Experimental Study of Cold-Formed Steel Built-Up Beams

Francisco J. Meza, Ph.D. 1; Jurgen Becque, Ph.D. 2; and Iman Hajirasouliha, Ph.D. 3

Abstract: This paper describes a comprehensive experimental program on cold-formed steel built-up beams with two different cross-
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

sectional geometries. The work aimed to experimentally investigate the interaction between the individual components under increasing
loading and to quantify the effect of the connector spacing on the cross-sectional moment capacity and the behavior of the beams. In total,
12 specimens were tested in a four-point bending configuration, with lateral restraints provided at the loading points in order to avoid global
instabilities. The built-up specimens were composed of three or four plain channels with nominal thicknesses of 1.2 and 1.5 mm, which were
joined together using M6 bolts. Each built-up geometry was tested with three different connector spacings. The specimens were designed to
fail by local buckling of their components. Additionally, strut buckling of the channel comprising the top flange in between connector points
was observed. The local buckling deformations and the beam deflections were recorded during the tests. The material properties of the test
specimens were determined by means of coupon tests and the geometric imperfections were measured prior to testing. The experimental
results showed clear evidence of interaction between the local buckling patterns of the components, with the interaction being affected by the
connector spacing and the type of geometry. However, the connector spacing showed a less significant effect on the ultimate capacity when
failure was governed by local instabilities of the components. DOI: 10.1061/(ASCE)ST.1943-541X.0002677. © 2020 American Society of
Civil Engineers.
Author keywords: Built-up column; Experiment; Cold-formed steel; Stability; Buckling; Imperfection measurements.

Introduction openings are needed in walls for windows or doors, or in floors

(e.g., for stairwells). Moreover, in light-weight portal frames the
The use of cold-formed steel (CFS) load-bearing members has be- rafters and columns are typically fabricated by connecting two
come increasingly popular during the past few decades. In many channel sections back-to-back to achieve a doubly symmetric cross
countries, particularly the US, common instances can be found section with increased capacity (Lim and Nethercot 2002; Dubina
of multistory commercial and residential buildings which are con- et al. 2009; Zhang and Rasmussen 2014; Blum and Rasmussen
structed entirely out of CFS, while CFS portal frames are finding 2019). An added advantage of using built-up sections is that, in prin-
popular application throughout the world in industrial and agricul- ciple, they can be assembled on site and therefore the benefits asso-
tural buildings. This evolution can be attributed to the numerous ciated with their light weight and ease of handling largely remain.
advantages CFS has to offer, such as high strength-to-weight and However, a lack of fundamental understanding of the behavior
stiffness-to-weight ratios; convenient handling, transportation, and of built-up members, as well as the absence of sufficient design
stacking; and a high construction speed, especially when combined guidance in the current design codes (e.g., AISI 2016; CEN 2005)
with a modular approach. However, due to their reduced wall thick- often lead designers to use overly conservative assumptions, which
ness the ultimate capacity of CFS members is often governed by prevents the exploitation of the real potential of these types of sec-
cross-sectional instabilities such as local and distortional buckling, tions. Eurocode 3 (CEN 2005) in particular provides no specific
which may occur in combination with global buckling. design rules for CFS built-up members. In the North American
The increasing structural demands on CFS members which en- Specification (AISI 2016) the provisions for built-up flexural mem-
sue from their frequent use as primary structural members are fun- bers only apply to I-shaped sections assembled from channels con-
damentally at odds with the inevitable limitations on their wall nected through their webs in a back-to-back configuration. These
thickness resulting from the fabrication process. Therefore, built-
provisions focus on limiting the maximum longitudinal connector
up members are a common solution in current practice when indi-
spacing to prevent excessive distortion of the components between
vidual sections are not able to carry the load or cover the required
connectors and avoid tensile failure of the connectors resulting
span. In structural framing systems this may be the case when large
from the tendency of the individual channels to rotate about their
1 shear center as the load is applied in the contact plane between
The Steel Construction Institute, Silwood Park, Ascot, SL5 7QN,
UK (corresponding author). Email: F.Meza@steel-sci.com the webs.
2
Lecturer, Dept. of Civil and Structural Engineering, The Univ. of Limited research has been dedicated to built-up CFS flexural
Sheffield, Sir Frederic Mappin Bldg., Mappin St., Sheffield S1 3JD, UK. members. Wang and Young (2015) experimentally investigated the
Email: j.becque@sheffield.ac.uk flexural behavior of built-up CFS members with open and closed
3
Senior Lecturer, Dept. of Civil and Structural Engineering, The Univ. cross-sectional geometries and with circular web holes, with the
of Sheffield, Sir Frederic Mappin Bldg., Mappin St., Sheffield S1 3JD, UK. aim of extending the direct strength method (DSM) to cover these
ORCID: https://orcid.org/0000-0003-2597-8200. Email: i.hajirasouliha@ types of members. The open geometry consisted of two lipped
sheffield.ac.uk
channels screw connected in a back-to-back configuration, while
Note. This manuscript was submitted on July 17, 2019; approved on
January 27, 2020; published online on April 25, 2020. Discussion period the closed cross section was assembled from two plain channels
open until September 25, 2020; separate discussions must be submitted screwed together through their flanges. The authors explored differ-
for individual papers. This paper is part of the Journal of Structural En- ent ways of calculating the elastic buckling stresses required as in-
gineering, © ASCE, ISSN 0733-9445. put to the DSM, in order to account for the effects of the connectors

