Design of Cold-Formed Steel Built-Up Closed Sections with

Intermediate Stiffeners
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.

Ben Young, M.ASCE1; and Ju Chen2

Abstract: A series of column tests on cold-formed steel built-up closed sections with intermediate stiffeners is presented in this paper.
The test specimens were first brake pressed from structural steel sheets to form open sections with intermediate web stiffeners, then two
of the open sections were connected at their flanges using self-tapping screws to form the built-up closed sections. The high strength
structural steel sheets had the measured 0.2% proof stress up to 586 MPa. Initial and overall geometric imperfections as well as material
properties and residual stresses of the test specimens were measured. Tests were performed over a range of lengths such that column
curves could be obtained. The test strengths are compared with the design strengths calculated using the direct strength method in the
North American Specification and Australian/New Zealand Standard for cold-formed steel structures. Three different methods were used
to obtain the local and distortional buckling stresses for the calculation of the direct strength method. Reliability analysis was performed
to assess the reliability of the direct strength method on cold-formed steel built-up closed section columns. It is shown that the direct
strength method using single section to obtain the buckling stresses is generally conservative.
DOI: 10.1061/共ASCE兲0733-9445共2008兲134:5共727兲
CE Database subject headings: Cold-formed steel; Columns; Buckling; Experimentation; Structural design; Thin wall structures;
Steel structures.

Introduction hanced by adding intermediate web stiffeners. In this study, the

The use of cold-formed steel structural members has increased in
recent years, especially in light-weight steel construction, such as
steel-framed housing, low-rise office buildings, factories, and
warehouses. Cold-formed steel members are either cold rolled or
brake pressed into structural shapes. As a result, cold-formed steel
sections are usually formed in singly-, point-, or nonsymmetric
open shapes. These open sections have a relatively small torsional
rigidity that is weak in twisting compared to closed sections.
Therefore, open sections would likely fail by twisting in addition
to other buckling modes depending on the dimension of the cross
sections and the length of the members. Closed sections such as
box-shaped sections made by connecting two channel sections tip
to tip are often found in use in cold-formed steel structures due to
their relatively large torsional rigidity and their favorable radius
of gyration about both principal axes 共Yu 2000兲. When the width-
to-thickness ratio of a stiffened compression element is relatively
large, local buckling will reduce the full strength of the member
共Hancock 1998兲. However, local buckling stress could be en-
1 cold-formed steel built-up closed sections with intermediate stiff-
Associate Professor, Dept. of Civil Engineering, Univ. of Hong
Kong, Pokfulam Rd., Hong Kong 共corresponding author兲. E-mail:
young@hku.hk
2
Lecturer, Dept. of Civil Engineering, Zhejiang Univ., Hangzhou,
China; formerly, Dept. of Civil Engineering, Univ. of Hong Kong,
Pokfulam Rd., Hong Kong.
Note. Associate Editor: Benjamin W. Schafer. Discussion open until
October 1, 2008. Separate discussions must be submitted for individual
papers. To extend the closing date by one month, a written request must
be filed with the ASCE Managing Editor. The manuscript for this paper
was submitted for review and possible publication on December 29,
2006; approved on November 5, 2007. This paper is part of the Journal
of Structural Engineering, Vol. 134, No. 5, May 1, 2008. ©ASCE, ISSN
0733-9445/2008/5-727–737/$25.00.
cold-formed steel built-up closed sections with web stiffeners are
investigated, as shown in Fig. 1. The section could be failed in
distortional buckling of the webs in addition to flexural buckling
for long columns, as shown in Fig. 2.
In the past decades, there were many test data on cold-formed
steel open section columns performed by researchers all over the
world, such as Klöppel and Bilstein 共1976兲, Rhodes and Harvey
共1977兲, Thomasson 共1978兲, Rhodes and Loughlan 共1980兲, Mulli-
gan and Peköz 共1984兲, Batista et al. 共1987兲, Lau and Hancock
共1988兲, Weng and Peköz 共1990兲, Kwon and Hancock 共1992兲,
Young and Rasmussen 共1998兲, Yan and Young 共2002兲, Young
共2005兲, and some other researchers as summarized by Yu 共2000兲.
However, not many test data have been reported on cold-formed
steel built-up closed section columns. De Wolf et al. 共1974兲 con-
ducted column tests on cold-formed steel box-shaped sections
built up by two plain channel sections connected at their flanges.
The webs of the box-shaped sections were flat and local buckling
occurred during the tests. The column strengths were influenced
by local buckling. However, the use of intermediate stiffeners
could improve the situation. Therefore, the behavior and design of
eners are investigated in this study.
The North American Specification 共NAS 2001, 2004兲 and
Australian/New Zealand Standard 共AS/NZS 2005兲 for cold-
formed steel structures have design provisions for intermediate
web stiffeners using effective width approach; however, the de-
sign provisions are not applicable to the sections that were tested
in this study. The advantage of using the direct strength method is
that it can accommodate for unusual shape sections. However, the
direct strength method developed by Schafer and Peköz 共1998兲
and Schafer 共2002兲 was based on open sections, such as the
simple lipped channel, lipped channel with web stiffeners,
Z-section, hat section, and rack upright section. Therefore, there

