Professional Documents
Culture Documents
Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws
a r t i c l e i n f o a b s t r a c t
Article history: To complement previous experimental studies on the performance of unprotected Elliptical Hollow
Received 31 October 2011 Section (EHS) steel columns, under the hydrocarbon re curve, a numerical analysis has been
Received in revised form performed and outlined in this paper. A three-dimensional Finite Element Method (FEM) model has
20 February 2012
been developed and calibrated against 12 experiments, comprising of six unrestrained and six axially
Accepted 20 February 2012
Available online 28 March 2012
restrained EHS columns of two slenderness, lz 40.1 and lz 50.8. The EHS temperature proles and
axial displacements, measured under three different loading levels, (aL 0.3, 0.45 and 0.6) were utilised
Keywords: in the calibration process. The mechanical and thermal properties for carbon steel at elevated
Elliptical Hollow Section temperatures, detailed in the Eurocode standard EC3 Part 12, design of steel structures [1] have been
Columns
applied. Axial displacement charts illustrate that a close agreement between the FEM model and the
Fire resistance
experiment results was achieved, while highlighting how critical the thermal expansion coefcient and
Axial restraint
Local buckling geometric imperfection was during the calibration process. Ultimately, the paper will detail appropriate
Finite Element Method recommendations for the thermal analysis of unlled Elliptical Hollow Section steel columns, of steel
Thermal analysis grade S355J2H, and provide the platform for comprehensive parametric re investigations to
Eurocode commence.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction required. One of the rst comprehensive studies was the cross
section classication by Gardner and Chan [4], which grouped
The Elliptical Hollow Section (EHS) is the newest member of each EHS cross section into four classes. Grouping into these
the Steel Hollow Section (SHS) family, and has been the focus of classes, identied the resistance and rotation capacity limits of
numerous studies over the last number years. This new hollow the cross section, before failure occurred by local buckling. This
section type provides an exciting alternative to the rectangular study concluded that the class limits for Circular Hollow Sections
and circular counterparts, for structural design engineers and (CHS) were deemed applicable, if an equivalent diameter for the
architects. The continuously changing outside curvature not only EHS was obtained. Further studies in this comprehensive experi-
provides elegant lines for aesthetic appeal, but numerous mental programme by Chan and Gardner, explored the exural
moments of inertia, that can be used to enhance and pioneer buckling [5], bending strength [6] and compressive resistance [7]
new structural designs. response of member length EHS columns. The elastic buckling of
Until recently, research into this new structural shape have Elliptical Hollow Sections in compression was also investigated by
been limited to a few studies in the 19 60s and early 1970s, which Ruiz-Teran and Gardner [8] and analysed by Silvestre [9].
observed the buckling behaviour of oval cylindrical shells, man- Filling the Elliptical Hollow Section void with concrete is a
ufactured from polyester lm. These investigations included practice widely used in Circular (CHS) and Rectangular Hollow
stability under compression [2], and a combination of bending Sections (RHS), this provides additional structural qualities to the
moments under axial force [3]. The downside to these studies section; these qualities include increased stiffness, compressive
realised that their plastic resistance could not be achieved. It was strength and additional re resistance. Yang et al. [10] investigated
recognised that to enhance the structural design guidance for the structural properties of concrete-lled elliptical hollow sections
this new section, a number of in-depth investigations would be (CFEHS) while suggesting a proposed design formula for member
design. Zhao and Packer [11] also investigated concrete lled EHS
stub columns. This investigation veried the suitability of circular
n
Corresponding author.
equivalent diameters, but in addition proposed effective RHS width
E-mail addresses: scullion-t1@email.ulster.ac.uk (T. Scullion), and depth dimensions, as local buckling observations resembled
f.ali@ulster.ac.uk (F. Ali), a.nadjai@ulster.ac.uk (A. Nadjai). plate buckling rather than cylindrical shell buckling.
