Engineering Fracture Mechanics 77 (2010) 452–469

Contents lists available at ScienceDirect

Engineering Fracture Mechanics

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

The Theory of Critical Distances to estimate the static strength

of notched samples of Al6082 loaded in combined tension and torsion.
Part I: Material cracking behaviour
Luca Susmel a,b,*, David Taylor b
a
Department of Engineering, University of Ferrara, Via Saragat 1, 44100 Ferrara, Italy
b
Department of Mechanical Engineering, Trinity College, Dublin 2, Ireland

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

Article history: The present paper is the first of a two-part series reporting an experimental and theoret-
Received 8 January 2009 ical study of the fracture of circumferentially notched samples of a commercial alumin-
Received in revised form 4 November 2009 ium alloy, i.e. Al6082, subjected to tension, torsion and mixed tension/torsion loading.
Accepted 28 November 2009
The overall aim of the work was to investigate the use of a particular method of failure
Available online 6 December 2009
prediction, known as the Theory of Critical Distances. This first part reports the experi-
mental data – load–deflection curves and observed material failure modes – and
Keywords:
discusses the consequences of these findings for the development of the theory, which
Aluminium
Mixed Modes I and III static loading
is covered in the second part. It was observed that relatively blunt notches loaded in
Material cracking behaviour tension failed by a conventional ductile fracture mode similar to plain (unnotched) spec-
Notches imens. However, in tensile specimens containing sharp notches, failure occurred via the
initiation, stable propagation and, finally, unstable propagation, of circumferential ring
cracks. Under torsional loading, and independent of the notch root radius, static failures
of the tested samples always occurred by the formation and stable propagation of ring
cracks. Under mixed-mode loading there was a gradual transition between the ductile
and brittle modes and between stable and unstable cracking. For all types of loading, it
was observed that crack initiation always coincided with peak loading conditions, and
that cracks invariably grew on the plane perpendicular to the specimen’s longitudinal
axis.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Due to their well-known weakening effect, stress concentration phenomena are always a matter of concern to structural
engineers engaged in performing the static assessment of real components. According to this fact, since the pioneering work
done by Galileo Galilei by considering plain materials, researchers investigating such a problem have made a big effort to
formalise sound methodologies suitable for estimating static strength when mechanical assemblies experience stress
concentration phenomena.
Examination of the state of the art shows that, since about the middle of the last century, a lot of work has been done
in this area of the structural integrity discipline to take full advantage of emerging Linear Elastic Fracture Mechanics
(LEFM). As to its accuracy and reliability, even if the use of this powerful theory was seen to be successful, strictly speaking,

* Corresponding author. Address: Department of Engineering, University of Ferrara, Via Saragat 1, 44100 Ferrara, Italy.
E-mail address: ssl@unife.it (L. Susmel).

0013-7944/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfracmech.2009.11.015
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 453

Nomenclature

dg gross diameter
dn net diameter
H degree of triaxiality
rn notch root radius
E Young’s modulus
Fu ultimate tensile force
Kt net stress concentration factor under tensile loading
Ktt net stress concentration factor under torsional loading
Mu ultimate torque
a notch opening angle
bi ri/rx (i = y, z)
rnom tensile nominal net stress
rUTS ultimate tensile stress
rY yield stress
rx, ry, rz normal stresses
snom torsional nominal net stress
sxy, sxz, syz shear stresses

