Cyclic behaviour of composite joints with reduced beam sections

Engineering Structures 136 (2017) 329–344

Contents lists available at ScienceDirect

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

Cyclic behaviour of composite joints with reduced beam sections Rui Li, Bijan Samali, Zhong Tao ⇑, Md Kamrul Hassan Centre for Infrastructure Engineering, Western Sydney University, Penrith, NSW 2751, Australia

a r t i c l e

i n f o

Article history: Received 5 June 2016 Revised 7 January 2017 Accepted 11 January 2017

Keywords: Concrete-filled steel tubular columns Reduced beam sections Through-diaphragms Beam-column joints Composite beams

a b s t r a c t Some design recommendations or standards, such as FEMA-350 and Eurocode 8 Part 3, recommend reduced beam section (RBS) connections to be used in earthquake-prone zones and practical design guidelines are provided accordingly. These recommendations, however, are mainly based on research conducted on joints without floor slabs. In reality, steel beams are often connected to reinforced concrete (RC) floor slabs by shear connectors. Thus, it is important to explore the performance of RBS joints with floor slabs. In this paper, cyclic test results of four such joint specimens are reported, where concretefilled steel tubular (CFST) columns were connected to the RBS beams by utilising through-diaphragms. To compare with the RBS joints, two reference joint specimens with or without slab were also tested. The experimental results are analysed to evaluate the influence of floor slabs on the cyclic performance of composite joints with RBS beams. It is found that composite joints with floor slabs still exhibit favourable seismic performance, and have good potential to be widely used in seismic regions. However, the presence of the floor slab should be considered in designing RBS connections. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Concrete-filled steel tubular (CFST) columns have many structural and constructional benefits [1]. The steel tube of a CFST column normally provides confinement to the concrete, and the concrete restrains the local buckling of the steel tube. In addition, the steel tube provides formwork for concrete pouring during construction, which reduces the construction cost and time. Therefore, the combined utilisation of CFST columns and steel/concrete composite beams in composite building construction has greatly increased in recent years. In using CFST columns, rigid beam-column joints are often recommended in seismic regions due to their good stiffness, high strength and easy transfer of bending moment. To form these rigid joints, through-diaphragms [2], internal diaphragms [3], and external diaphragms [4] are commonly used to connect steel beams and CFST columns. Ricles et al. [5] conducted cyclic tests on joints with square CFST columns and wide flange steel beams, where internal diaphragms were used in four specimens to connect the beams to the columns. They found that the use of internal diaphragms locally stiffened the joint, but also led to stress concentrations and fracture of the beam flanges at the weld access holes. To alleviate the stress concentrations at the beam ends, Qin et al. [6] proposed to add horizontal stiffeners at the beam ends to form tapered ⇑ Corresponding author. E-mail address: [email protected] (Z. Tao). http://dx.doi.org/10.1016/j.engstruct.2017.01.025 0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.

beam flanges. They conducted tests on two identical joints with internal diaphragms to investigate the influence of the horizontal stiffeners. It was found that fracture still occurred in a beam flange at the tip of the tapered plate. It happened at the 1st cycle of 3% or 4% rotation in either specimen. It should be noted that no floor slabs were provided by Ricles et al. [5] and Qin et al. [6] for their test specimens. In reality, shear connectors are often used to connect concrete floor to the top flange of the steel beam to develop composite action. This composite action between the two components can provide significant benefits to enhance both the stiffness and strength of the beam under sagging moment. However, it also increases the strain development of the bottom flange of the steel beam near the panel zone, and may lead to possible column failure [7]. Han and Li [8] tested 6 composite beam-CFST column joints with external diaphragms and RC slabs. Four out of the six specimens exhibited flexural failure in the beam under cyclic loading. When a beam failure occurred, fracture was observed at the weld between the bottom flange of the beam and the diaphragm corresponding to rotations as low as 2.5%. By comparing the test results reported by Han and Li [8] with those reported by Qin et al. [6], it might be inferred that the presence of slabs might promote fracture of the bottom flange of the beam. To delay or even avoid the brittle fracture near the beam flangeto-column welds, a design of reduced beam section (RBS) has been widely used by removing a portion of the beam flanges a short distance away from the column face [9]. RBS may also be used to

330

R. Li et al. / Engineering Structures 136 (2017) 329–344

make the beams weaker than the columns [10], which reduces the flexural resistance and inelastic deformation demands on the column. There have been extensive studies conducted on joints with RBS steel beams [11–13]. But concrete slabs were seldom provided, which could not reveal the realistic performance of RBS beams in construction. In particular, only very limited information is available for joints with CFST columns and RBS composite beams. To improve ductility of rigid joints, Wang et al. [10] tested five external diaphragm joints composed of RBS steel beams and CFST columns subjected to cyclic loading. No slabs were provided to the test specimens. It was shown that the RBS joints exhibited better seismic performance and ductility despite that the ultimate load reduced slightly compared with joints without RBS. In the presence of RBS, no fracture in the beam flanges was reported for the tested joints. Zhang et al. [14] studied the seismic behaviour of RBS composite joints with deep wide-flange columns, where composite floors with zinc-coated metal decks were used for the test specimens. For a typical test specimen with slab, a flange fracture at the centre of the RBS developed at 6% story drift. The reference specimen without slab developed a fracture in the heat affected zone of the beam bottom flange upon completing the first cycle of 5% story drift. Zhang and Ricles [15] conducted further finite element analysis and it was found that the presence of a floor slab can significantly reduce the column twist, leading to enhanced connection performance. Li et al. [16] proposed a type of exterior joints with extended endplates welded to RBS composite beams. Transverse ribs were welded to the circular column surface for providing planes to connect with the plane endplates. High strength bolts passing through PVC conduits in the steel tube were used to connect the RBS beams to the circular CFST columns. The test results showed that the RBS could relocate the plastic hinge from the endplate face to the reduced beam section even in the presence of the RC slab. The inelastic rotational angles of all specimens were more than 0.03 rad. Fracture of the RBS beam occurred at a rotational angle of 0.05 rad. Ciutina et al. [17] tested 6 moment-resisting beam-column connections, where three out of the six specimens had concrete slabs, and the RBS steel beam was directly welded to the steel column flange through full penetration weld. Significant RBS plasticisation was observed in all the test specimens. In the presence of the slab, the top flange of the steel beam in the RBS region was restrained from buckling. Huang et al. [18] developed an analytical formulation to investigate mechanical performance of joints with RBS composite beams, and an amplification factor was proposed to consider the slab contribution to the plastic moment of the beam section at the column face based on a parametric analysis. The above literature review clearly indicates that research on CFST column joints with RBS beams is still very limited, and further research is required to fill the following gaps: (1) Little attention has been paid to joints with throughdiaphragms to connect RBS beams and CFST columns. (2) Limited experimental data is available to clarify the influence of floor slabs on CFST joints with RBS beams. (3) There is no information on the influence of cut depth of RBS composite beams on the joint performance. FEMA-350 [19] and Eurocode 8 Part 3 [20] provide guidelines for RBS mainly based on studies of RBS steel beams without floor slabs [18]. Set against this background, six through-diaphragm joints composed of circular CFST columns and RBS composite beams were tested. These specimens included one joint with a steel beam, one joint with a composite beam and four joints with RBS composite beams. The use of through-diaphragms can achieve the following benefits: (1) Compared with external diaphragms, extrusion of the through-diaphragms out of the tube is very minimal and can

easily meet the architecture requirements. (2) The welding is easier and residual stresses resulting from welding are relatively moderate compared with using internal diaphragms. (3) The bond strength between the steel tube and concrete can be enhanced due to the presence of the diaphragms inside the steel tube [21]. The main variables considered in the test program were the beam height, cut depth of RBS and the presence of the RC slab. Effects of these parameters on the strength, ductility, strength degradation, stiffness degradation, energy dissipation capacity, and strain developments are analysed and compared in this paper based on the test results.

