Thin-Walled Structures 41 (2003) 543570

www.elsevier.com/locate/tws

Structural behaviour of cold-formed thin-walled

short steel channel columns at elevated
temperatures. Part 1: experiments
M. Feng, Y.C. Wang , J.M. Davies
Manchester Centre for Civil and Construction Engineering, University of Manchester, Manchester, UK

Received 2 May 2002; accepted 6 December 2002

Abstract

Cold-formed thin-walled steel structures are increasingly being used as primary load bearing
members. However, there is a lack of study of their performance in fire. This paper presents
a detailed description of an experimental study of the axial strength of cold-formed thin-walled
channel sections under ambient and uniform high temperatures. The objectives of this study
are two-fold: to gain an insight into the physical behaviour and failure modes of this type of
structure and to provide some experimental results for detailed numerical studies. A total of
52 strength tests were carried out on short cold-formed lipped channels with and without
service holes and unlipped channels at ambient and various elevated temperatures. From these
experimental studies, it has been observed that the failure mode of two nominally identical
columns can be different even though the failure loads are close. Depending on the thickness
of a channel and the location of the service hole, perforation can have an important effect on
the strength of the channel, irrespective of the temperature. The companion paper will describe
the results of design calculations and numerical studies.
2003 Elsevier Science Ltd. All rights reserved.

Keywords: Thin-walled columns; Elevated temperature tests; Fire resistance; Perforated section; Cold-
formed steel; Strength tests; Local buckling; Distortional buckling

Corresponding author. Tel.: +44-161-2008968; Fax: +44-161-2004646.
E-mail address: yong.wang@man.ac.uk (Y.C. Wang).

0263-8231/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0263-8231(03)00002-8
544 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

1. Introduction

Cold-formed thin-walled steel members provide a high strength to weight ratio

and are also easy to construct when compared to thicker hot-rolled steel members.
Because of these, cold-formed thin-walled steel sections as load-bearing structural
components are widely used in domestic and industry buildings and in almost any
imaginable location since it was first applied in building construction in about 1850
in USA. Its applications and our understanding of its behaviour have been further
expanded in various fields by leaps and bounds since the first specification for cold-
formed steel design was issued by the American Iron and Steel Institute (AISI) in
1946. However, one of the main disadvantages of cold-formed thin-walled steel as
a structural material is its low fire resistance because of a combination of the high
thermal conductivity of steel and the high section factor of the structural member,
both leading to rapid steel temperature rises in fire. Nevertheless, despite increasing
use, studies of the fire performance of cold-formed thin-wall steel members are rare
and available information in the literature is scarce [1-8,12,14,16,19,20].
There are a number of important differences between the fire behaviour of cold-
formed thin-walled steel members and that of hot-rolled ones. Firstly, the mechanical
properties at elevated temperatures are different. Cold-forming steel has a strengthen-
ing effect at ambient temperature, but this strengthening effect is reduced at elevated
temperatures. Secondly, due to the nature of fire protection to thin-walled steel mem-
bers and the fact that thin-walled members are usually exposed to fire attack on one
side, temperature distributions in thin-walled steel members under fire attack are
highly non-uniform and different from those in hot-rolled members. Thirdly, thin-
walled steel members usually have more complex buckling behaviour, involving
local, distortional, global buckling and their interactions. In addition, for easy instal-
lation of electrical wires and plumbing systems, pre-punched service holes are often
introduced in the web of a thin-walled steel member. Although the effect of a web
opening in a column at ambient temperature has been investigated by many
researchers [9,11,15,17,18,21], the effect of perforation under high temperatures is
not clear and there are very few studies of the fire performance of cold-formed thin-
walled steel columns with holes. As a result, no design method is available for this
type of construction.
At present, fire resistant design of cold-formed thin-walled steel structures is often
based on the results of manufacturers standard fire tests. Not only is this expensive,
but it also limits flexibility of the designer. In order to develop rational and general
fire engineering design methods for cold-formed thin-walled steel structures, a study
of the behaviour of cold-formed thin-walled steel structures in fire has been recently
started in the Structures and Fire Research Group at the University of Manchester
with financial support from the EPSRC. This paper reports the results of a number
of tests on cold-formed thin-walled steel short columns with and without holes under
ambient and uniform high temperature conditions. The objectives of these experi-
ments are to study the failure modes of short thin-walled columns at elevated tem-
peratures and to provide data for checking and validating design and numerical pre-
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 545

diction methods for local and distortional buckling. The companion paper will
describe the results of design calculations and numerical studies.

