Thin-Walled Structures 98 (2016) 154–168

Contents lists available at ScienceDirect

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

Review of the high temperature mechanical and thermal properties

of the steels used in cold formed steel structures – The case of the S280
Gd þ Z steel
Hélder D. Craveiro, João Paulo C. Rodrigues n, Aldina Santiago, Luís Laím
ISISE – Institute for Sustainability and Innovation in Structural Engineering, University of Coimbra, Portugal

art ic l e i nf o a b s t r a c t

Available online 2 July 2015 The accuracy of structural fire design of cold-formed steel structures depends on the available predictive
Keywords: models regarding mechanical and thermal properties. The deterioration of mechanical properties, such
Cold-formed steel structures as yield strength and modulus of elasticity, and the evolution of thermal properties, such as thermal
Steel conductivity, thermal elongation and specific heat, with temperature is a key issue in the assessment of
Mechanical properties the performance of cold-formed steel structural elements in fire. An experimental campaign was un-
Thermal properties dertaken to determine both mechanical and thermal properties of the S280GD þ Z steel used in cold-
formed steel building construction industry. Tensile coupon tests were carried out to determine the
mechanical properties of the S280 GD þZ steel with 2.5 mm thickness at temperatures ranging from 20
to 800 °C. Thermal properties were assessed using the Transient Plane Source (TPS) equipment. Test
results were compared with current design standards and with the ones available in the literature. The
proposed predictive equations based on the Ramberg–Osgood model show very good agreement with
the experimental results for temperatures beyond 300 °C. It was found that some predictive models
presented in the EN 1993-1-2:2005 regarding mechanical and thermal properties should be improved.
Hence new proposals are presented for both thermal and mechanical properties. Also a modified stress–
strain model based on the EN 1993-1-2:2005 formulation is proposed for the S280GD þZ steel.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction mechanical properties during cold forming, namely strain hard-

ening, strain aging and Bauschinger effect that refers to the fact
Due to the increasing use of cold-formed steel structural ele- that the longitudinal compression yield strength of the stretched
ments in building construction industry, the fire design of this steels is smaller than the longitudinal tension yield strength [1].
type of structures has become extremely important in the past few With temperature increase this additional cold forming strength is
years. Probably one of the most important aspects for an accurate gradually lost [2,3] and completely disappears above 500 °C.
structural fire design is the evolution of mechanical properties The currently available design codes such as BS 5950 Part 8 [4]
such as yield strength and modulus of elasticity with increasing and the EN 1993-1-2:2005 [5] provide reduction factors for me-
temperature. The deterioration of mechanical properties with in- chanical properties of cold-formed steels at elevated temperatures
creasing temperature results in a significant loss of load bearing but with some limitations. For instance the BS 5950 Part 8 [4] only
capacity of cold-formed steel structural elements. provides reduction factors for yield strengths corresponding to
Mechanical properties of cold-formed steels can be sig- 0.5%, 1.5% and 2.0% strain levels whereas the EN 1993-1-2:2005 [5]
nificantly different from those of the virgin steel sheet before provides the same reduction factors used for class 4 hot-rolled
forming and hot-rolled steels due to cold-forming procedure. The steels.
yield and tensile strengths are increased whereas the ductility Sidey and Teague [6] stated that the strength reduction of cold-
decreases [1] during these processes. The deformations imposed formed steels at elevated temperatures may be 10–20% higher
during fabrication process at the flat parts may be elastic whereas
than that of hot-rolled steels due to the metallurgical composition
at the corners the deformations are essentially plastic. Basically
and molecular surface effects [7,8]. Some relevant studies have
three phenomena are responsible for the changes in the
been conducted on the mechanical properties of cold-formed
steels at elevated temperatures for low and high strength
n
Corresponding author. Fax: þ351 239797242. cold-formed steels and with different thicknesses [2,3,7–13]. In the

http://dx.doi.org/10.1016/j.tws.2015.06.002
0263-8231/& 2015 Elsevier Ltd. All rights reserved.
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 155

Nomenclature k u,θ reduction factor for the ultimate strength of steel at

temperature θ
⁎
CFS cold-formed steel k u,θ reduction factor for the ultimate strength in relation to
Eθ slope of the linear elastic range for steel at tempera- the yield strength of steel at temperature θ
ture θ Δp required minimum distance between the edge of the
E20 modulus of elasticity of steel for ambient temperature sensor and the nearest sample boundary in the xy-
fy,20 yield strength of steel at ambient temperature plane
fy,θ yield strength of steel at temperature θ Δl/l relative elongation
fu,20 ultimate strength at ambient temperature θ temperature of steel
fu,θ ultimate strength at temperature θ considering the εus,θ plastic strain at the end of the uniform elongation (at
strain hardening maximum tension load, fu,θ) at temperature θ
fp,θ proportional limit stress of steel at temperature θ εr,θ strain at rupture at temperature θ
fθ stress at temperature θ εθ strain corresponding to a given stress, fθ, at tempera-
k thermal diffusivity ture θ
t time εy yield strain
ca specific heat of steel εy,θ yield strain at temperature θ
k y,θ reduction factor for the yield strength of steel at εp strain at the proportional limit
temperature θ εt,θ limiting strain for ultimate strength at temperature θ.
kE,θ reduction factor for the modulus of elasticity of steel εu,θ ultimate strain at temperature θ
at temperature θ nθ strain hardening coefficient
k p,θ reduction factor for the proportional limit of steel at β Ramberg–Osgood model parameter
temperature θ λ the thermal conductivity

great majority of these studies it was found that the yield strength accepted that within the time-scale of the accidental fires creep
and modulus of elasticity reduction factors available in the current may not be explicitly included, provided that the elevated stress–
design codes are not suitable for steels used in the cold-formed strain model presented in the EN 1993-1-2:2005 [5] is used
steel building construction industry [2,3,7–13]. Analysing the [19,21]. Some authors have already published some studies on the
available results in the literature it seems that the reduction fac- creep effect on steels at high temperatures [22–27]. Generally it
tors for the modulus of elasticity show little dependence on steel was found that for temperatures of 400 °C, for a time exposure of
grade [7,8] whereas the yield strength reduction factors are more 120 min and for a stress level of 0.9 fy,20 all tested steels present
dependent on steel grade. However the reduction factors obtained good creep resistance (creep strain below 0.5%) [22–27]. For in-
and the proposed equations still differ most likely due to the test stance for the S275JR [27] and ASTM A992 [25] steel the creep
method, strain rate, heating rate and material grade. Two test strain was about 0.5% and 0.25% respectively. At 500 °C, for the
methods are widely used, namely steady state and transient state ASTM A992 steel [25], for a stress level of 0.9 fy,20 and for a time
test method. Steady-state is more common and probably easier to exposure of 120 min creep strain is still reduced, around 0.3%,
conduct whereas the transient-state test method is considered whereas for the S275JR steel [27] and for a stress level of 0.8 fy,20
more realistic as it simulates a structural member under static the creep strain was about 4%. For temperatures above 600 °C it is
loading subjected to fire [13]. In the experimental investigation clear that creep effect may become more significant depending on
presented in this paper [14], the steady-state method was used. the steel tested and will increase with temperature increase. For
Steady-state tests are easier to conduct guaranteeing accurate data instance for the S275JR steel [27] it was observed that at 600 °C
acquisition. Consequently possible errors may be reduced during the steel was not creep resistant (creep strain of about 40% for a
the execution of the tensile tests and in the determination of the time exposure of about 80 min), whereas for the ASTM A992 steel
mechanical properties, since the stress–strain curves are plotted [25] at 800 °C for a stress level of 0.9 fy,20 and exposure time of
directly from the recorded data [2,3,7,10,12]. Moreover some re- 120 min the creep strain was about 25%.
searchers showed that the difference between the steady-state Additionally, the assessment of thermal properties such as
and transient-state methodologies was not very significant thermal conductivity, thermal elongation and specific heat capa-
[2,3,15]. For instance in the study conducted by Outinen [3,15] on city is extremely important for a better understanding of the be-
the S350GDþ Z steel the difference between both methodologies haviour of cold-formed steel structural elements in case of fire.
was on average about 5% in the temperature range from 300 to Thermal properties are crucial to understand heat transfer phe-
600 °C and almost identical for higher temperatures [2]. nomena and thermal deformations. However, so far, there is a lack
Steels subjected to elevated temperatures may also experience of research in the field of thermal properties of cold-formed steels
creep. The creep effect is time dependent and influenced by the at ambient and elevated temperatures. For instance some design
temperature exposure and applied load. The time dependent in- codes such as BS 5950-8 [4] and AS 4100-1998 [28] simply ignore
elastic deformation starts to occur when the temperature of the that thermal expansion coefficient is temperature dependent.
material exceeds 30–40% of the absolute melting temperature Some authors have already showed that the predicted thermal
[16,17]. According to some authors, the amount of creep may be elongation in the EN 1993-1-2:2005 [5] is conservative [10,15].
considered limited since steady state tests usually lasted less than Other thermal properties such as thermal conductivity and specific
one hour [2,7,8,18]. Despite the eventual reduced influence of heat may be determined using the Transient Plane Source method
creep in tensile tests, this does not mean that in real applications [29,30,31,32]. This equipment is widely used to measure thermal
creep should not be considered. However to include explicitly conductivity of bulk and slab specimens such as the ones that
creep strain in calculations is difficult and in most applications it is were tested in the scope of this research. Among the existing
simpler to use stress–strain relationships that include some transient techniques to determine thermal conductivity such as
amount of creep that might be expected [19,20,21]. It is currently hot wire and laser flash, the hot disk technique is probably the
156 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