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008 / 727

J. Struct. Eng., 2008, 134(5): 727-737

Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.
Fig. 1. Cross section of cold-formed steel member
is a need to investigate the appropriateness of the direct strength
method on the type of cold-formed steel built-up closed sections
with intermediate web stiffeners, as shown in Fig. 1.
The purpose of this paper is first to present a series of column
tests on the cold-formed steel built-up closed sections with inter-
mediate web stiffeners, as shown in Fig. 1. Second, the test
strengths are compared with the design strengths obtained using
the direct strength method in the North American Specification
共NAS 2001, 2004兲 and Australian/New Zealand Standard 共AS/
NZS 2005兲 for cold-formed steel structures. In the calculation of
the direct strength method, local buckling stress and distortional
buckling stress are required. In this study, three different methods
were used to obtain these stresses. The appropriateness of the
direct strength method on the cold-formed steel built-up closed
sections with intermediate web stiffeners is investigated.
Experimental Investigation
Test Specimens
Finite strip buckling analysis 共Papangelis and Hancock 1995兲 was
used in designing the test specimens. The test specimens of
built-up closed sections were first brake pressed from structural
steel sheets to form open sections with intermediate web stiffen-
ers, then two of the open sections were connected at their flanges
using self-tapping screws to form the built-up closed sections, as
shown in Fig. 1. The high strength zinc-coated grades G550
共1.0 mm兲 and G450 共1.5 mm and 1.9 mm兲 structural steel sheets
with nominal yield stresses of 550 and 450 MPa, respectively,
Fig. 2. Distortional buckling of the webs and flexural buckling
modes
728 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008
J. Struct. Eng., 2008, 134(5): 727-737
Fig. 3. Arrangement of screw spacing
were used for the test specimens. These structural steel sheets
conformed to the Australian Standard AS 1397 共AS 1993兲. Each
specimen was cut to a specified length of 300, 900, 1,500, 2,200,
and 3,000 mm, and both ends were milled flat by an electronic
milling machine. The ends of the specimen were then welded to
25 mm thick steel plates to ensure full contact between specimen
and end bearings. The screws were positioned 20 mm away from
the steel end plates and the longitudinal spacing of the screws is
approximately 100 mm apart, as shown in Fig. 3. The diameter
and length of the screws were 4.8 and 13 mm, respectively.
Three series of cold-formed steel built-up sections were tested,
each having the nominal plate thickness 共t兲 of 1.0, 1.5, and
1.9 mm. The three series are labeled T1.0, T1.5, and T1.9 accord-
ing to their nominal thickness. The base metal thickness 共t*兲 was
measured by removing the zinc coating by acid etching. The mea-
sured inside corner radius 共ri兲 at the flanges was 2.6 mm for all
specimens. The nominal widths of the web 共bw兲 and flange 共b f 兲
were 160 and 45 mm, respectively. The nominal dimensions of
the intermediate web stiffeners were w1 = 40 mm, w2 = 21 mm,
and w3 = 50 mm. The nominal depth of the web stiffeners 共d兲 was
15 mm. The web stiffener element w2 was inclined at 45 deg, as
shown in Fig. 1. Tables 1–3 show the measured cross-sectional
dimensions of the test specimens for Series T1.0, T1.5, and T1.9,
respectively, using the nomenclature defined in Fig. 1. Each
built-up closed section was formed by two open sections 共parts
“a” and “b”兲. The measured column length 共L兲 for each specimen
is also shown in Tables 1–3. The columns were compressed be-
tween fixed ends at various lengths from short column to long
column. The longest specimen length of 3,000 mm produced
ley / ry ratio of 86 for all column Series T1.0, T1.5, and T1.9,
where ley⫽effective length for buckling about the minor y-axis;
and ry⫽radius of gyration about the y-axis.
The test specimens are separated into three series according to
their plate thickness. The specimens are labeled such that the test
series and specimen length could be identified from the label. For
example, the labels “T1.0L2200” and “T1.5L300R” define the
specimens as follows:
• The first three letters indicate that the specimens belonged to
test Series T1.0 and T1.5, where the prefix letter “T” refers to
plate thickness of the sections, “1.0” and “1.5” refer to the
nominal plate thickness.
• The fourth letter “L” indicates the length of the specimen.
• The last three or four digits are the nominal column length of
the specimen in mm 共2,200 and 300 mm兲.
• If a test was repeated, then the letter “R” indicates the repeated
test.
 Table 1. Measured Specimen Dimensions for Column Series T1.0
Web Thickness
Length Flange Radius
L bw w1 w2 w3 d bf t t* ri
Specimen Part 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲
T1.0L300 a 298.2 162.0 40.9 21.3 50.2 15.0 45.6 1.02 0.98 2.6
b 298.2 163.1 41.3 21.3 50.4 15.0 45.6 1.01 0.97 2.6
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.

T1.0L900 a 898.1 160.8 40.9 21.0 49.9 15.1 46.1 1.02 0.98 2.6
b 898.1 161.4 40.9 21.1 50.2 15.0 46.1 1.02 0.98 2.6
T1.0L1500 a 1,497.2 162.2 40.7 21.4 50.2 14.9 45.4 1.01 0.97 2.6
b 1,497.2 162.2 40.8 21.4 50.2 15.0 45.4 1.01 0.97 2.6
T1.0L2200 a 2,196.9 160.0 40.3 21.2 49.6 15.1 45.7 1.01 0.97 2.6
b 2,196.9 162.4 40.8 21.3 50.6 15.0 45.7 1.01 0.97 2.6
T1.0L3000 a 2,996.8 162.7 41.2 21.3 50.2 15.0 45.5 1.01 0.97 2.6
b 2,996.8 161.9 40.8 21.3 50.2 15.0 45.5 1.01 0.97 2.6
Mean 161.9 40.8 21.2 50.2 15.0 45.7 1.01 0.97 2.6
COV 0.006 0.006 0.006 0.005 0.004 0.006 0.005 0.005 0.000
Note: 1 in.= 25.4 mm; COV⫽coefficient of variation; *base metal thickness.