2. Description of tests

2.1. Dimensions of test specimens

The ambient and elevated temperature tests were conducted in the fire-testing lab-
oratory of the University of Manchester. A total of 52 steady state tests were perfor-
med, of which 11 were on unlipped channels, 29 on lipped channels without a service
hole and 12 on lipped channels with a service hole. All the channels were pre-hot
dip galvanised to BS EN 10147 with a G275 coating. The steel grade is S350 GD+Z
with a minimum yield strength of 350 N/mm2. The results of tensile coupon tests
give much higher yield strengths. Two sets of lipped channels and one set of unlipped
channels were tested, in which the nominal width of the lipped channels and the
unlipped channels were 100 and 104 mm, respectively. The nominal thickness of
the two lipped channel sections and the one unlipped channel section were 1.2, 2.0
and 1.5 mm, respectively. The flange width was 54 and 56 mm for the lipped chan-
nels and 63 mm for the unlipped channel, respectively. The central service hole had

Fig. 1. Cross-Sections of unlipped channel and lipped channel with and without a service hole.
Table 1 546
Dimensions of test specimens

Type of section Temp. Name of Cross-section dimension Thickness t Length L

(C) specimens (mm) (mm)

b1 (mm) b2 (mm) b3 (mm) r (mm) d (mm)

Unlipped channel 104 63 1.5 Amb. Unlipa1 104.2 63.82 1.56 400
Unlipa2 103.8 63.68 1.57 398
Unlipc1 104.7 62.1 1.62 398
250 C Unlipc225 104.13 62.15 1.6 401.5
Unlipc325 103.89 62.59 1.6 401
400 C Unlipc440 104.91 62.2 1.6 401
Unlipc540 104.42 62.22 1.6 402.5
550 C Unlipd155 103.32 62.71 1.61 400
Unlipd255 100.76 53.23 1.59 399
700 C Unlipc670 104.09 62.5 1.59 400
Unlipb570 104.6 60.07 1.61 398.5
Lipped channel 100 54 15 1.2 Amb. Lip12a1 100.05 53.64 17.09 1.23 400.2
without hole Lip12a2 100.05 53.64 17.09 1.19 398
Lip12d5 99.29 53.29 16.05 1.195 402.5
Lip12b1 100.1 53.46 16.25 1.19 400
temperatures part 1

250 C Lip12b225 99.22 52.07 17.08 1.2 399

Lip12b325 100.27 53.46 17.21 1.19 400
Lip12d225 99.32 51.93 16.05 1.198 402
400 C Lip12c140 99.31 52.31 16.25 1.18 398
Lip12c240 100.56 51.25 17.31 1.18 397.5
550 C Lip12c355 99.96 52.32 16.56 1.18 399
M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Lip12c455 99.56 52.46 15.97 1.19 399.5

700 C Lip12d370 99.35 51.96 15.92 1.2 400
Lip12d670 100.31 52.44 15.75 1.2 402
Lip12e170 99.95 51.69 15.93 1.2 399
(continued on next page)
short steel channel columns at elevated
structural behavior of cold formed thin walled
Table 1 (continued)

Type of section Temp. Name of Cross-section dimension Thickness t Length L

(C) specimens (mm) (mm)

b1 (mm) b2 (mm) b3 (mm) r (mm) d (mm)

Lipped channel 100 56 15 2 with Amb. Lip2a1 100.08 54.82 16.32 1.97 397.8
hole Lip2a2 100.08 54.82 16.32 1.97 399.2
Lip2a5 99.8 54.64 16.82 1.97 399.8
Lip2b1 101.62 52.69 16.96 1.98 399
Lip2d1 100.17 53.92 15.5 1.96 401
250 C Lip2b225 100.63 54.02 15.54 1.995 398
Lip2b325 100.42 54.92 15.58 1.98 399.5
400 C Lip2c140 101.01 54.35 15.06 1.97 399.5
Lip2c240 100.98 55.2 15.2 1.98 399.5
550 C Lip2c355 100.87 54.95 15.5 1.99 399.5
Lip2a555 100.69 54.47 15.58 1.98 399.5
Lip2e155 100.78 54.36 16.15 1.98 399
700 C Lip2d270 100.55 53.51 15.72 2 400
Lip2d370 100.49 54.77 16.13 1.97 401
Lip2d470 100.43 54.87 15.33 1.97 401
Lipped channel 100 54 15 1.2 with Amb. Lip12a3 100.05 53.64 17.09 20 60 1.19 400.8
hole Lop12a4 100.05 53.64 17.09 19.97 59.94 1.19 400.4
400 C Lp12ch140 100.76 53.23 15.64 20 60 1.24 398
Lp12ch240 99.31 52.08 16.16 19.97 59.94 1.203 398.5
550 C Lp12bh155 99.43 51.92 16.15 19.95 59.95 1.188 399
Lp12bh255 99.87 51.04 16.17 20.00 59.88 1.19 399.00
Lipped channel 100 56 15 2 with Amb. Lip2a3 102.64 54.1 15.73 19.95 59.95 1.96 397.1
M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

