. 5), at the plastic
the target lateral strength. The knee braces were then chosen using hinge (strain gauges Set B), and at the knee brace (strain gauge
the approach previously discussed. Set C) are shown in Fig. 7. From these strain values, the sequence
The knee braces were designed assuming a probable over- of yielding can be deduced. The yielding first initiated in the knee
strength factor, ξ, of 1.25. This value was believed to be sufficient brace followed by yielding in the beam at the expected location.
for story drift levels of 3 to 4%. The value was estimated based on Significant yielding occurred from the drift level of 2% onwards
the overstrength factor for the design of other structural systems in agreement with the test observation. The beam-to-column con-
(AISC 2005b). The allowable bending moment at the connection nection remained below the yield level for the drift level up to 3%.
was selected to be 80% of the plastic moment of the beam Beyond the 3% story drift, the yielding area started to widen and
(γ ¼ 0:80). The length of the knee portion, Lk , was chosen to spread into the beam sections inside the knee regions. This was
be 20% of the span length (Lk =L ¼ 0:20), which corresponds to primarily attributable to the strain hardening at the hinges. This in-
the optimum value as discussed previously. For the knee brace, crease in the plastic hinge strength coupled with the reduction in the
because of its low slenderness ratio, the postbuckling strength postbuckling strength of the knee brace eventually resulted in the
reduction factor, α, was selected to be 1.0. The entire design is rise in the strain values at the beam-to-column connection, as

582 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011

J. Struct. Eng. 2011.137:579-588.


Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Details of specimen KBMF-1

shown in Fig. 7. The specimen continued to carry the load beyond


the design level up to the story drift of 4%. At this drift level,
the mechanism could be clearly identified, as can be seen in Figs. 8
and 9. The test was unexpectedly interrupted because of the fracture
of one of the bolts at the base plate and had to be temporarily
stopped because it was deemed unsafe to continue loading.
After the bolt had been replaced, the test was resumed, and the
frame was cycled at 4% story drift. In this stage of loading, the out-
of-plane deformation of the beam started to become visible. For this
test, the lateral-torsional buckling was prevented by the lateral brac-
ing system. In actual structures, the bracing must also be provided
by means of the floor slabs and cross beams. Because of the limited
number of test specimens, the influence of different design param-
eters on the lateral-torsional-buckling of the beam cannot be accu-
rately assessed at this time. Further study is needed on this issue.
Fig. 6. Hysteretic loops of specimen KBMF-1 The test was finally terminated when the net section fracture at the
brace-to-gusset plate connection was observed, as shown in Fig. 10.

Test Specimen 2 (KBMF-2)


During the testing of the KBMF-1 specimen, it was observed that
when the drift level became large, the yielding areas at the plastic

Fig. 7. Normalized average strain at key locations of specimen Fig. 8. Mechanism including plastic hinging and brace yielding in the
KBMF-1 test specimen at 4% story drift (photo by the authors)

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011 / 583

J. Struct. Eng. 2011.137:579-588.


hinges started to widen and penetrate into the portion of the beams
inside the knee regions. For the second test, the reduced beam
section (RBS) technique was applied to the beam to further confine
the inelastic activities at the plastic hinges. The RBS technique is
generally used at beam-to-column connections to relocate plastic
hinges away from the critical areas (AISC 2005a). A further advan-
tage of using RBSs with KBMFs is that the size of the knee brace
could be reduced according to Eq. (5).
The overall design of the KBMF-2 specimen followed the
approach previously described. The second test frame used the same
column and beam sizes as the first specimen. The RBS was designed
such that the plastic moment of the RBS was 75% of the original
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

