Theoretical and Applied Fracture Mechanics 69 (2014) 63–70

Contents lists available at ScienceDirect

Theoretical and Applied Fracture Mechanics

journal homepage: www.elsevier.com/locate/tafmec

Load sequence effects on fatigue crack growth in notched tubular

specimens subjected to axial and torsion loadings
Maksym Gladskyi a, Ali Fatemi b,⇑
a
Department of Manufacturing Engineering, Institute of Mechanical Engineering, National Technical University of Ukraine ‘‘Kyiv Polytechnic Institute’’,
37 Peremogy Avenue, Kyiv 03056, Ukraine
b
Mechanical, Industrial and Manufacturing Engineering Department, The University of Toledo, 2801 West Bancroft Street, Toledo, OH 43606, USA

a r t i c l e i n f o a b s t r a c t

Article history: Fatigue crack growth behavior of tubular specimens with a through thickness circular hole made of a
Available online 6 December 2013 carbon steel subjected to axial and torsional loads was investigate. Loading sequence effect on crack
growth rate was also studied by alternating between axial and torsion cycles in a loading block. Mode I
Keywords: crack growth was observed. Torsion fatigue crack growth lives were shorter and crack growth rates were
Fatigue crack growth higher than for axial loading. This is explained by a larger plastic zone size produced by a compressive tan-
Load sequence effect gential stress acting parallel to the crack growth path. In block loading with dominated torsion cycles crack
Notch effect
grown rate was slower in comparison with pure torsion, while in block loading with dominated axial
T-stress effect
cycles a faster crack growth rate occurred in comparison with pure axial loading. Effects of the stress state
on the plane of crack growth and of one pair of cracks on a second pair are considered to explain these
observations. Crack growth rates were correlated with stress intensity factor range with or without con-
sidering the T-stress effect. Short crack growth behavior near the threshold region is also discussed.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction high yield-strength steel were used with a thinned-down center

Many engineering structures are subjected to various combina-
tions of axial and torsion cyclic loading. In addition, many such
components often contain a variety of notches such as holes, fillets
and grooves. Therefore, the prediction of fatigue crack growth
behavior from notches due to stress concentrations is of signifi-
cance to a large number of engineering applications.
Mode I is typically found to control crack growth and fracture
mode of different materials under tension and torsion cycling.
However, mode I crack growth rate under cyclic torsion can be
higher than that under uniaxial loading [1]. Brown and Miller [2]
studied mode I fatigue crack growth under biaxial stress conditions
for AISI 316 austenitic stainless steel. Thin rectangular plates with
a through-thickness hole and two small opposite slots at the edge
of the hole were subjected to mode I cyclic loading. They found
that mode I crack growth accelerated by a negative T-stress and
the magnitude of this effect depended on the T-stress amplitude.
They proposed an equation to correlate experimental crack growth
rate data based on the Dugdale model [3] and the Paris equation
considering the size of plastic zone adjustment at the crack tip.
Smith and Pascoe [4] found similar results to Brown and Miller
observations. Flat cruciform-shaped specimens made from HY100
⇑ Corresponding author. Tel./fax: +1 419 530 8213.
E-mail address: afatemi@eng.utoledo.edu (A. Fatemi).
section and subjected to biaxial tension–compression loading.
They concluded that cyclic loads transverse to a normal or mode
I crack affects crack growth rate due to changes in crack tip plastic-
ity, where a tensile stress decreased crack rate and a compressive
stress has the opposite effect. Also, mode II crack growth rates
were found to be much greater than for the equivalent mode I
crack.
Similar observations were made by Shanyavskiy [5] for thin
sheet aluminum-based alloys with a central hole of 2 mm in diam-
eter subjected to biaxial loading. Compressive stress applied paral-
lel to the crack-growth plane caused a rotation effect resulting in
the formation of spherical or ellipsoidal particles and faster growth
rate. These particles are rolled between the opposite fracture-sur-
faces whose relative displacement is controlled by the KIII compo-
nent. The concluded that for constant amplitude cyclic loads,
including out-of-phase biaxial loads, crack growth rate during
the mode I crack opening stage may be successfully simulated by
equivalent stress intensity factor as a function of biaxial stress ratio
and phase angle.
Thomson and Sheppard [6] performed a series of experimental
studies on aluminum alloy 2024-T351 under uniaxial and torsion
cyclic loading for smooth and fillet notched cylindrical solid spec-
imens. They found torsional loading cases to result in higher slope
in Paris equation due to higher crack growth rate in comparison
with uniaxial loading. They discuss inapplicability of the LEFM

