You are on page 1of 10

Seismic Behavior of Composite Concrete Filled Steel Tube

Column-Wide Flange Beam Moment Connections

J. M. Ricles, M.ASCE1; S. W. Peng2; and L. W. Lu, M.ASCE3

Abstract: The results of an experimental research study involving the testing of ten full-scale moment resisting connections under
simulated seismic loading conditions are presented. Each test specimen modeled the interior joint of a moment resisting frame consisting
of square concrete filled steel tube columns and wide flange steel girders, where energy dissipation was designed to occur either primarily
Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

in the beams or in the connection panel zone during a severe earthquake. The results of the study indicate that moment resisting
connection details can be economically designed that enable more than 0.045 rad of inelastic story drift to develop under cyclic loading.
These details include split-tee connections without a shear tab, weak panel zone connections, and extended tee connections. Panel zone
shear yielding and local buckling are shown to be a ductile mode of response, with minimal strength deterioration occurring in the
connection. The use of interior diaphragms in the column is shown to locally stiffen the joint, but also lead to strain concentrations and
fracture of the beam flanges at their weld access holes. Furthermore, strain concentrations develop at connection details that lack a gradual
transition in geometry and result in a reduction in connection ductility.
DOI: 10.1061/共ASCE兲0733-9445共2004兲130:2共223兲
CE Database subject headings: Composite structures; Tubes; Beams; Connections; Seismic design; Concrete.

Introduction tute of Japan 共AIJ兲 provisions 共1987兲. Research by Kato 共1982兲,

Steel–concrete composite materials have been widely used in Matsui 共1985兲, and Yokoyama et al. 共1991兲 on welded beam-to-
framing systems for high-rise buildings over the past several de- column CFT connections having interior and exterior diaphragms,
cades. A common form of composite construction consists of a respectively, have shown that these elements develop a complex
steel frame with composite columns. This system has the at- stress state and are susceptible to fracture or local buckling. Cy-
tributes of a lightweight, ductile steel frame with the added stiff- clic tests on CFT moment connections were conducted by Kana-
ness of steel–concrete composite columns to control story drift. tani et al. 共1987兲, as well as Morino et al. 共1996兲, in which shear
One particular type of composite column is a concrete filled tube yielding of the steel tube within the connection’s panel zone oc-
共CFT兲. This type of composite column has the benefit of the tube curred. Their test results demonstrated that a ductile hysteretic
serving as a convenient formwork for placing the concrete and behavior in the specimens can be achieved, as long as shear buck-
providing confinement for the cured concrete. By confining the ling and fracture of the steel tube within the panel zone is inhib-
concrete in a CFT, an increase in the concrete’s compressive ited.
strength may be realized in addition to preventing the concrete Some of the above studies lead to the recommendations in the
from spalling while subjected to overload. Furthermore, the con- AIJ provisions for the panel zone seismic strength V t of a welded
crete inside the tube acts to restrain local buckling from occurring beam-to-column CFT with interior diaphragms:
in the wall of the steel tube.
Extensive research has been done in Japan to study the behav- b ␴y
V t ⫽2.5A c F c ⫹1.2A web (1)
ior of moment connections between a CFT column and wide d 冑3
flange 共WF兲 beams under seismic loading conditions. Seismic re-
sistant design recommendations appear in the Architectural Insti- The first term in Eq. 共1兲 is associated with the concrete contribu-
tion and the second is the steel tube contribution to the panel zone
Professor, Dept. of Civil and Environmental Engineering, ATLSS shear strength. In Eq. 共1兲 A c , A web , F c , ␴ y , b, and d⫽cross-
Center Faculty Associate, Lehigh Univ., 117 ATLSS Dr., Bethlehem, PA sectional area of the concrete, the area of the web of the steel
18015. E-mail: tube, the concrete shear resistance 共in kgf/cm2兲, the yield stress of
Former Graduate Research Assistant, ATLSS Center, Lehigh Univ., the steel tube, depth of the steel tube, and depth of the beam,
117 ATLSS Drive, Bethlehem, PA 18015. respectively. The concrete resistance F c is equal to
Professor, Dept. of Civil and Environmental Engineering, ATLSS
Center Faculty Associate, Lehigh Univ., 117 ATLSS Dr., Bethlehem, PA F c ⫽min共 0.12f ⬘c ;18⫹0.036f ⬘c 兲 (2)
Note. Associate Editor: Sherif El-Tawil. Discussion open until July 1, where f ⬘c ⫽concrete cylinder strength 共in kgf/cm2兲.
2004. Separate discussions must be submitted for individual papers. To
As an alternative to exterior and interior diaphragm connection
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted
details in a WF beam-to-CFT column connection, the studies by
for review and possible publication on September 11, 2002; approved on Kanatani et al. 共1987兲 included tests on specimens with bolted
January 27, 2003. This paper is part of the Journal of Structural Engi- connection details. The tests lead to the following design recom-
neering, Vol. 130, No. 2, February 1, 2004. ©ASCE, ISSN 0733-9445/ mendation for panel zone shear capacity for split-tee connection
2004/2-223–232/$18.00. details:


J. Struct. Eng. 2004.130:223-232.

␴y Table 1. Test Specimen Matrix
V t ⫽S 共 w⫺2t 兲 cos2 共 ␣ 兲 f ⬘c ⫹A web (3)
冑3 Spec Connection detail
1 Interior diaphragms 共four-sided CJP weld兲, weak beam
The first term in Eq. 共3兲 is associated with the concrete contribu- 2 Interior diaphragms 共three-sided CJP weld兲, weak beam
tion and the second is the steel tube contribution to the panel zone 1R Interior diaphragms 共four-sided CJP weld兲, weak panel zone
shear strength. In Eq. 共3兲 S, w, t, ␣, f ⬘c , A web , and ␴ y ⫽concrete 2R Interior diaphragms 共three-sided CJP weld兲, weak panel zone
compression strut width, width of the steel tube, thickness of the
3 Extended tee, weak beam
steel tube, concrete strut angle, concrete compressive strength, the
3R Extended tee with taper, weak beam
area of the web of the steel tube, and the yield stress of the steel
4 Bolted split-tee connection with shear tab, weak beam
tube, respectively. The strut width S is based on the vertical spac-
5 Bolted split-tee connection without shear tab, weak beam
ing of the bolts in the tee flange that pass through the column, and
6 Welded split-tee connection without shear tab, weak beam
␣ the angle of inclination between opposite diagonal split tees in
7 Welded split-tee connection without shear tab, weak beam
the connection.
Research on the seismic performance of CFT column-to-beam
Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

moment connections is not as extensive in the U.S. as that in

Japan. Schneider and Alostaz 共1998兲 performed a series of cyclic including: interior diaphragms, exterior extended structural tees,
tests on exterior beam-to-column connections involving a circular and split tees. The performance of the connection details is evalu-
CFT column and a beam on only one side of the column. Various ated by a comparison of the specimen cyclic strength, stiffness,
connection details were studied, where each connection detail was ductility, and energy dissipation capacity.
intended to develop the full flexural strength of the connected
beam. The connection detail that was determined to provide the
best cyclic strength and ductility consisted of a detail where the Experimental Program
beam was passed through the column. This was accomplished by
making a slotted hole in the walls of the tube to pass the steel The test matrix for the experimental program is given in Table 1.
beam section. Once the beam was passed through the column, it Ten interior connections were tested, where each specimen had a
was fillet welded to the outside wall of the tube. Additional tests two-sided moment connection 共i.e., beams on both sides of the
on this detail were performed by Elremainly and Azizinamini column兲. The specimens were full-scale replicas of connections in
共2000兲, who also found it to perform well under inelastic cyclic the middle to upper floors of a 6 –12 story MRF. Each of the test
loading. specimens consisted of two W24⫻62 beam sections attached to a
Experimental and analytical studies were conducted by 406⫻406⫻12.5 mm CFT column at midheight to form a
Koester 共2000兲 to study the panel zone behavior of square CFT cruciform-shaped test specimen. The connection details for the
column-to-WF beam split-tee bolted moment connections under ten tests specimens are shown in Fig. 1– 6. Specimens 1, 2, 1R,
cyclic loading. The objective of his research was to determine the and 2R had interior diaphragms in the joint region, where the
role of the concrete within the joint region in transferring joint beams of the latter two specimens were reinforced using tapered
shear forces due to earthquake loads. The results from their tests plates in order to study the strength and ductility of the composite
indicated that the failure of the panel zone was the result of a panel zone. Specimens 1R and 2R were therefore designed to
combination of shear yielding in the steel, bearing failure, and develop yielding primarily in the panel zone. The other specimens
some cracking within the concrete core. A modified model was 共Specimens 1, 2, 3, 3R, 4, 5, 6, and 7兲 were designed to have
suggested to predict the shear strength contributed by concrete for yielding develop primarily in the beams. It was assumed that a
a CFT panel zone with split-tee connection details, where the plastic hinge formed in the beam at the end of the connection. The
recommended panel zone strength is connection elements of these eight specimens were designed to

V t ⫽28Bh 冑 f ⬘c ⫹A web (4)
where in Eq. 共4兲 B and h⫽width and depth of the concrete core
within the CFT column panel zone; A web⫽area of the web of the
steel tube; f ⬘c ⫽concrete compressive strength 共in psi兲; and
␴ y ⫽yield stress of the steel tube.
In order to motivate designers to take advantage of the features
of a CFT column–WF beam moment resisting frame 共MRF兲 sys-
tem for seismic design, the connections between the beam and
column members must be reliable and economical. Otherwise, the
savings from reduced steel tonnage will be negated if the connec-
tion details are too costly to fabricate and erect. This paper pre-
sents the results of a study of the inelastic cyclic behavior of
connections in square CFT column–WF beam MRF systems. The
study was conducted in conjunction with the U.S.–Japan Coop-
erative Research Program on Composite and Hybrid Structures
共‘‘Recommendations,’’ 1992兲, with the complete details of the
study reported by Peng et al. 共2001兲. The purpose of this study
was to examine the effect that different connection details have on
Fig. 1. Specimens 1 and 2 connection details
cyclic performance. Various connection details were investigated,


