Journal of Constructional Steel Research 79 (2012) 140150

Contents lists available at SciVerse ScienceDirect

Journal of Constructional Steel Research

A comparison of the fatigue behavior between S355 and S690 steel grades
Ablio M.P. de Jesus a,, Rui Matos b, Bruno F.C. Fontoura a, Carlos Rebelo b,
Luis Simes da Silva b, Milan Veljkovic c
a
b
c

IDMEC/Engineering Department, School of Sciences and Technology, University of Trs-os-Montes and Alto Douro, Vila Real, Portugal
ISISE/Civil Engineering Department, University of Coimbra, Coimbra, Portugal
Lule University of Technology, Lule, Sweden

a r t i c l e

i n f o

Article history:
Received 2 November 2011
Accepted 24 July 2012
Available online 30 August 2012
Keywords:
Fatigue
Crack propagation
S690 steel grade
S355 steel grade
High strength steel
Mild steel

a b s t r a c t
The use of higher strength steels allows the design of lighter, slenderer and simpler structures. Nevertheless,
the increase of the yield strength of the steels does not correspond to a proportional increase of fatigue resistance, which makes the application of high strength steels on structures prone to fatigue, a major concern of
the design. This paper presents a comparison of the fatigue behavior between the S355 mild steel and the
S690 high strength steel grades, supported by an experimental program of fatigue tests of smooth specimens,
performed under strain control, and fatigue crack propagation tests. Besides the cyclic elastoplastic characterization, the fatigue tests of smooth small size specimens allow the assessment of the fatigue crack initiation
behavior of the materials. Results show that the S690 steel grade presents a higher resistance to fatigue crack
initiation than the S355 steel. However, the resistance to fatigue crack propagation is lower for the S690 steel
grade, which justies an inverse dependence between static strength and fatigue life, for applications where
fatigue crack propagation is the governing phenomenon. Consequently, the design of structural details with
the S690 steel should avoid sharp notches that signicantly reduce the fatigue crack initiation process.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
The use of high strength steels allows the design of lighter, slenderer
and simpler structures with high structural performance. The economic
factors are decisive concerning the choice of the steel for a structural
application. In general, the use of high strength steels contributes to
weight reduction which compensates the higher cost of the high
strength steels [1]. High strength steels are gaining competitiveness
with respect to the mild structural steels.
The application of high strength steels on steel bridges is becoming attractive. According to Miki et al. [2] and Jensen and Bloomstine
[3], the number of new bridges made of high strength steels is increasing signicantly in the last decades. New applications of high
strength steels are also being considered, such as windmill tower production [4]. The use of high strength steels allows the construction of
taller windmill towers with simple and cost effective joining systems
for tower assembling, contributing to the increase of the competitiveness of the wind energy generation.
Despite the important advantages of the increased yield strength
provided by the high strength steel grades, the use of these steels

Corresponding author at: School of Sciences and Technology, University of

Trs-os-Montes and Alto Douro, Quinta de Prados, 5001-801 Vila Real, Portugal. Tel.:
+351 259 350 306; fax: +351 259 350 356.
E-mail address: ajesus@utad.pt (A.M.P. de Jesus).
0143-974X/$ see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcsr.2012.07.021

faces important challenges. The weldability of the high strength steels

is lower than the weldability of mild steels, and decreases with strength
increasing [5]. The carbon and alloy element contents are, therefore,
limited to ensure weldability. Ductility, toughness and corrosion resistance are also desired characteristics for the high strength steels. A particular group of high strength steels is the high performance steels
(HPS) that combine high strength with enhanced ductility, toughness,
weldability and improved weathering ability [6].
Fatigue resistance of the high strength and HPS is a major concern,
since it is well known that fatigue resistance does not increase proportionally to the static strength of these steels. This is very often
the case for welded components [7]. The fatigue resistance of welded
joints made of high strength steels may be even lower than for
welded joints made of mild steels [2]. Nevertheless, SN curves proposed in design codes (e.g. Eurocode 3 [8]) do not show dependency
on material, which implies signicant safety margins. In general, high
strength steels and HPS are still less investigated than construction
mild steels, leading to a decient understanding of the fatigue behavior of the high strength steels and HPS. However, this topic has been
gaining much interest in the last decade [9,10]. Many fatigue studies
are focused on testing structural details rather than investigating
the plain material, which limits the extent of the ndings to the specic geometries under investigation. The investigation on plain material allows the assessment of the basic fatigue properties of the
materials, which are required to model the fatigue behavior of structural components. Particularly, the assessment of the fatigue crack

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

141

Fig. 1. Typical yielding and strain hardening behaviors of the S355 and S690 steel grades.

initiation and propagation behaviors is essential to model the fatigue

behavior of structural components [9,10].
This paper provides experimental assessment of the fatigue properties of two competing steel grades, namely the S355 mild steel and
the S690 high strength steel, both specied in the EN 10025 standard
[11]. Both fatigue crack initiation and fatigue crack propagation behaviors are investigated. The fatigue crack initiation behavior is evaluated through fatigue tests of smooth and small size specimens. The
fatigue crack propagation behavior is characterized by means of fatigue tests of compact tension (CT) specimens, covering several stress
ratios.