faster and more accurate thermal conductivity technique [29].

This paper presents an experimental investigation [14] on the
mechanical and thermal properties of the S280GD þZ steel (hot
dip galvanized with zinc (275 g/m2) on both sides with 2.5 mm
thickness) both at ambient and elevated temperatures. A series of
thermal property tests were undertaken to assess thermal con-
ductivity and specific heat using the Transient Plane Source
equipment from Hot Disk [33] at various temperatures ranging
from 20 °C to 800 °C. Thermal elongation was also evaluated using
the high temperature extensometer Epsilon model 3548HI [34].
Tensile coupon tests using the steady state method were under-
taken to assess the mechanical properties of the S280GD þZ cold-
formed steel both at ambient and elevated temperatures ranging
from 20 °C to 800 °C. In the scope of this investigation the creep
effect was not explicitly studied. The experimental data was then
compared with the current design standards [4,5,28] and with the
results available in the literature [2,3,7,8,10,12,13,15]. Also the Fig. 6. Comparison between the experimental results obtained with the high
reduction factors determined for the S280GD þZ steel were com- temperature extensometer Epsilon HI3548 and the extensometer FLA-6-11.
pared with some experimental results available at University of
Coimbra for hot-rolled steels. Based on the obtained results, im- Eurotherm controller (6) (Fig. 1). A constant heating rate of 10 °C/
proved predictive equations were developed to determine the min was used in the tests.
yield strength and modulus of elasticity reduction factors and The high temperature extensometer HI3548 [34] was pre-
stress–strain curves for the S280GD þ Z steel. Regarding thermal viously calibrated; however, in the first tests strain gauges FLA-6-
properties predictive equations were also developed based on the 11 were also used in order to compare the obtained results using
obtained experimental results. both instruments assessing their accuracy. In Fig. 6 it is possible to
observe that the results obtained using both solutions are in very
good agreement ensuring the accuracy of the high temperature
2. Experimental investigation extensometer. The tests were undertaken using the steady state
test method. During the heating stage the specimen freely ex-
2.1. Mechanical properties at ambient and elevated temperatures panded; after stabilized at the desired temperature level and po-
sitioned the high temperature extensometer in the tensile coupon,
2.1.1. Test set-up the test was ready to start. The load was applied using displace-
Tensile coupon tests were conducted at both ambient and ment control with a rate of 0.15 mm/min corresponding to a strain
elevated temperatures, ranging from 20 °C to 800 °C, to determine rate of 0.002 min  1 within the strain rate limit set to 0.003 min  1
mechanical properties of the S280GD þZ steel with 2.5 mm according to the EN 10002-5 [35].
thickness. A universal testing machine Servosis model ME402/20
with 200 kN capacity (1) was used to perform these tests. To 2.1.2. Test specimens
measure the strains of the tensile test specimens, the high tem- Each test specimen was cut in the longitudinal direction of
perature extensometer Epsilon model HI3548 (2) [34], with a gage forming of the lipped channel S280GD þZ profile according to the
length of 50.8 mm and travel of 25.4 mm in tension, was used. In dimensions proposed in the EN 10002-1 [36] (Fig. 2).
these tests, conical tip alumina rods were used since these type of The chemical composition of the cold-formed steel is presented
rods are suitable for flat specimens. The high temperature ex- in Table 1.
tensometer is water-cooled using a constant temperature bath Two 14 mm holes were made at each end of the specimen to fix
system Caron 2050 series (3). Also a data acquisition system TML it in refractory steel pull rods. The use of two holes was decided in
model TDS602 (4) was used to monitor the temperatures of the order to avoid localized crushing.
furnace chamber and the specimen in test. The thermal action was After cutting and machining, each specimen was measured
applied using an electrical split tubular furnace (5) controlled by a using a micrometer at three different points of the gauge length

Fig. 1. Test set-up. (a) Global view. (b) Detail of the high temperature extensometer. (c) Electrical tubular split furnace and high temperature extensometer.
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 157

the thermal conductivity, the diffusivity measurement and the

specific heat capacity. It is a highly accurate technique that can be
used in a wide range of material types, with a wide thermal con-
ductivity range, with easy sample preparation and non-destructive
[29–32]. The sensor is an insulated nickel double spiral which is
used for both transient heating and precise temperature readings.
The insulation to the nickel spiral is provided by two thin sheets of
Fig. 2. Tensile test specimens dimensions.
Kapton, Mica, or Teflon. In this experimental campaign Kapton and
Mica sensors were used (Fig. 3). The Hot Disk ref. 4922 [33] Kapton
and Mica sensors, with 14.6 mm radius, were used for ambient and
Table 1 high temperature tests ranging from 100 to 800 °C, respectively.
Chemical composition of the S280GD þZ steel. To perform these tests two square sample pieces of S280GD þZ
steel were used for each test. The sensor was placed between the
C [%] Mn [%] Si [%] S [%] P [%]
two sample pieces to be tested. Then a constant small electrical
S280GD þZ 0.06 0.43 0.026 0.007 0.012 current was supplied to the sensor which is able to monitor the
temperature increase through resistance measurement. The tem-
perature increase due to the electrical current depends on the
region. These measured values were then used for the determi- thermal transport properties of the two sample pieces surround-
nation of the cross sectional area of the specimens. ing the sensor. Thermal transport properties of the test samples
are determined by monitoring the temperature increase over a
2.2. Thermal properties at ambient and elevated temperatures short period of time [29–33] usually ranging from 5 to 20 s in the
experimental tests undertaken. Due to the reduced thickness of
Thermal conductivity, specific heat and thermal elongation the test samples (2.5 mm), the slab module of the software was
were experimentally determined in this investigation. Thermal used. Knowing the sample thickness this method only requires
elongation was assessed using the same test set-up of the coupon measurement time and output power as input data. In Fig. 4 the
tests at elevated temperatures. A 100 mm  60 mm rectangular slab test layout for ambient temperatures using the Kapton in-
plate was cut and fixed in one end. The other end of the plate sulation sensor is presented. The slab sample holder designed for
could freely expand with temperature increase. Thermal de- use in combination with the slab software module was used as
formations were measured with the high temperature ex- recommended in order to minimise heat losses to the sample
tensometer HI3548 [34]. holder.
The Transient Plane Source technique was used to determine Thermal tests layout comprises a Carbolite furnace (1) with a