hole Lip2a4 100.07 54.67 16.58 20 59.88 1.97 400.01

400 C Lp2bh140 101.97 54.96 16.63 20.00 59.98 2.00 399.70
Lp2bh240 101.27 55.43 16.14 19.98 59.86 1.99 396.70
550 C Lp2ch155 101.94 55.20 16.21 19.97 59.94 2.00 397.70
Lp2ch255 101.19 55.62 15.82 19.98 59.83 1.96 398.70
547
548 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

a nominal diameter of 40 mm and an overall length of 100 mm. The nominal test
length was 400 mm for all specimens. The reason of choosing such a length for the
short column test is to ensure that the column was short enough so that the influence
of overall buckling was minimised, but long enough so that the end effects could
be neglected. Two series of tests were performed for each lipped channel section,
one solid (without a service hole) and one with a service hole. Fig. 1 shows the test
specimens profile and Table 1 gives their measured dimensions.

2.2. Test rig and instrumentation

2.2.1. Ambient temperature tests

The ambient temperature tests were conducted in a purpose built rig. As shown
in Fig. 2, the test rig is made of a reaction frame and a loading jack. The test
specimen was fixed between the upper and lower supports, which were bolted to the
reaction frame. The support end plates were in direct contact with the test specimen,
allowing each individual plate of the test column to rotate out of plane but preventing
the in-plane rotation. As shown in Fig. 3, for the ambient temperature test, detailed
measurements of strains and deformations were taken at a number of locations of
the column. Four displacement transducers were used to record the longitudinal
movement of the column, two at the column top and two at the bottom. Displacement
transducers were also placed around the middle section of the column to measure

Fig. 2. Ambient temperature test rig.

 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 549

Fig. 3. Strain gauge and transducer locations for ambient tests.

550 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

lateral movements of the web and the flanges. A number of strain gauges were also
placed around the middle section to measure the longitudinal strains of the column.

2.2.2. High temperature tests

For high temperature tests, the test column and its supports were enclosed in an
electrically heated kiln, as shown in Fig. 4. Five thermocouples were used for solid
columns and six for columns with a service hole to measure the longitudinal distri-
bution of temperatures in the test column. To measure the column longitudinal dis-
placements, four displacement transducers were used, two at the top and two at the
bottom. These transducers were placed outside the kiln to avoid high temperature
damage. It is assumed that due to the high rigidity of the supporting blocks, relative
to the test specimen, the measured displacements were those of the test specimen.
To evaluate any rotation at the end of the test specimen, the displacement transducers
at each end were placed at two opposite corners of the supporting blocks. Due to
high cost and the perceived small benefit (see Fig. 12), no strain was measured in
the high temperature tests.

2.3. Test procedure

Each test specimen was placed between the top and bottom support blocks and
checked by eye to be square. However, due to inaccuracy in cutting and imprecision
in placing the specimen, it is inevitable that the specimen might not have been loaded
exactly in the axial direction and an initial bending moment might have been induced.

Fig. 4. High temperature test rig.

 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 551

During each ambient temperature test, the axial load was applied slowly using an
increment of 2.5 kN at the beginning and using smaller increments of about 1 kN
near the calculated failure load of the test specimen. After each load increment, the
measured data were recorded by a data logger.
The ambient temperature tests were performed in two stages. The first stage tests
were carried out independently in the ambient temperature test rig to compare the
performance of different types of columns. The second stage tests were performed
in the elevated temperature test rig to provide the basis of comparison for the elevated
temperature tests.
Elevated temperature tests were carried out at four temperature levels of 250, 400,
550 and 700 C. All elevated temperature tests were carried out under the steady
state condition. In other words, the kiln temperature was raised to the desired level
and then kept for about 20 min to ensure that the specimen reached uniform tempera-
ture. Before the specimen was heated, the specimen was made in contact with the
upper and lower support blocks. During the heating process, the specimen expanded
almost freely without any restraint to its thermal expansion because oil in the
hydraulic pump flowed back into the tank. Whenever any increase in the applied
load on the test column was detected, the oil pressure in the jack was manually
released to ensure that the specimen was stress free during heating. When the speci-
men temperature reached the desired level, axial compression was applied slowly
using an increment of 1 kN for end temperatures lower than 550 C and 0.5 kN for
tests at 700 C. The test was terminated when the applied load could not be main-
tained, indicating failure.

3. Auxiliary tests
3.1. Temperature distribution test
In order to check whether the test specimens would be uniformly heated, an
unloaded heating test was performed to monitor temperature distributions inside the

Fig. 5. Temperature distribution test results.

