Anda di halaman 1dari 16

Journal of Constructional Steel Research 102 (2014) 217–232

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Review

Local failures of coped steel beams — A state-of-the-art review


Michael C.H. Yam a,1, Cheng Fang b,⁎,2, Angus C.C. Lam c,3, J.J. Roger Cheng d,4
a
Department of Building & Real Estate, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
b
School of Civil Engineering and Geosciences, Newcastle University, Newcastle upon Tyne NE1 7RU, United Kingdom
c
Department of Civil & Environmental Engineering, University of Macau, Macau
d
Department of Civil & Environmental Engineering, University of Alberta, Edmonton, Alberta, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Coped (notched) steel beams are widely used in steel frame structures. A state-of-the-art review on local failures
Received 9 April 2014 of coped steel beams is presented with the main aim of providing a clear insight into the up-to-date research out-
Accepted 5 July 2014 comes and design approaches on the issues of local web buckling, block shear, and fatigue. With a comprehensive
Available online 1 August 2014
review of the available research database, the rationale behind the current design treatments for these failure
modes is evaluated. For the design of local web buckling for coped beams with bolted end connections, adopting
Keywords:
Coped (notched) steel beam
the lowest value of the resistances governed by elastic local web buckling, flexural yielding, and shear yielding
Local web buckling can be safe. A reduction factor of 0.9 on the design local web buckling resistance may be applied for the case of
Block shear welded connections due to the possible interactive effect of block shear. Use of the AISC–LRFD equations is sug-
Fatigue gested for the block shear design of coped beams with single-line bolted connections, and both the AISC–LRFD
Design and the CSA-S16-09 equations are suitable for the case of double-line bolted connections. For coped beams
Review with welded connections, the use of a reduced Ut of 0.7 in the CSA-S16-09 equations is recommended. For fatigue
design, use of a new category curve modified from the relevant equations in AISC–LRFD is proposed in this paper
to ensure a safe and consistent design. Illustrative design examples are presented based on the recommended
design methods. The expected future work is also outlined.
© 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2. Local web buckling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.1. Top flange coped beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.2. Double-coped beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
2.3. Reinforcing strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
3. Block shear. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
3.1. Coped beams with bolted connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
3.2. Coped beams with welded connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
3.3. Current design specifications and comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
4. Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
5. Comments, summary, and future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
6. Recommended design examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
. Illustrative Example 1 (bolted connections) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Local web buckling design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Block shear design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
. Illustrative Example 2 (welded connections) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Local web buckling design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

⁎ Corresponding author. Tel.: +44 191 208 7099.


E-mail address: cheng.fang@ncl.ac.uk (C. Fang).
1
Associate Professor.
2
Lecturer.
3
Assistant Professor.
4
C W Carry Professor.

http://dx.doi.org/10.1016/j.jcsr.2014.07.002
0143-974X/© 2014 Elsevier Ltd. All rights reserved.
218 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

Block shear design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231


. Illustrative Example 3 (fatigue design) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

1. Introduction behind experimental evidence. Equations have been proposed by early


researchers based on limited test data [1] for evaluating the local web
In steel structures, the identical elevation is often required for the buckling strength and fatigue strength of coped beams. The design
flanges of both secondary beams and primary beams to satisfy architec- guidance for block shear failure is currently available in major design
tural purposes at member intersections. In this case, the secondary standards, but some of these are partially based on earlier block shear
beams are usually coped (notched) at the top, bottom, or at both flanges tests on other types of connection (e.g. gusset plate). With newly
to avoid interference between the connected structural members, so emerging test data on the block shear failure of coped beams with
that sufficient clearance can be provided at the connections. From the bolted and welded connections, it is beneficial to re-evaluate the reli-
structural response perspective, however, the presence of copes ex- ability of the latest versions of these standards. The main aim of this
poses the beams to the risk of several types of local failure that are nor- paper therefore is to offer a comprehensive and critical review of the
mally not considered for an uncoped beam section. Firstly, due to the latest research outcomes on the local failures of coped beams and
reduced bending and torsional stiffness, local web buckling could preventive design methods, so as to benefit the communities of both
occur near the top coped edge (as shown in Fig. 1(a)) which is usually researchers and designers. The test database collected in this paper
under compression if the connection at the coped end has insignificant also facilitates future reliability analysis towards safer and making
rotational rigidity (and thus can be seen as a pinned connection). A high more rational designs for coped beams. For each failure mode, various
stress concentration near the cope corner due to the geometric discon- design methods are examined below, full comparisons with the avail-
tinuity could lead to the occurrence of local buckling after initial yield- able test data, and available reinforcement strategies for mitigating
ing, i.e. inelastic local web buckling. Secondly, due to the lack of local failures are discussed. Possible interactions between local web
restraint offered by the top flange, block shear failure can be a common buckling and block shear failure, which could lead to unsafe design,
type of coped beam failure, as shown in Fig. 1(b). This failure type is pos- are also identified. After the review, design examples are provided and
sible for coped beams with both bolted and welded connections. Third- future research directions are also highlighted. The symbols (defining
ly, when the connection at the cope end has a large rotational rigidity, a the coped beam geometry and details) frequently used in the following
negative moment can be induced and the cope edge is thus subjected to discussion are illustrated in Fig. 1(d).
tension. This could cause fatigue cracks, as shown in Fig. 1(c), if the
coped beams are used in bridges as stringers, where a cyclic load in- 2. Local web buckling
duced by passing vehicles is continuously applied during the whole
bridge service life. Another possible fatigue crack development scenario When the top flange of a coped beam is effectively restrained against
at the coped region is due to induced cyclic compressive stress in the lateral movement, local web buckling can be one of the failure modes
presence of a tensile residual stress (usually caused by the frame cut within the coped region, particularly when the cope length/depth is
procedure when producing the cope). large and the web thickness is small. Research studies of bolted beam
While the presence of beam copes inevitably increases the vulnera- end connections in the late 1970s and early 1980s [2–6] revealed that
bility of structures to local failure, relevant design guidance seems to lag local web buckling of coped beams in the coped region occurred

a b c

Web
buckling

Fig. 1. Common local failure modes and symbols for coped beams: a) local web buckling, b) block shear, c) fatigue cracking, d) symbols.
M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 219

subsequent to the development of either fracture or significant defor- where c = cope length. Since the classical plate buckling model did not
mation of the web around the bottom bolt line in the connection. consider the effects of stress concentration, shear stress, cope depth, and
Milek [7] warned that the reduced strength and stiffness of the coped moment variation (from the beam end to the end of the cope), a single
region should be carefully considered in the design of coped steel adjustment factor f was applied to the equation to conservatively con-
beams. Generally, secondary beams are coped at one or both flanges sider all these effects:
to avoid interference at member intersections, and different responses
may be expected for the two conditions, i.e. top flange only coped f ¼ 2c=D for c=D ≤1; or f ¼ 1 þ c=D for c=DN1 ð3Þ
beams and double coped beams, cases which are discussed in the
sections below. where D = beam depth. For practical design, it was required that the
critical stress σcr calculated from Eq. (1) should not exceed the design
2.1. Top flange coped beams bending stress calculated from conventional beam theory (known as
M/Z, where M = bending moment and Z = elastic section modulus).
A series of FE analysis was first performed by Cheng et al. [8] and The above design equations have been adopted by the AISC Steel Con-
Cheng and Yura [9] to investigate the local buckling behaviour of top struction Manual [10]. The SCI and BCSA publication ‘Joints in Simple
flange coped beams. Design recommendations were proposed based Construction: Simple Connections’ [11] also recommends this equation
on their parametric study. It was observed that when local buckling for a local buckling check, but it also states that local web buckling
occurred, lateral movement was considerable near the coped top edge, checking is not required provided that the cope depth and length are
whereas limited lateral movement was induced near the bottom flange. less than the stipulated limit, e.g. local web buckling checking is not re-
Considering this deformation mode, a classical plate bucking model quired when dc b D/2 and c b D, for D/tw b 48 for S355 steel where dc is
with simplified boundary conditions was proposed to predict the local the cope depth and tw is the beam web thickness as illustrated in
buckling response of the coped region, as reproduced in Fig. 2(a). It Fig. 1(d).
was assumed that the top edge of the plate was free, and the other It is possible that under certain conditions, e.g. when the d/tw ratio is
three edges, i.e. the bottom flange, the end of the beam, and the vertical low and the stress concentration is significant, the reduced section of a
line dividing the coped and uncoped regions, were simply supported. A coped beam can fail by inelastic local buckling. It was observed from the
linear stress distribution with a maximum value of σcr at the top of the parametric study [8] that the routine check of the bending stress and
cope and zero at the bottom was assumed. The critical stress was given shear stress could always offer a lower bound. In other words, explicitly
by: checking of inelastic buckling was not required, provided that flexural
yielding, shear yielding, and elastic buckling (Eq. (1)) were checked.
 2
π2 E tw To verify the reliability of the proposed design equations, ten tests on
σ cr ¼ k   f ð1Þ
12 1−v2 ho top flange coped beams were conducted by Cheng and Yura [9]. The
first five hot-rolled specimens (three with bolted angle cleats and two
where E = Young's modulus, ν = Poisson's ratio, tw = web thickness, with end-plate connections) were designed such that inelastic buckling
ho = beam depth at coped region, k = plate buckling coefficient given could govern the failure mode, while the remaining five (three hot-
by: rolled profile and two plate girders with end-plate connections) were
designed to fail by elastic buckling. It was generally found that for the
1:65
k ¼ 2:2ðho =cÞ for c=ho ≤1; or k ¼ 2:2ho =c for c=ho N1 ð2Þ case of elastic local web buckling, the overall response (i.e. load-
vertical deflection response) was almost linear until the occurrence of
local web buckling (after which the load started to drop quickly). For
those specimens which failed by inelastic buckling, on the other hand,
a nonlinear load-vertical deflection response, which indicates yielding,
was evident before buckling. The strain gauge readings also showed sig-
nificant stress concentration near the cope corner. The basic properties
of the specimens used for the group of elastic buckling tests are listed in
Table 1, while those which failed by inelastic buckling are summarised
in Table 2. All the geometric properties and calculated results listed in
the tables relate to, where available, the measured dimensions and
strength reported in the literature. The test-to-predicted ratios are
discussed below in this section.
The calculation model proposed by Cheng and Yura [9] is based on
the theory of plate bucking. From four more elastic local web buckling
tests and subsequent FE analysis on top flange coped beams, Yam
et al. [12] noticed that the buckling line in the web was usually oriented
at nearly 45° from a vertical line, particularly for those specimens (and
FE models) with relatively small ratios of c/ho. This indicates that
‘plate shear buckling’ might more realistically reflect the actual buckling
behaviour of the coped region. Attempting to seek an alternative solu-
tion, Yam et al. [12] proposed a new calculation model derived from
the perspective of plate shear buckling, and the critical shear stress τcr,
therefore, was considered to be the governing design parameter, as
given by:
 2
π2 E tw
τcr ¼ ks   ð4Þ
12 1−v2 ho

Fig. 2. Design models for local web buckling of coped beams (Cheng et al. [8]): a) top where ks is the shear buckling coefficientks = a(ho/c)b, and a = 1.38 −
flange coped beam, b) double flange coped beam. 1.79dc/D, b = 3.64(dc/D)2 − 3.36dc/D + 1.55. It was required in design
220 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

Table 1
Summary of elastic local web buckling tests.

