New Design Criteria for Gusset

Plates in Tension
STEVE G. HARD ASH and REIDAR BJORHOVDE

Gusset plates are common fastening elements used in

fabricated steel structures such as trusses or braced-frame
structures. In the latter case, their primary purpose is to
transfer either tensile or compressive loads from a bracing member to a beam and column joint.
Current gusset plate design is based primarily on elastic analyses for determining critical sections and stresses.
No known failures or adverse behavior have been noted,
but substantial differences in the factor of safety against
ultimate load exist because of the assumptions involved.^
Therefore, it is important to develop an improved design
method, with the goal of providing economy of design
and consistent levels of safety.
The ultimate-strength approach fulfills these requirements. To date, a few experimental studies have been
conducted to determine the behavior and ultimate strength
of gusset plate connections. However, additional tests
have been needed to develop a design method based on
ultimate strength behavior. It is the purpose of this paper
to describe a number of gusset plate tests that were performed, and to develop rational criteria that incorporate
all of the primary strength parameters.
PREVIOUS WORK
One of the first significant elastic experimental studies
of gusset plates was conducted by Whitmore on a Warren truss joint. '^ Analyses showed the maximum tensile
and compressive stresses were located in a region of the
plate at the ends of the tension and compression diagonals, respectively. It was determined the well-known effective width could be used to define the section subjected to maximum normal stress. Additional elastic

Steve G. Hardash is a Structural Engineer, Robin E. Parke & Associates, Plioenix, Arizona; and formerly graduate student, University of Arizona, Tucson, Arizona.
Reidar Bjorhovde is Professor of Civil Engineering and Engineering
Mectianics, University of Arizona, Tucson, Arizona.

SECOND QUARTER / 1985

analyses verified the locations of maximum normal stress

in the gusset plate are located in a region near the ends
of the connected members.^
Relatively few ultimate strength tests have been performed of gusset plate connections. The probable reason
is that behavior beyond the elastic range could not be
determined before the advent of finite element techniques. Chesson and Munse"^ conducted ultimate strength
tests of 16 large truss-type connections. Only one of these
failed at the gusset, which exhibited tearing across and
on the outer lines of bolts. Further work by the same
investigators^ provided ultimate strength data for 30 trusstype tensile connections. In this study, 10 connections
failed by tearing in the vicinity of the bolted or riveted
connection to the gusset plate.
More recently, tests were conducted at the University
of Alberta"* to determine the ultimate strength of gusset
plates used in diagonal bracing connections. The three
gusset plates tested to failure were full-sized; tearing was
observed across the last row of bolts and also in the
boundary connections, along with plate buckling.
Nonlinear finite element analyses of gusset plate behavior have been conducted in recent years.'"^'^^ The results confirm the maximum normal stresses are located
in a region near the end of the gusset plate connection.

THE BLOCK-SHEAR CONCEPT

In 1976, the Research Council on Riveted and Bolted
Structural Joints (now the Research Council on Structural Connections) increased the allowable bearing stress
in high-strength bolted joints from 1.35 F, to 1.5 F,, ,'^
where Fy and F^ are the yield and ultimate tensile stresses
of the material in the connected parts, respectively. This
was a radical alteration that prompted significant changes
in the connection design and detailing practices that had
been used previously. For example, for ASTM A36, the
most commonly used grade of steel for construction purposes, the increase in the bearing stress amounted to approximately 80%, from 48.6 ksi to 87.0 ksi (335 MPa
to 600 MPa).

77

Around the time when the changes in the bearing stress

criteria were implemented, research was conducted at the
University of Toronto on the shear force capacity of coped
beam webs. This work had been undertaken to verify the
shear design criteria of the new Canadian Hmit states design standard. ^^ The research involved a series of fullscale tests on uncoped and coped double-angle beam-tocolumn connections, loaded in as close to pure shear
conditions as possible.^ Beam action therefore did not
enter into the overall behavior of the connection.
It was found the coped web failed in a mode that involved a combination of a horizontal splitting of the beam
web at the lower bolt hole, along with an elongation of
all bolt holes in the direction parallel to the applied force.
This produced a shearing out of a block of the web, as
in Fig. 1, leading to the development of the concept of
a block-shear ultimate strength model. In the block-shear
model, the strength is developed by the shear resistance
of the web along line 1-1 (see Fig. 1), in addition to the
tensile resistance of the web along line 2-2.
As a result of the above studies, the AISC Specification^" incorporated design criteria for connections
aimed at covering the block-shear problem. The allowable strength is thus given by Eq. 1 as:
R,,^030AF,

+ 0.50 A,F,

(1)

where:
Ra = allowable resistance to block-shear, kips

Further testing of coped beam connections was conducted at the University of Texas at Austin. The findings
were used to develop modifications to the original blockshear model. ^''^
Based on the work with coped beam-to-column connections, the block-shear failure mode has been considered for application to gusset plates loaded in tension.
One such suggestion was advanced by the AISC Commentary;^^ subsequent evaluations of full-size gusset plate
tests^'^^ suggested a model in which the ultimate shear
resistance was developed along the last row of bolts, as
shown in Fig. 2. The agreement between tests and theory
was good; a maximum error of 7% was recorded.
Since this initial application of the block-shear concept to gusset plates only involved diagonal bracing gusset plates whose block-shear strength parameters were
limited, it would appear to be necessary to further modify the model to take into account all of them. Thus,
factors such as connection length, distance between outside bolt lines, plate thickness, bolt diameter, material
yield and ultimate strengths and the plate geometry should
be considered. This has been accomplished in the investigation presented in this paper.
EXPERIMENTAL STUDIES OF GUSSET PLATES
To develop an ultimate strength approach to the design
of gusset plates, a series of gusset plate tests was conducted at the University of Arizona.^^ In addition, the

A^ = net shear area along line 1-1, in.^

A^ = net tension area along line 2-2, in.^
F^ = specified minimum tensile strength, ksi.
This equation is based on a factor of safety against tension failure of 2.0 (hence, 0.5 F J , and against shear
failure of approximately 2.0. It is implied that the ultimate shear stress is related to the ultimate tensile stress
as 0.6 F.

