Bond-slip behaviour between GFRP I-section and concrete

Accepted Manuscript Bond-slip behaviour between GFRP I-section and concrete Jian Song Yuan, Muhammad N.S. Hadi PII:

S1359-8368(17)30232-9

DOI:

10.1016/j.compositesb.2017.07.060

Reference:

JCOMB 5196

To appear in:

Composites Part B

Received Date: 19 January 2017 Revised Date:

24 April 2017

Accepted Date: 29 July 2017

Please cite this article as: Yuan JS, Hadi MNS, Bond-slip behaviour between GFRP I-section and concrete, Composites Part B (2017), doi: 10.1016/j.compositesb.2017.07.060. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Submitted to Composite Part B

1

2 3

Bond-slip Behaviour between GFRP I-section and Concrete

5

Jian Song Yuan, Muhammad N. S. Hadi *

6

School of Civil, Mining and Environmental Engineering, University of Wollongong,

7

NSW 2522, Australia

SC

8 9

M AN U

10 11 12 13 14

TE D

15 16

Correspondence:

18 19 20 21 22 23 24

Muhammad N. S. Hadi School of Civil, Mining & Environmental Engineering University of Wollongong, Australia E-mail: [email protected] Telephone: + 61 2 4221 4762 Facsimiles: + 61 2 4221 3238

AC C

EP

17

25

RI PT

4

26

--------------------------------------------------------

27

* Corresponding author

28

1

ACCEPTED MANUSCRIPT Bond-slip Behaviour between GFRP I-section and Concrete

30

Jian Song Yuan, Muhammad N.S. Hadi

31

Abstract: This paper presents the results of an experimental study on the bond behaviour of

32

glass fibre reinforced polymer (GFRP) I-section embedded in concrete. A total of five

33

specimens with the same cross-section dimension were cast and tested using push-out test.

34

The main parameters investigated in this study were bond length (300 mm and 450 mm),

35

transverse stirrups and sand coating. The experimental results show that the ultimate bond

36

stress can be improved by a longer bond length and sand coating. However, the ultimate bond

37

stress was reduced when stirrups were used, and the reason may be because the application of

38

stirrups affected the vibration of the concrete, causing a weak bond at the interface. The bond

39

stress distribution at the web and the flange is analysed based on the strain of the I-section.

40

Finally, a bond stress-slip model is proposed for the GFRP I-section with a smooth surface.

41

Based on this model, the theoretical results are in good agreement with the experimental

42

results.

SC

M AN U

TE D

45

Keywords: Bond-slip; Theoretical model; GFRP pultruded profile; I-section; Concrete.

EP

44

AC C

43

RI PT

29

2

ACCEPTED MANUSCRIPT Research Highlights

47

Bond behaviour between GFRP I-section and concrete is assessed.

48

Push-out test is used to investigate the bond behaviour of GFRP I-section to concrete.

49

Effect of bond length and sand coating is investigated.

50

Theoretical model of bond stress-slip relationship is proposed.

AC C

EP

TE D

M AN U

SC

RI PT

46

3

ACCEPTED MANUSCRIPT 1. Introduction

52

Fibre Reinforced Polymer (FRP) pultruded profiles have been increasingly investigated in

53

recent years. Conventional FRP Pultruded profiles are usually made of glass fibres embedded

54

in a vinylester or polyester matrix (GFRP Pultruded profiles) [1, 2], and have some

55

advantages such as low self-weight, high strength and corrosion resistance [3]. Moreover,

56

GFRP pultruded profiles are recommended to be used in projects with a demand for faster

57

construction due to ease of installation [4]. The application of GFRP pultruded profiles

58

mainly includes two types, all GFRP structures [5] as well as hybrid structures [6, 7]. Among

59

them, the hybrid structures reinforced with GFRP pultruded profiles have recently gained

60

more attention due to the superior structural behaviour [8, 9].

M AN U

SC

RI PT

51

61

The performance of hybrid structures is dependent upon the properties of concrete and

63

reinforcement, as well as the bond behaviour between the two components [10]. Therefore,

64

an adequate bond between the concrete and the reinforcement is important for the

65

performance improvement of hybrid structures. Nevertheless, the bond behaviour of GFRP

66

pultruded profiles in concrete is traditionally weak due to the smooth surface, thus causing a

67

poor performance of the hybrid structures [7, 11-13]. Moreover, when compared with GFRP

68

bars or steel bars, GFRP pultruded profiles usually have a larger surface. Hence, the influence

69

of the bond behaviour at the interface is more significant. In order to achieve a good

70

composite action for the GFRP pultruded profiles used in the composite structures, it is

71

essential to understand the bond mechanisms and determine the bond-slip constitutive laws.

AC C

EP

TE D

62

72

73

The existing investigation of bond-slip model of FRP can be divided into two types, FRP

4

ACCEPTED MANUSCRIPT sheet/plates bonded to concrete [14-17] and FRP bars in concrete [10, 18, 19]. These two

75

types of bond-slip models are not suitable for the GFRP pultruded profiles. The bond-slip

76

model for FRP sheet/plate bonded to concrete cannot be used for GFRP pultruded profile due

77

to the different interface properties. Epoxy resin is usually used between FRP sheet/plate and

78

concrete to provide strong adhesion, while no adhesive is used to bond the GFRP pultruded

79

profiles and concrete. In terms of FRP bars, the size effect cannot be ignored since the

80

majority of GFRP pultruded profiles have a much larger surface (i.e. GFRP I-section, GFRP

81

tube) than FRP bars. Therefore, the investigation on the bond behaviour of the GFRP

82

pultruded profiles in concrete is needed both experimentally and theoretically.

M AN U

SC

RI PT

74

83

The bond behaviour of GFRP pultruded profiles in concrete was firstly investigated in this

85

experimental study. The GFRP pultruded profiles used was GFRP I-section (I-section). For

86

the experimental method, the common pull-out test [20, 21] was not used due to the difficulty

87

of fixing the I-section in the testing machine, and a push-out test [22, 23] was adopted in this

88

study. As a preliminary test, a total of five specimens with different configurations were

89

tested. The parameters investigated included bond length, transverse stirrups and sand

90

coating. Based on the experimental results, the failure modes, bond stress-slip curve and bond

91

stress distribution are presented. Afterwards, the effect of stirrups and bond length is

92

discussed, and the mechanism of the load transfer along the interface between the I-section

93

and the concrete is analysed. Finally, a bond stress-slip constitutive model is proposed. The

94

predictions from this model are in close agreement with the experimental results.