0263-8231/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tws.2012.02.012
T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236 23
Only a few viable Finite Element (FE) studies on Elliptical Full listings of the experimental results and comparisons with the
Hollow Sections have been carried out to date, the most signi- unrestrained EHS columns are documented in Scullion et al. [16].
cant performed by Zhu and Wilkinson [12]. They investigated the
local buckling behaviour of the EHS, while verifying that the
equivalent CHS diameters proposed by Gardner and Chan [13] 3. Finite element analysis
was a viable approach to adopt. Espinos et al. [14] produced a
Finite Element model to understand the behaviour of CFEHS, with Fire testing of structural elements is expensive, through an
proposed design guidance. The FEM model used in Espinoss accumulation of running costs and test specimen expenditure. As
investigation was calibrated with historical temperature proles test specimens have only a one test life time, the advantages of using
from concrete-lled circular hollow sections (CFCHS). This inves- Finite Element (FE) models become even more benecial. The Finite
tigation however, will use the experimental data obtained from Element Method (FEM) is a numerical modelling technique that nds
the rst ever studies into the re resistance of unrestrained [15] approximate solutions for partial and integral equations. With the
and axially restrained [16] unlled EHS steel columns performed advancements in computer technology both in FE software and
by Scullion et al., to calibrate the FEM model. computing processing speeds, the ability for solving complex elasti-
city and structural analysis problems has become extremely applic-
able. However, an FEM model for thermal analysis can only be
2. The experimental investigation successful and credible, if the correct input parameters and time
temperature varying values for material properties are correctly
A hydrocarbon re based experimental investigation, comprising detailed. In the absence of tensile coupon testing or experimental
of 12 unlled pin-ended Elliptical Hollow Section steel columns, was data, the most reliable source to date for obtaining thermal and
conducted at the FireSERT facilities of the University of Ulster. The mechanical properties for structural carbon steel is from the struc-
two part experimental programme tested two different section sizes, tural Eurocode (EC) publications. This investigation will use these EC
250 125 8.0 mm and 200 100 8.0 mm with slenderness material parameters to develop and calibrate an FE model, detailing
lz 40.1 and lz 50.8, namely EHS-A and EHS-B, respectively. The the unlled Elliptical Hollow Section steel column under various
rst stage tested six columns, three of each slenderness, unrestrained loading and restraint levels, subjected to re.
and under three different loading levels (aL 0.3, 0.45, 0.6). These rst The Finite Element Analysis (FEA) software used for this verica-
tests demonstrated the unique local and global buckling modes of the tion process was the software package TNO-DIANA Version 9.4,
EHS columns under compressive loads, full details of the results are with model geometry and mesh created using the windows based
available in Scullion et al. [15]. The second stage of the experimental graphical user interface (GUI) Midas FXVersion 3.0.0. A three-
program applied axial restraint to the EHS columns thermal expan- dimensional numerical model was created, utilising the staggered
sion. As columns are rarely used in isolation, the application of this heat ow-stress analysis modules of DIANA. The model rst demon-
axial restraint represents a more realistic condition of column strated and validated the maximum compressive capacity resistance
boundary constraints, found in construction. This study illustrated of the columns at room temperatures, before a structural thermal
the additional axial forces that are present when restraint is applied analysis commenced. The requirement for a heat ow-stress working
and demonstrated the accelerated failure rates of the EHS columns. domain environment determined the Finite Element (FE) selection.
24 T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236
During thermal analysis the eight-node isoparametric brick element 5.1. FEM model development at room temperatures
HX8HT was merged with the four nodes of the three-dimensional
(3D) linear boundary ow element BQ4HT. This boundary element The FEM model was initially evaluated under room tempera-
was used to describe the thermal convection and radiation to the ture conditions with theoretically hand calculations. This vali-
brick element surface. The temperature values resulting from the dated boundary end constraints, critical buckling mode shape,
heat ow analysis was automatically tabulated into a time and maximum compressive strength capacity, calculated using
temperature form, this was then used for the structural analysis. the EC method.