2. Overview of current approaches to fatigue

The fatigue approaches may be classed into SN, local and fracture
mechanics based approaches. The SN approach is the basis of current
design codes such as the Eurocode 3, part 1-9 [8]. This is a global approach that relates the stress range (e.g. nominal, structural or geometric) applied to the component with the fatigue life. With respect to
Eurocode 3, part 1-9, no distinction is made in procedures for welded
and non-welded components. The procedures do not account for the
material inuence. Also, the classication of complex details may be
problematic. Local approaches to fatigue and fracture mechanics can
be used as alternatives to the global SN approaches, which requires
the knowledge of the basic fatigue properties of the base materials.
The local approaches, recognizing the localized nature of the fatigue damage, propose the correlation of a local damage parameter
(e.g. strain, energy) with the number of cycles required to initiate a
macroscopic crack. The most well-known relations in this area are
the proposals by Basquin [12], Eq. (1), Cofn [13] and Manson [14],
Eq. (2) and Morrow [15], Eq. (3):


b

f 2Nf
2

c
P

f 2Nf
2

b


c
E P f

2N f f 2Nf
2
2
2
E

where f and b are, respectively, the fatigue strength coefcient and exponent; f and c are, respectively, the fatigue ductility coefcient and exponent; 2Nf is the number of reversals to failure; , E and P are,
respectively, the total, elastic and plastic strain ranges; is the stress
range and E is the Young's modulus. The constants in these relations
may be determined from fatigue tests of smooth specimens under
strain-controlled conditions. These tests also allow the identication of
the cyclic curve of the material which relates the stress amplitude with
the strain amplitude, corresponding to the stabilized behavior of the material. This relation is usually expressed using the RambergOsgood equation [16]:

Fig. 2. Microstructures of the investigated structural steels: a) S355 steel grade; b) S690
steel grade.



E P
1=n

2
2E
2
2
2K

142

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

Table 1
Comparison of the chemical composition between the S355 and S690 steels: EN 10025 standard recommendations [11] versus measured values (% weight).

EN 10025
standard
Measured
values

Steel
grade

C (%)

Si (%)

Mn (%)

Cr (%)

Cu (%)

Mo (%)

Ni (%)

V (%)

Nb (%)

Ti (%)

Al (%)

P (%)

S (%)

S355
S690
S355
S690

0.2 max
0.2 max
0.10
0.077

0.5 max
0.8 max
0.15
0.048

0.91.65 max
1.7 max
0.64
1.35

0.3 max
1.5 max
0.076
0.025

0.55 max
0.50 max
0.38

0.10 max
0.70 max
0.014

0.5 max
2.0 max
0.095
0.036

0.12 max
0.12 max
0.003

0.05 max
0.06 max

0.042

0.05 max
0.05 max

0.086

0.02 min
0.015 min

0.036

0.030 max
0.025 max
0.022
0.009

0.025 max
0.015 max
0.041
0.005

Fig. 3. Geometry of the smooth specimens used in the fatigue tests.

where K and n are, respectively, the strain hardening coefcient and exponent. The cyclic curve is required to perform an elastoplastic analysis of
components, either using a nite element formulation or using a simplied approach as proposed by Neuber [17]:
2nom K t 2

E

where nom is the nominal stress range, and are, respectively, the
local strain and stress ranges and Kt is the elastic stress concentration
factor.
The strainlife Eq. (3) is a general relation that does not account
for mean stress effects. In order to account for mean stress effects,
Smith, Watson and Topper [18] proposed the following alternative:
max

2

2b


bc

E f
2N f
f f E 2Nf
2

where max is the maximum stress of the cycle and the other nomenclatures are the same as Eq. (3).
Fracture mechanics may be also used as an alternative approach to
fatigue, based on the fatigue crack propagation phenomena. This approach may be used to complement the local approaches to fatigue
[9,10] allowing the residual life computation of a structural component with an initial crack. This approach is based on crack propagation laws, with Paris' law [19] being the most used:
m

da=dN C K

where C and m are material constants, da/dN is the fatigue crack

growth rate and K is the stress intensity factor range. The number
of cycles to failure is computed by integrating the crack propagation
law between an initial crack size (ai) and a critical crack size (af):
af

Nf
ai

da
:
C K m

Table 2
Nominal dimensions of the smooth specimens.
Material

W (mm)

T (mm)

L (mm)

L1 (mm)

H (mm)

R (mm)

S355
S690

30
16

7.5
4

26
13

200
110

12.5
8

8
4.5

3. Materials and experimental details

3.1. Basic materials' descriptions
A comparison of the fatigue properties between an S355 mild steel
and an S690 high strength steel is proposed in this research. These
steel grades are specied according to the EN 10025 standard [11]. Minimum yield stresses of 355 and 690 MPa are specied, respectively, for
the S355 and S690 steel grades, for thicknesses below 16 mm. The
S355 steel grade should exhibit a tensile strength within the range of
470 and 630 MPa and the S690 steel grade should present a tensile
strength between 770 and 940 MPa, also for thicknesses below 16 mm.
In order to verify the actual static strength properties of the two steel
grades used in the experimental program, quasi-static monotonic tensile
tests were performed, covering both steel grades. Average yield stresses
of 419 MPa and 765.7 MPa were measured, respectively for the S355 and
S690 steel grades. Average tensile strengths of 732 MPa and 823 MPa
were obtained, respectively, for the S355 and S690 steel grades. In general, these strength properties agree with the limits specied in the EN
10025 standard. However, the sample of the S355 steel grade used in
this research exhibited a tensile strength above the range specied in
the standard. Nevertheless, the trend of the monotonic stressstrain
curves, the chemical composition and the material microstructure are
typical of the S355 steel grade as will be veried hereafter. The tensile
tests were instrumented with strain gauges which allowed the assessment of the Young's modulus. Average values of 210.5 and 209.4 GPa
were measured, respectively, for the S355 and S690 steel grades. Fig. 1
shows typical monotonic stressstrain curves that were obtained for

Table 3
Summary of the fatigue tests of smooth specimens.
Specimens
(S355)

S
(mm2)

f
(Hz)

