Journal of Constructional Steel Research 64 (2008) 571–583

www.elsevier.com/locate/jcsr

Efficiency reduction due to shear lag on bolted cold-formed steel angles

Valdeir Francisco de Paula a , Luciano Mendes Bezerra b,∗ , William Taylor Matias b
a Federal Centre for Technological Education of Goias, CEFET-GO, Brazil
b Department of Civil Engineering, University of Brasilia, PECC/ENC/FT, Brazil

Received 7 February 2007; accepted 14 October 2007

Abstract

This work presents 66 new experimental tests carried out on cold-formed steel angles fastened with bolts and under tension. In order to calculate
tension-members’ ultimate capacity, net-section failure is considered. The shear-lag phenomenon reduces net-section capacity. This reduction is
computed through the reduction coefficient which is a function of two parameters: length of the connection and distance of the shear plane to
the centroid of the cross-section. This article examines the reduction coefficient performance based upon the new tests and data available in the
literature, comprising a total of 108 lab tests. A new expression for the net-section reduction coefficient is suggested.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Cold-formed steel angles; Shear lag; Reduction coefficient; Net-section failure

1. Introduction
In civil engineering steel construction, three main steel
sections are usually employed for beams, columns and tension-
members: (a) hot-rolled (b) welded and (c) cold-formed
sections. Cold-formed sections are obtained from cold bending
of thin steel sheets. The methods to manufacture cold-formed
sections allow for great freedom in the designing process and
offer to architects and engineers a wide variety of shapes
and sizes for steel sections. Before 1940, cold-formed steel
sections were not popular in the construction industry. Since
then, cold-formed steel sections have played a significant role
in residential construction and there has been a steady trend
to extend their use in low rise buildings [1]. In addition, cold-
formed sections are widely used in many other applications
such as in vehicle frames, storage racks in warehouses, bracing
members and chords of trusses, hangers for floors and roofs,
short-span floor beams for mezzanines, and panels for walls.
The primary advantages of cold-formed sections are price, light
weight, high strength and stiffness, uniform quality, ease of
prefabrication and mass production, economy in transportation
and handling, fast and easy erection and installation, among
others. In the United States, about 15,000 steel homes were built
∗ Corresponding author. Tel.: +55 61 33491441; fax: +55 61 32734644.
E-mail address: lmbz@unb.br (L.M. Bezerra).
in 1993 and 75,000 in 1996 [2]. In Brazil the use of cold-formed
steel sections has grown substantially.
One of the most useful and common structural members
in cold-formed steel is the tension member. The performance
of such structural members differs in several ways from those
of heavy steel sections [3]. The development of good design
specification for cold-formed members is highly desirable. The
North American Standard for USA, Canada and Mexico, AISI
LRFD [4], and the NBR-14762 [5] in Brazil are examples of
building codes corresponding to cold-formed members. Given
these specifications in order to determine tension-member
ultimate capacity when connected with bolts, the following
failure modes should be considered: (a) yielding of the entire
cross-section; (b) net-section failure; (c) block-shear failure
(d) bolt-shear fracture; and (e) yielding of the gusset plate
leading to a plate bulge in front of the bolt. The AISI LRFD
provisions [4] state that tension-member design strength is
the lesser of the values obtained among the verifications
mentioned before. This paper is concerned with the capacity
of cold-formed steel members under tension and connected
with bolts. This research is particularly interested in the shear-
lag phenomenon which influences the net-section capacity by
means of a reduction coefficient. Sixty-six specimens were
tested in the laboratory and the results are reported here together
with other experiments reported in the literature. Based on these

c 2007 Elsevier Ltd. All rights reserved.

0143-974X/$ - see front matter

doi:10.1016/j.jcsr.2007.10.008
572 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583
experiments, a mathematical model is developed and a new
expression for the reduction coefficient “U ” is suggested.
2. The problem
Among the verifications mentioned in the last section,
net-section failure (for nonflat sheets) takes into account
a debilitating phenomenon known as “shear lag”. Such a
phenomenon is due to the nonuniform straining of the tension
member when connected to the gusset plate. The distance
from the gusset plate to the centre of gravity of the steel
section causes reduction in the tension-member’s effectiveness
and is taken into consideration through the use of a reduction
coefficient [4,5]. The ultimate capacity (Tr ) is the nominal
resistance (Tn ) factored by a coefficient φ = 0.75. According
to the codes [4,5], net-section failure (ultimate capacity),
considering the shear-lag phenomenon, is given by
Tr = φTn with Tn = U An Fu (1)
where An is the net-section area, U is the net-section reduction
coefficient, and Fu is the ultimate tensile strength of the steel
material used. Depending on the type of steel section, there
are different expressions for U . The calculation of the net
section depends on hole patterns and whether these holes are
in line or staggered. The use of the staggered pattern is not very
common in cold-formed steel applications [6] and is out of the
scope of this article. The mathematical expression in Eq. (1)
for angle members [4] are relatively new and was established
based on research studies at the University of Missouri-Rolla
by LaBoube and Yu [7]. These studies proposed the following
expression so as to determine the reduction coefficient U for
angle members with two or more bolts in line with the direction
of the acting force:
 