AC C

EP

TE D

84

95

96

5

ACCEPTED MANUSCRIPT 2. Experimental Program

98

2.1 Test Specimens

99

A total of five specimens (Fig. 1) were fabricated and tested, and Fig. 2 shows two types of

100

cross-section for the five specimens, Section A-A (Specimens A and B) and Section B-B

101

(Specimens AS, BS and BSS). The design of the specimens, including the dimensions of the

102

cross-section as well as the space between the stirrups, has been justified by a flexural test

103

conducted by the authors [24]. Both types of cross-section have dimensions of 200 mm in

104

width and 350 mm in length. For all the specimens, the I-sections were placed at the centre of

105

the concrete, and the web was parallel with the long side of the cross-section. At the top end

106

of each specimen, part of the I-section (free end) was left outside of the concrete to push the

107

I-section out.

M AN U

SC

RI PT

97

108

A 50 mm clear distance was left for the debonding at the bottom of the I-section, and a layer

110

of plastic tape was employed on the surface of the I-section to debond the concrete and the I-

111

section within this region (debonding region). As shown in Fig. 3, the design of this

112

debonding region refers to the design of the specimens for the pull-out test of FRP bars as

113

recommended by ACI 440.3R-04 [25]. This debonding region is beneficial for the push-out

114

of GFRP I-section without the effect of the crushing of the concrete.

EP

AC C

115

TE D

109

116

The R10 steel bars with 10 mm nominal diameter and 250 MPa nominal tensile strength were

117

used as stirrups in Specimens AS, BS and BSS. Since the longitudinal bars cannot provide

118

any confinement for the concrete, it is believed that these bars have little effect on the bond

119

behaviour. Therefore, R10 bars were also employed as longitudinal reinforcement for ease

6

ACCEPTED MANUSCRIPT 120

fabrication of the specimens. Table 1 shows the test matrix of the specimens.

121

The label of the specimens consists of three parts. The first part is the letter A or B, which

123

indicates the different bond length of the specimen (300 mm for A and 450 mm for B). The

124

second part is the letter S indicating that the transverse stirrups are used in this specimen.

125

Lastly, the third letter S in the label means that sand coating was used on the surface of the I-

126

section.

SC

RI PT

122

M AN U

127

Specimen A was made of the I-section and concrete as shown in Fig. 1a, and the bond length

129

is 300 mm. Transverse stirrups were used in Specimen AS to investigate the effect of stirrups

130

on improving the bond behaviour (Fig. 1b), and four longitudinal bars were used to fix the

131

transverse stirrups. Specimen B (Fig .1c) was composed of the concrete and the I-section, and

132

the bond length is 450 mm. The longitudinal bars and transverse stirrups were used in

133

Specimen BS and Specimen BSS (Fig. 1d). Moreover, the I-section in Specimen BSS was

134

coated with sand to improve the friction at the interface.

EP AC C

135

TE D

128

136

2.2 Material Properties

137

Five samples of R10 steel bars were tested in tension based on the AS 1391 [26]. The average

138

tensile yield strength of the steel bars was 309 MPa and the elastic modulus was 192.5 GPa.

139

The concrete was supplied by a local company with a nominal compressive strength of 30

140

MPa and a slump of 120 mm, and the main composition of the concrete is given in Table 2.

141

Three concrete cylinders (100 mm × 200 mm/diameter × height) were cast to determine the

142

compressive strength of concrete. The cylinders were tested at 28 days and the average 7

ACCEPTED MANUSCRIPT 143

compressive strength of concrete was 31.8 MPa.

144

Fig. 4 shows the I-section used in this study, which was provided by Treadwell Group

146

Company [27] and manufactured by a pultrusion technology. The dimensions of the I-section

147

were 10 mm in thickness (both in flange and web), 100 mm in width and 200 mm in height.

148

Traditionally, the majority of GFRP fibres in the pultruded profiles are laid in the longitudinal

149

direction, therefore, only longitudinal strength of both web and flange were determined. The

150

tensile tests were conducted by using ISO 527 [28] and the compressive strength was

151

determined using ASTM D695 [29]. In total 20 coupons extracted from the I-section were

152

tested. Ten coupons (five from flange and five from web) were tested to determine the tensile

153

strength and the other ten coupons (five from flange and five from web) to determine the

154

compressive strength of the I-section. The coupons for the tensile strength test had nominal

155

dimensions of 25 mm in width and 250 mm in length, and for the compressive strength test,

156

the nominal dimensions of the coupon were 12.7 mm in width and 38.1 mm in length. The

157

test results are summarized in Table 3.

SC

M AN U

TE D

EP

158

RI PT

145

2.3 Preparation of Specimens

160

The preparation process of the specimens included attaching the strain gages and casting

161

concrete. Strain gages were first attached at the longitudinal direction of the flanges and

162

webs, and all the strain gages were set up within the bond region as shown in Fig. 5. A total

163

of 10 strain gages was attached at the I-section of Specimens A and AS, five strain gages (S1

164

- S5) at the flanges and five (S6 – S10) at the web (Fig. 5a). For Specimens B, BS and BSS,

165

seven strain gages (S11 – S17) were attached at the flange and the other seven strain gages

AC C

159

8

ACCEPTED MANUSCRIPT 166

(S18 – S24) at the web (Fig. 5b).

167

Afterwards, the I-section attached with strain gages was placed into the timber formwork. In

169

order to fix the I-section at the centre of the formwork, two tiny holes were drilled into the

170

bottom of the formwork as well as the bottom of the I-section. All the holes were 10 mm in

171

depth. Afterwards, two 20 mm long thin steel wires were inserted into the holes of the I-

172

section and the formwork to fix the I-section in the formwork (Fig. 6a), and the steel wires

173

were removed from the I-section before the test. No concrete cover was left at the bottom of

174

the specimens. After the I-section was fixed in the formwork, the steel cage was placed into

175

the formwork. Two steel wires with the same length as the cross-section of the specimens

176

were used to ensure the accurate location of the steel cage, and these two steel wires were

177

fixed at the top stirrup in the transvers and longitudinal directions, respectively (Fig. 6b).

M AN U

SC

RI PT

168

TE D

178

Vibration was carried out when the concrete was cast. In order to keep the moisture, a wet

180

hessian was placed over the specimens and the specimens were watered every day. After

181

seven days, the specimens were demolded (Fig. 7) and then cured in moist conditions until

182

the test day.

AC C

183

EP

179

184

2.4 Test Setup and Instrumentation

185

The push-out test was conducted using the 5000 kN testing machine. As shown in Fig. 8, the

186

specimen was vertically placed onto the testing machine. One steel plate was horizontally

187

placed at the top of the I-section to uniformly distribute the load. Two steel blocks were

188

placed under the bottom of the specimen, and adequate space under the specimen was left for 9

ACCEPTED MANUSCRIPT the slip of the I-section. The displacement of the loaded end were measured using two Linear

190

Variable Differential Transformers (LVDTs), which were set up at the corners between the

191

loading plates and supporting steel plate. In order to measure the displacement of the

192

unloaded end, one LVDT was vertically placed under the specimen. The loaded end and

193

unloaded end in this study refer to the ends of the I-section. The load and displacement data

194

were recorded by an electronic data-logger connected to a computer every 2 seconds. After

195

all these setups were completed, the specimens were loaded by a displacement controlled

196

load with a rate of 0.1 mm/min. When the I-section was pushed out and the load did not

197

increase, the test was terminated.

