Accepted Manuscript

Influence of microstructure on fatigue process in a low carbon

steel. Analysis and modelling

Donka Angelova, Rozina Yordanova, Svetla Yankova

PII: S1350-6307(17)30477-6
DOI: doi: 10.1016/j.engfailanal.2017.06.053
Reference: EFA 3218
To appear in: Engineering Failure Analysis
Received date: 13 April 2017
Revised date: 18 May 2017
Accepted date: 26 June 2017

Please cite this article as: Donka Angelova, Rozina Yordanova, Svetla Yankova , Influence
of microstructure on fatigue process in a low carbon steel. Analysis and modelling,
Engineering Failure Analysis (2017), doi: 10.1016/j.engfailanal.2017.06.053

This is a PDF file of an unedited manuscript that has been accepted for publication. As
a service to our customers we are providing this early version of the manuscript. The
manuscript will undergo copyediting, typesetting, and review of the resulting proof before
it is published in its final form. Please note that during the production process errors may
be discovered which could affect the content, and all legal disclaimers that apply to the
journal pertain.
 ACCEPTED MANUSCRIPT
INFLUENCE OF MICROSTRUCTURE ON FATIGUE PROCESS IN A LOW CARBON
STEEL. ANALYSIS AND MODELLING

Donka Angelova, Rozina Yordanova, Svetla Yankova

University of Chemical Technology and Metallurgy, 8 St. Kliment Ohridski, Blvd., 1756 Sofia, Bulgaria
donkaangelova@abv.bg, r.yordanova@uctm.edu, svetla_y@abv.bg

Corresponding author: Donka Angelova, donkaangelova@abv.bg

ABSTRACT. Fatigue in a low-carbon steel is investigated through observation on surface

crack propagation and on growth of cracks in preliminary notched specimens. Testing uses three

PT
groups of specimens. For surface crack observation there are two groups of samples consisting of
cylindrical specimens subjected to tension-tension and rotating-bending fatigue; in this case surface
microstructurally-short crack propagation is monitored by acetate-foil replica technique. For crack

RI
growth observation (in situ) in notched specimens there is a third group of samples including flat
specimens preliminary notched by FIB-technique and then subjected to pure-bending fatigue. Here
microstructurally-short crack propagation is examined at interruptions of each test at a given equal

SC
number of cycles for detailed observation of specimen surface by optical- and SEM-microscopy. The
study is focused on examining of crack paths in terms of interaction between the propagating short
cracks and the microstructure, and on a suitable mathematical description of crack growth in the
NU
investigated microstructure. The obtained data for pure-bending fatigue show higher crack growth
rates (dominated by the interaction with ferrite and pearlite grain boundaries and interfaces, ferrite
grains, pearlite colonies and non-metal inclusions) and shorter fatigue lifetimes than those found for
MA

rotating-bending fatigue. In comparison, the registered tension-tension fatigue data present the lowest
crack growth rates, due to much lesser loading than that applied at rotating-bending and pure-bending
fatigue. Based on data obtained, a Parabolic-linear model Crack growth rate Crack length is used
for describing and predicting adequately short crack propagation under the specified three types of
fatigue. The model is supported by a comparison between the predicted and the actual fatigue
ED

lifetimes.

Keywords: Fatigue; Short-crack propagation; Microstructure; Low-carbon steel; Fatigue modelling

PT

INTRODUCTION
CE

In the present work investigations on microstructurally-short fatigue crack growth are

performed in order to clarify the relationship between the typical features of the studied
microstructure and the specific behaviour of cracks propagating through it.
AC

Short cracks are recognized to have considerable influence on fatigue strength of a large number
of engineering components and structures. The standard approach usually uses the Paris relation to
describe crack growth data in terms of crack growth rate versus stress intensity factor range. But it
does not describe the propagation of sub-critical cracks which control a large part of the fatigue
lifetime of a component. These cracks grow at stress intensity ranges, smaller than the threshold value
of the stress intensity factor and show a specific behavior, with a widely fluctuating crack growth
rates higher than those of long cracks, described by the standard LEFM-procedure at the same
nominal driving force. In order to understand mechanics, mechanisms and microstructural effects on
short fatigue crack propagation, the mentioned specific crack growth behavior has been examined in
many different metallic materials. It is well known that usually there is a change of crack growth
direction and considerable slowdown in crack growth rate when cracks propagate across grain
boundaries and in the vicinity of an interface between phases with different mechanical properties.
These microstructural features can act as stress raisers or can cause shielding effect at the crack tip.
The barriers can reduce the effective driving force for crack propagation and that is why short crack

