Journal of Constructional Steel Research 148 (2018) 97–111

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Section moment capacity tests of hollow flange steel plate girders

Nilakshi Perera, Mahen Mahendran ⁎
Queensland University of Technology (QUT), Brisbane, QLD 4000, Australia

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

Article history: Hollow flange steel plate girder (HFSPG) is a new hollow flange I-section made using cold-formed rectangular
Received 13 July 2017 hollow sections (RHS) as flanges and a steel plate as web. Due to the increased torsional rigidity and unique
Received in revised form 6 April 2018 geometry, it can provide enhanced flexural capacities and thus are effective flexural members in long span
Accepted 30 April 2018
applications. The production of proposed HFSPGs contains welding currently available RHS sections to a web
Available online xxxx
plate, which allows engineers to form girders by varying dimensions, thicknesses and grades to suit their design
Keywords:
requirements. As the first step of this study, the section moment capacities of fully laterally restrained HFSPGs
Hollow flange steel plate girders were investigated. Despite earlier researches on the flexural capacities of hollow flange sections, HFSPGs have
Rectangular hollow sections not been tested previously. Therefore, twelve section moment capacity tests were conducted to study the flexural
Flexural members performance and capacities of HFSPGs. The ratios of ultimate moment capacity per unit area of tested beams were
Section moment capacity tests compared with conventionally used hot-rolled I-sections with similar cross-sectional area, which proved the
Design guidelines structural efficiency of HFSPGs over commonly used hot-rolled I-sections. The ultimate capacities of tested
HFSPGs were then compared with capacity predictions provided by the Australian, American and European
design standards (both hot-rolled and cold-formed). It was found that the current design standards under-
estimate the section moment capacities of HFSPGs in general and the level of under-estimation varied depending
on the section slenderness. Suitable recommendations are made regarding the appropriate use of these design
standards for HFSPGs. This paper presents the details of this research and its findings.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction use of LSBs was advantageous, they are no longer manufactured.

The applications of cold-formed steel have increased due to the
benefits it offers over conventionally used hot-rolled steel. To enhance
the structural performance of widely used cold-formed open sections,
hollow flange sections have been introduced in recent years. The use of
hollow flanges eliminates weaker elements with free edges and
increases the torsional rigidity of open I-girders. In addition, it provides
a higher second moment of area as a result of locating considerable
steel area away from its neutral axis, thus leading to an increased flexural
capacity about its major axis. After the introduction of the first
hollow flange section in 1990, various mono-symmetric and doubly-
symmetric hollow flange sections have been introduced and investigated.
Palmer Tube Mills Pty Ltd. produced the first hollow flange beam
(HFB) (see Fig. 1 (a)), which consisted of two closed triangular flanges.
HFBs were used for spans b10 m, however, connection difficulties with
triangular flanges was a weakness of HFBs. Following HFBs, mono-
symmetric LiteSteel beams (LSB) (see Fig. 1 (b)) were developed for
use as flexural members. Extensive research on the flexural perfor-
mance of LSBs was conducted by Anapayan and Mahendran [1] and
Anapayan et al. [2]. LSBs provided a solution to the connection problem
in HFBs and were used in spans up to 15 m. However, although the
⁎ Corresponding author.
E-mail address: m.mahendran@qut.edu.au (M. Mahendran).
Wanniarachchi and Mahendran [3] introduced screw fastened rectan-
gular hollow flange beams (RHFB) (see Fig. 1 (c)) for use in short to
intermediate span construction. However, this production method
lead to higher initial geometric imperfections and cannot be recom-
mended for larger unsupported spans. Following the applications of
single LSBs, designers proposed the use of built-up LSBs by connecting
two LSBs using bolts, in order to achieve improved performance.
Jeyaragan and Mahendran [4] studied the flexural behaviour of back
to back LSBs. Recently, Siahaan et al. [5] studied the flexural behaviour
of RHFCB, which is a rivet fastened channel section with rectangular
hollow flanges for use in medium span floor systems. As described
above, a range of innovative hollow flange sections has been investi-
gated by many researchers for use as flexural members in small span
applications (b15 m). However, studies on hollow flange sections for
use in large unsupported spans are limited to date.
Advanced architectural designs increase the demand for innovative
steel members with more favourable characteristics in large span appli-
cations. Hot-rolled and welded I-sections (universal beams-UB, welded
beams-WB) are the primary elements used in large span applications.
These I-sections are available in 9 to 30 m lengths, therefore, the previ-
ously mentioned hollow flange sections cannot be used as alternatives.
Due to the unique properties of hollow flange sections, development of
a suitable girder with hollow flanges is beneficial for long span applica-
tions. Dong and Sause [6] showed that hollow tubular flange girders

https://doi.org/10.1016/j.jcsr.2018.04.034
0143-974X/© 2017 Elsevier Ltd. All rights reserved.
98 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111

literature shows that the use of hollow flange sections provides efficient
Nomenclature
options over commonly used open I-girders.
The proposed hollow flange steel plate girder (HFSPG) (see Fig. 2) in
Ms, Mc,Rd Section moment capacity
this research is a built-up hollow flange steel plate I-girder made using
My First yield/elastic moment
commonly used cold-formed rectangular hollow sections (RHS) and a
Mp Plastic moment capacity
steel web plate. It is expected to achieve a steel girder with improved
ro Outer corner radius
performance. Based on the RHSs and steel plates available in the market,
t Element thickness
these sections allow designers to manufacture girders with different
Myw First yield/elastic moment when yielding commenced
thicknesses, dimensions and grades to suit their capacity requirements.
at the extreme fibre of the web element
The widths of the currently available RHSs vary from 50 to 400 mm and
Myf First yield/elastic moment when yielding commenced
the width/depth ratios commonly used for hollow flanges sections vary
at the extreme fibre of the flange element
from 0.25 to 0.48. The use of these RHS gives HFSPGs a maximum sec-
Ag Cross-section area
tion depth of 1600 mm, which have the potential to span up to 30 m.
Bf Width of the flange
Also, the ability to use different thicknesses and yield strengths for
D Total depth of a cross-section
web and flanges allows the designers to develop optimum sections.
Ze, Weff,min Effective section modulus
However, the flexural behaviour and capacities of the proposed
λs Slenderness of the section
HFSPGs have not been investigated. Welding of RHSs to a web plate
λsy Yield slenderness of the section
can induce unique geometric imperfections and residual stresses in
λsp Plastic slenderness of the section
the members, which will cause significant differences between HFSPGs
Z, Wel,min Elastic section modulus of the section
and the hollow flange sections investigated in the past. Hence a research
Zc, Wpl Plastic section modulus of the section
study was undertaken to investigate the structural performance and
b Width of the element excluding the radii
capacity of HFSPGs subjected to bending. The focus of this paper is the
ρ Effective width factor
section moment capacity of fully laterally restrained HFSPGs subject to
be Effective width of the element
local buckling and yielding effects.
E Modulus of elasticity
Twelve tests were conducted on HFSPGs formed using RHS flanges
k Plate buckling coefficient
and steel web plates. The section moment capacities of HFSPGs were
f* Applied stress
assessed by conducting four-point loading tests about the major axis
Mbl Nominal member moment capacity for local buckling
on simply supported HFSPGs. This paper presents this experimental
Mbe Nominal member moment capacity for laterally
study and its results. A brief comparison was conducted with the test
restrained members
results and the flexural capacities of the hot-rolled I-sections used in
Mol Critical elastic local buckling moment of the section
the industry, which proved the efficiency of the proposed sections over
γM0 Partial factor of resistance
conventionally used I-sections. Finally, test moment capacity results
were compared with the predictions of AS 4100 [10], AS/NZS 4600
[11] or AISI S100 [12], EN 1993-1-1 [13] and EN 1993-1-3 [14]. Details
(HTFG) can be used in bridge construction. Wimer [7] and Kim and of these comparisons and the findings are presented in this paper.
Sause [8] studied I-shaped girders with a concrete filled hollow flange
in compression and a flat plate in tension, known as concrete filled
tubular flange girders (CFTFG) (see Fig. 1 (d)). In their study Kim and 2. Experimental study
Sause [8] concluded that lateral torsional buckling capacity of CFTFGs
is immensely greater than that of I-girders if the cross-section distortion 2.1. General
is controlled. Hassanein and Silvester [9] used numerical analyses to
examine the shear and flexural performance of hollow tubular flange The section moment capacity, Ms, is influenced by local buckling
plate girders (HTFPG). In their study, they compared the performance and/or yielding of the cross-section elements. According to the cross-
of HTFPGs with I-section plate girders (IPGs) and confirmed the higher section classification given in the Australian hot-rolled steel design
capacities of HTFPGs over IPGs. However, it was limited to numerical standard AS 4100 [10], the section moment capacities of the slender
modelling. Among the novel ideas presented over the years, the sections are limited to their first yield moment (My), and are subjected