x
U = 1 − 1.20 < 0.9 but >0.4 (2)
L to the centroid of the angle section; and “L” is the distance
where L is the length of the bolted connection and is measured
“centre-to-centre” from the first to the last bolt in the line along
the direction of the applied force; x̄ is the distance from the
shear plane to the centroid of the cross-section of the tension
member. Eq. (2) takes into account the shear lag when the
applied load is not transmitted directly to all elements of the
cross-section.
Recent experiments reported in the literature [8,9] demon-
strate that Eq. (2) may be modified for a better representation of
the shear lag phenomenon [10] and, consequently, a better de-
termination of the cold-formed tension-member’s ultimate ca-
pacity when connected with bolts. Note that Eq. (2) does not
take into account the connected and nonconnected legs’ widths,
the angle section’s thickness, and bolt diameter, among other
parameters. It is difficult to believe that a complex phenomenon
such as the shear lag where the stress flow changes in direction
and in magnitude depends only on L and x̄.
3. Reduction coefficient “U” performance
In this section, the performance of the expression adopted
by AISI in establishing the reduction coefficient U for angle
Fig. 1. Typical cold-formed angle for a bolt connection.
members is examined. The ratio between the ultimate capacity
to the nominal net-section strength is used to evaluate the
coefficient U . The experimental ultimate capacity is here
named “Texp ”. The net-section strength “Tn ” is obtained from
Eq. (1) using measured proprieties for material and geometry.
Experimental data available in the literature are first used
for a preliminary evaluation of the reduction coefficient “U ”
expressed in Eq. (2). The experimental data from Holcomb
et al. [11] and Yip and Cheng [8] report recent experimental
tests over 42 cold-formed steel angle members under tension.
Table 1 indicates these angles’ geometrical parameters. All
angles reported in Table 1 have one bolt per cross-section.
The identification of each angle is in the third column and is
explained in Refs. [8,11]. The data from Holcomb et al. [11]
are marked with UMR (corresponding to the University of
Missouri-Rolla where the experiments were conducted); and
the data from Yip and Cheng [8] are marked as UofA (in
reference to the University of Alberta). “bc ” is the width
of the connected leg of the angle; “bd ” is the width of the
nonconnected leg; “t” is the uniform thickness; “d” is the
nominal bolt diameter; “x̄” is the distance of the shear plane
between extreme bolts along the applied force direction. Fig. 1
illustrates a typical cold-formed angle.
The angle in Fig. 1 is connected to a gusset plate by bolts.
In Table 1, the angles have two, three or four bolts per bolt
line. For the evaluation of Eq. (1), the ratio between Texp and
Tn is considered according to Eq. (3). When this ratio is less
than one, the nominal net-section strength Tn is greater than
the experimental value Texp (here considered as the true value).
In this case, Tn , determined according to Eqs. (1) and (2), is
against safety.
Texp Texp
= . (3)
Tn U An Fu
Table 2 reports the values for Texp , Fu , A (the gross area of
the angle member); An ; Tn ; (Texp − Tn ) which is the difference
in resistance and also expresses this difference in percentage
form with reference to Texp ; the ratio in Eq. (3) and UAISI (from
Eq. (2)). We use UAISI as most standards [5] follow AISI rec-
ommendations [4]. Table 2 also reports the ideal value for the
reduction coefficient U obtained from the value of the ultimate
 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583 573

Table 1
Characterization of cold-formed steel angles—Refs. [8,11]

Test no. Source Angle denomination bc (mm) bd (mm) t (mm) Number of bolt lines d (mm) x̄ (mm) L (mm)
1 LBN11-1 41.3 41.3 1.067 2 12.7 10.81 38.10
2 LBN11-2 41.3 41.3 1.067 2 12.7 10.81 38.10
3 LBN11-3 41.3 41.3 1.067 2 12.7 10.81 38.10
4 LCN11-1 41.3 41.3 1.067 3 12.7 10.81 76.20
5 LCN11-2 41.3 41.3 1.067 3 12.7 10.81 76.20
6 LCN11-3 41.3 41.3 1.067 3 12.7 10.81 76.20
7 LBN12-1 41.3 82.5 1.067 2 12.7 28.06 38.10
8 LBN12-2 41.3 82.5 1.067 2 12.7 28.06 38.10
9 LBN12-3 41.3 82.5 1.067 2 12.7 28.06 38.10
10 LCN12-1 41.3 82.5 1.067 3 12.7 28.06 76.20
11 LCN12-2 41.3 82.5 1.067 3 12.7 28.06 76.20
12 LBN13-1 82.5 41.3 1.067 2 12.7 7.34 38.10
13 LBN13-2 82.5 41.3 1.067 2 12.7 7.34 38.10
14 UMR (1995) LCN13-1 82.5 41.3 1.067 3 12.7 7.34 76.20
15 LCN13-2 82.5 41.3 1.067 3 12.7 7.34 76.20
16 LBN31-1 41.3 41.3 3.048 2 12.7 11.74 38.10
17 LBN31-2 41.3 41.3 3.048 2 12.7 11.74 38.10
18 LCN31-1 41.3 41.3 3.048 3 12.7 11.74 76.20
19 LCN31-2 41.3 41.3 3.048 3 12.7 11.74 76.20
20 LBN32-1 41.3 82.5 3.048 2 12.7 29.14 38.10
21 LBN32-2 41.3 82.5 3.048 2 12.7 29.14 38.10
22 LCN32-1 41.3 82.5 3.048 3 12.7 29.14 76.20
23 LCN32-2 41.3 82.5 3.048 3 12.7 29.14 76.20
24 LBN33-1 82.5 41.3 3.048 2 12.7 8.20 38.10
25 LBN33-2 82.5 41.3 3.048 2 12.7 8.20 38.10
26 LCN33-1 82.5 41.3 3.048 3 12.7 8.20 76.20
27 LCN33-2 82.5 41.3 3.048 3 12.7 8.20 76.20
28 12.2 102 102 2.657 2 19.1 26.72 95.50
29 12.3 102 102 2.657 3 19.1 26.72 191.00
30 12.4 102 102 2.657 3 19.1 26.72 191.00
31 UofA (1997) 14.2 50.8 50.8 1.897 2 15.9 13.57 63.30
32 14.3 50.8 50.8 1.897 3 15.9 13.57 126.60
33 16.2 38.1 38.1 1.519 2 12.7 10.22 38.10
34 16.3 38.1 38.1 1.519 3 12.7 10.22 76.20
35 A2-2 51 51 1.214 2 19.1 13.31 63.50
36 A2-2N 51 51 1.214 2 19.05 13.31 63.50
37 A2-3 51 51 1.214 3 19.05 13.31 127.00
38 A3-2 76 76 1.214 2 19.05 19.55 63.50
UofA (1999)
39 A3-3 76 76 1.214 3 19.1 19.55 127.00
40 A4-2 102 102 1.214 2 19.05 26.05 63.50
41 A4-3 102 102 1.214 3 19.05 26.05 127.00
42 A4-4 102 102 1.214 4 19.05 26.05 190.50

load Texp divided by An Fu , here named Uexp —see Eq. (4).

Tult
Uexp = . (4)
An Fu

In Fig. 2, Texp is compared to Tn . It can be observed that

for most of the specimens tested, Texp < Tn is against safety.
Fig. 3 plots the values of Uexp and compares them to UAISI from
Eq. (2). In looking at Fig. 3, it is clear that the values of Uexp
are under the values given by Eq. (2) for many points. Based
upon Figs. 1 and 2, it is also clear that reduction-coefficient
performance varies with the number of bolts and also according
Fig. 2. Reduction coefficient performance in net-section failure.
to the x̄/L parameter. From the plots, it can be observed that in
several instances, the nominal net-section strength Tn in Eq. (1)
and computed through the reduction coefficient U defined in In other words, the suggested Eq. (2) for the reduction
Eq. (2), turn out to be unsafe. coefficient U is not safe for almost all the situations reported in
574 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583