M AN U

SC

RI PT

189

198

199

3. Experimental Results

200

The average bond stress ( ) in this study is defined by:

=

TE D

201

(Eq. 1)

where P is the applied load at the loaded end, L is the bond length of the I-section and C is

203

the perimeter of the I-section.

204

The slip (S) in this study is defined as the relative slip between the I-section and the concrete

205

at the loaded end. Before the I-section is pushed out, the displacement of the unloaded end is

206

the vertical extension of the specimen based on the experimental results, which is explained

207

in the following discussion section. Therefore, the slip (S) before pushing out the I-section is

208

calculated taking into account the vertical extension of the specimen as below:

209

210

AC C

EP

202

=∆ −∆

(Eq. 2a)

where ∆ is the displacement of the loaded end and ∆ is the displacement of the unloaded 10

ACCEPTED MANUSCRIPT end.

212

After the I-section is pushed out, the displacement of the loaded end (∆ ) and the unloaded

213

end (∆ ) keep same increment, and the displacement of the unloaded end (∆ ) does not

214

represent the vertical extension of the specimen any more. Therefore, the slip (S) is equal to

215

the displacement of the loaded end (∆ ): =∆

216

(Eq. 2b)

SC

217

RI PT

211

3.1 Failure Modes

219

Five specimens were cast and tested, and the failure modes are shown in Fig. 9. The I-

220

sections in the four specimens (Specimens A, AS, B, BS) were pushed out. The surface of the

221

I-section was intact after the I-section was pushed out, which indicates that the shear failure

222

occurred on the interface between the concrete and the I-section. Few cracks were observed

223

on the concrete of Specimens A and B, while the development of cracks was delayed in

224

Specimens AS and BS due to the application of the stirrups. The I-section in Specimen BSS

225

could not be pushed out, and this I-section failed due to the premature compressive failure at

226

the loaded end (Fig. 10).

TE D

EP

AC C

227

M AN U

218

228

3.2 Bond Stress-slip Curves

229

The bond stress-slip curves of four specimens (A, AS, B, BS) are shown in Fig. 11a, and the

230

typical curve is shown in Fig. 11b. In the first branch (O-A), the initial bond stress increased

231

slowly. Afterwards, an almost linear increase of the bond stress was revealed from Point A to

232

the ultimate bond stress ( ) at Point B with a larger slope. After Point B, the bond stress

11

ACCEPTED MANUSCRIPT curve experienced a slight decrease to Point C, and then increased again to Point D where the

234

largest bond stress is reached. It should be noted that the ultimate bond stress ( ) is obtained

235

at Point B rather than Point D in this study, and the explanation of this is given in the sections

236

below. A descending branch could be observed after Point D. Finally, the slip showed a stable

237

increase and the residual bond stress ( r) of the four specimens almost remained constant

238

within 0.3-0.4 MPa. The experimental results of all the specimens are summarized in Table 4,

239

including the ultimate bond stress ( ), the residual bond stress ( r) as well as the ultimate

240

slip (Ss) and the residual slip (Sr).

SC

RI PT

233

M AN U

241

The bond stress-slip curve of Specimen BSS is shown in Fig. 11c, which has a different

243

stress-slip response compared with the other four Specimens (A, AS, B, BS). After the

244

fluctuation in the initial stage, the curve increased linearly to the maximum bond stress where

245

the premature failure of the I-section occurred. The largest bond stress among the five

246

specimens was observed in Specimen BSS. For all the specimens, the slip occurred inside the

247

specimen thus causing limited experimental observation, so the in-depth interpretation of the

248

bond stress-slip curves could not be given based on the experimental observation only. More

249

explanation about these curves is presented accompanied with the analysis of the strain of the

250

I-section in the following parts.

EP

AC C

251

TE D

242

252

3.3 Strain distribution of the I-section

253

The strain distribution taken from the strain gages at the flange and web of the I-section is

254

shown in Fig. 12 and Fig. 13. Due to the similarity, the strain distribution of the I-section in

255

Specimen A (Fig. 12) is analysed as a typical strain distribution for Specimen A and

12

ACCEPTED MANUSCRIPT Specimen AS, and the strain distribution of Specimen B (Fig. 13) is the typical distribution

257

for the Specimen B and Specimen BS. The strain distribution of the I-section in Specimen

258

BSS is not discussed in this study due to the premature failure at the loaded end of the I-

259

section.

RI PT

256

260

Fig. 12a and Fig 13a show the strain-load curves of the flange. All the strain increased with

262

the increase of the load. In general, the strain near the loaded end showed a more significant

263

increase than the strain near the unloaded end, the reason for this may be that the applied load

264

had been counteracted by the bond stress near the loaded end. Therefore, the load had little

265

effect on the unloaded end thus causing a small strain. However, it was observed that the

266

strain of S2 (or S12) was larger than that of S1 (or S11) at the flange, the reason of which

267

may be the stress concentration at the position of S2 (or S12). Since when the specimens

268

were loaded, the compressive force at the loaded end may have caused the expansion of the

269

web of the I-section, thus further resulting in a stress concentration to occur at the flange, in

270

the position of strain gages S2 (or S12). Therefore, the strain of S2 (S12) was abnormally

271

higher than S1 (or S11) in all of the specimens. The strain distribution along the flange under

272

different load is shown in Fig. 12b and Fig 13b. The similar strain-load curves and the strain

273

distribution were observed at the web for Specimen A (Fig. 12c and Fig. 12d) and Specimen

274

B (Fig. 13c and Fig. 13d).

M AN U

TE D

EP

AC C

275

SC

261

276

3.4 Bond stress distribution of the I-section

277

The bond stress distribution along the I-section is also studied based on the strain difference

278

between two strain gages. As shown in Fig. 14, the local bond force between two adjacent

13

ACCEPTED MANUSCRIPT cross-sections could be calculated by: −

280

281

where

282

sectional area of the I-section;

283

and

=

(Eq. 3)

are the stress at two adjacent cross-sections of the I-section;

is the local bond stress between two adjacent cross-sections;

is the length between two adjacent cross-sections.

and

could be calculated by the corresponding elastic modulus ( ) and the

286

compressive strain (

and

), so the local bond stress ( ) is calculated by:

M AN U

The stress

SC

284

285

is the cross-

RI PT

279

(

)

287

=

288

It is noted that the elastic modulus ( ) and the compressive strain (

289

experimentally determined in this study, therefore, these parameters were easily influenced

290

by the technical problems or the testing machine, thus affecting the accuracy of the

291

calculation for the local bond stress ( ). As a result, the local bond stress ( ) determined by

292

Eq. 4 was employed only for the investigation of the bond stress distribution in this study.

293

The comparison between the local bond stress and the average bond stress determined by Eq.

294

1 is given in Fig. 14b.

