Вы находитесь на странице: 1из 14

Pinned Connection Strength and Behavior

David Duerr, P.E., M.ASCE1

Abstract: Theoretical and experimental studies of pinned connections that have been published over the past 65 years are discussed.
Elastic stress distribution, service load deformations, and ultimate strength behavior are examined. The findings reported in the literature
are correlated and a single set of equations that define pinned connection strength and behavior are developed. The predictions of these
equations are compared to the reported test data and statistics are computed that may be used for the development of resistance factors or
design factors for use in design specifications. Recommendations for additional research are made.
DOI: 10.1061/共ASCE兲0733-9445共2006兲132:2共182兲
CE Database subject headings: Connections; Deformation; Design; Ultimate strength; Stress distribution.

Introduction Elastic Stress Distribution


The stress distribution in and failure mechanism of a pinned con- There are two concerns with respect to performance of pin-
nection are very complex. Resultantly, design code provisions are connected plates. These are elastic stress distribution prior to
typically empirical. For example, the American Institute of Steel yielding and fracture or instability at the ultimate load. Both of
Construction 共AISC兲 allowable stress design 共ASD兲 requirements these areas are affected by the plate geometry, material properties,
共AISC 1989兲 are based on an extensive test program conducted in and ratio of pin diameter to hole diameter. Peak elastic stresses
the 1930s 共Johnston 1939兲. The AISC load and resistance factor are generally only of concern in a connection that is to be
design 共LRFD兲 requirements 共AISC 2000兲 also consider the subjected to a high number of load cycles.
results of a second study 共Blake 1981兲. Both of these studies In order to discuss the existing data, we must first establish a
examined link plates of the configuration illustrated in Fig. 1. uniform notation. Some sources relate the tensile stress on the net
Pinned connections are commonly used in applications such as section to the ratio W / Dh and others use the ratio be / Dh. These
cranes and other lifting devices that require configurations that do two ratios are related as follows:
not conform to the simple pin-connected link plate on which these
specifications are based. Such configurations may be asymmetri- W Dh + 2be 2be
cal, have doubler plates to reinforce the area around the hole, or = =1+ 共1兲
Dh Dh Dh
have large clearances between the pin and the hole. In order to
adequately address these designs, one must develop an under-
standing of the behavior of pinned connections that goes beyond be W 1
= − 共2兲
the application of specification provisions. Dh 2Dh 2
A number of studies of pinned connection behavior have been
The tensile stress in the net section is distributed as shown in Fig.
published over the past 65 years. However, few of these studies
2 at a load low enough that the maximum tensile stress does not
reference one another and the details of the work and findings are
exceed the yield stress. The minimum stress f min may be either
not presented in a consistent format. The purpose of this paper is
tensile 共as shown兲 or compressive, depending on the proportions
to summarize the results of these various test programs and the-
of the plate. The maximum tensile stress f t max is typically the
oretical studies of pinned connections. The data are used to verify
value of interest.
previously developed expressions that quantify stress distribution
Elastic stress concentration factors have been reported in the
and ultimate strength or to develop new expressions, as appropri-
literature for a variety of plate configurations. Below are brief
ate, for each limit state. To make the information most useful to
discussions of three significant investigations of stress concentra-
both the researcher and the practicing engineer, the work from the
tions in pin-connected plates, followed by a summary of this
various sources is presented in a consistent notation. Last, recom-
work.
mendations are made for additional research needed to further
Frocht and Hill 共1940兲 report the results of two studies: one
improve pinned connection design methods.
using strain gages on aluminum specimens and the other using
photoelastic measurements. In addition to varying the proportions
1
President, 2DM Associates, Inc., Consulting Engineers, 9235 Katy of the plates under study, the aluminum plates were also loaded
Freeway, Suite 350, Houston, TX 77024-1526. E-mail: duerr@2dm.us by either steel or aluminum pins and lubrication of the pin was
Note. Associate Editor: Brian Uy. Discussion open until July 1, 2006. introduced in some tests. Stress concentration factors are pre-
Separate discussions must be submitted for individual papers. To extend
sented as both the ratio of maximum tensile stress to average
the closing date by one month, a written request must be filed with the
ASCE Managing Editor. The manuscript for this paper was submitted for tensile stress on the net section and the ratio of maximum tensile
review and possible publication on December 2, 2003; approved on May stress to average bearing stress between the pin and the edge of
23, 2005. This paper is part of the Journal of Structural Engineering, the hole.
Vol. 132, No. 2, February 1, 2006. ©ASCE, ISSN 0733-9445/2006/2- Consider the basic case of a pin and plate arrangement in
182–194/$25.00. which the pin and the plate are of the same material and the pin is

182 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006


Fig. 3. Ekvall 共1986兲 lug configuration
Fig. 1. Johnston 共1939兲 and Blake 共1981兲 specimen configuration

a close fit in the hole 共i.e., D p ⬇ Dh兲 with no lubrication. Six of the concentration factor as the ratio of the maximum tensile stress in
aluminum specimens fit this description. The proportions of these the net section divided by the average bearing stress between the
six specimens had ratios of a / be in the range of 0.68–9.18. The projected area of the pin and the edge of the hole. Further, the
results of the photoelastic tests were presented as a set of curves stress concentration factors are plotted with respect to the ratio
for different a / be ratios. Thirteen points from the curve for W / Dh for ␤ = 0° and with respect to the ratio 2Ro / Dh for
a / be = 1.00 are used in this examination. ␤ = 45°. These data must first be converted to express the stress
Scott and Stone 共1982兲 report stress concentration factor re- concentration factor as the ratio of maximum tensile stress to
sults based on photoelastic studies of pin-connected lugs. The average tensile stress on the net section and to relate these factors
primary stress concentration factor findings were based on six to the ratio be / Dh. Let the factor related to pin bearing stress be
photoelastic specimens of the configuration shown in Fig. 1 with Ktb and that related to average tensile stress be Kt
the corners clipped. Three of the specimens had bushings fitted in
the holes and three did not. Different pins were used to investigate f t max
the effect of pin-to-hole clearance. The ratio a / be ranged from Kt = 共3兲
P/共2bet兲
1.11 to 1.19. Of initial interest here are three data points for the
specimens with neat fit pins and no bushings.
Ekvall 共1986兲 reports the results of stress concentration factors f t max
determined by finite element analysis. Plate configurations and Ktb = 共4兲
P/共D pt兲
notation are shown in Fig. 3. Ekvall 共1986兲 expresses the stress

