Sunteți pe pagina 1din 10

Chaos, Solitons and Fractals 22 (2004) 589–598

www.elsevier.com/locate/chaos

Fractality in the threshold condition of fatigue crack


growth: an interpretation of the Kitagawa diagram
Andrea Spagnoli *

Department of Civil and Environmental Engineering & Architecture, University of Parma, Parco Area delle Scienze 181/A,
Parma 43100, Italy
Accepted 24 February 2004

Abstract
It has long been recognized that cracks having a length of the order of magnitude as that of the material micro-
structure size (the so-called short or small cracks) exhibit a fatigue growth behaviour which is remarkably different from
that of long cracks. In particular, the threshold condition of fatigue crack growth is seen to be correlated to the crack
length and the material microstructure. The well-known ‘Kitagawa diagram’ describes the variation of the threshold
stress intensity range against the crack length, showing the existence of a transition value of length beyond which the
threshold of fatigue crack growth is governed by linear elastic fracture mechanics. In the present paper, treating fracture
surfaces as self-similar invasive fractal sets (which are characterized by a uniform fractal dimension), owing to their
fractional physical dimensions, the stress intensity factor is shown to be a function of the crack length. Consequently,
the threshold stress intensity range appears to be also a function of the crack length. In the physical reality, the fractal
dimension of the fracture surfaces may change with the crack length and, thus, a varying fractal dimensional increment
(with respect to the Euclidean domain where the fractal set is contained) from 0 to 1 is here assumed. This allows us to
put forward a new interpretation of the Kitagawa diagram within the framework of the fractal geometry.
Ó 2004 Elsevier Ltd. All rights reserved.

1. Introduction

During last decades, the enhanced ability to detect and measure very short cracks, along with a great interest in using
fracture mechanics methods for smaller and smaller crack sizes, has pointed out the so-called problem of short (small)
cracks (e.g. see Refs. [1,2] for a review). Such cracks are characterized by an anomalous fatigue behaviour in com-
parison to their long counterparts, including: cracks growth rates da=dN higher than what would be predicted by a
long-crack law (e.g. the Paris law [3]) for a given stress intensity range DK; often a decrease in da=dN with increasing
DK; growth at DK values lower than the long-crack threshold; a growth rate strongly dependent on the material
microstructure.
Standard threshold data of stress intensity range are commonly determined for long cracks. According to the
implicitly governing similitude concept of linear elastic fracture mechanics (LEFM), such data are crack-size inde-
pendent. Therefore, an important consequence of the fatigue behaviour of short cracks is that a lower bound (in terms
of crack length) on the use of such data must be defined. Frost [4] firstly questioned the validity of LEFM-based
threshold stress intensity range in the region of short cracks. He found a crack size-dependence of the threshold stress
intensity range DKth in mild steel, aluminium and copper, where DKth decreases with decreasing crack length. A similar

*
Tel.: +39-0521-905927; fax: +39-0521-905924.
E-mail address: spagnoli@unipr.it (A. Spagnoli).

0960-0779/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.chaos.2004.02.034
590 A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598

Nomenclature

a projected (nominal) crack length


a0 intrinsic crack length according to the El Haddad model (see Eq. (2))
an crack length at the nth scale of observation of the fractal
a renormalized crack length
d fractal dimensional increment with respect to the Euclidean domain, 0 6 d 6 1
da=dN crack growth rate
E Young’s modulus
Gn fracture energy at the nth scale of observation of the fractal
G renormalized fracture energy
l0 characteristic (material) length
K stress intensity factor
KC critical stress intensity factor (fracture toughness)
K renormalized stress intensity factor
KC renormalized critical stress intensity factor (renormalized fracture toughness)
DK stress intensity range
DKth threshold stress intensity range
DKth0 threshold stress intensity range for long cracks

