Vol. 31, 2023, No.

2, 9 – 15
DOI: 10.2478/sjce-2023-0008

VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL

MODEL FOR THE FIRE DESIGN OF STEEL MEMBERS

Samer Nemer 1 *

Abstract Address

In this paper, the modified Ramberg-Osgood constitutive 1

 ept. of Structural and Geotechnical Engineering, Széchenyi
D
equations for calculating the stress-strain of steel at elevated István University, Győr, Hungary
temperatures using the parameters determined based on the
transient state tensile test results achieved at the Helsinki Uni- * Corresponding author: samer.nemer.91@gmail.com
versity of Technology are verified. This is done by numerically
comparing the global and local buckling capacities of I-shaped Key words
steel members incorporating the modified Ramberg-Osgood
model along with the material model given in the fire section ●● Ramberg-Osgood,
●● Fire design,
of Eurocode EN1993-1-2. For this purpose, a numerical model
●● Structural steel,
using the ABAQUS software was developed. Then, nonlinear
●● Material model,
analyses with imperfections (GMNIA) were performed to com-
●● Elevated temperatures,
pare the buckling capacities of the steel columns and beams of ●● Nonlinear analysis.
four different hot-rolled cross-sections (IPE160, IPE180, HE100B,
HE500B), made of steel grade S355, at three different temper-
atures (400°C, 500°C, 600°C). The results showed that adopting
the modified Ramberg-Osgood model can lead to the same
buckling capacities resulting when using the EN1993-1-2 ma-
terial model for steel temperatures of less than 400°C. However,
adopting this model for 600°C overestimates the buckling ca-
pacities in most cases.

1 Introduction In the literature available, the material models of steel at ele-

vated temperatures are usually given in the form of stress-strain
1.1 Literature study curves. The mechanical properties of steel (e.g., the modulus
of elasticity, yield stress and ultimate strength) can be obtained
Numerical modeling is considered nowadays to be an ef- from these curves. These data are obtained from two types of
fective tool for investigating and proposing new design meth- tensile tests of steel at elevated temperatures, namely, steady
ods. An extremely important step to build a reliable numerical state and transient state tests. In steady-state tests, the specimen
model, which is able to predict the actual buckling capacity of is first heated to the required temperature before straining to fail-
steel members at elevated temperatures, is the correct appli- ure. The specimen in transient-state tests is first subjected to the
cation of the accurate mechanical properties of the material/ load, and then the temperature is increased until failure. Kirby
materials used. It is a well-known fact that when the tempera- and Preston (1988) reported that transient-state tests give lower
ture increases, the mechanical properties of steel, especially its values of strengths than the values obtained from steady-state
strength and stiffness, reduce dramatically, and this consider- tests on carbon steel for a temperature range of (400 – 800) ℃.
ably affects the structural response of steel members and may There are many material models available in the literature. For a
lead to structural instability in steel structures as a whole. practical design, EN 1993-1-2 (2005) adopts Rubert and Schau-

© 2023 The Author(s). This is an open access article licensed under the Creative Commons Attribution
4 .0 License (https://creativecommons.org/licenses/by/4.0/)
 Vol. 31, 2023, No. 2, 9 – 15