During the structural analysis the boundary element BQ4HT was
deactivated, and the potential ow HX8HT element was converted to 5.1.1. Geometry and mesh application
the three-dimensional 20 node isoparametric solid brick structural Dia and Lam [23] investigated element sizes suggesting a
element CHX60 [17]. range of between 5 and 10 mm for steel. This range demonstrated
a good compromise, to maximise model accuracy with experi-
mental results and minimise computation runtime for cold
4. The Eurocode parameters analysis. However, thermal analysis is more computation exten-
sive therefore this range was increased at column ends with the
The Elliptical Hollow Sections are manufactured to technical smaller elements used in areas of interest. A number of mesh
delivery conditions of the European standard EN10210-1:2006 variations were used throughout this investigation (Fig. 1), for
[18], the tolerances and dimensions of the section are detailed in cold analysis mesh type 1 was selected. This mesh type used a
EN10210-2:2006 [19]. The Eurocode 3 BS EN 1993 contains 20 uniformly distributed seed along the length of the column and
parts detailing the structural design recommendations for steel three element layers for the section thickness. The mesh seed was
structures. EC3 Part 11 [20] details the general design rules for graded so mesh density was increased in the tighter corners of the
steel structures with EC3 Part 12 [21] outlining the structural EHS column.
re design rules for steel structures.
5.1.2. Maximum compressive capacity
4.1. Steel mechanical properties at cold temperatures EC3 Part 11 Section 6.3 [20] was applied to calculate the
buckling resistance of the EHS columns. This is the current
The parameters required to model steel under cold tempera- method for obtaining maximum compressive capacity for all hot
tures are the modulus of elasticity (Ea ), yield strength (f y ), nished hollow section columns of class 1, 2 or 3, and was veried
Poissons ratio (v) and density (ra ). The EC publications recom- by Chan and Gardner [5] to be suitable for EHS member columns.
mend values for these parameters for structural carbon steel. A To calculate the maximum compressive strength, a buckling curve
value of 210 kN/mm2 was used for the modulus of elasticity as and instability safety margin must be selected. Buckling curve a
recommended by EC3 Part 11 Section 3.26 [20]. Table B3 of the is the appropriate selection for hot rolled steel of grade S355J2H.
technical delivery conditions BS EN 10210-1:2006 [18] species For the purpose of this investigation an instability safety margin
for Steel Hollow Sections (SHS) with a thickness (T) less than gM1 1.1 was used. A value of 1.1 is considered a suitable value for
16 mm have a minimum yield strength of 355 N/mm2, dened steel structures and buildings with loads of wide scatter. Table 1
also in the EHS manufacturers datasheet [22]. The density of hot details the calculated maximum capacity loads for EHS-A and
rolled steel has been recommended as 7850 kg/m3 by EC3 Part EHS-B using these parameters.
12 Section 3.2.2 [21], with a Poisson Ratio of 0.3 assigned for hot
rolled steel by EC3 Part 11 [20].
5.1.3. Euler buckling analysis
Using the Euler buckling analysis module of DIANA, a buckling
4.2. Thermal and mechanical properties of steel at elevated value lB that represents the ratio Pcr =Pa was obtained. Where P cr
temperatures is the critical buckling load of a perfect column (Eq. (1)) and Pa the
applied load.
The thermal parameters required for structural carbon steel, at
elevated temperatures include, specic heat capacity, conductiv- p2 Ea I
Pcr 1
ity and thermal expansion. These parameters are appropriately Le 2
detailed in Section 3.4 of EC3 Part 12 [21]. The reduction values, Using this buckling value lB the FEM model was veried with
required for the modulus of elasticity and yield strength, are the theoretical critical load using Eq. (2). Fig. 2 illustrates the
presented in Table 3.1 Section 3.4 of EC3 Part 12 [21]. overall global buckling mode shape observed for EHS-B, this is the
lowest buckling mode and the most critical collapse mode. The
4.2.1. Density and Poissons ratio of structural carbon steel FEM model buckling values for both sections are compared with
at temperature theory in Table 1. The error tolerance indicates a margin of
The density of steel is considered independent of temperature accuracy; determining that a satisfactory correlation between
by EC3 Part 12 Section 3.2.2 [21], therefore the constant ambient FEM model results and theoretical values has been obtained.
