1 Full-scale Testing of Stiffened Extended Shear Tab

2 Connections under Combined Axial and Shear Forces

3 Mohammad Motallebi1, Dimitrios G. Lignos2, Colin A. Rogers3
4
5 1
Graduate Research Assistant, Department of Civil Engineering and Applied Mechanics, McGill University,
6 Montreal, QC. Email: [email protected]
7
8 2
Dimitrios G. Lignos, Associate Professor, School of Architecture, Civil and Environmental Engineering, Swiss
9 Federal Institute of Technology, Lausanne (EPFL), Lausanne, Switzerland, Email: [email protected]
10
11 3
Corresponding author
12 Colin A. Rogers, Associate Professor, Department of Civil Engineering and Applied Mechanics, McGill University,
13 Montreal, QC. Email: [email protected]
14 817 Sherbrooke Street West
15 Montreal QC, Canada, H3A 0C3
16 Tel. 514 398-6449
17 Fax. 514 398-7361
18

1
19 ABSTRACT

20 Owing to the lack of a comprehensive published procedure for the design of stiffened extended

21 shear tabs, practicing engineers usually follow design guides for unstiffened shear tabs. The results

22 of recent laboratory experiments and numerical analyses have demonstrated that improvements to

23 this design approach are warranted. Furthermore, design methods for this connection type under

24 loading scenarios including combined axial and shear forces are not well established. To address

25 these shortcomings, full-scale laboratory tests were carried out on the double-sided configuration

26 of stiffened extended beam-to-girder shear tabs with full depth shear plates. These experiments

27 were complemented by a thoroughly validated finite element (FE) study. Based on the results of

28 these experiments and FE simulations, the connection failure modes were characterized and the

29 axial force along with the other main parameters that affect the connection behaviour were further

30 examined. The current design practice for the double-sided configuration of the full-depth

31 extended beam-to-girder shear tab was also evaluated.

32

33 Keywords: extended shear tab, double-sided configuration, gross section yielding, plate out-of-

34 plane deformation, net section fracture

2
35 1 Introduction

36 Shear connections transfer end shear reactions of simply supported beams to supporting

37 columns or girders without transmitting significant flexural moment, i.e. less than 20% of the

38 nominal plastic moment resistance of the supported beam [1]. Furthermore, these connections must

39 have sufficient ductility to sustain rotational demands from a beam’s ends. Existing design

40 procedures [2] for shear connections consider only gravity-induced force shear force. However, a

41 simple shear connection may be subjected to an axial force due to wind and/or earthquake while it

42 is resisting gravity-induced shear force. Furthermore, extreme loading scenarios such as the loss

43 of a column develop a significant axial tension in these connections. As a conclusion, contrary to

44 traditional perspectives on simple shear connections, there exists a need for their design under

45 combined axial and shear forces. Despite this need, there is little guidance in the literature for the

46 design of shear connections under combined axial and shear forces [3, 4].

47 A shear tab is a common type of simple shear connection used in steel construction (Fig. 1).

48 The 15th edition of the AISC steel construction manual [2] considers 89 mm (3.5 in.) as the limit

49 to classify this connection into conventional and extended types based on the distance between the

50 support face and the vertical bolt line closest to the support. Referring to Fig. 1, this is noted as the

51 a distance. Extended shear tab connections are considered as a practical and economically

52 attractive solution to join a simply supported beam to a column or girder web. The long plate

53 moves the supported beam clear of the support; as such, there is no need for coping of the beam’s

54 flange(s). A common connection configuration is the extended shear tab with a full depth shear

55 plate. In this “stiffened” configuration, the shear plate is shop-welded to the girder web and both

56 flanges (Fig. 1a). In the case of a beam-to-column web connection (Figs. 1b and 1c), the shear

57 plate is welded to the column web and to two stabilizer plates, which in turn are welded to the
3
58 flanges of the column. Although the stiffened extended shear tab connection is common in steel

59 construction in North America, only a few recommendations [3, 4] have been published for its

60 design. The current AISC design approach for extended shear tabs [2] was originally developed

61 for unstiffened extended shear tabs (Fig. 1d). In this configuration, only the vertical edge of the

62 plate is welded to the support; its horizontal edges are laterally unrestrained.

a b c d

63 Fig. 1. Single-sided extended shear tab configurations: (a) stiffened beam-to-girder with full-depth shear plate
64 (hw definition based on CSA-S16 [5]), (b) stiffened beam-to-column, (c) stiffened beam-to-column with continuity
65 plates, (d) unstiffened beam-to-column

66 Prior studies demonstrated that plate buckling is the governing failure mode for stiffened full-

67 depth configurations of either beam-to-girder [6-10] or beam-to-column shear tab connections [11,

68 12]. The focus of these research programs was limited to the single-sided configuration of stiffened

69 extended shear tabs. Regarding the behaviour of stiffened extended shear tabs under combined

70 axial and shear forces, Thomas et al. [12] focused on the single-sided configuration, similar to that

71 shown in Fig. 1b. Nevertheless, this configuration would need to be modified if continuity plates

72 were incorporated into a fully restrained beam-to-column connection (Fig. 1c). Thomas et al. [12]

73 determined the shear plate’s out-of-plane deformation as the critical failure mode of all ten tests,

74 while the plate completely yielded prior to the connection failure. The range of the applied axial

75 force was limited because the single-sided shear tab experiences small axial force in real world

76 applications due to low stiffness of the girder’s weak-axis. In comparison to the single-sided shear

4
77 tab, the double-sided configuration may be subjected to much higher axial force because the this

78 force transfers through the girder.

79 This paper presents the results of a coordinated experimental-numerical study aiming to

80 deepen our understanding of the behaviour of the stiffened extended beam-to-girder shear tab

81 under combined axial and shear forces. The testing of full-scale connection specimens allowed for

82 an improved comprehension of the inelastic behaviour of the stiffened extended shear tab, while

83 the test results were relied on to validate the complementary detailed finite element (FE) models.

84 Based on the experimental and numerical results, probable failure modes and their influential

85 parameters were determined. The current design practice was evaluated and recommendations are

86 proposed to improve this design approach for double-sided stiffened extended beam-to-girder

87 shear tab connections with full depth shear plates.

88 2 Full-scale laboratory testing

89 Two full-scale connection specimens representing the current design practice in North

90 America were tested in the Jamieson Structures Laboratory at McGill University to examine the

91 behaviour of stiffened extended shear tabs under combined axial and shear forces. These

92 experiments were part of an extensive laboratory testing program [7, 8, 13-18] aiming toward

93 improving the current design and detailing provisions for shear tab connections. The test

94 specimens were chosen to represent the double-sided configuration of a beam-to-girder extended

95 shear tab with full-depth shear plates. The rationale behind choosing the double-sided

96 configuration was its ability to provide a rigid support, allowing the connection to experience a

97 wide range of axial and shear forces. Therefore, the shear-axial force interaction curve could be

98 developed for shear tab’s failure modes.