mann’s model, which is derived from transient-state tests on are able to predict the real behavior of steel members at elevat-
beams (Rubert and Schaumann, 1986). In addition to the dete- ed temperatures. Additionally, improving the accuracy of the
rioration of strength and stiffness when the temperature increas- material models incorporated in a fire design is also important
es, the clear yield point at a normal temperature disappears at with regard to the safety and economy of a steel structural de-
elevated temperatures, and the yield strength becomes difficult sign. Based on the available literature given above, it can be
to identify as the behavior becomes highly nonlinear with the stated that there is a lack of certainty in and many disadvan-
rising temperature. For this reason, and because of large strains tages of the current EN1993-1-2 material model. Therefore,
shown in steel members at elevated temperatures, it is more usu- this paper aims at evaluating the applicability of the modified
al to consider the effective yield stress at a total strain within a Ramberg-Osgood material model as an alternative material
range between 0.2% to 2.0% (ECCS, 1989; BS 5950, 2000). model for the fire design of steel members. The parameters for
EN 1993-1-2 adopts an effective yield strength at a 2.0% total this fire design model were presented by (Outinen et al., 1997)
strain for classes 1,2,3 cross-sections. For class 4 cross-sections, based on an actual test that was not investigated or verified
the 0.2% proof strength should be used (EN 1993-1-2, 2005). later by other researchers.
The material model investigated in this paper is that given by
(Outinen et al., 1997) it was proposed using the calculating meth-
od developed by (Ramberg and Osgood, 1943); the mechanical 1.3 The content of the paper
properties of S355 steel at elevated temperatures are calculated
using simple formulas based on transient state tensile test results In this paper, the EN1993-1-2 and the modified Ram-
carried out at the Laboratory of Steel Structures at the Helsin- berg-Osgood material models are first presented; then the nu-
ki University of Technology (SFS-EN 10 002-5, 1992; EN 10 merical model is explained. Subsequently, the local buckling
002-2, 1992; EN 10 002-4, 1992). These formulae are proposed capacity of a steel beam studied by (Prachar et al., 2016) and
instead of the reduction factors given in EN1993-1-2. Wang et al. the global buckling strengths of columns and beams are cal-
(2012) reported that the Ramberg-Osgood equations are capable culated by a geometrically and materially nonlinear analysis
of providing a convenient way to express stress-strain curves as (GMNIA) incorporating the two material models.
a continuous function. Very important research was conducted
by (Pauli et al., 2012), who carried out extensive experimental
investigations on the material behavior of carbon steel at elevat- 2 M ATERIAL MODELS AT ELEVATED
ed temperatures and on structural stub and slender columns on TEMPERATURES
fire. It was reported that the EN1993-1-2 material model for car-
bon steel has difficulties in describing the stress-strain relation- This section covers the description of two material models,
ships from tensile tests due to the fact that it overestimates the which represent the material response of carbon steel at elevat-
strain hardening for strains smaller than 2% and underestimates ed temperatures, i.e., (1) the EN1993-1-2 material model and
it for larger strains. Moreover, the shape of the modelled stress- (2) the Ramberg-Osgood material model.
strain curve of EN1993-1-2 cannot be adapted to the individu-
al stress-strain relationships of the experimental results. On the 2.1 The EN1993-1-2 material model
other hand, adapting the one-stage Ramberg-Osgood model and
its modification by Gardner-Nethercot [Gardner and Nethercot, The material model for carbon steel at elevated tempera-
2004), has led to the better modelling of experimentally obtained tures, as given in the fire section of Eurocode (EN1993-1-2,
individual stress-strain relationships of different steel grades and 2005), is presented in Fig. 1. The response of the material is
temperatures. Moreover, for high temperatures above 600 °C, it divided into four stages as follows:
has been shown that the stress-strain relationship has an almost The first stage is linear-elastic up to the proportional limit
bilinear shape (Pauli, 2012). Another disadvantage of the Euro-
code material model was reported by (Wang et al., 2012) who
stated that the rate of change of the gradients (the tangent modu-
lus) at the points between the limit of proportionality fp θ and yield
points fyθ is not continuous, even though the values of the stress
and tangent modulus are continuous. According to (Knobloch et
al., 2010), the effect of the nonlinear stress-strain relationship of
steel at elevated temperatures on the overall buckling strength is
of high importance. In addition, it was concluded that adopting
the temperature-dependent stress reached at a 2 % strain leads to
unreliable results for the cross-sectional capacity in pure com-
pression. Moreover, the unsafe nature of the EN1993-1-2 curve
was also reported by (Nemer and Papp, 2021).
Fig. 1 EN1993-1-2 material model for steel at elevated
temperature θ [1]
1.2 The aim of the current research

