Sunteți pe pagina 1din 10

Strength Determination of Heavy

Clip-Angle Connection Components


James A Swanson, Ph.D. and Xiaojiang Gao
University of Cincinnati
Dept. of Civil & Environmental Engineering
PO Box 210071 - 741 Baldwin Hall
Cincinnati, OH 45221-0071

ABSTRACT
Most existing research data on clip-angle connections centers around connections that use light
to medium weight angles (t = 1/4 to 5/8) for connecting the beam flanges to the column flange.
With the increased use of bolted connections in low to mid-rise structures, stronger and stiffer
top-and-seat angle connections are needed to fill the gap between T-stub or end-plate
connections and lighter angle connections. The results of ten cyclic tests of heavy angle
components (t = 1) are presented in this paper and are used to evaluate currently accepted
design procedures.
INTRODUCTION
Two recent seismic events, the Northridge and Kobe earthquakes, exposed several
weaknesses in fully welded beam-column connections. In the time since, several research
projects have been conducted to identify and correct flaws in welded connection design
procedures and to investigate alternative connection designs. One of those projects, SAC
subtask 7.03, was conducted at the Georgia Institute of Technology and focused on bolted Tstub and top-and-seat-angle connections as alternatives to fully welded connections for light to
medium-weight beam sizes. The investigation consisted of tests of T-stub and angle
components subjected to cyclic axial loads. The tests were designed to isolate flange
components of bolted moment connections so that an economical parametric investigation
could be conducted. A discussion of the experimental program was provided by Swanson (1)
and Swanson and Leon (2).
Although, the primary focus of the Georgia Tech project was the study of T-stub connections, a
series of ten angle components were also tested. The data from this series of tests, however,
was not thoroughly reduced and no design recommendations were made. This is unfortunate
because the angle components that were tested were much heavier than those used in typical
top-and-seat angle connections and represent an important addition to the existing base of data.
The objectives of this paper is to present the angle test data collected at Georgia Tech, to
evaluate two existing ultimate strength models for angle components, and to present an
alternative ultimate strength model.

SAC is a joint venture made up of the Structural Engineers Association of California (SEAOC), the
Applied Technology Council (ATC), and the California Universities for Research in Earthquake
Engineering (CUREe). SAC is funded by the Federal Emergency Management Association (FEMA).

BACKGROUND
The determination of the ultimate strength of an angle component subjected to an axial load is a
complex process. Of the possible failure modes, the most studied case is the development of a
bending mechanism in the upstanding angle leg followed by failure of the tension bolts (i.e. the
formation of a prying mechanism). In this paper, the prying models used by the AISC-LRFD (3)
and Eurocode (4) specifications will be evaluated by comparing their strength predictions to the
results of ten angle component tests that were conducted at the Georgia Institute of Technology
(1, 2). A new strength model, based on a combination of the first two models, will then be
presented and evaluated. All three of the models are based on the prying formulation
developed by Kulak et al. (5) for T-stub components. Although several other prying formulations
have been developed, including those developed by Chen et al. (6) and Jaspart and Maquoi
(7), they will not be addressed in this work.
In the discussion of the existing models, the notations used by the original authors will be
converted to that used in this work so that a clearer comparison can be made. The notation that
will be used is illustrated Figure 1. The analysis of an angle flange is made easier by
considering the width of the angle that is tributary to one bolt. This tributary width will be called
p and can be calculated as

p=

W Angle

EQ 1

n tb

where
W Angle = the width of the angle perpendicular to the beam axis
ntb
= the number of tension bolts

ta

a'
M

b
gt

b'

(a)

(b)
Figure 1: Clip Angle Geometry Notation

(c)

Other parameters that appear in the discussion of clip-angle components are:


T
B
Bn
Bo
Q
gt

=
=
=
=
=
=

a =
b =

the applied angle force per tension bolt


the force present in a tension bolt at any given time
the tensile capacity of a bolt
the initial pretension of a bolt.
the prying force per bolt.
the distance between the heal of the angle and the center line of
the tension bolts
the distance measured from the bolt centerline to the edge of the
upstanding angle leg
the distance measured from the bolt centerline to the face of the
angle leg that is bolted to the beam flange
the distance measured from the inside edge of the bolt to the edge
of the upstanding angle leg
the distance measured from the inside edge of the bolt to the face
of the angle leg that is bolted to the beam flange

