Recent development in cold-formed steel structures research

Krishanu Roy 1a*, Hieng Ho Lau 2b, *, Tina Ting Chui Huon 3c, James B.P. Lim 4a
aDepartment of Civil and Environmental Engineering, The University of Auckland, New Zealand
bFaculty of Engineering, Computing and Science, Swinburne University of Technology, Sarawak

Campus, Kuching, Sarawak, Malaysia

cFaculty of Engineering and Science, Curtin University Malaysia, Miri, Sarawak, Malaysia

*Corresponding authors Email: hhlau@swinburne.edu.my, kroy405@aucklanduni.ac.nz

Abstract
The use of cold-formed steel (CFS) in construction industry has become increasingly popular
over last few decades and the application of CFS channel sections as primary structural members is
common nowadays. Due the popularity of the back-to-back built-up CFS channel sections, many
research works are available in the literature. This paper presents some of the latest research and
results of collaborative works carried out by Swinburne University of Technology Sarawak Campus with
The University of Auckland, Curtin University, Malaysia Campus and University of Macau, which
includes study on collapse behaviour of a fire engineering designed single-storey cold-formed steel
building in severe fires, experimental and numerical work on cold-formed steel built-up columns and
also, experimental and finite element modelling work carried out for self-drilling screw connections for
high strength cold-formed steel. All the experimental works were carried out in Curtin University,
Malaysia Campus where a full-scale natural fire test was conducted to investigate the collapse
behaviour of a single storey cold-formed steel (CFS) building, in addition a total of 108 experimental
tests were conducted, of which 60 tests were for back-to-back built-up channels, 40 tests were for back-
to-back gapped built-up columns and the remaining 8 tests were conducted on face-to-face built-up
CFS box sections. Besides 25 experimental tests were also carried out on self-drilling screw
connections, to investigate the effect of screw patterns, number of screws, and screw spacing on the
connection strength. The finite element models as well as data analysis and design rules were
developed for the built-up columns, self-drilling screws in shear and collapse mechanism of single
storey CFS building in the University of Auckland.

Keywords: Cold-formed Steel, fire, finite element, built-up, self-drilling screw

⦁ Introduction
The use of cold-formed steel (CFS) in the construction industry has become increasingly popular
over the last few decades due to its many benefits which includes high strength to weight ratio, high
durability, termite resistance and etc. In Malaysia and East Asia, the use of CFS focused more on
single storey building which involves cold-formed steel wall frame systems. Thus, the structural
response of this type of CFS building in fire is of interest. Advancement in the CFS industry also led to
the introduction of built-up section to cater for the complex design requirement in the industry. As the
use of CFS structure increases, understanding the behaviour and performance of the CFS connection
becomes important. Screw connection is preferred over other types of connection for CFS because its
self-drilling ability simplifies the installation process of CFS structures. Researches on the issues
mentioned are still in progress. This paper presents some of the latest research and results of the
collaborative works carried out by the authors on single storey CFS building in fire, built-up column and
screw connection.

⦁ Collapse behaviour of a fire engineering designed single-storey cold-formed

steel building in severe fires
Single storey cold-formed steel (CFS) buildings are a popular form of construction in Malaysia
and South East Asia, where they are used for low-rise residential and light industrial construction. The
walls of these cold-formed steel buildings are normally constructed from cantilever ‘stud & track’ panels
and the roof of the building comprises cold-formed steel trusses pinned to the wall connection plates at
the top. This form of construction will be referred to as a cantilever wall/truss system throughout this
paper. Some of the advantages of cold-formed steel cantilever wall/truss system are light weight, ease
and precision in fabrication, speed of construction, non-combustibility and resistance to insect
infestation. With the growing use of such cold-formed steel wall frame systems, it has become
important to understand the structural response in fire.

1
2.1 Experimental investigations
A full-scale natural fire test on a building constructed entirely of cold-formed steel is described
with an 8 m span, height-to-eaves of 2.155 m, length of 10 m with trusses spacing at 2.5 m. The CFS
building structure tested had an opening factor of 0.027 m1/2 and subjected to a fire load density of 734
MJ/m2. Two adjacent walls were lined internally with fire resistant lining to achieve a structural fire
resistance of 30 minutes (R30) and the fire load was calculated to generate a structural fire severity of
30 minutes; i.e. matching the wall resistance. Thus, the two protected walls were expected to remain
vertical throughout the fire; the roof was expected to collapse first followed by pulling in of the
unprotected walls. The base connections were designed to support an overturning moment generated
by 0.5 kPa face loading on the wall during the fire, as is required by UK and New Zealand standard
design practice (see Fig. 1). The wall panels are assembled on-site using vertical studs fixed to the
bottom and the top tracks by self-drilling / self-tapping screws (Teks). The roof of the building
comprised cold-formed steel trusses pinned to the wall top plates, again connected through Tek
screws. The building was designed to achieve a specified performance based on a time equivalent
determination of structural fire severity assuming specified levels of roof venting and the actual
performance closely matched this. Further details can be found in Roy, Lim, Lau, Yong, Clifton,
Johnston, and Mei (2019).