Fig. 1. Hollow flange sections (a) hollow flange beam, (b) LiteSteel beam, (c) Rectangular hollow flange beam, (d) Concrete filled hollow flange beams.
 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111 99

Rectangular
hollow flanges
Continuous
weld

Fig. 2. Hollow flange steel plate girder.

to elastic local buckling effects. The compact sections do not fail until
they achieve their full plastic moment capacities (Mp). Non-compact
sections have Ms values between My and Mp, and are subjected to
inelastic local buckling effects. The HFSPG sections for the tests were
chosen such that they fall into the above categories.
Unlike the sections already available in the market, the proposed
HFSPGs do not have fixed dimensions. Therefore, the thicknesses,
widths and heights of the sections can be selected arbitrarily. However,
since it is intended to form HFSPGs by welding currently used high
strength and cold-formed rectangular hollow sections (RHS) to a web
plate, only the currently available RHSs were considered. After consider-
ing the feasibility of conducting the tests with respect to the available
loading capacity and test space, twelve HFSPG sections were selected
for the section moment capacity tests.
2.2. Test specimens
Twelve HFSPGs made of C450PLUS RHS and Grade 250 and 350 web
plates were tested in two series of tests, three tests in the first series and
nine tests in the second. The test specimen dimensions were measured
and are presented in Table 1. The base metal thicknesses measured
during the tensile coupon tests were used (Table 1). However, the
corner radii of RHS flanges were not measured, instead the nominal
corner dimensions provided by the manufacturers were used, i.e. the
outer radius ro is equal to twice the thickness (2 t) if t b 3 mm and 2.5 t
if t N 3 mm. As shown in Table 1, the specimens were labelled in order
to identify the sizes of the RHS used for flanges and the web plate. The
label consists of the beam type (i.e. HFSPG) followed by the RHS and
the web plate sizes used as flanges and web, respectively. The steel
grades are not included in this labelling system. For each tested speci-
men, tensile tests of coupons extracted from web and RHS flanges
were conducted to determine their mechanical properties.
Table 2 presents the compactness classification of HFSPGs based on
AS 4100 [10] and EN 1993-1-1 [13]. In Table 2, “C”, “NC” and “S” denote
compact, non-compact and slender sections, according to the AS 4100
[10] classification whereas, EN 1993-1-1 [13] uses four classes to classify
these sections. Both classifications are in agreement except for HFSPGs
9 and 10, ie. non-compact sections based on AS 4100 [10], but Class 2
(compact) sections based on EN 1993-1-1 [13].
2.3. Mechanical properties and initial imperfections
Accurate determination of the mechanical properties and initial
conditions is vital before using the proposed HFSPGs in the construction
industry. Mechanical properties of cold-formed steels are influenced by
the manufacturing process. When the steel sheet is cold-worked and
welded to produce rectangular hollow sections (RHS), residual stresses
are induced and the mechanical properties of steel can be altered.
2.3.1. Tensile coupon tests
Tensile coupons were extracted from grade 450PLUS RHSs and grade
250 and 350 web plates and tested according to the procedure in the
Australian Standard AS 1391 [15]. Tensile coupons of 1.6, 2.5 and
3 mm thicknesses extracted from RHSs and 1.6 and 3 mm thick web
plates were tested to failure. The coupons were extracted in the longitu-
dinal direction from different positions of the RHSs and web plates of
the HFSPGs. Rounded corners and the sides of welded seams of RHSs
were eliminated when cutting the coupons. The coupons were cut
using the water-jet method to minimize the heat and stresses induced
in the coupons during extraction. The paint layer and other impurities
on the surface within the 50 mm gauge length of the coupons were
removed using a sand paper and cleaned using Acetone to measure
the base metal thicknesses before attaching the strain gauges.

Table 1
Measured dimensions of tested HFSPGs.

Test/HFSPG Test specimen Steel grade bfo df dw tw tf

W F

1 HFSPG-75 × 25 × 2.5-100 × 3 250 450PLUS 75.00 24.50 101.00 2.97 2.47

2 HFSPG-75 × 25 × 2.5-100 × 3 350 450PLUS 75.00 24.75 100.00 2.99 2.47
3 HFSPG-75 × 25 × 2.5-150 × 1.6 250 450PLUS 75.00 24.60 151.00 1.58 2.47
4 HFSPG-75 × 25 × 2.5-200 × 1.6 250 450PLUS 75.00 25.00 201.00 1.58 2.47
5 HFSPG-75 × 25 × 1.6-100 × 3 250 450PLUS 75.00 23.50 101.00 2.97 1.57
6 HFSPG-75 × 25 × 1.6-100 × 3 350 450PLUS 75.00 24.00 101.00 2.99 1.57
7 HFSPG-75 × 25 × 1.6-150 × 1.6 250 450PLUS 75.00 24.25 151.00 1.58 1.57
8 HFSPG-75 × 25 × 1.6-200 × 1.6 250 450PLUS 75.00 23.75 200.50 1.58 1.57
9 HFSPG-65 × 35 × 2.5-100 × 3 250 450PLUS 65.00 34.50 100.00 2.97 2.45
10 HFSPG-65 × 35 × 2.5-100 × 3 350 450PLUS 65.00 34.75 100.00 2.99 2.45
11 HFSPG-65 × 35 × 2.5-150 × 1.6 250 450PLUS 65.00 34.50 150.50 1.58 2.45
12 HFSPG-65 × 35 × 2.5-200 × 1.6 250 450PLUS 65.00 35.00 201.00 1.58 2.45

Note: All dimensions are in mm.

bfo – outer flange width, df – flange depth, dw – web depth, d – total depth of the section, tw – web thickness and tf – flange thickness, W- web, F- flange.
100 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111

