Internal flange stiffened moment connections with low-damage capability under seismic loading

(1)

Internal

ﬂange stiffened moment connections with low-damage capability

under seismic loading

Chung-Che Chou

a,b,

⁎

, Sheng-Wei Lo

c

, Gin-Show Liou

c a

Department of Civil Engineering, National Taiwan University, Taipei, Taiwan

b_{National Center for Research on Earthquake Engineering, Taipei, Taiwan} c

Department of Civil Engineering, National Chiao Tung University, Hsinchu, Taiwan

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 30 October 2012 Accepted 14 April 2013 Available online 17 May 2013 Keywords:

IFS steel moment connection Low-damage capability Test

Finite element analysis

This study presents the seismic performance of steel moment connections using internalflange stiffeners (IFSs) welded at the face of the wide-flange column and inner side of the beam flange. The objective is to develop a steel moment connection that can achieve good seismic performance with low-damage capability during a large earthquake loading and minimize the repair cost. Four large-scale moment connections were tested to validate the cyclic performance. One connection which represented a welded-unreinforced flange-bolted web connection failed beforefinishing cyclic tests at a drift of 4%. Three IFS moment connections showed excellent performance and low damage after experiencing the AISC seismic load twice up to the target drift of 4%, without strength reduction. The specimens were also modeled using the computer program ABAQUS to further verify the effectiveness of the IFS in transferring beam moment to the column and to investigate potential sources of connection failure.

1. Introduction

The widespread damage of welded steel moment connections after the 1994 Northridge earthquake and 1995 Hyogoken-Nanbu (Kobe) earthquake initiates extensive research aimed at improving connection seismic performance. Many traditional steel moment connections, which were fabricated following pre-Northridge construction practices with a low notch toughness E70T-4 electrode, show minimal plastic deformation (e.g., 1% drift) before weld fracture at the beam-to-column

interface[1–4]. By using a high notch toughness electrode for connection

welds, strengthening or reducing the beam end section[5–10]are also

needed for most quali_{ﬁed moment connections to reach a required}

seismic performance. FEMA 350[11]lists some prequaliﬁed moment

connections for the special moment frame (SMF). These moment con-nections are capable of sustaining an interstory drift of at least 4% with

sufﬁcient ﬂexural resistance[12]. However, high damage in the beam

(e.g., buckling) after seismic loading leads to a large cost for repair.

Adding a pair of full-depth side plates or separate internalﬂange

stiffeners (IFSs) between the column face and beamﬂange inner side

has been demonstrated as an alternative to achieve good seismic

per-formance of moment connections[13,14]. This scheme not only

mini-mizes the interference from the composite slab but also reduces story height requirements in the building. Test results showed that the IFS moment connection experiences very minor beam local buckling (e.g.,

low damage) during the code-speci_{ﬁed cyclic loading}[12]in excess of

a 4% drift. The connection requires minor repair and has the capability to sustain the same cyclic loading again to a drift of 4% without failure,

showing repeatable seismic performance as observed in theﬁrst test.

However, previous studies focused only on the IFS moment connection

with a steel built-up box column and a wideﬂange beam, which are

commonly used in Asian countries to resist seismic loads in SMFs. The

use of a wide-ﬂange column is also very popular in SMFs, but the

load-transfer from IFSs to the box column is more effective than that

to the wide-ﬂange column due to two web plates in the box column.

Moreover, previous specimens used the ASTM A36 steel beam, which produces smaller stresses in connection welds than the ASTM A572

Gr. 50 steel beam. Therefore, the speci_{ﬁc connection conﬁguration in}

this study uses a wide-ﬂange column and a beam with various material

properties. To design a moment connection with low-damage capabili-ty under seismic loading, four IFSs, each of which is a rectangular or

tri-angularﬂat plate, are welded at the column face and beam ﬂange inner

side to help transfer some beamﬂange force to the column. The

objec-tive of the study is to examine alternaobjec-tive technique for the moment

connection with a wide-ﬂange column and beam to improve the

frac-ture resistance through strengthening of connections.

A total of four large-scale exterior moment connections were tested. Test parameters were IFS sizes and material properties of the beam. One

welded-unreinforcedﬂange-bolted web connection was tested as a

benchmark. Three moment connections with different IFSs and beam materials were tested to validate their cyclic performances. The study showed that all IFS moment connection specimens performed much better than a non-stiffened moment connection specimen, even being tested twice up to a 5% drift. These specimens were also modeled ⁎ Corresponding author. Tel.: +886 2 3366 4349; fax: +886 2 2739 6752.