Fig. 3. Hot disk sensors used in this investigation. (a) Kapton sensor model 4922. (b) Mica sensor model 4922 [21].
158 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

Fig. 4. Ambient temperature sample support with slab sample holder setup.

elasticity was determined from the stress–strain curve based on

the tangent modulus of the initial elastic linear curve.
The obtained stress–strain curves at both ambient (20 °C) and
elevated temperatures (100, 200, 300, 400, 500, 600, 700 and
800 °C) are given in Fig. 7. It is worth to mention that 3 repetitions
were undertaken for each temperature level. The obtained curves
for 20 °C and 100 °C show a small yield plateau whereas for
temperatures beyond 200 °C the stress–strain curves show a gra-
Fig. 5. Thermal properties test layout. dual yielding behaviour.
In Table 2 the obtained results regarding the yield strength
(fy,θ), modulus of elasticity (Eθ), ultimate strength (fu,θ) and pro-
Eurotherm controller, a laptop with the Hot Disk Thermal Constant
portional limit (fp,θ) are presented for each valid experimental test
Analyser software (2), the Transient Plane Source equipment
undertaken at both ambient and elevated temperatures.
model TPS 2500S (3) and a datalogger TML TDS 601 (4) used to
Although the creep effect was not explicitly studied in the
monitor the gas and the cold-formed steel specimen temperatures.
scope of this investigation it is worth noting that creep may have
Fig. 5 presents the test layout.
some influence on the obtained results for some temperature le-
As previously mentioned the special software module (thin slab
module) was used since the tested specimens had reduced thick- vels. No studies were found on the creep effect for the S280GD þ Z
ness (2.5 mm). The traditional analysis is based on the assumption steel. Using the available results in the literature it may be possible
that the sample dimensions are infinite and consequently that the to provide an estimation (as a guidance) on the amount of creep in
sample boundaries do not influence the temperature increase this experimental investigation. Assuming a similar behavior be-
measured by the used sensor. In this type of test, the outside lat- tween the S280GD þ Z steel and the S275JR steel [27], with similar
eral surfaces of the sample slabs must be insulated by a material chemical compositions, it was found that at 600 °C for a time ex-
(ceramic wool) with low thermal conductivity in order to reduce posure of 30 min and a stress level of 0.5 fy,20 the creep strain was
heat losses of these surfaces during measurement. Other para- about 10%. This stress level is usually a maximum for the servi-
meter to take into account is the probing depth (Δp [mm]) which ceability conditions of real cold-formed steel structures. In order to
is defined as the distance from the sensor edge to the nearest free accurately assess the influence of creep for temperatures above
surface of the sample. It was found that if Δp ≥ 4kt the influence 500 °C, time exposure of 120 min and different stress levels above
of the sample size on the results is negligible [29]. For the thin slab 0.5 fy,20 experimental creep tests must be carried out. Finally it is
module the probing depth depends on the thermal diffusivity
(k [mm2/s]) and measuring time (t [s]) of the experiment.

3. Mechanical properties

3.1. Experimental results

Based on the stress–strain curves obtained from experimental

tests the yield strength was determined considering the 0.2% proof
stress method at ambient and elevated temperatures. At elevated
temperatures the 0.5%, 1.5% and 2.0% total strain method is also
used by other researchers [7,8].
In Fig. 6 the results of an ambient temperature test are de-
picted. The stress–strain curves obtained with strain gauges and
high temperature extensometer are presented as well as the
method for determining the yield strength based on the 0.2% proof Fig. 7. Stress–strain curves of the S280GD þZ steel both at ambient and elevated
stress and 0.5%, 1.5% and 2.0% total strain method. The modulus of temperatures ranging from 20 to 800 °C.
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 159

Table 2
Experimental results.

Test Temp. [ °C] Eθ [GPa] Ēθ [GPa] r [%] fy,θ [MPa] f¯y, θ [MPa] r [%] fu,θ [MPa] f¯u, θ [MPa] r [%] fp,θ [MPa] f¯p, θ [MPa] r [%]

CT_20-1 20 205.23 204.18 0.94 302.54 306.81 3.70 422.47 424.04 2.25 208.28 212.50 5.15
CT_20-2 203.90 308.92 426.61 210.97
CT_20-3 203.42 308.98 423.04 218.24

CT_100-1 100 195.52 200.11 4.14 296.83 295.24 5.20 416.10 415.49 4.88 204.93 209.25 5.23
CT_100-2 201.24 299.46 410.34 207.77
CT_100-3 203.57 289.43 420.04 215.06

CT_200-1 200 176.90 171.80 4.48 275.71 275.75 3.05 469.11 471.50 8.95 170.91 170.68 2.26
CT_200-2 168.48 272.73 463.98 172.81
CT_200-3 170.03 278.82 481.40 168.31

CT_300-1 300 144.73 143.59 6.41 212.57 223.57 9.52 407.70 397.41 10.58 132.10 136.37 3.86
CT_300-2 149.35 229.23 386.57 139.62
CT_300-3 136.68 228.90 397.97 137.39

CT_400-1 400 120.49 121.21 1.39 181.05 181.68 1.64 301.32 291.73 8.56 98.41 98.71 0.53
CT_400-2 120.34 183.55 284.86 99.32
CT_400-3 122.81 180.45 289.01 98.39

CT_500-1 500 84.07 84.54 0.95 113.64 113.67 3.15 170.97 165.85 4.81 78.50 80.25 1.62
CT_500-2 85.63 110.53 161.42 80.57
CT_500-3 83.92 116.83 165.17 81.69

CT_600-1 600 62.58 62.46 0.32 77.36 77.83 2.58 84.86 87.00 5.18 52.28 54.57 2.95
CT_600-2 62.71 80.62 92.91 53.53
CT_600-3 62.10 75.52 83.24 57.90

CT_700-1 700 22.11 22.21 0.33 38.31 36.34 1.95 46.09 50.48 4.61 18.15 19.26 1.06
CT_700-2 22.58 36.28 50.05 20.25
CT_700-3 21.95 34.42 55.29 19.38

CT_800-1 800 17.47 17.45 0.33 19.45 19.91 0.41 35.89 31.47 3.83 16.81 16.49 0.35
CT_800-2 17.12 20.03 29.16 16.12
CT_800-3 17.77 20.24 29.35 16.54