552 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 6. Positions of tensile coupons.

kiln and in the test specimen. Nine thermocouples were attached to a lipped channel
specimen at three cross-sections. Every cross-section had three thermocouples, one
at the centre of the web and two near the junctions of the lip and the flange. During
the temperature distribution test, the kiln temperature was raised to 200, 500 and
700 C and then hold at these temperatures for a period of time to allow the kiln
and specimen temperatures to reach equilibrium.
The results of the temperature distribution test showed that the temperature distri-
bution within the kiln was relatively even after 15 min although at the beginning,
the kilns temperature was about 25% higher than those measured by the specimen
and master thermocouples. The difference between different measurements was less
than 5%, as shown in Fig. 5.

3.2. Tensile coupon tests

To enable accurate modelling (see the companion paper), tensile coupon tests were
carried out to determine the mechanical properties of steel. These tests were carried

Fig. 7. Determining elastic modulus, yield stress and ultimate stress of steel.
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 553

Table 2
Mechanical properties from tensile coupon test results at ambient temperature

Test No. Mean yield stress Mean tensile Mean elastic modulus
(N/mm2) stress (N/mm2) (N/mm2)

Lipped 100 54 15 x1.2 410.58 526.02 186.95

Lipped 100 56 15 2 406.00 494.59 186.65
Unlipped 104 63 1.5 494.00 552.91 188.38

out at ambient temperature only. For each type of channel section, two samples were
taken from the web and one sample from each flange, not including the corner. Fig.
6 shows the positions of the tensile test coupons in the web and the flanges. From
the resulting stress-strain curves, the mechanical properties, including the elastic
modulus, the yield strength based on the 0.2% plastic strain and the ultimate stress
can be determined. The evaluation procedure is illustrated in Fig. 7. Table 2 gives
the average test results. The difference in test results among different coupons of
the same test specimen is small, less than 8%.

4. Observations of column behaviour and failure modes

4.1. Ambient temperature tests

4.1.1. Unlipped channel 104 63 1.5

Two tests were performed in the ambient temperature test rig (without the electri-
cal kiln) and one test in the high temperature test rig. Typical local buckling was
observed in all unlipped channel tests. Even though there was a notable difference
between the strength of the first two tests, both experienced three stages of behaviour.

Fig. 8. Failure modes of lipped channel 100 54 15 1.2 at ambient temperature.

554 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 9. Failure modes of 100 56 15 2 at ambient temperature.

Firstly, a local buckling wave began to emerge on one flange and post-buckling
started when the applied load was about 15 kN. Afterwards, local buckling wave
appeared on the other flange when the load was increased to 37.5 kN. When the
load reached 54.80 kN for test specimen unlip15a1 and 58.02 kN for unlip15a2, a
local buckling wave appeared in the web and the column load could not be increased
any further. For the ambient temperature test carried out in the kiln, the deflected
shape of the column could not be observed during the test. However, the final shape
of the failed column also displayed local buckling with some local crush near the
top end.

4.1.2. Lipped channel 100 54 15 1.2 without a service hole

Two duplicate tests were performed in the ambient test rig and two in the high
temperature test rig. For the two tests performed in the ambient temperature test rig,

Fig. 10. Failure modes of lipped channel 100 54 15 1.2 with a service hole at ambient temperature.
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 555

Fig. 11. Failure modes of lipped channel 100 54 15 1.2 after different temperatures.
556 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 11. Continued

although two different failure modes, one being a combination of local buckling and
flexural bending and the other a combination of local buckling and torsional buckling,
the measured failure loads were almost the same. When these columns failed, local
buckling could also be found at one end of the column. The reason for different
failure modes may be attributed to the effects of different initial imperfect boundary
conditions. For test specimen lip12a1, the lips were slightly higher than the web,
causing flexural bending about the minor axis. For test specimen lip12a2, one flange
was lower than the other inducing torsional buckling. The two tests performed in
the high temperature rig failed in a combination of local, distortional buckling and
bending. The failure loads are slightly lower than the other two ambient temperature
tests. Fig. 8 shows a comparison of the failed test samples.

