Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Journal of Constructional Steel Research 67 (2011) 1760–1772

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

A balanced design procedure for special concentrically braced frame connections


Charles W. Roeder ⁎, Eric J. Lumpkin, Dawn E. Lehman
Department of Civil and Environmental Engineering, University of Washington, Seattle, WA 98195-2700, United States

a r t i c l e i n f o a b s t r a c t

Article history: Concentrically braced frames (CBFs) are stiff, strong structures that are suitable for resisting large lateral
Received 13 December 2010 loads. Special CBFs (SCBF) are used for seismic design and are designed and detailed to sustain relatively large
Accepted 26 April 2011 inelastic deformations without significant deterioration in resistance. Current AISC Seismic Design Provisions
Available online 14 June 2011
aim to ensure the brace sustains the required inelastic action, but recent research showed that current SCBF
design requirements lead to variable seismic performance, unintended failure modes, and limited
Keywords:
Seismic design
deformation capacity. To improve the seismic response of SCBFs, a balanced design procedure was proposed.
Braced frames The premise of the design methodology is to balance the primary yield mechanism, brace buckling and
Gusset plate connections yielding, with other, complementary ductile yielding mechanisms, such as gusset plate yielding. This balance
Special concentrically braced frames process maximizes ductile yielding in the frame thereby maximizing the drift capacity of the frame. Further,
Ductility the undesirable failure modes are balanced with the yield mechanisms and the preferred failure mode, brace
Inelastic behavior fracture, to ensure that the frame fails in the desired manner. To achieve the objectives of the design
methodology namely maximum drift capacity, and adherence to a desired yield and failure hierarchy, rational
resistance checks and appropriate balance factors (β factors) are used to balance each yield mechanism and
failure mode. These factors were developed, validated, and refined using the measured results from an
extensive test program. An SCBF connection design example to illustrate the application of the balanced
design method and to demonstrate differences from the current AISC design method is presented in an
appendix.
© 2011 Elsevier Ltd. All rights reserved.

Introduction connect the brace to the other framing members, as shown in Fig. 1,
because they are easier to construct and design than fully restrained
Concentrically braced frames (CBFs) are stiff, strong structures, brace-end connections. Design of gusset plate connections to achieve
which are suitable for resisting wind and seismic loading. CBF typically the design objectives requires significant effort [1], and recent research
are composed of diagonal bracing members connected to the beams and has shown that they may not lead to the intended response.
columns with gusset plate connection. The bracing may be placed in a Experimental investigations into modified design methods have
number of different configurations and geometries. CBF members are demonstrated small changes can improve the seismic performance of
initially designed assuming truss action, i.e. the members only carry the system [2]. These improvements can be implemented in practice
axial load. using a balanced-design procedure (BDP). Balanced design has been
Special concentrically braced frames (SCBF) are employed in high proposed as a method to control the sequence of yielding and to increase
seismic regions and use a Response Modification Factor, R, to reduce the inelastic deformation capacity in past seismic research (e.g., [3]), but it
seismic design force. During large, infrequent earthquakes, SCBFs must has not been widely used in practice because of the complexity and
sustain cyclic, inelastic tensile yielding and compressive buckling uncertainty in achieving the required component balancing in practice.
deformation of the brace without significant deterioration of stiffness In part, this uncertainty stems from the limited experimental data
and resistance for earthquakes exceeding the reduced design force. available to verify the design method and associated expressions. Here,
Brace buckling and tensile yield are the primary yield mechanisms of the this limitation is overcome using a large data set on braced frame system
system, which permit the frame to sustain the inelastic deformation and that simulates modern construction. The resulting BDP is robust and
energy dissipation demands needed to provide collapse-prevention available for immediate used by practitioners.
performance. This paper develops, presents, and justifies a simple BDP for SCBF
Although the brace is the primary member in the SCBF system, the gusset plate connections. The BDP identifies all possible yield mecha-
connections play an important role. Gusset plate connections typically nisms and failures modes associated with the SCBF system, and balances
their resistances to optimize the seismic performance of the system [4].
⁎ Corresponding author at: 233B More Hall, University of Washington, Seattle, WA
The BDP adapts currently accepted design expressions and uses balance
98195-2700, United States. Tel.: + 1 206 543 6199; fax: + 1 206 543 1543. factors (β factors) for each yield mechanism and failure mode to assure
E-mail address: croeder@u.washington.edu (C.W. Roeder). that (1) the primary yield mechanism occurs and is followed by

0143-974X/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2011.04.016
C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772 1761

Fig. 1. AISC gusset plate design; a) geometry, tension, buckling and net section checks; b) welds, free edge buckling, and weld checks; and c) geometry of tapered gusset.

secondary yield mechanisms, and (2) undesirable failure modes are the brace, Ry is the ratio of the expected yield strength to the
prevented. To maximize inelastic action and deformation capacity of the minimum specified yield strength, Fy, and Fcr is the critical stress
system, the BDP requires: (1) accurate models for predicting the associated with brace buckling.
resistance associated with each yield mechanism and failure mode, and
(2) balance factors (β factors) to assure the proper separation of the The gusset plate connecting the beam, column, and brace must be
primary and secondary yield mechanisms, to delay all failure modes and designed such that its resistance exceeds Put and Puc. Numerous
to suppress unacceptable failure modes. To achieve this, a review of design checks are needed to achieve this, and Table 1 provides the
current SCBF seismic design requirements is provided, and recent equations currently used in this evaluation. A cursory review of the
research that provides the basis for the proposed BDP is summarized. gusset plate design checks follows:
The results of this experimental research are then used to develop and 1. The net section resistance of the brace is evaluated adjacent to the
evaluate design expressions and β factors for relevant yield mechanisms gusset-to-brace connection and reinforcement is added if it cannot
and failure modes. Finally, a design example using the proposed BDP is achieve Put.
provided in the appendix. 2. The length of the brace-to-gusset interface is sized to suppress
block shear for Put.
Current design method 3. The AISC Uniform Force Method (UFM) may be used to select
gusset plate dimensions (dimensions a and b of Fig. 1a) and to
The premise of the current design method, as specified in the AISC establish the force or stress demands on the bolts or welds needed
Seismic Design Provisions [1] is to establish the force demands on the to maintain equilibrium with Put of the brace [5]. The welds or bolts
system and to use the LRFD AISC design provisions [5] to size the brace are then sized to sustain the computed demand.
and framing members. In this selection, the designer must: 4. The connection must be designed to either develop a fully
restrained brace connection or allowance must be made for brace
• Satisfy local and global slenderness limits for the brace and other end rotation during buckling. This end rotation capacity is
members. currently achieved with 2tp linear clearance, as shown in Fig. 1.
• Ensure that the tensile brace sustains no more than 70% of the lateral 5. The thickness of the gusset plate must be sufficient to develop both
force demand. Put and Puc. The tensile gusset plate resistance is determined using
• Design beams in frames with V-bracing and inverted V-bracing to the Whitmore width and specified strength of the plate [6]. The
sustain the vertical unbalanced load that results after brace buckling compressive resistance is determined using the Thornton method
occurs. which uses an average effective length of the gusset based upon the
• Design columns and column splices to resist the minimum of the centroidal length and the lengths at two ends of Whitmore width,
amplified seismic load and an amplified maximum force equal to the as shown in Fig. 1a [7]. The effective length coefficient, K, is
product of 1.1Ry and the maximum force the adjacent braces can commonly taken as 0.65 for corner gussets and 1.2 to 1.4 for
transfer to the column. midspan gussets. Edge buckling may also be evaluated.
• Establish the plastic capacity of the brace in tension, Put = RyFyAg,
and compression, Puc = 1.1RyFcrAg to complete the connection The design process is rational, however recent research has
design. In the expression, Ag is the gross cross-sectional area of demonstrated several shortcomings that have resulted in construction

Table 1
Limit states and corresponding resistance expressions for AISC and balanced design approaches.