DKth renormalized threshold stress intensity range
1
DKth asymptotic threshold stress intensity range for a ! þ1
Drth stress range at the threshold of crack growth
Drth0 fatigue limit for smooth specimens

tendency was also reported for several steels by Usami and Shida [5], and for a high-strength steel by Kitagawa and
Takahashi [6]. Kitagawa and Takahashi [7] later found that there exists a transition crack length, below which DKth is
smaller than that for long cracks, and that such a length is dependent on the material microstructure. The overall
dependence of threshold stress intensity range on the crack length (crack-size effect) is commonly described by a DKth
against a plot, known as ‘Kitagawa diagram’ (Fig. 1a).
With reference to metallic structural materials, a few investigations have been carried out to interpret the Kitagawa
diagram: models based on a critical value of the cyclic plastic zone size (e.g. see Ref. [8]), empirical approaches based on
an intrinsic crack length (e.g. see Ref. [9]), and models based on the interaction between crack-tip slip band and grain
boundary (e.g. see Ref. [10]). In the present paper, the dependence of the threshold stress intensity range on the crack
length is interpreted following an approach based on some fractal geometry concepts. Successful applications of fractal
geometry to size effect-related fatigue problems have recently been proposed by the present author [11,12]. In Ref. [11] a
theoretical explanation of size effect on fatigue limit has been deduced modelling the reacting cross-section of a
structural component through a lacunar fractal set. Accordingly, a so-called monofractal scaling law (related to self-
similar fractal topologies) and a multifractal scaling law (related to self-affine fractal topologies) have been derived. The
latter law accounts for a characteristic (material) length. In Ref. [12] a self-similar invasive fractal set has been exploited
to model fracture surfaces, and a size-dependent crack propagation law has been proposed.
In the following, some aspects of threshold condition related to fatigue growth of short cracks are firstly reviewed.
Then, definitions of fracture energy and stress intensity factor for self-similar fractal topologies (exploited to model
fracture surfaces) are given, and a general relationship of threshold stress intensity range versus crack length for self-
affine fractal topologies is presented. The DKth against a relationship here presented offers an interpretation of the
Kitagawa diagram within the framework of fractal geometry.

2. Some remarks on the threshold condition of fatigue crack growth

As is well known, the most common threshold parameter of fatigue crack growth is the fatigue limit derived from
S–N curves. In 1961, Paris [3] proposed the introduction of LEFM parameters to describe the fatigue crack propagation
under Mode I of loading. In particular, the stress intensity range was seen to be the driving parameter for fatigue crack
growth. Since the pioneering work of Paris, the fatigue limit has begun to be regarded as a threshold condition of non-
propagation of a pre-existing crack or notch. In 1963, McClintock [13] postulated the existence of a fatigue growth
A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598 591

Fig. 1. Schematic representation of the Kitagawa diagram: (a) threshold stress intensity range DKth as a function of the crack length a;
(b) stress range at the threshold of crack growth Drth as a function of the crack length a.

threshold, and as such the threshold stress intensity range DKth was introduced. Therefore, similarly to fracture
toughness defining the limit of unstable crack propagation, the threshold DKth defines the limit of stable fatigue crack
propagation. Experimentally, the threshold stress intensity range is calculated for a conventional very low propagation
rate da=dN (usually 1011 m/cycle) and is typically determined for long fatigue cracks. For such crack sizes, the
threshold stress intensity range DKth0 is found to be a property of the material independent of the crack length.
With reference to metallic structural materials, there are a number of theoretical models which attempt a correlation
between the threshold stress intensity range DKth0 and other material parameters (see Ref. [14] for a review). For instance,
energy-based models (e.g. see Ref. [15]) use a modification of a Griffith-type energy balance to define the crack growth
threshold; models based on dislocation dynamics consider the behaviour of dislocations in the region ahead of the crack
tip (e.g. see Ref. [16]); models based on crack-tip plasticity relate the plastic zone size ahead of the crack tip in the near-
threshold region to the grain size (e.g. see Ref. [17]). According to such models, the threshold stress intensity range can be
determined from material parameters such as Young’s modulus, fracture energy, yield stress and grain size.
As is mentioned in Section 1, the breakdown of LEFM-based threshold condition for short cracks is well sum-
marized by the Kitagawa diagram (see Fig. 1, where both DKth against a and Drth against a diagrams are depicted). A
number of theoretical models (e.g. see the previously-mentioned models of Refs. [8–10]) have been proposed to describe
the influence of the crack length a on DKth . According to the well-known El Haddad model [9], the Kitagawa diagram is
described by the following relationship:
DKth0
DKth ¼ pffiffiffiffiffiffiffiffiffiffiffi
ffi ð1Þ
1 þ aa0

where DKth0 is the crack-size independent threshold stress intensity range for long cracks, a0 is an intrinsic crack length
defined as follows
592 A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598
 2
1 DKth0
a0 ¼ ð2Þ
p Drth0

