You are on page 1of 9

Front. Struct. Civ. Eng.

2014, 8(2): 178–186


DOI 10.1007/s11709-014-0245-y

RESEARCH ARTICLE

Cyclic stress-strain behavior of structural steel with yield-


strength up to 460 N/mm2
Yiyi CHENa, Wei SUNb, Tak-Ming CHANc,*
a
State Key Laboratory of Disaster Reduction in Civil Engineering, Tongji University, Shanghai 200092, China
b
Architecture Design & Research Institute of Tongji University (Group) Co. Ltd., Shanghai 200092, China
c
School of Engineering, University of Warwick, Coventry CV4 7AL, UK
*
Corresponding author. E-mail: t.m.chan@warwick.ac.uk

© Higher Education Press and Springer-Verlag Berlin Heidelberg 2014


ABSTRACT This paper presents a constitutive model based on Ramberg-Osgood equation to describe the hysteresis
material behavior of structural carbon steel with nominal yield strength between 235 to 420 N/mm2. The proposed model
was calibrated against a series of cyclic material tests with strain amplitude varying from 0.5% to 2.0%. A simple
relationship between the modular parameter K and the yield strength fy was proposed. The calibrated Ramberg-Osgood
model revealed excellent agreement with the experimental results and captured further the experimental behavior of test
specimens with nominal yield strength of 460 N/mm2. The proposed constitutive model was also adopted in conjunction
with the combined kinematic/isotropic materials description in ABAQUS to mimic a full scale experimental test under
cyclic loading. The numerical results revealed close agreement with the experimental observations.

KEYWORDS constitutive model, finite element analysis, hysteresis curve, Ramberg-Osgood equation

1 Introduction research [11] is conducted in the development of hysteresis


stress-strain models based on Ramberg-Osgood relation-
Ramberg and Osgood [1] first proposed an analytical ships for structural carbon steel. This paper presents the
formulation to describe the nonlinear behavior of material calibration of the Ramberg-Osgood equations against a
under compression and tension. Based on Ramberg and series of cyclic materials tests [12] and proposes a
Osgood relations, numerous studies have been conducted constitutive model to describe the material hysteresis
to modify and improve the relationship to cover a wider performance.
range of strain for nonlinear materials under monotonic
loads [2–6] and to describe the stress-strain relationship at
elevated temperature [7,8]. Concerning hysteresis beha- 2 Calibration
vior, Skelton et al. [9] described the relationships between
the various Masing-type models and the Ramberg-Osgood A series of experimental tests on steel coupons under
stress-strain response. A new link was established between monotonic and cyclic loadings were conducted by Chen et
the Ramberg-Osgood strength and cyclic hardening al. [12]. Mean measured dimensions and key results from
parameters. Broggiato et al. [10] also described the use direct tensile tests are summarized in Table 1. The
of the Chaboche nonlinear kinematic hardening model for specimens are categorized according to the material
cyclic plasticity assessment. The tuned Chaboche model grade as shown in columns 1 and 2, measured yield
for elastoplastic material description was employed in the strength, measured ultimate strength, ratio of yield to
finite element simulation which was capable of following ultimate strengths and measured elongation are shown in
the cyclic elastoplastic response very closely up to large columns 3 to 6 respectively. Further experimental details
accumulated plastic strain levels. However, limited can be found in Chen et al. [12].
For the cyclic material tests, three different loading
Article history: Received Mar. 10, 2014; Accepted Apr. 22, 2014 protocols 1) cyclic ascend 2) cyclic alternate and 3) cyclic
Yiyi CHEN et al. Cyclic stress-strain behavior of structural steel with yield-strength up to 460 N/mm2 179

Table 1 Mean measured dimensions and key results from direct tensile tests
1
Group 2
Grade 3
yield strength, fy/MPa 4
ultimate strength, fu/MPa 5
fy/fu 6
elongation, δ%
1 Q235C 280 457 0.61 24.8
2 Q235B 308 478 0.64 24.7
3 Q345B 350 523 0.67 24.8
4 Q345B 358 552 0.65 20.7
5 Q345B 392 539 0.73 24.0
6 Q390C 403 559 0.72 22.8
7 Q420D 412 577 0.71 22.5

Notes: 1,2,3,4,5,6 mean the column

tensile were employed as shown in Fig. 1. As shown in Fig.