0167-8442/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.tafmec.2013.12.001
64 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70
Nomenclature
2a crack length including the hole diameter
A crack growth rate equation coefficient
da/dN crack-growth rate
da/dN1 biaxial crack-growth rate
F0 energy release correction function
n crack growth rate equation exponent
N0.2 fatigue life to 0.2 mm crack length
N0.5 fatigue life to 0.5 mm crack length
N0.55 fatigue life from 0.5 mm to 5 mm crack length
K biaxial stress ratio
assumption for torsional loading due to the presence of a biaxial
stress state and that applying the J-integral may be able to
account for the effect of biaxial loads on the crack tip plastic zone
size.
Tanaka et al. [7] carried out fatigue crack growth tests using
thin-walled tubular specimens with a circular hole made of
low-carbon steel. The loading included cyclic uniaxial tension–
compression, and cyclic torsion with or without superposed static
tension or with cyclic axial loading. In all cases cracks grew in
mode I. They found crack growth rates for torsion and combined
torsion with axial loading to be higher, as compared to uniaxial
loading. They attributed the acceleration of crack growth to larger
plasticity at the crack tip. The highest crack growth rate was
observed to be for the case of cyclic torsion with static tension
loading. The J-integral range determined from the relation
between load and displacement was shown to be an appropriate
parameter for crack growth analysis in the presence of significant
plasticity.
Endo et al. [8–10] studied crack growth behavior of 0.37%
carbon steel and a chrome-molybdenum steel under in-phase
and out-of-phase axial-torsion loading. Smooth solid round speci-
mens with a small drilled hole introduced into the surface were
used. The diameter of the hole was either 100 or 500 lm and
was equal to the hole depth. No significant effect of loading phase
was found on long crack growth rates. However, short crack behav-
ior was more sensitive to non-proportionality of the loading. This
was attributed to the abrasion of crack surfaces which reduces
the level of crack closure and, thereby, increases the net driving
force under out-of-phase loading. To consider the effect of biaxial
loading, they defined a net crack driving force parameter for fati-
gue crack propagation by modifying the expression for effective
stress intensity factor range DKeff. The effect of biaxial loading in
this parameter is reflected in the values of amplitudes ratio of
applied stress normal to the crack and applied stress parallel to
the crack. This parameter was found to correlate in-phase and
out-of-phase short crack growth rate data.
Nalla et al. [11] studied mixed-mode (mode I + II) fatigue crack
growth thresholds for large and short through-thickness cracks in a
Ti–6Al–4V turbine engine alloy. They observed fatigue thresholds
for through-thickness cracks which were large compared to micro-
structural dimensions to be strongly influenced by mode-mixity.
They concluded that although for large cracks fatigue cracking
under mixed-mode loading is predominately a mode I phenome-
non, mode-mixity influences thresholds by crack tip shielding
resulting from crack-surface interference.
Carpinteri et al. [12] examined the behavior of a part-through-
cracked notched shell with different notch configurations using 3D
FEA and concluded the effect of the stress concentrator on the
value of stress intensity factor to be significant. Mikheevskiy and
Glinka [13] used a fatigue crack growth model based on crack tip
r1 maximum principal stress
ra axial stress amplitude
ru ultimate tensile stress
rxx stress parallel to crack surface
ryy stress normal to crack surface
sa shear stress amplitude
DKI mode I stress intensity factor range
DKth threshold stress intensity factor range
region elastic–plastic stress/strain analysis where load-interaction
effects in a load spectrum was simulated by accounting for residual
compressive stresses produced by the reverse plastic deformation.
They applied their model to crack growth rate analysis of speci-
mens with a central through crack made of Al 7075-T6 alloy and
showed good correspondence between experimental and pre-
dicted results.
Susmel and Taylor [14] used sharply notched low carbon steel
specimens under in-phase mode I and II loading to study non-
propagating crack behavior. They observed the early stage of crack
growth to be mixed-mode, whereas the subsequent growth was
mainly mode I dominated. They also observed the non-propagating
crack length tended to increase as the mode II contribution to fati-
gue damage increased.
Makizaki et al. [15] investigated the effect of occasional mode II
loading on subsequent mode I fatigue crack growth behavior by
using thin-walled tubes made of 7075-T651 aluminum alloy. They
found the occasional mode II loading has two contradictory effects
on crack growth behavior, a retardation effect associated with the
plastic deformation near crack tip, and an acceleration effect
caused by mode II fatigue crack growth.
The overall goal of this study was to investigate fatigue crack
growth behavior of notched specimens made of a common carbon
structural steel subjected to axial and torsional loads. In addition,
loading sequence effect on crack growth rate was studied by alter-
nating between axial and torsion cycles in a loading block. In this
paper, first the experimental program is described. Then, experi-
mental results and observations are presented. This is followed
by crack growth rate correlations for different loading conditions.
Finally, conclusions based of the experimental observations and
crack growth rate analyses are presented.
2. Experimental program
The material used in the present study was a low-carbon steel
used for piping in nuclear power plants. The chemical composition
of the material by weight is 0.24% C, 0.25% Si, 0.45% Mn, 0.2% Cr,
and the balance Fe. The microstructure of the material was com-
posed of ferrite and pearlite structure with ferrite grain size of
about 20 lm. The yield strength is 365 MPa, the ultimate strength
is 506 MPa, elastic modulus is 185 GPa, and the cyclic yield stress is
311 MPa.
Tubular specimens with a through thickness circular hole at
the middle of gauge section were used. The gauge section of the
specimen had an outer diameter of 24.2 mm, an inner diameter
of 22 mm, and a wall thickness of 1.1 mm. The hole diameter
was 3.4 mm and the hole geometry resulted in stress concentra-
tion factors of 3.29 and 3.98 for axial and torsion loadings, respec-
tively [16].
 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70 65