and Drth0 is the fatigue limit for smooth specimens. For instance, the intrinsic crack length a0 ranges from 1–10 lm for
very high strength steels (yield stress ry up to 2000 MPa) to 100–1000 lm for very low strength steels (ry as low as 200
MPa) [14].
Since the following relationship
DKth
Drth ¼ pffiffiffiffiffiffi ð3Þ
Y pa

holds for the stress range at the threshold of crack growth (Y is a geometry and loading dimensionless factor), we obtain
the following expression by combining Eqs. (1) and (3):
DKth0
Drth ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4Þ
Y pða þ a0 Þ

As is clearly shown in Eq. (4), the El Haddad model introduces the concept of effective crack length which is the sum of
the actual crack length and the intrinsic crack length.
The early experimental results obtained by Tanaka and coworkers [10] for mild steel plate specimens under fully
reversed bending are herein interpreted through the El Haddad model. Such results are related to ferritic and pearlitic
steels with carbon content of 0.20% and grain size of the ferritic phase equal to 7.8 and 55 lm, respectively. For various
values of the crack length (ranging from 6 to 1383 lm), the threshold stress intensity range DKth was experimentally
evaluated. Fig. 2 reports the above experimental data along with the corresponding theoretical curves of the El Haddad
model (such curves are based on an experimentally-determined value of DKth0 and on the value of a0 obtained from

10
Threshold stress intensity

Grain size= 7.8 µm


range, ∆ Kth (MPa√m)

∆σth0 = 235 MPa


∆Kth0 = 5.21 MPa√m
a0 = 156 µm

0.1
1E-006 1E-005 0.0001 0.001 0.01
Crack length, a (m)
(a)

10
Threshold stress intensity

Grain size = 55 µm
range, ∆ Kth (MPa√m)

∆σth0 = 163 MPa


∆Kth0 = 6.20 MPa√m
a0 = 461 µm

0.1
1E-006 1E-005 0.0001 0.001 0.01
Crack length, a (m)
(b)

Fig. 2. Experimental data [10] in the a  DKth bilogarithmic plane and corresponding theoretical curve according to the El Haddad
model (see Eq. (1)): (a) small-grain size material; (b) large-grain size material.
A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598 593

Eq. (2) by calculating the fatigue limit for smooth specimens Drth0 according to an empirical expression, see Ref. [10],
function of the grain size).

3. Modelling of fracture surfaces using fractal geometry

Several theoretical investigations have been carried out in the field of fracture mechanics by considering the fractal
nature of materials (e.g. see Ref. [18] for a review). In Refs. [19–21], for instance, some arguments of fractal geometry
(see Refs. [22,23] for the basics of fractal geometry) have been exploited to explain size effect on both tensile strength
and fracture energy. In particular, by treating a fracture surface as an invasive fractal set (i.e. having a dimension higher
than that of the Euclidean domain where it is contained) [24], a renormalized (scale-invariant) fracture energy has been
determined [20]. Such a quantity applies to self-similar fracture surfaces which are characterized by a uniform fractal
dimension. Moreover, taking into account a non-uniform fractal dimension of the fracture surfaces [25], a so-called
multifractal scaling law for fracture energy has been proposed [21].
In the following, it is shown that modelling of fracture surfaces as self-similar invasive fractals yields the definition of
a renormalized (scale-invariant) threshold stress intensity range DKth . Accordingly, the nominal (apparent) stress
intensity range DKth turns out to be a function of the crack length through a power law. The concept of self-affinity is
then exploited to account for a decreasing influence of the crack length a on DKth with increasing a. Thus, modelling of
fracture surfaces using self-affine fractal topologies allows a bridging between the two experimentally observed
asymptotical tendencies of DKth against a curve (see Fig. 1a), for a ! 0 and a ! þ1.