worth mentioning that the existent amount of creep is included in modulus of elasticity at elevated temperature (kE,θ ¼Eθ/E20), ulti-
the experimental results and also implicit in the proposed for- mate strength (ku,θ ¼fu,θ/fu,20 and ku,θn ¼fu,θ/fy,20) and proportional
mulations based on the stress–strain model presented in the EN limit (fp,θ/fy,20) for the S280GD þZ steel.
1993-1-2:2005 [5]. In Fig. 9b the reduction factors for the modulus of elasticity at
The observed failure modes in the tensile tests both at ambient elevated temperatures are depicted. Between 100 °C and 700 °C
and elevated temperatures ranging from 100 °C to 800 °C are the degradation of the reduction factors is almost linear. In Fig. 9c
presented in Fig. 8. the obtained reduction factors for the ultimate strength (fu,θ/fu,20)
and for the proportional limit (fp,θ/fy,20) are also presented.
3.2. Reduction factors Regarding the ultimate strength, it was observed a singular
behavior at 200 °C. As observed in the investigations performed by
The reduction factors for each mechanical property (ky,θ, kE,θ, ku, Ranawaka [7] and Kankanamge [8], for low strength steels the
θ, ku,θn and kp,θ) at elevated temperatures were determined as the ultimate strength was higher at 200 °C (E11%) than at ambient
ratio between the value obtained at elevated temperature to that temperature (20 °C). This behavior can be discussed in relation to
obtained at ambient temperature except to the proportional limit the chemical composition of the steel tested. Increasing the tem-
which is determined in relation to the yield strength at ambient perature up to 200 °C some chemical reactions and transforma-
temperature. In Tables 3 and 4 the yield strength, modulus of tions in the steel base may occur leading to an increase of the
elasticity, ultimate strength and proportional limit reduction fac- ultimate strength [7,8]. When the temperature grows beyond
tors are presented. The yield strength reduction factors (ky,θ ¼ fy,θ/ 200 °C it seems that these chemical reactions are retarded and as a
fy,20) are presented for different strain values namely, 0.2% proof consequence the ultimate strength reduces.
stress, 0.5%, 1.5% and 2% strain levels. Regarding the ductility of the cold-formed steel tested, it was
In Fig. 9a the reduction factors for yield strength are plotted. found that the ductility increased with increasing temperature
The reduction factors based on 0.5% total strain are relatively si- from 200 °C to 800 °C. From Fig. 7 it can be stated that the tested
milar to the ones based on 0.2% proof stress. It seems that for yield steel is ductile as expected since it has low levels of carbon.
strengths based on higher strain levels (1.5% and 2.0%) the corre-
spondent reduction factors were higher and also close to the re- 3.3. Benchmarking of test results
duction factors obtained for the ultimate strength for some tem-
perature levels up to 500 °C. Regarding this observation, it is clear Some researchers have already presented some relevant results
that the yield strength based on the 1.5% and 2.0% total strain regarding the reduction factors for mechanical properties of low
should not be used in the design procedure. and high strength steels used in cold-formed steel structures at
The modulus of elasticity also decreases with increasing tem- elevated temperatures [2,3,7–13]. In this section those results will
perature. Table 4 presents the determined reduction factors for the be compared with the ones obtained in this experimental research
160 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

Fig. 8. Failure modes observed in the tensile tests both at ambient and elevated temperatures ranging from 20 to 800 °C.

Table 3 (Fig. 10a–c). The results presented by Kankanamge [8] for low
Yield strength reduction factors for different levels of strain. strength steel G250 with 1.95 mm are in good agreement with the
ones obtained in this investigation for the yield strength reduction
Yield strength reduction factors – Ky,θ
factors. Regarding the modulus of elasticity the reported values are
T (°C) fy,θ/fy,20 also relatively similar but slightly unsafe for some temperature
levels despite the difference in the steel grade and thicknesses
0.20% 0.50% 1.50% 2.00% used. It is worth to mention that the same displacement and
20 1 1 1 1
heating rates were used in both investigations. Ranawaka [7] re-
100 0.962 0.961 0.966 1.000 sults for the low strength steel G250 are conservative for the yield
200 0.898 0.945 1.047 1.086 strength reduction factors. Regarding the modulus of elasticity the
300 0.728 0.768 0.846 0.870 results presented by Ranawaka [7] are slightly unsafe for the low
400 0.592 0.615 0.654 0.656
strength steel G250 when compared with the results reported in
500 0.370 0.396 0.424 0.427
600 0.253 0.258 0.246 0.248 this paper. The results provided by Outinen [3,13] for the S350GD
700 0.118 0.130 0.145 0.145 steel with 2.0 mm thickness are unconservative beyond 200 °C
800 0.064 0.074 0.086 0.087 both for yield strength and modulus of elasticity. Mecozzi [12]
results are very overconservative for yield strength as well as for
the modulus of elasticity reduction factors for temperatures above
Table 4 200 °C. The results presented by Ranawaka, Outinen, Mecozzi and
Elastic modulus, ultimate strength and proportional limit reduction factors. Chen [7,3,13,12,10] are not in agreement for the cold-formed steel
considered in this study. Therefore most equations available in the
Reduction factors – KE,θ; Ku,θ; Ku,θn; Kp,θ
literature are not accurate to determine the reduction factors for
T (°C) Eθ/E20 fu,θ/fu,20 fu,θ/fy,20 fp,θ/fy,20 yield strength and modulus of elasticity of the S280GD þZ steel
tested at elevated temperatures. Generally, observing the results
20 1 1 1.382 0.692 available in the literature high strength steels present higher re-
100 0.980 0.979 1.354 0.682
duction factors both for yield strength and elastic modulus. Also it
200 0.841 1.111 1.537 0.556
300 0.703 0.937 1.295 0.444 is clear that the results provided by all researchers are more uni-
400 0.593 0.687 0.951 0.321 form for the modulus of elasticity than for the yield strength. It
500 0.414 0.391 0.541 0.261 seems that the elastic modulus reduction factors are less depen-
600 0.305 0.205 0.284 0.177 dent on the steel grade than the yield strength reduction factors,
700 0.108 0.045 0.165 0.062
800 0.085 0.038 0.103 0.053
hence the influence of the steel grade on the reduction factors of
the modulus of elasticity may be considered less important [7,8].
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 161

2:2005 [5] provides the same reduction factors used for class
4 hot-rolled steels for 0.2% proof stress whereas the BS 5950-Part 8
[4] provide the reduction factors for steels used in cold-formed
steel structures for yield strength at total strain levels of 0.5%, 1.5%
and 2.0%. Fig. 11a–c shows the comparison between the reduction
factors provided by the design standards against the reduction
factors determined in this investigation for both yield strength and
modulus of elasticity of a low strength cold-formed steel
S280GD þZ. From the results it is clear that all design standards
considered overestimate the reduction factors for yield strength.
The EN 1993-1-2:2005 [5] overestimates the reduction factors
beyond 200 °C and up to 600 °C. For instance for temperatures
ranging from 200 °C up to 300 °C the prediction presented in the
EN 1993-1-2:2005 [5] is about 8% higher than the reduction factor
determined in the experimental tests, whereas for 500 °C the
difference was about 30%. Above 600 °C the results are in good
agreement. The reduction factors provided by the BS 5950-Part 8
[4] and AS 4100 [28] are unsafe in the entire range of considered
temperatures, suggesting that their reduction factors should be
reviewed. Also reduction factors for different steel grades should
be introduced since the influence of the steel grade on the de-
termined reduction factors for yield strength has been proved by
other researchers [7,8].
Regarding the modulus of elasticity the BS 5959-Part 8 [4] does
not include reduction factors. The reduction factors provided by
the EN 1993-1-2:2005 [5] are unsafe up to 600 °C and the AS 4100
[28] provides unsafe reduction factors beyond 100 °C. The differ-
ence between the results from this experimental investigation and
the predictions presented in the EN 1993-1-2:2005 [5] is quite
relevant for temperatures ranging from 300 °C to 500 °C
(averageE20%).
Comparing the proportional limit reduction factors determined
in this experimental investigation with the ones provided in the
EN 1993-1-2:2005 [5], it was found that the ones presented in the
EN 1993-1-2:2005 [5] are very unconservative for temperatures
up to 500 °C. This observation was more or less expected since the
reduction factors predicted in the EN 1993-1-2:2005 [5] were
based on hot-rolled steels.