section. For the knee braces, it was decided to use the same brace
size as in the KBMF-1 specimen, even though calculations showed
that some size reduction was possible. The details of the KBMF-2
specimen are shown in Fig. 11. The second specimen was subjected
to the same loading history, as shown in Fig. 4.
The hysteretic response of specimen KBMF-2 is shown in
Fig. 9. Buckling of knee braces when specimen is subjected to Fig. 12. For the second test, the specimen was loaded up to a story
maximum test deformation (photo by the authors) drift of 5%. The specimen responded elastically up to the story drift
of approximately 1.2%. After this drift level, significant inelastic
behavior was observed. The specimen continued to carry the load
with the mechanism envisioned in the design concept. The yielding
of the RBS at this drift level is shown in Fig. 13. At this point, a
slight out-of-plane curvature can be observed during the test.
However, the system remained stable, which again illustrates the
importance of lateral braces.
Fracture eventually occurred in one of the knee braces near the
end of the second cycle of the 4% story drift level. The test was
continued even after the fracture of the knee brace to study the
behavior of the system in the event of extreme overloading. The
loss of the knee brace also provided a chance to observe the con-
tribution of the knee braces to the overall strength and stiffness of
the frame. As can be seen in Fig. 12, there is a decrease in lateral
strength and stiffness of approximately 20% after the loss of the
knee brace, which clearly shows the significant contributions of
the knee braces to the overall lateral strength. More importantly,
it was found that the frame continued to carry the load as a regular
MRF, with significant yielding occurring at the beam-to-column
Fig. 10. Failure of knee brace (photo by the authors) connections. The test was eventually terminated after the comple-
tion of the loading history.

Fig. 11. Details of specimen KBMF-2

584 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011

J. Struct. Eng. 2011.137:579-588.


Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Hysteretic loops of specimen KBMF-2 Fig. 15. Comparison of peak normalized strain values at the beam-to-
column connection

the first specimen. Despite reductions in the strength and stiffness,


the stable hysteretic response of the second specimen clearly indi-
cates that the RBS technique can be applied to KBMFs.
For this particular specimen, the average strain values indicated
that the yielding started in the RBS and was followed by yielding in
the knee braces. The peak normalized strain values at the beam-to-
column connections at each drift level are also shown (for compari-
son) in Fig. 15. As expected, the strain values were relatively
smaller in KBMF-2 because of the reduction in the overall strength
of the frame attributable to RBS. More importantly, the rate at
which the peak positive strain increased beyond the 3% drift level
was significantly smaller in the KBMF-2 specimen than in the
KBMF-1 specimen. This further demonstrates the effectiveness
of the RBS technique when applied to KBMFs. However, because
of additional fabrication costs, the use of RBSs should be reserved
for cases where a large level of deformation is expected.
Fig. 13. Flexural yielding of RBS in KBMF test specimen 2 at 4.0%
Example Application
story drift (photo by the authors)
An example KBMF structure was used to study the seismic
response of the proposed system. A three-story example structure
The response envelopes from the two tests are compared in was designed and subjected to inelastic static and dynamic analyses
Fig. 14. Specimen 2 has a smaller lateral strength because of to obtain the overall deformation demands and to analyze the
the reduction in the beam cross section. The elastic stiffness is response of key members. The elevation view of the study frame
is shown in Fig. 16. The seismic weights are 2,665 kN for the roof
only slightly affected by the RBS. The elastic stiffness of the
and 2,810 kN for the third and second stories, respectively. The
KBMF-2 specimen was approximately 5% lower than that of
frame was designed based on the ICC (2000) provisions. Important
constants used to calculate the design spectrum were S1 ¼ 0:8 g
and SS ¼ 1:2 g, Seismic Use Group I, and Soil Type B.

Fig. 14. Comparison of envelope curves of KBMF-1 and KBMF-2 test


frames Fig. 16. Elevation of example KBMF structure

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011 / 585

J. Struct. Eng. 2011.137:579-588.


A nonlinear finite-element code SNAP-2DX (Rai et al. 1996)
developed at the University of Michigan was used to perform
all analyses. The knee braces and frame members were modeled
using SNAP-2DX axial compression and beam-column elements.
The axial compression element in SNAP-2DX uses a hysteretic
model that is capable of modeling postbuckling behavior. The
beams and columns were modeled using lumped plasticity beam-
column elements. Elastic-plastic with strain hardening hysteretic
behavior was assumed for the plastic hinges. To capture the
postelastic response of the relatively short beam segments between
the knee braces, a fairly large strain hardening value of 8% was
assigned for the plastic hinges in the beams.
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