value was used. Like density, any variation in the Poissons ratio Pcr P a lB 2
has been discounted under elevated temperatures, as the percen-
tage change in Poissons ratio is negligible compared to other
inuencing parameters. 5.1.4. Elastic critical buckling stress of member length EHS columns
The critical buckling stress (scr ) of Eq. (3) is dependent on the
modulus of elasticity Ea of the material and the slenderness ratio
5. FEM model development Le =r) of the column [24].
P cr p2 Ea
Two FEM models have been created detailing the geometry of scr 3
A Le =r2
both EHS section types, EHS-A and EHS-B, used in the experi-
mental investigation, with the EC material properties for struc- The buckling value obtained from the FEM Euler buckling
tural carbon steel assigned. analysis and Eq. (4), was used to determine the critical buckling
T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236 25
Mesh Type 1 (8640) Mesh Type 2 (1440) Mesh Type 3 (4972) Mesh Type 3 - Top
Table 1
Critical buckling loads at cold temperature.
stress. The FEM results are compared with the theoretical values
in Table 2.
Dl
scr Ea :e Ea :lB : 4
l
5.2.1. Temperature development in unprotected steel (EC method) heat transfer by convection and radiation, described in EC1-1-2
To determine the temperature rise of an unprotected steel Section 3.1, shown in Eq. (5).
member, the net heat ux (h_ net ) to which the member surface is
exposed requires calculation. The net heat ux is a combination of h_ net h_ net,c h_ net,r 5
26 T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236
1600
1400
1200
Axial Compressive Load (kN)
EHS-A Imperfection 10 mm
800 EHS-A Imperfection 15 mm
EHS-A Imperfection 20 mm
600
EHS-B Imperfection 5 mm
EHS-B Imperfection 15 mm
200
EHS-B Imperfection 20 mm
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Total Axial Displacement (mm)
Fig. 3. Elastic range for pin ended EHS column with geometric imperfection.
The convective net heat ux is dened by the linear relation- Replacing h_ net with the convection and radiation summation of
ship shown in Eq. (6), and the radiative net heat ux by Eq. (7). A Eq. (8) produces Eq. (12).
conguration factor (F) of 1.0 is used for members completely
fac Yf ,t Ym,t em :s:Y4f ,t Y4m,t gDt Dya,t :Am 1 :V:ca,t :ra 12
engulfed in re, this is true for furnace experiments. The surface
emissivity (em ) of the member is normally assumed as 0.8, if not Resolving Eq. (12) determines the temperature development of
otherwise stated. The gas temperatures (Yg ) and radiation tem- the member surface, (Ym,t ) through heat transfer by convection
peratures (Yr ) will be replaced with one parameter called furnace and radiation induced by furnace temperatures, surrounding the
temperatures (Yf ). The emissivity of re (ef ) is 1.0 and the Stefan member. An iterative based process is required to resolve this
Boltzmann constant (s) is 5:67E8 W=m2 K 4 . With temperatures in equation, a time step of 5 s was used, as recommended by EC3
Kelvin, the net heat ux now becomes Eq. (8). [21]. The furnace temperatures (Yf ,t ) were calculated using the
appropriate TTC with coefcient of heat transfer by convection
h_ net,c ac Yg Ym 6
(ac ) detailed in EC1 Part 12, Section 3.2. [25] Fig. 4 illustrates
the calculated temperature development for section EHS-B, under
h_ net,r F:em :ef :s:Yr 2734 Ym 2734 7 the hydrocarbon and standard temperaturetime curves using the
aforementioned method.
h_ net ac Yf ,t Ym,t em :s:Y4f ,t Y4m,t 8
5.2.2. Modelling the boundary heat transfer conditions
The temperaturetime curve (TTC) equations for gas tempera- To represent the convection and radiation heat transfer con-
tures in a compartment with corresponding coefcients of heat ditions of the furnace, the DIANA boundary ow element BQ4HT
transfer by convection are outline in EC1-1-2 Section 3.2. The was used to model the heat ux at the boundary of the EHS
hydrocarbon TTC for furnace gas temperatures is shown in Eq. (9), elements. This heat ux per unit area (qc ) at the boundary is
with a corresponding coefcient of heat transfer by convection represented by the linear Eq. (13).
