Thin-Walled Structures 54 (2012) 126139

Contents lists available at SciVerse ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Prediction of the exural strengths of welded H-sections with local buckling

Young Bong Kwon n, Gun Ho Seo
Department of Civil and Environmental Engineering, Yeungnam University, Gyongsan, Korea

a r t i c l e i n f o a b s t r a c t

Article history: This paper describes the exural strength of welded sections based on a series of exural tests
Received 5 January 2012 performed on H-sections fabricated from steel plates of thickness 6.0 mm with nominal yield stress of
Received in revised form 315.0 MPa. Thin-walled exural members undergo local, lateral-torsional or their interactive buckling
7 February 2012
according to the section geometries and lateral boundary conditions. Flexural members with the
Accepted 7 February 2012
Available online 22 March 2012
anges or the web of large width-to-thickness ratios may undergo local buckling before lateral-
torsional buckling and their interaction before the nal collapse of the section. The local buckling has a
Keywords: negative effect on the exural strength based on the lateral-torsional buckling. This phenomenon
Flexural tests should be considered in the estimation of the nominal exural strength of thin-walled exural
Flexural strength
members. Welded H-section beams composed of the anges and the web with various width-to-
Local buckling
thickness ratios were tested to failure. The initial imperfections in local and lateral buckling mode, and
Lateral-torsional buckling
Buckling interaction residual stresses were included in the FE analyses. Simple design exural strength formulas for the
Width-to-thickness ratio direct strength method (DSM) were proposed based on the test and FE results of welded sections to
Direct strength method account for interaction between local and lateral-torsional buckling. The design strength curves were
compared with the AISC specications (2005), Eurocode3 (2003) and test results. The adequacy of the
strength curve for the DSM was conrmed. A set of conclusions on the exural strength and structural
behavior of thin-walled welded H-sections was drawn from the experimental studies.
& 2012 Elsevier Ltd. All rights reserved.

1. Introduction strength reserve in the design strength, AISC specications (2005)

Flexural members composed of thin-walled plate elements
undergo local, lateral-torsional or their interactive buckling according
to width-to-thickness ratios of plate elements and the unsupported
length. Flexural members composed of compact elements where the
width-to-thickness ratios are smaller than a certain limit undergo
lateral-torsion buckling or yielding of material without local buckling.
Flexural members with large width-to-thickness ratio in the anges
or the web may undergo local buckling before lateral-torsional
buckling and interaction between them may occur before nal failure
[1,2]. Though there is a signicant post-buckling strength in a local
buckling mode, the local buckling has a negative effect on the
strength of exural members based on the overall lateral-torsional
buckling [3]. This phenomenon should be considered to predict the
ultimate exural strength of thin-walled exural members accu-
rately. However, in the case that the lateral rigidity is enough not to
allow later-torsional buckling in the unsupported length, local buck-
ling strength governs the moment capacity of exural members. The
effects of local buckling in the anges or the web on the moment
capacity of the exural members were studied by many researchers
[1,46]. To account for local buckling and post-local-buckling
n
Corresponding author. Tel.: 82 11 802 2418; fax: 82 53 810 4622.
E-mail address: ybkwon@ynu.ac.kr (Y.B. Kwon).
[7] and AASHTO specications (2007) [8] takes the minimum value
between ange local buckling strength and lateral-torsional buckling
strength as nominal moment capacity and uses load reduction factor
to account for web local buckling. NAS (2004) [9] and Eurocode 3
(2003) [10,11] adopt effective width method to account for ange and
web local buckling. KHBDS(2010) [12] has basic design strength
curves in separate to account for local buckling of stiffened plate and
unstiffened plate, and lateral-torsional buckling strength and takes
the minimum value as the nominal exural strength of the sections.
The direct strength method (DSM) has been developed by
Schafer and Pekoz [13] and studied further by many researchers
[1416]. The method has been developed to overcome the weak
points of the effective width method (EWM), which has been used
in thin-walled steel section design for over 60 years. These weak
points are the complication in accurate computation of effective
width and the difculty in consideration of the elements inter-
acting in isolation. Also, as cold-formed steel sections become
more complex with additional edge and intermediate stiffeners,
calculations of effective width can become very complex. In 2004,
the direct strength method was adopted formally as an alter-
native to the effective width method, which has been used for
thin section design by NAS Supplement 1 (AISI, 2004) [9] and the
Australian/New Zealand Cold-Formed Steel Structures Standard
AS/NZS 4600 [17]. The application of the DSM was studied for the
welded H-section and C-section columns with interaction between

0263-8231/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tws.2012.02.005
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139 127