Limit state AISC design Balanced design

Resistance (φRn) ϕ Resistance (βRn) β

Whitmore yielding ϕFyBwtp 0.90 βRyFyBwtp 1.0


Brace net section fracture ϕU(RtbFubAnb + FupAgp) 0.75 βU(RtbFubAnb + FupAgp) 0.95
Brace to gusset weld ϕ(0.6)FEXXNwLc(.707)w2 0.75 Same as AISC 0.75
Brace to gusset base metal ϕ(0.6)FuNsLctf 0.75 Same as AISC 0.75
Block shear ϕ{(0.6FuAnv + UbsFuAnt) ≤ (0.6FyAgv + UbsFuAnt)} 0.75 β{(0.6FuAnv + UbsFuAnt)} 0.85
Whitmore fracture ϕFuBwtp 0.75 βFuBwtp 0.85
Gusset plate buckling ϕBwtpFcr 0.90 βBwtpFcr 0.90
Interface welds ϕ(0.6)FEXX[1 + 0.5(sinγ)1.5]2 Lw(.707)w1 ≥ UFM 0.75 2(1.5)β(0.6)FEXX(.707)w1 ≥ RyFytp 0.75
1762 C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772

challenges and reduced drift capacity. A new approach for gusset plate multi-story bracing systems [21,22]. These experiments evaluated a
design is needed to remedy these deficiencies. range of gusset plate connection design parameters including:

Experimental studies • The current AISC design procedures.


• The range of different failure modes possible of SCBFs with gusset
An extensive body of research has been performed on gusset plate plate connections.
connections [8–15]. These connection tests are useful in establishing • The impact of different brace cross sections, configurations, and
and validating basic models for predicting their resistance, but they orientations.
are not good indicators of SCBF seismic performance because they do • Different connections (e.g., bolted and welded) between the gusset
not simulate the interaction of all of the braced frame components. In plate and framing members.
particular, these connection tests neglect the effect of brace buckling • Differences in the 2tp linear clearance and alternate clearance
deformation demands on the connection (see Fig. 2a), the significant models to achieve brace end rotation.
inelastic deformation demands on the beams and columns caused by • The gusset plate thickness and resulting relative strength and
frame action of the SCBF system (see Fig. 2b), and the inelastic stiffness of the brace, gusset and framing members.
deformation of the gusset caused by frame action. These differences • The impact of gusset plate shape by testing both tapered and
are significant, because these effects are needed to capture the full rectangular gusset plates.
seismic demands on the gusset plate connection and to accurately These tests realistically simulated the demands on and the
reflect the ductility and inelastic performance of the system. performance of these connections, and the results were used to
Experiments that include these effects simulate one or more bays develop the proposed BDP [2]. The specimens were subjected to a
of an SCBF frame [2,16–23]. A recent research program experimentally cyclic inelastic deformation history based on the ATC-24 testing
evaluated 28 full-scale, single-bay single-story SCBFs with member protocol [24], although a few tests employed a near-fault inspired
sizes typical of those used in the bottom story of a 3- or 4-story cyclic deformation history. Table 2 summarizes the general test
building (see Fig. 3a) [2,18–20,23]. In addition, six 2- and 3-story full- results. The drift range provided in this table is the total drift range
scale SCBF frames (see Fig. 3b) were tested at the National Center for including positive and negative deflections prior to achieving the
Earthquake Engineering (NCREE) Laboratory in Taiwan to evaluate initial failure mode. The drift ranges shown for multi-story frames in
this table are average drift ranges over the height of the structure.
Specimens with HSS in the identifier had rectangular HSS braces, and
a) Brace Buckling Demands on Gusset those with WF had wide flange braces. Test results presented in [2]
include the first eleven tests and some are reproduced here in brevity
for completion; the reader is referred to that paper for additional
information. When needed results from other tests are summarized
herein. More detailed information about all of the tests can be found in
the original reports (Specimens HSS1–HSS5 [18]; Specimens HSS6–
HSS12 [19]; Specimens HSS13–HSS19 [20]; Specimens HSS20–HSS28
[23]; Specimens TCBF1 [21]; Specimens TCBF2 1–3 [22]).
The research findings have broad implications on the current
seismic design methodology and mechanisms to improve it. Fig. 4
uses the force–drift response curves from specimens that have tested
with a wide range of connection properties to illustrate the diversity
in system performance observed in this research. Specifically,
differences in the measured response result from different controlling
yield mechanisms and failure modes. A few key observations are
provided.

• The current design methodology does not meet the performance


expectations. Specimens HSS-1 and HSS-12 both were designed to
meet current AISC specifications [1], where Specimen HSS-1 used
fillet welds at the gusset plate-beam and gusset plate-column
b) Local Yielding of Beams and Columns interfaces and Specimen HSS-12 used a complete joint penetration
(CJP). The interface welds in Specimen HSS-1 failed prior to brace
tearing, as shown in Fig. 4a. Although Specimen HSS-12 achieved
the desired failure mode (brace fracture), the drift was less than
1.5% in tension and far less than specimens design to meet all or part
of the BDP (see Fig. 4b). Specimen HSS-11, which had a large beam
and gusset plate and met current SCBF design methods, failed at
smaller inelastic deformation because the excessive strength and
stiffness of the gusset (encouraged by the current design philoso-
phy) limited the system drift capacity (see Fig. 4e). Specimen HSS11
greatly exceeded the BDP strength design requirements. However,
the BDP is intended to avoid excess overstrength, and hence this
connection meets the requirements of AISC but does not meet BDP
requirements.
• The 2tp linear clearance often results in relatively large, thick gusset
plates. Observations from nonlinear analyses and experiments
showed that a more compact gusset plate can achieve the brace
Fig. 2. Inelastic deformation demands on SCBF systems. end rotation demands and reduce construction challenges. Using
C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772 1763

a) Single Story Test b) Multi-Story Test


N
200 mm Composite SlabAcuators

W24x94

3.19 m
200 mm Composite Slab

W21x68

3.19 m

W12x106

W12x106
200 mm Composite Slab

W21x68

2.93 m
6.00 m

Fig. 3. Typical test specimens.

those results as a basis, an alternative elliptical clearance method capacity of the gusset plate to permit gusset yielding following brace
was developed for corner gusset plates, as indicated in Fig. 5 [2]. yielding. Comparison of Fig. 4a and c shows that this revision
Specimens HSS-2 through HSS-8 evaluated variations in the significantly improves connection inelastic deformation capacity
elliptical clearance distance. Based on those results, a clearance of and performance. A 6tp vertical clearance zone, as shown in Fig. 6,
8tp was selected. Specimen HSS-5 employed this 8tp elliptical was developed for gusset plates used at the midspan of the beam
clearance model and balanced the brace tensile capacity with the and was evaluated in Tests TCBF2-1 and TCBF2-2.

