University of Wollongong

Research Online

Faculty of Engineering and Information Faculty of Engineering and Information

Sciences - Papers: Part B Sciences

2019

Whitmore Tension Section and Block Shear

Matthew Elliott
University of Wollongong, mde813@uowmail.edu.au

Lip H. Teh
University of Wollongong, lteh@uow.edu.au

Follow this and additional works at: https://ro.uow.edu.au/eispapers1

Part of the Engineering Commons, and the Science and Technology Studies Commons

Recommended Citation
Elliott, Matthew and Teh, Lip H., "Whitmore Tension Section and Block Shear" (2019). Faculty of
Engineering and Information Sciences - Papers: Part B. 2927.
https://ro.uow.edu.au/eispapers1/2927

Research Online is the open access institutional repository for the University of Wollongong. For further information
contact the UOW Library: research-pubs@uow.edu.au
Whitmore Tension Section and Block Shear

Abstract
This paper examines the validity of the Whitmore net section tension capacity for the design of bolted
gusset plates. Using simple algebra, this paper first shows that the Whitmore criterion and the correct
block shear criterion would give similar results for a standard connection having approximately seven
rows of bolts. It then shows that the Whitmore criterion severely underestimates the actual capacities of
connections having two or three bolt rows tested by independent researchers. Conversely, it also shows
that the same criterion overestimates the capacities of connections having nine bolt rows that were
believed by the testing researchers to fail in the Whitmore section. Using finite-element analysis
incorporating fracture simulation, this paper shows that the apparent Whitmore tensile fractures only took
place because the tests were continued long after the ultimate limit state of block shear. This paper
proposes that the Whitmore section check be made redundant in light of the block shear check, which
accurately predicted the ultimate test loads of all the specimens.

Keywords
shear, block, whitmore, tension, section

Disciplines
Engineering | Science and Technology Studies

Publication Details
Elliott, M. & Teh, L. H. (2019). Whitmore Tension Section and Block Shear. Journal of Structural
Engineering, 145 (2), 04018250-1-04018250-10.

This journal article is available at Research Online: https://ro.uow.edu.au/eispapers1/2927

 1

1 The Whitmore Tension Section and Block Shear

2 Matthew D. Elliott 1 and Lip H. Teh 2 M.ASCE

3 Abstract:

4 This paper examines the validity of the Whitmore net section tension capacity for the design of
5 bolted gusset plates. Using simple algebra, this paper first shows that the Whitmore criterion and the
6 correct block shear criterion would give similar results for a standard connection having
7 approximately seven rows of bolts. It then shows that the Whitmore criterion severely underestimates
8 the actual capacities of connections having two or three bolt rows tested by independent researchers.
9 Conversely, it also shows that the same criterion overestimates the capacities of connections having
10 nine bolt rows that were believed by the testing researchers to fail in the Whitmore section. Using
11 finite element analysis incorporating fracture simulation, this paper shows that the apparent
12 Whitmore tensile fractures only took place because the tests were continued long after the ultimate
13 limit state of block shear. This paper proposes that the Whitmore section check be made redundant in
14 light of the block shear check, which accurately predicted the ultimate test loads of all the specimens.

15 Author keywords: bolted connection, block shear, connection capacity, connection design, gusset
16 plate, steel connection, Whitmore section

17

1
PhD Candidate, School of Civil, Mining & Environmental Engineering, University Of Wollongong, Wollongong, NSW
2500, AUSTRALIA.
2
Associate Professor, School of Civil, Mining & Environmental Engineering, University Of Wollongong, Wollongong,
NSW 2500, AUSTRALIA.
 2

18 Introduction

19 In Section J4.1 “Strength of Elements in Tension” of the Specification for Structural Steel Buildings
20 (AISC 2016), the effective net area Ae of a connection plate may be limited to that calculated using
21 the Whitmore section. The Whitmore section is defined by drawing 30o lines from the pair of outer
22 downstream bolts to their respective intersections with a line passing through the upstream row of
23 bolts, as illustrated in Figure 1(a) for a bolted gusset plate. The Whitmore section was described by
24 Whitmore (1952) as a reasonable method for approximating the maximum tensile (and compressive)
25 elastic stress incurred in a riveted gusset plate by the axial force in a connected brace, and found
26 widespread use from the late 1970s (Thornton & Lini 2011). However, the Whitmore net section
27 tension capacity was not explicitly mentioned in the AISC specifications until the 2010 edition
28 (AISC 2010) in the form of User Note in Section J4.1.

29 Incidentally, around the time the Whitmore section became widely known among the structural
30 engineering community, Birkemore & Gilmor (1978) discovered the block shear failure mode for
31 coped beam shear connections. The occurrence of this failure mode in bolted gusset plates and braces
32 were subsequently studied by Richard et al. (1983), Hardash & Bjorhovde (1985), Gross & Cheok
33 (1988), Cunningham et al. (1995), Aalberg & Larsen (1999), Menzemer et al. (1999) and Topkaya
34 (2004) among many others. Studies involving both the Whitmore net section and the block shear
35 failure modes of bolted gusset plates were conducted in recent years by Higgins et al. (2010), Liao et
36 al. (2011) and Rosenstrauch et al. (2013).