local and overall buckling [18] and for longitudinally stiffened plates
undergoing interaction between local and distortional buckling [19]
recently.
This paper aims to develop the design exural strength formulae
for welded sections. The application of the DSM to welded section
exural members was studied experimentally and theoretically. A
series of exural tests was performed on welded H-sections fabri-
cated from structural steel plates of thickness 6.0 mm with nominal
yield stress 315.0 MPa to develop the exural strength formulae for
the welded steel sections undergoing interaction between local and
lateral-torsional buckling. Nonlinear FE analyses of the sections
tested have also been conducted to compare their results with the
test results. The strength formulae for the DSM have been proposed
and compared with the current design specications. The strength
curves proposed have been proven accurate and efcient to predict
the nominal exural strength of welded sections when the local
buckling and the lateral-torsional buckling occur simultaneously or
nearly simultaneously.
2. Test sections and material properties
2.1. Material properties
The structural steel grade of the test sections chosen was SM490
(KSD 3515) [20]. The minimum specied yield and ultimate tensile
stresses were 315.0 MPa and 490.0 MPa, respectively. Tensile cou-
pon tests were conducted for at coupons cut from the fabrica-
ted sections. All coupons were tested in a 250 kN capacity UTM
(Schmazu AUTOGRAPH AG 250kNG) at a displacement rate
0.1 mm/min. The yield and ultimate tensile stresses of the
Fig. 1. Cross section geometry of test section.
coupons cut from near the ange-web joint lines were slightly
higher than those of the coupons cut from the ange tip and web
center of the specimens. Average yield and tensile stresses obtained
from the tensile coupon tests were 413.0 MPa and 555.0 MPa, which
were 31.1% and 13.3% higher than the specied values, respectively.
Average elongation measured was approximately 25.0%. The elastic
modulus measured was 1.95  105 which was slightly lower than
the nominal value of 2.05  105 MPa.
2.2. Section geometries
The test H-sections were fabricated by gas welding from
SM490 structural steel plates of thickness 6.0 mm. The geome-
tries of the test specimens were determined using the buckling
analysis program BAP [21], which assumes pinned end boundary
conditions. The program BAP can account for the inelasticity of
the material and the residual stress distribution for inelastic
buckling analysis. The test section geometries determined were
proportioned to undergo local buckling and to fail in a mixed
mode between local and lateral-torsional buckling.
Bearing stiffeners of 10 mm in thickness were attached by
continuous llet welding at both supporting ends and at the one-
third points of overall length of the sections where concentrated
knife edge loads were applied to prevent web crippling or yielding
and ange local deformation. Additional horizontal stiffeners
were attached at both sides of the web outside of pure bending
zone in order to restrain web local buckling due to shear and
bending moments near both end supports before local buckling or
lateral-torsional buckling at pure bending zone due to pure
bending moment. Test section geometry is shown in Fig. 1.
Detail dimensions of test specimens are summarized in
Table 1. Compact and non-compact limits for exural members
in AISC specications (2005) [7] are 0.38(E/Fy)1/2 and 0.95(kcE/
0.7Fy)1/2 for the ange, and 3.76(E/Fy)1/2 and 5.70(E/Fy)1/2 for the
web of H-sections, respectively. The yield limits of width-to-
thickness ratio (b/t) provided by KHBDS (2010) [12] which are
equivalent to non-compact limit in AISC specications are 11.2 for
ange and 131.2 for web, respectively. As shown in Table 1, the
width-to-thickness ratio of anges ranged from 12.5 to 33.3 and
those of web ranged from 66.7 to 100.0. Test sections were
composed of compact and non-compact stiffened elements for webs
and non-compact or slender unstiffened elements for anges
according to the AISC specications (2005). Flanges of test sections
were categorized as class 3 and class 4, and webs were class 2 and
class 3 according to EC3 [10]. Referring to the width-to-thickness
ratios, it was supposed that test sections might undergo mainly
ange local buckling during testing.

Table 1
Detail dimensions of test sections.

Specimens bf (mm) tf (mm) d (mm) tw (mm) bf/tf h/tw l (mm) Ix (mm4)

H150-400 150.0 6.0 412.0 6.0 12.5 66.7 3000.0 106,181,600.0

H150-500 150.0 6.0 512.0 6.0 12.5 83.3 3000.0 177,721,600.0
H150-600 150.0 6.0 612.0 6.0 12.5 100.0 3000.0 273,261,600.0
H250-400 250.0 6.0 412.0 6.0 20.9 66.7 3000.0 155,636,000.0
H250-500 250.0 6.0 512.0 6.0 20.9 83.3 3000.0 254,536,000.0
H250-600 250.0 6.0 612.0 6.0 20.9 100.0 3000.0 383,436,000.0
H250-400 250.0 6.0 412.0 6.0 20.9 66.7 5000.0 155,636,000.0
H250-500 250.0 6.0 512.0 6.0 20.9 83.3 5000.0 254,536,000.0
H250-600 250.0 6.0 612.0 6.0 20.9 100.0 5000.0 383,436,000.0
H350-400 350.0 6.0 412.0 6.0 29.2 66.7 10,000.0 205,090,400.0
H350-500 350.0 6.0 512.0 6.0 29.2 83.3 10,000.0 331,350,400.0
H350-600 350.0 6.0 612.0 6.0 29.2 100.0 10,000.0 493,610,400.0
H400-400 400.0 6.0 412.0 6.0 33.3 66.7 10,000.0 229,817,600.0
H400-500 400.0 6.0 512.0 6.0 33.3 83.3 10,000.0 369,757,600.0
H400-600 400.0 6.0 612.0 6.0 33.3 100.0 10,000.0 548,697,600.0
128 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139

2.3. Determination of overall length of specimens
The effects of interaction between local and lateral-torsional
buckling, and post-buckling strength were the focuses of the
exural tests on welded H-sections with slender web and anges.
In order to determine the optimum length of the test sections, it
was necessary to calculate the local and lateral-torsional buckling
stresses of the specimens accurately. The overall lengths of
the test specimens were determined using the elastic buckling
stresses versus buckle half-wavelength curves obtained using the
program BAP [21] so that the test sections might undergo local
buckling rst and then fail in a mixed mode between local
and lateral-torsional buckling. However, since some of the test
sections composed of less slender plate elements might undergo
inelasticity of material, inelastic buckling stress should also be
considered in determination of the overall length. The membrane
residual stress distribution due to welding process assumed in the
inelastic buckling analysis, which was based on the tendon force
concept, is shown in Fig. 2. Magnitudes of compressive and tensile
residual stresses were 0.3Fy and Fy, respectively. The same
residual stress distribution was also used in FE analysis further.
Typical elastic and inelastic buckling stress versus half-wave-
length of the H250-500 section, subjected to pure bending moment,
is illustrated in Fig. 3. The minimum elastic local buckling stress
occurred in the curve at the half-wavelengths of 400 mm (point A)
but the minimum inelastic local buckling occurred at the half-
wavelength of 480 mm. The minimum elastic and inelastic local
buckling stresses computed were 296.3 MPa and 161.4 MPa, respec-
tively. As the half-wave length was increased beyond this minimum
point, elastic local buckling stress was increased up to 2200 mm
which was the maximum point (point B). The lateral-torsional
buckling mode of the section between 2200 mm and 3000 mm in
half-wavelength was slightly interacted with the local buckling
mode. The pure elastic lateral-torsional buckling occurred at the
half-wavelengths longer than 3000 mm. Referring to elastic and
inelastic buckling stresses in Fig. 3, the overall length of H250-500
section could be determined to be in the range from 3000 mm to
5000 mm to have interaction between local and lateral-torsional
buckling occur.
It was supposed that those sections whose inelastic local buck-
ling was higher than inelastic lateral-torsional buckling stress would
buckle only in lateral-torsional mode before collapse. However,
those sections of which inelastic local buckling stress was lower
than inelastic lateral-torsional buckling stress was supposed to
buckle rst in lateral-torsional mode before lateral-torsional buck-
ling, and show post-local-buckling strength and interaction between
local and lateral-torsional buckling before nal collapse. The overall
beam lengths of test sections were determined as 3000 mm,
5000 mm or 10,000 mm accounting for the liability of buckling
interaction between local and lateral buckling.
Elastic buckling stresses of test sections computed by the
program BAP [18] are summarized in Table 2. The lateral-torsional
buckling stresses corresponding to half-wavelengths 3000 mm,
5000 mm and 10,000 mm were included in Table 2. Referring to
the elastic buckling stress in Table 2, all the elastic local buckling
stresses of the H150 sections were higher than the nominal yield
stress of material 315.0 MPa. However, since lateral-torsional buck-
ling stresses of 3000 mm or longer in length were lower than the
yield stress, only lateral-torsional buckling would occur before
collapse. Local buckling stresses of the H250 were lower than
lateral-torsional buckling stresses of the sections of 5000 mm in
length. Therefore, it was expected that local buckling would occur
rst and then lateral-torsional buckling did. For the H300 sections,
local buckling stresses were higher than lateral-torsional buckling
stresses of the sections of 10,000 mm in length. For the H350 and
H400 sections of 10,000 mm in length, local buckling stresses were
lower than lateral-torsional buckling stresses at 10,000 mm in length.
It meant that local buckling would occur rst, post-local-buckling
strength would be displayed before lateral-torsional buckling

