Sei sulla pagina 1di 8

CYCLIC TESTS OF DOUBLE-SIDED BEAM-TO-COLUMN JOINTS

By Dan Dubina,1 Adrian Ciutina,2 and Aurel Stratan3

ABSTRACT: This paper summarizes the results of the full-scale tests on beam-to-column joints, carried out
within the experimental program of the COPERNICUS European research project at the ‘‘Politehnica’’ University
of Timisoara, Romania. Two series of six double-sided joints with three different beam-to-column connection
typologies have been tested under symmetrical and antisymmetrical cyclic loading. Three types of connections
(extended end plate, welded, and with cover plates) have been investigated. The main parameters considered in
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

this study are: (1) initial stiffness; (2) moment capacity; and (3) plastic rotation capacity. The obtained experi-
mental values are compared to the theoretical predictions by means of Eurocode 3, Annex J. The antisymmetrical
loading triggers the participation of the panel zone to plastic mechanism, which brings the following changes
as compared to the symmetrical loading: (1) Increase of ductility; (2) decrease of moment capacity and initial
stiffness; and (3) more stable energy dissipation through hysteretic loops. In any of the loading cases, the quality
of welds is essential for appropriate cyclic behavior of the joints.

INTRODUCTION TEST SPECIMENS AND EXPERIMENTAL SETUP


Steel moment-resisting (MR) frames were traditionally con- Two series of specimens, including three different connec-
sidered as a suitable structural system for low- and moderate- tion typologies (Fig. 1) have been tested under symmetrical
height buildings in seismic areas, due to their inherent local and antisymmetrical loading, respectively. Beam-to-column
and global ductility. However, the Northridge earthquake in connections were of three types:
1994 and the Hyogoken-Nanbu earthquake in 1995 revealed a
series of undesirable brittle failure modes in beam-to-column
connections that undermined the high seismic performance of 1. Bolted with extended end plate (EP)—a typical Euro-
steel moment-resisting frames. This fact generated concern in pean connection. The extended end plate connections
the scientific community for the causes of the unexpected poor have been designed as semirigid and partially resistant
behavior of beam-to-column moment connections. Extensive [according to Eurocode 3 (EC3) (1992)].
laboratory studies have been carried out to study the poor per- 2. Beam directly welded to the column (W).
formance of beam-to-column joints and to develop improved 3. Connection with welded cover plates and web cleat,
connection details and methodologies (SAC 1995). Among the bolted for erection (CWP). The cover plates have been
causes of brittle fractures in welded joints the following designed so as to shift the plastic hinge in the uncleated
sources could be identified: (1) Workmanship (welding de- portion of the beam, away from the column face. This
fects); (2) detailing (stress concentration at the root or toe of connection typology belongs to the suite of so called
welds); (3) materials (low-toughness weld metal); and (4) un- post-Northridge-reinforced connections (SAC 1997).
usually high seismic input (high strain rates).
The global performance of steel MR frames in seismic areas The test setup for symmetrical (XS series) and antisym-
is strongly influenced by properties of the beam-to-column metrical (XU series) joints are shown in Fig. 2. Two specimens
joints (resistance, stiffness, and plastic rotation capacity). Tra- of each joint type have been tested. The specified welds of the
ditionally, beam-to-column joints were required to be of rigid beam flanges/cover plates to the column flanges/end plate are
and full strength type. However, there can be noted a renewed of full-penetration type, while the beam web is welded with
interest in bolted connections for seismic applications, which fillet welds. The detail of edge preparation for full penetration
are usually of semirigid and partial resistant type (Leon 1999). welds is shown in Fig. 3. However, a visual inspection showed
At the same time, while the building industry in United States that not all the welds had been fully penetrated. To see the
adopted site-welded beam-to-column connections as the stan- influence of the welding procedure—for the XS-W1 specimen
dard, bolted beam-to-column connections are preferred in Eu- —that presented visible unconformities compared to design
rope, due to higher quality of shop welds. Also, relatively detailing, the root of the full penetration weld between beam
weak panel zones allowed by modern seismic codes [Uniform flanges and column was rewelded. The other specimen from
Building Code (UBC) (1997)] will result in joints of the semi- the same series (XS-W2) remained as it was initially manu-
rigid and partially resistant types. Contribution of the panel factured. In the case of bolted end-plate connection, the bolts
zone to the resistance and stiffness of interior (double-sided) have been fully pretensioned.
joints will depend primarily on the type of loading applied to The design steel grade was S235 ( fy = 235 N/mm2, fu = 360
the joint, symmetrical or antisymmetrical, the last being the N/mm2). Mechanical characteristics according to mill certifi-
one that acts in the case of earthquake motion, and that im- cates and conducted tensile tests are presented in Table 1. The
poses high demands on the panel zone (Dubina et al. 2000). yield stress of the tested elements is higher than the specified
1
one, the actual steel grade of beam and column being rather
Prof., Facu. of Civ. Engrg., ‘‘Politehnica’’ Univ. of Timisoara, 1900 grade S275. At the same time a relatively low value of the
Timisoara, Romania.
2
PhD Student, Facu. of Civ. Engrg., ‘‘Politehnica’’ Univ. of Timisoara,
yield stress of the end plate could be noted.
1900 Timisoara, Romania. The loading history was applied according to the European
3
PhD Student, Facu. of Civ. Engrg., ‘‘Politehnica’’ Univ. of Timisoara, Convention for Constructional Steelwork (ECCS) (1985) rec-
1900 Timisoara, Romania. ommendations’ simplified procedure [Fig. 4(a)]. Prior to plas-
Note. Associate Editor: Brad Cross. Discussion open until July 1, tic cycles, the conventional yielding displacement ey and the
2001. To extend the closing date one month, a written request must be corresponding force Fy have been determined as the intersec-
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on April 20, 2000.
tion between the initial stiffness line and the tangent to the
This paper is part of the Journal of Structural Engineering, Vol. 127, F-e curve having a slope of 10% of the initial stiffness [Fig.
No. 2, February, 2001. 䉷ASCE, ISSN 0733-9445/01/0002-0129–0136/ 4(b)]. The failure of the specimen was considered when the
$8.00 ⫹ $.50 per page. Paper No. 22355. force applied to the joint fell below 50% of the maximum load
JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 129