Load-controlled fully-reversed fatigue tests were conducted on
a servo-hydraulic testing machine at room temperature with inde-
pendent controls of push–pull and torsion loads and with frequen-
cies in the range of 0.5–3 Hz. The loading conditions consisted of
constant amplitude axial and torsion loadings, as well as three
cases of sequential axial and torsion cycles in a load block. These
three loading cases included axial dominated case where 1000
axial cycles were followed by 100 torsion cycles, torsion domi-
nated case where 1000 (or 10,000) torsion cycles were followed
by 100 (or 1000) axial cycles, and the case were equal number of
axial and torsion cycles were applied. The stress amplitude for each
load sequence was such that the maximum principal stress
amplitude was nearly the same during the axial and torsion cycles.
Table 1 lists the applied stress amplitudes, the maximum prin-
cipal stress amplitudes, number applied of cycles to reach a crack
length of 0.2 mm (N0.2), a crack length of 0.5 mm, (N0.5), as well as
the number of cycles to grow the crack from 0.5 to 5 mm (N0.55)
for each test. The number of cycles reported in the table for block
loading cases are for the dominant load, except for the cases of
equal axial and torsion cycles in a block for witch both axial
and torsion cycles are counted. Crack initiation location around
the hole as well as crack orientation angles for each test is also
listed in the table.
The growth of fatigue cracks from the hole was continuously
monitored and recoded by using a digital video camera with 30X
optical zoom. Fatigue crack growth tests for most cases were
stopped when the total crack length (i.e. the lengths of the cracks
on the opposite sides of the hole plus the hole diameter) became
larger than about 10 mm. It should be noted that crack initiation
and life prediction aspects of the material and notched specimen
geometry investigated was the subject of another publication
[16]. Also, some crack growth experiments similar to those
reported in this paper were conducted on 7075-T6 and 2024-T3
aluminum alloys, the results and analysis of which are presented
in [17].
3. Experimental results and observations
Fig. 1(a) and (b) illustrate examples of observed crack orienta-
tions emanating from the hole in axial and torsion loadings,
respectively. Cracks initiated at the edge(s) of the hole perpendic-
ular to the specimen axis under axial loading and at ±45° locations
around the hole under torsion loading. These are maximum stress
locations (both shear stress and maximum principal stress) for
both loadings. As expected, mode I crack growth is observed with
cracks in uniaxial loading growing in the transverse direction and
in the ±45° direction to the transverse direction for torsion loading.
Crack initiation planes were observed to be on the maximum shear
plane(s) [16].
Fig. 1(c–e) shows crack growth paths for block loading tests
under similar applied axial and torsion stress amplitudes
contained in each block, resulting in nearly the same maximum
principal stress amplitude during axial and torsion cycles. As can
be seen from these figures, for block loading with dominant axial
cycles (A1000–T100), crack(s) grew in the same direction as for
pure axial loading. Similarly, for block loading with dominant
torsion cycles (A100–T1000), crack(s) grew in the same direction
as for pure torsion loading. This cracking behavior was identical
to that observed for similar testing of 2024-T3 and 7075-T6
aluminium alloys in [17].
For equal applied axial and torsion number of cycles (A1000–
T1000), crack(s) grew along ±45° path, corresponding to mode I
growth direction for pure torsion loading. The reason for this
behavior may be related to a higher crack driving force for torsion
cycles in comparison to axial cycles, resulting from a higher crack
tip plasticity due to the T-stress in torsion, as will be discussed
later. This is in spite of the fact that the applied axial and torsion
stress amplitudes result in nearly the same maximum principal
stress amplitude.
Crack growth histories for axial, torsion, and sequential block
loadings, all at nearly the same maximum applied principal stress
amplitude, are shown in Fig. 2. Crack growth lengths in this figure
are from an initial crack size of 0.5 mm to near fracture (overall
crack length of about 7 mm, or about 10 mm if the hole diameter
of 3.4 mm is included). As can be seen from the figure, torsion
fatigue crack growth lives are shorter and crack growth rates are
higher than for axial loading. The same behavior was also observed
in axial and torsion fatigue tests of 2024-T3 and 7075-T6 alumin-
ium alloys in [17]. For block loading cases with dominant or equal
torsion cycles to axial cycles, crack growth rates are lower than
pure torsion loading. This is surprising, as it implies addition of
axial cycles to the torsion load history has a beneficial effect on
crack growth behavior.
It should be noted that the choice of an initial crack size of
0.5 mm does not significantly alter the aforementioned results or
crack growth curves in Fig. 2. For example, if an initial crack size
of 0.2 mm is used, instead of 0.5 mm, the difference in fatigue lives
between these two crack sizes is less than 10%. Generally a crack is
considered to be out of the notch affected zone when the crack
length becomes longer than about 10–20% of the notch radius. This
corresponds to a crack length of about 0.17 to 0.34 mm for the
notch radius used (1.7 mm). However, since crack nucleation is