Ktb关P/共D pt兲兴 P 2bet


Kt = = Ktb 共5兲
P/共2bet兲 D pt P

2be 2be
Kt = Ktb ⬇ Ktb 共6兲
Dp Dh
The ratio 2Ro / Dh is converted to W / Dh when ␤ = 45° as follows:

Ro W 2Ro 1
W=2 ⇒ = 共7兲
cos 22.5° Dh Dh cos 22.5°
Ten data points can be taken from Ekvall 共1986兲 with respect to
the criteria discussed above for the previous sources. The ratio
a / be is equal to or less than 1.00 for all of the specimens.
The data points taken from the referenced sources are com-
piled in Fig. 4. The curve drawn through the figure is a graph of
a stress concentration factor curve Kt as defined by the following
Fig. 2. Net section elastic stress distribution equation:

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006 / 183


Fig. 4. Stress concentration factors Fig. 6. Increase in Kt relative to neat fit pin

Kt = 1.5 + 2.5
be
Dh
冉 冊
− 0.27
be
Dh
2
共8兲
acted at an angle of 45° from the axis of symmetry of the plate. In
the other three analyses, the angle of the load was 90°. The stress
concentration factor was given as the ratio of peak tensile stress to
The best agreement between Eq. 共8兲 and the test data occurs for average pin bearing stress. However, the location of the peak
specimens with neat fit pins and a / be ⬇ 1. The most pronounced tensile stress was not reported, so these data cannot be resolved
deviations from the Eq. 共8兲 curve are the Frocht and Hill 共1940兲 into the form of peak tensile stress divided by average tensile
tests for which a / be was significantly greater or less than 1.00. stress.
Two issues with respect to stress concentrations that are inves- Compared to the stress concentration factors for symmetrical
tigated in the literature only lightly are the effects of pin-to-hole loading 共i.e., load angle= 0°兲, the stress concentration factors at a
clearance and the application of the load at an angle to the axis of 45° load angle are reported to be 7–15% greater and the factors at
symmetry of the plate. a 90° load angle are 7–13% lower. The lower percentages occur at
Scott and Stone 共1982兲 define the pin-to-hole clearance by the values of be / Dh of 0.7 and greater, again showing very little
ratio ␩ change as the value of be / Dh increases.
In summary, the tensile stress concentration factor for a sym-
e Dh − D p Dp metrical pin-connected plate for which a ⬇ be, the load acts along
␩= = =1− 共9兲
r Dh Dh the axis of symmetry of the plate, Dh ⬇ D p, and the pin and plate
where 2e = c = Dh − D p and r = Dh / 2. The stress concentration are of the same material can be reasonably estimated by Eq. 共8兲.
factor is seen to increase sharply from D p / Dh ⬇ 1.00 to If the pin is other than a neat fit, the reported investigations indi-
D p / Dh ⬇ 0.98. Beyond this point, a further increase in clearance cate that the stress concentration factor may be as much as 17%
has a negligible effect. Fig. 5 is a curve plotted from the Scott and greater than the value given by Eq. 共8兲. As other deviations from
Stone 共1982兲 series with be / Dh = 0.50. Although the reported data this basic configuration are introduced, estimation of the stress
are not extensive, one trend is clear. The increase in stress con- concentration factor becomes increasingly uncertain due to the
centration factor due to pin clearance is greater for smaller values lack of data reported in the literature.
of be / Dh. The increase from a neat-fit pin to the plateau of the
curve is between 10 and 17% for values of be / Dh that are com-
mon in practice 共0.5 and greater兲. Fig. 6 is a plot of the increase in Bearing Stress and Deformation
Kt for a clearance fit pin relative to a neat fit pin as reported by
Scott and Stone 共1982兲 for a range of be / Dh values. Initial distress of a pin-connected plate typically manifests itself
Ekvall 共1986兲 reported stress concentration factors for six as local deformation of the plate immediately beyond the
analyses of a plate of the configuration shown in Fig. 3 with pin. Johnston 共1939兲 and Rex and Easterling 共2003兲 examined
␤ = 45° and asymmetric loading. In three of the analyses, the load deformations prior to failure.
Rex and Easterling 共2003兲 report results applicable to local
deformation of the plate at service loads. The research reported
therein includes tests of steel plate specimens and finite element
analyses. The observed deformations are attributable to a combi-
nation of local bearing deformation, shear deformation, and
bending deformation.
Rex and Easterling 共2003兲 propose a plate stiffness Ki defined
by Eq. 共10兲, which is based on the model illustrated in Fig. 7
1
Ki = 共10兲
1 1 1
+ +
Kbr Kb Kv
where Kbr = stiffness component attributable to local bearing;
Kb = stiffness component attributable to bending deformation
Fig. 5. Pin clearance effect on stress concentration of the plate beyond the hole; and Kv = stiffness component

184 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006


and plate, and ⌬ = Dh / 2 − D p / 2 + ⌬br. The compressive stress
throughout the full area of the bearing deformation is assumed to
exceed Fy
A p = sin共␣1兲D pt 共14兲