FAILURE BY SHEARING
OUT OF SHADED PORTION

l-l = FAILURE

BY ULTIMATE

SHEAR

2-2 = FAILURE

BY ULTIMATE

TENSION

AT
Fig. 1.

78

Block-shear model of failure for coped-beam web

Fig. 2.

Suggested block-shear model for gusset plates

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

results of the work conducted at the University of lUinois"^'^

and the University of Alberta^ have been incorporated
into the strength evaluations. Details of each of the testing programs are given in the following.
University of Arizona
To evaluate the application of the block-shear concept
to gusset plate behavior and design, it was decided to
test plates loaded in tension through two lines of bolts. ^^
The intent was to isolate the tested joint so as to observe
its behavior. Figure 3 shows the general configuration
of the gusset plate. The region of interest is delineated
by the measurements / and S.
As indicated in the figure, the following were considered as the strength parameters for the plates: (1) gage
between lines of bolts, S; (2) edge distance, e; (3) bolt
pitch, 5; and (4) number of bolts. The total connection
length / depends on the edge distance and the total number of bolts in a line.
A total of 28 gusset plates were tested, using 5-values
of 2, 3 and 4 in. (51, 76 and 101 mm), 1 and IV2 in.
(25 and 38 mm) for e, and 1V2 and 2 in. (38 and 51 mm)
for s. The bolts used for all the tests were V2 in. (13
mm) diameter ASTM A325, and the number of bolts in
a line ranged from 2 to 5.
Material properties of the plate steel were determined
by tension coupon tests, performed in accordance with
the procedures of ASTM 370.^ Due to a fabrication mix-

up, one plate was fabricated from a different steel grade.

The average material properties were as follows: For tests
plates Nos. 1 to 17 and 19 to 28, Fy = 33.2 ksi (229
MPa) and F = 46.9 ksi (323 MPa);
For test plate No. 18, Fy = 49.5 ksi (341 MPa)
(0.2% offset) and F , = 64.5 ksi (444 MPa).
Slight fabrication errors required that test connection
holes be redrilled to a diameter of ^Vie in. (17.5 mm)
[^/i6 in. (5 mm) oversize] for three of the test specimens
(Nos. 16, 20 and 26). However, neither of these changes
had any effect on the strength model that was developed.
The gusset plate tests were performed using a 200-kip
capacity Tinius-Olsen universal testing machine. Loaddeformation data were obtained by positioning a dial gage
between the testing machine crossheads, with deformation readings recorded at convenient load increments. The
final test set-up is shown schematically in Fig. 4, and a
photograph of an installed and instrumented plate in
Fig. 5. Details of the failure loads, connection geometries and material properties for the tested gusset plates
are given in Table 1.
All of the gusset plates behaved similarly, reflecting
the following characteristics: (1) slip took place during

o
o
o
o

o
o
o
o
I II

' I'

<r^

<*

oj o o o o o o

I
I
I

I I
I I
I

>

'J'

ip4=4

OI O O O Q O O

J:
o
o
o
p

o
o
o
^o

t*4

o o

-(^

- ^ - ^ ^

(|)---(|)--(|)--(|)-1.75 In

15 in

->

u
o
o
o
o

2.25 in
(TYP)

Fig. 3.

General gusset test plate configuration

SECOND QUARTER / 1985

Fig. 4.

'II I I
<^<>

II I
II
II

>

o
o
o
o
Front and side view of typical test setup

79

Table 1. Gusset Plate Test Data (University of Arizona Study^^)

Test
No.

(kips)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

54.6
55.2
67.6
73.6
71.5
81.1
76.2
83.4
80.6
89.9
84.2
91.6
79.5
95.0
85.2
99.8
88.1
154.5
92.9
119.7
105.0
114.9
109.6
118.0
105.1
131.2
112.0
125.7

Pw/r

Fig. 5.

80

Hole Dia.
(in.)

S
(in.)

e
(in.)

s
(in.)

/
(in.)

t
(in 0

^>

Fu

(ksi)

(ksi)

9/16

2.00

1.10
1.50
1.00
1.00
1.50
1.50
1.00
1.00
1.50
1.50
1.00
1.60
1.00
1.00
1.50
1.50
1.00
1.00
1.50
1.50
1.00
1.00
1.50
1.65
1.00
1.00
1.50
1.50

1.50

2.60
3.00
4.00
5.00
4.50
5.50
4.00
5.00
4.50
5.50
4.00
4.60
5.50
7.00
6.00
7.50
5.50
7.00
6.00
7.50
5.50
7.00
6.00
7.65
7.00
9.00
7.50