2. Experimental program 2.1. General Six composite cruciform joint specimens were designed and constructed based on the provisions of EC3 [22], EC4 [23], EC8 [20] and AS 2327.1 [24]. No slab was provided for specimen CN1 (reference specimen), whereas RC slabs were provided for the rest specimens. Meanwhile, RBS beams were used for four specimens, namely, CS-2, CS-3, CS-4, and CS-5. The geometric details of the four specimens with RBS beams are illustrated in Fig. 1(a)– (d), where different cut lengths (b) and cut depths (c) presented in Table 1 were adopted for the beams. The details of the other two joint specimens were the same as those of the joints with RBS beams except that no cut was made for the steel beams (CN1 and CS-1) and no slab was provided for CN-1. The specimens were designed representing joints at half scale. The profiles of circular steel tubes were 250 mm in diameter (D) with 6 mm wall thickness (ts), as shown in Fig. 1(c). Two types of universal steel beams (200UB25.4 and 250UB25.7) were selected with a length of 1500 mm from the centre of the column to the assumed inflection point of a beam. The width and depth of the RC slabs were 800 and 60 mm, respectively. Through diaphragms with a thickness of 10 mm were used to connect the column to the steel beam. The outer diameter, inner diameter and vent hole diameter of the through diaphragms were 300, 120 and 20 mm respectively, as shown in Fig. 1(b). Sixteen M19 headed shear studs with a length of 50 mm and a diameter of 19.3 mm were welded at a spacing of 200 mm along the steel beam to connect the steel beam to the floor slab, as shown in Fig. 1(a) and (d). A layer of reinforcement (/10 mm) was placed in the RC slab, which were longitudinally and transversely distributed along the slab at a spacing of 100 mm. The clear cover to the reinforcement was 20 mm. The CFST columns were the same for all six joints. Complete joint penetration (CJP) groove welds were used to connect the through diaphragm and the beam. Each CFST column with two end plates was 1600 mm in length, and was fabricated in three segments separated by the two through diaphragms, as shown in Fig. 2. The through diaphragms were welded to the steel tube by double-fillet welds. Then the beam flanges were welded to the through diaphragms, and the beam web was welded directly to the steel tube. The thicknesses of all weld seams were around 6 mm. Each end of the steel tube was welded to a 350
350 mm2 steel plate; the steel plate on the top end had a 160 mm diameter hole used for pouring concrete.

2.2. Design of the RBS beam Circular radius cuts were utilised in both top and bottom flanges of the beam to reduce the flange area near the ends of the beam, as shown in Fig. 3. FEMA-350 [19] proposes general

331

R. Li et al. / Engineering Structures 136 (2017) 329–344

Longitudinal bars bars Longitudinal 10@100 88 10@100

Distribution bars bars Distribution 34 10@100 10@100 34

100 100 40

40

65

50 50

65

A A

270 270

A

1625 1625

250

270 115 115 270

bb

800 800

aa

250 250 3500 3500

1625 1625

124 (133)

connectors Shear connectors 19.3@200 16 19.3@200

120

20

Holes for for loading loading Holes

R35 25

(b) Through diaphragm

(a) Plan (a) Plan view viewofofthe theslab slab 350 20

Endplate 160 hole P

C

Shear connectors

P

C

Reinforcement RC Slab

Right ts=6

250

250

250 C-C section 1375

250

Through Steel beam diaphragm

CFST column

20

B

800 B-B section 1375

1560 1600

Left

25

B

3500

(c) Elevation of a typical test specimen

(d) Section A-A

Fig. 1. Configuration of the specimens (units: mm).

Table 1 Details of test specimens. Specimen

Column D
ts

Beam hb
bf
tw
tf

a (mm)

b (mm)

c (mm)

n

km

Slab

CN-1 CS-1 CS-2 CS-3 CS-4 CS-5

s-250
6 s-250
6 s-250
6 s-250
6 s-250
6 s-250
6

248
124
5.0
8.0 248
124
5.0
8.0 203
133
5.8
7.8 203
133
5.8
7.8 248
124
5.0
8.0 248
124
5.0
8.0

  80 80 80 80

  180 180 220 220

0 0 26 34 25 31

0.4 0.4 0.4 0.4 0.4 0.4

1.81 0.97 1.38 1.41 1.16 1.19

No Yes Yes Yes Yes Yes

(0.2bf) (0.25bf) (0.2bf) (0.25bf)

guidelines for steel beams with RBS, where the dimensional requirements for the RBS are given in Eqs. (1)–(4).

0:2bf 6 c 6 0:25bf

a ¼ 0:50  0:75bf

ð1Þ

r ¼ ð4c2 þ b Þ=8c

b ¼ 0:65  0:85db

ð2Þ

where a is the distance from the surface of the column to the start of the cut, b is the length of the weakened region, c is the depth of

2

ð3Þ ð4Þ

332

R. Li et al. / Engineering Structures 136 (2017) 329–344

D

D

D

D

350

Through diaphragm

10 232 10 1600

654

20

654

20

676.3

20

(a) CN-1, CS-1, CS-4 and CS-5

10 187.4 10 1600

676.3

(b) CS-2, CS-3

160 350

20

(c) Section D-D

Fig. 2. Details of the steel tube (units: mm).

r

P where Mc is the sum of the flexural capacities of the upper and lower column segments (Mc1 þ Mc2 ), calculated according to the P EC4 code [23]; and Mb is the sum of the flexural capacities of the left and right beam segments (M b1 þ Mb2 ), calculated by the Australian standard AS 2327.1 [24]. Mc1, Mc2, Mb1 and Mb2 are shown in Fig. 4. The km-value of specimen CS-1 with a composite beam was 0.97, which was smaller than 1. Therefore, a column failure mode was expected for this specimen. For other specimens, their kmvalues were greater than 1 due to either the absence of the slab (CN-1) or the introduction of RBS (CS-2 to CS-5). Therefore, a beam failure mode was expected for these specimens.

RBS

c b

a Fig. 3. The RBS zone.

weakened parts, r is the radius of the circular radius cuts, and bf and db are the beam flange width and beam depth, respectively. Although the behaviour of RBS composite beams may be different from that of RBS steel beams, no design recommendations are currently available for RBS composite beams. Thus the recommendations provided by FEMA-350 [19] for RBS steel beams are utilised for the design of the current specimens. The obtained test results can be used to validate numerical models in future research, and optimal dimensions for RBS composite beams can be proposed. Parameters of the joint specimens are given in Table 1, in which n is the axial load ratio of the column, and km is the ratio of the column flexural capacity to that of the steel/concrete composite beam, which is calculated from Eq. (5).