Table 2
Reduction coefficient “U ” performance based on the lab tests [1,4]
Texp
Test no. Angle denomination Texp (kN) Fu (MPa) A (mm2 ) An (mm2 ) Tn (kN) Texp −T n Tn UAISI Uexp
(kN) (%)
1 LBN11-1 15.80 385 86.3 71.11 18.1 −2.3 −14.3 0.875 0.659 0.577
2 LBN11-2 16.20 385 86.3 71.11 18.1 −1.9 −11.4 0.897 0.659 0.592
3 LBN11-3 15.90 385 86.3 71.11 18.1 −2.2 −13.5 0.881 0.659 0.581
4 LCN11-1 19.60 385 86.3 71.11 22.7 −3.1 −15.9 0.863 0.830 0.716
5 LCN11-2 20.00 385 86.3 71.11 22.7 −2.7 −13.6 0.880 0.830 0.731
6 LCN11-3 20.90 385 86.3 71.11 22.7 −1.8 −8.7 0.920 0.830 0.763
7 LBN12-1 17.90 385 130.2 115.07 17.7 0.2 1.0 1.010 0.400 0.404
8 LBN12-2 19.30 385 130.2 115.07 17.7 1.6 8.2 1.089 0.400 0.436
9 LBN12-3 18.20 385 130.2 115.07 17.7 0.5 2.6 1.027 0.400 0.411
10 LCN12-1 21.90 385 130.2 115.07 24.7 −2.8 −12.9 0.886 0.558 0.494
11 LCN12-2 22.80 385 130.2 115.07 24.7 −1.9 −8.4 0.922 0.558 0.515
12 LBN13-1 25.30 385 130.2 115.07 34.1 −8.8 −34.6 0.743 0.769 0.571
13 LBN13-2 24.40 385 130.2 115.07 34.1 −9.7 −39.6 0.716 0.769 0.551
14 LCN13-1 29.80 385 130.2 115.07 39.2 −9.4 −31.5 0.761 0.884 0.673
15 LCN13-2 31.70 385 130.2 115.07 39.2 −7.5 −23.6 0.809 0.884 0.716
16 LBN31-1 49.00 366 236.5 193.21 44.6 4.4 9.0 1.099 0.630 0.693
17 LBN31-2 48.30 366 236.5 193.21 44.6 3.7 7.7 1.084 0.630 0.683
18 LCN31-1 58.50 366 236.5 193.21 57.6 0.9 1.5 1.015 0.815 0.827
19 LCN31-2 56.70 366 236.5 193.21 57.6 −0.9 −1.7 0.984 0.815 0.802
20 LBN32-1 52.00 366 362.1 318.79 46.7 5.3 10.2 1.114 0.400 0.446
21 LBN32-2 56.00 366 362.1 318.79 46.7 9.3 16.7 1.200 0.400 0.480
22 LCN32-1 62.90 366 362.1 318.79 63.1 −0.2 −0.4 0.996 0.541 0.539
23 LCN32-2 60.20 366 362.1 318.79 63.1 −2.9 −4.9 0.953 0.541 0.516
24 LBN33-1 80.90 366 362.1 318.79 86.6 −5.7 −7.0 0.935 0.742 0.693
25 LBN33-2 79.60 366 362.1 318.79 86.6 −7.0 −8.7 0.920 0.742 0.682
26 LCN33-1 88.30 366 362.1 318.79 101.6 −13.3 −15.1 0.869 0.871 0.757
27 LCN33-2 90.90 366 362.1 318.79 101.6 −10.7 −11.8 0.895 0.871 0.779
28 12.2 135.80 516 530.4 475.69 163.1 −27.3 −20.1 0.833 0.664 0.553
29 12.3 154.70 516 530.4 475.69 204.3 −49.6 −32.0 0.757 0.832 0.630
30 12.4 158.30 516 530.4 475.69 204.3 −46.0 −29.0 0.775 0.832 0.645
31 14.2 35.70 327 186.8 153.81 37.4 −1.7 −4.6 0.956 0.743 0.710
32 14.3 43.00 327 186.8 153.81 43.8 −0.8 −1.9 0.981 0.871 0.855
33 16.2 20.30 317 112.0 90.39 19.4 0.9 4.3 1.045 0.678 0.708
34 16.3 24.40 317 112.0 90.39 24.0 0.4 1.5 1.015 0.839 0.852
35 A2-2 27.90 316 121.4 96.40 22.8 5.1 18.3 1.224 0.749 0.916
36 A2-2N 24.00 316 121.4 96.46 22.8 1.2 4.9 1.052 0.749 0.787
37 A2-3 31.10 316 121.4 96.46 26.6 4.5 14.3 1.167 0.874 1.020
38 A3-2 32.80 316 182.1 157.16 31.3 1.5 4.5 1.048 0.630 0.660
39 A3-3 37.70 316 182.1 157.10 40.5 −2.8 −7.3 0.932 0.815 0.759
40 A4-2 34.00 316 245.2 220.29 35.3 −1.3 −3.9 0.962 0.508 0.488
41 A4-3 45.10 316 245.2 220.29 52.5 −7.4 −16.3 0.859 0.754 0.648
42 A4-4 49.40 316 245.2 220.29 58.2 −8.8 −17.8 0.849 0.836 0.710

safe values for the nominal net-section strength Tn in all

designing situations. With this in mind, this research conducted
66 more lab tests on cold-formed angle sections connected with
bolts. Such experiments together with the data available in the
literature [8,11] will provide sufficient data to be statistically
treated for a more precise equation of the reduction coefficient
“U ”.

4. The experimental programme

The experimental programme described here sought to

Fig. 3. Experimental coefficient U [8,11] and U as a function of x̄/L.
Table 2. Given this conclusion, Eq. (2) can be re-evaluated and
this article tries to get a better expression that could produce
gather both simple quantitative data as well as qualitative
information on cold-formed steel angle behaviour under tension
and when connected with bolts. The specimens tested have their
ends connected by a variable number of bolts and in different
arrangements. The results of these tests will be statistically
 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583 575

Fig. 4. Cross-section dimensions (mm) of angle specimens with equal and unequal legs.

Fig. 5. Configuration of bolted connections specimens with 600 mm long. Note the gusset plates and the number of bolt lines and the number of bolts per bolt line.