Fig. 1: Photograph of the cold-formed steel building investigated

2.3 Collapse Behaviour

Fig. 2 illustrates the schematic collapse of central part of the structure, demonstrating
displacement of the frame with time from ignition. The collapse mode was asymmetric with initial
thermal expansion of the frame followed by an inward collapse at 622.5 °C, at a time of 21 min 30 sec
from ignition. The fire rated south wall, was stable and remained vertical during the fire to resist the 0.5
kPa load as the linings protected the wall, studs and base connections from high temperature. These
fire rated walls have prevented the outward collapse hence spread of the fire to the adjacent buildings.
The same resistance did not occur on the unlined north and east walls and the roof which was also
unlined and which collapsed during the fire as expected. This failure was governed by an inward snap-
through of wall studs at one-third of the wall height. The walls protected by gypsum did not collapse in
the end of the test. A layer of 15mm gypsum board protected the cold-formed steel stud in the South
wall from 45% rise in temperature and a time delay of 46%. The semi-rigid connections at the base of
these walls, which were designed to resist an overturning moment from 0.5 kPa applied to the wall
panel in either direction, were satisfactory in maintaining the walls upright after the roof collapsed.

Fig. 2: Deformations of cold-formed steel cantilever wall/truss system

2.4 Numerical investigation

A non-linear elasto-plastic finite element model as shown in Fig. 3 was developed using
ABAQUS (2018). The FEA model comprised a slice through the building containing two trusses and

2
their supporting walls to predict the collapse behaviour of the central part of a portal building remote
from the gable walls, based on recommendations from similar scope of modelling undertaken
previously (O’ Meagher et.al., 1990). The centre line dimensions of the cross section were used in the
FE model.
DS4 shell elements were used for heat transfer analysis as it provides temperature degree of
freedom. Across the length and width, a mesh size of 10 mm × 10 mm was used in the FE model,
based on the convergence of the model. Surface to surface contact was applied to all of the intersect
surfaces in FE model. In order to reduce the computational time, the tangential behaviour of contact
surfaces was defined as frictionless, while the normal behaviour of contact surfaces was defined as
“Hard Contact”. The screw connections in the modelling of CFS cantilever wall / truss system were
simplified by using “Fasteners Builder” with “Cartesian and Rotation” connector type to allow all the
relative displacements and rotations of the screw constraint. Initial imperfections were not included in
the FE model to avoid the complexity of the analysis. This would affect the ultimate load capacity at
ambient temperature but would not affect the failure temperature of cold-formed steel portal frames in
fire (Johnston, 2015, Johnston et al. (2014, 2015, 2016)). Johnston et. al (2016) work won the
Institution of Civil Engineers (ICE) Palmer Prize in 2017.
Thermal properties of cold-formed steel and gypsum boards are crucial for heat transfer analysis
and thermal-mechanical analysis. For the case of cold-formed steel, the specific heat capacity, thermal
expansion and conductivity were adopted from Eurocode 3 Part 1-2 (2005). The stress-strain curve
obtained from the tensile coupon tests at ambient temperature were converted to stress-strain curves at
elevated temperatures using empirical equations proposed by Kankanamge and Mahendran (2011).
The value of Young’s modulus used in this study was 200.4 GPa; whereas, a poisson ratio of 0.3 was
used in this study (Gunalan, 2011, Eurocode 3, 2005). The density of cold-formed steel used was
considered as 7850 kg/m3. As for gypsum board, the density loss, specific heat capacity and thermal
conductivity was determined from the laboratory tests using thermogravimetric analysis and differential
scanning calorimetry analysis. The thermal conductivity curve used for the FE modelling was
determined using the proposed equation by Rahmanian and Wang (2012). The values of Young’s
modulus and thermal expansion coefficient were adopted from Cramer et al. (2003). Poisson ratio of
the gypsum board was taken as 0.2 (Rahmanian and Wang, 2012).
Thermal boundary condition and mechanical boundary condition are used in this model. Thermal
boundary condition involved the input values of heat conduction, boundary convection, and boundary
radiation; whereby, mechanical boundary condition involved the input of base fixity condition for the
CFS cantilever wall / truss system. Two types of thermal boundary conditions were adopted in this
study. For Condition A, cold-formed steel was protected by a layer of gypsum board. The hot surface of
gypsum board was considered as exposed surface, while the remaining surfaces were regarded as
ambient surfaces. For Condition B, cold-formed steel was directly exposed to fire. Hence all the
surfaces of cold-formed steel were considered as exposed surfaces. The coefficient of convection at
exposed and ambient surfaces was used as 25W/m2 and 10W/m2, respectively in the FE model. The
radiation emissivity value considered in the FE model for exposed and ambient surfaces was 0.8.
Boundary conditions were used to define the base fixity of the model at the initial state. As shown in
Fig. 9, pinned support (U1=U2=U3=0) condition was applied on the bottom trucks of the FE model.