P Mc km ¼ P Mb

ð5Þ

2.3. Material properties Coupon tensile tests were carried out to determine material properties for steel. The test results are presented in Table 2, where d denotes the diameter of shear studs or steel reinforcement, t denotes the thickness of other steel components, fy denotes the yield stress, fu denotes the ultimate strength, Es denotes the modulus of elasticity, and d denotes the elongation ratio. Grade C32 concrete was used for the infilled concrete and RC slab. Concrete cylinder tests were conducted according to Australian standards AS 1012.9 [25] and AS 1012.10 [26] to measure concrete properties. The measured concrete cylinder compressive strength (f0c ) and tensile strength (f0ct ) were 36.5 and 4.7 MPa, respectively, and the Young’s modulus of concrete (Ec) was 37,740 MPa at the time of testing. 2.4. Test setup and loading protocol

Mc1

A schematical illustration of the test setup is shown in Fig. 5, which was specially made to test large-scale columns and joints [27]. A constant axial compressive load of 1116 kN was applied to the CFST column by a hydraulic jack during the testing. The axial load level (n) of the column was 0.4, which is defined by Eq. (6).

Mb2

Mb1

n¼

Mc2

N0 Nu

ð6Þ

where N0 is the axial compressive load applied to the CFST column, and Nu is the axial compressive capacity of the CFST column determined by EC4 code [23]. Two hinges were attached to the top and bottom ends of the column to simulate a pin-pin boundary

Fig. 4. Forces and moments in the panel zone.

Table 2 The material properties of steel. Elements

d or t (mm)

fy (N/mm2)

fu (N/mm2)

ey (le)

Es (MPa)

d

Steel tube UB flange UB web Through diaphragm Shear stud Reinforcement

5.96 14.2 8.6 10.1 /12 /10

386.6 299.9 344.6 289.7 402.3 609.7

480.4 455.6 468.9 440.7 487.5 639.6

2315 1418 1537 3217 1841 3197

199,433 201,516 204,120 201,745 200,713 203,238

27.9 37.6 34.3 38.3 34.1 24.5

333

R. Li et al. / Engineering Structures 136 (2017) 329–344

Rigid beam

500 kN actuator 250

1500

Hinge

Column

RC slab

65

275

Through diaphragm

65

Beam

230

Rigid frame

Strong floor

10,000 kN hydraulic jack Fig. 5. Schematic of test setup (units: mm).

condition. The distance from the centre of a hinge to the nearest end of the specimen was 230 mm, as shown in Fig. 5. The two ends of the beam were free, where two actuators with a loading capacity of 500 kN were attached to apply cyclic loads. The distance from one loading point to the column centre was 1500 mm. The two actuators were arranged to apply equal but opposite displacements at the same time. The cyclic displacement amplitude followed the

loading protocol in SAC [28], as shown in Fig. 6. Drift angle (h) was used to control the loading history, which was defined as the beam deflection at the loading point divided by the beam span. The loading history was divided into several steps. Firstly, the loading started with six cycles at each load step of 0.00375, 0.005 and 0.0075 rad rotation, respectively. The next four cycles in the 4th load step were at 0.01 rad rotation, followed by two cycles in the

0.05 0.04

Rotation (rad)

0.05 0.03

0.03 0.01

0.00375

0.005

0.0075

0.01

0.015

-0.01 0.02

-0.03 -0.05

Number of cycles

6

6

6

4

2

Fig. 6. Cyclic loading protocol.

2

2

2

2

334

R. Li et al. / Engineering Structures 136 (2017) 329–344

5th load step of 0.015 rad rotation. The loading sequence completed two cycles at each rotation level, followed by increasing the rotation value to 0.02 rad, 0.03 rad, 0.04 rad. . .until the strength of the joint decreased to 85% of its ultimate flexural resistance. The cyclic loading speed was controlled at a rate of 0.5 mm/s. 2.5. Instrumentation Various measurements were used to quantify and elucidate the behaviour of the specimens, as shown in Fig. 7. Six linear variable differential transducers (LVDT) and four inclinometers (IM) were used to measure the vertical/horizontal displacements and rotation of the steel beams, respectively. Two linear potentiometers (LP) were used to measure the interface slip between the RC slab and the steel beam, as shown in Fig. 7(b). A total of forty-two strain gauges were attached to the reinforcement, steel beam, steel tube, and the top surface of the concrete slab, as shown in Fig. 7(a).

3. Experimental results 3.1. Joints without RBS (specimens CN-1 and CS-1) Specimen CN-1 failed by fracture of the steel beam at the end of the test. Initial yielding occurred during the first cycle of 2% drift ratio, which was observed in the left beam. Later, local buckling of the top flange occurred when the right beam reached a drift ratio of 3% and falling of the whitewash paint on the beam web was observed at the same time. When the displacement of the beam tip reached 75 mm (5% drift ratio), the ultimate flexural resistance of the joint was reached, which was 1.18 times its yielding bending moment. The applied load then started to drop and the local buckling of the beams became more obvious with the increase of the beam displacement. Finally, the test was terminated after applying cycles of 8% drift ratio, and the moment capacity of the joint decreased to 85% of its ultimate flexural resistance. Fig. 8 shows the final appearance of specimen CN-1, where fracture of the weld

35

a+0.5b

5 45

35

2

Top flange

135

26 60 60

Beam web

Reinforcement

Slab surface

a+0.5b

20

150

11 35

8

45

35

Bottom flange

45

Beam web

Steel tube

(a) Strain gauges

LVDT for horizontal displacement (HDT) LVDT for vertical displacement (VDT) Inclinometer (IM) 250

Linear potentiometer (LP)

70

1305

HDT1

LP2

LP1

VDT1

VDT3

50

VDT4

HDT2

(b) LVDTs, LPs and inclinometers Fig. 7. Setup of instrumentation (units: mm).

VDT2

335

R. Li et al. / Engineering Structures 136 (2017) 329–344

Fractured welding seam

Fig. 8. Specimen CN-1 after testing and fracture of the welding seam.

joining the top beam flange to the through diaphragm can be clearly seen, which was initiated on the second cycle of 7% drift ratio. This rotation level was relatively large, which might be attributed to the high welding quality and high ductility of the steel used for the steel beams and through diaphragms. The elongation ratios for the beam flange and through diaphragm were 37.6% and 38.3%, respectively. In contrast, the reported elongation ratio of beam flanges by Han and Li [8] was only 20.7%. Column failure was observed for specimen CS-1 as expected. The first flexural crack developed on the slab surface close to the column at a drift ratio of 0.75%, and then propagated outward. The length and width of cracks expanded along the longitudinal direction of the slab as the increase of the beam load. During successive loading cycles, the steel tube buckling occurred in the column of specimen CS-1 (on the side of the beam under hogging moment) about 40 mm below the bottom through diaphragm when the displacement reached 1.5% drift ratio. As the columnto-beam bending moment ratio (km) of specimen CS-1 was less than 1, local buckling occurred in the column before the beam flange buckled. At a drift ratio of 3%, the steel tube local buckling

(a) Joint after testing

became more obvious, and significant deformation of the column was observed afterwards, as shown in Fig. 9(a) and (b), accompanied by clear sounds of the infilled concrete being crushed. The test terminated when the strength of the joint decreased to 85% of ultimate strength on the second cycle at 3% drift ratio. The locally crushed concrete inside the tube can be observed in Fig. 9(c). It should be noted that some crushed concrete beneath the lower diaphragm spalled off when cutting and removing part of the steel tube. In contrast, only a few concrete cracks could be observed in the RC slab of specimen CS-1 after the testing, as shown in Fig. 10(a). This indicates the limited deformation of the composite beam. Since the composite beam was designed with full interaction, there was no obvious slip between the RC slab and the steel beam, as indicated by measurements of the relative slip. This was also the case for other joints with composite beams. 3.2. Joints with RBS (specimens CS-2 to CS-5) The final failure mode of all specimens with RBS composite beams is beam failure. Specimens CS-2 and CS-3 had steel beams

(b) Locally buckled column Fig. 9. Failure modes of specimen CS-1.

