Engineering Structures 32 (2010) 2034–2048

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

I-beam to box–column connection by a vertical plate passing through the column

Seyed Rasoul Mirghaderi, Shahabeddin Torabian ∗ , Farhad Keshavarzi
School of Civil Engineering, University of Tehran, P.O. Box 11365-4563, Tehran, Iran

article info abstract

Article history: Box–columns are convenient structural members for moment resisting frames, but providing a proper
Received 25 February 2009 beam to box–column connection is still under investigation due to the closed shape of box–columns
Received in revised form and difficult access to the column inside for developing a reliable load path. To resolve the problem,
26 February 2010
a new easy-to-fabricate moment connection is proposed for connecting I-beams to box–columns. The
Accepted 2 March 2010
Available online 30 March 2010
connection consists of a vertical plate passing through the column and welded to the column flanges,
and beams connected to the plate named the Through Plate. In this paper, the force transfer mechanism
Keywords:
of the connection is described and a design method to determine the dimensions of the through plate
Box–column and other pertinent parts is proposed. An experimental program consisting of two relatively identical
Internal continuity plate cyclically loaded specimens was conducted to evaluate the seismic performance of the connection. The
Through plate results showed that the specimens reached at least 0.06 rad total story drift before experiencing strength
Experimental program degradation. Moreover, the connection behavior for a range of parameters including member sizes and
Finite element analysis through plate dimensions were investigated by a verified finite element model to examine the proposed
Seismic performance basic design criteria for the connection.
© 2010 Elsevier Ltd. All rights reserved.

1. Introduction under investigation. Extensive studies have been carried out and
several new connection details have been proposed for the connec-
Box–columns are frequently employed as a part of special mo- tion of I-beams to wide flange columns since the 1994 Northridge
ment resisting frames in regions of high seismic risk such as most earthquake, but limited research for the connection of I-beams to
Asian countries and Iran. Cold-formed hollow sections are often box–columns has been conducted. In spite of the fact that some
used for low and medium rise buildings and built-up sections differences exist between beam to box–column connections and
made up of four plates welded together are used for heavier mo- beam to wide flange column connections due to connection of
ment frames. This practice is common in Japanese steel construc- the beam web to the out-of-plane component of box–column
tion [1]. Large bending stiffness and strength about any axis make flange and also different panel zone behavior in the presence of
box sections more efficient than conventional wide flange sections two webs, application of similar connection details with empha-
in flexural and compression members such as beam–columns. In sis on the insertion of continuity plates inside the box–columns
addition, large torsional stiffness without relaying upon warping at beam flange levels is generally recommended for the connec-
behavior decreases the required lateral bracing of the column and tions to box–columns [3]. Installing internal continuity plates is a
reduces the strength deteriorations which are commonly initi- difficult fabrication task and also a costly procedure, and there-
ated from column rotation such as lateral torsional buckling of the fore researchers have tried to improve this type of connection
connection [2,3]. High ductility and energy dissipation capability not only by improving the connection details in the presence of
because of low sensitivity to local buckling and possessing post- internal continuity plates, but also by avoiding internal continu-
buckling strength make box sections appropriate for columns of ity plates and providing new load paths via external features.
seismic moment resisting frames. Box–columns provide the most Rib-reinforced welded connections [4], column-tree connection
efficient use of material and minimize costs for painting and other with improved details including no weld access hole detail and
surface maintenance requirements [2]. widened flange of the stub beam detail [5], and horizontal hunch
Although box–columns potentially have the above-mentioned connection [6] recently addressed the improvement of connec-
advantages, finding a proper beam to column connection is still tions with internal continuity plates. Moreover, external T-angle
or triangular plate stiffeners for box–columns [7–10], T-stiffeners
for CFT columns [11–13], stiffening plates around the box–column
∗ Corresponding author. Tel.: +98 21 4405 3876; fax: +98 21 8874 1819.
[14,15], special shaped high-strength bolts [16], high-strength
E-mail addresses: rmirghaderi@ut.ac.ir (S.R. Mirghaderi), torabian@ut.ac.ir, blind bolts [17], and bolted moment end plate connections [18,19]
shtorabian@gmail.com (S. Torabian), farhadk72@gmail.com (F. Keshavarzi). were employed to provide new load paths for beam to box–column
0141-0296/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2010.03.002
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2035

Nomenclature θ Through plate corner element angle.

θs Relative rotation between the beam end and the
a Width of the through plate corner element. column.
Ab Beam area. θp Total plastic rotation of the connection.
Afe Beam flange effective net area. σc Axial stress of the through plate corner element.
Afn Beam flange net area including the attached cover τyp Shear yield strength of the through plate.
plates. 1γ Increased virtual shear deformation.
b Height of the through plate corner element. 1σc Increase in axial stress of the through plate corner
bf Beam flange width.
element.
E Young’s modulus of elasticity.
1V Increase in the through plate panel zone shear
EI Bending rigidity of the beam.
capacity.
F1,2 Opposite forces on column flanges.
Ψ Ratio of in-plane to out-of-plane portions of the
Fuf The specified minimum tensile strength of the beam
induced moment.
flange.
Fy Specified yield strength.
Fyp Minimum specified yield strength of the through connections by means of external features and eliminate internal
plate. continuity plates.
db Beam depth. Correspondingly, to achieve the desired performance of box–
dc Column depth. columns, this paper proposed Through Plate moment connection
ha Through plate depth at the critical section. to eliminate the continuity plates and facilitate the connection
Ks Stiffness of the connection at service load. construction. The force transfer mechanism and design method of
L Connection length of the cover plate to beam flange. the connection were described, and afterwards the seismic per-
Lb Beam length. formance was investigated both experimentally and analytically,
Lp Distance from the beam end to the plastic hinge. using finite element models. During the experimental program,
Ma Moment at the through plate critical section. two relatively identical specimens were subjected to cyclic load-
Mc Total moment at the column centerline. ing to examine the seismic behavior of the proposed connection.
Mcf Moment induced on each column flange. Thereafter, the results of the nonlinear finite element analysis of
Me Moment at the end of the through plate. the test specimens were verified against the experimental results
Mf Moment transmitted through the column flanges. and the models were used to investigate the connection behav-
Mi Beam moment at the location i. ior more precisely. Finally, the proposed method for design of the
Mp Plastic moment at the through plate critical section. connection was examined by performing a set of parametric fi-
Mp0 Reduced plastic moment due to shear force at the nite element analyses, and the connection behavior was discussed
through plate critical section. thereafter.
Mpr Beam probable plastic hinge moment.
Ms Moment of connection joint. 2. Proposed connection
Pc (x) Axial force per unit length along the horizontal side
of the corner element. 2.1. Connection configuration
Rt A coefficient accounting for expected tensile strength.
Ry A coefficient accounting for the difference between The through plate moment connection between I-beam and
the minimum specified yield strength Fy and the box–column is schematically proposed in Fig. 1. The main part of
expected yield strengths. this new connection is a vertical plate passing through the column
Tfu Tensile rupture capacity of the beam flange. faces and called the Through Plate in this study. This plate is as-
tf Beam flange thickness. sumed to answer all the requirements for a rigid moment connec-
tp Through plate thickness. tion between the beams and the box–columns.
tw Beam web thickness. As shown in Fig. 1(b), the through plate passes through two
U Shear lag factor. aligned slots on the column flanges and is connected to them. The
V Shear force at the through plate critical section, beam connection to the through plate is provided by longitudinal
shear capacity of panel zone region. connection of the flanges to both sides of the through plate, after
Vb Beam shear force. trimming the beam web in this region, and also two web connec-
Vp Plastic shear capacity at the through plate critical tion plates on both sides of the through plate. Moreover, two cover
section. plates are welded to both beam flanges before beam installation to
Vpz Design shear force of the through plate panel zone. avoid overhead welding in the bottom flange.
Vu Shear capacity of the through plate panel zone. As shown in Fig. 1(c), the through plate can also be used in or-
w Effective width of the corner element. thogonal moment frames by a cruciform through plate which can
Wext External work done by the shear capacity of panel help the column plates to be assembled more easily and precisely.
zone region. The step-by-step fabrication process of the connection is shown in
Wint Internal work. Fig. 1(d) by assembling sequence number. Box–columns including
x Distance between the through plate and the cen- the through plate can be fabricated in the shop and thereafter the
troid of flange net area, Afn . beams can be installed in the field.
xi Distance from the beam end to the point where Mi The longitudinal connections between column flanges and
is to be calculated. through plate transfer all demands of the through plate to the box–
Zb Beam plastic modulus. column. The beam flange connection transfers the flange tension
εc Axial strain of the through plate corner elements. and compression forces, and the beam web connection transfers
γ Virtual shear deformation of the through plate panel the beam shear forces and also the portion of the beam web mo-
zone. ment to the through plate. The cover plates are provided to over-
come the probable fracture at the net section of the beam flanges.
2036 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