Table 2
Buckling stresses of test sections.
Fig. 2. Residual stress distribution.
Specimens Local buckling Lateral-torsional buckling stress (MPa)

Stress Half- L 3000 mm L 5000 mm L 10,000 mm

(MPa) wavelength
(mm)

H150-400 725.1 300 299.5 136.7 53.9

H150-500 583.3 360 273.3 125.7 49.1
H150-600 447.3 420 251.3 117.4 47.2
H200-400 462.4 320 555.9 224.0 73.7
H200-500 425.1 360 517.4 210.1 68.7
H200-600 370.7 440 481.6 198.3 66.1
H250-400 309.7 360 893.5 337.1 103.8
H250-500 296.3 400 846.4 316.3 98.3
H250-600 278.1 440 793.6 298.7 94.9
H300-400 220.6 400 1253.1 490.6 140.9
H300-500 213.6 440 1233.7 467.0 134.9
H300-600 205.7 460 445.6 130.8
H350-400 165.1 460 666.0 185.1
H350-500 160.6 480 643.3 179.0
H350-600 156.2 500 620.9 174.2
H400-400 128.4 500 838.5 237.5
H400-500 125.1 540 827.9 225.7
H400-600 122.2 560 813.9 216.2
Fig. 3. Elastic and inelastic buckling stress versus half-wavelength.
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139 129

occurred and then failed in the mixed mode between local and
lateral-torsional buckling. For the H300-600, H350 and H400
sections of 3000 mm in length, the critical buckling was a local
buckling rather than a later-torsional buckling.

2.4. Numerical analysis results

Prior to exural tests, the material and geometrical nonlinear

analysis of the specimens selected for the exural test was
conducted using the program LUSAS [22], to investigate the ultimate
exural strength and the structural behavior of the specimens.
The 4-node thick shell element (QTS4) was used for the numerical
modeling of the section, and the loaded end was assumed as a
simply supported boundary condition in both vertical and horizontal
directions. Two knife-edge loads were applied at approximately
one-thirds points of top ange of the sections in the vertical
direction. A pair of small triggering load was applied laterally at
the center of the top ange at one-thirds points of the test sections
to cause lateral-torsional buckling. The initial imperfections in the
critical local buckling mode were also included. Since the average
yield stress obtained from the tensile tests of the coupons which
were cut from the anges and web of the welded H-sections was
approximately 413.0 MPa, which was higher than the nominal yield
stress of 315.0 MPa by 18.1%. Therefore, the yield stress of the
material was taken as 413.0 MPa in the further FE analysis of the
test sections. Youngs modulus was assumed to be 2.0  105 MPa
and Poissons ratio was taken as 0.3. The stress-strain relation of the
material was assumed elastic-perfectly plastic neglecting the strain-
hardening and the von-Mises yield criterion was applied for the
plasticity theory of the material. Numerical modeling of the test
section is illustrated in Fig. 4.
To study the effect of buckling interaction between local and
lateral-torsional modes on the ultimate exural strength of the
test sections, an elastic buckling analysis was conducted rst to
nd out the critical local buckling mode. Since the negative effect
on the exural strength capacity seemed to be quite sensitive to
the imperfections, the sensitivity of the magnitude of imperfec-
tions was studied previously. The load versus shortening curves of
the H150-400 section obtained by the nonlinear analysis using
the program LUSAS with the different initial imperfections which
were the rst local buckling mode multiplied by the magnica-
tion factor b/10, b/100 and b/500 (b: clear half-ange width), are Fig. 5. Effect of initial imperfections and triggering load. (a) Effect of initial
compared in Fig. 5(a). The effect of the magnitude of triggering imperfections (with triggering load 0.001P) and (b) effect of triggering loads
(with initial imperfections 0.01b).
loads on the ultimate exural moment is also shown in Fig. 5(b).
The ultimate loads of the H150-400 section computed with different
magnitudes of imperfections of 0.01b and 0.002b under the same
triggering load were nearly similar as shown in Fig. 5(a). However,

Fig. 6. Deformed shape of test section H350-600-10000.