J. Struct. Eng. 2004.130:223-232.

Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Specimens 1R and 2R connection details Fig. 4. Specimen 3R connection details

resist a beam moment of 1.5M pn at the end of the connection fer the beam’s tension flange force directly to the sides of the steel
共AISC 1997兲, where M pn is the nominal plastic flexural capacity tube within the connection’s panel zone, where the additional ta-
of the beam. pered plates in Specimen 3R were intended to provide a more
The diaphragms of Specimens 1 and 1R were welded around gradual transition in the geometry of the connection region.
their perimeter to the inside of the steel tube using the E70TG-K2 Specimens 4, 5, 6, and 7 had split-tee moment connection
filler metal. Only three sides of the diaphragm for Specimens 2 details 共Figs. 5 and 6兲. The split-tee connections were designed to
and 2R were welded, where the full-penetration weld on the north activate a diagonal concrete compression strut within the connec-
side adjacent to the panel zone 共i.e., web兲 of the steel tube was tion’s panel zone under the action of overturning moment. This
omitted, see Figs. 1 and 2. The purpose of this detail was to was achieved by the use of A490 bolts to develop a horizontal
evaluate the extent that this omitted weld would have on the tension force through the joint. These bolts were passed through
performance of the connection in a box column. In box column the column with the use of PVC conduits placed prior to casting
construction this weld is placed by electroslag welding as the concrete, and tensioned after curing of the concrete. The split-tee
section is closed during fabrication. detail was designed to avoid prying action in the A490 bolts
Specimens 3 and 3R had an extended-tee moment connection 共AISC 1994兲. In Specimens 4 and 5 the stem of the tees were
detail 共Figs. 3 and 4兲. The extended tees for Specimen 3 and 3R attached to the beam flanges using 22-mm-diam A325 bolts with
consisted of ST7.5⫻25 sections that were attached to the beam 2 mm oversized bolt holes, whereas in Specimens 6 and 7 a 12
flanges and column by complete joint penetration welds. Speci- mm fillet weld was used. Net section fracture of the flanges of
men 3R was identical to Specimen 3, except that the former had Specimens 4 and 5 was checked using the following criteria,
250-mm-long tapered plates that were attached to the end of the which is based on AISC LRFD Seismic Provisions 共AISC 1992兲
extended tees and edge of the beam flange using complete joint for gusset plates:
penetration welds. The extended-tee detail was designed to trans-

Fig. 3. Specimen 3 connection details Fig. 5. Specimens 4 and 5 connection details


J. Struct. Eng. 2004.130:223-232.

are 27–54 and 122–163 J, respectively, at about ⫺28°C. The
shear studs inside the tube in the joint region were designed to
transfer the beam shear force to the composite column associated
with flexural hinging at the ends of the prototype beam.
The bolts in the connections were pre-tensioned to the required
force in accordance with the AISC Specification 共2000兲, and
checked using a torque wrench. All of the specimens with shear
tabs 共i.e., Specimens 1, 2, 1R, 2R, 3, 3R and 4兲 had supplemental
fillet welds placed around the perimeter of the shear tab to de-
velop 20% of the beam web flexural capacity in accordance with
the 1992 AISC Seismic Specification. The shear tabs for the
beams of Specimens 5, 6, and 7 were omitted.
The material properties for the specimens are summarized in
Table 2, where the concrete cylinder strength f c⬘ at the day of
Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

testing as well as the yield stress ␴ y and ultimate stress ␴ u for the
steel connection components 共structural tees, extended tees, dia-
phragm plates, tapered plates兲 and the beams and columns are
given. Concrete cylinders were cored out of the ends of the speci-
Fig. 6. Specimens 6 and 7 connection details men columns 共where no damage occurred兲 after testing to deter-
mine f ⬘c of each specimen. The steel beams for the specimens
were fabricated from A36 steel and the columns from ASTM
A n␴ u A500 Gr. B steel. The yield strength for the beams of Specimens
␰⫽ ⭓1.2 (5) 6 and 7 was notable smaller than that for the other specimens and
A g␴ y
equal to 266 and 230 MPa for the web and flanges, respectively.
where A n , A g , ␴ u , and ␴ y ⫽flange net area, flange gross area, The structural tees were cut from a W24⫻146 section of ASTM
flange tensile strength, and flange yield strength, respectively. A572 Gr. 50 steel that had a measured yield strength of 342 MPa.
Sacrificial brass shims were placed in the shear plane between the The average measured yield strength of the ST7.5⫻25 sections
tee stem and beam flanges to prevent gouging of their surfaces from which the extended tees were cut was 216 and 224 MPa for
against each other. In addition, washer plates were used in Speci- the web and flanges, respectively.
mens 4 and 5 to prevent local beam flange buckling from occur- The test setup is shown in Fig. 7. This test setup, which in-
ring in the net section at the bolt line, which would result in a cluded calibrated load cells, enabled the member and connection
local flange curvature and high tensile strains. The washer plates forces to be determined. As noted previously, the joint in each
of Specimen 5 were fillet welded to the beam flanges. Specimens specimen simulated the connection between an interior column
6 and 7 were similar to each other. and the two adjacent wide flange beams in a MRF. The length L
The beam flanges of Specimens 1, 2, 1R, 2R, 3, and 3R were and height h of the test specimen were based on assuming points
welded to the column using complete joint penetration welds, of inflection at midspan and midheight of the prototype beams
where the backing bars were removed and the exposed area back and columns, respectively, where L⫽6096 mm and h
gouged and a profiled 6 mm fillet weld placed to improve the ⫽3658 mm. Through the use of cylindrical bearings in the test
quality of the flange weld detail. The geometry of the weld access setup, pin boundary conditions were created at the ends of the
holes for these specimens were in accordance with the AISC members where the assumed inflection points were located.
Specification 共2000兲. The complete joint penetration welds for A typical test was conducted by applying a constant 2000 kN
Specimens 1 and 2 were created using the E70TG-K2 filler metal. axial force to the top of the column using two hollow core jacks
E7018 filler metal was used for Specimens 1R, 2R, 3, and 3R. and a set of tension rods to simulate gravity loading 共see Fig. 7兲.
The E70TG-K2 and E7018 electrodes produce a weld metal of A lateral cyclic displacement history was then imposed to the top
high toughness, where typically the Charpy V-notch toughnesses of the column by an actuator to produce lateral seismic loading