(ac ) of 50 W/m2K. qc hc T F T S 13
Yf ,t 108010:325e0:167t 0:675e2:5t 20 9 Parameters T S and T F are the temperatures of the steel and
furnace, respectively, and hc is the convection coefcient in
A method for determining the increase in the member tem- W/m2K. Eq. (13) models the convective heat transfer of Eq. (6).
perature for unprotected steel members is dened in equa- To model the radiative heat transfer using the BQ4HT linear
tion 4.25 of EC3 Part 12, [21] this temperature development boundary element, the radiative quadratic equation dened in
with respect to time is dened as Eq. (10). Eq. (7), must be represented in the linear form as shown in
Am =V _ Eq. (14) [26].
Dya,t ksh h net Dt 10
ca ra qr hr T F T S 14
Similar to other hollow section types, for an EHS fully The radiative heat transfer coefcient (hr ) is therefore dened
embedded in re, the shadow effect does not inuence tempera- as Eq. (15), to enable this linear form.
ture development, therefore ksh is unity. Rearranging Eq. (10)
hr em :s:T F 2 T S 2 T F T S 15
(conservation of energy) denes Eq. (11).
A combined heat transfer coefcient is a summation of
h_ net Dt Dya,t :A1
m :V:c a :ra 11 the radiative heat transfer and the convective heat transfer
T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236 27
1200
1000
800
Temperature (C)
600
Hydrocarbon TTC
0
0.00 5.00 10.00 15.00 20.00 25.00 30.00
Time (Minutes)
Table 3
Combined heat transfer coefcients (hcom) for each re curve, W/m2K.
Temperature (1C) 20 50 100 200 300 400 500 600 700 800 900 1000
Hydrocarbon TTC 55 92 117 147 170 195 223 256 293 341 387 442
Standard TTC 30 48 60 79 97 119 143 173 209 260 321 400
Exp Furnace EHS-A 30 30 31 34 36 38 40 42 45 47 49 51
Exp Furnace EHS-B 32 33 35 38 41 45 48 52 55 58 62 65
1100
1000
900
800
700
Temperature (C)
600
500
coefcients, Eq. (16). Table 3 details the combined heat transfer coefcients in
W/m2K used to represent a hydrocarbon and ISO standard re
hcom hr hc 16 for the FEM model.
30
25
Axial Displacement (mm)
20
15
10
0
0 200 400 600 800 1000 1200
Temperature (C)
to match temperatures observed at the hotspot during the 5.3. FEM staggered heat-ow stress analysis (unrestrained)
experiments. The remaining section surface area was customized
to the average steel surface experiment temperatures, using a Applying the corresponding compressive loads to the cali-
variable convection coefcient parameter, specic to the furnace brated thermal model and buckling imperfection factors, the
and section type. The same variable convection coefcient para- stability of the column was analysed under both temperature
meters could not be used for both section types, as the furnace and compressive loads, and compared with the experimental
size was not scaled for each section. The proximity of the section ndings.
sides to the furnace walls was closer for the larger sections,
creating small differences in convection and radiation distribution 5.3.1. Validating FEM model with free thermal expansion theory
within the furnace. However, for the hotspot area, a constant When the EHS column was exposed to elevated temperatures,
convection coefcient of 110 W/m2K was deemed to satisfy the the column increased in length; this thermal elongation for
experimental results, for both sections. Table 3 outlines these carbon steel is described in Section 3.4.1 of EC3 Part 12. An
calibrated combined convection parameters. The experiment approximation of the thermal expansion can be calculated using
average steel temperature proles were obtained by using these Eq. (18). Before any compressive loads were applied to the FEM
convection parameters, paralleled with the hydrocarbon TTC of model, the thermal elongation of the model was compared with
Eq. (9). Fig. 5 illustrates the calibrated steel surface temperatures theoretical hand calculations to verify functionality. To simplify
at the hotspot and average steel surface temperatures from the the hand calculations a constant thermal expansion coefcient (a)
experiment results and FEM model. Fig. 6 displays the FEM of 1.2E 05 1C 1 was used to validate the FEM model, with the
transient heat prole observed for section EHS-B, the area of the theoretical equation (Fig. 7). Table 4 shows these comparisons
hotspot is evident in this illustration. illustrating an error tolerance of between 3% and 6%.