(a) CS-1

(b) CS-5

Fig. 10. Concrete cracks and spalled concrete around the column.

(c) Crushed concrete

336

R. Li et al. / Engineering Structures 136 (2017) 329–344

with a height of 203 mm. A small gap between the column and the RC slab was observed corresponding to 0.5% imposed drift ratio. The first flexural crack of concrete was developed at this time and grew with increasing cyclic loading. The beams of these two specimens yielded at around 2% drift ratio, and the corresponding yielding moments were 118.7 and 112.6 kN m under sagging moment, respectively. Ultimate moments subjected to sagging moment were reached at the drift ratio of 6%, and the values were 148.6 and 138.3 kN m, respectively. After that, flexural resistances under sagging moment remained stable until the end of testing. The maximum flexural resistance under hogging moment was reached at 3% drift ratio, which was smaller than that under sagging moment. This is owing to the fact that the contribution from the RC slab under hogging moment was slight when subjected to tension. The ultimate sagging moments of CS-2 and CS-3 are 26.3% and 23.6% higher than the moments under hogging flexure, respectively. Inelastic deformation of the bottom flange in the RBS zone was initiated at the loading cycle of 5% drift ratio. Buckling was also observed in the beam web at the same location, highlighted by the falling of the whitewash paint, as shown in Fig. 11 (a) and (b). This confirms the effectiveness of the RBS zone in developing plasticity. The concrete slab was then crushed and spalled around the column. Subsequently, buckling of the steel beam in the RBS zone was more serious at higher drift ratios. Specimens CS-2 and CS-3 finally failed by fracture of the bottom beam flange in the RBS zone during the 8% drift cycle, while the top beam flange in the RBS zone experienced less deformation. This is because of the support provided by the RC slab, which is consistent with the observation of Ciutina et al. [17]. Weld seams between the through diaphragms and the beam flanges were kept intact, indicating that the introduction of RBS eliminated the possible fracture of the groove welded connections. Compared with specimens CS-2 and CS-3, specimens CS-4 and CS-5 had a higher steel beam with a height of 248 mm. The phenomena observed in specimens CS-4 and CS-5 were quite similar to those of specimens CS-2 and CS-3. The progress of the tests led to initial elastic response, followed by slab cracking (at 0.75% drift ratio), bottom beam flange and web yielding (at 2% drift ratio), beam local buckling in the RBS region (at 5% drift ratio), and finally

concrete crushing in the slab and steel fracture of the beam bottom flange (at 7% drift ratio). The steel fracture occurred in the RBS region as shown in Fig. 11(c) and (d). The ultimate flexural resistances of specimens CS-4 and CS-5 subjected to sagging moment are about 43% higher than those under hogging moment. Compared to the previous two specimens CS-2 and CS-3, specimens CS-4 and CS-5 had increased yielding flexural resistances and ultimate flexural resistances under sagging moment due to the increased beam height. In contrast, only moderate strength increase was observed under hogging moment. For example, when compared to specimen CS-2, the yielding flexural resistance and ultimate flexural resistance of specimen CS-4 under sagging moment increase by 16% and 24%, respectively; whereas the increases under hogging moment are only 12% and 10%, respectively. Meanwhile, it was found that the fracture of the bottom beam flange in specimens CS-4 and CS-5 was slightly earlier (at 7% drift ratio) than that occurred in specimens CS-2 and CS-3 (at 8% drift ratio), indicating the strain demand increasing with increased beam height. Compared with specimen CS-1, other specimens with RBS beams experienced much larger lateral drift. For this reason, substantial concrete cracks and crushing were observed in the RC slabs of these specimens, which can be seen in the typical specimen CS-5 shown in Fig. 10(b). It should be noted that no fracture was observed for the reinforcement in all RC slabs. 4. Analysis of test results Moment-rotation (M-h) relations are very important to illustrate the performance of joints. In this paper, the moment (M) is calculated as the load P applied at the beam tip times the beam span (M = P
L), where L is the distance from the loading point to the column centre, taking as 1500 mm. The rotation of the joint (h) was calculated as the rotation of the beam (hb) minus the rotation of the column (hc). Two vertical differential transducers (VDT1 and VDT2) were used to measure the hb [hb ¼ ðD1  D2 Þ=2L, where D1 and D2 are the vertical displacements measured by VDT 1 and VDT 2, respectively]. Two horizontal differential transducers (HDT1 and HDT2) were used to monitor the hc [hc ¼ ðD3  D4 Þ=Ls , where D3 and D4 are the horizontal displacements measured by

(a) CS-2

(b) CS-3

(c) CS-4

(d) CS-5 Fig. 11. Specimens CS-2, CS-3, CS-4 and CS-5 after testing.

337

R. Li et al. / Engineering Structures 136 (2017) 329–344

200 150 100 50 0 -50 -100 -150 -200 -30

0

Moment (kN·m) 30

60

90

3

1

-30

0

30

-60

-30

0

30

60

90

200 150 100 50 0 -50 -100 -150 -200 -90

1

4 -60

-30

0

30

(c) CS-2

(d) CS-3

5

1

4 -30

0

30

60

90

200 150 100 50 0 -50 -100 -150 -200 -90

90

5

Rotation (mrad)

3

60

3

2

Rotation (mrad)

2

-60

1

(b) CS-1

4

-90

6

(a) CN-1

5

200 150 100 50 0 -50 -100 -150 -200

3

2

Rotation (mrad)

2

-60

200 150 100 50 0 -50 -100 -150 -200 -90

Rotation (mrad)

Moment (kN·m)

Moment (kN·m)

200 150 100 50 0 -50 -100 -150 -200 -90

Moment (kN·m)

-60

In the pre-design stage, specimen CS-1 was designed as a weak column-strong beam joint. Thus, compared with other specimens with slabs, specimen CS-1 had much smaller rotation capacity owing to the premature failure of the CFST column. The specimen was still in nearly elastic range at 2% drift ratio, and then the loadcarrying capacity decreased suddenly because of the serious local buckling of the steel tube at a drift ratio of 3%. The test was terminated before specimen CS-1 could deform further under the applied loads. After the test, part of the steel tube in specimen

3

2

-90

4.1. M-h hysteretic curves

Moment (kN·m)

Moment (kN·m)

HDT1 and HDT2, respectively; and Ls is the distance between HDT1 and HDT2], as shown in Fig. 7. Owing to the symmetry of the interior joints, hysteretic curves obtained from both sides of a specimen are very similar. Thus, only measured hysteretic curves from the right side are shown in Fig. 12. The marked points on the curves denote crack initiation (Point 1), yield point (Point 2), maximum strength (Point 3), local buckling of the web (Point 4), beam flange fracture (Point 5), and local buckling of the steel tube (Point 6), respectively. It should be noted that positive and negative moments in Fig. 12 indicate the measurements under sagging and hogging conditions, respectively.

60

90

3

2

5

1

4 -60

-30

0

30

Rotation (mrad)

Rotation (mrad)

(e) CS-4

(f) CS-5

60

90

(1. Crack initiation; 2. Yielding point; 3. Maximum strength; 4. Local buckling of the web; 5. Beam flange fracture; and 6. Local buckling of the steel tube) Fig. 12. M-h hysteretic curves of specimens.