1(a), cyclic ascend is a loading protocol with an equal
strain increment of +/ – 0.005 for each five-cycle loading
phase up to +/ – 0.02 strain amplitude. cyclic alternate
refers to the loading scheme as shown in Fig. 1(b) which is
similar to cyclic ascend, except the second and third
loading phases were interchanged. The loading scheme of
cyclic tensile is indicated in Fig. 1(c). All loadings are in
tension and it oscillates incrementally up to 0.08 strain
amplitude. Previous research [12] revealed that the loading
protocols in particular between cyclic ascend and cyclic
alternate affected the hysteresis behavior but the effect is
minimal. This paper primarily focuses on the hysteresis
response from cyclic ascend and cyclic alternate loading
protocols.

2.1 Ramberg-Osgood model

The most commonly used expression to describe nonlinear


material stress-strain behavior is that proposed by Ramberg
and Osgood [1] as shown in Eq. (1).
   n
ε¼ þ , (1)
E K
where, σ and ε are engineering stress and strain,
respectively, E is the material Young’s modulus which
was taken as 2  105 N/mm2 in this study, n is a strain
hardening exponent and K is a modular parameter.
Grouping terms, Eq. (1) can be simplified to
  n
^y ¼ , (2)
K

where ^y ¼ ε – .
E
The corresponding data pairs are σi and εi - σi/E where
i = 1, 2,…, k. Given that
i
yi ¼ εi – , (3)
E
  n
i
^yi ¼ , (4)
K
Fig. 1 Cyclic loading protocols. (a) Cyclic ascend; (b) cyclic
and adopting the least-square method, the target is to alternate; (c) cyclic tensile
180 Front. Struct. Civ. Eng. 2014, 8(2): 178–186

minimize the following function to obtain an optimum pair For a given set of σi, εi (i = 1,2,…, k), the corresponding K
of modular parameter K and strain-hardening exponent n. and n parameters can then be determined.

X
k
2.2 Cyclic stress-strain curves
f ðK,nÞ ¼ ðyi – ^yi Þ2 , (5)
i¼1 From Chen et al. [12], typical hysteresis loops are shown in
from Fig. 2. The hysteresis loops for Group 1 specimens under
8 cyclic ascend (1-1) and cyclic alternate (1-2) loading
>
> ∂f ðK,nÞ protocol are shown in Figs. 2(a) and 2(b) respectively.
< ¼ 0,
∂K In general, the hysteresis loops are bi-symmetric,
(6a)
>
> ∂f ðK,nÞ therefore, to simplify the calibration process against
: ¼ 0: Ramberg-Osgood equations, the hysteresis loops in the
∂n
first quadrant (I) are calibrated and the calibration results
It can be deduced are shown in Tables 2 to 8.
8 Tables 2 to 8 show the calibrated parameters for Group 1
>
> n X k
to 7 specimens under cyclic ascend and cyclic alternate
>
> ni ðyi – ^yi Þ ¼ 0,
>
> K nþ1 loading protocols at different amplitudes. The coefficient
< i¼1
(6b) of determination r2 between the experimental data and
>
> 1 Xk
proposed hysteresis models are also tabulated.
>
> ni ðln i – ln KÞðyi – ^yi Þ ¼ 0:
>
> n As shown in the above tables, all coefficient of
: K i¼1
determination r2 are close to 1 showing close relation
between experimental data and Ramberg-Osgood equa-
As K and n are both nonzeros, Eq. (6b) can be further tions. Under different strain amplitudes, for example in
simplified to Group 7 specimens (Table 8), the modular parameter K
8 k remains fairly constant while the strain-hardening expo-
> X
>
> ni ðyi – ^yi Þ ¼ 0, nent, n increases with the strain amplitude.
>
>
>
< i¼1 2.3 Proposed K-fy relationship
X (7)
>
>
k
>
> i ln i ðyi – ^yi Þ ¼ 0:
n A summary of modular parameters for hysteresis loops
>
>
: i¼1 under cyclic ascend loading protocol is shown in Table 9.
As shown in the table, under different imposed strain
Table 2 Group 1 specimens Ramberg-Osgood parameters
parameters specimen designation strain amplitude
0.005 0.010 0.015 0.020
K/MPa 1-1 1076 1016 1025 1054
1-2 1025 1053 1001 1048
n 1-1 8.43 11.66 13.77 15.50
1-2 9.66 12.68 15.72 17.59
2
r 1-1 0.998 0.997 0.995 0.993
1-2 0.999 0.996 0.994 0.995