(%)

Specimens
(S690)

S (mm2)

f (Hz)

(%)

S355_100_01
S355_050_01
S355_200_01
S355_040_01
S355_030_01
S355_035_01
S355_030_02
S355_040_02
S355_100_02
S355_200_02

55.94
56.08
60.16
57.65
57.86
59.48
61.06
60.75
56.61
57.75

0.400
0.800
0.200
1.000
1.333
1.143
1.333
1.000
0.400
0.200

1.00
0.50
2.00
0.40
0.30
0.35
0.30
0.40
1.00
2.00

S690_200_01
S690_100_01
S690_050_01
S690_050_02
S690_100_02
S690_200_02
S690_040_01
S690_040_02
S690_800_01
S690_840_01
S690_150_01

18.35
18.23
17.80
18.63
18.51
17.92
17.98
18.25
17.93
18.14
17.93

0.200
0.400
1.000
1.000
0.400
0.200
5.000
5.000
15.000
15.000
1.000

2.00
1.00
0.50
0.50
1.00
2.00
0.40
0.40
0.36
0.38
1.50

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

143

Fig. 4. Geometry of the CT specimens.

the S355 and S690 steel grades. These curves are truncated since strains
were measured using glued strain gauges that were not able to monitor
the test until nal failure. Nevertheless, the curves allow the comparison
of the yield region of both steels as well as the initial strain hardening behavior. It is clear that the S355 steel shows a yield plateau, after which a
very signicant strain hardening is veried. The S690 steel does not show
that yield plateau, and a relatively small strain hardening is observed. The
yield stress was determined for the S355 steel as the maximum stress observed in the yield plateau. For the S690 steel, the yield stress was dened as the stress corresponding to 0.2% permanent strain. According
to EN 10025, the S690 steel shows a minimum elongation after a fracture
of 14%; the S355 steel shows an elongation after a fracture of 22%.
The microstructures of both steel grades were observed using an optical microscope. Fig. 2 compares the microstructures of both steel
grades. It is very clear that the microstructure of the S690 steel grade
is signicantly more rened than the microstructure of the S355 steel,
which has a signicant impact on mechanical properties, including fatigue properties as will be discussed later. The S690 steel is obtained
by thermomechanical rolling and supplied in quenched and tempered
condition. The S355 steel shows a microstructure of ferrite and perlite
which is typical of non-alloyed structural mild steels. Both steels are
weldable steels; however the weldability of the high strength steel is
in general poorer than the weldability of mid steel due to the higher
level of alloy elements. Table 1 presents a comparison of the typical
chemical composition for both steels, according to EN 10025 [11]. The
table also presents the actual chemical composition measured on steel
samples using the spark emission spectrometry. The chemical compositions are in general according to the standard recommendations. It is
clear that there is a signicantly higher amount of manganese on the
S690 steel, contributing to the higher strength and hardenability of
this steel with respect to the S355 steel. The S690 steel also includes
Table 4
Nominal dimensions of the CT specimens.
Material

W
(mm)

B
(mm)

L
(mm)

H
(mm)

h
(mm)

D
(mm)

he
(mm)

an
(mm)

()

S355
S690

50
40

8
5

62.5
50

60
48

27.5
22

12.5
10

3
1.6

10
8

60
60

grain-rening elements such as aluminum, niobium and titanium. The

phosphorus and sulfur contents are lower in the S690 steel grade,
since these elements have adverse effects on ductility and toughness,
especially on quenched and tempered steels.
The hardness of both steel grades were measured using an INDENTEC
hardness machine, resulting in the values of 27.43.9% and 33.84.2%
HRC, respectively for the S355 and S690 steel grades. The S690 steel
grade exhibits a Rockwell C hardness that is 23% higher than the S355
steel grade, which is consistent with the higher yield stress of the S690
steel grade.
3.2. Experimental details
This research aims at comparing the fatigue behavior between the
S355 mid steel and the S690 high strength steel, based on experimental
results from fatigue tests of smooth specimens and fatigue crack propagation tests. The fatigue tests of smooth specimens were carried out
according to the ASTM E606 standard [20], under strain controlled conditions. The crack propagation tests were performed using CT specimens, in accordance with the procedures of the ASTM E647 standard
[21], under load controlled conditions.
Fig. 3 shows the general geometry of the smooth specimens and
Table 2 indicates the dimensions adopted for each steel grade. Distinct
sizes of specimens were considered for each steel grade, since specimens
were extracted from plates with different thicknesses. Nevertheless both
geometries are in accordance with the ASTM E606 recommendation [20].
The gauge length of the specimens was polished with an appropriate sequence of sandpapers. One series of 10 specimens were prepared for
each material. The strain was controlled using a dynamic clip gauge,
model INSTRON 2620-202, with a range of 2.5 mm. Specimens of the
S355 steel were instrumented with a reference gauge length of 25 mm;
specimens of the S690 steel were instrumented with a reference gauge
length of 12.5 mm. The fatigue tests of the smooth specimens were
conducted for a strain ratio, R, equal to 1, following a sinusoidal
waveform with a frequency adjusted to result in an average strain rate
of 0.8%/s1. Table 3 summarizes the parameters adopted in each test,
for both steel grades. The table includes the actual section size of each
specimen, S, the frequency of the tests, f, and the applied strain range,
. Exceptionally, specimens S690_800_01 and S690_840_01 were

144

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

Table 5
Summary of fatigue crack propagation tests.
Specimens (S355)

B (mm)

Fmax (N)

Fmin (N)

Specimens (S690)

B (mm)

Fmax (N)

Fmin (N)