Table 2 measured yield strengths of RHS flanges are higher than their nominal
Section classification of tested HFSPGs. yield strengths while they are less than their nominal yield strengths
Test/HFSPG Test Specimen Compactness Compactness for some web plates. The lowest point in the yield plateau was taken
using AS 4100 using EN 1993–1-1 as the yield strength of sharp yielding graphs whereas 0.2% proof stress
W F Overall W F Overall was taken as the yield strength in gradual yielding graphs. All the stress-
strain graphs obtained for RHS flanges exhibited a gradual yielding type
1 HFSPG-75 × 25 × 2.5-100 × 3 C NC NC 1 3 3
2 HFSPG-75 × 25 × 2.5-100 × 3 C NC NC 1 3 3 whereas the graphs obtained for the low grade web plates exhibited
3 HFSPG-75 × 25 × 2.5-150 × 1.6 NC NC NC 3 3 3 sharp yielding type except in one case. Table 3 depicts the results of
4 HFSPG-75 × 25 × 2.5-200 × 1.6 S NC S 4 3 4 these tensile coupon tests.
5 HFSPG-75 × 25 × 1.6-100 × 3 C S S 1 4 4
6 HFSPG-75 × 25 × 1.6-100 × 3 C S S 1 4 4
7 HFSPG-75 × 25 × 1.6-150 × 1.6 NC S S 3 4 4
2.3.2. Initial geometric imperfections and residual stresses
8 HFSPG-75 × 25 × 1.6-200 × 1.6 S S S 4 4 4 Higher residual stresses are likely to occur during the cold-forming of
9 HFSPG-65 × 35 × 2.5-100 × 3 C NC NC 1 2 2 RHS flanges and the welding process of hollow flanges with webs to form
10 HFSPG-65 × 35 × 2.5-100 × 3 C NC NC 1 2 2 HFSPGs. Therefore, accurate distribution and magnitudes of residual
11 HFSPG-65 × 35 × 2.5-150 × 1.6 NC NC NC 3 2 3
stresses are important for the numerical analysis of HFSPGs. However,
12 HFSPG-65 × 35 × 2.5-200 × 1.6 S NC S 4 2 4
due to the difficulties in measuring the residual stresses, an idealized re-
Note: W- web, F- flange, C – compact, NC – non-compact, S – slender. EN 1993–1-1 uses
sidual stress model was developed based on available literature [6,
the class of the element which varies from 1 to 4.
16–18] for use in the numerical modelling phase of this research.
Since both cold-formed and hot-rolled steel types were tested, both The initial geometric imperfections account for the inconsistencies
gradual and sharp yielding stress-strain graphs were observed in the of member dimensions from its perfect geometry. These can be the
tensile coupon tests (see Figs. 3 (a) and (b)). The results show that the consequences of manufacturing, transport, storage and setting up

500

450

400

350
Stress (MPa)

300

250

200

150

100 Web 1.6mm - G250

Web 3.0mm - G250
50 0.2% line
Web 3.0mm - G350
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Strain
(a)

600

500

400
Stress (MPa)

300

200
75x25x2.5-RHS
0.2% line
100 65x35x2.5 RHS
0.2% line
75x25x1.6 RHS
0 0.2% line
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Strain
(b)

Fig. 3. Stress-strain curves from tensile coupon tests (a) web plates and (b) RHS flanges.
 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111 101

Table 3 produced during the welding process of RHSs to web plates. Since
Tensile coupon test results. both the moment capacity and the failure mode can be influenced by
RHS or web plate Average Yield Ultimate Modulus the local and global geometric imperfections of the beam, these imper-
thickness strength strength of elasticity fections were measured before testing the HFSPGs. A depth gauge was
(mm) (fy) (MPa) (fu) (MPa) (E) (GPa) used to measure the local imperfections while a dial gauge was used
65 × 35 × 2.5 RHS 2.45 448 495 200 to measure the global imperfections. However, the measured local and
2.46 450 495 199 global imperfections were found to be less than the allowable limits
75 × 25 × 2.5 RHS 2.48 452 502 199
(depth or width/150 for local and span/1000 for global imperfections).
2.46 450 499 200
75 × 25 × 1.6 RHS 1.56 449 512 200
1.58 449 511 197 2.4. Test set-up and procedure
Web plate 1 (1.6 mm) 1.58 285 407 200
1.59 284 408 200 The section moment capacities of HFSPGs were determined by
Web PLATE 2 (3.0 mm) 2.98 243 349 206
2.97 234 349 200
conducting four-point bending tests, which simulated a uniform bend-
Web Plate 3 (3.0 mm) 2.99 363 471 203 ing moment in the middle part of the girder. Similar test arrangements
3.00 364 469 196 have been used by many other researchers to test both singly symmetric
and doubly symmetric hollow flange sections and channel sections
processes. Although the initial geometric imperfections are often [1,3,5 and 19].
neglected, they have a considerable influence on the member behaviour Fully laterally restrained HFSPGs were tested to failure in these tests.
[17]. In the case of HFSPGs, the geometric imperfections are mainly The overall test set-up is illustrated in Fig. 4. Due to the four point

(a)

Spreader beam

Steel rollers Test specimen

Loading
Support device
box frame

Roller support

Lateral restraints
Deflection transducers

(b)

Fig. 4. Test set-up (a) Schematic diagram and (b) Overall view.
102 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111

bending arrangement, the region between the loading points had
a constant bending moment. Here the specimen length was sufficiently
short to prevent unintended global buckling, and also adequately
long to display local buckling without any effects from the
supports. The 1800 mm long HFSPGs were considered for the first
and second test series while the support to loading point and
the loading point to loading point distances were 575 mm and
550 mm, respectively. With a constant overhang of 50 mm at
each end, the test girders resulted in a span of 1700 mm. Here, the
distance between the loading points was selected to be less than the
distance between the support and the loading point to prevent shear
buckling.
Special support boxes and loading parts were used to apply the load
at the shear centre of the cross-section. As shown in Fig. 5 (a), the test
HFSPGs were supported by a box frame which consisted of two halves,
and were clamped to the web of the section using M12 high tensile
bolts. The box frame was then kept on half rounds to allow the section
to rotate about its major axis. Accordingly, it was considered that the
simply supported end condition was simulated. Lateral restraints were
provided near the loading points using steel sections clamped to the
base table as shown in Figs. 5 (a) and (b). This prevented the lateral
distortional and/or lateral torsional buckling modes. Frictionless bolts
were used between the lateral restraints and the HFSPG to allow the
test specimen to deflect freely in the vertical direction. 300 and
500 kN load cells were used in the first and second test series, respec-
tively, with a data recording system known as UDAQ Control V1.9. The
load was first applied at a constant rate of 1 mm/min to the spreader
beam, which then distributed it symmetrically to the two loading
points. Rollers were used on top of the loading arms (see Fig. 5 (c)) to
ensure the verticality of the applied load. Three linear displacement
transducers were positioned below the loading points and at mid-
span to measure the vertical deflections. Initially, the HFSPGs were
pre-loaded to about 20% of their expected failure loads and unloaded
several times. This method ensured the normal test conditions and the
proper functioning of the test set-up. Each specimen was then loaded
to failure.
These tests focused on the section moment capacities of the HFSPGs
and the test specimens were expected to fail by local buckling due to the
lateral restraints provided. As illustrated in Fig. 5, test specimens were
restrained near the loading points and at the supports. During Test 4

Test 85 mm
specimen
100 mm

Support box frame

(10 mm thick)
400 mm
Lateral restraints
at the support

6 mm thick web
stiffener plates

Roller
Support

(a)

Spreader
beam
Loading
location
Loading
via rollers

Test Test
specimen specimen

230 mm
Loading frame
Lateral
(10 mm thick)
restraint
100 mm

(b) (c)