37 According to Kulak et al. (2001, pg 253), the design of a gusset plate is to be checked against both
38 the Whitmore section and the block shear failure mode, and the more severe requirement resulting
39 from them should then be applied. Similarly, a reviewer of a recent paper (Teh & Deierlein 2017)
40 argued that the design example presented therein was controlled by the Whitmore tension rupture,
41 and the block shear criterion was therefore irrelevant to the example.

42 However, the authors have not found any convincing experimental evidence indicating the failure
43 mode of a bolted gusset plate that corresponds to the Whitmore net section, which would involve
44 simultaneous (or near simultaneous) fractures on both sides of each bolt hole, as illustrated in Figure
45 1(b). It should be noted that, in cases where fracture can be observed to have taken place on the outer
46 side of the bolt hole(s), the prevailing failure mode was actually block shear rather than Whitmore
47 section fracture, as demonstrated later in this paper. The outer side of the bolt hole only fractured
 3

48 when testing continued well beyond the ultimate (block shear) limit state of the specimen, after the
49 inner region fractured completely. Such cases include Specimen 28 tested by Hardash & Bjorhovde
50 (1985), who correctly identified the failure mode to be block shear.

51 Bjorhovde & Chakrabarti (1985) presented laboratory test results of bolted gusset plate connections,
52 and concluded that their test results were in acceptable agreement with the Whitmore concept.
53 Rabern (1983) had earlier noted that his finite element stress contours suggested that a block shear
54 criterion, modified from that derived for a coped beam (Birkemoe and Gilmor 1978), might be
55 applicable to a bracing connection of a gusset plate. However, Rabern (1983) concluded that his
56 finite element studies supported the Whitmore criterion for gusset plate design. While this conclusion
57 was affirmed by Richard et al. (1983), they also stated that the block shear concept might lead to a
58 more compact and efficient connection.

59 The ambiguity regarding the finite element finding as described in the preceding paragraph can also
60 be found in the work of Williams (1988), cited by Williams & Richard (1996). However, Williams &
61 Richard (1996) argued that the block shear and the Whitmore criteria gave the same result. This
62 argument is consistent with the statement of Gross & Cheok (1988) that the Whitmore design
63 criterion was essentially the same as the block shear design concept.

64 Astaneh-Asl (1998), on the other hand, made a distinction between “fracture along the Whitmore’s
65 30-degree effective width” and “block shear failure”, both of which were claimed to have been
66 observed in the field after earthquakes or in laboratories.

67 More recently, Rosenstrauch et al. (2013) stated that the so-called direct tension method (FHWA
68 2009), which evaluates the yield or ultimate capacity of the Whitmore section, did not accurately
69 predict the onset of plasticity or the behaviour of gusset plates. They argued that their finite element
70 analysis showed that block shear and the so-called global section shear were possible failure
71 mechanisms for gusset plates in bridge connections. The guidance document issued by the Federal
72 Highway Administration (FHWA 2009) was meant to assist the rating process of bridges in the wake
73 of the 2007 I-35W Bridge collapse in Minneapolis, Minnesota. The document was believed by some
74 parties to yield overly conservative gusset plate ratings (AASHTO 2013, NCHRP 2013).

75 It is also noteworthy that the application of the Whitmore section for tension failures has often led to
76 confusions in practice when the effective width crosses a connected edge, as illustrated in Figure 2
 4

77 for the gusset plate where the Whitmore effective width runs into the horizontal member. More
78 potential confusions have been described by Thornton & Lini (2011).

79 This paper aims to establish that the Whitmore criterion for the design of a bolted gusset plate under
80 tension is redundant provided the correct block shear check is performed. The Whitmore tension
81 capacity will first be compared algebraically against the block shear equation to articulate their
82 numerical relationship in terms of the connection geometry. Estimates of the ultimate test loads for
83 laboratory specimens given by the Whitmore and the block shear criteria will then be verified against
84 the laboratory test results. The specimens include those for which the Whitmore tension capacity is
85 much lower than the block shear capacity, and those that were considered by the testing researchers
86 to have failed by tension in the Whitmore net section.

87 Finite element analysis including fracture propagation will be presented to show that the fractures
88 across the Whitmore net section only took place long after the ultimate limit state of block shear had
89 passed, and well after the complete fracture of the net section within the block shear zone. In
90 addition, it will be demonstrated that the ultimate load of a bolted gusset plate failing in block shear
91 is often reached due to necking of the net tensile section, before the occurrence of fracture.

92 Additional supporting test data and analysis results are provided in the last three tables pursuant to
93 the comments of reviewers of the original manuscript. This paper is not concerned with the
94 Whitmore effective width for the design against buckling of a gusset plate under compression, which
95 has been shown to be grossly inaccurate (Cheng et al. 2000, Sheng et al. 2002). Astaneh-Asl (1998)
96 has also suggested that the Whitmore concept was intended for gusset plates in tension only.