1
 ACCEPTED MANUSCRIPT
growth is so microstructurally sensitive; all the variations in the microstructure surrounding the crack
tip are responsible for its specific growth behavior [1, 2].
In the present work short fatigue crack propagation behavior is investigated under tension-
tension, (T-T), rotating- bending (RB) and pure-bending (PB) loading conditions [3-10]. Three
different groups of specimens are used: for surface crack observation two groups of hour-glass
specimens for T-T and RB fatigue tests [3]; and for crack growth observation (in situ) in notched
specimens a group of flat samples notched by Focused Ion Beam (FIB) technique, tested under PB
[4, 5, 10]. A plastic-foil replication is used for short fatigue-crack growth monitoring at T-T and RB
fatigue and direct observation by optical- and SEM-microscopy at PB fatigue. The obtained data are
presented by microstructural photos, and plots Crack length Cycles and Crack growth rate
Crack length; some comparisons between them are made. A mathematical model of fatigue crack
propagation Parabolic-Linear Model, PLM, is proposed in [6] and employed for mathematical

PT
propagation description of data obtained.

MATERIAL, SPECIMENS AND TESTING

RI
A rolled low-carbon, low-alloyed steel, ROLCLAS, marked as 09Mn2 Steel (according to the
Bulgarian Construction Steel Standard), used mostly for offshore applications and in shipbuilding,
was subjected to tension-tension, rotating-bending and pure-bending fatigue.

SC
The chemical composition of ROLCLAS and its mechanical and microstructural characteristics
are given in Table 1.
ROLCLAS was available in sheets of 8 mm thickness. Its microstructure revealed a sequence
NU
of long and uniform pearlite and ferrite bands, as shown in Fig. 1a. The bands are wider in the middle
of the sheet but loose and thinner close to the surface. Two groups of hour-glass specimens were
under investigation: Group A consisting of specimens for T-T fatigue with geometry shown in Fig.
MA

1b; and Group B including specimens for RB fatigue with geometry presented in Fig. 1c.
A third group (Group C) of flat specimens with shape and sizes shown in Fig. 1d was specially
machined in the Erich Schmid Institute of Material Science, Leoben-Austria for microstructural
investigations, [5]. On each specimen four micro-notches were machined by FIB technique in
different position in the microstructure, Fig. 2a, 2b, and the notch geometry presented in Fig. 2c,
ED

2d. Three of the notches were central (made on the longitudinal axis of specimen and perpendicular
to it) and located in-between the pearlite bands (notch 1 is very close to one of the pearlite band); the
forth one was an edge notch aligned with the top notch from the group of the central notches. The
PT

distance between notches was 200 m.

Chemical Composition
CE

C, % Si, % Mn, % Cr, % Ni, % P, % S, % Cu, % Al, % As, %

0,09 0,28 1,63 0,05 0,04 0,017 0,026 0,13 0,12 0,014
Mechanical and Microstructural Properties
Tensile Strength Proof strength Cross section Hardness Average grain
AC

B, MPa 0,2, MPa contraction, , % HB, MPa size, [ m]

482 382 62,3 148 25,6

Table 1: Chemical Composition, Mechanical and Microstructural Properties of ROLCLAS 09Mn2

The T-T tests were carried out on Amsler Fatigue Machine in Acad. A. Balevski Institute of
Metals Science, Bulgarian Academy of Sciences, Sofia. The chosen loading conditions for the
specimens from Group A were: applied stress range = 396 MPa, stress ratio R = 0,1, frequency
190 Hz. For RB tests a table model Fatigue Rotating-Bending Machine, FATROBEM-2004 was
used, designed and assembled in Fracture and Fatigue Laboratory in UCTM-Sofia, [4]. The
specimens from Group B were tested at stress ranges = 580 MPa and 620 MPa, applied stress
ratio R = 1, frequency 11 Hz. The PB tests ( with flat specimens of Group C) was carried out on
Schenk Fatigue Machine in Erich Schmid Institute of Materials Science, Austrian Academy of
Sciences, Leoben under constant stress control at: stress level 580 MPa, stress ratio R = 1, frequency
5Hz [5].