it applies only to those situations involving sharp notches, that is, to those stress concentrators which behave like long
cracks.
On the contrary, as far as conventional geometrical features are concerned, the standard methodology suggests follow-
ing different strategies according to the type of material to be assessed. In more detail, when brittle materials are involved,
static strength is usually estimated by using the maximum principal stress criterion together with linear-elastic peak
stresses.
The problem of performing the static assessment of notched ductile materials is instead much more tricky.
Nominal net stresses are commonly used, rather than the estimated elastic peak stresses at the notch root, due to
the fact that the stress fields in the vicinity of stress raiser apices can redistribute themselves as soon as the material
begins to deform plastically. Further, when multiaxial loadings are involved, care must be taken to choose the
proper equivalent stress (i.e. Von Mises’ equivalent stress, Tresca’s equivalent stress, etc.) according to the actual
fracture behaviour of the material to be assessed. Unfortunately, this approach has several difficulties, including the
definition of a reference section to be used to calculate nominal stresses for components having complex geom-
etry.
The above problems could partially be overcome by taking full advantage of either elasto-plastic Finite Element (FE) mod-
els or classical rules suitable for estimating elasto-plastic notch root stresses and strains (as, for instance, the well-known
rule devised by Neuber [1]), even though, as far as the authors are aware, the scientific community has not yet agreed a
unique method suitable for predicting static failure of notched ductile materials by directly post-processing the outputs from
elasto-plastic analysis.
In this complex scenario, and according to the authors’ understanding of the static assessment issue, any reliable engi-
neering method suitable for estimating static strength of notched ductile materials should be capable of efficiently and
simultaneously modelling both the damaging effect of the stress/strain field acting on the process zone, including the effect
of multiaxial stresses, and also the role of stress concentration phenomena. At the same time, due to the needs of industrial
reality, the calibration of such a method should be based on pieces of experimental information which can easily be gener-
ated by using standard testing equipment. Finally, to reduce the time of the design process itself, the stress analyses needed
to estimate the necessary engineering quantities should, ideally, be done by directly post-processing simple linear-elastic FE
models.
The present investigation ambitiously aims to devise and validate an engineering method which meets all the require-
ments mentioned above. The specific hypothesis to be investigated is that the Theory of Critical Distances, which has pre-
viously been applied to other problems in failure prediction [2], can safely be used also to perform the static assessment
of notched ductile materials when multiaxial loadings are involved. In other words, aim of the present investigation is to
check whether static strength of notched ductile materials under multiaxial loading can be predicted by directly post-pro-
cessing the entire linear-elastic stress field acting on the process zone without modelling the cracks explicitly. It is evident
that, in order to soundly formalise such an approach, the initial logical step is the systematic investigation of the cracking
behaviour of notched ductile materials subjected to multiaxial quasi-static loading paths. Unfortunately, it is also obvious
that the experimental outcomes obtained by studying in depth the response to multiaxial static loading of one single
notched aluminium alloy cannot directly be extended to all the ductile materials which are commonly used to make real
mechanical components. Bearing in mind these limitations, the present investigation then represents nothing but a first at-
tempt to test the above hypothesis.
454 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

Tension - Plain
400
350
300
Stress [MPa]
250
200
150
σUT S=367 MPa
100 σY=347 MPa
50 E=69090 MPa
0
0 0.1 0.2 0.3 0.4 0.5
Strain [mm/mm]
Fig. 1. Plain stress–strain curve under tensile loading.

rn=4.0 mm (Kt=1.33; Ktt=1.12) Plain specimen

φ10 φ10 φ10

φ6
φ6

80 R4 80 60° 80 5

φ6 R50

rn=0.44 mm (Kt=2.94; Ktt=1.71)

rn=0.5 mm (Kt=2.76; Ktt=1.64)
rn=1.25 mm (Kt=1.92; Ktt=1.32)
Fig. 2. Geometries of the tested samples (dimensions in millimetres).

2. Experimental details

The material investigated in the present study was commercial aluminium alloy Al6082. The material was supplied in
bars having diameter, dg, equal to 10 mm. The plain samples used to determine the static properties of such an aluminium
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 455

alloy were machined to obtain a gauge length of 5 mm with a diameter equal to 6 mm. The material was found to have an
ultimate tensile stress, rUTS, equal to 367 MPa, a yield stress, rY, equal to 347 MPa and a Young’s modulus, E, equal to
69,090 MPa. Fig. 1a shows one of the stress–strain curves generated from plain (i.e. unnotched) specimens, plotted in terms
of engineering quantities.
The tested V-notched cylindrical samples (Fig. 2) had gross diameter, dg, equal to 10 mm and net diameter, dn, ranging be-
tween 6.1 mm and 6.2 mm. The notch opening angle, a, was equal to 60° and four different values of the notch root radius, rn,
were investigated, i.e. 0.44 mm (tensile stress concentration factor Kt = 2.94; torsional stress concentration factor Ktt = 1.71);
0.50 mm (Kt = 2.76; Ktt = 1.64); 1.25 mm (Kt = 1.92; Ktt = 1.32) and; 4.00 mm (Kt = 1.33; Ktt = 1.12). Static tests, conducted
according to standard methods [3–5] were run by using a conventional tensile/torsional testing machine, investigating the
following five different values of the ratio between nominal tensile and nominal torsional net stress: rnom/snom = 0 (torsion),