3.5. Comparison with hot rolled steels tested at University of

Coimbra

Regarding the assessment of the deterioration of mechanical

properties of steel with temperature increase, some experimental
tests were in the past undertaken on hot-rolled S355 steel at
University of Coimbra: (i) thicknesses lower than 16 mm and (ii)
thicknesses upper than 16 mm. Those results are also compared
with the ones obtained in this experimental investigation for the
S280GD þZ steel and with the available design standards, namely
Fig. 9. Reduction factors taken from the experimental tests. (a) Yield strength. (b) the EN 1993-1-2:2005 [5] (Fig. 12).
Elastic modulus. (c) Ultimate strength and proportional limit. Observing the results depicted in Fig. 12 it was found that for
high temperatures, the thickness of hot-rolled specimens is not an
important parameter for the reduction factors of the mechanical
The reduction factors for the ultimate strength determined in the
properties; exception was observed at low temperatures (100–
different investigations are also depicted in Fig. 10c. The same
200 °C), where some discrepancies were noted. Comparing the
behaviour was observed at about 200 °C in several experimental
investigations [7,8] predominantly for low strength steels. This reduction factors determined for yield strength for both cold-
behaviour was already explained in this paper. formed and hot-rolled steels it is clear that the differences are
quite significant of about 20%, for temperatures ranging from 200
3.4. Comparison with design standards to 300 °C, and increasing up to 60% for temperatures about 700 °C.
The differences are relatively similar for the reduction factors
The experimental results obtained in this research were com- predicted in the EN 1993-1-2:2005 [5] for hot-rolled steels.
pared with the current design standards, such as the EN 1993-1- Regarding the modulus of elasticity the results are less scattered
2:2005 [5], BS 5950-Part 8 [4] and AS 4100 [28]. The EN 1993-1- despite the different types of steel considered.
162 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

Fig. 10. Comparison of the reduction factors at elevated temperatures determined in this research study with other authors. (a) Yield strength reduction factors. (b) Modulus
of elasticity reduction factors. (c) Ultimate strength.

3.6. Proposal for mechanical properties prediction formed steels. For instance the equations proposed are similar to
the ones provided by Ranawaka [7] for the low strength steel
3.6.1. Yield strength G250. The proposed predictive equations depicted in Fig. 13b are
Based on the experimental results empirical equations were in good agreement with the test results.
proposed to predict the evolution of the reduction factors for yield
Eθ
strength with temperature ranging from 20 °C to 800 °C. The = − 2.5 × 10−4 × θ + 1.005, 20 °C ≤ θ ≤ 100 °C
empirical equations (Eq. (1)) derived for the yield strength re- E20 (2a)
duction factors were determined based on the 0.2% proof stress
method. In Fig. 13a it is possible to observe the good agreement Eθ
= − 1.4 × 10−3 × θ + 1.118, 100 °C < θ ≤ 700 °C
between the proposed model and the experimental results. Ad- E20 (2b)
ditionally, already existent proposals that better fit the results of
this investigation are presented. Eθ
= − 5.3 × 10−4 × θ + 0.509, 700 °C < θ ≤ 800 °C
E20 (2c)
f y, θ
= − 5.5 × 10−4 θ + 1.011, 20 °C ≤ θ ≤ 200 °C
f y,20 The previous proposal (Eq. (2)) may be slightly conservative. As
(1a)
an alternative a different model may be proposed (Eq. (3)).The
f y, θ proposal presented based on this investigation may be slightly

f y,20
(
= 0.070876 × 0.4882 − θ 0.482 , ) 200 °C < θ ≤ 800 °C
(1b)
conservative since it is possible to propose a model based only in
two equations (Eq. (3)). The proposals presented in this paper are
very similar to the ones presented by Ranawaka and Kakanamge
The predictive formulations proposed by Ranawaka and Kaka-
[7,8].
namge [7,8] are conservative in comparison to the ones here
presented. Nevertheless the differences are not very relevant Eθ
= − 2.5 × 10−4 × θ + 1.005, 20 °C ≤ θ ≤ 100 °C
meaning that each model may be capable to accurately predict the E20 (3a)
reduction factors for yield strength of low strength cold-formed
steels. Eθ
= − 1.345 × 10−3 × θ + 1.109, 100 °C < θ ≤ 800 °C
E20 (3b)
3.6.2. Modulus of elasticity
As for the yield strength, elevated temperatures leads to de-
gradation of the modulus of elasticity. In order to predict the de- 3.6.3. Stress–strain relationship to Ramberg–Osgood methodology
gradation of the modulus of elasticity with the temperature a new There are two types of stress–strain curves, namely sharp
set of empirical equations (Eq. (2)) was developed for the yielding and gradual yielding. In this investigation sharp yielding
280GD þZ steel. This equation is valid for low strength cold- stress–strain curves, with a small yield plateau, were obtained for
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 163

Fig. 11. Comparison between the reduction factors determined in this research study with the ones of the current design standards. (a) Yield strength reduction factors. (b)
Modulus of elasticity reduction factors. (c) Proportional limit reduction factors.

lower temperatures (20 °C and 100 °C) and gradual yielding curves ⎛ ⎞nθ
f f ⎟
for higher temperatures (200 °C, 300 °C, 400 °C, 500 °C, 600 °C, εθ = + 0.002 × ⎜⎜ ⎟
Eθ ⎝ f y, θ ⎠ (4)
700 °C and 800 °C). The stress–strain relationship of the tested
steel coupons under steady state test was modelled considering
The Ramberg–Osgood coefficient was determined based on the
the methodology presented by Ramberg and Osgood [37] and then
provisions presented in MMPDS-01 [40]:
compared with the proposals presented by other researchers. The
Ramberg–Osgood [37] model was developed to describe stress– ⎛ fu, θ ⎞
strain curves at ambient temperature. Based on this model Ola- εus, θ = 100⎜⎜εr, θ − ⎟⎟
⎝ Eθ ⎠ (4a)
wale and Plank [38], Outinen [3] and Lee et al. [2] proposed stress–
strain models at elevated temperatures respectively for hot-rolled εus , θ
steels, for S355 hot-rolled steel and for light gauge steel. The ex- nθ =
ln ( )0.2

pression based on the Ramberg–Osgood model used in this in- ⎛f ⎞

ln⎜ fu, θ ⎟
⎝ y, θ ⎠ (4b)
vestigation is [39, 40]

Fig. 12. Comparison with hot-rolled S355 steel tested at elevated temperatures.
164 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

Fig. 13. Comparison between proposed models and experimental results. (a) Yield strength. (b) Modulus of elasticity.

where εθ is the strain at temperature θ (°C), εr, θ is the strain at

rupture, εus, θ is the plastic strain at maximum tension load, f is the
stress, Eθ is the modulus of elasticity, fu,θ is the ultimate strength
and fy,θ is the yield strength at temperature θ (°C). The nθ para-
meter in this formulation is reciprocal to the strain hardening
coefficient [40]. The evolution of the parameter nθ with tempera-
ture is presented in Table 5 for each temperature level. The stress–
strain curves obtained using this formulation are compared with
the experimental results obtained (Fig. 14).
For low temperatures (20 °C and 100 °C) the proposed equa-
tions are not suitable since the original equations were developed
only for gradual yielding stress–strain curves and for lower tem-
peratures sharp yielding stress–strain curves with a visible yield
plateau were obtained. Between 200 °C and 300 °C the proposed
equations fit reasonably well up to 1.6% strain however the Fig. 14. Comparison between experimental stress–strain curves and the proposed
equations may be considered slightly unsafe in the strain range of stress–strain model at elevated temperatures based on the Ramberg–Osgood
model.
0.2–1.3%. For temperatures above 300 °C the proposed equations
are in very good agreement with stress–strain curves determined
in the experimental tests.
The proposed equations in this investigation for the stress–
strain curves were also compared with the models proposed by
Ranawaka (Eq. (5)) [7], Chen and Young (Eq. (6)) [10] and Kanka-
namge [8] (Fig. 15). The model proposed by Kankanamge [8] is
very similar to the one proposed by Ranawaka [7]. The only change
is the β parameter which has assumed the value 1.5 instead of
0.86.

⎛f ⎞ ⎛ f ⎞ T
η
fT
εT = ⎜
+β×⎜
y, T
⎟ ⎜ T ⎟
ET ⎟×⎜ ⎟
⎝ E T ⎠ ⎝ f y, T ⎠ (5)

ηT = 0.000138 × T2 − 0.085468T + 19.212, 350 °C ≤ T ≤ 800 °C

β = 0.86 (5a)

Fig. 15. Comparison of the proposed stress–strain model with other predictive
stress–strain models proposed by other authors.