Thorough knowledge and proper implementation of the point fp,θ (the end of the elastic stage), at which the stress is
behavior of materials at elevated temperatures is extremely proportional to the strain. At this stage, only two basic material
important for developing accurate numerical models, which parameters are needed, i.e., the slope of the linear elastic range

10 VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL MODEL FOR THE FIRE DESIGN...

 Vol. 31, 2023, No. 2, 9 – 15

for steel at elevated temperatures (Eθ) and also the proportional σi , εi: represent the stress and corresponding strain, respec-
limit for steel at elevated temperatures (fp,θ). Then, the linear tively, at temperature θ
elastic range is followed by an elliptical curve between the nθ: coefficient that enables the curvature to be adjusted;
proportional limit fp,θ to the effective yield strength of steel at β = 6/7.
an elevated temperature (fy,θ), which is defined as the strength
at a 2% total strain. In the third stage, the stress remains con- Outinen et al. (1997) proposed simple formulas to deter-
stant between εy,θ = 2% and the limiting strain for the yield mine the parameters of the Ramberg-Osgood equation (2,3,4)
strength, εt,θ = 15% . In the last stage the stress drops to zero at based on the test results for steel grade S355, as follows:
the ultimate strain εy,θ = 20%, which means a fracture.
Fig. 2 shows the stress-strain curve according to the fire For 200℃ ≤ θ ≤ 700℃:
section of Eurocode EN1993-1-2 for S355 steel at three diffe­
rent temperatures (400, 500, 600 ℃). (2)

In EN1993-1-2, the strength and stiffness degradation of For 200℃ ≤ θ ≤ 700℃:

(3)

For 200℃ ≤ θ ≤ 700℃:

(4)

Fig. 2 EN1993-1-2 stress strain curve at elevated temperatures

for S355

steel at an elevated temperature is presented using reduction

factors, where the property at an elevated temperature is nor-
malized with respect to the equivalent property at an ambient
temperature. Table 1 presents the reduction factors for the ef-
fective yield strength (ky,θ = fy,θ / fy ), proportional limit (kp,θ = fp θ /
fy and elastic modulus (kθ = Eθ /E) at the three temperatures
(400, 500, 600 ℃) investigated, where fy E are the yield
strength and the modulus of elasticity of the normal tempera- Fig. 3 Modified Ramberg-Osgood material model for S355 steel
ture design, respectively.

Tab. 1 Reduction factors at steel temperature θ as given in

EN1993-1-2 3 NUMERICAL MODEL

Temperature 3.1 General

400 ℃ 1.00 0.42 0.70
An advanced geometrically and materially nonlinear analy­
500 ℃ 0.78 0.36 0.60 sis with imperfections included (GMNIA) using ABAQUS
software (ABAQUS 3DEXPERINECE r2018x) was used for
600 ℃ 0.47 0.18 0.31
calculating the ultimate load capacity of the members studied.
The validation of this program was published in (Hajdú and
Papp, 2018) and (Hajdú, 2021). The members are modelled
2.2 T he Modified Ramberg-Osgood model based on using the general purpose S4R shell elements.
a real transient test

The equation for calculating the stress-strain curve of steel

at elevated temperatures proposed by (Ramberg and Osgood, 3.2 Study program
1943), is the following: Four sections were studied to recognize the effect of mate-
rial models on steel members with different sections subject-
(1) ed to an axial force or a bending moment. The dimensions of
these cross-sections are presented in Table 2.

VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL MODEL FOR THE FIRE DESIGN... 11

 Vol. 31, 2023, No. 2, 9 – 15

Tab. 2 Characteristics of different sections investigated in this paper The load was modelled by applying the distributed forces
on the flanges and on the web of the loaded end using the mod-
Cross-section h b tw tf r ified RIKS tool, which is available in the ABAQUS library
IPE160 160 82 5.0 7.4 9 (ABAQUS, 2018).