Fig. 5. Test rig components (a) Support box frame, (b) Lateral restraints and (c) Loading via rollers.
 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111 103

of HFSPG-75 × 25 × 2.5-200× 1.6, slipping of the specimen was to 8), which had slender flanges. They were categorized as “S” (Slender)
observed at one of the end half-round supports due to inadequate or “Class 4” in Table 2. When comparing the failures of tested specimens
lateral restraints. Therefore, in the later tests more lateral restraints shown in Fig. 7, the deformations in some flanges are more visible and
were provided near both loading and support locations. However, severe. These failures were observed in girders with slender flanges or
during Tests 10, 11 and 2 of HFSPG-65 × 35 × 2.5-100 × 3 (G350 webs while the least buckling deformations were observed in the com-
web), HFSPG-65 × 35 × 2.5-150 × 1.6 (G250 web) and HFSPG-75 pact sections. Tondini and Morbioli [21] observed similar failure modes
× 25 × 2.5-100 × 3 (G350 web), lateral deformations were observed during their moment capacity tests of cold-formed steel rectangular
after the ultimate load was reached. Wang et al. [20] also noted this hollow flange beams. They tested two slender beams of 2 mm and
phenomenon in their investigation on the flexural and ductility behav- 3 mm thicknesses. Their beams were also of hybrid form with flanges
iour of hybrid high performance steel I-girders. In their investigation and webs made of different steel grades. Since they used slender
they used both three-point and four-point loading test set-ups with sections, they observed a sudden decrease after the ultimate bending
varying number of lateral supports. moment.
Another issue encountered during the section moment capacity Due to the different yield strengths of elements in hybrid HFSPGs,
tests was the premature shear web buckling failure before flexural elastic (first yield) (My) and plastic (Mp) moment capacities were
failure. This failure was observed for HFSPG-75 × 25 × 2.5-150 × 1.6 calculated by considering the different yield strength values of web and
(Test 3), which had a non-compact web. Two plates of 6 mm thickness flange elements. The typical stress distribution diagram used in these cal-
were bolted to the web of the HFSPG on either side between the loading culations is shown in Fig. 8. Note that initial yielding takes place at the ex-
point and the support to stiffen the web. This method prevented the treme web location except in HFSPGs 2, 6 and 10 since the web yield
web shear failures in the other HFSPGs with non-compact and slender strength of all other HFSPGs tested are considerably less than the flange
webs until the ultimate bending failure. yield strength (Table 4). Therefore, Myw (see Fig. 8) was determined
when yielding commenced at the extreme fibre of web element. How-
ever, Myw does not indicate the elastic moment capacity of flanges since
3. Test results and discussion they have not reached their yield point. Therefore, Myf was calculated
based on the attainment of elastic limit (first yield) in flanges. The corre-
All tested HSPGs failed mainly by local buckling of plate elements in sponding stress distribution is presented in Fig. 8. These elastic (first
compression. Since the verticality of the applied loads was maintained yield) and plastic moment capacities are given in Table 5 and are also
throughout by using the rollers on loading arms, the measured applied shown in Fig. 6 (a) to (g).
load was multiplied by 575 mm, the distance between the support and Comparison of calculated elastic and plastic moment capacities
the loading point to compute the applied bending moment. Here the of HFSPGs with experimental ultimate moment capacities in Table 5
applied load at one loading point was taken as equal to 0.5 times the and Fig. 6 provided many useful observations. HFSPGs 9 and 10
load reading from the data system. Table 4 displays the ultimate were classified as compact sections (Class 1 web and Class 2 flange)
moment capacities of the tested specimens while Figs. 6 (a) to (g) based on EN 1993-1-1 [13] while AS 4100 [10] classified them as
illustrate the typical moment versus deflection curves of seven HFSPGs non-compact sections (compact web and non-compact flange)
with varying slenderness values. (Table 2). Their capacities were slightly higher than their Mp values,
Fig. 7 shows the typical failure modes of the tested HFSPGs. The fail- which appears to confirm the EN 1993-1-1 [13] classification
ures of all the specimens were due to local buckling of compression method. The higher moment capacities beyond Mp values are likely to
flange within the uniform bending moment region (see Fig. 7). Post- be due to strain hardening effects. HFSPGs 11 and 12 had the same flanges
failure observations of HFSPGs 4 and 11 (Tests 4 and 11) showed as HFSPGs 9 and 10, but had a non-compact web (class 3) and a slender
some shear web buckling deformations in the region between the (class 4) web, respectively. However, their ultimate moment capacities
support and loading point. However, they appear to have not affected were closer to their Mp values, indicating the greater influence of compact
the flexural capacity results. Compression flange buckling occurred flanges. HFSPGs 1 and 2 had compact web and non-compact flanges and
closer to the loading points than towards the mid-point of the girder thus were expected to have a section moment capacity between My and
except for HFSPG-75 × 25 × 2.5-100 × 3 with G350 web plate Mp. This was confirmed in the case of HFSPG 1 while HFSPG 2 reached a
(Test 2) where the top compression flange buckled locally at the mid- capacity higher than Mp (strain hardening effects). Fig. 6 (a) to
span (see Fig. 7 (a)). For the girders with slender webs, web local buck- (c) exhibit a long plateau for HFSPGs 9, 11 and 1 as expected.
ling was also experienced following local buckling of compression HFSPG 4 had the same flanges as HFSPG 1 and a slender web,
flange (see Fig. 7 (c) and (d)). All the tested HFSPGs had compact or however, it reached a higher capacity than Myf. This observation again
non-compact flanges except in four girders (HFSPGs 5 to 8 in Tests 5 indicates the greater influence of compact flanges despite the presence
of slender web elements and the need to use Myf in the moment capacity
calculations. HFSPGs 5 and 6 had a compact web and slender flanges,
Table 4
however, they also reached a capacity closer to Myf, showing the
Ultimate moment capacities of HFSPGs from experiments.
beneficial role of compact web element. HFSPGs 7 and 8 had the same
Test/ Test specimen Flange Web Ultimate moment flanges as HFSPGs 5 and 6 but had a non-compact and a slender web,
HFSPG fy (MPa) fy (MPa) capacity (kNm)
respectively, hence could not reach Myf, particularly HFSPG 8 that had
1 HFSPG-75 × 25 × 2.5-100 × 3 451.0 238.5 27.41 slender web and flange elements. HFSPG 8 with a slender web reached
2 HFSPG-75 × 25 × 2.5-100 × 3 451.0 363.5 31.72 a moment capacity higher than Myw, indicating the need to use Myf in
3 HFSPG-75 × 25 × 2.5-150 × 1.6 451.0 284.5 ⁎25.72
the moment capacity calculations. Figs. 6 (d) to (g) exhibit the
4 HFSPG-75 × 25 × 2.5-200 × 1.6 451.0 284.5 50.64
5 HFSPG-75 × 25 × 1.6-100 × 3 449.0 238.5 16.23 moment-deflection behaviour of HFSPGs as their elements become
6 HFSPG-75 × 25 × 1.6-100 × 3 449.0 363.5 17.60 more slender. For HFSPG 8 with slender elements, the moment dropped
7 HFSPG-75 × 25 × 1.6-150 × 1.6 449.0 284.5 22.16 rapidly after reaching the ultimate moment with increasing deflection
8 HFSPG-75 × 25 × 1.6-200 × 1.6 449.0 284.5 28.29
(Fig. 6 (g)).
9 HFSPG-65 × 35 × 2.5-100 × 3 449.0 238.5 31.90
10 HFSPG-65 × 35 × 2.5-100 × 3 449.0 363.5 34.11 Fig. 9 depicts the moment versus deflection curves for HFSPGs with
11 HFSPG-65 × 35 × 2.5-150 × 1.6 449.0 284.5 42.90 the same geometric dimensions and flange steel grade but different
12 HFSPG-65 × 35 × 2.5-200 × 1.6 449.0 284.5 ⁎⁎51.75 web steel grades (G250 and G350). This comparison shows that the use
Note: “*”-Section failed in shear, “**”- Only the ultimate capacity was estimated based on of higher grade steel for web increases the ultimate moment capacity
observations (Full moment versus deflection curve was not recorded). by about 10%.
104 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111

35 Compact web and Non-Compact flange

30
Mp

Bending moment (kNm)

25
Myf
Myw
20

15
Theoretical

10 HFSPG-65x35x2.5-100x3

0
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0
Deflection (mm)
(a) HFSPG 9

Mp
40
Myf
35
Myw
Bending moment (kNm)

30
Non-Compact web and Non-Compact flange
25

20

15

10 Theoretical

5 HFSPG-65x35x2.5-150x1.6

0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
Deflection (mm)
(b) HFSPG 11

30
Mp
25
Bending moment (kNm)

Myf

20 Compact web and Non-Compact flange

Myw

15

10

5 HFSPG-75x25x2.5-100x3.0
Theoretical

0
0 10 20 30 40 50
Deflection (mm)
(c) HFSPG 1

Fig. 6. Moment versus vertical deflection curves for HFSPGs.