Tensile Coupon Tests
Tensile coupon tests were conducted to obtain the material prop-
erties of the test specimens. The coupons were taken from the
center of the web plate, which is the center of element w3 as
shown in Fig. 1, in the longitudinal direction of the finished speci-
mens belonging to the same batch as the column test specimens.
The coupon dimensions conformed to the Australian Standard AS
1391 共AS 1991兲 for the tensile testing of metals using 6 mm wide
coupons. The coupons were also tested in accordance with the AS
1391 共AS 1991兲 in a displacement controlled testing machine
using friction grips. A calibrated extensometer of 25 mm gauge
length was used to measure the longitudinal strain. A data acqui-
sition system was used to record the load and the readings of
strain at regular intervals during the tests. The static load was
obtained by pausing the applied straining for 1.5 min near the
0.2% tensile proof stress 共␴0.2兲 and the ultimate tensile strength
共␴u兲. This allowed the stress relaxation associated with plastic
straining to take place. The values of Young’s modulus 共E兲, static
0.2% proof stress 共␴0.2兲, static tensile strength 共␴u兲, and elonga-
tion after fracture 共␧ f 兲 based on a gauge length of 25 mm are
shown in Table 4. The 0.2% proof stress was used as the corre-
sponding yield stress in calculating the design strength of the
columns.
Residual Stress Measurements
The membrane and bending residual stress measurements were
conducted on half of the cold-formed steel built-up closed section
of plate thickness 1.9 mm for the column Series T1.9. The
longitudinal residual strains were measured by the method of
sectioning, and the strains were converted to residual stresses. The
cold-formed steel specimen was cut into strips of 8 mm width,
and the length of the specimen was 300 mm. The residual strains
were measured using a Cambridge in situ electrical demountable
extensometer as described by Denston and White 共1977兲. A gauge
length of 100 mm was marked on the outer and inner surfaces of

Table 2. Measured Specimen Dimensions for Column Series T1.5

Web Thickness
Length Flange Radius
L bw w1 w2 w3 d bf t t* ri
Specimen Part 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲
T1.5L300 a 298.0 161.3 41.1 21.2 49.8 15.1 45.6 1.52 1.48 2.6
b 298.0 162.1 41.5 21.1 50.0 15.1 45.6 1.52 1.48 2.6
T1.5L300R a 298.1 162.0 40.9 21.5 49.8 15.1 44.9 1.52 1.48 2.6
b 298.1 162.7 40.8 21.5 50.0 15.1 45.3 1.52 1.48 2.6
T1.5L900 a 898.0 161.6 41.1 21.1 50.1 15.0 45.2 1.53 1.49 2.6
b 898.0 163.6 41.2 21.6 49.9 15.0 45.4 1.52 1.48 2.6
T1.5L1500 a 1,497.1 162.1 41.0 21.4 49.9 14.9 45.5 1.53 1.49 2.6
b 1,497.1 162.3 41.2 21.5 50.0 14.9 45.4 1.52 1.48 2.6
T1.5L2200 a 2,197.3 162.5 40.9 21.5 50.5 15.0 45.7 1.52 1.48 2.6
b 2,197.3 162.4 40.5 21.4 50.4 15.0 45.7 1.52 1.48 2.6
T1.5L3000 a 2,997.0 161.9 41.3 21.5 49.5 15.0 45.7 1.52 1.48 2.6
b 2,997.0 162.5 41.2 21.5 50.1 15.0 46.7 1.52 1.48 2.6
Mean 162.3 41.1 21.4 50.0 15.0 45.6 1.52 1.48 2.6
COV 0.004 0.006 0.008 0.005 0.005 0.009 0.003 0.003 0.000
Note: 1 in.= 25.4 mm; COV⫽coefficient of variation; *base metal thickness.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008 / 729

J. Struct. Eng., 2008, 134(5): 727-737

 Table 3. Measured Specimen Dimensions for Column Series T1.9
Web
Length Flange Thickness Radius
L bw w1 w2 w3 d bf t t* ri
Specimen Part 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲 共mm兲
T1.9L300 a 298.0 162.6 40.9 21.6 50.2 15.2 45.6 1.93 1.88 2.6
b 298.0 160.8 40.4 21.4 49.9 15.2 45.6 1.93 1.88 2.6
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.

T1.9L300R a 298.0 159.9 40.3 21.4 49.5 15.3 45.8 1.93 1.88 2.6
b 298.0 161.2 40.4 21.5 50.2 15.3 45.8 1.93 1.88 2.6
T1.9L900 a 898.2 160.6 40.7 21.3 49.0 15.0 45.4 1.93 1.88 2.6
b 898.2 161.5 40.6 21.6 49.4 15.0 45.4 1.93 1.88 2.6
T1.9L1500 a 1,496.8 163.4 41.1 21.5 50.5 15.0 45.6 1.93 1.88 2.6
b 1,496.8 163.4 41.1 21.6 50.5 15.1 45.6 1.93 1.88 2.6
T1.9L2200 a 2,197.1 163.0 41.0 21.5 50.3 15.0 45.7 1.93 1.88 2.6
b 2,197.1 161.4 40.5 21.4 50.2 15.1 45.7 1.93 1.88 2.6
T1.9L3000 a 2,997.2 161.1 40.6 21.5 49.2 15.0 45.7 1.93 1.88 2.6
b 2,997.2 160.6 40.4 21.4 49.4 15.0 45.8 1.93 1.88 2.6
Mean 161.6 40.7 21.5 49.9 15.1 45.6 1.93 1.88 2.6
COV 0.007 0.007 0.004 0.011 0.008 0.003 0.000 0.000 0.000
Note: 1 in.= 25.4 mm and COV⫽coefficient of variation; *base metal thickness.