Fig. 3. FE model

2.5 Analysis procedure

Heat transfer analysis and thermal-mechanical analysis of cold-formed steel truss were carried
out to model the collapse behaviour of cantilever wall / truss system under fire loading.
In heat transfer analysis, the rate of thermal energy flow through a surface per unit of time with
the consideration of heat radiation and convection, known as heat flux, is calculated using Eqs. 1 and 2.
(1)

(2)
where,
R is thermal radiation constant; r is the radiation emissivity; is known as the Stefan Boltzmann
constant (5.68 x 10-8 W/m2K4); T is the surface temperature; T0 indicates the absolute zero
temperature; Ts indicates the sink temperature and h is denoted as the thermal convection coefficient.

3
 The sink temperature for unexposed surface or ambient surface used was 20 °C. In case of fire
exposed surface, the sink temperature followed the temperature-time curve for roof, northern and
southern wall recorded in the full-scale fire test. It is important to note that the temperature against time
curve recorded in the full-scale fire test are intended to validate the structural response of the CFS
building structure.
Thermal stress analysis was performed to predict the failure mechanism, time and temperature
of the CFS cantilever wall/truss system. Mesh type of S4R was selected for thermal stress analysis.
S4R can analyse rotations, local buckling and torsion which is suitable to be used in this study. Two
steps were used to perform thermal stress analysis. The first step was loading step to simulate
permanent loads while the second step was temperature step. All of the applied loads in step 1 were
propagated to step 2. Dynamic implicit incorporated with quasi-static solver was used in step 2 because
it can handle the instabilities of structure and able to capture the snap-through effect of cold-formed
steel members at elevated temperatures. Quasi-static non-linear analysis with time period of 4800 was
used in the dynamic implicit analysis. Maximum number of increments were 1000000 with initial
increments of 0.0001 and maximum increment size of 10 was used in the finite element analysis.
2.6 Validation of the Finite Element Model
The non-linear elasto-plastic finite element shell model for cold-formed steel cantilever wall/truss
system under elevated temperature is validated using the full-scale fire test results. The thermal and
structural performance of cold-formed steel northern and southern side walls were assessed using
temperature-displacement relationships.
According to the results of FEA, snap-through buckling was formed at approximately one third
from the top of the northern side wall with collapse temperature of 628.2℃ and 22 minutes while the
temperature of south wall was 87℃ and able to withstand more than 22 minutes of fire severity. In Figs.
4(a) and 4(b), the failure modes from FEA were compared against the experimental failure modes for
both northern and southern walls, respectively. The final collapse mechanism of FE model and the fire
test is shown in Fig. 5. The FE model predicted an asymmetrical collapse at 628.2℃ which was similar
to the asymmetrical collapse of full-scale fire test at 622.5℃ and 22 minutes. The collapse temperature
predicted by the numerical model is found to be similar to that of the full-scale test; therefore, the FE
model can be used by engineers to assist in the design of this type of cold-formed steel cantilever wall /
truss system in severe fire conditions.
The FE model developed was able to simulate the performance of cold-formed steel cantilever
wall/truss system at elevated temperatures. The FEA results were in good correlation with the full-scale
fire test results, in terms of failure modes, rotational displacement and collapse temperature. The
investigation also shows that gypsum board has significantly improved the thermal and structural
performance of cold-formed steel wall panel assemblies. The investigation presented herein, can form
the basis of a performance-based approach for the design of cold-formed steel cantilever wall/truss
system in fire boundary conditions.