5
 99 2.1 Description of test specimens

100 The specimens varied with respect to the number of horizontal bolt lines and the dimensions

101 of the shear plate including its depth, length, and thickness (Fig. 2). The specimen ID, e.g. BG3-

102 2-13-F-200C, identifies the following: BG stands for beam-to-girder configuration, 3 represents

103 the number of horizontal bolt lines, 2 shows the number of vertical bolt lines, 13 demonstrates the

104 thickness of shear plate (mm), F indicates that a full-depth shear plate was used, and 200C

105 represents the magnitude (200 kN) and direction (Compression) of the applied axial force.

a) b)

106
107 Fig. 2. Double-sided configuration of test specimens: (a) BG3-2-13-F-200C, (b) BG6-2-19-F-500C

108 In both specimens, the slenderness ratio (bf/2tpl) of the shear plate satisfied the CSA-S16

109 compactness requirement [5] for plate girder stiffeners ( 200 / Fy  10.7 ). However, this is not a

110 requirement for the existing AISC design method because local buckling is not a concern for an

111 unstiffened extended shear tab. Prior studies [7-10] demonstrated the influence of the shear plate

112 compactness on the ductile response of single-sided shear tab connections.

113 Considering the symmetry of a double-sided shear tab along the girder axis, the laboratory

114 specimens consisted of only half of the connection (Fig. 3), i.e. a single beam connected to a

115 simulated girder. Prior research indicated that the behaviour of single- and double-sided shear tabs

116 is different due to the distortion of the girder web [9]. To simulate one side of the girder two steel
6
117 plates were joined to the column flange using a complete joint penetration (CJP) weld. The plate

118 dimensions were chosen to be representative of the half width of the girder flange. The shear plate

119 was connected to the girder flanges, as well as to the column flange, through a fillet weld, which

120 was detailed based on the AISC’s requirements [2] for the weld of the extended shear tab. The in-

121 plane displacement of the column was restricted using two back braces, which were attached to

122 the strong-floor of the laboratory as described in Section 2.2. These braces, in addition to the

123 strong-axis stiffness of the column, provided a rigid support to the connection being tested and

124 prevented all possible failure modes of the simulated girder.

125 Furthermore, the bottom flange of both beams was coped to increase the beam-plate gap, and

126 consequently delay beam binding, i.e. contact between the beam’s bottom flange and the edge of

127 the shear tab. Preliminary FE analyses suggested that these short copes would not affect the

128 connection global response, although the out-of-plane deformation of the beam and plate might

129 increase slightly.

a) b)

130
131 Fig. 3. Details of test specimens: (a) BG3-2-13-F-200C, (b) BG6-2-19-F-500C

132 The beams and girders were fabricated from ASTM A992 Grade 50 (Fy = 345 MPa) steel [19]

133 while the shear plates were made of ASTM A572 Grade 50 (Fy = 345 MPa) steel [20]. To attach

134 the shear tab to the fabricated supporting girder, an E71T electrode (Xu = 490 MPa) [21] was used
7
135 in a flux-cored arc welding process with additional shielding gas (CO2) to provide a fillet weld on

136 both sides of the plate. Each beam was snug tightened to the shear tab using ASTM F3125 Grade

137 A490 bolts [21] in standard size holes, 2mm (1/16”) larger in diameter than the bolts. Figure 4

138 shows these two specimens prior to testing.

a b

139 Fig. 4. Specimens: (a) BG3-2-13-F-200C, (b) BG6-2-19-F-500C

140 Table 1 shows the nominal and expected strength of the connection components along with

141 their measured material properties obtained by ancillary tests in the form of steel and all-weld

142 tensile coupon tests. The test coupons of the shear plates and beams (including web and flanges)

143 were extracted from the same batch of full-scale test components. For each beam, four coupons

144 were cut from the flanges while three were cut from the web. Six coupons were taken from each

145 plate thickness, three along and three perpendicular to the grain direction.

146 Table 1. Material properties of connection components

Nominal Probable 1 Measured
Connection components Fy Fu Fy Fu Fy Fu
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
W310×74 Flange 345 448 379 493 374 490
(W12×50) Web 345 448 379 493 379 495
W610×415 Flange 345 448 379 493 372 513
(W24×279) Web 345 448 379 493 377 507
13mm (1/2”) plates 345 448 379 538 432 508
19mm (3/4”) plates 345 448 379 538 377 527
E71T electrode 400 490 -- -- 548 620
A490 bolts 896 1034 -- -- -- --
1
147 RyFy and RTFu; for steel plates 1.1 Fy and 1.2 Fu while 1.1 Fy and 1.1 Fu for hot-rolled structural shapes [25]

8
148 All steel coupons were tested based on ASTM A370 [22], while two all-weld coupons were

149 tested based on AWS A5.20 [23]. All-weld coupons were extracted from a groove welded

150 assembly of two plates, fabricated from the same weld electrodes used for the shear tab specimens

151 [23, 24]. As neither bolt fracture, nor bolt deformation was observed in these tests, bolt shear tests

152 were not conducted.

153 The connection specimens were designed based on the current AISC procedure [2] for

154 unstiffened extended shear tabs. This method contains an assumption that the inflection point forms

155 at the support face; the geometric eccentricity (e), distance between the support face and the centre

156 of the bolt group, was chosen as the bolt group eccentricity. As such, the bolt group was designed

157 for the beam end shear reaction (R) and its eccentric bending moment (R × e). The weld line was

158 designed to concentrically resist the beam end reaction (R). To ensure sufficient ductility of the shear

159 tab connection, the weld throat and the plate thickness were detailed such that yielding can develop

160 over the full height of the shear plate’s extended portion (he in Fig. 1) in advance of bolt shear fracture

161 and weld tearing. The buckling strength of the shear plate was calculated using both the current [7]

162 and previous [26] versions of the AISC design method. To address the higher probability of

163 occurrence of shear plate instability, because of its large eccentricity, the latest AISC design method

164 [2] estimates the shear tab’s buckling strength based on the rectangular plate buckling model [1, 27],

165 while its earlier editions [26] used models representative of the flexural buckling of a doubly coped

166 beam [28-30]. To calculate the buckling strength, the distance between the girder web and the interior

167 bolt line (a distance) was conservatively chosen to be the unbraced length of the shear plate. Both

168 methods predicted that buckling would not prevent the shear plate from reaching its fully plastic

169 flexural capacity (Mp=FyZp). Contrary to the findings from prior research [6-12], the current AISC

170 design method predicted the bolt shear fracture as the connections’ governing failure mode.

9
171 In addition to the nominal and expected material properties, the measured properties of the

172 steel beam, girder, plate, and weld were used to conduct these AISC-based calculations, whereas

173 the nominal properties of the bolts were relied on in this process. Table 2 contains a summary of

174 the calculated connection strengths corresponding to the probable failure modes. The axial force

175 was not considered in these calculations because the AISC shear tab design procedure is limited

176 to connections that carry shear alone.

177 Table 2. AISC predicted strength of shear tab test specimens

BG3-2-13-F BG6-2-19-F
Design Expected Expected Design Expected Expected
Failure mode strength strength1 strength2 strength strength1 strength3
(kN) (kN) (kN) (kN) (kN) (kN)
Flexural and shear yielding of shear plate 293 349 391 1088 1278 1251
Shear yielding of shear plate 616 678 761 1835 2018 1976
Bolt bearing 257 377 377 1172 1875 1771
Buckling of shear plate 333 407 456 1351 1651 1616
Rupture at net section of shear plate 430 688 648 1207 1931 1824
Bolt shear 228 337 337 789 1169 1169
Weld tearing 1497 1995 2524 2616 3489 4451
1
178 Expected strength based on probable material properties i.e.RyFy (1.1 Fy) and RTFu (1.2 Fu) for steel plates [25]
179 2
Expected strength based on measured material properties i.e Fy=432MPa and Fy=508MPa for 13mm plate
180 3
Expected strength based on measured material properties i.e Fy=377MPa and Fy=527MPa for 19mm plate
181 2.2 Test setup

182 The test setup (Fig. 5a) consisted of a 12 MN and a 445 kN hydraulic actuator, a lateral bracing

183 system for the steel beam, supporting elements for the connection, and an axial load application

184 system. The 12 MN actuator was located near the shear tab connection and it developed the main

185 shear force in the connection. The 445 kN actuator, placed near the far end of the beam, facilitated

186 the vertical displacement control of the beam tip, as well as the connection rotation. The lateral

187 bracing system was installed to restrict the lateral displacement of the beam, without affecting its

188 vertical displacement. The overall setup has been successfully used in prior research [7, 8, 13-16,

189 31].