Table 2
Test summary.

Specimen brace type Specimen description Gusset and clearance Failure mode Drift range%

HSS-1 AISC design — fillet welds by UFM. 13 mm–2tp linear Weld fracture 2.8
HSS-2 HSS1 w/fillet weld sized to cap. of plate 13 mm–6tp ellipse Brace fracture 4.0
HSS-3 BDP — fillet weld sized to capacity of plate 13 mm–6tp ellipse Brace fracture 5.0
HSS-4 BDP — fillet weld sized to capacity of plate 13 mm–9.4tp ellipse Brace fracture 4.8
HSS-5 BDP — fillet weld sized to capacity of plate 10 mm–8tp ellipse Brace fracture 5.5
HSS-6 BDP — HSS5 except fillet welds reinforced 10 mm–8tp ellipse Brace fracture 4.8
HSS-7 BDP — fillet weld sized to capacity of plate 22 mm–6tp ellipse Brace fracture 4.0
HSS-8 BDP — fillet weld sized to capacity of plate 10 mm–3tp ellipse Brace fracture 4.6
HSS-9 BDP — CJP weld 13 mm–6tp ellipse Brace fracture 3.7
HSS-10 BDP — tapered gusset–fillet welds to plate cap. 13 mm–7tp ellipse Brace fracture 4.5
HSS-11 Heavy beam — fillet welds to plate capacity 22 mm–6tp ellipse Brace fracture 2.6
HSS-12 AISC design — CJP weld 13 mm–2tp linear Brace fracture 3.5
HSS-13 BDP — CJP weld 13 mm–7tp ellipse Brace fracture 4.1
HSS-14 No net section reinf — fillet welds to pl. cap. 10 mm–8tp ellipse Brace fracture 3.9
HSS-15 BDP — min. block shear–fillet welds to pl. cap. 10 mm–6tp ellipse Brace fracture 4.1
HSS-17 BDP — tapered gusset–fillet welds to plate cap. 10 mm–9tp ellipse Brace fracture 4.9
HSS-18 BDP — bolted shear pl.–fillet welds to plate cap. 10 mm–8tp ellipse Brace fracture 4.2
HSS-20 BDP — bolted end plate 10 mm–7tp ellipse Brace fracture 4.0
HSS-21 BDP — bolted end plate 10 mm–7tp ellipse Bolt fracture 4.2
HSS-22 BDP — tapered gusset — unwelded beam flanges 10 mm–8tp ellipse Gusset tearing 4.0
WF-23 BDP — W6x25 wide flange brace 10 mm–8tp ellipse Weld fracture 5.6
HSS-24 BDP-3/8″ gusset, 6 tp elliptical 10 mm–8tp ellipse Brace fracture 4.4
HSS-25 Heavy beam — no net section reinf. — CJP weld 22 mm–6tp ellipse Brace fracture 3.3
HSS-26 Heavy beam — no net section reinf. — near fault 22 mm–6tp ellipse Net section 1.7
HSS-27 No net section reinforcement — near fault 10 mm–8tp ellipse Net section 2.5
HSS-28 BDP — tapered gusset 19 mm–2tp linear Brace fracture 4.7
TCBF1-HSS BDP — two story 10 mm–8tp ellipse Brace fracture 4.3 Avg.
TCBF1-WF BDP — two story 10 mm–8tp ellipse Brace fracture 6.2 Avg.
TCBF1-T BDP — two story — tapered gusset 20 mm–2tp linear Brace fracture 5.6 Avg.
TCBF2-HSS BDP — three story 10 mm–varies Brace fracture 3.8 Avg.
TCBF2-WF BDP — three story 10 mm–varies Brace fracture 4.9 Avg.
TCBF2-IP BDP — three story — in plane buckling 20 mm–2tp linear Brace fracture. 3.5 Avg.
1764 C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772

Fig. 4. Cyclic force–deflection behaviors; a) HSS-1: AISC reference, (b) HSS-12: AISC reference CJP weld, c) HSS-5: rectangular plate meeting BDP method, d) HSS-10: tapered plate
meeting BDP method, e) HSS-11: thick tapered plate and heavy beams, and f) WF-23: wide flange brace meeting BDP method.

• Specimen HSS-1 showed that the use of the UFM for design of the • Both rectangular and tapered gusset plates are used in practice.
interface welds is not adequate, and improved performance was Tapered gusset plates result in similar drift capacity as rectangular
achieved with interface welds designed to suppress weld fracture by gusset plates (see Fig. 4c, rectangular, and 4d, tapered), but they
ensuring the weld strength can develop the plastic capacity of the sustain more concentrated strain and deformation demands, which in
gusset plate [2,5]. Specimen HSS-5 used welds sized to achieve the turn increase the deformation demands on the gusset plate interface
capacity of the plate where Specimen HSS-1 used welds design to welds. As a result, tapered gussets may have increased weld tearing
meet the UFM. Comparing the measured responses of Specimens compared to rectangular gusset plates at large deformations.
HSS-1 and HSS-5 (Fig. 4a and c, respectively) shows that weld • Experiments show that gusset plate connections must be stiff and
damage is reduced and specimen deformation capacity is increased strong enough to develop the expected resistance and deformation
using the balanced approach. capacity of the brace, but additional strength and stiffness is
detrimental to SCBF performance. Specimen HSS-11 met current
SCBF criteria, but had thicker gussets as commonly occurs with
current design methods. Comparison of Fig. 4e (specimen with thick
gusset plate) to 4c (specimen with thinner plate meeting the BPD)
shows that the additional gusset plate stiffness and resistance
significantly reduced SCBF system deformation capacity.
• Prior research, including that conducted by the authors, shows that
rectangular HSS tubular braces achieve less deformation capacity
than other brace cross sections. For example, Specimen WF-23 was
identical to Specimen HSS-5 except that a wide flange brace was

Fig. 5. Elliptical offset to achieve end rotation [2]. Fig. 6. Midspan gusset plate buckling geometry.
C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772 1765

used; in both cases the specimens met the BDP. Comparison of The results from the experimental research programs were used to
Fig. 4c and f shows that the specimen with the wide flange brace establish the β factors. Two sets of β factors (βdesign and βactual) were
provided greater deformation capacity. computed from the test results. Design β factors, βdesign, were based
on the nominal material properties, while actual β factors, βactual,
These results show that the proposed BDP improved the ductility
were based upon actual material properties measured in coupon tests
and increased the inelastic deformation capacity of SCBF connections,
as illustrated in Eqs. (3) and (4).
while significantly delaying the undesirable failure modes.