E-mail address:[email protected](C.-C. Chou).

http://dx.doi.org/10.1016/j.jcsr.2013.04.005

Contents lists available atSciVerse ScienceDirect

(2)

using the computer program ABAQUS[15]to further verify the effec-tiveness of the IFS in transferring beam moment to the column and to investigate potential sources of connection failure. This paper presents experimentally and analytically the cyclic behavior of the IFS moment connection, and provides recommendations for seismic design of such connections.

2. IFS moment connection 2.1. Connection design

Fig. 1shows a moment connection with IFSs. The purpose in using

IFSs is to transfer some not all beamﬂange force to the column

be-cause existing beamﬂange groove welded joints conducted by the

high toughness electrode can sustain modest inelastic deformation

before fractures. Moment demand, Mdem, along the beam is shown

in theﬁgure, assuming that a plastic hinge is located at a quarter

beam depth from the IFS end. This location is used based on

previous connection test results[13,14]. The moment at the column

face, determined by projecting moment capacity MPHat the plastic

hinge section, is Mdem¼ Lb Lb− Lðsþ db=4Þ MPH¼ Lb Lb− Lðsþ db=4Þ βRyσyn Zb ð1Þ

where Lbis the distance from the actuator to the column face; Lsis the

IFS length, which assumes half the beam depth in initial design; dbis

the beam depth; Zbis the plastic section modulus of the beam;σynis

the speci_{ﬁed yield strength of the steel; R}y is the material

over-strength coefﬁcient, and coefﬁcient β accounts for strain hardening[11].

Moment capacity near the beam-to-column interface increases

due to presence of IFSs. Theﬂexural capacity of the stiffened beam,

Mcap, is the summation ofﬂexural strengths of the beam, Mpb, and

the IFSs, Mps[13]: Mcap¼ Mpbþ Mps¼ ZbRyσynþ 2 2 ffiffiffi 1 2 r −1 ! db−2tf Ryσyndsts ð2Þ

where tfis the beamﬂange thickness; dsis the IFS depth, and tsis the

IFS thickness. Assuming that the stiffened beam moment capacity–

demand ratio,_{α (=M}cap/Mdem), is larger than 1.05, the IFS size can

be determined by: dsts≥ αMdem−Mpb 2 2 ﬃﬃ1 2 q −1 db−2tf Ryσyn : ð3Þ

Since the force in the IFS, PSI, is transferred through shear on the

groove welded joint between the IFS and beamﬂange inner side,

the length of the IFS, Ls, is determined based on shear strength of

the IFS: Ls≥ PSI 0_{:9 0:6R}yσyn ts ¼ 2 ﬃﬃ 1 2 q −1 Ryσyntsds 0_{:9 0:6R}yσyn ts ¼ 0:77ds: ð4Þ e PSI Top IFS Moment Diagram IFS Ls Plastic Hinge VCu VCL MCL MCu

P

ACT db/4 ds MPH= βxMpb Mpb Mcap Mdem Lb-e Lb Lp ds bf db Section A-A

A

Fig. 1. Moment capacity and demand of the beam.

Table 1

Member sizes and properties. (a) Specimen sizes

Specimen Column size Beam size IFS size (ts× ds× Ls) ∑M_pc

∑M pb UR H428 × 407 × 20 × 35 (A572 Gr. 50) H702 × 254 × 16 × 28 (A36) – 1.72 IFS1 R25 × 308 × 300 1.53 IFS2 T20 × 308 × 300 1.47 IFS3 H702 × 254 × 16 × 28 (A572 Gr.50) T28 × 308 × 300 1.19

∑ Mpc⁎ = sum of the column nominal moments at the top and bottom of the panel zone.

∑ Mpb⁎ = sum of the beam moments resulting from the beam plastic hinge moment.

(b) Material properties

Specimen Column strength (MPa) Beam strength (MPa) IFS strength (MPa)

Flange Web Flange Web

σy σu σy σu σy σu σy σu σy σu

UR 357 521 390 510 251 463 285 453 –

IFS1 409 528

IFS2 272 469 275 440 421 527

IFS3 388 531 417 564 388 531

(3)

The IFS size can be determined based on Eqs.(3) and (4). Iterate

over a new Lsby returning to Eq.(1)if Eq.(4)is not satisﬁed.