4.1.3. Lipped channel 100 56 15 2 without a service hole

Four tests were performed, two replicate tests each in the two different test rigs.
Until the applied load reached the column failure load, there was very little detectable
lateral deformation. The final failure mode of the two tests performed in the ambient
temperature test rig was clearly distortional buckling and two distinct distortional
buckling waves could be seen in the two flanges as shown in Fig. 9. The final failure
mode of the two tests performed in the high temperature test rig was a combination
of predominantly distortional buckling and some flexural bending, with slightly lower
failure loads. It is possible that the initial imperfections induced some initial bending
moment in the test specimens, causing a different failure mode and reduced column
ailure loads.
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 557

Table 3
Failure modes and loads for solid sections

Section type Temperature Name of Test strength Observed failure mode

(C) specimens (kN)

Lipped Ambient Lip12a1 55.99 Local and flexible buckling

channel 100 Lip12a2 53.46 Local and torsion buckling
54 15 Lip12d5 53.86 Local, distortional buckling and bending
1.2 without Lip12b1 50.97 As above
hole 250 Lip12b225b 40.04 Local buckling & local crush
Lip12b325 53.16 Local, distortional buckling and bending
Lip12d225 47.86 As above
400 Lip12C140 45.75 Local buckling & bending
Lip12C240 47.01 As above
550 Lip12C355 23 Distortional buckling
Lip12C455 27 Local buckling & bending
700 Lip12D370a 5.52 Local, distortional buckling and bending
Lip12D670a 5.7 As above
Lip12e170 8.85 As above
Lipped Ambient Lip2a1c 110.21 Local buckling
channel 100 Lip2a2 129.16 Distortional buckling
56 15 Lip2a5 124.66 As above
2without Lip2b1 116.79 Local, distortional buckling and bending
hole Lip2d1 108.97 As above
250 Lip2b225 123.69 As above
Lip2b325 123.23 As above
400 Lip2C140 101.87 As above
Lip2C240 101.56 As above
550 Lip2C355 45.52 As above
Lip2a555 43.33 As above
Lip2e155 45.03 As above
700 Lip2D270a 12 As above
Lip2D370a 13.83 As above
Lip2D470 15.77 As above
Unlipped Ambient Unlip15a1 55 Local buckling
channel 104 Unlip15a2 58.02 Local buckling
63 1.5 Unlipc1 54.5 Local buckling & local crush
250 Unlipc225 49.58 As above
Unlipc325 50.77 As above
400 Unlipc440 45.7 As above
Unlipc540 44.67 As above
550 Unlipd155 27.07 As above
Unlipd255 28.38 As above
700 Unlipc670 9.28 As above
Unlipb570 9.21 As above

a
Results unreliable due to high temperature effect on load cell.
b
Faulty tests.
c
Faulty test due to movement of the reaction frame.
558 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Table 4
Failure modes and loads for sections with a service hole

Section type Temperature Name of Test Observed failure mode

(C) Specimens results
Pu (kN)

Lipped Amb Lip2a3 111.1 Local buckling

channel 100 Lip2a4 116.94 Local buckling
56 15 2 400 Lp2bh140 73.62 Distortional buckling and bending
with hole Lp2bh240 83.78 Distortional buckling and bending
550 Lp2ch155 48.01 Distortional buckling
Lp2ch255 47.99 Distortional buckling
Lipped Amb Lip12a3 54.86 Local buckling
channel 100 Lip12a4 53.48 Local buckling
54 15 1.2 400 Lp12ch140 39.93 Local buckling and bending
with hole Lp12ch240 43.02 Local buckling and bending
550 Lp12bh155 22.64 Local, distortional buckling and bending
Lp12bh255 25.17 Local, distortional buckling

4.1.4. Lipped channel 100 54 15 1.2 and 100 56 15 2 with a service

hole
Two replicate tests were performed in the ambient test rig for lipped channel
100 54 15 1.2 and lipped channel 100 56 15 2. Due to a shortage of
specimens, no ambient temperature tests were conducted in the high temperature test
rig. Distinct local buckling waves could be found near the service hole in either
lipped channel 100 54 15 1.2 or lipped channel 100 56 15 2 as shown
in Fig. 10. After failure, the edges of the perforated web in one test (lip12a3) of
lipped channel 100 54 15 1.2 moved in the same direction and in the other
test (lip12a4) moved in opposite directions. No other difference could be detected.

Fig. 12. Load-strain curves for lipped channel Lip12a1 under ambient temperature.
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 559

From the above discussions, it appears that due to different initial imperfections,
the failure mode of nominally identical thin-walled steel columns can be different.
However, as will be shown later, the column strength does not seem to be sensitive
to the column failure mode, reaching similar values for different buckling modes.