␣1 = cos−1 冉 0.25D2h − 0.25D2p − ⌬2


D p⌬
冊 共15兲

It is noted that Rex and Easterling 共2003兲 used A p = ␣1D pt in


their derivation of Kbr. For the test specimens considered,
sin共␣1兲 ⬇ 0.95␣1, so the difference is relatively small. However,
the difference becomes significant at smaller pin clearances.
Eq. 共14兲 is considered to produce better results in these configu-
rations. The bearing area A p defined by Eq. 共14兲 has an upper
bound width equal to the diameter of the pin.
An initial bearing deflection ⌬br of 0.10 mm is assumed. The
pin load P = A pFy and the stiffness Kbr = P / ⌬br. The following
equation is derived from these relationships:
Fig. 7. Rex and Easterling 共2003兲 plate deformation model
Kbr = 10C sin共␣1兲D ptFy 共16兲

attributable to shear deformation of the plate beyond the hole in which C = 1.0 in SI units and 25.4 in USCU. It is noted that
10C sin共␣1兲 ⬇ 120 in USCU for the range of specimen dimensions
Kbr = 120tFy共D p/25.4兲0.8 共11兲 reported in Rex and Easterling 共2003兲, so Eq. 共16兲 gives results
very similar to those of Eq. 共11兲 relative to the test data.
Kb = 32Et共R/D p − 1/2兲3 共12兲 Examination of the load-deflection test data that underlie Rex
and Easterling 共2003兲 shows that the plate deformations are ap-
Kv = 6.67Gt共R/D p − 1/2兲 共13兲 proximately linear up to deflections of about 0.25 mm. This is
consistent with the load-deformation curves in Johnston 共1939兲.
The coefficient 25.4 in Eq. 共11兲 is dropped when using US Cus- Thus, the stiffness value given by Eq. 共16兲 can be used to com-
tomary Units 共USCU兲. Examination of the three stiffness compo- pute local bearing deformations ⌬br up to about 0.25 mm with
nents for a range of plate proportions and material strength grades reasonable reliability.
shows that Kb and Kv are generally an order of magnitude or more Johnston 共1939兲 measured the longitudinal deformation be-
greater than Kbr. tween the pin and the plate relative to a point on the body of the
Eq. 共11兲 is based on the model shown in Fig. 8 and the results plate about 215 mm from the center of the pin and plotted a
of finite element analyses in which the initial bearing deformation load-deformation curve for each test. These curves were used to
was taken as 0.10 mm. The validity of the equation was further determine the general yield load of each plate, defined as the load
checked by comparison to test data for which D p = 19, 22, or corresponding to the point at which the slope of the curve is three
25 mm, and Dh = D p + 1.6 mm. These limits are practical for the times the initial slope. Johnston 共1939兲 proposed the following
bolted connections of interest in Rex and Easterling 共2003兲, but equation for the calculation of the general yield load in terms of
are too limited for the wider range of proportions common to the average bearing stress on the projected area of the pin:

冋 冉 冊 册
pinned connections. A more general form of the bearing stiffness
2
can be derived using the Rex and Easterling 共2003兲 model. Fy a a c
fp = 3 − −2 共17兲
The local bearing stiffness model assumes that the pin bearing 2 Dh Dh Dh
area A p is given by Eq. 共14兲, in which ␣1 is defined by Eq. 共15兲
and illustrated in Fig. 8. Note in Fig. 8 that ⌬br is the local bearing Application of Eq. 共17兲 is limited to details with small pin clear-
deformation, 2␣1 is the angle of the arc of contact between the pin ances, i.e., values of c not significantly greater than 0.067D p 关the
maximum value in the Johnston 共1939兲 tests兴. Inspection of Eq.
共17兲 shows that the results will be negative at large values of c.
Comparison of the initial bearing load P = A pFy with the mea-
sured general yield loads raises an important point. On average,
the bearing load is about 81% of the general yield load for the 106
Johnston 共1939兲 specimens. However, this calculated bearing load
is greater than the general yield load for 24 specimens, all of
which had pin clearances less than 0.024D p. This indicates that
the bearing area given by Eq. 共14兲 and the stiffness given by
Eq. 共16兲 are most likely overestimated for plates with very close
fitting pins.
A second point is also noted with respect to deformation be-
havior. The methods by which the plate deformations were mea-
sured by Johnston 共1939兲 and by Rex and Easterling 共2003兲 were
not identical. Likewise, pin deflection measurements reported in
Blake 共1981兲 and Duerr and Pincus 共1985兲 were also made in
Fig. 8. Rex and Easterling 共2003兲 bearing stiffness model different manners. Therefore, comparisons of test results from