Table 5 ⎧ ⎛ f ⎞nT
n Parameter determined for each temperature level. ⎪ fT ⎜ T ⎟
⎪E + 0.002 × ⎜f ⎟ for fT ≤ f y, T
⎪ T ⎝ y, T ⎠
Temperature [°C] ET = ⎨
⎪f −f ⎛ f −f ⎞mT
+ εu, T × ⎜⎜
y, T ⎟
20 100 200 300 400 500 600 700 800 ⎪ T y, T T
for fT > f y, T
⎪ ⎟ × εy, T
⎩ E y, T ⎝ fu, T − f y, T ⎠ (6)
n 14.602 13.724 8.586 8.353 10.411 13.397 47.024 16.113 11.701
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 165

and The Eurocode models were compared with the experimental

stress–strain curves as depicted in Fig. 16a and b for both models
ET
E y, T = ET presented in the EN 1993-1-2:2005 [5]. As expected these models
1 + 0.002 × nT × are not suitable to describe the stress–strain curves of cold-formed
f y, T (6a)
steels since these have been created for hot-rolled steels. Again,
nT = 20 − 0.6 × T (6b) this shows that new stress–strain models should be developed
specifically for steels used in cold-formed steel structures.
T Based on the stress–strain model presented in the EN 1993-1-2
mT = 1 + [5] for hot-rolled steels a modified predictive model is proposed
350 (6c)
for the S280GD þZ steel. The available strain hardening model
Comparing all proposed models it was found that above 500 °C formulation was also used and modified in the new proposal. The
all models are in good agreement with the experimental results. proposed model is now thoroughly described (Eq. (9)):
Chen and Young [10] model seem to be the less accurate. Kanka-
namge [8] model is unsafe for lower levels of temperature (300 °C  for ε ≤ εy, θ and 20 °C r θa r800 °C, where εy, θ = f y, θ
:
Ea, θ
and 400 °C), as well as the model proposed by Ranawaka [7].
Overall, the proposed model in this investigation is the one that
σ = ε⋅Ea, θ (9a)
fits best to the experimental stress–strain curves.  for εy, θ < ε ≤ 0.1:
▪ 20 °C r θa o200 °C, the strain hardening is as follows:
3.6.4. Stress–strain relationship of EN 1993-1-2
The EN 1993-1-2:2005 [5] also proposes a stress–strain curve fu, θ = 1.3 × f y, θ (9b.1)
model which is based on hot-rolled steel, this could be the reason
▪ 200 °C r θa o 400 °C, the strain hardening is as presented
why it does not accurately reproduce the behavior of low strength
in Eq. (9b.2). This behavior may be explained by the ob-
cold-formed steel S280GD þZ at elevated temperatures. The
served singularity in the behavior of low strength cold
stress–strain model proposed by the EN 1993-1-2:2005 [5] is gi-
formed steels for temperatures between 200 °C and 300 °C.
ven by (Eq. (7)):
At 200 °C the ultimate strength was higher (E11%) than at
⎧ εE for ε ≤ εp, θ ; ambient temperature (20 °C) and at 300 °C the ultimate
⎪ a, θ strength was close to the one obtained at ambient tem-
⎪ ⎛ b ⎞⎡ 2 2⎤0.5
⎪ fp, θ − c + ⎜ ⎟⎣a − (εy, θ − ε) ⎦ for εp, θ ≤ ε ≤ εy, θ ; perature. This behavior has been already explained in this
⎪ ⎝ a ⎠
paper and observed by other authors [7,8].
⎪
σ = ⎨ f y, θ for εy, θ ≤ ε ≤ εt, θ ;
⎪ fu, θ = 1.6 × f y, θ (9b.2)
⎪ ⎡ (ε − εt, θ ) ⎤
⎪ f y, θ ⎢1 − (ε − ε ) ⎥ for εt, θ ≤ ε ≤ εu, θ ; ▪ 400 °C r θa o 600 °C, the strain hardening is as follows:
⎪ ⎣ u, θ t, θ ⎦
⎪0 for ε = εu, θ ; fu, θ = f y, θ (2.8 − 0.003θa)
⎩ (7) (9b.3)

and ▪ For both temperature ranges (20 °Cr θa o300 °C and

300 °C r θa o 600 °C) the stress is determined as follows:
f p, θ
εp, θ = ; εy, θ = 0.02; εt, θ = 0.15; εu, θ = 0.20;
Ea, θ 1
(0.1 − εy,θ ) ( u,θ y,θ )
σ= × f −f ×ε
⎛ c ⎞
(
a2 = εy, θ − εp, θ ) ⎜⎜εy, θ − εp, θ +
⎝
⎟⎟
Ea, θ ⎠ ⎡ ⎛ ε ⎞⎤ ⎛ ε ⎞
+ ⎢1 + ⎟⎟⎥ × f y, θ − ⎜⎜
y, θ y, θ
⎜⎜ ⎟⎟ × fu, θ
(
b2 = c εy, θ − εp, θ Ea, θ + c 2 ) ⎢⎣ ⎝ 0.1 − εy, θ ⎠⎦⎥ ⎝ 0.1 − εy, θ ⎠ (9b.4)
2
(f ) − f p, θ ▪ 600 °C r θa r 800 °C, strain hardening is no longer ob-
y, θ
c= served, hence, fu, θ = fy, θ and consequently σ = fy, θ .
(εy,θ − εp,θ )E − 2(f
a, θ y, θ
− f p, θ ) (7a)  for 0.1 ≤ ε < εt, θ and considering σ = fu :
▪ 20 °C r θa r 300 °C the strain range is as follows
This model does not take into account the strain hardening of 0.1 ≤ ε ≤ 0.18;
steel, but accordingly to the EN 1993-1-2:2005 [5] the stress– ▪ 300 °C o θa r500 °C the strain range is as follows
strain model can be extended by the strain hardening model for 0.1 ≤ ε ≤ 0.2;
temperatures below 400 °C given in Annex A (Eq. (8)): ▪ 500 °C o θa r800 °C the strain range is as follows
⎧ 0.1 ≤ ε ≤ 0.3.
( )
⎪50 fu, θ − f y, θ ε + 2f y, θ − fu, θ for 0.02 < ε < 0.04;  for εt , θ ≤ ε < εu, θ :
⎪
⎪f ▪ 20 °C r θa r300 °C the strain range is as follows
for 0.04 ≤ ε ≤ 0.15;
σa = ⎨ u, θ
⎪ f [1 − 20(ε − 0.15)] 0.18 < ε ≤ 0.2 and the stress is determined according to Eq.
for 0.15 < ε < 0.2;
⎪ u, θ (9c.1):
⎪
⎩0 for ε ≥ 0.2; (8)
σ = fu, θ × ⎡⎣1 − 50 × (ε − εt, θ)⎤⎦, where εt, θ = 0.18 (9c.1)
where the ultimate strength at elevated temperatures should be
determined as follows: ▪ 300 °C o θa r500 °C the strain range is as follows
0.20 < ε ≤ 0.25 and the stress is determined according to
⎧1.25f for θa < 300 °C ; Eq. (9c.2):
⎪
⎪
y, θ

fu, θ = ⎨ f y, θ (2 − 0.0025θa) for 300 °C ≤ θa < 400 °C; σ = fu, θ × ⎡⎣1 − 20 × (ε − εt, θ )⎤⎦, where εt, θ = 0.2
⎪ (9c.2)
⎪
⎩ f y, θ for θa ≥ 400 °C; (8a) ▪ 500 °C o θa r800 °C the strain range is as follows
166 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

Fig. 16. Comparison of stress–strain curves predicted by EN 1993-1-2:2005 model with test results. (a) Stress–strain model without strain hardening. (b) Stress–strain model
with strain hardening.