Table 1
Stress levels, fatigue lives, and crack growth rate properties in constant amplitude and block loading fatigue tests.

Loading type ra (MPa) sa (MPa) Dr1/2 (MPa) A (m/cycle) n N0.2 N0.5 N0.55 Cracka initiation Crackb initiation
(cycles) (cycles) (cycles) location (°) orientation (°)
Axial 144 – 144 1  1010 2.8 20,900 24,040 26,700 0 2
Axial 106 – 106 4  1010 2.7 140,500 202,400 107,400 0 0
Torsion – 149 149 6  1012 6.8 5,407 5697 1275 42 45
Torsion – 149 149 8  1012 5.8 5,910 6820 2010 45 45
Torsion – 121 121 9  1011 6.1 34,700 37,500 26,850 145 45
Torsion – 87 87 5  1011 5.7 120,000 145,100 276,900 45 45
Torsion – 87 87 – – 276,000 298,260 325,040 40 48
Blocks (A1000–T100) 144 149 144–149 5  1011 4.0 23,610 23,970 6800 0 0
Blocks (A1000–T100) 144 149 144–149 9  1011 3.9 19,650 20,640 17,510 3 5
Blocks (A1000–T1000) 144 149 144–149 4  1012 5.1 14,800 15,010 2590 45 47
Blocks (A1000–T1000 144 149 144–149 3  1012 5.9 18,200 19,220 4980 138 48
Blocks A100–T1000) 144 149 144–149 7  1012 4.1 5220 5510 4690 136 45
Blocks (A100–T1000) 144 149 144–149 4  1011 4.5 6570 6715 2230 49 40
Blocks (A1000–T10000) 98 102 98–102 5  1011 4.4 112,300 119,200 61,100 38 49
a
Location of crack initiation around hole circumference with 0° and 180° being at the hole edges.
b
Orientation of surface crack initiation with ±90° being along the specimen axis.
66 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70