3. Test program 3.1. Specimens

The experimental program consisted of tests of four specimens. Each specimen represented an exterior moment connection with

one steel beam (H702 × 254 × 16 × 28) and one wide-ﬂange

column (H428 × 407 × 20 × 35).Table 1shows specimen sizes and

material properties obtained from coupon tensile tests. ASTM A572

Gr. 50 steel was utilized for all columns and internal_{ﬂange stiffeners.}

ASTM A36 steel was utilized for the beams of Specimens UR, IFS1, and IFS2; ASTM A572 Gr. 50 steel was utilized for the beam of Specimen IFS 3. These two types of steel were manufactured in Taiwan, conforming to chemical and mechanical properties of ASTM

stan-dards[16]. All connections were welded using the ER70S-G electrode,

which is similar to the high-toughness E71T-8 or E70TG-K2

electrodes and provides a minimum speciﬁed Charpy V-Notch value

of 27 J at −29 °C (20 ft-lb at −20 °F). The steel backing bars

projected 30 mm beyond both sides of the beamﬂange and no weld

tabs were used. The steel backing bar was left in place and a_ﬁllet

weld, helping to reduce the notch effect of a left in place backing

bar[11], was not made between the backing bar and column. Each

pass of_{ﬂange groove welds was initiated and terminated at a point}

outside the ﬂange. This was done to prevent poor-quality welds,

which normally occur at the initiation of the weld. All specimens were made by a fabrication shop welder, using weld positions typical

tofield welding. More specifically, beam flange groove welds were

made with the specimen oriented to permitﬂat position welding.

Ultrasonic tests (UT) were conducted for allﬂange groove welds, and

they all satisﬁed the prescribed acceptance criteria [17]. Only A490

high-strength bolts were used to connect the column shear tab and beam web.

Specimen UR used a welded-unreinforced _{ﬂange-bolted web}

connection (Fig. 2(a)). Specimen IFS1 was identical to Specimen UR,

except that the 25-mm thick rectangular IFSs were used at the beam

ﬂange edges of Specimen IFS1 [Fig. 2(b)]. Specimen IFS2 was identical

to Specimen IFS1, except that the 20-mm thick triangular IFSs were

40 60 6@70 40 80 40

A

B

View A-A Weld View B-B H428×407×20×35 (ASTM A572 Gr.50) _{H702×254×16×28} (A36) Detail A Doubler PL (12mm for each)

Shear PL (20mm) Bolt (A490 22mm ) 20 15 Continuity PL (28mm) 28 50 50 14 7 R50 R10 Detail A 45 300 308 (t=25mm) IFS (25mm) 30° Detail A View A-A 9 9 IFS

A

45° 15 H428×407×20×35 (ASTM A572 Gr.50) H702×254×16×28 (A36) IFS (ASTM A572 Gr.50 25mm) Shear PL (20mm) Bolt (A490 22mm ) 20 Detail A Doubler PL (12mm for each)

Detail A View A-A 70 70 9 9 300 308 (t=20mm) IFS (20mm) 30° IFS

A

45° 15 H428×407×20×35 (ASTM A572 Gr.50) H702×254×16×28 (A36) IFS (ASTM A572 Gr.50 20mm) Shear PL (20mm) Bolt (A490 22mm ) 20 Detail A

Doubler PL (12mm for each)

70 70 View A-A

A

45° 15 H428×407×20×35 (ASTM A572 Gr.50) H702×254×16×28 (ASTM A572 Gr.50) 9 9 300 308 IFS Detail A IFS (ASTM A572 Gr.50 28mm) 30° Shear PL (20mm) Bolt (A490 27mm ) 20 (t=28mm) Detail A Doubler PL (20mm for each)

IFS (28mm)

60 90 50

50

5@90

(a)

Specimen UR

(b)

Specimen IFS1

(c)

Specimen IFS2

(d)

Specimen IFS3

(4)

used at the beamﬂange edges of Specimen IFS2 [Fig. 2(c)]. Thinner IFSs in Specimen IFS2 leaded to less welding and smaller beam moment

capacity–demand ratio, α, as listed inTable 2. Specimen IFS3 [Fig. 2(d)]

was identical to Specimen IFS2, except that Specimen IFS3 had the

ASTM A572 Gr. 50 steel beam and thick triangular IFSs (Table 1(a)).

The beam moment capacity_{–demand ratio, α (=M}cap/Mdem), ranged

from 1.06 to 1.20 (Table 2) to study the effects of IFSs on the connection

behavior. Doubler plates were added in the column to maintain a strong panel zone; in other words, the panel zone shear computed based on the

beam plastic hinge moment, MPH, was less than 60% panel zone shear

strength, Vp,[12]: Vp¼ 0:6σyndcttotal 1þ 3bcft 2 cf dhdcttotal " # ð5Þ

where tcfis the columnﬂange thickness; bcfis the columnﬂange width;

dhis the panel zone depth; dcis the column depth, and ttotalis the total

thickness of the column web and doubler plates. 3.2. Test setup and loading protocol

The exterior connection specimens were tested as shown inFig. 3.