Table 2. Material Properties

Steel–yield stress/ultimate stress 共MPa/MPa兲
Structural tees Extended tees Beam Column
f c⬘ Diap. Tapered
Spec 共MPa兲 Web Flange Web Flange plates plates Web Flange Web Flange
1 50 — — — — 280/463 — 342/475 293/456 376/492 352/477
2 58 — — — — 280/463 — 325/473 282/456 384/492 352/467
1R 61 — — — — 280/463 346/503 342/475 293/456 376/492 352/477
2R 61 — — — — 280/463 346/503 325/473 282/456 384/492 352/467
3 52 — — 216/405 224/430 — — 325/473 282/456 388/503 352/469
3R 58 — — 216/405 224/430 — — 325/473 282/456 388/503 352/469
4 58 342/517 343/519 — — — — 310/465 283/469 382/496 352/471
5 58 342/517 343/519 — — — — 299/452 264/437 382/496 352/471
6 58 342/517 343/519 — — — — 266/450 230/421 382/496 352/471
7 58 342/517 343/519 — — — — 266/450 230/421 382/496 352/471


J. Struct. Eng. 2004.130:223-232.

V p⫽ A web (6)
in which A web⫽web area 2(b⫺t)t of the steel tube of depth b and
thickness t. V t for specimens with interior diaphragms and ex-
tended tees 共i.e., Specimens 1, 2, 1R, 2R, 3, and 3R兲 is based on
Eq. 共1兲, while V t for the specimens with split-tee connection de-
tails 共i.e., Specimens 4, 5, 6, and 7兲 is based on both Eqs. 共3兲 and
共4兲, 共see columns 10 and 11 in Table 3兲. For determining V t based
on Eq. 共3兲, values of 200 mm and 56° were used for S and ␣,
The subassembly stiffness K 0 was found to range from 9826
kN/m 共Specimen 2兲 to 11,733 kN/m 共Specimen 2R兲, where the
19% increase in stiffness in Specimen 2R is associated with the
Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

addition of the tapered plates to the interior diaphragm connection

Fig. 7. Test setup
detail. The split-tee connection detail had a 3% 共Specimen 4兲 to
12% 共Specimen 7兲 greater stiffness than the interior diaphragm
detail of Specimen 1, and 5–15% greater stiffness than the inte-
effects. The displacement history was similar to that recom- rior diaphragm detail of Specimen 2. This indicates that the split-
mended by ATC-24 共ATC 1992兲, and consisted of six initial elas- tee connection detail provided a comparable stiffness to that of
tic cycles, followed by a series of inelastic sets of cycles having a the traditional Japanese rigid connection for WF beams-to-CFT
gradual increase in amplitude. columns, which uses interior diaphragms similar to Specimen 1.
The testing of the weak panel zone specimens 共Specimens 1R
and 2R兲 resulted in minor beam flange yielding at the column face
Test Results prior to the occurrence of significant shear yielding and cyclic
local shear buckling of the steel tube in the panel zone. The extent
Response quantities for the test specimens are summarized in of shear deformation and yielding of the panel zone is evident in
Table 3, and include subassembly initial elastic stiffness K 0 at Fig. 8共a兲, where the discoloration of the panel zone is caused by
0.75% story drift, maximum beam moment M max and maximum the flaking of whitewash from the panel zone. This yielding was
plastic rotation ␪ p,max at the end of the connection, inelastic story accompanied by diagonal shear cracks in the concrete within the
drift ␪ in,n at 80% of the specimen nominal capacity following panel zone 关see Fig. 8共b兲兴. The cyclic shear force-deformation
local beam flange buckling, maximum panel zone shear V max , (V⫺␥) response of the panel zone is given in Fig. 9 for Speci-
panel zone maximum plastic shear deformation ␥ p,max , and dissi- mens 1R and 2R. The hysteretic response of both specimens was
pated energy E d . In Table 3 the specimen maximum beam mo- stable, with the specimens sustaining their shear capacity through-
ments M max are compared to the plastic flexural capacity (M p ) out the test. However, a deterioration of the joint capacity oc-
and the nominal plastic flexural capacity (M pn) of the beam, and curred near the end of testing 共during the cycle with a story drift
the maximum panel zone shear force V max to the shear capacity of of 5%兲 of Specimen 2R. This loss of capacity was due to cracking
the steel tube within the panel zone V p and the total panel zone of the steel tube near the north edge of the tapered beam flange
shear resistance V t . M p and V p are based on measured material plates, where the diaphragm was not welded to the steel tube. It
properties and dimensions, where was determined from strain gage reading that a nonuniform strain