S355_00_01
S355_00_03
S355_25_01
S355_25_02
S355_50_01
S355_50_02
S355_75_01

7.79
7.81
7.47
7.37
7.52
7.41
7.80

5764
6118.6
7246.2
7288.3
9872.4
9345.9
19938.7

58
61.8
1811.5
1822.1
4936.2
4672.9
14954.0

0.0
0.0
0.25
0.25
0.50
0.50
0.75

S690_00_01
S690_00_02
S690_25_01
S690_25_02
S690_50_01
S690_50_02
S690_75_01
S690_75_02

4.36
4.37
4.36
4.36
4.36
4.34
4.36
4.37

3292.8
3089.9
3842.5
3575.4
4967.2
4524.6
7636.7
6862.5

32.9
30.9
960.6
893.9
2483.6
2262.3
5727.5
5146.9

0.0
0.0
0.25
0.25
0.5
0.5
0.75
0.75

tested under stress control in the elastic regime. For these two
specimens, the testing frequency was signicantly increased to result in
an appropriate testing time. Specimen S690_150_01 was tested previously as specimen S690_800-01 but, since it did not fail, the specimen
was re-tested under a high strain range.
Fig. 4 illustrates the geometry of the CT specimens adopted in the
crack propagation tests. The nominal dimensions of the specimens are
summarized in Table 4. Distinct specimen dimensions were adopted for
each material, due to the reasons mentioned before. The crack propagation tests were carried out covering four stress ratios, namely R =0.0,
R =0.25, R =0.5 and R =0.75. Two specimens were tested per stress
ratio, with one exception: only one specimen made of S355 steel was
tested under R =0.75. Table 5 summarizes the experimental program
of fatigue crack propagation tests. During tests, cracks were measured
on both side faces of the CT specimens, by direct observation through a
magnication system (resolution of 1 m). The crack propagation tests
were performed under a frequency of 20 Hz, which was reduced as
soon as the crack achieved high crack growth rates (approximately
0.3 mm/1000 cycles).
Both types of fatigue tests (smooth specimens and crack propagation tests) were performed in an INSTRON 8801 servohydraulic machine, rated to 100 kN, at room temperature and in air.

4. Results and discussion

The results of the experimental work are presented and discussed in
this section. In particular, using the results of the fatigue tests on smooth
specimens, a comparison between the mid and high strength steels is
performed taking into account their cyclic elastoplastic and strainlife
behaviors. Also, the fatigue crack growth behaviors are compared for
several stress ratios, taking into account data from the fatigue crack
propagation tests. Finally, comparisons with data available in the literature are performed.
Tables 6 and 7 summarize the results of the fatigue tests carried
out with smooth specimens, under strain controlled conditions, respectively for the S355 and S690 steel grades. These tables include
the controlled strain range and the resulting number of cycles to failure, Nf, for each specimen. Also, the parameters of the stabilized cyclic
stressstrain hysteresis loops are shown namely, the stress range and

Table 6
Fatigue test results obtained using smooth specimens of the S355 steel (R = 1).
Specimens (S355)

(%)

P (%)

E (%)

(MPa)

Nf (cycles)

S355_100_01
S355_050_01
S355_200_01
S355_040_01
S355_030_01
S355_035_01
S355_030_02
S355_040_02
S355_100_02
S355_200_02

1.00
0.50
2.00
0.40
0.30
0.35
0.30
0.40
1.00
2.00

0.570
0.219
1.443
0.093
0.021
0.051
0.003
0.072
0.637
1.427

0.429
0.281
0.557
0.307
0.279
0.299
0.297
0.328
0.363
0.573

817.39
569.54
975.47
615.40
536.34
581.51
646.56
661.21
663.57
968.21

4805
16,175
336
29,501
861,304
278,243
191,940
64,244
2009
542

the elastic and plastic strain components. These hysteresis loops were
dened for half-life.
4.1. Cyclic elastoplastic behavior
Fig. 5 presents the stabilized cyclic stressstrain hysteresis loops
obtained for both steel grades. It is clear that the S355 steel shows a
higher scatter than observed in the S690 steel. The scatter increases in
the S355 steel with the decrease in the amount of cyclic plasticity. The
hysteresis loops presented in Fig. 5 are superimposed in Fig. 6, making
their lower tips coincident with the origin of the graph. This alternative
representation of the hysteresis loops allows the assessment of the
Masing behavior [22]. The Masing behavior is observed if the upper
branches of the hysteresis loops are all coincident. For a material obeying the Masing behavior, the relationship between the cyclic stress and
elastoplastic strain ranges and the shape of the hysteresis loops may be
both described by the cyclic curve of the material. Both steels show
some degree of deviation from the Masing behavior. However, the
S690 steel may be considered a quasi Masing material, since the small
deviations observed in the upper branches of the hysteresis loops may
be attributed to scatter in the material, rather than a phenomenological
characteristic. In the case of the S355 steel, the deviations from the
Masing behavior are signicant, this material being considered a nonMasing material.
The hysteresis loops presented in Figs. 5 and 6 were determined
using a half-life criterion. This criterion may coincide with a cyclic stabilized behavior criterion, for those tests that showed stabilization. However, some tests did not show stabilization and, in those cases, the
half-life criterion corresponds to a pseudo-stabilized behavior. Fig. 7
shows the evolution of the stress amplitudes with the number of cycles
and applied strain range. It is clear that for some applied strain ranges,
no stabilization of the cyclic behavior is observed, mainly for the S355
steel. The S690 steel shows a quasi-stabilized cyclic behavior just after
the rst cycles. Only small amounts of cyclic hardening is observed for
high strain ranges ( > 1%).
Using the stabilized cyclic stressstrain hysteresis loops (see Fig. 5),
the elastic and plastic strain ranges were computed and results are listed
in Tables 6 and 7. The plastic strain amplitude was evaluated as the
width of the hysteresis loops measured over the axis, = 0. The elastic