10
 a b

190 Fig. 5. Laboratory tests: (a) test setup, (b) axial load application system

191 The axial load application system (Fig. 5b) was used to maintain a constant axial force on the

192 connection, while following the beam end rotation to maintain a force normal to the beam’s cross-

193 section. Slots on the column flanges allowed two threaded 31.8 mm (1 ¼”) steel rods to pass

194 through and transfer the axial load to a heavily reinforced region of the beam. Further, these rods

195 passed through the moving plate and half cylinder, which allowed for control of the rods’ rotation

196 and vertical displacement, respectively. The axial force was generated by two horizontal Enerpac

197 RRH-3010 hydraulic jacks while the vertical displacement of the moving plate was controlled by

198 a vertical 31.8 mm (1 ¼”) steel rods pass through an Enerpac cylinder.

199 2.3 Instrumentation

200 The implemented test setup was similar to that used in prior research [16], other than the beam

201 lateral bracing system. The new bracing system provided enough free space to implement an

202 optical Coordinate-Measuring Machine (CMM) for 3D measurement of the connection

203 deformation at discrete points (Fig. 6a). Linear Variable Differential Transformers (LVDTs) were

204 installed to measure the shear plate out-of-plane as a backup of the optical CMM system (Fig. 6b).

11
205 Inclinometers measured the in-plane rotation of the beam, top girder flange, shear plate, and

206 column. The out-of-plane rotation of the shear plate and beam was measured as well. String

207 potentiometers were used to measure the vertical deformation of the beam and shear plate, as well

208 as the horizontal displacement of the column capping plate. In order to determine the yielding

209 pattern of the connection, it was whitewashed and strain gauges were installed on the shear plate,

210 beam web and flanges adjacent to the connection (Fig. 6c). Load cells were used to monitor the

211 applied vertical and horizontal forces. Vishay Model 5100B scanners and the Vishay System 5000

212 StrainSmart software were used to record the measured data.

a) b) c)

213
214 Fig. 6. Instrumentation of Specimen BG3-2-13-F-200C: (a) targets of optical CMM system, (b) LVDTs, (c)
215 strain gauges

216 2.4 Loading protocol

217 The loading protocol aimed to simulate end demands of a simply supported beam when

218 subjected to coupled axial and shear force demands. As such, each test specimen was first

219 subjected to its service level of shear load followed by the application of the axial force. From this

220 point in the loading protocol, the axial force was kept constant while the shear demand was

221 increased until failure of the connection. As prior research [16] suggested that shear tab

222 connections locally yielded only in small areas under the service shear load, the axial force was

12
223 applied in advance of yielding onset based on real time monitoring of strain gauge data. For both

224 specimens, axial force was applied at a connection rotation of approximately 0.0085 rad.

225 To resemble the rotational demand at the end of a simply supported beam under gravity

226 induced shear force, 0.02 rad relative rotation between the beam and column was set as a target.

227 This was deemed a rational approach based on prior research [31, 32]. This target rotation should

228 be achieved at the connection probable shear resistance, which was calculated based on the

229 expected material properties in lieu of measured ones, as coupons tests could be conducted only

230 after full-scale tests. To follow the loading protocol, the ratio between the displacement rates of

231 the actuators was adjusted constantly up to the target rotation / load point; after reaching this level,

232 the ratio between displacement rates of the actuators was held constant.

233 2.5 Experimental results

234 Figure 7 shows the response of both specimens versus the connection rotation, relative rotation

235 between the beam and girder (i.e. the girder top flange). The measured connection shear force was

236 normalized by the shear force corresponding to the plastic shear resistance of the plate’s gross

237 section (he in Fig. 1), which is equal to 761 kN and 1976 kN for Specimens BG3-2-13-F and BG6-

238 2-19-F, respectively.

a 1.4 b 35
BG3-2-13-F-200C BG3-2-13-F-200C
1.2 30
LED4 Displacement [mm]

BG6-2-19-F-500C BG6-2-19-F-500C
Shear Force [V/V ]
GP

1 25

0.8 20

0.6 15

0.4 10

0.2 5

0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Connection Rotation [rad] Connection Rotation [rad]
239 Fig. 7. Measured response vs. connection rotation: (a) connection shear force, (b) shear plate out-of-plane
240 deformation

13
241 Referring to Fig. 8a, the axial load was applied to Specimen BG3-2-13-F-200C prior to the

242 plate yielding. The extended portion of the shear plate started to yield along its bottom edge (Strain

243 gauge 13 in Fig 6c) where the compression stress was developed due to the eccentric shear force

244 and the axial compression. Then, plate yielding was observed along the interior bolt line (Strain

245 gauges 14 and 15 in Fig 6c). The top edge of the shear plate yielded after the bottom because the

246 compression force counterbalanced a portion of the developed tensile stress due to the eccentric

247 shear. The connection stiffness reduced at 0.026 rad due to yielding of the extended portion of the

248 shear plate.

249 The connection shear force still increased and yielding propagated toward the girder web at

250 the stiffener upper portion. Strain gauges P6 and P7 indicated that there was flexural yielding due

251 to the eccentric shear force. The stiffener strain gauges, installed adjacent to the girder web,

252 demonstrated the non-uniform distribution of the shear force along the stiffener. Strain gauges P1,

253 P2, and P3 reported yielding stress, while the recorded shear strain of strain gauges P4 and P5 was

254 negligible. The connection stiffness decreased again when the slope of the curve representing the

255 out-of-plane deformation of the plate bottom edge (LED4, Fig. 6a) largely increased. The

256 connection shear force still increased, while the out-of-plane deformation of the plate increased.

257 Following a shear strength plateau (Figs.8b and 8c), binding between the shear plate and the

258 bottom edge of the beam web slightly increased the shear resistance of Specimen BG3-2-13-F-

259 200C. The test was terminated when the beam’s bottom flange started to bind on the shear plate

260 (Fig. 9a). The out-of-plane deformation of the shear plate was obvious at the end of the test (Figs.

261 9b-9d). The tested specimens responded similarly to the combined axial and shear forces other

262 than the strength plateau, which was precluded by binding in Specimen BG6-2-19-F-500.

