Sunteți pe pagina 1din 38

International Journal of Fracture (2005) 134:267–304

DOI 10.1007/s10704-005-1088-2 © Springer 2005

A review of critical plane orientations in multiaxial fatigue failure


criteria of metallic materials

ALEKSANDER KAROLCZUK and EWALD MACHA∗


Department of Mechanics and Machine Design, Technical University of Opole, ul. Mikolajczyka 5,
45-271 Opole, Poland

Author for correspondence (E-mail: karol@po.opole.pl)

Received 30 July 2004; accepted 18 July 2005

Abstract. The paper presents a review of multiaxial fatigue failure criteria based on the critical plane
concept. The criteria have been divided into three groups, according to the fatigue damage parameter
used in the criterion, i.e. (i) stress, (ii) strain and (iii) strain energy density criteria. Each criterion
was described mainly by the critical plane orientation. Multiaxial fatigue criteria based on the crit-
ical plane concept usually apply different loading parameters in the critical plane whose orientation
is determined by (a) only shear loading parameters (crack Mode II or III), (b) only normal loading
parameters (crack Mode I) or sometimes (c) mixed loading parameters (mixed crack Mode). There
are also criteria based on few critical plane orientations and criteria based on critical plane orienta-
tions determined by a weighted averaging process of rotating principal stress axes.

Key words: Critical and fracture plane, fatigue failure criteria, multiaxial loading.

Nomenclature

Loading parameters
J J integral
R stress ratio R = σmin /σmax
Ta generalised shear stress amplitude
W strain energy density
WII,A , WII,B strain energy density for Mode II of cases A and B cracks,
respectively
χ angle determining the shear direction in the critical plane
described by ζ , ξ angles
ε normal strain
γ shear strain
ρ∗ stabilised stress tensor
σ ,τ normal stress and shear stress, respectively
ζ, ξ angles in spherical coordinate system describing the normal
direction to the critical plane in relation to the basic Cartesian
coordinate system Oxyz
Material coefficients
b, c fatigue strength exponent and fatigue ductility exponent, respec-
tively
E, G Young’s modulus and Kirchhoff’s modulus, respectively
268 Aleksander Karolczuk and Ewald Macha

n′ , K ′ exponent and coefficient of cyclic hardening, respectively


γf′ shear fatigue ductility coefficient

εf fatigue ductility coefficient
ν Poisson’s ratio
σf′ fatigue strength coefficient
σy , σu yield stress and ultimate strength, respectively

τf shear fatigue strength coefficient
General
Nf number of cycles to failure
t time
Subscripts and others
a amplitude
b bending
af fatigue limit
c critical value
e, p elastic, plastic
eff, eq effective and equivalent, respectively
h hydrostatic
ij (i, j = x, y, z) components in the Cartesian coordinate system Oxyz
m, max, min mean, maximum and minimum value in time domain, respec-
tively
n in the plane with normal vector n
ns along direction s on the plane with normal vector n
oct octahedral
s along direction s
µ microscopic
I, II, III, ou crack Mode I, Mode II, Mode III and mixed Mode, respec-
tively
* range

1. Introduction

Multiaxial fatigue failure criteria according to applied loading state parameter may
be divided into three groups i.e. stress, strain and energy based criteria. Among these
criteria, one type called the critical plane approach can be distinguished. Significance
of this approach has increased during last years, because of its effectiveness and
broad application range. The general aim of criteria based on the critical plane con-
cept is the reduction of a multiaxial stress state to an equivalent uniaxial one. The
idea to combine the normal and shear stresses in a fixed plane (critical) within a
material was first proposed by Stanfield (1935):

Another possible criterion which was not referred to, was that in which the two
components of stress acting across any plane, i.e. shear and direct stress, might be
taken as each contributing a definite quota to “disruption” combined by a sim-
ple arithmetical relation. ( . . . ) The planes on which such effect were maximum
would not be the principal planes ( . . . )
Multiaxial fatigue failure criteria of metallic materials 269

but this concept was not developed until the Fifties when Findley (1959), Stulen and
Cummings (1954) introduced the word ”critical” and verified fatigue stress criteria
based on the critical plane approach. The critical plane concept first used in stress
criteria, was subsequently introduced to strain criteria and then to energy criteria.
Criteria based on the critical plane concept allow to calculate fatigue life as well
as fatigue fracture plane position. Together with the evolution of research on fatigue
phenomena, many models of crack types have been proposed. The algorithms for
calculation of fatigue life on the basis of the critical plane approach are specified
depending on a dominating crack type.
The process of fatigue failure may be divided into different number of phases.
In uniaxial fatigue tests, one of the most popular division is that into four phases
(Brown and Miller, 1979): (i) nucleation of fatigue crack (defined as initiation); (ii)
crack growth on a plane of maximum shear; (iii) crack propagation normal to the
tensile stress; and (iv) final rupture of the specimen.
Forsyth (1961) presented the crack propagation model of two stages in the fatigue
process. In stage I fracture planes coincide with the maximum shear stress direction.
Stage II is predominated by the maximum normal stress value and macroscopic crack
lines are perpendicular to that stress. Stage I and stage II are equivalent to phase (ii)
and (iii), respectively.
Other well known descriptions of crack types are Modes I, II and III, distin-
guished by Irwin, and cases of cracks A and B, distinguished by Brown and Miller
(1973).
Socie (1987) distinguished three categories of fatigue damage depending on the
predominating damage mechanism. They are: type A in which cracks grow along the
maximum shear strain (shear crack growth, stage I), type B in which crack grows
perpendicular to the maximum normal strain direction predominates (tensile crack
growth, stage II) and type C in which only crack nucleation exists. Socie proposed a
different fatigue criterion for each type of cracks. For type A, the criterion of max-
imum shear and normal strains on the maximum shear strain plane was proposed.
Socie proposed, for type B, the criterion of the maximum normal stress and strain
on the maximum normal strain plane and, for type C, the criterion of the maximum
shear and normal stress on the maximum shear stress plane.
Many researchers agree that crack planes in the initiation phase coincide with the
maximum shear stress plane and in a propagation period coincide with the maximum
normal stress plane. The crack length and fatigue life of each phase distinguished by
different authors depend on many factors, i.e. stage I (Forsyth, 1961), type A (Socie),
Mode II and III (Irwin) dominate in ductile materials; stage II (Forsyth), type B (Socie),
Mode I dominate in brittle materials. Domination of each phase depends also on
material sensitivity to damage mechanism caused by normal and shear stresses. Some
authors stated that macroscopic fatigue fracture plane is a shear type when the ratio of
the amplitudes of shear stress to normal one is greater than 0.63 (Cox and Field, 1952;
McDiarmid 1985); 1.0 (Susmel and Petrone, 2001) or when the ratio of ranges of shear
strain to normal one, *γ /*ε, is greater than 1.5 Brown and Miller, 1973); 1.7(Ogata
et al., 1989); 3.0 (Sakane et al., 1987); 4.0 (Brown and Miller, 1973). Values of this
ratio depend on material, stress level, correlation between loading components and
temperature. Macroscopic fracture plane orientation should not in general be identi-
fied with the critical plane orientation. These orientations may be compared only when
270 Aleksander Karolczuk and Ewald Macha

the fatigue crack directions in an initiation and a propagation period coincide. This
case occurs for specific loading and materials: for example, for the materials in brittle
state (cast irons, Marquis and Karjalainen-Roikonen, 2001) as well as for materials in
intermediate state under specific cases of loading (Savaidis and Seeger, 1997), the crack
directions in an initiation and a propagation period coincide with the maximum nor-
mal stress direction. The choice of the critical plane orientation in case when the crack
directions in an initiation and a propagation period do not coincide is not obvious. In
that case the lengths of the initiation and propagation periods should be taken into
account, or fatigue life should be computed using more than one critical plane posi-
tion. This approach is used in a few multiaxial fatigue criteria (Das and Sivakumar,
1999, 2000; Zolochevsky, 2000). In the paper (Berard et al. 1993) authors noticed that,
under non-proportional loading, the number of cracks with different orientations in
an initiation period is greater than under proportional loading, when only two crack
directions were noticed, which coincided with the maximum shear strain plane. That is
caused by the rotation of the principal stress directions which activate a greater number
of slip bands in grains. The aim of this paper is to review critical plane orientations
used in multiaxial fatigue failure criteria.

2. Critical planes in criteria based on stresses

2.1. Generalised hypothesis of material resistance (socie and marquis, 2000)

The first criteria of multiaxial fatigue come from static hypotheses of material resis-
tance. Many researchers tried to adapt the static hypotheses of material resistance to
fatigue by replacing stress static values with amplitudes or ranges of dynamic load-
ing. Among multiaxial fatigue failure criteria, which come from static hypotheses, the
most frequently verified with experimental data were: criterion of maximum normal
stress, criterion of maximum shear stress, and criterion of octahedral shear stress.
These criteria formally do not belong to fatigue criteria based on the critical plane
concept. However, we can exactly determine the plane related to the considered stress.

2.1.1. Criterion of maximum normal stress


According to this criterion, the maximum normal stress range is responsible for
fatigue of materials. The hypothesis of static effort applied to cyclic loading leads to
the following equation for the equivalent stress range

*σeq = *σ1 . (1)

Multiaxial stress state is reduced to uniaxial one by means of equivalent stress range
*σeq , which is equal to the maximum normal stress range *σ1 . In this model, the
critical plane is assumed as the plane with the maximum normal stress range.

2.1.2. Criterion of maximum shear stress


In this criterion, it is assumed that the maximum value of shear stress range is
responsible for fatigue of materials (Tresca hypothesis), as follows:
*σeq *σ1 − *σ3
*τ13 = = . (2)
2 2
Multiaxial fatigue failure criteria of metallic materials 271

Since the maximum shear stress occurs always in two perpendicular planes, we have
two perpendicular critical planes.

2.1.3. Criterion of maximum octahedral shear stress


In static resistance hypothesis of the maximum octahedral shear stress τoct , the
shear stress occurs on the plane equally inclined to the principal stresses directions.
From this hypothesis adapted to variable loading, we have
1 ! "1/2 3
*σeq = √ (*σ1 − *σ2 )2 + (*σ2 − *σ3 )2 + (*σ3 − *σ1 )2 = *τoct √ . (3)
2 2
The critical plane is the octahedral plane with the maximum shear stress range. The
stress ranges are calculated as differences between the maximum and minimum val-
ues of stresses in time domain. In the cyclic loading we can calculate a stress range
when the maximum and minimum values are specified in the same plane. As is well
known, principal directions of stresses change even for the simplest variable loading
state and this fact makes impossible to determine the stress ranges in unambiguous
way. Nevertheless, there were many trials carried out especially for proportional load-
ing states, wrongly establishing a stationary position of principal stress directions.

2.2. Findley criterion (1956) (findley, 1959)

Findley (1959) proposed a linear combination of normal stress σn and shear stress
τns,c in the critical plane for a given number of cycles to failure

τns,c + kσn = f, (4)

where f and k are material coefficients. A similar concept was proposed by Stulen
and Cummings (1954). This criterion was formulated and tested for proportional
combination of cyclic bending and torsion. Then the relation was expanded in order
to take into account the influence of a mean stress value

τns,a,c + kσn,max = f. (5)

This criterion was effective for proportional combination of bending and torsion
under the same ratio of normal to shear stress amplitudes for variable loading and
static loading (σa /τa = σm /τm ). The critical plane is defined as the plane experienc-
ing the biggest value f . According to Findley, orientation of the critical plane in the
case of zero mean stress values depends on the direction of the maximum principal
stress σ1 and material coefficient k. Findley noticed that k value was small for ductile
materials and the position of the critical plane for these materials approaches to the
direction of maximum shear stress (α = π/4). A high k value (α = 0) is characteris-
tic for brittle materials like cast iron, and the critical plane position is then compat-
ible with the position of maximum principal stress direction σ1 . In the case of mean
stresses different from zero, the position of the critical plane depends not only on the
direction of maximum principal stress σ1 and k value, but also on variable and static
stresses. When the mean stresses from bending and torsion are high in comparison
with variable stresses, the angle α approaches zero, that is, the critical plane is the
272 Aleksander Karolczuk and Ewald Macha

plane of the maximum principal stress σ1 . Assumption made by Findley that the crit-
ical plane is the plane experiencing the biggest value f , Equation (5) is correct from
mathematical point of view and our intuition. But, from experimental observations it
is evident that under uniaxial tension-compression tests cracks are formed only in the
plane of the maximum shear or normal stress. According to Equation (5) the shear
stress τns always influences fatigue life, it leads us to conclusion that even under ten-
sion-compression tests the critical plane orientation will never coincide with the plane
of the maximum normal stress, what is not true for some type of materials (e.g. brit-
tle materials). Findley did not defined a mathematical formula for coefficient f . Some
researchers (Park and Nelson, 2000; Backstrom and Marquis, 2001) assume that it
can be determined from the shear-mode cracking