Table 7
Fatigue test results obtained using smooth specimens of the S690 steel (R = 1).
Specimens (S690)

(%)

P (%)

E (%)

(MPa)

Nf (cycles)

S690_200_01
S690_100_01
S690_050_01
S690_050_02
S690_100_02
S690_200_02
S690_040_01
S690_040_02
S690_800_01
S690_840_01
S690_150_01

2.00
1.00
0.50
0.50
1.00
2.00
0.40
0.40
0.36
0.38
1.50

1.199
0.295
0.015
0.010
0.307
1.194
0.001
0.003
4.72E04
1.48E03
0.760

0.800
0.707
0.484
0.490
0.692
0.807
0.399
0.396
0.360
0.382
0.739

1555.20
1401.88
1105.02
1062.68
1435.58
1590.57
867.39
879.30
797.95
835.56
1565.25

190
1272
60,505
44,819
1920
160
131,000
371,000
3,807,939
1,545,579
410

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

145

Fig. 5. Stabilized stressstrain hysteresis loops: a) S355 steel; b) S690 steel.

Fig. 7. Evolution of the stress amplitude with the number of cycles and applied strain
range: a) S355 steel; b) S690 steel.

strain amplitude was evaluated from the total strain decomposition into
elastic and plastic components. The stabilized hysteresis loops were also
used to determine the stress range. The relation between the plastic
strain amplitude and the stress range denes the cyclic curve of the material. Fig. 8 compares the cyclic curves of the S355 and S690 steel
grades. The cyclic curve of the S690 steel is determined with a high determination coefcient; the determination coefcient of the cyclic curve
of the S355 steel is relatively low, which is consistent with the scatter in
the hysteresis loops observed for this material. The comparison of the
two cyclic curves shows a signicantly higher cyclic strain hardening
(cyclic strain hardening coefcient) of the S690 steel. Concerning the
slopes of the cyclic curves, they are essentially parallel, which means
very similar slopes (cyclic strain hardening exponent).

Fig. 6. Superposition of the hysteresis loops with the lower tip at the origin: a) S355
steel; b) S690 steel.

Fig. 8. Comparison of the cyclic curves between the S355 and S690 steel grades.

146

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

Table 8
Summary of cyclic elastoplastic and fatigue properties.
Material

K (MPa)

f (MPa)

S355
S690
S690 [7]
HPS 485W
[10]
A7 [10]

595.85
1282.65

956

0.0757
0.0921

0.113

952.2
1403
1191
851

0.089
0.087
0.09
0.069

0.7371
0.7396
0.9113
0.775

0.664
0.809
0.674
0.701

7095
675
5809
3686

1139

0.248

760

0.121

0.196

0.486

50,119

2NT

4.2. Strainlife behavior

Table 8 summarizes the cyclic constants and the parameters of the
Morrow's relation for the two steel grades under investigation the
S355 and S690 steel grades. The Morrow's constants resulted from
the individual tting of the Basquin [12] and CofnManson [13,14]

Fig. 10. SN curve prediction for generic structural components.

relations. The analysis of the results shows that the number of transition reversals (2NT) is very distinct between the two steels. The S690
steel shows a very small number of transition reversals, which means
that for fatigue lives above 337 cycles the fatigue behavior of the steel
is governed by fatigue strength properties rather than fatigue ductility properties. As a consequence, plastic deformation is more fatigue
damaging for the S690 steel than for the S355 steel. Fig. 9 compares
the effects of the elastic, plastic and total strains on fatigue lives, between the S355 and S690 steels. Fig. 9a) illustrates the higher fatigue
resistance of the S690 steel; Fig. 9b) shows the higher fatigue ductility
of the S355 steel; nally, Fig. 9c) shows an improved fatigue behavior
of the S690 steel only for total strain amplitudes below 3.27E 3
(lives above 6720 cycles).
The extrapolation of the comparisons based on strainlife data to
the fatigue behavior of structural components, in particular with respect to the SN curves of those components is not straightforward.
However, adopting a simplied approach based on Neuber's analysis
[17] supported by the RambergOsgood description of the cyclic
curve of the material [16], it is possible to derive SN curves, for generic structural components, characterized by an elastic stress concentration factor, Kt:


2
1=n
K 2 2nom
2
t

E
E
2K



1=n

:
2
2E
2K

For a given nominal stress range applied to a structural component

characterized by an elastic stress concentration factor, Kt, one may compute the total strain range, , at the notch root. Using this total strain
range, the cycles to failure are computed using the strainlife relations
of the materials. Fig. 10 shows the resulting SN curves generated for
generic structural components, considering both steel grades under investigation. The signicant advantage of the S690 high strength steel

Fig. 9. Comparison of the strainlife data between the S355 and S690 steel grades:
a) elastic strainlife data; b) plastic strainlife data; c) total strainlife data.

Fig. 11. Comparison of the SmithTopperWatson relations between the S355 and
S690 steel grades.