analysed later on to obtain a mathematical expression for the
reduction coefficient or shear-lag effect on the net section. To
achieve this goal, 88 cold-formed steel angle members were
tested until rupture [12]. From the 88 specimens tested, 66
showed net-section failure.
The specimens prepared for the laboratory tests measure
600 mm long and were cut and manufactured from narrow
plate sheets 3000 mm long. Different sheet thicknesses were
used. For the same thickness, the specimens were taken from
the same roll of steel sheet and cold-formed by appropriate
machinery. The steel is known in the Brazilian industry as
COR-420 which has anticorrosive properties, yielding stress of
300 MPa and ultimate stress of 420 MPa. The angle members
have different thickness dimensions: 2.25, 3.35 and 3.75 mm,
with equal or different legs. The dimensions (in mm) of the
angle legs are: 50 × 50, 80 × 80, 100 × 100, 50 × 80 or
80 × 100. Different configurations for the bolted connections
were worked out and the angles are connected with one or
two bolt lines. Each bolt line has one, two, three or four bolts.
Only one angle leg is fastened. Figs. 4 and 5 show the angle
sections and give a picture of how the bolts are distributed in the
connections. It is observed here that only the Brazilian code [5]
recognizes the possibility of using one bolt in the connection if
desired.
The angle members were tested with an Electromechanical
EMIC Tensile Testing Equipment, Model DL 60000. The
EMIC equipment is connected to a microcomputer for
automatic data acquisition with 2 Hz frequency for data
transmission. The positioning of the angle specimens to the
EMIC equipment for the tensile testing is achieved by gusset
plates which are placed within the grips of the equipment. The
gusset plates are chunky, made of steel, and 12.7 mm (1/2”)
thick. In Fig. 6, the gusset plate is placed within the tension
grips of the EMIC DL-60000 Testing Equipment. Fig. 7 shows
the gusset plate geometry and dimensions.
The load applied to the specimens is controlled by a load
cell. The axial deformation is recorded from a displacement
translator fixed at the bolts. The bolts are ASTM A325 with
12.7 mm (1/2”) in diameter and tightened with a torquing
moment of 100 N m. Washers are not employed. The holes
for the bolt are 1.5 mm larger than the bolt diameter. All
holes were drilled. The distance between sequential bolts is
3d (38.1 mm). Here, “d” is the nominal bolt diameter. The
distance bolt-to-edge, in the direction of the applied load, is
576 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583

Table 3
Geometrical data of the angle specimens tested at UnB [12]

Test Angle denomination bc (mm) bd (mm) t (mm) Number of bolt lines No. of holes per bolt line d (mm) x (mm) L (mm)
1 A121 50 50 2.23 2 1 12.7 13.53 38.10
2 A131 50 50 2.26 3 1 12.7 13.54 76.20
3 A141 50 50 2.34 4 1 12.7 13.58 114.30
4 A221 50 50 3.51 2 1 12.7 14.13 38.10
5 A231 50 50 3.49 3 1 12.7 14.12 76.20
6 A241 50 50 3.57 4 1 12.7 14.15 114.30
7 A321 50 50 3.7 2 1 12.7 14.22 38.10
8 A331 50 50 3.72 3 1 12.7 14.23 76.20
9 A341 50 50 3.66 4 1 12.7 14.20 114.30
10 B131 80 80 2.4 3 1 12.7 21.10 76.20
11 B141 80 80 2.26 4 1 12.7 21.04 114.30
12 B221 80 80 3.54 2 1 12.7 21.63 38.10
13 B231 80 80 3.55 3 1 12.7 21.63 76.20
14 B241 80 80 3.68 4 1 12.7 21.69 114.30
15 B321 80 80 3.86 2 1 12.7 21.78 38.10
16 B331 80 80 3.81 3 1 12.7 21.76 76.20
17 B341 80 80 3.76 4 1 12.7 21.73 114.30
18 B122 80 80 2.43 2 2 12.7 21.11 38.10
19 B132 80 80 2.43 3 2 12.7 21.11 76.20
20 B142 80 80 2.43 4 2 12.7 21.11 114.30
21 B212 80 80 3.49 1 2 12.7 21.61 0.00
22 B222 80 80 3.5 2 2 12.7 21.61 38.10
23 B232 80 80 3.53 3 2 12.7 21.62 76.20
24 B242 80 80 3.53 4 2 12.7 21.62 114.30
25 C131 100 100 2.25 3 1 12.7 26.03 76.20
26 C141 100 100 2.56 4 1 12.7 26.17 114.30
27 C221 100 100 3.51 2 1 12.7 26.61 38.10
28 C231 100 100 3.49 3 1 12.7 26.60 76.20
29 C241 100 100 3.69 4 1 12.7 26.70 114.30
30 C331 100 100 3.91 3 1 12.7 26.80 76.20
31 C341 100 100 3.87 4 1 12.7 26.78 114.30
32 C122 100 100 2.66 2 2 12.7 26.22 38.10
33 C132 100 100 2.42 3 2 12.7 26.11 76.20
34 C142 100 100 2.45 4 2 12.7 26.12 114.30
35 C212 100 100 3.58 1 2 12.7 26.64 0.00
36 C222 100 100 3.58 2 2 12.7 26.64 38.10
37 C232 100 100 3.59 3 2 12.7 26.65 76.20
38 C242 100 100 3.56 4 2 12.7 26.64 114.30
39 C312 100 100 3.86 1 2 12.7 26.77 0.00
40 C322 100 100 3.86 2 2 12.7 26.77 38.10
41 C332 100 100 3.85 3 2 12.7 26.77 76.20
42 C342 100 100 3.84 4 2 12.7 26.76 114.30
43 D121 50 80 2.41 2 1 12.7 25.84 38.10
44 D131 50 80 2.43 3 1 12.7 25.85 76.20
45 D141 50 80 2.36 4 1 12.7 25.82 114.30
46 E121 50 100 2.49 2 1 12.7 34.67 38.10
47 E131 50 100 2.58 3 1 12.7 34.72 76.20
48 E141 50 100 2.38 4 1 12.7 34.61 114.30
49 F121 80 100 2.34 2 1 12.7 28.90 38.10
50 F131 80 100 2.34 3 1 12.7 28.90 76.20
51 F141 80 100 2.3 4 1 12.7 28.88 114.30
52 F122 80 100 2.46 2 2 12.7 28.96 38.10
53 F132 80 100 2.48 3 2 12.7 28.97 76.20
54 F142 80 100 2.38 4 2 12.7 28.92 114.30
55 D121-L 80 50 2.29 2 1 12.7 10.61 38.10
56 D131-L 80 50 2.26 3 1 12.7 10.59 76.20
57 D141-L 80 50 2.24 4 1 12.7 10.58 114.30
58 D112-L 80 50 2.22 1 2 12.7 10.57 0.00
59 D122-L 80 50 2.21 2 2 12.7 10.57 38.10
60 D132-L 80 50 2.23 3 2 12.7 10.58 76.20
61 D142-L 80 50 2.23 4 2 12.7 10.58 114.30
62 E131-L 100 50 2.25 3 1 12.7 9.29 76.20
63 E141-L 100 50 2.29 4 1 12.7 9.31 114.30
 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583 577

Table 3 (continued)

Test Angle denomination bc (mm) bd (mm) t (mm) Number of bolt lines No. of holes per bolt line d (mm) x (mm) L (mm)
64 E122-L 100 50 2.27 2 2 12.7 9.30 38.10
65 E132-L 100 50 2.29 3 2 12.7 9.31 76.20
66 E142-L 100 50 2.27 4 2 12.7 9.30 114.30

Fig. 6. (a) Electromechanical EMIC DL-60000 Testing Equipment. (b) Specimens and the positioning of the displacement translator.