each strip. The residual strains were measured for both outer and Geometric Imperfection Measurements
inner surfaces of each strip. The initial readings before cutting
The initial local and overall geometric imperfections of the test
were recorded for each strip together with the corresponding
specimens were measured prior to testing. Initial local geometric
temperature. The specimen was then cut into strips using an AGIE
imperfections were measured on the three short column Speci-
wire cut machine with an accuracy of 0.005 mm, and a wire diam
mens T1.0L300, T1.5L300, and T1.9L300. The imperfections
was 0.25 mm. The specimen was cut by a wire-cutting method
were measured on all faces of the sections, as shown in Fig. 5,
under water to eliminate the additional stresses resulting from the
and the measured local imperfection profiles are shown in Figs.
cutting process. The readings were taken after cut and the
6–8. The measurements were taken at midlength of the speci-
corresponding temperature was also recorded. The readings were
mens, which is approximately 40 mm away from the screw con-
corrected for temperature difference before and after cutting.
nections. A Mitutoyo coordinate measuring machine with an
The measured strains were converted to residual stresses by
multiplying the measured Young’s modulus obtained from the
tensile coupon tests. The membrane and bending residual stresses
were calculated as the average and the difference in residual stress
measurements at the two surfaces, respectively. Two sets of
measurements were recorded, and the average values were used to
plot the membrane and bending residual stress distributions in the
cold-formed steel section, as shown in Fig. 4. The negative value
indicates a compressive membrane stress, and the negative value
of bending stress indicates higher tensile 共or lower compressive兲
residual strain at the inner surface of the cross section. The
maximum values of the membrane and bending residual stresses
were −99 and −101 MPa at the corners of the web-flange
junctions, which are 19.2 and 19.6% of the measured 0.2% tensile
proof stress 共␴0.2兲, respectively.

Fig. 4. Measured membrane and bending residual stresses

distributions in half of the section for column Series T1.9
Table 4. Nominal and Measured Material Properties Obtained from
Tensile Coupon Tests
Nominal Measured
␴0.2 E ␴0.2 ␴u ␧f
Test series 共MPa兲 共GPa兲 共MPa兲 共MPa兲 共%兲
T1.0 550 200 586 596 10
T1.5 450 202 514 539 11
T1.9 450 203 515 540 11
Note: 1 ksi= 6.89 MPa. Fig. 5. Location of local geometric imperfection measurements

730 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008

J. Struct. Eng., 2008, 134(5): 727-737


Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.
Fig. 6. Measured local geometric imperfection profiles of Specimen
T1.0L300
accuracy of 0.001 mm was used to measure the initial local geo-
metric imperfections. Readings were taken at approximately
2 mm intervals across the plate widths. The negative values of
local imperfection measurements indicated concave profiles and
the positive values indicated convex profiles. The maximum mea-
sured local geometric imperfections were 0.162, 0.178, and
0.071 mm, which are 16.2, 11.9, and 3.7% of the plate thickness
for Specimens T1.0L3000, T1.5L3000, and T1.9L3000, respec-
tively. The initial overall geometric imperfections were measured
along the flange junction of the specimens. Theodolites were used
to obtain readings at midlength and near both ends of the speci-
mens. The measured overall geometric imperfections at midlength
Fig. 7. Measured local geometric imperfection profiles of Specimen
T1.5L300
Fig. 8. Measured local geometric imperfection profiles of Specimen
T1.9L300
JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008 / 731
J. Struct. Eng., 2008, 134(5): 727-737
Table 5. Measured Overall Geometric Imperfections at Midlength
Specimen ␦/L
T1.0L900 1/9,000
T1.0L1500 1/15,000
T1.0L2200 1/7,333
T1.0L3000 1/4,286
T1.5L900 1/15,000
T1.5L1500 1/5,000
T1.5L2200 1/2,750
T1.5L3000 1/3,000
T1.9L900 1/3,273
T1.9L1500 1/5,000
T1.9L2200 1/5,500
T1.9L3000 1/4,000
共␦兲 over the specimen length 共L兲 are shown in Table 5. The maxi-
mum minor axis flexural imperfections at midlength were 1/4286,
1/2750, and 1/3273 of the specimen length for column Series
T1.0, T1.5, and T1.9, respectively.
Test Rig and Operation
The test rig and the test setup of the column tests are shown in
Figs. 9–11. A DARTEC servocontrolled hydraulic testing ma-
chine was used to apply compressive axial force to the column
specimens. Two 25 mm thick steel end plates were welded to the
ends of the specimen. A movable upper end support allowed the
tests to be conducted at various column lengths. A rigid flat bear-
ing plate was connected to the upper end support, and the top end
plate of the specimen was bolted to the rigid flat bearing plate,
which was restrained against the minor and major axis rotations
as well as twist rotations and warping. The bottom end plate of
the specimen was bolted to a fixed-ended bearing. The fixed-
ended bearing was considered to restrain both minor and major
axis rotations as well as twist rotations and warping. All columns
were tested between the top and bottom fixed-ended bearings.
Three displacement transducers were positioned on the top end
plate of the specimen to measure the axial shortening of the col-
umn. In addition, four linear variable displacement transducers
共LVDT兲 were also used to measure the deformation of the column
at midlength, and the location of the transducers is shown in
Fig. 12. Displacement control was used to drive the hydraulic
actuator at a constant speed of 0.2 mm/ min for all test specimens.
The use of displacement control allowed the tests to be continued
into the postultimate range. The static load was recorded by paus-
ing the applied straining for 1.5 min near the ultimate load of all
the specimens. A data acquisition system was used to record the
applied load and the readings of the transducers at regular inter-
vals during the tests.
Column Test Results
The test ultimate loads 共PTEST兲 and failure modes of the columns
are shown in Table 6. Two tests were repeated and the test results
for the repeated tests are very close to their first test values, with
differences ranging from 0.1 to 0.4%. The small differences be-
tween the repeated test values and their first test values demon-
strated the reliability of the test results. The failure modes at the
ultimate load of the columns involved local buckling 共L兲, distor-
tional buckling of the webs 共D兲, and flexural buckling 共F兲. The
failure modes were determined by the deformed test specimens at
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.
Fig. 9. Column test of Specimen T1.0L1500
ultimate load. The deformation of the Specimen T1.0L300 at
midlength of the column is shown in Fig. 13. The location of the
LVDT transducers is shown in Fig. 12.
Design Rules
General
In this study, the type of intermediate web stiffeners as shown in
Fig. 1 are not really covered by the North American Specification
共NAS 2001, 2004兲 and Australian/New Zealand Standard 共AS/
NZS 2005兲 for cold-formed steel structures using effective width
method. However, the direct strength method in the NAS Speci-
fication 共NAS 2004兲 and AS/NZS Standard 共AS/NZS 2005兲 can
be used to calculate the column strength. As mentioned in the
Introduction, the direct strength method was developed based on
open sections rather than built-up closed sections. Hence, the ap-
732 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008
J. Struct. Eng., 2008, 134(5): 727-737
Fig. 10. Column test of Specimen T1.9L300
propriateness of the direct strength method on the cold-formed
steel built-up closed sections with intermediate web stiffeners is
investigated.
In the calculation of the column strengths, the effective length
for minor axis flexural buckling 共ley兲 was assumed equal to one-
half of the column length 共L兲 for the fixed-ended columns. The
measured specimen dimensions and material properties were used
to calculate the column strengths. The base metal thickness was
used in the calculation.
Direct Strength Method
The direct strength method in the Australian/New Zealand Stan-
dard 共AS/NZS 2005兲 for cold-formed steel structures was adopted
from the North American Specification 共NAS 2004兲 for the design
of cold-formed steel structural members. The design rules of the
direct strength method in the AS/NZS Standard 共AS/NZS 2005兲
are identical to those in the NAS Specification 共NAS 2004兲.
Hence, the design column strengths obtained from the AS/NZS
Standard are identical to those obtained from the NAS Specifica-
tion. The design rules of the direct strength method for compres-
sion members in the NAS Specification 共NAS 2004兲 and AS/NZS
Standard 共AS/NZS 2005兲 are summarized in this paper.
The nominal axial strength or unfactored design strength
共PDSM兲 is the minimum of the nominal axial strengths for flexural
buckling 共Pne兲, local buckling 共Pnl兲, and distortional buckling
共Pnd兲, as shown in Eq. 共1兲
PDSM = min共Pne, Pnl, Pnd兲 共1兲
The nominal axial strength 共Pne兲 for flexural buckling is
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Location of LVDT transducers