τns,a,c + kσn,max = τf′ (2Nf )b . (6)

2.3. Mcdiarmid criterion (1972, 1990) (mcdiarmid, 1985, 1987; diebmann and
bhongbhibhat, 1991)

McDiarmid noticed that important parameters of loading in high cycle fatigue


(HCF) regime are the maximum shear stress range and the normal stress in the crit-
ical plane. The critical plane in this criterion is the plane of maximum shear stress
range. McDiarmid proposed a few fatigue criteria based on the critical plane of the
maximum shear stress amplitude. In 1972, he proposed
1.5
τns,a = C1 − C2 σn,a , where C1 = τaf , C2 = (1 − σaf ,b /2)/(σaf ,b /2)1.5 , (7)

valid for 0.5 < τaf /σaf ,b < 1. This expression describes the relation between shear stress
amplitude and normal stress amplitude in the critical plane for a given number of
cycles to failure. For non-proportional loading the proposed criterion does not pre-
cisely describe experimental test results. In 1990 McDiarmid proposed a new criterion
including mean stress value σn,m = σn,max − σn,a in the following formula:

τns,a = τaf − [τaf /2σu ]σn,max or τns,a /τaf + σn,max /2σu = 1, (8)

valid for 0.5 < τaf /σaf ,b < 1 and 0 ! σn,max ! σT . This criterion distinguishes case A and
B of fatigue cracks (described in strain criterion by Brown and Miller). Distinction
consists in applying different fatigue limits τaf determined separately for crack case A
and B. The proposed criterion correlated experimental data for proportional bending
with torsion very well, as well as for non-proportional loading with zero and non-
zero mean value of loading. Only for one case of loading, namely τa /σa = 0.5 and
phase shift π/2, where any plane experiences a maximum shear stress range, the cri-
terion was ineffective. From the criterion (8) damage parameter can be deduced (Park
and Nelson, 2000) as

τns,a + [τaf /2σu ] σn,max = τf′ (2Nf )b . (9)


Multiaxial fatigue failure criteria of metallic materials 273

2.4. Dietman et al. criterion (1974), (dietmann et al. 1991)

Dietman et al. were among the first researchers who paid attention to the interac-
tion between changes of principal stress directions and fatigue life. They proposed to
modify the criterion of octahedral shear stress to take into account the changes of
direction of principal stress axes. This criterion assumes that material damage occurs
when the shear stress amplitude, τns,a , in the critical octahedral plane reaches its crit-
ical stress value, τns,a,c , characteristic for a given material
τns,a = τns,a,c . (10)
Note that search of the critical plane orientation takes place only among the octahe-
dral planes, which means planes equally inclined to principal axes of stresses. In case
of fatigue loading, especially under out-of-phase loading, the principal stress axes
rotate, so the position of the octahedral plane changes as well. The critical plane is
the octahedral plane at time tmax , for which octahedral shear stress τoct,max achieves
the maximum value. Unfortunately, this criterion was used only to determine the
fatigue limit and the results were not compared to any standard fatigue characteristic
(uniaxial S–N curves, e.g Wöhler characteristic).

2.5. Simbürger and grubisic criterion (1976), (grubisic and simburger, 1976;
zenner et al. 2000)

Simbürger proposed a criterion including mean stress value and rotation of principal
stress directions. In plane stress state, all possible orientations of the considered plane
can be described by the angle α. The proposed fatigue parameter S is formulated as
follows
#
$ &π
$8
$
S =% Sn2 dα, (11)
π
0

where:
σeq
Sn = , (12)
σa,c
' '
σeq = σeq,a − mσeq,m , σeq,m = a1 'τns,m ' + a2 σn,m , σeq,a = a1 τns,a + a2 σn,a . (13)
Material coefficients a1 and a2 are functions of fatigue limits σaf and τaf , whereas σa,c
represents the critical stress amplitude for a given number of cycles to failure. Coeffi-
cients m allows to take into account the mean stress value σeq,m . Simbürger defines
the parameter Sn for different kinds of materials. For brittle materials, the parameter
Sn is a function of normal stress only. For materials revealing intermediate features of
brittle and ductile materials, the parameter Sn depends on normal and shear stresses.
This criterion in principle does not belong to the critical plane approach because
the parameter Sn is independent of a specific plane orientation. However, Simbürger
determined the position of fatigue fracture plane as a plane with maximum value of
Sn parameter. Simbürger did not define a fatigue characteristic (Nf − S) which should
be used to calculate fatigue life basing on Equation (11).
274 Aleksander Karolczuk and Ewald Macha

2.6. Matake criterion (1977), (matake, 1977)

Matake proposed a criterion assuming that fatigue damage of material is due to the
maximum shear stress along with the normal stress σn in plane of maximum shear
stress τns = τ13
τns,a + kσn,a = τf′ (2Nf )b . (14)
This criterion was formulated to analyze cyclic torsion, bending and proportional tor-
sion with bending. For such cases of loading, constant directions of principal stresses
are assumed. With this assumption Equation (14) can be written as follows
σ1 − σ3
+ kσn,a = τf′ (2Nf )b . (15)
2
Coefficient k may be calculated from uniaxial fatigue tests
τaf − σaf /2 τaf
k= =2 − 1. (16)
σaf /2 σaf
The critical plane is one of two the planes of maximum shear stress with a higher
normal stress value.

2.7. Macha criterion (1979), (macha, 1979)

Macha formulated a generalized criterion of maximum shear τns (t) and normal stress
σn (t) in the critical plane for multiaxial random loading. The particular assumptions
are as follows:
– fatigue crack is controlled by normal stress σn (t) and shear stress τns (t) occur-
ring in s direction on the critical plane with normal n,
– direction s on the plane with normal n coincides with the mean direction of maxi-
mum shear stress, max {τns (t)}. Direction of the maximum shear stress max {τns (t)}
s s
rotates, in general case, on the fixed plane with normal n at time t. Hence,
direction s is assumed to be a mean direction from all directions of max {τns (t)}
s
occurring in the analyzed time,
– for a given fatigue life, the maximum value of a linear combination of stresses
σn (t) and τns (t) under multiaxial random loading satisfies the following equation
max {Bτns (t) + Kσn (t)} = F, (17)
t

where B, K, F are constants which depend on the choice of the particular


criterion version. The position of the critical plane is determined by normal
direction n, and shear direction on that plane is determined by the vector s,
(Figure 1).
Directions n and s may be computed by three methods: the weight function
method, the damage accumulation method, the variance method. The weight func-
tion method (Macha, 1988, 1989, Carpinteri et al., 1999a, b) consists in the weighted
averaging process of the parameters which describe the instantaneous locations of the
principal stress axes. Averaged directions of the principal stress axes determine aver-
aged directions of maximum shear stresses as well. In the case of material sensitive
Multiaxial fatigue failure criteria of metallic materials 275

σ (t) t
n

τ (t)
ns
s

t
Figure 1. Normal stress history σn (t) and shear stress history τns (t) in the critical plane.

to the Mode I loading, the principal directions should be averaged through weight
functions based on normal stresses. For materials sensitive to Mode II or III load-
ing, the weight functions should depend on shear stresses. The first case is charac-
teristic for brittle materials, the second one for semi-ductile materials. Directions of
vectors n and s are established with respect to the averaged positions of principal
stress axes and these with respect to a fixed coordinate system. Damage accumulation
method (Macha, 1988, 1989) consists in fatigue damage accumulation on all possible
planes by finally choosing that with the maximum damage degree. As a result, the
orientation of the critical plane and the fatigue life are obtained. Vector of the nor-
mal direction n and vector of the shear stress s are searched in an iterative way or
by optimization methods. In the variance method (Macha, 1988, 1999) it is assumed
that the plane in which variance of equivalent stress according to a chosen fatigue
criterion achieves maximum is critical for material. Equation (17) is a function of two
vectors determining the normal direction n of the critical plane and the shear direc-
tion s in that plane. Hence, Equation (17) is a function of three independent vari-
ables, as well as in the case of the damage accumulation method.

2.8. Dang van criterion (1982), (dang van, 1989, 1993; froustey and
lasserre, 1989)

This criterion is based on the concept of micro-stresses in the critical volume of mate-
rial. Such a model was built on the basis of observations of the local processes of
the crack nucleation in grains. As the result of plastic deformation, the intracrystal-
line slipping bands are formed in grains, which begin the crack process. Dang Van
proposed that an important parameter responsible for crack nucleation along slipping
bands is microscopic shear stress τµ in grain area. The second also very important
parameter according to the author is the microscopic hydrostatic stress σµ,h , which
influences crack opening process. This two proposed fatigue parameters are involved
in a linear function

τµ (t) + a1 σµ,h (t) = a2 , (18)


276 Aleksander Karolczuk and Ewald Macha

where a1 , a2 are constants determined from cyclic uniaxial fatigue tests. Microscopic
shear stress τµ is computed from the microscopic principal stresses according to
Tresca hypothesis
! "
τµ (t) = 21 σµ,1 (t) − σµ,3 (t) . (19)
The microscopic principal stresses σµ,1 (t), σµ,3 (t) are calculated from microscopic
stress tensor σµ,ij (t). This tensor is calculated as the sum of macroscopic stress tensor
σij (t) and deviatoric part of the stabilised residual stress tensor, devρ ∗
σµ,ij (t) = σij (t) + devρ ∗ . (20)
Hence, the critical plane position (corresponding to the maximum shear plane)
depends on two quantities one of which is time independent. Because of that, the
rotation of maximum shear plane calculated from microscopic stress tensor is smaller
than that calculated from macroscopic stress tensor. The use of the stabilized residual
stress tensor devρ ∗ distinguished this proposal from the other fatigue criteria.

2.9. Robert criterion (1992), (vidal et al., 1996; kenmevgne et al., 1996;
weber et al., 1999a,b)

Robert presented a criterion which takes into account the shear stress τns (t), the nor-
mal stress σn (t) and the mean value of the normal stress σn,m in the critical plane
σeq (t) = ∥τns (t)∥ + a1 (Nf )σn (t) + a2 (Nf )σn,m , (21)
where a1 (Nf ), a2 (Nf ) are criterion parameter depending on uniaxial fatigue charac-
teristic: fully reversed axial and torsion loading (R = −1), a zero to maximum tensile
test (R = 0). The fatigue criterion is defined by
max σeq (t)
n,t
= 1, (22)
a3 (Nf )
where a3 is the third criterion parameter. All criterion parameters a1 , a2 , a3 were iden-
tified (Vidal et al., 1996; Kenmevgne et al., 1997; Weber et al. 1999a,b). The number
of cycles to failure Nf is the solution of the Equation (22) and it is obtained from
an iterative process. The mean value of shear stress τns,m in the critical plane was
neglected in the proposal because of its little influence on fatigue life in comparison
with mean normal stress σn,m . The critical plane is the plane in which the equivalent
stress σeq (t) reaches the maximum value. The procedure presented by Robert is com-
plicated, it requires iterative procedure to compute the number of cycles to failure Nf
and an another procedure to determine the critical plane orientation. But, there are
two advantages: (i) the criterion parameters a1 , a2 , a3 were identified with the use of
three uniaxial S–N curves and (ii) the criterion can be applied for random loading.

2.10. Papadopoulos criterion (1993), (papadopoulos et al., 1997;


papadopoulos, 2001)

Papadopoulos proposed the criterion based on a linear combination of generalized


shear stress amplitude Ta in the critical plane and the maximum value of hydrostatic
stress σh,max
Multiaxial fatigue failure criteria of metallic materials 277

Figure 2. Shear stress history τns (t) in the critical plane.

max Ta + α∞ σh,max ≤ γ∞ , (23)


where γ∞ , α∞ are material parameters, and
σh,max = max [σkk (t)/3] (24)
t

is the hydrostatic stress. The shear stress amplitude τns,a is defined in direction s
(Figure 2) described by the angle χ in the plane with normal vector n established
by ζ and ξ angles in spherical coordinate system with respect to the fixed Cartesian
coordinate system Oxyz
( )
τns,a (ζ, ξ, χ ) = 21 max τns (ζ, ξ, χ , t) − min τns (ζ, ξ, χ , t) . (25)
t t

For this definition of shear stress amplitude τns,a , the generalised shear stress ampli-
tude Ta is searched in the critical plane defined by the vector n
#
$ &2π
$
$1 2 (ζ, ξ, χ) dχ.
Ta (ζ, ξ ) = % τns,a (26)
π
χ =0

The quantity Ta is directly connected to the mean square value of shear stress ampli-
tude τns,a in the critical plane. The critical plane is the plane where Ta achieves its
maximum value. Papadopoulos made the hypothesis that any finite-life locus is rep-
resented by a plane in the space (σh,m , σh,a , max Ta ). Orientation of this plane can
be determined according to standard fatigue tests (fully reversed torsion or bend-
ing tests, (Papadopoulos, 2001). Formula of S–N curve was assumed as a power law
function of the number of cycles to failure. Nevertheless, Papadopoulos does not
impose any particular form of the basic uniaxial S–N curve to be used for identi-
fication of the parameters.