© ASCE 04020126-1 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

 and the web openings. This research was further extended by the
same researchers (Wang and Young 2016a, b) by carrying out an
experimental and numerical investigation of flexural members with
four different built-up cross-sectional geometries failing by cross-
sectional instabilities. Each built-up geometry was assembled from
two identical sections, screw connected either in a back-to-back
configuration to form an open section or through the flanges to
form a closed section. The authors assumed in their elastic stability
analysis that the built-up geometries behaved as one solid cross
section and found that, while for some built-up geometries the
DSM provided acceptable predictions, for others the predictions
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

were either overly conservative or unsafe. They resolved to modify

the DSM equations and proposed tailored equations for each built- Fig. 1. Cross-sectional geometries.
up geometry based on their experimental and numerical data. More
recently, a study on the flexural rigidity of built-up beams assembled
from lipped channels connected back-to-back was presented by improved bending efficiency. The second built-up geometry was
Dang and Rasmussen (2019). The authors demonstrated experimen- selected in consultation with CFS manufacturers/contractors, who
tally, numerically, and analytically that the flexural rigidity of a pointed out that this built-up geometry is regularly used as a sol-
built-up beam increases when more intermediate rows of connectors ution to bridge large openings in structural framing systems made
are provided and the connector stiffness increases, particularly when of CFS. The back-to-back channels thereby work as a lintel, while
the connectors are located toward the end of the member. the top channel is used as a track to receive the studs of the wall
Other research on CFS built-up flexural members has spurred above the opening. Due to the lack of design guidance, however,
some noteworthy innovations. The built-up LiteSteel beam (LSB), only the capacity of the back-to-back channels is currently counted
consisting of two channels with hollow rectangular flanges con- on in practice.
nected back-to-back, was investigated numerically and experimen- The channels were manufactured from pregalvanized steel
tally by Jeyaragan and Mahendran (2008a, b). The authors found plates of grade S450 according to CEN (2016), which have a nomi-
that the detrimental effect of lateral-distortional buckling, to which a nal yield stress of 450 MPa, and with a 0.04 mm Z275 zinc coating
single LSB is susceptible, can be mitigated by connecting two LSBs applied to both sides, in accordance with CEN (2004). The nominal
back-to-back, resulting in a moment capacity which is more than dimensions of the cross sections are listed in Table 1. The symbols
twice the capacity of the individual LSB section. The modular in the table refer to the dimensions of the individual channels,
light-weight cold-formed beam (MLC beam) is another innovative as clarified in Fig. 2. All beams had a total length of 3,400 mm,
built-up beam, consisting of two lipped channels with hollow flanges with a nominal distance between the end supports of 3,000 mm.
and a web containing stiffened openings. The channels are laser The specimens were loaded in a four-point bending configuration.
welded together in a back-to-back configuration with reinforcing The constant moment span, corresponding to the distance be-
plates placed inside the flanges. The MLC beam was designed to pro- tween the loading points, had a length of 1,600 mm. The portions
vide high resistance to lateral-torsional buckling and has been exten-
sively investigated both numerically and experimentally (Di Lorenzo
et al. 2004, 2006; Landolfo et al. 2009, 2008; Portioli et al. 2012). Table 1. Nominal dimensions of the component sections
This paper presents the results of a detailed experimental pro- Beam Section h (mm) b (mm) t (mm) rint (mm)
gram consisting of 12 CFS built-up beams with two different B1 T12915 129 48 1.5 3.0
cross-sectional geometries. The specimens were constructed by T10412 104 42 1.2 2.4
joining either three or four plain channel sections together using
M6 bolts. The specimens were tested in a four-point bending configu- B2 T12915 129 48 1.5 3.0
ration and were designed to fail by local buckling of the components T10412 104 42 1.2 2.4
within the constant moment span. Each built-up geometry was tested
with three different connector spacings and each test was repeated.
The program sought to investigate the various cross-sectional insta-
bilities in the beams, the interaction between the cross-sectional com-
ponents during buckling, and the effect of the connector spacing on
the buckling modes and the ultimate capacity of the built-up beams.
This experimental work fits within a wider program investigating
CFS built-up members, which also includes CFS built-up column
tests (Meza et al. 2016; 2020), with the eventual goal of developing
practical design rules for these types of structural members.

Specimen Geometry

The two different cross-sectional geometries considered in this

experimental program are illustrated in Fig. 1. The first geometry
was chosen because it resembles the traditional I-shaped cross sec-
tion widely used for beams. Based on the commonly encountered
Fig. 2. Nomenclature used to refer to the dimensions of the component
back-to-back channel arrangement, this built-up geometry included
sections.
two additional channels to increase the flange area and provide

© ASCE 04020126-2 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Location of connectors in beams with Geometries 1 and 2.

of the beam between the end supports and the loading points, fur- Finally, the components were bolted together using grade 8.8 M6
ther referred to as the shear spans, had a length of 700 mm. The bolts, and the clamps were removed. All bolts were tightened with a
built-up specimens were designed with either zero, two, or three controlled torque of 10 Nm. This is representative of the torque
equally spaced rows of connectors along the constant moment span, exerted by a spanner when hand-tightening the bolts and did not
in addition to the connectors placed in the cross sections under the introduce any significant pretensioning in the bolts. Fig. 4 shows
loading points. A close connector spacing along the shear spans of the specimens after the assembly process.
100 mm was maintained in all test specimens, as shown in Fig. 3, in
order to promote load sharing between all components, as well as to
avoid failure outside the constant moment region. Two identical Material Properties
beams were fabricated for each configuration in order to gain in-
creased confidence in the results. A series of coupon tests was carried out in order to determine the
A labeling system was used for both the individual components material properties of the beams. The coupons were cut along the
of the cross section (as indicated in Fig. 1) and the test specimens. rolling direction of the material (i.e., the longitudinal direction of
The latter consisted of a B followed by a 1 or a 2 to indicate the the beam) near the end section of one of the test specimens after
cross-sectional configuration (with reference to Fig. 1), the number testing. This location was chosen because the material had not been
of intermediate rows of connectors between the loading points, and subjected to significant stress in this region during the test. For each
the letter a or b to differentiate between the first and the second of type of channel component, two flat coupons were cut along the
twin test specimens. The actual cross-sectional dimensions of each center line of the web and two corner coupons were taken from
component were measured at several locations along the length be- the web-flange junction. The flat coupons had a nominal width of
fore assembly. The dimensions of the web and the flanges were mea- 12.5 mm, while the corner coupons had a nominal width of 6 mm.
sured using a digital Vernier caliper with a precision of
0.03 mm. All coupons were instrumented with an extensometer of 50 mm
All measurements were taken of the outside dimensions. The thick- gauge length. In addition, each flat coupon was instrumented with
ness of the cross section was measured with a micrometer with one 5-mm strain gauge on each side of the coupon, while each cor-
a precision of
0.002 mm. The measured cross-sectional dimen- ner coupon was instrumented with a 5-mm strain gauge on the
sions are listed in Tables 2 and 3 for all the specimens belonging outside of the corner. Due to the asymmetric shape of the corner
to Geometries 1 and 2, respectively. It should be noted that the val- coupons they were tested in pairs with a square bar placed between
ues reported in the tables are those obtained after accounting for the the gripped ends of the coupons, as illustrated in Fig. 5.
0.04 mm nominal thickness of the zinc coating. The cross-sectional area of the curved corner coupon was
In order to assemble the specimens, holes with a nominal diam- determined by taking a macro photograph of the cross section
eter of 6.25 mm were first drilled in one of the components at the using the reversed lens technique. The images were then imported
appropriate locations, as indicated in Fig. 3. The components were into AutoCAD software version 19.1 and scaled based on the
then positioned in their built-up configuration and secured with measurement of the width of the gripped end of the coupon, as
clamps. Next, the first component (containing the holes) was used illustrated in Fig. 6. The measured width of the coupon along the
as a template to drill the holes in the adjacent components. This gauge length was then superimposed on the photograph, allowing
allowed for an easy assembly, avoiding any mismatch of the holes. the area to be automatically calculated by the software. The process

