B Rocks 2004

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Engineering Fracture Mechanics 71 (2004) 127–146

www.elsevier.com/locate/engfracmech

Parametrizing ductile tearing resistance by four parameters


W. Brocks *, P. Anuschewski
Institute for Materials Research, GKSS Research Center, Geesthacht, D-21502, Germany
Received 11 September 2002; received in revised form 19 November 2002; accepted 22 November 2002

Abstract
The energy dissipation rate, R, is considered as a measure of resistance to crack extension in elasto-plastic fracture
mechanics. It can be re-evaluated from JR test records of bend and tensile specimens. Three types of RðDaÞ-curves are
identified. If crack initiation occurs close to or at maximum load, the energy dissipation rate is decreasing with crack
extension and approaches a stationary value. This type of RðDaÞ-curves can be described by an exponential function
with three parameters, namely the initial value, the final stationary value, and a transition length. The cumulative JR -
curves for different specimen geometries can be derived by integration. The three parameters of the RðDaÞ-fit together
with an integration constant, the initiation value, Ji , characterize ductile fracture resistance both quantitatively and
physically interpretable. Constraint effects on R-curves can be quantified in terms of these parameters. A procedure for
transferring JR -curves from one geometry to another is proposed.
 2003 Elsevier Ltd. All rights reserved.

Keywords: Ductile tearing; Dissipation rate; Constraint effects

1. Introduction

The prediction of failure by unstable crack extension is one of the main issues in the assessment of safety
and reliability of components for industrial applications. For ductile materials, crack instability may be
preceded by some amount of stable crack extension. The ductile tearing resistance of a material is con-
ventionally characterized by a J -resistance curve which is obtained from bend-type specimens, i.e. C(T) or
SE(B), by standard procedures [1]. It characterizes, within certain limits set forth in the standards, the
resistance against slow stable crack extension. These requirements are severe and may inhibit the appli-
cation to structural components, which in general will not meet the ‘‘validity’’ conditions.

1. Standard JR -curves are derived from deeply notched bend specimens. It is well known, e.g. [2–4], that the
slopes of the JR -curve generally depend on the specimen geometry and loading configuration.

*
Corresponding author.
E-mail address: wolfgang.brocks@gkss.de (W. Brocks).

0013-7944/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0013-7944(02)00287-4
128 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

2. The ASTM standard for measurement of fracture toughness [1] requires, that the permitted crack exten-
sion does not exceed 25% of the remaining ligament, W  a0 . This limit has even been set to 10% in the
past, and the reasons for extending it to 25% has not been made clear.
3. The power law curve fitting procedure, which is prescribed in the standard, has no physical basis and,
hence, does not allow for extrapolation or transfer to other structural configurations.

Crack propagation in pipelines with crack extensions of some thousand meters is one important example
for the inadequacy of the conventional JR -curve characterization of ductile tearing resistance [5]. Klemm [6]
therefore introduced a new parameter characterizing the material resistance against fast ductile fracture in
pipeline steels called crack propagation energy, defined as the dissipated energy, dUdis , per increment of
crack area, dAcr . This definition is a straight transfer of GriffithÕs elastic energy release rate to plastic
processes, which is consistent with the incremental theory of plasticity.
In fact, the cumulative quantity J, which rises with increasing crack length, is not the true driving force
for ductile tearing as Turner [7] has pointed out. He, too, proposed to define tearing resistance in terms of
the energy dissipation rate,
dUdis dWext dUel
R¼ ¼  ; ð1Þ
dAcr dAcr dAcr
where Wext is external work and Uel the (recoverable) elastic strain energy. For specimens with constant
thickness B and straight crack front (‘‘plane’’ problems) the increment of crack area is dAcr ¼ Bda. The
dissipation rate has the same dimension as J and characterizes the increment of non-recoverable mechanical
work per crack extension increment. It falls with increasing crack length in gross plasticity and consists of
two contributions, namely work of remote plastic deformation and local work of separation,
dUpl dUsep
R¼ þ : ð2Þ
dAcr dAcr
Several studies have produced evidence, that the dissipation rate, R, is more appropriate for characterizing
crack extension in plastically deformed structures than the conventionally used J integral [8–11]. It is, in
fact, the true ‘‘driving force’’ which has to equalize the structural resistance in order to propagate the crack
by some amount, Da, whereas J accumulates the plastic work done along a given loading path. It is,
however, not a material but a structural property, as it contains the work of remote plastic deformation.
Some remarks with respect to nomenclature and terminology seem necessary at this point. Whereas
Turner [7] started denoting the energy dissipation rate by the letter R, which has been adopted by Memhard
et al. [8], he switched to D later [9,10], which has been taken up by Sumpter [11]. As D commonly denotes
‘‘damage’’ in the literature on continuum damage mechanics, the authors of the present contribution retain
R for dissipation rate. A distinction should be made however between an applied and a resistance quantity,
though both have to be of the same type, of course. Up to now, the term ‘‘dissipation rate’’ has been used to
indicate the ‘‘type’’ of the quantity. As defined by Eq. (1), the dissipation rate is calculated from measured
quantities obtained in fracture mechanics testing and is insofar an applied value, Rappl ðF ; v; DaÞ, denoted by
C in [11], which will be regarded as an appropriate measure of the driving force for crack extension in a
given structure, in the following. With the crack extending, the equation
Rappl ðF ; v; DaÞ ¼ Rres ðDaÞ ð3Þ
holds as a criterion for ductile tearing, thus defining ductile tearing resistance, Rres ðDaÞ, denoted by D in
[11], in terms of the dissipation rate. There is no way of measuring Rres ðDaÞ than by calculating
Rappl ðF ; v; DaÞ from fracture mechanics test data and assuming the validity of Eq. (3). Note once more, that
Rres ðDaÞ is a structural but not a material property. For the sake of simplicity of notation, the subscripts
‘‘appl’’ and ‘‘res’’ will be omitted in the following sections, keeping in mind that R is understood as an
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 129