 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111 105

Mp
50
Myf

Bending moment (kNm)

40
Myw

30 Slender web and Non-Compact flange

20

10 HFSPG-75x25x2.5-200x1.6
Theoretical

0
0.0 10.0 20.0 30.0 40.0 50.0 60.0
Deflection (mm)
(d) HFSPG 4
20
Mp
18
16
Myf
Bending moment (kNm)

14
Myw
12
Compact web and Slender flange
10
8
6
HFSPG-75x25x1.6-100x3
4
Theoretical
2
0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Deflection (mm)
(e) HFSPG 5
30
Non-Compact web and Slender flange
25
Myf
Bending moment (kNm)

20
Myw

15

10

HFSPG-75x25x1.6-150x1.6
5 Theoretical

0
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0
Deflection (mm)
(f) HFSPG 7

Fig. 6 (continued).
106 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111

Myf
30
Slender web and Slender flange

Bending moment (kNm)

25
Myw
20

15

10
Theoretical
5
HFSPG-75x25x1.6-200x1.6
0
0 5 10 15 20 25 30 35 40
Deflection (mm)
(g) HFSPG 8

Fig. 6 (continued).

The discussions above have shown that the moment versus
deflection curves and ultimate moment capacities obtained for HFSPGs
confirmed the expected outcomes for compact, non-compact and slen-
der steel girders in general. In the moment versus deflection curves,
nonlinearity was observed before reaching the first yield moment.
This is likely to be due to the existence of initial geometric imperfections
and residual stresses in the test girders.
4. Comparison of proposed HFSPGs with industrial sections
The HFSPG is a new section which has not been used earlier in the
industry. However, due to the benefits it offers such as light weight,
enhanced flexural capacity and double symmetricity, these sections
could become an alternative to the commonly used I-sections (UBs
and WBs). In order to emphasize the importance of introducing this
section to the industry, a comparison was undertaken with the conven-
tionally used sections in the construction field.
In order to compare the two sections a suitable criterion is needed to
find an equivalent HFSPG to a particular UB or WB section. However, the
tested HFSPG sections are of smaller dimensions than that of WBs and
UBs as they were formed after considering the feasibility of conducting
the tests with respect to the loading capacities and space available in the
laboratory. Therefore, in this study, the section moment capacity per
unit area (Ms/Ag) was calculated to compare UBs and HFSPGs. WBs

Flange buckling

(b)
(a)
Flange and web buckling

(c)
(d)

Fig. 7. Observed failure modes in HFSPGs (a) flange buckling near the mid-span, (b) flange buckling near the loading points, (c) and (d) web and flange buckling.
 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111 107

σ < fyf σ = fyf σ = fyf

σ = fyw

σ = fyw σ = fyw

fyf ≥ fyw

Myw Myf Mp

Fig. 8. Stress distribution considered in calculating the yield and plastic moment capacities of HFSPGs.

were not considered here since the sizes of the tested HFSPGs are much
smaller than WB sizes. However, a comprehensive parametric study
could be used to compare larger HFSPGs with conventionally used WBs.
Table 6 presents the section moment capacity per unit area of
some of the tested HFSPGs with their gross cross-sectional areas
and outer dimensions. As shown in Table 6, the gross cross sectional
area and Bf/D ratio of the smallest UB are larger than those of tested
HFSPGs due to smaller thicknesses of HFSPGs. However, 150UB14.0
has a closer Ag (1780 mm2) for HFSPGs 1, 2, 9 and 10 (1218 mm2).
As seen in Table 6, Ms/Ag values of HFSPGs are higher than those
of UB sections, which, confirms the efficiency of tested HFSPGs over
conventionally used UBs in terms of their section moment capacity.
Also, the steel areas of HFSPGs are less than that of 150UB14.0, which
leads to an economical section. Comparison of capacities per unit area
can be considered as an important parameter, which gives an indication
of the cost efficiency of the section.
Due to the sizes of the tested HFSPGs, the above comparison
was limited to a few sections. However, a detailed comparison of the
capacity and cost efficiency of proposed HFSPGs with conventionally
used UBs and WBs will be presented following a detailed parametric
study.
5. Comparison with current steel design standards
In this section, the experimental capacity results were compared
with the section moment capacity predictions from the relevant design
equations given in AS 4100 [10], AS/NZS 4600 [11] or AISI S100 [12], EN
1993-1-1 [13] and EN 1993-1-3 [14]. Measured dimensions and yield
strengths were used in the moment capacity calculations, but the
rounded corners of RHS flanges were not considered.
5.1. AS 4100: 1998 – Australian standard for the design of steel structures
Clause 5.2.1 of AS 4100 [10] defines the section moment capacity Ms
as shown next.
Ms ¼ Z e f y ð1Þ
where, Ze = Effective section modulus and fy = yield strength.
The calculation of Ze depends on the section classification in AS 4100
[10]. Table 2 provides the classification of tested HFSPGs determined
using the measured dimensions and mechanical properties. The equa-
tions for Ze are given in AS 4100 based on the section slenderness as

18

16

14
Bending moment (kNm)

12

10

4 HFSPG-75x25x1.6-100x3.0-G250 web

2 HFSPG-75x25x1.6-100x3.0-G350 web

0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Deflection (mm)

Fig. 9. Moment versus vertical deflection curve for HFSPGs 5 and 6.

108 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111

Note: “*”-Section failed in shear-therefore not used, Ms – Section moment capacity, My – Elastic moment capacity, Mp – Plastic moment capacity, Mol – Elastic local buckling moment from Thin-wall. All the capacities are in kNm, “**”- Only the ultimate
follows;

EN 1993–1-3
λs ≤λsp : Compact section Z e ¼ S b 1:5Z ð2aÞ

1.03
0.99
1.22
1.28
1.16
1.11
1.15
0.09
1.15
1.33

1.10
1.11
1.18



–
λsy −λs
EN 1993–1-1 λsp b λs ≤λsy : Non−compact section Z e ¼ Z þ ðZ c −Z Þ 
λsy −λsp
ð2bÞ

1.22
1.11
1.13
0.08
1.14
1.32

1.10
1.11
1.18
1.03
0.99
1.07
1.11
–
2

λsy λsy
λsy b λs : Slender section Ze ¼ Z  or Z e ¼ Z 
AISI S100 DSM

λs λs
ð2cÞ
1.09
1.26

1.20
1.17
1.25
1.10
1.04
1.22
1.28
1.19
1.20
1.18
0.06
where λs is the section slenderness defined in terms of element clear
–

width, thickness and the yield strength (fy); Z and Zc are elastic and
AS/NZS 4600 – DSM

plastic section moduli, λsp and λsy are section plasticity and yield
slenderness limits given in Table 5.2 of AS 4100 [10].
sffiffiffiffiffiffiffiffiffi
b fy
λs ¼ ð3Þ
1.04
1.30
1.38
1.22
1.20
1.21
0.08
1.14
1.32

1.20
1.17
1.25
1.10

t 250
–
AS/NZS 4600 – EWM

In order to account for the different yield strengths of web and flange
Exp. Ms/design prediction

elements in hybrid HFSPGs, Z, Zc and Ze in Eqs. (2a)–(2c) were replaced

with Myf, Mp and Ms in calculating the section moment capacities of
capacity was estimated based on observations (Full moment versus deflection curve was not recorded). FLB – Flange local buckling, WLB –Web local buckling.