© ASCE 04020126-3 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

 Table 2. Measured dimensions of specimens with Geometry 1
Web channels Flange channels
Beam Section Web (mm) Flange (mm) Thickness (mm) Section Web (mm) Flange (mm) Thickness (mm)
B1-0a T12915-1 129.36 43.38 1.493 T10412-1 104.35 39.82 1.141
T12915-2 129.15 43.49 1.487 T10412-2 104.10 39.92 1.139
B1-0b T12915-3 129.31 43.48 1.495 T10412-3 104.24 39.95 1.136
T12915-4 129.20 43.45 1.493 T10412-4 104.05 39.72 1.137
B1-2a T12915-5 129.26 43.45 1.489 T10412-5 103.95 39.94 1.137
T12915-6 129.13 43.53 1.496 T10412-6 104.06 40.04 1.139
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

B1-2b T12915-7 128.95 43.78 1.501 T10412-7 104.04 39.99 1.139

T12915-8 129.02 43.66 1.496 T10412-8 104.01 39.97 1.144
B1-3a T12915-9 128.93 43.74 1.498 T10412-9 103.95 39.98 1.141
T12915-10 128.90 43.69 1.501 T10412-10 104.18 39.97 1.140
B1-3b T12915-11 128.83 43.70 1.506 T10412-11 104.22 39.86 1.147
T12915-12 128.87 43.69 1.497 T10412-12 103.96 39.93 1.146
Average — 129.08 43.59 1.496 — 104.09 39.92 1.141
Standard Deviation — 0.183 0.136 0.005 — 0.129 0.087 0.004

Table 3. Measured dimensions of specimens with Geometry 2

Web channels Flange channels
Beam Section Web (mm) Flange (mm) Thickness (mm) Section Web (mm) Flange (mm) Thickness (mm)
B2-0a T12915-1 129.28 43.42 1.486 T10412-1 104.11 39.71 1.141
T12915-2 129.25 43.45 1.487 — — — —
B2-0b T12915-3 129.41 43.40 1.489 T10412-2 104.04 39.92 1.136
T12915-4 129.13 43.44 1.490 — — — —
B2-2a T12915-5 129.05 43.65 1.495 T10412-3 103.99 39.71 1.137
T12915-6 129.14 43.49 1.494 — — — —
B2-2b T12915-7 128.80 43.69 1.495 T10412-4 103.89 39.76 1.137
T12915-8 128.84 43.72 1.495 — — — —
B2-3a T12915-9 129.02 43.69 1.507 T10412-5 103.98 39.78 1.133
T12915-10 128.85 43.62 1.504 — — — —
B2-3b T12915-11 128.91 43.63 1.506 T10412-6 103.95 39.70 1.134
T12915-12 129.07 43.61 1.508 — — — —
Average — 129.06 43.57 1.496 — 103.99 39.76 1.136
Standard Deviation — 0.201 0.123 0.007 — 0.076 0.083 0.003

Fig. 4. Built-up beams after assembly.

© ASCE 04020126-4 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Corner coupon test setup.

Fig. 6. Macro photograph of the cross section of a corner coupon.

was repeated with pictures taken from the other end of the coupon Fig. 7. Stress-strain curves of flat and corner coupons.
and a difference in the calculated areas of less than 1.5% was ob-
tained for all coupons.
The tensile coupons were tested in a 300-kN Shimadzu AGS-X
(Kyoto, Japan) universal testing machine, while applying a displace-
by taking readings of the nominally flat table in the absence of a test
ment rate of 2 mm=min. Each test was halted for 2 min at regular
specimen, and these imperfections were deducted from the readings.
intervals in order to allow the load to settle down to static values and
Thus, an accuracy of approximately 0.06 mm was obtained, as this
eliminate strain rate effects. The static stress-strain curves are shown
was the guaranteed flatness of the table. Readings were taken along
in Fig. 7 for all coupons. Table 4 lists the (static) engineering values
of the material properties obtained for each coupon, as well as aver- different longitudinal lines on the specimen, as illustrated in Fig. 9,
age values for corresponding pairs. σ0.2% corresponds to the 0.2% within the constant moment span. Whenever possible, the imperfec-
proof stress, σu is the ultimate tensile strength, and εf is the elon- tions were recorded in the assembled state (as indicated by the red
gation after fracture measured over a gauge length of 50 mm. arrows in Fig. 9). However, the imperfections in the webs of the
built-up specimens were recorded before assembling the top and
bottom channels, as access was partially blocked in the assembled
Imperfection Measurements state (black arrows in Fig. 9).
The laser sensor was moved at a constant speed of 5 mm=s,
The capacity and behavior of thin-walled structural members in while readings were taken with a sampling rate of 50 Hz, resulting
many cases exhibit pronounced sensitivity to geometric imperfec- in a reading every 0.1 mm.
tions (Van der Neut 1969; Becque 2014). For this reason the im- The recorded imperfections were used to determine represen-
perfections of all specimens were recorded before testing using the tative out-of-plane imperfections in the web (δ web ) and the flanges
measuring rig developed by Meza et. al (2018), which consisted of (δ flange ) of the channels. δ web was determined as the out-of-plane
a nominally flat steel table and a traverse system powered by two imperfection recorded along the center line of the web relative to
electric motors (Fig. 8). The motors allowed the movement of a the imperfections recorded along the corners. δ flange was deter-
laser displacement sensor in two orthogonal directions along guid- mined by first computing the out-of-plane imperfections along the
ing bars. The imperfections in the guiding system were determined flange tip relative to the imperfections recorded along the corner.