Balanced design procedure (BDP) Pnom ½P or Puc nom


βdesign = βd = = ut ð3Þ
Rnom Rnom
The proposed BDP seeks to maximize yielding in and therefore
deformation of the system while suppressing undesirable failure modes Pactual ½P or Puc actual
βactual = βa = = ut : ð4Þ
[4]. The BDP has similarities and differences with current design Ractual Ractual
methods. Table 1 provides an overview of the relevant resistance
expressions for both the proposed BDP and the current AISC procedure.
The βdesign values are appropriate for development of the BDP,
As evident in the table, the same yield mechanisms and failure modes are
because the design process is based upon nominal material properties.
used for both design procedures. Although there are other similarities,
However, βactual is considered, because differences between βactual
there are important, fundamental differences in the proposed BDP
and βdesign illustrate the reserve capacity in the proposed procedure.
method. With both methods, the brace is sized using reduced seismic
The β factors were selected to maximize the inelastic deformation
loads. Both use the expected plastic strength of the brace (Put and Puc) to
capacity from the SCBF system, where the deformation capacity is
quantify the demand on the gusset plate connections. Slenderness limits
defined based upon the total drift range prior to initial failure obtained
and geometric requirements for all members are retained from the
from the test specimens. It is logical that the β factors will often be
current AISC Provisions [1]. The most important differences in the two
larger than the ϕ factors employed in LRFD design [5], because the
methods are that 1) the current procedure uses resistance factors to
seismic loads for the gusset plate are based upon Put and Puc, which
design and detail the gusset plate where the balance design method uses
are much larger than the factored seismic design loads used for
balance factors to balance the yield mechanisms and failure modes to
member design. The procedures and data used to establish the
maximize the drift capacity of the system, and 2) the balance factors are
selection of β factors are provided in subsequent text.
based upon the inelastic deformation provided by the mechanism rather
than predicted strength. In addition, there are differences in some of the
Determination of balance factors
resistance expressions, as shown in Table 1.
The BDP has the objective of developing a connection design
The predicted nominal yield and failure resistances, Ryield and Rfail,
methodology to balance yielding of the brace and the gusset plate and
and associated β factors for welded connections are provided in Table 1.
maximize system ductility. This is accomplished by balancing the
Their derivation is presented in this section. Although the resistance
resistances associated with the yield mechanisms, as illustrated in
equations for bolted connections are not explicitly described, the same
Eq. (1):
design principles apply. Several experiments have been completed on
bolted connections, and their performance was variable but sometimes
ðPut or Puc Þ ≤ βyield 1 Ry Ryield;1 ≤ βyield 2 Ry Ryield;2 ≤ βyield i Ry Ryield;i : ð1Þ
quite good. However, the experimental data for bolted connections are
limited and therefore actual beta values were not derived.
Eq. (2) is the failure mode balance procedure:
Primary yield mechanism
Ryieldmean = Ry Ryield b
· βfail;1 Rfail;1 b
· βfail;2 Rfail;2 and βyield b βfail : ð2Þ
The primary yield mechanism(s) for SCBF systems are tensile
In these equations, Ry is the ratio of the expected yield stress to the yielding and compressive buckling of the brace. The values of Put and
specified yield stress and Ryield and Rfail are the nominal resistance Puc are the expected tensile and buckling yield resistances, and therefore
values of the yield mechanism and failure mode in question, they serve as the mean yield resistance, Ryield mean, in Eqs. (1),(2) and
respectively. The hierarchy of yielding and prevention of undesirable (3). The general dimensions of the gusset plate (dimensions a and b of
failure modes is achieved through the use of balance factors called beta Fig. 5) are selected to be the minimum required to accommodate the
(β) factors as shown within these expressions. These β factors appear geometry of the brace, the connection length of the brace to the gusset
similar to the resistance factors, φ, used in AISC LRFD design [5], but β plate, and the elliptical clearance, rather than those recommended by
factors are based on achieving ductility and inelastic deformation UFM and other design provisions for AISC [1].
capacity rather than simply designing to achieve a specified strength
under statistically extreme loads [4]. The failure mode resistance Secondary yield mechanism: gusset plate yielding
expressions do not include the material overstrength factors, Ry,
because doing so would increase the likelihood of critical failure Controlled tensile yielding of the gusset plate is the preferred
modes and reduce the safety of the connection. secondary yield mechanism. The Whitmore width, Bw, is used to
Each yield mechanism and failure mode has its own β factor, estimate the maximum nominal stress in these connections [6]. The
which was established using measured values from specimens that method assumes a 30-degree angle relative to the length of the
achieved the target ductility and inelastic performance corresponding connection as shown in Fig. 1. Fig. 7 shows the deformation capacity
to the specified yield mechanism or failure mode. The subscripts 1, 2, achieved in past experiments as a function of the gusset plate yielding β
3, etc. reflect the preferred sequence of yielding hierarchy and factor (relative to the tensile capacity of the brace). This figure compares
suppressing of specific failure modes within the system. The β factors the empirically derived beta factors to the measured maximum
are less than or equal to 1, and they are smaller for less desirable yield deformation in order to derive all of the β factors. It should be noted
mechanisms and failure modes. For example, the preferred yield that each specimen has multiple variables, and this variability results in
mechanism of gusset plate yielding has a βyield value of 1.0 where an increase in the scatter in the test results. However, the tests are
undesirable failure modes have βfail values that are smaller than 1.0, expensive, and there are insufficient test data to permit isolation of the
because the consequence of these failures are more severe. variables and therefore all available data were used in this evaluation.
1766 C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772

Fig. 7. Design β factors for Whitmore yielding.


Fig. 9. Design β factors for brace net section fracture.

Fig. 7 shows that the inelastic deformation capacity increases as


the β factor for gusset plate yielding approaches 1.0. This is consistent where the variables are shown in the figure. This AISC procedure
with recent research results showing that yielding of the gusset typically requires significant net section reinforcement if applied to
following brace yielding maximizes the drift capacity of the system SCBF bracing.
[2]. Those results show that excess strength and stiffness of the gusset Fig. 9 shows that only three (3) specimens had net-section damage
reduces the inelastic deformation capacity of the system [2]. (two sustained net section fracture); 10 specimens did not sustain net
The βdesign values in Fig. 7 range from 0.38 to 1.1. Data points that are section fracture although they did not have net-section reinforcement
enveloped in a square exhibited yielding over more than 50% of the with β factors greater than 1.0. Net-section fracture is preceded by
surface area of the gusset plate; on average these systems achieved limited yield deformation. Further, the figure shows that larger β
greater deformation capacities than the others. The figure also shows a factors increased system deformation capacity if net-section fracture
trend where smaller β factor reduces the drift range. One specimen is avoided. Nevertheless, net-section fracture is sudden and results in
(HSS-15) had βdesign values that exceeded 1.0, but still achieved a frame complete loss of brace resistance and therefore must be avoided. The
drift range of approximately 4.1% without sustaining connection failure. tests suggest that net section fracture is more likely to occur when the
This occurs because thinner gusset plates reduce the connection brace is subjected to cycles that results in large tensile strains and very
stiffness and strain accumulations at the center of the brace, which small compressive strains which reflects the demand history that
reduces the potential for brace fracture [2]. To encourage Whitmore might be expected with near fault seismic events [23]. The data
yielding of the gusset plate connection after brace yielding, a β factor of demonstrate that the recommended β factor of 0.95 is conservative to
1.0 is recommended. mitigate brace net-section fracture.
While the AISC LRFD ϕ factor is 0.75, a user note exception
Brace net-section failure mode provided in the AISC Seismic Provisions permits the use of the
expected tensile strength (RtFu) when evaluating the net section
The reduced net area of the brace at the gusset plate joint may lead to resistance. This exception is based on engineering judgment rather
net section fracture and is an issue in seismic design [25]. Rectangular than experimental results. The Rt for ASTM A500 HSS tubes is 1.3, and
HSS tubes are commonly used as braces, and Fig. 8 illustrates the net so this exception implies an equivalent β value of 1.3 0.75 or 0.975.
section adjacent to the brace-to-gusset interface, typical net section This research generally supports this practice.
reinforcement and geometry for establishing the net effective area. The
AISC Specification [5] employs a resistance factor, ϕ, of 0.75 using the net Brace-to-gusset interface failure modes
effective area, Ae, where
Welds joining the brace to the gusset plate require consideration of
Ae = UAn ð5Þ two basic failure modes. The welds (or heat affected zone) joining the
brace to the gusset may fracture or block shear failure may occur. For the
which uses the correction factor U to account for non-uniform tensile proposed BDP, the AISC expressions for evaluating the resistance of the
stress due to shear lag, and weld and base metal were adopted, as shown in Table 1. Although this
failure mode was not observed in any test, the consequences of weld
x failure are severe and over-strength of the welds had no adverse effect
U = 1− ð6Þ
Lc on SCBF performance. Therefore, the β factor for failure of the weld in
the base metal and the filler metal is 0.75, identical to the ϕ factors used
A in AISC [1].
Agp Net Section Reinforcement
Block-shear failure may occur if the combined shear and tensile
C.G. stress demand at the brace-to-gusset interface exceeds the capacity
X
around the perimeter of the joint. Block-shear resistance is the
Lc summation of the capacity in tension (Ant), region adjacent to the end
Anb Gusset of the brace, and shear (Anv), as indicated by the resistance expression
Net Section Fracture provided in Table 1. Design for block shear frequently controls the
Considered at Section A-A A connection design with the BDP, because the procedure encourages
thinner gusset plates. If the block-shear perimeter is shortened, the
Fig. 8. Net section evaluation for rectangular HSS brace. plate thickness will be controlled by block shear. Thick gusset plates
C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772 1767

