Moment Connections to Circular Concrete-Filled

Steel Tube Columns

Atorod Azizinamini, M.ASCE,1 and Stephen P. Schneider, M.ASCE2
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

Abstract: As part of the U.S.-Japan Cooperative Research Program on Composite/Hybrid Structures 共1992兲, investigations were carried
out at the University of Nebraska-Lincoln and University of Illinois, to develop moment connection details and accompanying design
provisions for connecting steel beams to circular concrete filled tube 共CFT兲 columns. This paper provides an overview of the work carried
out. Work at the University of Illinois consisted of evaluating the inelastic response of six possible connection details. Six large-scale
connections were tested to failure using the quasi-static test method. Connections welded to the skin of the tube created large distortions
of the tube wall, and were susceptible to weld, flange, or tube wall fracture. External diaphragm and continuous web details exhibited
more favorable inelastic behavior, but the flexural strength of these connections began to deteriorate relatively early in the imposed
deformation history. Continuing the flanges through the composite column showed adequate strength, however due to excessive slipping
this connection did not dissipate significant inelastic dynamic energy. One of the details tested at the University of Illinois was passing the
girder section through the CFT columns. This through beam detail provided the most effective detail to achieve the ideal rigid connection
condition. Work at the University of Nebraska-Lincoln concentrated on comprehending the behavior of the through beam connection
detail. The behavior of CFT columns under seismic loads was studied through testing six columns which were subjected to a constant
axial load in addition to a cyclic lateral load. Failure modes for through beam connection detail were identified through testing seven,
two-thirds scale connection specimens. These tests were used to comprehend the force transfer mechanisms between steel beams and CFT
columns and develop design provisions that could estimate the capacity of the various elements of the through beam connection detail.
DOI: 10.1061/共ASCE兲0733-9445共2004兲130:2共213兲
CE Database subject headings: Composite columns; Beams; Connections; Cyclic tests; Full-scale tests.

Introduction Work at the University of Illinois consisted of evaluating seis-

mic performance of six different details, capable of connecting
Concrete-filled steel tube 共CFT兲 columns combine the advantages steel beams to CFT columns. Experimental results indicate that
of ductility generally associated with steel structures with the the inelastic connection behavior was very dependent on the type
stiffness of concrete components. The advantages of the concrete- of connection detail. Connections welded to the skin of the tube
filled steel tube column over other composite systems includes: created large distortions of the tube wall, and were susceptible to
the steel tube provides formwork for the concrete, the concrete weld, flange, or tube wall fracture. External diaphragm and con-
prolongs local buckling of the steel tube wall, the tube provides tinuous web details exhibited more favorable inelastic behavior,
excessive concrete spalling, and composite columns add signifi- but the flexural strength of these connections began to deteriorate
cant stiffness to a frame compared to more traditional steel frame relatively early in the imposed deformation history. Deformed
construction. While many advantages exist, the use of CFTs in bars welded to the girder flange and embedded into the concrete
building construction has been limited due, in part, to a lack of core exhibited promising ductility with large flexural strength.
construction experience and to the complexity of connection de- Continuing the flanges through the composite column showed
tailing. This paper summarizes part of the investigation carried adequate strength, however due to excessive slipping this connec-
out at the University of Nebraska-Lincoln 共UNL兲 and the Univer- tion did not dissipate significant inelastic dynamic energy. Con-
sity of Illinois to develop connection details and accompanying tinuing the girder section through the composite column provided
design provisions for connecting steel beams to circular CFT col- the most effective detail to achieve the ideal rigid connection
umns. condition.
Work at the University of Nebraska-Lincoln concentrated on
1
Professor, Dept. of Civil Engineering, Univ. of Nebraska, Lincoln, comprehending the behavior of the through connection detail.
NE 68588 共corresponding author兲. E-mail: aazizi@unl.edu Failure modes were identified followed by development of the
2
Kramer-Gehlen & Assoc., 400 Columbia St., Suite 240, Vancouver, design provisions. The failure modes for through beam connec-
WA 98660. tion detail were identified by investigating the behavior of CFT
Note. Associate Editor: Sherif El-Tawil. Discussion open until July 1, columns and connection tests. Six CFT columns were tested
2004. Separate discussions must be submitted for individual papers. To under constant axial load and cyclic lateral loads. These tests
extend the closing date by one month, a written request must be filed with
were used to develop design provisions for estimating flexural
the ASCE Managing Editor. The manuscript for this paper was submitted
for review and possible publication on May 8, 2002; approved on June capacity of the CFT columns. A complete summary of the CFT
23, 2003. This paper is part of the Journal of Structural Engineering, column tests is provided in Elremaily and Azizinamini 共2000兲.
Vol. 130, No. 2, February 1, 2004. ©ASCE, ISSN 0733-9445/2004/2- Developing design provisions for any connection requires
213–222/$18.00. knowledge of the possible failure modes and ability to estimate