Panel zones (PZs) were modeled according to a method pro-


posed by Schneider and Amidi (1998). The shear resistance of a
joint was simulated by two rotational springs corresponding to
Fig. 17. Pushover curve and sequence of inelastic activities
the contributions of the column web and the column flanges sur-
rounding the PZ. Rigid eccentricities were used to specify the finite
dimension of the elements. In the analyses, small gravity loads suit-
The base-moment frame was first designed using the design able for a perimeter frame were applied as concentrated forces on
base shear equal to 70% of the design base shear calculated by the beams at every one-fifth of the span length.
using a response modification coefficient (R) of 8.0. This resulted The plot of roof drift versus base shear from the pushover analy-
in a design base shear coefficient (V=W), including an accidental sis using IBC lateral force distribution is shown in Fig. 17. Also
torsion of 0.084. The knee braces were then designed and added to shown in the figure is the sequence of inelastic activities up to a
the base-moment frame using the capacity-design concept previ- roof drift of 2.5%. The response of the frame was elastic up to
ously described. The 30% reduction in the design base shear of the a drift level of approximately 0.8%, when the first set of plastic
base-moment frame was made to ensure that in the end, the result- hinges formed. The inelastic activity then quickly spread out,
ing KBMF would have approximately the same strength as a con- resulting in a significant reduction in lateral stiffness. The full
ventional SMRF designed by ICC (2000), with R ¼ 8:0. The RBS mechanism formed when all the braces and beams had yielded
technique was not used in the design of this example frame. at a roof drift of about 2.5%. Beyond this drift level, the resistance
The length of the knee portion was chosen to be 20% of the span came primarily from the strain hardening of the material at plastic
length. The overstrength factor (ξ) was taken to be 1.35. For the hinges. Plastic hinges occurred only at the designated locations, as
knee braces, the postbuckling strength reduction factor, α, was intended in the design. No yielding in the panel zones or the col-
taken to be 0.80. In this study, the nominal strength values for umns was detected primarily because the moments at beam-to-
the beams and columns were assumed to be 248 MPa (36 ksi) column connections were limited by the presence of the knee
and 345 MPa (50 ksi), respectively. The actual yield strength for braces.
the braces and the beams used in the capacity design was taken In the nonlinear dynamic analyses, the frame was subjected to
as 338 MPa (49 ksi). The actual yield strength of the columns seven selected earthquake records scaled to represent two levels of
was taken as 379 MPa (55 ksi). These actual yield stress values ground motion intensities: the design basis earthquake (DBE) and
were also used in all of the analyses. The members were selected the maximum considered earthquake (MCE) levels. These ground
to satisfy the compactness requirements (AISC 2005b). motions are taken from another study (Somerville et al. 1997).

Fig. 18. Maximum interstory drifts and inelastic activities attributable to DBE and MCE ground motions

586 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011

J. Struct. Eng. 2011.137:579-588.


The seven ground motions are referred to as LA02, LA06, LA08, properly considered when analyzing KBMFs under earthquake
LA10, LA14, LA16, and LA18. For the DBE level, the ground ground motions.
motions were scaled such that their average spectra values between 5. Lateral bracing in KBMFs is an important issue and must be
the periods of 0.2T and 1.5T were not less than those obtained from properly addressed. To this end, it may be advantageous to pro-
the IBC elastic design spectrum. In this study, the MCE ground vide lateral supports at or near the plastic hinges, at certain
motions were defined as the DBE ground motions amplified by points along the beam, and at the brace-to-column joint to
a factor of 1.5. A mass proportional damping of 2% was assumed. ensure stability during inelastic deformations.
P-delta was not included in the analysis. 6. The experiments show that the connections can be designed
The results including the overall deformation demands and to remain elastic. However, it is still prudent to use post-
the response of key members were obtained from the nonlinear Northridge detailing practices for enhanced ductility in case
dynamic analyses. Fig. 18 shows the values of maximum interstory of an extreme event not anticipated by the design.
drifts under the seven selected records along with their average 7. The response analysis of the example structure show that plas-
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