date. The resistance factor of ϕ = 0.75 is used to compute tensile


fracture. However for the BDP, a relatively conservative β factor of 0.85
was chosen to calculate the Whitmore fracture resistance based upon
the test results. Additional experimental evaluation is warranted for
bolted brace-to-gusset connections, because boltholes again impact the
potential for crack initiation in the gusset.

Gusset plate buckling

The gusset plate is checked for buckling relative to Puc of the brace.
True buckling of the gusset plate, not out-of-plane bending to support the
end rotation of the brace, concentrates the inelastic deformation and
damage to the connection preventing development of the inelastic
capacity of the brace. The Thornton method is commonly used to evaluate
gusset plate buckling [6,7]. This method uses a 30-degree angle for
establishing the area used for determination of the buckling capacity, and
a average buckling length, Lavg, of the three buckling lengths (L1, L2, L3), as
Fig. 10. Design β factors for block shear failure mode.
illustrated in Fig. 1a for a corner gussets and Fig. 6 for a midspan gussets.
An effective length coefficient of 0.65 is proposed for the corner gussets.
may be required, and the resulting plate may be too stiff and strong to For mid-span gussets, a 6tp vertical clearance zone is employed from the
encourage yielding. Several specimens had βdesign values significantly top of the brace to the beam flange, as shown in the right hand portion of
larger than the AISC ϕ factor, and block shear fractures or yield lines, Fig. 6. These clearance models often force the L1 or L3 length into the beam
both of which would suggest imminent problems, were not observed. or column web to cause a negative length as shown in Fig. 6, and those
Fig. 10 shows that ϕ = 0.75 significantly underestimates the block negative lengths are permitted to be used in calculating the average
shear resistance for welded brace-to-gusset connections, and that buckling length. The TCBF-2 experiments showed that the interior vertical
βactual is consistently smaller than βdesign value. As a result, a β factor stiffeners provide stability to the midspan gusset plate connections while
of 0.85 was conservatively chosen to suppress block-shear failure, and permitting brace end rotation, and the buckling length, L1, in Fig. 6 is
it is expected that an even larger value may be possible for welded terminated at the vertical stiffener instead of extending it to the beam
joints, however the limited test data to date does not permit this with this midspan clearance model. However, a more conservative
increase at present. Further, only a single equation considering block effective length coefficient of 1.5 is used for these midspan gussets,
shear fracture is employed, since yielding in the gusset is encouraged. because sideway buckling is possible due to the reduced edge restraint
Additional testing is required for bolted brace-to-gusset joints, provided by the bottom flange of the beam. This recommendation is
because the boltholes may facilitate crack initiation and crack growth based upon experimental results and nonlinear analysis [4,22,26]. This
for block shear failure. method may also be used with a single stiffener in the midspan gusset, but
midspan connections without vertical stiffeners are not recommended. In
Whitmore fracture failure mode either case, once the Whitmore width and effective length of the gusset
plate are established, the AISC column provisions are used to determine
Whitmore fracture must be discouraged, because gusset plate the critical buckling stress (Fcr) as noted in Table 1.
fracture causes a dramatic loss of lateral resistance. For this evaluation, The proposed BDP encourages thinner, more compact connections,
the area associated the Whitmore width was again employed, and the which usually have shorter effective lengths relative to gusset plates
failure mode was evaluated with respect to Put. Tensile fracture depends design to meet the 2tp offset suggested in [1], and buckling is less
on the effective net section and ultimate tensile strength, Fu, of the plate, likely to control their design. Significant gusset plate deformation was
as shown in Table 1. Fig. 11 shows that increasing βd values for the noted in all experiments, which is prescribed by both AISC and the
Whitmore fracture failure mode led to larger drift ranges, and Whitmore BDP design methods, to accommodate the end rotations of the
fracture did not occur in any test nor were there any indications of buckling brace. Fig. 12 shows the βdesign values computed for these
distress or imminent problems with β factors as large as 0.9. Net-section tests. None of the specimens exhibited gusset plate buckling, and
fracture of the gusset was not sustained in any of the tests conducted to specimens with larger β factors showed somewhat greater inelastic

Fig. 11. Design β factors for Whitmore fracture mode. Fig. 12. Design β factors for gusset plate buckling.
1768 C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772