J. Struct. Eng. 2001.127:129-136.


Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. Connection Configurations: (a) Bolted with Extended EP; (b) W; (c) Welded with CWP

FIG. 2. Testing Setup for (a) Symmetrical Loading and (b) Antisymmetrical Loading

Symmetrical Loading Case


Bending moment at the column face is determined as M =
P(L ⫺ hc)/4. Transducers 1 and 2 (displacements ␦1 and ␦2)
are used in order to determine the total rotation of the joint
(Fig. 5), as given by (Grecea 1999)

␾j,t =
1
Lb 冉
␦1 ⫹ ␦ 2
2

P ⭈ L 3b
6 ⭈ E ⭈ Ib

P ⭈ Lb

2 ⭈ G ⭈ A␯b
(1)

FIG. 3. Detail of Edge Preparation


The term P ⭈ L 3b /6 ⭈ E ⭈ Ib ⫹ P ⭈ Lb /2 ⭈ G ⭈ A␯b accounts for the
elastic rotation of the beam on the length Lb. Due to the pres-
applied during the loading history. The load was applied qua- ence of transversal stiffeners, the rotation due to panel zone
sistatically. deformation was insignificant. The global rotation of the joint
␾j,t represents the average rotation of the two connections (left
and right).
DATA PROCESSING

Depending on the connection typology and type of loading, Antisymmetrical Loading Case
the inelastic deformations are expected to occur in the panel
zone, connection, or in the beam. Therefore, the relevant ro- The bending moment at the column face is computed as M
tations and bending moments should be computed at different = H ⭈ P ⭈ Lb /L. Rotations that may be defined in the case of
locations along the beam axis (at the middle of the panel zone, antisymmetrically loaded joints are:
at the column face, or at the end of the cover plates). In order
to make possible a direct comparison of the different joints, 1. Panel zone rotation ␥ has two components, ␥1 and ␥2
moment and rotations have been considered at the column face [Fig. 6(a)]. It is difficult to separately estimate these com-
for all configurations. Basic arrangement of displacement ponents, therefore the overall panel zone rotation angle
transducers used to compute joint rotations are presented in ␥ = ␥1 ⫹ ␥2 is determined from transducers 3 and 4
Fig. 5. (displacements ␦3 and ␦4) as
130 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:129-136.