2.11. Zolochevski et al. criterion (2000), (zolochevsky et al. 2000)

Zolochevski, Obataya and Betten proposed a criterion taking into account the mean
stress value and the material anisotropy. The criterion is based on the stress intensity
amplitude, σ̄a
*
3
σ̄a = sij,a sij,a , (27)
2
278 Aleksander Karolczuk and Ewald Macha

where sij,a are amplitudes of deviatoric part of stress tensor. Stress √ intensity ampli-
tude is proportional to octahedral shear stress amplitude τa,oct (σ̄a = 3 2/2τa,oct ). The
criterion takes into account the mean stress value by adding the amplitude of stress
intensity and two mean stress quantities computed in two directions (directions: n
and m). The amplitude of equivalent stress, σeq,a , is given by
σeq,a = λ1 σ̄a + a1 λ2 nk σkl,m nl + a2 λ3 mk σkl,m ml , (28)
where λ1 , λ2 , λ3 are material parameters, a1 , a2 are weight coefficients, σkl,m is the
mean stress tensor, nk , nl , mk , ml are vector n and m coordinates. Vectors n, m deter-
mine two fatigue crack directions. Vector n is perpendicular to the family of Mode
I cracks. Vector m is perpendicular to the family of Mode II cracks. If for a given
body and condition only Mode I cracks exist, then the coefficient values are a2 = 0
and a1 = 1. For Mode II cracks: a2 = 1, a1 = 0. When crack directions do not coincide
with crack Mode I and II, then the material isotropy and the lack of mean stress
influence are assumed, i.e. a1 = a2 = 0. In the case of uniaxial tension-compression
test with stress ratio R = σmin /σmax = 0 and assuming the existence of Mode I and II
cracks. Equation (28) has the following form
+ ,
σeq,a = 21 σmax λ1 + a1 λ2 + 21 a2 λ3 . (29)
Direction of vector n coincides with the maximum normal stress direction, and it
means that, for tensile phase of loading, the coordinates of vector n[n1 , n2 ] are n1 = 1
and n2 = 0, and for compressive phase of loading, n1 = 0 and n2 = 1. Direction of
vector m coincides with
√ the maximum shear stress direction and its coordinates are
equal to m1 = m2 = 2/2. The proposed criterion is equivalent to Huber – Mises –
Hencky’s criterion in the case of loading with zero mean stress value. The critical
plane in that case is the octahedral plane. In the case of loading under non-zero
mean stress value, the criterion distinguishes the two additional critical planes with
the orientation determined by vectors n and m on which the mean normal stresses
are computed (the influence of mean shear stresses is neglected). In general the cri-
terion proposed by Zolochevski, Obataya and Betten is based on the three differ-
ent orientations of the critical planes. Equivalent stress amplitude (Equation (28)) is
computed by summing the parameters determined for each orientation of the criti-
cal planes. Zolochevski et al. assumed the fatigue damage evolution equation in the
following form
- .
dω σeq,a − b2 b4
= (1 − ω)(1−b1 )
f (σeq,a ), f (σeq,a ) = , (30)
dN b3 − σeq,a
where ω ∈ [0, 1] characterizes the damage state, b1 , b2 , b3 , b4 are materials constants.
For constant amplitude loading fatigue Equation (30) can be integrated to compute
the number of cycles to failure Nf .

2.12. Carpinteri and spagnoli criterion (2001), (carpinteri and spagnoli, 2003;
carpinteri et al., 2003; spagnoli, 2001)

The criterion proposed by Carpinteri and Spagnoli links to paper (Carpinteri et al.,
1999a, b) since the critical plane orientation is correlated with the averaged princi-
pal stress directions 1̂, 2̂, 3̂. The averaged prinipal stress directions are computed using
Multiaxial fatigue failure criteria of metallic materials 279

a weight function which depends on the maximum principal stress σ1 (t) and two
material parameters deduced from the S–N curve for fully reversed axial loading.
Carpinteri and Spagnoli proposed to compute the critical plane orientation n with
respect to the averaged maximum principal stress direction 1̂ in the plane of 1̂, 3̂ by
the following relationship
/ - .2 0
3 τaf
*α = 45 1 − , (31)
2 σaf

where the angle *α is expressed in degrees. According to Equation (31), the √angle
*α is equal to 0◦ for τaf /σaf = 1 (hard metals) and *α = 45◦ for τaf /σaf = 3/3
(border between hard and mild metals). The equivalent stress amplitude is computed
from
1
- .2
2
σaf 2 .
σeq,a = σn,max + τns,a (32)
τaf
Authors did not propose any fatigue characteristic (Nf − σeq,a ). The equivalent stress
amplitude σeq,a was compared only with the fatigue limit σaf . The criterion initially
proposed for cyclic loading has been expanded to random loading (Carpinteri, et al.,
2003).

2.13. Küppers and sonsino criterion (konos) (2003), (kueppers and sonsino,
2003)

Küppers and Sonsino proposed a criterion for plane stress state including phase shift
δ between normal and shear stress
S(ϕ ∗∗ , δ)
σeq (δ) = σeq (δ = 0◦ ) , (33)
S(ϕ ∗ , δ = 0◦ )
where ϕ is the angle determining the plane with normal vector n, S is the parameter
based on Huber – Mises – Hencky’s criterion
2
S(ϕ, δ) = σn2 (ϕ, δ) + 3τns
2 (ϕ, δ). (34)

The expression S(ϕ ∗∗ , δ) represents the value of S parameter computed for the phase
shift δ and plane orientation ϕ ∗∗ for which S parameter is maximum. Analogously
for expression S(ϕ ∗ , δ = 0◦ ), where angle ϕ ∗ determines the plane orientation under
loading with assumed δ = 0◦ phase shift of stresses. The equivalent stress σeq (δ = 0)
is computed on the basis of Huber – Mises – Hencky’s criterion under loading with
zero phase shift
2
σeq (δ = 0◦ ) = σx2 + σy2 − σx σy + 3τxy
2 . (35)

For proportional loading (δ = 0◦ ), the proposed criterion (Equation (33)) is equal to


the Huber–Mises–Hencky criterion (Equation (35)) because ϕ ∗ = ϕ ∗∗ . In that case of
loading, the critical plane coincides with the octahedral plane. For non-proportional
loading, the parameter S(ϕ ∗∗ , δ) is smaller than S(ϕ ∗ , δ = 0◦ ). As a consequence of the
280 Aleksander Karolczuk and Ewald Macha

inequality S(ϕ ∗∗ , δ) < S(ϕ ∗ , δ = 0◦ ), the equivalent stress σeq (δ) is smaller than σeq (δ =
0◦ ). Hence, the fatigue life Ncal calculated according to σeq (δ) is greater than fatigue
life Ncal calculated using σeq (δ = 0◦ ). In general, the proposed criterion corresponds
the Huber–Mises–Hencky’s criterion modified by the coefficient taking into account
the phase shift of stresses. To compute the value of S(ϕ ∗∗ , δ)/S(ϕ ∗ , δ = 0◦ ) parameter,
it is necessary to determine two values of angle ϕ: ϕ ∗ and ϕ ∗∗ . These angles can be
identified with the critical plane orientations but only as the planes where to com-
pute the value of the damage parameter (Equation (33)), and not with the fracture
plane orientation. Unfortunately, authors did not propose any formula for fatigue
life calculation using the fatigue parameter (33). Only correlation between the fatigue
parameter and experimental fatigue life was investigated. The proposed criterion was
verified on the basis of fatigue tests related to aluminum welded joints under com-
bined bending and torsion with phase shifts δ = 0 and δ = π/2, respectively.

2.14. Summary

Among stress criteria based on the critical plane approach we can distinguish crite-
ria which assume that fatigue failure is due to (i) the linear combination of shear
τns and normal σn stresses acting on the critical plane; (ii) the linear combination of
shear parameter (τns or Ta ), acting on the critical plane, with hydrostatic stress σh ;
(iii) the nonlinear combination of shear τns and normal σn stresses acting on the crit-
ical plane.
The criteria proposed by: Findley, McDiarmid, Matake, Macha and Robert belong
to the first group. These criteria vary in the critical plane orientations and mate-
rial constants. Findley assumed that the critical plane orientation coincides with the
plane orientation, where the linear relation τns,a + kσn,max achieves maximum value.
McDiarmid assumed that the critical plane, used in the linear relation similar to
Findley’s τns,a + kσn,max , was the plane of the maximum shear stress τns . Matake like
McDiarmid assumed that shear stress τns determines the critical plane orientation,
but since there are two planes of the maximum values of shear stresses, the criti-
cal one is the plane where the normal stress value σn is higher. Macha assumed the
critical plane orientation can be determined in reference to the averaged principal
stress directions, and the fatigue parameter in the criterion is given by Bτns + Kσn .
It is a very general assumption, because the directions of averaged principal stresses
depend on the weight functions, and orientation of the critical plane can coincide e.g.
with the maximum shear stress plane. Robert assumed that the critical plane is the
plane where the linear relation ∥τns (t)∥ + a1 σn (t) + a2 σn,m reaches the maximum value,
likewise in Findley criterion. If we assume that a1 = a2 = 0.5k, then we will obtain
the criteria proposed by Findley, McDiarmid and Matake. The material coefficients
k, B, K, a1 , a2 used in the first group of the linear criteria are important parameters.
Findley and Robert assumed that material coefficients are a function of the number
of cycles to fracture and the type of material, but the relation between number of
cycles to fracture and the value of material coefficients was not proposed in Findley
criterion. In McDiarmid criterion, k coefficient depends on a type of material and
cases A and B of fatigue cracks. Coefficient k depends only on the type of material in
Matake proposal. Material constants B, K in Macha proposal depend on the choice
Multiaxial fatigue failure criteria of metallic materials 281

of the particular criterion version, which depends on the type of material. For exam-
ple: B = 0 and K = 1 for brittle materials.
Other linear criteria proposed by Dang Van and Papadopoulos, assume that two
quantities have an influence on fatigue fracture. One of them is the hydrostatic stress
value σh , which is independent of the critical plane orientation and the other one is
a shear stress parameter. In Dang Van criterion, such a parameter is the microscopic
shear stress τµ in the plane of microscopic maximum shear stress. In Papadopoulos
criterion, it is the quantity Ta , which is connected with the mean square value of
shear stress amplitude τns,a in the critical plane. The critical plane is the plane where
Ta achieves its maximum value. Material coefficients in both criteria depend on mate-
rial properties.
The third group is characterized by nonlinear combinations of shear τns and nor-
mal σn stresses acting on the critical plane. The most popular nonlinearity adopted
in these criteria comes from Huber–Mises–Hencky’s (HMH) 2 hypothesis. The stress
version of the HMH hypothesis used in plane stress state σxx 2 + 3τ 2 was applied
xy
3
2 2
to the stresses acting on the critical plane σn + kτns . In Carpinteri et al. criterion,
the critical plane orientation is correlated with the averaged principal stress direc-
tions 1̂, 2̂, 3̂, and the material coefficient depends on the kind of material. In Küppers
and
3 Sonsino criterion, the critical plane position is the plane where the parameter
σn2 + 3τns2 achieves maximum value. The value of this parameter depends only on

the history of stresses.


There are three stress criteria, described above which cannot be easily assigned to
one of the mentioned groups. A very interesting criterion was proposed by Dietman
et al., where the critical plane is the octahedral plane at time tmax , for which octa-
hedral shear stress τoct achieves the maximum value. In every time instant, the angle
φ between fixed direction of maximum octahedral shear stress τoct,max and direction
of instant octahedral shear stress τoct is computed. The value of this angle is used
to determine the damage parameter: τoct (φ) = cos(φ)τoct . This criterion is an inter-
esting compromise between Huber–Mises–Hencky’s (HMH) hypothesis (octahedral
shear stress τoct ) and the critical plane approach. Unfortunately, the criterion pro-
posed by Dietman et al. was not verified enough. Simbürger and Grubisic proposed
a criterion, in which the damage parameter does not depend on the critical plane ori-
entation. However, the authors defined the fracture plane orientation
' ' as a plane with
maximum value of the linear relation a1 τns,a + a2 σn,a − m(a1 'τns,m ' + a2 σn,m ). Mate-
rial constants a1 , a2 , m depend on the type of material. In Zolochevski et al. criterion
only the mean stress values σij,m are reduced to the critical plane. 2 The other com-
ponents of stress tensor are reduced to the stress intensity, σ̄a = 23 sij,a sij,a (HMH
hypothesis), which does not depend on the critical plane orientation.
In Table 1, fatigue failure criteria based on stresses are listed together with the
used critical plane orientation, successfully applied loading and included or not (in
the criterion) mean stress value. If more than one critical plane orientation is marked
in Table 1, it means that the critical plane orientation depends on e.g. type of mate-
rial. There is no universal criterion based on stresses, and the successful application
of a particular stress criterion, in a large degree, depends on experimentally estab-
lished material coefficients. The most promising fatigue criteria seem to be those
which can be used under the most general type of loading, i.e. the multiaxial random
Table 1. Critical plane criteria based on stresses.