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006 / 185


Fig. 10. Duerr and Pincus 共1985兲 specimen 1-A

Fig. 9. Plate failure modes


be
Pult = f pD pt = 2Fu D pt 共19兲
Dh
For close-fit pins, Dh ⬇ D p, so Eq. 共19兲 can be simplified to the
following equation, which is simply the net area multiplied by the
different studies can appear inconsistent. The information pres-
ultimate tensile strength of the plate material:
ently available, particularly from the older work, is not detailed
enough to allow normalization of the reported data to allow direct Pult = Fu2bet 共20兲
comparisons.
It is intuitively obvious that Eq. 共20兲 will not give accurate results
for a very wide plate. At some point, additional material to the
sides of the hole will not provide a corresponding increase in
Static Strength Limit States of Plate strength. To account for this behavior, the strength of the plate in
net section tension is calculated using an effective width bef f . The
Pinned connections exhibit six strength limit states. These are effective width is a function of the ductility of the material and of
tension in the net section through the hole 关Fig. 9共a兲兴, splitting on the stress distribution. The limit state of tension in the net section
a single plane beyond the hole 关Fig. 9共b兲兴, shear on two planes is thus expressed as the following equation:
beyond the hole 关Fig. 9共c兲兴, out-of-plane instability, also called
“dishing” 关Fig. 9共d兲兴, shear of the pin, and bending of the pin. Pult = Fu2bef f t 共21兲
The following sections treat each strength limit state of the plate
The ratio of ultimate tensile strength to yield stress Fu / Fy is a
by presenting relevant test data, examining existing strength
practical indicator of the material ductility using values that are
equations, and, where appropriate, proposing new limit state
normally available in practice. The stress concentration factor of
expressions.
Eq. 共8兲 can be used to provide an indication of the effect of initial
yielding on the effective width. An expression for effective width
Tension in Net Section is derived as follows.
Consider Duerr and Pincus 共1985兲 specimen 1-A, shown after
Johnston 共1939兲 proposed Eq. 共18兲 for the calculation of the ulti- failure in Fig. 10. Dimensions and other pertinent data are given
mate load of a plate that fails in tension through the net section. in Table 1. Eq. 共21兲 is solved to find a value of bef f = 27.7 mm.
This equation expresses the ultimate load in terms of the average Given be / Dh = 0.687 共based on the average be兲, Eq. 共8兲 gives
bearing stress f p at failure between the pin and the edge of the
hole based on the projected area of the pin on the plate
Table 1. Specimen 1-A, derived from Duerr and Pincus 共1985兲

be Dh 50.6 mm a 35.9 mm
f p = 2Fu 共18兲
Dh Dp 50.4 mm Fy 673 MPa
t 10.5 mm Fu 754 MPa
Left be 34.5 mm Pult 437 kN
Eq. 共18兲 can be written in terms of the failure load by multiplying
Right be 35.0 mm — —
through by the pin bearing area

186 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006


Fig. 13. Net section Cr curve and test values
Fig. 11. Effective width for specified material

Kt = 3.09. At the failure load of 437 kN, the average tensile stress
Given an asymmetrical plate in which the difference in net
on the net section is 601 MPa and the 共fictitious兲 peak stress is
widths on the two sides of the hole is not great, use of the average
1,858 MPa.
value of bef f in Eq. 共21兲 is reasonable. If the asymmetry is great
The proposal is to calculate an effective width for which the
enough that the ratio of the larger width be to the smaller be is
limit state load computed with Eq. 共21兲 will give a peak stress
greater than about 1.30, then use of the smaller value of bef f
equal to that shown above for this particular material. The follow-
appears to give superior results.
ing equation is proposed:
Tests of connections in which the pin is smaller than the hole
bef f
be
= 0.6
Fu
Fy
冑 Dh
be
⇒ bef f = be0.6
Fu
Fy
冑 Dh
be
艋 be 共22兲
by varying amounts 共Blake 1981; Duerr and Pincus 1985兲 show
that the strength of the plate diminishes as the pin-to-hole clear-
ance increases. This observation is generally consistent with the
There are 12 tests reported in which the plate failed in net section increase in the peak tensile stress discussed in the previous sec-
tension. Of these, only two plates with Dh ⬇ D p failed at loads tion, although the elastic stress increase and ultimate strength de-
significantly lower than that predicted by Eq. 共20兲. Both were of crease do not appear to exhibit the same pattern. As suggested by
high-strength steel with Fu / Fy = 1.121 共Duerr and Pincus 1985 Tolbert and Hackett 共1974兲, a practical means of accounting for
specimens 1-A and 5-A兲. The curve of Fig. 11 is a plot of Eq. 共22兲 this effect is to multiply the plate strength as determined using
for Fu / Fy = 1.121 through a range of 0.50艋 be / Dh 艋 1.50. The two Eq. 共21兲 with a capacity reduction factor Cr. The following equa-
points plotted are from the test specimens. tion defines a reasonable value for Cr as a function of the ratio of
The curve of Fig. 12 illustrates the effect of Eq. 共22兲. The ratio the pin and hole diameters. A comparison of this curve to four test
of fictitious peak tensile stress f t max to ultimate tensile strength is data points is shown in Fig. 13
calculated for Fu / Fy = 1.121 over a range of values of be / Dh as


follows. Pult is computed using Eq. 共21兲 with bef f from Eq. 共22兲
and the peak tensile stress is taken as f t max = Kt Pult / 2bet. By re- D2p
Cr = 1 − 0.275 1− 共23兲
lating this expression with Eq. 共21兲, the stress ratio plotted in D2h
Fig. 12 reduces to f t max / Fu = Ktbef f / be. Once the proportions of
the plate reach the point at which bef f ⬍ be, the curve becomes The strength of a pin-connected plate in the limit state of tension
nominally flat. in the net section is given by

Pn = Cr2bef f tFu 共24兲


where Cr is as given by Eq. 共23兲 and bef f is as given by Eq. 共22兲,
using either the average bef f or minimum bef f , as discussed above.
Table 2 shows a comparison of the test results to the predicted
limit state loads given by Eq. 共24兲. The ratios of test verses pre-
diction for the 12 specimens are plotted against the ratio be / Dh in
Fig. 14.