of four or five characters, forming individual labels for each

Fig. 7. Gusset plates (12.7 mm thick, all dimensions in mm) and bolt holes to
hold the cold-formed angle specimens during tensile testing.
2.5d (31.75 mm). For angles with different leg dimensions,
the connected leg is the short one—unless otherwise stated. A
system of identification for the specimens is established. The
identification uses letters (L) and numbers (N ) in a sequence
situation. If the attached leg is the long leg, there will be the
letter “L” attached to the end of the identifier and separated
from the other four characters by a hyphen. The general form
of a specimen identifier is therefore “LNNN-L”. Each character
at a certain position is explained as follows:
• First position (L). Letters A to F, represent the dimensions
in mm of the angle legs. A = 50 × 50, B = 80 × 80, C =
100 × 100, D = 50 × 80, E = 50 × 100 and F = 80 × 100.
• Second position (N). Integer numbers 1–3 represent the
nominal thickness of the angle, and: 1 = 2.25 mm,
2 = 3.35 mm and 3 = 3.75 mm.
• Third position (N). Integer numbers 1–4 represent the
number of bolts along each bolt line.
• Fourth position (N). Integer numbers one or two represent
the number of bolt lines.
• Fifth position (L). Optional. The letter L after a hyphen
means the long leg is connected.
Examples of specimen identifiers: (a) Specimen C345:
Angle 100 × 100 mm, thickness 3.75 mm, four bolts per
bolt line, and two bolt lines. (b) Specimen E231-L: Angle
50 × 100 mm, thickness 3.35 mm, three bolts per bolt line,
and one bolt line, and the long leg is connected to the machine
attachment.
Table 3 summarizes all the geometrical information on
the 66 angle specimens tested in the Structures Lab at the
Department of Civil Engineering in the University of Brası́lia
(UnB) and at CEFET-GO labs. The loading is applied slowly
578 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583

Table 4
Performance of the Efficiency Coefficient according to the tests at UnB [12]