HFSPGs. The section moment capacities of the tested HFSPGs calculated

1.15
1.33

1.09
1.10
1.18
1.03

1.30
1.38
1.22
1.10
1.17
0.11
0.97

using AS 4100 [10] are given in Table 5, which also gives the elastic (first
–

yield) and plastic moment capacities of HFSPGs.

AS 4100

5.2. AS/NZS 4600:2005 – Australian/New Zealand standard for the design of

1.47

1.13
1.17
1.14
1.53
1.36
0.15
1.10
1.27

1.50
1.57
1.68

1.39
–

cold-formed steel structures

Experiment
Ms (kNm)

AS/NZS 4600 [11] uses the same equation in AS 4100 [10] (Eq. (1))
⁎⁎51.75
⁎25.72

34.11
42.90
27.41
31.72

50.64
16.23
17.60
22.16

31.90
28.29

for defining Ms, however, Ze is based on the first yield in the extreme
compression fibre, thus it excludes any inelastic reserve capacity.
FLB/WLB

AS/NZS 4600 uses both Effective Width Method (EWM) and Direct
Buckling
Element

Strength Method (DSM) to calculate the section moment capacities of

WLB
WLB
WLB
WLB
FLB
FLB

FLB
FLB
FLB

FLB
FLB

members subject to bending.

Mol (Thin-Wall)

5.2.1. EWM
33.78/34.04

In this method, Ze is calculated with the extreme compression or

(kNm)

tension fibre at fy. The effective width of each element has to be calcu-
65.77
65.33
74.49
54.76
17.33
17.42
24.95

84.94
85.21
83.27
59.39

lated with reference to Clause 2.2 of AS/NZS 4600 [11] to determine

Ze. The effective width (be) of uniformly compressed stiffened elements
Mp (kNm)

depends on the element slenderness λ (Eqs. (4a)–(4b)).

Comparison of section moment capacities of HFSPGs with available design guidelines.

30.84
28.25
29.07
39.60
52.40
18.54
19.67
26.31
34.75
29.76

41.05
53.92

For λ ≤ 0:673; be ¼ b ð4aÞ

Myf (kNm)

For λ N 0:673; be ¼ ρb ð4bÞ

24.14
24.09

16.11

35.03
47.07
34.89
47.06
15.89

23.12
31.15
24.56
24.67

Where; b is the element width excluding the radii and ρ is the

Myw (kNm)

effective width factor defined as,

19.03

29.24
37.26
12.45

19.42
24.61
22.06

32.39
40.30


0:22
–

1−
λ
HFSPG-65 × 35 × 2.5-150 × 1.6
HFSPG-65 × 35 × 2.5-200 × 1.6
HFSPG-75 × 25 × 2.5-150 × 1.6
HFSPG-75 × 25 × 2.5-200 × 1.6

HFSPG-75 × 25 × 1.6-150 × 1.6

HFSPG-75 × 25 × 1.6-200 × 1.6

ρ¼ ð5Þ
HFSPG-65 × 35 × 2.5-100 × 3
HFSPG-75 × 25 × 2.5-100 × 3
HFSPG-75 × 25 × 2.5-100 × 3

HFSPG-75 × 25 × 1.6-100 × 3

HFSPG-65 × 35 × 2.5-100 × 3

λ
HFSPG-75x25x1.6-100 × 3
Specimen designation

Unlike AS 4100 [10], the plate element slenderness (λ) depends on

the applied stress ( f ⁎) as shown in Eq. (6).
 rffiffiffiffiffi
1:052 b f
λ ¼ pffiffiffi ð6Þ
k t E

where, t is the plate thickness and E is the modulus of elasticity.

Test/HFSPG

Here, k, the plate buckling coefficient = 4 for stiffened compression

elements supported along each longitudinal edge. In the HFSPGs, all
Mean
Table 5

COV

the elements are stiffened with uniform or varying stress distributions.

10
11
12
7
8
9
1
2
3
4
5
6

For stiffened elements with a stress gradient, k is given by the following

 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111 109

Table 6
Comparison of HFSPGs with hot-rolled I-sections.

UB Ag (mm2) Ms (kNm) Ms/Ag HFSPG Ag (mm2) Ms (kNm) Ms/Ag % difference

(× 10−3 kNm/mm2) (× 10−3 kNm/mm2)

150UB14.0 1780 32.56 18.3 1 1218 27.41 22.50 22.90

2 1218 31.72 26.04 42.30
9 1218 31.90 26.19 43.10
10 1218 34.11 28.00 53.00

UB Bf (mm) D (mm) tf (mm) tw (mm) Bf/D HFSPG Bf (mm) D (mm) tf (mm) tw (mm) Bf/D

150UB14.0 75 150 7.00 5.00 0.50 1 75 150 2.50 3.00 0.50

2 75 150 2.50 3.00 0.50
9 65 170 2.50 3.00 0.38
10 65 170 2.50 3.00 0.38

equation. The section moment capacities of tested HFSPGs calculated using

this equation are given in Table 5.
k ¼ 4 þ 2ð1−ΨÞ3 þ 2ð1−ΨÞ ð7Þ ANSI/AISC 360-10 [22] the American specification for structural steel
buildings, provides suitable equations only for open sections, and hence
f it cannot be used in calculating the section moment capacities of
where, Ψ ¼ f 2, f ∗1 is compression (+) and f ∗2 can be either tension (−) or
1
HFSPGs.
compression (+).
The section moment capacities of HFSPGs were calculated
5.4. EN 1993-1-1 – Eurocode 3: design of steel structures – Part 1–1:
based on effective widths and the results are given in Table 5. Here
General rules and rules for buildings
also, they were determined in terms of moments using first principles
due to the non-uniform stress distribution of the section as shown in
EN 1993-1-1 [13] design equations are similar to the AS 4100 [10]
Fig. 8.
equations. It classifies the cross-sections into four classes depending
AISI S100 [12] also uses the same procedure as AS/NZS 4600 [11] in
on the width to thickness ratio of elements in compression (Clause
EWM to calculate the section moment capacities of flexural members,
6.2.5) and then provides the section moment capacity equations as
thus not described separately in this section.
follows.

5.2.2. Direct strength method (DSM) W pl f y

Section 7 of AS/NZS 4600 [11] provides the DSM based design Mc;Rd ¼ Mpl;Rd ¼ for Class 1 and 2 cross−sections ð10Þ
γM0
equations. Clause 7.2.2.3 of AS/NZS 4600 defines the nominal member
moment capacity for local buckling (Mbl) as;
W el; min f y
Mc;Rd ¼ Mel;Rd ¼ for Class 3 cross−sections ð11Þ
For λl ≤0:776; M bl ¼ Mbe ð8aÞ γM0