338

R. Li et al. / Engineering Structures 136 (2017) 329–344

semi-rigid joints: between the above criteria. In classifying the joint stiffness, Wang et al. [32] used a value of 6 m based on engineering practice. In this paper, Lb is taken as 3 m retrieved from the joint specimens since the joints are not full-scale ones. (b) Strength classification. Full-strength joints: M u P M j;Rd ; Nominally pinned joints: M u 6 0:25M j;Rd ; and partialstrength joints: between the above criteria. In the comparisons, Mu is taken as the measured ultimate flexural resistance of the joint, and Mj,Rd is the calculated plastic flexural resistance of the joint according to the Australian standard AS 2327.1 [24]. For joints demonstrating a beam failure mode, Mj,Rd is equal to the design plastic flexural resistance of the beam Mb. Since specimen CS-1 demonstrated a column failure mode, it is not included in the comparison shown in Fig. 13.

CS-1 was removed and significant damage was detected in the concrete core just below the bottom through diaphragm, as shown in Fig. 9(c). This highlights the fact that the contribution from the slab should be considered in determining the km ratio of joints for design, and a weak column-strong beam design should be avoided in practice. This might be realised by providing a RBS region for the beam. All other specimens demonstrated very good deformation capacity. Their hysteretic curves are plump without obvious pinch phenomenon. Due to the presence of the RC slab, not only the sagging flexural resistance is higher than the hogging flexural resistance, but also the strength degradation is improved under sagging moment. The more obvious strength degradation under hogging moment is owing to the fast development of local buckling of the bottom beam flanges under that condition. 4.2. Flexural stiffness and flexural resistance Envelope curves of the M-h hysteretic curves are obtained by connecting the peak load points of each cycle. Comparable M-h envelope curves are obtained for both left and right beam segments in each specimen, and the curves are averaged and shown in Fig. 13. Generally, joints can be classified by their stiffness or strength [22,29–32] based on their M-h curves. The criteria suggested by EC3 Part 1-8 [33] are selected to classify joints, as shown in Fig. 13.

Moment (kN·m)

200 150 100 Rigid braced frame 50 0 -50 -100 -150 -200 -90 -60 -30

Mj,Rd

Pinned

CN-1 0

30

60

90

200 150 Rigid braced 100 frame 50 0 -50 -100 -150 -200 -90 -60 -30

Mj,Rd

Pinned

CS-2 CS-3 0

30

Rotation (mrad)

Rotation (mrad)

(a)

(b)

Moment (kN·m)

Moment (kN·m)

(a) Stiffness classification. Rigid joints: Sj;ini P aEIb =Lb (a is taken as 8 and 25 for braced and unbraced frames, respectively); Nominally pinned joints: Sj;ini 6 0:5EIb =Lb ; and

According to the measured curves, the proportional limit point for each joint is around 0.4 Mu. Thus, the secant stiffness at 0.4 Mu could reasonably represent the initial flexural stiffness Sj,ini of the specimens [32]. The obtained values of Sj,ini are presented in Table 3, where (+) and () denote sagging moment and hogging moment, respectively. Steel beams used in specimens CN-1, CS-4 and CS-5 had the same dimensions. At the presence of the slab, an increase in Sj,ini was observed under sagging moment for CS-4 and CS-5. For example, Sj,ini of CS-4 under sagging moment increased by 24.3% when compared with that of CN-1. This is illustrated in Fig. 14(a), where ‘S’ and ‘H’ represent sagging moment and hogging moment, respectively. The presence of the slab, however, had only moderate influ-

200 Mj,Rd 150 100 Rigid braced 50 frame 0 -50 -100 -150 -200 -90 -60 -30 0

Pinned

CS-4 CS-5 30

Rotation (mrad) (c) Fig. 13. M-h envelope curves.

60

90

60

90

339

R. Li et al. / Engineering Structures 136 (2017) 329–344 Table 3 Summary of measured flexural stiffness and resistance. Specimen

CN-1 CS-2 CS-3 CS-4 CS-5

Sj,ini (103 kN m/ rad) (+)

()

5.97 7.13 6.68 7.40 6.82

5.49 4.29 4.12 5.61 5.49

25EIb/Lb (103 kN m/rad)

8EIb/Lb (103 kN m/rad)

0.5EIb/Lb (103 kN m/rad)

56.5 38.4

18.1 12.3

1.13 0.77

56.5

18.1

1.13

CS-5

CS-5

CS-4

CS-4

CS-3

CS-3 H

CS-2

S

CN-1 -10

-5

0

5

10

CS-2

H

CN-1

S

-200

-100

(a) Initial flexural stiffness

0

Mu (kN m)

Mj,Rd (kN m)

(+)

()

(+)

()

147.1 148.6 138.1 184.3 171.1

139.1 117.7 111.7 127.2 120.5

92.6 139.1

92.6 110.2

161.1

133.1

100

200

(b) Flexural resistance

Fig. 14. Comparison of flexural resistance and initial flexural stiffness between different joints.

capacities of specimens CS-4 and CS-5 under hogging moment are smaller than that of CN-1 without RBS and slab by up to 13.4%. This is due to the utilisation of RBS beams in CS-4 and CS-5 while the strength contribution from the RC slab in these two specimens was minimal under hogging moment. As shown in Fig. 7, VDT3 and VDT4 were used to measure the panel zone rotation. A typical specimen CS-5 is selected to illustrate the difference in experimental stiffness based on the measured panel zone rotation and total rotation (from VDT1 and VDT2), as shown in Fig. 15. It should be noted that only curves in the initial stage of rotation are shown since the measurements from VDT3 and VDT4 became unreliable after the steel beam in the RBS zone buckled. Obviously, the flexural stiffness of the joint determined based on the total rotation (Sj,ini) is much lower than that determined based on the panel zone rotation (Sj,panel). For CS-5 under sagging moment, the value of Sj,ini is 6.82
103 kN m/ rad, whereas Sj,panel has a value of 11.23
103 kN m/rad. Compared with the stiffness Sj,panel, a reduction of 39.3% is found for Sj,ini. Meanwhile, the values of Sj,ini and Sj,panel under hogging moment are 5.49
103 and 12.01 kN m/rad, respectively. In this case,

Moment (kN·m)

ence on Sj,ini under hogging moment, which can be attributed to the fact that the RC slab was in tension under hogging condition. The cut depth of RBS slightly impacts the initial flexural stiffness of joints. Comparing specimen CS-2 having a cut depth of 26 mm (0.2bf) with CS-3 having a cut depth of 34 mm (0.25bf), the value of Sj,ini for CS-2 under sagging moment is 6.7% higher, and that under hogging moment is 4.1% higher. Similar observation is found when comparing CS-4 and CS-5. The beam height also affects the initial flexural stiffness of joints. Compared with CS-2 with an hb of 203 mm, specimen CS4 with a higher hb of 248 mm had an increased Sj,ini by 30.8% under hogging moment. At this moment, the contribution from the slab to Sj,ini was minimal. Under sagging moment, however, the increase was only 3.7% since the significant contribution from the slab under this circumstance outweighs the influence of the beam height. As shown in Fig. 13, the initial flexural stiffness of all tested specimens is less than the limit of 8ELb/Lb, which indicates that the specimens can be classified as semi-rigid joints. In general, the chosen joint type is widely recognised as a rigid one. Han et al. [8] and Wang et al. [10] conducted tests on steel beam to CFST column joints with external diaphragms. It is found that their specimens can also be classified as semi-rigid joints by comparing their test results with the EC3 classification criteria. Generally, the tested joints are small-scale specimens and the column size is relatively small compared with the beam size. This may be responsible for the relatively low initial flexural stiffness measured in the test. Further studies should be carried out to investigate the stiffness of full-scale joints with diaphragms in the future to clarify this. It can also be seen from Fig. 13 that all specimens shown in this figure can be classified as full-strength joints. Their maximum flexural capacities (Mu) are higher than or at least very close to Mj,Rd under both sagging and hogging conditions. As shown in Fig. 14 (b), the effects of the presence of the RC slab, cut depth of RBS and beam height on the joint flexural resistance are similar to those on the initial flexural stiffness. However, the flexural

200 150 100 50 0 -50 -100 -150 -200 -90

Panel zone rotation Total rotation

-60

-30

0

30

60

90

Rotation (mrad) Fig. 15. Comparison of total rotation and panel zone rotation (CS-5).