0.237

33,2

46.9

^
0.253
0.237

i
49.5
33.2

64.5
46.9

9.50

3.00

4.00
4.00
2.00

>f
11/16
9/16

\f

^/

3.00

11/16
9/16

4.00

>

3.00

11/16
9/16
9/16

>

>f

f_

Test setup with plate No. 28 installed

2.00
1.50
2.00
1.50
2.00
1.50
2.00
1.50

2.00
1.50
2.00
1.50
2.00
1.50
2.00
1.50
2.00
1.50
2.00
1.50
2.00
1.50
2.00

the elastic loading phase; (2) the plates exhibited a long

yield plateau to ultimate load; and (3) the load subsequently dropped to a second strength plateau. As an example, a typical load-deformation curve is that for test
plate No. 28, shown in Fig. 6.
Out of the 28 gusset plates tested, 12 exhibited some
slipping during the elastic loading portion of the loaddeformation curve. As the loading progressed, it was
found the occurrence of the first yield lines accurately
signaled the beginning of the yield plateau on the loaddeformation curve. After reaching the ultimate strength,
recording of the load-deformation data was continued until
the load was stabilized at a second (lower) strength plateau. This plateau would be reached when the tensile
failure across the last row of bolts was complete.
The plate failure modes only depended on the types
of steel in the plate, as shown in Fig. 7. For the test
plates fabricated of mild structural steel (test specimen
Nos. 1 to 17 and 19 to 28), the basic failure mode consisted of a tensile failure across the last row of bolts,
along with an elongation of the bolt holes, as shown in
Figs. 7a and 8b. None of the test specimens showed significant tearing across the lines of bolts in the direction
of the applied load. Oversizing the connection holes (test
plate Nos. 16, 20 and 26) did not influence the failure
mode for this type of steel. For test specimen No. 18,

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

140 -1

0.4

TOTAL DEFORMATION
Fig. 6.

TENSILE

0.8

(INCHES)

Load-deformation curve for test plate No. 28

-SHEAR TEARING

SHEAR YIELD ZONES-

APPROX

0.6

> c ^ ^ ' APPROX.

TEAR-^

AT
(a) Tensile failure with shear yielding
(mild, hot-rolled structural steel)
Fig. 7.

SECOND QUARTER / 1985

(b) Tensile failure with shear tearing

(cold-rolled structural steel)

Two basic failure modes for gusset test plates

81

Fig. 9.

Test plate No. 18 at end of loading cycle

fabricated from cold-rolled structural steel, the failure

mode also included the tearing across the last row of
bolts. However, along with the elongation of the bolt
holes, the specimen also exhibited some tearing along
the bolt lines as indicated by Figs. 7b and 9. It was observed during the testing that the shear failure along the
bolt lines occurred after the ultimate load was reached.

(a)

University of Illinois
The 1958 study"^ resulted in only one connection failure
at the gusset plates, and the 1963 study^ gave 10 additional sets of data. These tests incorporated riveted and
bolted joints in either single- or double-plane connection
configurations. Two gusset plate connections consisted
of 5 lines of bolts with 7 bolts per line; this gave a total
distance between outside bolt lines of 12 in. (305 mm)
and a total connection length of 17 in. (432 mm). The
remaining test connections had 2 lines of bolts, with either
7 or 10 bolts per line; the overall connection length was
either 19V4 (489 mm) or 24V2 in. (622 mm), with the
shorter connections having a gage between bolt lines of
4 in. (101 mm). The holes were either punched or drilled,
and the influence of these parameters will be examined
in the following. Failure modes consisted of tearing across
the last row of bolts with some tearing along the bolt
lines.
(b)
Fig. 8. Test plate No. 28 at end of loading cycle
(a) Load =110 kips (point A on curve of Fig. 6)
(b) Specimen removed from the testing machine

82

University of Alberta
The University of Alberta study^ yielded three failures
of diagonal bracing connection gusset plates. All three

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

Table 2. Gusset Plate Test Data for University of Illinois'*'^ and University of Alberta^ Studies
New
Test No.

Previous
Test No.

(kips)

29
30
31
32
33
34
35
36
37
38
39
40
41
42

ADl
A-l-DB^
SA-l-PR^
SA-2-PR'
SA-l-PB^
SA-2-PB'^
SA-2-DB^
30 Gusset^
45 Gusset''
60 Gusset'
SE-l-DR^
SE-2-DB^
SE-l-PR^
SE-2-PR^

617.5
640.0
483.8
476.5
481.4
482.0
504.1
142.7
148.1
158.4
772.0
778.0
576.0
582.0

Hole Dia.
(in.)

S
(in.)

13/16

12.0
12.0
4.0

(in.)

(in.)

(kips)

(kips)

17.0

0.50

34.2
36.2
34.9

60.0
59.0
61.1

Y
55.7

Fu

5.0

19.25

Y
0.125

Y
42.4

5.25

24.5

0.50

35.9

Tor last two letters: D = drilled holes, P = punched holes, R = rivets used in the connection, and B = bolts used in the connection.
'Angle measured from the beam axis.

plates had the same geometry for the connection region,

which consisted of 2 lines of bolts with 9 bolts per line;
the gage between bolt lines was 5 in. (127 mm), with a
total connection length of 19V4 in. (489 mm). However,
each plate reflected a different bracing member angle,
using inclinations of 30, 45 and 60 with respect to the
beam axis. All three gusset plates failed by tearing of
the plate at the last row of bolts in the tension connection, in a direction perpendicular to the applied tensile
loads.
Details of the failure loads, connection geometries and
material properties for the University of Illinois and the
University of Alberta studies are presented in Table 2.
It should be noted the test specimens have been renumbered here, in order of increasing connection length.