TABLE 1. Characteristics of Main Steel Components

Yield Stress, fy (N/mm2) Tensile Strength, fu (N/mm2)


Element Mill certificate Coupon tests Mill certificate Coupon tests
(1) (2) (3) (4) (5)
Beam IPE360 285.0 329.8 flange, 348.4 web 427.0 463.2 flange, 464.0 web
Column HEB300 311.3 313.0 flange, 341.8 web 446.0 449.8 flange, 464.4 web
End plate (t = 20 mm) 281.0 248.3 424.7 416.0
Cover plate (t = 14 mm) 296.0 273.2 443.0 436.7

3. Total joint rotation ␾j,t [Fig. 6(c)] is the rotation between


Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

beam and column axes at the panel zone boundary, and


is given by
␾j,t = ␥ ⫹ ␾c (4)

TEST RESULTS
The number of loading cycles to which the specimens have
been subjected is presented in Table 2, together with the total
energy dissipated throughout the loading history. Fig. 7 pre-
sents some characteristic moment rotation curves for the XS
FIG. 4. Recommended ECCS Procedure: (a) Loading History; series, while Fig. 8 shows the same curves for the XU series.
(b) Determination of Yield Force and Displacement
XS-EP Specimens
The end plate was the weakest component, showing visible
bending deformations at the level of beam flange in tension at
cycles of ⫾2ey. Cycles of ⫾4ey were characterized by cracking
of the weld between the beam bottom flange and end plate
(initiated in the root of the weld) and local buckling of the
upper beam flange. In the first cycle of ⫾6ey complete rupture
of the bottom beam flange to end-plate weld occurred, the
crack propagating into the base metal (end plate on one side
and beam flange on the other). After the complete rupture of
the beam bottom flange weld, the beam web starts cracking,
near the weld to the column flange. In the case of specimen
XS-EP2, one of the bolts from the second bolt row (at the
bottom beam flange) failed in tension, leading to large defor-
mations of the end plate at the bottom beam flange [Fig. 9(a)].
Due to loosening of bolts, a degradation of the joint stiffness
occurred for the cycles of ⫾4ey to ⫾8ey. Within the cycles of
FIG. 5. Basic Joint Instrumentation for Determination of Ro- the same amplitude of ⫾4ey and ⫾6ey a degradation of the
tations dissipated energy could be observed, especially for the ⫾6ey
cycles.
The essentially different behavior of the end plate connec-
兹a2 ⫹ b2 ⭈ (␦3 ⫺ ␦4)
␥= (2) tion at the bottom and upper parts, due to detailing of the
2⭈a⭈b
welds, is to be noted. Both upper and bottom flanges of the
2. Connection rotation ␾c is determined from transducers 5 beam have been welded in the downward position. In this way,
and 6 for the left side, and from transducers 7 and 8 for the root of the full penetration weld is at the interior side of
the right side [Fig. 6(b)] the upper flange and at the exterior part of the bottom flange.
Due to smaller stresses at the interior side of the beam flange
␦5 ⫺ ␦6 ␦8 ⫺ ␾7
␾ left
c = , ␾ right
c = (3) and restraining caused by the beam web, the upper beam
b b flange is less prone to crack initiation at the root of the weld.
The rotation determined in this way includes the contri-
butions in the portion of the beam components adjacent XS-W Specimens
to the column. The positive rotation is considered to be The XS-W1 specimen was deliberately modified by reweld-
clockwise. ing the roots of the beam flange to column flange welds. In

FIG. 6. Definition of Rotations at Boundary of Panel Zone: (a) Panel Zone; (b) Connection; (c) Total Joint

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 131

J. Struct. Eng. 2001.127:129-136.


TABLE 2. Loading History of Symmetrical and Antisymmetrical Specimens

Number of Plastic Cycles (XS Series) Number of Plastic Cycles (XU Series)
Plastic range EP1 EP2 W1 W2 CWP1 CWP2 EP1 EP2 W1 W2 CWP1 CWP2
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
ey to 2ey 4 4 5 3 3 * 7 — 4 — 4 1
⫾2ey 3 3 3 3 3 * 3 3 3 3 3 3
⫾4ey 4 3 3 4 3 * 3 3 3 3 3 3
⫾6ey — 6 3 — 3 * 3 3 3 3 3 3
⫾8ey — — 2 — 9 * 18 28 17 14 3 6
⫾10ey — — — — — * — — — — 34 15
[Total] 11 16 16 10 21 * 34 37 30 23 50 31
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

[Total energy (KNm rad)] 76.7 120.2 125.2 64.7 390.0 * 661.5 924.6 721.0 611.7 1,666.8 1,051.2
Note: * = accidental failure of column-loading end plate.