(a) (b)

σ σ
σ

σ σ

A1000-T100 A1000-T1000 A100-T1000

(c) (d) (e)
Fig. 1. Fatigue crack growth path for notched specimens under (a) uniaxial, (b) torsion, and block loading with (c) 1000 axial and 100 torsion cycles, (d) 1000 axial and 1000
torsion cycles and (e) 100 axial and 1000 torsion cycles.

Fig. 2. a–N curves for different loading condition at similar values of maximum principal stress amplitude of Dr1/2 = 144 MPa.

on maximum shear stress plane while crack growth is on maxi-
mum principal plane, a transition between crack planes takes
place. This transition was found to be complete at a crack length
of about 0.5 mm. A schematic representation of the transition
between maximum shear surface plane for crack formation and
through-the-thickness maximum shear plane for small crack
growth during crack initiation phase is shown in Fig. 3 for axial
loading.
As can be seen from Fig. 2, more than 20 times shorter crack
growth life is observed for torsion loading in comparison with uni-
axial loading. Such difference may be explained by a compressive
tangential stress which acts parallel to the crack growth path.
The stress states for mode I growth direction for axial and torsion
loadings are shown in Fig. 4, where the compressive tangential
stress is observed in torsion. Smith and Pascoe [4] attributed the
increase in torsion crack growth rate to the effect of this stress
component parallel to the crack. They argue that, although a paral-
lel stress component does not change the magnitude of the crack
tip stress singularity, the increased plastic zone size produced by
a compressive parallel stress increases the crack driving force,
thereby increasing the growth rate. Similar observations and con-
clusions were reported in [2,5,6].
Another factor to consider is the effect of one pair of cracks on
another, if more than one pair is present, such as in cyclic torsion.
As mentioned previously, two pairs of opposite cracks grew from
the hole under torsion loading, see Fig. 1(b). In order to estimate
the influence of the second pair of cracks on growth of the first pair,
FEA was used. Analysis of cracked specimens with one and two
pairs of cracks at ±45° orientations with 0.25, 0.5, 1.0 and
2.0 mm lengths was carried out using ABAQUS software [18]. The
maximum principal stress ratio, defined as the maximum principal
stress for two pairs of crack divided by the maximum principal
 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70 67

Fig. 3. Schematic of through thickness crack initiation plane orientation under axial
loading.

stress for one pair of cracks, versus crack length is presented in

Fig. 5. As can be seen from this figure, existence of the second pair
of cracks increases the crack tip stress for the first pair of cracks.
This effect becomes more significant with a longer length of the
second crack pair. This can further explain a higher crack growth
rate in torsion, as compared to axial loading at the same stress
intensity factor range.
Crack growth curves for sequential block loading conditions
shown in Fig. 2 indicate longer observed crack growth lives for
A100–T1000 loading block, in comparison with pure torsion load-
ing. This means axial sequential loading retards growth in torsion.
However, an opposite effect is observed for loading blocks A1000–
T100, where a faster crack growth rate occurs in comparison with
pure axial loading. This means torsion sequential loading acceler-
ates growth under axial loading.
There are several aspects to consider in explaining such obser-
vations. The stress state on the plane of crack growth paths pro-
vides a reasonable explanation for the observed behavior. Fig. 6
shows stress states for the axial and torsion loadings on the 0°
and 45° planes. For axial dominated loading, where crack growth
is on the horizontal plane (0° plane), the shear stress in torsion
on 0° plane does not affect fatigue crack growth rate in transverse
direction. However, the normal stress which is acting on 45°
through-the-thickness plane opens the shear crack initiated during
each block of axial cycles. In [16] it was shown that through
thickness failure surface corresponds to 45° plane under uniaxial
Fig. 5. The ratio of maximum principal stress for one pair and two pairs of cracks as
a function of crack length under torsion loading.
loading. The magnitude of the normal stress from torsion cycles
on 45° plane is equal to the opening stress for uniaxial loading. This
normal stress accelerates short crack growth rate in the thickness
direction initiated under uniaxial loading.
In the case of loading block 100A–1000T with dominated tor-
sion cycles, the torsion controlled crack growth rate on 45° plane
is slower in comparison with pure torsion. Every sequence of axial
loading produces a tension T-stress at the notch tip which
decreases the plastic zone size, thereby, decreasing the overall
fatigue crack growth life. Somewhat between the two mentioned
cases can be found for loading blocks A1000–T1000 with equal
number of axial and torsion cycles.
4. Crack growth rate correlations
Fatigue crack growth rate correlations were made based on
mode I stress intensity factor range:
da
¼ AðDK I Þn ð1Þ
dN