No. Criterion Critical plane ori- Successfully applied under loading Mean
entation based on stress
stress crack Mode
I II/III mixed

1 Maximum normal stress + − − Cyclic, proportional (Sonsino, 1995) −


(Rankine)
2 Maximum shear stress − + − Cyclic, proportional (Sonsino, 1995) −
(Tresca)
3 Octahedral shear stress − − + Cyclic, proportional (Sonsino, 1995; Spagnoli, 2001; Park and Nelson, 2000) −
(HMH)
4 Findley (1956) − + + Cyclic, proportional (Findley, 1959; Alfredsson and Olsson, 2001; Spagnoli, 2001) +
5 McDiarmid (1972) − + − Cyclic, proportional −
(1990) − + − Cyclic, non-proportional (McDiarmid, 1985, 1987, 1990; Park and Nelson, 2000; +
Spagnoli, 2001)
282 Aleksander Karolczuk and Ewald Macha

6 Dietman and Issler − − + Cyclic, non-proportional (Dietmann et al., 1976) +


(1974)
7 Simbürger and Grubisic + + + Cyclic, non-proportional (Grubisic and Simbürger, 1976) +
(1976)
8 Matake (1977) − + − Cyclic, proportional (Spagnoli, 2000) −
9 Macha (1979) + + + Random, non-proportional (Lagoda and Macha, 1994, 1997) −
10 Dang Van (1982) − + − Cyclic, non-proportional (Froustey and Lasserre, 1989; Dang Van et al., 1989) +
11 Robert (1992) − + + Random, proportional (Vidal et al., 1996; Weber et al., 1999b) +
12 Papadopoulos (1993) − + − Cyclic, non-proportional (Papadopoulos et al., 1997; Papadopoulos, 2001) −
13 Zolochevski et al. (2000) + + + Cyclic, uniaxial (Zolochevsky et al., 2000) +
14 Carpinteri et al. (2001) + + + Cyclic, non-proportional, (Carpinteri and Spagnoli, 2001; Spagnoli, 2001) +
(2003) + + + Random, proportional (Carpinteri et al., 2003) +
15 Küppers and Sonsino − − + Cyclic, non-proportional (Küppers and Sonsino, 2003) −
(2003)
Multiaxial fatigue failure criteria of metallic materials 283

loading. Unfortunately, only a few stress criteria were experimentally verified under
random loading. The fatigue failure criteria based on stresses are not able to take
into account the effect of cyclic hardening or softening. If the fatigue tests are car-
ried out under stress controlled system, the effect of cyclic hardening or softening is
visible only in strain history, which is not taken into account in the fatigue failure
criteria based on stresses.

3. Critical planes in criteria based on strains

3.1. Generalised hypothesis of material resistance

Like stress criteria, the first strain criteria of multiaxial fatigue were formulated on the
basis of static hypothesis of material resistance. The most popular criteria of generalised
hypothesis of material resistance are: criterion of maximum normal strain, criterion of
maximum shear strain, and criterion of maximum octahedral shear strain.

3.1.1. Criterion of maximum normal strain


According to this criterion (based on static hypothesis by de Saint–Venant), maxi-
mum normal strain ε1 is responsible for material fracture. This hypothesis of static
resistance applied to cyclic loading leads to the following formula for equivalent
strain range

*εeq = *ε1 . (36)

In this case, the critical plane is the plane with maximum normal strain range.

3.1.2. Criterion of maximum shear strain


To follow Coulomb–Tresca–Guest static hypothesis of material resistance, the
maximum shear strain ε13 is responsible for material fatigue fracture, which can be
expressed by means of strain range as follows:

*εeq *ε1 − *ε3


*ε13 = = . (37)
2 2
Consequently to this assumption, the critical plane is the plane with maximum shear
strain range *ε13 .

3.1.3. Criterion of maximum octahedral shear strain


The hypothesis of octahedral shear strain applied to fatigue loading is based on
the range of this strain

2 ! "1/2
*εeq = (*ε1 − *ε2 )2 + (*ε2 − *ε3 )2 + (*ε3 − *ε1 )2 = (38)
2(1 + ν)

3 2
= (1 + ν)*γoct .
4
and the critical plane is the octahedral plane with maximum shear strain range *γoct .
284 Aleksander Karolczuk and Ewald Macha

3.2. Brown et al. criteria (1973, 1982, 1993), (brown and miller, 1973,
kandil et al., 1987; wang and brown, 19931)

Brown and Miller proposed a criterion for multiaxial fatigue, which assumes that
fatigue life is generally a non-linear function of the strain state. The contour for a
given fatigue life may be presented as follows:
4 5
ε1 − ε3 ε1 + ε3 γ13
=f or = f [εn ] . (39)
2 2 2
The critical plane is the plane with maximum shear strain γ13 . This definition of crit-
ical plane is correct only in case of proportional loading assuming fixed principal
strain directions. The criterion written in a general way by Equation (39) was reduced
to a linear form by Kandil–Brown–Miller in 1982 (Kandil et al. 1987)
*γ13
+ S*εn = C, (40)
2
where S is a coefficient determined by experiments, called coefficient of normal strain
influence. Note that only in case of proportional loading, when shear strains and nor-
mal strains are in phase and their extremes occur at the same time instant, the appli-
cation of strain ranges in Equation (40) is justified. In order to take into account
the influence of non-proportionality and variable-amplitude strains, Wang and Brown
(1993) proposed to modify the criterion (Equation (40)) in the following way
*γns σ′
+ S*εn∗ = (1 + νe + (1 − νe )S) f (2Nf )b + (1 + νp + (1 − νp )S)εf′ (2Nf )c . (41)
2 E
The difference between the two criteria above (Equations (40) and (41)) is based on
different definitions of the normal strain range. The normal strain *εn∗ (called normal
strain excursion by authors) in Equation (41) is calculated in the plane of maximum
shear strain range *γns in the cycle with maximum range of *γns . The range of shear
strain is calculated in direction s in the plane with normal vector n. If the beginning
of the cycle is determined by point A and the end by point B, we have
*εn∗ = max (εn (t)) − min (εn (t)) = εn,max − εn,min . (42)
tA <t<tB tA <t<tB

The critical plane is one of the planes of maximum shear strain range *γns with
higher value of normal strain range *εn∗ . Equation (41) contains A and B crack cases
*γns
γeq,a = + S*εn for case A, (43)
2
*γns
γeq,a = for case B. (44)
2
For case B the authors assumed no influence of non-proportional loading (S = 0).
The criterion proposed by Wang and Brown (Equation (41)) was used by Kim et al.
(1999) for torsion, tension–compression and their combination for proportional and
non-proportional, constant–amplitude and variable-amplitude loading. They obtained
satisfactory results with different definitions of the critical plane. The calculated
fatigue life obtained in the maximum shear strain plane and in the critical plane of
maximum damage according to the Brown–Wang criterion are comparable.
Multiaxial fatigue failure criteria of metallic materials 285

3.3. Lohr and ellison criterion (1980) (lohr and ellison, 1980)

Lohr and Ellison proposed a criterion to calculate fatigue life in low-cyclic fatigue
regime. This criterion assumes that fatigue life and crack growth rate can be assessed
by a linear combination of shear γns,a and normal strain εn,a amplitudes in the crit-
ical plane

εeq,a = γns,a + kεn,a , (45)

where k is material coefficient (k = 0.2 for 1Cr–Mo–V steel). The critical plane is the
plane inclined by 45◦ to the free surface of material, and this is one of the four
planes of the maximum shear strain. This assumption concerns the case when two
principal strain axes are on the free surface of material. This orientation of the crit-
ical plane coincides with the plane of maximum shear strain γ13 when the ratio of
minimum principal strain (ε3 or ε2 ) to maximum principal strain (ε1 ) on the free
material surface ranges between −ν and +1. In the case of pure torsion when the
ratio of principal strains, ε3 /ε1 , on the material surface is equal to −1, the critical
plane is a plane determined not through maximum shear strains γ13 , but through
shear strain γ12 . Correlation of the experimental test results (fatigue life, Nf ) with the
calculated equivalent strain parameter εeq,a based of the proposed criterion was sat-
isfactory. However, authors did not propose any mathematical formula for Nf − εeq,a .
The influence of normal strain εn in the critical plane turned out to be less important
because coefficient k for the best estimation of the experimental results was equal to
about 0.2.

3.4. Socie, fatemi et al. criterion (1985, 1988) (socie et al., 1985; fatemi and
kurath, 1988; fatemi and socie, 1988; socie et al., 1987)

Socie et al. observing fatigue fractures came into conclusion similar to those by
Brown and Miller, that is, the normal strain εn in the plane of maximum shear
strain accelerates the fatigue damage process through crack opening. Crack opening
decreases the friction force between slip planes blocking crack growth. Hence, mean
normal stress σn,m in the plane of maximum shear strain amplitude γns,a was included
in the following criterion

σn,m τ′
γns,a + εn,a + = γf′ (2Nf )c + f (2Nf )b . (46)
E G
On the basis of fatigue tests related to different materials, Fatemi and Socie noticed
that Brown and Miller criterion based only on strain values does not include addi-
tional material hardening occurring under non-proportional loading. In order to take
into account this effect, they modified the Brown and Miller criterion by replacing
the value of normal strain εn in the critical plane with the maximum normal stress
value σn,max . The critical plane is the plane of maximum shear strain amplitude γns,a .
For a given number of cycles to failure Nf , the proposed criterion is given by
+ ,
γns,a 1 + nσn,max /σy = constant, (47)
286 Aleksander Karolczuk and Ewald Macha

ε (t)
n
t

ε (t)
ns s

Figure 3. Normal strain history εn (t) and shear strain history εns (t) in the critical plane.

where n is an experimental coefficient. The proposed criterion includes the mean


stress value through the maximum value of normal stress in the critical plane

σn,max = σn,a + σn,m . (48)

For low cyclic fatigue regime, the criterion is given by



+ , σ′ n σ2
γns,a 1 + nσn,max /σy = (1 + νe ) f (2Nf )b + (1 + νe ) f (2Nf )2b +
E 2 Eσy
n ε′ σ ′
+(1 + νp )εf′ (2Nf )c + (1 + νp ) f f (2Nf )b+c . (49)
2 σy

3.5. Macha criterion (1988) (macha, 1988)

Macha formulated a generalised multiaxial random fatigue criterion of the maxi-


mum shear strain εns (t) and the normal strain εn (t) in the critical plane (Figure 3).
Detailed assumptions of the proposed criterion are as follows:
– fatigue fracture occurs under the influence of normal strain εn (t) and shear
strain εns (t) in the direction s on the fracture plane with normal n,
– the direction s in the fracture plane coincides with the mean direction of the
maximum shear strain max {εns (t)}. Direction of the maximum shear strain,
s
max {εns (t)} rotates, in a general case, on the fixed plane with normal n at
s
time t. Hence, direction s is assumed to be a mean direction from all directions
of max {εns (t)} occurring in the analyzed time,
s
– for a given fatigue life, the maximum value of a linear combination of strains
εn (t) and εns (t) under random loading satisfies the following equation:

max {bεns (t) + kεn (t)} = q, (50)


t

where b, k, q are constants used to select a particular form of Equation (50).


Multiaxial fatigue failure criteria of metallic materials 287

Orientation of the critical plane is determined by normal vector n, and shear vector
s in this plane may be calculated in the same way as for the Macha stress criterion
(Equation (17)) by means of one of three methods: the weight function method, the
damage accumulation method, the variance method (Macha, 1988, 1989; Carpinteri
et al., 1999a,b).