) were

EP

TE D

and

AC C

295

(Eq. 4)

296

The bond stress distribution at the flange and the web for Specimen A and Specimen B is

297

shown in Fig. 15. It is clear that the bond stress distribution is not uniform along the flange or

298

web. In the initial stage of the test, the majority of the bond stress was distributed near the

299

loaded end, and it was small near the unloaded end. As the increase of the load, the bond

300

stress near the unloaded end was gradually increased until the failure of the specimen. 14

ACCEPTED MANUSCRIPT 4. Discussion and Analysis

302

4.1 Slip process

303

The interaction between the I-section and concrete is similar to steel-concrete composite

304

systems, so partial-interaction theory [30] could be referred to develop an in-depth

305

mechanical analysis for the interaction between two elements. Based on the analysis of the

306

strain and bond stress distribution as above-mentioned, the preliminary analysis about the slip

307

process of the I-section is presented in Fig. 16. Traditionally, the mechanics of stress transfer

308

by the bond between reinforcements (e.g., Steel/FRP bars) and concrete is mainly controlled

309

by three factors [10, 31]: (a) chemical adhesion provided by the concrete; (b) friction due to

310

the roughness of the reinforcements; (c) mechanical interlocking offered by the deformation

311

of the reinforcements. The three mechanisms are not isolated during the slip process, and

312

each mechanism has different performance in the different stages of test. The surface of the I-

313

section is traditionally smooth, therefore, mechanical interlocking is ignored in this

314

experimental study, and only chemical adhesion as well as friction is considered.

SC

M AN U

TE D

EP

315

RI PT

301

In this study, the interface between the I-section and the concrete was divided into two

317

regions, bond region and slip region. The interface in bond region is intact without slip, and

318

the bond force in bond region was dependent upon both chemical adhesion and friction. In

319

slip region, the chemical adhesion was degraded due to the slip at the interface, therefore,

320

only friction was contributed to the bond. The letters in Fig. 16 indicate the different stages of

321

the test, which have the same meaning as the letters in Fig. 11b.

AC C

316

322

323

When the I-section was loaded in the initial stage (O-A), the entire interface between the

15

ACCEPTED MANUSCRIPT concrete and the I-section was bond region which provided the bond force to counteract the

325

applied load. Afterwards, the different deformation between the concrete and the I-section at

326

the loaded end was increased with the increment of the load, thus causing a sudden relative

327

slip at the interface. Therefore, the slip region occurred at the loaded end of the specimen, and

328

it was also the reason why a fluctuation of the bond stress-slip curve at Point A (Fig. 11b) was

329

observed. When the bond stress reached the ultimate bond stress (Point B), the I-section

330

could not provide larger bond stress, therefore, the forces were unbalanced and the original

331

interface was totally broken. The slip region was extended to the entire interface (Loading

332

stage B-C) with a slight drop at the bond stress-slip curves, and I-section was pushed out at

333

the same time (Point B).

M AN U

SC

RI PT

324

334

The sudden slip at Point B caused a new interface which had a coarse surface. This new

336

interface could provide a larger friction to balance the applied load. Hence, the applied load

337

increased again from Point C to Point D. Although maximum stress was observed at Point D,

338

this stress could not reflect the bond behaviour of the original interface due to the damage of

339

the interface at Point B. With the increase of the slip, the interface was smoothed and the

340

friction was decreased. Finally, the load and the friction force reached the equilibrium state

341

again, and the I-section was gradually pushed out.

EP

AC C

342

TE D

335

343

4.2 Effect of stirrups, bond length and sand coating

344

Based on the analysis of the experimental results and the failure modes, it is clear that the

345

application of stirrups did not improve the bond strength as expected. As shown in Fig. 17,

346

the ultimate bond stress ( ) is decreased by using the stirrups, the possible reason for this

16

ACCEPTED MANUSCRIPT 347

might be that the application of stirrups affected the vibration of concrete during the casting,

348

thus causing a decrease of the bond strength at the interface of the I-section. The development

349

of cracks was reduced by the stirrups in Specimens AS, BS and BSS.

RI PT

350

The influence of the bond length was investigated by comparing the specimens with different

352

bond length (Fig. 18). For specimens with the same bond length, the same initial stiffness was

353

observed even though stirrups were used in one of the specimens (Fig. 17). For specimens

354

with different bond length, the ultimate bond stress ( ) of the specimen was improved by the

SC

351

longer bond length. For example, the ultimate bond stress ( ) was increased from 0.46 MPa

356

in Specimen A to 0.51 MPa in Specimen B due to the increase of the bond length.

M AN U

355

357

The I-section in Specimen BSS was coated with sand to investigate the influence of sand

359

coating on the bond behaviour. Nevertheless, the I-section crushed at the loaded end and

360

could not be pushed out. Although the accurate ultimate bond stress ( ) could not be

361

obtained in Specimen BSS, the bond stress in Specimen BSS had exceeded more than 1.3

362

MPa, which had been more than two times the ultimate bond stress of the I-sections without

363

sand coating. Therefore, the bond strength could be significantly improved by using sand

364

coating. More tests should be conducted to accurately estimate the influence of sand coating

365

on the bond stress.

AC C

EP

TE D

358

366

367

4.3 Theoretical Modelling

368

In this study, only the initial ascending stage (Stage O-B) of the bond stress-slip curves was

369

investigated. The main reasons for this include: (a) the I-section was pushed out at Point B 17

ACCEPTED MANUSCRIPT (Fig. 11b), therefore, Stage O-B can accurately reflect the bond behaviour of the original

371

interface of the specimens; (b) the randomness of the descending stage from B to C could not

372

be accurately predicted; (c) after the I-section was pushed out after Point B, the bond

373

behaviour of the interface is obviously different from the original interface.

RI PT

370

374

As the material properties of GFRP bars is similar to the I-section, the bond stress-slip

376

relationship of GFRP bars in concrete is investigated to understand the bond behaviour of the

377

I-section in concrete. Several bond stress-slip constitutive models for FRP bars have been

378

reported and summarized in Table 5. Among these models, the BPE model proposed by

379

Eligehausen et al. [32] is the classical model. To start with, this model was applied to the

380

bond of steel bars to concrete, and then successfully used for the bond behaviour of FRP bars

381

to concrete by Rossetti et al. [33]. The bond stress-slip curve in this model is divided into

382

different parts based on some representative parameters, such as the ultimate bond stress ( ),

383

the ultimate slip ( ) and the parameters !" , !# , $ and %.

TE D

M AN U

SC

375

384

Using curve fitting on the experimental results, the parameter α in this model was determined

386

as 2.5. Therefore, the bond stress-slip relationship in the curvilinear ascending branch is

387

proposed as:

AC C

388

EP

385

=

(

'( )

is the slip at the loaded end and

.+

(0 < ! ≤ ! )

Eq. 5

389

where

is the average bond stress. The experimental

390

results of the ultimate bond stress ( ) and ultimate slip ( ) were used in this calculation. The

391

comparison between the theoretical model and the experimental results are presented in Fig.