4. Fracture energy and stress intensity factor for self-similar fracture surfaces

Let us consider a two-dimensional problem of a body containing a crack. From a macroscopic point of view, such a
crack resembles a straight cut of length a (projected or nominal length). By treating the crack as a mathematical self-
similar invasive fractal curve (the crack is assumed to be fractal along its length, and smooth along its tip), such as the
von Koch curve (e.g. see Ref. [22]), its morphology appears to be similar at different scales of observation (or, in other
words, at different steps in the fractal generation procedure). At the nth scale of observation the measured crack length,
e.g. using the ‘yardstick’ method or the projected divider method, is equal to an , so that the measured length of the
fractal crack at the 0th scale of observation is equal to its projected length a.
At the nth scale of observation of the fractal crack, the total energy Ws dissipated at the surface of the crack in a two-
dimensional body of unit thickness is given by

Ws ¼ Gn an ð5Þ

where Gn is the fracture energy at the nth scale of observation. A new definition of fracture energy has been proposed in
order to obtain a material property, the renormalized fracture energy G , which is independent of the scale of obser-
vation [19,26]. Accordingly, the total energy dissipated at the surface of the crack can be written as a function of G

Ws ¼ G a ð6Þ

where a ¼ a1þd is the fractal length of the crack, having physical dimensions equal to ½L1þd , with d being the fractal
increment with respect to the Euclidean domain where the fractal set is contained. By applying dimensional analysis to
Eq. (6), the following fractional physical dimensions of the renormalized (scale-invariant) fracture energy G can be
determined: ½F ½Lð1þdÞ .
Note that for the limit case of d ¼ 0 (smooth cracks) the physical dimensions of G are the conventional ones
(½F ½L1 ), while G has physical dimensions ½F ½L2 for the limit case of d ¼ 1. This points out the dichotomy between
surface energy dissipation (d ¼ 0) corresponding to brittle fracture, and bulk energy dissipation (d ¼ 1) corresponding
to plasticity [27].
Now let us reconsider, within the context of fractal cracks, the well-known Griffith’s energetic approach to the
problem of an infinite two-dimensional body of unit thickness, containing a crack of projected length 2a and subjected
to a remote tensile stress r (Fig. 3) [19,26]. The elastic strain energy release caused by the crack depends on its projected
length, that is:

pr2 a2
We ¼ ð7Þ
E
594 A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598

da* da*

2a*
da 2a da

Fig. 3. Griffith’s problem for fractal cracks (in shadow is the zone of elastic strain energy release).

On the other hand, the energy dissipated at the surface of the fractal crack can be expressed using the renormalized
fracture energy G (see Eq. (6)):
Ws ¼ 2G a ð8Þ

Consequently, the variation dWe of elastic strain energy release due to an incremental increase in the crack length is
defined with respect to the projected crack length 2a, while the corresponding variation dWs of the energy dissipation
refers to the fractal crack length 2a . Differentiating Eqs. (7) and (8) against a and a , respectively, we obtain (e.g. see
Ref. [20] for details on the differentiability of fractal quantities):
2pr2
dWe ¼ a da ð9aÞ
E
dWs ¼ 2G da ¼ 2ð1 þ dÞG ad da ð9bÞ

The energy balance dWe ¼ dWs yields:


r2 pa1d ¼ ð1 þ dÞG E ð10Þ

An extension to fractal cracks of the classical definition of stress intensity factor (SIF) K by Westergaard [28] for the
pffiffiffiffiffiffi
problem under consideration (K ¼ r pa), and that of critical SIF (fracture toughness) KC by Irwin [29] for plane stress
pffiffiffiffiffiffiffi
states (KC ¼ GE) has been proposed in Ref. [20], and the following renormalized quantities have been defined:
1 d
K  ¼ rðpa1d Þ2 ¼ Ka2 ð11aÞ
1
KC ¼ ½ð1 þ dÞG E 2
ð11bÞ

Thus, Eq. (10) reads:


K 2 ¼ KC2 ð12Þ
3þd
The renormalized K  and KC have the following physical dimensions: ½F ½L 2
.

5. Threshold stress intensity range for self-similar and self-affine fracture surfaces

Modelling of fracture surfaces via fractal geometry represents a mathematical formalism which clearly does not
address all relevant physical details of the processes occurring away from the fracture surfaces themselves, although it
well established that, particularly in the threshold condition of fatigue crack growth, such processes (energy dissipation
at the cyclic plastic zone ahead of the crack tip or at slip bands, etc.) might be fundamental (e.g. see the theoretical
models reported in Refs. [8,10,15–17]). Nevertheless, by modelling the roughness of fracture surfaces using fractal
geometry it is implicitly assumed that the energy dissipated at the nominal fracture surfaces is supplemented by an
energy dissipation in the bulk (see the third paragraph of Section 4). In this context, it might be worth recalling the
model of Suresh [30] which accounts for the toughening effects of crack roughness (together with the combined
influence of roughness-induced crack closure) on the near-threshold fatigue crack growth. In particular, it is shown that
considering periodic kinks of a given angle in the fatigue crack path results in an increase of the crack growth resistance.
A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598 595