Restraint to lateral-torsional buckling of the beam was provided near the actuator and at a distance of 2000 mm from the column center-line. Displacements were imposed on the beam by actuators at a distance of 4000 mm from the column centerline. The AISC cyclic

dis-placement history[12]was used and run under displacement control.

The intersory drift, which was computed by the actuator displacement

divided by the distance to the column centerline, was used as the con-trol variable. Specimens were tested until connection failure occurred. 4. Test results

4.1. Welded-unreinforcedﬂange-bolted web connection

Fig. 4(a) shows the global response of Specimen UR; the moment

computed at the column face is normalized by the nominal plastic

moment of the beam, Mnp (=Zbσyn). Whitewash ﬂaking was

ob-served in the beamﬂange at an interstory drift of 0.75%, indicating

beam yield. A minor fracture occurred in the beam topﬂange groove

weld at an interstory drift of −2%, but the peak strength was

maintained at this drift level. A signiﬁcant reduction in strength

occurred toward the second cycle of an interstory drift of−4% due to

beam top_{ﬂange fracture (}Fig. 5). No yielding of the column or panel

zone was observed throughout the test. Although Specimen UR utilized

the ASTM A36 beam, the connection failed before ﬁnishing cyclic

tests at a drift of 4%.

4.2. Internalﬂange stiffened moment connection

All IFS connections performed well under the ﬁrst cyclic test,

exhibiting no groove-weld failures at an interstory drift of 4%. The low damage (e.g., minor buckling) in the beam did not need repair

after theﬁrst test, so all IFS specimens were tested again using the

same loading protocol, exhibiting similar cyclic performances as

ob-served in theﬁrst test up to an interstory drift of 4% [Fig. 4(b)–(d)].

Specimens IFS1 and IFS2 were identical to Specimen UR, except that (1) the 25-mm thick rectangular IFSs were used for Specimen IFS1, and (2) the 20-mm thick triangular IFSs were used for Specimen

IFS2. Specimens IFS1 and IFS2, which had beam capacity_–demand

ratios of 1.2 and 1.06, respectively, were used to evaluate the effects of IFS sizes on the connection behavior. Two specimens showed

similar cyclic behaviors during theﬁrst test. Yielding, observed by

whitewashﬂaking, occurred at an interstory drift of 0.75%,

concen-trated outside the IFS. Afterﬁnishing 4% drift cycles, yielding

extend-ed more than 1000 mm from the column face with sign of minor

ﬂange buckling (Figs. 6(a) and7(a)). A minor fracture occurred at

the end of welds between the IFS and beam topﬂange. The weld

crack was repaired before conducting the second test. For subsequent loading cycles, Specimen IFS1 achieved a maximum interstory drift of

5% with beam local buckling (Fig. 6(b)) and no groove weld fracture.

For Specimen IFS2 in the second test, a minor crack occurred in the

beam top_{ﬂange near groove welds at a drift of −1%, but it did not}

af-fect the connection performance afterﬁnishing the ﬁrst cycle of 5%

drift [Fig. 7(b)]. The beam topﬂange fractured when the connection

moved toward the second cycle of−5% drift. This indicates that the

connection with thicker IFSs can provide better cyclic performance in the second cyclic test.

Specimen IFS3 used the ASTM A572 Gr. 50 steel beam, so its IFS size was thicker than other specimens with the ASTM A36 beams

to maintain similar beam capacity–demand ratios (Table 2). Since

beam local buckling was minor and no strength degradation was

observed after theﬁrst cyclic test (Fig. 8(a)), Specimen IFS3 was

also retested using the same AISC loading protocol[12]. Beam local

buckling became obvious at an interstory drift of 3%, but the peak

strength was maintained afterﬁnishing two cycles of 5% drift without

failure (Fig. 4(d)). Beam buckling accompanied by twisting resulted

in a small reduction in beamﬂexural strength at a ﬁrst cycle of 6%

drift [Figs. 8(b) and4(d)]. Meanwhile, a minor fracture in the groove

weld was observed near the beam bottomﬂange to the column face.

A signiﬁcant reduction in strength occurred toward a second cycle of

6% drift due to beam bottomﬂange fracture (Fig. 8(c) and4(d)). No

yielding of the column or panel zone was observed throughout the test. The performance of Specimen IFS3 in the second cyclic test also Table 2

Beam moment capacity–demand ratio.

Specimen Mpb Mps Mcap Mdem MPH α β

(kN-m)

IFS1 1679 1685 3364 2810 2457 1.20 1.46 IFS2 1763 1388 3151 2983 2608 1.06 1.48 IFS3 2556 1791 4347 3663 3203 1.19 1.25 Note: Moment is calculated based on the actuator force at an interstory drift of 4%.