97 Comparisons between the Whitmore section and the block shear mode

98 According to Equation (J4-2) of the specification (AISC 2016), the Whitmore tension capacity of the
99 bolted gusset plate in Figure 1(a) is equal to

Rn = Fu Ae = Fu Ww t
100
[ { }]
= Fu (nl − 1)(g − d h ) + 2(nr − 1) p tan 30o − d h t
(1)

101 in which Fu is the material tensile strength, Ae is the effective net area, Ww is the Whitmore net width,
102 t is the plate thickness, nl is the number of bolt lines in the direction of loading (equal to 2 in Figure
103 1), g is the gauge, dh is the bolt hole diameter, nr is the number of bolt rows perpendicular to the
104 loading direction (equal to 4 in Figure 1), and p is the pitch.
 5

105 Teh & Deierlein (2017) have provided the following block shear capacity

Rn = Fu Ant + 0.6 Fu Aev

106  ( ) 
= Fu (nl − 1)(g − d h ) + 1.2(nr − 1) p + e1 − 2nr − 1 d h  t
(2)
  4 

107 in which e1 is the end distance defined in Figure 3. The tensile and shear resistance planes in the
108 block shear mode are indicated in the figure. It should be noted that the shear resistance area Aev is
109 the mean between the gross and the net shear areas defined in the specification (AISC 2016).

110 Equation (2) has been demonstrated by Teh & Deierlein (2017) to be accurate and reliable through
111 verifications against 161 bolted gusset plate specimens tested by independent researchers around the
112 world. The first term in the first line of Equation (2) represents the tensile resistance, while the
113 second term is the shear resistance of the block. It should be noted that the use of the material
114 strength Fu in the shear resistance term does not account for shear fracture, but for shear yielding at
115 full strain hardening (Teh & Uz 2015).

116 From Equations (1) and (2), it can be derived that if the pitch p is equal to three times the bolt hole
117 diameter dh, and the end distance e1 is 1.5 times dh, then the Whitmore tension capacity will be equal
118 to the block shear capacity when the number of bolt rows nr is equal to 6.7:

( )
2(nr − 1)(3d h ) tan 30o − d h = 1.2(nr − 1)(3d h ) + (1.5d h ) − 2nr − 1 d h 
119  4  (3)
nr = 6.7

120 Therefore, the ultimate capacity of a typical gusset plate connection with approximately seven rows
121 of bolts may be accurately estimated using the Whitmore criterion even if it fails in block shear,
122 giving the false impression that the Whitmore tension section were valid.

123 If the connection has only a few rows of bolts, then the Whitmore tension capacity will be
124 significantly lower than the block shear capacity. This algebraic outcome enables the verification of
125 the Whitmore criterion against the laboratory test results of such specimens.

126 Table 1 shows the results of Equations (1) and (2) for the specimens tested by Aalberg & Larsen
127 (1999), which were known to have failed in block shear. The variable Fy is the (measured) yield
128 stress of the steel material, given here for the purpose of the finite element analysis discussed in the
129 next section. The variable Pt denotes the ultimate load obtained in the laboratory test, and the ratio
 6

130 Pt/Rn is known as the professional factor. An empty cell in the following tables indicates that the
131 entry in the cell above applies.

132 It can be seen from Table 1 that the ultimate test loads of the specimens with two rows of bolts were
133 up to 90% higher than the computed Whitmore tension capacity. The extent of underestimation
134 decreases with increasing number of bolt rows. In any case, it is clear that, had the Whitmore tension
135 fracture mode existed, the specimens would have failed at loads significantly lower than their actual
136 ultimate loads. The results presented in Table 1 is an unambiguous indication that the Whitmore
137 tension capacity does not exist.

138 Bjorhovde & Chakrabarti (1985) presented laboratory test results of bolted gusset plates that were
139 believed to have failed in the Whitmore section. Photographs of two specimens seem to indicate
140 fractures in the Whitmore zone similar to that illustrated in Figure 1(b), i.e. the tension fracture
141 extends into the outer Whitmore zone beyond the block shear perimeter indicated in Figure 3.
142 However, there are two points worth noting regarding this apparent indication. First, the number of
143 bolt rows nr in each specimen is equal to 9, so the Whitmore tension capacity is greater than the
144 block shear capacity. It will therefore be instructive to compare the estimates of Equations (1) and (2)
145 against each other, knowing that the latter governs (in contrast to Table 1). Second, it was not clear
146 whether the fractures in the outer Whitmore zone took place before the inner zone (the block shear
147 zone) completely fractured, or after it. The first point is investigated in this section, while the second
148 in the next.

149 Table 2 shows that the professional factors of Equation (2) are noticeably closer to unity compared to
150 Equation (1) for both specimens while all of them are less than or equal to unity, suggesting that the
151 specimens failed in block shear. It is therefore quite possible that the fractures in the outer Whitmore
152 zone took place after the tests were continued well beyond the respective block shear failures,
153 associated with fractures in the net tensile section of the block.

154 Finite element investigations

155 Finite element analysis is an excellent tool to investigate the tensile stress contours at the ultimate
156 limit state of a bolted gusset plate, and to corroborate the inference made in the preceding paragraph.
157 The use of the hexahedral reduced integration brick element C3D8R available in ABAQUS 6.12
158 Standard (ABAQUS 2012) also enables the demonstration that the ultimate load is due to (out-of-
159 plane) necking in the net tensile section, not fracture. It should be noted that this phenomenon and
 7

160 the associated gradual drop in resistance cannot be captured by the use of 2D elements such as that
161 employed by Wen & Mahmoud (2017).