values. The maximum average interstory drifts are approximately tic hinges occurred only at the designated locations, as in-
1.6% and 2.3% for the DBE and MCE ground motions, respec- tended in the design. The results indicate that the proposed
tively. The peak story drifts are 2.3% and 3.6% for the DBE concept is viable and can be applied to multistory structures.
and MCE ground motions, respectively. These drifts are smaller Although further studies may be required before this system can
than the 4.0% ultimate story drift observed during the test. be used in practice, the results of this study strongly illustrate the
Based on the response observed during the test of specimen importance and the potential of the KBMF system, particularly in
KBMF-1, it can be expected that under the DBE ground motions cases where obstructions because of conventional braces and strin-
(1.6% story drift), a significant yielding will occur in the knee gent, demanding quality assurance procedures are issues under
braces along with a slight yielding at the plastic hinges. The consideration. In addition, because the tests conducted in this study
beam-to-column connections are expected to remain in the elastic were carried out on relatively large-scale specimens, they demon-
range. For the MCE ground motions (2.3% story drift), it can be strate that real and practical solutions can actually be developed.
expected that a significant yielding in the braces will occur with a Research on the dynamic behavior of KBMF systems as well as
ductility demand in excess of 2.0. This is accompanied by a sig- studies to improve the details are being carried out by the authors
nificant yielding at the plastic hinges. The stress at the beam-to- to further investigate and improve the behavior of KBMFs.
column connection remains only slightly below the yield level.
Inelastic activities of the frame are also shown in Fig. 18. Again,
it can be seen that the plastic hinges occurred only at the designated Acknowledgments
locations, which is consistent with the design concept.
This research work was supported by funding from the Royal Thai
Government through King Mongkut’s University of Technology–
Conclusions Asian Institute of Technology joint research program and from the
National Research Council of Thailand. The authors gratefully
This paper presents the design concept and seismic behavior of acknowledge the support from the above agencies. Generous assis-
KBMFs. For this system, the frames are designed such that the tance from the Italian-Thai Development Public Company Limited
knee braces will yield and buckle under seismic loads along with in the fabrication of the test specimens is also gratefully acknowl-
the plastic hinging of beams at the ends of the beam segments out- edged. The conclusions in this paper are those of the authors and
side the knee portions. The design concept and design procedure may not necessarily reflect the views of the sponsors.
were presented. Experiments were carried out to validate the design
concept, as well as the details of various key elements. An example
was provided to illustrate the application of the proposed concept to Notation
multistory structures. The main findings can be summarized as
follows:
1. Cyclic tests of relatively large-scale specimens indicate that The following symbols are used in this paper:
KBMFs behave in a ductile manner with a stable hysteretic E = modulus of elasticity of steel;
characteristic. The test frames showed a ductile behavior with F y = yield strength;
all inelastic behavior confined to only the designated elements k = effective length factor;
in the frame. Based on these results, it can be concluded that Lb = length of knee brace;
the proposed system represents a viable alternative to existing Lc = clear span length, length of beam between the knee
structural systems. braces;
2. The plastic analysis and capacity-design procedure presented Lk = length of knee portion, length of beam inside the knee
in this paper can be used to determine the member sizes of region;
KBMFs. The concept was used in the design of test specimens M C = moment at the face of beam-to-column connection;
that performed as envisioned in the design process. M max = maximum moment that can be developed at the plastic
3. The RBS technique can be used to further enhance the perfor- hinges;
mance of KBMFs. The RBS confines inelastic activities to de- M p = plastic moment of beam;
signated locations. The RBS can potentially reduce the size of Pcr = buckling strength of knee brace;
key members. However, the increase in fabrication costs needs r = radius of gyration;
to be carefully assessed. V max = maximum shear force that can be developed at the plastic
4. The KBMF test specimen with RBS showed reduced lateral hinges;
strength and stiffness. The reduction in strength can be signif- α = postbuckling strength reduction factor;
icant depending on the design. However, the reduction in stiff- γ = numerical factor indicating the level of the allowable
ness is marginal. Nevertheless, these reductions should be moment at the connections;

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011 / 587

J. Struct. Eng. 2011.137:579-588.


δp = plastic story drift of the frame; Inouel, K., Suita, K., Takeuchi, I., Chusilp, P., Nakashima, M., and Zhou, F.
θ = angle that the knee braces make with the beam; (2006). “Seismic-resistant weld-free steel frame buildings with
θp = plastic rotation of the beam; mechanical joints and hysteretic dampers.” J. Struct. Eng., 132(6),
λe = effective slenderness ratio; and 864–872.
ξ = overstrength factor. International Code Council (ICC). (2000). International building code,
ICC, Washington, DC.
Lee, K., and Bruneau, M. (2005). “Energy dissipation of compression
members in concentrically braced frames: review of experimental data.”
References J. Struct. Eng., 131(4), 552–559.
Rai, D. C., Goel, S. C., and Firmansjah, J. (1996). “SNAP 2D-X: A general
American Institute of Steel Construction (AISC). (2005a). “Prequalified
purpose computer program for nonlinear structural analysis.” UMCEE
connections for special and intermediate steel moment frames for
seismic applications.” ANSI/AISC 358-05, AISC, Chicago. 96-21, Dept. of Civil and Environment Engineering, Univ. of Michigan,
American Institute of Steel Construction (AISC). (2005b). “Seismic provi- Ann Arbor, MI.
Downloaded from ascelibrary.org by University of Illinois at Chicago on 07/29/13. Copyright ASCE. For personal use only; all rights reserved.