Notch tip
y circumference
O

r
x

σnom

rn=0.44 mm rn=0.5 mm
1.5 1.5
H Hy H Hy
1 1
Hz Hz
0.5 0.5
Linear-elastic Linear-elastic
stress fields stress fields
0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
r [mm] r [mm]
rn=1.25 mm rn=4.0 mm
1.5 1.5
H H
1 Hy 1

Hy
0.5 Hz 0.5
Linear-elastic Linear-elastic
Hz
stress fields stress fields
0 0
0.0 1.0 2.0 3.0 0.0 1.0 2.0 3.0
r [mm] r [mm]
Fig. 3. Degree of triaxiality of the linear-elastic stress fields in the vicinity of the investigated notches.
456 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

(a) 18 Tension
rn =0.5mm
16
14 rn =0.44mm rn =1.25mm
Load [kN] 12 rn =4mm
10 Plain
8
6
4
2
0
0 0.5 1 1.5 2 2.5
Extension [mm]

(b) Torsion
16
14
12
rn =4mm
Torque [Nm]

10
8
6
rn =0.44mm
4 rn =1.25mm
2 rn =0.5mm
0
0 10 20 30 40 50 60
Twist angle [°]
Fig. 4. Load vs. extension and torque vs. twist angle curves generated by testing the notched samples under tensile loading (a) as well as under torsional
loading (b).

rnom/snom = 0.23, rnom/snom = 0.55, rnom/snom = 1 and rnom/snom = 1 (tension). Both uniaxial and biaxial tests were carried
out under an axial displacement rate equal to 0.002 mm/s, by accordingly setting the twist angle rate to obtain the wanted
ratio between the two nominal stress components. The tests under pure torsion instead were run under a twist angle ratio
equal to 1°/s.
Examination of the state of the art shows that the most efficient approaches suitable for predicting static strength
of notched ductile materials make full use of the degree of triaxiality of the stress fields in the vicinity of crack
initiation sites [6–8]. According to that, and to better show the features of the investigated stress concentrators, the
diagrams in Fig. 3 plot the degree of triaxility, H, against the distance, r, from the notch tip. In more detail, the
above charts were built by post-processing the linear-elastic stress fields along the notch bisector (calculated,
under uniaxial fatigue loading, through refined FE models) and by defining the degree of triaxility as follows [9]
(see Fig. 3):

2bi ðrÞ þ 1
Hi ðrÞ ¼
3½1  bi ðrÞ
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 457

Twist angle [°]

(a) 0 10 20 30
2.5 22
Fu =2.11 kN 20
2 18
Tensile curve 16
M u =13.52 Nm

Torque [Nm]
14
Load [kN]

1.5
Torsional 12
curve 10
1
8
6
0.5 rn =4 mm
4
σ nom /τ nom =0.23 2
0 0
0 0.02 0.04 0.06 0.08 0.1
Extension [mm]

(b) 4 20
rn =4 mm 18
3.5
σ nom /τ nom =0.23
16
3 M u =13.52 Nm
14

Torque [Nm]
2.5 12
Load [kN]

Fu =2.11 kN
2 10

1.5 8
Torsional
curve 6
1
Tensile curve 4
0.5 2
0 0
0 5 10 15 20 25 30
Time [sec]
Fig. 5. Load vs. extension and torque vs. twist angle curves generated by testing the notched samples having rn = 4 mm under combined tension and torsion
(rnom/snom = 0.23) (a) and corresponding load vs. time and torque vs. time curves (b).

where

ri ðrÞ
bi ðrÞ ¼ ðfor i ¼ y; zÞ
rx ðrÞ
The diagrams of Fig. 3 fully confirm that, as expected, the degree of multiaxiality of the linear-elastic stress fields in-
creases with increasing of the sharpness of the tested notch.
For tests conducted under pure tensile and pure torsional loading, the failure force, Fu, and failure torque, Mu, were de-
fined as the maximum value recorded during each test. Fig. 4 shows the profile of some load vs. extension and torque vs.
twist angle curves generated by testing the investigated notched geometries: the above charts make it evident that, under
simple loading paths, it was straightforward to determine Fu and Mu according to the above definition.
458 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