162 In order to reduce the analysis time and minimise possible numerical precision errors, advantage was
163 taken of the symmetry of the bolted gusset plates studied in the present work. The modelling of
164 symmetry conditions for such plates has been described by Clements & Teh (2013). Furthermore,
165 each of the bolts was modelled as a 3D analytical rigid body revolved shell as their deformations had
166 little effects on the gusset plate’s failure mode. The bolts were displaced together to simulate loading
167 of the gusset plate as the displacement would be resisted by the surface contact between the bolt and
168 the bolt hole at the downstream end, in the same manner as conducted by Clements & Teh (2013).

169 However, unlike the work of Clements & Teh (2013), fracture initiation and propagation were
170 simulated in the present work in order to investigate the Whitmore section fracture postulated in
171 Figure 1(b) and apparently found by researchers in the literature. The simulations using
172 ABAQUS/Standard also enable the development sequence of fractures and their relationships to the
173 resistance level of the gusset plate be studied. Element deletion was activated when the maximum
174 degradation was reached at an integration point.

175 The present finite element models are validated in Table 1, where it can be seen that there are
176 excellent agreements in the ultimate loads between the test results and the FEA results. Modelling of
177 the stress-strain curve and the damage parameters of a certain specimen is described in the following
178 subsection, where validation of the load-deflection graph can also be found.

179 Specimen underestimated by the Whitmore criterion

180 As shown in Table 1, the ultimate test load of Specimen T-16 tested by Aalberg & Larsen (1999) was
181 27% higher than the computed Whitmore tension capacity. If the underestimation was simply a
182 numerical inaccuracy, then there would have been fractures in the outer Whitmore zone as illustrated
183 in Figure 1(b). However, such fractures are not evident in the photograph of the tested specimen
184 provided by Aalberg & Larsen (1999), even though the test was continued until the shear planes
185 fractured. The conditions of the specimen at the ultimate limit state and beyond are studied using the
186 present finite element analysis.
 8

187 The true stress-strain curve used in the analysis is plotted in Figure 4. Up to the horizontal portion,
188 the engineering stress-strain relationship was first defined using the Ramberg-Osgood power model
189 (Ramberg and Osgood 1943)
n

ε=
σ
+ 0.002
σ 
190  (4)
E
F  y

191 in which ε is the engineering strain, σ is the engineering stress and E is the Young’s modulus of
192 elasticity. The power term n is determined from

ln ( ε us 0.2 )
193 n= (5)
ln ( F u F y )

194 in which εus is defined as

195 ε us 100 ε u − ( F u E ) 

= (6)

196 The variable εu is the engineering strain at the ultimate stress. For Specimen T-16, it is assumed to be
197 10%.

198 Having defined the engineering stress-strain relationship as given by Equation (4), the true stress-
199 strain curve (up to the horizontal portion shown in Figure 4) was plotted from

200 ε=
true ln [1 + ε ] (7)

201 and

202 σ=
true σ [1 + ε ] (8)

203 The damage initiation parameters used in the present work are shown in Table 3, which have been
204 obtained by trial and error to reasonably match the response of the spliced (bolted) tension coupon
205 tested by Aalberg & Larsen (1999) and depicted in Figure 5(a). The equivalent plastic displacement
206 at failure is set to be 0.2. The calibration result is shown in Figure 5(b). It should be noted that the
207 elastic portion (initial stiffness) of the experimental curve has been adjusted to account for the fact
208 that there were small misalignments in the bolted coupon as no attempt was made by Aalberg &
209 Larsen (1999) to achieve perfect alignment of the bolt holes during the experiment. The plasticity of
210 the steel material was handled through the von Mises yield criterion and the Prandtl-Reuss flow rule
 9

211 with isotropic hardening. The elastic modulus is assumed to be 200 GPa, and the Poisson’s ratio is
212 0.3.

213 In the following Figures 6 through 8 for Specimen T-16, the mirror images of the symmetric-half
214 FEA models are added to facilitate illustration. Figure 6(a) shows the out-of-plane necking of the net
215 tensile section within the block shear zone at the ultimate limit state, which occurs at a simulated
216 load of 956 kN, or 0.5% lower than the ultimate test load obtained by Aalberg & Larsen (1999). It
217 can be seen from the out-of-plane displacement contours that necking is confined between the two
218 bolt holes only, not extending into the outer Whitmore zone despite the applied load being 27%
219 higher than the Whitmore capacity. The corresponding longitudinal normal stress contours are shown
220 in Figure 6(b).

221 Shear displacement of the “block” is also evident in Figure 6. The existence of the block is indicated
222 by the von Mises stress contours in Figure 7. It can be seen that the von Mises stresses around the
223 block shear perimeter are significantly higher than in the outer Whitmore zone, vindicating the use of
224 the block shear criterion rather than the Whitmore criterion despite the latter’s lower capacity. Shear
225 displacement of the block becomes more pronounced following the net tensile section rupture, as
226 shown in Figure 8(a).

227 Figure 8(b) shows that, even after the net tensile section of the block fractured completely, and the
228 block continued to shear, there is no evidence of necking or fracture in the outer Whitmore zone.