in flexural buckling in this study, since the doubly symmetric

built-up closed sections only failed by flexural buckling. The tor-
sional buckling or flexural-torsional buckling is not the governing
buckling mode for the doubly symmetric built-up closed sections.
The critical elastic column buckling load 共Pcre兲 can be computed
in accordance with Section C4.1 of the NAS Specification 共NAS
2001兲 and Section 3.4.2 of the AS/NZS Standard 共AS/NZS 2005兲
using the elastic flexural buckling stress multiplied by the gross
cross-sectional area. The modified slenderness approach in Sec-
tion C4.5 of the NAS Specification 共NAS 2001兲 was used to
calculate the critical elastic column buckling load for the built-up
compression members. The nominal axial strength 共Pnl兲 for local
buckling is

冦冋 冣
Pne for ␭l ⱕ 0.776
Pnl =
1 − 0.15 冉 冊 册冉 冊
Pcrl
Pne
0.4
Pcrl
Pne
0.4
Pne for ␭l ⬎ 0.776

共3兲
where ␭l = 冑Pne / Pcrl and Pcrl = Af ol. Pcrl⫽critical elastic local col-
umn buckling load; f ol⫽elastic local buckling stress of the cross
section; and Pne is defined in Eq. 共2兲.
The nominal axial strength 共Pnd兲 for distortional buckling is

冦冋 冣
Py for ␭d ⱕ 0.561
Pnd =
1 − 0.25 冉 冊 册冉 冊
Pcrd
Py
0.6
Pcrd
Py
0.6
Py for ␭d ⬎ 0.561

Fig. 11. Column test of Specimen T1.0L3000
共4兲
where ␭d = 冑Py / Pcrd and Pcrd = Af od. Pcrd⫽critical elastic distor-
tional column buckling load; f od⫽elastic distortional buckling
stress of the cross section; and Py is defined in Eq. 共2兲.
In calculating the nominal axial strengths for local buckling
共Pnl兲 and distortional buckling 共Pnd兲 as shown in Eqs. 共3兲 and 共4兲,
respectively, the elastic local buckling stress 共f ol兲 and distortional
buckling stress 共f od兲 of the cross section are required. These elas-
tic buckling stresses were obtained from a rational elastic finite
strip buckling analysis 共Papangelis and Hancock 1995兲. In this
study, three different methods were used to obtain the elastic local
and distortional buckling stresses. In the finite strip buckling
analysis, three different sections were considered, namely, the
single section, single section restrained at the flanges, and double
section with the flange thickness two times the web thickness, as
shown in Fig. 14. It should be noted that there is some degree of