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004 / 213

J. Struct. Eng., 2004, 130(2): 213-222

 the contribution of each connection element to overall connection tube was subjected to high local distortions adjacent to the con-
capacity. Failure modes for through beam connection detail were nected region. Fracture initiated in the connection stub at a rota-
identified through testing seven, two-thirds scale connection tion of 1.25%, and propagated into the tube wall by approxi-
specimens. The experimental work for identifying the failure mately 2.75% rotation. This tearing propagates from the tips of
modes, analyzing the experimental work, comprehending the the flange toward the web. Degradation in flexural strength began
force transfer mechanism between steel beam and CFT column, once the flange fractured, and continued upon subsequent defor-
and developing the design provision for through beam connection mation cycles. This fracture precipitated high shear demand on
detail are discussed in Elremaily and Azizinamini 共2000兲. the web plate, which led to fracture of the weld between the web
and the pipe wall. Consequently, the girder lost all of its gravity-
load-resisting capacity shortly after flange failure. The stiffness of
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

Possible Connection Details this connection was 85% of the ideal rigid connection condition.
Further, this connection exhibited the earliest decay in the effec-
Test data exist on a variety of moment connection details that
tive stiffness, clearly due to the large distortions of the tube wall.
could be used to connect steel beams to CFT columns. Details
The external diaphragms of connection Type II were intended
that connect the girder to the steel tube only include: welding the
to alleviate the severe distortion of the steel tube skin. These
girder directly to the skin of the steel tube 共Valbert 1968兲, using
diaphragms improved the cyclic behavior of the simple connec-
web angles or shear tabs to connect the girder to the column
tion significantly. Each diaphragm fractured at approximately
共Shakir-Khalil 1992; Bridge and Webb 1992兲, providing external
2.0% rotation. Upon subsequent imposed deformation cycles, the
or internal diaphragms 共Kato et al. 1992; Morino et al. 1992兲, and
fracture propagated in each diaphragm, eventually tearing the
variations of these details 共Ansourian 1976兲. Conclusions from
tube wall at a rotation of 3.5%. Deterioration of the M ⫺␪ behav-
these studies suggest that connections loading the skin of the steel
ior occurred at the onset of diaphragm fracture. At large cyclic
tube only can cause excessive deformation demands on the tube
deformations, the diaphragm buckled and the tube wall fractured
wall and connection components. Connections that attempted to
along the depth of the girder. External diaphragm plates also ex-
improve this behavior include: through bolting girder end plates
hibited reasonably stiff behavior, with relatively little deteriora-
共Prion and McLellan 1992; Kanatani et al. 1987兲 and continuing
tion before 3.0% total rotation.
the structural steel shape through the column 共Aziznamini and
The intent of the continuous web Type III detail was to im-
Prakash 1992兲. Comparison of test data suggests that embedding
connection components into the concrete core alleviates high prove the flexure and the shear behavior of the simple Type I
shear demand on the tube wall, which may improve the seismic connection. However, significant web tearing was still observed
performance of the connection. once the flange fractured. Similar to the simple connection, the
Fig. 1 shows the details of the six large-scale moment connec- flange fracture exacerbated the high strain demands on the ex-
tions tested at the University of Illinois. In general, the type of treme fiber of the web. Because of the embrittlement of the web
connection detail needed for a frame may depend on the relative in the heat affected zone, tearing initiated close to the fillet weld
size of the column and the girder at the joint. For example, the that attached the web to the pipe wall. Fracture initiated at ap-
simple welded Type I connection may be more suitable for large proximately 2.0% rotation, creating a 20% decrease in the peak
diameter pipes, or perhaps thick tube walls, while the diaphragm flexural strength by the end of the imposed deformation cycles.
Type II detail may be more appropriate for small diameter tubes. Eventually, this fracture propagated from each flange toward the
Each tested connection consisted of a 356-mm-diam pipe with center of the web. The effective stiffness deterioration of the con-
a 6.4 mm wall thickness, and a W14⫻38 for the girder. The yield nection detail was reasonably gradual beginning at approximately
strength of the pipe and the girder was 320 MPa, with an approxi- 1.5% total rotation.
mate concrete strength of 35 MPa. Connection Type IV was identical to connection Type I, except
The primary interest of the work carried under the U.S.-Japan four 20-mm-diam (Fy⫽420 MPa) weldable deformed bars were
Cooperative work was to investigate connections that develop the embedded in the concrete core through holes drilled in the steel
full flexural strength of the connected girder. To accomplish this, tube wall. Each deformed bar was then welded to the girder
it was important to detail connections to just develop the plastic flange, and the weld was sufficient to develop 50% more bar force
bending strength of the girder, as might be done in practice. than the minimum required weld. Embedment lengths were suf-
Therefore, all of these connections were intended to just develop ficient to develop the deformed bar under tested conditions. The
M p of the girder at the face of the CFT column. Since connection M ⫺␪ behavior of the detail was a significant improvement com-
behavior was the primary interest, and not joint or panel zone pared to connection Type I. Initial wall tearing was observed at
behavior, the girder connected to only one side of the CFT col- 3.0% total rotation, however, this tearing was located in the tube
umn was sufficient. Consequently, a tube column and single wall between the openings for the deformed bars. The minor wall
girder specimen were tested, and cyclic deformations were im- tearing did not affect the subsequent inelastic performance of the
posed on the tip of the cantilever at a distance of 2.75 m from the connection. Local flange buckling was also observed at a 3.0%
face of the CFT column. The cyclic deformation history followed total rotation, which occurred in the girder beyond the connection
the recommendations provided by the ATC-24 Guidelines for Cy- region. This suggests that the connection was strong enough to
clic Seismic Testing of Components 共ATC-24 1992兲. initiate significant yield in the girder flange. Failure of the con-
nection was caused by fracture of the deformed bars at approxi-
mately 3.75% total rotation. No significant stiffness or strength
Performance of Various Connection Details loss was observed prior to the fracture of the deformed bars. The
effective stiffness of this connection ranged between 98 and
Figs. 2 and 3 show the moment envelope and the measured effec- 106% of the ideal rigid connection conditions, and did not appear
tive stiffness, respectively, for each tested connection. to deteriorate over the range of imposed deformations. This in-
For the simple Type I connection, the girder was welded di- creased connection stiffness was likely due to the added stiffness
rectly to the skin of the steel tube wall. Consequently, the steel of the deformed bars on the top of each flange.