area between the two outside bolts in the last row. From
the test results for the 28 plates in the present study, it
was found the drop in strength from the ultimate load to
the second strength plateau corresponded approximately
to the ultimate tensile strength of the net area at the last
row of bolts. That is, as the plate tore, the load was
reduced by the magnitude FJ_t(S - d)]. This is shown
by the schematic load-deformation curve in Fig. 10.
Ultimate shear resistance is more difficult to define,
since the shear behavior varied among the 42 test specimens. For instance, the 28 plates of the present study
did not display significant tearing along the bolt lines.
Only the test plate made from the cold-rolled steel (No.
18) was observed to tear along the bolt lines, but this
occurred after the ultimate strength was reached. This
suggests the shear stress distribution is not uniform, as

DEVELOPMENT OF ULTIMATE
STRENGTH CRITERIA
Block-Shear Modeling
The relationship between the ultimate load and the observed failure mode must be considered to develop a
strength model that accurately reflects the true behavior
at ultimate strength. For a tensile gusset plate connection, it appears the strength model must incorporate two
terms: one reflecting the tensile resistance developed at
the last row of bolts, and one reflecting the shear resistance developed along the outside bolt lines.
For all 42 gusset plate test specimens, a tensile tear
across the last row of bolts was observed, regardless of
the strength parameters, hole size or plate material. This
suggests that to model the connection behavior accurately, the ultimate strength model must incorporate the
ultimate tensile stress of the plate material F^ over the

SECOND QUARTER / 1985

TENSILE RESISTANCE
1 (APPROX. F^J [ t ( S - d ) : )

SHEAR RESISTANCE

DEFORMATION

Fig. 10.

General load-deformation
for gusset plates

diagram

83

THEORY

THEORY

"THEORY

n-HEORY

s
1^

^1
I

>

li

1
1
ii 1f

(TYP)

\\

I
t

(a) Tensile gross

area; shear
gross area

Y T

(b) Tensile net

area; shear
gross area
Fig. 11.

t t

(c) Tensile gross

area; shear net
area

(d)

l i t

i_i

Tensile net
area; shear net
area

The four basic free-body sections of gusset plate

has been indicated in early examinations, but rather depends on the particular connection geometry and material.
The contribution of each of these terms (tensile resistance and shear resistance) in the ultimate strength model
is shown in Fig. 10.
Detailed Block-Shear Model Development
Based on the discussion of the previous section, four basic free body diagrams can be constructed for the connection region; these diagrams are shown in Fig. 11. The
basic differences between the four is the method of considering the areas (either gross or net) over which the
tensile stress and the shear stress act. The ultimate tensile strength F^ is the assumed level of stress on the tensile area at ultimate strength. The shear stress T is of
unknown magnitude, but is assumed to be distributed
uniformly along the shear area.
The shear-yield stress of steel has been determined to
lie within the range of one-half to five-eighths of the
tensile-yield stress. Using the von Mises yield criterion
for plane stress gives the shear-yield stress as T^ ==
Fy / V s . This relationship is based on a mechanistic failure model for a ductile material such as steel, and will
be used in this study. Therefore, the shear stress magnitude used here is T = F / V 3 , where F represents an
unknown tensile stress.
From the diagrams in Fig. 11, the following equations

84

describe the ultimate capacity Ptheory of the four connection models:

1. Gross-gross: P,heory = FuSt + 2{F/\3)lt

(2a)

2. Net-gross:

P,,,^ = F5f + 2(F/V3)/r

(2b)

3. Gross-net:

P^^ory = F,St + 2(F/V3)/r

(2c)

4. Net-net:

P,,,,^ = FJJ + 2(F/V3)/f

(2d)

The value for the shear stress F / V s has purposely been

left in general terms to allow for variations of this undefined stress term.
To compare the actual failure loads to those obtained
by the models of Fig. 11 and Eq. 2, it is convenient to
use the non-dimensional term P, which is known as the
professional factor in Load and Resistance Factor Design
(LRFD) terminology.^'^^ The professional factor is an indication of the accuracy of the model and is given by
the expression:
Test ultimate strength
P=
Theoretical ultimate strength
A value of F = 1.0 would indicate perfect agreement
between the strength model and the observed strength.
The results of the test versus theory comparisons for
the different connection models are shown in Figs. 12a
through d, corresponding to the results using Eqs. 2a
through d. In each figure, two extremes of shear stress
are used: T = T and T = T .

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

1.2

-,

I.I

A A A ^

A A

1.0

cc
o
I-

<

LL

0.9

A
D

-I

<
g

QnQn

0.8

nD

0.7

CL

LEGEND

0.6 -\

0.5

1^
20

10

GUSSET

Fig. 12a.

1.4

PLATE

1^
40

30

TEST

NUMBER

FROM

TABLE

50

7.2

Professional factor values, based on tensile

gross and shear gross areas

-I

A
A

1.2

A ^ A A

^
^

A^

^^^^

o
<
<
Q Q

CO

</)

LU
U_

O
^

LEGEND
0.6

At=T^

Q Q

QT=t,

0.4

~T~
10

20

GUSSET PLATE

Fig. 12b.

SECOND QUARTER / 1985

TEST

40

30

NUMBER

FROM

TABLE

50

7.2

Professional factor values, based on tensile net

and shear gross areas

85

1.6

_,

1.4
A

^^A4

cr
o
O

<

A
1.2

<
O

(/)

1.0

CO
LU
LiO

cr
CL

0.8 H

LEGEND

A t=T^

0.6

"1~
10

30

20
GUSSET

PLATE

TEST

40

NUMBER

FROM

TABLE

50
7.2

Fig. 12c. Professional factor values, based on tensile

gross and shear net areas

L8

-,

.6

H
A

A
A

(T
C)

14

A
D

<
LL

1.2

<
2

D O

AA

1-

_l

^ .