" W eff; min f y

 #
 Mc;Rd ¼
M ol 0:4 Mol 0:4
For λl N0:776; M bl ¼ 1−0:15 Mbe
Mbe Mbe
where Mol is the critical elastic local buckling moment of the section, Mbe
is the nominal member moment capacity for laterally restrained
members, which is equal to Myf and λl is a non-dimensional slenderness
(Mbe/Mol)0.5. Mol values of HFSPGs were determined using a finite strip
analysis program Thin-Wall and are given in Table 5.
The section moment capacities of HFSPGs were calculated using the
above equations and the results are given in Table 5.
5.3. AISI S100–2012: North American specification for the design of cold-
formed steel structural members
The EWM design method is the same as in AS/NZS 4600 [11] as
described earlier. However, the DSM design procedure was improved
to include the inelastic reserve capacity available in non-compact and
compact sections. Clause 1.2.2.1.2.1.2 of AISI S100 [12] defines the
inelastic reserve local buckling strength Mbl as follows.
For λl ≤ 0.776 and Mbe ≥My, sections symmetric about the axis of
bending or sections with first yield in compression,
!
1   − If the effective section modulus Weff is equal to the gross elastic
Mbl ¼ M y þ 1− Mp −My
C 2yl
for Class 4 cross−sections ð12Þ
γ M0
ð8bÞ
where, Wpl, Wel,min and Weff,min refer to the plastic section modulus,
minimum elastic section modulus and minimum effective section mod-
ulus, respectively. EN 1993-1-5 [23] provides the necessary calculations
to determine Weff,min using the effective widths allowing for local buck-
ling effects. γM0 is the partial factor of resistance, which can be taken as
1 for HFSPGs. Table 5 provides the section moment capacities of tested
HFSPGs calculated using the equations in EN 1993-1-1 [13].
5.5. EN 1993-1-3 – Eurocode 3: Design of steel structures – Part 1-3:
General rules – supplementary rules for cold-formed members and sheeting
Clause 6.1.4.1 of EN 1993-1-3 [14] provides the design bending
moment resistance (Mc,Rd) of a cross-section for bending about one
principal axis using the following equations.
− If the effective section modulus Weff is less than the gross elastic
section modulus Wel,
W eff f yb
Mc;Rd ¼ ð13Þ
γ M0
ð9Þ section modulus Wel,