Axial Torsion 144 MPa

σ=144 MPa τ=144 MPa 144 MPa

144 MPa Mode I

Mode I

(a) (b)
Fig. 4. Normal stresses at the maximum stress location for (a) axial and, (b) torsion loading.
68 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70

σ 144 MPa σ 144 MPa

72 MPa

Axial cycles Torsion cycles Torsion cycles Axial cycles

on 0° plane on 0° plane on 45° plane on 45° plane

144 MPa 144 MPa 144 MPa 72 MPa 72 MPa

144 MPa
72 MPa

(a) (b)
Fig. 6. Normal and shear stresses under torsion and axial loading with principal stress amplitude Dr1/2 = 144 MPa on (a) 45° plane and, (b) 0° plane.

Sander’s energy release rate integral for a circumferentially
cracked cylindrical shell loaded in both tension and torsion [19]
was used to derive the mode I stress intensity factor solution based
on AFGROW procedure as [20]:
pffiffiffiffiffiffiffiffi
K I ¼ F 0 r1 pa
where F0 is energy release correction function, as given in the
Appendix A.
Crack growth rate correlations plotted against the stress inten-
sity range, DKI, are shown in Fig. 7(a). As can be seen from this fig-
ure, a factor of about 2.5 times higher crack growth rate is observed
for torsion or torsion dominated loading cases in block loadings, as
compared with axial loading. Similar observations were made for a
low-carbon steel in [8], where the authors reported about 2 times
higher crack growth rates for torsion and combined torsion with
axial cyclic loading in comparison with uniaxial loading. As dis-
cussed earlier, this can be attributed to a higher crack tip plasticity
due to the T-stress in torsion, as well as a higher local stress due to
ð2Þ existence of two pairs of cracks in torsion, as compared to typically
one pair under axial loading.
In order to consider the faster crack growth rate under torsion
loading, Brown and Miller [2] proposed an equation to correlate
experimental data by considering adjustment of the plastic zone
size at the crack tip. This adjustment is based on Dugdale’s model
extended to biaxial loading of a mode I crack by adding the T-stress
term, resulting in:

Fig. 7. Crack growth rate versus stress intensity factor range (a) without considering biaxiality stress ratio and, (b) with considering biaxiality stress ratio.
 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70 69