2
 ACCEPTED MANUSCRIPT

PT
RI
b c d
Figure 1: Microstructure of cross section of ROLCLAS specimens (a), hour-glass fatigue specimens for T-T

SC
(b) and RB fatigue (c), flat fatigue specimen for RB fatigue (d); all dimensions are in mm
NU
MA

a b
ED
PT

c d e
CE
AC

f
Figure 2: FIB- machined micronotches on a specimen of Group C: central notches (a), edge notch (b);
geometry of notches (c) and (d); cracks starting from notches under stress range 580 MPa (e); two parts of
the fracturing crack accordingly to the axis perpendicular to the top central notch (f) - right part with cracks
1, 2 (started from the edge notch aligned with the top central notch) and crack 3 (started from the top central
notch), and left part with crack 4 (started from the top central notch); crack bifurcation at point A and
merging of cracks 2 and 3 at point B

3
 ACCEPTED MANUSCRIPT
The surface short-crack propagation on the hour-glass specimens was monitored by acetate-foil
replica technique during a fixed interval of fatigue cycles and observed on the replicas by an optical
microscope for measuring registered surface crack lengths. The short fatigue-crack experiments with
the flat specimens included interruptions of the test at every 1000 cycles for in-situ examination of
the specimen surface under optical microscope and SEM-microscope. The crack lengths were
measured by using an image analyzer, as described in [5].

RESULTS, ANALYSES, DISCUSSION

Data obtained from the fatigue experiments with specimens from Groups A, B, C the crack
lengths a, at the corresponding numbers of cycles N are plotted as:
a) functions Crack length, a Number of cycles, N, {a-N}, in log-log scale, shown for Group
C in Fig. 3a for cracks 1-4;

PT
b) functions Number of cycles, N Crack length, a, {N-a}, in semi-log scale (N in log scale and
a in ordinary scale); they are presented
- for Group C in Fig. 3b (cracks 1-4), Fig. 4-1 (crack 2), Fig. 4-2 (showing the whole length of

RI
crack 2 (a) and its first 300 m (b)), Fig. 4-3 (cracks 1, 2), Fig. 5-1 (crack 4), Fig. 5-2
(showing the whole length of crack 4 (a) and its first 300 m (b)), Fig. 5-3 (crack 3, 4), Fig. 6
(part of crack 1, crack 2 and part of crack 3; this crack combination form the combined right crack

SC
part from the two-parts notch-specimen fracturing crack, the left part consists only of crack 4); in
Figs. 4-3a, 5-3a a decreased image of Fig. 2f is represented only to show cracks location
under discussion in the whole microstructure picture,
NU
- for Group B in Fig. 7 (showing: at = 580 MPa the whole length of the major crack (a) and
its first 500 m (b); at = 620 MPa the major crack length (c) and its first 500 m (d),
- for Group A in Fig.8 ( showing the whole length of the major crack at = 396 MPa (b);
MA

c) functions Crack growth rate da/dN Crack length a, {da/dN -a}, in semi-log scale (N in log
scale and a in ordinary scale); they are presented for Group C in Fig. 3c, and for Groups A,
B, C in Figs. 4-1, 4-2, 4-3, 5-1, 5-2, 5-3, 6, 7, 8 as it is fully explained in b);
d) functions Crack growth rate da/dN Crack length a, {da/dN -a}, in semi-log scale (N in log
ED

scale and a in ordinary scale) represented by a Parabolic-Linear Model, PLM, proposed and
described in detail in [6]; PLM consists of three sections two parabolas and a linear part
and is applied to the data of all presentations {da/dN -a} and shown in Fig. 4-3, Fig. 5-3, Fig.
6; the three sections of PLM are located in the following a-intervals [0, d1], [d1, d2], [d2 , a])
PT

where d1, d2 are known as microstructural barriers connecting with microstructural influence on
crack propagation;
e) functions Crack growth rate da/dN Crack length a, {da/dN -a}, in log-log scale, shown in
CE

Fig. 9 for Groups A, B, C; the same PLM from [6] is applied here to the data of {da/dN -a}
and shown in the same Fig. 9;
f) functions Cycles to failure calculated by PLM Nfmodel Cycles to failure obtained
AC