© ASCE 04020126-5 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

 Table 4. Material properties of tensile coupons
E (GPa) σ0.2% (MPa) σu (MPa) εf (%)
Type Section Individual Average Individual Average Individual Average Individual Average
Flat T10412-a 212 208 432 426 471 466 14 14
Flat T10412-b 204 419 461 14
Flat T12915-a 201 204 544 531 628 619 13 13
Flat T12915-b 207 518 610 12
Corner T10412-a 193 198 — 460 — 479 — 3
Corner T10412-b 202 — — —
Corner T12915-a 211 218 — 585 — 645 — 5
— — —
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Corner T12915-b 225

Connector Behavior

A series of single lap shear tests with two fasteners in the line of
stress was carried out to determine the behavior of the connectors
used to assemble the built-up beams. The specimens were fabri-
cated from steel strips taken from spare sections and a total of
six tests were carried out. The specimens labeled CB15-12 com-
prised ply thicknesses and connectors identical to the configuration
encountered in the flanges of the built-up beams, while specimens
CB15-15 were representative of the connectors located in the web.
Three identical specimens were fabricated for each configuration in
order to account for statistical variability in the results (i.e., six tests
were carried out in total). Fig. 11 shows the test setup. The dimen-
sions of the test specimens were chosen following the recommen-
dations given in ECCS TC7 (2009) and the specimens were
assembled in the same way as the built-up beams. In particular,
Fig. 8. Imperfection measurement rig. the same controlled torque of 10 N · m was used. All specimens
were tested in a 300-kN Shimadzu universal testing machine and
were loaded to failure at a displacement rate of 0.5 mm=min. Fig. 12
shows the typical recorded load-elongation behavior. Further infor-
mation about the connector tests, as well as complete results for all
However, because it was not possible to place the web of the connector specimens, can be found in Meza (2018).
channel in a vertical position with perfect accuracy during the read-
ings, these initial values still contained an arbitrary constant offset,
which was eliminated by expressing δ flange relative to its average Test Setup
value along the beam. Fig. 10 shows typical imperfection profiles in
a representative beam. The vertical dashed lines indicate the loca- All 12 specimens were bent about their major axis using a four-point
tions of the connectors. The complete imperfection data of all spec- bending configuration, as illustrated in Fig. 13(a). The test speci-
imens can be found in Meza (2018). The maximum and the average mens were supported at their ends on rollers located 3,000 mm
values of δweb and δ flange are listed in Table 5 for the components of apart. The actuator was connected to a spreader beam, which exerted
each geometry. concentrated loads onto the specimen through loading points which

Fig. 9. Locations of the imperfection measurements.

© ASCE 04020126-6 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Out-of-plane imperfections of beam B2-2a.

Fig. 12. Load-elongation behavior of connector test specimens.

Table 5. Maximum and average imperfections

Imperfection (mm)
The cross sections of the test specimens above the supports
were packed with wooden blocks tightly fitted within the web
Specimen Section Maximun Average and flanges. This eliminated possible bearing failure and also pre-
B1 T12915 δweb 0.24 0.07 vented a distortion of the cross section characterized by a lateral
T10412 δweb 0.25 0.07 displacement of the compression flange combined with bending
δ flange 0.49 0.00 of the web about a horizontal axis in its plane. This distortion
B2 T12915 δweb 0.24 0.07 originated from the tendency of the beam to fail in lateral-torsional
T10412 δweb 0.35 0.11 buckling, and this phenomenon was observed in beam B1-0a
δ flange 0.51 0.00 (which was tested without wooden blocks) after local buckling took
place. This resulted in the beam being unable to reach the load at-
tained by its twin specimen, and therefore the ultimate capacity
obtained for specimen B1-0a was discarded.
Three LVDTs with a stroke of 50 mm were used to record the
deflections of the beams under the loading points and at midspan,
as illustrated in Fig. 13(a). Eight potentiometers with a stroke of
25 mm were also used to measure the buckling deformations
of the components. These potentiometers were mounted on two
identical yokes, which were attached to the bottom flange of the
built-up cross section, as shown in Fig. 15. As the test specimens
underwent overall bending, the bottom flange of the built-up cross
section experienced tension and was not at risk of local buckling,
while the yokes were able to move with the cross section as a
whole. Fig. 16 illustrates the positions of the yokes for each type
of test specimen.
The specimens were tested in a reaction frame using an actuator
with a capacity of 160 kN and a 50-kN load cell. The actuator
was steered using a Cubus controller, and a displacement rate of
1 mm=min was imposed. The tests were halted for 4 min slightly
before the peak load was reached in order to allow the load to settle
to its lower bound static value. The specimens were deformed well
beyond the peak load in order to identify the location where the
plastic deformations localized. The data was also collected using
Fig. 11. Single lap shear test setup. the Cubus system, using a sampling rate of 1 Hz.

Test Results
were implemented as simple supports (one roller and one pin)
located 1,600 mm apart. An adjustable lateral support system was
used to restrain the spreader beam against any out-of-plane move- Geometry 1
ment, as shown in Fig. 13(b). Nylon blocks were used as bearing All test specimens with Geometry 1 failed within the constant mo-
pads between the spreader beam and the uprights of the support ment span as a result of local buckling, with significant interaction
system in order to reduce friction. The loading points under the taking place between the top channel and the channels comprising
spreader beam consisted of top and bottom assemblies (Fig. 14), the web of the built-up cross section. The top channel consistently
which were bolted to the spreader beam and the test specimen, re- buckled first, before buckling was observed in the channels com-
spectively, and contained vertical flanges to prevent out-of-plane prising the webs. Fig. 17 illustrates the local buckling pattern,
displacements of the top flange of the test specimen. which typically contained 15 or 16 half-waves of fairly similar

© ASCE 04020126-7 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. (a) Four-point bending test rig; and (b) lateral support system.

Fig. 14. Loading devices: (a) pin support; and (b) roller support.

Fig. 15. Potentiometer layout within the cross section.

half-wavelength along the constant moment span. These buckles the connectors. It is important to note that the observed half-
were estimated to have an average half-wavelength of around wavelength of 100 mm was very close to the natural local buckle
100 mm. A slight exception occurred in beams B1-3, which dis- half-wavelength of 110 mm, calculated using CUFSM version 4.05
played 17 half-waves, with two of the buckles being noticeably (Schafer 2006) for the top channel in isolation. The web channels
smaller than the others. This was attributed to the presence of had a similar (calculated) natural local buckle half-wavelength of

© ASCE 04020126-8 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 16. Location of the potentiometers along the test specimens.