214 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004

J. Struct. Eng., 2004, 130(2): 213-222

Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Connection details

The Type V connection detail was tested to study the effect of approximately 1.0% rotation, and the flange plate welds on the
continuing only the flanges through the CFT column. Flange opposite column side remained intact. This resulted in a tube wall
plates were welded to the tube wall on each side of the column to fracture on the CFT face opposite the girder. Subsequent cycles
transfer girder flange forces. The resistance was to have been only increased the tearing of the tube wall. The stiffness of this
induced by bearing of the steel tube against the concrete core, connection detail was larger than the simple welded Type I con-
provided the continuity flange plate weld remained intact with the nection; however, stiffness deterioration was significant and
steel tube skin. No attempt was made to enhance the bond be- began when the flange weld fracture was first observed.
tween the embedded flange plate and the concrete core. The lack Results of connection Type VI exhibited quite stable inelastic
of bond impacted the hysteretic performance of this connection behavior. Local flange buckling was observed at approximately
significantly compared to other connection types. Both flange 2.75% total rotation, and the web buckling was observed at about
welds on the girder side of the column connection fractured at 3.0% total rotation. Deterioration of the inelastic characteristics

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004 / 215

J. Struct. Eng., 2004, 130(2): 213-222

 signed such that possible modes of failure and the connection
strength could be identified. The main test variables were the
beam-to-column flexural capacity ratio, the type of weld used to
attach the beam to the tube, and the presence or absence of some
of the connection elements within the panel zone. Table 1 shows
the properties of the test specimens. The height of the column, H,
shown in the table represents the distance between the lateral
reaction points. The total beam length, L, between the loading
points for all the specimens was 13 ft, 6 in. 共4,115 mm兲. The
beam and column flexural strengths given in Table 1 were calcu-
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

lated based on measured material properties. The measured ma-

terial properties are given in Table 2. The moment capacity of the
column was calculated based on the axial load applied on column
during the test using the analytical model developed by Elremaily
and Azizinamini 共2000兲. The squash load, P 0 , was calculated as
the summation of the ultimate axial capacities of both the steel
Fig. 2. Moment-rotation envelope curves for tested specimens and concrete and is given by the following equation:

P 0 ⫽A s F y ⫹A c f c⬘ (1)
was observed after the onset of local web buckling. Failure of the An analytical investigation was also carried out using nonlin-
connection was caused by fracture of the beam flange in the con- ear finite element modeling techniques to provide a supplement to
nection stub region. This flange tearing eventually propagated the experimental data in understanding the joint behavior and to
into the web. Although the flexural strength decreased approxi- quantify some of the parameters that could not be measured ex-
mately 30% compared to the peak values, the hysteretic behavior perimentally. The results of the analytical investigation are briefly
remained stable even at large rotations. No crushing of the con- discussed in this paper. Greater details are provided in Elremaily
crete was observed, and the tube wall showed no apparent signs and Azizinamini 共2000兲. The experimental and analytical results
of local distress. The effective stiffness remained equivalent to were used to identify the force transfer mechanism within the
that of the ideal rigid connection condition throughout the im- joint. A model was then developed to estimate the joint strength.
posed deformation history. The ability to predict the CFT column capacity, beam capacity,
and joint capacity was used as a basis to develop step by step
design guidelines for the through beam connection detail as out-
Behavior of Through Beam Connection Detail lined in Elremaily and Azizinamini 共2000兲.
The research study described above demonstrated the merits of
using through beam connection detail in seismic regions. The Failure Modes
investigation at the University of Nebraska-Lincoln 共UNL兲 共Elre- In general, the development of design criteria for a certain con-
maily and Azizinamini 2000兲 concentrated on comprehending, in nection involves first identifying possible failure modes and then
more detail, the behavior of through beam connection detail and establishing appropriate design checks to prevent undesirable fail-
developing design guidelines. The remaining portion of this paper ure modes. The results of the experimental study presented by
will provide part of the work carried out at UNL. Elremaily and Azizinamini 共2000兲 have shown that, in the case of
As part of the work carried out at UNL, seven two-thirds scale through beam connection detail, subassembly failure could take
connection specimens were tested. Each test specimen consisted place in either one of the following modes: 共a兲 column failure, 共b兲
of a CFT column and a steel beam passed through the column to beam failure, or 共c兲 joint failure.
represent an interior joint in a building. The specimens were de- Column failure takes place when the column flexural strength
is lower than the beam flexural strength and the joint shear capac-
ity. In the experimental study conducted in Specimens NSF1 and
NSF2 共see Table 1兲 the column was weaker than the beam and
fillet weld was used to attach the beam to the tube. Failure of
these two specimens initiated by fracture of the weld attaching the
beam flange to the tube was followed by tube tearing on the
tension side and buckling on the compression side. When fillet
weld is used to attach the beam to the steel tube, the capacity of
the column is reduced due to the reduction in the effective tube
area as a result of the presence of the fillet weld. Although no
experimental data are available on the case when full penetration
weld is used with the column being weaker than the beam, it is
expected that, in such case, failure would still take place in the
column, but probably in a more ductile way.
Beam failure takes place when the beam capacity is smaller
than the column flexural capacity and shear capacity of the joint.
Such mode of failure was observed in Specimens NSF4, NSF5,
and NSF8. In Specimens NSF4 and NSF5, the column-to-beam
Fig. 3. Variation of effective stiffness
strength ratio was about 1.5 and full penetration weld attached the

216 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004

J. Struct. Eng., 2004, 130(2): 213-222

 Table 1. Comparison Between Experimental and Analytical Results
Specimen NSF1 NSF2 NSF4 NSF5 NSF6 NSF7 NSF8
Beam W18⫻50 W18⫻50 W18⫻35 W18⫻35 Built up Built up W16⫻31
d (in.) 17.99 17.99 17.7 17.7 17.5 17.5 15.88
b f (in.) 7.495 7.495 6 6 8 8 5.525
t f (in.) 0.57 0.57 0.425 0.425 1.75 1.75 0.44
t w (in.) 0.355 0.355 0.3 0.3 0.5 共0兲a 0.5 共0.25兲a 0.275
M p (kip in.) 4,429 4,429 3,312 3,451 10,721 10,721 2,619
Column 12 ⭋⫻1/4 in. 12 ⭋⫻3/8 in. 16 ⭋⫻1/4 in. 16 ⭋⫻1/4 in. 16 ⭋⫻1/4 in. 16 ⭋⫻1/4 in. 16 ⭋⫻1/4 in.
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

H (ft) 7 7 8 8 8 8 8
P axial / P 0 0.17 0.16 0.18 0.19 0.22 0.19 0.21
M n (kip in.) 2,668 3,548 5,642 5,682 11,375 11,345 5,680
M n /M p 0.60 0.80 1.70 1.65 1.06 1.06 2.17
Weld type Fillet Fillet Full Full Full Full Fillet
penetration penetration penetration penetration
Special No concrete Rebars Weak joint Weak joint
details in panel zone no web thin web
P test 共kip) 51 48 50 52 94 107 37
Failure mode Weld fracture Weld fracture Beam failure Beam failure Joint shear Joint shear Beam failure
& tube tearing & tube tearing
V wy (kips) 126 126 149 133 共124兲b 0 共0兲b 116 共113兲b 112 共105兲b
V f (kips) 6 6 3 3 共1兲b 74 共75兲b 74 共68兲b 3 共1兲b
V csn (kips) 211 0 338 341 共168兲b 331 共413兲b 331 共416兲b 334 共152兲b
V tn (kips) 153 228 242 242 共122兲b 242 共273兲b 242 共266兲b 242 共91兲b
P j (kips) 74 54 108 106 85 101 89
P bn (kips) 59 59 45 47 147 147 36
P cn (kips) 42 56 85 86 172 171 84
P pred 共kips) 42 54 45 47 85 101 36
Note: 1 in.⫽25.4 mm, 1 ft⫽305 mm, 1 kip⫽4.448 kN, 1 kip in.⫽0.113 kN m.
a
Thickness of the portion of beam web inside the tube.
b
Values from the finite element model.