340

R. Li et al. / Engineering Structures 136 (2017) 329–344

the absolute value of Sj,ini is 54.3% lower than that of Sj,panel. The significant stiffness reduction is attributable to the additional flexural deflection of the beam, especially developed in the RBS zone. This comparison highlights that both the reduced beam section and panel zone significantly affects the flexural stiffness of the joint. Meanwhile, the RBS influence is more significant under hogging moment than under sagging moment. 4.3. Rotation ability and ductility ratio Rotation ability and ductility ratio (l) are compared in this section to evaluate the joint performance. The rotation (hf) when the moment decreased to 0.85Mu is used to represent the rotation ability of a joint, as shown in Fig. 16. The ductility ratio (l) is defined as hf/hy, where hy is the rotation corresponding to the yield point on the M-h envelope curve, which is determined according to Fig. 16. Values of hf and l for each specimen are presented in Table 4. FEMA-350 [19] suggests that the minimum story drift angle at connection failure (hf) should be at least 0.03 rad for ordinary moment frames. As can be seen from Table 4, all current specimens satisfy this criterion except specimen CS-1 due to the premature column failure. The range of hf-values for all specimens except CS-1 is from 0.058 to 0.080 under sagging moment, whereas the corresponding range is from 0.049 to 0.064 under hogging moment. Compared with the reference specimen CN-1 without a slab, specimens CS-2 to CS-5 demonstrated increased rotation ability under sagging moment due to the presence of the slab. But the presence of the slab slightly decreased the rotation ability under hogging moment. Meanwhile, it is found that the influence of the cut depth of RBS on hf is not obvious. By comparing the ductility ratios shown in Table 4, it is found that the effects of different parameters on l are generally very similar to those on hf, as discussed earlier. Specimen CS-1 had very limited ductility due to the column failure mode. In contrast, l-values of other specimens were 2.1–3.3 times that of CS-1 under sagging moment, and 1.5–1.8 times that of CS-1 under hogging moment.

M Mu Mf

0.15Mu

My

θy

θu

θ

θf

Fig. 16. Definition of yield, ultimate and failure points.

This is due to the change from the column failure mode to beam failure mode, which further confirms the effectiveness of using RBS beam in promoting the beam failure mode. Another finding is that l increased slightly with increasing beam height under hogging condition. This can be attributed to the slightly reduced flange slenderness ratio for the steel beams used in CS-4 and CS-5, which had a higher height. In contrast, specimens CS-4 and CS-5 had smaller l-values than CS-2 and CS-3 under sagging condition. This is attributable to the presence of the RC slab, which was under compression in this circumstance. The higher the composite beam, the larger the compressive strains developed in the concrete slab at a same curvature. Due to the strain-softening behaviour of concrete, steeper declining trend would be expected for the M-h envelope curves of CS-4 and CS-5 under sagging condition, leading to reduced ductility. 4.4. Strength degradation Because of cumulative damages in the specimens under cyclic loading, the peak force attained during loading cycle at a same displacement level will decrease. The strength degradation coefficient mj is defined by Eq. (7).

v j ¼ F ij =F 1j

ð7Þ

where F ij and F 1j are the maximum loads under the ith and 1st loading cycle, respectively, when the corresponding loading cycle equals j [34]. Fig. 17 depicts strength degradation coefficient versus rotation relations (mj-h), where the value of mj is an average value calculated from beam segments on both sides. Since the strength did not obviously degrade before yielding, the curves shown in Fig. 17 start at 1.5% drift ratio. As only two loading cycles at the same displacement level were applied to the joints at 1.5% drift ratio and beyond, mj in this paper is taken as the ratio of the ultimate load of the second cycle to that of the first cycle (F 2j =F 1j ). It is found that the strength degradation coefficient of a specimen decreases as the rotation increases due to the cumulative damages. Before reaching the maximum strength, values of strength degradation coefficients are generally higher than 0.94 and 0.90 under sagging moment and hogging moment, respectively, indicating minor strength degradation. After the maximum strength was reached, mj decreased continuously. Once the beam flange fractured or the column was crushed, a sharp decrease in mj was observed, as can be seen from Fig. 17. To illustrate the influence of RC slab and RBS on the strength degradation of joints, Fig. 17(a) compares mj-h relationships of specimens CN-1 (steel beam), CS-1 (composite beam without RBS), and CS-5 (RBS composite beam), where all specimens used 250UB steel beams. Specimen CN-1 demonstrated gradual strength degradation, whereas specimen CS-1 had a sharp drop at 3% drift ratio due to the premature column failure. Obviously, the presence of the slab increased the flexural capacity of the beam, leading to a weak column-strong beam condition for CS-1. Due to the presence of the slab, specimen CS-5 demonstrated faster strength degrada-

Table 4 Summary of the measured results. Specimen

CN-1 CS-1 CS-2 CS-3 CS-4 CS-5

hy (mrad)

hu (mrad)

hf (mrad)

l

(+)

()

(+)

()

(+)

()

(+)

()

(+)

()

(+)

()

123.4 143.2 118.7 112.6 138.1 136.3

114.7 91.3 94.8 93.8 115.2 98.9

153.1 149.9 148.6 138.1 184.3 171.1

139.1 112.1 117.7 111.7 127.2 120.5

23 24 22 20 23 24

25 17 26 28 21 19

49 25 60 70 59 58

49 20 39 41 30 30

58 29 80 79 69 69

64 25 58 59 53 49

2.52 1.21 3.61 3.98 3.00 2.88

2.56 1.47 2.24 2.15 2.52 2.58

My (kN m)

Mu (kN m)

l

Et (kN m)

2.54 1.34 2.93 3.06 2.76 2.73

135.02 12.62 129.30 123.23 106.03 97.79

341

1.2

1.2

1.1

1.1

1

1

0.9

0.9

νj

νj

R. Li et al. / Engineering Structures 136 (2017) 329–344

0.8

0.6 -0.09 -0.06 -0.03 0

0.8

CN-1 CS-1 CS-5

Column failure

0.7

Flange fracture

0.6 -0.09 -0.06 -0.03 0

0.03 0.06 0.09

θ (rad)

1.2

1.1

1.1

1

1

0.9

0.9

νj

νj

(b) CS-2 and CS-3

1.2

0.8

Flange fracture

0.7

0.03 0.06 0.09

θ (rad)

(a) CN-1, CS-1 and CS-5

0.8

CS-2 CS-3

0.7

0.6 -0.09 -0.06 -0.03 0

CS-4 CS-5

CS-2 CS-4

0.7 0.6 -0.09 -0.06 -0.03 0

0.03 0.06 0.09

0.03 0.06 0.09

θ (rad)

θ (rad)

(c) CS-4 and CS-5

(d) CS-2 and CS-4 Fig. 17. mj-h relationships.

tion than specimen CN-1. The mj-value of the former is about 13% lower than that of the latter at 7% drift ratio. Despite this, the strength degradation of CS-5 was relatively slow due to the introduction of the RBS. Fig. 17(b)–(d) shows the development of mj as a function of rotation for specimens with RBS composite beams (CS2 to CS-5). It can be seen that both cut depth (Fig. 17(b) and (c)) and beam height (Fig. 17(d)) have only moderate impact on the strength degradation if no flange fracture occurs. However, the increased cut depth indeed propagates the flange fracture at a later stage, leading to sharp strength degradation. 4.5. Stiffness degradation The stiffness degradation of specimens under cyclic loading can be evaluated by the stiffness degradation coefficient (Sj) [35], which is calculated by Eq. (8).