Table 3. Response Quantities of Test Specimens

K0 M max ␪ p,max ␪ in,n V max ␥ p,max Ed
Spec 共kN/m兲 共kN-m兲 M max /M p M max /M pn 共rad兲 共rad兲 共kN兲 V max /Vp V max /Vt V max /Vt 共rad兲 共kN m兲
1 10,053 953 1.25 1.53 0.028 0.019a 2,581 1.35 0.83c 0.006 1,026
2 9,826 809 1.18 1.30 0.035 0.003a 2,304 1.11 0.66c 0.001 184
1R 11,384 808 1.03 1.29 0.003 0.046b 3,296 1.76 1.04c 0.044 1,954
2R 11,733 544 0.89 0.87 0.005 0.057b 2,838 1.49 0.88c 0.055 1,674
3 10,158 870 1.21 1.40 0.025 0.011a 2,731 1.38 0.85c 0.003 630
3R 11,032 976 1.23 1.57 0.065 0.047 3,190 1.61 0.96c 0.011 1,387
4 10,333 974 1.34 1.57 0.055 0.042 3,034 1.53 0.88d 0.65e 0.001 1,780
5 10,858 878 1.29 1.41 0.051 0.033 2,758 1.59 0.91d 0.67e 0.001 1,825
6 11,208 840 1.44 1.35 0.054 0.033 2,709 1.45 0.97d 0.61e 0.001 1,908
7 11,295 899 1.56 1.45 0.055 0.040 2,878 1.45 0.97d 0.61e 0.001 1,918
Test stopped after flange fracture occurred.
Specimen plastic story drift dominated by panel zone deformations with minimal specimen strength degradation.
V t based on Eq. 共1兲.
V t based on Eq. 共3兲.
V t based on Eq. 共4兲.


J. Struct. Eng. 2004.130:223-232.

Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Specimen 2R: 共a兲 panel zone deformation at 5% story drift,

and 共b兲 exposed concrete core in panel zone following test
Fig. 10. Yielding and local buckling in beam flange and web at end
of connection at 5% story drift, Specimen 7
distribution developed across the width of the beam flange at the
column face, and that a strain concentration developed in the tube
near the corner adjacent to the north wall prior to when it devel- the complete joint penetration weld. Prior to fracture, the beams
oped a crack. developed appreciable yielding in the flanges and web. Specimen
The maximum joint shear developed in Specimen 1R was 1 developed a maximum beam moment and plastic rotation at the
3296 kN, which exceeded V p by 76% and the AIJ joint shear end of the connection of 1.25M p and 0.028 rad, respectively.
capacity V t by 4%. The lack of welding all of the edges of the After completing the typical elastic series of cycles during the
diaphragm to the steel tube in Specimen 2R reduced its panel testing of Specimen 2 关Fig. 11共b兲兴, an equipment malfunction
zone shear capacity. The maximum joint shear developed in occurred, resulting in an unintentional inelastic half cycle corre-
Specimen 2R was 2838 kN, exceeding V p by 49% but only equal sponding to 5% story drift being imposed in the east direction. In
to 0.88V t . Specimens 1R and 2R developed a maximum beam an attempt to maintain a symmetric series of cyclic displace-
moment at the end of the tapered plates of 1.03 and 0.89M p , ments, a deformation corresponding to 5% of story drift was im-
respectively. Consequently, the plastic beam rotation was small, posed in the west direction. During this latter half cycle a fracture
and equal to 0.003 and 0.005 rad for these specimens. The panel occurred in the east beam’s tension flange, which originated be-
zone maximum plastic shear deformation ␥ p,max for Specimens neath the access hole of the web and propagated across the flange
1R and 2R was appreciable and equal to 0.044 and 0.055 rad, width. This lead to a loss of connection capacity. The maximum
respectively. The shear yield strain ␥ y is equal to about 0.0025 rad beam moment and plastic rotation at the end of the connection for
for all of the specimens. Specimen 2 was equal to 1.18M p and 0.035 rad, respectively,
The remaining specimens 共i.e., Specimens 1, 2, 3, 3R, 4, 5, 6, which occurred during the one large cycle of loading.
and 7兲, all of which are weak beam configurations, developed Specimen 3 关Fig. 11共c兲兴 developed a fracture in the beam ten-
inelastic response in the beams outside of the connection. An sion flange, adjacent to an extended tee at the end of the connec-
example of this behavior is shown in Fig. 10, where yielding and tion during the first half cycle of 3% story drift. Fig. 3 shows the
local buckling in the beam web and flanges occurred in the plastic location of this fracture. The test was then stopped, with the
hinge near the end of the connection. The moment-plastic rotation specimen having achieved a maximum beam moment and plastic
(M ⫺␪ p ) response for these specimens is given in Fig. 11. Speci- rotation at the end of the connection of 1.21M p and 0.025 rad,
men 1 关Fig. 11共a兲兴 achieved a maximum story drift of about 4% respectively. An examination of the beam flange in the fractured
when a fracture initiated at the fusion line of the beam flange and area revealed that the material had necked at the crack, indicating

Fig. 9. Panel zone shear-deformation response for weak panel zone specimens


J. Struct. Eng. 2004.130:223-232.

Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Beam moment-plastic rotation response at end of the connection for weak beam specimens

that a significant amount of strain had developed. Readings from testing. The retrofit involved the use of tapered plates welded to
strain gages located near the crack revealed a strain had devel- the beam flanges of Specimen 3, which became Specimen 3R
oped that exceeded 20,000 microstrain before cracking occurred. after being repaired 共see Fig. 4兲. During testing, Specimen 3R
Inelastic finite element analyses were conducted by Ricles et al. developed significant beam yielding outside the connection, as
共1998兲 to verify the development of this strain concentration and well as in the flanges of the extended-tees during the inelastic
design a retrofit for Specimen 3 to avoid such behavior in further displacement cycles. Shear yielding initiated in the steel tube


J. Struct. Eng. 2004.130:223-232.

tion (M ⫺␪ p ) relationship plotted in Figs. 11共e兲–11共h兲. Each test
was stopped after a story drift of 6% was imposed to the top of
the column of these specimens.
Pinching occurred in the M ⫺␪ p relationship of Specimen 4
关see Fig. 11共e兲兴, and is associated with bolt hole elongation that
developed from the bolt bearing on the beam flanges. The elon-
gation in the bolt holes led to a slippage between the beam and
the connection under cyclic loading. At the end of the test a net
section fracture occurred in the flange bolt line, leading to a de-
terioration in strength. The welding of the washer plates in Speci-
men 5 and tee stems to the beam flanges in Specimens 6 and 7
served to reinforce the bearing capacity and increase the net area
in the beam flanges. This avoided hole elongation from develop-
ing, hence, no pinching occurred in the M ⫺␪ p relationship for
Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

these other split-tee connection specimens 关see Figs. 11共f兲–11共h兲兴.