Fig. 18. Axial load versus lateral displacements (B1-3a).

Fig. 17. Local buckling pattern (specimen B1-0b).

the potentiometers (inward displacements of the channel). In the

100 mm, but the number of buckles generated in the web channels specimens with two or three sets of intermediate connectors along
and the associated half-wavelength could not be experimentally de- the constant moment span, the buckling pattern of the top channel
termined as the magnitude of these buckles was not large enough to was affected by subsequent buckling of the web channels, as illus-
allow accurate measurements. trated by Fig. 18. In order to accommodate the buckling pattern of
The potentiometers mounted on the yokes consistently revealed a the web channels, a modification of the buckle half-wavelength oc-
locally buckled shape in the top channel which was symmetric about curred in the top channel. This is reflected in a rather sudden change
the plane of bending in the initial stages of buckling, with both lips of the amplitude of the buckles of the top channel recorded locally
buckling either inward or outward. The web channels, on the other by the potentiometers (followed in some cases by a reversal in di-
hand, were recorded to buckle symmetrically with respect to the rection). The closer the connector spacing, the more pronounced this
plane of bending only in those specimens without any connectors effect was.
in the constant moment span (B1-0), while in the remaining spec- Despite the similarity of the initial buckling pattern observed in
imens the buckling pattern was antisymmetric. The symmetric buck- each pair of twin specimens, upon reaching the peak load the plastic
ling pattern of the web channels in the former specimens could be deformations generally localized in different locations. Yield lines
attributed to the fact that the channels were able to buckle largely simultaneously developed in the top channel and in the web and the
independently from each other, without any connectors in the con- top flange of the channels comprising the web of the built-up spec-
stant moment span forcing them to remain in contact. In the spec- imens, resulting in a hingelike behavior. In general, the yield line
imens with 2 and 3 rows of connectors, on the other hand, the pattern in the top channel was symmetric with respect to the plane
connector spacings were short enough to force the web channels to of bending, while that in the web channels was antisymmetric, as
buckle in a synchronous manner. This is illustrated in Fig. 18, which shown in Fig. 19. An exception to this occurred in specimens B1-0b
shows the out-of-plane displacements of the components of beam and B1-2b, in which the top channel developed an asymmetric
B1-3a against the applied load. The solid and dashed lines in this yield line pattern near midspan. This was accompanied by the top
figure represent the readings obtained from the set of potentiometers flange undergoing limited out-of-plane displacements, combined
placed on each of the two yokes. Positive values indicate displace- with twisting of the cross section along the constant moment span,
ments toward the potentiometer (outward displacements of the while a second hinge appeared near one of the loading points,
channel), while negative values indicate displacements away from as shown in Fig. 20. However, these out-of-plane displacements

© ASCE 04020126-9 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 19. Yield line mechanism in B1-2a: (a) top channel; and (b) channels comprising the web.

Fig. 21. Local buckling pattern (specimen B2-3a).

Fig. 20. Asymmetric yield line mechanism in built-up specimen B1-2b.

Fig. 22. Flexural-torsional deformations in specimen B2-0a, pictured

did not occur until after the peak load was reached, and therefore all
after the peak load.
beams could be considered fully laterally restrained until the test
specimens had reached their ultimate capacity.

was more significant, causing the top channel to twist with a half-
Geometry 2
wavelength of approximately 750 mm before the specimen reached
In general, the specimens with Geometry 2 failed within the con- its ultimate capacity, as shown in Fig. 23.
stant moment span by local buckling. Fig. 21 illustrates the typical In all test specimens the top channel buckled locally before the
local buckling pattern. However, in the beams without intermediate web channels, and the amplitude of its buckles was significantly
connectors (B2-0), flexural-torsional buckling of the top channel as larger. The top channel displayed between 14 and 16 half-waves
an individual component was observed to interact with local buck- along the constant moment span with fairly similar half-wavelength.
ling near the peak load. In the case of beam B2-0a this occurred Although the number of half-waves generated in the web channels
shortly after the local buckling deformations in the top channel was not easily discerned due to their small amplitude, it was noted
localized near midspan, resulting in localized twisting, as illustrated that in specimens B2-2 one of the buckles had a half-wavelength
in Fig. 22. However, in specimen B2-0b the interaction between significantly larger than the others. Despite the difficulty in distin-
local buckling and flexural-torsional buckling of the top channel guishing the half-waves in the web channels, it was estimated to

© ASCE 04020126-10 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 25. Axial load versus lateral displacements of B2-3a.

Fig. 23. Flexural-torsional deformations in specimen B2-0b, pictured
after the peak load.
The potentiometers consistently recorded a buckling pattern in
the top channels which was initially symmetric with respect to the
plane of bending, while the web channels were recorded to buckle
in an antisymmetric shape. As a typical example, Fig. 25 shows the
measured out-of-plane deformations of the components of beam
B2-3a. The potentiometers mounted on the yoke located to the left
(solid black lines) and to the right (dashed black lines) of midspan
recorded both flanges of the top channel to buckle inward. In the
yoke located to the right of midspan one of the potentiometers re-
corded an inward deformation in the web, while the other potenti-
ometer recorded an outward deformation.
In general, the readings from the potentiometers did not show
the buckling pattern of the top channels to be affected by sub-
sequent buckling of the web channels to the same degree as in
the specimens with Geometry 1. A noticeable interaction between
the components leading to a definite change in the locally recorded
amplitude of the buckles in the top channel was only observed in
beam B2-3b.
Fig. 24. Local buckling pattern in specimen B2-2b. Although the buckling pattern observed in twin specimens was
generally similar until the ultimate capacity was reached, localiza-
tion of the buckling pattern often occurred at different locations
along the constant moment region. As the built-up specimens
also be between 14 and 16 within the constant moment span. This deformed beyond their ultimate capacity, yield lines appeared in
resulted in half-wavelengths in the top channel and the web chan- the top channel and in the top flange of the web channels. As this
nels ranging from approximately 94 to 107 mm. These values were mechanism developed further, the built-up cross sections generally
very close to the natural local buckling half-wavelengths calculated experienced some twisting and lateral displacement in the neighbor-
using CUFSM version 4.05 (Schafer 2006) for the top channel hood of the resulting hinge. These deformations gradually spread
and the web channels in isolation, which were 120 and 100 mm, over the whole constant moment span as the specimen continued to
respectively. deflect. However, this phenomenon only occurred during the de-
In the beams without intermediate connectors (B2-0), a gap de- scending part of the load-deflection curve, and the beams could be
veloped along the constant moment span between the top channel considered to be sufficiently laterally restrained until they reached
and the web channels while the specimens deformed. This gap was their ultimate capacity. Fig. 26 shows the yield line mechanism
more pronounced around midspan, allowing a larger amplitude of which developed in a typical specimen.
the buckles in the top flange of the web channels to develop in this
region. On the other hand, toward the ends of the constant moment
span, where the components were in closer contact with each other, Critical Buckling Stresses
the amplitude of the buckles in the top flange of the web channels The buckling stresses of the individual cross-sectional components
was considerably smaller, to the point that it was difficult to ob- were determined from the readings of the potentiometers, using the
serve them. prediction of classical perturbation theory that the initial postbuck-
In beams B2-2, the top channel also tended to separate from ling deformations, when plotted against the load, follow a parabolic
the web channels, especially at midspan. However, the intermediate shape (Venkataramaiah and Roorda 1982). This is illustrated in
connectors opposed this separation, thereby pulling the flanges of Fig. 27. Because the top channel was the first component to buckle
the web channel upward, as illustrated in Fig. 24. in all of the test specimens, its critical buckling stress could be es-
In beams B2-3, no gap originated. In these specimens the top timated from the recorded bending moment, assuming that no slip
channels displayed exactly four buckled half-waves between con- occurred in the connections prior to buckling, so that plane sections
nectors, resulting in 16 half-waves along the constant moment span. remained plane:

© ASCE 04020126-11 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

Fig. 26. Yield line mechanism in B2-2a: (a) top channel; and (b) channels comprising the web.

of the connections is not accounted for. After buckling of the top

channels occurred, the stress distribution in these channels was ex-
pected to become highly nonlinear, so that the critical buckling
stress of the web channels could no longer be determined using
an analogous procedure.
For each beam the experimentally derived buckling stress of the
top channel was compared to the theoretical buckling stress ob-
tained from an elastic stability analysis carried out using CUFSM
version 4.05 (Schafer 2006). In this analysis the buckling stresses
were determined based on the average measured cross-sectional di-
mensions and the Young’s modulus obtained from the flat tensile
coupons. The top channel was considered in isolation, without any
interaction with the rest of the cross section, but under a stress gra-
dient consistent with the plane section assumption and using the
buckle half-wavelength observed during the test.
In order to confirm that no slip occurred in the connections be-
tween the top channel and the web channels prior to buckling, the
Fig. 27. Local buckling out-of-plane deformations.
shear force in the connectors located along the shear span was cal-
culated using Zhuravsky’s equation (Timoshenko 1940):
V · Atop · ȳ
M y V con ¼ Ls ð2Þ
σcr ¼ cr max ð1Þ 16I xx
I xx
where V con = shear force in the connector; V = shear force in the
where I xx = second moment of area of the built-up cross section beam; Atop = cross-sectional area of the top channel; ȳ = distance
about the horizontal axis through its centroid; ymax = distance from from the centroid of the top channel to the centroid of the built-up
the centroid of the built-up section to the extreme compression cross section, as illustrated in Fig. 28 for both geometries; and Ls =
fiber; and M cr = moment at which the potentiometers indicated length of the shear span. The resulting shear force is compared to
buckling to occur. The stress given by Eq. (1) provides an upper the slip force of the connectors, obtained from the “Connector
bound to the actual critical stress because the finite shear stiffness Behavior” section, in Table 6. The comparison indicates that no

Fig. 28. Distance between the centroid of the top channel and the centroid of the built-up cross section.

© ASCE 04020126-12 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

 Table 6. Shear force in shear span connectors
Specimen Vcon (N) Vslip (N) V slip =V con
B1-0a 1,528 2,600 1.702
B1-0b 1,881 2,600 1.382
B1-2a 2,010 2,600 1.294
B1-2b 1,836 2,600 1.416
B1-3a 1,573 2,600 1.653
B1-3b 1,490 2,600 1.745
B2-0a 1,044 2,600 2.490
B2-0b 954 2,600 2.725
B2-2a 1,091 2,600 2.383
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

B2-2b 1,041 2,600 2.498

B2-3a 1,087 2,600 2.392
B2-3b 1,035 2,600 2.512

Fig. 29. Moment versus relative deflection at midspan for specimens

with Geometry 1.
Table 7. Buckling stresses of the top channel
Buckling
Theoretical Buckling stress from
Component buckling stress from test/theoretical
Specimen section stress (MPa) test (MPa) buckling stress
B1-0a T10412-1 107 138 1.29
B1-0b T10412-3 111 170 1.53
B1-2a T10412-5 106 170 1.60
B1-2b T10412-7 109 154–166 1.41–1.52
B1-3a T10412-9 107 142 1.33
B1-3b T10412-11 108 126–134 1.17–1.24
B2-0a T10412-1 126 127 1.01
B2-0b T10412-2 126 120–142 0.95–1.13
B2-2a T10412-3 127 133–152 1.05–1.20
B2-2b T10412-4 127 121–136 0.95–1.07
B2-3a T10412-5 128 133 1.04
B2-3b T10412-6 128 128 1.00
Fig. 30. Moment versus relative deflection at midspan for specimens
with Geometry 2.