500 1500 2500 4500 3500 5500 6500

4000

2000

IFS

Column

Beam

835 1835 2835 3835 4835

Lateral Support

Reaction Wall Floor

Actuator

Unit:mm

35 20 407 H428 H702 254 16 28

(5)

exceeds stringent requirements based on AISC seismic provisions

[12]. The test results indicate that as long as the IFS is designed

properly, it can reduce stress concentration in groove welds and delay weld fractures to a drift level much higher than 4%. Specimens

IFS1 and IFS3 with a similar beam capacity demand ratio,α = 1.2

(Table 2), showed comparable deformation capacities irrespective

of shapes of the IFS and beam material properties. The maximum

moment developed at the assumed plastic hinge location was 1.25–

1.48 times the beam's actual plastic moment (Table 2); the value of

strain hardening,β, was larger for the ASTM A36 beam (Specimens

IFS1 and IFS2) than for the ASTM A572 Gr. 50 beam (Specimen IFS3). The strain hardening of around 1.5 for the ASTM A36 beam

exceeded that calculated based on FEMA 350 [11] due to minor

beam local buckling at the plastic hinge location in the test.

4.3. Beamﬂange strains

The effectiveness of the IFS in decreasing beamﬂange tensile strain

can be observed from the measured strain at a distance of 60 mm from

the column face (Fig. 9(a)). At an interstory drift of 4%, the tensile

strains in the beam topﬂange of Specimen UR range from 6 to 12%,

much higher than those of the IFS moment connections. The maximum tensile strain of Specimens IFS1-3 at an interstory drift of 4% was about

Beam Deflection (mm)

-4 -2 0 2 4

Moment (

1000 kN-m)

-240 -160 -80 0 80 160 240 -6 -4 -2 0 2 4 6 Interstory Drift (%) -2 -1 0 1 2

M/M

np

(a)

UR

Beam Deflection (mm)

-4 -2 0 2 4 Interstory Drift (%) -2 -1 0 1 2

M/M

np

(b)

IFS1

Beam Deflection (mm)

M/M

np

(c)

IFS2

Beam Deflection (mm)

-4 -2 0 2 4 Interstory Drift (%) -1 0 1

M/M

np

(d)

IFS3

1st Test 2nd Test 1st Test 2nd Test 1st Test 2nd Test

Moment (

1000 kN-m)

Moment (

1000 kN-m)

Moment (

1000 kN-m)

-240 -160 -80 0 80 160 240 -6 -4 -2 0 2 4 6 -240 -160 -80 0 80 160 240 -6 -4 -2 0 2 4 6 -240 -160 -80 0 80 160 240 -6 -4 -2 0 2 4 6

Fig. 4. Beam moment-deﬂection responses.

Fig. 5. Specimen UR beam topﬂange fracture (ﬁrst test).

(a)

First Test to +4% Drift (Second Cycle)

(b)

Second Test to -5% Drift (Second Cycle)

(6)

(a)

First Test to +4% Drift (Second Cycle)

(b)

Second Test to +5% Drift (First Cycle)

Fig. 7. Specimen IFS2 observed performance.

(a)

First Test to +4% Drift (Second Cycle)

(c)

Second Test to +6% Drift (Second Cycle)

(b)

Second Test to -6% Drift (First Cycle)

Fig. 8. Specimen IFS3 observed performance.

Distance from Beam Ceterline (mm)

0 3 6 9 12

Strain (%)

UR IFS1 IFS2 IFS3 3% Drift 4% Drift

(a)

Strain Profiles across Beam Width

-127 -63.5 0 63.5 127

0 100 200 300 400 500 600 700 800

Distance from Column Face(mm)

0 3 6 9 12

Strain (%)

UR IFS1 IFS2 IFS3 3% Drift 4% Drift

(b)

Strain Profiles along a Beam Axis

(7)

1%, indicating that the IFS was effective in reducing strain demand and thereby delayed brittle fracture of the connection to a drift higher than 4%.

Fig. 9(b) showsﬂange strains along the beam axis of all specimens.

Maximum tensile strains of Specimens UR and IFS1-3 occurred near the column face and beyond the end of the IFS, respectively. At an interstory drift of 4%, maximum strain at the assumed location of the beam plastic hinge in IFS connections, which was about

476 mm (= Ls+ db/4) away from the column face, was about 9εyin

tension and 6εyin compression, demonstrating successful relocation

of the plastic hinge away from the column face.