the ultimate load with 0.1b was lower than other those with 0.01b
and 0.002b by 6.1%. As shown in Fig. 5(b), the ultimate loads under
Fig. 4. FE analysis modeling. different triggering loads of 0.1%, 0.5% and 1.0% of reference vertical
130 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139
load were quite similar. The post peak load behavior under 0.1%
triggering load of vertical load were more stable than other two
cases. The ultimate loads and the structural behavior after the peak
load were nearly the same in the cases of 0.1% and 0.5%, triggering
load as shown in Fig. 5(b). In consideration of the effect of the
magnitude of the initial imperfections and the triggering load, the
initial imperfections could be assumed as the rst local buckling
mode multiplied by the factor of b/100 and the triggering load could
be taken as 0.5% of the vertical reference load in the further analysis.
Typically deformed shapes of the test section H350-600-10000
obtained by FE analysis and exural test are shown in Fig. 6.
Interaction between local and lateral-torsional buckling mode in
the pure bending zone between intermediate stiffeners is clearly
shown in the gures obtained by FE analysis and exural test.
Final failure mode was similar to the buckled shape where several
short half-wave of local buckling was interacted with long half-
wave of lateral-torsional buckling. Vertical load (transformed in
moment term) versus deection curves of the test sections of
600 mm in web depth obtained by the FE analysis are shown in
Fig. 7. Referring to the moment versus displacement curves in
Fig. 7, sections failing in lateral-torsional buckling mode with or
without local buckling except H300-600-1000 section showed
gradual drop of load after peak load. The sections of 10,000 mm in
overall length showed similar behavior except the section H300-
600-10000. The trace of local buckling for those sections could not
be found in the load versus deection curves in Fig. 7. However,
considering low elastic local buckling stress, those sections with
anges of large width-to-thickness ratio should show a signicant
Fig. 7. Vertical load versus deection curves.
Fig. 8. Test set-up and auxiliary device for loading.
post-local-buckling strength before peak load. The FE results were
compared with test results in the next sections.
3. Flexural tests of H-section beams
3.1. General
Pseudo-static exural test on the H-sections listed in Table 1
were executed using a 5000 kN universal testing machine (Shi-
madzu Co., Ltd.). Typical exural test set-up is shown in Fig. 8.
Both end boundary conditions in the tests were simple supports
in vertical and lateral directions. Lateral displacements at the
supports were effectively restrained by strong fabric strips shown
in Fig. 8. The auxiliary device for loading was a hinge composed of
three pieces, two grooved plates and a round bar. The round bar
was inserted between grooved plates as shown in Fig. 8. To minimize
the friction between round bar and grooved plates of the device for
loading, small amount of grease was applied. Bearing stiffeners were
attached to the web at the end supports and internal loading points
to prevent web crippling and local buckling. Horizontal stiffeners
were attached at both sides of the web outside pure bending zone to
prevent web buckling of side parts near supports due to shear and
bending as shown in Fig. 9. Two concentrated (knife-edge) loads were
applied at approximately one-third points of top ange in vertical
direction as shown in Fig. 9. Downward loading at 0.01 mm/s was
controlled by the displacement control method. The vertical displace-
ment was measured by two displacement transducers, which were
attached at the center of the test specimens. The horizontal displace-
ment at the beam center was measured using three LVDTs to
determine the loading point corresponding to the local and the
Fig. 9. Locations of LVDT, strain gauges and loading.
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139 131

Fig. 10. Buckled modes of test sections.

Fig. 11. Failure modes of test sections.

lateral-torsional buckling load. Since the local buckling occurred 3.2. Test results
mainly in the pure bending zone, three strain gauges were attached
at the top anges and the web at the beam center. Locations of strain After initial take up, the vertical deection at the center of the
gauges and displacement transducers are shown in Fig. 9. test specimen was linearly increased according to the load
132 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139
increment before local buckling occurred. For the H150-400-3000,
H150-500-3000, H150-600-3000, H350-400-10000, H350-500-
10000 and H350-600-10000 sections, as the load was increased,
a local buckling in two half-waves occurred rst and a signicant
post-buckling strength was observed before the commencement
of the lateral-torsional buckling. Then the buckling interaction
between local and lateral-torsional bucklings occurred and the
test section collapsed in the mixed mode of local buckling and
lateral-torsional buckling as shown in Fig. 10. Even if the local
buckling occurred rst, the lateral-torsional buckling governed
the exural strength. For the other sections, since the local
buckling stresses were much lower than lateral-torsional buck-
ling strengths, the interaction between local and lateral-torsional
bucklings was not observed. As the load was increased, local
buckling occurred in the pure bending zone and showed a
signicant post-buckling strength before nal collapse as shown
in Fig. 11. The nal collapse was resulted from local failure at the
local buckle waves as shown in Fig. 11.
The vertical load versus central deection curves for the speci-
mens are shown in Fig. 12(a)(c). The vertical load versus horizontal
displacement curves are shown in Fig. 13(a)(c). The horizontal
displacements were measured at the LVDT attached to top of the
web at the test beam center. However, the measurement point was
moved according to deection of beams. The magnitude of horizontal
displacement was increased abruptly near the peak load point where
lateral-torsional buckling occurred. The vertical load versus strain
curves are shown in Fig. 14(a)(e). The strains in Fig. 14(a)(c) were
the values of the gauges attached at compression ange center but
those in Fig. 14(d),(e) were the measured values of the side gauges as
shown in Fig. 9. At the intersection point on the vertical load versus
deection curves and the vertical load versus strain curves where the
Fig. 12. Load versus central deection curves. (a) d 400 mm, (b) d 500 mm and (c) d 600 mm.
slope changed sharply, a local buckling commenced at the top ange
and the web of the specimens. At the intersection point on the
vertical load versus horizontal deection curves where the slope
changed sharply, a lateral-torsional buckling occurred. However,
since the central deection was too large, the horizontal LVDT could
not stand at the same locations. For all the test sections, a signicant
post-local-buckling strength reserve was observed up to the ultimate
load on the load versus deection curves. The magnitude of post-
buckling strength was increased according to the increase of width-
to-thickness ratio of anges and web.
3.3. Comparison of test and FE results
The experimental results are summarized in Table 3 and com-
pared with the critical buckling and ultimate moments obtained by
the program LUSAS [22]. The FE buckling moments were determined
on the load versus displacement curves obtained by the LUSAS in
the same manner as used to determine the buckling load. The test
ultimate moments for H150 and H250 sections of less slender anges
were generally lower than the FE results by maximum 24.0%.
However, for H350 and H400 sections of slender anges, test results
were larger than FE results by 4.027.0%. For the H150 sections,
experimental lateral-tortional buckling moments were coincident
with numerical ones within maximum 11.0% difference. For the other
sections with anges of large width-to-thickness ratio, the numerical
buckling moment was slightly lower or higher than the test result.
There was little difference between test buckling and test maximum
moments for H150 sections. However, for the other sections, since
local buckling was critical buckling, the difference between experi-
mental buckling and maximum moments ranged from 68.3 MPa to
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139 133

Fig. 13. Load versus horizontal displacement curves. (a) d 400 mm, (b) d 500 mm and (c) d 600 mm.