By considering continuous kinks in the fatigue crack path, Lung and coworkers [31] linked Suresh’s model to fractal
geometry.
Having said this, is the definition of the renormalized quantity in Eq. (11a) applied to the fatigue threshold value of
the stress intensity range, we obtain
 2 d
DKth ¼ DKth a ð13Þ

which describes the variation of the nominal threshold stress intensity range DKth as a function of the crack length a (see
the a  DKth bilogarithmic plane shown in Fig. 4 where Eq. (13) describes a straight line with slope d=2). In other words,
Eq. (13) shows that considering the projected crack length a, i.e. neglecting crack roughness, yields an apparent
threshold stress intensity range DKth whose value is a function of the crack length a, while the renormalized quantity

DKth based on the fractal crack length turns out to be the true material constant.
It is worth noticing that a correlation between DKth and d has experimentally been pointed out, for instance, by
Ivanova and Vstovskii (see data reported in Ref. [18]). On the other hand, it is interesting to note that the power law of
Eq. (13) is formally identical to the empirical relationship reported in the previously-cited work of Frost [4] and to the
theoretical law of the Murakami–Endo model [32]. The former assumes an exponent equal to 1/6 for a (corresponding
to d ¼ 0:33 according to Eq. (13)), and is based on experimental data related to crack lengths in carbon steel specimens
ranging from 100 to 20000 lm. The latter considers an exponent 1/3 (corresponding to d ¼ 0:66 according to Eq. (13)),
and applies to DKth values determined for crack lengths (in carbon steel specimens) in the range 5–200 lm. Hence, such
exponents appear to decrease as the limit values of crack length defining the range of applicability of the aforemen-
tioned laws increase (see the a  DKth bilogarithmic plane shown in Fig. 5 where the exponents in the power laws of
Frost and of Murakami–Endo are indicated together with the relevant experimental data).
The derivation of Eq. (13) is based on mathematical self-similar fractal topologies, that is to say, to fractals having a
uniform fractal dimension at any scale range (this leads, for instance, to the paradoxical conclusion that an infinite
threshold stress intensity range would occur for a crack length tending to infinity). However, when fracture surfaces are
modelled by fractals, their scale invariance has to be understood (in a statistical sense) within a limited scale range [the
lower bound of such a range can be associated to a characteristic material length (e.g. grain size for metals), while the
upper bound can be associated to the size of the body (e.g. specimen or structural component)]. In order to overcome
Threshold stress intensity
range, ∆Kth (log-scale)

2
d

Crack length, a (log-scale)

Fig. 4. Monofractal scaling law (see Eq. (13)) in the a  DKth bilogarithmic plane.
Threshold stress intensity

100
range, ∆ Kth (MPa√m)

Frost [4]
Murakami-Endo [32]

3
10 1
1
6

1
1E-006 1E-005 0.0001 0.001 0.01 0.1
Crack length, a (m)

Fig. 5. Experimental data [4,32] in the a  DKth bilogarithmic and slope of the power laws of Frost [4] and of Murakami–Endo [32].
596 A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598

Fig. 6. Multifractal scaling law (see Eq. (14)) in the a  DKth bilogarithmic plane.