3000 2000 1000

0.03 0.06 0.09

n X Dij

ð8Þ

i¼1

where Dij denotes the peak drift of the ith cycle; Pij denotes the peak load applied to the beam at the ith cycle; and n denotes the cycle time of loading. Fig. 18 shows the stiffness degradation coefficient versus rotation relationships (Sj-h), where the values of Sj are average values calculated from both beam segments of a specimen. As shown in Fig. 18(a), the influence of the RC slab on the joint stiffness degradation is similar to that on the strength degradation. Due to the presence of the RBS, specimen CS-5 under hogging moment demonstrates similar trend as that of specimen CN-1 in the development of Sj. But CS-5 has higher flexural stiffness than CN-1 under sagging condition because of the slab contribution.

4000

CN-1 CS-1 CS-5

0 -0.09 -0.06 -0.03 0

, Pij

i¼1

S j (kN/mm)

S j (kN/mm)

4000

Sj ¼

n X

3000

CS-2 CS-3 CS-4 CS-5

2000 1000 0 -0.09 -0.06 -0.03 0

0.03 0.06 0.09

θ (rad)

θ (rad)

(a) CN-1, CS-1 and CS-5

(b) CS-2 to CS-5

Fig. 18. Sj-h relationships.

342

R. Li et al. / Engineering Structures 136 (2017) 329–344

of RBS leads to a slight decrease in energy dissipation. Further research is required to optimise the RBS design for composite beams.

Fig. 18(b) compares the stiffness degradation of the joints with RBS composite beams (CS-2 to CS-5). It can be seen that both the cut depth of RBS and steel height only have moderate influence on the flexural stiffness degradation of joints. The degradation of Sj for specimens CS-4 and CS-5 with a higher hb of 248 mm is slightly faster than that of specimens CS-2 and CS-3 with a lower hb of 203 mm. This is owing to the fact that a higher beam develops more significant plasticity at a same rotation level.

4.7. Strain development The test data indicates that strain developments for specimens CS-2 and CS-3 are quite similar to each other, which is also true for specimens CS-4 and CS-5. The effects of RBS on the strain development in the joints are negligible. Thus, only typical specimens CN1, CS-1, and CS-5 are selected to analyse the strain development in this section, where positive and negative strains represent tension and compression, respectively. Fig. 19 shows the strain developments in specimens CN-1, CS-1, and CS-5 at different drift ratios, where strain gauges No. 2 and No. 5 were on the top flange, and strain gauges No. 8 and No. 11 were on the bottom flange. It is observed from Fig. 19(a) that the strain developments in both top and bottom flanges are quite similar for specimen CN-1 since no slab is available. The flange of this specimen yielded at 3% drift ratio. For specimen CS-1 with slab, the steel beam remains elastic since column failure occurs. After the introduction of a RBS region, strains of section D-D close to the column face are relatively small for specimen CS-5, as shown in Fig. 19(c). But the strains (eb) of section E-E in the RBS region are much higher than the yield strain (ey) at 2% drift ratio, indicating the early formation of a plastic hinge in the RBS region. This avoids the possible column failure. When CS-5 reaches a 3% drift ratio, the bottom flange strain is 2.5 times that of the top flange strain at section E-E; this confirms the existence of composite action between the RC slab and steel beam.

4.6. Energy dissipation The capacity of energy dissipation is an important performance measure for joints. The bigger the area of hysteretic loops of the PD relationship, the better the seismic behaviour of the joint. Table 4 shows the total accumulated energy (Et) for different specimens, where Et is the sum of the energy dissipation in each cycle, calculated as the area of each P-D hysteretic loop. Specimen CN-1 with steel beams absorbed the most amount of energy (Et) of 135.0 kN m during the test. In contrast, specimen CS-1 (composite beams without RBS) failed very early due to the column failure, thus it could absorb very limited amount of energy (12.6 kN m). In general, the presence of the RC slab and RBS leads to a slight reduction in Et, but the influence is not very significant. The current test results indicate that increasing beam height could reduce the energy dissipation capacity. The total accumulated energy absorbed by the specimens CS-4 and CS-5 (with 248 mm height beam) decreases on average by 24.3% compared with the specimens CS-2 and CS-3 (with 203 mm height beam). This can be explained by the earlier bottom flange failure in the RBS region of the beam. In addition, the increase in the cut depth

20

20 1% 2% 3%

16 12 8

1% 2% 3%

16

D

E

2

5

8

11

D

E

D

12 8

εy

E

2

5

8

11

D

E

εy

4

4

0

0 2

5

8

11

2

5

8

11

Strain gauge number

Strain gauge number

(a) CN-1

(b) CS-1

20 1% 2% 3%

16

D

12 8

E

2

5

8

11

D

E

εy

4 0

2

5

8

11

Strain gauge number (c) CS-5 Fig. 19. Strain development along the beam span.

343

Moment (kN·m)

200 Section F-F 150 Section G-G 100 50 0 26 -50 F F εy -100 G -150 G 20 -200 -15000 -10000 -5000

εy

0

5000

200 Section F-F 150 Section G-G 100 50 0 -50 26 F F -100 εy -150 G G 20 -200 -15000 -10000 -5000

εc (με)

εc (με)

(a) CN-1

(b) CS-1

Moment (kN·m)

Moment (kN·m)

R. Li et al. / Engineering Structures 136 (2017) 329–344

200 Section F-F 150 Section G-G 100 50 0 26 -50 F F -100 G -150 G εy 20 -200 -15000 -10000 -5000

εy 0

5000

εy 0

5000

εc (με) (c) CS-5 Fig. 20. Strain development in the steel tube.

Two sections, i.e., section F-F and section G-G, are selected to analyse the strain developments in the columns of typical specimens CN-1, CS-1, and CS-5. Fig. 20 shows the column moment (Mc) versus strain (ec) relationship, in which Mc is calculated by Eq. (9).

Mc ¼

PL1 H1 H

ð9Þ

where P is the applied load on the right or left beam ends, L1 is the distance between the two loading points on the beam; H1 is the distance from the selected section to the closer column end and H is the height of the column. It is found that the strain developments in both sections F-F and G-G are similar for specimen CN-1 without slab, as shown in Fig. 20(a). The column remains elastic and the maximum strain developed is 1760 le, which is smaller than the yield strain of 2315 le. For the joints with RC slab, local buckling of the steel tube occurred near section G-G of specimen CS-1 during the loading. The steel tube above the slab remained intact, whereas the steel tube at section G-G developed significant plasticity after the ultimate moment was reached, as shown in Fig. 20(b). Specimen CS-5 developed a beam failure mode. Its steel tube above the slab remained elastic, whereas section G-G below the slab developed obvious plasticity, as shown in Fig. 20(c). This is due to the additional axial stress resulted from the bending moment exerted on the column.