CO

(/)
LU

LO

LJ_

O
Q:

LEGEND

QL

0.8-4

A l = t^

0.6
10

GUSSET

1^
20

1^
30

PLATE TEST NUMBER FROM

40

"1
50

TABLE 7.2

Fig. 12d. Professional factor values, based on tensile net

and shear net areas

86

ENGINEERING J O U R N A L / A M E R I C A N INSTITUTE OF STEEL CONSTRUCTION

Figure 12a shows that using the ultimate shear stress

value along the bolt lines gives a much larger theoretical
failure load (small professional factor), while using the
shear yield stress gives a smaller theoretical failure load
for the majority of the tests. In Fig. 12b, using the ultimate shear stress appears to work well for the shorter
connections (smaller gusset plate test number), while the
shear yield stress gives good results for the longer connections. Both Figs. 12c and 12d show that using the
net area for the shear effect underestimates the failure
load by a large margin for a majority of the test specimens.
The conclusion is that the block-shear model using the
net tensile area and the gross shear area, as in Fig. l i b ,
is the most acceptable. It can also be concluded that, as
the gusset plate connection length increases (corresponding to an increasing test specimen number, as they have
been arranged), the professional factor decreases. This
indicates the effect of varying the connection length is
important, and it must be incorporated in a rational and
complete gusset plate model.
Figure 12b shows that for short connections, the ultimate shear stress acting on the gross connection length
1.6

area would be appropriate, while for longer connections,

the tendency is to approach the shear yield stress. This
shows the assumed uniformly distributed shear stress must
be expressed as a function of the connection length. This
can be accomplished by considering an interpolation between the yield and ultimate shear stress, expressed in
terms of the tensile stress (T^^ = F^^/V3), as the following:
eff-

(1 - C,)Fy + Q F ,

(3)

where:
F^ff = Effective tensile stress
Ci = Connection length factor
The variable C/ is the linear interpolation factor; if C/
equals zero, then F^^ equals the tensile yield stress, and
if C/ equals one, then F^^equals the tensile ultimate stress.
Using the net tensile area-gross shear area block-shear
model, it is possible to determine the required value for
C/ to give exact agreement with the observed ultimate
strength for each test. Figure 13 illustrates this result as
a function of the connection length /.

-I

LEGEND

a:
o

ARIZONA

UNIVERSITY

OF

UNIVERSITY

OF ILLINOIS

UNIVERSITY

OF ALBERTA

C = 0.9383-0.04l63i

X
IUJ

^
O

0.2-^

LU
O
O

CONNECTION
Fig. 13.

SECOND

QUARTER

1985

LENGTH, I

(INCHES)

Relationship between connection length factor,

Ci, and connection length (42 tests)

87

1.4 -n

1.2 -J

cr
o
i-

1.0

^ ^si!: \

0.8 H

o
C/)

(/)

LEGEND

Ui
LL

0.6 H

Q.

0.4

UNIVERSITY OF ARIZONA

UNIVERSITY OF ILLINOIS

UNIVERSITY OF ALBERTA

10
CONNECTION LENGTH, i

Fig. 14.

15

20

I
25

(INCHES)

Values of the professional factor vs. connection

length, using 42 data points

Many possible curves could be fit through the points,

but a least squares linear regression has been used for
this study. The equation of this line in Fig. 13 is:
Ci = 0.9383 - 0.04163 x /

(4)

where:
C/ = Connection length factor to be used in Eq. 3
/ = Total connection length, in. (Note that Eq. 4
must be rephrased if metric units of length are
used.)
It is interesting to note that for very short connections,
a value of F^^ approaching F^ is obtained, and for connections longer than 22.5 in. (572 mm), the value for
Fgjis less than Fy. This result appears intuitively correct,
since longer connections would tend not to slip into bearing at mid-length of the connection.
Using Eq. 4 to determine the connection length factor,
the effective uniform shear stress, expressed in terms of
the effective tensile stress, can be obtained from Eq. 3.
Using this effective stress in Eq. 2b, the theoretical ultimate strength can be obtained. Figure 14 shows the re-

sulting professional factor vs. connection length; the results of this figure can be compared to the data in Fig.
12b. For the 28 gusset plates tested during the present
study, the mean value for the professional factor P is
1.000, with a coefficient of variation Vp of 0.0439. For
all 42 tests, P,, is 1.003 with a coefficient of variation
of 0.0716.
Refinement of Strength Model
Figure 14 shows three of the 42 test plates exhibited much
larger observed strengths than would be expected. Test
plate No. 1 had almost the same observed strength as
plate No. 2, although the only difference between the
two was that plate No. 1 had an edge distance 0.40 in.
(10 mm) less than No. 2. It is believed that since plate
No. 1 was the first to be tested, some testing error might
have evolved. Test plates Nos. 39 to 42 were plates with
similar geometry and material, but the fabrication of plates
Nos. 39 and 40 involved drilling of the bolt holes, while
the holes were punched for Nos. 41 and 42. The 34%
strength increase for the drilled specimens cannot be attributed completely to the hole preparation, as other

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

1.6

-I
LEGEND

POINT DISCARDED

1.4

UNIVERSITY OF ARIZONA
O
Q:

o
<

1.2 -i

-H UNIVERSITY

UNIVERSITY OF ALBERTA

1.0 - |

0.8 J

X
h-

OF ILLINOIS

C. = 0 . 9 3 8 3 - 0 . 0 4 l 6 3 i
(42 DATA POINTS)

0.6 H

LLI

0.4

o
2:

J
2 POINTS
i
DISCARDED-^

0.2 J
0.0 J

0^ = 0 . 9 4 6 7 - 0 . 0 4 6 5 B X
(39 DATA POINTS)

-0.2 H

-r

-0.4

10

CONNECTION
Fig. 15.