(a) Northern wall (b) Southern wall

Fig. 4. Comparison of FEA and experimental failure modes

4
 Fig. 5. Final collapse

⦁ Cold-formed steel built-up columns

The gaining popularity in the use of built-up columns leads to complex design problems. The
complexity is due to the unique buckling characteristic of built-up columns under load, either buckling
as one single member or two individual members. In order to account for the unique buckling
behaviours, the AISI specifications highlighted two important criterions for the design of built up
columns: the modified slenderness ratio, and the minimum fastener requirement. However, the
modified slenderness ratio is adopted from hot-rolled steel design and the minimum fastener
requirement can be conservative. Thus, further investigation is necessary.
3.1 Experimental Investigation
The built-up lipped channels as shown in Fig. 6 were tested under compression for different column
lengths starting from stub (length of 300 mm) to slender (length of 2000 mm) columns. Thickness of the
CFS channels were 1.2 mm. Figure below shows the cross section of the built-up columns investigated
by the authors, to be referred to as B, BU and GBU respectively. Two sizes were tested for each cross-
section type with web width of 75mm and 90mm.

(a) Box Section (B) (b) Back-to-back Section (BU) (c) Gapped Back-to-back Section (GBU)
Fig. 6. Investigated Built-up Columns
In total, 8 box (B) specimens, 60 back-to-back (BU) specimens and 40 gapped back-to-back (GBU)
specimens were tested, covering column heights of (i) 0.5m and 1.5m for box specimens, and (ii) 0.3m,
0.5m, 1m, and 2m for other specimens. Type of the built-up section, fastener spacing, nominal
specimen length and test specimen number was coded by the specimen labelling. Tensile coupon tests
were conducted for longitudinal and transverse directional coupons, average values of the Modulus of
elasticity and yield stress were 207 N/mm2 and 560 N/mm2, respectively. All the built-up columns were
loaded with the help of a 500kN Universal Testing Machine (UTM). Prior to testing an LVDT with 0.11

5
mm accuracy was used to measure initial geometric imperfections present in the channel sections. The
imperfections are included in FEA models. Further information on box column can be found in Roy, K.,
Ting, T. C. H., Lau, H. H., and Lim, J. B. P. (2019a). Similar studies on built-up face-to-face column can
be found in Roy, K., Mohammadjani, C., and Lim, J. B. P. (2019).
3.2 Numerical Investigation
ABAQUS 6.14-2 was used to develop a finite element model for CFS built-up columns under
axial load (Fig. 7). Centre line dimensions were used for all FEA models. A linear 4-noded quadrilateral
thick shell element (S4R5), available in ABAQUS element library, was used to model the built-up
section. A mesh size of 5 mm × 5 mm was used for the convergence of the model. Fasteners between
two back-to-back un-lipped channels were modelled using MPC beam connector elements available in
the ABAQUS library. MPC beam connector elements were assigned a stress of 6210 MPa to
incorporate the stiffness of the fasteners. “Surface to surface” contact was used for modeling the
interaction between the webs of back-to-back un-lipped channels. The web of one un-lipped channel
was modeled as slave surface, while the web of other un-lipped channel section was considered as
master surface. Two types of finite element analysis were performed for buckling of built-up sections:
Eigen value analysis and load-displacement analysis. Eigen values of the built-up columns were
determined to model the geometric imperfections. A load displacement nonlinear analysis was, then,
carried out using RIKS algorithm available in the ABAQUS library. The geometric imperfections and
non-linear material properties were included in the FEA model. From this analysis, the failure loads,
buckling modes and load-axial shortenings are determined.

(a) FE mesh (b) FE boundary condition & loading (c) Initial imperfection contours from FEA
Fig. 7. FE Model for built-up channels

3.3 Modified Slenderness Ratio

Comparison of FEA and AISI & AS/NZS shows that AISI & AS/NZS standard is over-
conservative while predicting the axial capacity of B75 and B90 columns. For all screw configurations,
the FEA and test strengths were higher than the design strengths predicted in accordance with the AISI
& AS/NZS. It is also noticeable that increasing the number of screws from 5 to 10, makes the difference
between the design strengths and FEA strengths higher. Therefore, it can be concluded that for both
B75 and B90 series, modified slenderness approach, as prescribed in the AISI & AS/NZS are
conservative by 17% for all built-up box columns.
Similarly, for axially loaded back-to-back built-up CFS channel columns, comparison shows that
the AISI and AS/NZS standards are conservative to the FE results for short, intermediate and slender
columns, however for stub columns, the AISI and AS/NZS standards, are overdesign by 10%. AISI and
AS/NZ standards safely predicted the strength of back-to-back built-up CFS channel section columns
which were failed mainly by overall buckling (modified slenderness greater than 55), but are
unconservative for stub columns which failed mainly by local buckling (modified slenderness ratio is
less than 30).
The modified slenderness ratio may not be completely applicable for cold-formed steel built-up
columns. Thus, the current design requirements on the modified slenderness ratio for built-up columns
need to be further evaluated. Important parameters to the investigation include the spacing of
intermediate fasteners, the slenderness of the column’s cross section, and the gap in the gapped built-
up column.
3.4 Effect of Intermediate Fasteners Spacing
Investigation on the effect of fastener spacing on the axial capacity of CFS built-up box sections
involves a wide range of slenderness, covering stub to slender columns was considered. Two different
cross-sections i.e. B75 and B90 were considered in the parametric study. The parametric study was
further extended by considering three different number of fasteners: 3, 5 and 10. Length of the built-up
box columns was varied from 300 mm to 2000 mm at an increment of 100 mm. It is shown that as the
fastener spacing increased, axial capacity is reduced. This effect of fastener spacing is more obvious
for slender columns for both BU75 and BU90 when the largest spacing with intermediate fasteners only