Specimen code Coping details (mm) Rtest(kN) Ry (kN) Rvy(kN) Rcheng(kN) Ryam (kN) Rtest Rtest Note
Rcheng Ryam

Cheng et al. [8] RB12B c = 304.8, dc = 28.7 123.7 121.2 331.2 93.0 118.9 1.33 1.04 EP
RB12C c = 450.7, dc = 25.4 74.7 84.0 335.2 51.9 71.6 1.44 1.04 EP
RB12D c = 454.6, dc = 79.5 57.4 54.5 263.2 40.9 61.5 1.40 0.93 EP
PB26Aa c = 332.7, dc = 28.7 62.7 356.3 510.4 30.4 37.2 2.06 1.69 EP
PB26Ba c = 207.3, dc = 23.1 92.5 572.6 514.8 62.5b/64.6c 62.5b/77.0d 1.48b 1.48b EP
Yam et al. [12] 406d005 c = 342.9, dc = 19.9 165.9 224.1 479.2 135.1 164.9 1.23 1.01 EP
406d01 c = 342.9, dc = 39.8 161.1 202.1 454.0 126.5 152.4 1.27 1.06 EP
406d03 c = 342.9, dc = 119.4 111.9 127.0 353.1 101.3 108.7 1.10 1.03 EP
457d02 c = 388.0, dc = 90.0 268.9 221.0 541.8 210.1 244.8 1.28 1.10 EP

Mean= 1.40 1.15


CoV= 0.195 0.219

Note:
a
Plate girder with the h/tw ratio of 197.0,
b
Failure governed by elastic web shear buckling away from copes,
c
Capacity predicted by Cheng's method,
d
Capacity predicted by Yam's method, EP = End-plate connection.

practice that the critical shear stress τcr calculated from Eq. (4) should on the critical bending stress σcr proposed by Cheng and Yura [9], the
not exceed the design shear stress calculated from conventional beam critical shear stress τcr proposed by Yam et al. [12], the bending yield
theory (known as V/twho, where V = shear force, and other symbols strength, and the shear yield strength are given for comparison. Based
have been illustrated in Fig. 1(d)). Details of the relevant test specimens on these design approaches, the calculated reactions are designated as
and the local buckling resistances obtained (ultimate reaction Rtest) are Rcheng, Ryam, Ry, and Rvy, respectively. The design capacities of specimens
given in Table 1. which fail by elastic buckling are expected to be governed by Rcheng or
Aalberg and Larsen [13] conducted six tests to study the local buck- Ryam, while the design capacities of the specimens that failed by inelastic
ling response of top flange coped beams, where one test was on an buckling are expected to be governed by Ry or Rvy based on simple beam
uncoped beam and the remaining five were on coped beams with theory. It can be found from Table 1 that local web buckling governs the
different coping details. Instead of using actual end-plate or angle design resistance of all the specimens listed. Using Cheng's method, the
cleat connections at the coped beam end, these specimens were tested test-to-predicted ratios obtained ranged from 1.10 to 2.06. The mean
as simply supported beams with beam ends seated in a small steel test-to-predicted ratio was 1.40 with an associated coefficient of varia-
block supported by a cylindrical bearing which provided free rotation. tion (CoV) of 0.195. The design reactions obtained from Yam's method
The test results showed that the uncoped beam S1 and the coped were closer to the test results, where the test-to-predicted ratios ranged
beam S2 with the smallest cope length (c = 100 mm) failed by com- between 0.93 and 1.69, and with a mean test-to-predicted ratio of 1.15
pressive web buckling at the supported region. The remaining four with associated CoV of 0.219. It was noticed that both methods tend to
specimens, S3–S6, failed by inelastic local web buckling. It is worth be overly conservative for the two thin web girder specimens, where
mentioning that compressive web buckling may be less likely to occur the web depth to thickness ratio is large. For the remaining specimens,
in actual situations, where the vertical free edge of the web is connected Yam's method led to good agreement between the test and predicted
to the supporting primary beam/column via connections, rather than results, while Cheng's method could be slightly conservative. According
the test case where the bottom flange was just simply seated on a sup- to the literature reporting these test data, all the specimens shown in
port. Nevertheless, the test specimens S3–S6 extended the test database Table 2 failed by inelastic local web buckling. For specimen RB12A, how-
for the study of local web buckling. Through comparisons with the ever, the design reaction was governed by the critical reaction obtained
uncoped specimen, it was concluded that, at most, half of the load car- from Cheng's method. This indicates that the failure mode of this
rying capacity could be lost for the largest practical cope dimensions. specimen might fall between the cases of elastic buckling and inelastic
All the available test data are summarised in Table 1 for the cases of buckling. For the other specimens, failure was governed by either bend-
elastic local web buckling and Table 2 for the cases of inelastic local web ing or shear yielding. The test-to-predicted ratios ranged between 0.88
buckling. For both tables, the ultimate beam reactions calculated based and 1.36, with a mean value of 1.20 and associated CoV of 0.142.

Table 2
Summary of inelastic local web buckling tests.

Specimen code Coping details (mm) Rtest(kN) Ry(kN) Rvy(kN) Rcheng(kN) Ryam(kN) Rtest Note
Rdesign

Cheng et al. [8] W-1 c = 88.9, dc = 31.75 512.6 1021.7 510.0 2635.5 2018.6 1.01 DAB
W-2 c = 165.1, dc = 31.75 496.2 550.0 510.0 949.0 884.6 0.97 DAB
W-3 c = 254.0, dc = 31.75 440.6 357.8 510.0 466.2 498.2 1.23 DAB
RB18A c = 457.2, dc = 77.7 206.9 161.5 454.8 169.7 196.8 1.28 EP
RB12A c = 226.1, dc = 31.75 168.7 157.5 320.0 133.8 158.4 1.26 EP
10-4 c = 203.2, dc = 29.7 209.2 141.5 281.7 230.5 248.8 1.47 DAB
10-7 c = 127.0, dc = 44.45 262.6 198.9 267.4 451.1 369.4 1.32 DAB
18-14 c = 203.2, dc = 38.1 670.6 612.8 623.0 1763.2 1619.6 1.09 DAW
18-15 c = 127.0, dc = 38.1 730.7 980.3 623.0 3829.1 2981.2 1.17 DAW
Aalberg & Larsen [13] S3 c = 200.0, dc = 110.0 243.0 198.4 380.2 269.6 221.8 1.22 SS
S4 c = 300.0, dc = 110.0 166.0 124.9 380.2 169.8 159.9 1.33 SS
S5 c = 200.0, dc = 70.0 254.0 287.3 460.7 351.8 290.7 0.88 SS
S6 c = 200.0, dc = 150.0 170.0 124.8 299.8 215.0 156.3 1.36 SS

Mean= 1.20
CoV= 0.142

Note: DAB = Double angle cleat bolted connection, DAW = Double angle cleat welded connection, EP = End-plate connection, SS = Simply-supported.
M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 221