deformation capacity. However, the figure shows that these tests did than rectangular gusset plates. Rectangular gussets have material
not place severe demands on the AISC buckling design provisions, and outside of the Whitmore width, which increases the weld resistance
therefore the balance factor, β, for gusset plate buckling is 0.9, which and decreases the strain concentrated at the interface [20]. Wide flange
is the same value as the ϕ factor for AISC LRFD compression member braces also cause greater damage to the gusset plate welds, because
design. wide flange braces sustain larger out-of-plane deformation demands
than HSS tubes prior to brace fracture, and this increased deformation
Interface welds increases the demands on the welds [21–23].
Fig. 13 also shows that specimens with a β value above 0.75, which
Experiments have demonstrated that welds joining the gusset plate is the value of the AISC resistance factor, may sustain excessive weld
to beams and columns are important to the ductility and deformation tearing. Further, there is not a clear increase in ductility associated
capacity of the system [18–20]. Current AISC Seismic Provisions require with larger β factors. One specimen (HSS-08) had a βd value below
that these welds be designed to develop the tensile force and the 0.75 limit, and it experienced weld tearing that was beyond
deformation capacities of the brace [1]. The UFM is commonly used to acceptable limits (weld tearing N 60%). However, this specimen was
establish the demands on the weld corresponding to Put of the brace, but designed with a very tight 3tp elliptical clearance, which does not
research shows that these welds are subjected to stresses and meet the proposed BDP elliptical design clearance. The 3tp elliptical
deformations much larger than those predicted by the UFM. Further, clearance was deemed too small a clearance for the elliptical method,
crack initiation will occur at larger inelastic deformations [18–20]. The and thus is not recommended, because the close proximity of the
UFM is an equilibrium method that was originally developed for wind brace to the interface welds dramatically increases the demands on
loading without consideration for inelastic deformation, and it was the welds. Therefore, for the 8tp elliptical clearance as suggested by
adapted to current seismic design by designing for the expected Put the BDP, a β of 0.75 is adopted for the BDP.
rather than the factored design load. The UFM does not consider the
stresses and strains caused by inelastic rotational demands due to brace Beam-column connection
end rotation and the opening and closing moments at the beam-column
intersection and can underestimate the actual demand. The beam-column connection must also be designed. This design
Cracking of the weld and the gusset plate adjacent to the welds must must consider any diaphragm forces that are brought to the braced
still be expected after large brace buckling deformations and frame story frame through this connection. The authors recommend welded-web
drifts [2]. Experiments show crack initiation at about 2% story drift welded-flange connections, because they provide better inelastic
regardless of the connection design. Experiments show that these cracks performance. However, bolted-web welded-flange and shear plate
should remain stable, because demand critical weld metal is used and (with free flanges) connections have been tested. The inelastic
this limits the rate and length of crack growth. Limited ductile tearing is deformation capacity of these connections was reduced, but the
not detrimental to the performance and permits larger inelastic drifts reduction was smaller than noted from other aspects of the gusset
under these conditions. However, complete weld fracture must be plate connection design.
prevented, and therefore the crack length must be controlled. Hence, in
the BDP, the gusset plate interface welds are designed to develop the Conclusions and design recommendations
expected plastic capacity of the plate rather than Put of the brace. This
design philosophy protects against interface weld fracture prior to brace A BDP for SCBF gusset plate connections is presented. The procedure
fracture, while allowing for yielding and inelastic deformations of the promotes ductility and inelastic deformation capacity of the SCBF
plate. To accomplish this, a new resistance equation is proposed, as system by balancing the resistances associated with the yield
shown in Table 1. This equation employs the AISC tensile strength of mechanisms and the primary failure mode, while suppressing undesir-
fillet welds but this capacity is compared to the tensile yield capacity of able failure modes to assure yielding from secondary mechanisms as
the gusset plate, rather than the tensile capacity of the brace, since the well as the primary yield mechanism, and while achieving the desired
cyclic buckling of the brace induces cyclic yielding as the gusset plate failure mode. A coordinated testing program has shown that the BDP can
deforms due to brace buckling end rotation and this demand can result greatly increase the deformation capacity of an SCBF system [2]. Overall,
in weld cracking and failure. SCBFs with rectangular gusset plates designed using the BDP showed an
Fig. 13 shows the measured inelastic deformation capacity of SCBF average 46% increase in drift range when compared with similar
specimens as a function of the βd for the interface welds. Several key specimens designed to meet the current SCBF specification. The benefit
observations can be made from this figure and observations from prior shown with tapered gusset plates was significant but smaller, because
research [2,18–23]. Tapered gusset plates suffer greater crack damage tapered gusset plates have less reserve capacity and sustain greater
damage than rectangular gussets at a given drift level.
The resistance equations and balance factors (β factors) for the BDP
were presented (see Table 1) and verified using experimental results.
Values for the β factors were developed to encourage gusset plate
yielding by ensuring the plate was not overdesigned but still has
adequate resistance to restrict undesirable failure modes. The values of
the β factor and the current AISC ϕ factors are identical for gusset plate
buckling and the interface welds. However, the design philosophies and
resistance expressions are different in that the BDP sizes the welds
joining the gusset plate to the beam and column to develop the expected
plastic capacity of the plate rather than for Put of the brace by the UFM, as
is required for the current specification. A β factor of 1.0 is
recommended for Whitmore yielding, and 0.85 is recommended for
the gusset plate fracture limit state. A factor of 0.95 is recommended for
net-section fracture of the brace, and a value of 0.85 is recommended for
block shear. These values are larger than the resistance factors currently
used in SCBF gusset plate design, and they are recommended because
Fig. 13. Design β factors for interface weld failure mode. they result in increased system ductility while preventing undesirable
C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772 1769

failure modes. The β factor of 0.95 for designing brace net-section Appendix A. Design example
reinforcement provides a rational basis for the 0.9 exception provided
for the ϕ factor in current AISC seismic design, and this value is also A design example is presented to illustrate the proposed BDP. As
adopted as part of the BDP [1]. An additional yet equally important part with current design methods, the structural members (brace, beam
of the design procedure is the clearance models for corner and midspan and column) are designed to resist factored loads. The illustration
gusset plats. Two clearance models are proposed: 8tp elliptical and 6tp employs an HSS200x200x12 (HSS8x8x1/2) A500 Grade C rectangular
vertical clearance gusset plate clearances for corner and middle gusset tube for the brace. The brace is oriented at 40° to the horizontal, and it
plate, respectively, because they lead to thinner, more compact gusset is connected to the W24x162 beam and the W14x370 column with a
plates, and permit increased deformation capacity from the system. corner gusset plate as illustrated in Fig. 14a. The beam and column
These clearances delay initiation of interface weld tearing, and reduce members are A992 sections, and plates used in the example are A572
secondary damage to the beam and column and improve construct- grade 50 steel. The brace is connected to the framing elements with
ability of the system. The resulting BDP will improve the deformation two identical corner gusset plates as shown in Fig. 14a. The final
capacity and predictability of the seismic performance of SCBFs. gusset plate design for the BDP is presented in Fig. 14b. The associated
resistance values for each yield mechanism and failure mode are
presented in Table 3. The bold text in Table 3 identifies the controlling
Symbol definitions design criteria for that connection.
Ag and Agp Gross cross-sectional brace area of brace and reinforcing The design procedure is as follows:
plate, respectively
1. Establish the expected yield capacity of the brace. The properties of
Agv Gross area of brace in shear
the brace are Ag = 8653 mm2 (13.5 in2) and rx = 75.9 mm (3.04 in).
Anb Net area of brace
The expected tensile capacity of the brace is: Put = RyFy Ag = 1.4
Bw Whitmore width
(317.2) 8653 = 3843 kN (864 kips).
E Modulus of elasticity 2. Establish the brace-to-gusset connection. Here, a slotted connection
Fcr Critical buckling stress is used with four (4) fillet welds, as shown in Fig. 14b. With E-70 fillet
Fe Euler buckling stress welds and a weld thickness equal to the thickness of the tube or
FEXX Electrode strength 12 mm, the connection length, Lc, is computed using the resistance
Fu and Fy Yield and ultimate strength of steel expression from Table 1:
K Effective length coefficient
L Brace length Put
Lc = ðA:1Þ
Lavg Average buckling length βð0:6ÞFEXX NW ð:707Þw2
Lc Brace-to-gusset connection length 3843
L1, L2, L3 Buckling lengths as defined in Figs. 1a and 5 = = 522 mmð20:6 inÞ
0:75ð0:6Þ483ð4Þ:707ð12Þ
Ns Number of shear planes
Nw Number of welds
Puc and Put Maximum expected capacity of brace in compression and 3. A check of the brace base material is also made using the
tension, respectively connection length of 522 mm: Put b 0.75 (0.6) Fu Ns Lc tf = 0.75
Rn Nominal resistance (0.6) 399.9 (4) 522 (12) = 4509 kN (1013 kips), which is found to
Rt and Ry Ratio of the expected to minimum specified ultimate satisfy the inequality.
tensile and yield stress of steel 4. Size the plate thickness to permit plate yielding after brace yielding
U AISC Shear lag factor and restrict tensile rupture and block shear. The 522-mm
rx Radius of gyration about x-axis connection length is used to establish the Whitmore width, Bw,
tf Brace wall thickness i.e., Bw = 200 + 2 (522) tan30 o = 803 mm, The yield, tensile
tp Gusset plate thickness rupture and block shear requirements establish minimum gusset
w1 Interface weld leg size thickness, tp, follow:
w2 Brace-to-gusset weld leg size Put 3843
x̅ Connection eccentricity tp ≥ = = 12:62 mm ðA:2Þ
βRy Fy Ag 1:0ð1:1Þ803ð345Þ
φ Resistance factor
β Balance factor
βa and βd Actual and design balance factors, respectively Put 3843
tp ≥ = = 12:57 mm ðA:3Þ
βFu Bw 0:85ð448Þ803