3.6. Summary

Among strain criteria based on the critical plane approach we can distinguish crite-
ria which assume that fatigue failure depends on (i) the linear combination of shear
γns and normal εn strains acting on the critical plane; (ii) the nonlinear combination
of shear strain τns and different kind of tensile parameters acting on the critical
plane. The criteria of Brown et al. (1982, 1993), Lohr and Ellison, Socie et al. (1985),
Macha belong to the first group. In Brown et al. criteria (1982, 1993) multiaxial
fatigue is divided into two distinct classes, characterized by different directions of
crack growth: one along the surface (case A) and the other one inwards and away
from the surface (case B). The critical plane is one of the planes of maximum shear
strain range *γns with the highest value of normal strain range *εn∗ . Lohr and
Ellison assumed that the critical plane orientation is inclined by 45◦ to the free sur-
face of material, and this is one of the four planes of the maximum shear strain. This
assumption is true only for the plane state of stress. The criterion proposed by Socie
et al. (1985) is based on the critical plane of the maximum shear strain γns but the
authors included in the criterion the influence of the mean normal stress σn,m act-
ing in the critical plane. Macha criterion was proposed for random loading and is
based, in general, on the same assumption as the stress criterion proposed by him,
i.e. the critical plane orientation can be determined in reference to the averaged prin-
cipal strain directions. This criterion was verified only by comparison of the critical
plane orientations with the experimental fatigue fracture plane orientations.
Criteria proposed by Brown and Miller (1973) and Socie et al. (1988) belong to
the second group of strain criteria. Brown and Miller proposed a criterion, which
assumes that fatigue life is generally a non-linear function of the maximum shear
strain 0.5(ε1 − ε3 ) = ε13 and normal strain 0.5(ε1 + ε3 ) = εn on the plane of maximum
shear strain ε13 . A non-linear function of the two mentioned parameters was not
mathematically expressed, but only the experimentally determined contours of con-
stant life for this function were presented. A very interesting criterion was proposed
by Socie et al. (1988), which assumes that fatigue fracture is due to the shear strain
amplitude γns,a influenced by the maximum
+ normal
, stress σn,max . This assumption
leads to a non-linear function: γns,a 1 + nσn,max /σy = γns,a + nγns,a σn,max /σy . The cri-
terion proposed by Socie et al. in 1988 was verified (Table 2) with satisfactory results.
It should be noted that this criterion was more effective if orientation of the critical
plane coincided with the plane experiencing the maximum value of Socie parameter.
The criterion proposed by Brown et al. evolved from fatigue criteria (1973, 1982)
applicable only for cyclic loading to the criterion applicable for random loading.
But, as many other criteria based on strains, this criterion is not able to take into
account cyclic hardening or softening in fatigue tests carried out under strain con-
trolled system. Socie et al. criterion has this feature and includes the influence of the
normal stress on fatigue process. This criterion was not verified for random loading.
288 Aleksander Karolczuk and Ewald Macha

Table 2. Critical plane criteria based on strains.

No. Criterion Critical plane ori- Successfully applied under loading Mean stress
entation based on
crack Mode

I II/III mixed

1 Maximum normal strain + − − Cyclic, proportional −


(de Saint-Venant)
2 Maximum shear strain − + − Cyclic, proportional −
(Coulomb-Tresca-Guest)
3 Octahedral shear strain − − + Cyclic, proportional −
(HMH)
4 Brown et al. (1973) − + − Cyclic, proportional (Brown and −
Miller, 1973)
(1982) − + − Cyclic, proportional (Kandil −
et al., 1982; Wang and Yao,
2004)
(1993) − + − Random, non-proportional (Smith −
et al., 1976; Wang and Brown,
1993; Kim and Park, 1999)
5 Lohr and Ellison (1980) − + − Cyclic, proportional (Lohr and −
Ellison, 1980)
6 Socie et al. (1985) − + − Cyclic, non-proportional (Socie +
et al., 1985; Socie 1987)
(1988) − + − Cyclic, non-proportional (Fatemi +
and Kurath, 1988; Fatemi and
Socie, 1988; Das and Sivakumar,
2000*; Lykins et al., 2000; Park and
Nelson, 2000; Han et al., 2002)
7 Macha (1988) + + + Random, non-proportional −
(Bedkowski et al., 1995)


Critical plane is defined as the plane experiencing the maximum damage.

In Table 2, fatigue failure criteria based on strains are presented together with the
used critical plane orientation, successfully applied loading and included or not (in
the criterion) mean loading value. In Table 2 it can be noted that the critical plane
of maximum shear strain γns dominates. The reason of this is that the fatigue failure
criteria based on strains are usually applied at low cyclic fatigue (LCF) regime for
non brittle materials, where Mode II and Mode III of fracture dominate.

4. Critical planes in criteria based on strain energy density

The first criteria based on strain energy density came from static hypotheses of
material resistance. The most applied criteria from this group are based on the
Multiaxial fatigue failure criteria of metallic materials 289

Beltrami and Huber–Mises–Hencky’s (HMH) generalized hypotheses (Macha and


Sonsio, 1999; Macha, 2001). The Beltrami’s generalized criterion is based on total
strain energy, that is the summation of all products of stress and strain tensor com-
ponents. Similarly, the generalized Huber–Mises–Hencky’s hypothesis based on shear
strain energy takes into account the summation of all products of deviatoric part
of stress and strain tensor components. It follows that the two generalized hypoth-
eses do not belong to criteria based on the critical plane approach. In case of elas-
tic strain state, the Huber–Mises–Hencky’s generalized criterion can be described by
means of shear stress or shear strain in octahedral plane. This fact was presented
previously for stress and strain criteria. The octahedral planes calculated in stress or
strain state coincide only under elastic state of material, and only for this case the
two above generalized criteria can be applied.

4.1. Smith, watson and topper criterion (1970) (smith et al., 1970)

Smith and co-authors proposed a simple form of a damage parameter, SWT,


described as stress and strain product σmax εa for fatigue life determination under uni-
axial tension-compression.

σ2
Wa = σmax εa = f (2Nf )2b + σf′ εf′ (2Nf )b+c . (51)
E
The SWT parameter after the following modification was used under multiaxial propor-
tional and non-proportional loading for materials with cracks according to Mode I. The
modification consists in taking into consideration only stresses and strains occurring in
the critical plane. The most popular form of SWT parameter in the critical plane is that
proposed by (Socie, 1987) in the plane of maximum normal strain range *ε1

*ε1 σf 2
Wa = σn,max = (2Nf )2b + σf′ εf′ (2Nf )b+c . (52)
2 E
4.2. Nitta, ogatta and kuwabara criterion (1988), (nitta et al., 1989)

Nitta, Ogata and Kuwabara applied parameters based on strain energy density to
calculate fatigue life depending on crack modes and loading type. For proportional
loading and crack propagation according to Mode I, they proposed normal strain
energy density calculated as the product of maximum normal stress amplitude *σ1 /2
and maximum normal strain range *ε1 .
*σ1 −β
*WI = *ε1 = A1 Nf 1 , (53)
2
where A1 and β1 are material constants.
For proportional loading and crack propagation according to Mode II, they pro-
posed shear strain energy density calculated as product of maximum shear stress
range *τ13 and maximum shear strain range *γ13 .
−β2
*WII = *τ13 *γ13 = A2 Nf , (54)

where A2 and β2 are material constants.


290 Aleksander Karolczuk and Ewald Macha

Generally, the directions of principal strains and stresses do not coincide. Hence,
the critical plane position is not strictly determined in that criterion.
Under non-proportional loading, cracks with mixed Modes I and II were noticed.
So, they proposed to calculate fatigue life Nou for non-proportional loading in the
following form
1 1 1
= + . (55)
Nou NI NII
Depending on phase shift of loading, the fatigue life calculated according to crack
Mode I (NI ) and crack Mode II (NI I ) is determined in a different way. For phase
shift equal to π/2, cracks were perpendicular to longitudinal specimen axis, and nor-
mal strain energy density was calculated according to the following equation

*WI = *ε, (56)
2
and shear strain energy density was calculated according to
*WII = *τ *γ . (57)
The numbers of cycles to failure calculated according to Mode I (Equation (53)) and
Mode II (Equation (54)) are then transformed according to Equation (55) in order
to calculate fatigue life Nou . For the tested phase shifts (π/6, π/4 and π/3) the frac-
ture planes coincide with the planes of maximum shear strain range *γ13 , and fatigue
parameters are calculated according to
*σn
*WI = *εn , *WI I = *τ13 *γ13 . (58)
2
The critical plane is the plane of maximum shear strain range *γ13 . Fatigue param-
eters were determined on the basis of observed crack propagation of 304 stainless
steel tested at high temperature of 550 ◦ C under low-cyclic regime (LCF). For propor-
tional loading, different crack Modes depending on strain ratio *γ /*ε were noticed.
Mode I occurred for *γ /*ε < 1.7, and Mode II for *γ /*ε ≥ 1.7. For non-propor-
tional loading fatigue cracks were of mixed mode.

4.3. Liu criterion (1993) (liu, 1993; liu and wang, 2001)

Liu proposed an energy method to estimate the fatigue life, based on virtual strain
energy (VSE). The parameters of virtual strain energy are associated with two differ-
ent Modes of fatigue cracks. The first parameter is the sum of the maximum nor-
mal strain energy density (*Wn )max and the shear strain energy density *Wns , in the
plane of maximum normal strain energy density (*Wn )max , and is used for fatigue
life determination according to Mode I fracture
*WI = (*Wn )max + *Wns = (*σn *εn )max + *τns *εns . (59)
The critical plane is the plane on which the product (*σn *εn ) achieves the maxi-
mum value. The second parameter is the summation of the maximum shear strain
energy density (Wns )max and the normal strain energy density Wn in the plane of max-
imum shear strain energy density (Wns )max , and is used for fatigue life determination
Multiaxial fatigue failure criteria of metallic materials 291

of materials with cracks according to Mode II. In the material cracking according to
Mode II, we can distinguish Case A and Case B of fatigue fractures, and two other
parameters were formulated:
Mode II and case A
(60)
*WII,A = (*Wns )max + *Wn = (*τns *εns )max + *σn *εn ,

Mode II and case B


(61)
*WII,B = (*Wns )max + *Wn = (*τns *εns )max + *σn *εn .
The critical plane is the plane on which the product (*τns *εns ) achieves the maxi-
mum value. Despite formal similarity of presented formulas for *WII,A and *WII,B ,
shear and normal stresses and strain ranges are calculated according to Mohr’s cycle
in a different way: for Case A, from maximum and minimum normal stresses and
strains σ1 , σ3 and ε1 , ε3 ; for Case B, from σ1 , σ2 and ε1 , ε2 . In uniaxial stress state,
the parameter VSE can be written as follows:
+ ,
*W = *σ · *ε = E*εe *εp + *εe = *Wp + *We . (62)
The parameter VSE contains the elastic *We and plastic strain *Wp energy density.
It can be described by the area of the rectangles defined by the hysteresis loop ranges.
In uniaxial stress state, the energy density *W is equivalent to the Smith–Watson–
Topper parameter and, as a function of fatigue life Nf , it is written as follows:
*εp = B1 Nf−a1 , *εe = B2 Nf−a2 , (63)

*W = *Wp + *We = EB1 B2 Nf−(a1 +a2 ) + EB22 Nf−2a2 , (64)


where B1 , B2 , a1 , a2 are materials constants. The described procedure for fatigue life
calculation and energy parameter determination is correct only for proportional load-
ing since stress and strain tensor components are in phase. Hence, it is easy to find
the time instant when the product of stress and strain is maximum. For that time
instant, the VSE parameter is calculated. Under non-proportional loading, the maxi-
mum values of stress and strains are not in phase. For this case of loading, Liu pro-
posed another definition of the VSE parameter. Biaxial sinusoidal cyclic loading can
be described in a plane of loading by a path covered in general by a rotated ellipse.
Liu assumes that this loading path can be replaced by two virtual paths (path P1
and P2) of proportional loading (Figure 4). The paths are defined by the two axes
of ellipse. The VSE parameter is the summation of two VSE parameters computed
from two virtual paths in the same critical plane:
Mode I
(65)
*ŴI = *ŴI (P1) + *ŴI (P2),

Mode II
(66)
*ŴII = *ŴII (P1) + *ŴII (P2).
The criterion proposed by Liu is limited only to some special kind of loading. We
cannot agree that all non-proportional paths of loading are represented by a rotated
ellipse. Liu did not define the fatigue parameter for random loading or even for tri-
axial state of stress.
292 Aleksander Karolczuk and Ewald Macha

Figure 4. Replacement of non-proportional loading path (ellipse) by two virtual paths P1 and P2 of
proportional loading.

4.4. Glinka et al. criterion (1995, 1999), (glinka et al., 1995a,b;


pan et al., 1999)

Glinka et al. proposed an energy parameter, being a part of total strain energy den-
sity, expressed by stress and strain in the critical plane
*γns *τns *εn *σn
Wa∗ = + . (67)
2 2 2 2
This came from Brown and Miller criterion (Brown and Miller, 1973), where fatigue
life Nf is a function of normal and shear strains in the critical plane. Instead of shear
and normal strains, authors apply shear and normal strain energy density in the crit-
ical plane. The critical plane in both criteria is the plane of maximum shear strain
range *γns . In order to take into account the mean value, authors (Glinka et al.,
1993b) have modified the above criterion, Equation (67) as follows
4 5
*γns *τns 1 1
Wa∗ = + , (68)
2 2 1 − τns,max /τf′ 1 − σn,max /σf′
where τns,max and σn,max are maximum absolute values of shear and normal strains
in the critical plane of maximum shear strain range *γns . In 1999 (Pan et al., 1999)
noticed that the influence of strain energy in shear direction, (*γns /2)(*τns /2), on
fatigue life is different than the influence of strain energy calculated in normal direc-
tion, (*εn /2)(*σn /2). For this reason, they proposed to modify Glinka criterion by
applying two coefficients determined by experiments
*γns *τns *εn *σn
Wa∗ = + κ 1 κ2 , (69)
2 2 2 2
σ′ γ′
where the coefficients are equal to κ1 = τ f′ and κ2 = εf′ . The fatigue parameters (67)–
f f
(69) were used to correlate experimental fatigue life Nf . However, any formula to
determine relation between fatigue life Nf and the proposed strain energy parameters
was not proposed.