The maximum plastic beam rotations developed at the end of
the connection in the Specimens 4, 5, 6, and 7 were 0.055, 0.051,
Fig. 12. Extended-tee strain distribution for Specimen 3R east beam 0.054, and 0.055 rad, respectively. The maximum panel zone
at selected displacement ductility ␮ in east direction; view is from the plastic shear deformation ␥ p,max that developed in these speci-
end of the connection 共note: ␮⫽1 corresponds to story drift ␪ p mens was 0.001 rad. The maximum plastic rotation that devel-
⫽0.01 rad) oped at the end of the connection of these four test specimens
coincided with the end of the test, where the strength of each
specimen had degraded to about 50% of the maximum strength
achieved during the test. This was also the case for Specimen 3R,
within the panel zone. Strain gage readings revealed that a shear whereas the maximum plastic rotation at the end of the connec-
lag phenomenon developed across the beam’s tension flange tion for the remaining specimens 共i.e., Specimens 1, 2, 1R, 2R,
within the connection, where the longitudinal strain was largest in and 3兲 occurred at their maximum capacity. The inelastic story
the flanges of the extended tees 共see Fig. 12兲. This indicated that drift ␪ in,n that developed at the end of the connection when the
a significant amount of the beam’s tension flange force was being specimen strength had degraded to the nominal specimen strength
transferred from the extended tees to the sides of the steel tube are reported in Table 3 for Specimens 3R, 4, 5, 6, and 7, in
that formed the connection’s panel zone. Specimen 3R developed addition to specimens 1R and 2R, ␪ in,n . is seen to exceed 0.033
full plastic flexural hinges in the beams at the end of the connec- rad.
tion, where pronounced flange and web yielding occurred and The beam moments that developed in the weak beam speci-
was followed by local flange and web buckling. Due to the effects mens resulted in a maximum panel zone shear force V max , which
of strain hardening, a maximum beam moment of 1.23M p devel- exceeded the plastic shear capacity V p of the structural steel tube
oped at the end of the connection in Specimen 3R prior to local by a factor of 1.11 共Specimen 2兲 to 1.61 共Specimen 3R兲 共see Table
buckling. Following local buckling in the beam flanges and web, 3兲. Specimen 3R, having the extended-tee connection detail, de-
the moment capacity of Specimen 3R degraded in subsequent veloped some inelastic response in the panel zone, where ␥ p,max is
cycles of story drift, as seen in Fig. 11共d兲. The maximum plastic equal to 0.011 rad. The specimens with the split-tee connections
beam rotation at the end of the connection ␪ p,max and panel zone did not develop panel zone shear yielding. This is attributed to the
plastic deformation ␥ p,max was equal to 0.065 and 0.011 rad, re- enhancement of the joint shear resistance by the diagonal com-
spectively. pression strut that developed in the concrete under the action of
During the testing of the remaining specimens, which had the overturning moment. The split-tee connection detail allows a
split-tee connection details 共i.e., Specimens 4, 5, 6, and 7兲, sig- better mobilization of the diagonal compression strut. The shear
nificant yielding occurred in the inelastic displacement cycles at capacity V t of the composite joint using Kanatani’s model 关i.e.,
the base of the tee’s stem. These specimens also developed full Eq. 共3兲兴 is 3–12% greater than the maximum panel shear V max
plastic flexural hinges in the beams at the end of the connection, developed in the test specimens having split-tee details 共see col-
where pronounced flange and web yielding occurred and was fol- umn 10, Table 3兲. The panel zone capacity predicted by Koester
lowed by local flange and web buckling 共similar to that shown in 共2000兲 关i.e., Eq. 共4兲兴 is shown in Table 3 共see column 11兲 to be
Fig. 10兲. Prior to local buckling, the beams of these specimens significantly less than that predicted by Kanatani’s model for the
developed a maximum moment M max at the end of the connection split-tee connection test specimens.
ranging from 1.29M p 共Specimen 5兲 to 1.56M p 共Specimen 7兲, The maximum beam moment (M max) at the end of the connec-
where the latter is unusually high compared to the other speci- tion is shown in Table 3 to range from 1.3 to 1.57M pn in speci-
mens. The large values of M max compared to M p is due to the mens with the weak-beam configuration. This implies that using
yield-to-tensile strength ratio (␴ y /␴ u ) of the beam flange being M max of 1.5M pn for the A36 steel beams to design the connections
equal to 0.54 for Specimens 6 and 7, while the average value was was reasonable.
0.6 for Specimens 4 and 5, and 0.63 for the remaining specimens. The A490 bolts that anchored the tees to the column of the
Lower values for ␴ y /␴ u result in a higher value for the ratio of specimens with the split-tee connection detail were found from
M max /M p . Following beam initial local web and flange buckling measurements not to develop any significant increase in bolt force
in the specimens with split-tee connections, the moment capacity due to prying action. Some loss of bolt pretension force was
of each of these specimens began to deteriorate in subsequent found to occur, amounting to less that 10% of the pretension
loading cycles from the effects of cyclic local web and flange force. The loss of bolt tension force was determined to have been
buckling. This is apparent in the specimen moment-plastic rota- caused by the bearing forces exerted by the tee flanges on the


