Professional Documents
Culture Documents
T.J. Cunningham
PA Dept of Transportation, Montoursville, PA 17701, USA
&
J. G. Orbison & R. D. Ziemian
Departraent of Civil Engineering, Bucknell University, Lewisburg, PA 17837, USA
(Received 15 April 1994; received in revised form 5 October 1994; accepted 5 October 1994)
ABSTRACT
Block shear is known to be a potential failure mode for several different types of
connections. In this study,failure loads determined by experimental block shear tests
performed in North America are compared with the corresponding capacities
predicted by the current American Institute of Steel Construction Load and
Resistance Factor Design and Allowable Stress Design Specifications. The two
different design procedures are shown to provide a reasonable level of accuracy,
although they both exhibit a fairly wide variation in experimental-to-predicted load
capaciJ:y ratios. In an effort to identify and in turn, attempt to quantify the influence
of additional connection parameters that may influence block shear behavior, several
failure load design equations are developed to fit the experimental data. It is shown
that ( I ) the variation in the capacity ratios can be significantly reduced by
incorporating the "aspect ratio' of the block and the in-plane load eccentricity
measured to the bolt line, and (2)failure loads can be predicted with equal accuracy
using either the material yield strength or the ultimate strength.
NOTATION
323
324 T. J. Cunningham, J. G. Orbison, R. D. Ziemian
Ci Optimized coefficients
e~ In-plane shear eccentricity
et In-plane tension eccentricity
Fu Material ultimate stress
Fy Material yield stress
it, Gross length of tension plane
1,n Net length of tension plane
Iv, Gross length of shear plane
lw Net length of shear plane
Pexp Experimental failure load
Ppred Predicted load capacity
~scrv Service load capacity
Pult Ultimate load capacity
U Shear lag coefficient
Resistance factor
1 INTRODUCTION
A block shear failure can control the load capacity of several different
types of bolted connections, including shear connections at the ends of
coped beams, tension member connections and gusset plates. In all cases,
failure develops as yielding and/or rupture along two or more planes
bounding the connection. A typical block shear failure mechanism for a
single angle tension member is shown in Fig. 1. The 'block' of the
t P
Tension
Plane
~a
Pl__s
connected leg bounded by the fastener holes 'tears out' in this failure
mode; a net tensile force is developed along the upper edge of the block
(the tension plane) and a net shear force develops along the bolt line (the
shear pla:ae).
American provisions for determining design load capacities for this
failure mode first appeared in the American Institute of Steel Construction
(AISC) Specification in 1978.1 Currently, both the Allowable Stress De-
sign 2 (ASD) and the Load and Resistance Factor Design 3 (LRFD)
Specifications contain methods for estimating block shear design loads.
Both are based on one or more equations that estimate the resistance as
the sum of the tensile plane and shear plane forces at failure.
Over tlhe past two decades, many important experimental block shear
tests have been conducted by several researchers at different laboratories
in North America. In the present study, test data were obtained from
studies including research done at the University of Arizona, 4'5 the
University of Connecticut, 6 the University of Texas at Austin, 7'8 the
University of Alberta, 9 the University of Toronto 1° and the Western Area
Power Administration.11 Using a total of 77 individual tests or specimen
geometries that are known to have failed in pure block shear, their
corresponding experimental failure loads are compared in this study with
those predicted by the current AISC ASD and LRFD procedures.
It is shown that although these two specification procedures provide
overall reasonable levels of accuracy, there tends to be a significant
amount of variation in the experimental-to-predicted load capacity ratios.
Because of this variation, the influence of additional geometric parameters,
suggestecl by Hardash and Bjorhovde s and Epstein, 6 on block shear load
capacities was further studied. In an attempt to quantify their relative
significa~Lce, several load capacity equations were developed using the
above experimental data. It was determined that incorporating geometric
parameters, such as in-plane load eccentricity and the 'aspect ratio' of the
block, into the design procedures provided a means for significantly
reducing the variability between experimental and predicted behavior.
2 ,CURRENT A M E R I C A N D E S I G N P R O C E D U R E S
The current AISC LRFD procedure postulates two block shear failure
mechanisms. In the first, it is assumed that the failure load is reached when
rupture occurs along the tension plane and full yield is developed along
the shear plane. The corresponding load capacity is given by
The second failure mechanism assumes that the shear plane ruptures and
full yield is reached on the tension plane. In this case the load capacity is
given by
Am, Atg = net and gross areas of the tension plane, respectively;
Aw, Avg= net and gross areas of the shear plane, respectively;
Fy, Fu = material yield and ultimate strengths, respectively.
The AISC LRFD procedure permits the use of the larger failure load
predicted by these two equations. In both cases it is important to note that
the gross area is employed in predicting the resistance of the fully yielded
plane. The AISC LRFD Specification Commentary notes that this pro-
cedure is consistent with that used for tension member design, although in
the authors' opinion only net areas physically exist on the block shear
failure planes.