4.5. Lagoda and macha criterion (1998) (lagoda et al., 1999)

Lagoda and Macha formulated a generalised criterion of the normal Wn (t) and shear
strain energy density Wns (t) in the critical plane. Detailed assumptions of the pro-
posed criterion are as follows:
Multiaxial fatigue failure criteria of metallic materials 293

– fatigue crack is formed by the part of strain energy density which corresponds
to work Wn (t) of normal stress σn (t) on normal strain εn (t), and work Wns (t) of
shear stress τns (t) on shear strain εns (t) occurring in s direction on critical plane
with normal n,
– direction s in plane with normal n coincides with the mean direction of maxi-
mum shear strain energy density parameter, max {Wns (t)}. Direction of the max-
s
imum shear strain energy density parameter, max {Wns (t)} rotates, in a general
s
case, at time t on the fixed plane with normal n. Hence, direction s is assumed
to be the mean direction from all directions of max {Wns (t)} occurring in the
s
analyzed time,
– for a given fatigue life, the maximum value of linear combination of strain
energy density parameters Wn (t), Wns (t) under multiaxial random loading sat-
isfies the following equation

max{βWns (t) + κWn (t)} = Q or max{Weq (t)} = Q, (70)


t t

where β, κ, Q are constants to select a particular criterion version. The param-


eter of strain energy density for uniaxial stress state is defined as follows:
sgn [σ (t)] + sgn [ε(t)]
W (t) = 21 σ (t)ε(t)sgn [σ (t), ε(t)] = 21 σ (t)ε(t) . (71)
2
For multiaxial stress state history, the equivalent parameter of strain energy density
is calculated in the critical plane on normal n and shear s directions as
1 1
Weq (t) = βτns (t)εns (t)sgn [τns (t), εns (t)] + κσn (t)εn (t)sgn [σn (t), εn (t)] . (72)
2 2
Amplitude of the equivalent strain energy density parameter (72) is used to calculate
fatigue life Nf
σf′2 1
Weq,a = (2Nf )2b + σf′ εf′ (2Nf )b+c . (73)
2E 2
Orientation of the critical plane may be established by one of three methods
already described, that is: the weight function method, the variance method or the
damage accumulation method (Carpintei et al., 1999a, b; Macha 1998,1999). The cri-
terion of the normal and strain energy density in the critical plane (Equation (72))
may be used for random loading in high and low cyclic fatigue regimes.

4.6. Rolovic and tipton criterion (1999) (rolovic and tipton, 1999)

Rolovic and Tipton proposed a criterion for multiaxial cyclic proportional and non-
proportional fatigue loading including the mean value of normal stress. The criterion
in general way is written as follows
! " ! "
τns,a + f1 (σn,max ) γns,a + σn,a + f2 (σn,max ) εn,a = f3 (Nf ). (74)

The first part of Equation (74) takes into account fatigue process caused by the shear
stress and strain amplitudes, τns,a , γns,a , modified by a function f1 (σn,max ). The second
294 Aleksander Karolczuk and Ewald Macha

Wn (t)
t

Wns (t)
s

Figure 5. History of the normal strain energy density parameter Wn (t) and shear strain energy density
parameter Wns (t) in the critical plane.

part of Equation (74) takes into account the influence of strain energy density calcu-
lated in normal direction to the critical plane, with a modified function f2 (σn,max ).
The right side of Equation (74) is a function of the uniaxial energy life relation. The
specific form of Equation (74) was proposed as

σf′2
(τns,a + 0.3σn )γns,a + σn,max εn,a = (2Nf )2b + σf′ εf′ (2Nf )b+c . (75)
E
The critical plane is determined by Equation (74) as the plane with the highest calcu-
lated damage level. The model uses standard uniaxial data and requires no additional
testing to determine multiaxial parameters. Uniaxial and biaxial in phase and out-of-
phase fatigue data from three materials were used to verify the proposed model. The
proposed model can be used for multiaxial random loading. Unfortunately, this very
interesting model was very little verified (only by authors). The problem is to deter-
mine the formula of functions: f1 (σn,max ) and f2 (σn,max ) for other materials.

4.7. Chen et al. criterion (1999), (chen et al; 1999)

Chen et al. proposed two criteria: the first one for materials characterised by Mode
I crack, and the second one for materials characterised by Mode II crack. Both cri-
teria assume the influence of normal and shear stresses and strains in critical plane
on fatigue life. For materials characterised by Mode I crack, the critical plane is the
plane of maximum normal strain range *εn

σf′2
*εn *σn + *γns *τns = 4 (2Nf )2b + 4σf′ εf′ (2Nf )b+c . (76)
E
For materials with Mode II crack, the critical plane is the plane of maximum shear
strain range *γns

τf′2
*γns *τns + *εn *σn = 4 (2Nf )2b0 + 4τf′ γf′ (2Nf )b0 +c0 . (77)
E
Multiaxial fatigue failure criteria of metallic materials 295

Equation (77) is equivalent to Equation (67) which corresponds to the criterion pro-
posed by Glinka and co-authors. In the criterion the different influence of the normal
and the strain energy density on fatigue life is not included.

4.8. Hoffmeyer et al. criterion (2001), (hoffmeyer et al., 2001)

Orientation of the critical planes can be determined on the basis of fracture mechan-
ics. The critical plane can be assumed as the plane with the highest fatigue crack
growth rate. Fatigue crack growth rate can be determined on the basis of the Paris’s
law using the effective range of integral J .
da
= C1 (*Jeff )C2 , (78)
dN
where C1 , C2 are material constants. In order to take into account mixed mode
cracks, the effective range of integral, *Jeff , is calculated as a summation of an effec-
tive range of integral *JI,eff for Mode I crack and an effective integral range *JII,eff
for Mode II crack. Equations describing the calculation method of effective integral
range J were taken from Dowling’s works:
*Jeff = *JI,eff + *JII,eff , (79)

6 Mode I 7
(*σn )2 *σn *εn,p (80)
*JI,eff = 1.24 E + 1.02 √ n′
a,

6 Mode II 7
(*τns )2 *τ *γ (81)
*JII,eff = 1.24 G + 1.02 ns√n′ ns,p a,

where a is the crack length.

4.9. Varvani - farahani criterion (2002) (varvani-farahani, 2000;


varvani-farahani and topper, 2000)

Varvani-Farahani proposed a fatigue parameter as the summation of the normal


*σn *εn and shear *τns,max * γns,max 2
strain energy density ranges calculated in the criti-
cal plane of maximum shear strain at the time instant when Mohr’s circles of stresses
and strains are the largest during the loading and unloading parts of a cycle. Energies
of normal and shear strains are weighted through material coefficients. This param-
eter includes the mean value of normal stress σn,m . The proposed parameter can be
applied only for cyclic loading and plane stress state
1 1 + σn,m /σf′ 6γ 7
ns,max
*σn *εn + *τ ns,max * = f (Nf ) . (82)
σf′ εf′ τf′ γf′ 2
This criterion does not include the fact that positions of principal axes of strains and
stresses do not coincide in case of elastic–plastic strain state. The fatigue parameter
(82) was used to correlate experimental fatigue life Nf , but the function f (Nf ) was
not formulated.
296 Aleksander Karolczuk and Ewald Macha

4.10. Lee et al. criterion (2003) (lee et al., 2003)

Lee et al. proposed an energy parameter for multiaxial fatigue. The parameter
includes elastic, We∗ , and plastic Wp∗ strain energy densities in the critical plane

W ∗ = We∗ + Wp∗ . (83)

The parameter of elastic strain energy density We∗ in the critical plane is computed
from summation of normal and shear strain energy density with coefficient w
- . - .
∗ 1 *σn 2 1 *τns 2
We = +w . (84)
2E 2 2G 2
The parameter of plastic strain energy density Wp∗ in the critical plane is a sum of
normal and shear strain energy density dissipated in a cycle of loading with coeffi-
cient w
& &

Wp = p
σn dεn + w τns dγnsp . (85)
cykl cykl

Fatigue damage is computed from two equations:


+ ′ ,2
σf − σn,m

We = (2Nf )2b , (86)
2E

6 ′ 7 ′ c−b
Wp∗ = 4 σf − σn,m εf (2Nf )b+c . (87)
c+b
Equation (87) came from the area of hysteresis loop, and the quantity (c − b)/(c + b)
refers to (1 − n′ )/(1 + n) according to Ramberg–Osgood equation. In the case of con-
stant amplitude loading, the critical plane is the plane of maximum fatigue dam-
age. Under variable amplitude loading, the loading cycle was defined. This cycle was
used to determine all quantities in Equations (84) and (85). The cycles are extracted
by the rainflow (ASTM, 1997) method on the basis of two courses: according to
normal strain εn (t) and shear strain γns (t), respectively. The following analysis con-
cerns the cycles computed from shear strain history γns (t). The range of shear strain
*γns in each cycle is calculated as (γns,max − γns,min ). From elastic shear strain range,
the range of shear stress *τns is determined. The range of normal strain, *εn , for
this cycle is computed as the parameter εn∗ in Wang and Brown criterion (Wang and
Brown, 1993). Damage was computed for each cycle (Equations (86) and (87)) and
then summed according to the linear hypothesis of Palmgren–Miner. Total damage
Dγ computed from cycles determined by shear strain history γns (t) is compared with
the total damage Dε computed according to cycles determined by normal strain his-
tory εn (t)

κ= . (88)

If κ > 1, then the history of shear strain γns (t) is selected for rainflow counting. Oth-
erwise, the history of normal strain εn (t) is chosen. The critical plane orientation is
Multiaxial fatigue failure criteria of metallic materials 297

determined by the maximum damage Dγ or Dε . The parameter κ can be written as


a function of plane orientation
Dγ (n)
κ (n) = . (89)
Dε (n)
The coefficient w is determined from uniaxial fatigue tests and its value is higher
than 0.5. The criterion was verified on the basis of fatigue tests performed under
combined tension-compression and torsion with variable and constant amplitude
loading in low cycle fatigue regime.

4.11. Summary

Among strain energy density criteria based on the critical plane approach we can
distinguish criteria which assume that fatigue failure is due to (i) only the shear
strain energy density parameter Wns or the normal strain energy density parameter
Wn computed on the critical plane; (ii) the linear combination of shear Wns and nor-
mal Wn strain energy parameters computed on the critical plane. Criteria of Smith
et al. (1970, 1987), Nitta et al., belong to the first group. The SWT damage param-
eter assumes that only the parameter of normal strain energy density Wn computed
in the plane of maximum normal strain range *ε1 responds for fatigue fracture. Sat-
isfactory results were obtained also when the SWT damage parameter was computed
in the plane experiencing the maximum value of SWT damage parameter (Table 3).
Anyway, SWT damage parameter is applicable for brittle materials. Nitta et al., pro-
posed two energy parameters, one based on the normal strain energy density Wn for
materials with Mode I cracks, the other based on the shear strain energy density Wns
for materials with Mode II cracks.
Criteria proposed by Liu, Glinka et al., Lagoda and Macha, Rolovic and Tipton,
Chen, Varvani–Farahani, Lee belong to the second group. This group can be divided
into two parts, i.e. (a) criteria with a simple mathematic expression e.g: aWns + bWn ,
where a, b are material coefficients, and (b) criteria where a linear relation between
Wns and Wn is perturbed by the mean values of stresses. The fatigue failure criteria of
group (a) differ mainly in the critical plane orientation. Liu assumed that the critical
plane orientation depends on the observed crack Mode in the material being exam-
ined. For materials with Mode I cracks, the critical plane orientation coincides with
the plane of the maximum range of normal strain energy density *Wn , and for mate-
rials with Mode II cracks, the critical plane orientation coincides with the plane of
the maximum range of shear strain energy density *Wns . Additional feature of Liu
criterion is that Liu distinguishes the cases A and B of fatigue fractures in Mode II.
Glinka et al. criterion is not defined with respect to energy but with respect to the
plane of maximum shear strain range *γns . Macha and Lagoda criterion was suc-
cessful verified by assumming that the critical plane orientations coincides with the
plane of the maximum normal strain energy density parameter Wn or with the plane
of the maximum shear strain energy density parameter Wns (Table 3). The energy
parameter proposed by Macha and Lagoda includes signs of strain and stress while
determining energy and allows to distinguish tension from compression at loading
Table 3. Critical plane criteria based on strain energy density.