6
at both ends of the column is used.
Similarly, capacity reduction with the increase of intermediate fastener spacing is also noticed in
the gapped built-up columns for both GBU75 and GBU90. Strength of the gapped built-up sections
reduced significantly for all columns beyond 1000 mm length for both GBU75 and GBU90 with two
intermediate link-channels. On the other hand, for GBU75 with no intermediate link-channel, significant
reduction in axial strength was observed for column length higher than 500 mm. This is because the
intermediate link-channels holds the individual back-to-back channels together.
There is a trend towards lower ultimate strength as fastener spacing increases. However, the
same effect increases when the fasteners are spaced further apart. This is not accounted for by the
design standard since the slenderness ratio is limited by restrictive and conservative maximum
allowable spacing. Further information on this research is available in Ting, Roy, Lau, and Lim (2018)
and Roy, K., Ting, T. C. H., Lau, H. H., and Lim, J. B. P. (2018b).
3.5 Effect of Thickness
The validated FE model is also used to investigate the effect of thickness on the axial strength of
built-up columns. BU75 was considered, covering a wide range of lengths, from stub to slender
columns. Spacing between the screws was kept constant to investigate the effect of only one parameter
(thickness). The considered spacing of screws for stub, short, intermediate and slender columns were
50 mm, 100 mm, 225 mm and 475 mm, respectively. Thickness of the channel sections were varied
from 0.75 mm to 2.5 mm at an increment of 0.05 mm.
From the parametric study results, the effect of thickness was observed on axial strength of
columns. It was found that the AISI and AS/NZ design standards underestimated the axial capacity for
short and intermediate columns with thickness higher than 1.55mm. Similar trend was found for slender
columns with thickness higher than 1.15mm. The difference in strength predicted from the FEA and
AISI and AS/NZ standards remained almost constant with increasing thicknesses for all other columns.
Further information on this research is available in Roy, Ting, Lau, and Lim (2018a).
3.6 Effect of Gap
Gapped back-to-back built-up column is similar to the usual back-to-back built-up columns but
has greater lateral stiffness than individual columns due to larger cross section. Current design
guidelines do not provide comprehensive methods to design such columns with complex cross section.
Comparing the experimental and finite element results with the AISI and AS/NZS shows that the design
standard is conservative by 53%. This is because the beneficial effect of the gap is ignored in the AISI
and AS/NZS. Results also show that the presence of gap in built-up column has increased the strength
of the column when the column length is approximately 700 mm. As expected, for the stockier and
slenderer sections, the increase in strength is less. Further information on this research is available in
Roy, Ting, Lau, and Lim (2018c).
3.7 Design Recommendations
The current design guidelines by the AISI (2016) and AS/NZ (2018) standards propose either
the Effective width method (EWM) or the Direct Strength Method (DSM) while determining the axial
strength of compression members. The Effective width method (EWM) is used in this study to calculate
the design strengths of built-up columns.
Comparison is made on the experimental and FE results with existing design equations for back-
to-back built-up CFS channel columns subjected to axial compression. It is found that the AISI and
AS/NZ standards is over conservative for intermediate and slender columns which failed through
flexural buckling. However, for stub columns, which failed through local buckling, AISI and AS/NZ
standards are overdesigned, which emphasize the need for improved design rules. Therefore, new
equations, Eqs. 3 and 4, for back-to-back built-up columns are proposed (Roy, Ting, Lau, Lim, 2019b).
Close agreement is achieved between the test and FEA results with the proposed design equations.
:
(3)
:
(4)
For gapped built-up column, comparison of experimental and finite element results shows that
design in accordance with the American Iron and Steel Institute (AISI) and Australian and New Zealand
Standards (AS/NZS) is very conservative. This is because the beneficial effect of the gap is ignored in
the AISI and AS/NZS. Johnston (1971) introduced a theoretical equation for the elastic critical bucking
load (P) to account for the gap. The theoretical equation is as shown in Eqs. 5 and 6 (Roy, Ting, Lau,
and Lim, 2018c). Solving Eq. 5 numerically for P, leads to a value for the non-dimensional slenderness,
λc,GAP.
(5)
where,

(6)
Using a modification to the non-dimensional slenderness, that considers the gap, results in the
design standards being 5% difference between the experimental and finite element results. Hence it is
7
recommended to use λc,GAP while calculating the axial strength of back-to-back gapped built-up CFS
columns.