beam to the steel tube. In Specimen NSF8, the column-to-beam Design Philosophy
strength ratio was approximately 2 and fillet weld attached the
beam to the steel tube. For design purposes, it is desirable that failure would take place
Test results have shown that failure can take place in the joint by plastic hinging in the beam outside the joint. Both column
panel when the joint is weaker than the attached members. Such failure and joint failure should be avoided. This is consistent with
mode of failure was observed in Specimens NSF6 and NSF7. the strong column-weak beam concept implied by building codes.
Failure of these two specimens was due to excessive shear distor- To prevent column failure, the column should be stronger than the
tion in the connection panel. beam. The experimental results of Specimens NSF4, NSF5, and

Table 2. Material Properties of Connection Test Specimens

Specimen NSF1 NSF2 NSF4 NSF5 NSF6 NSF7 NSF8
Beam
flange
F y (ksi) 43.9 43.9 49.8 51.9 42.3 42.3 48.5
F u (ksi) 64.2 64.2 72.7 72.5 75.1 75.1 68.5
Beam web
F y (ksi) 49.4 49.4 51.7 46.3 56.9 56.9 共48.4兲a 42.6
F u (ksi) 67.8 67.8 72.4 71.7 75.2 75.2 共71.7兲a 68.6
Column
F y (ksi) 54.3 53.9 64.3 64.3 64.3 64.3 64.3
F u (ksi) 72.3 69.9 80.4 80.4 80.4 80.4 80.4
f c⬘ (ksi) 6.0 5.6 4.9 5.0 4.7 4.7 4.8
Note: 1 ksi⫽6.895 MPa.
a
Value for the portion of beam web inside the tube.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004 / 217

J. Struct. Eng., 2004, 130(2): 213-222

Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Deformed shape of joint

NSF8 showed that column failure was prevented when the

column-to-beam flexural strength ratio was approximately 1.5 for Fig. 5. Forces at panel boundaries
full penetration weld and approximately 2.0 for fillet weld. These
ratios are higher than the 1.2 ratio recommended by the American
Concrete Institute 共ACI兲 specifications 共ACI Committee 318 chanics of the joint shear mechanisms may be visualized by con-
1995兲. It is recommended that these conservative values should sidering their role in resisting the horizontal flange forces. The
be used as lower limits on the column-to-beam strength ratio for experimental results presented in Elremaily and Azizinamini
through beam connection detail until additional experimental data 共2000兲 indicated a drop in the axial tensile strains measured in the
become available to justify lower values. beam flange right at the outside and the inside of the tube wall.
To prevent joint failure, the force transfer mechanism through The variation of the axial stress in the flange along the beam
the joint and the joint strength need to be established. The follow- longitudinal axis as obtained from the finite element model of
ing sections discuss the joint behavior. Specimen NSF5 is shown in Fig. 7. It is noticed that, on the
tension side, there is a significant stress drop in the beam flange
Force Transfer Mechanism once the flange enters the tube. The axial stress in the tension
flange just inside the tube is approximately 70% of the stress
Moment and Shear Force Transfer value just outside the tube. These observations indicate that a
A moment connection should be designed to resist the forces portion of the axial tensile force in the flange is transferred to the
transferred to the joint by adjacent members. The forces acting on tube wall, V t . The gradual reduction in the axial tensile stresses
an interior joint in a frame subjected to lateral loads are shown in in the flange, inside the tube, is due to transfer of the flange force
Fig. 4. In the discussion to follow, the effect of the column axial to the concrete core by friction, V c f , and to the beam web by
load is ignored. It should be noted that high levels of axial load shear, V w . The rest of the tension flange force is applied as com-
could lower the joint shear capacity. The effect of axial load on pression force to the backside of the tube. At the compression
joint behavior is a subject that requires additional investigation. side, the stress drops almost to zero right at the inside of the tube
For simplicity, the forces from both beams are assumed to be of wall. This suggests that the axial compression force in the flange
equal magnitudes. Also, the forces from the top and bottom col- is transferred to the concrete core by bearing of the tube wall
umns are assumed to have equal magnitudes. Such kind of load- against the concrete. The same behavior was also observed from
ing causes larger shear forces and shear deformations in the con- experimental results 共Elremaily and Azizinamini 2000兲. The force
nection panel as shown in Fig. 4. The mechanism of moment
transfer could be visualized from the free body diagram shown in
Fig. 5, which represents the external forces acting on the bound-
ary of the connection panel. In this figure, the beam moment is
resolved into flange forces, T f and C f , while the column mo-
ments are replaced with the vertical tension in the tube, T s , ver-
tical compression in the tube, C s , and the concrete compressive
stress block, C c .