No. Criterion Critical plane Successfully applied under loading Mean strain
orientation based energy
on crack Mode

I II/III mixed

1 Smith et al. (SWT, 1970) + − − Cyclic, uniaxial (Smith et al., 1970) +


Socie (1987) + − − Cyclic, non-proportional (Socie, 1987; Das and Sivakumar, 2000*) +
2 Nitta et al. (1988) + + − Cyclic, non-proportional (Nitta et al., 1989) −
3 Liu (1993) + + − Cyclic, non-proportional (Liu, 1993; Liu and Wang, 2001) −
4 Glinka et al. (1995) − + − Cyclic, proportional (Glinka et al., 1995a) −
Glinka et al. (1995) − + − Cyclic, non-proportional (Wang and Yao, 2004; Glinka et al., +
1995b)
Pan et al. (1999) − + − Cyclic, non-proportional (Pan et al., 1999; Han et al., 2002*) −
5 Lagoda and Macha (1998) + + + Random, non-proportional (Lagoda et al., 1999; Lagoda 2001; −
Banvillet et al., 2004; Itoh et al., 2004)
298 Aleksander Karolczuk and Ewald Macha

6 Rolovic and Tipton (1999) − − + Cyclic, non-proportional (Rolovic et al., 1999) +


7 Chen et al. (1999) + + Cyclic, non-proportional (Chen and Huang, 1999) −
8 Hoffmeyer et al. (2001) + + − Cyclic, non-proportional (Hoffmeyer et al., 2001) −
9 Varvani-Farahani (2002) − + − Cyclic, non-proportional (Wang and Yao, 2004; Varvani-Farahani +
and Topper, 2000; Varvani-Farahani, 2000; Han and chen, 2002*)
10 Lee et al. (2003) − − + Cyclic, non-proportional (variable amplitude loading) (Lee et al., +
2003)

Critical plane is defined as the plane experiencing the maximum damage.
Multiaxial fatigue failure criteria of metallic materials 299

path. Chen proposed two fatigue parameters, one is for materials characterized by
Mode I crack, and the critical plane is the plane of maximum normal strain range
*εn . The other one is for materials characterized by Mode II crack, and the critical
plane is the plane of maximum shear strain range *γns , which corresponds to the cri-
terion proposed by Glinka and co-authors. In Lee criterion the critical plane orien-
tation is determined by the maximum damage given by the proposed parameter. The
parameter proposed by Lee is original since it is composed of two separately com-
puted parts, i.e. elastic and plastic strain energy density parameters computed in the
critical plane.
The criteria proposed by Glinka et al. (1995), Rolovic and Tipton, Varvani–Farahani
belongs to the group (b). In Glinka et al. criterion proposed in 1995, it was assumed
that the maximum absolute values of shear τns,max and normal strains σn,max in the crit-
ical plane of maximum shear strain range *γns have influence on fatigue fracture. On
the other hand, Rolovic and Tipton criterion assumes only the influence of the maxi-
mum normal stress σn,max in the critical plane with the highest calculated damage level.
Similarly, Varvani–Farahani proposed a criterion, which includes the mean values of
the normal stress σn,m in the critical plane of the maximum shear strain.
In Table 3, a fatigue failure criteria based on strain energy density are presented
together with the used critical plane orientation, successfully applied loading and
included or not (in the criterion) mean loading value. The most verified strain energy
criteria based on the critical approach are criteria of Glinka et al., Lagoda and
Macha, Varvani–Farahani. Among them only the criterion proposed by Lagoda and
Macha was verified under random loading. The fatigue failure criteria based on
energy are able to take into account the effect of cyclic hardening or softening since
they use both strain and stress histories of loading.

5. Summary of the research state

From the present review of the multiaxial fatigue failure criteria based on the critical
plane approach, it appears that the critical plane concept is very wide and variously
used by different authors and in different period of time. In Tables 1–3 a classi-
fication of multiaxial fatigue failure criteria is presented depending on the critical
plane orientation. The first models used to describe fatigue life by stress state and
the critical plane were those based on the plane of maximum value of damage
parameter (Findley). Then, criteria used a strain state to analyse fatigue process.
Maximum shear strain plane is the most often used critical plane orientation in
strain criteria. Together with development of fatigue tests, different authors proposed
parameters based on the strain energy density, which is a scalar that allows add-
ing the elements of strain energy density without a conflict with physics laws. There
were few models based on different strain energy density parameters. Generally, we
can assume that the critical plane orientation in multiaxial fatigue failure criteria
using strain energy density is connected with a part of shear strain energy density
and normal strain energy density. The maximum normal stress plane, maximum nor-
mal strain plane or the plane of maximum normal strain energy density are used
in multiaxial fatigue failure criteria based on the critical plane for materials and
loading with dominating Mode I cracks. For materials and loading characterized by
Mode II cracks, the multiaxial fatigue failure criteria are based on the critical plane
300 Aleksander Karolczuk and Ewald Macha

determined as maximum shear stress plane, maximum shear strain plane or plane
of maximum shear strain energy density. For materials and loading characterized by
mixed Mode cracks, the octahedral or different planes were accepted. There were also
proposals of fatigue life calculations using two critical planes. First, the critical plane
connected with Mode II and, secondly, the critical plane connected with Mode I,
on which damage levels were calculated, respectively. Then, both damage levels were
summed.
Through the literature review of the critical plane approach to multiaxial fatigue,
the following comments can be drawn:
1. Along with science development of multiaxial fatigue failure criteria, the mean-
ing of basic quantities as maximum value, amplitude and range of stresses and
strains were changing. The methodologies applied to the cases of static or pro-
portional loading turned out to be false for non-proportional loading.
2. The rotation of principal stress and strain directions is not always taken into
account.
3. In a few criteria of multiaxial fatigue failure, it is assumed that the critical
plane position is defined in relation to the averaged directions of principal
stresses and strains
4. There is not enough information concerning the influence of mean stress and
strain on the critical plane and fracture plane orientation.
5. Information about fatigue fracture plane positions under multiaxial random
loading is very poor.
6. There are attempts to include different mechanisms of fatigue process (Mode I,
II, III, stage I, stage II, etc.) to fatigue life calculation through:
– application of several criteria based on different fracture mechanism;
fatigue lifetime is established by a criterion based on the highest damage
level (Das and Sivakumar, 1999),
– summation of damage levels calculated according to several multiaxial cri-
teria (Socie, 1987, Nitta et al., 1989).
7. One part of the multiaxial criteria considers a few critical plane orientations
for calculation of a single damage parameter (Küppers and Sonsino 2003;
Zolochevski et al., 2000).

References

Alfredsson, B. and Olsson. M. (2001). Applying multiaxial fatigue criteria to standing contact fatigue.
International Journal of Fatigue 23, 533–548.
ASTM E 1049-85 (REAPPROVED 1997): Standard practices for cycle counting in fatigue analysis. In:
Annual Book of ASTM Standards, Vol. 03.01, Philadelphia (1999), pp. 710–718.
Banvillet, A., Lagoda, T., Macha, E., Nieslony, A., Palin-Luc, T. and Vittori, J.-F. (2004). Fatigue life
under non-Gaussian random loading from various models. International Journal of Fatigue 26, 349–
363.
Backstrom, M. and Marquis, G. (2001). A review of multiaxial fatigue of weldments: experimental
results, design code and critical plane approaches. Fatigue of Engineering Materials and Structures
24, 279–291.
Bedkowski, W., Macha, E., Ohnami, M. and Sakane, M. (1995). Fracture plane of cruciform specimen
in biaxial low cycle fatigue – estimate by variance method and experimental verification. Journal of
Engineering Materials and Technology 117, 183–190.
Multiaxial fatigue failure criteria of metallic materials 301

Berard, J.Y., McDowell, D.L. and Antolovich, S.D. (1993). Damage observation of a low-carbon steel
under tension-torsion low cycle fatigue. In Advances in Multiaxial Fatigue (Edited by McDowell,
D.L., R. Ellis, R.). American Society for Testing and Materials STP 1191. Philadelphia pp. 326–344.
Brown, M.W. and Miller, K.J. (1973). A theory for fatigue failure under multiaxial stress–strain condi-
tions, Proceedings of the Institute of Mechanical Engineers 187, 745–755.
Brown M.W. and Miller, K.J. (1979). Initiation and growth of cracks in biaxial fatigue. Fatigue of Engi-
neering Materials and Structures 1, 231–246.
Carpinteri, A., Macha, E., Brighenti, R. and Spagnoli, A. (1999a). Expected principal stress directions
under multiaxial random loading. Part I: Theoretical aspects of the weight function method. Inter-
national Journal of Fatigue 21, 83–88.
Carpinteri, A., Macha, E., Brighenti, R. and Spagnoli, A. (1999b). Expected principal stress direc-
tions under multiaxial random loading. Part II: Numerical simulation and experimental assessment
through the weight function method. International Journal of Fatigue 21, 89–96.
Carpinteri A. and Spagnoli, A. (2001). Multiaxial high-cycle fatigue criterion for hard metals. Interna-
tional Journal of Fatigue 23, 135–145.
Carpinteri, A., Spagnoli A. and Vantadori, A.S. (2003). A multiaxial fatigue criterion for random load-
ing. Fatigue of Engineering Materials and Structures 26, 515–522.
Chen, X., Xu, S. and Huang, D. (1999). A critical plane-strain energy density criterion for multiaxial
low-cycle fatigue life under non-proportional loading. Fatigue of Engineering Materials and Struc-
tures 22, 679–686.
Constantinescu, A., Dang Van, K. and Maitournam, H. (2003). A unified approach for high and low cycle
fatigue based on shakedown concepts. Fatigue of Engineering Materials and Structures 26, 561–568.
Cox, H.L. and Field, J.E. The initiation and propagation of fatigue cracks in mild steel pieces of square
section. The Aeuronautical Quarterly IV, 1–19.
Dang Van, K. (1983). Macro-micro approach in high-cycle multiaxial fatigue. In Advances in Multiaxial
Fatigue. (Edited by McDowell, D.L. and Ellis, R.) American Society for Testing and Materials STP
1191. Philadelphia (1993) pp. 120–130
Dang Van, K., Cailletaud, G., Flavenot, J.F., Le Douaron, A. and Lieurade, H.P. (1989). Criterion for
high cycle fatigue failure under multiaxial loading. In Biaxial and Multiaxial Fatigue (edited by
Brown, M. and Miller, K.J.). Mechanical Engineering Publications, London, pp. 459–478.
Das, J. and Sivakumar, S.M. (1999). An evaluation of multiaxial assessment methods for engineering
components. International Journal of Pressure Vessels and Piping 76, 741–646.
Das, J. and Sivakumar, S.M. (2000). Multiaxial fatigue life prediction of a high temperature steam tur-
bine rotor using a critical plane approach. Engineering Failure Analysis 7, 347–358.
Dietmann, H., Bhongbhibhat, T. and Schmid, A. (1976). Multiaxial fatigue behaviour of steels under
in-phase and out-of-phase loading, including different wave forms and frequencies. In K. Kussmaul,
D. McDiarmid, and D. Socie (eds.) Fatigue Under Biaxial and Multiaxial Loading, ESIS10 (edited
by Kussmaul, D.). Mechanical Engineering Publications, London, pp. 449–469.
Fatemi, A. and Kurath, P. (1998). Multiaxial fatigue life predictions under the influance of mean-stresses.
Transaction of ASME Journal of Engineering and Technology 110, 380–388.
Fatemi, A. and Socie, D.F. (1988) A critical plane approach to multiaxial fatigue damage including out-
of-phase loading. Fatigue of Engineering Materials and Structures 11, 149–165.
Findley, W.N. (1959). A theory for the effect of mean stress on fatigue of metals under combined torsion
and axial load or bending. Journal of Engineering for Industry November, 301–306.
Forsyth, P.J.E. (1961). A two-stage process of fatigue crack growth. Proceedings of the Symposium on
Crack Propagation, Cranfield, England, 76–94.
Froustey, C. and Lasserre, S. (1989). Multiaxial fatigue endurance of 30NCD16 steel. International Jour-
nal of Fatigue 11, 169–175.
Glinka, G., Shen, G. and Plumtree. A. (1995a). A multiaxial fatigue strain energy density parameter
related to the critical fracture plane. Fatigue of Engineering Materials and Structures 18, 37–46.
Glinka, G., Shen, G. and Plumtree, A. (1995b). Mean stress effects in multiaxial fatigue. Fatigue of Engi-
neering Materials and Structures 18, 755–764.
Grubisic, V. and Simbürger, A. (1976). Fatigue under combined out-of-phase multiaxial stresses. Inter-
national Conference on Fatigue, Testing and Design. Society of Environmental Engineers, London,
8 pp.
302 Aleksander Karolczuk and Ewald Macha