4.4. Internalﬂange stiffener strains

Fig. 10presents the measured longitudinal strains along the

stiff-ener depth, 35 mm from the column face. Experimental observations were that (1) longitudinal strains beyond the neutral axis of the IFS

have values opposite those of the IFS side connecting the beamﬂange,

and (2) longitudinal strains near the beam_{ﬂange are greater than}

yield strain at a drift of 4%. Because Specimen IFS2 had weaker stiff-eners than Specimen IFS1, the tensile strain in the IFS near the

beamﬂange was higher in Specimen IFS2 than in Specimen IFS1.

For Specimen IFS3 with the ASTM A572 Gr. 50 beam, much higher tensile strain could be observed as compared to Specimens IFS1 and IFS2 with the ASTM A36 beam. The maximum measured tensile strain

at an interstory drift of 4% was 1.5εyin Specimen IFS3.

5. Analytical study

Theﬁnite element models were prepared for Specimens UR, IFS1,

IFS2, and IFS3 using theﬁnite element analysis program ABAQUS[15]

to study the effectiveness of the IFS in transferring beam moment

to the column and sources of potential failure mode. Fig. 11(a)

shows theﬁnite element model consisting of eight-node brick

ele-ments C3D8R that use standard integration. The groove welds joining

the beamﬂange and column were also modeled (Fig. 11(b)). The

geometry of beamﬂange groove welds in the model was considered

based on theﬂange bevel angle and gap between the beam ﬂange

and the column. The steel backing was not modeled. Coordinates common to components joined by the shear tab and beam web were constrained such that they had identical displacements. Materi-al properties used for the models were taken from coupon tensile

tests (Table 1(b)). The stress–strain curve was approximated by a

bi-linear relationship. No residual stresses of groove welds were taken into account in the modeling. The analyses accounted for mate-rial nonlinearities, using the von Mises yield criterion. Combined isotropic and kinematic hardening was assumed for the cyclic analy-sis; the parameters for modeling were obtained based on previous

research[18].

Fig. 12shows comparisons of beam moment-deﬂection hysteretic

responses from the test and analysis. Both initial stiffness and

post-yield results show reasonable agreement with test data.Fig. 13

shows longitudinal strains in the IFS from the test and analysis, indi-cating that the force transfer from the IFS to the column can be

corre-lated well from theﬁnite element model. Moment, Ms, transferred

through the IFS to the column was computed from longitudinal stresses along the IFS depth, the respective sectional area, and

dis-tance to beam web centerline. The ratio of Msto connection moment,

MABA, computed at the column face, increased with drift (Fig. 14).

Specimen IFS1 showed higher moment resistance of the IFS than Specimen IFS2 because a stiffener with increased thickness helps

transfer a larger moment from the beamﬂange to the column. At an

Top Stiffener 35 [email protected] 35 R1R2R3R4 35 R6 R7 R8 R5 [email protected] Bottom Stiffener 1 2 3

(a)

Strain Gauge Location

-350 -175 0 175 350

Distance from Beam Web Ceterline (mm)

-0.4 -0.2 0 0.2 0.4

Strain (%)

R1-1 R2-1 R3-1 R4-1 R5-1 R6-1 R7-1 R8-1 IFS1 IFS2 IFS3 3% Drift 4% Drift

(b)

Strain Profiles

Fig. 10. IFS strain proﬁles (35 mm away from the column face).

(a)

Global Model

(b)

Beam-to-Column Interface

(8)

interstory drift of 4%, this moment ratio was about 20–25%, lower than that obtained from the IFS moment connections with a steel built-up box column and beam. It suggests that the web plate located at both sides of the box column is more effective than that located in

the center of the wide-ﬂange column to transfer the IFS moment from

the beam to the column.

The rupture index (RI) is computed at different locations of the connection from ABAQUS results to assess the possible source of frac-ture. The RI equals the product of a material constant and the PEEQ (plastic equivalent strain) divided by the strain at the ductile fracture,

εr, which is given by Hancock and Mackenzie[19]:

RI¼aPEEQ εr ¼ ffiffiffiffiffiffiffiffiffiffiffiffi 2 3ε p ijε p ij q =εy exp −1:5σm σeff ð6Þ

whereεijpis the plastic strain components;σmis the hydrostatic stress,

andσeffis the von Mises stress. Therefore, locations in a connection

with high RI values have a high potential for fracture.Fig. 15(a)

shows three possible fracture locations observed in the tests: the

beamﬂange top surface located 60 mm from the column face (Line

A), the groove-weld top surface near the column face (Line B), and

the beam ﬂange inner side along the weld between the IFS and

beamﬂange (Line C). The RI values can be signiﬁcantly reduced at

the beam _{ﬂange near the column face by providing the IFS}

[Fig. 15(b)]. The maximum RI value for Specimens IFS1-3 at both

ends of the beamﬂange groove weld is higher than that for Specimen

UR [Fig. 15(c)] because the IFSs are positioned at both edges of the

beamﬂange to transfer beam ﬂange force to the column. Moreover,

Specimen IFS3 with the ASTM A572 Gr. 50 beam hasﬂexural capacity

much higher than other specimens with the ASTM A36 beam, so it has the highest RI value among all specimens. Although the RI value at the IFS location increases, it is still lower than the fracture limit due to no weld fractures before an interstory drift of 5% in the test. The RI value