Fig. 1. The through plate moment connection: (a) Planar-form Configuration, (b) Planar-form Assembly, (c) Orthogonal-form configuration, (d) Step-by-step fabrication
process (by assembly sequence number).

By the through plate connection, subsequent advantages and char- lateral instabilities. However, weak axis of the through plate can
acteristics are expected based on the connection configuration, be additionally reinforced by horizontal stiffeners between the col-
• In this connection, conventional components such as weld ac- umn face and the through plate.
cess holes, cover plates, hunches and triangular plate stiffen-
ers are replaced by the through plate in the fabrication process. 2.2. Connection force transfer mechanism
Moreover, as installation of internal diaphragms in conven-
tional beam to box–column connections is inherently difficult It is clear based on the connection configuration that the con-
because of internal welds [20,21], use of the through plate nection load transfer mechanisms are quite different from the
with accessible welds eliminates this costly and labor intensive
load transfer mechanisms in the conventional connections with
process.
internal continuity plates. However, the connection load transfer
• A continuous access to the column inside has been provided for
mechanisms can be explained clearly based on the behavior of
filling the column with concrete.
• In conventional beam to box–column connections, the beam connection components.
web connection is not so effective in transferring the bending At the expected plastic hinge region, right after the cover plates,
moment but in this connection the portion of the beam web the plastic hinge moment and the plastic shear force are developed
moment is transferred to the column by means of the through and transferred to the through plate by means of flange and web
plate. connections. As depicted in Fig. 2(a), all the forces resulting from
• The proposed connection is applicable for both cold-formed and the beam plastic hinge are in the through plate plane. These in-
built-up closed shaped sections. duced in-plane forces tend to rotate the through plate relative to
• The flexural stiffness and strength of beam and column sec- the column but two mechanisms, including out-of-plane and in-
tions are increased in the connection region about the beam’s plane resistance of the column flanges, are resisting against the
stronger axis. relative rotation and providing the load path for transferring the
It should be noted that the flexural stiffness and strength about moment of the beam to the box–column. As shown in Fig. 2(b),
the beam weak axis may decrease at the column face, therefore ad- the out-of-plane resisting mechanism provides a force distribu-
equate beam and column lateral braces should be applied accord- tion perpendicular to the column faces and the in-plane resist-
ing to code provisions as provided in this study to prevent beam ing mechanism provides two sets of opposite in-plane forces along
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2037

a b

Fig. 2. Connection force transfer mechanism: (a) Connection mechanism, (b) In-plane and out-of-plane resisting forces.