290.8 MPa. As the width-to-thickness ratio of anges became larger, was determined as the minimum value among the compressive
the difference between buckling and maximum moments became stresses based on lateral-torsional and local buckling. The KHBDS
larger, which indicated that a signicant post-local-buckling strength cannot account for the interaction between local buckling and
was displayed before failure. In case that the elastic lateral-torsional lateral-torsional buckling and post-buckling strength of web and
buckling moments of the H150 and H250 sections were higher than ange in local buckling mode.
90% of the yield moment, the compressive residual stress might cause The nominal moment is given by based on local and lateral-
the inelastic behavior and affected signicantly on the ultimate torsional buckling is given by:
moments of the test specimens. However, since the local moments
and lateral-torsional buckling moments were comparatively low for Mn F cr S 1
the H350 and H400 specimens with slender anges, the residual
stress did not affect on the ultimate moment of these sections The critical stress Fcr based on lateral buckling is given by:
signicantly.
F cr F y a r 0:2 2a

4. Design methods for exural moments

F cr f1:00:412a0:2gF y a 40:2 2b

4.1. Current specications where

The design exural compressive stress was dened in KHBDS q 2k r

Fy l
[12] (Korean Highway Bridge Design Specications, 2010) account- a F y =F cr 2c
p Eb
ing for lateral-torsional buckling and local buckling, separately.
The design compressive stress based on the web local buckling
and ange local buckling was dened as stiffened element and p2 E
F cr 2d
unstiffened element, respectively. Design exural compressive stress L=r i 2
134 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139

Fig. 14. Load versus strain curves. (a) d 400 mm (at center), (b) d 600 mm (at center), (c) d500 mm (at center), (d) d500 mm (at left side) and (e) d500 mm (at right side).

 2
b b kp2 E t
r i r
  2k 2e F crl 3d
12 1 16 Awc A 121n2 b

Since the KHBDS (2010) does take a half of elastic local buckling
p
k 3 Aw =2Ac Z2:0 2f stress as a basic design strength curve, it gives too conservative
predictions of exural members based on local buckling. It neglects
The critical stress Fcr based on the local buckling is given by:
also the warping torsional rigidity in deriving the basic design curve
F cr F y R r0:7 3a based on lateral-torsional buckling. Therefore the curves cannot
effectively be used for the exural members with non-negligible
0:5 warping rigidity.
F cr R 4 0:7 3b
R In the AISC specications (2005) [7], lateral-torsional buckling
where and ange local buckling were treated separately. Web local buck-
s ling is treated in a different manner. To account for web local
Fy buckling, the reduction factors of RPC and RPG are computed accord-
R 3c
F crl ing to width-to-thickness ratio of the web of sections and used to
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139 135

Table 3
Comparison of test and FE results (kNm).

Specimens Tests FE results B/D

Buckling Maximum BA Buckling Ultimate

moments (A) moments (B) moments (C) moments (D)

H150-400-3000 172.0 239.9 67.9 154.4 (LTa) 247.3 0.97

H150-500-3000 190.8 236.6 45.8 189.7 (LT) 263.3 0.84
H150-600-3000 238.1 257.9 19.8 224.4 (LT) 338.1 0.76
H250-400-3000 211.7 294.9 83.2 234.0 (L) 294.2 1.00
H250-500-3000 269.0 356.4 87.4 294.6 (L) 378.6 0.94
H250-600-3000 317.5 385.8 68.3 348.5 (L) 405.0 0.95
H250-400-5000 205.8 284.3 78.5 234.0 (L) 320.0 0.89
H250-500-5000 281.3 353.0 71.7 294.6 (L) 394.3 0.90
H250-600-5000 343.0 417.5 74.5 348.5 (L) 485.5 0.86
H350-400-10000 180.5 311.8 131.3 164.4 (L) 251.9 1.24
H350-500-10000 232.0 392.5 160.5 207.9 (L) 314.3 1.25
H350-600-10000 227.5 483.9 256.4 252.0 (L) 380.0 1.27
H400-400-10000 160.5 328.3 167.8 143.2 (L) 314.4 1.04
H400-500-10000 188.5 426.6 238.1 180.7 (L) 384.8 1.11
H400-600-10000 203.4 494.2 290.8 219.1 (L) 447.9 1.10

a
L: local buckling; LT: lateral-torsional buckling.