Table 3 Group 2 specimens Ramberg-Osgood parameters


parameters specimen designation strain amplitude
0.005 0.010 0.015 0.020
K/MPa 2-1 1003 1032 1054 1071
2-2 1118 1099 1058 1096
n 2-1 9.06 10.94 12.52 13.30
2-2 7.95 9.69 11.48 12.48
2
r 2-1 0.997 0.998 0.997 0.997
2-2 0.995 0.998 0.997 0.997
Yiyi CHEN et al. Cyclic stress-strain behavior of structural steel with yield-strength up to 460 N/mm2 181

Table 4 Group 3 specimens Ramberg-Osgood parameters


parameters specimen designation strain amplitude
0.005 0.01 0.015 0.02
K/MPa 3-1 1146 1152 1159 1177
3-2 1179 1211 1165 1177
n 3-1 9.75 11.57 13.54 14.81
3-2 9.13 11.17 13.24 14.46
2
r 3-1 0.998 0.998 0.997 0.997
3-2 0.993 0.998 0.997 0.997

Table 5 Group 4 specimens Ramberg-Osgood parameters


parameters specimen designation strain amplitude
0.005 0.010 0.015 0.020
K/MPa 4-1 1260 1263 1256 1263
4-2 1270 1269 1268 1282
n 4-1 8.49 10.14 11.84 12.95
4-2 8.08 10.46 11.18 12.10
2
r 4-1 0.998 0.998 0.997 0.997
4-2 0.998 0.998 0.997 0.997

Table 6 Group 5 specimens Ramberg-Osgood parameters


parameters specimen designation strain amplitude
0.005 0.010 0.015 0.020
K/MPa 5-1 1138 1143 1137 1142
5-2 1141 1177 1145 1159
n 5-1 10.98 11.19 13.23 14.42
5-2 11.20 11.31 13.84 15.11
2
r 5-1 0.995 0.998 0.996 0.996
5-2 0.995 0.997 0.997 0.997

Table 7 Group 6 specimens Ramberg-Osgood parameters


parameters specimen designation strain amplitude
0.005 0.010 0.015 0.020
K/MPa 6-1 1320 1318 1309 1317
6-2 1315 1380 1307 1315
n 6-1 8.95 9.40 10.69 11.42
6-2 8.83 8.89 10.48 11.12
2
r 6-1 0.996 0.998 0.997 0.997
6-2 0.996 0.998 0.997 0.997

amplitude, the variation of modular parameter K is trend. It is therefore proposed a linear relationship between
minimal, for example for specimen 2-1, the maximum modular parameter K and yield strength as follows:
modular parameter is 1071 MPa while the minimum K ¼ 1:79fy þ 532: (8)
modular parameter is 1003 MPa, so an average value
was used. It is also observed that as the yield stress The predicted K values show good correlation with the
increases, the modular parameter K displays an increasing experimental results as summarized in Table 10.
182 Front. Struct. Civ. Eng. 2014, 8(2): 178–186

Table 8 Group 7 specimens Ramberg-Osgood parameters


parameters specimen designation strain amplitude
0.005 0.01 0.015 0.02
K/MPa 7-1 1250 1254 1246 1255
7-2 1220 1242 1195 1209
n 7-1 10.82 10.87 12.24 13.47
7-2 10.80 11.28 13.75 14.96
2
r 7-1 0.995 0.998 0.996 0.996
7-2 0.996 0.998 0.997 0.996