FIG. 7. Moment-Rotation Curves for XS Specimen Series: (a) XS-EP2; (b) XS-W2; (c) XS-CWP1

FIG. 8. Moment-Rotation Curves for XU Specimen Series: (a) XU-EP1; (b) XU-W1; (c) XU-CWP1

the case of the XS-W1 specimen, local buckling of the beam There were significant differences between the behavior of
flanges was observed at the first cycles of ⫾4ey. Plastic de- the two XS-W specimens. Due to rewelding of the root of the
formations in the beam flanges and web increased during the weld in the case of XS-W1 specimen, crack initiation was
cycles of ⫾6ey and the first cycle of ⫾8ey. At the second cycle precluded in the bottom flange. Plastic deformations were
of ⫾8ey the complete brittle rupture of the bottom beam flange mainly due to local buckling of both the beam flanges and the
occurred, followed by rupture of the other flange and the web. web. The XS-W2 specimen showed early crack initiation at
Both flanges of the beam have been ruptured at the corner of the root of the bottom flange weld. Local flange buckling oc-
the weld access hole. curred only at the upper flange, while deformations at the bot-
For the XS-W2 specimen, at the first cycle of ⫾4ey a crack tom flange were mainly due to cracking of the weld and the
initiated at the weld root of the beam bottom flange from the heat-affected zone. Even failure was brittle in both cases; the
left connection and in the beam web at the corner of the weld XS-W1 specimen resisted to considerably more cycles, and
access hole. During the third cycle of ⫾4ey a crack in the was characterized by a larger rotation and moment capacity.
bottom-flange weld propagated on the full length and thickness
of the flange. After buckling of the upper flange during the XS-CWP Specimens
same cycle, it ruptured in a brittle manner starting from the
corner of the weld access hole, and followed by rupture of the The upper beam flange buckled next to the cover plates in
beam web in the next cycle [Fig. 10(a)]. The XS-W1 specimen the first positive half-cycle of ⫾4ey, followed by the lower
showed some energy degradation only in the groups of cycles beam flange in the negative half-cycle. Local buckling of beam
of ⫾6ey and ⫾8ey. The other specimen was characterized by flanges [Fig. 11(a)] continued to increase during the cycles of
a sudden degradation only in the last cycle, when the brittle ⫾6ey, followed by local buckling of the beam web at angles
failure occurred. of approximately 45⬚ during the cycles of ⫾8ey.
132 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:129-136.


Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 9. Cracking of (a) Bottom Flange and End Plate (Specimen XS-EP2) and (b) Rupture of End Plate (Specimen XU-EP1)

FIG. 10. Cracking of (a) Upper Beam Flange (Specimen XS-W2) and (b) Tearing of Panel Zone (Specimen XU-W1)

FIG. 11. Local Buckling of (a) Beam Flange and Web (Specimen XS-CWP1) and (b) Cracking of Column Flange and Web (Specimen
XU-CWP2)

No cracks were observed in the welds between the flange (2) higher quality welds than for the other specimens from the
cleats and column flanges. Two factors are responsible for this XS series (root of the welds were rewelded by the fabricator).
behavior: (1) The smooth flow of stress from the beam to the Some of the cycles are unsymmetrical due to attainment of
column due to trapezoidal shape of the cleats, extending over the actuator limit in tension. The XS-CWP2 specimen was lost
the full width of the column flange at the column face; and due to accidental failure of the column-loading plate welds.
JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 133