slip occurred in the connections prior to buckling of the top chan-

nels. Consequently, the plane section assumption used in the cal- Table 8. Ultimate capacities
culation of both the theoretical and the experimental buckling stress
Ultimate moment Average ultimate
of the top channel was valid in good approximation.
Beam (kN=m) moment (kN=m)
Table 7 compares the theoretical critical buckling stresses of the
top channels with the experimentally derived values. For some of B1-0a — 11.84
the test specimens the potentiometers located on the two different B1-0b 11.84
yokes captured the onset of local buckling in the top channel at B1-2a 12.51 12.17
B1-2b 11.83
slightly different stress levels, in which case both readings were
B1-3a 13.28 13.14
included in Table 7. In the beams with Geometry 1 the experimental B1-3b 13.01
buckling stresses were, on average, 40% higher than the theoretical B2-0a 9.08 8.63
buckling stresses of the isolated channel. This demonstrated that B2-0b 8.19
the top channel benefited significantly from the restraint provided B2-2a 10.41 10.53
by the unbuckled web channels. In the beams with Geometry 2, B2-2b 10.64
on the other hand, the experimentally derived buckling stresses B2-3a 11.66 11.70
generally coincided well with the theoretical ones, indicating that B2-3b 11.74
the top channel in these specimens did not noticeably benefit from
any restraint provided by the unbuckled web channels.
While no slip occurred in the connectors before buckling of the and 2, respectively. The midspan deflection was thereby considered
top channel took place in any of the specimens, once the specimens relative to the deflections of the beam under the loading points.
deformed beyond their ultimate capacity some relative movement Static moments were plotted, which were obtained after account-
between the components was generally noticed in the end sections ing for the decrease in moment observed when halting the test for
of the built-up beams. This slip measured between 1 and 2 mm at 4 min shortly before the peak load in order to eliminate strain-rate-
the end of the test. dependent effects.
The ultimate moment capacities obtained for all beams, as well
as the average values of the ultimate capacity for each set of twin
Ultimate Capacity
specimens, are listed in Table 8. Good agreement in the ultimate
Figs. 29 and 30 plot the bending moment against the relative de- capacity was generally achieved within each pair of twin speci-
flection at midspan for all specimens with built-up Geometries 1 mens. As previously discussed, beam B1-0a failed prematurely

© ASCE 04020126-13 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

 as a result of being tested without wooden blocking over the sup- restraint exerted by the web channels onto the top channel. A ten-
ports, causing the end sections to distort, accompanied by in-plane dency of the top channel to separate from the web channels was also
bending of the top flange. Consequently, the ultimate capacity re- observed, which was counteracted by the connectors.
corded for test B1-0a was discarded. The twin specimens of beams In both geometries it was observed that the components of the
B1-2 and B1-3 showed small differences in the ultimate capacity of cross section were forced to adjust their buckling patterns relative to
6% and 2%, respectively. their preferred patterns as isolated components, as a result of the
Regarding the specimens with Geometry 2, the B2-0 speci- connections and the contact between components. This manifested
mens showed the maximum difference in ultimate capacity, which itself in a reduction of the amplitude of the buckles or a constraint on
amounted to 11%. This can be attributed to the fact that in these the direction of the buckling displacements due to contact between
specimens flexural-torsional buckling of the top channel was ob- the components, and a minor adjustment of the half-wavelength
served to interact with local buckling. The interaction of global and of the buckles due to the constraints provided by the connectors.
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

local modes is notoriously imperfection sensitive, which might The specimens with Geometry 1 showed a relatively modest
explain why the flexural-torsional mode appeared slightly earlier increase in the ultimate capacity when reducing the connector spac-
over the loading history in specimen B2-0b and the interaction ing, while the ultimate capacity of the specimens with Geometry 2
was somewhat more pronounced, resulting in a lower ultimate was more significantly affected by the connector spacing. More
capacity. In beams B2-2 and B2-3, on the other hand, the difference specifically, reducing the connector spacing from 1,500 mm to
in ultimate capacity between twin specimens was 2% and 1%, 375 mm resulted in increases in the ultimate capacity of 11% and
respectively. 36% for the specimens with Geometries 1 and 2, respectively.
The test results showed that reducing the connector spacing The presented research is part of a wider investigation into built-
in the constant moment span resulted in a modest increase in the up cold-formed structural members (Meza et al. 2016; Meza 2018;
ultimate capacity of the specimens with Geometry 1. The beams Meza et al. 2020). The data gathered from the experiments will be
with two rows of intermediate connectors experienced an average used to develop accurate FE models which will serve as a tool to
increase in ultimate capacity of 3% relative to the beam without further study the behavior of built-up CFS members through para-
connectors, while the specimens with three rows of connectors ex- metric studies. The eventual goal is to develop accurate, safe, and
hibited an average increase of 11%. reliable design guidelines for these types of structural members for
The specimens with Geometry 2 showed a more substantial in- the benefit of practitioners.
crease in ultimate capacity when the spacing between the con-
nectors was reduced. The beams with two intermediate rows of
connectors along the constant moment span experienced an average Data Availability Statement
increase in ultimate capacity of 22% relative to the beams without
intermediate connectors, while the beams with three rows of inter- Some or all data, models, or code generated or used during the
mediate connectors displayed an average increase of 36%. This can study are available from the corresponding author by request,
in part be attributed to the elimination of the flexural-torsional including material properties data, geometric imperfections data,
mode for shorter connector spacings. connector behavior data, and out-of-plane deformation data.

Conclusions Acknowledgments

An experimental program investigating CFS built-up beams, en- The authors are grateful for the financial support provided by the
compassing 12 specimens with two different cross-sectional geom- EPSRC through Grant No. EP/M011976/1.
etries, is presented in this paper. The built-up specimens were
composed of plain channel components assembled using M6 bolts,
while varying the connector spacing among specimens. All beams References
were tested in a four-point bending configuration and lateral re-
straints were provided at the loading points in order to avoid global AISI (American Iron and Steel Institute). 2016. North American specifica-
instability of the built-up specimen. The geometric imperfections of tion for the design of cold-formed steel structural members. AISI S136.
Toronto: AISI.
the test specimens were measured using a laser sensor prior to test-
Becque, J. 2014. “Local-overall interaction buckling of inelastic columns:
ing, while the material properties of the constituent channels were
A numerical study of the inelastic Van der Neut column.” Thin Walled
determined by testing a series of tensile coupons. Struct. 81 (Aug): 101–107. https://doi.org/10.1016/j.tws.2013.07.010.
The specimens with Geometry 1 all failed by local buckling, Blum, H. B., and K. J. R. Rasmussen. 2019. “Experimental investigation of
with the top channel consistently being the first component to long-span cold-formed steel double channel portal frames.” J. Constr.
buckle. The observed buckle half-wavelengths were close to the Steel Res. 155 (Apr): 316–330. https://doi.org/10.1016/j.jcsr.2018.11
natural half-wavelengths of the isolated components, regardless of .020.
the connector spacing. However, a comparison of the experimen- CEN (European Committee for Standardization). 2004. Continuously
tally derived critical buckling stresses of the top channel with the hot-dip coated strip and sheet of structural steels—Technical delivery
theoretical buckling stress of the channel in isolation revealed that conditions. EN 10326. Brussels, Belgium: CEN.
the top channel benefited significantly from the restraint provided CEN (European Committee for Standardization). 2005. Design of steel
structures, Part 1.1: General rules and rules for buildings. EN 1993-
by the web channels.
1-1. Brussels, Belgium: CEN.
The specimens with Geometry 2 failed by local buckling, while
CEN (European Committee for Standardization). 2016. Designation sys-
flexural-torsional buckling of the top channel in between connector tems for steels, Part 1: Steel names. EN 10024-1. Brussels, Belgium:
points was also observed in the specimens with the largest connector CEN.
spacing. The experimentally measured buckle half-wavelengths and Dang, K. P., and K. J. R. Rasmussen. 2019. “Flexural rigidity of cold-formed
critical buckling stresses of the top channel agreed well with those steel built-up members.” Thin Walled Struct. 140 (Jul): 438–449. https://
calculated for the isolated channel, indicating a minimal amount of doi.org/10.1016/j.tws.2019.03.051.