In practice, to precisely distinguish the elastic and inelastic web It should be noted that in Eq. (7), the unit of σy is ksi, and 1 ksi =
buckling modes in the absence of advanced analysis (e.g. FE analysis) 6.89 MPa. To date, no physical test data is available to validate the
or tests may be difficult because of the effects of stress concentration above design recommendations. Therefore, further experimental work
near the cope corner. A design strategy that uses the lowest value on double-coped beams is desirable.
of Rbuckling (either Rcheng or Ryam), Ry, and Rvy for the local web buckling
design of top flange coped beams would cover both buckling modes, 2.3. Reinforcing strategy
and it would seem to be safe for a wide range of practical steel beam
profiles and cope details (the cope details with c ≤ 1.5D and dc ≤ 0.5D As beam capacities can be significantly reduced by the cope, it may
were covered by the available test database). For elastic local web buck- be necessary to reinforce the web in the coped region under certain con-
ling, Yam's method, which is a modified plate shear buckling model, ditions. The most straightforward way would be to increase the mem-
predicts the critical resistance reasonably well, whereas Cheng's meth- ber section size. Alternatively, local reinforcement can be applied in
od can be slightly conservative. Cheng et al. [8] believed that even a the coped region to effectively prevent local buckling. Following FE
slight rotational stiffness provided by the connections at the coped analysis, three types of reinforcement, i.e. types A to C as shown in
beam end could be beneficial to the local web buckling capacity, and Fig. 3(a), were first proposed by Cheng et al. [8]. It was preliminarily
this might, in part, explain the conservative predictions (noting that suggested that for hot-rolled sections, where d/tw (d is the net depth
Cheng's method was initially proposed based on numerical results of the web for the uncoped section) is normally less than 60, reinforce-
where connection rigidities were not considered). ment types A or C could be sufficiently effective, provided that LRx is
larger than dc. For plate girders where the ratio of beam depth to web
2.2. Double-coped beams thickness can be excessively large (thin web), e.g. d/tw N 60, reinforce-
ment type B with LRx ≥ c/3 was recommended. Satisfying the above re-
Double-coped beams are usually used at beam-to-beam intersec- inforcement arrangement details, requires only material yield checking
tions where both top and bottom flanges of the secondary beam are for the coped region based on conventional beam theory. This work has
coped to maintain the same top and bottom flange elevations for floor been adopted by the AISC Steel Construction Manual [10].
and ceiling support, respectively. The local web buckling strength of To provide test evidence, Yam et al. [15] first conducted three full
double flange coped beams has been studied numerically by Cheng scale tests on top flange coped beams with reinforcement type A pro-
et al. [8] following which, design recommendations were proposed. A posed by Cheng et al. [8]. Among the three tests, two specimens were
simple lateral–torsional plate buckling model was used, with an considered with different longitudinal stiffener lengths (c + LRx) and
unbraced length c and a linearly increasing moment diagram as compared with an unreinforced specimen. It was shown that the maxi-
shown in Fig. 2(b), to deal with the local buckling problem for double- mum beam end resistance was improved by the applied stiffeners, but
coped beams. The critical moment Mcr was determined by web crippling, as shown in Fig. 3(b), prevented the full development
of cross-section yielding. In addition, it was found that the stiffener
πqffiffiffiffiffiffiffiffiffiffiffi
Mcr ¼ f d EIy GJ ð5Þ extension length of LRx had no obvious influence on local web buckling
c capacity provided that the minimal stiffener extension length LRx equal
to dc was ensured. Ten more full scale tests were subsequently conduct-
where c = cope length, E = elastic modulus, Iy = hot3w/12 (ho is the ed by Yam et al. [16] on top flange coped beams to investigate reinforce-
depth of a beam at coped region; tw is the web thickness), G = shear ment types A and B and a new reinforcement type D as shown in
modulus, J = hot3w/3, fd = an adjustment factor taken as fd = 3.5 − Fig. 3(a). Among these, two specimens without reinforcement failed
7.5(dc/D) to account for other parameters such as stress concentration, as expected by local web buckling, and the failure modes for the remain-
cope depth, and shear stress. Eq. (5) can be also expressed in terms of ing eight reinforced specimens varied. For the specimens with longitu-
critical bending stress σcr = 0.62πE(t2w/cho)fd. dinal stiffeners only (type A), flexural yielding of the full beam section
The proposed equations are applicable to copes with a length less at the location of maximum bending moment occurred first, which
than twice the beam depth (c ≤ 2D) and a depth less than one-fifth of was followed by web crippling. For the specimens with combined longi-
the beam depth (dc ≤ 0.2D). For practical design, it was suggested that tudinal and transverse stiffeners (types B and D), flexural yielding of the
the calculated value of Mcr should not exceed the flexural yield moment full beam section occurred first, but web crippling was avoided due to
of the coped region, and in addition, yielding in shear of the coped the presence of vertical stiffeners. As the load increased, local buckling
region should also be checked. Similar to the case of top flange coped of the compressive flange near the loading position was observed, as
beams, inelastic buckling strength in double-coped beams could be shown in Fig. 3(b). A summary of the available tests on reinforced
conservatively covered by the conventional bending and shear stress coped beams is provided in Table 3, which includes the coping details
calculations at the reduced section against material yielding. Compari- and main test results (ultimate reaction and failure mode). The ultimate
sons with numerical study results, found the design recommendation reaction ratios of the reinforced specimens over their unreinforced
to be conservative. counterpart, which indicate the effectiveness of the reinforcements,
The AISC Steel Construction Manual [10] employs the same formulae are also shown in the Table.
for the local web buckling design of double-coped beams with the cope The experimental work was followed by a comprehensive numerical
details satisfying c ≤ 2D and dc ≤ 0.2D. For the beams with cope dimen- study [17], where the structural response predicted by the FE models
sions beyond this limit, a conservative procedure based on classical agreed well with the test results. A parametric study was performed to-
plate buckling theory, later simplified by Muir and Thornton [14], can wards practical recommendations for reinforcements which best cater
be used: to various cope details. Two reinforcement types, A and D, were recom-
mended for hot-tolled coped beams for most of the available sizes. Rein-
σ cr ¼ σ y Q ð6Þ forcement type A with extension length LRx = 2dc was recommended
for coped beams with d/tw ≤ 52.7, dc/D ≤ 0.3 and c/D ≤ 0.75, and the re-
Where σy = yield strength, Q = 1 for λ ≤ 0.7; Q = 1.34–0.486 λ for inforcement type D with the extension length LRx = dc and LRy = dc for
0.7 b λ ≤ 1.41; Q = 1.30/λ2 for λ N 1.41, and coped beams with 52.7 b d/tw ≤ 57.1, dc/D ≤ 0.3 and c/D ≤ 0.9. For cope
details beyond the above ranges, e.g. for thin web thickness girders with
pffiffiffiffiffiffi
ho σ y d/tw N 57.1, the need for stricter reinforcement details was suggested,
λ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7Þ
but no further details were given. The recommendations proposed by
10t w 475 þ 280ðho =cÞ2
Yam and Chung [17] did not include the use of reinforcement types B
222 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

Web
crippling

Flange
buckling

Fig. 3. Reinforcement of coped beams against local web buckling: a) reinforcing strategies, b) possible failure modes of reinforced coped beams (Yam et al. [16]).

and C and neither was there any information or discussion on the rein- shown in Fig. 4(a). End distance is defined as the horizontal distance be-
forcement details for double-coped beams. tween the vertical beam edge to the first bolt line, and edge distance is
defined as the vertical distance between the horizontal coped edge to
3. Block shear the first bolt row, as illustrated in Fig. 4(a).
The block shear type of failure in coped beams was first revealed in a
Block shear failure can occur within various structural details includ- double angle steel beam connection test conducted by Birkemoe and
ing steel gusset plates, angles, and coped beams. The failure pattern nor- Gilmor [2]. This test was initially intended to study the allowable bolt
mally involves tensile fracture on a plane transverse to the loading bearing capacity of steel beams and to evaluate the effect of the cope
direction combining shear failure (yield or fracture) on a plane parallel that might potentially reduce the bearing capacity of the bolt hole due
with the loading direction, resulting in a fully or partially torn out of a to the removal of the constraint provided by the top flange. At ultimate
block of material. Although block shear can occur in both bolted and load, prior to attaining the nominal ultimate bearing stress, the web ma-
welded connections, relevant research on the former type, because of terial at the location of the lowest bolt was significantly torn with neck-
reduced net area concerns was initiated much earlier. In this section, ing followed by fracture. Local web buckling was also observed at the
the study of block shear failure in coped beams only is discussed. ultimate state. The block shear failure mode observed in that study
Block shear in other types of members, e.g. gusset plates and tension an- raised significant research interest. A preliminary design equation,
gles, which have been well reported in [18–25] and by others are not which included the parameters, the sum of the ultimate tensile strength
covered in this paper. over the net tension area and the ultimate shear strength over the net
shear area, was first proposed for calculating the block shear capacity
3.1. Coped beams with bolted connections of coped beams.
Following the work of Birkemoe and Gilmor [2], Yura et al. [5] per-
It is necessary to unify the definitions of bolt row/line orientations formed eight tests on top flange coped beams with bolted double
and end/edge distances, which are usually used inconsistently among angle connections, where six tests included standard bolt holes and
different researchers. In this paper, bolt lines and rows are defined as two tests using slotted holes. Both single and double bolt line arrange-
the vertical line A-A and horizontal line B-B, which are parallel and per- ments were taken into account. All the coped beam specimens failed
pendicular to the line of action of the reaction force, respectively, as by block shear, except for one specimen which failed by general web

Table 3
Summary of tests on reinforced coped beams.

Specimen code Coping details (mm) c/D dc/D Rtest Reinforced/unreinforced capacity ratio Reinforcement type Local failure mode
(kN)

Yam et al. [15] 406d03sa c = 343.0, dc = 119.0 0.87 0.30 172.1 1.54 Type A Web crippling
406d03sb c = 334.0, dc = 120.0 0.84 0.30 173.2 1.55 Type A* Web crippling
Yam et al. [16] A2 c = 207.0, dc = 59.7 0.59 0.17 305.6 1.51 Type A Web crippling
A3 c = 210.0, dc = 62.0 0.60 0.18 329.0 1.63 Type A Web crippling
A4 c = 202.8, dc = 64.3 0.58 0.18 327.5 1.62 Type B Flange local buckling
A5 c = 209.7, dc = 64.3 0.60 0.18 340.3 1.69 Type D Flange local buckling
B2 c = 207.2, dc = 105.0 0.59 0.30 293.9 1.97 Type A Web crippling
B3 c = 315.3, dc = 105.0 0.90 0.30 240.7 1.61 Type A Web crippling
B4 c = 212.3, dc = 107.7 0.61 0.31 318.8 2.13 Type B Flange local buckling
B5 c = 209.3, dc = 106.3 0.60 0.30 333.0 2.23 Type D Flange local buckling

Note: * shorter stiffener length with a clearance equal to dc at beam end.


M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 223

Fig. 4. Block shear failure of coped beams: a) definition of bolt layout, b) definition of eccentricity (Fang et al. [35]), c) design model for coped beams with welded connections (Yam et al.
[38]).