experimentally Nfexp, {Nfmodel -Nfexp}, in log-log scale, plotted in Fig. 10 for Groups A, B, C;
the adequacy of PLM-s are proved through a comparison between Nfmodel and Nfexp in {Nfmodel
-Nfexp} presentation, Fig. 10, [6-9]; = |100(, , )/, | coefficient showing the
precision of PLM.
Propagation of some of the observed short cracks through the microstructure (Group C) is
presented in Fig. 2e, 2f, Figs. 4-1, 4-3, Figs 5-1, 5-3, Fig. 6, Fig. 8.
The first analysis begins with a notched specimen from Group C. This specimen shows
appearance of 8 cracks started from the notches, Figs. 2e, 2f, [5]. Our main interest is focused on
cracks 1, 2, 3, 4 which fracture the specimen, Fig. 2f, Fig. 3a, Figs. 4-1, 4-2, 4-3, Figs. 5-1, 5-2, 5-3,
Fig. 6, [7, 8].The longest crack is crack 2, which originates from crack 1 as its branch at 15000 cycles.
At this moment crack 1 has a length of 107.73 m. After the bifurcation (point A in Fig. 2f) the crack
2 grows faster than the other cracks during the rest of fatigue lifetime and causes the specimen failure,
merging with crack 3 (point B in Fig. 2f) which starts from the central notch 1. Finally, crack 1

4
 ACCEPTED MANUSCRIPT
showing high growth rate during its propagation, Fig. 3c, becomes a non-propagating crack at 27000
cycles with a length 367.36 m.
The central notch 1 is machined in the ferrite band, so that its right side is very close to the next
pearlite band, Figs. 2a and 2f. Thus the crack 3 starts its growth through the pearlite band at the very
early loading cycles without significant retardation, [5]. On the other side of notch 1 crack 4 begins
its growth into the ferrite band; in the beginning its growth is slower than that of the crack 3.
Afterwards both cracks continue to grow faster and faster through the microstructure having higher
growth rate than those of the other central-notches cracks, Fig. 3c [5].
It can be clearly seen in Figs. 3b, 3c, Fig. 4-3, Fig. 5-3 and Fig. 6 that short fatigue crack growth
data exhibit considerable fluctuations and scattering and alternating decreases and increases in crack
propagation rates. This effect is caused by the interactions between the crack tip and associated
microstructural obstacles, and the crack geometry. Such a behavior can be described by the PLM

PT
presented in Fig. 4-3, Fig. 5-3 and Fig. 6, [6, 9]. In more details, the highest growth rates belong to
cracks 2 and 3 (Fig. 3c) which merge with crack 4 and cause the specimen failure. Crack 2 starts its
propagation as a branch of crack 1 relatively late; it shows the highest growth rate and has different

RI
behavior in comparison with all the other cracks. In the beginning crack 2 starts with permanently
increasing growth rate reaching a length of 53.79 m. At the same time all the other cracks show

SC
mostly decreasing growth rates due to an intensifying crack closure effect.
NU
MA
ED
PT

a b
CE
AC

c
Figure 3: Plots Crack length, a Number of cycles, N for cracks from Groups A, B, C T-T, RB, PB
fatigue (a), Number of cycles, N Crack length, a for cracks 1, 2, 3, 4 from Group C specimen PB
fatigue (b), Crack growth rate da/dN Crack length a for the first 100 m of cracks 1, 2, 3, 4 from
Group C specimen PB fatigue (c) ; plots from (b) and (c) show a coordinate system with Y-axis located
in the middle of the central notches, which specifies left and right propagation of the cracks
accordingly to the central notches

5
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
Figure 4-1: Microstructural path of crack 2 (Group C specimen), together with its corresponding growth
MA

rates and its propagation in N-a terms

ED
PT
CE
AC

a b
Figure 4-2: Combined plots Crack growth rate da/dN Crack length a & Number of cycles, N Crack
length, a {da/dN a & N a} for crack 2 (Group C specimen) in semi-log scale presentations for the
whole length of crack 2 (a), and for the first 300 m of its propagation (b); d1 and d2 are the microstructural
barriers from the Parabolic-Linear Model, PLM, proposed in [6] and presented in Fig. 4-3

6
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
MA
ED
PT

Figure 4-3: Decreased image of Fig. 2f representing cracks location under discussion in the
CE

whole microstructure picture (a); microstructural paths of cracks 1 and 2 (Group C specimen) with their
corresponding growth rates in da/dN -a terms, crack propagation in N-a terms, and PLM applied
to the fatigue data of presentation da/dN a (b, c)
AC