冦冉 冣
2
共0 . 658␭c 兲Py for ␭c ⱕ 1.5 restraint in the flanges of the built-up sections. The single section
Pne = 0.877
␭2c
冊 Py for ␭c ⬎ 1.5
共2兲 restrained at the flanges and the double section provided some
degree of restraint in the flanges, but this is not the case for the
single section. The finite strip analysis results are shown in Table
where ␭c = 冑Py / Pcre and Py = Af y. Py⫽squash load; A⫽gross 7. The elastic local and distortional buckling stresses obtained
cross-sectional area; f y⫽yield stress, which is the static 0.2% from the finite strip buckling analysis were multiplied by the
proof stress 共␴0.2兲; and Pcre⫽critical elastic column buckling load gross cross-sectional area of the built-up closed sections.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008 / 733

J. Struct. Eng., 2008, 134(5): 727-737

 Table 6. Comparison of Test Strengths with Design Strengths
Test Design Comparison

PTEST Failure PDSM−S PDSM−SR PDSM−D

Specimen 共kN兲 mode 共kN兲 共kN兲 共kN兲 PTEST / PDSM−S PTEST / PDSM−SR PTEST / PDSM−D
T1.0L300 193.9 L+D 131.6 214.5 214.9 1.47 0.90 0.90
T1.0L900 179.8 L+D 125.5 204.2 204.5 1.43 0.88 0.88
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.

T1.0L1500 132.6 L+D+F 114.1 185.0 185.3 1.16 0.72 0.72

T1.0L2200 113.1 L+D+F 96.2 154.6 154.8 1.18 0.73 0.73
T1.0L3000 90.1 L+D+F 72.8 115.2 115.4 1.24 0.78 0.78
T1.5L300 319.2 L+D 244.8 395.5 396.0 1.30 0.81 0.81
T1.5L300R 320.4 L+D 244.8 395.5 396.0 1.31 0.81 0.81
T1.5L900 313.8 L 234.7 378.1 378.7 1.34 0.83 0.83
T1.5L1500 235.5 L+F 215.7 334.4 334.4 1.09 0.70 0.70
T1.5L2200 212.5 L+F 185.1 264.4 264.4 1.15 0.80 0.80
T1.5L3000 165.8 L+F 144.3 181.3 181.3 1.15 0.91 0.91
T1.9L300 416.2 L 366.0 515.8 515.8 1.14 0.81 0.81
T1.9L300R 415.8 L 366.0 515.8 515.8 1.14 0.81 0.81
T1.9L900 415.8 L 350.8 483.4 483.4 1.19 0.86 0.86
T1.9L1500 332.6 L+F 322.1 424.6 424.6 1.03 0.78 0.78
T1.9L2200 288.7 F 275.9 336.3 336.3 1.05 0.86 0.86
T1.9L3000 201.2 F 214.4 231.2 231.2 0.94 0.87 0.87
Mean, Pm 1.19 0.82 0.82
COV, V P 0.118 0.075 0.075
Reliability index, ␤ 3.38 2.13 2.12
Note: 1 kip= 4.45 kN; L⫽local buckling; D⫽distortional buckling; and F⫽flexural buckling.

Reliability Analysis M m = 1.10, Fm = 1.00, V M = 0.10, and VF = 0.05, which are the mean
values and coefficients of variation for material properties and
A reliability analysis was performed to assess the current direct fabrication factors. The statistical parameters Pm and V P are the
strength method for the cold-formed steel built-up closed sections mean value and coefficient of variation of tested-to-predicted load
with intermediate web stiffeners. The reliability of the design ratios, respectively, as shown in Table 6. A correction factor C P is
rules is measured by a reliability index 共␤兲. A target reliability used in the reliability analysis to account for the influence due to
index of 2.5 for structural members in the NAS Specification
共NAS 2001兲 is recommended as a lower limit The resistance fac-
tor 共␾c兲 of 0.8 was used in the analysis as specified in the NAS
Specification 共NAS 2004兲 and AS/NZS Standard 共AS/NZS 2005兲
for the columns in this study. A load combination of 1.2 DL
+ 1.6 LL as specified in the American Society of Civil Engineers
Standard 共ASCE 2006兲 was used in the reliability analysis, where
DL is the dead load and LL is the live load. The statistical param-
eters are obtained from Table F1 of the NAS Specification 共NAS
2001兲 for concentrically loaded compression members, where

Fig. 14. Different methods of determining buckling stresses

Table 7. Finite Strip Analysis Results

Single section Single restraint section Double section

f ol f od f ol f od f ol f od
Section 共MPa兲 共MPa兲 共MPa兲 共MPa兲 共MPa兲 共MPa兲
T1.0 88 — 342 1,010 344 930
T1.5 189 — 785 1,577 788 1,440
Fig. 13. Deformation measurements of Specimen T1.0L300 T1.9 302 — 1,245 2,028 1,251 1,854