IPE180 180 91 5.3 8.0 9

HE100B 100 100 6.0 10.0 12 3.5 Imperfections
HE500B 500 300 14.5 28.0 27 The residual stresses were taken into account by introduc-
ing the ECCS type model for hot-rolled cross-sections (ECCS,
1976) into the numerical model. The amplitude of the initial
stress depends on the height-to-width ratio of the section in-
3.3 Meshing size vestigated, as shown in Fig. 6.
The initial geometric imperfections of the members were
The mesh size was determined to be 16 elements in the taken into account by performing a linear buckling analysis
flange, and 16 elements in the web depth, while along the (LBA) on the perfect prismatic member with the given bound-
member’s length, the size of the elements was 20 mm, as ary conditions; then the relevant normalized global buckling
shown in Fig. 4. mode was extracted. Thus, the first global buckling mode shape
derived from the linear buckling analysis was introduced into
the non-linear finite element model (GMNIA), and multiplied
by the amplitude of the initial geometrical imperfection. The
nodal coordinates of the model were updated by adding the
nodal imperfections established. The amplitude of the initial
geometrical imperfection of the column is considered to be
equal to L/1000, where L is the member length. The ampli-
tude of the initial geometrical imperfection of the column is
taken to be L/1000, which is widely used in the literature, and
corresponds to 75% of the recommended tolerance value of
L/750 for a steel column in Annex D of EN1090-2:2008 (BS
EN 1090 2, 2008).

Fig. 4 Meshing of the ABAQUS model

3.4 Load and boundary conditions

For the sake of simplicity, the members investigated in this

study were simply-supported; i.e., two reference points, one at
each end, were coupled with the nodes of both end surfaces of
the members using kinematic coupling restraints. Then, these
reference points were restrained against all degrees of freedom
except for the displacement in the direction of the load applied
at the loaded end and the rotations about the axes of the buck-
ling at both ends, as shown in Fig. 5.
Fig. 6 Considered residual stress patterns for h/b<1.2 (left) and
h/b>1.2 (right) [16]

It is worthwhile mentioning that in the finite element mod-

els, the true stress and plastic strain were adopted instead of
the engineering stress and strain. Therefore, the following
equations were used to represent the relationship between the
true stress and plastic strain:

(5)

(6)

Fig. 5 ABAQUS model: Boundary conditions and load application (7)

12 VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL MODEL FOR THE FIRE DESIGN...

 Vol. 31, 2023, No. 2, 9 – 15

Where: 4.2 Comparative study of columns

σtrue: true stress
σnom: engineering or nominal stress Following the comparison of the lateral torsional buck-
εnom: engineering or nominal strain ling of the steel beam presented in the previous section, the
εtrue: true strain two material stress-strain curves were incorporated in the
εpl: plastic strain ABAQUS model to calculate the capacities (N) of steel com-
pressed members in a fire with varying member lengths (slen-
derness ratios) and under three different temperatures.
4 Results Fig. 8 shows the results of the numerical models for the
columns made of IPE160 and HE100B cross-sections at three
4.1 Comparison with the numerical simulation given different temperatures (400 ℃, 500 ℃, and 600 ℃), where
by Prachar et al. Npl,t,Rd is the plastic compressive resistance of the cross-section
at elevated temperatures.
A number of class 4 (slender) welded steel beams were It can be seen that adopting the Ramberg-Osgood model
investigated numerically and experimentally at elevated tem- can result in similar buckling capacities to those predicted
peratures by (Prachar et al., 2016). The two material models by using the EN1993-1-2 material model for steel columns
investigated in this paper were employed to simulate a heat- at 400℃ and 500℃. However, for columns under 600℃, the
ed Class 4 steel beam under lateral torsional buckling at 450 Ramberg-Osgood model overestimates the buckling capacities
℃ presented in (Pauli, 2012) as (TEST 6). It is worthwhile for various slenderness ratios and cross-sections.
mentioning that in this model, the yield strength is based on
the strain at 0.2% proof strength for both materials used. The
dimensions of the cross-section of the beam along with the
capacities calculated by Prachar et al., and by the numerical
model described above incorporating the two material models
are presented in Table 3.