Acknowledgments
Put
tp ≤   ðA:4Þ
This research was funded by the National Science Foundation β 0:6Fu Agv + Ubs Fu B
under grants CMS-0301792, Performance-Based Design of Concentri-
cally Braced Frames and CMS-0619161, NEESR-SG International 3843
≥ = 12:21 mm
Hybrid Simulation of Tomorrow's Braced Frames. Supplemental 0:85f0:6ð448Þ2ð522Þ + 1ð448Þ200Þ
funding was provided by AISC. International testing, fabrication and
collaboration were conducted at the NCREE Laboratory in Taiwan with
Dr. Keh-Chyuan Tsai, Laboratory Director. Frequent advice and Tensile yielding controls and a 13-mm thickness is chosen.
guidance were provided through the research Advisory Group, 5. The 8tp elliptical clearance illustrated in Fig. 4 is used to size the
composed of Tom Schlafly (AISC), Tim Fraser (CANRON), Walterio plan dimensions of the plate. This clearance can be derived
Lopez (Rutherford and Chekene), Larry Muir (CIVES) and Rafael graphically or by the approximate theoretical method derived in
Sabelli (Walter P. Moore). The advice and financial support of these [2]. The gusset has a length of a = 1038 mm adjacent to the beam
individuals and institutions are greatly appreciated. and b = 737 mm adjacent to the column.
1770 C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772

a 6.12 m
b

1028 13 mm Gusset Plate 635


40°

40° 12
12
737
HSS200x200x12 522
Brace
126
64 HSS200x200x12
5.15 m Brace, L=6156 mm
104 14
14
4 mm
Reinf. Plate

457

c d
1947 635 1004 17 mm Gusset Plate 635
40° 34

40° 56 12
28 mm Gusset 12
Plate 732
522
1382 12 522
12
126
HSS200x200x12
30° Brace, L=5637 mm
HSS200x200x12 64
Brace, L=4149 mm
64 7
8 7 16 mm
8 126
Reinf. Plate
17 mm Reinf. Plate
457

457

Fig. 14. Design example; a) frame geometry, b) rectangular gusset by BDP, c) rectangular gusset by current SCBF, and d) tapered gusset by current SCBF.

6. The buckling capacity of the brace can now be established. Using Puc = 1:1Ry Fcr Ag = 1:1ð1:4Þð171Þð8653Þ = 2275 kNð511 kipsÞ ðA:7Þ
the gusset plate geometry and frame the length of the brace, L, can
be established as shown in Fig. 14a.

π2 E π2 ð200; 000Þ 7. The gusset plate must have adequate buckling capacity to resist
Fe =  2 =   = 300 MPað43:2 ksiÞ ðA:5Þ this buckling load. The average length of the gusset is defined using
KL 1ð6156Þ 2
the Thornton method, as illustrated in Fig. 2a.
r 75:9
 Ry Fy
  
1:4ð317Þ
L1 + L2 + L3
Fcr = 0:658 Fe
Fy = 0:658 300 317 = 171 MPað24:6 ksiÞ ðA:6Þ Lavg = = 222 mmð8:75 inÞ: ðA:8Þ
3

Table 3
Final connection resistances.

BDP (rect. gusset) AISC design (rect. gusset) AISC design (tapered gusset)
βRn (kN) φRn (kN) φRn (kN)

Put 3843 3843 3843


Puc 2275 3190 2515
Whitmore yielding 3957 6973 4234
Whitmore fracture 3976 7556 4587
Block shear 4093 6417 3896
Gusset plate buckling 2907 3428 3144
Net section fracture 3906 3883 3879
Gusset plate-to-brace welds 3847 3847 3847
Gusset plate-to-brace base material 4509 4509 4509
C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772 1771

The buckling capacity of the gusset plate is compared with the than the capacity of the gusset plate, and research has shown that this
plastic bucking strength of the brace. The results show that the 13- may lead to premature weld fracture [2]. The difference in the weld
mm plate is sufficient. size is notable; 7 mm for the AISC-compliant design; 14-mm for the
weld design to meet the BDP. Differences are also notable in the net
KLavg 0:65ð222Þ section reinforcement, for which the AISC-compliant net section
= pffiffiffiffiffiffi = 38:5 ðA:9Þ
r 13 = 12 reinforcement is approximately 2.5 times the thicker than the BDP.
The plan geometry of the tapered gusset plate in Fig. 14d is
π2 E π2 ð200; 000Þ comparable to the gusset in Fig. 14b, but the welds joining the gusset
Fe =  2 = = 1332 MPað191:6 ksiÞ ðA:10Þ
KLavg 38:52 to the beam and column are sized to the expected tensile capacity of
r the brace rather than the plastic capacity of the gusset, and premature
weld fracture may occur. The tapered gusset is thicker and provides
 
345 greater resistance to brace end rotation. The current SCBF gusset
Puc ≤ βðBw Þtp ðFcr Þ = 0:9ð803Þ13 0:6581332 345 = 2907kN ð653kipsÞ
designs result in shortened brace lengths and large connections in
ðA:11Þ which the gusset plate thickness must be increased to satisfy the
buckling limit state. These factors result in stiffer and less ductile
8. The next step is to design the interface welds, i.e., the welds joining connections [2]. It should be noted that edge buckling or edge
the gusset plate to the beam and column, to develop the plastic stiffeners were not considered in these AISC or BDP designs, because
capacity of the gusset plate. Therefore, prior research shows that the measured buckling capacity of gusset
plate connections varies between 0.25 and 8 times the capacity
Ry Fy tp predicted for edge buckling [27].
w1 ≥ ðA:12Þ
2ð1:2Þβð0:6ÞFEXX ð0:707Þ