(1) The introduction of the reduced beam section (RBS) region in composite beams can promote the plastic hinge formation in the RBS region, and avoid the weld fracture at the junction between the through diaphragm and the beam flange. Through-diaphragms are suitable to be used to connect RBS beams to concrete-filled steel tubular columns. (2) Both the RBS and panel zone significantly affect the flexural stiffness of the joint. The RBS influence is more significant under hogging moment than under sagging moment. (3) The presence of the RC slab improves the flexural strength and stiffness of the composite joints under sagging moment, but reduces the energy dissipation capacity of joints. In general, the cut depth of RBS has only slight impact on the seismic behaviour of composite joints. But increased cut depth propagates the flange fracture at a later stage, leading to sharp strength degradation. Further research is required to optimise the design of RBS in composite beams. (4) According to Eurocode 3, the tested joints can be classified as full-strength joints, but are semi-rigid. Further studies should be carried out to investigate the stiffness of fullscale joints with diaphragms. The tested joints with RBS composite beams also have reasonable high rotation capacity, which is greater than 0.03 rad if the column failure can be avoided. The through-diaphragm joints are suitable to be applied in seismic regions, and care should be taken to ensure a strong column-weak beam design.

5. Conclusions An experimental research was conducted to investigate the cyclic behaviour of composite joints with reduced beam sections. It can be concluded that:

It should be noted that the number of test specimens is still very limited in this study. Further numerical study, e.g. finite element analysis, should be conducted to extend the range of test data, and to investigate the influence of other variables not investigated

344

R. Li et al. / Engineering Structures 136 (2017) 329–344

in the present study. This experimental work has provided a very good basis for the development of theoretical models and design guidelines. References [1] Han LH, Li W, Bjorhovde R. Developments and advanced applications of concrete-filled steel tubular (CFST) structures: members. J Constr Steel Res 2014;100:211–28. [2] Qin Y, Chen ZH, Wang XD. Elastoplastic behaviour of through-diaphragm connections to concrete-filled rectangular steel tubular columns. J Constr Steel Res 2014;93:88–96. [3] Nie JG, Qin K, Cai CS. Seismic behaviour of connections composed of CFSSTCs and steel–concrete composite beams–-experimental study. J Constr Steel Res 2008;64(10):1178–91. [4] Zhang DX, Gao S, Gong J. Seismic behaviour of steel beam to circular CFST column assemblies with external diaphragms. J Constr Steel Res 2012;76:155–66. [5] Ricles JM, Peng SW, Lu LW. Seismic behavior of composite concrete filled steel tube column–wide flange beam moment connections. J Struct Eng 2004;130 (2):223–32. [6] Qin Y, Chen Z, Wang X. Experimental investigation of new internal–diaphragm connections to CFT columns under cyclic loading. J Constr Steel Res 2014;98:35–44. [7] Li W, Han LH. Seismic performance of CFST column to steel beam joint with RC slab: Analysis. J Constr Steel Res 2011;67(1):127–39. [8] Han LH, Li W. Seismic performance of CFST column to steel beam joint with RC slab: experiments. J Constr Steel Res 2010;66(11):1374–86. [9] Shen J, Kitjasateanphun T, Srivanich W. Seismic performance of steel moment frames with reduced beam sections. Eng Struct 2000;22(8):968–83. [10] Wang WD, Han LH, Uy B. Experimental behaviour of steel reduced beam section to concrete–filled circular hollow section column connections. J Constr Steel Res 2008;64(5):493–504. [11] Han SW, Moon KH, Stojadinovic B. Design equations for moment strength of RBS-B connections. J Constr Steel Res 2009;65(5):1087–95. [12] Naghipour M, Javadi N, Naghipour A. Investigation of RBS connection ductility in eccentrically braced frame. Proc Eng 2011;14:743–52. [13] Oh K, Lee K, Chen L, Hong SB, Yang Y. Seismic performance evaluation of weak axis column-tree moment connections with reduced beam section. J Constr Steel Res 2015;105:28–38. [14] Zhang XF, Ricles JM, Lu LW, Fisher JW. Analytical and experimental studies on seismic behavior of deep column-to-beam welded reduced beam section moment connections. In: 13th World conference on earthquake engineering. Vancouver (BC, Canada). [15] Zhang XF, Ricles JM. Seismic behavior of reduced beam section moment connections to deep columns. J Struct Eng 2006;132(3):358–67. [16] Li X, Xiao Y, Wu Y. Seismic behavior of exterior connections with steel beams bolted to CFT columns. J Constr Steel Res 2009;65(7):1438–46.

[17] Ciutina A, Dubina D, Danku G. Influence of steel-concrete interaction in dissipative zones of frames: I-Experimental study. Steel Compos Struct 2013;15(3):299–322. [18] Huang Y, Yi WJ, Zhang R, Xu M. Behavior and design modification of RBS moment connections with composite beams. Eng Struct 2014;59:39–48. [19] Federal Emergency Management Agency. Recommended seismic design criteria for new steel moment-frame buildings. FEMA-350, Washington (DC, USA); 2000. [20] British Standards Institution. Design of structures for earthquake resistance  Part 3: assessment and retrofitting of buildings, BS EN 1998–3, Eurocode 8. London (UK): BSI; 2005. [21] Tao Z, Song TY, Uy B, Han LH. Bond behavior in concrete-filled steel tubes. J Constr Steel Res 2016;120:81–93. [22] British Standards Institution. Design of steel structures  Part 1. 1: general rules and rules for buildings, BS EN 1993-1-1, Eurocode 3. London (UK): BSI; 2005. [23] British Standards Institution. Design of composite steel and concrete structures  Part 1. 1: general rules and rules for buildings, BS EN 1994-1-1, Eurocode 4. London (UK): BSI; 2006. [24] Standards Australia. Composite structures, Part 1: simply supported beams, AS 2327.1. Sydney (Australia); 2003. [25] Standards Australia. Methods of testing concrete, Method 9: compressive strength tests-concrete, mortar and grout specimens, AS 1012.9. Sydney (Australia); 2014. [26] Standards Australia. Methods of testing concrete, Method 10: determination of indirect tensile strength of concrete cylinders (‘Brazil’ or splitting test), AS 1012.10. Sydney (Australia); 2014. [27] Tao Z, Hassan MK, Song TY, Han LH. Experimental study on blind bolted connections to concrete-filled stainless steel columns. J Constr Steel Res 2017;128:825–38. [28] SAC Joint Venture. Protocol for fabrication, inspection, testing and documentation of beam-column connection tests and other experimental specimens, SAC/BD-97/02. Sacramento (CA, USA); 1997. [29] American Institute of Steel Construction. Specification for structural steel buildings, ANSI/AISC 360-05. Chicago (IL, USA); 2005. [30] Hasan R, Kishi N, Chen WF. A new nonlinear connection classification system. J Constr Steel Res 1998;47(1):119–40. [31] Nethercot DA, Li TQ, Ahmed B. Unified classification system for beam-tocolumn connections. J Constr Steel Res 1998;45(1):39–65. [32] Wang ZB, Tao Z, Li DS, Han LH. Cyclic behaviour of novel blind bolted joints with different stiffening elements. Thin-Walled Struct 2016;101:157–68. [33] British Standards Institution. Design of steel structures  part 1. 8: design of joints, BS EN 1993-1-8, Eurocode 3. London (UK): BSI; 2005. [34] JGJ101-96. Specification for test methods of seismic buildings. Beijing: Architecture Industrial Press of China; 1997 [in Chinese]. [35] Wang JF, Han LH, Uy B. Hysteretic behaviour of flush end plate joints to concrete-filled steel tubular columns. J Constr Steel Res 2009;65(8):1644–63.