the column flanges. The stiffness of the out-of-plane mechanism 3.1. Test specimens
mainly results from the flexural stiffness of the box–column walls,
whereas the in-plane mechanism stiffness is basically results from Two 2/3-scale relatively identical specimens demonstrating an
the axial stiffness of the column walls. As box–columns commonly interior connection of a moment resisting frame were assumed
have thin wall thicknesses, the resulting out-of-plane stiffness is to be tested. As shown in Fig. 3, each specimen consisted of a
relatively small in comparison to the stiffness resulting from the built-up box–column and two I-shaped hot-rolled beams which
in-plane stiffness. Hence, a factor Ψ = (Mc − Mf )/Mc , is consid- are connected to opposite sides of the column. The cruciform shape
ered to calculate the in-plane and the out-of-plane portions of the of the specimen was concluded from the deflected shape of the
induced moments, where Mc is the total moment at the column frames under lateral loads. Due to the moment distribution in such
centerline, and Mf is the moment transmitted through the column a frame, inflection points can be assumed near the mid-span of
flanges. Ψ values are estimated to be between 0.8 and 0.9, in Sec- the beams and mid-height of the columns. By this assumption,
tion 4.3. Therefore, the opposite forces on the column flanges F1,2 , the subassembly was extracted from the inflection point at mid-
points of the members with the hinged end condition. The total col-
and the moment induced on each column flanges Mcf can be cal-
umn height was 2150 mm and the beam span was 3290 mm. Both
culated as follows,
columns of the specimens had 200×160×10 rectangular cross sec-
±Ψ tion and beam sections were 180-mm-depth I-sections which were
P
Mpr X
F1,2 = + Vb (1) commonly used in Iranian practice. A light IPE section (IPE180-A)
dc
was used for the beams of the first specimen and IPE 180 (Euro-
1 X pean standard section) was used for the second one. The beam and
Mcf = (1 − Ψ ) Mpr (2)
2 column sizes were generally governed by the capacity and stroke
of the actuator. Also, the specimens were proportioned to remain
where Mpr is the probable plastic hinge moment of the beam pro-
relatively elastic until 1% drift ratio. Moreover, the strong-column
jected to the column centerline, dc is the column depth, and Vb is weak-beam criterion was satisfied to ensure plastic hinge forming
the beam shear force. Based on this model, the plastic hinge mo- at the beam ends. Beams, columns and the other connection plates
ment of the beams, after transferring to the through plate plane, is were all A36 steel. Standard coupons were extracted from the pro-
converted into a couple of opposite forces on the column flanges files and plates used for construction of the test specimens. The ma-
and also two column flange moments, and the resultant of these terial properties resulting from the tensile test are given in Table 1.
forces develops the bending moment in the column. From the construction point of view, the box–columns were
fabricated by welding four plates by means of fillet welds. Prior
3. Experimental program to the column construction, the column flanges were slotted and
beveled in a limited zone on the flange centerline to pass the
through plate. The through plate was prepared beforehand and
The behavior of the moment connections under severe cyclic crossed throughout the slots and welded to the column flanges
loading, particularly in regard to the initiation and propagation of by PJP groove welds of the column flange thickness groove depth.
fracture, cannot be reliably predicted by analytical means alone. The prepared beam (as shown in Fig. 1(b)) was welded to the
Consequently, the satisfactory performance of connections must through plate with full penetration single-bevel-groove welds and
be confirmed by laboratory testing [22]. Since the important issues the cover plates were fillet welded thereafter on the beam flange.
in seismic design are ductility, hysteretic energy dissipation and The web plates were installed on the through plate and the beam,
failure modes, two relatively identical test specimens have been and then welded by circumferential fillet welds.
designed to be tested under incremental cyclic loading protocol
to well capture and monitor the connection seismic behavior and 3.2. Connection design method
related issues. Therefore, the testing procedure and test results
for global and local seismic behavior of the test specimens are The proposed connection should be designed for the demands
discussed in the following sections. of the beam plastic hinge, based on the knowledge about the
2038 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

Fig. 3. Specifications and dimensions of the test specimens.

Table 1
Mechanical properties of the tested steel coupons.
Test spec. Member Coupon Yield strength (MPa) Tensile strength (MPa) Elongation (%)

1 Beam (IPE 180-A) Web 361.0 423.0 28

1 Beam (IPE 180-A) Flange 303.0 411.0 35
2 Beam (IPE 180) Web 345.3 483.0 32
2 Beam (IPE 180) Flange 303.6 469.3 25
1&2 Column (Built up 200 × 160 × 10) Web & Flange plates 253.0 370.0 41
1&2 Through plate Plate 244.2 374.3 30

load transfer mechanisms (see Section 2.2). Hence, a set of design Based upon AISC, Ry for A36 steel hot-rolled structural shapes and
checks are arranged for the design of the through plate and other plates is equal to 1.5 and 1.3, respectively [22]. The multiplier 1.1
pertinent components. The connection design criteria are generally accounts for the peak connection strength including limited strain
based on the AISC Seismic Provision recommendations [22]. hardening and other types of over strength.
To obtain the bending moment demand at the critical sections Consequently, four main steps in designing the through plate
of the through plate, a moment diagram resulted from static equi- connection are considered as follows,
librium has been considered as shown in Fig. 4. The bending mo- Step 1. Design of the through plate net section at the end of the
ment value can be calculated along the beam length based on the beam
following equation: The shear and bending moment capacity of the through plate
xi at critical section should be checked, where the beam ends and
Mi = Mpr (3) the through plate only has to provide the load path. Although the
Lp
interaction of shear and bending is not so critical in controlling
where Mi is the moment at location i as illustrated in Fig. 4, Lp is the I-sections, the effect of shear force on plastic moment of rectangu-
distance from the beam end to the plastic hinge moment, xi is the lar sections cannot be ignored. Hence, based on either the Tresca
distance from the beam end to the point where Mi is to be calcu- or von Mises yield criteria, the following lower bound interaction
lated, and Mpr = 1.1Ry Zb Fy is the probable plastic hinge moment. formula was given by Horne [23].
Ry is multiplied to account for the difference between the mini- 2
Mp0

mum specified yield strength Fy and the expected yield strengths, 3 V 2
and also Zb is the beam plastic modulus at the plastic hinge region.
+ = 1, V ≤ Vp (4)
Mp 4 Vp 3
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2039

Table 2
Design summary of the test specimens.
Beam section properties
Spec Section Ab (m2 × 10−4 ) db (m × 10−2 ) bf (m × 10−2 ) tf (m × 10−2 ) tw (m × 10−2 ) Zb (m3 × 10−6 )

1 IPE180-A 21.6 18 8.8 0.7 0.5 147

2 IPE180 23.9 18 9.1 0.8 0.53 166

(a): Step 1. Design of the through plate net section at the end of the beam
Spec Zb (m3 × 10−6 ) Fy (MPa) xpr (m) xa (m) Mpr (kN m) Ma (kN m) Vb (kN) tp (m × 10−2 ) ha (m × 10−2 ) Mp (kN m) Vp (kN) Mp0 (kN m) Ma /Mp0

1 147 240 1.295 1.520 58.2 68.3 45.0 1.50 30.0 81.0 623.2 80.7 0.85
2 166 240 1.295 1.520 65.7 77.2 50.8 1.50 30.0 81.0 623.2 80.6 0.96

(b): Step 2. Design of connection between the beam and through plate
Spec xe (m) Me (kN m) db (m × 10−2 ) Tf (kN) L (m × 10−2 ) x (m × 10−2 ) U Afn (m2 × 10−4 ) Afe (m2 × 10−4 ) Fu (MPa) Rt Tfuc (kN) Tf /Tfuc

1 1.345 60.5 18.24 331.5 18.0 2.2 0.88 8.1 7.1 370 1.2 316.5 1.05
2 1.345 68.3 18.24 374.3 18.0 2.3 0.87 9.1 7.9 370 1.2 351.6 1.06

(c): Step 3. Design of the through plate panel zone

Spec Me (kN m) db (m × 10−2 ) ψ Vpz (kN) w (m × 10−2 ) tp (m × 10−2 ) dc (m × 10−2 ) Fyp (MPa) τyp (MPa) Vu (kN) Vpz /Vu
1 60.5 18.24 0.9 596.6 7.0 1.5 19 240 138.5 588 1.01
2 68.3 18.24 0.9 673.8 7.0 1.5 19 240 138.5 588 1.15

