Available online at www.sciencedirect.

com

Procedia Engineering 10 (2011) 270275

ICM11

Fatigue of Constructional Steel S460 under Complex Cyclic Stress

and Strain Sequences
Michael Vormwald a *
a

Technische Universitt Darmstadt, D-64287 Darmstadt, Germany

Elsevier use only: Received date here; revised date here; accepted date here

Abstract

The constructional fine-grained steel S460 shows considerably high fatigue performance. In the last two decades this material has
been investigated experimentally in great detail. These results served as basis for developing models and algorithms aiming at
improvement of fatigue life assessments. The paper summarizes both the cyclic deformation and failure behaviour. The experimental results are compared with the results obtained by applying the models. Simple uniaxial loading is addressed also as a
complex multiaxial, non-proportional elastic-plastic straining. The deformation behaviour is described by an enhanced plasticity
model based on continuum mechanics principles. Cyclic softening and hardening, ratcheting and cyclic mean stress relaxation as
well as non-proportional hardening are taken into account. The fatigue cracking behaviour is modelled based on a critical plane
approach where the short crack growth calculations determine the critical plane of shortest fatigue life.
Keywords: cyclic plasticity; fatigue; constructional steel

1. Introduction
This article describes the deformation and failure behaviour of construction steel S460 in non-welded condition
in the case of fatigue load. Table 1 contains the typical chemical composition of the steel. The majority of the experimental results reported below stem from the batch of Table 1. The texture is ferritic-pearlitic with average grain
size of about 10 mm.
Table 1: Chemical composition (in weight percentage)

S460N

C
0.18

Si
0.44

Mn
1.54

P
S
Al
Cr
N
Ni
0.016 0.001 0.013 0.022 0.019 0.27

* Corresponding author. Tel.: +49-6151-163645; fax: +49-6151-163038

E-mail address: vormwald@wm.tu-darmstadt.de

1877-7058 2011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of ICM11 doi:10.1016/j.proeng.2011.04.047

Cu
V
0.009 0.17

271

Michael Vormwald / Procedia Engineering 10 (2011) 270275

2. Deformation behaviour
2.1. Experimental results
The fatigue load starts with the first reversal. Figure 1 shows the stress-strain relationship determined with tensile
tests at different maximum strains. Then a load reversal was performed during which the stress-strain curve was
recorded further.
800
600

V [MPa]

400
200
0
-200
-400
-600
-800
0.00

simulation
experiment
0.02

0.04

0.06

0.08

0.10

H [-]
Fig. 1: Tensile test curve with unloading curves at different maximum strains from [1]

It has been established that after the pronounced yield point is exceeded once it does not appear again in the subsequent reverse plastification. Plastic reverse deformations take place at compressive stress values which are significantly smaller than the yield point. This phenomenon is known as Bauschinger effect. The Bauschinger effect is the
first sign of the transient processes of cyclic softening or cyclic hardening taking place along with the further fatigue
cycling. When a number of fatigue tests are performed on samples of materials with different strain amplitudes, almost stabilized stress-strain hysteresis loops always set in. A function is designated as a stabilized cyclic stressstrain curve relating these stabilized stress and strain amplitudes.
1

Ha

Va

V n'
 a .
E K'

(1)

The cyclic strength coefficient and exponent are K ' 1145MPa and n ' 0.164 .
In load-controlled experiments with a mean load, an additional deformation effect takes place, namely the cyclic
creep or ratcheting. Figure 2 shows several ratcheting curves of the construction steel S460. The ratcheting occurs in
particular also in the case of a multiaxial non-proportional load. In the case of multiaxial non-proportional load, another phenomenon - the non-proportional hardening takes place. In the case of uniaxial and proportional load, the
plastic deformation takes place through slipping in relatively small number of preferentially oriented slip planes, i.e.
in planes with greater shear stress. In the case of non-proportional load, the planes with greater shear stress are not
anymore stationary during a given stress cycle. In this way, more slip systems are activated. Since now slip systems
can block each other, it comes macroscopically to a hardening phenomenon. The plasticity model according to [2],
with which all phenomena presented in this section can be considered with sufficient precision, is presented in a
general outline in the next section.