applied value, Rappl ðF ; v; DaÞ, if its determination from test records is considered, and as a resistance value,
Rres ðDaÞ, if its relation to JR -curves is addressed.
In the present paper, procedures to calculate Rres ðDaÞ-curves from existing JR -curves and a simple ex-
ponential curve fit are described and discussed. Applying this curve fit, Rres ðDaÞ-curves can be characterized
by three parameters, namely the initial value, R0 ¼ Rres ðDa ¼ 0Þ, a final stationary value, R1 , which actually
equals the crack propagation energy as introduced by Klemm [6], and a characteristic transition length,
1=k. The shapes of the cumulative JR -curves for different specimen geometries can be derived by integration
and do not, in general, appear to follow a power law.
The three parameters of the Rres ðDaÞ-fit together with an integration constant, the initiation value, Ji
characterize ductile fracture resistance both quantitatively and in a manner, which has a physical inter-
pretation. Constraint effects on JR -curves can be quantified in terms of these four parameters.
Finally, the question how Rres ðDaÞ-curves can be scaled in a way, which allows transferring the data from
one specimen to another and, finally, to a component is addressed.

2. Experimental determination of the dissipation rate

The dissipated energy can be calculated locally or globally.

• The local dissipation rate results from the plastic work consumed in the wake of crack extension, only. It
can be determined experimentally from temperature measurements in the crack tip region with thermo-
couples or infrared camera [6,8]. It must not be confused with the ‘‘separation energy’’ as defined in Eqs.
(2) and (15) or (16) below, as it also includes a substantial amount of plastic strain energy.
• The global dissipation rate is determined according to Eq. (1) from the total irreversible external work
per crack extension. It may also include plastic strain energy in zones far remote from the crack tip, e.g.
around load points or supports. For quasi-static processes, the dissipated work is simply evaluated from
the area under the measured load, F , versus displacement, v, curve,
DUdis F Dvpl
R ¼ lim ¼ lim : ð4Þ
Da!0 Bn Da Da!0 Bn Da

It was shown by Memhard et al. [8] that local and global dissipation rate become equal if stationary
crack extension is reached.
If the respective information for evaluating Eq. (4) is not available any more, R can be simply re-
evaluated from existing JR -curves by just inverting the recursion formulas for calculating J from test data
[8]. The value JðiÞ at a point corresponding to aðiÞ , vðiÞ , FðiÞ on the specimen load versus load-line displace-
ment record is generally split into an elastic and a plastic contribution,
JðiÞ ¼ JelðiÞ þ JplðiÞ ; ð5Þ
2
where the elastic part results from JelðiÞ ¼ KðiÞ =E0 and the plastic part is determined from the area under the
load versus plastic load-line displacement record.
For bend-type specimens, C(T) and SE(B), we have [1]
    
gði1Þ AplðiÞ  Aplði1Þ aðiÞ  aði1Þ
JplðiÞ ¼ Jplði1Þ þ 1  cði1Þ ; ð6Þ
W  aði1Þ Bn W  aði1Þ

with the geometry factors


130 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

2:0 þ 0:522ð1  a=W Þ for CðTÞ;
g¼ ð7aÞ
2:0 for SEðBÞ;