(5)

Using this expression for C/ to determine the effective

shear stress, Eq. 2b can be used to obtain the theoretical
strength.
The professional factor for each of the remaining 39
tests is plotted in Fig. 16. The mean value of P for the
remaining 27 of the University of Arizona test results is
1.000, with a coefficient of variation of 0.0338. For all
39 tests, P, is 1.001, with a coefficient of variation of
0.0322. This can be compared to the values of P^ =
1.003 and K = 0.0716 for all 42 tests.

SECOND QUARTER / 1985

LENGTH,i

I
25

(INCHES)

Linear regression lines for the connection length

factor, using 39 and 42 data points

identical plates with punched or drilled holes only showed

a slight increase in strength (5%) for the drilled specimens. Based on these evaluations, it is justifiable to discard the results from test plate Nos. 1, 39 and 40, since
they introduce test parameters not quantifiable and comparable to the other plates.
Figure 15 shows the regression line that has been developed on the basis of the remaining 39 gusset plate
test results and the expression for the connection length
factor is given as:
Ci = 0.9467 - 0.04658/

I
20

15

The effects of the other strength parameters, both those

directly used in the equations and those not included,
were analyzed to determine if any additional relationships exist.'' It has been concluded the effect of each
strength parameter contained in the proposed block-shear
equations is properly considered, and no others need be
included.
To simplify the proposed equations, all constants should
be rounded off to two decimal places. This results in the
following set of equations, which gives the nominal ultimate resistance of a gusset plate loaded in tension:

R, = F,S,,t+ \.\5F,fflt
F,ff={l-C,)F^

+ C,F,

C, = 0.95 - 0.047 /

(6)
(7)
(8)

where:
Rn = Nominal ultimate resistance of connection
Snet ^ Net gage between outside bolts
^ S - (no. of holes - 1) x hole diameter
t = Plate thickness

89

1.4

-I

1.2 H

a:
o
ho

1.0

0.8 H

^M^r

CO
CO
LU
U-

LEGEND

o
oc

0.6 H

UNIVERSITY OF ARIZONA

4- UNIVERSITY OF ILLINOIS

0.4

1^
5

UNIVERSITY OF ALBERTA

10

CONNECTION

15

LENGTH, i

20

25

(INCHES)

Fig. 16. Values of the professional factor vs. the

connection length, using 39 data points

This simplification of the original equation changes

the professional factor data only slightly, to P^ = 1.00
and Vp = 0.033. It is therefore seen that performance
of the model is excellent also in statistical terms, giving
a low coefficient of variation, and a mean of 1.0 for
results that reflect a wide range of strength parameters.
LOAD AND RESISTANCE FACTOR DESIGN OF
GUSSET PLATES
Development of the Resistance Factor, c|)
The motivation behind developing an ultimate strength
model for gusset plates is its eventual incorporation into
a limit states design procedure. The development of the
resistance factor will be presented in the following. For
a more detailed treatment of the subject of LRFD, Refs.
8 to 10 give further documentation.
The mean strength R^ and the coefficient of variation
Vi^ are given by the expressions:^'^
^m

^n^m^^^m^

V, = \/vl -^vl + vl
90

The coefficient F, is the mean value of the professional

factor, and the statistics of this term have been determined previously as P^ = 1.00 and Vp = 0.033. The
coefficient M^ represents the mean value of the ratio of
the actual static yield stress to the specified minimum
yield stress. The data for the statistics of this coefficient
have been determined as M^ = 1.10 and V^ 0.11.^
The coefficient F^ represents the mean value of the fabrication factor, and its statistics have been determined as
F^ - 1.00 and Vj. = 0.05.^ The fabrication factor is
representative of the geometric accuracy of the component in question.
Incorporating the above data for M, F and P and their
coefficients of variation gives the following values for
R^ and Vy^, using Eqs. 9 and 10, respectively:
R^=

lAOR^

y^ - 0 . 1 2 5
The resistance factor, ^, is given by the expression^''^

(9)
(10)

(t) = ^ e x p ( - 0 . 5 5 p V ^ )

(11)

Rn

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

10

15

20

CONNECTION LENGTH,!

Fig. 17.

25

(INCHES)

Gusset plate design curves for ASTM A36 steel

plate material

All terms in Eq. 11 have been determined in the preceding, except p, which represents the reliability index;
an increasing value of p represents a decreasing probability of failure. It has been considered good practice to
make connections stronger than the parts being joined,
to give ample warning of impending failure. On this basis, connections in general have been assigned a value
of P = 4.5, while the members they connect (beams,
columns, etc.) have been assigned a value of p = 3.0.^
Both of these values of p will be used to calculate ^ to
obtain a range of values for this factor. From Eq. 11,
the values for cj) are:
Forp = 4.5: (t) = 0.81
Forp - 3.0: (j) - 0 . 8 9
A suitable value would depend on the level of reliability
that is needed. For example, cj) = 0.85 appears to be a
reasonable choice.