yielding. The main finding was that increasing the end distance im- greater portion of the reaction contributed to by the tension plane. Com-
proved the block shear capacity. As a follow-up to the study of Yura bining the FE analysis results with the test observations, a new design
et al. [5], Ricles and Yura [6] reported four more tests on top flange model was proposed, as given by:
coped beams with double bolt lines carried out to further study the ef-
fects of end and edge distances and bolt layouts. The new test results P u ¼ 0:5f u Ant þ 0:6 f y Agv ð8Þ
showed that all the specimens failed in block shear with fracture
along the entire tension plane and significant yield along the shear where Pu = block shear capacity (unfactored), fu = ultimate tensile
plane. The hole deformations along the tension plane of fracture were strength, fy = yield strength, Ant = net tension plane, Agv = gross
much more significant than for the remaining holes, which indicated shear plane. This equation indicates that the block shear capacity is
that the loads resisted by the bolts were not equal. Local web buckling the sum of the fracture strength of the net tension plane assuming a tri-
was also observed in some of the tests, but this only occurred after angular stress distribution and the full yield strength of the gross shear
web fracture, and was therefore not considered to have affected the plane. This equation formed an important basis of the more reasonable
ultimate block shear capacity. In addition, increasing the end distance design models employed later in the North American Standards. The
was more effective in improving the block shear capacity than increas- above test results were summarised by Kulak and Grondin [26,27],
ing the edge distance. Elastic FE analysis was subsequently performed to and compared with the relevant design specifications (US, Canadian,
obtain a clearer insight into the elastic stress distribution within the European, and Japanese standards) used at that time. Inconsistent com-
web, particularly along the tensile and shear planes. It was shown that parisons were generally observed among the different specifications.
increasing the end distance could change the tensile stress distribution Eq. (8) proposed by Ricles and Yura [6], because it led to good compar-
along the tension plane from triangular (the maximum stress was at the isons, was recommended by Kulak and Grondin [26,27].
beam end) to bilinear (the maximum stress was under the outer bolt), Aalberg and Larsen [28] performed eight block shear tests on four
and that the shear stress distribution along the shear plane was less in- sets of coped beams, where one normal steel beam web and one high
fluenced by the bolt layout. The influence of angle connection flexibility strength steel beam web were tested for each set. All specimens
was also studied. The more flexible angle connection could lead to a underwent block shear failure at the end of the tests, and it was noticed
224 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

that the ultimate load was reached before the occurrence of tensile frac- All the above studies focused on coped beams with double clip angle
ture for the high strength steel specimens, whereas tensile fracture oc- bolted connections, in the absence of any information on those with
curred during the load increasing stage for the normal strength steel single-sided connections using tees or single angles. The main differ-
specimens. The deformations at ultimate load for the normal steel ence between the two connection types is the out-of-plane loading
specimens were generally larger than those for the high strength steel eccentricity induced between the centreline of the beam web and the
specimens. The test-to-predicted ratios of the ultimate capacities for centroid of the connection when single-sided connections are used, as
the normal strength steel specimens were higher than those for the shown in Fig. 4(b). This effect was recently investigated by Fang et al.
high strength steel specimens, and this difference could be due to the [35], where ten full-scale coped steel I-beam tests were conducted,
influences of material ductility and yield-to-tensile strength ratio. It including seven specimens with single-sided connections and the
should be noted that four double-coped beams were examined remaining three with double angles. It was found that in general,
among the eight, but no obvious differences were observed between the out-of-plane loading eccentricity had no detrimental effect on the
the block shear responses of single flange and double flange coped block shear capacity of the specimens. Further numerical studies con-
beams. firmed this observation, although under some conditions. For example,
Franchuk et al. [29,30] significantly enriched the test database by specimens with double-line/row (two by two) bolt layouts in combina-
conducting seventeen more block shear tests on coped beams connect- tion with very flexible connections in rotation exhibited a reduced block
ed by double angle connections. One of those was a double-coped beam shear capacity for beams with single-sided connections. It was believed
and the remaining sixteen specimens were top flange coped beams. All that more friction (mainly caused by the out-of-plane rotation of the
top flange coped beams failed in block shear, whereas the double-coped web), which might reduce the load carried by the block, could be in-
beam failed in pure shear yielding over the web due to the reduced duced between the angle and beam web when single-sided connections
shear resistance of the full cross-section. The main test parameters con- were employed, such that the possible detrimental effects of out-of-
sidered in this programme were the effects of edge and end distances, plane loading eccentricity could be counteracted by the benefit of
end rotation, gross shear area, bolt diameter, and connection depth. friction. It was preliminarily concluded that the effect of single-sided
The main conclusions were that 1) end rotation had a limited effect on connections on block shear strength could be ignored in practical
block shear capacity and ductility of the beams concerned; 2) the actual design.
shear resisting area was greater than the net shear area, but might be Tables 4 and 5 summarise the details of the available tests for single
slightly less than the gross shear area; 3) doubling the end and edge dis- bolt line and double bolt line connections, respectively. The designa-
tances (from 25 mm to 50 mm for both cases) led to an increase in block tions used for typical bolt layout types are illustrated in Fig. 4(a). The
shear capacity of 20% and 9%, respectively; 4) bolt diameter did not reason for separating the two types of layout (single bolt line and double
affect the connection behaviour; and 5) predictions deriving from dif- bolt line) is mainly because of the difference in e stress distribution
ferent specifications were generally inconsistent and the predicted along the failure plane, particularly the tensile failure plane. The test-
block shear capacities for two bolt line connections might be unsafe. to-predicted ratios are also shown, which are further discussed in
Based on all the available test data (thirty-six tests by then), a reliability Section 3.3.
analysis was undertaken by Franchuk et al. [31] to evaluate the level of
safety of the major design standards. A new design equation, which was 3.2. Coped beams with welded connections
later adopted by the latest Canadian Standard CAN/CSA-S16-09 [32],
was proposed: Besides the coped beams with bolted connections, those with
welded connections, e.g. double angle cleats welded to the web
 
fy þ fu (Fig. 4(c)), are also susceptible to block shear failure even though
P u ¼ Rt Ant f u þ Agv pffiffiffi ð9Þ
2 3 there are no bolt holes. In such cases, the block of the web material
can be torn out by a combination of tensile and shear failure along the
where Rt is a stress correction factor, and Rt = 0.3 for two bolt line perimeter adjacent to the weld toes. Relevant research including exper-
connections, and Rt = 0.9 for one bolt line connections. This equation imental work on block shear failure in tension members using welded
assumes that the block shear capacity is governed by fracture of the connections can be traced back to the 1990s [36], but investigations
net tension area (but assuming a non-uniform stress distribution) on the case of coped beams have occurred only recently and was first
combined with failure of the gross shear area with average yield and attempted by Yam et al. [37,38]. Ten full-scale coped beam tests were
ultimate shear strengths. A reliability analysis model involving all conducted, out of which two specimens failed by block shear, six speci-
thirty-six available tests found the required resistance factors to be mens by local web buckling combined with significant ‘block shear type’
0.78, 0.72, and 0.66 in achieving the target safety indices of 3.5, 4.0, deformations, and no failure was achieved in the remaining two
and 4.5, respectively. This equation was also recommended by Driver specimens because of loading capacity limitations. For the specimens
et al. [33] for coped beams. In addition, the stress correction factors in featuring block shear failure, partial tear-out of the web block with a
Eq (9) were modified to extend its validity to more types of block tensile fracture underneath the bottom weld line of the clip angles
shear, e.g. block shear in angles, tees, and gusset plates. Topkaya [34] was exhibited, and no shear fracture was observed. The effects of
presented a nonlinear FE analysis study on block shear of coped block aspect ratio, web thickness, beam section depth, cope length,
beams, and compared the FE results with the available experimental and connection position were explored. By comparing with the predic-
works. Numerical modelling was adopted for the subsequent paramet- tions of the major standards, it was preliminarily found that none of
ric study comprising ninety analyses. Based on the test data and the these standards accurately reflected the actual failure modes observed
parametric study, an alternative block shear model, which is applicable in the tests. Following this experimental work, a FE study was per-
to both single-line and double-line connections was developed. The formed to examine the stress distribution along the shear and tension
simplified equation is given by: planes and the fracture mechanism of the web. Good agreement was
found between the FE predictions and test results in terms of both ulti-
mate capacity and failure modes. Subsequent parametric studies were
P u ¼ f y Ant þ 0:5f y Agv ð10Þ
conducted, based on which, a new block shear model for coped beams
with welded end connections was proposed. The basic form of the
The use of this equation is intended as an alternative to the more tra- block shear capacity model normally used, which sums up the tensile
ditional code equations, and good comparisons, on the conservative strength of the tension plane and yield strength of the shear plane,
side, with the test results were found. was modified to account for the effect of non-uniform stress distribution
M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 225

Table 4
Summary of block shear tests — single bolt line.

Specimen code Bolt layout Rtest (kN) Test-to-predicted ratio

AISC–LRFD CSA-S16-09 EN1993-1-8 AIJ

Birkemoe & Gilmor [2] I-2 1-3 498.4 1.00 0.87 1.51 1.24
Yura et al. [5] 18-3 1-5 943.4 1.22 1.03 1.89 1.59
18-4 1-5 894.5 1.28 1.03 1.85 1.73
18-5 1-5 769.9 1.18 0.95 1.75 1.62
18-8 1-5-s 645.3 0.91 0.75 1.39 1.23
18-9 1-3 676.4 0.88 0.74 1.19 1.02
Franchuk et al. [30] A1 1-4 439.0 1.02 0.86 1.55 1.42
A2 1-4 496.0 1.15 0.97 1.75 1.60
B1 1-4 514.0 1.19 1.01 1.79 1.64
B2 1-4 475.0 1.10 0.93 1.65 1.51
C1 1-3 402.0 0.99 0.85 1.40 1.28
D1 1-3 448.0 1.01 0.88 1.53 1.37
E1 1-4 568.0 1.07 0.95 1.69 1.39
E2 1-4 517.0 1.08 0.92 1.58 1.46
F1 1-3 324.0 1.02 0.87 1.55 1.38
G1 1-3 379.0 1.20 1.00 1.71 1.53
G2 1-3 387.0 1.22 1.02 1.75 1.56
H1 1-3 324.0 1.23 0.99 1.64 1.48
H2 1-3 341.0 1.29 1.04 1.73 1.55
Aalberg & Larsen [28] S1-S355 1-2 401.0 1.18 1.08 1.92 1.46
S2-Weldox 1-2 523.0 1.08 0.90 1.42 1.13
S3-S355 1-3 563.0 1.25 1.10 2.02 1.63
S4-Weldox 1-3 716.0 1.13 0.89 1.42 1.20
S5-S355 1-3 662.0 1.23 1.08 1.84 1.55
S6-Weldox 1-3 823.0 1.02 0.85 1.24 1.09
S7-S355 1-3 636.0 1.18 1.04 1.76 1.49
S8-Weldox 1-3 836.0 1.04 0.86 1.26 1.11
Fang et al. [35] A1-1-3-a 1-3 304.7 1.16 0.92 1.64 1.45
T1-1-3-a 1-3 331.9 1.27 1.00 1.79 1.59
A1-1-3-b 1-3 392.6 1.20 1.01 1.80 1.42
T1-1-3-b 1-3 414.9 1.26 1.06 1.89 1.50
T2-1-3-a 1-3 358.3 1.33 1.03 1.81 1.62
T2-1-3-b 1-3 484.6 1.40 1.16 2.04 1.62
A1-1-3-a-S 1-3 319.2 1.21 0.96 1.70 1.51

Mean= 1.15 0.96 1.66 1.44


CoV= 0.106 0.101 0.130 0.126

along the failure planes. The proposed design equation, illustrated in tension area, and Av = (b + s + 0.5p)tw is the shear area, in which
Fig. 4(c), assumes a trapezoidal tensile stress distribution and uniform φ =0.0342a − 0.8231, where a is the length of the tension area and
shear stress distribution which linearly decreases to zero from the top 40 mm ≤ a ≤ 100 mm, b is the length of the shear area, tw is the thick-
of the angle towards the top edge of the web, as expressed by: ness of the beam web, s is the size of the fillet weld, and p is the distance
between the angle and the top web edge.
P u ¼ Rt f u At þ Rv 0:6 f y Av ð11Þ With the need for more experimental evidence, Wei et al. [39] con-
ducted twelve more tests to study the block shear failure of coped
where Rt = (1.05 − φa/2fu) is a tension correction factor, and Rv = beams with welded connections. Eight specimens were found to fail
1.5528 − 0.0078a is a shear correction factor, At = (a + s)tw is the by block shear, two specimens with the longest cope lengths failed by

Table 5
Summary of block shear tests — double bolt line.