⦁ Experiments and finite element modelling of self-drilling screw connections for

high strength cold-formed steel
The increasing use of cold-formed steel (CFS) sections as structural members leads to the
increase use of self-drilling screw connections. Self-drilling screws are essentially externally threaded
fasteners, as their name implies, are capable of drilling their own hole and form their own mating
threads. These screw connections are often used for structures such as portal frames and houses, as
well as roofing and walling systems, which involves multiple screws with varying configuration at
different location of the structure. There is a lack of knowledge on the performance and behaviour for
these screw connections. Therefore, the authors have investigate the effects of number of screw and
screw pattern on the behaviour of screw connections.
4.1 Experimental Investigation
A total of 25 self-drilling screw connections for G550 cold-formed steel were tested covering:
three pattern arrangement to study the effects of screw patterns, and one to four screws to study the
effects of number of screws. The test setup is as shown in Fig. 8. CFS sheets of 400mm x 80mm at
thickness of 1.0mm and 1.2mm were used connected by different screw type involving ASTeks A2S
12 - 14 x 20 HWF self-drilling screws along with gauge 14 screws of Tenn JX, ASTeks A3, and ASTeks
XT. Both ends of the test specimens were glued with packing shims of 60 mm long. All the test
specimens in this research are designed to meet both AISI (2016) and AS/NZS 4600 (2018)
specification on end and edge distance. 50 kN Instron 5900 machine is used for the experimental tests
on the standard lap-joint. The loading rate was kept as 0.5 mm/min for all test specimens. Four LVDTs
were used to measure the displacements of the test specimens from both the top and bottom of the
connections. Further information on this research is available in Roy, Lau, Ting, Masood, Kumar, and
Lim. (2019).

Fig. 8. Test Setup for Single Shear Lap Test

4.2 Numerical Investigation
A non-linear explicit dynamic finite element model as shown in Fig. 9 was developed which
includes a fracture criterion for the steel sheets. Coupon tests were performed to investigate the
material properties for both the screws and steel sheets. Yield strength of 547 MPa and ultimate stress
of 659 MPa, along with Young’s modulus of 208 GPa, was used in the numerical modelling of the steel
plates. For the screws, the material properties were also determined in the experimental tests and used
in the FEA models. Three dimensional eight noded solid elements with reduced integration point
(C3D8R) were used to model the screws and steel plates. Based on mesh sensitivity analysis, element
size of 1.0 mm by 1.0 mm was found to be appropriate for screws and 2.0 mm by 2.0 mm for the steel
plates. Mesh refinement was made around the screws and near the holes in the steel plates to
incorporate the fracture behaviour accurately. The finite element model includes 3D modelling of the
screws with both actual dimensions and half width screws using symmetry boundary conditions were
modelled which also included a ductile metal damage criterion based on fracture strain formulas. In the
ductile damage model, the stress triaxiality and strain rate were included (Choung, Shim, Song, 2012).
Surface to surface interaction was considered between two sheets along with screw heads and upper
steel plate, and screw threads and drilled hole. Interlocking between two steel plates and the screw
thread was modelled by using a friction coefficient. The finite element model could effectively simulate
the different failure modes including bearing, tilting, pull-out and shear failure of the screw fasteners,
which were similar to the experimental failure modes.

8
Fig. 9. FE Model for Screw Connection
4.3 Effect of Screw Patterns
Three different screw patterns were designed for the experiments, which include specimens with
three, four and five screws (Fig. 10). The connection strengths of the screw patterns for 3, 4, and 5
screws are slightly lesser when compared to the number of screws. Although the differential
percentages are small, the screw arrangement does contribute to some effects for the screw
connection strength.