229 The load-deflection graph obtained by the finite element analysis can be compared to that obtained
230 by Aalberg & Larsen (1999) in Figure 9. States corresponding to Figures 6, 8(a) and 8(b) are
231 annotated along the curve.

232 As indicated in Figure 9, the ultimate block shear load is due to necking of the net tensile section, not
233 due to fracture. This point has been previously explained by Teh & Clements (2012). In fact, fracture
234 only took place after extended gradual softening of the response under quasi-static loading of the
235 high-strength steel specimens, as annotated in the figure.

236 Specimen showing apparent Whitmore fracture

237 As mentioned in the section “Comparisons between the Whitmore section and the block shear
238 mode”, two nine-row bolted gusset plates of Bjorhovde & Chakrabarti (1985) were shown in
 10

239 photographs (see Figure 10 for example) to fracture in the outer Whitmore zone. However, Table 2
240 shows that the ultimate test loads of both specimens, which were nominally identical to each other,
241 were closer to the block shear capacity given by Equation (2) than to the Whitmore tension capacity.
242 It should also be noted that all the computed capacities were on the same side of (un)conservatism.
243 The development of fractures in the specimens is studied in the present finite element analysis, which
244 incidentally models the gusset plate having its loading direction inclined at 45 degrees to the adjacent
245 member.

246 The true stress-strain curve used in the finite element analysis is plotted in Figure 11, employing the
247 procedure expressed by Equations (4) through (8) based on the assumption that the engineering strain
248 εu of the mild steel is 40%. Since Bjorhovde & Chakrabarti (1985) did not provide any coupon test
249 results, the damage initiation and damage evolution parameters obtained from the preceding
250 calibration against the test results of Aalberg & Larsen (1999) were used in the present analysis.

251 The simulated ultimate load is 690 kN, or 1.9% lower than the block shear capacity given by
252 Equation (2). Figure 12 shows that the drop from the ultimate load was less gradual compared to that
253 of Specimen T-16 tested by Aalberg & Larsen (1999), shown in Figure 9. However, as shown in
254 Figure 13, the ultimate load taking place at the displacement of 9.3 mm was still due to necking of
255 the net tensile section, although fracture was imminent for the 3.2 mm thick gusset plate with nine
256 rows of bolts.

257 Figure 14 shows the complete fracture of the net tensile section within the block when the
258 displacement is equal to 15.2 mm, as annotated in Figure 12. Even at this stage, there is no fracture in
259 the outer Whitmore zone. It is only when the displacement reaches 20.8 mm (corresponding to a load
260 of 603 kN) that fracture initiates in the outer Whitmore zone. Figure 15 shows the fracture at a
261 displacement of 24 mm.

262 The present FEA results, coupled with the comparison results shown in Table 2, clearly indicate that
263 the two gusset plate specimens of Bjorhovde & Chakrabarti (1985) did not fail in the Whitmore
264 section but in block shear.

265 Additional test data and modified Whitmore sections

266 Tables 4 and 5 compare Equations (1) and (2) for the specimens tested by Huns et al. (2002) and
267 Mullin (2004), respectively. It can be seen that the outcome is consistent with that for Table 1. All
 11

268 the ultimate test loads of Huns et al. (2002) were more than 25% higher than the Whitmore estimates
269 given by Equation (1), and were significantly closer to the block shear capacities given by Equation
270 (2).

271 For the specimens of Mullin (2004) which had six to eight rows of bolts, the Whitmore capacities
272 given by Equation (1) are close to the block shear capacities given by Equation (2), supporting the
273 finding of Equation (3) that the two equations will give similar results for connections with
274 approximately seven rows of bolts. For the specimen of Mullin (2004) which had two rows of bolts,
275 the ultimate test load exceeded the Whitmore capacity by 44%, but was only 7% higher than the
276 block shear capacity computed using Equation (2).

277 Table 6 presents the results of the modified Whitmore sections proposed by Irvan (1957), Chesson &
278 Munse (1963), Yamamoto et al. (1985), Cheng et al. (2000) and Dowswell (2013) for the specimens
279 listed in Table 1. It should be noted that not all of the cited authors necessarily referred to the
280 Whitmore tension section, and were in some cases concerned with the “dispersion angle” under
281 compression. In the case of Irvan (1957), the 30o lines are projected from the geometric centre of the
282 rivet group, resulting in an effective width that is as narrow as one tenth of the Whitmore width.
283 Readers should consult the references for details of the modified Whitmore sections.

284 In any case, it can be seen from the results given in Table 6 that there is no reliable method for
285 determining the dispersion angle of the Whitmore section, even if the section existed. For some
286 variants, the errors are even more severe than those obtained using the well-established dispersion
287 angle of 30o proposed by Whitmore (1952).

288 Conclusions

289 The Whitmore criterion has been used by structural engineers to determine the tension capacity of
290 connected steel plate elements. Some authorities suggested or still require that the design of a gusset
291 plate be checked against both the Whitmore and the block shear criteria.