4.2. High temperature tests

Since all high temperature tests were carried out in the kiln, only the final
deformed shape of the column could be observed after removal from the kiln. From
the final shapes, it appears that the behaviour of columns under high temperatures
was similar to that at ambient temperature.
For the unlipped channels, the final shapes of the specimen indicate local buckling
with some local crush at the top of the test specimens. The degree of local crush

Fig. 13. Load-lateral deflection curves for lipped channel Lip12a1 at ambient temperature.
560 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 14. Load-axial deformation relationships for lipped channel Lip2a5 at ambient temperature.

deformation appears to be higher than that at ambient temperature. This is caused

by longer post failure deformation in the kiln.
For the 1.2-mm thick solid lipped channels, since the ambient temperature failure
mode involved local and distortional buckling and bending, it is difficult to detect
any marked difference in failure mode for the high temperature tests. However, it
does appear that at lower temperatures (400 C), the failure mode was predomi-

Fig. 15. Load-axial displacement curves for unlipped channel 104 63 1.5 at ambient temperature.
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 561

nantly local buckling. At higher temperatures (550 C), the failure mode was pre-
dominantly distortional buckling. Fig. 11 shows a comparison of deflected shapes
after different elevated temperatures.
For the 2-mm solid channel sections, the predominant failure mode was distor-
tional buckling, and in some cases, this was accompanied by some local buckling
and flexural bending about the minor axis due to initial imperfections.
Due to a lack of test specimens of lipped channels with a service hole, only four
tests of each size were conducted, two replicate tests at 400 and 550 C. For the
1.2-mm thick channels, the failure mode from the two replicate tests at the same
temperature was the same. The failure mode was local buckling at 400 C and a
combination of local and distortional buckling at 550 C. For the 2-mm thick channel
specimens, the failure mode was the same at different temperatures, predominantly
distortional buckling. Bending was also observed due to initial imperfect contacts at
the supports.
Tables 3 and 4 summarise the test results for sections with and without a service
hole respectively. In general, unlipped channels and the 1.2-mm thick lipped channels
failed mainly in local buckling. The 2-mm thick lipped channels failed predominantly
in distortional buckling. Because the test columns were short (400 mm), no global
buckling was observed in any test.

Fig. 16. Load-axial displacement curves for lipped channel 100 54 15 1.2 with and without a
service hole at ambient temperature.
562 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 17. Load-axial displacement curves for lipped channel 100 56 15 2 with and without a service
hole at ambient temperature.

5. Discussions of column test results

5.1. Load-deflection and load-strain behaviour

Fig. 12 shows the recorded load-strain relationships for one 1.2 mm solid lipped
channel. It can be seen that there are large variations in the longitudinal strains of
the test column as measured by different strain gauges. This highly non-uniform
strain distribution is due to flexibility of the thin-walled steel plates affected by initial
imperfections. Similar behaviour has also been observed for other columns. However,
due to unknown initial imperfections, it is very difficult to predict these strains with
any accuracy. Therefore, it was decided that in elevated temperature tests, strains
would not be recorded.
Fig. 13 shows the recorded load-lateral deflection relationships for one 1.2-mm
solid lipped channel. As with the behaviour of column strains in Fig. 12, there is a
large scatter of results in the column lateral deflections. Again, similar behaviour
has been observed for other columns. Due to the impracticality of measuring lateral
deflections in the high temperature setup with the electrical kiln in place, no measure-
ment of lateral deflection of the columns was taken for the elevated temperature tests.
Fig. 14 shows the recorded loadaxial deformation relationships for a 2-mm lipped
channel at ambient temperature at the four deformation measuring locations. There
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 563

is very little difference in the measured deformations by each pair of transducers at

each end of the column, indicating very little rotation of the column ends and good
contact between the column ends and the loading system. Other columns behaved
in the same way.
Figs. 1517 show the recorded loadaxial deformation relationships for ambient
temperature tests for unlipped channels, 1.2-mm thick lipped channels and 2-mm
thick lipped channels. Due to initial lack of fit in the column supporting system,
there was some large initial column deformation. After these initial movements, the
slopes of all curves are quite close to each other, indicating that the axial deformation
of the nominally identical columns followed the same behaviour. In the companion
paper, only this behaviour will be used to compare numerical predictions with the
test results. The fact that Figs. 16 and 17 show that there is very little difference in
the slopes of the loadaxial deformation curves for solid and perforated columns
indicates that the service hole had very little effect on the overall axial stiffness of
the column.
Figs. 1822 compare the loadaxial deformation relationships of the five different
channel sections at different temperatures. From these figures, it can be seen that
the effect of elevated temperature is to reduce the column stiffness as expected.

Fig. 18. Load-axial deformation relationships of unlipped channel 104 62 1.5 under different tem-
peratures.
564 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 19. Load-axial deformation relationships of lipped channel 100 54 15 1.2 without a service
hole under different temperatures.

Considering the high flexibility of thin-walled columns, the replicate tests at the same
temperature give similar trends in deflection and close results in strength, confirming
the repetibility of the elevated temperature test procedure. Attempts were made to
find the stiffness of the columns under elevated temperatures to check whether these
test results could be used to confirm the reported values of Youngs modulus of
cold-formed steel at elevated temperatures. However, this proved to be difficult
because most loadaxial deformation relationships show non-linear behaviour from
the start of loading and no predominantly straight line in the relationship could be
found. All elevated temperature tests show much greater deformation due to highly
non-linear stress-strain relationship of steel. As a result, the post-failure deformations
of elevated temperature tests were much higher than those of ambient temperature
tests. This explains the much more severely deformed shape of the failed samples
after elevated temperature tests. For unlipped channels, Fig. 18 clearly shows post-
buckling strength at all temperatures.