Twist angle [°]

(a) 0 5 10 15
2.5 20
Tensile curve 18
Fu =1.99 kN
2 16
M u =13.39 Nm 14

Torque [Nm]
Load [kN]

1.5 12
10
Torsional
1 curve 8
6
0.5 rn =0.44 mm 4
σ nom /τ nom =0.23 2
0 0
0 0.01 0.02 0.03 0.04 0.05 0.06
Extension [mm]
Twist angle [°]
(b) 0 1 2 3 4 5 6 7 8 9
9 16
8 Fu =7.65 kN 14
7 12
6

Torque [Nm]
10
Load [kN]

Torsional
5
curve M u =11.86 Nm 8
4
6
3
Tensile curve
2 4
rn =0.44 mm
1 σ nom /τ nom =1 2
0 0
0 0.05 0.1 0.15 0.2
Extension [mm]
Fig. 6. Load vs. extension and torque vs. twist angle curves generated by testing the notched samples having rn = 0.44 mm under combined tension and
torsion by setting the rnom to snom ratio equal to 0.23 (a) as well as to unity (b).

On the contrary, much more tricky was the determination of the ultimate tensile loading and the ultimate torque un-
der mixed tension/torsion loading. As an example, Fig. 5a shows the loading vs. extension and the torque vs. twist angle
curves generated by testing, under a rnom to snom ratio equal to 0.23, a notched sample having notch root radius equal to
4 mm. In the above chart the ultimate tensile force, Fu, and the ultimate torque, Mu, are not aligned along a unique ver-
tical line. This is due to the fact that, to keep the ratio between the two nominal stress components constant during the
test (see Fig. 5b), the axial strain rate had to be different from the corresponding torsional strain rate. Moreover, for any
controlled channel, the deformation rate had to vary to keep the rnom to snom ratio unchanged when the material began
to deviate from the linear-elastic behaviour. The direct analysis of all the gathered channels showed that, under combined
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 459

Table 1
Summary of the generated results.

rn (mm) dg (mm) dn (mm) a (°) rnom/snom Fu (kN) Mu (N m) Kt Ktt

0.44 10.0 6.2 60 0 – 11.92 2.94 1.71
0 – 12.43
0 – 12.95
0.23 2.40 16.19
0.23 1.86 12.52
0.23 1.99 13.39
0.55 4.50 12.69
0.55 4.25 11.98
0.55 6.10 17.19
1 7.47 11.58
1 7.94 12.31
1 7.65 11.86
1 15.75 –
1 16.32 –
1 16.59 –
0.50 10.0 6.1 60 0 – 10.89 2.76 1.64
0 – 11.85
0 – 11.37
0.23 1.87 12.61
0.23 1.90 12.82
0.23 1.92 12.92
0.55 4.40 12.39
0.55 4.36 12.30
0.55 4.27 12.04
1 7.38 11.44
1 7.56 11.72
1 7.77 12.05
1 16.85 –
1 15.69 –
1 15.82 –
1.25 10.0 6.2 60 0 – 13.01 1.92 1.32
0 – 12.82
0 – 13.14
0.23 2.00 13.45
0.23 2.24 15.10
0.23 2.05 13.82
0.55 4.82 13.59
0.55 4.56 12.86
0.55 4.64 13.09
1 8.34 12.92
1 8.01 12.41
1 8.27 12.82
1 16.32 –
1 15.08 –
1 16.03 –
4.00 10.0 6.1 – 0 – 13.89 1.33 1.12
0 – 14.01
0 – 13.83
0.23 2.47 16.63
0.23 2.11 13.52
0.23 2.09 14.08
0.55 4.47 12.61
0.55 4.77 13.44
0.55 4.81 13.55
1 7.81 12.10
1 7.32 11.35
1 7.30 11.32
1 12.96 –
1 13.13 –
1 13.29 –

loading, the maximum point was always reached sooner in the torque vs. twist angle curve than in the corresponding
load vs. extension curve; this was true independently of both sharpness of the notch and ratio between the applied nom-
inal loadings. For this reason, the time corresponding to the maximum value in the torque vs. twist angle curve was used
to determine both Fu and Mu as illustrated in Fig. 5b. It is worth noticing here also that the difference between the Fu
460 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