292 Using simple algebra, the paper has shown that the Whitmore criterion only gives a similar result to
293 the (correct) block shear criterion under certain conditions. For a standard bolted connection
294 satisfying the AISC recommendations for the bolt spacing and the end distance, the two criteria
295 would lead to similar results if there are approximately seven rows of bolts.
 12

296 The Whitmore criterion has been shown in the paper to be excessively conservative for connections
297 having two or three rows of bolts. The ultimate test load obtained by Aalberg & Larsen (1999) for a
298 connection having two rows of bolts was 90% higher than that predicted by the Whitmore criterion.
299 If the Whitmore criterion were valid, such an outcome would not have been possible. The ultimate
300 test loads of the specimens tested by Aalberg & Larsen (1999) were accurately determined using the
301 block shear equation proposed by Teh & Deierlein (2017).

302 Conversely, for connections having nine rows of bolts tested by Bjorhovde & Chakrabarti (1985), the
303 Whitmore criterion overestimated the ultimate capacities even though the gusset plates were thought
304 by the researchers to have failed in the Whitmore section. The ultimate test loads were closer to the
305 block shear capacity, suggesting that the failure mode was block shear. The actual failure mode has
306 been confirmed through the finite element analysis presented in this paper to be block shear. The
307 finite element analysis has also shown that fractures in the Whitmore zone outside the block only
308 took place because the connection test was continued well beyond the ultimate limit state.

309 Additional test data and modified Whitmore sections have also been investigated, the results of
310 which confirm the conclusion that the Whitmore section check is not a viable criterion. The paper
311 has demonstrated that the Whitmore section check for the design of a bolted gusset plate under
312 tension is redundant provided the correct block shear check is performed.

313 By not requiring the Whitmore section check, the design of standard gusset plates having bolt rows
314 less than seven will be more economical. Furthermore, difficulties in applying the Whitmore section
315 check in geometries where the Whitmore section crosses into another member will be obviated.

316 Acknowledgment

317 This research has been conducted with the support of the Australian Government Research Training
318 Program Scholarship for the first author, administered by the University of Wollongong.
 13

319 References

320 AASHTO (2013) Proposed Changes to AASHTO Specifications, AASHTO Bridge Committee
321 Agenda Item, Technical Committee T-18 Bridge Management, Evaluation, and
322 Rehabilitation/ T-14 Steel, American Association of State Highway & Transportation
323 Officials, Washington DC.
324 ABAQUS (2012) ABAQUS Analysis User’s Manual, Version 6.12, Dassault Systèmes, Providence
325 RI.
326 AISC (2010) Specification for Structural Steel Buildings, ANSI/AISC 360-10, American Institute of
327 Steel Construction, Chicago IL.
328 AISC (2016) Specification for Structural Steel Buildings, ANSI/AISC 360-16, American Institute of
329 Steel Construction, Chicago IL.
330 Aalberg, A., and Larsen, P.K. (1999) Strength and Ductility of Bolted Connections in Normal and
331 High Strength Steels, Report N-7034, Dept. of Structural Engineering, Norwegian University
332 of Science and Technology, Trondheim, Norway.
333 Astaneh-Asl, A. (1998) Seismic Behaviour and Design of Gusset Plates, Structural Tips, Structural
334 Steel Educational Council, Moraga CA.
335 Birkemoe, P. C., and Gilmor, M. I. (1978) “Behavior of bearing-critical double-angle beam
336 connections.” Engineering Journal AISC, 15 (3), 109–115.
337 Bjorhovde, R., and Chakrabarti, S. K. (1985) “Tests of full-size gusset plate connections.” J. Struct.
338 Eng., 111 (3), 667-684.
339 Cheng, J. J. R., Grondin, G. Y., and Yam, M. C. H. (2000) “Design and behavior of gusset plate
340 connections.” Connections in Steel Structures IV, 22-25 October, Roanoke, VA.
341 Chesson, E. and Munse, W. H. (1963) “Riveted and bolted joints: Truss-type tensile connections.” J.
342 Struct. Div. ASCE, 89 (ST1), 67–106.
343 Clements, D.D.A., and Teh, L.H. (2013) “Active shear planes of bolted connections failing in block
344 shear.” J. Struct. Eng., 139 (3), 320-327.
345 Cunningham, T. J., Orbison, J. G., and Ziemian R. D. (1995) “Assessment of American block shear
346 load capacity predictions.” J. Construct. Steel Res., 35 (3), 323-338.
347 Dowswell, B. (2013) “Calculation of stress trajectories using fracture mechanics.” Engineering
348 Journal AISC, 50, 3-20.
349 FHWA (2009) Load Rating Guidance and Examples for Bolted and Riveted Gusset Plates in Truss
350 Bridges, Federal Highway Administration Publication No. FHWA-IF-09-014, National
351 Academy of Sciences.
 14