(d): Step 4. Design of connection between the through plate and column
τwy τwx τw Fw (E60) τw /Fw
P P
Spec Mpr Vb F1,2 Mcf hw tw e Awe Jw
(kN m) (kN) (kN) (kN m) (m × 10−2 ) (m × 10−2 ) (m2 × 10−4 ) (m3 × 10−6 ) (MPa) (MPa) (MPa) (MPa)
1 147.8 0.0 700.1 7.4 48.0 1.4 67.2 537.6 104.2 13.75 105 185.4 0.57
2 167.0 0.0 791.0 8.35 48.0 1.4 67.2 537.6 117.7 15.53 119 185.4 0.64
H: weld height, twe : effective throat thickness of weld, Awe : effective throat area of weld, Jw : torsional constant of weld, τwx = Mcf (H /2)/Jw : weld shear stress on effective
throat area in x direction, τwy = F1,2 /Awe : weld shear stress on effective throat area in y direction, τw = (τw2 x + τw2 y )0.5 : resultant weld shear stress on effective throat area,
Fw = φ(0.6FEXX ): weld nominal strength, φ = 0.75.

at the end of the through plate can be calculated as follows (see

Fig. 3),
Tfu = Afe Rt Fuf (5)
where Afe = Afn U is the flange effective net area, Afn is the flange
net area including the attached cover plates, U = 1 − x/L is the
shear lag factor, x is the distance between the through plate and the
flange net area centriod, L is the connection length, Rt is a factor for
calculating the expected tensile strength which is equal to 1.2 for
A36 steel hot-rolled structural shapes and plates [22], and Fuf is the
specified minimum tensile strength of the beam flange. It should
be noted that the steel grade of the beam flange and the cover
plates are assumed identical. Therefore, it should be checked that
the tensile demand on the flange net effective area, Tf = Me /db , is
less than the tensile rupture capacity, Tfu , where, Me is the moment
at the end of the through plate (see Fig. 4), and db is the beam depth.
To have full load path between the beam web and through plate,
the dimensions of the web plates are determined to provide the
beam web moment and shear capacity. The design summary is
presented in Table 2(b) for the connection between the beam and
through plate.
Fig. 4. Bending moment diagram of the connection.
Step 3. Design of the through plate panel zone
Based on the equilibrium, the central part of the through plate
where Mp0 is the reduced plastic moment due to a shear
experiences high shear demands. This part plays the role of panel
force V (=Mpr /Lp ), Mp (=tp h2a Fyp /4) is the plastic moment, zone in conventional connections and the through plate should be
Vp (=0.577Fyp tp ha ) is the plastic shear capacity, tp is the through proportioned to provide the expected required strength.
plate thickness, ha is the through plate depth at the critical sec- Consequently, a mathematical model is assumed to interpret
tion, and Fyp is the minimum specified yield strength of the through the shear resistance mechanism of the through plate. This model
plate. Therefore, the moment at the through plate critical section consists of three main parts as depicted in Fig. 5(a). The central part
Ma should be less than the reduced plastic moment Mp0 calculated is a shear panel with pure shear behavior, the corner parts are uni-
based on Eq. (4), as presented in Table 2(a). axial links with tension and compression behavior, and the rigid
Step 2. Design of connection between the beam and through plate parts are representing the beam and column flanges assumed to
The connection between the beam and through plate consists have rigid behavior in comparison with the other parts. The de-
of two parts: beam flanges to the through plate connection and formed shape of the mathematical model under external force is
the beam web to the through plate connection. As the beam web shown in Fig. 5(b) and used to calculate the ultimate shear capac-
is removed according to Fig. 1(b), tensile rupture on the flange net ity of this region. Up to total shear yielding of the central part,
effective area is probable. Therefore, the tensile rupture capacity the shear strength is provided by the central and corner parts, and
2040 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

Fig. 6. Test setup configuration and overall dimensions of the specimens

(photograph of test specimen 2).

Step 4. Design of connection between the through plate and the

b column
According to Section 2.2, the vertical welded connection be-
tween the through plate and column flange should be designed for
the combination of in-plane force Fi and column flange moment
Mcf as expressed in Eqs. (1) and (2), respectively. These forces in-
duce perpendicular shear stresses on effective throat thickness of
the partial-joint-penetration groove weld and the weld strength
should be checked against the resultant shear stress, as summa-
rized in Table 2(d).

3.3. Test setup and instrumentation

According to the cruciform shape of the specimen, a test setup

was prepared to simulate the boundary conditions of the beam to
column interior subassembly in a laterally loaded moment frame.
The column was supported at the base by a real hinge and beam
Fig. 5. Through plate panel zone: (a) Representation of main parts, (b) Deformed ends were simply supported in a manner to enable free movement
shape of the mathematical model.
in the direction along the length of the beam. The beams were lat-
erally braced in the vicinity of the plastic hinge and also near the
thereafter increase in shear strength can only be attributed to re- beam end. Lateral bracings were located 0.6 m and 1.35 m from the
sistance of the corner parts. As completely explained in Appendix, column centerline where the maximum spacing of the beam lat-
the shear capacity of the through plate panel zone can be calculated eral bracing was about 0.9 m, according to AISC seismic provisions.
as follows, The column was also laterally supported near the point of loading
to limit the column movement in a direction perpendicular to the
3.464w 2
 
plane of the specimen. The general configuration and dimensions
Vu = τyp dc tp 1+ (6)
dc db of the test setup and test specimens are shown in Fig. 6.
The cyclic displacement was applied at the tip of the column
where τyp = 0.577Fyp is shear yield strength of the through plate, by a hydraulic actuator. The hydraulic loading system, mounted
dc is the column depth, and w is the effective width of the corner between the reaction frame and the column tip, consisted of a
element as shown in Fig. 5(a), and all other parameters are as pre- 100 kN hydraulic actuator with 200 mm equal stroke for both
viously defined. The shear strength Vu must be equal to or larger forward and backward directions and also a load cell placed on the
than the design shear force Vpz which can be computed as follows, tip of a hydraulic actuator to record the applied loads. The actuator
P  displacement is measured via both an internal displacement
Me transducer of the actuator and an external LVDT. As shown in Fig. 7,
Vpz = Ψ (7)
0.95db two diagonal LVDTs (no. 1 and 2) were installed to measure the
P shear deformation of the through plate in the panel-zone region.
where Me is the summation of bending moments imposed on Additionally, total rotation of the through plate was measured with
the through plate (Eq. (3)), and Ψ is conservatively taken as 0.9 as respect to the column centerline by means of two horizontal LVDTs
explained in Section 4.3. It should be noted that beam depth db in (nos. 3 and 4) placed at the tips of the through plate and connected
Eq. (7), includes the thickness of cover plates (if needed). to the column centerline by rigid rods. The specimens were cleaned
The panel zone design summary is presented in Table 2(c). As and strain gauges were pasted at several locations as depicted in
depicted in the table, the ratio of Vpz /Vu is 1.01 for the first spec- Fig. 7. Finally, a whitewash coating was provided to monitor the
imen and 1.15 for the second one. The different values of Vpz /Vu mode of progression of yielding during the test.
are considered to examine the accuracy and sensibility of the panel
zone design relationship. Therefore, it is expected that the first 3.4. Loading history
specimen shows lower nonlinear shear deformation in the panel
zone region as compared to the second specimen with a relatively The specimens were subjected to the loading sequence
weak panel zone. proposed by AISC seismic provisions for testing of beam-to-column
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2041