J. Struct. Eng. 2001.127:129-136.


XU-EP Specimens occurred first for the right beam, and continued in a smaller
extent on the left side. During the successive cycles top flanges
The first signs of inelastic deformations were observed in ruptured at the corner of the weld access hole. At the end of
the panel zone, where paint started to blister at the ⫾ey cycles. the 16th cycle of ⫾8ey, top beam flanges were completely
Plastic deformations in the panel zone increased progressively ruptured, and bottom ones caused extensive pullout of the col-
with the number of cycles. Deformations of the end plate were umn flange and tearing of the web. Minor buckling of beam
observed starting with cycles of ⫾2ey, a gap being formed flanges occurred during the test.
between the end plate and column flange in the tension zone. The XU-W2 specimen showed a similar behavior, but it was
The first cracks in the welds between the beam bottom flange stopped in the 14th ⫾8ey cycle due to problems at one of the
and end plate appeared at the ⫾6ey and ⫾4ey cycles for the supports. Beam bottom flange welds cracked extensively at the
XU-EP1 and XU-EP2 specimens, respectively. Limited buck- end of the 11th ⫾8ey cycle. During the last cycles, cracks were
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

ling of the beam flanges was also observed. Deformation of observed at the welds between transversal stiffeners and col-
the end plate was also given by the loosening of bolts, which umn flange, showing initiation of column flange pullout.
greatly decreased the stiffness of the connection. Cracking of Shear of the panel zone brought the main contribution to
welds appeared at the top flange only at ⫾8ey displacement the energy dissipation in the joint. Only minor plastic defor-
levels. After a number of plastic excursions at ⫾8ey, complete mations occurred at the beam end, mainly during the last cy-
rupture of the extended part of the end plate occurred. Starting cles, when effective rupture of the beam top flanges and col-
from this point, the extended end plate transformed practically umn pullout at the bottom beam flanges occurred. Energy
into a flush end-plate [Fig. 9(b)]. begins to degrade starting with the fourth cycle of ⫾8ey in the
The panel zone showed stable hysteresis loops over the en- case of the XU-W1 specimen, and in the sixth cycle of ⫾8ey
tire loading history, with an important strain hardening. It was in the case of the XU-W2 one. During anterior loading history,
the main source of ductility and resistance up to the rupture energy dissipation had a stable character.
of the end plate. The inelastic demand on the panel zone
started to decrease at this point, leading to its ‘‘relaxation.’’ XU-CWP Specimens
On the other hand, extended end plate connection showed a
continuous degradation of both stiffness and moment over the Paint in the panel zone began to blister at deformation levels
loading history. The dissipated energy is quite constant in the of ⫾ey. Through the loading history up to ⫾10ey panel zone
groups of three cycles up to ⫾6ey; it begins to degrade when continued to show increasing distortion without any visible
displacement levels of ⫾8ey are reached, mainly due to rupture cracks. At the 14th cycle of ⫾10ey the first cracks appeared
of the extended end plate. in the welds at the lower part of bottom cover plates for the
In the case of the XU-EP2 specimen, cracking of the beam XU-CWP1 specimen. Similar cracks appeared at the XU-
flange in the heat-affected zone at the corner of the weld access CWP2 specimen already at the second cycle of ⫾8ey. This is
hole occurred. The specimen failed by complete fracture of the explained by the fact that the root of the weld between the
beam web and top flange at the right connection and rupture cover plate and beam flange was rewelded at the XU-CWP1
of two bolts below the tensioned flange for the left connection. specimen.
An important drop in moment capacity accompanied it. In the 22nd cycle at ⫾10ey column web in the panel zone
was slightly buckled at the XU-CWP1 specimen. A crack ap-
XU-W Specimens peared at the left inferior part of the column web in the 27th
cycle of ⫾10ey. In the following cycles the crack spread along
The panel zone was again the weakest component. It the entire bottom edge of the panel zone, and then also on the
showed important deformations (blistering of paint) at dis- two lateral sides. A significant drop in load capacity was no-
placements exceeding ⫾ey. The first cracks appeared at ⫾6ey ticed. At the end of the test column the web in the panel zone
in the root of welds between the beam bottom flange and col- was completely torn on three sides, being deformed out of its
umn flange. Top flange welds cracked only at the first ⫾8ey plane.
cycles. In the case of the XU-CWP2 specimen a sudden and deep
In the case of the XU-W1 specimen, starting with the ninth crack occurred in the fourth cycle of ⫾10ey at the root of the
cycle at ⫾8ey, cracks in beam bottom flanges propagated pro- weld between the right bottom-cover plate and column flange.
gressively into the column flange, which was ruptured on the Three cycles later a similar sudden crack accentuated at the
beam flange width. Column flanges were pulled out, together left connection. In the 11th cycle of ⫾10ey a crack was formed
with the column web tearing [Fig. 10(b)]. This phenomenon between the bottom transversal stiffener and the column