Specimen code Bolt layout Rtest (kN) Test-to-predicted ratio

AISC–LRFD CSA-S16-09 EN1993-1-8 AIJ

Yura et al. [5] 18-10 2-2-a 494.0 0.94 0.88 1.08 0.80
18-11 2-2-a-s 449.5 0.75 0.74 0.84 0.59
18-12 2-2-a 676.4 1.16 1.14 1.31 0.92
Ricles & Yura [6] 18-16 2-2 494.0 1.00 0.93 1.15 0.84
18-17 2-2 583.0 1.06 1.03 1.20 0.84
18-18 2-2-b 449.5 1.00 0.95 1.26 0.88
18-19 2-2-b 596.3 1.09 1.06 1.31 0.92
Franchuk et al. [30] C2 2-3 537.0 1.13 1.08 1.40 1.06
J1 2-3 667.0 1.37 1.32 1.70 1.29
J2 2-2 338.0 1.24 1.29 1.48 1.04
Fang et al. [35] A2-2-2-a 2-2 384.0 1.66 1.61 1.88 1.31
T1-2-2-a 2-2 379.7 1.63 1.58 1.85 1.29
T2-2-2-a 2-2 329.0 1.49 1.45 1.71 1.20

Mean= 1.19 1.16 1.40 1.00


CoV= 0.230 0.235 0.224 0.224
226 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

local web buckling, one specimen failed in the welds, and no failure the coped beam case, Ut = 0.7 was recommended. It is worth noting
occurred in the remaining specimen. The eight specimens which failed that Ut = 0.9 leads to a test-to-predicted ratio close to unity, but given
by block shear featured a similar failure mode, i.e. tensile fracture with the relatively large scatter in the test-to-predicted ratio, a decreased Ut
significant shear yielding. The test results were compared against the is employed based on the reliability analysis in order to secure a suffi-
major design equations and Eq. (11), and it was found that CAN/CSA- cient level of safety. Eq. (12) offers a simpler alternative to the
S16-09 [32] (using Ut = 0.9) and Eq. (11) provided the most reasonable Eq. (11) earlier proposed by Yam et al. [38], and importantly, Eq. (12)
estimations. With extended test data, further FE analysis and parametric is consistent with the equations used for bolted connections.
analyses were subsequently conducted by Yam et al. [40]. One of the im-
portant findings was that the block shear capacity increased with in- 3.3. Current design specifications and comparisons
creasing connection rotational stiffness. This is explained by the fact
that when the connection rotational stiffness is small, the relatively General equations for block shear have been stipulated in various
large end rotation leads to less significant shear yielding over the major standards, including the American Standard AISC–LRFD [42],
shear plane, as confirmed by the FE analysis results. In this respect, the Canadian Standard CSA-S16-09 [32], Eurocode 3 [43], and Japanese
above proposed design equation (Eq. (11)) was improved by taking standard AIJ [44]. Although some studies have attempted to evaluate
into account the effect of connection rotational stiffness. The main strat- the reliability of these standards in predicting the block shear capacity
egy was to change the shear correction factor Rv, and thus lead to a more [26,27], only older versions of standards were available to those studies.
reasonable agreement with the FE observation. The modified Rv was re- Noting that AISC–LRFD, CSA-S16-09, and Eurocode 3 have all been
expressed asRv = 1.087 − 0.00549a when the g/t ratio ≥ 5.6, where g is updated in the last decade, it is now necessary to re-evaluate the latest
the bolt gauge as shown in Fig. 4(c) and t is the thickness of the angle versions of these standards based on the latest test database. The design
part connected to the supporting face. A larger g/t ratio indicates a rela- equations employed in those standards are summarised in Table 6.
tively smaller rotation stiffness. It was also suggested that when the g/t As indicated in Table 6, AISC–LRFD assumes that in combination
ratio b 5.6, the previous equation for Rv can still be used. It was shown with tensile fracture, either gross yielding or net rupture on the shear
through a reliability analysis that when the modified design equation area governs the block shear capacity and the utilisation level of the ten-
was employed in conjunction with the use of a resistance factor of sile fracture resistance depends on bolt layouts. CSA-S16-09 considers
0.76, a safety index β of 4.0 could be achieved. that block shear capacity is governed by fracture of the net tension
Summarising all the observations from the tests and the FE analyses area combined with failure of the gross shear area based on average
undertaken by Yam and colleagues, some important conclusions can be yield and ultimate shear strengths and that the tensile fracture
drawn: 1) block shear capacity is affected by the geometry of the resistance also depends on bolt layouts. Eurocode 3 assumes the block
welded angle, and in general, the capacity increases with increasing shear capacity to be governed by tensile fracture on the net tension
web block aspect ratio (ratio of shear area over tension area); 2) in- area and shear yielding on the net shear area, and a factor of 0.5 is
creasing the tension area increases the in-plane loading eccentricity applied to the tensile fracture resistance, regardless of bolt layouts. AIJ
(expressed as the distance from the centroid of the weld group to considers two possible failure modes, namely, tensile fracture in combi-
the supporting column face), which leads to decreased shear yielding; nation with net shear yielding, or tensile yielding in combination with
3) to improve block shear capacity, increasing the shear area is net shear fracture, although the latter case has hardly been observed
preferred over increasing the tension area; 4) the beam depth, cope in any coped beam test. In view of the above, all standards share a sim-
depth, and connection position seem to have little influence on the con- ilar design strategy (i.e. block shear capacity = resistance of tension
nection capacity; 5) increasing the connection rotational stiffness is plane + resistance of shear plane), but they differ in two main aspects:
beneficial to block shear capacity. 1) the determination of the efficiency of the net tension plane in
Recently, Oosterhof and Driver [41] recommended the use of unified resisting fracture (an efficiency coefficient is usually used), and 2) the
block shear equations to estimate the block shear capacities of lap choice of yield or ultimate shear strengths over the net or gross shear
plates, slotted HSS-to-gusset plates, and coped beams – all with welded plane.
connections. With the available test database, reliability analysis was Tables 4 and 5 give the test-to-predicted ratios of the available test
employed to achieve this goal. The unified equation, following the specimens with single bolt line and double bolt lines, respectively. For
form of that in CAN/CSA-S16-09 [32], was expressed as: the case of the single bolt line, AISC–LRFD and CSA-S16-09 provide
sound predictions with average test-to-predicted ratios relatively
 
fy þ fu close to unity. The associated CoVs, indicating the level of scatter of
P u ¼ U t f u Ant þ 0:6 Agv ð12Þ
2 the test data, are also within reasonable ranges. Both Eurocode and AIJ
are conservative, and the CoVs are slightly larger than those of AISC–
where Ut is the equivalent tensile stress factor which has a similar LRFD and CSA-S16-09. For the case of the double bolt line, the CoVs
meaning to the tension correction factor proposed by Yam et al. [38]. for all four considered standards are large. It seems that AIJ provides a
Different values of Ut were proposed for various connection types. For good prediction of the average test-to-predicted ratio, which is unity.

Table 6
Design equations for block shear in major standards.

Codes Design procedures

AISC–LRFD [42] Pr = ϕ(UbsFuAnt + 0.6FuAnv) ≤ ϕ(UbsFuAnt + 0.6FyAgv)


where ϕ = resistance factor, Ubs = 1.0 if the tensile stress is uniform (single bolt line or welded connections); Ubs = 0.5 if
the tensile stress is non-uniform (double or multiple bolt lines); the other symbols have been defined in the previous section.
Canadian Standard CSA-S16-09 [32] Pr = ϕu[UtFuAnt + 0.6Agv(Fy + Fu)/2]
where ϕu = the resistance factor; and Ut = the efficiency factor: Ut = 0.9 for coped beams with one bolt line; Ut = 0.3 for coped
beams with two bolt lines.
 pffiffiffi
Eurocode 3 EN1993-1-8 [43] P r ¼ 0:5F u Ant =γM2 þ 1= 3 F y Anv =γM0
where γM0 = the partial safety factor for resistance of cross-section; γM2 = the partial safety factor for resistance of cross-section
in tension to fracture.
h  pffiffiffi i h  pffiffiffi i
AIJ [44] P r ¼ φ F u Ant þ 1= 3 F y Anv or P r ¼ φ F y Ant þ 1= 3 F u Anv
and the lesser governs the design capacity. The symbols are similar to those in the above codes.
M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 227

Table 7
Summary of block shear tests — welded connection.