Fig. 7. Instrumentation of the test specimens (all dimensions are in mm).

moment connections [22]. The loading history consists of six cycles of the deformed specimen with the lateral braces. During the test
each of 0.375, 0.5, and 0.75% total story drift angle, sequentially. process, no yielding points or defects were found in the connection
The next four cycles are at 1% story drift, followed by two cycles, region and the column. Fig. 8(a) and (b) illustrate the first specimen
each of successive increasing drift percentages (i.e., 2, 3, 4 . . . %). The at 6% story drift.
loading was quasi-statically applied in displacement controlled
mode of the actuator by low rate to identify the behavior of the 3.5.2. Second specimen observations
connection during the load cycles and record all the deformations In the second specimen, a flange yielding appeared during 1.5%
and probable events. story drift and during 2% story drift, yielding extended into the
beam web and spread about 200 mm along the beam flanges.
During the last cycle of the 3% story drift, a minor flange lateral
3.5. General test observations
movement of the left beam was initiated from the plastic hinge
location. With cycles to 4% and 5% story drifts, the spread of yield-
General test observations include incidents throughout the test ing and the amplitude of flange lateral movement increased. In 6%
procedure for both specimens. story drift, flange lateral movement of the right beam was initiated
and followed by the onset of local buckling in both beam flanges.
3.5.1. First specimen observations Thereafter, a minor yielding occurred in the through plate at the
The yielding of the first specimen was initiated in flanges on the end of the beam, and the web of the right beam buckled. During
beams during the first cycle of 1.5%, and at 2% story drift flaking of 7% story drift, the web buckling occurred in the left beam and the
whitewashes was spread about 250 mm along the beam flanges. amplitude of the beam flange and web buckles grew rapidly. At the
During the last cycles of the 3% story drift the local yielding at the end of 8% story drift cycles, the specimen experienced local buckles
end of the cover plates was followed by the onset of simultaneous with large amplitudes and the beam compression flange severely
lateral torsional buckling (LTB) of the right beam and one way moved laterally. Finally, the test was terminated during the first
local buckling of the right beam bottom flange. During the 4% cycle of 9% story drift due to excessive deformations and strength
and 5% story drift cycles, instabilities became more extensive and degradation of the specimen. Fig. 8(c) and (d) depict the second
when the specimen tried to complete 5% story drift, the beam web specimen at 9% story drift.
buckled as well as the flanges and LTB of the beams was followed
3.6. General evaluation of the connection behavior
by minor twisting of the column. During the 6% story drift, global
and local buckling of the beams became so extensive that the test The hysteretic curve of the column tip load versus the col-
was stopped in the last cycle of 6% story drift due to interference umn tip displacement and the moment at the end of the through
2042 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

Fig. 8. Tested specimens at the last cycles of the cyclic test. (a) and (b) The first specimen at 6% story drift, (c) and (d) the second specimen at 9% story drift (L.B.: Local
buckling; LTB: Lateral torsional buckling).

(a) Test specimen 1 (b) Test specimen 2

(c) Test specimen 1 (d) Test specimen 2

Me Me
ZFy ZFy

Fig. 9. (a) and (b) Column tip load versus the column tip displacement, (c) and (d) Normalized moment at the end of the through plate versus total inelastic rotation.

plate versus total inelastic rotation are presented in Fig. 9 for both any strength deterioration and exceed the AISC seismic provision
specimens. The total rotation is computed as the ratio of the col- requirements, while this provision accepts maximum 20% strength
umn tip displacement to the column height. Also, the total plastic degradation until 4% total story drift for qualifying a connection
rotation θp is obtained by subtracting the elastic deformations from for special moment resisting frames [22]. Based on Fig. 9(c) and
the total rotation. The moment at the end of the through plate is (d), the first specimen developed at least 4% total inelastic rotation
calculated based on the overall equilibrium of the specimen. and the second one developed at least 7% total inelastic rotation
According to Fig. 9(a) and (b), the specimens exhibit stable (with maximum 17% strength degradation) which is more than the
and reliable hysteretic behavior until 4% total story drift without required 3% story inelastic drift for special moment frames in the
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2043

Test specimen 1 Test specimen 2

Ks=20EI/L (Rigid) Ks=20EI/L (Rigid)
Ks=20EI/L (Simple) Ks=20EI/L (Simple)

Fig. 10. Determination of the connection rigidity.

Test specimen 1 Test specimen 2

Fig. 11. Envelope of the normalized strains along the beam length versus total story drifts (See Fig. 7 for numbering of the strain gauges).

preceding AISC seismic provisions [24]. It should be noted that the increased rapidly due to formation of plastic hinge at the expected
strength degradation of the specimens resulted from ductile local region. The results verified that the first yielding was occurred be-
and global buckles during the cyclic loading and no fracture was tween the 1% and 2% drift angle and the beam full plastic moment
occurred in the specimens. developed until 3% drift angle.
To examine the rigidity of the connection, the connection stiff- The result of bending moment at the plastic hinge location ver-
ness has been evaluated in accordance with the method proposed sus story drift presented in Fig. 9(c) and (d) revealed that the plas-
in AISC specifications [25]. Consequently, based on the value of tic hinge of both specimens experienced moments which were
Ks Lb /EI as an index of stiffness, the connections are categorized 1.55 and 1.85 times more than the nominal plastic hinge moment,
into three different stiffness types. For index values below 2 the respectively in the first and the second specimen, due to the strain
connection considered to be simple and for the values above 20 hardening phenomenon. Subsequently, the required strength of
the connection assumed to be fully constrained or rigid. The con- the connection parts was increased according to the increase in the
nections having an index between these two limits are considered expected plastic hinge moment. Since the plastic hinge moments
partially restrained or semi-rigid. In this index, Ks = Ms /θs is the were calculated in accordance with AISC seismic provision, an over
stiffness of the connection at service load, where Ms and θs are the strength factor of 1.1Ry = 1.65 was previously considered to cal-
moment and the relative rotation between the beam end and the culate the expected moment at the plastic hinge location, which
column, respectively, at service loads and L and EI are the length nearly estimated the ultimate plastic hinge moments. The exper-
and bending rigidity, respectively, of the beam. A plot of the mo- imental results verified that the estimated plastic hinge moment
ment versus the relative rotation, which was measured by means based on AISC seismic provision was adequate for the connection
of LVDT3 and LVDT4 data sets, has been shown in Fig. 10. The com- design and no damage in the connection region was observed in
paring of the connection stiffness at service loads with the thresh- the tests.
old values validates the assumption that the connection is in the
category of rigid connections. 3.8. Evaluation of panel zone behavior