The AISC ASD procedure for estimating block shear service load
capacity employs the following equation:
3 C O M P A R I S O N O F AISC P R O C E D U R E S W I T H
EXPERIMENTAL DATA
The AISC LRFD procedure was first used to predict the block shear
capacities of the aforementioned experimental test specimens. Resistance
factors were not incorporated, as direct failure predictions were desired. In
accordance with the October 1992 draft of the AISC LRFD Specifica-
tion, 12 bolt hole sizes were taken as 1/16 in. greater than the nominal hole
diameter. In Fig. 2, the predicted ultimate loads are plotted against the
Block shear load capacity predictions 327
350kips • uw.~,~tT,~ s a ~ •
v u=,=-=tx ,~ An=*= • •
D unh~-~ ~ Cean~t~t.
300kips • u~=r=tT =r e , * m = ~ t t ~ t /
~'='!''* 17 ] ~ k ~ I /
250kips
3 - - E = = ~
o 200kips
'~ A zPm
150kips
0kips
0kips 50kips 100kips 150kips 200kips 250kips 300kips
Test Ultirnote Lood
experimental failure loads. Data points appearing below the diagonal line
indicate tests for which the AISC LRFD procedure is conservative (that
is, the load capacity is underestimated), while points above the line
indicate unconservative load predictions.
In Fig. 3, the resistance factor of ~b=0.75 was applied to the AISC
LRFD equations. With this factor, the AISC LRFD procedure more
conservatively predicts block shear load capacities. It is interesting to note
300kips . . . . * . . . . . . . . u . . . . i . . . . n . . . .
• Unlwraat7 c~ A.Ib~r~ • / /
V U=tmr=tLX~ Art=on~ • /
o u=~.~t7ol e==,.=u=ut • /
BkPe,lt ~ 0==17
250klps • u=,=.=t, .~ c m . u . =
Z~ Un~perllty =l
0 Un~mrmty~' 1~.oaLo / "
.~ 150klps L~ ^
5
~ 100klps
Test U l t i r n o t e Lood
Fig. 3. Experimental vs predicted failure loads (LRFD Procedure with resistance factor
=0"75).
328 T. J. Cunningham, J. G. Orbison, R. D. Ziemian
that five of the eight cases that are unconservative (above the diagonal
line) consisted of coped beam shear connections with two rows of bolts.
For each experimental test, a professional factor, s Pe.p/Pp,ed, was
computed using the .experimental failure load, Pexp, and the predicted
AISC LRFD load capacity, Pp,ea. As such, perfect agreement between
predicted and experimental failure loads is indicated by a professional
factor of unity; factors less than unity indicate tests where the design
procedure overestimates the failure load (and is thus unconservative),
while factors greater than unity indicate conservative load estimates.
Statistical measures of the professional factors are given in Fig. 4 and
Table 1. The resistance factor of ~b= 0.75 was not used in computing these
professional factors. In Figs 2-4, Table 1 and subsequent figures and
tables where indicated, several additional specimens are included that were
reported by Epstein. 6 These tests failed predominantly in block shear
although they also exhibited a partial second failure mode, such as net
section rupture. These test results are included in the 'All Data' statistics,
but not in the 'Block Shear' results.
Figure 4 and Table 1 indicate that the AISC LRFD procedure produces
an impressive professional factor mean of 1.005. However, a considerable
24 i i i
18
16
14
c
~2 77~
10 ~
//z
8 ,//,
6 !
~////~ z / / / .... ,//-/
0
0.4 0.6 0.8 1.0 1 i2 1.4 1.6
Professlonol Foctor
TABLE 1
Professional Factor Statistics for AISC LRFD Procedure
350kips
Unl'nmCLyof Albertt
V Unlver~ty of/u'lzoaa • •
Q tlnlvetmtty o~ connecu~t
300kips
• o.,~".",,,."7~.~" .... • S
a. 100kips
50klps
()kips
Okips 50kips lOOkips 150kips 200kips 250kips 300kips
Test Ultimote Lood
24 i i i J i
20
18
16
14
12
10
0
0.4 0.6 0.8 1.0 1.2 1.4 1.6
Professional Factor
TABLE 2
Professional Factor Statistics for AISC ASD Procedure
loads. As in Figs 2 and 3, data points appearing above the diagonal line
indicate unconservative load predictions.
Professional factors were also computed using the AISC ASD procedure
to calculate Ppr=d (the factor of 2"0 was included). Statistical measures of
the AISC ASD professional factors are provided in Fig. 6 and Table 2. A
mean of 1.075 and a standard deviation of 0.201 were obtained. Similar to
the AISC LRFD procedure, the histogram (Fig. 6) has a fairly wide
distribution. The AISC ASD procedure was also shown to significantly
Block shear load capacity predictions 331
overestirrtate failure loads in several cases (P.F. < 1.0) and in one particular
case overestimates the load capacity by 41.3%. Relatively conservative
load capacities were also frequently obtained.