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

Fig. 12. Comparison of strainlife relations between several construction steels.

over the mild steel is clear, in terms of fatigue performance, for a wide
range of applied nominal stress ranges and elastic concentration factors.
However, the benet of using the S690 is higher for lower elastic stress
concentration factors, which means that components made of S690
should present smoother notches. This result is justied by the signicantly higher yield stress of the S690 steel, with respect to the S355
steel, which reduces the plastic deformation on components made of
this steel. The strainlife approach is usually applied to assess the
crack initiation life for structural details. Therefore, the benecial effect
of the S690 steel grade will be limited to structural components with
dominating fatigue crack initiation. Welded joints made of S690 steel
may not show any advantage over welded joints made of S355 steel
since crack initiation may have a marginal impact on the total fatigue
life.
The Morrow's equation plotted in Fig. 9c) does not account for mean
stress effects. One alternative is the relation proposed by Smith, Watson
and Topper [18] for positive maximum stresses. This model was
assessed using the data available for the two steels. Fig. 11 compares
the SmithWatsonTopper relation with the experimental data. A satisfactory agreement is veried between the model and the available experimental data for both steel grades. In this comparison, the high
strength steel shows better performance for fatigue lives above, approximately, 500 cycles.
Table 8 also presents the constants proposed in Ref. [7], for the
S690 steel grade. Also, data concerning two structural steel grades
specied in ASTM standards are presented, from Ref. [10], including
one high performance steel the HPS 485W steel (ASTM A709
[23]) and one carbon structural steel (relatively low yield strength)
as specied in the former ASTM A7 standard (replaced by ASTM
A36 standard [24]). Fig. 12 illustrates the strainlife relations generated using the Morrow's equation with constants from Table 8, allowing
a more precise comparison between materials.

147

Fig. 13. Fatigue crack propagation rates obtained for the S355 steel: analysis of stress
ratio effects.

The comparison of the results shows a discrepancy between the

properties proposed in Ref. [7] for the S690 steel and the properties
that resulted from this investigation. The discrepancy is more marked
for the fatigue ductility properties. Nevertheless, properties proposed
in Ref. [7] did not result from an experimental program on S690 steel.
Instead, the proposed fatigue properties were extrapolated from
other materials with similar monotonic strength properties, reducing
the condence on those properties.
Excluding the data from Ref. [7], an analysis of the number of transition reversals, 2NT, shows an inverse dependency between the yield
strength and the number of transition reversals. The A7 steel which is
characterized by a minimum yield strength of about 250 MPa, shows
a very high number of transition reversals, meaning a fatigue behavior
dominated by ductility properties for a wider fatigue domain than
other steels.
A comparison of the fatigue behavior between the S355, HPS 495W
and S690 steel grades, shows a remarkable trend. The strainlife curves
intersect each other at about 1.4 104 reversals (7 103 cycles) (see
Fig. 12). For lives above this intersection point, the fatigue resistance increases with the yield strength of the material; inversely, for lives bellow the intersection point the fatigue performance increases with
decreasing yield strength of the materials. It is worth noting that the
HPS 495W steel exhibits a minimum yield strength of 495 MPa which
represents an intermediate strength between the S555 and S690 steels.
4.3. Fatigue crack propagation rates
The results of the fatigue crack propagation tests are presented in
this section. The crack growth rates, da/dN, are plotted as a function of
the stress intensity factor range, K. The crack growth rates were

Table 9
Constants of Paris' law for the S355 and S690 steel grades.
Material

Ca

R2

S355

0.0
0.25
0.50
0.75
0.25 + 0.50 + 0.75
0.0
0.25
0.50
0.75
0.25 + 0.50 + 0.75
0
0.5

2.5893E 15
2.5491E 15
8.2764E 16
4.9643E 14
2.1111E 15
6.8261E 13
2.3196E 12
2.9529E 14
1.2956E 15
2.2607E 13
6.39E14
1.07E13

3.5622
3.7159
3.8907
3.2328
3.7447
2.8789
2.7592
3.4517
3.9595
3.1255
3.12
3.14

0.9716
0.9841
0.9810
0.9504
0.9872
0.9713
0.9549
0.9703
0.9902
0.9533

S690

HPS 485W [10]

a

da/dN in mm/cycle and K in N.mm1.5.

Fig. 14. Fatigue crack propagation rates obtained for the S690 steel: analysis of stress
ratio effects.

148

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

Fig. 15. Comparison of fracture surfaces for the compact tension specimens at the region of stable propagation (R = 0.0): a) S355 steel; b) S690 steel.

computed using the seven point incremental polynomial technique,

as proposed in the ASTM E647 standard [21]. The stress intensity factor ranges were computed using the formulation proposed in ASTM
E647 for the CT specimens [21]:

F 2

2
3
4
K p
0:886 4:6413:32 14:72 5:6
=
B W 1
10
3

where: = a / W, a is the crack size, B is the thickness of the specimen,

W is the width of the specimen and F is the applied load range. The
experimental crack sizes are computed as the average value of the
two measurements performed on both faces of the CT specimens.
The experimental crack propagation data was correlated with Paris'
law, since only the crack propagation regime II was covered by the
present experimental research.
Table 9 summarizes the constants of Paris' law for both steel
grades and for combinations of several stress ratios. The table also includes the determination coefcients resulting from the adjustment
of Paris' law to the experimental data. All determination coefcients
are above 0.95, which represents very high correlations.
Fig. 13 illustrates the effects of the stress ratio on fatigue crack propagation rates, for the S355 steel grade. An increase in fatigue crack propagation rates is clear, when the stress ratio changes from 0 to any
positive stress ratios considered in the experimental program. Also, it
is clear that all the positive stress ratios resulted in similar crack propagation rates. This behavior is consistent with a crack closure effect that