392

19. A good agreement is observed in the ascending branch for the four specimens, especially 18

ACCEPTED MANUSCRIPT 393

in Stage A-B.

394

The prediction of the bond stress in Eq. 5 requires the given ultimate bond stress ( ) and the

396

corresponding loaded end slip ( ). For GFRP bars, some empirical equations were proposed

397

to obtain these two parameters. Nevertheless, in this experimental study, the number of

398

specimens was not sufficient for an accurate empirical model to predict these two parameters.

399

Therefore, more studies should be conducted to estimate the ultimate bond stress ( ) and the

400

corresponding loaded end slip ( ).

M AN U

SC

RI PT

395

401

5. Conclusion

403

In this investigation, the experimental results and the bond stress-slip model on bond

404

behaviour of the GFRP I-section in concrete were reported. Five specimens with different

405

configurations were tested using push-out test. Based on the experimental results, the

406

following conclusions are drawn:

407

1. Push-out test is an effective method to investigate the bond behaviour of the GFRP

408

pultruded profiles in concrete.

409

2. The ultimate bond stress is improved by longer bond length and by using sand coating.

410

Although the I-section with sand-coating could not be pushed out, the larger bond stress of

411

this specimen had proved that sand coating is an effective measure to improve the bond

412

strength.

413

3. The ultimate bond stress was reduced when using stirrups to confine the concrete, the

414

reason may be because the stirrups affected the vibration of concrete, causing weak bond at

AC C

EP

TE D

402

19

ACCEPTED MANUSCRIPT the interface between the I-section and the concrete.

416

4. The bond stress distribution at the web and flange was investigated based on the strain of

417

the GFRP I-section, and two components showed similar bond stress distribution. The bond

418

stress performed a nonuniform distribution and is mostly distributed in the loaded end.

419

5. An empirical model was proposed to predict the curvilinear ascending branch of the bond

420

stress-slip curve. The results of the proposed model were in good agreement with the

421

experimental results. Nevertheless, this model is based on the ultimate bond stress ( ) and

422

the corresponding loaded end slip (

423

parameters needs to be established.

SC

RI PT

415

M AN U

), therefore, a method for predicting these two

424

As a preliminary experimental study, this study provides a significant reference for

426

investigating the bond behaviour of the I-section or other pultruded profiles with respect to

427

the test method (push-out test) and the design of the specimens. More variables should be

428

investigated such as the compressive strength and the type of concrete, as well as the shape of

429

the profiles, thus developing a more accurate bond stress-slip model. Moreover, the study in

430

future should focus on improving the bond strength by using sand coating or other

431

roughening treatment due to the small ultimate bond stress.

EP

AC C

432

TE D

425

433

Acknowledgement

434

The authors acknowledge the Senior Technical Officer Mr. Cameron Neilson and Mr. Ritchie

435

McLean for their contribution in the aspect of the technique. The first author also thanks the

436

China Scholarship Council and the University of Wollongong, Australia, for providing the

437

Ph.D. scholarship.

20

ACCEPTED MANUSCRIPT References

439

[1] F. Nunes, J.R. Correia, N. Silvestre, Structural behavior of hybrid FRP pultruded beams:

440

Experimental, numerical and analytical studies, Thin-Walled Structures 106 (2016) 201-217.

441

[2] L.C. Hollaway, A review of the present and future utilisation of FRP composites in the

442

civil infrastructure with reference to their important in-service properties, Construction and

443

Building Materials 24(12) (2010) 2419-2445.

444

[3] J.G. Teng, T. Yu, Y.L. Wong, S.L. Dong, Hybrid FRP-concrete-steel tubular columns:

445

Concept and behavior, Construction and Building Materials 21(4) (2007) 846-854.

446

[4] J.R. Correia, F.A. Branco, J.G. Ferreira, Flexural behaviour of GFRP–concrete hybrid

447

beams with interconnection slip, Compos. Struct. 77(1) (2007) 66-78.

448

[5] H. Nguyen, H. Mutsuyoshi, W. Zatar, Hybrid FRP-UHPFRC composite girders: Part 1 –

449

Experimental and numerical approach, Compos. Struct. 125 (2015) 631-652.

450

[6] C.A. Neagoe, L. Gil, M.A. Pérez, Experimental study of GFRP-concrete hybrid beams

451

with low degree of shear connection, Construction and Building Materials 101 (2015) 141-

452

151.

453

[7] R. El-Hacha, D. Chen, Behaviour of hybrid FRP–UHPC beams subjected to static flexural

454

loading, Composites Part B: Engineering 43(2) (2012) 582-593.

455

[8] M.N.S. Hadi, J. Youssef, Experimental Investigation of GFRP-Reinforced and GFRP-

456

Encased Square Concrete Specimens under Axial and Eccentric Load, and Four-Point

457

Bending Test, Journal of Composites for Construction (2016) 04016020.

458

[9] A. Abouzied, R. Masmoudi, Structural performance of new fully and partially concrete-

459

filled rectangular FRP-tube beams, Construction and Building Materials 101, Part 1 (2015)

460

652-660.

461

[10] E. Cosenza, G. Manfredi, R. Realfonzo, Behavior and modeling of bond of FRP rebars

462

to concrete, Journal of Composites for Construction 1(2) (1997) 40-51.

AC C

EP

TE D

M AN U

SC

RI PT

438

21

ACCEPTED MANUSCRIPT [11] H. Nordin, B. Taljsten, Testing of hybrid FRP composite beams in bending, Composites

464

Part B: Engineering 35(1) (2004) 27-33.

465

[12] T. Keller, H. Gürtler, Composite action and adhesive bond between fiber-reinforced

466

polymer bridge decks and main girders, Journal of Composites for Construction 9(4) (2005)

467

360-368.

468

[13] W.H. Kwan, M. Ramli, Indicative performance of fiber reinforced polymer (FRP)

469

encased beam in flexure, Construction and Building Materials 48 (2013) 780-788.

470

[14] S.T. Smith, J.G. Teng, FRP-strengthened RC beams. I: Review of debonding strength

471

models, Engineering Structures 24(4) (2002) 385-395.

472

[15] J.F. Chen, J.G. Teng, Anchorage strength models for FRP and steel plates bonded to

473

concrete, Journal of Structural Engineering 127(7) (2001) 784-791.

474

[16] H.C. Biscaia, C. Chastre, M.A.G. Silva, Bond-slip model for FRP-to-concrete bonded

475

joints under external compression, Composites Part B: Engineering 80 (2015) 246-259.

476

[17] J. Pan, Y.F. Wu, Analytical modeling of bond behavior between FRP plate and concrete,

477

Composites Part B: Engineering 61 (2014) 17-25.