In detail as it is described in [5] the micronotch surfaces do not contact during the loading, hence in
the beginning the propagating crack is open during the complete load amplitude (even at stress ratio
R = -1); with the growth of crack length, new-generated fracture surfaces come to an effective contact
and crack closure load increases.
A careful examining of combined fracture-crack path (of cracks 1, 2, 3, 4) shows the following.
Crack 1 starts from the edge notch, Figs. 2b, 2e, 2f, Fig. 4-3 and Fig. 6 at a length of 68.11 m
it reaches an obstacle in the ferrite band that leads to decrease in its growth rate [5]. To overcome this
obstacle crack 1 changes its direction at almost 90 degrees. Five thousand cycles later when it reaches
a length of 107.73 m, a branch appears as crack 2, Fig. 4-1, Fig. 4-3 and Fig. 6. Even when crack 1
shows steep slope on its growth curve (Figs. 3a and 3b), and has significantly high growth rate before
the appearance of crack 2 (Fig. 3c), finally it stops without any further propagation. At the same time
crack 1 does not have pronounced growth-rate drops through the microstructure, because of its start
from the edge notch and its propagation near to the specimen surface, where the microstructure does
not have well defined ferrite-pearlite bands and looks more homogeneous, Fig. 1a.

7
 ACCEPTED MANUSCRIPT
Although the crack 2 shows the highest growth rate there are some large slowdowns during
its propagation, Fig. 4-1, Fig. 4-3 and Fig. 6. These slowdowns can be associated with crack entrances
into the pearlite structure. Anyway, at a length of 467.43 m its growth stops; at this point crack 2
leaves a pearlite band and enters into a ferrite band. This is an example for crack retardation when it
approaches the immediate vicinity of the strong-week (pearlite-ferrite) metal interface. There are
more examples of this type of behaviour, shown by crack 4 (Fig. 5-1, Fig. 5-3, Fig. 6) and cracks 5
and 8 [5]. According to Pippan [10] the driving force of crack propagation in such a non-uniform
material is not only depends on crack length, applied load, geometry of microstructural elements, but
also on the physical properties of the different phases and their geometrical arrangement.
In all the cases of 1-8 cracks, they exhibit relatively high growth rates traversing a pearlite band.
It shows that microstructurally short fatigue cracks can sometimes propagate faster in the
mechanically stronger material which in this case is the pearlite phase [5].

PT
Crack 3 starts its growth immediately from the central notch 1 into a pearlite band while the
other cracks start propagation into the ferrite grains, Fig. 2f, Fig. 5-3 and Fig. 6. Its initial growth rate
is not quite different than those of the other cracks how it can be seen in Fig. 3c. It seems that there

RI
is no difference in the initial crack growth rate, if the crack starts either in the pearlite or in the ferrite
structure. Crack 3 shows slowdowns either starting its growth into a pearlite band or having its
direction changes when it passes trough a ferrite band; some slowdowns can be observed later when

SC
the crack approaches the next pearlite band and shows increasing growth rate inside it [5].
Crack 4 starts from the central notch 1 and forms the left part of the two-parts fracturing crack
of the notch specimen (the right part of the notch specimen consists of the following combination
NU
crack 1 to the bifurcation point A, crack 2, and crack 3 to the merging point B, and can be observed
in Fig. 6). The microstructural path of crack 4 through the ferrite-pearlite microstructure can be seen
in Fig. 5-1, Fig. 5-3. After a considerable retardation in front of the pearlite bands, crack 4 enters into
MA

a wide pearlite band. Since this happens, the crack growth rate da/dN increases rapidly and reaches
its highest values during the whole fatigue lifetime. The described high growth rate is caused by a
row of nonmetal inclusions which are used by crack 4 for easier propagation along them, even that
they are perpendicular to its propagation direction. Such nonmetal inclusions consist of plastic MnS
ED

which has been preliminary deformed during the hot rolling process of the studied steel.
Part of crack 1, crack 2 and part of crack 3 form together the combined right part of the two-
parts notch-specimen fracturing crack, which can be clearly seen in Fig. 6 (the left part of the two-
parts fracturing crack consists of crack 4 - Fig. 2f, Fig. 5-1, Fig.5-3). The line L1L2, Fig.6 effectively
PT

represents the merging process of cracks 2 & 3 against the actual oscillations of the growth rates of
crack 2, and crack 3 to the merging point B.
So, the major fracturing crack of the studied notch-specimen, Group C, is a complicated
CE

combination of crack 4, and part of crack 1, crack 2 and part of crack 3.