Additional parameters that are specific to particular models will be introduced as needed. It is
crucial to understand that T is the applied load per tension bolt. Therefore, the total applied load
is equal to T ntb.
In both of the models considered in this work, a prying force is assumed to develop as the
flange deforms. This prying force is added to the conventional force present in the tension bolts
and effectively reduces the load that can be safely applied to the angle. The basic mechanism
is shown in Figure 1c and fundamental equilibrium shows that the bolt tension, B, is the sum of
the prying force, Q, and applied load, T. The prying forces can generally be minimized by
reducing the tension bolt gage, gt, or by increasing the flange thickness. The assumption that
the prying force acts at the tip of the flange is generally accepted and is considered accurate
until the length of the flange exterior to the bolt becomes large or until the flange thickness
becomes small (5). Figure 2 shows an angle specimen just prior to a tension bolt fracture.
AISC-LRFD
The prying model use in the AISC-LRFD specification for the ultimate strength determination of
clip-angle components is based on the model developed by Kulak et al (5). Although the model
was originally developed to predict prying forces in T-stub components, the model can be
applied to clip angles with minor modification. In this model, the bolt force is assumed to act at
the inside edge of the bolt shank as opposed to acting at the centerline of the bolt. This premise
is based on the assumption that as the flange deforms, more of the bolt force is transmitted to
the angle flange under the inside of the bolt head than under the outside. This is a result of the
stiffness of the bolt head and the degree of bending present in the flange and bolt. To
accommodate this idea in the model, equilibrium is based on the dimensions a and b instead of
a and b. The parameters a and b are defined in Equations 2 and 3. The magnitude of the
length a is limited to 1.25b in this model in recognition of the prying force, Q, may not act as a
concentrated force for wider flanges.

Figure 2: Angle Component During Testing Prior to Tension Bolt Fracture

db
2
d
b' = b b
2

a' = a +

EQ 2
EQ 3

The parameter (Figure 1b) is defined by Kulak et al. as the ratio of the moment at the bolt line
to the moment at the face of the angle leg that is bolted to the beam flange (hereafter referred to
as the angle stem), and is an indicator of the level of prying present. Physically, is limited to
values between 0 and 1. A value of 1 is achieved if the bolt is stiff enough to cause the flange
to act as a fixed-fixed beam and a value of 0 results when the flange separates completely from
the column. In calculating the prying capacity, however, is not limited. If 0 then the flange
is in single curvature, the prying forces are zero, and the bolts are subjected to conventional
tension only. If 1 then the flange is in double curvature and the prying forces are
maximized. When 0 1, a combination of flange yielding and bolt prying will occur (8). M is
the moment at the face of the stem and is the ratio of the net section of the flange at the bolt
line to the gross section at the face of the stem, excluding the fillet. can be written as

= 1

dh
p

EQ 4

Moment equilibrium of the flange between the face of the stem and the bolt line, using b, results
in Equation 5, moment equilibrium of the flange to the exterior of the bolt line, using a, results in
Equation 6, and force equilibrium of the entire flange results in Equation 7.

T b' = (1 + )M
Q a' = M
B = T +Q

EQ 5
EQ 6
EQ 7

Equation 5 can be solved for as shown in Equation 8.

1 T b'
=
1
M

EQ 8

At failure, M will be the plastic moment capacity of the flange and can be written as

p t 2a
Fy
Mp =
4

EQ 9

Substitution of Mp into Equation 8 yields as a function of the applied load, T, as shown in


Equation 10.

1 4T b'
=
1
2

p t a Fy

EQ 10

Manipulation of the equilibrium equations provides the prying force, Q, as

b'
Q = T

1 + a'

EQ 11

The capacity of an existing angle is then calculated based on the minimum value resulting from
Equations 12, 13, and 14 which correspond to a flange mechanism, mixed mode failure, and a
tension fracture, respectively. These three equations represent the possible failure modes of
the flange and tension bolts. As was previously mentioned, a flange mechanism will develop if
1, bolt prying combined with flange yielding will govern if 0 1, and conventional bolt
strength with no prying governs if 0. Note, however, that the capacity of an existing angle
can be computed without calculating . This is convenient because is a function of the
applied load per bolt, T, and the solution would otherwise be iterative.

T=

(1+ ) p F

2
y ta
4b'
2
B n a' p Fy t a
T=
+
a'+b' 4(a'+b')
T = Bn

EQ 12
EQ 13
EQ 14

A solution space for a typical angle is shown in Figure 3. The solution space is the result of
plotting an angles flange capacity as a function of the angle thickness. The bold line OABC
defines the capacity of the flange and tension bolts and the region below this line, OABCD,
represents an adequate design. Segment OA defines the flange mechanism strength and is
calculated using Equation 12 which assumes that = 1, segment AB is defined by the bolt
capacity including the effects of prying and is computed using Equation 13, and segment BC
represents the conventional strength of the bolts without prying and is computed as shown in
Equation 14.