478

[18] Z. Achillides, K. Pilakoutas, Bond behavior of fiber reinforced polymer bars under direct

479

pullout conditions, Journal of Composites for Construction 8(2) (2004) 173-181.

480

[19] D.Y. Yoo, H. O. Shin, J. M. Yang, Y. S. Yoon, Material and bond properties of ultra high

481

performance fiber reinforced concrete with micro steel fibers, Composites Part B:

482

Engineering 58 (2014) 122-133.

483

[20] I. Vilanova, M. Baena, L. Torres, C. Barris, Experimental study of bond-slip of GFRP

484

bars in concrete under sustained loads, Composites Part B: Engineering 74 (2015) 42-52.

485

[21] M. Baena, L. Torres, A. Turon, C. Barris, Experimental study of bond behaviour between

486

concrete and FRP bars using a pull-out test, Composites Part B: Engineering 40(8) (2009)

487

784-797.

AC C

EP

TE D

M AN U

SC

RI PT

463

22

ACCEPTED MANUSCRIPT [22] C. Xu, H. Chengkui, J. Decheng, S. Yuancheng, Push-out test of pre-stressing concrete

489

filled circular steel tube columns by means of expansive cement, Construction and Building

490

Materials 23(1) (2009) 491-497.

491

[23] A. Bouchair, J. Bujnak, P. Duratna, A. Lachal, Modeling of the Steel-Concrete Push-Out

492

Test, Procedia Engineering 40 (2012) 102-107.

493

[24] M.N.S. Hadi, J.S. Yuan, Experimental investigation of composite beams reinforced with

494

GFRP I-beam and steel bars, Construction and Building Materials 144 (2017) 462-474.

495

[25] ACI. 440.3R, Guide test methods for Fiber-Reinforced Polyners (FRPs) for reinforcing

496

or strengthening concrete structure. American concrete Institute (ACI). 2004, 2004.

497

[26] AS 1391-2007. Metallic materials-Tensile testing at ambient temperature, Standards

498

Australia Limited, NSW; 2007.

499

[27] Treadwell, “Custom Fibreglass Reinforced Plastic (FRP) & Underfoot Products.” 58

500

Deeds Rd, North Plympton, SA Australia.< http://www.treadwellgroup.com.au/> (April. 02.

501

2016). 2016).

502

[28] ISO 527-1. Plastics: Determination of tensile properties. European Committee for

503

Standardisation, Brussels, Belgium; 1996.

504

[29] ASTM D695. Standard test method for compressive properties of rigid plastics. United

505

States: ASTM international. 2002.

506

[30] E. Martinelli, Q.H. Nguyen, M. Hjiaj, Dimensionless formulation and comparative study

507

of analytical models for composite beams in partial interaction, Journal of Constructional

508

Steel Research 75 (2012) 21-31.

509

[31] M.N.S. Hadi, Bond of high strength concrete with high strength reinforcing steel, The

510

Open Civil Engineering Journal 2 (2008) 143-147.

511

[32] R. Eligehausen, E.P. Popov, V.V. Bertero, Local bond stress-slip relationships of

512

deformed bars under generalized excitations, Earthquake Engineering Research Center,

AC C

EP

TE D

M AN U

SC

RI PT

488

23

ACCEPTED MANUSCRIPT University of California; 1983 (1982).

514

[33] V. Rossetti, D. Galeota, M. Giammatteo, Local bond stress-slip relationships of glass

515

fibre reinforced plastic bars embedded in concrete, Materials & Structures 28(6) (1995) 340.

516

[34] L.J. Malvar, Bond stress-slip characteristics of FRP rebars, California: Naval Facilities

517

Engineering Service Center (1994).

518

[35] B. Zhang, B. Benmokrane, Pullout bond properties of fiber-reinforced polymer tendons

519

to grout, Journal of Materials in Civil Engineering 14(5) (2002) 399-408.

520

[36] E. Cosenza, G. Manfredi, R. Realfonzo, Analytical modelling of bond between FRP

521

reinforcing bars and concrete., Proc., 2nd Int. RILEM Symo. (FRPRCS-2). (1995).

522

[37] B. Tighiouart, B. Benmokrane, D. Gao, Investigation of bond in concrete member with

523

fibre reinforced polymer (FRP) bars, Construction and Building Materials 12(8) (1998) 453-

524

462.

M AN U

SC

RI PT

513

AC C

EP

TE D

525

24

526

List of Tables

527

Table 1. Configuration of Specimens

528

Table 2. Composition of concrete

529

Table 3. Material Properties of GFRP I-section

530

Table 4. Experimental results

531

Table 5. Existing bond–slip models for FRP bars

AC C

EP

TE D

M AN U

SC

532

RI PT

ACCEPTED MANUSCRIPT

25

ACCEPTED MANUSCRIPT List of Figures

534

Fig. 1. Schematic diagram of specimens (mm) (a) Specimen A, (b) Specimen AS, (c)

535

Specimen B, (d) Specimen BS, (e) Specimen BSS

536

Fig. 2. Cross-section of specimens (mm) (a) Cross-section of Specimens A and B (Section A-

537

A), (b) Cross-section of Specimens AS, BS and BSS (Section B-B)

538

Fig. 3. Explanation of debonding region (a) Specimen for the pull-out of FRP bars, (b)

539

Specimen for push-out of I-section

540

Fig. 4. GFRP I-section

541

Fig. 5. Layout of the strain gages (mm) (a) Strain gages at Specimen A and AS, (b) Strain

542

gages at Specimens B, BS, BSS

543

Fig. 6. Layout of the I-section and steel cage (a) Fixing I-section (b) Fixing steel cage

544

Fig. 7. Formwork and specimens

545

Fig. 8. Test setup (a) Schematic diagram of push-out test, (b) Setup of test

546

Fig. 9. Failure modes of specimens (a) Specimen A, (b) Specimen AS, (c) Specimen B, (d)

547

Specimen BS, (e) Specimen BSS

548

Fig. 10. Compression failure of I-section

549

Fig. 11. Bond stress-slip curves at loaded end (a) Bond stress-slip curves of Specimens A, AS,

550

B, BS, (b) Typical bond stress-slip curve, (c) Bond stress-slip curve of Specimen BSS

551

Fig. 12. Analysis of strain (Specimen A) (a) Strain-load curves of flange, (b) Strain

552

distribution along flange, (c) Strain-load curves of web, (d) Strain distribution along web

553

Fig. 13. Analysis of strain (Specimen B) (a) Strain-load curves of flange, (b) Strain

554

distribution along flange, (c) Strain-load curves of web, (d) Strain distribution along web

555

Fig. 14. Local bond stress (a) Calculation of local bond stress, (b) Comparison between local

556

bond stress and average bond stress

557

Fig. 15. Typical bond stress distribution (a) Bond stress distribution at flange (Specimen A)

AC C

EP

TE D

M AN U

SC

RI PT

533

26

ACCEPTED MANUSCRIPT (b) Bond stress distribution of web (Specimen A), (c) Bond stress distribution at flange