Test Angle denomination Texp (kN) Fu (MPa) A (mm2 ) An (mm2 ) Tn (kN) Texp /Tn UAISI Uexp
1 A121 54.830 502.00 214.83 183.16 52.774 1.039 0.574 0.596
2 A131 64.588 502.00 217.61 185.51 73.269 0.882 0.787 0.694
3 A141 78.063 502.00 225.00 191.77 82.546 0.946 0.857 0.811
4 A221 89.145 463.00 330.75 280.91 72.195 1.235 0.555 0.685
5 A231 97.756 463.00 328.98 279.42 100.612 0.972 0.778 0.756
6 A241 102.002 463.00 336.05 285.36 112.488 0.907 0.851 0.772
7 A321 81.653 457.00 347.50 294.96 74.443 1.097 0.552 0.606
8 A331 97.461 457.00 349.26 296.43 105.122 0.927 0.776 0.719
9 A341 109.573 457.00 343.98 292.01 113.559 0.965 0.851 0.821
10 B131 93.941 502.00 374.53 340.45 110.166 0.853 0.668 0.550
11 B141 92.009 502.00 353.21 321.11 125.597 0.733 0.779 0.571
12 B221 108.141 463.00 545.80 495.54 91.773 1.178 0.400 0.471
13 B231 130.501 463.00 547.29 496.88 150.295 0.868 0.659 0.567
14 B241 142.299 463.00 566.54 514.29 183.881 0.774 0.772 0.598
15 B321 115.369 457.00 593.11 538.30 98.401 1.172 0.400 0.469
16 B331 140.249 457.00 585.74 531.64 159.212 0.881 0.657 0.577
17 B341 142.917 457.00 578.36 524.97 185.175 0.772 0.772 0.596
18 B122 98.393 502.00 379.10 310.08 62.264 1.580 0.400 0.632
19 B132 103.964 502.00 379.10 310.08 103.902 1.001 0.667 0.668
20 B142 109.181 502.00 379.10 310.08 121.155 0.901 0.778 0.701
21 B212 86.811 463.00 538.38 439.27 54.306 1.599 0.267 0.427
22 B222 119.380 463.00 539.87 440.47 81.574 1.463 0.400 0.585
23 B232 128.952 463.00 544.32 444.07 135.586 0.951 0.659 0.627
24 B242 131.619 463.00 544.32 444.07 158.925 0.828 0.773 0.640
25 C131 110.309 502.00 441.68 409.73 103.281 1.068 0.590 0.536
26 C141 115.232 502.00 501.23 464.88 156.681 0.735 0.725 0.494
27 C221 122.175 463.00 681.75 631.91 99.067 1.233 0.400 0.418
28 C231 152.145 463.00 677.98 628.42 147.755 1.030 0.581 0.523
29 C241 164.914 463.00 715.62 663.22 208.296 0.792 0.720 0.537
30 C331 160.511 457.00 756.87 701.35 163.391 0.982 0.578 0.501
31 C341 179.144 457.00 749.39 694.43 215.626 0.831 0.719 0.564
32 C122 99.560 502.00 520.37 444.83 89.321 1.115 0.400 0.446
33 C132 115.899 502.00 474.38 405.65 119.909 0.967 0.589 0.569
34 C142 122.420 502.00 480.13 410.55 149.576 0.818 0.726 0.594
35 C212 94.274 463.00 694.94 593.26 54.909 1.717 0.200 0.343
36 C222 139.592 463.00 694.94 593.26 109.872 1.270 0.400 0.508
37 C232 154.195 463.00 696.82 594.86 159.835 0.965 0.580 0.560
38 C242 161.746 463.00 691.17 590.07 196.804 0.822 0.720 0.592
39 C312 95.265 457.00 747.51 637.89 58.362 1.632 0.200 0.327
40 C322 143.976 457.00 747.51 637.89 116.606 1.235 0.400 0.494
41 C332 170.161 457.00 745.64 636.30 168.201 1.012 0.578 0.585
42 C342 171.612 457.00 743.77 634.71 208.555 0.823 0.719 0.592
43 D121 62.058 502.00 303.75 269.53 54.122 1.147 0.400 0.459
44 D131 71.061 502.00 306.20 271.69 80.861 0.879 0.593 0.521
45 D141 78.397 502.00 297.65 264.13 96.656 0.811 0.729 0.591
46 E121 64.363 502.00 363.31 327.95 65.852 0.977 0.400 0.391
47 E131 76.700 502.00 376.06 339.42 77.223 0.993 0.453 0.450
48 E141 81.545 502.00 347.69 313.89 100.314 0.813 0.637 0.517
49 F121 68.570 502.00 412.20 378.97 71.608 0.958 0.400 0.360
50 F131 80.770 502.00 412.20 378.97 103.664 0.779 0.545 0.425
51 F141 89.253 502.00 405.31 372.65 130.349 0.685 0.697 0.477
52 F122 92.244 502.00 432.85 362.99 72.888 1.266 0.400 0.506
53 F132 99.442 502.00 436.29 365.86 99.881 0.996 0.544 0.541
54 F142 107.592 502.00 419.09 351.50 122.880 0.876 0.696 0.610
55 D121-L 67.011 502.00 289.08 256.56 70.078 0.956 0.666 0.520
56 D131-L 77.514 502.00 285.41 253.31 103.740 0.747 0.833 0.610
57 D141-L 85.418 502.00 282.95 251.15 112.066 0.762 0.889 0.678
58 D112-L 62.274 502.00 280.50 217.45 38.511 1.617 0.353 0.570
59 D122-L 78.612 502.00 279.27 216.51 72.502 1.084 0.667 0.723
60 D132-L 86.605 502.00 281.73 218.40 91.369 0.948 0.833 0.790
61 D142-L 92.734 502.00 281.73 218.40 97.457 0.952 0.889 0.846
62 E131-L 89.626 502.00 329.18 297.23 103.281 0.868 0.854 0.601
63 E141-L 102.708 502.00 334.88 302.36 136.948 0.750 0.902 0.677
 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583
Table 4 (continued)
Test Angle denomination Texp (kN)
64 E122-L 81.672
65 E132-L 93.411
66 E142-L 96.059
Fig. 8. Details of net-section failure of a cold-formed angle with two bolts.
Fig. 9. Typical load–displacement plots from the angle specimens tested.
through a controlled displacement set to 2 mm/min. Fig. 8
shows a typical specimen after the net-section’s collapse—in
this case specimen A121. From the 88 specimens tested, 66
specimens presented collapse of the net section and are listed
in Table 4.
Fig. 9 shows typical plots of load versus displacement
obtained from various angles tested. By examining these plots,
the load at failure is recognized as the point where the load
starts to decrease.
5. Test results
From the 66 specimens tested, four angles have one bolt
and were tested just to compare the results foreseen in the
Brazilian code NBR [5] which also allows connection with
one bolt. The other 62 specimens have more than one bolt
per bolt line as recommended by the codes [4,5]. The 66
specimens clearly showed that the main source of rupture
was the net-section failure. The results reported in Table 4
demonstrate that for a great number of the specimens the
experimental (true) value Texp is less than the nominal net-
section strength Tn calculated according to Eq. (1). In these
cases, as Texp /Tn < 1, the experimental reduction coefficient
579
Fu (MPa) A (mm2 ) An (mm2 ) Tn (kN) Texp /Tn UAISI Uexp
502.00 332.03 267.56 94.961 0.860 0.707 0.608
502.00 334.88 269.85 115.598 0.808 0.853 0.690
502.00 332.03 267.56 121.198 0.793 0.902 0.715
Uexp is inferior to the reduction coefficient UAISI from Eq. (3).
The UAISI relies just on two basic parameters: (1) connection
eccentricity (x̄); and (2) length of the bolted connection (L).
Results in Table 4 demonstrate that the net-section failure
may depend on other parameters too. The links with other
parameters can be incorporated in an alternative equation for
“U ” that can be obtained through the use of statistics and
multiple regression, as is done in the next section.
6. Statistical analysis of the experimental test results
The stress distribution in the vicinity of the bolted
connection is very complex. This is a region of nonuniform
strain, high stress concentration, plastic zone formation and
stress-direction change. The shear-lag phenomenon has to do
with the transfer of such stress from the tension member to bolts
under such a complex stress distribution. The efficiency factor
expresses the reduction on the tension member’s effectiveness
due to the shear-lag phenomenon. It is important for engineers
to be able to use safety expressions in calculating the reduction
coefficient. Table 4 demonstrates how the bolt-connection
efficiency changes when certain parameters are modified. It is
important to examine Table 4 in light of the information given
by the identifier on the specimen dimensions, connected legs,
number of bolts and bolt arrangements. In fact, by studying the
identifiers, it can be noted that the width of the non-connected
leg (bd ) and the number of bolts per section (n) – besides
the traditional x̄ (connection eccentricity) and L (connection
length) – control the experimental reduction coefficient (last
column). Therefore, other parameters, in addition to x̄/L in
Eq. (2) may be considered in the construction of an alternating
expression for “U ”. The proposed alternating expression
depends on more than one independent variable and for that
reason statistical regression is applied.
In statistics, multiple regression is used to establish the
effect of each independent variable when the other independent
variable is kept constant [13]. Here a multiple linear regression
is conducted. The general purpose is to find if there is
a correlation between a certain independent variable (or
parameter) and the reduction coefficient (U ). The proposed
equation for coefficient U is based on the independent variables
X i1 , X i2 , X i3 , . . . , X i , p−1 associated with the geometry of the
angle and the bolted connection’s characteristics. The proposed
equation for the linear regression is [13,14]:
Ui = b0 + b1 X i1 + b2 X i2 + · · · + b p−1 X i, p−1 + εi (5)
where: Ui is the value of the dependent variable in test number
“i”; b0 , b1 , . . . , b p−1 are regression coefficients; and εi is the
error between the predicted Ui and the true experimental value
Uexp at test “i”. It is assumed that the distribution of εi is
580 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583

Table 5
Performance of the Efficiency Factor (U ) given by Eqs. (2) and (6) applied to the experimental tests carried out at UnB [12]