occurs between R = 0.0 and R = 0.25. For R = 0.0 there is some

crack closure, the applied stress intensity factor range being not fully effective. For R = 0.25 and higher, there is no crack closure, the applied
stress intensity factor range being fully effective.
Fig. 14 shows the effects of the stress ratios on fatigue crack growth
rates for the S690 steel grade. Similar to the S355 steel, the S690 steel
shows distinct behaviors between R = 0.0 and the other tested positive
stress ratios. No signicant differences are found on crack growth rates
for R = 0.25, R = 0.5 and R = 0.75. This behavior is consistent with
the crack closure phenomena. Nevertheless, the crack closure effect is
less accentuated than observed for the S355 steel. One possible explanation is the fact that the crack closure may be induced by the roughness
of crack surfaces, which facilitates the crack closure before the load
reaches the null value. This roughness is closely related to the material
grain sizes, which is higher for the S355 steel. Effectively, Fig. 15 compares the fracture surfaces obtained for the CT specimens, under stable
crack propagation, for R = 0.0. It is clear that the fracture surface of the
S355 steel shows a higher roughness than that observed for the S690
steel. Therefore, the fatigue crack growth rates of the S355 steel are
much more inuenced by the crack closure effects than expected for
the S690 steel.
Fig. 16 compares the fatigue crack propagation rates between the
two steel grades under investigation. The comparison is performed individually for each stress ratio. It is clear that the S690 steel shows the
highest fatigue crack growth rates for all tested stress ratios. This result
may be justied by the ner grain of the S690 steel which facilitates the
fatigue crack propagation. Fig. 17 compares all crack growth data generated for the S355 and S690 steels. Besides the higher crack growth rates
observed for the S690 steel, it is clear that the crack growth rates of the
S690 steel, for R = 0.0, are similar to the crack growth rates of the S355
steel for R = 0.25 and higher.
The crack propagation constants for the HPS 485W steel are also included in Table 9, which were determined by Chen et al. [10]. Fig. 18
compares the fatigue crack propagation rates between the S355, HPS
485W and S690 steels, for a wide range of stress intensity factors and
for two stress ratios, namely R = 0 and R = 0.5. It is interesting to
note that fatigue crack propagation rates increase with the yield
strength of steels, for both stress ratios.
In order to allow a further comparison between the two steels, three
notched details were considered with distinct stress concentrations
(Kt = 1.90, 3.26, 4.52) and the same thicknesses of the respective tested
CT specimens. These details were subjected to a remote uniform stress
with sinusoidal shape and R = 0. Using available crack propagation
data for R = 0, SN curves were generated for the details. Paris' law
was integrated assuming an initial crack size of 0.5 mm (see Eq. (8)).
The required stress intensity factors were evaluated using nite element analysis and the J-integral method. In the three details, the same
resisting section length was considered (95 mm). Fig. 19a) compares
the resulting SN curves that take into consideration the crack propagation only. Unstable crack propagation was accounted considering the
maximum stress intensity factor registered in crack propagation tests
for each material (see Fig. 16a). The lower crack growth rates of the
S355 steel leads to higher fatigue lives for medium to low stress ranges.
However, for high stress ranges, a reverse condition due to the lower
toughness of the S355 with respect to the S690 steel is observed.
Fig. 19b) presents the simulations of the SN curves considering both
crack initiation and crack propagation phases. The crack initiation was
computed using the strainlife approach presented in the previous section and was assumed a crack initiation criterion of a 0.5 mm deep
crack. The resulting global SN curves show the same trend observed
in Fig. 10 for the crack initiation curves the notched details made of
the S690 steel show a higher fatigue resistance than the details made
of the S355 steel. This result means that crack initiation is dominant
for details under consideration. Crack propagation signicance is limited for the notched details under consideration, but its importance increases if larger member sections are assumed.

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

149

Fig. 16. Comparison of the crack growth rates between the S355 and S690 steel grades: a) R = 0.0; b) R = 0.25; c) R = 0.50; d) R = 0.75.

5. Conclusions
The fatigue behavior of the S355 mild steel and S690 high strength
steel grades were evaluated by means of an experimental program
which included fatigue tests of smooth specimens as well as fatigue
crack propagation tests. The analysis of the results leads to the following
conclusions:
The fatigue tests on smooth specimens showed that the S690 high
strength steel grade exhibits a lower fatigue resistance than the
S355 steel, for strain amplitudes higher than 0.33% or fatigue lives
bellow 6720 cycles, which represents the low cycle fatigue regime.
In the high cycle fatigue regime, the S690 steel shows a higher fatigue resistance than the S355 steel. This superior fatigue resistance,

Fig. 17. Comparison of all crack propagation data obtained for the S355 and S690 steel
grades.

based on smooth specimen test data, corresponds to a higher resistance to fatigue crack initiation.
Concerning the fatigue crack propagation rates, the S690 steel shows
systematically higher propagation rates than the S355 steel, for any
tested stress ratio. This implies that there is no advantage of using
the S690 steel for structural components whose fatigue life is dominated by crack propagation, rather than crack initiation. Nevertheless, the fracture toughness of the S690 steel was demonstrated to
be superior to that of the S355 steel (for the tested thicknesses)
which beneciate the crack propagation resistance of the S690
steel for high stress levels.
Both S355 and S690 structural steels showed crack propagation rates
clearly affected by the crack closure, the S355 steel being more sensitive to this phenomenon due to the higher grain size which leads
to higher roughness on fracture surfaces.
The fatigue properties of the S690 steel were assessed with a lower

Fig. 18. Comparison of fatigue crack propagation data between three structural steels
with distinct strength properties.

150

A.M.P. de Jesus et al. / Journal of Constructional Steel Research 79 (2012) 140150

As a concluding remark, the design of structural details in high

strength steels should take advantage of the superior resistance of
these steels to fatigue crack initiation. The utilization of high strength
steels increases the fatigue sensitivity of the structural detail to sharp
notches reducing the fatigue life with respect to details made of mild
steel.