(a) Tension - r n=4 mm

16
14 Fu =12.96 kN
12
10
Load [kN]

8 Notch tip

6
4
2 No crack!
0
0 0.2 0.4 0.6 0.8 1
Extension [mm]

(b) 20 Tension - r n=0.5 mm

18 Fu =16.85 kN
16
14
Load [kN]

12
10
8
6
4
2
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Extension [mm]
No crack!
Crack

Fast Fracture

Notch tip Notch tip

Ring Stable
crack
Fig. 7. Cracking behaviour under tensile loading of the notched specimens having rn equal to 4 mm (a) and to 0.5 mm (b).
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 461

Torsion - r n=4 mm
16
M u=13.91 Nm
14
12
Torque [Nm]

10
8
6
4
2
0
0 10 20 30 40 50 60 70
Twist angle [°]

Crack Crack
Notch tip

Notch tip
Notch tip
Crack

Fig. 8. Cracking behaviour under torsion (rn = 4 mm).

value estimated under mixed-mode loading according to the above definition and the maximum value of the tensile
loading recorded during each test was always less than 8%, with an average value equal to approximately 3.5%.
Lastly, the reanalysis of the gathered channels proved that as soon as the maximum value of the torsional loading
was reached, the material became unstable, rapidly coming to final breakage (see Fig. 5b). These considerations should
fully confirm the validity of the assumptions made to define the ultimate strength of the tested samples under complex
loading paths.
Fig. 6 shows two other examples of the static curves obtained by testing the notched specimens with rn = 0.44 mm under
a rnom to snom ratio equal to 0.23 and 1.0.
Three tests were carried out for every investigated geometry/loading configuration and the obtained results are listed in
Table 1.
Finally, in order to investigate in detail the cracking behaviour of this material, several tests were conducted dur-
ing which the specimens were closely observed using a optical microscope with digital camera, allowing the physi-
cal appearance of the specimens to be correlated to the applied stress and strain. Finally, crack initiation was defined
as the formation on the specimen surface of a technically detectable crack, that is, a crack having length of about
0.5 mm.

3. Results: material cracking behaviour

Fig. 1 shows the fracture profile of a plain specimen tested under tensile loading. This picture confirms that failure
was characterised by a moderately ductile fracture following some conventional necking. In other words, the maximum
462 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

Torsion - r n=0.5 mm
14
M u =11.81 Nm
12
10
Torque [Nm]

6
4

2
0
0 10 20 30 40
Twist angle [°]
Crack
Crack
Crack

Notch tip
Notch tip
Notch tip
Fig. 9. Cracking behaviour under torsion (rn = 0.5 mm).

value in the stress vs. strain curve corresponded to the onset of necking. No stable crack propagation was observed before
final breakage occurred, therefore it is very probable that failure of these specimens occurred in a conventional manner
for ductile materials, beginning internally with the formation and coalescence of small cavities [10–15], followed by
shear yielding, creating a fracture surface with a central flat region surrounded by shear lips at 45° to the applied
loading.
A similar cracking behaviour, producing a conventional cup-and-cone fracture surface, was observed also in the notched
tensile samples having rn = 4 mm (Fig. 7a): again, it should be emphasised that no cracking was observed before final break-
age occurred, so that, the maximum load in the load-extension curve corresponded again to the conventional onset of
necking.
On the contrary, in the specimens having a sharper notch root radius of 0.5 mm, the observed reduction of the tensile
load was seen to coincide with the formation of a circumferential, ring crack which initiated in the notch at the point of
maximum stress. Observing both the specimen surface during testing and the fracture surface afterwards, it can be con-
cluded that this crack propagated in a stable manner in the radial direction during the time that the applied load was
decreasing. Final breakage was seen to take place without any evident deformation of the net section by rapid unstable
crack growth occurring on a material plane perpendicular to the applied force (Fig. 7b). Finally, it is interesting to
observe that, since the tested aluminium was characterised by a moderate ductility, the yield stress under uniaxial load-
ing (but also the corresponding ultimate tensile force) tended to increase with increasing of the notch sharpness due to
the well-known ‘‘notch strengthening” phenomenon [18]: in the sharpest samples the bulk of the sample constrained
that material in the notch region which attempted to deform plastically, resulting in an increase of the yield stress
(Fig. 4a).
The observed material cracking behaviour under torsional loading is shown in Fig. 8 (for the blunt notch of
rn = 4 mm) and Fig. 9 (for the sharp notch of rn = 0.5 mm). These two figures make it evident that, independent
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 463

Twist angle [°]

0 10 20 30
9 22
Tensile curve 20
8 Fu =7.32 kN
18
7
16
6

Torque [Nm]
14
Load [kN]

5 M u =11.35 Nm 12
4 Torsional 10
3 curve 8
6
2 rn =4 mm 4
1 σ nom /τ nom =1 2
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6
Extension [mm]
No Crack!