5.2. Strength

Tables 3 and 4 give the recorded failure loads for all test columns. It may be
found that despite differences in failure modes of some nominally identical columns,
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 565

Fig. 20. Load-axial deformation relationships of lipped channel 100 54 15 1.2 with a service hole
under different temperatures.

the strengths of replicate column tests are quite close. Fig. 23 plots changes in column
strengths, the column strength being normalized relative to the appropriate average
column strength at ambient temperature. It can be seen that with the exception of
the 2-mm thick solid lipped columns at 250 C, the effect of elevated temperature
is to reduce the column failure load. The curves for different types of columns are
reasonably close, in some cases, the difference in two identical columns of the same
type being almost as large as the total difference between different types of columns.
This indicates that it is possible to use a single curve to represent all these results.
The two replicate tests on the 2-mm thick lipped channel at 250 C give higher
column failure loads than those at ambient temperature. The reason for this is not
entirely clear. It may be that the low elevated temperature helped relieve the residual
stress in the cross-section due to cold-forming without significantly reducing the steel
strength, which can affect the thick section more seriously than the thin one.
Table 5 shows a comparison between the test results of lipped channels with and
without a service hole. For lipped channels 100 56 15 2, the load carrying
capacities of solid sections are about 10 and 20% higher and 5% lower than those
of the ones with a service hole at ambient temperature, 400 and 550 C, respectively.
The strengths of lipped channels 100 54 15 1.2 with a service hole are about
0, 10 and 5% lower than those without a service hole at the above temperatures.
The difference for the thinner section is smaller because the service hole is almost
entirely within the ineffective section of the solid section. For both sections, the
effect of a service hole at 400 C appears to be particularly large. This may be due
to the relatively slow reduction in the elastic modulus compared to the yield stress
566 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Fig. 21. Load-axial deformation relationships of lipped channel 100 56 15 2 without a service
hole under different temperatures.

of steel at this temperature, thus increasing the effective area of the web. At 550
C, the failure loads of lipped channels 100 56 15 2 with a service hole are
slightly higher than those without a service hole, and the failure mode for the perfor-
ated sections is distortional buckling and for the solid sections a combination of local
buckling, distortional buckling and bending due to initial imperfections. The initial
imperfections in the solid lipped channel induced an initial bending moment on the
solid specimen and is the main reason to give rise to the lower load carrying capacity.
Discarding the initial imperfections, it may be concluded that service holes can
have a significant effect on the load carrying capacity of a short column regardless
of the column temperature, this effect becoming more pronounced if at elevated
temperatures, the percentage of reduction in the Youngs modulus of steel is less
than that in its yield strength.

6. Conclusions

This paper has presented the results of an experimental investigation into the
strength of cold-formed thin-walled steel short columns under uniform high tempera-
tures. A total of 52 short column tests were conducted to assess their local and
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 567

Fig. 22. Load-axial deformation relationships of lipped channel 100 56 15 2 with a service hole
under different temperatures.

Fig. 23. Reduction of column load as a function of temperature.

568 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

Table 5
Comparison of test results of short columns with and without a service hole

Section T (C) With a service hole Without a service hole

type

Name of Failure Failure mode Name of Failure Failure mode

specimens load specimens load

Lipped Amb. Lip2a3 111.1 Local buckling Lip2a2 129.16 Distortional

channel buckling
100 56
15 2
Lip2a4 116.94 Local buckling Lip2a5 124.66 Distortional
buckling
400 Lp2bh140 73.62 Distortional Lip2c140 101.87 Local, distortional
buckling and buckling and
bending bending
Lp2bh240 83.78 Distortional Lip2c240 101.56 Local, distortional
buckling and buckling and
bending bending
550 Lp2ch155 48.01 Distortional Lip2c355 45.52 Local, distortional
buckling buckling and
bending
Lp2ch255 47.99 Distortional Lip2a555 43.33 Local, distortional
buckling buckling and
bending
Lipped Amb. Lip12a3 54.86 Local buckling Lip12a1 55.99 Local buckling and
channel flexible buckling
100 54
15 1.2
Lip12a4 53.48 Local buckling Lip12a2 53.46 Local buckling and
torsion buckling
400 Lp12ch140 39.93 Local buckling Lip12c140 45.75 Local buckling and
and bending bending
Lp12ch240 43.02 Local buckling Lip12c240 47.01 Local buckling and
and bending bending
550 Lp12bh155 22.64 Local, Lip12c355 23 Distortional
distortional buckling
buckling and
bending
Lp12bh255 25.17 Local, Lip12c455 27 Local buckling and
distortional bending
buckling

distortional buckling behaviour. From the experimental results, the following con-
clusions may be drawn:

Depending on initial imperfections, the failure mode of nominally identical col-

umns can be different. This applies to both ambient and elevated temperature tests.
 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570 569

Despite the difference in failure modes, the failure loads of nominally identical
columns were very close.
The service hole in lipped channels did not seem to affect the overall column
axial stiffness. However, it changes the column buckling mode and can have a
significant effect on the column load carrying capacity. Depending on the
reduction in the yield strength relative to that in the elastic modulus of steel at
elevated temperatures, the effect of holes can become much higher than that at
ambient temperature.

Acknowledgements

This research is sponsored by the UKs Engineering and Physical Science

Research Council (EPSRC) under grant GR/M56319. The test samples were provided
free of charge by METSEC. The authors would like to thank the technicians in the
Structures and Fire Research Group for help with the tests.

References

[1] Ala-Qutinen T, Myllyaki J. The local buckling of RHS members at elevated temperatures. Espoo
1995, Technical Research Centre of Finland, VTT Tiedotteita, Meddelanden, Research Nodes 1672,
37 pp. + appendix 42 pp. 1995.
[2] Alfawakhiri F, Sultan M. Fire resistance of loadbearing LSF assemblies. In: The 15th International
Specialty Conference on Cold-formed Steel Structures, St Louis, MO, USA. 2000.
[3] Alfawakhiri F, Sultan M. Numerical modelling of steel members subjected to severe thermal loads.
In: Proceedings of the Fire and Materials 2001 Conference. 2001. p. 48394.
[4] Gerlich JT. Design of loadbearing light steel frame walls for fire resistance, fire engineering research
report 95/3. New Zealand: School of Engineering, University of Canterbury, 1995.
[5] Klippstein KH. Preliminary study on the column-strength of cold-formed steel studs exposed to
elevated temperatures. Washington DC: American Iron and Steel Institute, 1978.
[6] Klippstein KH. Strength of cold-formed studs in fire tests. Washington DC: American Iron and Steel
Institute, 1980.
[7] Klippstein KH. Strength of cold-formed studs exposed to fire. Washington DC: American Iron and
Steel Institute, 1980.
[8] Lawson RM. Building design using cold formed steel sections: Fire protection. Publication P129.
The Steel Construction Institute, 1993.
[9] Loov R. Local buckling capacity of C-shaped cold-formed steel sections with punched webs. Can
J Civ Engrg, Ottawa 1984;11(1):17.
[10] Miller TH, Pekoz T. Unstiffened strip approach for perforated wall studs. J Struct Engrg, ASCE
1994;120(2):41021.
[11] Narayanan R, Chow FY. Ultimate capacity of uniaxially compressed perforated plates. Thin-Walled
Structures 1984;2:24164.
[12] Outinen J. Mechanical properties of structural steels at elevated temperature. Licentiate thesis, Hel-
sinki University of Technology, Espoo, 1999.
[13] Outinen J. A study for the development of the design of steel structures in fire condition. In: Proceed-
ings of the First International Workship on Structures in Fire, Copenhagen. 2000. p. 26981.
[14] Outinen J. Effect of high temperature on mechanical properties of cold-formed structural steel. In:
Puthli, Herion, editors. Tubular structures IX. 2001.
570 M. Feng et al. / Thin-Walled Structures 41 (2003) 543570

[15] Pu Y. Prediction of ultimate capacity of perforated lipped channels. Journal of Structural Engineering
1999;125:5104.
[16] Ranby A. Structural fire design of thin walled steel sections. J Construct Steel Res 1998;46:3034.
[17] Ritchie D, Rhodes J. Buckling and post-buckling behaviour of plates with holes. Aeronautical Quar-
terly 1975;26:28196.
[18] Sivakumaran KS. Local capacity of uniformly compressed cold-formed steel section with punched
web. Can J Civ Engrg, Ottawa 1987;14(4):5508.
[19] Sultan MA, Alfawakhiri F, Benichou N. A model for predicting heat transfer through insulated steel-
stud wall assemblies exposed to fire. In: Proceedings of the Fire and Materials 2001 International
Conference, San Francisco. 2001. p. 495506.
[20] Wang YC, Davies JM. Design of thin-walled steel channel columns in fire using Eurocode 3 Part
1.3. In: Proceedings of the First International Workship on Structures in Fire, Copenhagen. 2000.
p. 18193.
[21] Yu WW, Davies CS. Cold-formed steel members with perforated elements. J Struct Div ASCE
1973;99(10):206177.