Specimen code a/b (mm) Rtest (kN) Test-to-predicted ratio

AISC–LRFD CSA-S16-09 EN1993-1-8 AIJ Yam

Yam et al. [37] C2-457T 50/120 630.0 1.06 1.03 1.43 1.08 1.02
E1-356T 50/120 601.2 1.18 1.11 1.62 1.20 1.13
Wei et al. [39] A1 50/100 532.3 0.91 0.88 1.26 0.93 1.03
A2 51/100 582.2 0.99 0.95 1.37 1.00 1.12
A3 50/100 632.1 1.08 1.04 1.50 1.10 1.04
A4 50/101 652.0 1.12 1.08 1.56 1.14 1.07
A5 50/120 671.2 1.08 1.03 1.47 1.11 1.23
S2 51/161 631.8 1.09 1.02 1.42 1.11 1.03
S3 50/158 638.2 1.12 1.05 1.45 1.14 1.06
S4 51/157 591.9 1.04 0.98 1.35 1.06 0.99

Mean= 1.07 1.02 1.44 1.09 1.07


CoV= 0.071 0.066 0.070 0.071 0.066

The predictions from AISC–LRFD and CSA-S16-09 are also acceptable, On the other hand, as shown in Table 8, the ultimate capacities of the
but Eurocode 3 gives a quite conservative prediction. It is of interest to coped beam specimens that fail by inelastic local buckling and signifi-
note that the test-to-predicted ratios for the double bolt line specimens cant ‘block shear type’ deformation are overestimated by the predic-
investigated by Fang et al. [35] are generally higher than those for the tions deriving from both local web buckling checking (lowest value of
remaining specimens (Table 5). A possible explanation is that the Ry, Rvy, and Rcheng/Ryam governed by flexural yielding, shear yielding,
shear areas of the specimens in Fang et al. [35] were clearly less than and local buckling, respectively) and block shear checking (particularly
those for the other specimens, and this may result in a more efficient for AISC–LRFD, CSA-S16-09, and AIJ), which leads to unsafe design. This
utilisation of the shear area, thus leading to increased test-to- is possibly due to the interaction between the inelastic shear deforma-
predicted ratio. This speculation, however, needs further verification. tion (along shear plane due to block shear) and local web buckling in
The codified design equations in Table 6 are mainly based on re- the coped web welded to angle cleats. In other words, the significant
search on bolted connections, so their applicability to welded conditions ‘block shear type’ deformation along the shear plane may interactively
is not made clear. In practice, these equations may be applied to welded induce earlier inelastic local web buckling. This possible interaction,
connections with a reasonable level of reliability, although appropriate which has never been considered in any existing studies, needs further
efficiency factors (to allow for non-uniform stress distributions) for investigation. To ensure a safe design, a reduction factor may be applied
tension planes are generally lacking. The latest version of AISC–LRFD to both the local web buckling resistance and the block shear resistance
stipulates the tensile stress distribution factor Ubs, to be taken as 1.0, for coped beams with welded connections, but this suggestion should
for coped beams with welded connections. This assumes a uniform be subjected to further studies. It is noted that unsafe predictions
tensile stress distribution along the tensile plane. Table 7 shows the were not made for coped beams with bolted connections, as confirmed
test-to-predicted ratios for the available coped beam tests when welded in Table 2 (specimens with double angle cleat bolted connections). This
connections apply. For CSA-S16-09, the efficiency factors are taken as is probably due to the fact that the plane of shear deformation was
those used for the case of single bolt line (Ut = 0.9). The test-to- restrained by the angle legs and, therefore, interaction between block
predicted ratios obtained from Yam's design equation [38], which was shear and local web buckling is less likely to occur.
explicitly proposed for welded connections, are also included in the
table. It should be noted that Table 7 only includes the specimens failing 4. Fatigue
in block shear. Those failing in inelastic local buckling but with signifi-
cant ‘block shear type’ deformation (as reported in Yam et al. [37,38]) Fatigue is another common failure type for coped beams/stringers
are listed in Table 8 separately. It is observed from Table 7 that AISC– used in bridges. The case studies reported by Fisher [45] raised research
LRFD, CSA-S16-09, AIJ, and Yam's design equation provide good predic- interest in this issue, and experimental studies have been subsequently
tions, with the mean test-to-predicted ratios slightly above 1.0 and the undertaken. Nine tests were first conducted by Yam and Cheng [46,47]
CoVs near 0.07, whereas Eurocode 3 is again conservative. The main rea- to evaluate the fatigue strength of coped beams, where one specimen
son why Eurocode 3 is conservative is that a factor of 0.5 is applied to the had a square cut cope, five specimens had flame cut copes with nominal
tensile fracture resistance, regardless of the bolt or weld layouts actually radii from 10 mm to 30 mm, two specimens used flame cut copes and
proposed. ground smooth edges. The cope of the remaining specimen was flame

Table 8
Summary of block shear test specimens with welded connections — failed by inelastic local buckling with significant ‘block shear type’ deformation.

Specimen code a/b (mm) Rtest (kN) Test-to-predicted ratio

Inelastic local buckling AISC-LRFD CSA-S16-09 EN1993-1-8 AIJ Yam

Yam et al. [37] A1-406r3 50/160 395.2 0.85 0.94 0.87 1.22 0.96 0.88
A2-406r2 70/140 437.4 0.94 0.95 0.92 1.32 0.97 1.02
B1-406t 50/120 394.0 0.84 1.05 1.00 1.42 1.07 1.01
B2-406r1 90/110 390.3 0.84 0.81 0.80 1.20 0.85 1.00
D1-457R1 90/120 623.0 0.74 0.81 0.81 1.18 0.84 0.99
E2-356P 50/120 581.0 1.09 1.05 0.98 1.41 1.07 1.04
Wei et al. [39] C1 100/100 448.0 0.89 0.71 0.71 1.09 0.77 0.97
C2 83/130 506.7 1.01 0.83 0.80 1.20 0.85 0.97

Mean= 0.90 0.89 0.86 1.26 0.92 0.99


CoV= 0.123 0.137 0.115 0.092 0.122 0.048
228 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

cut from a 10 mm pre-drilled hole. The span of each beam was 3.67 m, drilled hole with inserted bolt, and bolt removal, as illustrated in
and a cyclic point load was applied at mid-span. The up-side-down test Fig. 5(a). The first two methods eliminate the crack tip because of the
beams were simply supported at both ends, such that the coped edge additional hole drilled at the end of the crack development path. The
was in tension under the downward point load. The nominal stress last method reduces the negative moment by removing several bolts/
ranges were 40 MPa, 50 MPa, and 70 MPa with a nominal mean stress rivets near the top edge of the coped web. The tests mainly showed
of 70 MPa in tension for all specimens. It was observed that fatigue life that: 1) the copes with square cut or notched flame cut were more sus-
increased with an increase of cope radius and decrease of the stress ceptible to fatigue crack initiation, but the crack growth rate was inde-
range. The crack growth rate after initial cracking was not dependent pendent of the cope surface finish; 2) connection rotational stiffness
on the cope details, but increased with increasing nominal stress had an important influence on the fatigue resistance of the cope;
range. An analytical model was developed to estimate the fatigue life 3) the drilled hole DLM was less effective than the drilled hole with
of coped beams, and the predictions were shown to be conservative. inserted bolt and bolt removal; and 4) a new crack could occur when
Sixteen more fatigue tests were performed by Roeder et al. [48,49] the drilled hole with inserted bolt DLM was used, and this crack might
on coped beams. The bolted connection geometry was arranged to en- propagate very quickly. Finally, the test data were used to propose a de-
sure sufficient connection rotational stiffness, such that negative mo- sign based approach associated with the S-N curve (i.e. Stress range vs.
ment could be induced and the coped edge of the beam became under Number of cycle to allowable crack) for evaluating the fatigue strength
tension. The span of each beam was 7.0 m, and a cyclic load was applied of coped beams.
at 3.31 m from the coped end. Among the sixteen specimens, two spec- As an alternative solution to the conventional damage limitation
imens used square cut copes, six specimens used flame cut copes with methods mentioned above, a new repair method employing carbon
22 mm radius, and the remaining eight specimens used 22 mm-radius fibre reinforced polymer (CFRP) patches was proposed recently by
flame cut copes with additional notches to speed up crack initiation. Holden [50]. Six tests were conducted, including one unrepaired speci-
The test programme included two stages: the first stage was the study men with the other five repaired by CFRP patches. For the repaired
of the initial fatigue crack, and the second stage was to evaluate the cases, the bare coped beams were subjected to cyclic loading until a
effectiveness of three damage limitation methods (DLM) applied after crack length of 10 mm was evident, and CFRP patches were then applied
initiation of the crack. The three DLMs consisted of a drilled hole, a before for further fatigue testing with the same cyclic loading. The test

Fig. 5. Fatigue failure of coped beams: a) repair strategies, b) comparisons against S-N design curves.
M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 229