Fast fracture

Notch tip

Fig. 10. Cracking behaviour under mixed-mode loading (rn = 4 mm, rnom/snom = 1).

of the sharpness of the tested notch, the maximum torque recorded during the test corresponded to the initiation
of a ring crack, similar to that seen in the sharply notched tensile specimens. These torsional cracks were also seen
to propagate radially on a plane perpendicular to the specimen axis, though the resulting fracture surface was
somewhat different in appearance from that in the tensile tests, being characterised by circular concentric marks
created by the rubbing action of the two emerging fracture surfaces (see Figs. 8 and 9). Unlike the tensile testing,
however, no unstable crack propagation occurred in this case: rather, there was a gradual reduction in the applied
torque with increasing twist angle, corresponding to a gradual decrease in the remaining cross-section as the ring crack
grew.
It is interesting to recall here that the fracture surfaces observed in the above notched samples loaded in torsion
were different to that commonly observed in brittle materials subjected to Mode III loading [16,17]. In more detail,
in pre-cracked specimens of glass under anti-plane stress final breakage is seen to occur due to the formation of cracks
which propagate on those material planes experiencing the maximum Mode I loading. This particular cracking behaviour
results in fracture surfaces showing characteristic markings usually called ‘‘lances” (see Ref. [16] and references reported
therein). We have recently observed that a similar mechanism is active also in both bluntly and sharply notched cylin-
drical samples of PMMA subjected to torsion, final failure being caused by the formation and propagation of multiple
Mode I cracks [17]. It is interesting to note here also that the above Mode I cracks were seen to initiate at different
points along the notch tip circumference and propagate by interacting with each other: this retarded the final failure
due to an energy dissipation phenomenon and such a retardation resulted, after the initial elastic behaviour, in an
almost horizontal plateau in the corresponding torque vs. twist angle curves.
464 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

Twist angle [°]

0 1 2 3 4 5 6 7 8 9
9 18
8 Fu =7.56 kN 16
7 14
Torsional
6 M u =11.72 Nm 12

Torque [Nm]
curve
Load [kN]

5 10
4 Tensile curve 8
3 6
2 rn =0.5 mm 4
1 σ nom /τ nom =1 2
0 0
0 0.05 0.1 0.15 0.2 0.25
Extension [mm]
Crack

Fast fracture

Notch tip
Ring stable crack

Fig. 11. Cracking behaviour under mixed-mode loading (rn = 0.5 mm, rnom/snom = 1).

Contrary to the cracking mechanisms observed in brittle materials, the direct inspection of our samples of
Al6082 tested under torsion suggests that the initiation and propagation phenomena occurred on planes perpendicu-
lar to the specimen axis simply because the most important mechanisms leading to initiation and growth of
cracks were mainly shear stress governed, the tested material being characterised by a relatively ductile behaviour
(Fig. 1).
Finally, it is interesting to observe that, since the tested aluminium displayed a negligible level of plastic deformation
when loaded in pure torsion, the notch strengthening effect [18] was not observed: as shown by the chart of Fig. 4b, the mea-
sured yield shear stress was practically independent of notch sharpness.
Cracking behaviour was also monitored in the mixed tension/torsion experiments carried out under a rnom to snom ratio
equal to unity. Two different notch geometries were considered, i.e. rn equal to 4 mm and to 0.5 mm.
As Fig. 10 shows, for this mixed-mode loading condition, failure in the blunt notch was similar to that previously observed
when this notch was tested in pure tension, i.e. no crack was detected before final breakage occurred and the fracture surface
was seen to be characterised by two different areas: a shear-like fracture surface at 45° to the specimen axis and a brittle-like
fracture surface perpendicular to the axis of the sample itself.
On the contrary, failure from the sharper notches (Fig. 11) was preceded by the formation of the ring crack at the time of
maximum torque, which propagated first in a stable manner and then, at final failure, in an unstable manner.
To conclude, Fig. 12 summarises the different fracture surfaces obtained by reducing the notch root radius value from
4 mm down to 0.44 mm as well as by increasing the ratio between the applied nominal loadings from zero up to infinity:
the above matrix makes it evident that, as the geometry/loading configuration changed, the morphologies of the fracture
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 465