sions for structural steel buildings.” ANSI/AISC 341-05, AISC, Chicago. Remennikov, A. M., and Walpole, W. R. (1998). “A note on compression
Aristizabal-Ochoa, J. D. (1986). “Disposable knee bracing: improvement in strength reduction factor for a buckled strut in seismic-resisting braced
seismic design of steel frames.” J. Struct. Eng., 112(7), 1544–1552. system.” Eng. Struct., 20(8), 779–782.
Balendra, T. (1991). “Preliminary studies into the behavior of knee braced Sabelli, R. (2001). “Research on improving the seismic behavior of
frames subject to seismic loading.” Eng. Struct., 13(1), 67–74. earthquake-resistant steel braced frames.” EERI/FEMA NEHRP Profes-
Balendra, T., Yu, C. H., and Lee, F. L. (2001). “An economical structural sional Fellowship Rep., Earthquake Engineering Research Institute,
system for wind and earthquake loads.” Eng. Struct., 23(5), 491–501. Oakland, CA.
Black, R. G., Wenger, W. A., and Popov, E. P. (1980). “Inelastic buckling of Sarraf, M., and Bruneau, M. (1996). “Cyclic testing of existing and retro-
steel struts under cyclic load reversal.” Rep. No. UCB/EERC-80/40, fitted riveted stiffened seat angle connections.” J. Struct. Eng., 122(7),
Earthquake Engineering Research Center, Univ. of California, Berkeley, 762–775.
Berkeley, CA. Schneider, S. P., and Amidi, A. (1998). “Seismic behavior of steel frames
Engelhardt, M. D., and Popov, E. P. (1989). “On design of eccentrically with deformable panel zones.” J. Struct. Eng., 124(1), 35–42.
braced frames.” Earthquake Spectra, 5(3), 495–511. Seo, Y., and Kim, J. (2003). “Seismic design of steel structures with
Engelhardt, M. D., and Popov, E. P. (1992). “Experimental performance of buckling-restrained knee braces.” J. Constr. Steel Res., 59(12),
long links in eccentrically braced frames.” J. Struct. Eng., 118(11), 1477–1497.
3067–3088. Somerville, P. G., Smith, M., Punyamurthula, S., and Sun, J. (1997).
Federal Emergency Management Agency (FEMA). (2000a).
“Development of ground motion time histories for phase 2 of the
“Recommended seismic design criteria for new steel moment-frame
FEMA/SAC steel project.” SAC/BD-97/04, SAC Joint Venture,
buildings.” FEMA-350, Federal Emergency Management Agency,
Sacramento, CA.
Washington, DC.
Uang, C. M., and Nakashima, M. (2004). “Steel buckling-restrained
Federal Emergency Management Agency (FEMA). (2000b).
“Recommended specifications and quality assurance guidelines for frames.” 16, Earthquake engineering: From engineering seismology
steel moment-frame construction for seismic applications.” FEMA to performance-based engineering, Y. Bozorgnia and V. V. Bertero,
353, Federal Emergency Management Agency, Washington, DC. eds, CRC Press, Boca Raton, FL.
Fell, B. V., Kanvinde, A. M., Dierlein, G. G., and Myers, A. T. (2009). Uriz, P., Filippou, F. C., and Mahin, S. A. (2008). “Model for cyclic
“Experimental investigation of inelastic cyclic buckling and fracture inelastic buckling of steel braces.” J. Struct. Eng., 134(4), 619–628.
of steel braces.” J. Struct. Eng., 135(1),19–32. Uriz, P., and Mahin, S. A. (2008). “Toward earthquake-resistant design of
Goel, S. C., and Chao, S.-H. (2008). Performance-based plastic design: concentrically braced steel-frame structures.” PEER-2008/08, Pacific
Earthquake-resistant steel structures, International Code Council, Earthquake Engineering Research Center (PEER), Univ. of California,
Washington, DC. Berkeley, Berkeley, CA.

588 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2011

J. Struct. Eng. 2011.137:579-588.

You might also like