the limitations of mathematical self-similar fractals in modelling fracture surfaces, natural self-affine fractals (e.g. see
Ref. [25]), which are characterized by a statistical anisotropic scale invariance defined by the Hurst’s exponent and by a
non-uniform scaling, might be considered. Self-affine fractals present a characteristic internal length, absolutely lacking
in any self-similar fractal set, which defines a transition from a fractal regime, characterized by a certain fractal
dimension, to an ‘Euclidean’ regime, characterized by classical integer dimensions. As has been pointed out in Ref. [33],
the length measurement, e.g. by applying the ‘yardstick’ method or the projected divider method, of experimentally-
determined self-affine profiles of fracture surfaces shows that the aforementioned two regimes of self-affinity (fractal
regime and Euclidean regime) are only the asymptotics of a continuous topological transition. As has been proposed in
Ref. [21], this kind of non-uniform scaling is defined as ‘geometrical multifractality’, since an infinity of fractal
dimensions is necessary to describe the entire range of scaling.
In the light of the above, in order to describe the DKth against a relationship from small to large values of a, we
model fracture surfaces as natural self-affine invasive fractal surfaces. Now we can assume that the Euclidean regime
(fractal dimensional increment d ¼ 0) is asymptotically attained for a ! þ1 and that a fractal regime characterized by
the theoretical upper bound of the fractal dimensional increment d ¼ 1 is attained for a ! 0þ . If a continuous topo-
logical transition is assumed in line with the definition of geometrical multifractality [21], the DKth vs. a relationship can
assume the following two-parameter expression
1
DKth
DKth ¼ qffiffiffiffiffiffiffiffiffiffiffi ð14Þ
1 þ la0

1
where DKth is the asymptotic threshold stress intensity range for a ! þ1, and l0 is a characteristic (material) length.
Eq. (14) describes the fact that fractality decreases as the crack length a increases (fractal dimensional increment d
ranges from 1 for a ! 0þ to 0 for a ! þ1), namely as a becomes larger and larger with respect to some characteristic
(material) length. In the a  DKth bilogarithmic plane shown in Fig. 6, the slope of Eq. (14) tends to 1/2 for a ! 0þ and
the continuous transition from fractal regime to Euclidean one (where the slope of Eq. (13) tends to zero) becomes
evident.
It can be noted that Eq. (14) is formally identical to Eq. (1) describing the Kitagawa diagram according to the El
1
Haddad model [9]. The comparison shows that the renormalized quantity DKth can be read as the threshold stress
intensity range DKth0 for long cracks (which is crack-size independent), and the characteristic (material) length l0 as the
intrinsic crack length a0 . Thus, Eq. (14) does not provide a new relationship for calculating the threshold stress intensity
range for a given short crack length knowing some material parameters (DKth0 and Drth0 according to Eqs. (1) and (3)),
but it demonstrates that the El Haddad relationship describing the Kitagawa diagram can be well fitted within the non-
conventional framework of fractal geometry.

6. Conclusions

Some concepts of fractal geometry are exploited to describe the topology of fracture surfaces. In particular, treating
such surfaces as mathematical self-similar invasive fractals (which are characterized by a uniform fractal dimension), a
renormalized (scale-invariant) threshold stress intensity range is firstly derived. This implies a power-type relationship
between the nominal threshold stress intensity range and the crack length.
A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598 597

Then, modelling fracture surfaces as natural self-affine invasive fractals (which are characterized by a non-uniform
fractal dimension), might lead to a general relationship between the threshold stress intensity range and the crack length
which is formally identical to that of the El Haddad model [9] describing the threshold fatigue behaviour of short
cracks. Hence, the present investigation offers a new theoretical basis within the framework of the fractal geometry,
according to which the Kitagawa diagram can be interpreted.