Splitting Failure Beyond Hole


Johnston 共1939兲 proposed Eq. 共25兲 for the calculation of the fail-
ure load of a pin plate beyond the hole, again written in terms of
the average bearing stress between the pin and the plate. A modi-
fication of the Johnston equation to be in terms of the failure load
is shown as Eq. 共26兲 and a modification of this expression for a
Fig. 12. Peak stress ratio versus width ratio close-fit pin 共Dh ⬇ D p兲 is shown as Eq. 共27兲

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006 / 187


Table 2. Comparison of Test Strengths with Calculated Strengths
Net section Beyond Double plane
tensiona splittingb shearc Dishingd
Number of specimens 12 48 18 80
Minimum Test/Prediction 0.985 0.809 0.735 0.651
Maximum Test/Prediction 1.071 1.144 1.040 1.387
Average Test/Prediction 1.042 1.024 0.952 0.961
Standard deviation 0.029 0.082 0.086 0.170
Coefficient of variation 0.028 0.080 0.090 0.177
a
Test data from Johnston 共1939兲, Blake 共1981兲, Duerr and Pincus 共1985兲.
b
Test data from Johnston 共1939兲, Tolbert 共1970兲, Blake 共1981兲, Duerr and Pincus 共1985兲, and Rex and Easterling 共2003兲.
c
Test data from Tolbert 共1970兲, Blake 共1981兲, Duerr and Pincus 共1985兲, and Rex and Easterling 共2003兲.
d
Test data from Johnston 共1939兲, Duerr and Pincus 共1985兲, and Rex and Easterling 共2003兲.

冢 冣
be the capacity reduction factor Cr defined by Eq. 共23兲 gives close
0.92 agreement between the test results and the predicted failure loads
a Dh
f p = Fu 1.13 + 共25兲 for these specimens. A comparison of this curve to these two test
Dh be data points is shown in Fig. 15.
1+
Dh The ratios of test failure load verses limit state load predicted
by Eq. 共27兲 exhibit a somewhat greater scatter for the group of 48

冢 冣
be specimens as compared to the 23 Johnston 共1939兲 specimens.
0.92
a Dh However, the scatter using the 48 specimens is not excessive and
Pult = f pD pt = Fu 1.13 + D pt 共26兲 supports continued use of this equation for design.
Dh be
1+ The strength of a pin-connected plate in the limit state of
Dh single plane splitting beyond the hole is given by the following
equation, where Cr is as given by Eq. 共23兲

冢 冣
0.92be
Pult = Fu 1.13a + t 共27兲

冢 冣
be 0.92be
1+ Pb = CrFu 1.13a + t 共28兲
Dh be
1+
Both the distance beyond the hole, a, and the distance to the side Dh
of the hole, be, affect the strength with respect to a splitting fail- The third column of Table 2 shows a comparison of the test re-
ure. In the case of an asymmetric plate, examination of the test sults to the predicted limit state loads given by Eq. 共28兲. The
data shows that the smaller value of be should be used in Eq. 共27兲. ratios of test verses prediction for the 48 specimens are plotted
Eq. 共25兲 is empirical, based on the results of 23 specimens that against the ratio a / Dh in Fig. 16.
failed in this splitting mode. The proportions of these specimens
are in the ranges 0.500艋 be / Dh 艋 1.157, 0.313艋 a / Dh 艋 0.905,
and 0.289艋 a / be 艋 1.206. All specimens were of mild steel Double Plane Shear Failure
and were loaded through close fitting pins 共D p / Dh 艌 0.938兲. The second limit state of the plate in the region beyond the hole is
The addition of specimens from Tolbert 共1970兲, Blake 共1981兲, that of shear on two planes that are parallel to each other and to
Duerr and Pincus 共1985兲, and Rex and Easterling 共2003兲 the line of the applied load 关Fig. 9共c兲兴. The locations off center of
brings the total to 48 and expands the ranges of dimensional the two shear planes can be defined by an angle ␾, as shown in
proportions to 0.340艋 be / Dh 艋 2.448, 0.313艋 a / Dh 艋 1.655, Fig. 17. The point at which a radial line drawn from the center of
0.273艋 a / be 艋 1.206, and D p / Dh 艌 0.496. the pin intersects the edge of the hole defines the position of the
Only two specimens in this group 共Duerr and Pincus 1985
specimens 1-D and 5-B兲 had large pin-to-hole clearances. Use of

Fig. 14. Test/prediction for net section failures Fig. 15. Beyond splitting Cr curve and test values

188 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006


Fig. 18. Shear failure with large pin-to-hole clearance

Fig. 16. Test/prediction for beyond splitting failures

Define the angle ␾ as


shear plane. The magnitude of ␾ for the case of a neat fit pin as
expressed in the literature ranges from 40° 共Shanley 1944兲 to an Dp
implied 90° 共AISC 2000兲. Test results published in Tolbert 共1970兲 ␾ = 55 共29兲
showed shear failure along two planes defined by an angle ␾ of Dh
about 55° for specimens with a neat fit pin. The length of the shear plane Z 共Fig. 18兲 is defined by Eq. 共30兲 for
The ultimate strength of the plate diminishes as the clearance a plate with a straight edge beyond the hole
between the pin and the hole increases 共all other dimensions re-
maining the same兲. Tolbert 共1970兲, Blake 共1981兲, and Duerr and
Pincus 共1985兲 are used to develop an expression for ␾ that agrees Dp
Z=a+ 共1 − cos ␾兲 共30兲
with the observed shear failure loads and that accounts for the 2
pin-to-hole clearance.
Calculation of the shear strength of the plate requires knowl- If the end of the plate is curved 共Fig. 19兲, as is often the case for
edge of the ultimate shear strength Fus of the material. This prop- pin-connected elements in practice, the length Z must be adjusted
erty is not measured in the standard suite of physical tests. In the accordingly. For the plate configuration shown in Fig. 19, in
studies cited herein, only Tolbert 共1970兲 measured the ultimate which the plate is symmetrical about the axis defined by the line
shear strength of the specimen material. Evaluation of the speci- of the applied load, the length of the shear area of each plane lost
mens from other sources must rely on typical tensile-to-shear to the curvature Z⬘ is given by the following equation, where R is
strength relationships. Lyse and Godfrey 共1933兲 report a series of the radius of curvature of the plate edge:
tests on mild steel in which the ultimate shear strength was mea-
sured to be 81% or more of the ultimate tensile strength. The
percentage was at the lower end of the observed range for higher
strength steels. Bibber et al. 共1952兲 report that the ultimate shear
strength of T-1 steel 共ASTM A514兲 is about 70% of the ultimate
tensile strength.