0.30 < ε ≤ 0.35 and the stress is determined according to forces and consequently to poor fitting between experimental [41]
Eq. (9c.3) and fu, θ = fy, θ : and numerical results.
The thermal elongation determined with free thermal elonga-
σ = fu, θ × ⎡⎣1 − 20 × (ε − εt, θ )⎤⎦, where εt, θ = 0.3 (9c.3) tion was compared with the available design predictions pre-
sented in the EN 1993-1-2:2005 [5] and BS 5950-Part 8 [4]
In Fig. 17 the modified predictive stress–strain curve model
(Fig. 18). A new predictive equation based on the obtained results
(P_θ) is compared with the experimental results obtained in the
was proposed (Eq. (10)).
scope of this investigation for the S280GD þZ steel. The initial part
of the stress–strain curves is presented in Fig. 17a while the Δl
= 1.10235 × 10−9 × θ 2 20 °C ≤ θ ≤ 740 °C
complete stress–strain curves are presented in Fig. 17b. The com- l
parison indicates that the proposed model conservatively predicts + 0.68575 × 10−5 × θ
the stress–strain curves for the S280GD þ Z steel at elevated tem-
− 0.79712 × 10−4
peratures. At this point the model may be valid for the S280GD þZ
Δl
steel and eventually for low strength steels up to 350 MPa. How- = 1.1031 × 10−2 740 °C < θ ≤ 890 °C
l
ever further investigations must be conducted and the proposed
Δl
model further developed. Nevertheless this proposal for the = 2.16443 × 10−5 × θ 890 °C < θ ≤ 1000 °C
S280GD þZ steel is a significant improvement against the current l
model. − 8.04389 × 10−3 (10)

4. Thermal properties 4.2. Thermal conductivity

4.1. Thermal elongation A suitable model for thermal conductivity for the S280GD þ Z
steel may be important to improve design methodologies at ele-
The thermal elongation predicted in the available design vated temperatures. Rigorous thermal conductivity models will
standards [4,5] is conservative when compared with experimental lead to an accurate prediction of the heat transfer phenomenon,
results obtained in this investigation and others previous studies hence the prediction of the ability of the S280GD þZ steel to
[3,10]. The accurate assessment of this property is fundamental for transport heat energy from high to low temperature regions. With
structural fire design. For instance while performing numerical increasing temperature the thermal conductivity decreases as it
analysis of cold-formed steel columns with restrained thermal was observed in the experimental results. BS 5950-8 [4] assumes
elongation in case of fire, overestimated coefficient of thermal that the thermal conductivity is constant with temperature in-
elongation will lead to overestimated generation of restraining crease 37.5 W/m °C. The proposed model for thermal conductivity

Fig. 17. Comparison between the new proposed model based on the EN 1993-1-2 formulation with test results. (a) Initial part of the stress–strain curves. (b) Complete
stress–strain curves.
 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168 167

Fig. 20. Specific heat experimental results and comparison with the model pre-
sented in the EN 1993-1-2:2005.

Fig. 18. Thermal elongation with increasing temperature. Comparison of experi-

mental results and proposed model with design standards predictions and other presented and compared with other investigations and with the
proposed models.
available predictions in design standards on this subject.
In the experimental campaign for the assessment of mechan-
ical properties of the S280GD þZ steel with 2.5 mm thickness,
tensile coupon tests were conducted at different temperatures,
ranging from 20 °C to 800 °C, using the steady state test method.
Based on the stress–strain curves obtained for each temperature
level, the yield strength, modulus of elasticity, proportional limit
and ultimate strength were determined and the reduction factors
for each property were obtained for each value of temperature.
These results were then compared with the ones presented by
other authors and with the design standards predictions such as
EN 1993-1-2:2005 [5], BS 5950-8 [4] and AS 4100 [16]. It was
found that for low strength cold formed steels such as the
S280GD þZ, the yield strength predicted in the design standards
are unconservative in the entire range of temperatures tested. The
Fig. 19. Thermal conductivity test results and new predictive model. Comparison same conclusion can be observed for the modulus of elasticity and
with the EN 1993-1-2:2005 model. proportional limit. However the BS 5950-8 [4] does not provide
any reduction factors for the modulus of elasticity and propor-
presented in the EN 1993-1-2:2005 [5] was found unconservative tional limit whereas the AS 4100 [16] does not present reduction
for the S280GD þZ cold-formed steel tested. The difference be- factors for the proportional limit. Nevertheless it is worth to
tween the two models was about 20% in the entire range of mention that the reduction factors predicted in the EN 1993-1-
temperatures tested (Fig. 19). The difference may be explained by 2:2005 [5] are the ones that best fit to the provided experimental
the chemical composition of the steel tested. Steels with higher results. In the scope of the research study a new set of predictive
levels of ferrite (Fe) and with lower levels of manganese (E0.3%) equations was presented for the yield strength, modulus of elas-
(Mn) tend to have values of thermal conductivity higher than ticity and proportional limit reduction factors.
70 W m  1 K  1 [42]. These results show that the EN 1993-1- Creep may have influence for temperatures of 600 °C and above
2:2005 [5] should adopt different thermal conductivity models and for higher stress levels hence specific creep tests must be
according to the chemical composition of the steel to be con- conducted for the S280GD þZ steel at elevated temperatures
sidered in design procedures, establishing lower and upper limits (above 400 °C) in order to accurately assess the influence of creep
to the thermal conductivity. and eventually understand how creep may affect the fire safety
The proposed thermal conductivity model for the S280GD þZ design of structures. In this experimental investigation the ex-
steel tested in this study is defined as follows (Eq. (11)): istent amount of creep is included in the experimental results as
well as in the proposed formulation for stress–strain curves at
λ a = − 3.332 × 10−2 × θ + 65.206, 20 °C ≤ θ ≤ 800 °C (11)
elevated temperatures.
Also new predictive equations for stress–strain curves based on
4.3. Specific heat the Ramberg–Osgood model were proposed and compared with
other proposals from other researchers. It was found that the
The obtained experimental results for the specific heat (cp) proposal presented by Ranawaka [7] also predict reasonably well
were compared with the model available in the EN 1993-1-2:2005 the stress–strain curves for the S280GD þ Z steel tested. The
[5] (Fig. 20). Clearly the obtained results are in good agreement stress–strain curve model proposed by the EN 1993-1-2:2005 [5]
with the model available in the design standard hence can be used does not represent the behavior of low strength cold-formed steel
for structural fire design. at elevated temperatures even when the strain hardening model
(Annex A, EN 1993-1-2:2005 [5]) is included in the original for-
5. Conclusions mulation. Hence a new improved formulation of the stress–strain
model is presented for the S280GD þZ steel. This new formulation
An experimental investigation on the mechanical and thermal conservatively predicts the stress–strain curves of the S280GD þZ
properties of the S280GD þ Z at elevated temperatures was and may be valid for low strength steels (up to 350 MPa).
168 H.D. Craveiro et al. / Thin-Walled Structures 98 (2016) 154–168