 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi 
5. Conclusions
sec p= ð4ru =DrÞ2  3ðK  1Þ2 þ ðK  1Þ  1
da=dN
¼
da=dN1 secðpDr=4ru Þ  1 (1) Mode I crack growth was observed for both axial and torsion
ð3Þ loading conditions. For block loading with dominant axial
cycles crack(s) grew in the same direction as for pure axial
loading, while for block loading with dominant torsion
where da/dN1 is equibiaxial value of crack growth rate, K = rxx/ryy
cycles crack(s) grew in the same direction as for pure torsion
is a measure of biaxiality with ryy being normal to the crack plane,
loading.
and ru is ultimate tensile strength.
(2) Much shorter crack growth life (by a factor of 20) was
Fig. 7(b) shows results of crack growth rate correlations
observed for torsion loading in comparison to axial loading,
employing Eq. (3). Better correlations of crack growth rates are ob-
explained by a compressive tangential stress which acts par-
served for axial and torsion loadings. However, the slope A in Paris
allel to the crack growth path. Existence of a second pair of
equation for torsion loading is still somewhat higher than for axial
cracks increases the crack tip stress for the first pair of
loading. This may be due to the existence of two pairs of cracks in
cracks. This can further explain a higher crack growth rate
torsion, as opposed to one pair in axial loading. As discussed ear-
in torsion, as compared to axial loading at the same stress
lier, the second pair of cracks in torsion increase the stresses on
intensity factor range.
the first, therefore, increased crack driving force. This increase is
(3) For the loading block with dominated axial cycles a higher
not accounted for by the stress intensity factor used for data corre-
crack growth rate occurred in comparison to pure axial load-
lations in Fig. 7(a), or by Brown-Miller’s crack growth ratio (Eq. 3)
ing, attributed to the effect of stress state on the crack
accounting for the T-stress as shown in Fig. 7(b). It should also be
growth plane. For the loading block with dominated torsion
noted that a higher scatter is observed in Fig. 7(a) and (b) for crack
cycles, crack growth rate was lower in comparison to pure
growth rates in block loadings tests. This is due to the load se-
torsion, attributed to the tensile T-stress at the notch tip
quence effects discussed previously.
during subsequent axial cycles.
A clear threshold region at crack growth rates below 5  109 m/
(4) Short crack growth behavior with reduced crack growth
cycle may be observed in Fig. 7, corresponding to very slow growing
rates with increasing crack length (retardation) was
or non-propagating cracks. The threshold stress intensity factor va-
p observed near the threshold region, explained by a change
lue is about DKth = 8 MPa m. Fig. 8 shows crack growth data for
in fracture mechanism and crack orientation as the initiated
tests with data in the threshold region in an expanded scale. Short
crack at the notch root starts to growth through the notch
crack growth behavior with reduced crack growth rates with
plastic zone and out of the notch affected zone.
increasing the short crack length (retardation) is observed, before in-
creased crack growth rates with further increase in crack length
(acceleration). This transitional change in growth rate may be
Acknowledgements
explained by the change in fracture mechanism and crack orienta-
tion, as the initiated crack at the notch root starts to growth through
This research was conducted under the Fulbright Scholar
the notch plastic zone (see Fig. 3). Once the crack grows out of the
Program with financing of the U.S. Department of State and
notch affected zone, a switching of the critical plane occurs, which
technical support of National Technical University of Ukraine ‘‘Kyiv
slows down crack growth rate. A similar observation regarding a
Polytechnic Institute’’.
transition between tensile and shear mode crack growth resulting
from an overload was made in [5] for Al-alloy sheet and plate
Appendix A. Stress intensity factor calculation
specimens subjected to biaxial loading.

The solution for energy release rate of a circumferential

through-crack in a cylindrical shell was derived by Sanders [19].
In integral form, the energy release rate may be written as:
Z p
@
I¼ uc dh
@a p

where uc is the dimensionless crack displacement in the axial

direction. With curve fits to both long and short crack solutions,
Forman [20] derived the following expressions:
" #
pffiffiffi pb2 a2
2
I0 ¼ 8ðf  1Þ þ
b k

where

Fig. 8. Relation between crack growth rate and stress intensity factor range near
threshold region.
70 M. Gladskyi, A. Fatemi / Theoretical and Applied Fracture Mechanics 69 (2014) 63–70