Tension plane gross area, Atg Shear plane gross area, A,~
Tension plane gross length, Its Shear plane gross length, lvs
Tension plane net area, Atn Shear plane net area, Avn
Tension plane net length, ltn Shear plane net length, l~n
Tensile yield strength, Fy Ultimate tensile strength, Fu
Several other parameters were also studied. For example, the effects of
in-plane shear eccentricity, es and tension eccentricity, e,, were considered.
As shown in Fig. 7, these eccentricities are defined as the in-plane distance
from the line of action of the resultant shear or tensile force developed on
~ P
Sheor
Eccentricity, es
Tension
I Eccentricity, et
350kips . . . . , . . . . . . . . ~ . . . . , . . . . ~ . . . .
• Unlve~ityol Alberta
V e.~ver,Jty a ~:*n, • • •
[ ] Um~r~L7 of Co~Ucut
300kips mockm, , ~ ~i~,
P~domtn~uymockSh~r
/k UalV~lity of T|I~
250kips Q um~ty of Toronto
~es~rn~ Power Admmis~aUon
8
J
200klps
E
150klps
100kips
50kips uation I0
p u = 0,55AtnFu+ [ 1.55(ltn/lvn)0.25_ O, 1 es]AvnFy
Oklps . . . . L . . . . , . . . . , , , , , I , , i , I , i , ,
24 1" "1" 1 T r
20
18
16
14
12
10
4 -
0
0.4 0.6 0.8 1.0 1.2 1.4 1.6
Professionol Foctor
TABLE 3
Professional Factor Statistics for all Developed Equations (Block Shear Data Only)
5 CONCLUSIONS
With the exception of one connection having two rows of bolts, the
current AISC LRFD procedure, with resistance factors included, reason-
ably predicts block shear load capacities when compared with previously-
published experimental results. As the professional factor mean, range and
standard deviation from the AISC ASD procedure are similar to those
from the AISC LRFD procedure, the same conclusion can be extended to
the AISC ASD procedure. However, considerable variation in the profes-
sional factors from the two approaches is apparent.
The present study indicates that both the aspect ratio of the block and
the in-plane shear eccentricity may have a significant influence on the
block shear load contribution from the shear plane. Conversely, these
parameters appear to have no influence on the tension plane load
contribution. With respect to out-of-plane eccentricity, Epstein 6 suggests
applying the shear lag coefficient,3 U, to the tension plane area with angles
to approximately account for this affect. The statistical data observed in
Block shear load capacity predictions 337
the present study indicate that the load contribution of the shear plane is
affected by the bending moments being applied to that plane through load
eccentricity, and other factors that may be influential with relatively wide,
short connection blocks.
Numerical comparisons with the AISC LRFD and ASD procedures on
hypothetJLcal connections having aspect ratios in excess of 5.0 indicate the
equation:; developed to quantify the significance of the above effects may
be substantially conservative relative to the current procedures. The
authors note that three tests on coped beam simple connections having
block aspect ratios ranging from 16.1 to 17-9 are reported by Yura. s While
none of the connections failed in pure block shear, both the AISC LRFD
and ASD, block shear procedures predict failure loads within 16% of the
experimental failure loads, while the equations used in this study under-
estimatedL these loads by as much as a factor of two. Hence, the aspect
ratio and shear eccentricity may have no influence on the block shear load
capacity of connections with large aspect ratios.
It shotdd be noted that the experimental data obtained had net aspect
ratios (lvn/ltn) ranging from less than 1.0 to a maximum of slightly less than
5.0; no experimental pure block shear failure data was found that exceeded
this range. While actual connection aspect ratios smaller than those
included in the test data would probably be rare, aspect ratios greater than
5.0 can be obtained with relatively common connection geometries,
particularly in heavier coped beam shear connections. As this study has
been based upon the available experimental results, its conclusions,
particularly with respect to the aspect ratio and in-plane shear eccentricity,
cannot reasonably be applied to connections whose aspect ratios exceed
the experimental range. Regardless, the authors recommend that tests on
connections having large aspect ratios resulting in block shear failures be
conducted.
This study also indicates that the accepted 'shear rupture' constant of
0.6 may not accurately predict the shear plane load contribution. How-
ever, tests of bolted connections to verify this directly have not been found.
It has also been shown that various combinations of yield strength and
ultimate tensile strength in the tension and shear terms produce results of
similar accuracy. Use of the ultimate strength in the tension term and yield
strength JLnthe shear term would, however, appear to best reflect the actual
experimental behavior of the connections. In further regard to material
properties, it was also found that all of the experimental tests available
employed steel having yield strengths below 45 ksi, with the exception of
one, single, cold-rolled, high-strength gusset plate having a yield strength
of 49"5ksi. 5 The authors also recommend that block shear tests be
338 T. J. Cunninghara, J. G. Orbison, R. D. Ziemian
REFERENCES