u a:L:DT s 18
14
13
12
11
10
9
8
7
Displacement (mm)
6
5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
-1
-2
-3 EHS-A2-U Experiment FEM_A_U_45_5_EC FEM_A_U_45_10_EC
-4 FEM_A_U_45_15_EC FEM_A_U_45_20_EC FEM_A_U_45_25_EC
-5 FEM_A_U_45_30_EC FEM_A_U_45_30_EC -1.2E-05
-6
Time (Minutes)
maximum axial displacement of the experimental data. The axial coefcient with a constant value of 1.2E-05 1C 1, has helped to
displacement is inuenced more by the thermal elongation align the FEM model with the maximum axial displacement.
coefcient. From the experimental results, any measurement of For the EHS-A section type, an imperfection of approximately
axial displacement is delayed until the steel temperature is 30 mm and a constant thermal expansion coefcient of 1.2E-
approximately 100 1C. This delay in measurement relates to 05 1C 1 produced a close approximation with experimental
slackness and bedding within in the test rig, during the start of results, an exception for EHS-A1 applied as an imperfection
the heating process. Therefore, delaying the onset of thermal increase to 32.5 mm was required, Fig. 9. For the more slender
expansion, and replacing the variable thermal expansion section type EHS-B, a geometric imperfection of 25 mm for all
10
5
Displacement (mm)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
-1
-2
-3
-6
Time (Minutes)
10
7
6
5
Displacement (mm)
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
-1
-2
-3
Table 5
Unrestrained experiment and FEM model comparisons.
Load ratio Imp. (mm) Time (min) Temperature (1C) Max. displacement (mm)
(aL) x Exp. FEM Error Exp. FEM Error Exp. FEM Error
EHS-A-U 0.60 30.0 10.20 10.16 0.996 389 455 1.170 5.12 6.18 1.207
0.45 30.0 12.58 12.67 1.007 504 532 1.056 6.75 6.69 0.991
0.30 32.5 14.83 15.50 1.045 586 618 1.055 8.52 8.39 0.984
EHS-B-U 0.60 25.0 10.25 9.42 0.919 460 477 1.037 5.69 6.05 1.063
0.45 25.0 11.50 10.92 0.950 519 532 1.025 7.08 7.09 1.001
0.30 25.0 14.08 13.75 0.977 597 627 1.050 8.37 8.34 0.996
10
5
Displacement (mm)
0
0 100 200 300 400 500 600 700
-1
-2
-5
Steel Average Temperature (C)
three loading levels illustrated close agreement with experiment heating-up program does not inuence the ultimate failure
axial displacements, Fig. 10. Table 5 details the comparisons for temperature of the section.
each experiment with corresponding model results and error
tolerances dened.
5.5. FEM staggered heat-ow stress analysis (restrained)
10
5
Displacement (mm)
0
0 100 200 300 400 500 600 700
-1
-2
-3
EHS-B1-U Experiment EHS-B2-U Experiment EHS-B3-U Experiment
-4 FEM_B_U_60_25 FEM_B_U_45_25 FEM_B_U_30_25
-5
Steel Average Temperature (C)
10
5
Displacement (mm)
0
0 100 200 300 400 500 600 700
-1
-2
-3
EHS-B1-U Experiment FEM_B_B_U_60_25_HS FEM_B_U_60_25_UH
-4
-5
Steel Average Temperature (C)
restrained experiment data. Fig. 14 and Fig. 15 show the axial column type EHS-A, the accuracy of the approximate solution
displacements for the restrained EHS columns of EHS-A-R and deviated as the loading level increased from 30% to 60%. Table 6
EHS-B-R, respectively. The FEM model provided an extremely outlines the experimental results with the FEM model counter-
close approximate solution for the more slender column type parts with error tolerance. Fig. 16 and Fig. 17 compare induced
EHS-B, for the three ranges of loading. However, for the stockier restraint forces between the experiment and FEM model results.
T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236 33
2
Displacement (mm)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-3
-5
Time (Minutes)
3
Displacement (mm)
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1
-2
-4
Time (Minutes)
Table 6
Restrained experiment and FEM model comparisons.