Flange Force Transfer

Fig. 6 shows an isolated portion of the top flange within the
connection panel. As shown in Fig. 6, the axial flange forces, T f
Fig. 6. Isolated portion of top flange 共vertical elevation兲
and C f , tend to push the beam flange through the column. Me-

218 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004

J. Struct. Eng., 2004, 130(2): 213-222

 reducing joint shear forces. Assuming that the beam bending mo-
ment is carried entirely by the flanges, the tensile and compres-
sive forces in the beam flange, T f and C f , can be estimated as
Mb
T f ⫽C f ⫽ (2)
db
Referring to Fig. 8, the effective horizontal shear force acting
on the joint panel, V u , may be calculated as
2M b
V u⫽ ⫺V c (3)
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

db

Joint Nominal Strength

From the equilibrium of the horizontal forces shown in Fig. 8, the
horizontal shear force in the joint is carried by: 共1兲 shear in the
steel web, V w ; 共2兲 shear in the concrete panel, V cs ; and 共3兲 shear
in the steel tube, V t . The joint shear capacity is reached when all
the contributing mechanisms have reached their individual shear
strengths. Test results indicated that the shear strengths of indi-
vidual mechanisms are achieved at different joint deformations.
In the test series conducted, the web panel was always the first
mechanism to reach its capacity, followed by tube yielding, and
then finally concrete crushing. It should be noted that both of the
web and tube panels maintained their strength after yielding up to
the joint global failure. Based on this discussion, the nominal
Fig. 7. Variation of axial stress in top flange along beam axis for joint shear strength, V n , is calculated as the sum of the individual
NSF5 nominal shear strengths of the contributing mechanisms. This can
be expressed as follows:
V n ⫽V wn ⫹V tn ⫹V csn (4)
C b represents the summation of the forces transferred from both
of the tension and compression flanges to the concrete core by where V wn , V csn , and V tn ⫽nominal shear strengths of the web,
bearing. concrete core, and the steel tube, respectively.
At the point of joint failure, the applied horizontal joint shear
Joint Shear Strength is equal to the joint shear capacity. Equating V u from Eq. 共3兲 and
Flange forces transferred through the joint produce large shear in V n from Eq. 共4兲 results in the following expression:
the panel zone of a CFT-to-beam moment connection. If the joint 2M b
is unable to resist such shear, failure will take place due to exces- ⫺V c ⫽V wn ⫹V tn ⫹V csn (5)
sive shear deformations in the connection panel. The joint shear db
strength is evaluated through studying the equilibrium of the hori- In the above expression, V c can be related to M b based on the
zontal forces on a horizontal plane at the midheight of the joint as geometry of the beam-column subassembly. Once the individual
shown in Fig. 8. terms on the right hand side are established, the joint strength can
be calculated in terms of the maximum beam moment, M b , that
Applied Joint Shear could be resisted by the connection. The applied beam moment
Column shear transferred through the joint increases joint shear should be less than this nominal value in order to prevent joint
strength by reducing the beam flange forces transferred to the failure.
joint. In through beam connection detail, since the column is con-
tinuous through the joint and is directly attached to the beam, it is Contribution of Beam Web Panel, Vwn. The major portion of
reasonable to consider that the entire column shear is effective in web resistance to joint shear forces is provided by means of panel
shear. Additional resistance to joint shear is provided by the frame
mechanism shown in Fig. 4. Such mechanism is formed due to
local bending of the beam flanges about their own axis right at the
beam column interface as shown in Fig. 4. Results obtained from
finite element analysis of specimen NSF7 confirms this finding.
Fig. 9 shows the distribution of the axial stresses at the top and
bottom surfaces of the beam tension flange for specimen NSF7 as
obtained from finite element analysis. The stresses at the bottom
of the tension flange reverse to compression near the column face
indicating local bending of the beam flange. Similar observations
have been reported by Sheikh 共1987兲.
The web panel strength, V wn , is provided by means of web
shear yielding, V wy , and flexural rigidity of the flanges at the
connection panel boundaries, V f . The web panel strength is given
Fig. 8. Equilibrium of horizontal forces
by the following equations:

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004 / 219

J. Struct. Eng., 2004, 130(2): 213-222

 ␲d c t t
V tn ⫽0.6F yt (10)
2

In the above expression, the resistance provided by the tube

wall is calculated based on an average yield shear stress of 0.6F yt
acting over the effective shear area of the tube given by ␲d c t t /2
共Boresi, Schmidt, and Sidebottom 1993兲.

Contribution of Concrete Core Panel, Vcsn. The concrete panel

provides resistance to the applied joint shear forces by the forma-
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

tion of a diagonal compression strut. The formation of the com-

pression strut was observed experimentally through the formation
of diagonal cracks in the concrete core within the panel zone of
Specimens NSF6 and NSF7, as shown in Fig. 10. The formation
of the compression strut is confirmed by the direction of the mini-
mum principal stress in the concrete panel obtained from the fi-
Fig. 9. Axial stress (␴ x ) in tension flange in specimen NSF7 nite element model as shown in Fig. 10. This compression strut is
similar to that used to model shear in reinforced concrete joints
V wn ⫽V wy ⫹V f (6) 共ACI-ASCE 1985兲. The horizontal force resisted by the strut,
V csn , is calculated by the following:
V wy ⫽0.6F yw d c t w
冉 冊
(7)
␲d 2c
4M p f V csn ⫽24冑 f ⬘c 共 f ⬘c in psi兲
Vf⫽ (8) 4
(11)
db