3.7. Evaluation of the plastic hinge behavior The shear deformation of the interior part of the through plate
was measured by means of two diagonal LVDTs as shown in Fig. 7.
In both specimens, a plastic hinge formed along the beam af- Therefore, the average shear strain of the panel zone region was
ter the cover plates and the web connection plates, as shown in calculated about 1.55 and 3.40 times of shear yield strain respec-
Fig. 8. The envelope of normalized strains of the beam top and bot- tively in the first and second specimens at 6% story drift. The story
tom flanges along the beam length in and out of the connection drift developed by shear deformation of the through plate is shown
zone and also the normalized strains of the beam web, measured in Fig. 12 for both specimens. The maximum story drifts caused by
by means of pasted strain gauges, are plotted versus the story drift the through plate deformation were 0.28% and 0.62% at 6% story
for both specimens as shown in Fig. 11. According to these figures, drift, respectively, for the first and second specimens. As the second
while the strains in the connection zone were nearly under yield specimen was designed with a relatively weak panel zone, larger
strain, the beam flange and web strains out the connection zone shear deformation was expected in the second specimen and this
2044 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

Test specimen 1 Test specimen 2

Fig. 12. Total story drift developed by shear deformation of the panel zone.

phenomenon was verified by the experimental results. Therefore, nodes were constrained to each other to have the same displace-
the proposed formula for the panel zone shear capacity (Eq. (6)) ments and, therefore, no connecting welds were modeled explic-
can estimate the balance point between the panel zone mechanism itly in the model. The plastic hinge area in the beam, the column
and plastic hinge mechanism in the beam and more shear demands adjacent to the through plate, Through Plate and the connection
result in large inelastic distortions in the panel zone. Accordingly, area were modeled with a finer mesh to get more accurate results.
the different beam inelastic behavior observed throughout the first The boundary conditions in the test specimen models were set to
and second tests can be justified by the difference panel zone be representative of boundary conditions in the test setup. A rep-
behaviors. As the first specimen had a relatively elastic panel zone resentation of the 3D finite element model of the second test spec-
and experienced low levels of plastic shear strain, most of the in- imen is shown in Fig. 13.
elastic demand led to the beam and severe global and local insta- To account for material nonlinearities, the von Mises yield cri-
bilities was observed, whereas, the relatively weak panel zone of terion was employed. In cyclic analyses, the kinematic hardening
the second specimen led to large inelastic deformation of the panel rule and a bilinear stress–strain relationship were considered for
zone and reduced the inelastic demands of the beam plastic hinge. all components. Young’s modulus of elasticity, E, was considered
Excessive panel zone yielding may increase vulnerability of col- to be 203 GPa up to the yielding strength and thereafter the tan-
umn, connecting welds and beam flanges to fracture and therefore gent modulus was assumed to be 0.03E up to the ultimate strength
of the material. The yield and ultimate strengths of the materi-
a panel zone with limited yield strain is usually recommended for
als were as summarized in Table 1. The Poisson’s ratio was taken
ductile moment frames [3,20].
as 0.3 for all materials throughout the analyses. To mobilize the
geometric nonlinearities in the model, a geometric imperfection
4. Evaluation of the connection behavior by finite element was imposed with the distribution similar to the first eigenvector
analysis resulted from the eigenbuckling analysis of the connection.

Nonlinear finite element analysis was conducted to more pre- 4.2. Finite element results
cisely investigate the various aspects of the connection seismic
behavior, and also to evaluate the proposed design method of The finite element analysis was performed by imposing a cyclic
the connection. Although finite element analyses cannot com- displacement to the column tip similar to the loading history of
monly model the actual material behaviors, imperfections, residual Section 3.4, whereas elastic cycles were ignored to save the time of
stresses and defects, the general behavior of the connection and the computations. The cyclic outcomes are compared with the cyclic
stress and strain states can be captured with a reasonable level of experimental results in Fig. 14 for both specimens. As shown, the
accuracy. Therefore, the general behavior of the subassemblies un- experimental and finite element results are in good agreement.
der cyclic loading, and also the expected nonlinear mechanism in While the ultimate load and initial stiffness are well evaluated, the
the beam and through plate have been verified by the experimental extant differences between the two data sets are justified by ge-
results. Thereafter, the verified model has been utilized to investi- ometric differences between the finite element models and spec-
gate the nonlinear behavior of several designed subassemblies un- imens, uncertainties in the material model, and also unavoidable
der monotonic loading to validate the proposed designed method residual stresses.
for the connection. This study used the finite element package In Fig. 15, deformed shape of the second specimen at 6% story
ANSYS [26] for all analyses. drift, von Mises plastic strain distribution, and also the plastic
mechanism of the through plate are presented. Fig. 15(a) shows
that the beam plastic hinge forms at the expected location after the
4.1. Finite element model through plate and the yielding and buckling pattern is consistent
with the experimental observations. Based on the finite element
A 3D finite element model has been constructed on the basis results, all parts of the connection except the through plate have
of the geometric dimensions of the test specimens including sec- not experienced plastic strains and remained relatively elastic. As
tion sizes, beam and column lengths, dimensions of the through shown in Fig. 15(b), the plastic regions of the through plate inside
plate, and other parts of the connection. As the specimens sub- and outside the column verify the assumed mechanism which was
stantially consisted of planar plates with relatively high width to explained in Fig. 5 for calculating the ultimate shear capacity of the
thickness ratios, a quadrilateral four-node shell element having through plate. Consequently, the corner parts and also the shear
six degrees of freedom per node (SHELL43 in ANSYS) with plastic- panel at the center of the through plate experienced most of the
ity, large deflection, and large strain capabilities were employed in plastic strains while the other areas remained relatively elastic and
the model. At the joint location of the components, shell element constituted the rigid parts in the mathematical model.
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2045

Fig. 13. 3D finite element model of the second test specimen.

Fig. 14. Comparison of the experimental and analytical column tip force versus total story drift.

Moreover, comparison of Fig. 15(a) with Fig. 15(b) shows that

a the column web contribution to the load transfer mechanism
was much less than the through plate in the panel zone region.
Therefore, the through plate effectively provided load paths for
constituting moment connection between the beam and the
box–column.