559

(Specimen B), (d) Bond stress distribution of web (Specimen B)

560

Fig. 16. Slip process and bond stress distribution of I-section

561

Fig. 17. Typical effect of stirrups on bond stress

562

Fig. 18. Typical effect of bond length on bond stress

563

Fig. 19. Comparison of bond stress-slip curves (a) Specimen A, (b) Specimen AS, (c)

564

Specimen B, (d) Specimen BS

AC C

EP

TE D

M AN U

SC

RI PT

558

27

ACCEPTED MANUSCRIPT

Table 1. Configuration of Specimens Bond length2 (mm)

Height of debonding region (mm)

GFRP I-section (mm)

Stirrups (mm)

Longitudinal bars (mm)

Surface of the I-section

200×100×10

-

-

Smooth

200×100×10

Steel R10 @ 60

Steel 4 R10

Smooth

200×100×10

-

-

Smooth

50

200×100×10

Steel R10 @ 60

Steel 4 R10

Smooth

50

200×100×10

Steel R10 @ 60

Steel 4 R10

Sand coated

A

350×200

400

50

300

50

AS

350×200

400

50

300

50

B

350×200

600

100

450

50

BS

350×200

600

100

450

BSS

350×200

600

100

450

1

EP

TE D

Free end is the part of the I-section out of the concrete. Bond length = height of the concrete – height of debonding region (See Fig. 1 for the details)

2

SC

Group A

M AN U

Specimen

RI PT

Height of free end1 (mm)

Group

Group B

567 568 569

Total height (mm)

CrossSection （mm） ）

AC C

565 566

28

ACCEPTED MANUSCRIPT Table 2. Composition of concrete Constituent (kg/m3)

Values

Cement

285

Fly ash

100

Coarse aggregate

1135

Coarse sand

543

Fine sand

217

Water

170

RI PT

570 571

AC C

EP

TE D

M AN U

SC

572 573 574 575

29

ACCEPTED MANUSCRIPT Table 3. Material Properties of GFRP I-section Dimensions of coupon (mm)

Property

Averages and Sample Standard Deviations

Flange

25 ×250

Tensile strength (MPa)

381.5 ± 8.1

Tensile elastic modulus (GPa)

38.5 ± 4.2

12.7 ×37.1

Web

RI PT

Position

Compressive strength (MPa)

214.2± 17.4

Compressive elastic modulus (GPa)

26.9 ± 1.5

Tensile strength (MPa)

353 ± 30

25 ×250

Tensile elastic modulus (GPa)

32.88 ± 1.8

Compressive strength (MPa)

233.8 ± 18.4

M AN U

12.7 ×37.1

SC

576 577

Compressive elastic modulus (GPa)

EP

TE D

Note: Tensile properties were determined based on ISO 527 (1997); Compressive properties were determined based on ASTM D695 (2002).

AC C

578 579 580 581

30.2 ± 8.5

30

ACCEPTED MANUSCRIPT

Table 4. Experimental Results Ultimate bond load (Ps)1 (kN)

Ultimate load (Pu)2 (kN)

Ultimate bond stress (/0 ) (MPa)

Group A

A

109.8

116.6

0.46

AS

72.1

99.7

0.30

B

184.6

193.5

0.51

BS

122.2

138.1

0.34

BSS

-

474.2

1

Ultimate bond load is reached at Point B as shown in Fig. 11b. Ultimate load is reached at Point D as shown in Fig. 11b.

TE D

2

-

EP

584 585 586 587 588 589 590 591 592 593 594 595

M AN U

Group B

31

Residual bond stress (/1 ) (MPa)

Ultimate slip (Ss) (mm)

Residual slip (Sr) (mm)

0.32

1.09

4.35

0.29

1.04

4.45

0.36

1.61

5.02

0.25

1.40

4.74

-

-

RI PT

Specimen

SC

Group

AC C

582 583

ACCEPTED MANUSCRIPT

Table 5. Existing bond–slip models for FRP bars

Eligehausen et al. model (BPE model) [32] a

a

BPE modified model [10]

Zhang et al. model [35]a

597

=

=

+ 9 :1 − exp (−

=

! ( )E !

=

! ( )E !

K1 − (

! − 1) O !

CMR model [36]a

=

! E K1 − exp (− )O !

Tighiouart et al. [37] a

=

P1 − exp (4!)RS.+

a

The values of

>?

"8

)@; ! = A + B#

=

−F

− #

=

! − !" I !# − !"

"G H

=%

! K1 − L M − 1NO ! #

=

RI PT

"8

! ! 1 + (2 − 2) '! ) + 4 ' ) !

=%

TE D

7

! ! 2 '! ) + (4 − 1) '! )

shapes of curves

M AN U

=

Malvar model [34]

Descending branch

SC

Ascending branch

−(

−

EP

Model

AC C

596

" )(

! − !" ) !# − !"

as above

parameters A,B,C,D,E,F,G = empirical constants determined for each bar type

B# = confining axisymmetric radial pressure CD = tensile concrete strength

$, % = curve-fitting parameter

$, % , p = curve-fitting parameter

-

$= curve-fitting parameter

-

-

as above

, ! , !" , # , !# was calibrated on the basis of the experimental results. 32

-

ACCEPTED MANUSCRIPT

200

200

I-section

Bond region

A

B

SC

50

(b) Specimen AS

(c) Specimen B

100

I-section Free end

Sand coating R10 @ 60 4 R10

4 R10

Bond region

Debonding region

B

450

TE D

450

100 350

200

R10 @ 60

Bond region

Debonding region

B

B

50

B

50

A

Free end

EP

Debonding region

AC C

100

Bond region

50

350

I-section

Free end

A

B Debonding region

200

I-section

450

M AN U

350

200

4 R10

Bond region

A Debonding region

(a) Specimen A

R10 @ 60

Free end

RI PT

50 300

Free end

50

300

50

I-section

350

(d) Specimen BS

350

(e) Specimen BSS

Fig. 1. Schematic diagram of specimens (mm) (See Fig. 2 for sections A-A and B-B) 33

ACCEPTED MANUSCRIPT

I-section

SC

350

RI PT

R10 @ 60

200

4 R10

M AN U

(a) Cross-section of Specimens A and B (Section A-A)

TE D

200

I-section

350

(b) Cross-section of Specimens AS, BS and BSS (Section B-B)

AC C

EP

Fig. 2. Cross-section of specimens (mm) (See Fig. 1 for elevation views)