Tab. 3 Numerical results using the two material models

Cross-section (mm) Load capacities (kN)

NO. hw X tw Prachar GMNIA GMNIA

bf X tf et al. (EN1993-1-2) (R-O)
Test 6 446 X 4 150 X 7 151.84 151.63 149.14

The total load-displacement curves of the same numeri-

cal simulations are shown in Fig. 7. Based on the results pre-
sented in Table 3 and Fig. 7, it can be said that adopting the
modified Ramberg-Osgood material model in the numerical
model results in a capacity similar to that calculated using the
EN1993-1-2. However, there is a clear difference in the de-
flection, which is due to the difference in the Young’s modulus
value, as follows:
According to EN1993-1-2, the Young’s modulus at 450 ℃
is Eθ = 136500 N/mm2; while according to the test results giv-
en by (Outinen et al., 1997), Eθ = 116750 N/mm2.

Fig. 8 Comparison of the buckling capacities for columns made of

IPE160 (above) and HE100B (under)

4.3 Comparative study of beams

In this section, the two material models were used for calculat-
ing the buckling capacities of steel members under pure bending.
The results of the comparison of the lateral torsional buck-
ling capacities (M) for members made of IPE180 and HE500B
Fig. 7 Comparison of the buckling capacity-displacement curves of cross-sections at three different temperatures 400 ℃, 500 ℃,
the beam investigated and 600 ℃ are presented in Fig. 9, where Mpl,t,Rd is the plastic

VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL MODEL FOR THE FIRE DESIGN... 13

 Vol. 31, 2023, No. 2, 9 – 15

Fig. 9 Comparison of the buckling capacities for columns made of IPE160(Left) and HE100B (Right)

moment resistance of the cross-section at elevated tempera- Acknowledgments

tures.
Again, it can be seen that the two material models lead to This research did not receive any specific grant funding from
almost the same buckling resistances for steel beams at 400℃ agencies in the public, commercial or not-for-profit sectors.
and 500℃. However, for the beams at 600℃, the Ramberg-Os-
good material model overestimates the buckling capacities
compared to the capacities calculated with the EN1993-1-2 Disclosure statement
material model mainly for short beams, while this difference
reduces for long beams (1.5), where both material models lead The authors declare that they have no known competing
to similar capacities. financial interests or personal relationships that could have ap-
peared to influence the work reported in this paper.

5 CONCLUSION

This paper presents a numerical investigation of the veri-

fication of the Ramberg-Osgood material model with param-
eters calculated based on a transient test achieved at the Hel-
sinki University of Technology. A finite element model was
developed and verified; then three different cases were investi-
gated, i.e., a Class 4 steel beam at 450℃, a comparative study
on columns with different slenderness ratios and three different
temperatures (400, 500, 600℃), and a similar study on beams.
It can be concluded that the modified Ramberg-Osgood leads
to buckling capacities that correspond to those given by adopt-
ing the EN1993-1-2 material model for beams and columns at
400℃ and 500℃. However, at 600℃, the Ramberg-Osgood
model overestimates the buckling capacities for most cases.
Therefore, the parameters should be modified, and more con-
servative values should be adopted.