[2] M.W. Brown, K.J. Miller, Mode I fatigue crack growth under biaxial stress at
hð1  a cot aÞ room and elevated temperature, Multiaxial Fatigue, ASTM STP 853, 1985, pp.
f ¼1þ
2a 135–152.
[3] D.S. Dugdale, Yielding of steel sheets containing slits, J. Mech. Phys. Solids 8
pffiffiffi (1960) 100–104.
2 [4] E.W. Smith, K.J. Pascoe, Fatigue crack initiation and growth in a high-strength
h¼ n h i pffiffiffi o
ðpaÞ ductile steel subject to in-plane biaxial loading, Multiaxial Fatigue, ASTM STP
cot pffiffi2 þ 2 cot a 853, 1985, pp. 111–134.
[5] A. Shanyavskiy, Fatigue cracking simulation based on crack closure effects in
Al-based sheet materials subjected to biaxial cyclic loads, Eng. Fract. Mech. 78
a (2011) 1516–1528.
b¼
2k [6] K.D. Thomson, S.D. Sheppard, Fatigue crack growth in notched and plain
shafts subjected to torsion and axial loading, Eng. Fract. Mech. 43 (1) (1992)
rffiffiffi 55–71.
t
14 [7] K. Tanaka, H. Takahash, Y. Akiniwa, Fatigue crack propagation from a hole in
k¼ 12ð1  m2 Þ
R tubular specimens under axial and torsional loading, Int. J. Fatigue 28 (2006)
324–334.
p [8] M. Endo, A.J. McEvily, Fatigue crack growth from small defects under out-of-
2 3
b¼1þ b  0:0293 b for b 6 1 phase combined loading, Eng. Fract. Mech. 78 (2011) 1529–1541.
16 [9] M. Endo, I. Ishimoto, The fatigue strength of steels containing small holes
under out-of-phase combined loading, Int. J. Fatigue 28 (2006) 592–597.
pffiffiffiffiffiffi!0:5  0:885 [10] M. Endo, I. Ishimoto, Effects of phase difference and mean stress on the fatigue
8b 0:179 strength of small-hole-containing specimens subjected to combined load, J.
b¼ þ for b > 1 Solid Mech. Mater. Eng. 1 (3) (2007) 343–354.
p b
[11] R.K. Nalla, J.P. Campbell, R.O. Ritchie, Mixed-mode, high-cycle fatigue-crack
growth thresholds in Ti–6Al–4V: role of small cracks, Int. J. Fatigue 24 (10)
a (2002) 1047–1062.
a¼ [12] A. Carpinteri, R. Brighenti, S. Vantadori, Notched shells with surface cracks
R under complex loading, Int. J. Mech. Sci. 48 (2006) 638–649.
The dimensionless stress intensity factor for individual cases of [13] S. Mikheevskiy, G. Glinka, Elastic–plastic fatigue crack growth analysis under
variable amplitude loading spectra, Int. J. Fatigue 31 (11–12) (2009) 1828–
tension and torsion can be determined from: 1836.
[14] L. Susmel, D. Taylor, Non-propagating cracks and high-cycle fatigue failures in
F 0 ¼ ðI0 =2paÞ1=2 sharply notched specimens under in-phase Mode I and II loading, Eng. Fail.
Anal. 14 (2007) 861–876.
The resulting formula for stress intensity factor can then be [15] M. Makizaki, H. Matsunaga, K. Yanase, M. Endo, Effect of occasional shear
written as: loading on fatigue crack growth in 7075 aluminum alloy, in: S. Zhu, B. Ni, D.Y.
pffiffiffiffiffiffiffi Ju (Eds.), Advanced Materials Science and Technology (IFAMST-8) Book Series:
KI ¼ F0 r1 p a Materials Science Forum, vol. 750, 2013, pp. 264–267.
[16] M. Gladskyi, A. Fatemi, Notched fatigue behavior including load sequence
where effects under axial and torsional loadings, Int. J. Fatigue 55 (2013) 43–53.
[17] A. Fatemi, N. Gates, D.F. Socie, N. Phan, Fatigue crack growth behavior of
a ¼ aR tubular aluminum specimens with a transverse hole under axial and torsion
loadings, Eng. Fract. Mech., in press.
[18] ABAQUS v6.11, SIMULIA. RI, USA, 2008.
[19] J.L. Sanders Jr., Circumferential through-cracks in cylindrical shells under
References tension, ASME J. Appl. Mech. 49 (1982) 103–107.
[20] R.G. Forman, Stress intensity factors for circumferential through cracks in
[1] A.T. Yokobori Jr., T. Yokobori, K. Sato, K. Syoji, Fatigue crack propagation under hollow cylinders subjected to combined tension and bending loads, Eng. Fract.
mixed modes I and II, Fatigue Fract. Eng. Mater. Struct. 8 (1985) 315–325. Mech. 21 (1985) 563–571.