14
 a b c
1.2
Strength plateau Binding-BF
1 LED 4
Shear Force [V/V ]

SG P6
GP

0.8 Binding-BW
SG P7
SG P1-P2-P3
0.6 Stiffness reducion
SG P12
0.4
SG P15
0.2 SG P13 SG P14
Axial load application
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Connection Rotation [rad]
263 Fig. 8. Specimen BG3-2-13-F-200C: (a) damage propagation, (b and c) deformed shape at strength plataeu

264 Through post-test examination, bolt bearing was obvious along the interior vertical bolt line

265 of both shear plates. Referring to Fig. 10, the bearing deformation was larger at the upper portion

266 of the plate where the tensile and shear stress developed simultaneously due to the applied bending

267 moment and shear force, respectively. In comparison to Specimen BG6-2-19-F (Fig. 11), small

268 fractures and larger bearing deformation were observed along the interior bolt holes in Specimen

269 BG3-2-13-F (Figs. 10). After unloading the specimens, a diagonal crack was observed at the

270 bottom re-entrant corner of the shear plate (Figs. 10c and 11c). It is believed that this occurred due

271 to the out-of-plane deformation of the shear plate and binding between the beam web and the shear

272 plate.

a b c d

273 Fig. 9. Specimen BG3-2-13-F-200 : (a) binding between beam flange and shear plate, (b-d) deformed shape at
274 end of test

15
 a b c

275 Fig. 10. Bearing deformation and fracture along the interior bolt line of specimen BG3-2-13-F-200C at: (a) top
276 bolt hole, (b) middle bolt hole, (c) bottom bolt hole

a b c

277 Fig. 11. Specimen BG6-2-19-F-500C: (a) bolt bearing at plate top half, (b) bolt bearing at plate bottom half, (c)
278 diagonal crack at bottom re-entrant corner

279 To evaluate the accuracy of the current design procedure for extended shear tab connections

280 **, its predictions were compared with laboratory test observations. Referring to Table 2, the

281 current design method suggests that bolt shear fracture should be the governing failure mode.

282 However bolt fracture was not observed in the laboratory tests. Furthermore, no evidence of bolt

283 deformation was observed through post-test examination. The connection stiffness started to

284 decrease at a shear force, which was much larger than the expected resistance corresponding to the

285 flexural and shear yielding of the shear plate. These discrepancies were due to the current design

286 method assumption that the inflection point formed at the support face; hence, the design strength

287 was calculated based on the geometric eccentricity.

16
288 The out-of-plane deformation of the shear plate started to increase rapidly when yielding

289 propagated into the stiffened portion of the plate, which resulted in a reduction of the connection

290 strength. This deformation would likely have been more severe if the shear plate had not satisfied

291 the CSA-S16 compactness requirements [5] for the plate girder stiffeners. Of note, the observed

292 out-of-plane deformation was the result of the combined compression and flexural moment of the

293 shear tab, as demonstrated later on in subsequent FE analyses (Section 3).

294 In addition to the plate yielding, the bolt bearing contributed to the connection ductility.

295 Although the bearing deformation was quite large along the interior vertical bolt line of the shear

296 plate, bearing failure was not considered to have occurred based on observations. The connection

297 shear force became larger than the predicted strength corresponding to the net section fracture,

298 while minor tearing around the bolt holes were observed only in Specimen BG3-2-13-F-200C.

299 This could be attributed to the compressive force influence and the inherent conservatism of the

300 design equation for net section fracture. Furthermore, it was not possible to determine the

301 connections’ ultimate failure mode because binding between the beam web and shear plate

302 changed the load transfer mechanism at the end of the test. The ultimate failure mode could,

303 however, be determined through finite element simulations by excluding the beam binding

304 (Section 3).

305 3 Complementary finite element simulations

306 Complementary finite element (FE) simulations were conducted to further understand the load

307 transfer mechanism in stiffened extended shear tab connections subjected to coupled gravity and

308 axial loads. Several parameters were interrogated that were not evaluated through experiments,

309 including the axial force and the connection’s ultimate failure mode. The FE models were

310 developed in the commercial software ABAQUS-6.11-3 [33]. The features of the FE models were
17
311 chosen to be representative of those seen in the laboratory experiments; including geometry,

312 boundary conditions, material properties, element size and element type, contacts and interactions,

313 and the imposed loading protocol [9, 10]. The employed material properties were defined based

314 on true stress-strain curves of the various components shown in Fig. 12. Other than the bolt’s

315 characteristic response, the implemented stress-strain curves were obtained from testing of the

316 tensile coupons. The bolt’s material properties were defined based on typical stress-strain curves

317 reported in Kulak et al. [34], which were scaled to meet the minimum specified values for ASTM

318 F3125 Grade A490 bolts [21].

319
320 Fig. 12. Finite element model specifics: (a) overall model, (b) column mesh (typical element size of 40 mm),
321 (c) shear plate mesh (typical element size of 3 mm), (d) bolt mesh (typical element size of 1.5 mm), (e) mesh of the
322 beam in the vicinity of connection (typical element size of 20 mm), (f) beam mesh (typical element size of 40 mm)

323 First-order fully-integrated 3D solid elements (C3D8) were utilized to mesh the components.

324 Based on a mesh refinement analysis, the element size (Fig. 11) was determined. Frictionless

325 interaction was defined for surface-to-surface contact pairs between the load cubes and the beam

326 flanges. For all other components in contact, surface-to-surface contact pairs with a friction

327 coefficient of 0.3 was used to allow transmission of tangential force. Furthermore, possible local

18
328 instabilities of the shear tab connection were triggered by the introduction of local imperfections into

329 the shear plate. These local imperfections were proportioned to the limits of manufacturing

330 tolerances for the web and flange of W-sections [35-37]. This approach has been successfully

331 implemented in prior FE studies concerned with member and local instabilities [38].

332 3.1 Model validation

333 To evaluate the accuracy of the numerical analyses, the FE model predictions were compared

334 with the experimental measurements. The developed connection shear force and the out-of-plane

335 deformation of the shear plate were chosen as the FE model verification criteria.

a b
1.4 35
BG6-2-19-F-500C-Exp BG6-2-19-F-500C-Exp
1.2 30
LED4 Displacement [mm]
BG6-2-19-F-500C-FE
BG6-2-19-F-500C-FE
Shear Force [V/V ]
GP

1 25
BG3-2-13-F-200C-Exp
0.8 20
BG3-2-13-F-200C-Exp
0.6 15
BG3-2-13-F-200C-FE
0.4 10

0.2 5
BG3-2-13-F-200C-FE
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Connection Rotation [rad] Connection Rotation [rad]
336 Fig. 13. FE model verification: (a) shear force, (b) shear plate out-of-plane deformation

337 Referring to Fig. 13, the FE model predicted reasonably well the connection response up to

338 the point where the beam web started bearing on the stiffened portion of the shear plate. This

339 discrepancy was due to the uncertainties related to the contact between beam web bottom edge and

340 the shear plate. In addition to the fabrication tolerance and installation of the respective test

341 specimens, these uncertainties arise because of the imperfections introduced into the FE model.

342 The applied imperfections were an estimate based on the connection bifurcation buckling and

343 allowable manufacturing tolerance of W sections. Of note, structural engineers typically neglect

19
344 the over-strength in a connection due to beam binding because it is neither desirable nor

345 dependable.

346 As a snug-tightened connection, the initial response of a shear tab connection depended greatly

347 on the contact between shanks of the bolts and the bolt holes. Because the initial position of each

348 bolt in its hole could not be controlled in the laboratory tests, the bolts were placed at the centre of

349 the bolt hole in the FE model, resulting in a 1 mm (1/32 in.) gap around the entire perimeter.

350 Therefore, the real contact conditions of the bolts may be different from those assumed in the FE

351 models. Due to this discrepancy, the FE model predictions for the connection shear force deviated

352 from the test measurements in the initial increments of the applied loading.