Fig. 12. Matrix of the observed fracture surfaces.

surfaces varied according to the different mechanisms summarised in the above paragraphs. For the sake of clarity, the above
complex behaviour is summarised also by the schematic chart reported in Fig. 13.

4. Discussion

It is evident that the introduction of a notch, and the orientation of this notch with respect to the applied loading,
plays a crucial role in determining the mode of failure as well as the critical load. Here we have demonstrated transi-
tions between three mechanisms of failure – ductile failure (with no cracking); stable crack growth and unstable crack
growth – as a result of changing notch sharpness and loading mode, for the particular case of the circumferential notch.
It is well-known that notches have the effect of ‘‘embrittling” a ductile material, especially one which has a relatively
low fracture toughness, such as this aluminium alloy; here we show that the embrittling effect is stronger, i.e. it occurs
for blunter notches, if a torsional loading component is present. On the other hand, we see that, in this geometry at
least, torsion has the effect of stabilising crack growth, causing a more gradual loss of strength with increasing
deformation.
466 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

Fig. 13. Schematic chart summarising the observed cracking behaviour.

Some of these findings have important consequences for the development of a suitable theory for the prediction of failure
in this situation – a theory which will be developed in full in the second part of this two-part series of papers. The first impor-
tant finding is that crack initiation did not begin until peak loading conditions occurred. As noted above, the definition of
peak loading is not a straightforward matter when both tension and torsion are being applied simultaneously; nevertheless,
using the definition developed above, it is clear that no cracks appeared until the maximum stress was reached. This is a very
useful observation: since the aim of the theory is to predict the strength of the specimen, i.e. its maximum load-bearing
capacity, then it is clear that one does not need to take crack propagation into account: rather, failure can be defined in terms
of crack initiation. This simplifies the theoretical model considerably as it means that it can be based on a stress analysis of
the initial, uncracked, specimen geometry.
The second important finding is that, whatever the degree of tension or torsion applied, crack propagation always took
the form of a circumferential crack which lay in a plane perpendicular to the specimen axis. The initiation of this crack is
obviously dictated by the need to form in the notch root, where stresses are highest. In high-cycle fatigue of metals, and
in the static fracture of some brittle materials, torsional loading causes cracks to grow at 45° to the specimen axis, but this
behaviour did not occur in the present case. This finding is important for theoretical developments because it confirms that
the crack does not follow the path which maximises the Mode I crack-opening stress, implying that a maximum principal
stress criterion will not be appropriate to describe this type of failure.
A complicating factor for the development of any theoretical model is the change in mode from ductile to brittle failure.
This suggests that a single unified approach may not be able to predict failure loads for all the notch types and loading modes
used here. Approaches based on LEFM might be expected to work well when cracking precedes failure: the notches might, for
example, be modelled as pre-existing cracks. But such an approach would clearly not be expected to apply when ductile fail-
ure occurred: this type of failure might be controlled by an effective stress such as that of Von Mises or Tresca. Despite these
reservations, we decided to proceed on the basis of developing a single approach to predict all types of failure, at least as an
initial hypothesis to be tested.
Finally, one further observation is perhaps worthy of mention. We noticed that some of the fracture surfaces
displayed a series of parallel lines: these occurred on the flat areas of the fracture surface, i.e. those regions correspond-
ing to crack propagation, and tended to be more obvious if the tensile component was relatively large. This type of
fracture surface appearance is very similar to that commonly displayed by fatigue failures: Figs. 14 and 15 illustrate
this, showing fracture surfaces from fatigue test specimens in the same material. If one was conducting a failure
investigation of an engineering component it would be easy to jump to the conclusion that the failure was due to
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 467

Fig. 14. Static vs. fatigue failures in specimens with rn = 4 mm and loaded in simple tension as well as in simple torsion.

fatigue, based on the flat, planar nature of the surface and presence of these parallel lines, but such a conclusion could
be erroneous.