reduce the moment capacity by multiplication. Structural sections accounted for in estimation of nominal exural strength of
are categorized in compact section, non-compact section and exural members and post-local-buckling strength of anges is
slender sections according to the width-to-thickness of anges to also not considered.
account for local buckling of anges of exural members. The Eurocode3 (2003, 2006) [10,11] adopted effective width concept
nominal exural strength is dened as plastic moment (MP ZFy) to account for local buckling of plated structural elements. Sections
for compact section, linearly interpolated values between plastic are categorized in four classes according to width-to-thickness ratio
moment and yield moment (Mr S(Fy  Fr)) accounting for the of plated elements. Plastic sectional modulus based on gross section
residual stress for non-compact section as is used for class 1 and 2 sections, elastic sectional modulus based on
  gross section is used for class 3 and effective sectional modulus
llr
M n RPG M p M p M r 4a based on effective area is used for class 4 section in estimation of the
lp lr
design buckling resistance moment. The design buckling resistance
where moment formula is given by:
s!
Fy
aw hc E Mb,Rd X LT bW 7a
RPG 1 5:7 r1:0 4b gM1
1200 300aw tw Fy
where, gM1 is the partial safety factor (1.1 for buckling strength);
hc t w bW Wpl,y is the plastic section modulus: Class 1, 2 sections,
aw r 10 4c
bf t f bW Wel,y is the elastic section modulus: Class 3 sections, bW Weff,y
is the effective section modulus: Class 4 sections, gM1 is the partial
lp and lr are limit width-to-thickness ratios for plastic and elastic safety factor, XLT is the reduction factor for the relevant buckling
bucklings, respectively. For slender ange sections, nominal exural mode:
strength is based on the elastic ange local buckling stress and is
given by 1
q r1:0 7b
2
0:9Ekc fLT fLT 2 l LT
M n RPG S 5a
bf =2t f 2
fLT 0:51 aLT l LT 0:2 l2 LT  7c
where
s
4 bW F y
0:35 r kc p r 0:76 5b l LT 7d
h=t w M cr
While, according to the laterally unsupported beam length where aLT is the imperfection factor (0.21 for rolled sections, 0.49 for
between lateral supports, nominal exural strength was also welded sections), Mcr is the elastic lateral-torsional buckling
divided into three categories: plastic moment, linearly interpo- moment accounting for moment gradient.
lated values between plastic moment and yield moment account- The effective area of class 4 sections can be computed according
ing for the residual stress in the same manner with replacing l for to Eq. (4.1)(4.3) in EC3 part15 (2006)[11].
l in Eq. (4a). The elastic lateral-torsional buckling moment given
by: 4.2. Direct strength method (AISI)
2
Cbp E
M n RPG S 6 The effective width method (EWM) adopted in EC3 [11] or NAS
Lb =r i 2
(2004) [9], which can account for the post-local-buckling strength
Nominal exural strength is taken as a minimum value among reserve and interaction between local and overall buckling in the
the moments and based on width-to-thickness ratio and lateral design strength, requires slightly complex calculations to com-
unsupported length. As shown above, the interaction between pute the effective width of the composing element in separate.
ange local buckling and lateral-torsional buckling is not To overcome this problem, the direct strength method (DSM) for
136 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139
cold-formed steel sections has been developed by Shafer and
Pekoz [13] and has been further studied by many researchers
[14,15]. The method was extended to the cold-formed stainless
steel sections [23] and welded steel sections [18,19]. The method
incorporates the empirical formulae and the elastic local or
distortional buckling stress obtained by the rigorous buckling
analysis or reliable strength formulas. The direct strength method
was formally adopted in the NAS (AISI, 2004) [9] and AS/NZS
(2005) [17] recently. The nominal exural strength for local
buckling is for l r0.776.
M nl M ne
For l 0.776
  !  sectional modulus referenced to the extreme ber in rst yield. The
M crl 0:4 Mcrl 0:4
M nl 10:15 M ne
M ne M ne
where l (M ne /M crl ) 1/2 x; M nl Sf fnl is the exural strength
accounting for local buckling and lateral-torsional buckling;
Mcrl SfFcrl is the critical elastic local buckling moment; Mne is
the nominal exural strength based on the lateral-trosional
buckling; Sf is the the gross sectional modulus referenced to the
extreme ber in rst yield.
The direct strength method has been adopted as an alternative
design procedure for the design of cold-formed steel structural
members by NAS (2004) [9] and AS/NZS (2005) [17]. The relia-
bility of the direct strength method equals or betters the tradi-
tional effective width method for a large database of tested beams
and columns [16]. Extensive design aids for cold-formed steel
sections are now available for designers and expansion of the
direct strength method is underway.
4.3. Proposed direct strength method
The application of the direct strength method to welded
section columns undergoing interaction between local and overall
buckling was studied by Kwon et al. [18] recently. The nominal
compressive strength formulas for the direct strength method
considering the interaction between local and overall buckling
was used for the prediction of nominal compressive strength for
welded H-section and C-section columns tested. The direct strength
formulas adopted by NAS (2004) [9] and AS/NZS (2005) [17] were
slightly modied to account for the specic characteristics of welded
sections which are quite different from cold-formed steel sections.
The nominal compressive strength Pnl accounting for the interaction
between local buckling and overall buckling for the welded section
columns proposed by Kwon et al. [18] are given by
For l r 0:816 Pnl Pne
  !  yield stress was used for Fig. 16(a) and (b) instead of nominal
P crl 0:5 P crl 0:5
For l 4 0:816 Pnl 10:15 Pne
P ne P ne
where ll (Pne/Pcrl)1/2; Pcrl is the critical elastic local buckling load;
Pne is the overall column strength based on the overall failure mode,
which is determined from the minimum of the elastic exural,
torsional, and exural-torsional buckling stresses. Overall column
strength Pne can be computed by the current specications
such as AISC (2005) [8], EC3 (2003) [10] or KHBDS (2010) [12].
The 0.5 exponent in Eq. (9b), rather than 0.4 in Eq. (8b) as used for
cold-formed steel sections, reects lower post-buckling strength
reserve for welded sections than cold-formed sections.
Based on the ndings for welded section columns, a similar set of
equations can be derived for welded section exural members
undergoing interaction between local buckling and lateral-torsional
buckling. The design exural strength Mnl accounting for the
interaction between local buckling and lateral-torsional buckling
for welded section exural members can be given by
For l r0:816
M nl Mne 10a
For l 40:816
  ! 
M crl 0:5 M crl 0:5
M nl 10:15 M ne 10b
M ne M ne
where l (Mne/Mcrl)1/2; Mnl Sffnl is the exural strength accounting
for local buckling and lateral-torsional buckling; Mcrl SfFcrl is the
8a
critical elastic local buckling moment; Mne is the nominal exural
strength based on the lateral-trosional buckling; Sf is the the gross
8b nominal exural strength Mne can be calculated from the AISC [8],
the Eurocode 3 [10] or the KHBDS [12]. The critical elastic local
buckling stress Fcrl under pure bending moment can be computed by
the rigorous Finite Element Method (FEM) or the Finite Strip Method
(FSM) [13].
The design exural strength Eqs. (10a) and (10b) based on the
nominal exural strength Mne of AISC specications [8] and EC3
[10] were compared with FE and test results in Figs. 15(a),(b) and
16(a),(b), respectively. The nominal yield stress 315 MPa was used
in calculation of nominal exural strength Mne for Fig. 15(a) and test
yield stress for Fig. 15(b). The EC3 [10] species that the reduction
factor c or d for the buckling curve should be used for welded
I-sections with h/br2 or h/b42, respectively. The exural strength
curve based on EC3 curve c or d was compared with test and FE
results in Fig. 16(a). However, exural strength predictions by the
DSM based on EC3 curve a are also compared with test results in
Fig. 16(b) for reference. Referring to the moment predictions by
DSM in Figs. 15(a),(b) and 16(a),(b), the resulting exural strength
predicted by the proposed formulas were very conservative for all
the test sections in comparison with FE and test results. In
comparison between Fig. 15(a) and (b), it can be known that the
test yield stress affected the maximum test moment and the
design exural strength signicantly. Since the nominal exural
strengths based on the gross section area for exural members
based on the EC3 and AISC specications are different, there are
slight differences between the comparison results shown in
Figs. 15(a),(b) and 16(a),(b). The nominal exural strength Mne
based on the EC3 was slightly more conservative than that based
on AISC specications in the plastic and elastic buckling region.
The reason is that the nominal exural strength Mne calculated by
EC3 without accounting for the local buckling is generally lower
than that obtained by AISC specications [24]. Referring to the
curves Fig. 16(a) and (b), the nominal exural strength Mne based
9a on the EC3 buckling curve c or d is much more conservative
than that based on the EC3 buckling curve a. However, if the test
9b yield stress, the predictions could be less conservative.
Referring to the similar comparison for welded section col-
umns (Kwon et al., 2007 [18]), the proposed formulas predict
more conservative strength for exural members than for com-
pression members. One main reason of signicantly conservative
prediction was the yield and tensile stresses of test specimens
which were much higher than nominal stresses by 30.8% and
6.1%, respectively. Another reason was that the interaction effect
between local buckling and lateral-torsional buckling on the
bending moment capacity might be less signicant than that of
between local and column bucklings on the compressive strength.
The post-local-buckling strength reserve of exural members of
large local buckling slenderness may also be more signicant than
that of the welded compression members due to mainly the
different stress distribution between compression member and
exural member.
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139 137

Fig. 15. Comparison of DSM based on AISC and test results. (a) Fy 315.0 MPa (nominal yield stress) and (b) Fy 413.0 MPa (test yield stress).