353 3.2 Simulation results

354 Figures 14 and 15 show the normalized predictions of the FE models. Referring to Figs. 14a

355 and 15a, the shear force along the outer end of the shear plate re-entrant corners was normalized

356 based on the plastic shear resistance of the gross section ( VGP  0.6F y A g ), while the plate’s plastic

357 shear resistance of the net section ( VNP  0.6F y A net ) was implemented to normalize the shear force

358 along the bolt line (Figs. 14b and 15b). The plastic bending moment resistance of the gross section

359 ( M GP  Fy Z g ) was used to normalize the bending moment at the plate’s gross section, as shown in

360 Figs. 14c and 15c. The bending moment along the plate’s interior bolt line (Figs. 14d and 15d) was

361 normalized based on the flexural capacity of the plate’s net section ( M NP  Fy Z net ). The plastic

362 section modulus was defined for an odd number of horizontal bolt lines as

363 Znet  1/4tpl (s - dh )(n2s  dh ) , while Znet  1/4tpl (s - d h )(n2s) was used for an even number of

364 horizontal bolt lines [39]. In these equations, n=number of horizontal bolt lines, s=bolt spacing,

365 dh=diameter of bolt hole, tpl=plate thickness, and dpl=plate depth. The aforementioned plastic
20
366 capacities of the shear plate, shown in Table 3, were calculated based on its measured dimensions

367 and yield stress.

368 Table 3. Calculated plastic capacities of shear tab test specimens

Specimens BG3-2-13-F BG6-2-19-F
PGP (Fy A g  Fy d pl t pl ) 1268 kN 3294 kN
PNP ( Fy A net  Fy (d pl  nd h )t pl ) 950 kN 2331 kN
VGP (0.6F y A g  0.6F y d pl t pl ) 761 kN 1976 kN
VNP ( 0.6F y A net  0.6F y (d pl  nd h )t pl ) 570 kN 1398 kN
MGP (Fy Zg  F t d /4) 2
y pl pl 72.5 kN.m 376.5 kN.m
M NP ( Fy Z net ) 54.0 kN.m 256.8 kN.m
369

370 Regarding Specimen BG3-2-10-F, a comparison between the normalized shear flow and the

371 connection rotation (Figs. 14a and 14b) demonstrated that only a fraction of the connection shear

372 force was transferred through the net section along the centerline of the bolt holes, the critical

373 section with the smallest cross-sectional area along the plate. Referring to Fig. 14a, Specimen

374 BG3-2-13-F experienced the connection shear force equal to 614 kN (V/VGP =0.81) at 0.04 rad

375 rotation while the net section was subjected to only 463 kN shear force (VN/VNP =0.81 in Fig. 14b).

376 Figures 15a and 15b show a similar trend for Specimen BG6-2-19-F. This observation, which

377 coincided with prior research studies [40], was due to the bearing mechanism between the bolt

378 shanks and the bolt holes. This is further elaborated in Section 4.2. A larger bending moment

379 developed at the gross section (Figs. 14c and 15c) in comparison to the net section (Figs. 14d and

380 15d) because the inflection point (Figs. 14e and 15e) formed far from the column face, farther from

381 the bolt group centroid.

382 To evaluate the influence of the axial load on the observed connection behaviour and failure

383 modes, additional FE analyses were carried out for each specimen. Only gravity-induced shear

384 force was applied to the connection in the first FE analysis, while the connection was subjected to

21
385 combined tensile and shear forces in the second one. These FE models were subjected to the

386 representative experiment loading protocols; to maintain simplicity, the magnitude of the tensile

387 force in the analysis was set equal to the magnitude of the compression force used during testing.

a b

c 1 d 1
BG3-2-13-F
]
Bending Moment [M/M ]

NP

BG3-2-13-F-200T
GP

0.8 0.8
Bending Moment [M /M

BG3-2-13-F-200C
N

0.6 0.6

0.4 0.4

0.2 BG3-2-13-F 0.2

BG3-2-13-F-200T
BG3-2-13-F-200C
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Connection Rotation [rad] Connection Rotation [rad]

e 1 f 14
BG3-2-13-F
12
LED4 Displacement [mm]

BG3-2-13-F-200T
0.8
Shear Force [V/V ]

BG3-2-13-F-200C
GP

10
0.6
8

0.4 6

BG3-2-13-F 4
0.2
BG3-2-13-F-200T 2
BG3-2-13-F-200C
0 0
0 50 100 150 200 250 300 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Effective Eccentricity (e ) [mm] Connection Rotation [rad]
eff
388 Fig. 14. Simulated response of Specimen BG3-2-13-F: (a) connection shear force, (b) net section shear force,
389 (c)gross section bending moment, (d) net section bending moment, (e) effective eccentricity, (f) plate out-of-plane
390 deformation

391

22
 a 1.2 b 1.2

1 1

]
Shear Force [V/V ]

NP
GP

Shear Force [V /V
0.8 0.8

N
0.6 0.6

0.4 0.4
BG6-2-19-F BG6-2-19-F
0.2 BG6-2-19-F-500T 0.2 BG6-2-19-F-500T
BG6-2-19-F-500C BG6-2-19-F-500C
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Connection Rotation [rad] Connection Rotation [rad]

c 1 d 1
BG6-2-19-F

]
Bending Moment [M/M ]

BG6-2-19-F-500T

NP
GP

0.8 0.8

Bending Moment [M /M
BG6-2-19-F-500C

N
0.6 0.6

0.4 0.4

0.2 BG6-2-19-F 0.2

BG6-2-19-F-500T
BG6-2-19-F-500C
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Connection Rotation [rad] Connection Rotation [rad]

e 1.2
BG6-2-19-F
f 25
BG6-2-19-F
BG6-2-19-F-500T BG6-2-19-F-500T
LED4 Displacement [mm]

1
20
Shear Force [V/V ]

BG6-2-19-F-500C BG6-2-19-F-500C
GP

0.8
15
0.6
10
0.4

0.2 5

0 0
0 100 200 300 400 0 0.02 0.04 0.06 0.08
Effective Eccentricity (e ) [mm] Connection Rotation [rad]
eff
392 Fig. 15. Simulated response of Specimen BG6-2-19-F: (a) connection shear force, (b) net section shear force,
393 (c)gross section bending moment, (d) net section bending moment, (e) effective eccentricity, (f) plate out-of-plane
394 deformation

395 In all FE models, gross and net section yielding of the shear plate were observed and the net

396 section fracture along the plate interior bolt line was determined as the connection’s ultimate

397 failure mode. Referring to Figs. 14 and 15, the axial force affected the connection’s response

23
398 slightly because the level of the applied axial load was small (P/PGY=0.16 and 0.15 for Specimens

399 BG3-2-13-F and BG6-2-19-F, respectively).

400 4 Discussion

401 4.1 Shear plate yielding

402 Referring to Fig. 16, Neal’s interaction equation [41] was used to account for the interaction

403 of axial, shear, and flexural loads at the plate gross and net sections. It was observed that the results

404 of Neal’s [41] and the AISC’s [2] interaction equations (Eqs. 1 and 2, respectively) were almost

405 equal. Of note, Astaneh proposed Eq. (2) as a simplified version of Neal’s interaction equation

406 [42]. Regarding the shear tab design, the AISC considers the interaction of the shear and bending

407 moment using an elliptical interaction equation (Eq. (3)).