Load ratio Imp. (mm) Time (min) Temperature (1C) Max. displacement (mm)
(aL) x Exp. FEM Error Exp. FEM Error Exp. FEM Error
EHS-A-R 0.60 30.0 10.13 9.60 0.948 417 434 1.041 3.31 4.44 1.341
0.45 30.0 11.40 10.50 0.921 470 470 1.000 4.35 4.90 1.126
0.30 30.0 12.27 12.50 1.019 501 534 1.066 5.51 5.42 0.984
EHS-B-R 0.60 25.0 10.03 9.08 0.905 430 459 1.067 3.37 3.80 1.128
0.45 25.0 10.88 9.58 0.881 475 480 1.011 4.32 4.34 1.005
0.30 25.0 13.02 12.58 0.966 556 590 1.061 5.56 4.92 0.885
400
EHS-A1-R Experiment
250
200
150
100
50
0
0 100 200 300 400 500 600 700
Steel Average Temperature (C)
proles, axial displacement development, induced restraint force introduction of a slack coefcient to the restrained FEM models,
measurements and ultimate failure times. Using the mechanical allowed for the recommended EC variable thermal expansion
and thermal properties dened for structural carbon steel in the coefcient to be reintroduced. A combination of non-uniform
EC publications, a number of Finite Element models were created heating and inherent test rig slackness, contributed to these
and compared with the collated experimental data. anomalies in the thermal expansion of the EHS columns, between
Using the EC thermal parameters, the FEM model demon- the FEM model and experimental data.
strated an excellent approximation with the recorded experimen- For the axially restrained FEM models the same geometric
tal temperatures for unprotected steel. The EC mechanical imperfections were used. A very close agreement with the
properties for carbon steel at elevated temperatures were then experiment axial displacement was displayed for EHS-B, the more
utilised, for the potential ow structural analysis. Calibrating the slender column type. For the stockier EHS-A column, the FEM
unrestrained EHS FEM model axial displacements, with the model accuracy deviated as the loading level increased. A number
experimental data, highlighted how critical the geometric imper- of reasons may contribute to this divergence between experi-
fection is in the ultimate failure time of the heated EHS column. mental and FEM model results. First, a perfect pin ended
Experiment loading and alignment eccentricities were accounted condition is assumed for the FEM model, any increase in the
for in this geometric imperfection. Imperfection values of 30 mm friction coefcient on the graphite end bearings was not consid-
for EHS-A and 25 mm for EHS-B produced a close approximation ered. This friction would increase with thermal expansion
with the unrestrained ultimate failure times. directly inuencing the columns effective length during the
The thermal expansion parameter had a signicant impact on experiment. This increase in friction at the end bearings also
the axial displacement of the EHS columns, substituting the produces additional moments that are not considered in the
variable thermal expansion coefcient suggested by EC 3, with a FEM model.
constant value of 1.2E-05 1C 1 help to aligned the FEM maximum Overall, the FEM model calibration process with the experi-
model displacements with the experiment values. The ment data has demonstrated that a very good numerical
T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236 35
400
EHS-B1-R
350 EHS-B2-R
EHS-B3-R
FEM_B_R_60_25
300
FEM_B_R_45_25
FEM_B_R_30_25
Restraint Force (kN)
250
200
150
100
50
0
0 100 200 300 400 500 600 700
Steel Average Temperature (C)
400
EHS-A3-R Experiment
350 FEM_A_R_30_30
FEM_A_R_30_30_EC -1.9 mm Slack
300
Restraint Force (kN)
250
200
150
100
50
0
0 100 200 300 400 500 600 700
Steel Average Temperature (C)
Fig. 18. EHS-A3-R restraint force development, EC parameters with 1.9 mm slack.
approximate solution can be obtained, on careful consideration to on the EHS columns behaviour. Ultimately, the investigation
the thermal expansion coefcient, and geometric imperfection of recommends that the thermal and mechanical properties, of
the column. However, the investigation has highlighted that structural carbon steel, outlined in the EC publications, are
further FEM numerical analysis is needed to successfully model deemed suitable, for the thermal analysis of unprotected unlled
thermal creep, to scrutinise the impact of the heating-up program Elliptical Hollow Section steel columns, in re.
36 T. Scullion et al. / Thin-Walled Structures 55 (2012) 2236
Acknowledgments [12] Zhu Y, Wilkinson T. Finite element analysis of structural steel elliptical
hollow sections in pure compression. Welding in the World 2006;50(no.