J. Struct. Eng. 2004.130:223-232.

Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 14. Inelastic story drift capacity of specimens

Fig. 13. Variation in bolt force, Specimen 6

specimens 共Specimens 4, 5, 6, and 7兲 achieved an inelastic story

column face, which slightly distorted the composite column’s drift that exceeds that expected maximum inelastic drift of 0.031
cross section. However, there was sufficient residual tension re- and 0.04 rad that would occur during the design basis earthquake
maining in the bolts at the end of the test that no separation 共DBE兲 and maximum considered earthquake 共MCE兲 共ICC 2000兲,
between the surface of the tee flanges and face of the column respectively. The inelastic drift demand corresponding to the DBE
occurred. The bolt tension-panel zone shear force response for a and MCE events was determined by conducting inelastic time
bolt is given in Fig. 13 for Specimen 6, where a small loss of bolt history analysis of several CFT moment resisting frames, subject-
tension is apparent under the cyclic panel zone shear. ing them each to several ground motion records 共Ricles et al.
All of the specimens dissipated energy during testing, where 2002兲. The AISC Seismic Provisions 共1997兲 requires an inelastic
specimens with the weak panel zone configuration dissipated en- story drift capacity for connections of 0.03 rad prior to degrading
ergy primarily in shear in the panel zone and specimens with the to 80% of the nominal capacity, while the new FEMA design
weak beam configuration dissipated energy primarily in flexure in recommendations for steel frames 共FEMA 2000兲 require a total
the beam plastic hinges at the end of the connections. The stable story drift capacity of 0.04 rad prior to degrading to the nominal
hysteretic behavior of Specimen 1R involving inelastic panel capacity. The latter of 0.04 rad is comparable to about 0.03 rad of
zone response resulted in the specimen having an energy dissipa- inelastic drift. Hence, based on a comparison of specimen re-
tion E d of 1954 kN m, which was the maximum among all the sponse with the DBE and MCE demands, as well as with the
specimens. The reduced shear capacity of Specimen 2R resulted AISC Seismic Provisions and FEMA recommendations 共see ␪ in,n
in E d equal to 1674 kN m. Specimen 1, a weak beam configura- in Table 3兲, a weak panel zone connection, split-tee connection
tion, was tested to only 4% story drift and did not develop beam detail, and extended-tee connection detail with tapered tee flanges
local flange or web buckling. It dissipated 1026 kN m of energy. all appear to be suitable for seismic resistant design. The connec-
For Specimens 2 and 3, which were also a weak beam configu- tion detail consisting of a weak beam configuration with interior
ration and had fewer displacement cycles imposed, had no local diaphragms 共Specimens 1 and 2兲 or an extended tee without a
beam web or flange buckling occur and their energy dissipation tapered tee flange 共Specimen 3兲 does not appear to be adequate
was equal to 184 and 630 kN m, respectively. Although some of for seismic resistant design.
the weak beam specimens developed cyclic beam web and flange
local buckling 共Specimens 3R, 4, 5, 6, and 7兲 which led to a Summary and Conclusions
deterioration in their capacity, these specimens had a considerable
amount of energy dissipation occur during testing. Specimen 3R, The results of a full-scale experimental study show that split-tee
which had the extended-tee detail, dissipated energy in the connections and extended-tee connections with tapered tee
amount of E d equal to 1387 kN m. Specimens 5, 6, and 7 had an flanges provide adequate cyclic joint stiffness and strength for a
E d of 1825, 1908, and 1918 kN m during testing, respectively, weak beam–strong CFT column system, and therefore offer a
which was 1.03, 1.07, and 1.08 times the value of 1780 kN m for viable alternative to connections with diaphragms for seismic re-
Specimen 4, which had hysteretic pinching. sistant design. These connection details should be designed to
resist the maximum expected moment 共FEMA 2000兲 that devel-
ops in the beam plastic hinge at the end of the connection under
Assessment of Specimen Performance cyclic loading. Reinforcing the bolt holes in the beam flanges of
the bolted split-tee connections prevented hole elongation and
A summary of the contribution of the beam, column, panel zone, fracture, reducing the slip and pinching in the hysteretic response.
and connector elements 共i.e., split tees兲 to the total maximum The development of a diagonal compression strut in the concrete
inelastic story drift for each specimen is given in Fig. 14. These of the panel zone enhanced the shear resistance of the joint with
results are based on the last successful cycle that was achieved split-tee connection details. The split-tee connection detail allows
prior to either fracture or the end of the test. The inelastic story a better mobilization of this concrete strut. In these connections
drift for each specimen was determined by removing the elastic the presence of beam shear tabs was found to have only a minor
drift from the specimen lateral load-total story drift response 共see effect on overall behavior.
Peng et al. 2001 for complete details兲. The results show that the Weak panel zone specimens having connections with interior
weak panel zone specimens 共Specimens 1R and 2R兲, extended-tee diaphragm details were found to have stable hysteretic behavior
specimen with tapered plates 共Specimen 3R兲 and the split-tee without any significant loss in cyclic strength. However, the col-