V 4
) (
M P 2 VP
408 ( )( ) ( ) 1 (1)
MP PP P 2
1 ( )
PP

M P V
409 ( )  ( )2  ( )4  1 (2)
MP PP VP

M 2 V
410 ( )  ( )2  1 (3)
MP VP

411 The behaviour of the FE model of Specimens BG3-2-13-F-200C and BG6-2-19-F-500C was

412 similar to the test specimens. Yielding began from the re-entrant corners of the shear plate, then

413 propagated toward the interior bolt line. The FE models showed that the connection stiffness

414 slightly decreased when a large portion of the shear plate along the interior bolt line yielded. The

415 full depth of the shear plate along the net section yielded after yielding of the gross section of

416 Specimen BG3-2-13-F-200C, while they occurred at the same time in Specimen BG6-2-19-F-

24
417 500C. Following the shear plate yielding, its out-of-plane deformation increased. Furthermore, the

418 FE models demonstrated that the net section fracture would determine the connection’s ultimate

419 strength in the absence of beam binding. Referring to Fig. 17, the maximum plastic strain

420 developed at the bottom re-entrant corner and at the bolt holes of the plate’s upper portion.

a b
1.6 1.6
1.4 1.4

Neal's Interaction Equation

Neal's Interaction Equation

1.2 1.2
1 1
0.8 0.8
0.6 0.6
0.4 BG3-2-13-F 0.4 BG3-2-13-F
0.2 BG3-2-13-F-200T 0.2 BG3-2-13-F-200T
BG3-2-13-F-200C BG3-2-13-F-200C
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14
Connection Rotation [rad] Connection Rotation [rad]
c d
2.5 2.5
Neal's Interaction Equation

Neal's Interaction Equation

2 2

1.5 1.5

1 1

0.5 BG6-2-19-F 0.5 BG6-2-19-F

BG6-2-19-F-500T BG6-2-19-F-500T
BG6-2-19-F-500C BG6-2-19-F-500C
0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Connection Rotation [rad] Connection Rotation [rad]
421 Fig. 16. Neal Interaction equation (Eq. (1)) at: (a and b) gross and net sections of Specimen BG3-2-13-F,
422 respectively, (c and d) gross and net sections of Specimen BG6-2-19-F, respectively

423

25
 a b

424 Fig. 17. Shear plate plastic strain corresponding to the net section fracture at: (a) BG3-2-1-13-F-200C, (b) BG6-
425 2-1-9-F-500C

426 4.2 Shear plate internal forces along the interior bolt line

427 Figures 14 and 15 show that the net section, the section along the bolt line centerline, was subjected

428 to only a portion of the connection shear force. Furthermore, applying the axial force changed the

429 shear demand at the net section (Figs. 14b & 15b). To clarify this fact, the net shear and axial

430 forces were compared with corresponding values from the gross section of the plate, Fig. 18.

431 Referring to Figs. 18a and 18b, the tensile force increased the ratio between the shear force at the

432 net and gross sections, while the compression force decreased it. Referring to Figs. 18c and 18d,

433 the axial force along the net section was compared with the applied axial force (Pa), 200 kN and

434 500 kN for Specimens BG3-2-13-F and BG6-2-19-F, respectively. In comparison to the tensile

435 force, the net section was subjected to a smaller portion of the applied axial force in the presence

436 of the compression force. Furthermore, Figs. 18c and 18d show that the tensile force was developed

437 along the net section even under gravity-induced shear force.

26
 a b
1.2 1

1 0.8
Shear Force [V /V ]

Shear Force [V /V ]
G

G
0.8
N

N
0.6
0.6
0.4
0.4
BG3-2-13-F 0.2 BG6-2-19-F
0.2 BG3-2-13-F-200T BG6-2-19-F-500T
BG3-2-13-F-200C BG6-2-19-F-500C
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0 0.02 0.04 0.06 0.08
Connection Rotation [rad] Connection Rotation [rad]
c d
2 2
BG6-2-19-F-500T
BG3-2-13-F-200T
1.5 1.5

Axial Force [P /P ]

Axial Force [P / P ]

BG6-2-19-F

a
a

1 1

N
N

BG3-2-13-F
0.5 0.5
BG3-2-13-F-200C
0 0

-0.5 -0.5
BG6-2-19-F-500C
-1 -1
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0 0.02 0.04 0.06 0.08
Connection Rotation [rad] Connection Rotation [rad]
438 Fig. 18. FE model predictions for: (a) shear force of BG3-2-13-F models, (b) shear force of BG6-2-19-F
439 models, (c) Axial force of BG3-2-13-F models,(d) Axial force of BG6-2-19-F models

440 The bearing mechanism between the bolt shanks and the bolt holes was thoroughly studied to

441 explain the reasons for the aforementioned observations. Figure 19a shows the bolt group, which

442 was subjected to the eccentric shear force. In addition to the vertical shear force, a horizontal force

443 was developed in the top and bottom bolts due to the eccentric shear force and its consequent

444 bending moment. Referring to Fig 19b, the horizontal force moved the top bolt away from the

445 centerline of the bolt hole, while the bottom bolt moved closer to the support.

27
 a b c

446 Fig. 19. Bolt group under an eccentric shear force, (a) applied shear force, (b) resultant force at each bolt

447 The middle bolt (Fig. 20a) transferred a shear force to the plate while it was placed along the

448 centerline of the bolt hole. Therefore, half of the bolts’ shear force was transferred through the net

449 section. In the presence of the tensile force (the top bolt), the net section was subjected to a larger

450 portion of the shear and axial forces as the bolt moved away from the support and crossed the bolt

451 line centerline (Fig. 20b). Therefore, the horizontal force of the top bolt subjected the net section

452 to the tensile force (Fig. 19c). That was the reason behind development of an extra tension in Figs.

453 18c and 18d. In contrast, compression pushed the bottom bolt toward the support (Fig. 20c) and

454 the net section resisted a smaller component of the shear and axial force.

455

a b c

456 Fig. 20. Bolt under: (a) shear force, (b) shear and tension, (c) shear and compression

457

458

28
459 4.3 Effect of axial force

460 Referring to Figs. 14a and 15a, the axial tensile force decreased the ultimate shear resistance

461 of the connection, while the axial compression force increased it. This occurred because the tensile

462 force increased the force demands on the interior bolt line of the shear plate, while the compression

463 force decreased those demands (Figs. 14b and 15b). Then, the tensile force hastened the onset of

464 the connection’s ultimate failure mode, i.e. net section fracture of the shear plate, while the axial

465 compression force delayed the onset of this failure mode. The same observations held true for the

466 connection resistance corresponding to the net section yielding. Referring to Table 4, the tension

467 force caused the net section yielding to precede the gross section yielding. However, the difference

468 between the yielding strength of the net and gross sections was small; hence, the connection could

469 still resist much larger shear after the gross section yielding. In addition to the axial force, the ratio

470 between the gross and net section areas affected the yielding sequence of the gross and net sections.

471 In model BG3-2-13-F, the net section yielded shortly after the gross section, while they occurred

472 at the same time in the BG6-2-19-F model. The aforementioned ratio, Anet/Ag, was equal to 0.73

473 and 0.69 for Specimens BG3-2-13-F and BG6-2-19-F, respectively.