Fig. 10. Test Specimens’ Screw Patterns

4.4 Effect of Number of Screws
Generally, the ultimate failure for the gauge 14 screw was pull-out failure. Some of the pull-out
failures were observed from the gauge 12 for 3 and 4 screws only. Hole elongation and sheet piling
behaviour was observed for both the screw gauges of 12 mm and 14 mm series and are shown in
Figure 14. All the screws which failed under the combination of tilting and bearing failures, they were
not influenced by the number of screws for both the steel thicknesses of 1.0 mm and 1.2 mm.
The strength of the screw connections increases with the increasing number of screws. For
gauge 14 screws of ASTeks A3, screw connections yielded at much higher ultimate loads compared to
the Tenn JX screws for both the single or double screw connections. There was no significant group
effect for gauge 12 series and only minor effect was observed for gauge 14 series. This is because the
low ductility steel has lower Fu/Fy ratio, and thus contributed to a lower stress distribution, which
eventually improved the performance of the screws within the steel plates. Majority of the test results for
one to four screws showed slightly lower connection strength for gauge 12 screws as compared to
gauge 14 screws. This is because gauge 12 and 14 screws had a nominal diameter of 5.35 mm and
6.30 mm, respectively.
5. Conclusion
A full-scale natural fire test on a single storey cold-formed steel building, designed to behave in
a specified way in a severe fire, both with and without gypsum board. A non-linear elasto-plastic FEA
model was also developed, which comprised a slice through the building containing two trusses and
their supporting walls to predict the collapse behaviour of the central part of a portal building remote
from the gable walls. The collapse temperature of the building from the fire test and that predicted using
the finite element shell model are 622.5 °C and 628.2 °C, respectively. The finite element model closely
matched the experimental behaviour. Therefore, the FE model can be useful in understanding and
predicting the behaviour of this type of building in severe fire conditions. Further research will
investigate the effect of initial imperfections and quantify actual base fixity. The first author is currently
investigating the effect of different material properties of CFS and gypsum board, thickness of primary
structural members and effect of joint stiffness though an extensive parametric study using the FE
model described in this paper. The aim of the parametric study is to develop a performance based
design method for CFS cantilever wall/truss system in fire boundary conditions, which can be used by
researchers and practicing engineers.
Current design guidelines for built-up columns are conservative due to the conservative
intermediate fastener spacing used in the calculation of modified slenderness ratio adopted from hot-
rolled steel design. The effects on axial capacity is investigated on several parameters including:
modified slenderness ratio, intermediate fastener spacing, thickness and gap between the individual
Channels. Generally, there is a trend towards lower ultimate strength as fastener spacing increases.

9
However, the same effect increases when the fasteners are spaced further apart. As for thickness, the
design standards underestimated the short and intermediate, and slender columns at thickness higher
than 1.55mm, and 1.15mm respectively. In terms of gapped built-up columns, the presence of gap has
increased the strength of the column. These effects are conservative or not accounted for by the AISI
and AS/NZS design standard. Based on the investigation, design recommendations were presented for
the design of back-to-back built-up columns with and without gap.
The performance of different types of screws connections in the steel plates was investigated in
terms of number of screws, group effect for more than one screw, and screw pattern. It was found that
the behaviour of the connection varies with different types of screws. The strength of the screw
connections increases with the increasing number of screws. The screw arrangement contributes some
effects to the screw connection strength. It was also observed from the experimental tests that the
screw connections experienced multiple screw failures in different sequence. The mixed failure modes
observed in the experimental tests include either single or combined modes of failures as tilting,
bearing, tearing of the steel plates, pullover and pull-out failure of the screws. Both the screw and plate
failures were observed in the experiments.
Acknowledgements

The authors would like to acknowledge:

⦁ ICE Research and Development Fund, and IStructE Research Fund for the financial support on fire
research;

⦁ Malaysia Ministry of Education (MoE) under the Fundamental Research Grant Scheme (FRGS),
FRGS/1/2018/TK01/CURTIN/01/1 for supporting the screw research; and

⦁ Ecosteel Sdn. Bhd., Kuching, Malaysia for the financial support and for sponsoring the test
specimens for these research works.