TABLE 3. Comparison between Computed and Experimental Joint Characteristics


Weakest
M exp⫹
j,max M exp⫹
j,pl M th
j,pl S exp
j,ini S th
j,ini ␾ exp
u ␾ exp
pl component
Specimen (kNm) (kNm) (kNm) (kNm/rad) (kNm/rad) (rad) (rad) M th
j,pl /Mb,pl (Annex J)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
XS-EP1 334.2 255.6 262.7 69,539 142,932.2 0.033 0.029 0.78 End plate
XS-EP2 337.9 288.9 261.3 44,205 140,886.8 0.038 0.032 0.77 End plate
XS-W1 437.7 305.6 309.3 333,953 ⬁ 0.029 0.028 1.00 Beam
XS-W2 412.1 277.8 317.6 321,569 ⬁ 0.016 0.015 1.00 Beam
XS-CWP1 542.0 316.7 452.1 366,309 ⬁ 0.038 0.037 1.35 Beam
XS-CWP2 * * 449.0 * ⬁ * * 1.35 Beam
XU-EP1 263.7 146.7 169.2 44,081 43,727.2 0.060 0.056 0.50 Web panel
XU-EP2 256.3 157.8 169.1 49,004 43,718.2 0.062 0.059 0.50 Web panel
XU-W1 248.8 113.3 163.6 63,102 68,792.1 0.052 0.050 0.48 Web panel
XU-W2 252.1 131.1 164.1 49,681 69,062.1 0.052 0.049 0.50 Web panel
XU-CWP1 287.2 131.1 178.6 60,712 75,597.2 0.064 0.062 0.53 Web panel
XU-CWP2 301.5 164.4 177.4 58,453 74,963.1 0.060 0.058 0.52 Web panel
Note: * = accidental failure of column-loading end plate.

134 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:129-136.


flange. One cycle later the two cracks at the interior and ex- Comparing the experimental and computed values of joint
terior of the column flange increased suddenly and formed a plastic moments, it can be observed that generally close values
single one. At the end of the test, the right column flange was are obtained for the XS series. An exception is the XS-CWP
ruptured, with the crack extending into the column web [Fig. joint, which showed considerably lower experimental value.
11(b)]. Krawinkler (1995) showed that very large panel zone In the case of the XU series, all experimental values are lower
distortions cause very high strains at the corners of the joints than the ones computed by Annex J of EC 3. The difference
where beam flanges are welded to the column flanges, leading between the computed and measured plastic moments could
to the so-called local kinking. These local kinks seem to be be explained by the conventional procedure used to determine
responsible for column flange cracking in the case of the XU- the experimental yielding moments, the numerical results be-
CWP2 joint. ing sensitive at high initial stiffness. Lower values of experi-
The XU-CWP1 specimen showed a very stable energy dis- mental joint plastic resistance M exp⫹j,pl do not imply unconser-
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