Figure 3: General Solution for Angle Capacity

The line segment OB represents the case of = 0. The region OBCD represents a design with
negligible prying effects, as would be desired when considering fatigue. The angle thickness
associated with the point B is often referred to as the critical thickness, tc, because an angle with
a thickness greater than tc will have negligible prying and will develop the full tensile strength of
the bolts. The point A is generally considered to represent a balanced failure because the full
strength of the flange is exhausted at the same time that the bolt forces, including prying,
become critical (9). The critical thickness and balanced load are written as

4B n b'
p Fy

EQ 15

Bn
b'
1+

1 + a'

EQ 16

tc =

To =

Table 1 shows a comparison of the experimental angle capacities of the ten specimens from the
SAC investigation with the model predictions. The column labeled Experimental Capacity
contains the maximum loads recorded during the testing of each angle and the percent
difference was calculated using Equation 17. As a result, a positive percent difference
represents a conservative prediction of the angle capacity. All capacity predictions listed in the
table were computed using the actual material properties and are based on either the combined
flange mechanism and bolt capacity (0 1) determined by Equation 13 or the conventional
bolt strength without prying ( 0) determined by Equation 14. No attempt was made to identify
the load at which a flange mechanism formed ( 1).

% Difference =

Experimental Capacity - Predicted Capacity


Experimental Capacity

EQ 17

Table 1: Comparison of Experimental and Predicted Angle Capacities


Exp.
Test ID

AISC - LRFD

Capacity

Capacity

Kip

Kip

CA-01

108.3

131.1

CA-02

125.3

CA-04

84.3

CA-17

120.9

CA-18

119.2

CA-09

% Diff

Eurocode
Capacity

Proposed Model

% Diff

Kip

Capacity

% Diff

Kip

-21.0%

93.7

13.4%

112.4

133.1

-6.2%

93.6

-11.0%

132.8
133.4

125.2

167.7

CA-10

159.3

CA-12

109.2

CA-14
CA-16

-3.8%

106.0

15.4%

120.5

3.8%

73.2

13.2%

85.1

-0.9%

-9.8%

94.9

21.5%

113.8

5.9%

-11.9%

95.3

20.1%

114.4

4.1%

-33.9%

115.3

7.9%

143.7

-14.8%

171.2

-7.5%

133.9

15.9%

156.0

2.0%

120.5

-10.4%

91.5

16.3%

109.6

-0.3%

136.8

149.1

-9.0%

116.5

14.9%

135.1

1.2%

95.2

104.8

-10.1%

80.0

15.9%

95.3

-0.1%

Ave:

-13.1%

Ave:

15.5%

Ave:

-0.3%

Var:

0.7%

Var:

0.1%

Var:

0.3%

The capacities predicted by AISC-LRFD model were all higher than the observed experimental
capacities. The average percent difference was -13.1% with a coefficient of variation of 0.7%.
The over prediction of strength is likely related to the location of the plastic hinge near the angle
stem. The AISC-LRFD model assumes that a plastic hinge forms at the face of the stem. Yield
lines and deformation of the test specimens (Figure 2), however, indicated that the hinge formed
in the stem of the angle and not in the upstanding leg. By calculating the angle capacity under
the assumption that the hinge is located in the upstanding leg when it is actually in the stem, the
moment arm, b is underestimated which leads to a lower moment in flange and lower prying
forces in the tension bolts.

The experimental capacity of angle CA-09 is consistently lower than the calculated values for all models examined.
No explanation for this, besides random variability of material properties, is suggested.

Eurocode Model
Annex J of the Eurocode 3 (4) addresses the design of beam-to-column connections. The
model used by the Eurocode for clip-angle strength closely resembles the theory developed by
Kulak et al. The code recognizes the same three failure modes documented by Kulak, et al.,
shown as Equations 18, 19, and 20, which predict a flange mechanism, mixed mode failure, and
simple tension bolt fracture, respectively. The primary differences between the Eurocode and
the AISC-LRFD model are the way that the dimensions are defined and the fact that the
Eurocode makes no strength reduction for flange material lost to the drilling of the bolt holes.
The geometric definitions for the Eurocode model are illustrated in Figure 4. Like the dimension
a in the AISC-LRFD model, n is limited to a value no greater 1.25m. The length m is defined as
the distance from the centerline of the bolt to the face of the stem, minus 80% of the radius of
the fillet in the K-zone. No modification is made to the location of the concentrated bolt forces
as is done in AISC-LRFD model.