272

Michael Vormwald / Procedia Engineering 10 (2011) 270275

'F
A = 750MPa

mean strain Hm [-]

20
R= -0.5

0
'F
A = 900MPa

R= -2

-20

'F
A = 1200MPa

-40
0

2000

4000 n 6000

8000

10000

Fig. 2: Uniaxial ratcheting (cyclic creep) for S460 from [1]

2.2. Modelling
A description of the cyclic plasticity can be obtained from the constitutive equations of the plasticity theory [2].
Since in practical cases the strains resulting from a fatigue load can hardly become greater than 1%, it is sufficient
for modelling purposes to proceed within the framework of linear deformation kinematics and additive decomposition of the strain tensor,
(2)

= el + pl

The elastic strains are linked with the strains through the stress-strain relation,
=  el

(3)

whereby  is the elasticity tensor for the isotropic behaviour of materials. The use of the von Mises yield criterion is
widely accepted for metal materials and thus also for the S460.
'( s, , r (0) ) = s   r (0) = 0

(4)

Here s is the deviatoric part of the stress tensor . The centre of the yield surface is described by the backstress
tensor whose evolution equation describes the kinematic hardening the most important item in cyclic plasticity.
The backstress tensor is divided into several partial tensors according to Chaboches [3] suggestion;
M

(5)

(i )

i 1

The following condition is specified for the evolution equation,

d

(i )

(i )
c r (i ) n  (i )

(i )

F (i )

(i )
(i ) dp  (i ) dr ( i )

(6)

273

Michael Vormwald / Procedia Engineering 10 (2011) 270275

3. Failure behaviour
3.1. Experimental results
The failure behaviour of materials and structures under fatigue is normally given in the form of Whler lines. For
S460 the strain-life curve may be written as
V 'f
b
c
Ha
(7)
 2 N   H 'f  2 N 
E
with the parameters V 'f 1218MPa , H 'f 0.452 , b 0.104 , and c 0.536 . The failure is defined as the occurrence of a surface crack with a length of 0.5mm. Figure 3 shows the mean load effect on the amplitudes based on a
Haigh-type diagram. As it can be expected, the mean tensile loads V m reduce the tolerable strain amplitudes.
0.6

S460
PSWT

Ha [%]
0.4

RV

PJ

2
RV

N 10

1

RV

0.5

0.2

0.0
-200

N 105

-100

100

200

RV

300

400

Vm [MPa]
Fig 3: Mean stress diagram from [4]

In this phase of the initiation of technical cracks, the material shows just like all ductile metal materials a severe sequence effect in the loading with variable amplitudes. In order to make this clearer, the results of two constant amplitude tests with a 0.5% and 0.2% strain amplitude are juxtaposed with the result of a two-level test with
one cycle of alternating strain load with a 0.5% and 26 cycles with a 0.2% strain amplitude. All strains are set up in
such a way so that no mean stress is acting. The fatigue life of the constant amplitude tests amounting to 6500 and
190000 cycles follow from the Whler line, eq. (24). The Miner rule [5] is usually used for calculation of the lives in
the case of a variable amplitude load, which implies independence of the damaging action of a cycle from the loading history. After such a calculation, a life of 93000 cycles is obtained for the two-level test which, however, requires in the course of the experiment only 26000 cycles until an initiation occurs.
The representation of the crack growth behaviour belongs also to the complete characterization of the failure behaviour of a given material under the effect of the fatigue loading. If the crack opening loads are measured, one can
determine from them the so-called effective ranges i.e. the ranges, during which the crack is open and relate the
crack growth rates to the corresponding effective ranges of the stress intensity factor, and thus the coherent picture
in Figure 4 is obtained. The material behaves with respect to the relationship to the crack growth rates and the effective ranges exactly in the same way as all other known steels: The effective stress intensity factors
5
4
'K eff 9 MPa m and 20 MPa m correspond to the crack growth rates of 10 and 10 mm/cycle. The threshold value with respect to the fatigue crack growth lies at 3 MPa m , also expressed in effective ranges. When the
crack initiation phase is analyzed more precisely, it is established that it is determined in more than 90% by a growth
phase of short fatigue cracks, see [4].

274

Michael Vormwald / Procedia Engineering 10 (2011) 270275

3.2. Modelling
The crack growth rates of these short cracks can be represented in a uniform manner along with the rates of the
longer cracks when the cyclic J-integral is selected as a crack tip parameter instead of the cyclic stress intensity factor,
da
dn

C  'J eff 

C  PJ a 

(8)

with
2

'V eff
1.02


'V eff 'H pl,eff a
1.24

E
nc

'J eff

(9)

PJ a

for short surface cracks. The damage parameter PJ describes the mean stress effect through the mean stress dependant effective range, maximum value minus opening value. The latter is calculated by using the approximation formula
V op
V
S V
0.535 cos max 
0.344 R max ; V F 0.5  Rcp 0.2  R m . ; R d 0
.
(10)
VF
V max
2 VF
10