qffiffiffiffiffiffi   λemax
M pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi f yb W el þ W pl −W el 4 1−
where, λl ¼ Moly , C yl ¼ 0:776=λl ≤ 3 and other terms are as defined λe0 W pl f yb
Mc;Rd ¼ ≤ ð14Þ
earlier. γM0 γ M0
110 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111
where, λemax is the slenderness of the section elements, which gives
the highest λe/λe0 value. For double supported plate elements λe =
qffiffiffiffiffiffi
λp ¼ b=tpffiffiffiffi where ε ¼ 235 and kσ is the buckling factor calculated
28:4ε kσ f y
in accordance with Table 4.1 of EN 1993-1-5 [23] based on Ψ value,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and λe0 = 0.5+ 0:25−0:055ð3 þ ΨÞ, where Ψ is the stress ratio.
For the calculations of the effective widths of plate elements, EN
1993-1-5 [23] is used. However, EN 1993-1-3 [14] considers the
inelastic reserve capacities of sections unlike AS/NZS 4600 [11].
The section moment capacities of tested HFSPGs were calculated
using the above equations and are given in Table 5. Similar to the
calculation procedure used in AS 4100 [10], Myf and Mp were used
in Eq. (14), instead of fybWel and fybWpl, respectively, to include the
different yield strengths of web and flange elements.
5.6. Comparisons
Table 5 presents the ultimate section moment capacities (Ms )
obtained by testing HFSPGs and by using the design equations
given in the steel design standards discussed in the previous sec-
tions. As concluded in Section 3, Myf should be used as the elastic mo-
ment capacity of hybrid HFSPGs as flanges play a vital role in their
flexural behaviour.
Cold-formed steel hollow sections are designed using AS 4100 [10]
and thus HFSPGs can also be designed using AS 4100. Comparisons
in Table 5 show that AS 4100 is able to predict the section moment
capacities adequately for compact and non-compact sections (HFSPGs
1, 2, 9-11) with an overall mean of 1.16 for the ratio of experimental
to predicted moment capacities. However, AS 4100 design equations
are unsuitable for HFSPGs with slender elements (overall mean of 1.52
for HFSPGs 4-8, 12). This is because of the use of a reduction factor
based on the worst slender element on the overall section modulus.
Although HFSPGs 5 and 6 have compact webs and reached higher
capacities, their overall section capacities are significantly reduced
by AS 4100 Eq. (2c) due to the presence of slender flanges. A similar
observation can also be made in the case of HFSPGs 4, 7, 8 and 12
with one or more slender elements.
Since RHS flange elements and web element can be subjected to
significant local buckling effects, AS/NZS 4600 [11] and AISI S100
[12] design equations can be more suitable and were used in this
study. As expected, Table 5 comparisons show that AS/NZS 4600
and AISI S100 EWM are able to predict the section moment capacities
adequately for slender sections (overall mean for HFSPGs 4-8, 12 is
1.08). However, since it does not include the inelastic reserve capac-
ity, the predictions are found to be over-conservative for compact
and non-compact sections with an overall mean of 1.28 for HFSPGs
1, 2, 9-11. In comparison to EN 1993-1-3 [14] predictions, AS/NZS
4600 equations are less conservative for HFSPGs with slender webs
due to the differences in the effective width calculations between
them.
The relatively new DSM in AS/NZS 4600 [11] does not include the
inelastic reserve capacity and limits the section moment capacity to
the first yield moment capacity (Myf ), and hence was found to be
over-conservative for compact and non-compact sections (overall
mean of 1.27 for HFSPGs 1, 2, 9-11). Although DSM equations are ad-
equate for slender HFSPGs as observed with AS/NZS 4600 EWM, they
under-estimate the section moment capacity of HFSPGs severely
when they have one element considerably more slender than others.
For example, the overall mean for the ratio of experimental to pre-
dicted moment capacities is 1.16 for HFSPGs 4-8 and 12 (worst
case with HFSPG 12), while this ratio was 1.08 for AS/NZS 4600
EWM. The use of lower Mol obtained for the whole section is the rea-
son for this.
AISI S100 [12] DSM equations include the inelastic reserve capacity
and thus are more suitable than AS/NZS 4600 [11] DSM equations for
compact and non-compact sections. Table 5 results show that AISI
S100 DSM predictions have improved in comparison with AS/NZS
4600 predictions for compact and non-compact sections (compare the
moment capacity ratios for HFSPGs 1, 2, 9-11). However, no section
could achieve the full plastic moment capacity with the use of AISI
S100 DSM equations and thus they were found to be over-
conservative for class 2 (compact) HFSPGs 9 and 10.
EN 1993-1-3 [14] EWM design equations are similar to AS/NZS 4600
[11] equations for slender sections, and hence their predictions are
similar to those of AS/NZS 4600 as seen in Table 5. However, the inclu-
sion of inelastic reserve capacity to a limited extent helps to improve the
comparison for compact and non-compact sections. For example, their
predictions for HFSPGs 9 to 11 were improved in comparison to AS/
NZS 4600 equations and were similar to those of AISI S100 [12] DSM
equations, although still less than the more accurate AS 4100 [10]
predictions. This is evident when the moment capacity ratios of HFSPGs
9 to 11 in Table 5 are compared.
EN 1993-1-1 [13] and EN 1993-1-3 [14] design equations for
slender sections are the same as they both refer to EN 1993-1-5
[23] for effective width calculations. However, the former gives Mp
for compact sections (classes 1 and 2) and My for non-compact sec-
tions (class 3). Hence it is an improvement over EN 1993-1-3 [14]
for compact sections. However, it limits the capacity of class 3 sec-
tions to My, and thus it is over-conservative in comparison with AS
4100 [10] and even EN 1993-1-3 design equations that allow some
inelastic reserve capacity for class 3 sections. Despite this limitation,
EN 1993-1-1 design equations are easier to use. Unlike with AS 4100
that classifies HFSPGs 9 and 10 as non-compact sections, EN 1993-1-
1 design equations classify them as class 2 sections and give Mp as
their capacity.
When the experimental results of HFSPGs 5 and 6 are considered,
although they consists of slender flanges, due to the presence of
compact webs, they have reached ultimate capacities above their elastic
moment capacities. This capacity increment was observed in some
beams with slender and non-compact elements as well. But all the
design guideline considered, limit the section moment capacity of a
slender section below their elastic moment capacity.
Overall comparison of moment capacity results in Table 5 show that
all the steel design standards considered in this study are conservative
and thus safe to use for HFSPGs. AS 4100 [10] appears to under-
estimate the capacities more severely than other standards as reflected
by the overall mean value of experimental to predicted moment
capacities. However, AS 4100 gives the best estimates for compact and
non-compact HFSPGs. Hence it is recommended that AS 4100 is used
for compact and non-compact HFSPGs while AS/NZS 4600 [11] and
AISI S100 [12] EWM are used for slender sections. DSM can also be
used conservatively for slender sections. Its accuracy will improve if
the slenderness of elements does not vary significantly among them.
Similarly, EN 1993-1-1 [13] is also recommended for any type of
HFSPG section although it is over-conservative for class 3 sections.
Also, the section classification method given in EN 1993-1-1 appears
to correlate well with experimental ultimate moment capacities than
AS 4100. It is not known why EN 1993-1-1 does not allow a higher
value than Myf for class 3 sections while even EN 1993-1-3 [14] allows
it by using Eq. (14), which is similar to the method used for non-
compact sections in AS 4100 (Eq. (2b)).
Most of the current steel design standards are based on the struc-
tural behaviour of open I-sections or open cold-formed sections, and
this could be a reason for the conservative predictions of the enhanced
section moment capacities of HFSPGs. Since cold-formed RHS sections
are currently designed using hot-rolled steel design standards, it is rec-
ommended that AS 4100 [10] and EN 1993-1-1 [13] design equations
are used for HFSPGs. However, for slender HFSPGs, AS 4100 should
refer to AS/NZS 4600 [11] as is done in EN 1993-1-1. Further improve-
ments can be recommended to improve the accuracy of predicting
the section moment capacities of HFSPGs after conducting a detailed
parametric numerical study.
 N. Perera, M. Mahendran / Journal of Constructional Steel Research 148 (2018) 97–111
6. Conclusions
In recent years several forms of hollow flange sections have been
developed and used for some years. However, the production of most
of these sections has been terminated due to numerous reasons. This
paper has proposed a Hollow Flange Steel Plate Girder (HFSPG) section,
which is cost-effective and structurally efficient than the sections intro-
duced in the past. It has presented the section moment capacity tests
carried out to examine the flexural performance of HFSPGs.
Four point loading tests were carried out to measure the section
moment capacities of HFSPGs. Twelve HFSPGs were tested with a span
of 1800 mm, which included compact, non-compact and slender
sections. The failure of all the HFSPG specimens was by the local
buckling of plate elements in compression in the uniform moment
region. Bending moment versus mid-span vertical deflection graphs
are provided with the ultimate failure moments of the tested HFSPGs.
A brief comparison has been presented between the section
moments capacities of HFSPGs and commonly used UB sections. It
proved that HFSPGs have a significant potential to replace the industrial
UBs and WBs. The ultimate section moment capacities obtained from
the tests were compared with the capacities predicted by the relevant
design equations in the current steel design standards and discussed
in detail. This comparison showed that all the current steel design
standards gave conservative predictions in general and the level of
inaccuracy varied depending on the section type. This paper has made
suitable recommendations on the appropriate use of currently used
steel design standards for HFSPGs.
Acknowledgement
The authors acknowledge the financial support and the laboratory
facilities provided by the Queensland University of Technology to com-
plete this research study.
References
[1] T. Anapayan, M. Mahendran, Improved design rules for hollow flange sections sub-
ject to lateral distortional buckling, Thin-Walled Struct. 50 (2012) 128–140.
[2] T. Anapayan, M. Mahendran, D. Mahaarachchi, Section moment capacity tests of
LiteSteel beams, Thin-Walled Struct. 49 (2011) 502–512.
111
[3] K.S. Wanniarachchi, M. Mahendran, Experimental study of the section moment ca-
pacity of cold-formed and screw-fastened rectangular hollow flange beams, Thin-
Walled Struct. 119 (2017) (2017) 499–509.
[4] S. Jeyaragan, M. Mahendran, Experimental Investigation of the New Built-up
LiteSteel Beams, Proceedings of the 5th International Conference on Thin-walled
Structures, Gold Coast, Australia, 2008 433–442.
[5] R. Siahaan, M. Mahendran, P. Keerthan, Section moment capacity tests of rivet fas-
tened rectangular hollow flange channel beams, J. Constr. Steel Res. 125 (2016)
252–262.
[6] J. Dong, R. Sause, Flexural strength of tubular flange girders, J. Constr. Steel Res. 65
(2009) 622–630.
[7] M.R. Wimer, R. Sause, Rectangular tubular flange girders with corrugated and flat
webs, Lehigh University, ATLSS Report No. 04-18, 2004 Bethlehem, PA, USA.
[8] B.G. Kim, R. Sause, Lateral torsional buckling strength of tubular flange girders, J.
Struct. Eng. (2008) 902–910.
[9] M.F. Hassanein, N. Silvestre, Lateral-distortional buckling of hollow tubular flange
plate girders with slender unstiffened webs, Eng. Struct. 56 (2013) 572–584.
[10] Standards Australia, Australian Standard AS 4100 Steel Structures SydneyAustralia
1998.
[11] Standards Australia, Australia/New Zealand Standard AS/NZS 4600 Cold-Formed
Steel Structures, Sydney, Australia, 2005.
[12] AISI, North American Specification for the Design of Cold-Formed Steel Structural
Members, American Iron and Steel Institute, Washington, D.C., USA, 2012.
[13] European Committee for Standardization (ECS) Eurocode 3 EN 1993-1-1, Design of
Steel Structures – Part -1-1: General Rules and Rules for Buildings, Brussels, May
2005.
[14] European Committee for Standardization (ECS) Eurocode 3 EN 1993-1-3, Design of
Steel Structures, General Rules – Supplementary Rules for Cold-formed thin Gauge
Members and Sheeting, Brussels, Belgium, 2006.
[15] Standards Australia, Australian Standard Metallic Materials – Tensile Testing at Am-
bient Temperature, Sydney, Australia, 2007.
[16] Peter W. Key, Gregory J. Hancock, A theoretical investigation of the column behav-
iour of cold-formed square hollow sections, Thin-Walled Struct. 16 (1993) 31–64.
[17] L. Gardner, D.A. Nethercot, Numerical modelling of stainless steel structural compo-
nents – a consistent approach, J. Struct. Eng. 130 (2004) 1586–1601.
[18] M. Balarupan, Structural behaviour and design of cold-formed steel hollow section
columns under simulated fire conditions, PhD Thesis, Queensland University of
Technology, Brisbane, Australia, 2015.
[19] C. Pham, G. Hancock, Experimental investigation and direct strength design of high-
strength, complex C-sections in pure bending, J. Struct. Eng. 139 (2013) 1842–1852.
[20] Chun-Sheng Wang, Lan Duan, Y. Frank Chen, Shi-Chao Wang, Flexural behaviour
and ductility of hybrid high performance steel I-girders, J. Constr. Steel Res. 125
(2016) 1–14.
[21] N. Tondini, A. Morbioli, Cross-sectional flexural capacity of cold-formed laterally-
restrained steel rectangular hollow flange beams, Thin-Walled Struct. 95 (2015)
196–207.
[22] ANSI/AISC 360-10, Specification for Structural Steel Buildings, American Institute of
Steel Construction, 2010.
[23] European Committee for Standardization (ECS) Eurocode 3 EN 1993-1-5, Design of
Steel Structures – Part -1-5: Plated Structural Elements, Brussels, October 2006.