parameters included single/double sided repair, orientation of the The work presented in this study not only offers a clear view of the
patch, and total thickness of CFRP. The nominal stress range at the current treatments for these local failure modes, but also facilitates
cope was 80 MPa. All copes were flame cut with a radius of 15 mm, future reliability analysis studies towards more reliable design equa-
with an additional notch 0.5 mm long for all cases to hasten crack initi- tions. The following comments are noted based on this work:
ation. The test results showed that applying double sided CFRP patches
• For the design against elastic local web buckling of top flange coped
were quite efficient in increasing fatigue life, whereas single sided CFRP
beams, two methods are currently available. For most commercially
patches proved marginal with earlier adhesive bond failure. Angled re-
available hot-rolled sections, Yam's method [12] can lead to predic-
pairs were more effective than the other orientations. Increasing the
tions very close to the test results, and closer than Cheng's method
CFRP patch thickness did increase fatigue life, but the improvement
[8] which tends to be conservative. Both methods tend to be conserva-
was not significant. For instance, applying the double sided CFRP
tive for plate girders with ‘thin’ webs. Where inelastic web buckling
patches at a crack of 10 mm improved the remaining fatigue life by
is possible, the design strategy that uses the lowest value of Rbuckling
three times before a 30 mm crack was reached when compared with
(either Rcheng or Ryam), Ry, and Rvy can be safe for coped beams with
the unrepaired specimen. When angled single sided patches were
bolted connections. For coped beams with welded connections, a
employed, the remaining fatigue life was approximately 1.5 times that
reduction factor of 0.9 is suggested for the local web buckling design
of the unrepaired one. For the other patch orientations, the remaining
resistance due to the possible interactive effect of block shear. For
fatigue life was only 1.15 times that of the unrepaired one. A supple-
coped beams with insufficient local web buckling resistance, the
mentary study showed that earlier repair achieved more efficient
reinforcement strategies using stiffeners were found to be effective.
usage of CFRP patches. Finally, an analytical model was proposed to
For the design of double flange coped beams, the current design meth-
evaluate the fatigue life of CFRP repaired coped beams [51].
od, which was based on numerical studies, is on the conservative side.
In design practice, S-N curves with various categories are used by
No test evidence is currently available for the case of double flange
engineers to conduct design against fatigue. Different categories are
coped beams.
applicable to different scenarios as stipulated in relevant standards
• Design equations for block shear failure of coped beams are available
(AISC–LRFD [42], Eurocode 3 [43], AS4100 [52], and CAN-S16-09 [32]).
in the current major design standards, but inconsistent test-to-
Although categorization is slightly different in the different standards,
predicted strength ratios were generally found, based on the available
the basic principle is similar, i.e. linking the design stress range to the
test database. For coped beams connected by bolted connections with
predicted fatigue life. Fig. 5(b) shows the S-N relationships for the test
single bolt line, AISC–LRFD's design equations are recommended as
specimens after initial crack (crack first observed by visual inspection)
they provide good predictions (mean test-to-predicted ratio = 1.15)
and a 30 mm crack, using the relevant AISC–LRFD category curves. For
and an acceptable CoV. Although CSA-S16-09 can also provide sound
some test specimens subjected to variable load amplitude, the nominal
predictions (mean test-to-predicted ratio = 0.96), the level of scatter
stress ranges reported by Roeder et al. [49] were, in fact, effective stress
of the test data might lead to unsafe design. For the case of bolted
ranges using Miner's rule [53]. In addition, when applying S-N curves,
connections with a double bolt line, although relatively large levels
Yam and Cheng [47] and Roeder et al. [49] used modified stress ranges
of CoV have been observed for all of the four design codes considered,
(allowing for stress concentrations) and nominal stress ranges, respec-
the AISC–LRFD and CSA-S16-09 equations are recommended, with a
tively. Most of the standards recommend that for the application of
mean test-to-predicted ratio of 1.19 and 1.16, respectively. For
the category curves, the nominal stress range should be used without
coped beams with welded connections, AISC–LRFD, CSA-S16-09, AIJ,
stress concentration factors (SCFs). Eurocode 3, however, also states
and Yam's method (Eq. (11)) provide sound and relatively consistent
that the stress concentration effect at details other than the stipulated
predictions with mean test-to-predicted ratios slightly above 1.0. But
cases should be accounted for when determining the stress range. Con-
based on the reliability analysis conducted by Oosterhof and Driver
sidering the fact that a precise determination of the SCF for various cope
[41], a decreased Ut might have to be employed to secure a sufficient
details is difficult, the stress ranges shown in Fig. 5(b) are the nominal
level of safety, bearing in mind that there are possible interactions be-
values, which may be convenient for engineers. It was shown that for
tween local web buckling and block shear. The authors recommended
square cut copes or copes flame cut with radius and notch, the test
that Ut is taken as 0.7 in the CSA-S16-09 equation, as suggested by
data lay significantly below the category E' curve. These cope details
Oosterhof and Driver [41].
should be avoided in practical design. For flame cut copes with a certain
• For fatigue design of coped beams, a new category S-N curve
range of radii (which are more practical cases) and an allowable crack
(allowing a 30 mm crack length) is proposed in this paper based on
length of 30 mm, it seems that category E' is not appropriate for the fa-
the available test database. The new curve is a modification of the
tigue strength estimation. Therefore, a new category curve modified
relevant equations in AISC–LRFD with a newly proposed constant
from relevant AISC–LRFD equations in, as shown in Fig. 5(b), is proposed
Cf = 2.0 × 108. The nominal stress range without considering the ef-
in this paper, and given by:
fect of stress concentration should be used when the new category
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi curve is referred to. To strengthen or repair a coped beam with
3 C f  329 existing cracks, the two DLMs of drilled hole with inserted bolt and
F SR ¼ ð13Þ
nSR bolt removal are adequately effective. Using CFRP patches is also an
available alternative, but these have to be placed at both sides of the
cracked web, preferably with angled repairs, to ensure a satisfactory
where FSR = nominal stress range (MPa), nSR = fatigue life (maximum
efficiency. It is noted that a smooth surface of the cope radius effec-
number of load fluctuation), and Cf = 2.0 × 108. This curve could be pre-
tively postpones the occurrence of the initial crack.
liminarily used for the coped beam case, with an allowable crack length
of 30 mm. Based on the above, the recommended design solutions are
summarised in Table 9. Shortcomings in the scope of current research
5. Comments, summary, and future research and design approaches are also identified. The following future studies
are required:
A comprehensive overview of the updated research outcomes and
design approaches to control the local failures of coped beams is pre- • Modified local web buckling design equations are desirable for coped
sented in this paper. The failure modes considered in this study include beams with d/tw ratio larger than 60 (e.g. plate girders with
local web bucking, block shear, and fatigue. The available test database thin webs), in which case the Cheng and Yam methods can be conser-
in the literature and corresponding design solutions are also discussed. vative.
230 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

Table 9
Summary of recommended design solutions.

Failure mode Recommended design solutions Comments

Local web buckling – bolted Yam's method [12] Tends to be conservative for thin web girder specimens. May need to be revised
connection (single cope) for thin web girder specimens.
Local web buckling – welded A coefficient of 0.9 to be applied on the More tests are required to investigate the possible interaction effect between
connection (single cope) result obtained from Yam's method [12] local web buckling and block shear.
Local web buckling – double cope Cheng's method [8] Test evidence is necessary.
Block shear – single bolt line AISC–LRFD [42] –
Block shear – double bolt line AISC–LRFD [42] or CSA-S16-09 [32] Tends to be conservative for the cases where the shear area is relatively small
compared with the tension area. More test data are required.
Block shear – welded connection CSA-S16-09 [32] equation modified by Ut = 0.7 was recommended
Oosterhof and Driver [41]
Fatigue The newly proposed category curve Nominal stress range should be used without considering stress concentration
(Eq. (13) in this paper) factors (SCFs). The allowable crack length is 30 mm for the new curve.

• Test evidence on the failure and behaviour of double-coped beams is Illustrative Example 1 (bolted connections)
necessary, and possible reinforcement strategies are also required.
• The effect of out-of-plane eccentricity on coped beams with single- Local web buckling design
sided welded connections is worth investigating. Assume that the applied reaction is 250 kN.
• The possible interaction effect between local web buckling and block
1) Check for flexural yielding:
shear on coped beams with welded connections may lead to unsafe
design, and this aspect should, therefore, receive attention. 2 5 3
• To improve the fatigue life of coped beams, more investigations are M co;x ¼ f y Z co;x ¼ 360N=mm  1:34  10 mm ¼ 48:24kN:m;
desirable on the application of CFRP patches, which seems to be
efficient especially when they are applied on both sides. It is also
worthwhile to optimize the timing of the application of CFRP patches. Ry ¼ M co;x =c ¼ 48:24kN:m=200mm ¼ 241:2kN:
The feasibility of using other types of FRP (e.g. GFRP) is also worth
investigating. 2) Check for shear yielding:
pffiffiffi pffiffiffi 2
pffiffiffi
Rvy ¼ f y Av = 3 ¼ f y t w ho = 3 ¼ 360N=mm  6mm  299mm= 3
6. Recommended design examples ¼ 372:9kN:

Three illustrative examples are provided covering local web buck- 3) Check for local web buckling:
ling, block shear, and fatigue design. Grade S355 UB356 × 127 × 33 According to the design recommendations discussed above, Yam's
coped steel beams are considered, for which the geometric and material method is used:
properties are given in Fig. 6. Resistance factors/partial factors are not
considered in the calculation. a ¼ 1:38−1:79dc =D ¼ 1:38−1:79  50=349 ¼ 1:1235;

Fig. 6. Geometric dimensions and material properties for illustrative design examples.
M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232 231

2
b ¼ 3:64ðdc =DÞ −3:36dc =D þ 1:55 Block shear design
2
¼ 3:64ð50=349Þ −3:36ð50=349Þ þ 1:55 ¼ 1:1442: The CSA-S16-09 equation with the decreased tensile factor Ut = 0.7
(Oosterhof and Driver 2011) is recommended for block shear design of
coped beams with welded connections:
The shear buckling coefficient ks =a(ho/c)b =1.1235(299/200)1.1442 =
 
1.78, and thus: fy þ fu 2 2
R ¼ 0:6 Agv þ U t f u Ant ¼ 0:6  430N=mm  1032mm
2  2   2
π E t π2 2:0  105 N=mm2 6 2 2 2
τcr ¼ ks  2
 w ¼ 1:78  2
 þ0:7  500N=mm  372mm ¼ 396:5kN;
12 1−v ho 12 1−0:3 299
2
¼ 129:4N=mm ;
noting that the tensile plane and shear plane are positioned along the
outer perimeter of the welds.
Rbuckling = τcrtwh0 = 129.4N/mm2 ⋅ 6mm ⋅ 299mm = 232.1kN, (it is
noted that Rbuckling predicted by Cheng's method is 238.5 kN). Illustrative Example 3 (fatigue design)
 Therefore, the  local web buckling design is governed by min
Ry ; Rvy ; Rbuckling ¼ Rbuckling ¼ 232:1kN, which is less than the applied Assume that the negative bending moment Mc over the critical sec-
reaction of 250 kN, so stiffeners are required. Since d/tw = 311.6/6 = tion fluctuates between 0 kN.m and 10 kN.m (Fig. 6), and hence the
51.9 b 52.7, dc/D = 50/349 = 0.14 b 0.3, and c/D = 200/349 = range of nominal stressΔMc/Zco,x = 10 kN.m/134,000 mm3–0 =
0.57 b 0.75, stiffener type A (Fig. 3(a)) with LRx = 2dc = 100 mm is pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
74.7 MPa. According to the proposed equation F SR ¼ 3 C f  329=nSR ,
recommended. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i.e. 74:7MPa ¼ 2  108  329=nSR , the fatigue life nSR controlled by a
3

Block shear design 30 mm crack length is 157,857 cycles. Alternatively, directly employing
AISC–LRFD (2010) is recommended for block shear design of coped the S-N design curve given in Fig. 5(b), the obtained fatigue life nSR is
beams with bolted connections: approximately 150,000 cycles.

2 2
R ¼ 0:6f u Anv þ U bs f u Ant ≤0:6 f y Agv þ U bs f u Ant ¼ 0:6  500N=mm  480mm Acknowledgement
2 2 2 2
þ1:0  500N=mm  228mm ≤ 0:6  360N=mm  840mm
2 2 The work described in this paper was partially supported by a grant
þ1:0  500N=mm  228mm ¼ 258:0kN ≤ 295:4kN:
from the Research Grants Council of the Hong Kong Special Administra-
tive Region, China (project no. PolyU 5288/13E).