Fig. 16. Comparison of DSM based on EC3 and test results. (a) Buckling curve c and d and (b) buckling curve a.

4.4. Comparison with current specications
A comparative study of AISC and EC3 strength limit states for
columns and exural members was executed recently [24]. Since
partial safety factor and combination factor in EC3, load factor
and resistance factor in AISC, and safety factor in KHBDS are
different, direct comparison of design exural strength for all
sections is difcult. The different way how to account for the local
buckling in current specications makes general comparison
between them impossible. However, comparison of basic
design moment curves based on the lateral-torsional buckl-
ing without local buckling can give some meaningful information
for design philosophy. Since the design moment is based on
laterally unbraced length (Lb/ri) for AISC specications and the
lateral buckling slenderness factor lLT (My/Mcr)1/2 for EC3 and
KHBDS, unbraced length was transformed into lateral buckling
slenderness. For comparison, the H150-400 section where lateral-
torsional buckling governed the design moments was selected.
For the H150-400 section, ange is categorized as class 3 and web
is class 2 in the EC3. The anges and the web are classied as non-
compact section in the AISC specications. The lateral buckling
slenderness and the shape factor (x Mp/My) of the H150-400
were computed and considered. The comparison in Fig. 17 shows
that the reduction factor (buckling curve) of the section in EC3 is
too conservative in comparison with the AISC specications.
The KHBDS gives less unconservative predictions than EC3 in
lateral buckling region but both specications are well coincident
in the laterally stable region. For the compact section in the AISC
and class 1 or 2 section in the EC3, since the plastic moment is
nominal moment, the difference may become smaller than that in
Fig. 17. The AISC specications takes account for local buckling of
the web and the anges of sections separately and the post-local-
138 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139

Table 4
Comparison of tests, DSM and current specications (kNm).

Specimens Tests DSM (EC3) DSM (AISC) AISC EC3 KHBDS

H150-400-3000 239.9 87.4/107.8 147.1 147.1 75.6 109.0

H150-500-3000 236.6 111.9/138.0 192.6 192.6 96.8 143.3
H150-600-3000 257.9 137.0/168.7 239.9 239.9 118.5 179.7
H250-400-3000 294.9 175.5/213.5 217.9 157.3 160.7 74.8
H250-500-3000 356.4 225.3/279.2 281.3 185.2 214.3 98.0
H250-600-3000 385.8 274.4/348.9 344.9 213.1 243.1 124.1
H250-400-5000 284.3 136.2/167.4 189.0 157.3 121.9 74.8
H250-500-5000 353.0 173.6/213.8 242.6 185.2 157.4 98.0
H250-600-5000 417.5 212.2/261.6 294.7 213.1 168.7 124.1
H350-400-10000 311.8 118.8/151.9 158.4 105.8 106.8 49.3
H350-500-10000 392.5 150.6/192.4 196.3 123.0 135.6 64.0
H350-600-10000 483.9 183.5/234.8 234.7 139.9 165.4 79.8
Fig. 17. Comparison of EC3, AISC and KHBDS for H150-400. H400-400-10000 328.3 132.4/203.2 176.4 90.8 130.8 24.9
H400-500-10000 426.6 166.5/254.3 223.8 105.1 168.5 32.2
H400-600-10000 494.2 201.4/306.6 272.4 119.1 207.9 40.0