4.3. Parametric study on the through plate connection

The geometry and thickness of the through plate can be deter-

mined based on the connection design method which was pro-
posed in Section 3.2. Therefore, three specimens with different
b beam and column sizes have been totally proportioned as pre-
sented in Table 3, and then employed for a sensitivity study on the
through plate thickness. The beam section properties in the ana-
lytical specimens are tabulated in Table 4.
The nonlinear behavior of the specimens has been investigated
by means of monotonic nonlinear analysis of 3D finite element
models constructed based on the geometric properties of the de-
signed specimens. Finite element models were prepared accord-
ing to Section 4.1 with a difference in hardening rule, while the
Fig. 15. Deformed shape and von Mises plastic strain distribution at 6% story drift.
(a) The second specimen, (b) Through plate.
isotropic hardening rule was implied to better represent the be-
havior of the material in monotonic loading condition. The through
2046 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

Table 3
Specifications of the designed specimens for parametric study on the through plate connection.
Ma Tf Vpz
Analytical specimen Beams Column Cover plates Web plates Through plate Mp0 c
Tfu Vu

tp L1 L2 R w
P1 IPE 360 Built up 350 × 40 × 18 220 × 220 × 12 35 250 250 80 940 0.85 0.94 1.00
370 × 370 × 20
P2 IPE 450 Built up 470 × 80 × 18 300 × 240 × 12 35 350 355 100 140 0.82 0.94 1.01
450 × 450 × 25
P3 IPE 600 Built up 550 × 75 × 18 450 × 300 × 12 45 400 450 150 166 0.87 1.0 1.04
560 × 560 × 30
All dimensions are in mm.

Table 4
Beam section properties in the analytical specimens.
Spec Section Ab (m2 × 10−4 ) db (m × 10−2 ) bf (m × 10−2 ) tf (m × 10−2 ) tw (m × 10−2 ) Zb (m3 × 10−6 )

P1 IPE360 72.7 36 17 1.27 0.8 1019

P2 IPE450 98.8 45 19 1.46 0.94 1702
P3 IPE600 156 60 22 1.9 1.2 3512

plate thickness, tp , was changed in increments of ±0.2tp to exam-

ine the accuracy of the design method, and to investigate the ef-
fect of the through plate thickness on the nonlinear behavior of the
connection. Accordingly, the moment at the end of the through
plate versus total story drift of the specimens with different Me
through plate thicknesses is shown in Fig. 16. Generally, the re- ZFye
sults have revealed that the behavior of the connections with the
designed through plate thickness, tp , was an acceptable nonlinear
behavior for seismic applications. These specimens experienced
more than 4% story drift without any strength degradation and pro-
vided the required moment capacity until maximum 2% story drift.
Generally, the specimens with the thicker through plate degraded
at earlier stages of the story drift and the specimens with the thin-
ner through plate could not provide the required moment capacity
even until large story drifts. Therefore, it can be concluded that the
proposed connection design method provides an adequate thick- Me
ness for proportioning of the through plate connection. ZFye
According to the observed behavior of the connections with dif-
ferent through plate thicknesses, the behavior of the connection
can be categorized into three distinct behaviors. In the specimens
with the thick through plate, as the through plate remains rela-
tively elastic, a series of serious local buckling and also a type of
lateral buckling instability occurred in the beam plastic hinge re-
gion and resulted in more strength degradation specially in higher
story drifts. In contrast to thick through plates, the specimens with Me
thin through plates experience high inelastic deformations in the ZFye
through plate via the plastic mechanism explained in the previous
sections. As the panel zone contributed to the nonlinear behavior of
the connection, the demand of inelastic rotation on the beam plas-
tic hinge reduced and, therefore, the beam local and global instabil-
ities decrease dramatically. As a result, strength degradation is de-
ferred to large story drifts, while the required moment capacity of
the connection is not provided as the through plate shear capacity
is less than the required shear capacity and the beam plastic hinge
formed with difficultly at the large story drifts. Finally, the spec- Fig. 16. Normalized moment at the end of the through plate versus total story drift
imens with the designed through plate thickness have a balanced for different beam depths and different through plate thicknesses (Fye = Ry Fy ).
behavior between the strong through plate and weak through plate
behavior. 8% (Ψ = 0.92) up to 1% story drift and less than 17% (Ψ = 0.83)
Furthermore, the contribution of the column flanges and the up to 6% story drift. Therefore, the assumed value of 0.9 for Ψ in
through plate to transmission of bending moment from the beam Eq. (7) had been selected conservatively.
to the box column was examined analytically, using finite element
models. For this purpose, the moment at the inside and outside 5. Conclusions
faces of the column was numerically calculated by integrating the
longitudinal stresses over the through place height. The difference (1) The proposed connection can effectively provide a rigid mo-
between these two moments is the moment transmitted by means ment connection between I-beams and box–columns by a new
of the column flanges. As shown in Fig. 17, the contribution of the load transfer mechanism which was different from that of con-
column flanges to the transmission of bending moment is less than ventional connections.
 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048 2047

Mf
Mc

Fig. 17. Bending moment portion of the column flange (column width to thickness
ratio ≈ 17.0).

(2) As the through plate completely transfers the loads to the col-
umn, no continuity plates are needed inside the column for
beam concentrated forces.
(3) The through plate transfers the loads between beams and
columns by in-plane actions and most welds are loaded
parallel-to-axis which is different to vulnerable flange CJP weld
in conventional connections loaded normal-to-axis. Therefore,
no type of weld fracture was observed during the experimental
program.
(4) Based on the finite element and experimental results, the
Fig. 18. Corner part of the through plate.
specimens designed in accordance with the proposed design
method, reached at least 0.06 rad total story drift which was
more than 0.04 story drift criterion specified by AISC seismic The internal work Wint is obtained as the summation of the work
provisions for qualifying connections for seismic use. done by the central shear panel through virtual shear deformation
(5) The proposed mathematical model for the through plate panel Wint ,p and the work done by the corner elements through virtual
zone was effectively estimated the nonlinear mechanisms and axial deformation Wint ,c as follows,
ultimate shear capacity of this region. According to the test re-
sults and also parametric study on the through plate thickness, 1
Wint ,p = γ τyp db dc tp (9)
the panel zone design formulae provide balanced panel zone 2
behavior.
Z
1
(6) Although more ductile behavior was seen in experimental and Wint ,c =4 Pc (x)δc (x)dx. (10)
2
analytical studies in connections with weaker panel zone, it is
recommended to limit yield strain in the panel zone (as pro- The axial strain of the corner elements εc can be expressed as,
vided in the balance panel zone) to prevent probable fracture
λ
 
in the connection. εc = γ (11)
(7) Based on the parametric study, the portion of column flange 1 + λ2
bending moment was found to be small compared to the total
where λ = a/b = cot θ . As the axial strain is constant throughout
moment transferred to the column.
the corner element, the axial stress σc is also considered to remain
(8) Although the finite element and the experimental results
constant. Therefore, the axial force Pc (x) per unit length along the
showed that the design method proposed for proportioning of
horizontal side of the corner element is σc tp sin θ and the axial dis-
the connection was acceptable, additional research should be
placement of the corner element is εc x/ cos θ . Accordingly, Eq. (10)
done to further investigate the adequacy of the design method
can be written as,
especially concerning the connection fracture behavior.
Z a
1
Wint ,c = 4 × (σc tp sin θ )(εc x/ cos θ )dx. (12)
Appendix 2 0