352 Gross, J. L., and Cheok, G. (1988) Experimental Study of Gusseted Connections for Laterally Braced
353 Steel Buildings, NISTIR 88-3849, National Institute of Standards and Technology,
354 Gaithersburg, MD.
355 Hardash, S. G., and Bjorhovde, R. (1985) “New design criteria for gusset plates in tension,”
356 Engineering Journal AISC, 22 (2), 77-94.
357 Higgins, C., Senturk, A. E., and Turan, O. T. (2010) “Comparison of block-shear and Whitmore
358 section methods for load rating existing steel truss gusset plate connections.” J. Bridge Eng.,
359 15 (2), 160-171.
360 Huns, B. B. S., Grondin, G. Y., and Driver, R. G. (2002) Block Shear Behaviour of Bolted Gusset
361 Plates, Structural Engineering Report No. 248, University of Alberta, Edmonton, AB.
362 Irvan, W. G. (1957) Experimental Study of Primary Stresses in Gusset Plates of a Double Plane
363 Pratt Truss, Engineering Research Station Bulletin No. 46, University of Kentucky, KY.
364 Kulak, G. L., Fisher, J. W. and Struik, J. H. A. (2001) Guide to Design Criteria for Bolted and
365 Riveted Joints, 2nd ed., American Institute of Steel Construction, Chicago, IL.
366 Liao, M., Okazaki, T., Ballarini, R., Schultz, A. E., and Galambos, T. V. (2011) “Nonlinear finite-
367 element analysis of critical gusset plates in the I-35W Bridge in Minnesota.” J. Struct. Eng.,
368 137 (1), 59-68.
369 Menzemer, C. C., Fei, L., and Srivatsan, T. S. (1999) “Design criteria for bolted connection elements
370 in aluminum alloy 6061,” Journal of Mechanical Design, 121 (9), 348-358.
371 Mullin, D. (2004) “Ductile gusset plates – Test and analyses,” Pacific Structural Steel Conference,
372 Long Beach, California, American Institute of Steel Construction.
373 NCHRP (2013) Guidelines for the Load and Resistance Factor Design and Rating of Riveted and
374 Bolted Gusset-Plate Connections for Steel Bridges, National Cooperative Highway Research
375 Program, The National Academies Press.
376 Rabern, D. A. (1983) Stress, Strain and Force Distributions in Gusset Plate Connections, MSc
377 thesis, The University of Arizona, AZ.
378 Ramberg, W., and Osgood, W. R. (1943) Description of Stress–Strain Curves by Three Parameters,
379 Technical Note No. 902, National Advisory Committee For Aeronautics, Washington, DC.
380 Richard, R. M., Rabern, D. A., Hormby, D. E., and Williams, G. C. (1983) “Analytical models for
381 steel connections,” Behavior of Metal Structures - Proceedings of the W. H. Munse
382 Symposium, American Society of Civil Engineers, 128-155.
383 Rosenstrauch, P. L., Sanayei, M., and Brenner, B. R. (2013) “Capacity analysis of gusset plate
384 connections using the Whitmore, block shear, global section shear, and finite element
385 methods.” Eng. Struct., 48, 543-557.
 15

386 Sheng, N., Yam, M. C. H., and Iu, V. P. (2002) “Analytical investigation and the design of the
387 compressive strength of steel gusset plate connections.” J. Construct. Steel Res., 58, 1473-
388 1493.
389 Teh, L. H., and Clements, D. D. A. (2012) “Block shear capacity of bolted connections in cold-
390 reduced steel sheets,” J. Struct. Eng., 138 (4), 459-467.
391 Teh, L. H., and Deierlein, G. G. (2017) “Effective shear plane model for tearout and block shear
392 failure of bolted connections.” Engineering Journal AISC, 54 (3), 181-194.
393 Teh, L. H., and Uz, M. E. (2015) “Block shear failure planes of bolted connections — Direct
394 experimental verifications.” J. Construct. Steel Research, 111, 70-74.
395 Thornton, W. A., and Lini, C. (2011) “The Whitmore section: How to use the Whitmore method for
396 tension and compression strength checks.” Modern Steel Construction, July 2011.
397 Topkaya, C. (2004) “A finite element parametric study on block shear failure of steel tension
398 members.” J. Construct. Steel Res., 60 (11), 1615-1635.
399 Yamamoto, K., Akiyama, N. and Okumara, T. (1985) “Elastic analysis of gusseted truss joints.” J.
400 Struct. Eng., 111 (12), 2545–2564.
401 Wen, H., and Mahmoud, H. (2017) “Simulation of block shear failure in bolted connections.” J.
402 Construct. Steel Res., 134, 1-16.
403 Whitmore, R. E. (1952) Experimental Investigation of Stresses in Gusset Plates, Bulletin No. 16, The
404 University of Tennessee, Knoxville, TN.
405 Williams, G. C. (1988) Steel Connection Design Based on Inelastic Finite Element Analysis, PhD
406 thesis, The University of Arizona, AZ.
407 Williams, G. C., and Richard, R. M. (1996) “Analysis and design of large diagonal bracing
408 connections.” Struct. Eng. Rev., 8 (1), 1-27.
Figure 1 The Whitmore section: (a) Geometric variables; (b) Whitmore fracture
Figure 2 Difficulty in the Whitmore section concept
Figure 3 Geometric variables of the block shear capacity
 1000
900

800

700
600
Stress (MPa)

500

400

300
200

100

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Strain

Figure 4 True stress-strain curve for Specimen T-16

Figure 5 Calibration of damage initiation parameters for Specimen T-16 (Aalberg & Larsen
1999)
 (a)

(b)