sipation capacity up to the 30th cycle of ⫾10ey. For the other vative design, the maximum experimental moment M exp⫹ j,max
one, it begins to degrade already at the fourth cycle of ⫾10ey, always being higher than the theoretical plastic moment
but the degradation is not so steep. It has to be emphasized M thj,pl. In what concerns the initial stiffness, numerical and ex-
that for this type of joint and loading, plastic deformations perimental results agree fairly well for the XU series, while
concentrated in the panel zone, the total joint rotation being significant differences are noticed for the XS series.
practically equal to the rotation of the panel zone. The key characteristics of the joints (initial stiffness, mo-
ment capacity, and rotation capacity) could be followed in Fig.
COMPARISON TO EC3 12(a), which shows a comparison of envelope curves for each
type of joint. The relevant characteristics agree with the code
Table 3 comprises the results of the experimental tests com- predictions and could be summarized as follows:
pared to those of EC 3, Annex J, in terms of joint bending
moments, rotational stiffness, and ultimate rotation. The table • High stiffness of W and CWP connections, and reduced
also presents the weakest component for each of the joints and stiffness of EP connection (XS series)
the nondimensional joint movement M thj,pl /Mb,pl. It should be • High moment capacity of CWP connection, and good ro-
noted that for this comparison, the joint characteristics were tation capacity for EP and CWP connection (XS series)
computed with the measured strengths and dimensions of the • A drop of stiffness and moment capacity for the XU se-
joint components, considering the partial safety factors equal ries, and an increase of rotation capacity as compared to
to one. The reduction of the moment capacity of the beam due the XS series
to the presence of shear force was taken into account. The
experimental plastic moment of the joint M exp⫹ j,pl is computed Fig. 12(b) shows a typical comparison between an experi-
according to the ECCS procedure, as the intersection of the mental envelope and moment-rotation relationship computed
initial stiffness S exp exp
j,ini line and the S j,ini/10 tangent to the envelope according to EC3, Annex J. It is noticed that the experimental
curve [analogous to Fy in Fig. 4(b)]. curve is above the theoretical one; lower values of the experi-
Most of the joints resulted in being partially resistant, with mental plastic moment M exp⫹
j,pl in the case of the XU series being
the exception of the XS-W and XS-CWP joints. The antisym- caused by the procedure used to determine these moments.
metrical loading has led to a 50% drop of the theoretical joint Two alternatives for modeling beam-to-column joints are
plastic moment with respect to the plastic moment of the con- mentioned in EC3, Annex J (Fig. 13), but only the simplified
necting beam, due to the sheared web panel. modeling is supported by detailed design provisions. The sim-

FIG. 12. (a) Moment-Rotation Envelopes; (b) Comparison between Experimental and Computed Curves

FIG. 13. Joint Modeling Reflecting (a) Actual Behavior and (b) Simplified Modeling

JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001 / 135

J. Struct. Eng. 2001.127:129-136.


plified procedure models each of the joints as a separate ro- rad for special MR frames (UBC 1997). For the XS series, the
tational spring, which takes into account the behavior of the EP and CWP joints are at the limit, while the W joints have
web panel. Following the EC3, Annex J procedure, one obtains inadequate plastic rotation capacities.
two sets of moment-rotation characteristics for symmetrical
(gravity) and antisymmetrical (earthquake) loading for the ACKNOWLEDGMENTS
same joint configuration. Though it is simple to implement one Support of the European Commission for the entire project, and par-
of the joint characteristics into structural analysis programs, it ticularly for the acquisition of equipment and specimens, is gratefully
will not reflect the real joint behavior under different loading acknowledged.
types. A model of the type shown in Fig. 13(a) should be used
instead, but is not supported directly by EC3, Annex J. APPENDIX I. REFERENCES
Dubina, D., Ciutina, A., and Stratan. (2000). ‘‘Influence of loading asym-
Downloaded from ascelibrary.org by INDIAN INSTITUTE OF TECHNOLOGY on 06/08/14. Copyright ASCE. For personal use only; all rights reserved.