T=

p Fy t a2

EQ 18

2m
2
n B n p Fy t a
T=
+
m + n 4(m + n)
T = Bn

EQ 19
EQ 20

The most significant difference between the Eurocode model and the AISC-LRFD model is that
the Eurocode recognizes that a plastic hinge is likely to form in the angle stem instead of the
upstanding leg when the gap between the end of the beam and the column face, the beam setback, is sufficiently large. When the beam set-back is less than or equal to 40% of the angle
thickness, the length m is measured from the bolt centerline to the face of the stem minus 80%
of the fillet radius (Figure 4a). When the beam set-back if greater than 40% of the angle
thickness, however, m is measured from the bolt centerline to the center of the angle stem
(Figure 4b). This difference in the definition of m leads to an increased capacity prediction when
the beam set-back is less than 0.40ta and to reduced capacity predictions when the set-back is
greater than 0.40ta when compared to the AISC-LRFD model.
ta

ta

m
m

r
0.8r

ta / 2

(a)

(b)
Figure 4: Eurocode Geometry

Flange capacities predicted using the Eurocode model are compared to the experimental
capacities and AISC-LRFD predictions in Table 1. The average percent difference of the
predictions is 15.5% with a coefficient of variation of 0.1%. No resistance factors were used for
the Eurocode capacity predictions.

Proposed Model
A combination of the AISC-LRFD and Eurocode models was considered next. The values
shown in the columns labeled Proposed Model in Table 1 were obtained by using the
dimension definitions shown in Figure 5. The parameters a and b are identical to m and n of the
Eurocode. These dimensions were then used in the AISC-LRFD equations (Equations 1-14).
The resulting model provided better results than either of the other two models alone. An
average percent difference of -0.3% with a coefficient of variation of 0.3% was obtained.
ta

ta

a'

b'

a'

b'

0.8r
ta / 2

(a)

(b)
Figure 5: Proposed Model Geometry

DISCUSSION AND CONCLUSIONS


Both the AISC-LRFD model and the Eurocode model provided consistent results when their
predictions were compared to the experimental results. The coefficients of variation of the
models were 0.7% and 0.3%, respectively. The strength predictions obtained using the AISCLRFD model were unconservative by an average of approximately 13% while predictions
obtained using Eurocode were conservative by an average of approximately 15%. The
proposed model, which was developed by combining ideas from the first two models, yielded
results that were both consistent and accurate when compared with the experimental results.
It should be restated that the experimental data used for this evaluation was taken from tests of
angle components with thicknesses significantly greater than those used in typical angle
connections. Furthermore, all of the specimens used for comparison in this work had beam setbacks greater than 40% of the angle thickness. As a result, the geometries in Figures 4a and
5a were not used in any calculations. An evaluation using results from a broader, more typical
base of data would lead to stronger conclusions.

ACKNOWLEDGEMENTS
The experimental data used for this research was generated as part of a SAC research project
(SAC Task 7.03) that was conducted at the Georgia Institute of Technology under the
sponsorship of FEMA. Portions of the analytical work associated with this project were funded
by the Mid-America Earthquake (MAE) Center as part of the work under Task ST7.
REFERENCES
1
2
3
4
5
6
7

8
9

Swanson, J. A. 1999. Characterization of the Strength, Stiffness and Ductility Behavior of Tstub Connections, Ph.D. Dissertation, Georgia Institute of Technology, Atlanta, GA.
Swanson, J. A. and Leon, R. T., 2000. Bolted Steel Connections: Tests on T-stub
Components, J. Struc. Engrg., ASCE, Vol. 126, No. 1, pp. 50-56.
AISC, 1994. Load and Resistance Factor Design, 2nd Ed., American Institute of Steel
Construction, Chicago, IL
Eurocode 3, 1993. Design Procedures to C-EC3 - Concise Eurocode 3 for the Design of
Steel Buildings in the United Kingdom, The Steel Construction Institute, Ascot, UK.
Kulak, G. L., Fisher, J. W., and Struik, J. H. A., 1987. Guide to Design Criteria for Bolted and
Riveted Joints, 2nd Ed., John Wiley & Sons, New York.
Chen, W. F., Goto, Y., and Richard Liew, J. Y., 1996, Stability Design of Semi-Rigid Frames,
John Wiley & Sons, Inc.
Jaspart, J. P., and Maquoi, R., 1991, Plastic Capacity of End-Plate and Flange Cleated
Connections - Predictions and Design Rules Proceedings of the 2nd Intnl Workshop on
Connections in Steel Structures: Behavior, Strength and Design, Pittsburgh, USA, April
Thornton, W. A., 1985, Prying Action - A General Treatment, Engineering Journal,
American Institute of Steel Construction, Vol. 22, No. 2
Astaneh, A., 1985 Procedure for Design and Analysis of Hanger-type Connections,
Engineering Journal, American Institute of Steel Construction, Vol. 22, No. 2

S-ar putea să vă placă și