J. Struct. Eng. 2004.130:223-232.

umn damage caused by inelastic panel zone response is expensive Federal Emergency Management Agency 共FEMA兲. 共2000兲. ‘‘Recom-
to repair. The cyclic ductility of the specimens with welded con- mended seismic design criteria for new steel moment-frame build-
nection details having abrupt geometric changes 共i.e., Specimen ings.’’ Rep. No. FEMA 350, FEMA, Washington, D.C.
3兲 was hindered by the development of strain concentrations that ICC. 共2000兲. International building code, International Code Council
led to beam flange fracture. The use of tapered plate details in 共ICC兲, Falls Church, Va.
Kanatani, H., Tabuchi, M., and Kamba, T. 共1987兲. ‘‘A study on concrete
these connections is recommended 共i.e., Specimen 3R兲 in order to
filled RHS column to H-beam connections fabricated with HT bolts in
reduce strain concentrations and avoid fracture. Connection de- rigid frames.’’ Proc., 1st Composite Construction in Steel and Con-
tails with interior diaphragms and a weak beam configuration crete Conf., Engineering Foundation, New York, 614 – 635.
were found not to perform well due to fracture of the beam flange Kato, B. 共1982兲. ‘‘Beam-to-column connection research in Japan.’’ J.
at either the weld access hole or the weld fusion line. Struct. Eng., 108共2兲, 343–360.
Koester, B. 共2000兲. ‘‘Panel zone behavior of moment connections be-
tween rectangular concrete-filled steel tubes and wide flange beams.’’
Acknowledgments PhD dissertation, Dept. of Civil and Environmental Eng., Univ. of
Texas, Austin, Austin, Tex.
Matsui, C. 共1985兲. ‘‘Local buckling of concrete filled steel square tubular
Downloaded from by Seattle University on 09/15/13. Copyright ASCE. For personal use only; all rights reserved.

The research program reported herein was supported by the Na-

tional Science Foundation 共Dr. Shih-Chi Liu and Dr. Ken Chong兲, columns.’’ Proc., IABSE–ECCS Symp. on Steel in Buildings, IABSE
the American Institute of Steel Construction 共Nestor Iwankiw兲, Rep. No. 48.
Morino, S., Sakino, K., Mukai, A., and Yoshioka, K. 共1996兲. ‘‘U.S.–Japan
and Nippon Steel Corporation 共Dr. Mamoru Iwata, Isao Kimura,
cooperative earthquake research program on CFT column systems.’’
and Mitsuhiko Yazaki兲. Donations of materials were provided by Proc., 5th Int. Colloquium on Stability of Metal Structures, SSRC,
Bethlehem Steel 共Robert Alpego兲 and the Steel Tube Institute Univ. of Florida, Gainesville, Fla., 83–92.
共Frederick Palmer兲. The support by these sponsors is gratefully Peng, S. W., Ricles, J. M., and Lu, L. W. 共2001兲. ‘‘Seismic resistant
acknowledged. connections for concrete filled column-to-WF beam moment resisting
frames.’’ Rep. No. 01-08, ATLSS Eng. Res. Center, Lehigh Univ.,
Bethlehem, Pa.
References ‘‘Recommendations for U.S.–Japan cooperative research program—
Phase 5: Composite and hybrid structures.’’ 共1992兲. Rep. No. UMCEE
AISC. 共1992兲. Seismic provisions for structural steel buildings, American 92-29, Dept. of Civil and Environmental Eng., Univ. of Michigan.
Institute of Steel Construction, Chicago. Ricles, J. M., Lu, L. W., and Sooi, T. K. 共1998兲. ‘‘Seismic behavior of
AISC. 共1994兲. Manual of steel construction—Load and resistant factored concrete filled tube column-to-WF steel beam moment resisting con-
design, 2nd Ed., American Institute of Steel Construction Chicago, 2. nections.’’ American Concrete Institute Special Publication: Hybrid
AISC. 共1997兲. Seismic provisions for structural steel buildings, American and Composite Structures, SP-174, 1–36.
Institute of Steel Construction, Chicago. Ricles, J. M., Sause, R., Muhummud, T., and Herrera, R. 共2002兲. ‘‘Mod-
AISC. 共2000兲. Load and resistant factored design specification for struc- eling and behavior of composite MRFs with concrete filled steel tu-
tural steel buildings, American Institute of Steel Construction, Chi- bular columns subject to earthquake loading.’’ Proc., Int. Conf. on
cago. Advances of Building Technology, Hong Kong, Hong Kong Polytech-
Applied Technology Council. 共1992兲. ‘‘Guidelines for cyclic seismic test- nic University, Elsevier, London.
ing of components of steel structures.’’ Redwood City, Calif. Schneider, S. P., and Alostaz, Y. M. 共1998兲. ‘‘Experimental behavior of
Architectural Institute of Japan. 共1987兲. ‘‘Standard for structural calcula- connections to concrete-filled steel tubes.’’ J. Constr. Steel Res., 45共3兲,
tion of steel reinforced concrete structures.’’ Tokyo. 321–352.
Elremainly, A. F., and Azizinamini, A. 共2000兲. ‘‘Development of detail Yokoyama, Y., Morita, K., Kawamata, Y., and Matsumura, H. 共1991兲.
and design criteria for steel beam to concrete filled tube column con- ‘‘Structural behaviors of steel-beam to concrete-filled square tube col-
nections in seismic regions.’’ Final Rep. submitted to National Sci- umn connections reinforced with inner ring stiffener.’’ Proc., 3rd Int.
ence Foundation, (Award No. 9520280), Dept. of Civil Eng., Univ. of Conf. on Steel–Concrete Composite Structures, Fukuoka, Japan, 165–
Nebraska–Lincoln, Lincoln, Neb. 170.


J. Struct. Eng. 2004.130:223-232.