References

Fig. 19. SN curve prediction for three generic notched components: a) accounting fatigue crack propagation; b) accounting fatigue crack initiation and propagation.

scatter than resulted for the S355 steel, which means higher quality
of the high strength steel with respect to the mild steel.
Despite the fact that strainlife data have shown a better fatigue performance of the S355 steel for low-cycle fatigue regimes, this result
may not be extrapolated directly to structural components under
load/stress control due to the superior yield strength of the S690
steel. It was shown that the S690 steel, due to its superior yield
strength, shows a higher resistance to fatigue crack initiation for
structural components under stress control, for a wide range of stress
concentration factors and applied stress ranges, covering both low
and high cycle fatigue regimes. The superior fatigue crack initiation
resistance of the S690 steel grade for structural details may not be
relevant for welded joints, since fatigue life is often dominated by fatigue crack propagation.
The comparison of the fatigue data from this study with the fatigue
data published in literature for the HPS 485W steel, which is a steel
grade with intermediate yield strength between the S355 mild steel
and the S690 high strength steel, lead to the following conclusions:
The increase of the yield strength of the steel promotes the rotation of
the strainlife curves around the fatigue life of about 7103 cycles,
increasing the fatigue resistance in the high cycle regime and decreasing the fatigue resistance in the low-cycle fatigue regime.
The fatigue crack propagation rates increase with the yield strength
of the structural steels, independently of the stress ratio.

[1] Sperle J-O. High strength sheet steels for optimum structural performance. Conference on Iron and Steel Today, Yesterday and Tomorrow, Proc. 250th Anniversary of The Swedish Ironmasters Association; 1997.
[2] Miki C, Homma K, Tominaga T. High strength and high performance steels and
their use in bridge structures. J Constr Steel Res 2002;58:320.
[3] Jensen L, Bloomstine ML. Application of high strength steel in super long span
modern suspension bridge design. Proceedings of the Nordic Steel Construction
Conference (NSCC 2009), September 24, Malm, Sweden; 2009.
[4] Veljkovic M, Feldmann M, Naumes J, Pak D, Rebelo C, Simes da Silva L. Friction
connection in tubular towers for a wind turbine. Stahlbau 2010;79:660-8.
[5] Willms R. High strength steel for steel constructions. Proceedings of the Nordic
Steel Construction Conference (NSCC 2009), Malm, Sweden; 2009. p. 597-604.
[6] Kayser CR, Swanson JA, Linzel DG. Characterization of material properties of HPS-485W
(70W) TMCP for bridge girder applications. J Bridg Eng 2006;11(1):99108.
[7] Beretta S, Bernasconi A, Carboni M. Fatigue assessment of root failures in HSLA steel
welded joints: a comparison among local approaches. Int J Fatigue 2009;31:102-10.
[8] European Committee for Standardization (CEN). EN1993-1-9: Eurocode 3: design
of steel structures, part 19: fatigue. Brussels: European Standard; 2004.
[9] Chen H, Grondin GY, Driver RG. Fatigue resistance of high performance steel.
Structural engineering report no 258. Canada: University of Alberta; 2005.
[10] Chen H, Grondin GY, Driver RG. Characterization of fatigue properties of ASTM
A709 high performance steel. J Constr Steel Res 2007;63:838-48.
[11] European Committee for Standardization (CEN). EN 10025: hot rolled products of
structural steels. Brussels: European Standard; 2004.
[12] Basquin OH. The exponential law of endurance tests. Proc Am Soc Test Mater
1910;10:625-30.
[13] Cofn LF. A study of the effects of the cyclic thermal stresses on a ductile metal.
Trans ASME 1954;76:931-50.
[14] Manson SS. Behaviour of materials under conditions of thermal stress, NACA
TN-2933. USA: National Advisory Committee for Aeronautics; 1954.
[15] Morrow JD. Cyclic plastic strain energy and fatigue of metals. Int. Friction,
Damping and Cyclic Plasticity, ASTM, STP 378; 1965. p. 45-87.
[16] Ramberg W, Osgood WR. Description of stressstrain curves by three parameters, NACA TN-902. USA: National Advisory Committee for Aeronautics; 1943.
[17] Neuber H. Theory of stress concentration for shear-strained prismatic bodies with
arbitrary nonlinear stressstrain law. J Appl Mech 1961;28:544-50.
[18] Smith KN, Watson P, Topper TH. A stressstrain function for the fatigue of metals.
J Mater 1970;5(4):767-78.
[19] Paris PC, Gomez MP, Anderson WE. A rational analytic theory of fatigue. Trend
Eng 1961;13(1):914.
[20] American Society for Testing, Materials. ASTM E606: standard practice for strain
controlled fatigue testing. Annual Book of ASTM Standards, Vol. 03.01; 1998.
p. 557-71. West Conshohocken, PA.
[21] American Society for Testing, Materials. ASTM E647: standard test method for
measurement of fatigue crack growth rates. Annual Book of ASTM Standards,
Vol. 03.01; 1999. p. 591-629. West Conshohocken, PA.
[22] Abdel-Raouf HA, Plumtree A. Cyclic stressstrain response and substructure. Int J
Fatigue 2001;23:799-805.
[23] American Society for Testing, Materials. ASTM A709/A709M: standard specication for carbon and high-strength low-alloy structural steel shapes, plates, and
bars and quenched-and-tempered alloy structural steel plates for bridges. Annual
Book of ASTM Standards, Vol. 01.04; 2004. West Conshohocken, PA, USA.
[24] American Society for Testing, Materials. ASTM A36/A36M: standard specication
for carbon structural steel. Annual Book of ASTM Standards, Vol. 01.04; 2004.
West Conshohocken, PA, USA.