torsional buckling governed the exural strengths. The ange

Fig. 18. Comparison of EC3 and DSM
buckling strength is not considered. However, EC3 adopts effec-
tive width concept to account for local buckling and post-local-
buckling strength of the web and anges of the section, and uses
different section modulus according to the category of composing
elements of sections. Therefore, the comparison illustrated in
Fig. 17 cannot tell which specication produces higher moment
capacity than other specications.
The local buckling and post-local-buckling strength of slender
plate elements can be accounted for by the conventional effective
width method in EC3 or NAS. The effective width formulas for
internal and outstand compression element for welded exural
members in EC 3, and Eqs. (10a) and (10b) are compared in
Fig. 18. EC3-A curve is used for internal element (web) and EC3-B
curve is applied for outstand element (anges) in EC3. As shown
in Fig. 18, the DSM nominal exural strength curves are slightly
higher than EC3 effective width curves in the buckling range of
elements where the local buckling slenderness factor is between
0.673 and 2.0. The reduction factors for the EC3 and the DSM are
quite similar in plastic range and elastic buckling range where the
local buckling slenderness factor is smaller than 0.673 and larger
than 2.0, respectively. The comparison can tell that if the nominal
exural strength for the DSM based on the EC3, the DSM predicts
higher design exural strengths than the EC3.
The design exural strengths of test sections predicted by the
DSM proposed and the current specications are summarized in
Table 4. The predictions by the DSM based on EC3 and AISC
specications were included. The design exural strengths pre-
dicted by the DSM based on EC3 buckling curve a were also
included at the lower part of DSM (EC3) column for comparison.
For the the AISC specications and KHBDS, the nominal exural
strengths were determined by the ange local buckling strength
for the sections except for the H150 sections where lateral-
elements of the sections except the H150 were categorized in
class 4 and therefore the section modulus based on the effective
area was used for the calculation of the nominal exural
strengths. The ange of H150 sections were categorized in class
3. The web of 400 mm in depth was categorized in class 2 and
those of 500 mm and 600 mm were in class 3.
All the predictions by the DSM and current specications were
conservative in comparison with the test results. One main reason
was the measured yield and tensile strengths were much higher
than the nominal strengths. For the H150 sections, the EC3 predicted
too conservative nominal exural strengths compared with the AISC
and the DSM and test results. Since the DSM accounted for post-
local-buckling strength of anges but the AISC did not, the DSM
produced slightly higher predictions for the sections except for H150
sections than the AISC specications. In concerning with the H250
sections, the EC3 predicted higher exural strength for 3000 mm
long sections than the AISC, but for section of 5000 mm in length,
vice versa. It resulted from that while the nominal exural strengths
by the EC3 were based on the lateral-torsional buckling, the
predictions by the AISC specications were governed by the ange
local buckling rather than lateral-torsional buckling and therefore
they were the same. Even if the AISC specications could not
account for buckling interaction, it produced very conservative
predictions for the H350 and H400 sections undergoing signicant
interaction between local and lateral-torsional buckling. Since the
EC3 can account for post-local-buckling strength but the AISC
cannot, the predictions by the EC3 were higher than those by the
AISC for the H350 and H400 sections with local buckling. Since the
KHBDS had too conservative design exural strength curves based
on local buckling, it generally produced too conservative moments
for all sections except the H150 sections where LTB governed the
nominal moment in comparison with test results and other speci-
cations. Referring to the comparison in table 4, it can be concluded
that the DSM proposed can predict fairly reasonable design exural
strengths for welded H-section beams with or without interaction
between local buckling and lateral-torsional buckling.
5. Conclusions
The experimental and numerical study for the application of
the direct strength method (DSM) to the welded H-section
exural members undergoing interaction between local buckling
and lateral-torsional buckling was conducted. The local buckling,
which occurs prior to the lateral-torsional buckling and has a
signicant post-local-buckling strength reserve, deteriorates the
 Y.B. Kwon, G.H. Seo / Thin-Walled Structures 54 (2012) 126139
exural moment capacity to some extent. This phenomenon
should be considered appropriately to predict the nominal ex-
ural strengths of the welded sections accurately.
The direct strength formulas, which were adopted for the design
of the cold-formed steel sections, have been modied to take into
account the effect of the interaction between local buckling and
lateral buckling and post-local-buckling strength reserve for the
design exural strengths of welded H-sections. The bending moment
capacity from the direct strength formulas proposed were compared
with the test results of welded H-section beams fabricated from the
steel plates with nominal yield stresses of 315.0 MPa. The bending
moment capacity predicted by the direct strength method was also
compared with predictions estimated by the AISC specications, EC3
and KHBDS. The comparison of the nominal exural strengths
predicted by the design specications and the test results has proven
the reliability of the direct strength method. The direct strength
formulas proposed have been proven designer-friendly to apply and
accurate to predict the design exural strengths of the H-section
members that undergo the interaction between local buckling and
lateral-torsional buckling and fail in the mixed mode between local
buckling and lateral-torsional buckling. However, detailed provisions
as to the width-to-thickness ratio limits for the ange and web
should be prepared and the proposed direct strength formulas
should be further veried and calibrated against the test results of
various kinds of sections for practical use.
Acknowledgment
This paper forms a part of the project entitled Development of
Direct Strength Method for Welded Sections being conducted in
the Department of Civil Engineering at the Yeungnam University.
This research was supported by the 2011 research fund of
National Research Foundation of Korea.
References
[1] Kemp AR. Inelastic lateral and local buckling in design codes. Journal of
Structural Engineering, ASCE 1996;122:374482(4) 1996;122:374482.
[2] Hancock GJ. Local, distortional and lateral buckling of I-beams. Journal of
Structural Division, ASCE 1978;104(11):178798.
[3] Kwon YB, Seo GH The Bending Moment Capacity of Flexural Members with
Interaction of Local Buckling and Lateral-Torsional Buckling. In: Proceedings
139
of Thirteenth IABSE-IASS 2011 Conference; Taller, Longer, Lighter-Meeting
Crowing Demand of Limited Resources, London, England; 2011.
[4] Cherry S. The stability of beams with buckled compression anges. Structural
Engineering 1960;38(9):77285pp2 1960;38:77285.
[5] Bradford MA, Hancock GJ. Elastic interaction of local and lateral buckling in
beams. Thin-Walled Structures 1984;2(1):125.
[6] Basler K, Thuliman B. Strength of plate girders in bending. Journal of
Structural Division, ASCE 1992;87(ST 6):15381.
[7] American Iron and Steel Construction (AISC). Load and Resistance Factor
Design specication for Steel Structural Buildings, Chicago, IL, USA; 2005.
[8] American Association of State Highway and Transportation Ofcials. AASHTO
LRFD Bridge Design Specications, 4th ed., Washington, D.C., USA; 2007.
[9] American Iron and Steel Institute. North American Specications for the
Design of Cold-Formed Steel Structural Members, Supplement no. 1,
Washington DC, USA; 2004.
[10] European Committee for Standardisation (ECS). Eurocode 3: Design of Steel
Structures, Part 1-1: General rules and rules for buildings, Brussels, Belgium;
2003.
[11] European Committee for Standardisation (ECS). Eurocode 3: Design of Steel
Structures, Part 15: Plated structural elements, Brussels, Belgium; 2006.
[12] Korean Ministry of Land and Maritime Affairs. Korean Highway Bridge Design
Specications (KHBDS), Seoul, Korea, 2010.
[13] Schafer BW, Pekoz T. Direct strength prediction of cold-formed steel
members using numerical elastic buckling solutions. In: Shanmugan NE,
Liew JYR, Thevendran V, editors. Thin-Walled Structures, Research and
Development. Elsevier; 1998 p. 13744.
[14] Hancock GJ, Murray TM, Ellifrit DS. Cold-Formed Steel Structures to the AISI
Specications. Marcel Dekker, Inc.; 2001.
[15] Silvestre M, Camotim D Direct strength method for lipped channel columns
and beams affected by local-plate/distortional interaction. In: Proceedings of
Eighteenth International Specialty Conference on Cold-Formed Steel Struc-
tures, Orlando, FL, USA; 2006, p. 1737.
[16] Schafer BW. Review: the direct strength method of cold-formed steel
member design. Journal of Constructional Steel Research 2008;64:76678.
[17] Standard Australia. Cold-Formed Steel Structures AS/NZS 4600: 2005, Syd-
ney, NSW, Australia; 2005.
[18] Kwon YB, Kim NG, Hancock GJ. Compression tests of welded section columns
undergoing buckling interaction. Journal of Constructional Steel Research
2007;63:1590602.
[19] Kwon YB, Park HS. Compression tests of longitudinally stiffened plates
undergoing distortional buckling. Journal of Constructional Steel Research
2011;67:121224.
[20] Korean Standards. Welded Structural Steel, KS D 3515, Seoul, Korea, 2003.
[21] Kwon YB Buckling Analysis Program (BAP) users manual, version 2.0,
Department of Civil Engineering, Yeungnam University, Korea; 2000.
[22] FEA Co., Ltd. Lusas Element reference manual & users manual, version 14.3-7;
2010.
[23] Lecce M, Rasmussen KJR. . Distortional buckling of cold-formed stainless steel
sections: nite-element modelling and design. Journal of Structural Engi-
neering, ASCE 2006;132(4):50514.
[24] Topkaya C, Sahin SA. Comparative study of AISC and EC3 strength limit states.
International Journal of Steel Structures 2011;11(1):1327.