474 Table 4. FE model predictions for connection resistance

BG3-2-13-F BG6-2-19-F
Axial Load 200C 0 200T 500C 0 500T
Measured Measured Measured Measured Measured Measured
Failure mode strength strength strength strength strength strength
(kN) (kN) (kN) (kN) (kN) (kN)
Gross section yielding 507 518 517 1674 1676 1631
Net section yielding 631 545 450 1767 1676 1544
Out-of-plane deformation 662 --- --- 1995 2021 ---
Net section fracture 688 666 634 2120 2103 2046
475

476 Referring to Figs 14f and 15f, the axial compression force increased the plate’s out-of-plane

477 deformation, while the tension force decreased it. This observation suggested that the compression
29
478 could trigger the shear plate buckling and change the connection’s ultimate failure mode,

479 especially in the case of a slender shear plate or larger compressive force.

480 4.4 Evaluation of the current design procedure of extended shear tab connections

481 Various failure modes were observed in the studied connection configurations, both tested and

482 numerical, including the gross and net section yielding of the shear plate, the shear plate out-of-

483 plane deformation, and the net section fracture. Of note, the shear plate yielded at its gross and net

484 sections because of the interaction of moment, shear and axial force. Referring to Table 5, to

485 evaluate the accuracy of the current AISC design method [2], the results obtained from it were

486 compared with those determined from the experimental measurements and the FE model. The

487 design method became more accurate if the geometric eccentricity was replaced with the measured

488 eccentricity corresponding to the gross section yielding of the shear plate. Furthermore, the current

489 design method correctly predicted the governing failure mode when the measured eccentricity was

490 implemented. Referring to Table 5, although the AISC elliptical moment-shear interaction

491 equation (Eq. (3)) resulted in a conservative estimate of the moment-shear-axial force yielding of

492 the shear plate gross section, it might overestimate the shear plate yielding strength in the presence

493 of a large axial force. Based on Eqs. (1) and (2), the shear plate gross section of Specimens BG3-

494 2-13-F-200C and BG6-2-19-F-500C yielded at a connection shear force equal to 496 kN and 1595

495 kN, respectively. Furthermore, the current design procedure might significantly overestimate the

496 buckling strength of Specimen BG6-2-19-F-500C, because it neglected the detrimental effects of

497 the axial and shear forces on the plate’s flexural capacity. To address this issue, Dowswell &

498 Whyte [27] used Eq. (1) to determine the available flexural buckling strength in the presence of

499 the shear and axial forces. If this advice was taken for the test specimens, the buckling strength of

500 the extended portion of the shear plate was equal to the applied force corresponding to the gross
30
501 section yielding of the shear plate. To calculate the weld group capacity under an eccentric shear

502 force, the Instantaneous Centre of Rotation (ICR) method was implemented for the C-Shape weld

503 group, while only the vertical weld lines were considered in the calculation of the weld group

504 capacity under a concentric shear force.

505 Table 5. Connection resistance to different failure modes

BG3-2-13-F-200C BG6-2-19-F-500C
Expected Expected Measured Expected Expected Measured
Failure mode strength1 strength2 strength strength1 strength2 strength
(kN) (kN) (kN) (kN) (kN) (kN)
Plate moment-shear-axial force yielding 391 4823 507 1251 16303 1674
Plate Shear yielding 761 761 -- 1976 1976 1976
4
Bolt bearing 377 978 -- 1771 4202 --4
Plate buckling 456 6255 6626 1616 28855 19956
Rupture at net section of shear plate 648 648 687 1824 1824 2120
Bolt shear 337 874 >687 1169 2774 >2120
7 7
Weld tearing 2524 2334 -- 4451 4777 --
1
506 Expected strength based on geometric eccentricity (e)
507 2
Expected strength based on measured eccentricity
508 3
Yielding strength of the extended portion of the shear plate based on elliptical yield criterion (Eq. (3))
4
509 Although large bearing deformation was observed, bearing failure did not occur
510 5
Buckling strength of the extended portion of the shear plate
511 6
Shear resistance corresponding to the shear plate out-of-plane deformation
512 7
Strength of C-shape weld group
513

514 Among the observed failure modes, the gross section yielding of the shear plate occurred earlier

515 under a smaller shear force. Furthermore, other failure modes occurred when the connection

516 underwent large deformation and rotation, which negatively affected the supported beam’s

517 serviceability. Therefore, the moment-shear-axial force yielding of the shear plate’s gross section

518 should be considered as a conservative estimate of the connection’s capacity. In the presence of

519 the axial tensile force, yielding of the net section preceded yielding of the gross section (i.e. BG3-

520 2-13-F-200T and BG6-2-19-F-500T). However, the yield strength of the gross section was still a

521 conservative estimate of the connection’s capacity because the difference between the yield

31
522 strength of the gross and net sections was small and the connection was able to resist a much larger

523 shear force.

524 5 Conclusions

525 Two full-scale specimens were tested in order to deepen our understanding of the behaviour

526 of the double-sided configuration of the full-depth extended beam-to-girder shear tab under

527 coupled gravity and axial force demands. The test specimens were constructed of different

528 features, including shear plate dimensions, bolt size, bolt group configuration, geometric

529 eccentricity, beam and girder sizes. Furthermore, validated finite element models were adopted to

530 investigate the dependency of the connection’s behaviour on critical parameters including the axial

531 force direction and the force distribution along the plate net section. The main findings of the paper

532 are summarized as follows:

533  The double-sided configuration of the full-depth extended beam-to-girder shear tab

534 yielded through its net section along the bolt line, the closest to the girder. Furthermore,

535 the gross section yielding of the shear plate occurred along the outer end of its re-

536 entrant corners.

537  The net section fracture was determined as the ultimate failure mode of the studied

538 connections.

539  The net section along the centerline of the plate’s interior bolt line was subjected to a

540 portion of the connection axial and shear forces. This amount depended on the number

541 of vertical bolt lines, bolt hole diameter, the distance between bolt holes, the axial load

542 direction and magnitude, and the initial position of the bolt in its hole.

32
543  The compressive axial load increased the out-of-plane deformation of the shear plate,

544 which could result into plate buckling in the case of the slender shear plate or a larger

545 compression force. The axial compression force decreased the shear force demand on

546 the net section.

547  The tensile axial force accelerated the plate yielding and fracture along the interior bolt

548 line by increasing the force demands on the shear plate’s net section. Furthermore, the

549 tensile force decreased the shear plate’s out-of-plane deformation and delayed the plate

550 buckling.

551  The gross section yielding strength of the shear plate could be considered as a

552 conservative estimate of the connection capacity as the connection resisted much larger

553 shear force following the gross section yielding of the shear plate. Further analyses are

554 needed to validate this finding in the presence of a large tensile force.

555  The current design method significantly underestimated the connection shear capacity

556 due to the assumption that the inflection point formed at the girder web’s face. In

557 contrast, the inflection point formed far away from the girder web, farther from the

558 bolt group centroid.

559 6 Acknowledgments

560 The authors would like to thank the ADF Group Inc. and DPHV Structural Consultants for their

561 generous technical and financial support, as well as the Natural Sciences and Engineering Research

562 Council of Canada. The finite element computations were conducted on the McGill University

563 supercomputer Guillimin, which is managed by Calcul Québec and Compute Canada. The

33
564 supercomputer operation is funded by the Canada Foundation for Innovation (CFI), NanoQuébec,

565 RMGA and the Fonds de recherche du Québec - Nature et technologies (FRQ-NT).

566 7 References

567 [1] AISC 360-16, Specification for structural steel buildings, American Institute of steel

568 Construction, Chicago, IL, 2016.

569 [2] AISC 325-17, Steel construction manual, 15th edition, American Institute of steel Construction,

570 Chicago, IL, 2017.

571 [3] K. Thomas, Design and behaviour of extended shear tabs under combined loads, Master Thesis,

572 University of Alberta, Edmonton, Alberta., 2014.

573 [4] A.R. Tamboli, Handbook of structural steel connection design and details, Third edition,

574 McGraw-Hill, New York, NY., 2016.

575 [5] CSA-S16-14, Design of steel structures, Canadian Standards Association, Mississauga, ON.,

576 2014.