14 VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL MODEL FOR THE FIRE DESIGN...

 Vol. 31, 2023, No. 2, 9 – 15

References

British Standard BS 5950, Part 8, Code of Practice for Fire Resis- Outinen, J. - Kesti, J. - Mäkeläinen, P. (1997) Fire Design Mod-
tance Design. British Standard Institution. 2000. el for Structural Steel S355 Based Upon Transient State
BS EN 1090 2 (2008) Execution of steel structures and aluminium Tensile Test Results. Journal of Constructional Steel Research.
structures Part 2: Technical requirements for steel structures. 42(3), pp 161-169.
2008. BSI, London, UK. Pauli, J. - Somaini, D. - Knobloch, M. - Fontana, M (2012) Ex-
EN 1993-1-2 (2005) Eurocode 3, Design of steel structures – Part periments on steel under fire conditions. ETH Zurich, Institute
1–2: General rules Structural fire design. In: European of Structural Engineering. IBK test report No. 340, Zurich, Swit-
committee for standardization. Brussels, Belgium; 2005. zerland.

European Convention for Construction Steelwork. Calculation for Pauli, J. (2012) The Behavior of Steel Columns in Fire.. IBK Report
Fire Resistance of Composite Structures. Technical Note No. 55, No. 343, ETH Zurich: Zurich.
ECCS Technical Committee 3, 1989. Prachar, M. - Hricak, J. - Jandera, M., - Wald, F. - Zhao, B. (2016)
Experiments of Class 4 open section beams at elevated tempera-
European Convention for Constructional Steelwork (ECCS).
ture. 2016. Thin-Walled Structures, Vol. 98, Part A, pp. 2-18,
(1976) Manual on stability of steel structures, 2nd ed. Technical
ISSN 0263-8231. https://doi.org/10.1016/j.tws.2015.04.025
Committee 8, Structural stability. 1976. www.eccspublications.eu
Gardner, L. - Nethercot, N. A. (2004) Experiments on stainless steel hol- Ramberg, W. - Osgood, W. R. (1943) Description of stress-strain
low sections – Part 1: Material and cross-sectional behavior. curves by three parameters. NACA Technical note 902.
Journal of constructional steel research. 2004. 60(9), pp. 1291-1318. Rubert, A. - Schaumann, P. (1986) Structural steel and plane frame
Hajdú, G. - Papp, F. (2018) Safety assessment of different stability assemblies under fire action. Fire Safety Journal. 10, pp. 173 –
design rules for beam-columns. Structures. 2018. (14), pp. 376– 184.
388. https://doi.org/10.1016/j.istruc.2018.05.002. Standard EN 10 002-2 (1992) Metallic materials. Tensile testing.
Hajdú, G. (2021) Buckling resistance of steel beam-columns - the Part 2: Verification on the load cell of tensile testing machine,
verification of Overall Imperfection Method with numerical Brussels, 1992. 3.
method. (In Hungarian), dissertation, 2021. Standard EN 10 002-4 (1992) Metallic materials. Tensile testing.
Kirby, B.R. - Preston, R.R. (1988) High temperature properties of Part 4: Verification of extensometers used in uniaxial testing,
hot-rolled structural steels for use in fire engineering studies. Fire Brussels, 1992.
Safety Journal. 13(1), pp. 27–37. Standard SFS-EN 10 002-5 (1992) Metallic materials. Tensile test-
Knobloch, M. - Somaini, D. - Pauli, J. - Fontana, M. (2010) Stabil- ing. Part 5: Method of testing at elevated temperature (in Finn-
ity of steel columns subjected to fire. In: SDSS‘Rio 2010 Stabil- ish), Helsinki, 1992. 2.
ity and Ductility of Steel Structures, Rio de Janeiro, Brazil, Sept. Wang, Y. - Burgess, I. - Wald, F. - Gillie, M (2012) Perfor-
8–10, 2010. mance-Based Fire Engineering of Structures. 2012. 10.1201/
Nemer, S. - Papp, F. (2021) Numerical Investigation on Flexural b12076.
Buckling Behavior of Hot-rolled Steel Columns at Elevated Tem-
peratures. 2021. Periodica Polytechnica Civil Engineering,
65(3), pp. 918–927. https://doi.org/10.3311/PPci.17799.

VERIFICATION OF THE RAMBERG-OSGOOD MATERIAL MODEL FOR THE FIRE DESIGN... 15