Figure 6 Contours of Specimen T-16 at the ultimate limit state: (a) Out-of-plane
displacements; (b) Longitudinal normal stresses
Figure 7 von Mises stress contours of Specimen T-16
Figure 8 Specimen T-16: (a) Fracture imminent; (b) No Whitmore fracture
Figure 9 Load-deflection graphs of Specimen T-16 (Aalberg & Larsen 1999) 
Figure 10 Specimen showing apparent Whitmore tension fracture (Bjorhovde & Chakrabarti
1985)

   
  

600

500

400
Stress (MPa)

300

200

100

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Strain
 

Figure 11 True stress-strain curve for Bjorhovde & Chakrabarti (1985)

Figure 12 Load-deflection graph for Bjorhovde & Chakrabarti (1985)
 

Figure 13 Necking at the ultimate limit state of Bjorhovde & Chakrabarti (1985)
Figure 14 Complete tensile fracture within the block only
Figure 15 Fracture in the outer Whitmore zone
Table 1. Comparison of Whitmore and block shear predictions for Aalberg & Larsen (1999)
Pt/Rn
e1 p g dh t nr nl Fy Fu
Spec Whitmore Block Shear
(mm) (mm) (mm) (mm) (mm) (MPa) (MPa) FEA
Eqn (1) Eqn (2)
T-1 50 60 65 21 8.4 2 3 373 537 1.58 1.07 1.05
T-3 1.55 1.05 1.03
T-2 7.7 786 822 1.54 1.05 1.02
T-4 1.51 1.02 1.00
T-7 38 47.5 47.5 19 8.4 2 2 373 537 1.90 1.07 1.05
T-8 7.7 786 822 1.79 1.01 0.99
T-9 8.4 3 2 373 537 1.40 1.04 1.02
a
T-15 1.32 0.99 1.03
T-10 7.7 786 822 1.31 0.98 0.96
a
T-16 1.27 0.95 1.00
T-11 8.4 4 2 373 537 1.18 1.00 0.98
T-12 7.7 786 822 1.11 0.94 0.92
Mean 1.46 1.01 1.00
COV 0.162 0.043 0.035
a
These I-section specimens had their flanges removed.

Table 2. Comparison of Whitmore and block shear predictions for Bjorhovde & Chakrabarti
(1985)

Pt/Rn
e1 p g dh t nr nl Fy Fu
Spec Whitmore Block Shear
(mm) (mm) (mm) (mm) (mm) (MPa) (MPa)
Eqn (1) Eqn (2)
30o 31.8 57.1 127 22.2 3.2 9 2 294 383 0.95 1.00
45o 0.89 0.94
Table 3. Ductile damage parameters

Stress Triaxiality Fracture Strain

0 2
0.20 1.25
0.45 0.5
0.50 0.15
0.57 0.5
0.68 1
1 2
Table 4. Comparison of Whitmore and block shear predictions for Huns et al. (2002)

Pt/Rn
e1 p g dh t nr nl Fy Fu
Spec Whitmore Block Shear
(mm) (mm) (mm) (mm) (mm) (MPa) (MPa)
Eqn (1) Eqn (2)
T1B 38 76 51 21 6.6 3 2 336 450 1.26 1.03
T1C 1.31 1.06
T1A 6.5 1.29 1.05
b
T2B 25 4 1.26 1.12
T2C 1.27 1.12
b
While the ultimate test load was cited in some places of the report to be 756 kN, it was given as 691 kN in page
137 of the report. An inspection of the load-deflection graph in page 43 confirms that the lower value is the
correct one.

Table 5. Comparison of Whitmore and block shear predictions for Mullin (2002)

Pt/Rn
e1 p g dh t nr nl Fy Fu
Spec Whitmore Block Shear
(mm) (mm) (mm) (mm) (mm) (MPa) (MPa)
Eqn (1) Eqn (2)
4U 38 76 51 21 6.8 2 2 317 415 1.44 1.07
8U 4 1.13 1.02
12U 6 1.00 0.97
14U 7 0.99 0.99
16U 8 0.93 0.94
Table 6. Comparison of Whitmore variants and block shear for Aalberg & Larsen (1999)

Pt/Rn

Spec Block Shear Whitmore Irvan Chesson & Munse Yamamoto et al. Cheng et al. Dowswell
Eqn (2) (1951) (1957) (1963) (1985) (2000) (2013)

T-1 1.07 1.58 15.8 1.75 1.87 1.15 1.16

T-3 1.05 1.55 15.4 1.71 1.82 1.12 1.13
T-2 1.05 1.54 15.5 1.72 1.83 1.13 1.13
T-4 1.02 1.51 15.1 1.67 1.78 1.10 1.11
T-7 1.07 1.90 4.45 2.27 2.55 1.17 1.18
T-8 1.01 1.79 4.19 2.14 2.40 1.10 1.11
T-9 1.04 1.40 5.84 1.70 1.93 0.83 0.84
T-15 0.99 1.32 5.50 1.60 1.82 0.79 0.79
T-10 0.98 1.31 4.63 1.44 1.65 0.70 0.70
T-16 0.95 1.27 4.37 1.36 1.55 0.66 0.66
T-11 1.00 1.18 5.52 1.60 1.82 0.79 0.79
T-12 0.94 1.11 5.31 1.54 1.76 0.76 0.76