1:1ð345Þ13
= = 13:37 mm References
2ð1:2Þð0:75Þð0:6Þ483ð0:707Þ
[1] American Institute of Steel Construction (AISC). Seismic provisions for structural
The design specifics 14-mm fillet welds on each face of the plate steel buildings. AISC/ANSI Standard 341-08; 2005. Chicago.
[2] Lehman DE, Roeder CW, Herman D, Johnson S, Kotulka B. Improved seismic
over the full length of the interface (Fig. 14). performance of gusset plate connections. Journal of Structural Engineering, ASCE
9. The final step is to design the net section reinforcement at the tip of 2008;134(6):890–901.
the slot in the HSS tube. This reinforcement will be placed as a flat [3] Prahuangsit D, Goel SC, Hanson RD. Axial hysteresis behavior with end restraints.
Journal of the Structural Division ASCE 1978;104(6):883–96.
plat attached to each flange as shown in Fig. 14b. The workable flat [4] Roeder, C.W., Lehman, D.E., and Yoo, J-H. “Improved seismic design of steel frame
area on the top of the brace, upon which the net section connections.” International Journal of Steel Structures, Korean Society of Steel
reinforcement is place including the dimension for the fillet Construction, 5(2), 141–153, Seoul.
[5] American Institute of Steel Construction (AISC). Steel Construction Manual13th
welds, is approximately 146-mm wide. Therefore, a 126-mm
Ed. ; 2005. Chicago.
wide plate will be used with 10 mm free on each side to place fillet [6] Whitmore RE. Experimental investigation of stresses in gusset plates, a thesis
welds. The slot in the brace will be cut 5 mm thicker than the submitted in partial fulfillment of the MS degree. Knoxville, Tennessee: University.
of Tennessee; 1952. MS.
gusset plate thickness for ease of erection, and the length of the
[7] Thornton WA. On the analysis and design of bracing connections. Proceedings,
reinforcement will be taken as the same length as the brace-to- 1991 National Steel Construction Conference, AISC, Section, 26; 1991. p. 1–33.
gusset connection (522 mm). [8] Bjorhovde, R., and Chakrabarti, S.K. Test of full size gusset plate connections,
Journal of Structural Engineering, ASCE, 111(3), 667–684.
2 [9] Brown VLS. Stability of gusseted connections in steel structures, a dissertation
Anb = 8653−2ð13 + 5Þ12 = 8222 mm ðA:13Þ submitted in partial fulfillment of the PhD. Department. of Civil Engineering.
Delaware: Univ. of Delaware; 1998.
2 [10] Cheng JJR, Yam MCH, Hu SZ. Elastic buckling strength of gusset plate connections.
Agp = 126ð4Þ2 = 1008 mm ðA:14Þ
Journal of Structural Engineering, ASCE 1994;120(2).
[11] Grondin GY, Nast TE, Cheng JJR. Strength and stability of corner gusset plates
x 67:9 under cyclic loading. Proceedings, Annual Technical Session and Meeting.
U = 1− = 1− = 0:87 ðA:15Þ Structural Stability Research Council; 2000.
L 522
[12] Hu SZ, Cheng JJR. Compressive behavior of gusset plate connections. Structural
  Engineering Rep. No. 153. Canada: Univ. of Alberta; 1987.
Put = 3843 ≤ βU Rtb Fub Anp + Fup Agp ðA:16Þ [13] Rabinovitch JS, Cheng JJR. Cyclic behavior of steel gusset plate connections.
Structural Engineering Rep. No. 191. Canada: University of Alberta; 1993.
= 0:95ð0:87Þ½1:3ð400Þ8221 + 448ð1008Þ = 3906kN [14] Yam MCH. Compressive behavior and strength of steel gusset plate connections, a
thesis submitted in partial fulfillment of Doctor of Philosophy degree. Canada:
University of Alberta; 1994.
The condition is met, and the 4 mm × 126 mm reinforcing plate is [15] Yam MCH, Cheng JJR. Behavior and design of gusset plate connections in
adequate for net section reinforcing. compression. Journal of Constructional Steel Research 2002;58(5–8):1143–59
Elsevier.
To serve as a comparison, an AISC-designed rectangular and [16] Tremblay R, Archambault RM-H, Filiatrault A. “Seismic response of concentrically
tapered gusset plates are also presented in Fig. 14c and d. The braced steel frames made with rectangular hollow bracing members”, ASCE.
Journal of Structural Engineering 2003;129(12):1626–37.
calculations for these SCBF designs are not provided but typical results [17] Uriz P, Mahin S. Seismic performance of concentrically braced steel frame
are tabulated in Table 3. The balanced design rectangular gusset is buildings. Proceedings, 13th World Conf. on Earthquake Engineering; 2004. Paper
more compact than a comparable rectangular plate designed by the 1639.
[18] Johnson S. Improved seismic performance of special concentrically braced frames,
BDP. More importantly, the connection design is governed by the a thesis submitted in partial fulfillment of the MS degree. Seattle: University. of
Whitmore yielding limit state, which encourages gusset plate yielding Washington; 2005.
and is beneficial at increasing system deformation capacity. Fig. 14c [19] Herman D. Further improvements on and understanding of SCBF systems, a thesis
submitted in partial fulfillment of the MS degree. Seattle: University of Washington;
shows that the current SCBF criteria results in extremely large and
2006.
uneconomical rectangular gusset plate because of the geometric [20] Kotulka BA. Analysis for a design guide on gusset plates used in special
constraints imposed by the UFM and the 2tp linear clearance model. concentrically braced frames, a thesis submitted in partial fulfillment of the MS
The thickness of the plate is almost twice that of the plate designed by degree. Seattle: University of Washington; 2007.
[21] Clark KA. Experimental performance of multi-story X-brace systems, a thesis
the BDP. Further, the welds joining the gusset to be beam and column submitted in partial fulfillment of the MS degree. Seattle: University of Washington;
are sized to develop the expected tensile resistance of the brace rather 2009.
1772 C.W. Roeder et al. / Journal of Constructional Steel Research 67 (2011) 1760–1772

[22] Lumpkin EJ. Enhanced seismic performance of multi-story special concentrically [26] Roeder, C.W., Lehman, D.E., Clark, K., Powell, J., Yoo, J-H, Tsai, K-C, Lin, C-H, and
brace frames using a balanced design procedure, a thesis submitted in partial Wei, C-Y. Influence of gusset plate connection and braces on seismic performance
fulfillment of the MS degree. Seattle: University of Washington; 2009. of X-braced frames, Earthquake Engineering and Structural Dynamics, Vol. 40, No. 4,
[23] Powell JA. Evaluation of special concentrically braced frames for improved seismic pgs 355–74, Wiley.
performance and constructability, a thesis submitted in partial fulfillment of the [27] Roeder CW, Lehman DE, Yoo J-H. Performance based seismic design of braced
MS degree. Seattle: University of Washington; 2010. frame connections. Proceedings, 7th Pacific Structural Steel Conference, Long
[24] Applied Technology Council. Guidelines for Testing Steel Components, ATC-24. Beach, CA; March 24–27 2004.
Redwood City CA: Applied Technology Council; 1992. 57 pp.
[25] Fell BV, Kanvinde AM, Deierlein GG, Myers AT, Fu X. Buckling and fracture of
concentric braces under inelastic cyclic loading. Structural Steel Educational Council;
2006.

You might also like