734 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008

J. Struct. Eng., 2008, 134(5): 727-737

Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.
Fig. 15. Column curves for Series T1.0
a small number of tests 共Peköz and Hall 1988兲, and the factor C P
is given in Eq. F1.1-3 of the NAS Specification 共NAS 2001兲.
Comparison of Test Strengths with Design
Strengths
The test strengths 共PTEST兲 of the cold-formed steel built-up closed
section columns are compared with the nominal 共unfactored兲 de-
sign strengths 共PDSM兲 obtained using the direct strength method in
the North American Specification 共NAS 2001, 2004兲 and
Australian/New Zealand Standard 共AS/NZS 2005兲 for cold-
formed steel structures, as shown in Table 6. Three different
sections were considered in the finite strip buckling analysis;
therefore, three different design strengths were calculated for each
column. The design strengths are PDSM−S for buckling stresses
obtained using single section, PDSM−SR for buckling stresses ob-
tained using single restrained section, and PDSM−D for buckling
stresses obtained using double section with the flange thickness
two times the web thickness, as shown in Table 6.
The test strengths are plotted against the effective length for
minor axis flexural buckling 共ley兲 for column Series T1.0, T1.5,
and T1.9, as shown in Figs. 15–17, respectively. The effective
length 共ley兲 was assumed equal to one-half of the column length
共L兲 for the fixed-ended columns 共ley = L / 2兲. The design strengths
are also plotted in Figs. 15–17. It is expected that the values of the
design strength PDSM−SR are similar to those of the PDSM−D, as
shown in Table 6. It is because the flanges of both sections are
restrained, and the elastic local and distortional buckling stresses
having similar values.
Fig. 16. Column curves for Series T1.5
JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008 / 735
J. Struct. Eng., 2008, 134(5): 727-737
Fig. 17. Column curves for Series T1.9
It is shown that the design strengths PDSM−S for buckling
stresses obtained using single section are conservative, except for
test Specimen T1.9L3000. However, the design strengths PDSM−SR
for buckling stresses obtained using single restrained section and
PDSM−D for buckling stresses obtained using double section are
unconservative for all columns, as shown in Figs. 15–17. In Table
6, the mean value of tested-to-predicted load ratios PTEST / PDSM−S,
PTEST / PDSM−SR, and PTEST / PDSM−D are 1.19, 0.82, and 0.82 with
the coefficients of variation 共COV兲 of 0.118, 0.075, and 0.075,
and the corresponding values of reliability index 共␤兲 are 3.38,
2.13, and 2.12, respectively. Hence, the direct strength method
using single section to obtain the buckling stresses is reliable.
Conclusions
A test program on cold-formed steel built-up closed sections with
intermediate stiffeners has been described. The test specimens
were fabricated using high strength zinc-coated grades G550 and
G450 structural steel sheets with nominal yield stresses of 550
and 450 MPa, respectively. Material properties of the test speci-
mens were determined using tensile coupon tests. The membrane
and bending residual stresses of a test specimen was measured,
and the residual stresses distributions in the cold-formed steel
section have been presented. The initial local and overall geomet-
ric imperfections of the test specimens were measured. The mea-
sured local geometric imperfection profiles of the test specimens
have been presented. The fixed-ended columns were tested at
various lengths. The failure modes of the columns involved local
buckling, distortional buckling of the webs, and flexural buckling.
The test strengths were compared with the design strengths
obtained using the direct strength method in the North American
Specification and Australian/New Zealand Standard for cold-
formed steel structures. In this study, three different methods were
used to obtain the elastic local and distortional buckling stresses
of the cross sections as required in the calculation of the direct
strength method. The appropriateness of the direct strength
method on cold-formed steel built-up closed sections with inter-
mediate web stiffeners has been assessed. It is shown that the
direct strength method using single section to obtain the elastic
buckling stresses are generally conservative and reliable.
Acknowledgments
The writers are grateful to BlueScope Lysaght 共Singapore兲 Pte.
Ltd. for supplying the test specimens. The research work de-
 scribed in this paper was supported by a grant from the Research ␧ f ⫽ elongation 共tensile strain兲 after fracture
Grants Council of the Hong Kong Special Administrative Region, based on gauge length of 25 mm;
China 共Project No. HKU6106/04E兲. ␭c, ␭d, ␭l ⫽ nondimensional slenderness used in direct
strength method;
␴0.2 ⫽ static 0.2% tensile proof stress;
Notation ␴u ⫽ static ultimate tensile strength; and
␾c ⫽ resistance 共capacity兲 factor for compression
member.
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.

The following symbols are used in this paper:

A ⫽ gross cross-sectional area;
bf ⫽ width of flange;
bw ⫽ width of web; References
CP ⫽ correction factor in reliability analysis;
d ⫽ depth of web stiffeners; AS/NZS. 共2005兲. “Cold-formed steel structures.” AS/NZS No. 4600:2005,
E ⫽ Young’s modulus; Australian/New Zealand Standard, Standards Australia, Sydney,
Fm ⫽ mean value of fabrication factor; Australia.
ASCE. 共2006兲. “Minimum design loads for buildings and other struc-
f od ⫽ elastic distortional buckling stress of cross
tures.” ASCE/SEI 7-05, Reston, Va.
section; Australian Standard 共AS兲. 共1991兲. “Methods for tensile testing of metals.”
f ol ⫽ elastic local buckling stress of cross section; AS 1391, Standards Association of Australia, Sydney, Australia.
fy ⫽ yield stress is taken as the static 0.2% proof Australian Standard 共AS兲. 共1993兲. “Steel sheet and strip—Hot-dipped
stress 共␴0.2兲; zinc-coated or aluminium/zinc-coated.” AS 1397, Standards Associa-
L ⫽ length of column specimen; tion of Australia, Sydney, Australia.
l ⫽ plate width for local geometric imperfection Batista, E., Rondal, J., and Maquoi, R. 共1987兲. “Column stability of cold-
measurements; formed U and C sections.” Proc., Int. Conf. on Steel and Aluminium
ley ⫽ column effective length for buckling about Structures, Cardiff, U.K., 419–427.
the minor y-axis; Denston, R. J., and White, J. D. 共1977兲. “An electrical demountable ex-
Mm ⫽ mean value of material factor; tensometer.” CUED Technical Report No. CUED/C-Struct/TR.61.
Pcrd ⫽ critical elastic distortional column buckling DeWolf, J. T., Peköz, T., and Winter, G. 共1974兲. “Local and overall buck-
load; ling of cold-formed members.” J. Struct. Div., 100共10兲, 2017–2036.
Pcre ⫽ critical elastic column buckling load in Hancock, G. J. 共1998兲. Design of cold-formed steel structures (To
flexural buckling; Australian/New Zealand Standard AS/NZS 4600:1996), 3rd Ed., Aus-
tralian Institute of Steel Construction, Sydney, Australia.
Pcrl ⫽ critical elastic local column buckling load;
Klöppel, K., and Bilstein, W. 共1976兲. “Untersuchungen zur linearen und
PDSM ⫽ nominal axial strength calculated using the nichtlinearen Beultheorie mit Beulwerttafeln fur dunnwandige U,
direct strength method 共unfactored design C- und Hut-Profile und Tafeln fur mitwirkende Breiten und Tragspan-
strength兲; nungen von dreiseitig und vierseitig gelenkig gelagerten Reckteck-
PDSM−S ⫽ PDSM for buckling stresses obtained using platten nach der nichtlinearen Beultheorie.” Stahlbau, 45共2兲, 33–38.
single section; Kwon, Y. B., and Hancock, G. J. 共1992兲. “Tests of cold-formed channels
PDSM−SR ⫽ PDSM for buckling stresses obtained using with local and distortional buckling.” J. Struct. Eng., 118共7兲, 1786–
single restrained section; 1803.
PDSM−MOD−S ⫽ PDSM for buckling stresses obtained using Lau, S. C.W., and Hancock, G. J. 共1988兲. “Strength tests and design
single section; methods for cold-formed channel columns undergoing distortional
buckling.” Research Rep. No. R579, School of Civil and Mining En-
Pm ⫽ mean value of tested-to-predicted load
gineering, Univ. of Sydney, Australia.
ratios;
Mulligan, G. P., and Peköz, T. 共1984兲. “Locally buckled thin-walled col-
Pnd ⫽ nominal axial strength for distortional umns.” J. Struct. Eng., 110共11兲, 2635–2654.
buckling; NAS. 共2001兲. North American specification for the design of cold-formed
Pne ⫽ nominal axial strength for flexural buckling; steel structural members, American Iron and Steel Institute, Washing-
Pnl ⫽ nominal axial strength for local buckling; ton, D.C.
PTEST ⫽ experimental ultimate load 共test strength兲; NAS. 共2004兲. Supplement 2004 to the North American specification for
Pu ⫽ ultimate load; the design of cold-formed steel structural members, American Iron
Py ⫽ squash load; and Steel Institute, Washington, D.C.
ri ⫽ inside corner radius at flanges of specimen; Papangelis, J. P., and Hancock, G. J. 共1995兲. “Computer analysis of thin-
ry ⫽ radius of gyration about the minor y-axis; walled structural members.” Comput. Struct., 56共1兲, 157–176.
Peköz, T. B., and Hall, W. B. 共1988兲. “Probabilistic evaluation of test
t ⫽ plate thickness of specimen;
results.” Proc., 9th Int. Specialty Conf. on Cold-Formed Steel Struc-
t* ⫽ base metal plate thickness of specimen;
tures, Univ. of Missouri-Rolla, Rolla, Mo.
VF ⫽ coefficient of variation of fabrication factor;
Rhodes, J., and Harvey, J. M. 共1977兲. “Interaction behaviour of plain
VM ⫽ coefficient of variation of material factor;
channel columns under concentric or eccentric loading.” Proc. 2nd
VP ⫽ coefficient of variation of tested-to-predicted
Int. Colloquium on the Stability of Steel Structures, ECCS, Liege,
load ratios; Belgium, 439–444.
w 1, w 2, w 3 ⫽ dimensions of intermediate web stiffeners; Rhodes, J., and Loughlan, J. 共1980兲. “Simple design analysis of lipped
z ⫽ measurement coordinate for local geometric channel columns.” Proc., 5th Int. Specialty Conf. on Cold-formed
imperfection measurements; Steel Structures, St. Louis, 241–262.
␤ ⫽ reliability index 共safety index兲; Schafer, B. W. 共2002兲. “Local, distortional, and Euler buckling of thin-
␦ ⫽ measured initial overall geometric walled columns.” J. Struct. Eng., 128共3兲, 289–299.
imperfections at midlength; Schafer, B. W., and Peköz, T. 共1998兲. “Direct strength prediction of cold-

736 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008

J. Struct. Eng., 2008, 134(5): 727-737

 formed steel members using numerical elastic buckling solutions.”
Proc., 14th Int. Specialty Conf. on Cold-Formed Steel Structures,
Univ. of Missouri-Rolla, Rolla, Mo., 69–76.
Thomasson, P. O. 共1978兲. “Thin-walled C-shaped panels in axial com-
pression.” Document D1:1978, Swedish Council for Building Re-
search, Stockholm, Sweden.
Weng, C. C., and Peköz, T. 共1990兲. “Compression tests of cold-formed
steel columns.” J. Struct. Eng., 116共5兲, 1230–1246.
Downloaded from ascelibrary.org by Kolhapur Institute Of Technology College Of Engineering (KIT on 03/25/19. Copyright ASCE. For personal use only; all rights reserved.
Yan, J., and Young, B. 共2002兲. “Column tests of cold-formed steel chan-
nels with complex stiffeners.” J. Struct. Eng., 128共6兲, 737–745.
Young, B. 共2005兲. “Experimental investigation of cold-formed steel
lipped angle concentrically loaded compression members.” J. Struct.
Eng., 131共9兲, 1390–1396.
Young, B., and Rasmussen, K. J. R. 共1998兲. “Design of lipped channel
columns.” J. Struct. Eng., 124共2兲, 140–148.
Yu, W. W. 共2000兲. Cold-formed steel design, 3rd Ed., Wiley, New York.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / MAY 2008 / 737

J. Struct. Eng., 2008, 134(5): 727-737