Illustrative Example 2 (welded connections) References


[1] Cheng JJR. Design of steel beams with end copes. J Constr Steel Res 1993;25:3–22.
Local web buckling design
[2] Birkemoe PC, Gilmor MI. Behavior of bearing critical double-angle beam connections.
Again assume that the applied reaction is 250 kN. Eng J 1978;15(4):109–15.
[3] Ricles JM, Yura JA. The behavior and analysis of double-row bolted shear web
1) Check for flexural yielding: connections. Phil. M. Ferguson Structural Engineering Laboratory Thesis No. 80-
1University of Texas at Austin; 1980.
2 5 3 [4] Shelton BG, Yura JA. Tests of bolted shear web connections. Department of Civil
M co;x ¼ f y Z co;x ¼ 360N=mm  1:34  10 mm ¼ 48:24kN:m; Engineering, Report No. 80-1Phil M. Ferguson Structural Engineering Laboratory,
University of Texas at Austin; 1981.
[5] Yura JA, Birkemoe PC, Ricles JM. Beam web shear connections: an experimental
study. J Struct Div 1982;108(ST2):311–25.
Ry ¼ M co;x =c ¼ 48:24kN:m=100mm ¼ 482:4kN: [6] Ricles JM, Yura JA. Strength of double-row bolted-web connections. J Struct Eng
1983;109(ST12):126–42.
[7] Milek WA. A cautionary note on beam copes. Eng J 1980;17(3):72–3.
2) Check for shear yielding: [8] Cheng JJ, Yura JA, Johnson CP. Design and behavior of coped beams. Ferguson
Structural Engineering Laboratory Report No. 84-1Department of Civil Engineering,
pffiffiffi pffiffiffi 2
pffiffiffi University of Texas at Austin; July 1984.
Rvy ¼ f y Av = 3 ¼ f y t w ho = 3 ¼ 360N=mm  6mm  299mm= 3 [9] Cheng JJR, Yura JA. Local web buckling of coped beams. J Struct Eng
¼ 372:9kN: 1986;112(10):2314–31.
[10] American Institute of Steel Construction. Steel Construction Manual. One East
Wacker Drive, Suite 700, Chicago, Illinois14th ed. ; 2011.
3) Check for local web buckling: [11] The Steel Construction Institute and the British Constructional Steelwork Association
Limited. Joints in Steel Construction: Simple Connections. Publication No. 212, UK;
2002.
b 1:1442
ks ¼ aðho =cÞ ¼ 1:1235ð299=100Þ ¼ 3:933; [12] Yam MCH, Lam ACC, Iu VP, Cheng JJR. The local web buckling strength of coped steel
I-beam. J Struct Eng 2003;129(1):3–11.
[13] Aalberg A, Larsen PK. Local web buckling of coped beams. Proceedings of Nordic
Steel Construction Conference NSCC 2001, Helsinki, Finland; 2001.
2  2 2 5 2  [14] Muir LS, Thornton WA. A technical note: a direct method for obtaining the plate
π E tw π 2:0  10 N=mm 6 2
τ cr ¼ ks   ¼ 3:933   buckling coefficient for double-coped beams. Eng J 2004;41(3):133–4.
12 1−v2 ho 12 1−0:32 299 [15] Yam MCH, Lam ACC, Wei F, Chung KF. The local web buckling strength of stiffened
2 coped steel-I-beam. Int J Steel Struct 2007;7(2):129–38.
¼ 286N=mm ; [16] Yam MCH, Ma HW, Lam ACC, Chung KF. Experimental study of the strength and be-
haviour of reinforced coped beams. J Constr Steel Res 2011;67:1749–59.
[17] Yam MCH, Chung KF. A numerical study of the strength and behaviour of reinforced
Rbuckling = τcrtwh0 = 286N/mm2 ⋅ 6mm ⋅ 299mm = 513.1kN, (it is
coped beams. J Constr Steel Res 2013;80:224–34.
noted that Rbuckling predicted by Cheng's method is 748.5 kN). [18] Clements DDA, Teh LH. Active shear planes of bolted connections failing in block
shear. J Struct Eng 2013;139(3):320–7.
The local web buckling design is governed by min(Ry, Rvy, Rbuckling) = [19] Huns BBS, Grondin GY, Driver RG. Tension and shear block failure of bolted gusset
Rvy = 372.9kN. For welded connections, because a reduction factor of plates. Can J Civ Eng 2006;33(4):395–408.
0.9 is suggested to consider the possible interactive effect of block [20] Topkaya C. A finite element parametric study on block shear failure of steel tension
members. J Constr Steel Res 2004;60(11):1615–35.
shear, the final local web buckling resistance is 0.9 × 372.9 kN = [21] Topkaya C. Block shear failure of gusset plates with welded connections. Eng Struct
335.6 kN, and no stiffeners are required. 2007;29:11–20.
232 M.C.H. Yam et al. / Journal of Constructional Steel Research 102 (2014) 217–232

[22] Epstein HI, McGinnis MJ. Finite element modeling of block shear in structural tees. [38] Yam MCH, Zhong YC, Lam ACC, Iu VP. An investigation of the block shear strength of
Comput Struct 2000;77:571–82. coped beams with a welded clip angle connection-part II: numerical study. J Constr
[23] Oosterhof SA, Driver RG. Effects of connection geometry on block shear failure of Steel Res 2007;63(1):116–34.
welded lap plate connections. J Constr Steel Res 2011;67:525–32. [39] Wei F, Yam MCH, Chung KF, Grondin GY. Tests on block shear of coped beams with a
[24] Orbison JG, Wagner ME, Fritz WP. Tension plane behavior in single-row bolted welded end connection. J Constr Steel Res 2010;66(11):1398–410.
connections subject to block shear. J Constr Steel Res 1999;49:225–39. [40] Yam MCH, Grondin GY, Wei F, Chung KF. Design for block shear of coped beams with
[25] Hardash SG, Bjorhovde R. New design criteria for gusset plates in tension. Eng J, 22. a welded end connection. J Struct Eng 2011;137(8):811–21.
AISC; 1985 77–94. [41] Oosterhof SA, Driver RG. Performance of the unified block shear equation for
[26] Kulak GL, Grondin GY. Block shear failure in steel members — a review of design common types of welded steel connections. Eng J 2011;48(2):77–92.
practice. In: Leon RT, Easterling WS, editors. Proceedings of fourth international [42] AISC-LRFD. Load and resistance factor design specification for structural steel build-
workshop on connections in steel structures IV: steel connections in the new millen- ings. Chicago, IL, USA: American Institute of Steel Construction; 2010.
nium. Chicago (III): AISC; 2000. [43] EN 1993-1-8:2005. Eurocode 3: Design of Steel Structures – Part 1-8: Design of
[27] Kulak GL, Grondin GY. AISC LRFD rules for block shear—a review. Eng J Joints. Brussels, Belgium: European Committee for Standardization; 2005.
2001;38(4):199–203. [44] Architectural Institute of Japan (AIJ). Standard for limit state design of steel
[28] Aalberg A, Larsen PK. Strength and ductility of bolted connections in normal and structures; 1990.
high strength steels. Proceedings of the seventh international symposium on struc- [45] Fisher JW. Fatigue and fracture in steel bridges: case studies. John Wiley & Sons;
tural failure and plasticity; 2000. 1984.
[29] Franchuk CR, Driver RG, Grondin GY. Block shear behaviour of coped steel beams. [46] Yam MCH, Cheng JJR. Fatigue strength of coped steel beams, Structural Engineering
Structural engineering report no. 244. Department of Civil and Environmental Engi- report no. 160. University of Alberta, Department of Civil & Environmental Engineer-
neering. Edmonton (AB, Canada), University of Alberta; 2002. ing; 1988.
[30] Franchuk CR, Driver RG, Grondin GY. Experimental investigation of block shear [47] Yam MCH, Cheng JJR. Fatigue strength of coped steel beams. J Struct Eng
failure in coped steel beams. Can J Civ Eng 2003;30(5):871–81. 1990;116(9):2447–63.
[31] Franchuk CR, Driver RG, Grondin GY. Reliability analysis of block shear capacity of [48] Roeder CW, MacRae GA, Kalogiros AY, Leland A. Fatigue crack of riveted, coped,
coped steel beams. J Struct Eng 2004;130(12):1904–13. stringer-to floorbeam connections. Final report no. WA-RD494.1Olympia, Washing-
[32] Canadian Standards Association (CSA). CAN/CSA-S16-09 Limit states design of steel ton: Washington Department of Transportation; 2001.
structures; 2009. Canadian Standards Association, Toronto, ON, Canada. [49] Roeder CW, MacRae G, Leland A, Rospo A. Extending the fatigue life of riveted coped
[33] Driver RG, Grondin GY, Kulak GL. Unified block shear equation for achieving consis- stringer connections. J Bridg Eng 2005;10(1):69–76.
tent reliability. J Constr Steel Res 2006;62:210–22. [50] Holden TA. Fatigue rehabilitation of coped steel beams using carbon fibre reinforced
[34] Topkaya C. Finite element modeling of block shear failure in coped steel beams. J polymers. [MSc thesis] Department of Civil and Environmental Engineering, Uni-
Constr Steel Res 2007;63(4):544–53. versity of Alberta; 2012.
[35] Fang C, Lam ACC, Yam MCH, Seak KS. Block shear strength of coped beams with [51] Holden TA, Cheng JJR, Lam ACC, Yam MCH. Analytical study of fatigue repair of coped
single-sided bolted connection. J Constr Steel Res 2013;86:153–66. beams using carbon fibre reinforced polymers. The International Association for
[36] Zhao XL, Hancock GJ. Longitudinal fillet welds in thin cold-formed RHS members. J Bridge Maintenance and Safety (IABMAS; 2010 [2010].
Struct Eng 1995;121(11):1683–90. [52] AS 4100-1998. Steel structures. Sydney, Australia: Australian Standard; 1998.
[37] Yam MCH, Zhong YC, Lam ACC, Iu VP. An investigation of the block shear strength of [53] Miner MA. Cumulative damage in fatigue. J Appl Mech 1945;67:A159–64.
coped beams with a welded clip angle connection-part I: experimental study. J
Constr Steel Res 2007;63(1):96–115.

Anda mungkin juga menyukai