-3

da/dn [mm/cycle]

-1 b R b 0.7

10

-4

10

-5

10

20

30

40

50

60

70

80 90

'K = Kmax - Kop = 'Keff [MPa m]

Fig. 4: Fatigue crack growth relationship in effective ranges for S460

An algorithm was proposed in [4] which permits the simulation of the sequence-dependent position of the crackopening strain. In this way, it is possible to achieve to a large extent the consideration of the above-described sequence effects. The important characteristics of this algorithm include the fact that the compressive strains lead to
immediate reduction of the crack-opening strain, while the cycles in the tensile mean stress regime cause only a
gradual increase of the crack-opening strain. The formulation of the mean stress effect takes place mostly through
the so-called damage parameter. The most widely used damage parameter goes back to Smith et al. [6]. With the
introduction of such a parameter it is determined which combination of mechanical parameters of a hysteresis loop
is responsible according to the hypothesis for the fatigue damage. In the case of the above damage parameter,
this is the product of the maximum stress and the strain amplitude. The mean stress effect represented with this
damage parameter is shown in Figure 3. One can see a good representation of the effect except for the effect of the
mean compressive loads. Results from the application of the damage parameter PJ are also shown.

275

Michael Vormwald / Procedia Engineering 10 (2011) 270275

PSWT

V maxH a E

(11)

One can see in Figure 5, on the one hand, the differences in the calculated crack initiation lives and, on the other
hand the comparison with the test results. In the calculations with the Smith parameter, the endurance limit of the
material was ignored which leads to a steep progression of the calculated life-line for lower amplitudes. When the
short crack PJ-parameter is used, a continuous decrease of the endurance limit according to [7] was taken into account, whereby the life-line approaches the constant amplitude endurance limit. At the high amplitudes the sequence
effect is clearly visible there as it is manifested in the sequence-dependent crack opening strain.

nominal stress Sa, S a [N/mm2]

500

Sa
Sa

400

Gaussian sequence
R = -1

0
10 4

300

PJ

P SWT

notch strain simulation

for K t = 2.5

V
200

H S460

const. ampl.

R = -1
140 4
10

10

10

10

N, N

10

Fig. 5: Comparison of experimentally determined and calculated fatigue lives in the case of loads with variable amplitudes

4. Summary
Construction steel S460 in a non-welded condition has a good to very good (depending on its static strength) resistance capacity against fatigue. In the last two decades, it was subject to intensive experimental studies and served
thereby as a model material in the development of simulation methods. The present paper outlines in details both the
cyclic deformation behaviour and also the failure behaviour in the case of fatigue. The experimental results are juxtaposed consistently with the calculation result by applying the current model. Thereby the range of issues extends
from the simple uniaxial load to the complex multiaxial non-proportional load.
5. References
[1] E. Herz, R. Thumser, J. W. Bergmann, M. Vormwald: Endurance limit of autofrettaged Diesel-engine injection tubes with defects, Engineering Fracture Mechanics 73, (2006) p. 3-21
[2] R. Dring, J. Hoffmeyer, T. Seeger: A plasticity model for calculating stress-strain sequences under multiaxial nonproportional cyclic loading, Computational Materials Science 28 (6), (2003) p. 587-596
[3] J. L. Chaboche: Modelization of the strain memory effect on the cyclic of 316 stainless steel, 5th Int. Conf. Struct. Mech. in Reactor Techn.,
Div. L, L11/3 (1979)
[4] M. Vormwald: Anrilebensdauervorhersage auf der Basis der Schwingbruchmechanik fr kurze Risse, Report 47 Institut fr Stahlbau und
Werkstoffmechanik, TH Darmstadt, (1989)
[5] M. A. Miner: Cumulative Damage in Fatigue, Transactions of ASME, Journal of Applied Mechanics 12(3), (1945) p. A159-A164
[6] K. N. Smith, P. Watson, T. H. Topper: A stress-strain function for fatigue of metals, Journal of Materials 5, (1970) p. 767778
[7] M. Vormwald, T. Seeger: Consideration of fatigue damage below the endurance limit in life predictions for variable amplitude loading, Proc.
4th Int. Conf. FATIGUE '90, Materials and Component Engng. Publ. Ltd., (1990), p. 517-522

Acknowledgement
For decades, the results presented in this paper have been achieved in research work supported by many grants of the Deutsche Forschungsgemeinschaft. The author expresses his severe gratitude for continuously backing this research work.