References

[1] Miller KJ. The short crack problem. Fatigue and Fracture of Engineering Materials and Structures 1982;5:223–32.
[2] Suresh S, Ritchie RO. Propagation of short fatigue cracks. International Metals Reviews 1984;29:445–76.
[3] Paris PC, Gomez MP, Anderson WP. A critical analysis of crack propagation laws. Journal of Basic Engineering 1961;85:528–34.
[4] Frost NE. The growth of fatigue cracks. In: Yokobori T, editor. Proceedings of the First International Conference on Fracture,
The Japan Society for Strength and Fracture of Materials, Sendai, 1966. p. 1433–59.
[5] Usami S, Shida S. Elastic–plastic analysis of the fatigue limit for a material with small flaws. Fatigue and Fracture of Engineering
Materials and Structures 1979;1:471–81.
[6] Kitagawa H, Takahashi S. Applicability of fracture mechanics to very small cracks or the cracks in the early stage. In: Proceedings
of Second International Conference on Mechanical Behaviour of Materials, American Society for Metals, Metal Park, 1976. p.
627–31.
[7] Kitagawa H, Takahashi S. Fracture mechanical approach to very small fatigue cracks and to the threshold. Transactions of Japan
Society of Mechanical Engineers 1979;45:1289–303.
[8] Ohuchida H, Usami S, Nishioka A. Fatigue limit of steel with cracks. Bulletin of the Japan Society of Mechanical Engineers
1975;18:1185–93.
[9] El Haddad MH, Topper TH, Smith KN. Prediction of nonpropagating cracks. Engineering Fracture Mechanics 1979;11:573–84.
[10] Tanaka K, Nakai Y, Yamashita M. Fatigue growth threshold of small cracks. International Journal of Fatigue 1981;17:519–33.
[11] Carpinteri An, Spagnoli A, Vantadori S. An approach to size effect in fatigue of metals using fractal theories. Fatigue and Fracture
of Engineering Materials and Structures 2002;25:619–27.
[12] Carpinteri An, Spagnoli A. A fractal analysis of size effect on fatigue crack growth. International Journal of Fatigue 2004;26:125–
33.
[13] McClintock FA. In: On the plasticity of the growth of fatigue cracks.Drucker DC, Gilman JJ, editors. Fracture of solids, 20. New
York: Wiley; 1963. p. 65–102.
[14] Taylor D. Fatigue thresholds. London: Butterworths; 1981.
[15] Purushotaman S, Tien JK. Generalized theory of fatigue crack propagation. Materials Science and Engineering 1978;34:241–6.
[16] Yokobori T, Yokobori AT, Kamei A. Dislocation dynamics theory for fatigue crack growth. International Journal of Fracture
1975;11:781–8.
[17] Yoder GR, Cooley YA, Crooker TW. A critical analysis of grain size and yield strength dependence of near-threshold fatigue
crack growth in steels. In: Fracture Mechanics, Special Technical Publication 791, vol. 1, American Society for Testing and
Materials, Philadelphia, 1982. p. 348–65.
[18] Cherepanov GP, Balankin AS, Ivanova VS. Fractal fracture mechanics––a review. Engineering Fracture Mechanics 1995;51:997–
1033.
[19] Carpinteri Al. Scaling laws and renormalization groups for strength and toughness of disordered materials. International Journal
of Solids and Structures 1994;31:291–302.
[20] Carpinteri Al, Chiaia B. Crack-resistance behavior as a consequence of self-similar fracture topologies. International Journal of
Fracture 1996;76:327–40.
[21] Carpinteri Al, Chiaia B. Multifractal scaling laws in the breaking behaviour of disordered materials. Chaos, Solitons and Fractals
1997;8:135–50.
[22] Mandelbrot BB. The fractal geometry of nature. New York: W.H. Freeman and Company; 1982.
[23] Falconer K. Fractal geometry: mathematical foundations and applications. Chichester: Wiley; 1990.
[24] Mandelbrot BB, Passoja DE, Paullay AJ. Fractal character of fracture surfaces of metals. Nature 1984;308:721–2.
[25] Mandelbrot BB. Self-affine fractals and fractal dimension. Physica Scripta 1985;32:257–60.
[26] Goldshtein RV, Mosolov AB. Fractal cracks. Journal of Applied Mathematics and Mechanics 1992;56:563–71.
[27] Carpinteri Al. Decrease of apparent tensile and bending strength with specimen size: two different explanations based on fracture
mechanics. International Journal of Solids and Structures 1989;25:407–29.
[28] Westergaard HM. Bearing pressure and cracks. Journal of Applied Mechanics 1939;6:49–53.
[29] Irwin GR. Analysis of stresses and strains near the end of a crack traversing a plate. Journal of Applied Mechanics 1957;24:109–
14.
[30] Suresh S. Crack deflection: implications for the growth of long and short fatigue cracks. Metallurgical Transactions
1983;14A:2375–85.
[31] Fu R, Mu ZQ, Lung CW, Ai SH. Fractal characterization of fatigue threshold. Scripta Metallurgica et Materialia 1991;25:1647–
50.
598 A. Spagnoli / Chaos, Solitons and Fractals 22 (2004) 589–598

[32] Murakami Y, Endo M. Effects of hardness and crack geometries on DKth of small cracks emanating from small defects. In: The
Behaviour of Short Fatigue Cracks, Publication 1 of the European Group of Fracture, London: Mechanical Engineering
Publications, 1986. p. 275–93.
[33] Weiss J. Self-affinity of fracture surfaces and implications on a possible size effect on fracture energy. International Journal of
Fracture 2001;109:365–81.

S-ar putea să vă placă și