Gusset Plate Design Curves

The proposed strength model lends itself well to the preliminary design of gusset plates. For instance, Eqs. 6, 7

SECOND QUARTER / 1985

and 8 can be combined into one expression, relating the

plate thickness to the variables /?, S^t^ I, Fy, and F^.
For a given type of steel, Fy and F are known, so the
expression relates the gusset plate thickness directly to
nominal strength R^ and the connection size S^t and /.
The above concepts have been applied to gusset plates
of ASTM A36 steel [Fy = 36 ksi (248 MPa) and F , =
58 ksi (400 MPa)], and the resulting curves are shown
in Fig. 17 in which R^/t has been plotted as a function
of / and Snet' The result is a family of parabolas, since
the strength model is a function of /^. Interpolation between parabolas (fixed values for S^^^) is linear, since the
strength model is a linear function of Snet- These design
curves can be constructed for any type of steel, but A36
is probably the steel grade most commonly used for detail material.
Figure 17 shows that as the length of a connection
increases, the rate of increasing gusset plate capacity decreases. This suggests that for longer connections it is
more efficient to increase the outside bolt gage (5^^).
Figure 17 can be used in either of two ways: (1) if ^ has
previously been chosen, then the connection size (/ and
Snet) can be determined; or (2) if the connection size has
been determined, then the necessary gusset plate thickness can be obtained. Either way, many combinations of
gusset plate thickness and connection size can be checked
quickly to determine the best design for the gusset plate
connection.
Design Example. For illustration, consider the design of the gusset plate connection shown in Fig. 18. It
has been found that two angles 8x6x1/2 with 12 A325
bolts in two lines (six bolts per line) will be adequate,
using the proposed LRFD Specification.^^ To aid in the
gusset plate design, it has also been determined that the
minimum connection length for spacing requirements is
14.5 in. (368 mm), and the possible range of S^^^ is 1.4375
in. (36.5 mm) < S^et ^ 4.6875 in. (119 mm), due to
spacing requirements.^^ Many combinations are possible, but the maximum value for S^et will be used. With
/ = 14.5 in. (368 mm) and S^et = 4.6875 in. (119 mm).
Fig. 17 gives a value of RJt = 970 kips/in (170 kN/mm).
Therefore:
S 7 . 6 . - 490

<^Rn ^^^^. .
= 970 kips-m.
t
t
A value of c|) = 0.85 as determined previously is used
to obtain the necessary plate thickness:
<

t>

490

490

(1)970

(0.85)970

0.594 in. (15.1 mm)

Therefore, a gusset plate thickness of ^/g in. (0.625 in.,

16 mm) is adequate, and a total connection length of
14.5 in. (368 mm) and an outside bolt gage of 5.5 in.
(140 mm) could be used.

91

GUSSET PLATE
A36 STEEL -

3/4

A325

BOLTS
2-L8x6xl/2
A572 GRADE 50

~M

! O O O O O O
>

<>

o o o o o o

I.}f.Q . = 4 9 0 KIPS
1 ni

Fig. 18.

Details of gusset plate connection

for design example

If a thinner gusset plate is needed, say, ^/le in. (14

mm), the required value of RJt is:

SxQ.

490

^t

(0.85)(9/16)
1,025 kips/in. (180 kN/mm)

R.

With this value for R,/t and S^,, = 4.6875 in. (119 mm).
Fig. 17 gives the required total length as / = 16.5 in.
(419 mm).
The design with the ^/le-in. thick gusset plate will now
be checked. From Eq. 8,
C/ = 0.95 -0.047(16.5) = 0.175
The effective stress becomes, from Eq. 7,
Feff= (1 - 0.175)(36) + (0.175)(58)
= 39.9 ksi (275 MPa)
The nominal strength is given by Eq. 6 as:
R, = (58)[5.5 - (13/16)1(9/16)
+ (1.5)(39.9)(16.5)(9/16)
= 578.8 kips (2576 kN)
By the LRFD criterion, S^.Gm ^ K490 kips < (0.85)(578.8 kips) = 492 kips
Therefore, the gusset plate thickness and connection size
are adequate.

92

The above computations demonstrate the ease with

which gusset plate connections can be sized using the
proposed procedure. Naturally, for the complete connection design, the limit state of yielding on the gross
cross-section just below the last row of bolts and the limit
state of tensile failure on the net section at the last row
of bolts must also be checked.

SUMMARY AND CONCLUSIONS

This paper has presented the following observations and
conclusions for tensile gusset plate connections:
1. All ultimate failure modes consist of a tensile tear
across the last row of bolts, with various stages of
shear yielding along the outside lines of bolts. The
extent of the latter depends on the connection length.
2. The governing block-shear model is shown to be the
one incorporating tensile ultimate stress on the net
area between the last row of bolts, and a uniform
effective shear stress acting on the gross area along
the outside bolt lines.
3. The effective shear stress acting along the outside bolt
lines can be assumed to be uniformly distributed, with
the magnitude given as a function of the total connection length, and material yield and ultimate stress
levels.
From the above, three equations have been developed
which accurately predict the ultimate strength of 39 gusset plate tests. The accuracy of the proposed model is
not affected by such factors as plate boundaries, fastener

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION

size or edge distance to the first bolt line.

Typical design curves for A36 steel are presented to
demonstrate the ease with which tensile gusset plate connections can be sized and gusset plate thicknesses selected to give maximum connection efficiency.
The 39 gusset plate tests considered in the development of the final strength model have included a wide
range of strength parameters. However, it may prove
worthwhile to conduct additional tests, especially with
connection lengths within the range of 10 to 16 in. (254
to 406 mm).
This study did not address the problems of compressive gusset plate connections, nor the related problem of
gusset plate buckling, which are important future considerations. Also, the combined effect of multiple members framing into one gusset plate, and gusseted connections in close proximity to boundary elements are
important areas to investigate further.

VF
Vp

Vn

7
T

Tf#

1.
2.