Fig. 2 Hysteresis loops for Group 1 specimens. (a) Cyclic Ascend (1-1); (b) cyclic alternate (1-2)

Table 9 Modular parameters for hysteresis loops


modular parameter specimen designation
K/MPa
1-1 2-1 3-1 4-1 5-1 6-1 7-1
maximum 1076 1071 1177 1263 1143 1320 1255
minimum 1016 1003 1146 1256 1137 1309 1246
mean Kmean 1043 1040 1159 1261 1140 1316 1251
standard deviation 27.46 29.38 13.43 3.32 2.94 4.83 4.11
COV 0.026 0.028 0.016 0.003 0.003 0.004 0.003

Table 10 Modular parameter and yield strength relation


yield stress/MPa Kmean/MPa Kpredicted/MPa Kmean/Kpredicted
280 1043 1033 1.01
308 1040 1083 0.96
350 1159 1159 1.00
358 1261 1173 1.08
392 1140 1234 0.92
403 1316 1253 1.05
412 1251 1269 0.99
Yiyi CHEN et al. Cyclic stress-strain behavior of structural steel with yield-strength up to 460 N/mm2 183

3 Constitutive model
Based on the discussion from Section 2, the proposed
constitutive model of cyclic stress-strain behavior is
summarized as follows:

3.1 Primary skeleton stress-strain curve


8  n
>
>  
< þ , ³0,
E  K
ε¼  – n (9)
>
>
: – , <0,
E K
where K and n can be evaluated based on the
methodology as shown in Section 2.1 for cyclic ascend
loading protocol.

3.2 Cyclic stress-strain curve


8   Fig. 3 Hysteresis loops for Group 1 specimen under cyclic
>
>  – u  – u n
þ εu ,  loading,  i:e:,   – u > 0,
ascend loading protocol
< þ
ε ¼  –E   –K n
>
>
: u
– u
þ εu ,  unloading,  i:e:,   – u <0,
E K
(10)
where σu, εu are the unloading reference point at each
cycle, E is taken as 2  105 MPa, K can be obtained from
the relationship proposed in Eq. (8) and n can be defined by
the approach in Halford and Morrow [13]. Hence, to define
full hysteresis stress-strain behavior of structural carbon
steel, two key parameters which are K and n are needed.
The parameters K and n also depend on the strength of
steel, and the latter changes significant with the increase of
maximum strain.
Based on Eqs. (9) and (10), the derived hysteresis stress-
strain curves were compared with the experimental results.
For Group 1 specimens, as shown before, the correspond-
ing parameters are K = 1067 MPa (primary skeleton), n =
4.34 (primary skeleton), E = 2  105 MPa and K = 1033
MPa (hysteresis behavior as shown in Table 10). For
specimens under cyclic ascend loading protocol, the Fig. 4 Hysteresis loops for Group 1 specimen under cyclic
straining history is 0 ! 0.005 ! -0.005 ! 0.01 ! alternate loading protocol
-0.01 ! 0.015 ! -0.015 ! 0.02 ! -0.02 ! 0.02 and for
specimens under cyclic alternate loading protocol, the indicated at strain amplitude of 0.01, the cyclic peak values
straining history is 0 ! 0.005 ! -0.005 ! 0.015 ! for specimen 1-2 is greater than 1-1, which illustrates the
-0.015 ! 0.015 ! -0.01 ! 0.01 ! -0.01 ! 0.02 ! effect of loading protocol but as discussed in Chen et al.
-0.02 ! 0.02. The comparison between the proposed [12], the effect is insignificant toward seismic applications.
constitutive model and the experimental results are In general, the calibrated Ramberg-Osgood equation
shown in Figs. 3 and 4. shows excellent agreement with the experimental results
The above figures revealed the proposed hysteresis on the primary skeleton and the cyclic stress-strain curves.
models correlate closely with the experimental data. The
proposed material stress-strain model can also capture the
effect of different loading protocols. Figure 5 shows the 4 Application to Q460 specimens
hysteresis loops for Group 1 specimens under cyclic
ascend (1-1) and cyclic alternate (1-2). The difference is The proposed constitutive model in Section 3 was
184 Front. Struct. Civ. Eng. 2014, 8(2): 178–186