CONCLUSIONS metry on the cyclic behaviour of beam-to-column joints.’’ Proc., 3rd


Int. Conf.
Loading type (symmetrical or antisymmetrical) significantly Eurocode 3 (EC3). (1992). General rules and rules for buildings, Part
affects the response parameters of beam-to-column joints. The 1.1, CEN, Brussels, Belgium.
European Convention for Constructional Steelwork. (ECCS). (1985).
main component that introduces the difference is the panel ‘‘Recommended testing procedures for assessing the behaviour of
zone in shear. The most important consequences on the cyclic structural elements under cyclic loads.’’ Tech. Comm. 1, TWG 1.3—
behavior of beam-to-column joints are the reduced moment Seismic Des., No. 45.
capacity and stiffness and (in general), increased ductility with Grecea, D. (1999). ‘‘Caracterisation du comportament sismique des os-
more stable hysteresis loops in the case of antisymmetrical satures metalliques—Utilisation d’assemblages a resistance partielle.’’
loading. It becomes necessary to use an appropriate model for PhD thesis, INSA, Rennes, France.
Krawinkler, H. (1995). ‘‘Systems behaviour of structural steel frames sub-
interior (double-sided) beam-to-column joints that would make jected to earthquake ground motion.’’ Rep. No. SAC-95-09, SAC Joint
it possible to reflect the different behavior of these joints under Venture.
gravitational and earthquake loading. EC3, Annex J provides Leon, R. T. (1999). ‘‘Developments in the use of partial restraint frames
a design tool for determining joint properties under the two in the United States.’’ Proc., Int. Conf., 95–104.
types of loading, but two sets of joint properties will result for SAC. (1995). ‘‘Connection test summaries.’’ Rep. No. SAC-96-02, SAC
each loading. This makes it difficult to implement the EC3 Joint Venture, Sacramento, Calif.
SAC. (1997). ‘‘Interim guidelines advisory No. 1. Supplement to FEMA-
model for structural analysis, as different joint properties 267 interim guidelines: Evaluation, repair, modification, and design of
should be used for different loading types. welded steel-moment frame structures.’’ Rep. No. SAC-96-03, SAC
Unbalanced moments at the boundary of the panel zone im- Joint Venture, Sacramento, Calif.
pose high demands on the column web panel, reducing the Suita, K., Nakashima, M., and Morisako, K. (1998). ‘‘Tests of welded
stiffness and moment capacity of the antisymmetrically loaded beam-column subassemblies.’’ J. Struct. Engrg., ASCE, 124(11), 1236–
beam-to-column joints. A joint that is classified as rigid and 1252.
Uniform Building Code (UBC). (1997). Structural engineering design
full strength under symmetrical loading, may become partially provisions, Vol. 2, International Conference of Building Officials, Whit-
resistant and semirigid under antisymmetrical loading. tier, Calif.
Investigation of the different joint typologies revealed the
importance of connection detailing and welding procedure, as APPENDIX II. NOTATION
well as its quality. Defective welding (lack of complete pen-
etration) was responsible for such phenomena as early crack The following symbols are used in this paper:
initiation. Rewelding of the weld root is particularly important A␯b = beam shear area;
at the exterior part of the beam flanges (bottom beam flange a and b = dimensions of panel zone (a = hc ⫺ tfc; b = hb ⫺ tfb);
in the case of downward welding). The weld access hole is E = modulus of elasticity;
another negative factor, leading to stress concentration and fi- ey = conventional yield displacement;
nally to beam flange rupture. Bolted end-plate connections Fy = conventional yield force;
showed good rotation capacity and more ductile behavior with fu = tensile strength;
respect to welded connections, though at a reduced initial stiff- fy = yield stress;
ness. These type of connections should be designed so as to G = shear modulus;
prevent brittle failure by bolt rupture. Loosening of bolts dur- H = column height;
ing cycle reversals has led to stiffness degradation. Another hb = beam depth;
aspect characteristic of antisymmetrical bolted joints is the dis- hc = column depth;
tribution of ductility demands between the end plate (connec- Ib = moment of inertia of beam;
tion) and the panel zone. L = horizontal distance between beam supports;
Generally, failure was brittle for welded joints (XS-W) and Lb = clear length of beam;
ductile for the other ones in the case of symmetrical loading. M = bending moment at column face;
The ductile behavior was due to connection (bolted joint) and Mb,pl = theoretical beam plastic moment, according to Euro-
to shifting of the plastic hinge away from the column face in code 3;
M exp⫹
j,max = maximum positive experimental moment;
the case of the CWP joint. Participation of panel zone to the
M exp⫹
j,pl = experimental positive joint plastic moment;
plastic mechanism significantly increased the ductility of the
M thj,pl = theoretical joint plastic moment, according to Annex
antisymmetrically loaded joints. In any case, welded joints J of Eurocode 3;
failed in a brittle manner here also. P = actuator force;
The joint with cover and web plates showed good behavior. S exp
j,ini = experimental initial stiffness of joint;
However, in the case of antisymmetrical loading, due to in- S thj,ini = theoretical joint initial stiffness, according to Annex J
creased moment at the column face combined with relatively of Eurocode 3;
weak panel zone, increased demands were exerted on the col- tfb = thickness of beam flange;
umn flanges (local kinks). This led to fracture of the column tfc = thickness of column flange;
flange, though after a considerable number of cyclic excursions. ␾ exp
pl = experimental plastic rotation of joint;
Plastic rotations of the XU series of joints are fairly higher ␾ exp
u = experimental ultimate total rotation of joint; and
than the generally accepted plastic rotation demand of 0.03 ␾ exp
y = experimental yield rotation of joint.

136 / JOURNAL OF STRUCTURAL ENGINEERING / FEBRUARY 2001

J. Struct. Eng. 2001.127:129-136.

Potrebbero piacerti anche