As it is already mentioned above, a mathematical Parabolic-linear model, PLM, proposed in
[6] is used to describe fatigue data from all the presentations {da/dN -a} in semi-log scales (shown in
AC

Fig. 4-3, Fig. 5-3, Fig. 6); PLM consists of two parabolas and a final linear part, divided by two
microstructural barriers d1 and d2 at which cracks significantly slow down their growth rates. Each
parabola and the linear part describe three different stages of crack development. The first two stages
are greatly influenced by the studied different microstructures, when the third one is independent and
can be described by the linear fracture mechanics.
The complete mathematical description of the major fracturing crack of the notch-specimen
consists of PLM-s for
- crack 4, shown in Fig. 5-3c and,
- the complicated combination of part of crack 1, crack 2 and part of crack 3, shown in Fig.
6 and for the first time in the present paper.
All the minimums of crack growth rates in Figs. 4-1, 4-3, Figs. 5-1, 5-3 and Fig.6 are connected
by arrows with the corresponding elements of the microstructure which slow down cracks
propagation.

8
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
MA

Figure 5-1: Microstructural path of crack 4 (Group C specimen), together with its corresponding growth
rates and its propagation in N-a terms
ED
PT
CE
AC

a b
Figure 5-2: Combined plots {da/dN a & N a} for crack 4 (Group C specimen) in semi-log scale
presentations for the whole length of crack 4 (a), and for the first 300 m of its propagation (b); d1 and d2 are
the microstructural barriers from the Parabolic-linear model, PLM, proposed in [6] and presented in Fig. 5-3

9
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
MA
ED
PT
CE

Figure 5-3: Decreased image of Fig. 2f representing cracks location under discussion in the
whole microstructure picture (a); microstructural paths of cracks 3 and 4 (Group C specimen) with their
corresponding growth rates in da/dN -a terms, propagation in N-a terms, and PLM applied to the
AC

fatigue data of presentation da/dN a (b, c)

The same elements are connected by other arrows with the corresponding crack lengths from the
newly inserted Plot {N a}, built in semi logarithmic scale to correspond to crack growth through the
microstructure. In all figures there is an example of a perpendicular dash line to the a-axis, which
connects a minimum of the crack growth rate from the Plot {da/dN a} with the corresponding crack
length from the Plot {N a}. The dash line is a resulting line which is connected with the projection
(on the a-axis) of the real crack length (between two successive interruptions of the cycles); usually
crack propagates at an angle to the a-axis. Only in the case when the real crack length between two
successive interruptions of the cycles is zero, which means that the crack is perpendicular to the a-
axis, the resulting dash line coincides completely with the two arrows (for example see fig 5-1).
Propagation of some of the observed fatigue cracks through the microstructure (Group A, B)
is respectively presented in Figs. 7 and 8, and their corresponding PLM-s in Figs. 9a and 9b; the
microstructural barriers d1, d2 of PLM-s are shown in Figs. 7 and 8b.

10
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
MA
ED

Figure 6: Microstructural paths of part of crack 1, crack 2 and part of crack 3 (which together form the
PT

combined right part from the two-parts notch-specimen fracturing crack, the left part consists only of crack 4
- Figs. 2f, 5-1, 5-3) with their corresponding growth rates in da/dN -a terms, propagation in N-a terms, and
PLM applied to the data of da/dN a; here the part of crack 1 is to the A-bifurcation, and the part of crack 3
CE

is to the merging point B of cracks 2 and 3

Analyzing all fractured specimens two important remarks should be made concerning the
three groups of samples, beginning with the most interesting one involving in-situ observation:
AC

- Group C (Fig. 4-1, Fig. 4-2a, Fig. 4-3c, and especially Fig. 4-2b; Fig. 5-1, Fig. 5-2a and Fig.
5-3, and especially Fig. 5-2b; Fig. 6),
- Group B (Fig. 7a, and especially Fig. 7b as well as Fig. 7c, and especially Fig. 7d) and
- Group A (Fig. 8a, and especially Fig. 8b).
(1) All crack-growth-rate minimums from the Plot {da/dN a}, connected by resulting lines
(dash lines) with the corresponding crack lengths from the Plot {N a} show change in direction of
propagating crack presented in the Plot {N a} that is inserted to correspond to crack growth through
the microstructure. So when there are not direct observations of cracks propagation through the
microstructure, the combined plots Crack growth rate da/dN Crack length a & Number of cycles,
N Crack length, a, {da/dN a & N a}, can give information about presence of some
microstructural obstacles or elements. It is of special interest for the interval from crack initiation to
crack length coinciding with the first microstructural barrier d1, (Figs 4-2b, 4-3, Figs. 5-2b, 5-3, Fig.
6, Fig. 7, Fig. 8b) as in this interval the influence of microstructure on crack propagation is the most
pronounced one;