at the end of the IFS-to-beamﬂange also increases [Fig. 15(d)],

indi-cating another possible source of fracture as observed in the ﬁrst

test of Specimens IFS1 and IFS2. 6. Conclusions

Four large-scale exterior moment connection specimens, each composed of the ASTM A572 Gr. 50 H428 × 407 × 20 × 35 column and the H702 × 254 × 16 × 28 beam, were tested and analyzed to verify their seismic performance. The objective was to evaluate the

-240 -160 -80 0 80 160 240

Beam Deflection (mm)

(a)

UR

Beam Deflection (mm)

(b)

IFS1

Beam Deflection (mm)

(c)

IFS2

Beam Deflection (mm)

-4 -2 0 2 4 Interstory Drift (%) -1 0 1

(d)

IFS3

ABAQUS

Test

Moment (

1000 kN-m)

Moment (

1000 kN-m)

Moment (

1000 kN-m)

Moment (

1000 kN-m)

M/M

np

M/M

np -240 -160 -80 0 80 160 240 -240 -160 -80 0 80 160 240 -6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6 -240 -160 -80 0 80 160 240

M/M

np

M/M

np

Fig. 12. Comparison of hysteresis responses from theﬁrst test and ABAQUS analysis.

Distance from Beam Web Ceterline (mm)

-0.4 -0.2 0 0.2 0.4

Strain (%)

4.0% Drift -2 -1 0 1 2

Nomalized Strain

-350 -175 0 175 350 -350 -175 0 175 350

Distance from Beam Web Ceterline (mm)

-0.4 -0.2 0 0.2 0.4

Strain (%)

4.0% Drift -2 -1 0 1 2

Nomalized Strain

(a)

Specimen IFS2

(b)

Specimen IFS3

1st Test

ABAQUS

(9)

IFS moment connection with low-damage capability under the

code-speciﬁed loading protocol[12]. Test parameters were IFS sizes and

material properties of the beam, made by either the ASTM A572 Gr. 50 or A36 steel. The ER70S-G electrode, which is similar to the high-toughness E71T-8 or E70TG-K2 electrodes, was used to make

beam ﬂange groove welds in all specimens. Ultrasonic tests (UT)

were conducted for all ﬂange groove welds, and they all satisﬁed

the prescribed acceptance criteria [17]. Steel backing was left in

place for the top and bottomﬂanges and no ﬁllet welds were made

between the steel backing and column face. Web joints were made with only slip-critical, high-strength bolts connecting the beam web to a shear tab welded to the column face. Finite element models of specimens were prepared using solid elements to verify

IFS effectiveness and identify the possible sources of failure mode. The following conclusions are based on experimental results and associated analytical studies.

1. Specimen UR used a welded-unreinforcedﬂange-bolted web

con-nection. Although Specimen UR utilized the ASTM A36 beam and

high-toughnessﬂange groove welds, brittle fracture of the beam

ﬂange occurred before ﬁnishing the second cycle of 4% drift under

theﬁrst cyclic test. It is expected that if the welded-unreinforced

ﬂange-bolted web connection had the ASTM A572 Gr. 50 beam, the connection failure would have occurred at a low drift.

2. Three IFS connection specimens, which had beam capacity–

demand ratios larger than 1.05, experienced excellent perfor-mance and minor beam local buckling (e.g., low damage) under

theﬁrst cyclic loading test up to a drift of 4%. These specimens

were retested using the same loading protocol[12]and also

expe-rienced low damage in the beam up to a drift of 4%, leading to

sim-ilar hysteretic responses as observed in theﬁrst cyclic test. Minor

strength degradation due to beam buckling was noticed in the

sec-ond test beysec-ond drift of 4%. As long as the beam capacity–demand

ratio,α (Table 2), was near 1.2 (e.g., Specimens IFS1 and IFS3), the

IFS moment connection performed well in the second cyclic test

up to a drift of 5–6%, irrespective of the ASTM A36 or A572 Gr. 50

steel beam.

3. Maximum moment developed at a quarter beam depth from the IFS end (plastic hinge location) was 1.25 and 1.48 times the actual plastic moment of the ASTM A572 Gr. 50 and A36 beams, respec-tively. The factor of around 1.5 for the ASTM A36 beam accounted for strain hardening that was accompanied by large inelastic

Drift(%)

0 10 20 30 40 50

Ms/M

ABA

(%)

1 2 3 4

IFS1

IFS2

IFS3

Fig. 14. IFS moment contribution ratio (positive bending).