Han, C., Chen, X. and Kim, K.S. (2002). Evaluation of multiaxial fatigue criteria under irregular load-
ing. International Journal of Fatigue 24, 913–922.
Hoffmeyer, J., Döring, R., Seeger, T. and Vormwald, M. (2001). Short fatigue crack growth under mul-
tiaxial nonproportional loading. In 10th International Conference on Fracture. Elsevier Science, CD,
8 pp.
Itoh, T., Karolczuk, A., Lachowicz, C.T. and Macha, E. (2004) Energy models of fatigue life of steels
and an aluminium alloy under nonproportional loading, Proceedings of the 7th International Confer-
ence on Biaxial/Multiaxial Fatigue and Fracture, DVM, Berlin, pp. 57–62.
Kandil, F.A., Brown M.W. and Miller, K.J. (1988). Biaxial low-cycle fatigue fracture of 316 stainless steel
at elevated temperatures. Book 280. The Metals Society, London, pp. 203–210.
Kenmeugne, B., Weber, B., Carmet, A. and Robert, J.L. (1997). A stress-based approach for fatigue
assessment under multiaxial variable amplitude loading. In Proceedings of 5th International Confer-
ence on Biaxial/Multiaxial Fatigue & Fracture (Edited by Macha, E. and Mróz, Z.). Technical Uni-
versity of Opole, pp. 557–573.
Kim, K.S. and Park, J.C. (1999) Shear strain based multiaxial fatigue parameter applied to variable
amplitude loading. International Journal of Fatigue 21, 475–483.
Küppers, M. and Sonsino, C.M. (2003). Critical plane approach for the assessment of the fatigue behav-
iour of welded aluminium under multiaxial loading. Fatigue of Engineering Materials and Structures
26, 507–513.
Lagoda, T. and Macha. E. (1997). Estimated and experimental fatigue lives of 30CrNiMo8 steel under
in- and out-of-phase combined bending and torsion with variable amplitudes. Fatigue & Fracture of
Engineering Materials & Structures 17, 1307–1318.
Lagoda, T. and Macha. E. (1997). Fatigue life under biaxial stress state with different cross-correlation
coefficients of normal stresses. In Proceedings of 9th International Conference of Fracture, Vol. 3.
(Edited by Karihaloo, B.L., Mai, Y-W., Ripley, M.I. and Ritchie R.O.) Pergamon, pp. 1371–1377.
Lagoda, T., Macha, E. and Bedkowski, W. (1999). A critical plane approach based on energy concepts:
Application to biaxial random tension-compression high-cycle fatigue regime. International Journal
of Fatigue 21, 431–443.
Lagoda, T. (2001). Energy models for fatigue life estimation under uniaxial random loading. Part II:
Verification of the model. International Journal of Fatigue 23, 481–489.
Lee, B.L., Kim, K.S. and Nam, K.M. (2003). Fatigue analysis under variable amplitude loading using
an energy parameter. International Journal of Fatigue 25, 621–631.
Liu, K.C. (1993). A method based on virtual strain-energy parameters for multiaxial fatigue life predic-
tion. in Advances in Multiaxial Fatigue (edited by McDowell, D.L. and Ellis, R.). American Society
for Testing and Materials STP 1191, Philadelphia, pp. 67–84.
Liu, K.C. and Wang, J.A. (2001). An energy method for predicting fatigue life, crack orientation, and
crack growth under multiaxial loading condition. International Journal of Fatigue 23, S129–S134.
Lohr, R.D. and Ellison, E.G. (1980). A simple theory for low cycle multiaxial fatigue. Fatigue of Engi-
neering Materials and Structures 3, 1–17.
Lykins, C.D., Mall, S. and Jain, V. (2000) An evaluation of parameters for predicting fretting fatigue
crack initiation. International Journal of Fatigue 22, 703–716.
Macha, E. (1979). Mathematical models of the life to fracture for materials subjected to random complex
stress systems. Scientific Papers of the Institute of Materials Science and Applied Mechanics of Wro-
claw Technical University 41, Monographs 13, Wroclaw (1979) 99 pp. (in polish)
Macha, E. (1989). Simulation investigations of the position of fatigue plane in materials with biaxial
loads. Material-wissenschaft und Werkstofftechnik 20, 132–136 and 153–163.
Macha, E. (1998). Generalization of Strain Criteria of Multiaxial Cyclic Fatigue To Random Loadings.
Technical University of Opole, Vol. 23, Opole, 89 pp. (in polish)
Macha, E. and Sonsino, C.M. (1999). Energy criteria of multiaxial fatigue failure. Fatigue of Engineering
Materials and Structures 22, 1053–1070.
Macha, E. (2001). A review of energy-based multiaxial fatigue failure criteria. The Archive of Mechanical
Engineering XLVIII, 71–101.
Marquis, G.B. and Karjalainen-Roikonen, P. Long-life multiaxial fatigue of SG cast iron. In Proceedings
of 6th International Conference on Biaxial/Multiaxial Fatigue & Fracture (Edited by Freitas, M.). In-
stituto Superior Tecnico, Lisboa, pp. 151–158.
Multiaxial fatigue failure criteria of metallic materials 303

Matake, T. (1977). An explanation on fatigue limit under combined stress. Bulletin of the The Japan
Society of Mechanical Engineers 20, 257–263.
McDiarmid, D.L. (1985). Fatigue under out-of-phase biaxial stresses of different frequencies. In Multiax-
ial Fatigue. (Edited by Miller, K.M. and Brown, M.W) American Society for Testing and Materials
STP 853, 606–621.
McDiarmid, D.L. (1987). Fatigue under out-of-phase bending and torsion. Fatigue of Engineering Mate-
rials and Structures 9, 457–475.
McDiarmid, D.L. (1990). A general criterion for high cycle multiaxial fatigue failure. Fatigue of Engi-
neering Materials and Structures 14, 429–453.
Nitta, A. Ogata, T. and Kuwabara, K. (1989). Fracture mechanisms and life assessment under high-strain
biaxial cyclic loading of type 304 stainless steel. Fatigue of Engineering Materials and Structures 12,
77–92.
Ogata, T., Nitta, A. and Kuwabara, K. (1989). Biaxial low cycle fatigue failure of type 304 stainless
steel under in-phase and out-of-phase straining conditions. In Fatigue under Biaxial and Multiaxial
Loading (Edited by kussmaul, K.F., McDiarmid, D.L. and Socie D.S.). MPA Univ. Stuttgart, pp.
377–392.
Pan, W., Hung, C. and Chen, L. (1997). Fatigue life estimation under multiaxial loadings. International
Journal of Fatigue 21, 3–10.
Papadopoulos, I.V. (2001). Long life fatigue under multiaxial loading. International Journal of Fatigue
23, 831–849.
Papadopoulos, I.V., Davoli, P., Gorla, C., Filippini, M. and Bernasconi, A. (1997). A comparative study
of multiaxial high-cycle fatigue criteria for metals. International Journal of Fatigue 19, 219–235.
Park, J. and Nelson, D. (2000). Evaluation of an energy-based approach and a critical plane approach
for predicting constant amplitude multiaxial fatigue life. International Journal of Fatigue 22, 23–39.
Rolovic, R. and Tipton, S.M. (1999). An energy based critical plane approach to multiaxial fatigue analy-
sis, In Fatigue and Fracture Mechanics. American Society for Testing and Materials STP 1332 (Edited
by Panontin, T.L. and Shepard, S.D.). West Conshohocken, PA, pp. 599–613.
Sakane, M., Ohnami, M. and Sawada, M. (1987). Fracture modes and low cycle biaxial fatigue life at
elevated temperature. Journal of Engineering Materials and Technology 109, 236–243.
Savaidis, G. and Seeger, T. (1997). Material behaviour and life evaluation under cyclic multiaxial pro-
portional loading. In Biaxial/Multiaxial Fatigue & Fracture (Edited by Macha, Z. and Mróz, Z.).
Technical University of Opole, Opole pp. 81–98.
Smith, K.N., Watson, P. and Topper, T.H. (1976). A stress-strain function for the fatigue of metals. Jour-
nal of Materials JMLSA 5, 767–778.
Socie, D.F. (1987). Multiaxial fatigue damage models. Journal of Engineering Materials and Technology
109 292–298.
Socie, D.F. and Marquis, G.B. (2001). Multiaxial Fatigue. Society of Automotive Engineers, Inc. War-
rendale, Pa, 484 pp.
Socie, D.F., Waill, L.A. and Dittmer, D.F. (1985). Biaxial fatigue of Inconel 718 including mean stress
effects. In Multiaxial Fatigue (edited by Miller, K.J. and Brown, M.W.). American Society for Testing
and Materials STP 853, Philadelphia, pp. 463–481.
Sonsino, C.M. (1995). Multiaxial fatigue of welded joints under in-phase and out-of-phase local strains
and stresses. International Journal of Fatigue 17, 55–70.
Spagnoli, A. (2001). A new high-cycle fatigue criterion applied to out-of-phase biaxial stress state. Inter-
national Journal of Mechanical Sciences 43, 2581–2595.
Stanfield, G. (1935). Discussion on “The strength of metals under combined alternating stresses”, by
H. Gough and H. Pollard. Proceeding of the Institution of Mechanical Engineers 131, 93.
Stulen F.B. and Cummings, H.N. (1954). A failure criterion for multiaxial fatigue stresses. Proceedings
of the ASTM 54, 822–835.
Susmel, L. and Petrone, N. (2001). Fatigue life prediction for 6082-T6 cylindrical specimens subjected
to in-phase and out-of-phase bending/torsion loadings. In Proc. 6th International Conference on Biax-
ial/Multiaxial Fatigue & Fracture (edite by freitas, ed.). Instituto Superior Tecnico, Lisboa pp. 125–132.
Taira, S., Inoue T. and Takabashi, M. (1969). Low-cycle fatigue under multiaxial stresses (in the case
of combined cyclic tension-compression and cyclic torsion at room temperature), Trans. The Japan
Society of Mechanical Engineers 35, 525–532.
304 Aleksander Karolczuk and Ewald Macha

Tipton, S.M. and Nelson. D.V. (1997). Advances in multiaxial fatigue life prediction for components with
stress concentrations. International Journal of Fatigue 19, 503–515.
Varvani-Farahani, A. (2000). A new energy-critical plane parameter for fatigue life assessment of vari-
ous metallic materials subjected to in-phase and out-of-phase multiaxial fatigue loading conditions.
International Journal of Fatigue, 22, 295–305.
Varvani-Farahani, A. and Topper, T.H. (2000). A new multiaxial fatigue life and crack growth rate model
for various in-phase and out-of-phase strain paths. In Multiaxial Fatigue Deformation: Testing and
Prediction. American Society for Testing and Materials STP 1387, West Conshohocken, PA, pp. 305–
322
Vidal, E., Kenmeugne, B., Robert, J.L., Bahuaud, J. (1996). Fatigue life prediction of components using
multiaxial criteria. In Multiaxial Fatigue and Design, ESIS 21 (Edited by Pineau, A., Cailletaud, G.
and Lindley, T.C.) Mechanical Engineering Publications, London, pp. 365–378.
Wang, C.H. and Brown, M.W. (1993). A path-independant parameter for fatigue under proportional and
non-proportional loading. Fatigue of Engineering Materials and Structures 16, 1285–1293.
Wang, Ying-Yu, and Yao, Wei-Xing (2004). Evaluation and comparison of several multiaxial fatigue cri-
teria. International Journal of Fatigue 26, 17–25.
Weber, B., Clement, J.C., Kenmeugne, B. and Robert, J.L. (1999a). On a global stress-based approach
for fatigue assessment under multuiaxial random loading. In Engineering Against Fatigue (Edited by
Beynon, J.H., Brown, M.W., Lindley, T.C., Smith, R.A. and Tomkins, B.). Rotterdam, pp. 407–414.
Weber, B., Clement, J.C., Kenmeugne, B. and Robert, J.L. (1999b). A stress-based approach for fatigue
assessment under multiaxial variable amplitude loading. In Multiaxial Fatigue and Fracture ESIS
Publications 25, Elsevier (Edited by Macha, E., Bedkowski W. and Lagoda T.). Amsterdam, pp. 218–
231.
Zolochevsky, A., Obataya, Y., Betten, J. (2000). Critical plane approach with two families of microcracks
for modelling of uniteral fatigue damage. Forschung im Ingenieurwesen Vol. 66, pp. 49–56.
Zenner, H., Simbürger, A. and Liu, J. On the fatigue limit of ductile metals under complex multiaxial
loading. International Journal of Fatigue 22, 137–145.

S-ar putea să vă placă și