5. Conclusions

The most important mechanisms leading to final static breakage of the tested notched samples made of Al6082 can be
summarised as follows:

(1) In the presence of blunt notches subjected to uniaxial tensile loading, failure occurs due to a ductile-like fracture
without any initial stable crack propagation.
(2) In the presence of sharp notches subjected to uniaxial tensile loading, failure occurs due to a brittle-like Mode I fast
fracture preceded by the growth of initial stable cracks.
(3) Under torsional loading, and independent of notch sharpness, failure occurs due to the propagation, on material
planes perpendicular to the specimen axis, of stable shear cracks.
(4) Under mixed Mode I and III loading, failure occurs under a combination, depending on the mutual interaction
between sharpness of the notch and nominal tensile stress to nominal shear stress ratio, of the above three funda-
mental mechanisms.
(5) Crack initiation always coincides with the peak loading conditions, as defined from the tensile and shear force/
deflection curves.
468 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469

Fig. 15. Static vs. fatigue failures in specimens with rn = 0.5 mm and loaded in simple tension as well as in simple torsion.

(6) Crack propagation always follows the plane perpendicular to the specimen axis.
(7) In some cases, crack propagation gave rise to flat fracture surfaces displaying a series of parallel lines, very similar in
appearance to the fracture surfaces generated in fatigue tests on the same material.

References

[1] Neuber H. Theory of stress concentration for shear-strained prismatical bodies with arbitrary nonlinear stress–strain law. J Appl Mech
1961;28:544–50.
[2] Taylor D. The Theory of Critical Distances: a new perspective in fracture mechanics. Oxford (UK): Elsevier; 2007.
[3] Anon. Standard test method for Young’s modulus, tangent modulus, and chord modulus. ASTM E111-04; 2004. doi:10.1520/E0111-04.
[4] Anon. Standard test methods for tension testing of metallic materials. ASTM E8/E8M-08; 2008. doi:10.1520/E0008_E0008M-08.
[5] Anon. Standard test method for shear modulus at room temperature. ASTM E143-02; 2002. doi:10.1520/E0143-02.
[6] Hancock JW, Mackenzie AC. On the mechanism of ductile failure in high strength steels subjected to multi-axial stress states. J Mech Phys Solid
1976;24:147–69.
[7] Mackenzie AC, Hancock JW, Brown DK. On the influence of state of stress on ductile failure initiation in high strength steels. Engng Fract Mech
1977;9:167–88.
[8] Hancock JW, Brown DK. On the role of strain and stress state in ductile failure. J Mech Phys Solid 1983;31:1–24.
[9] Scheider I. Derivation of separation laws for cohesive models in the course of ductile fracture. Engng Fract Mech 2009. doi:10.1016/
j.engfracmech.2008.12.006.
[10] Puttik KE. Ductile fracture in metals. Philos Mag 1959;4:964–9.
[11] Rogers HC. The tensile fracture of ductile metals. Trans Met Soc AIME 1960;218:498–506.
[12] Gurson AL. Continuum theory of ductile rupture by void nucleation and growth. Part I: Yield criteria and flow rules for porous ductile media. Trans
ASME J Engng Mater Technol 1977;99:2–15.
[13] Tvergaard V. On localization in ductile materials containing spherical voids. Int J Fract 1982;18:237–52.
 L. Susmel, D. Taylor / Engineering Fracture Mechanics 77 (2010) 452–469 469

[14] Needleman A, Tvergaard V. An analysis of ductile rupture in notched bars. J Mech Phys Solid 1984;32:461–90.
[15] Callister WD. Fundamentals of materials science and engineering – an integrated approach. 2nd ed. USA: John Wiley & Sons; 2005. ISBN: 0-471-47014-
7.
[16] Sommer E. Formation of fracture ‘lances’ in glass. Engng Fract Mech 1969;1:539–46.
[17] Susmel L, Taylor D. The Theory of Critical Distances to predict static strength of notched brittle components subjected to mixed-mode loading. Eng Frac
Mech 2008;75(3–4):534–50.
[18] Hertzberg RW. Deformation and fracture mechanics of engineering materials. 3rd ed. Singapore: John Wiley & Sons; 1989.