Fig. 17. Shear failure with neat fit pin Fig. 19. Shear area with curved edge

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006 / 189


trends that indicate decreasing accuracy as the indicated ratios
change, with the exception of four Blake 共1981兲 specimens. In
these cases, the calculated shear strength is greater than the test
failure load. Possible reasons for these differences may include:
共1兲 the shear area as defined by the calculated value of Z is not
fully effective for larger values of a / Dh; 共2兲 the estimated values
of the ultimate shear strength in this analysis overestimated the
true strength 共these four specimens, but none of the others, were
cut from the same two plates兲; or 共3兲 a combination of 共1兲 and 共2兲.
In the absence of more data and noting the lack of a trend in this
direction among the other specimens, there does not appear to be
a sound basis for altering Eq. 共30兲.
The loss of ultimate strength due to pin-to-hole clearance is a
function of the reduction in the length of the shear plane area, as
defined by Eq. 共30兲. This loss is not consistent for a given D p / Dh
ratio for plates of all configurations. As can be inferred from Eq.
Fig. 20. Test/prediction for shear failures
共30兲, the loss is smaller for plates with a larger a / Dh ratio. The Cr
term that is applied to the net section and splitting limit states is

冑 冉 冊 2
not applicable to the double shear limit state.
Dp
Z⬘ = R − R2 − sin ␾ 共31兲
2
Failure by Out-of-Plane Instability „Dishing…
The strength of a pin-connected plate in the limit state of double
plane shear is given by Slender pin-connected plates may fail by out-of-plane buckling,
commonly called dishing. Johnston 共1939兲 presents an empirical
Ps = 2ZtFus 共32兲 formula based on test results that can be used to predict the dish-
ing failure load 共the following equation兲. As with the previous
where Z is as given by Eq. 共30兲, modified if necessary to account
Johnston equations, the following equation is written in terms of
for the shape of the plate edge.
the average bearing stress between the pin and the hole:
The references cited report 18 specimens that failed in double
plane shear. D p / Dh ranged from 0.50 to 1.00, a / Dh from 0.37 to
1.86, be / Dh from 0.40 to 2.78, and a / be from 0.21 to 1.03. Most
of the specimens were of mild or high-strength, low-alloy steels
共Fu from 370 to 513 MPa兲; two specimens 共Duerr and Pincus
f p = 20 + 315
t
Dh
tbe a
+ 75 2 + 20 − 20
Dh Dh
a
Dh
冉 冊 2
共33兲

1985兲 were of high-strength quenched and tempered alloy steel


Eq. 共33兲 provides reasonably good results when compared to the
共Fu = 754 MPa兲. The ultimate shear strength was reported for the
Johnston 共1939兲 tests, but not when expanded to results from a
Tolbert 共1970兲 specimens. The ultimate shear strength was taken
wider range of plate proportions 共e.g., Rex and Easterling 2003兲.
as 0.70 Fu for the Duerr and Pincus 共1985兲 and Rex and Easter-
This section presents the derivation of an analysis method based
ling 共2003兲 specimens. The ultimate shear strength was estimated
on column buckling theory using the model shown in Fig. 22.
as varying percentages of Fu for the Blake 共1981兲 specimens
The proposed dishing model assumes that the plate above the
based on the Fu to Fus relationships shown in Lyse and Godfrey
pin is analogous to a cantilevered column. The critical buckling
共1933兲.
stress can be conveniently based on a slenderness ratio KL / r of
The fourth column of Table 2 shows a comparison of the test
the plate above the hole, where L is equal to plate dimension a
results to the predicted limit state loads given by Eq. 共32兲. The
and r is the radius of gyration of the plate in the through thickness
direction 共t / 冑12兲. Observations of plates that failed by dishing
scatter of the results with respect to D p / Dh is illustrated in Fig. 20
and with respect to a / Dh in Fig. 21. Neither of these plots shows
共e.g., Johnston 1939, Fig. 6兲 show that the buckled length of plate
may be something other than plate dimension a. That is, the
length of the cantilever is not necessarily equal to a. Plates that
are wide 共relatively larger be兲 are seen to provide less support to
the area of plate in compression above the hole, thus resulting in
a larger effective length 共i.e., the wider the spread of the centroids
of the tensile stress areas to either side of the hole, the greater the
tendency of the plate to buckle兲. An effective length factor that
accounts for this behavior can be expressed as

K=2 冑 be
a
共34兲

The plate may buckle elastically or inelastically. Plates for which


Cc = 冑2␲2E / Fy ⬍ KL / r will fail inelastically. The inelastic critical
plate buckling stress is defined by Eq. 共35兲 and the elastic critical
plate buckling stress by Eq. 共36兲, where ␷ is Poisson’s ratio 共0.3
Fig. 21. Test/prediction for shear failures for steel兲

190 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006


Fig. 23. Test/prediction for dishing failures

Easterling 共2003兲 specimens, so the nominal value of


200,000 MPa was used in the calculation of the predicted dishing
loads. Last, examination of the data show that no dishing failures
were observed in plates for which the proportions were as defined
by the following equation:

a Dh
t Dp
⬍ 0.19 冑 E
Fy
共40兲

Only three of the test specimens in this group had large pin-to-
Fig. 22. Dishing analysis model hole clearances and all three failed in the inelastic range. The
smaller pin diameter is accounted for in this case by Eq. 共37兲 and