SPEC. ISS.):16370.
The authors would like to take this opportunity to thank the [13] Gardner L, Chan TM. Structural design rules for elliptical hollow sections.
DELNI Department for Employment and Learning, Northern presented at the 12th International Symposium on Tubular Structures
Ireland for providing funding for this research, and extend their 2009:3239.
[14] Espinos A, Gardner L, Romero ML, Hospitaler A. Fire behaviour of concrete
thanks also, for the guidance provided by TNO-Diana technical lled elliptical steel columns. Thin Walled Structures 2011;49(2):23955.
support team. [15] Scullion T, Ali F, Nadjai A. Experimental study on performance of elliptical
section steel columns, under hydrocarbon re. Journal of Constructional Steel
Research 2011;67(6):98691.
[16] Scullion T, Ali F, Nadjai A. Effect of axial restraint on the performance of
References elliptical hollow section steel columns, in hydrocarbon re. Engineering
Structures 2011;33(12):315561.
[17] DIANA Finite Element Analysis Users Manual Element Library Release 9.4.2,
[1] Eurocode 3 BS EN 1993-1-4:2006 design of steel structures. General 1st edn. 2010.
Rules. Supplementary Rules for Stainless Steels. [18] BS EN 10210-1:2006 Hot nished structural hollow sections of non-alloy and
[2] Feinstein G, Erickson B, Kempner J. Stability of oval cylindrical shells. ne grain steels. Technical Delivery Requirements BSI British Standards.
Experimental Mechanics 1971;11(11):51420. [19] BS EN 10210-2:2006 Hot nished structural hollow sections of non-alloy
[3] Chen YN, Kempner J. Buckling of oval cylindrical shells under compression and ne grain steels. Tolerances, Dimensions and Sectional Properties BSI
and asymmetric bending. AIAA Journal 1976;14(9):123540. British Standards.
[4] Gardner L, Chan TM. Cross-section classication of elliptical hollow sections. [20] Eurocode 3 BS EN-1993-1-1:2005 Design of steel structures: General
Steel and Composite Structures 2007;7(3):185200. rules and rules for buildings.
[5] Chan TM, Gardner L. Flexural buckling of elliptical hollow section columns. [21] Eurocode 3 BS EN 1993-1-2:2005 design of steel structures. General
Journal of Structural Engineering 2009;135(5):54657. Rules. Structural Fire Design.
[6] Chan TM, Gardner L. Bending strength of hot-rolled elliptical hollow sections. [22] CT29 Celsius 355s Ovals Sizes and Capacities Eurocode Version. Corus
Journal of Constructional Steel Research 2008;64(9):97186. Tubes, 2004.
[7] Chan TM, Gardner L. Compressive resistance of hot-rolled elliptical hollow [23] Dai X, Lam D. Numerical modelling of the axial compressive behaviour of
sections. Engineering Structures 2008;30(2):52232. short concrete-lled elliptical steel columns. Journal of Constructional Steel
[8] Ruiz-Teran AM, Gardner L. Elastic buckling of elliptical tubes. Thin Walled Research 2010;66(7):93142.
Structures 2008;46(11):130418. [24] Timoshenko S. Theory of Elastic Stability. 2nd edn. New York: McGraw-Hill;
[9] Silvestre N. Buckling behaviour of elliptical cylindrical shells and tubes under 1988.
compression. International Journal of Solids and Structures 2008;45(16): [25] Eurocode 1 BS EN 1991-1-2:2002 Actions on structures. General actions.
442747. Actions on Structures Exposed to Fire BSI British Standards. BSI, 2002.
[10] Yang H, Lam D, Gardner L. Testing and analysis of concrete-lled elliptical [26] H.G. Ransing, How To Undertake Finite Element Based Thermal Analysis.
hollow sections. Engineering Structures 2008;30(12):377181. NAFEMS, 1999.
[11] Zhao XL, Packer JA. Tests and design of concrete-lled elliptical hollow [27] Twilt L. Strength and deformation properties of steel at elevated tempera-
section stub columns. Thin Walled Structures 2009;47(67):61728. tures some practical implications. Fire Safety Journal 1986;13(1):915.