11
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
MA
ED
PT
CE
AC

Figure 7: Combined plots {da/dN a & N a} for Group B specimens in semi-log scale presentations for
the whole length of the major crack: at = 580 MPa (a) and for the first 500 m of its propagation (b); at
= 620 MPa (c) and for the first 500 m of its propagation (d); d1 and d2 are the microstructural barriers
from the PLM presented in Fig. 9a

the interval between d1 and d2 (Fig. 4-2a, Fig. 5-2d, Fig. 6, Fig. 7, Fig. 8b) is interesting too, as in this
interval cracks change their mode from shear to tensile. In this sense Fig. 4-2b includes only one deep
minimum which coincides with d1; all the other minimums marked by the previous five dash lines do
not influence significantly crack 2 propagation, as they belong to the growth of crack 1, from which
crack 2 branches at the fifth minimum (107.73 m). In the 5-minimums interval, crack 1 develops
from the edge notch and is not so sensitive to the influence of microstructure.

12
 ACCEPTED MANUSCRIPT

PT
RI
SC
a b
Figure 8: Microstructural crack propagation (a), and Combined plots {da/dN a & N a} in semi-log scale
NU
for Group A presentation for the whole length of the major crack at = 396 MPa (b); d1 and d2 are the
microstructural barriers from the PLM presented in Fig. 9b

1.E+00 1.E-01
MA

rotating bending fatigue

1.E-01
for cylindrical specimens 1.E-02
da/dN, 10 m/cycle
da/dN, 10 m/cycle

1.E-02
-6

-6

1.E-03
ED

1.E-03 620 MPa tension-tension fatigue

620 MPa
580 MPa 1.E-04 for cylindrical specimen
1.E-04 580 MPa
PT

1.E-05 1.E-05
1.E+01 1.E+02 1.E+03 1.E+04 1.E+01 1.E+02 1.E+03 1.E+04
a, 10-6 m a, 10-6 m

a b
CE

1.E+00

pure bending fatigue

AC

1.E-01
da/dN, 10 m/cycle

1.E-02
rotating bending fatigue
-6

1.E-03

1.E-04
tension-tension fatigue
1.E-05
1.E+01 1.E+02 1.E+03 1.E+04
a, 10-6 m
c
Figure 9: Description of fatigue data by PLM proposed in [6] for all Groups of specimens

13
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
MA
ED
PT

Figure 10: Graphical presentation of adequacy of PLM for Group A (a), Group B (b), Group C (c)

(2) When cycle intervals (Nm and Nm+1) between measuring of two successive crack lengths (am
CE

and am+1) are equal and (am+1-am) is significantly bigger in comparison with the other crack length
intervals, that means that there is probably a real change in crack propagation direction and it is near
to 90 degrees, (see the interval [100 200] m in Fig. 5-2b).
As it is already mentioned above, a mathematical Parabolic-linear model, PLM, proposed in [6] is
AC

used to describe fatigue data from all the presentations {da/dN -a} in log-log scale, Fig. 9.
The adequacy of models applied to the presentations {da/dN -a} in semi-log and log-log scales
is accordingly proved in [6-9] where the procedure is explained in detail; the graphical results can be
seen in Fig. 10. The symbols presenting cycles to failure, obtained by PLM, determine the straight
lines symmetrically surrounding the diagonals and showing the intervals of model precision.

CONCLUSIONS
Three groups of low-carbon-steel specimens with ferrite and pearlite microstructure are
subjected to tension-tension, rotating-bending and pure-bending fatigue. Two of the groups use
cylindrical specimens subjected to tension-tension and rotating-bending fatigue; surface crack
propagation is monitored by acetate-foil replication. The third group of flat specimens are preliminary
notched by FIB-technique alongside their longitudinal axis and at one of the edges, and then subjected
to pure-bending fatigue; interaction between surface cracks and the microstructure is directly (in situ)
observed by optical- and SEM-microscopy.