Fig. 5 The effect of loading protocols on hysteresis loops


Fig. 6 Hysteresis loops for Group 8 specimen under cyclic
ascend loading protocol
calibrated by experimental coupons with nominal yield
strengths between 235 to 420 MPa. To investigate their
applicability to a higher grade of steel, this section presents
the comparison between the proposed constitutive model
and the experimental result on coupons with nominal yield
strength of 460 MPa (Q460 specimens). The direct tensile
tests from the additional Group 8 specimens were
conducted by the Authors on Grade Q460E materials
with the measured yield strength of 491 MPa, ultimate
strength of 611 MPa, yield-ultimate strength ratio of 0.8
and the elongation of 22.7%. Other Group 9 specimens on
Grade Q460C materials were tested by Shi et al. [14] with
the reported yield strength of 466 MPa.
Based on the proposed constitutive models (Eqs. (8)‒
(10)), Figs. 6 and 7 illustrate the comparison between the
proposed models and the experimental observations which
reveal close agreement between the two.

5 Numerical validation
Fig. 7 Hysteresis loops for Group 9 specimen under cyclic
This section presents a numerical study based upon the ascend loading protocol
proposed constitutive model in Section 3.
shown in Fig. 1(a). As shown in Fig. 9, Δ1 (0.7 mm) is the
5.1 Validation experiment yield displacement. Further experimental details are
presented in Zhao et al. [15].
The validation experiment refers to a cyclic test of tubular
column of steel grade Q235B with nominal yield strength 5.2 Numerical simulation
of 235 N/mm2. The geometry of the test arrangement is
shown in Fig. 8. The bottom part of the specimen was A numerical modeling investigation, employing the finite
anchored to the strong floor facilities at the Structures element (FE) package ABAQUS [16], was carried out to
laboratory of Tongji University. The loading was applied replicate the experimental results based on the proposed
through the top part of the specimen under displacement- constitutive model in Section 3. The element chosen for the
controlled loading scheme. The loading scheme (Displace- FE model was a general purpose linear brick element, with
ment ratio Δ/Δ1 vs. number of cycle n) is shown in Fig. 9 reduced integration, designated as C3D8R in the ABA-
which is similar to the cyclic ascend loading protocol as QUS element library. The mesh density was carefully
Yiyi CHEN et al. Cyclic stress-strain behavior of structural steel with yield-strength up to 460 N/mm2 185

Fig. 8 Specimen dimension (unit: mm)


Fig. 10 Finite element model

Fig. 9 Loading protocol

chosen by carrying out a mesh convergence study with the


aim of achieving accurate results while minimizing
computational effort. The meshed finite element model is
shown in Fig. 10.
The true material stress-strain relationships were
generated from the engineering stress-strain curves
obtained from the tensile coupon tests and cyclic coupons Fig. 11 Hysteresis loops
[12]. A kinematic/isotropic combined model [16] was
employed in which the kinematic hardening parameters of test results was found to be satisfactory with the
employed were determined through the full stress-strain numerical models able to successfully capture the initial
response of specimen 1-1 from Eqs. (8) and (9) under four primary stress-strain behavior and general hysteresis
different strain amplitudes. Boundary conditions were response.
applied to the model fixed end support conditions at both
ends but allowing free vertical displacement at the top. The
vertical displacement at top end was monitored throughout 6 Conclusions
the analysis. Geometric and material nonlinearity was
incorporated in the models which also enabled the cyclic This paper proposes a constitutive model based on
behavior to be traced. The numerical derived hysteresis Ramberg-Osgood equations to describe the primary
curves are shown in Fig. 11 and are compared with the skeleton stress-strain behavior and the cyclic stress-strain
corresponding experimental hysteresis results. Replication response for structural carbon steel with yield strength up
186 Front. Struct. Civ. Eng. 2014, 8(2): 178–186