冤 冥
the calculated results agreed very well with the test loads.
共KL/r兲2
1−
2C2c
Fcr = Fy 共35兲 Summary of Plate Strength Limit States
1 − ␷2
Given the above expressions for plate strength in each of four
␲ 2E limit states, the strength and expected mode of failure is deter-
Fcr = 共36兲 mined in a straightforward manner. The limit state load is com-
共KL/r兲2共1 − ␷2兲
puted using Eqs. 共24兲, 共28兲, and 共32兲, and, for plates in the range
This critical buckling stress acts on an effective area of the plate, defined by Eq. 共40兲 as susceptible to dishing, Eq. 共39兲. The lowest
equal to Wef f t, where Wef f is an effective width 共Fig. 22兲. Wef f is value defines both the predicted failure load and the mode of
the smaller of the values given by Eqs. 共37兲 or 共38兲. Eq. 共37兲 is failure. The 158 specimens considered in this study have been so
analogous to the effective width model used for some edge- analyzed. The results are summarized in Table 3.
loaded plate buckling problems. The upper limit of Wef f defined The mean and the coefficient of variation of the ratios of Test/
by Eq. 共38兲 is based on the test data. In the case of an asymmetric Prediction are consistent with similar results reported in the lit-
plate, the smaller value of be is used in Eqs. 共34兲 and 共38兲 erature for bolted and welded structural connections. The larger
Wef f = D p + a 共37兲 errors and the disagreements in predicted mode generally appear
attributable to issues discussed above, such as the lack of test
measurements of the shear strength or modulus of elasticity of
Wef f = Dh + 1.25be 共38兲
some of the specimen material.
The strength of a pin-connected plate in the limit state of dishing
is given by
Pd = Wef f tFcr 共39兲 Table 3. Analysis Results
where Wef f is as given by Eqs. 共37兲 or 共38兲 and Fcr is as given by
Number of Specimensa 158
Eqs. 共35兲 or 共36兲.
The three references cited report 80 specimens that failed by Minimum Test/Prediction 0.651
dishing. D p / Dh ranged from 0.49 to 1.00, a / Dh from 0.32 to 2.32, Maximum Test/Prediction 1.387
be / Dh from 0.49 to 2.04, and a / be from 0.28 to 2.17. All of the Average Test/Prediction 1.013
specimens were of mild or high-strength, low-alloy steels. Standard deviation 0.141
The fifth column of Table 2 shows a comparison of the test Coefficient of variation 0.140
results to the predicted limit state loads given by Eq. 共39兲. The Correct limit state prediction 76.6%
a
scatter of the results with respect to a / t is illustrated in Fig. 23. Test data from Johnston 共1939兲, Tolbert 共1970兲, Blake 共1981兲, Duerr and
The modulus of elasticity was not measured for the Rex and Pincus 共1985兲, Rex and Easterling 共2003兲.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006 / 191


Fig. 24. Main plate reinforced with doubler plates

Pin Hole Doubler Plates

A common practice in the design of pinned connections is the use


of doubler plates 共also called cheek plates or side plates兲 on one
or both sides of the main plate to improve the strength of the
region around the hole 共Fig. 24兲. Among the specimens reported
herein, only two were fabricated with doubler plates 共Duerr and
Pincus 1985 specimens 5-A and 5-B兲. Calculation of the strength
of these two specimens was accomplished by calculating the
strengths of the individual plates and summing the values. This
approach yielded good agreement with the test results.
An alternate approach used in practice assumes that the ap-
plied load is distributed between the pin and the plate as uniform
bearing. Using this approach, the calculated strength of each plate
is compared to the applied load so determined. In the absence of
test data, the performance of this approach cannot be established.
Regardless of the method used to design the plates and their con-
nections to one another, the design of the pin and the design of the
plates must be based on the same loading model.

Pin Strength

The pin is often evaluated using a simple loading model 关Fig.


25共a兲兴. The most common pin loading model used in practice
共Blake 1974兲 considers the connection force to be transferred be-
tween plates and pin in uniform bearing across the full thick-
nesses of the plates 关Fig. 25共b兲兴. Given this loading, the resulting
shear and moment in the pin are calculated and compared to
appropriate limit states using simple beam theory. In many prac-
tical design situations, this method provides satisfactory results.
While the assumption of uniform bearing is a convenient ap-
proximation, it is intuitively obvious that the bearing stresses are Fig. 25. Pin loading models
not uniform. Bending deflection of the pin and bearing deforma-
tion of the plates will create a varying bearing stress distribution
with the load concentrated toward the edges of the plates. Melcon
and Hoblit 共1953兲 present a method of pin analysis that accounts the strength of the center plate relative to the applied load and the
for this behavior. An effective width of bearing between the pin stiffness of the center plate, expressed as the value r = a / t.
and the center plate of a symmetrical connection is defined as Application of this method is as follows. The strength of the
␥t / 2 关Fig. 25共c兲兴. This analysis method is based on the observa- center plate Pult is computed as previously discussed. The value
tion that the transfer of the load in bearing between the pin and of ␥ is calculated using the following equation when r 艌 0.55.
the center plate builds 共or “peaks”兲 toward the sides of the plate. When r ⬍ 0.55, ␥ must be determined from the curves of Fig. 26
The value of ␥, referred to as the peaking factor, is a function of 共based on Melcon and Hoblit 1953 Fig. 4兲.