3.
ACKNOWLEDGMENTS
The work presented in this paper formed part of a research project funded by the American Institute of Steel
Construction, Inc. Special thanks are due the Institute
for this support, as well as the individuals who assisted
in the research. Messrs. M. F. Leong and A. E. Moussa
provided invaluable work in the laboratory, and the shop
work of Messrs. W. D. Lichtenwalter and L. Lujan was
essential to the success of the work. Mrs. Carole Goodman did an excellent job in typing of the manuscript.

NOMENCLATURE
A,
A,

c,
e

Feff
Fn.
Pu
Fy

I
^net

M
P
P.
P
* theory

Ra
Rm
Rn
S

S
"^net

= net tension area

= net shear area
= connection length factor
== edge distance
= effective tensile stress
= mean value of the fabrication factor
= ultimate tensile strength
tensile yield stress
= total connection length
= net connection length
= mean value of the material factor
= professional factor
= mean value of the professional factor
= theoretical block-shear capacity
= generalized load effect
= nominal resistance to block-shear
= mean resistance
= nominal resistance
= bolt pitch
= gage between outside bolt lines
= net distance between outside bolt lines
= gusset plate thickness

SECOND QUARTER / 1985

4.
5.

6.

7.
8.

9.

10.
11.

12.
13.

= coefficient of variation of fabrication uncertainties

= coefficient of variation of theoretical assumptions
= coefficient of variation of resistance
= reliability index
= load factor
= shear stress
= effective shear stress
== resistance factor
REFERENCES
Annual Book of ASTM Standards ASTM, Section 1,
Vol. 01.04, Easton, Md., 1984.
Birkemoe, P. C. andM. I. Gilmor Behavior of Bearing Critical Double-Angle Beam Connections
AISC Engineering Journal, 4th Qt., 1978, pp. 109115.
Bjorhovde, Reidar and S. K. Chakrabarti Tests of
Gusset Plate Connections Research Report, Dept. of
Civil Engineering, Univ. of Arizona-Tucson, April 1983.
Chesson, E., Jr. and W. H. Munse Behavior of Riveted Truss-Type Connections ASCE Transactions, Vol.
123, 1958, pp. 1,087-1,128.
Chesson, E., Jr. and W. H. Munse Riveted and Bolted
Joints: Truss-Type Tensile Connections ASCE
Journal of the Structural Division, Vol. 89, No. STl,
February 1963, pp. 67-106.
Fisher, J. W., T. V. Galambos, G. L. Kulak and M.
K. Ravindra Load and Resistance Factor Design Criteria for Connectors ASCE Journal of the Structural
Division, Vol. 104, No. ST9, September 1978, pp.
1,427-1,441.
Fisher, J. W. and J. H. A. Struik Guide to Design
Criteria for Bolted and Riveted Joints John Wiley and
Sons, Inc., New York, N.Y., 1974.
Galambos, T. V. and M. K. Ravindra Load and Resistance Factor Design for Steel ASCE Journal of the
Structural Division, Vol. 104, No. ST9, September 1978,
pp. 1,337-1,353.
Galambos, T. T. and M. K. Ravindra Properties of
Steel for Use in LRFD ASCE Journal of the Structural Division, Vol. 104, No. ST9, September 1978,
pp. 1,459-1,468.
Galambos, T. V. Load and Resistance Factor
Design AISC Engineering Journal, 3rd Qtr., 1981,
pp. 74-82.
Hardash, S. G. and Reidar Bjorhovde Gusset Plate
Design Utilizing Block-Shear Concepts Research
Report, Dept. of Civil Engineering, Univ. of ArizonaTucson, August 1984.
American Institute of Steel Construction, Inc. Manual
of Steel Construction 8th Ed., 1980, Chicago, III.
American Institute of Steel Construction,
Inc.
Proposed Load and Resistance Factor Design Specification for Structural Steel Buildings Sept. 1, 1983,
Chicago, III.

93

14. Rabern, D. A. Stress, Strain and Force Distributions

in Gusset Plate Connections Thesis presented to the
Univ. of Arizona-Tucson, 1983, in partial fulfillment
of the requirements for the degree of Master of Science
in Civil Engineering.
15. Richard, R. M., D. A. Rabern, D. E. Hormby and G.

C.

Williams

Analytical

Models

for

Steel

Connections Behavior of Metal Structures, Proceedings of the W.H. Munse Symposium, W. J. Hall and
M. P. Gaus, Eds., ASCE, May 17, 1983, pp. 128155.
16. Rides, J. M. and J. A. Yura Strength of Double-Row
Bolted-Web Connections ASCE Journal of the Structural Division, Vol. 109, No. 1, January 1983, pp.
126-142.

94

Steel Structures for

17. Canadian Standards Association
BuildingsLimit States Design CSA Standard S16.11974, Rexdale, Ontario, Canada, December 1974.
18. Struik, J.H. A. Applications of Finite Element Analysis to Nonlinear Plane Stress Problems Thesis presented to Lehigh Univ., Bethlehem, Pa., 1972, in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Civil Engineering.
19. Whitmore, R. E. Experimental Investigation of Stresses
in Gusset Plates Univ. of Tennessee Engineering Experiment Station Bulletin 16, May 1952.
20. Yura, J. A., P. C. Birkemoe and J. C. Rides
Double Angle Beam Web Connections: An Experimental Study ASCE Journal of the Structural Division, Vol. 108, No. ST2, February 1982, pp. 311-325.

ENGINEERING JOURNAL/AMERICAN INSTITUTE OF STEEL CONSTRUCTION