577 [6] D.R. Sherman, A. Ghorbanpoor, Design of extended shear tabs, University of Wisconsin-

578 Milwaukee, Milwaukee, WI, 2002.

579 [7] J. Hertz, Testing of extended shear tab connections subjected to shear, Master's Thesis, McGill

580 University, Montreal, QC, 2014.

581 [8] N. Goldstein Apt, Testing of extended shear tab and coped beam-to-girder connections subject

582 to shear loading, Master's Thesis, McGill University, Montreal, QC, 2015.

583 [9] M. Motallebi, D.G. Lignos, C.A. Rogers, Behavior of stiffened extended shear tab connections

584 under gravity induced shear force., J. Constr. Steel Res., Under Review (2018).

585 [10] M. Motallebi, D.G. Lignos, C.A. Rogers, Stability of stiffened extended shear tab connections

586 under gravity induced shear force, J. Constr. Steel Res., Under Review (2018).
34
587 [11] W. Goodrich, Behavior of extended shear tabs in stiffened beam-to-column web connections,

588 Thesis, Vanderbilt University, Nashville, TN, 2005.

589 [12] K. Thomas, R.G. Driver, S.A. Oosterhof, L. Callele, Full-scale tests of stabilized and

590 unstabilized extended single-plate connections, Structures, 10 (2017) 49-58.

591 [13] M. Marosi, Behaviour of single and double row bolted shear tab connections and weld

592 retrofits, Master's Thesis, McGill University, Montreal, QC, 2011.

593 [14] M. Marosi, M. D’Aronco, R. Tremblay, C.A. Rogers, Multi-row bolted beam to column shear

594 tab connections, 6th European Conference on Steel and Composite Structures, Budapest,Hungary,

595 2011.

596 [15] M. D'Aronco, Behaviour of double and triple vertical rows of bolts shear tab connections and

597 weld retrofits, Master's Thesis, École Polytechnique de Montréal, Montreal, QC, 2013.

598 [16] A. Mirzaei, Steel shear tab connections subjected to combined shear and axial forces, PhD

599 Thesis, McGill University, Montreal, QC, 2014.

600 [17] J. Hertz, D.G. Lignos, C.A. Rogers, Full scale testing of extended beam-to-column and beam

601 to-girder shear tab connections subjected to shear, 8th International Conference on Behavior of

602 Steel Structures in Seismic Areas, Shanghai, China, 2015.

603 [18] C.A. Rogers, M. Marosi, J. Hertz, D.G. Lignos, R. Tremblay, M. D’Aronco, Performance of

604 weld-retrofit beam-to-column shear tab connections, 8th Int. Workshop on Connections in Steel

605 Structures, Boston, MA., 2016.

606 [19] ASTM A992 / A992M-11(2015), Standard specification for structural steel shapes, ASTM

607 International, West Conshohocken, PA, 2015.

608 [20] ASTM A572 / A572M-15, Standard specification for high-strength low-alloy columbium-

609 vanadium structural steel, ASTM International, West Conshohocken, PA, 2015.

35
610 [21] ASTM F3125 / F3125M-15a, Standard specification for high strength structural bolts, steel

611 and alloy steel, heat treated, 120 ksi (830 mpa) and 150 ksi (1040 mpa) minimum tensile strength,

612 inch and metric dimensions, ASTM International, West Conshohocken, PA, 2015.

613 [22] ASTM A370-17, Standard test methods and definitions for mechanical testing of steel

614 products, ASTM International, West Conshohocken, PA, 2017.

615 [23] AWS A5.20/A5.20M:2005 (R2015), Carbon steel electrodes for flux cored arc welding,

616 American Welding Society, Miami, FL., 2015.

617 [24] A.M. Kanvinde, I.R. Gomez, M. Roberts, B.V. Fell, G.Y. Grondin, Strength and ductility of

618 fillet welds with transverse root notch, J. Constr. Steel Res., 65 (2009) 11.

619 [25] AISC 341-16, Seismic provisions for structural steel buildings, American Institute of steel

620 Construction, Chicago, IL, 2016.

621 [26] AISC 325-11, Steel construction manual, 14th edition, American Institute of steel

622 Construction, Chicago, IL, 2011.

623 [27] B. Dowswell, R. Whyte, Local stability of double-coped beams, Eng. J. AISC, 51(1) (2014)

624 43-52.

625 [28] J.-J. Cheng, J. Yura, C. Johnson, Design and behavior of coped beams, University of Texas

626 at Austin, Austin, TX, 1984.

627 [29] L. Muir, W. Thornton, A direct method for obtaining the plate buckling coefficient for double-

628 coped beams, Eng. J. AISC, 41 (2004) 133-134.

629 [30] L.S. Muir, C.M. Hewitt, Design of unstiffened extended single-plate shear connections, Eng.

630 J. AISC, 46(2) (2009) 67-80.

631 [31] A. Astaneh, K.M. McMullin, S.M. Call, Behavior and design of steel single plate shear

632 connections, J. Struct. Eng. ASCE, 119(8) (1993) 2421-2440.

36
633 [32] A. Astaneh, Demand and supply of ductility in steel shear connections, J. Constr. Steel Res.,

634 14(1) (1989) 1-19.

635 [33] ABAQUS 6.11-3, [Computer software], Dassault Systemes Simulia Corp., Providence, RI.

636 [34] G.L. Kulak, J.W. Fisher, J.H. Struik, Guide to design criteria for bolted and riveted joints,

637 AISC, Chicago,IL, 2001.

638 [35] ASTM A6 /A6M, General requirements for rolled structural steel bars, plates, shapes, and

639 sheet piling, ASTM International, 2004.

640 [36] CSA-G40.20-13/G40.21-13, General requirements for rolled or welded structural quality

641 steel/ structural quality steel, Canadian Standards Association, Toronto, ON., 2013.

642 [37] CISC, Handbook of steel construction, Canadian Institute of Steel Construction, Markham,

643 ON., 2016.

644 [38] A. Elkady, D.G. Lignos, Analytical investigation of the cyclic behavior and plastic hinge

645 formation in deep wide-flange steel beam-columns, Bull. Earthq. Eng., 13(4) (2015) 1097-1118.

646 [39] B.A. Mohr, T.M. Murray, Bending strength of steel bracket and splice plates, Eng. J. AISC,

647 45(2) (2008) 97-106.

648 [40] P. Salem, Unified design criteria for steel cantilever plate connection elements, PhD Thesis,

649 University of Alberta, Edmonton, AL., 2016.

650 [41] B.G. Neal, The effect of shear and normal forces on the fully plastic moment of a beam of

651 rectangular cross section, Journal of Applied Mechanics, 28(2) (1961) 269-274.

652 [42] A. Astaneh, Seismic behavior and design of gusset plates, Steel Tips, Structural Steel

653 Education Council, Moraga, CA., 1998.

654

655

37