Regarding thermal properties of the S280GD þ Z steel, it was of steel structures in fire conditions, in: Proceedings of the 1st International
found that the predictions available in the EN 1993-1-2:2005 [5] Workshop Structures in Fire, Copenhagen, June 2000, pp. 267–282.
[16] N. Ottosen, M. Ristinmaa, The Mechanics of Constitutive Modeling, Elsevier,
are overconservative for the thermal elongation and thermal Amsterdam, 2005.
conductivity whereas for the specific heat the proposed model in [17] M. Cowan, K. Khandelwal, Modelling of high temperature creep in ASTM A992
the design standard fits well with the obtained results. An im- structural steels, Eng. Struct. 80 (2014) 426–434.
[18] S. Kesawan, V. Jatheeshan, M. Mahendran, Elevated temperature mechanical
proved predictive equation was proposed for thermal elongation properties of flange channel sections, Constr. Build. Mater. 87 (2015) 86–89.
and thermal conductivity of the steel tested. Finally, it is re- [19] Wei Lu, P. Makelainen, Advanced Steel Structures: 1. Structural Fire Design; 2.
commended additional thermal properties tests in different types Fatigue Design, Helsinki University of Technology, Laboratory of Steel Struc-
tures Publications, 2003.
of steel in order to establish upper and lower limits for thermal [20] J.M. Barsom, S.T. Rolfe, Fatigue and Fracture Control in Structures: Application
conductivity as a function of the chemical composition of each of Fracture Mechanics, American Society for Testing and Materials (ASTM),
steel. USA, 1999.
[21] T. Lennon, D.B. Moore, Y.C. Wang, C.G. Bailey, Designers’ Guide to EN 1991-1-2,
EN1992-1-2, EN 1993-1-2 and EN 1994-1-2. Handbook for the Fire Design of
Steel, Composite and Concrete Structures to the Eurocodes, Thomas Telford,
Acknowledgements 2007.
[22] J. Brnic, et al., Comparison of material properties: steel 20MnCr5 and similar
steels, J. Constr. Steel Res. 95 (2014) 81–89.
The authors gratefully acknowledge the Portuguese Foundation [23] J. Brnic, M. Canadija, G. Turkalj, D. Lanc, Behavior of S355JO steel subjected to
for Science and Technology–FCT (www.fct.mctes.pt) and the cold- uniaxial stress at lowered and elevated temperatures and creep, Bull. Mater.
Sci. 33 (2010) 475–489.
formed steel profile maker PERFISA S.A. (www.perfisa.net) for
[24] J. Brnic, G. Turkalj, M. Canadija, D. Lanc, AISI 316Ti (1.4571) steel—mechanical
their support under the framework of research project PTDC/ECM/ creep and fracture properties versus temperature, J. Constr. Steel Res. 67 (12)
116859/2010, as well as to the Human Potential Operational Pro- (2011) 1948–1952.
gramme (POPH), the European Social Fund and National Strategic [25] M. Cowan, K. Khandelwal, Modeling of high temperature creep in ASTM A992
structural steels, Eng. Struct. 80 (2014) 426–434.
Reference Framework (QREN). [26] J. Brnic, et al., Behavior of AISI 316L steel subjected to uniaxial state of stress at
elevated temperatures, J. Mater. Sci. Technol. 25 (2009).
[27] J. Brnic, G. Turkalij, J. Niu, M. Canadija, D. Lanc, Analysis of experimental data
on the behavior of steel S275JR – reliability of modern design, Mater. Des. 47
References (2013) 497–504.
[28] AS 4100-1998: Steel structures. Standards Australia Limited, Sydney, Australia,
[1] W.W. Yu, Cold-formed steel design, 4th edition,. Wiley, New York, 2010. 1998.
[2] J.H. Lee, M. Mahendran, P. Makelainen, Prediction of mechanical properties of [29] Y. He, Rapid thermal conductivity measurement with a hot disk sensor. Part 1.
light gauge steels at elevated temperatures, J. Constr. Steel Res. 59 (2003) Theoretical considerations, Thermochim. Acta 436 (2005) 122–129.
1517–1532. [30] H. Zhang, M. Li, W. Fang, D. Dan, Z. Li, W. Tao, A numerical study on the
[3] J. Outinen, Mechanical Properties of Structural Steels at Elevated Temperatures theoretical accuracy of film thermal conductivity using transient plane source
(Licentiate thesis), Helsinki University of Technology, Finland, 1999. method, Appl. Therm. Eng. 72 (2014) 62–69.
[4] BS 5950-8:1990, Structural Use of Steel Work in Building–Part 8: Code of [31] V.K.R. Kodur, A.M. Shakya, Effect of temperature on thermal properties of
Practice for Fire Resistant Design, British standards institution (BSI), British spray applied fire resistive materials, Fire Saf. J. 61 (2013) 314–323.
Standard BS, 1998. [32] M. Gustavsson, E. Karawacki, S.E. Gustafsson, Thermal conductivity, thermal
[5] EN 1993-1.2, Design of Steel Structures. General Rules. Structural Fire Design, diffusivity, and specific heat of thin samples from transient measurements
European Committee for Standardization, Brussels, Belgium, 2005. with hot disk sensors, Rev. Sci. Instrum. 65 (1994) 3856–3859.
[6] M.P. Sidey, D.P. Teague Elevated Temperature Data for Structural Grades of [33] Hot Disk, 〈http://www.hotdisk.se/〉, 2015.
Galvanized Steel. British Steel (Welsh Laboratories) Report, UK, 1988. [34] Epsilon, 〈http://www.epsilontech.com/〉, 2015.
[7] T. Ranawaka, M. Mahendran, Experimental study of the mechanical properties [35] EN 10002-5: Metallic Materials-Tensile Testing – Part 5: Method of Testing at
of light cold-formed steels at elevated temperatures, Fire Saf. J. 44 (2009) Elevated Temperature, European Committee for Standardization (CEN), Brus-
219–229. sels, 1992.
[8] N.D. Kankanamge, M. Mahendran, Mechanical properties of cold-formed [36] EN 10002-1: Metallic Materials-Tensile Testing: I. Technique of Test at Am-
steels at elevated temperatures, Thin-Walled Struct. 49 (2011) 26–44. bient Temperature, European Committee for Standardization (CEN), Brussels,
[9] P. Makelainen, J. Outinen, J. Kesti, Fire design model for structural steel S420M 2001.
based upon transient-state tensile test results, J. Constr. Steel Res. 48 (1998) [37] W. Ramberg, W.R. Osgood Description of Stress–Strain Curves by Three Para-
47–57. meters, NACA Technical Note 902, 1943.
[10] J. Chen, B. Young, Experimental investigation of cold-formed steel material at [38] A.O. Olawale, R.J. Plank, The collapse analysis of steel columns in fire using a
elevated temperatures, Thin-Walled Struct. 45 (2007) 96–110. finite element strip method, Int. J. Numer. Methods Eng. 26 (1988) 2755–2764.
[11] J. Chen, B. Young, Corner properties of cold-formed steel sections at elevated [39] B. Faggiano, G.D. Matteis, R. Landolfo, F.M. Mazzolani, Behaviour of aluminium
temperatures, Thin-Walled Struct. 44 (2006) 216–223. alloy structures under fire, J. Civil Eng. Manag. X (3) (2004) 183–190.
[12] E. Mecozzi, B. Zhao, Development of stress–strain relationships of cold-formed [40] FAA AR MMPDS-01 – Metallic Material Properties Development and Stan-
lightweight steel at elevated temperatures, in: Proceedings of the Eurosteel dardization, US Department of Transportation, Federal Aviation Administra-
Conference, 5.1, 2005, pp. 41–49. tion, Washington DC, 2003.
[13] J Outinen, P. Makelainen, Mechanical properties of structural steel at elevated [41] H.D. Craveiro, J.P. Rodrigues, L. Laim, Cold-formed steel columns made with
temperatures and after cooling down,in: Proceedings of the 2nd International open cross-sections subjected to fire, Thin-Walled Struct. 85 (2014) 1–14.
Workshop Structures in Fire, Christchurch; March 2002, pp. 273–290. [42] M.J. Peet, H.S. Hasan, H.K.D.H. Bhadeshia, Prediction of thermal conductivity of
[14] S. Pinheiro, Experimental analysis of mechanical and thermal properties of the steel, Int. J. Heat Mass Transf. 54 (2011) 2602–2608.
S280GD þZ steel (M.Sc. thesis in civil engineering), University of Coimbra,
2015 (in Portuguese).
[15] J Outinen, O Kaitila, P. Makelainen, A study for the development of the design