34

RI PT

ACCEPTED MANUSCRIPT

Pull

Push

M AN U

SC

FRP bar

Debonding tube

TE D

(a) Specimen for the pull-out of FRP bars

I-section

Debonding region

(b) Specimen for push-out of I-section

AC C

EP

Fig. 3. Explanation of debonding region

35

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Fig. 4. GFRP I-section

36

ACCEPTED MANUSCRIPT

S10

S5

50

Debond region

Bond region

Debond region

TE D

Unloaded end

(a) Strain gages at Specimens A and AS

S21

S14

S22

S15

S23

S16

S24

S17

40 20

S13

80

60

S20

80

S4

S12

80

S9

S11

S19

SC

S3

450

S8

60

S2

80

S1

S7

80

Bond region

S6

40 20

Free end

M AN U

50

300

50

Loaded end

S18

80

Free end

RI PT

100

Loaded end

Unloaded end

(b) Strain gages at Specimens B, BS, BSS

AC C

EP

Fig. 5. Layout of the strain gages (mm)

37

RI PT

ACCEPTED MANUSCRIPT

GFRP I-section

SC

Tiny Hole

M AN U

Formwork

Steel Wire

AC C

EP

TE D

(a) Fixing I-section

Steel wires

(b) Fixing steel cage

Fig. 6. Layout of I-section and steel cage

38

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Fig. 7. Formwork and specimens

39

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

LVDT

AC C

EP

TE D

(a) Schematic diagram of push-out test

LVDT (b) Setup of test Fig. 8. Test setup

40

(b) Specimen AS

(c) Specimen B

EP

TE D

M AN U

(a) Specimen A

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

(d) Specimen BS

(e) Specimen BSS

Fig. 9. Failure modes of specimens

41

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

Fig. 10. Compression failure of I-section

42

ACCEPTED MANUSCRIPT Specimen A Specimen AS Specimen B Specimen BS

0.4

0.2

RI PT

Bond stress (MPa)

0.6

0 2

4

6 8 10 12 Slip (mm) (a) Bond stress-slip curves of Specimens A, AS, B, BS

TE D

M AN U

SC

0

(b) Typical bond stress-slip curve

Specimen BSS

1.05

AC C

Bond stress (MPa)

EP

1.4

0.7

0.35

0 0

3

6 Slip (mm)

9

12

(c) Bond stress-slip curve of Specimen BSS

Fig. 11. Bond stress-slip curves at loaded end 43

−150

−150

S5 S4 S3 S2 S1

−450 −600 0

20

40

100kN 75kN 50kN 25kN

−450 −600 60 80 Load (kN)

100

(a) Strain-load curves of flange

0

0

120

50

100 150 200 250 Bond length (mm)

300

(b) Strain distribution along flange

0

−600 0

20

40

TE D

−450

EP

S10 S9 S8 S7 S6

−300

Strain (10 -6 )

−150

60 80 Load (kN)

AC C

Strain (10 -6 )

−300

SC

−300

RI PT

0

Strain (10 -6 )

0

M AN U

Strain (10 -6 )

ACCEPTED MANUSCRIPT

100

(c) Strain-load curves of web

120

−150 −300

100kN 75kN 50kN 25kN

−450 −600 0

50

100 150 200 250 Bond length (mm)

300

(d) Strain distribution along web

Fig. 12. Analysis of strain (Specimen A)

44

ACCEPTED MANUSCRIPT 0

S11 S12 S13 S14 S15 S16 S17

−600 −800 0

50

−400 −600 −800

100 Load (kN)

150

0

200

90

0

180 270 360 Bond length (mm)

450

(b) Strain distribution along flange

M AN U

(a) Strain-load curves of flange

0

−600 −800 0

50

100 Load (kN)

150

EP

−400

Strain (10 -6 )

S18 S19 S20 S21 S22 S23 S24

TE D

−200

(c) Strain-load curves of web

200

−200 −400 −600

186 kN 150 KN 100 kN 75 kN 50 kN 25 kN

−800 0

90

180 270 360 Bond length (mm)

450

(d) Strain distribution along web

AC C

Strain (10 -6 )

186kN 150kN 100kN 75kN 50kN 25kN

SC

−400

−200

RI PT

−200

Strain (10 -6 )

Strain (10 -6 )

0

Fig. 13. Analysis of strain (Specimen B)

45

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

(a) Calculation of local bond stress

(b) Comparison between local bond stress and average bond stress Fig. 14. Local bond stress

46

ACCEPTED MANUSCRIPT

0.2

0

0.4

0.2

0

0

60

120 180 240 Bond length (mm)

300

0

(a) Bond stress distribution at flange (Specimen A)

60

120 180 240 Bond length (mm)

300

(b) Bond stress distribution at web (Specimen A)

TE D

186 kN 150 kN 100 kN 75 kN 50 kN 25 kN

0.4

0 0

AC C

EP

0.2

90

180 270 360 Bond length (mm)

Bond stress (MPa)

0.6

0.6 Bond stress (MPa)

100kN 75kN 50kN 25kN

SC

0.4

Bond stress (MPa)

100kN 75kN 50kN 25kN

RI PT

0.6

M AN U

Bond stress (MPa)

0.6

0.4

186 KN 150 kN 100 kN 75 kN 50 kN 25 kN

0.2

0 450

(c) Bond stress distribution at flange (Specimen B)

0

90

180 270 360 Bond length (mm)

450

(d) Bond stress distribution at web (Specimen B)

Fig. 15. Typical bond stress distribution

47

D C

Slip region

A

O

Load

Slip Bond stress

Concrete

TE D

I-section

Bond region

Loading stage A

Loading stage B

Loading stage C

EP

Loading stage O

SC

B

M AN U

Bond stress

RI PT

ACCEPTED MANUSCRIPT

Fig. 16. Slip process and bond stress distribution of I-section

AC C

(See Fig. 11 for explanation of loading stages)

48

SC

RI PT

ACCEPTED MANUSCRIPT

Specimen B Specimen BS

M AN U

Bond stress (MPa)

0.6

0.4

0.2

TE D

0

0

2

4

6 8 Slip (mm)

10

12

AC C

EP

Fig. 17. Typical effect of stirrups on bond stress

49

RI PT

ACCEPTED MANUSCRIPT

0.6

SC

0.4

0.2

0 2

4

6 8 Slip (mm)

10

12

TE D

0

M AN U

Bond stress (MPa)

Specimen A Specimen B

AC C

EP

Fig. 18. Typical effect of bond length on bond stress

50

ACCEPTED MANUSCRIPT

0.4

0.3 0.2

1

2 Slip (mm)

(a) Specimen A

3

0

1

2

3

Slip (mm)

(b) Specimen AS

0.4

0.4

TE D

0.3 0.2 0.1 0 0

1

EP

Theoretical result B Test result B 2

3

0.3 0.2 0.1

Theoretical result BS Test result BS

0 0

1

Slip (mm)

AC C

Bond stress (MPa)

0.5

0

M AN U

0

0.1

Theoretical result AS Test result AS

Theoretical result A Test result A

0

0.2

SC

0.1

0.3

RI PT

Bond stress (MPa)

0.4

Bond stress (MPa)

Bond stress (MPa)

0.5

(c) Specimen B

2

3

Slip (mm) (d) Specimen BS

Fig. 19. Comparison of bond stress-slip curves

51