Test Number Specimen Experimental Uexp Today UAISI Difference (UAISI − Uexp ) Proposed UUnB Difference (UUnB − Uexp )
1 A121 0.596 0.574 −0.022 0.653 0.057
2 A131 0.694 0.787 0.093 0.7 0.006
3 A141 0.811 0.857 0.047 0.718 −0.093
4 A221 0.685 0.555 −0.13 0.692 0.007
5 A231 0.756 0.778 0.022 0.74 −0.016
6 A241 0.772 0.851 0.079 0.759 −0.013
7 A321 0.606 0.552 −0.053 0.698 0.092
8 A331 0.719 0.776 0.057 0.747 0.028
9 A341 0.821 0.851 0.03 0.762 −0.06
10 B131 0.55 0.668 0.118 0.544 −0.006
11 B141 0.571 0.779 0.208 0.565 −0.006
12 B221 0.471 0.4 −0.071 0.492 0.021
13 B231 0.567 0.659 0.092 0.566 −0.001
14 B241 0.598 0.772 0.175 0.593 −0.004
15 B321 0.469 0.4 −0.069 0.498 0.029
16 B331 0.577 0.657 0.08 0.571 −0.006
17 B341 0.596 0.772 0.176 0.595 −0.001
18 B122 0.632 0.4 −0.232 0.572 −0.06
19 B132 0.668 0.667 0 0.644 −0.024
20 B142 0.701 0.778 0.077 0.668 −0.033
21 B222 0.585 0.4 −0.185 0.592 0.006
22 B232 0.627 0.659 0.032 0.666 0.039
23 B242 0.64 0.773 0.133 0.69 0.05
24 C131 0.536 0.59 0.054 0.479 −0.057
25 C141 0.494 0.725 0.231 0.514 0.02
26 C221 0.418 0.4 −0.018 0.408 −0.01
27 C231 0.523 0.581 0.058 0.498 −0.025
28 C241 0.537 0.72 0.183 0.532 −0.005
29 C331 0.501 0.578 0.077 0.505 0.004
30 C341 0.564 0.719 0.154 0.535 −0.03
31 C122 0.446 0.4 −0.046 0.476 0.03
32 C132 0.569 0.589 0.02 0.562 −0.007
33 C142 0.594 0.726 0.132 0.592 −0.002
34 C222 0.508 0.4 −0.108 0.489 −0.019
35 C232 0.56 0.58 0.02 0.58 0.02
36 C242 0.592 0.72 0.128 0.61 0.018
37 C322 0.494 0.4 −0.094 0.493 −0.001
38 C332 0.585 0.578 −0.007 0.584 −0.001
39 C342 0.592 0.719 0.127 0.614 0.022
40 D121-L 0.52 0.666 0.146 0.605 0.084
41 D131-L 0.61 0.833 0.224 0.64 0.031
42 D141-L 0.678 0.889 0.211 0.652 −0.026
43 D122-L 0.723 0.667 −0.056 0.703 −0.02
44 D132-L 0.79 0.833 0.043 0.74 −0.05
45 D142-L 0.846 0.889 0.043 0.752 −0.094
46 E131-L 0.601 0.854 0.253 0.621 0.021
47 E141-L 0.677 0.902 0.226 0.632 −0.044
48 E122-L 0.608 0.707 0.099 0.67 0.062
49 E132-L 0.69 0.853 0.164 0.702 0.012
50 E142-L 0.715 0.902 0.187 0.712 −0.003
51 D121 0.459 0.4 −0.059 0.473 0.014
52 D131 0.521 0.593 0.072 0.562 0.041
53 D141 0.591 0.729 0.138 0.589 −0.002
54 E121 0.391 0.4 0.009 0.347 −0.043
55 E131 0.45 0.453 0.003 0.469 0.019
56 E141 0.517 0.637 0.119 0.502 −0.016
57 F121 0.36 0.4 0.04 0.375 0.014
58 F131 0.425 0.545 0.12 0.473 0.049
59 F141 0.477 0.697 0.22 0.505 0.028
60 F122 0.506 0.4 −0.106 0.477 −0.029
61 F132 0.541 0.544 0.002 0.576 0.035
 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583
Table 5 (continued)
Test Number Specimen Experimental Uexp Today UAISI
62 F142 0.610 0.696
Maximum residual error
Minimum residual error
Mean of residual error
Standard deviations of residual error
normal, that is, N (0, σ 2 ) with standard deviation σ 2 . The tests
of the specimens with just one bolt (a particular case allowed
in [5]) are not considered. Therefore, only 62 data on specimen
from Table 4 are analysed. The statistical study of these 62 tests
in Table 4 considered the parameters noticed as significant to
explain all the test results. Special attention must be paid to the
information condensed by the identifiers.
The difference between Uexp value and the predicted U
is the residual error. If there is no relationship between the
independent variable X ik and the predicted variable U , the
residual variability is 1. In the case in which the independent
variables X ik are perfectly related and the variable is Ui , the
residual variability is zero. In the real world, the variability
rate is between zero and one. Lower variability rate of
the residual error means better estimation of the dependent
variable (U ). The statistical analysis undertaken with the
software Statistica [15] allowed us to identify that the following
parameters play a significant role in the computation of the
experimental efficiency factor (Uexp ).
• Connection eccentricity (x̄)
• connection length (L)
• width of the angle connected leg (bc )
• net width of the angle connected leg (bcn ) which is bc minus
the hole diameters at the cross-section of the angle
• width of the nonconnected leg (bd )
• nominal bolt diameter (d)
• angle thickness (t).
Cases in which the difference between estimated and
experimental values is over two-times the standard deviation
of the whole set of experiments were excluded. In such cases,
problems with testing equipment grips and/or data acquisition
were observed. The elimination of such values is a normal
procedure in statistics to account for roughly 95% of the values
in the set. This procedure is also suggested by the software
used for the statistical analysis. By employing the least square
root fitting, the following equation for the computation of the
efficiency factor UUnB is proposed:
x̄
UUnB = 1.19 − 0.26 − (0.63bcn + 0.17bd
L
− 0.47d − 1.70t)/bc .
The statistical parameters obtained from Statistica show that
the proposed Eq. (6) explains 91.9% (multiple R 2 = 0.919) of
the residual variability in the predicted values and that all the
independent variables (parameters) selected for the regression
are statistically significant. Eqs. (2) and (6) are applied to
581
Difference (UAISI − Uexp ) Proposed UUnB Difference (UUnB − Uexp )
0.087 0.607 −0.003
0.253 0.092
−0.232 −0.094
0.061 0.001
0.107 0.036
the specimens tested at University of Brası́lia [12] (Table 4)
and the results reported in Table 5. It is noticed that Eq. (6)
results are compatible with the test results. Among the 62 cases
considered [12] in Table 5, it is evident that the difference
between UUnB and Uexp is in general much less than UAISI
and Uexp . In fact at the end of Table 5, this conclusion can
be perceived in relation to (a) the maximum and the minimum
residual errors, (b) the residual error mean and (c) the standard
deviation of the residual error. A similar performance is also
obtained when applying Eqs. (2) and (6) to the experimental
data from References [8,11], reported in Table 6. Once more,
the proposed formula in Eq. (6) traces a better performance
than Eq. (2). In the last four lines of Table 6, it is possible to
compare the values of the maximum, the minimum, the mean
and the standard deviation of the residual errors, when Eqs. (2)
and (6) are compared to the Uexp values.
7. Conclusions
In this study, the results of 66 experimental tests at the
University of Brası́lia on cold-formed steel angles with bolted
connections were presented. These 66 specimens showed net-
section failure with two or more bolts in the cross-section
of cold-formed angles. The specimens tested have equal or
different legs, different cross-sections, various thicknesses
and a varied number of bolts and bolt lines. By using
multiple regression, the parameters (x̄/L , bcn /bc , d/bc , t/bc
and bd /bc ) were significant in explaining the reduction of the
net-section capacity, and, therefore, were selected for a new
proposed alternative equation in order to compute the reduction
coefficient “U ” due to shear lag. The proposed equation is
UUnB = 1.19 − 0.26 (x̄/L) − (0.63bcn + 0, 17bd
− 0.47d − 1.70t)/bc .
This equation depends on the following geometrical charac-
teristics: (1) connection eccentricity (x̄); (2) connection length
(L); (3) width of the angle connected leg (bc ); (4) net width of
the angle connected leg (bcn ); (5) width of the nonconnected leg
(bd ); (6) nominal bolt diameter (d); and (7) angle thickness (t).
The equation proposed matches the results from the lab tests
(6)
carried out at the University of Brası́lia, in Brazil, as well as the
results available in References [8,11] from tests performed at
the University of Missouri-Rolla (USA) and the University of
Alberta (Canada). A total of 104 (62 + 42) experiments were
checked based upon the formulae proposed. The estimated val-
ues for the reduction coefficient derived from the same formulae
582 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583