© ASCE 04020126-14 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126

 Di Lorenzo, G., F. Portioli, and R. Landolfo. 2004. “On the characterization Meza, F. J. 2018. “The behaviour of cold-formed steel built-up structural
of the material properties for cold-formed steel members.” In Proc., members.” Ph.D. thesis, Dept. of Civil and Structural Engineering,
4th Int. Conf. on Thin-Walled Structures, 251–258. Loughborough, Univ. of Sheffield.
UK. Meza, F. J., J. Becque, and I. Hajirasouliha. 2020. “Experimental study of
Di Lorenzo, G., F. Portioli, and R. Landolfo. 2006. “Experimental inves- cold-formed steel built-up columns.” Thin Walled Struct. 140: 106291.
tigation on the load bearing capacity of built-up cold-formed steel https://doi.org/10.1016/j.tws.2019.106291.
beams.” In Proc., Int. Colloquium on Stability and Ductility of Steel Portioli, F., G. Di Lorenzo, R. Landolfo, and F. M. Mazzolani. 2012.
Structures. Lisbon, Portugal. “Contact buckling effects in built-up cold-formed steel beams.” Proc.,
Dubina, D., A. Stratan, and Z. Nagy. 2009. “Full-scale tests on cold-formed 6th Int. Conf. on Coupled Instabilities in Metal Structures, 1–8.
Glasgow, Scotland.
steel pitched-roof portal frames with bolted joints.” Adv. Steel Constr.
Schafer, B. W. 2006. “Buckling analysis of cold-formed steel members
5 (2): 175–194.
using CUFSM: Conventional and constrained finite strip methods.”
Downloaded from ascelibrary.org by "National Institute of Technology, Raipur" on 08/08/23. Copyright ASCE. For personal use only; all rights reserved.

ECCS TC. 2009. The testing of connections with mechanical fasteners in

In Proc., 18th Int. Specialty Conf. on Cold-Formed Steel Structures.
steel sheeting and sections. ECCS TC7. Brussels, Belgium: European Orlando, Florida.
Convention for Constructional Steelwork Technical Committee. Timoshenko, S. 1940. Strength of materials—Part 1: Elementary theory
Jeyaragan, S., and M. Mahendran. 2008a. “Experimental investigation and problems. 2nd ed. New York: D. Van Nostrand Company.
of the new built-up LiteSteel beams.” In Proc., 5th Int. Conf. on Van der Neut, A. 1969. “The interaction of local buckling and column fail-
Thin-Walled Structures. Brisbane, Australia. ure of thin-walled compression members.” In Proc., 12th Int. Congress
Jeyaragan, S., and M. Mahendran. 2008b. “Numerical modelling and of Applied Mechanics, 389–399. Delft, Netherlands: Delft Univ. of
design of the new built-up LiteSteel beams.” In Proc., 5th Int. Conf. Technology.
on Coupled Instabilities in Metal Structures, Sydney, Australia. Venkataramaiah, K. R., and J. Roorda. 1982. “Analysis of local plate
Landolfo, R., O. Mammana, F. Portioli, G. Di Lorenzo, and M. R. buckling experimental data.” In Proc., 6th Int. Specialty Conf. on
Guerrieri. 2008. “Laser welded built-up cold-formed steel beams: Cold-Formed Steel Structures, 45–73. St. Louis, MO.
Experimental investigations.” Thin Walled Struct. 46 (7–9): 781–791. Wang, L., and B. Young. 2015. “Beam tests of cold-formed steel built-up
https://doi.org/10.1016/j.tws.2008.03.009. sections with web perforations.” J. Constr. Steel Res. 115 (Dec): 18–33.
Landolfo, R., O. Mammana, F. Portioli, G. Di Lorenzo, and M. R. https://doi.org/10.1016/j.jcsr.2015.08.001.
Guerrieri. 2009. “Experimental investigation on laser welded connec- Wang, L., and B. Young. 2016a. “Behavior of cold-formed steel built-up
tions for built-up cold-formed steel beams.” J. Constr. Steel Res. sections with intermediate stiffeners under bending. I: Tests and numeri-
65 (1): 196–208. https://doi.org/10.1016/j.jcsr.2008.03.015. cal validation.” J. Struct. Eng. 142 (3): 04015150. https://doi.org/10
Lim, J. B. P., and D. A. Nethercot. 2002. “Design and development of a .1061/(ASCE)ST.1943-541X.0001428.
Wang, L., and B. Young. 2016b. “Behavior of cold-formed steel built-up
general cold-formed steel portal framing system.” Struct. Eng. 80 (21):
sections with intermediate stiffeners under bending. II: Parametric study
31–39.
and design.” J. Struct. Eng. 142 (3): 04015151. https://doi.org/10.1061
Meza, F., S. Cheng, J. Becque, and I. Hajirasouliha. 2016. “Experimen-
/(ASCE)ST.1943-541X.0001427.
tal investigation of the cross-sectional stability of cold-formed steel Zhang, X., and K. Rasmussen. 2014. “Tests of cold-formed steel portal
built-up columns.” In Proc., Int. Colloquium on Stability and Ductility frames with slender sections.” Steel Constr. 7 (3): 199–203. https://doi
of Steel Structures SDSS2016. Timisoara, Romania. .org/10.1002/stco.201410030.

© ASCE 04020126-15 J. Struct. Eng.

J. Struct. Eng., 2020, 146(7): 04020126