-127 -63.5 0 63.5 127

Distance from BeamWeb Centerline (mm)

0 0.5 1 1.5 2

Rupture Index

(b)

Line A

UR

IFS1

IFS2

IFS3

Distance from Beam Web Centerline (mm)

0 0.5 1 1.5 2

Rupture Index

(c)

Line B

UR

IFS1

IFS2

IFS3

UR

IFS1

IFS2

IFS3

0 100 200 300 400 500

Distance from Column face (mm)

0 0.5 1 1.5 2

Rupture Index

(d)

Line C

60 Weld

Detail A

35

500 Line A

Line B

Line C

LineB _LineA 60

Detail A

A

View A-A

Column

Top Flange

Weld

Top View

(a)

Locaton

(10)

deformations with minor beam buckling, and this value was

higher than that calculated based on FEMA 350[11].

4. Finite element analyses showed that the IFSs transferred about

20–25% connection moment to the column, lower than that

obtained from the IFS moment connection with the built-up box

column and wide-_{ﬂange beam. The IFSs were effective in reducing}

the RI demands on the beamﬂange and groove-welded joint of

the beamﬂange excluding both ends.

References

[1] Lu L-W, Ricles J, Mao C, Fisher J. Critical issues in achieving ductile behavior of welded moment connections. J Constr Steel Res 2000;55:325–41.

[2] Nakashima M, Roeder CW, Maruoka Y. Steel moment frames for earthquakes in Unites States and Japan. J Struct Eng ASCE 2000;126(8):861–8.

[3] Chi B-C, Uang C-M, Chen A. Seismic rehabilitation of pre-northridge steel moment connections: a case study. J Constr Steel Res 2006;62:783–92.

[4] Kim T, Stojadinovic B, Whittaker A. Seismic performance of pre-Northridge welded steel moment connections to built-up box columns. J Struct Eng ASCE 2008;134(2):289–99.

[5] Engelhardt MD, Winneberger T, Zekany AJ, Potyraj T. The dogbone connection: part II. Mod steel constrAISC; 1996.

[6] Uang C-M, Yu Q-S, Noel S, Gross J. Cyclic testing of steel moment connections rehabilitated with RBS or welded haunch. J Struct Eng ASCE 2000;126(1):57–68. [7] Yu Q-S, Uang C-M, Gross J. Seismic rehabilitation of steel moment connections

with welded haunch. J Struct Eng ASCE 2000;126(1):69–78.

[8] Chou C-C, Uang C-M. Cyclic performance of a type of steel beam to steel-encased reinforced concrete column moment connections. J Constr Steel Res 2002;58: 637–63.

[9] Kim T, Whittaker AS, Gilani ASJ, Bertero VV, Takhirov SM. Cover-plate and ﬂange-plate steel moment-resisting connections. J Struct Eng ASCE 2002;128(4):474–82. [10] Lee CH. Seismic design of rib-reinforced steel moment connections based on

equivalent strut model. J Struct Eng ASCE 2002;128(No. 9):1121–9.

[11] Federal Emergency Management Agency (FEMA). Recommended seismic design criteria for new steel moment-frame buildings. Rep. No. FEMA 350. Washington, D.C.: Federal Emergency Management Agency; 2000

[12] American Institute of Steel Construction (AISC). Seismic provisions for structural steel buildings, Chicago, IL; 2010.

[13] Chou C-C, Jao C-K. Seismic rehabilitation of welded steel beam-to-box column connections utilizing internalﬂange stiffeners. Earthquake Spectra 2010;26(4): 927–50.

[14] Chou C-C, Tsai K-C, Wang Y-Y, Jao C-K. Seismic rehabilitation performance of steel side plate moment connections. Earthquake Eng Struct Dyn 2010;39:23–44. [15] HKS. ABAQUS user's manual version 6.3. Pawtucket, RI: Hibbitt, Karlsson &

Sorensen; 2009.

[16] ASTM. Speciﬁcation for structural steel. Philadelphia, PA: American Society for Testing and Materials; 1988.

[17] Structural welding code-steel. Miami, Fla: Am. Welding Soc; 2006.

[18] Chou C-C, Wu C-C. Performance evaluation of steel reducedﬂange plate moment connections. Earthquake Eng Struct Dyn 2007;36:2083–97.

[19] Hancock JW, Mackenzie AC. On the mechanism of ductile fracture in high-strength steel subjected to multi-axial stress states. J Mech Phys Solids 1976;24: 147–69.