Table 6
Performance of the Efficiency Factor (U ) given by Eqs. (2) and (6) applied to the experimental tests available in References [8,11]

Test number Specimen Experimental Uexp Today UAISI Difference (UAISI − Uexp ) Proposed UUnB Difference (UUnB − Uexp )
1 LBN11-1 0.577 0.659 0.082 0.698 0.121
2 LBN11-2 0.592 0.659 0.068 0.698 0.107
3 LBN11-3 0.581 0.659 0.079 0.698 0.118
4 LCN11-1 0.716 0.830 0.114 0.735 0.019
5 LCN11-2 0.731 0.830 0.099 0.735 0.005
6 LCN11-3 0.763 0.830 0.066 0.735 −0.028
7 LBN12-1 0.404 0.400 −0.004 0.411 0.007
8 LBN12-2 0.436 0.400 −0.036 0.411 −0.025
9 LBN12-3 0.411 0.400 −0.011 0.411 0.000
10 LCN12-1 0.494 0.558 0.064 0.507 0.013
11 LCN12-2 0.515 0.558 0.043 0.507 −0.008
12 LBN13-1 0.571 0.769 0.198 0.616 0.045
13 LBN13-2 0.551 0.769 0.218 0.616 0.065
14 LCN13-1 0.673 0.884 0.212 0.641 −0.031
15 LCN13-2 0.716 0.884 0.169 0.641 −0.074
16 LBN31-1 0.693 0.630 −0.063 0.774 0.081
17 LBN31-2 0.683 0.630 −0.053 0.774 0.091
18 LCN31-1 0.827 0.815 −0.012 0.814 −0.014
19 LCN31-2 0.802 0.815 0.013 0.814 0.012
20 LBN32-1 0.446 0.400 −0.046 0.485 0.040
21 LBN32-2 0.480 0.400 −0.080 0.485 0.005
22 LCN32-1 0.539 0.541 0.002 0.585 0.046
23 LCN32-2 0.516 0.541 0.025 0.585 0.069
24 LBN33-1 0.693 0.742 0.049 0.651 −0.042
25 LBN33-2 0.682 0.742 0.060 0.651 −0.031
26 LCN33-1 0.757 0.871 0.114 0.679 −0.078
27 LCN33-2 0.779 0.871 0.092 0.679 −0.100
28 12.2 0.553 0.664 0.111 0.568 0.014
29 12.3 0.630 0.832 0.202 0.604 −0.026
30 12.4 0.645 0.832 0.187 0.604 −0.041
31 14.2 0.710 0.743 0.033 0.742 0.032
32 14.3 0.855 0.871 0.016 0.770 −0.085
33 16.2 0.708 0.678 −0.030 0.755 0.046
34 16.3 0.852 0.839 −0.013 0.790 −0.062
35 A2-2 0.916 0.749 −0.167 0.788 −0.128
36 A2-2N 0.787 0.749 −0.039 0.787 −0.001
37 A2-3 1.020 0.874 0.814
38 A3-2 0.660 0.630 −0.030 0.613 −0.048
39 A3-3 0.759 0.815 0.056 0.654 −0.106
40 A4-2 0.488 0.508 0.019 0.509 0.021
41 A4-3 0.648 0.754 0.106 0.562 −0.086
42 A4-4 0.710 0.836 0.126 0.580 −0.130

Maximum residual error 0.218 0.121

Minimum residual error −0.167 −0.130
Mean of Residual Error 0.050 −0.005
Standard deviations of residual error 0.087 0.065

reflected a good performance vis à vis the performance of the
equation currently in use.
References
[1] Chung KF, Lau L. Experimental investigation on bolted moment
connections among cold formed steel members. Engineering Structures
1999;21:898–911.
[2] Yu WW. Cold-formed steel design. 3rd ed. New York: John Wiley & Sons;
2000.
[3] AISC. LRFD specification for structural steel buildings. Chicago (IL).
2005.
[4] AISI. North American specification for the design of cold-formed steel
structural members. Washington (DC). 2001.
[5] ABNT. NBR 14.762 – Dimensionamento de estruturas de aço constituı́das
por perfis formados a frio – Procedimentos. Rio de Janeiro. 2001 [in
Portuguese].
[6] Gaylord Jr EH. Design of steel structures. New York: McGraw-Hill; 1991.
[7] Laboube RA, Yu WW. Tensile and bearing capacities of bolted
connections. Final summary report. Civil Engineering Study 95-6.
University of Missouri-Rolla; 1995.
[8] Yip ASM, Cheng JJR. Shear lag in bolted cold-formed steel angles and
channels in tension. Structural engineering report no. 233. Edmonton
(Canada): University of Alberta; 2000.
[9] Maiola CH. Ligações parafusadas em chapas finas e perfis de aço
formados a frio. Tese (doutorado). EESC – USP, São Carlos, SP. Brasil.
2004 [in Portuguese].
[10] Kulak GL, Wu EY. Shear lag in bolted angle tension members. Journal of
Structural Engineering, ASCE 1997;123:1144–52.
 V.F. de Paula et al. / Journal of Constructional Steel Research 64 (2008) 571–583 583

[11] Holcomb BD, Yu WW, Laboube RA. Tensile and bearing capacities of
bolted connections. Second summary report. Civil Engineering Study 95-
1. University of Missouri-Rolla; 1995.
[12] De Paula VF. Análise Experimental e Numérica de Cantoneiras de
Aço Formadas a Frio, sob Tração e Conectadas por Parafusos. Tese
(doutorado). DF (Brasil): Universidade de Brası́lia – UnB; 2006 [in
Portuguese].
[13] Kennedy JB, Neville AM. Basic statistical methods for engineering and
scientists. New York: Harper & Row Publishers; 1986.
[14] Spiegel MR, Schiller JJ, Srinivasan RA. Schaum’s outline of theory and
problems of probability and Statistics. 2nd ed. New York: McGraw-Hill;
2000.
[15] STATSOFT, Inc. STATISTICA (data analysis software system). version
6. 2001.