192 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006


2. Local plate deformation behavior is reasonably understood
for details with pin clearances in the range of about
0.01D p – 0.10D p. Additional study is required to better quan-
tify this behavior for plates with large pin clearances.
3. Very little experimental data exist by which definition of the
effective net width for tensile strength 关Eq. 共22兲兴 can be veri-
fied. Further testing of specimens proportioned to fail in the
net section and of various grades of steel should be con-
ducted.
4. Very little experimental data exist by which the effect of
pin-to-hole clearance 关Eq. 共23兲兴 can be verified. Further test-
ing of specimens with a range of pin-to-hole diameter ratios
should be conducted. Specimens should be proportioned to
fail in each of the four modes 共Fig. 9兲 and should be fabri-
Fig. 26. Peaking factor curves 共from Melcon and Hoblit 1953兲
cated from a range of material strengths.
5. Only two specimens in the references cited were fabricated
with doubler plates around the hole. Testing of specimens
Pult
␥ = 0.5 共41兲 fabricated with doubler plates should be conducted. Such
F uD pt tests should be designed to investigate both the plate behav-
The effective width of bearing at each side of the center plate is ior and the interaction of the pin and the plates.
taken as ␥t / 2. The shear and moment in the pin are computed 6. The past work has questioned the effect of the method of
using this loading configuration. cutting the edge beyond the hole on the strength of the plate
Melcon and Hoblit 共1953兲 also present an extension of this 共e.g., sawed, sheared, flame-cut兲. The present data are not
concept that is applicable to the case where the center plate is extensive enough to allow study of this effect. Tests of iden-
significantly stronger than is required by the design application tical specimens fabricated using different cutting methods
共referred to as being “overstrong”兲. In this case, the method should be conducted to quantify this effect.
makes the assumption that the full thickness of the plate is not 7. Given the available data, the limit state expressions devel-
required and, therefore, a center portion of the plate can be ig- oped herein provide an adequate degree of accuracy for prac-
nored and only the outer widths considered in the analysis. The tical design use. The statistics reported can be used to de-
load is assumed carried only by two strips of some effective velop resistance factors for limit state standards or design
thickness on either side. factors for allowable stress standards.
The strength Pult of the center plate is calculated based on the
net effective thickness, ␥ is computed, and the pin analyzed as
above. The thickness of the “inactive” section of plate is a some-
what randomly selected value. Therefore, this method may re- Acknowledgment
quire a number of iterations to arrive at a solution that demon-
strates adequate strength in both the pin and the plates. The analysis method for the limit state of out-of-plane buckling
Melcon and Hoblit 共1953兲 report that this pin analysis method 共dishing兲 is based on unpublished work by George Pincus, Ph.D.,
is based on a series of tests on pins and plates of varying propor- P.E. of the University of Houston.
tions and strengths. However, the results of individual tests are
not given, nor is a source cited for these data.
The Melcon and Hoblit 共1953兲 method does not appear to be
References
acceptable for use in a connection in which the center plate is
reinforced with doubler plates. The pin must be of adequate
American Institute of Steel Construction 共AISC兲. 共1989兲. Specification
strength and stiffness to distribute the applied load among the
for structural steel buildings—allowable stress design and plastic
main plate and the doubler plates in a manner consistent with the design, 9th Ed., Chicago.
assumptions made in the design of the plates. American Institute of Steel Contruction 共AISC兲. 共2000兲. Load and resis-
tance factor design specification for structural steel buildings, 3rd
Ed., Chicago.
Conclusions Bibber, L. C., Hodge, J. M., Altman, R. C., and Doty, W. D. 共1952兲. “A
new high-yield-strength alloy steel for welded structures.” Trans.
This paper presents a review of theoretical and experimental stud- ASME, 74共3兲, 269–285.
ies of pinned connections and correlates the findings of these Blake, A. 共1974兲. “Structural pin design.” Des. News, 29共2兲, 88–94.
sources to develop a single consistent set of equations that define Blake, G. T. 共1981兲. Structural tests of large pin-connected links, U.S.
pinned connection strength and behavior. Both elastic stress and Steel Research, Monroeville, Pa.
ultimate strength equations are presented. Based on this informa- Duerr, D., and Pincus, G. 共1985兲. “Experimental investigation of pin
plates.” Research Rep. No. UHCE 85-3, Univ. of Houston, Houston.
tion, the following conclusions and recommendations are made.
Ekvall, J. C. 共1986兲. “Static strength analysis of pin-loaded lugs.”
1. Elastic stress concentrations are affected by numerous char-
J. Aircr., 23共5兲, 438–443.
acteristics of the pin/plate arrangement. Available research Frocht, M. M., and Hill, H. N. 共1940兲. “Stress-concentration factors
and design guides provide useful practical solutions, but around a central circular hole in a plate loaded through pin in the
must be regarded as approximate. The level of accuracy is hole.” Trans. ASME, J. Appl. Mech., 62共1兲, A5–A9.
suitable for design using appropriate resistance factors or de- Johnston, B. G. 共1939兲. “Pin-connected plate links.” Trans. Am. Soc. Civ.
sign factors. Eng., 104, 314–339.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006 / 193


Lyse, I., and Godfrey, H. J. 共1933兲. “Shearing properties and Poisson’s the critical stresses of eye bars under load: An evaluation by photo-
ratio of structural and alloy steels.” Proc., ASTM International, elastic modeling.” Proc., Joint Conf. on Experimental Mechanics of
Vol. 33, West Conshohocken, Pa., 274–292. the Society for Experimental Stress Analysis, Honolulu.
Melcon, M. A., and Hoblit, F. M. 共1953兲. “Developments in the analysis Shanley, F. R. 共1944兲. Basic structures, Wiley, New York.
of lugs and shear pins.” Prod. Eng. (N.Y.), 24共6兲, 160–170. Tolbert, R. N. 共1970兲. “A photoelastic investigation of lug stresses and
Rex, C. O., and Easterling, W. S. 共2003兲. “Behavior and modeling of a failures.” Master of Science thesis, Vanderbilt Univ., Nashville, Tenn.
bolt bearing on a single plate.” J. Struct. Eng., 129共6兲, 792–800. Tolbert, R. N., and Hackett, R. M. 共1974兲. “Experimental investigation of
Scott, R. G., and Stone, J. C. 共1982兲. “The effects of design variables on lug stresses and failures.” Eng. J., 11共2兲, 34–37.

194 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / FEBRUARY 2006

Вам также может понравиться