M pf⫽
F y f t 2f b f
4
(9)
V csn ⫽1.99冑 f ⬘c 冉 冊
␲d 2c
4
共 f ⬘c in MPa兲

The web shear yield is calculated based on an average yield The term 24冑 f ⬘c psi (1.99冑 f c⬘ MPa) in Eq. 共11兲 is the limiting
shear stress of 0.6F yw acting over the web area within the joint horizontal shear stress over the horizontal projection of the strut
panel. The shear yield stress of 0.6F yw is based on von Mises region. This term is the same as that recommended by the ACI-
yield criteria and is the same value used in the American Institute ASCE 共1985兲 for reinforced concrete joints confined due to
of Steel Construction 共AISC兲 LRFD specifications 共AISC 1994兲. beams framing into all four sides of the joint. Since the joint in a
The flanges’ resistance is calculated based on a frame mechanism CFT column is highly confined by the tube wall, it is reasonable
where plastic hinges form in the beam flanges at four locations as to use the same value recommended for confined joints. The ex-
shown in Fig. 4. Sheikh 共1987兲 and Deirlein 共1988兲 used the same perimental data reported by Sheikh 共1987兲 for connections be-
expression given in Eq. 共8兲 to calculate the force V f . tween steel beams and reinforced concrete columns indicated that
the shear stress term varied between 24冑 f ⬘c and 36冑 f ⬘c psi
Contribution of Steel Tube, Vtn. The tube contribution to the (1.99冑 f ⬘c to 2.99冑 f ⬘c MPa). Experimental and analytical results
joint shear strength is calculated as the maximum horizontal shear have shown that connection panel deforms as a monolithic unit.
force that can be resisted by the tube cross section, V tn , and is Thus, in Eq. 共11兲, it is assumed that the entire area of the concrete
expressed as follows: core is effective in resisting joint shear.

Fig. 10. Formation of compression strut

220 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004

J. Struct. Eng., 2004, 130(2): 213-222

 This paper discusses the performance of various connection
details that could be used to connect steel beams to CFT columns
and elaborate on the behavior of one detail—through beam con-
nection detail—that is suitable for high seismic areas.

Acknowledgments

The funding for the work carried out at the University of

Nebraska-Lincoln was provided by the National Science Founda-
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

tion, under award No. 9520280, with Dr. S. C. Liu as the program
director. The test specimens for the UNL investigation were do-
nated by Valmont industry. Additional support for the work com-
pleted at UNL was provided by the National Bridge Research
Fig. 11. Horizontal joint shear force distribution for NSF7 Organization 共NaBRO兲 at UNL. Investigation conducted at the
University of Illinois is based upon research funded by the Na-
tional Science Foundation under Grant No. NSF CMS 93-00682
with Dr. S. C. Liu as the program director. The writers gratefully
acknowledge the support provided. Opinions expressed in this
Relative Contributions
paper are those of the authors, and do not necessarily reflect those
The horizontal forces resisted by individual mechanisms were of NSF or other sponsors.
calculated from the finite element model and are shown in Fig. 11
for Specimens NSF7. The additional resistance to the joint shear
provided by the frame action resulting from local bending of the Notation
flange about its own axis was calculated by summing up the local
bending moments in the beam flanges at the column face and The following symbols are used in this paper:
dividing the resultant by the beam depth to get an equivalent A c ⫽ cross sectional area of the concrete core;
horizontal couple. Fig. 11 shows that all the elements contributing A s ⫽ cross sectional area of the steel tube;
to the joint shear strength are activated from the beginning of the d ⫽ total beam depth;
loading. The web was the first mechanism to reach its capacity. d b ⫽ distance between centroids of beam flanges;
After web yielding, the increase in the shear force is distributed d c ⫽ diameter of the tube wall;
between the tube and the concrete until the tube yields. After tube f ⬘c ⫽ Concrete compressive strength;
yielding, the additional shear force in the joint is resisted solely F y ⫽ yield stress;
by the concrete core. F y f ⫽ yield stress of the beam flange;
F yt ⫽ yield stress of the steel tube;
Design Approach F yw ⫽ yield stress of the beam web;
H ⫽ column height between points of lateral support;
The section on ‘‘Force Transfer Mechanism’’ outlines mecha- M n ⫽ nominal moment capacity of the column;
nisms by which the beam forces are transferred to column joint M p ⫽ beam plastic moment;
for through beam connection detail. This section also presents M test ⫽ moment due to ultimate load P test ;
equations that could be used to calculate the shear capacity of the P axial ⫽ axial load applied on column during the test;
joint for through beam connection detail. Elremaily and Aziz- P bn ⫽ beam load required to develop beam plastic
inamini 共2000兲 provide an approach that could be used to calcu- moment;
late the CFT column capacity. Calculating the beam capacity is P cn ⫽ beam load required to develop column flexural
discussed in steel text books. Using the ability to predict column, strength;
beam, and joint capacity, Elremaily and Azizinamini 共2000兲 pro- P j ⫽ beam load required to fail the joint in shear;
vide step by step procedures that could be used to design through P 0 ⫽ nominal pure axial capacity of the column;
beam connection detail. The same reference also provides an ex- P pred ⫽ least of P j , P cn , and P bn ;
ample problem demonstrating the application of the suggested P test ⫽ ultimate load obtained during test;
design provisions. t f ⫽ beam flange thickness;
t w ⫽ beam web thickness; and
t t ⫽ thickness of the tube wall.
Summary