to 420 N/mm2. The proposed model was calibrated against Constructional Steel Research, 2004, 60(9): 1291–1318
a series of cyclic material tests with nominal yield strength 5. Gardner L, Ashraf M. Structural design for nonlinear metallic
between 235 to 420 N/mm2 which revealed close agree- materials. Engineering Structures, 2006, 28(6): 926–934
ment between the proposed model and the experimental 6. Quach W M, Teng J G, Chung K F. Three-stage full-range stress-
results. A linear relation between modular parameter K and strain model for stainless steels. Journal of Structural Engineering,
yield strength fy was also proposed. Two key parameters K 2008, 134(9): 1518–1527
and n were defined which were used to describe the full 7. Chen J, Young B. Effects of elevated temperatures on the
stress-strain response under cyclic load. The proposed mechanical properties of cold-formed steel. Proceedings in the
model was also able to capable the material hysteresis International symposium on cold formed steel metal structures,
performance of coupons with nominal yield strength of 2004, 139–158
460 N/mm2. The proposed constitutive model was then 8. Chen J, Young B. Stress-strain curves for stainless steel at elevated
employed in a numerical validation which has showed temperatures. Engineering Structures, 2006, 28(2): 229–239
excellent agreement between the numerical and the 9. Skelton R P, Maier H J, Christ H J. The Bauschinger effect, Masing
experimental results. This research forms part of an on- model and the Ramberg-Osgood relation for cyclic deformation in
going research program at Tongji University to investigate metals. Materials Science and Engineering, 1997, 238(2): 377–
the material hysteresis response of structural carbon steel 390
with strain amplitude up to 10%. 10. Broggiato G B, Campana F, Cortese L. The Chaboche nonlinear
kinematic hardening model: Calibration methodology and valida-
Acknowledgements The authors wish to kindly acknowledge Kwanghua tion. Meccanica, 2008, 43(2): 115–124
Foundation from Tongji University, the Warwick China Partnership Fund 11. Shi G, Wang M, Bai Y, Wang F, Shi Y J, Wang Y Q. Experimental
from the University of Warwick, Engineering Physical Sciences Research
and modeling study of high-strength structural steel under cyclic
Council (EP/I020489/1) for their support toward the third author as a visiting
scholar at Tongji University. The support from the National Natural Science loading. Engineering Structures, 2012, 37: 1–13
Foundation of China (Grant No. 51038008) is also greatly appreciated. 12. Chen Y Y, Sun W, Chan T M. Effect of loading protocols on the
hysteresis behaviour of structural carbon steel with yield strength up
to 460 N/mm2. Advances in Structural Engineering, 2013, 16(4):
References 707–719
13. Halford G R, Morrow J. Low-cycle fatigue in torsion. Proceedings
1. Ramberg W, Osgood W R. Description of stress-strain curves by of the American Society for Testing and Materials, 1962, 62: 695–
three parameters. National Advisory Committee for Aeronautics, 709
Technical Note No. 902, 1943 14. Shi G, Ban H Y, Shi Y J, Wang Y Q. Engineering application and
2. Mirambell E, Real E. On the calculation of deflections in structural recent research progress on high strength steel structures. Industrial
stainless steel beams: An experimental and numerical investigation. Construction, 2012, 42(1): 1–7 (in Chinese)
Journal of Constructional Steel Research, 2000, 54(1): 109–133 15. Zhao X Z, Ge L J, Chen Y Y, He M X. Research on the effect of
3. Rasmussen K J R. Full-range stress-strain curves for stainless steel assembling weld form on mechanical behavior of box brace. Journal
alloys. Journal of Constructional Steel Research, 2003, 59(1): 47–61 of Building Structures, 2008, 29(3): 88–95 (in Chinese)
4. Gardner L, Nethercot D A. Experiments on stainless steel hollow 16. ABAQUS. Analysis User’s Manual I–V. Version 6.7. USA:
sections. Part 1: Material and cross-sectional behaviour. Journal of ABAQUS, Inc, 2007

You might also like