By substituting Eq. (11) into Eq. (12) and assuming a = w/ sin θ

In this appendix, the detailed calculation of the through plate (Fig. 18), Eq. (12) can be summarized as follows,
shear capacity in the panel zone region is presented. Up to total
shear yielding of the central part, the shear strength is provided Wint ,c = σc γ tc w 2 . (13)
by the central and corner parts and thereafter, increase in shear
strength can only be attributed to resistance of the corner parts. By equating internal and external works,
The panel zone shear strength is calculated by the principle of
virtual work. The external work Wext done by the shear capacity of Wext = Wint ,p + Wint ,c . (14)
panel zone region V (up to total shear yielding of the central part)
The shear capacity of panel zone region V (up to total shear yielding
through virtual shear deformation γ is
of the central part) is as follows,
1
Wext = V γ db . (8) V = τyp dc tP + 2σc tp w 2 /db . (15)
2
2048 S.R. Mirghaderi et al. / Engineering Structures 32 (2010) 2034–2048

The increase in the shear capacity 1V can be calculated based on [9] Ting LC, Shanmugam NE, Lee SL. Design of I-beam to box–column connections
the corner elements only as follows, stiffened externally. Eng J, AISC 1993;(Fourth quarter):141–9.
[10] Shanmugam NE, Ting LC, Lee SL. Behavior of I-beam to box–column
connections stiffened externally and subjected to fluctuating loads. J Construct
1
1γ 1Vdb = 1σc 1γ w 2 (16) Steel Res 1991;20(2):129–48.
2 [11] Kang CH, Shin KJ, Oh YS, Moon TS. Hysteresis behavior of CFT column to
H-beam connections with external T-stiffener and penetrated elements. Eng
1V = 21σc tp w 2 /db . (17) Struct 2001;23:1194–201.
[12] Shin KJ, Kim YJ, Oh YS, Moon TS. Behavior of welded CFT column to H-beam
Therefore, the ultimate shear capacity Vu can be expressed as, connections with external stiffeners. Eng Struct 2004;26:1877–87.
[13] Shin KJ, Kim YJ, Oh YS. Seismic behavior of composite concrete-filled tube
2(σc + 1σc )w 2 column-to-beam moment connections. J Construct Steel Res 2008;64:118–27.
 
Vu = V + 1V = τyp dc tp 1+ . (18) [14] Park JW, Kang SM, Yang SC. Experimental studies of wide flange beam to
τyp dc db square concrete-filled tube column joints with stiffening plated around the
column. J Struct Eng, ASCE 2005;131(12):1866–76.
As the maximum axial stress in the corner elements can reach the [15] Kurobane Y. Connections in tubular structures. Prog Struct Eng Mater 2002;2:
yield strength of the through plate, the term (σc + 1σc ) in Eq. (18) 35–43.
[16] Tanaka A, Masuda H, Kadoya H, Ito A. Behavior of WF beam-to-SHS column
is assumed to be Fyp and the equation is rearranged as follows, connections using special shaped high strength bolts. In: Proceedings of
the fourth international workshop on connections in steel structures. 2000.
3.464w 2
 
p. 205–12.
Vu = τyp dc tp 1+ . (19) [17] Korol RM, Ghobaran A, Mourad S. Blind bolted W-shaped beam to HSS
dc db columns. J Struct Eng, ASCE 1993;119(12):3463–81.
[18] Wheeler AT, Clarke MJ, Hancock GJ, Murray TM. Design model for bolted
moment end plate connections joining rectangular hollow sections. J Struct
References Eng, ASCE 1998;124(2):164–73.
[19] Wheeler AT, Clarke MJ, Hancock GJ. FE modeling of four-bolt tubular moment
[1] Nakashima M, Roeder CW, Maruoka Y. Steel moment frames for earthquakes end-plate connections. J Struct Eng, ASCE 2000;126(7):816–22.
in United States and Japan. J Struct Eng, ASCE 2000;126(8):861–8. [20] Roeder CW. General issues influencing connection performance. J Struct Eng,
[2] Sherman DR. Designing with structural tubing. Eng J, AISC 1996;33(3):101–9. ASCE 2002;128(4):420–8.
[3] FEMA-355D. State of the art report on connection performance. Washington [21] FEMA-355D. State of the art report on connection performance. Washington
(DC): Federal Emergency Management Agency; 2000. (DC): Federal Emergency Management Agency; 2000.
[4] Chen CC, Lin CC, Tsai CL. Evaluation of reinforced connections between steel [22] AISC/ANSI 341-05. Seismic provisions for structural steel buildings. Chicago
beams and box columns. Eng Struct 2004;26:1889–904. (IL): American Institute of Steel Construction, Inc.; 2005.
[5] Chen CC, Lin CC, Lin CH. Ductile moment connections used in steel column-tree [23] Horne Michael R. Plastic theory of structures. UK: Thomas Nelson and Sons;
moment-resisting frames. J Construct Steel Res 2006;62:793–801. 1971.
[6] Tanaka N. Evaluation of maximum strength and optimum haunch length of [24] American Institute of Steel Construction (AISC). Seismic provisions for
steel beam-end with horizontal haunch. Eng Struct 2003;25:229–39. structural steel buildings. Chicago (IL); 1997.
[7] Shanmugam NE, Ting LC. Welded interior box–column to I-beam connections. [25] AISC/ANSI 360-05. Commentary on the specification for structural steel
J Struct Eng, ASCE 1995;121(5):824–30. buildings. Chicago (IL): American Institute of Steel Construction, Inc.; 2005.
[8] Ting LC, Shanmugam NE, Lee SL. Box–column to I-beam connections with [26] ANSYS (Revision 5.4), User’s manual, theory, vol. IV. Swanson Analysis
external stiffeners. J Construct Steel Res 1991;18(3):209–26. Systems, Ine; 1992.