14
 ACCEPTED MANUSCRIPT
Microstructure. The most interesting data are obtained at pure-bending fatigue of a notched
specimen with major fracturing crack represented by a complicated combination of cracks and part
of cracks (crack 4, and part of crack 1, crack 2 and part of crack 3). These data show higher crack
growth rates (dominated by the interaction with ferrite and pearlite grain boundaries and interfaces,
ferrite grains, pearlite colonies and non-metal inclusions) and shorter fatigue life in comparison with
the specimens subjected to rotating-bending fatigue. Crack propagation rates at pure-bending fatigue
decrease in the vicinity of (i) interface between ferrite and pearlite bands when a crack propagates
into the ferrite or pearlite colony, (ii) ferrite-ferrite or pearlite-pearlite grain boundary, (iii) obstacles,
where a crack changes its propagation direction or bifurcates. In this case, the observed rows of
longitudinal nonmetal inclusions (MnS) increase crack growth rate serving as crack paths. At the
same time the registered tension-tension fatigue data represent the lowest crack growth rates, due to
much lesser loading than that applied at rotating-bending and pure-bending fatigue.

PT
Data presentations. The newly inserted Plots Number of cycles, N Crack length, a in semi-
logarithmic scale is a base for constructing new Combined Data Presentations {Crack growth rate,
da/dN - Crack length, a} with Parabolic-Linear Model describing the da/dN-a data Real

RI
Microstructure Image {Number of cycles, N Crack length, a} for more successful analyzing of
microstructural crack paths together with some fatigue characteristics; also for getting useful
information even without detailed microstructural observations.

SC
Models. The applied Parabolic-linear model can describe and predict adequately short crack
behaviour under conditions of tension-tension, rotating-bending and pure-bending fatigue. For the
first time a complete Parabolic-linear model is represented for describing the complicated major
NU
fracturing crack of the notch specimen at pure-bending fatigue. The inserted Parabolic-linear models
allow comparison between the fatigue characteristics of different metallic materials at different kinds
of loading.
MA

AKNOWLEGEMENTS
The authors thank the University of Chemical Technology and MetallurgySofia, Bulgaria for
its valuable support.
ED

REFERENCES
[1] S. Suresh, Fatigue of Materials, Cambridge Univ. Press, Cambridge, UK, 1998.
[2] N. Dowling, Mechanical Behavior of Materials. Engineering Methods for Deformation, Fracture,
PT

and Fatigue, 3th edition, Prentice Hall, 2006.

[3] D. Angelova, R. Yordanova, Bending Fatigue in a Low-Carbon Steel, in: V.V.Panasyuk (Ed.)
Fracture Mechanics of Materials and Structural Integrity, Karpenko Physico-Mechanical Institute of
CE

NASU, Lviv, 2009, pp. 309-314.

[4] A. Davidkov, On factors influencing fatigue in 09Mn2 steel, PhD Thesis, University of Chemical
Technology and Metallurgy Sofia, 2007.
AC

[5] A. Davidkov, R. Pippan, Studies on short fatigue crack propagation through a ferrite-pearlite
microstructure, CD ROM, 9-th International Fatigue Congress, Atlanta, Georgia, USA, 2006.
[6] R. Yordanova, Modeling of fracture process in a low-carbon 09Mn2 steel on the bases of short
fatigue crack growth experiments. Comparative analyses on the fatigue behaviour of other steels, PhD
Thesis, University of Chemical Technology and Metallurgy Sofia, 2003.
[7] D. Angelova, R. Yordanova, S. Yankova, Fatigue crack paths in a low-carbon steel. Modelling of
fatigue behaviour, Proceedings of The Fifth International Conference on Crack Paths CP 2015,
Ferrara, Italy, 16 - 18 September, 2015.
[8] D. Angelova, R. Yordanova, S. Yankova, Fatigue crack development in a low carbon steel.
Microstructure influence. Modelling, Procedia Structural Integrity, 2 (2016) 2726-2733.
[9] D. Angelova, A. Davidkov, in: Proceedings of Second International Conference Deformation,
Processing and Structure of Materials, Belgrade, Serbia and Montenegro (2005) 179-184.
[10] R. Pippan, K. Flechsig, F. O. Riemelmoser, Mater. Sci. Eng., A283 (2000) 225.

15
 ACCEPTED MANUSCRIPT
Highlights

Tension-tension, rotating-bending, pure-bending fatigue are investigated and compared

Short fatigue crack length, a, in notch specimens is observed in situ and measured
Presentations Number of cycles N/Crack growth rate da/dN a are plotted and modeled
Combined Presentations {da/dN-a and its model}Microstructure Image{N a}are studied
Combined Presentations show the relation fatigue parametersmicrostructure elements

PT
RI
SC
NU
MA
ED
PT
CE
AC

16