The work reported in this paper is part of an investigation con-

References
ducted under Phase five of the U.S.-Japan Cooperative work in
Earthquake Engineering and supported by the National Science
American Concrete Institute/American Society of Civil Engineers
Foundation 共NSF兲. The results presented are a brief summary of
Committee 352 共ACI-ASCE 352兲. 共1985兲. ‘‘Recommendations for de-
work carried out at the University of Nebraska-Lincoln and Uni- sign of beam-column joints in monolithic reinforced concrete struc-
versity of Illinois. The responsibility of these two universities was tures.’’ ACI J., 83共3兲, 266 –283.
to concentrate on development of connection detail and accom- American Concrete Institute Committee 318 共ACI 318兲. 共1995兲. Building
panying design provisions for connecting steel beams to circular Code Requirements for Structural Concrete (ACI 318-95) and Com-
CFT columns. mentary (ACI 318R-95), American Concrete Institute, Detroit.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004 / 221

J. Struct. Eng., 2004, 130(2): 213-222

 American Institute of Steel Construction 共AISC兲. 共1994兲. ‘‘Load and re- Kanatani, H., Tabuchi, M., Kamba, T., Hsiaolien, J., and Ishikawa, M.
sistance factor design.’’ American Institute of Steel Construction, Chi- 共1987兲. ‘‘A study on concrete-filled RHS columns to H-beam connec-
cago. tions fabricated with HT bolts in rigid frames.’’ Composite construc-
Ansourian, P. 共1976兲. ‘‘Connections to concrete-filled tube columns.’’ Int. tion of steel and concrete, ASCE, New York, 614 – 635.
Assoc. Bridge Struct. Eng., 36-I, 1–22. Kato, B., Kimura, M., Ohta, H., and Mizutani, N. 共1992兲. ‘‘Connections
Applied Technology Council 共ATC-24兲. 共1992兲. ‘‘Guidelines of cyclic of beam flange to concrete-filled tubular columns.’’ Composite con-
seismic testing on components for steel structures.’’ Applied Technol- struction of steel and concrete II, ASCE, New York, 528 –538.
ogy Council, Redwood City, Calif. Morino, S., Kawaguchi, J., Yasuzaki, C., and Kanazawa, S. 共1992兲. ‘‘Be-
Aziznamini, A., and Prakash, B. 共1992兲. ‘‘Tentative design guidelines for havior of concrete-filled steel tubular three dimensional subassem-
a new steel beam connection detail to composite tube columns.’’ Eng. blages.’’ Composite construction of steel and concrete II, ASCE, New
J., 30共3兲, 108 –115. York, 726 –741.
Downloaded from ascelibrary.org by UFMG - Universidade Federal de Minas Gerais on 04/10/22. Copyright ASCE. For personal use only; all rights reserved.

Boresi, A. P., Schmidt, R. J., and Sidebottom, O. M. 共1993兲. Advanced Prion, H. G. L., and McLellan, A. B. 共1992兲. ‘‘Connecting steel beams to
mechanics of materials, 5th Ed., Wiley, New York. concrete-filled steel columns.’’ Proc., ASCE Structures Congress on
Bridge, R., and Webb, J. 共1992兲. ‘‘Thin walled circular concrete-filled Composite Compression Members, New York, 918 –921.
steel tubular columns.’’ Composite construction of steel and concrete Shakir-Khalil, H. 共1992兲. ‘‘Full-scale tests on composite connections.’’
II, ASCE, New York, 634 – 649. Composite construction of steel and concrete II, ASCE, New York,
Deirlein, G. G. 共1988兲. ‘‘Design of moment connections for composite 539–554.
framed structures.’’ PhD Dissertation, Department of Civil Engineer- Sheikh, T. M. 共1987兲. ‘‘Moment connections between steel beams and
ing, The University of Texas at Austin, Austin, Tex. concrete columns.’’ PhD Dissertation, Department of Civil Engineer-
Elremaily, A., and Azizinamini, A. 共2000兲. ‘‘Development of detail and ing, The University of Texas at Austin, Austin, Tex.
design criteria for steel beam to concrete filled tube column connec- Valbert, G. 共1968兲. ‘‘Essais d’assemblages soudes d’une solive sur un
tions in seismic regions.’’ A final Rep., submitted to National Science poteau en tube rempli de beton.’’ Constr. Metal., 4 Dec., 27–38. 共in
Foundation. French兲.

222 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2004

J. Struct. Eng., 2004, 130(2): 213-222