References
ABAQUS Analysis User’s Manual-Version 6.14-2. ABAQUS Inc., USA, (2018).
AISI, North American Cold-formed Steel Specification for the Design of Cold-formed Steel Structural
Members, AISI S100. 2016, American Iron and Steel Institute: Washington DC, U.S.A.
Australian/New Zealand Standard, Cold-Formed Steel Structures, AS/NZS 4600:2018. 2018, Standards
Australia/Standards New Zealand: Sydney, Australia.
Choung, J., Shim, C-S., Song, H-C., (2012) Estimation of failure strain of EH36 high strength marine
structural steel using average stress triaxiality. Marine Structures. 29(1) (2012), 1-21. 19
Cramer S.M., Friday O.M., White R.H., and Sriprutkiat G. (2003) Mechanical properties of gypsum
board at elevated temperatures. In Fire and materials 2003: 8th International Conference, January
2003, San Francisco, CA, USA. London: Inter science Communications Limited, c2003: pages
33-42.
European Committee for Standardization (CEN) (2005) Design of steel structures—Part 1-2: General
rules—Structural fire design. Eurocode 3.
Gunalan S (2011) Structural behaviour and design of cold-formed steel wall systems under fire
conditions. PhD Thesis, Queensland University of Technology.
Johnston, B.G. (1971) Spaced steel columns, Journal of the Structural Division, American Society of
Civil Engineers, Volume 97, Issue 5, Page 1465-1479.
Johnston, R.P.D. (2015) Investigation of cold-formed steel portal frames at ambient and elevated
temperatures. PhD thesis, Queen’s University Belfast, Belfast, UK.
Johnston RPD, Sonebi M, Lim JBP, Armstrong CG, Wrzesien AM , Abdelal G , Hu Y (2015) The
collapse behaviour of cold-formed steel portal frames at elevated temperatures, Journal of
Structural Fire Engineering 6 (2): 77–102.
Johnston RPD, Lim JBP, Lau HH, Xu Y, Sonebi M, Armstrong CG, Switzer C, Mei CC (2014) Cold-
formed steel portal frames in fire: Full-scale testing and finite element analysis, The Structural
Engineer 92(10): 44-50.
Johnston RPD (2015) Investigation of cold-formed steel portal frames at ambient and elevated
temperatures. PhD thesis, Queen’s University Belfast, Belfast, UK.
Johnston RPD, Lim JBP, Lau HH, Xu Y, Sonebi M, Armstrong CG, Mei CC (2016) Finite-element
investigation of cold-formed steel portal frames in fire, Proceedings of the Institution of Civil

10
 Engineers – Structures and Buildings 169(1): 3–19.
Kankanamge ND and Mahendran M (2011) Mechanical properties of cold-formed steels at elevated
temperatures. Thin-Walled Structures 49(1): 26–44.
O’ Meagher, A.J. et.al. (1990) Fire Protection of Steel Framing – AUBRCC Research Report AP25.
BHP Research and New Technology. Melbourne, Australia.
Pekoz, T. (1990) Design of cold-formed steel screw connection. Proceedings of 10th International
Specialty Conference on Cold-Formed Steel Structures. 23-24 October, 1990, St. Louis, Missouri,
USA.
Rahmanian I and Wang YC (2012) A combined experimental and numerical method for extracting
temperature-dependent thermal conductivity of gypsum boards. Construction and Building
Materials, 26(1), 707-722.
Roy, K., Lau H.H., Ting, T.C.H., Masood, R., Kumar, A. and Lim, J.B.P. (2019), Experiments and finite
element modelling of screw pattern of self-drilling screw connections for high strength cold-formed
steel, Thin-Walled Structures, Revised.
Roy, K., Lim, J. B. P., Lau, H. H., Yong, P. M., Clifton, G. C., Johnston, R. P. D., and Mei, C. C. (2019).
Collapse behaviour of a fire engineering designed single-storey cold-formed steel building in severe
fires. Thin-Walled Structures, 142, 340-357.
Roy, K., Mohammadjani, C., and Lim, J. B. P. (2019). Experimental and numerical investigation into the
behaviour of face-to-face built-up cold-formed steel channel sections under compression. Thin-
Walled Structures, 134, 291-309.
Roy, K., Ting, T. C. H., Lau, H. H., and Lim, J.B.P. (2018a). Effect of thickness on the behaviour of
axially loaded back-to-back coldformed steel built-up channel sections - Experimental and numerical
investigation. Structures, 16, 327-346.
Roy, K., Ting, T. C. H., Lau, H. H., and Lim, J. B. P. (2019a). Experimental and numerical investigations
on the axial capacity of cold-formed steel built-up box sections. Journal of Constructional Steel
Research, 160, 411-427.
Roy, K., Ting, T.C.H., Lau, H.H., and Lim, J.B.P. (2019b) Improved Design Rules on the Buckling
Behaviour of Axially Loaded Back-To-Back Cold-Formed Steel Built-up Channel Sections, 9th
International Conference on Steel and Aluminium Structures (ICSAS19) Bradford, UK, 3-5, July,
2019.
Roy, K., Ting, T. C. H., Lau, H. H., and Lim, J. B. P. (2018b). Nonlinear behavior of axially loaded back-
to-back built-up cold-formed steel un-lipped channel sections. Steel and Composite Structures, 28
(2), 233-250.
Roy, K., Ting, T. C. H., Lau, H. H., and Lim, J. B. P. (2018c). Nonlinear behaviour of back-to-back
gapped built-up cold-formed steel channel sections under compression. Journal of Constructional
Steel Research, 147, 257-276.
Ting, T. C. H., Roy, K., Lau, H. H., and Lim, J. B. P. (2018). Effect of screw spacing on behavior of
axially loaded back-to-back cold-formed steel built-up channel sections. Advances in Structural
Engineering, 21 (3), 474-487.

11