1:0 þ 0:76ð1  a=W Þ for CðTÞ;
c¼ ð7bÞ
1:0 for SEðBÞ:
The increment of area, DApl ¼ AplðiÞ  Aplði1Þ , for a crack extension increment, Da ¼ aðiÞ  aði1Þ , between
lines of constant plastic displacement is
DApl ¼ AplðiÞ  Aplði1Þ ¼ 12ðFðiÞ þ Fði1Þ ÞðvplðiÞ  vplði1Þ Þ ¼ 12ðFðiÞ þ Fði1Þ ÞDvpl : ð8Þ
Rearranging the formula for Jpl yields
    !
FðiÞ þ Fði1Þ Dvpl Da W  aði1Þ DJpl cði1Þ
1  cði1Þ ¼ þ Jplði1Þ ; ð9Þ
2BN Da W  aði1Þ gði1Þ Da gði1Þ

from which the dissipation rate is obtained according to Eq. (4) for Da ! 0, i.e. ði  1Þ ! i,
 
W  a dJpl c
R¼ þ Jpl : ð10Þ
g da g
The conversion formula given by Kolednik [12,13] for bend-type specimens misses the second term of
Eq. (10). As this term may contribute significantly to R, see Fig. 1, the respective results will be wrong
without it.
For tensile-type specimens, M(T) and DE(T), we have [14,15]
2DU 
JplðiÞ ¼ Jplði1Þ þ ; ð11Þ
Bn ð2W  ai1  ai Þ

4000
20 Mn Mo Ni 5 5
Rtot
3000
(W -a)/η (dJpl/da)

(γ/η) Jpl
R [N/mm]

2000

1000

0
0 2 4 6
∆a [mm]

Fig. 1. Evaluation of RðDaÞ from a Jpl ðDaÞ-curve for a C(T) specimen (W ¼ 50 mm; B ¼ 20 mm, Bn ¼ 16 mm, a0 =W ¼ 0:62) of 20 Mn
Mo Ni 5 5; contributions of dJpl =da and Jpl according to Eq. (10).
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 131

where the increment DU  for a crack extension increment, Da ¼ aðiÞ  aði1Þ , is determined from the area
under the load versus elongation record,
Z vðiÞ
1
DU  ¼ F dv  FðiÞ ðvðiÞ  vði1Þ Þ: ð12Þ
vði1Þ 2

The dissipation rate is again obtained from Eq. (4)


dJpl
R ¼ ðW  aÞ : ð13Þ
da
The evaluation of RðDaÞ-curves by Eqs. (10) and (13), respectively, requires a numerical differentiation,
which is generally sensitive to oscillations of data, in particular, if the data result from experimental records.
A beveling procedure for precursory smoothing of J ðDaÞ data and following differentiation has been
successfully applied in [16] for the analysis of numerous JR -curves. An ‘‘n-point bevel’’ is calibrated through
a global power law fit, which is obtained from a linear fit, y ¼ c1 x þ c2 , of the logarithmic JR -curve,
ln½J ðDa þ c0 Þ. As lnðxÞ goes to minus infinity for x ! 0, a positive constant, c0 , has to be added to the Da
values before taking the logarithm. If the global shape of the curve after re-transformation coincides sat-
isfactorily with the measured JR -curve, the fit is proper, see Fig. 2a. The global fit may be improved by
changing the value of c0 , which defines the initial point of the bevel. The RðDaÞ-curve can be calculated
according to Eqs. (10) or (13) with the derivative of the global fit curve, see Fig. 2b. Now, the n-point bevel
is conducted like a flexible ruler over the N test points of the JR -curve by calculating the fit parameters over
n data points, n  N , starting with the first (global) test point, j ¼ 1, and taking the first point, i ¼ 1, and
the corresponding slope of the local fit as the respective values for the global curve. The bevel is then shifted
to the second data point, j ¼ 2, and the procedure is repeated until the whole test curve has been passed
through, i.e. j ¼ N  n, see Fig. 2a. The missing data points, N  n < j 6 N can be calculated by starting
from the end of the curve. The RðDaÞ-curve is again calculated according to Eqs. (10) or (13) with the
derivative of the local fit curve, see Fig. 2b. The value n of the bevel determines the smoothing effect of the
procedure. Its optimum value depends on the specific shape of the curve, the density of data points and
their distance, respectively.

1. 104
20 Mn Mo Ni 5 5 20 Mn Mo Ni 5 5

8. 103 from global fit


1000

from local fit


6. 103
Jpl [N/mm]

R [N/mm]

500 4. 103

test [16]
2. 103
global fit
local fit
0 0
0 5 10 15 0 5 10 15
(a) ∆a [mm] (b) ∆a [mm]

Fig. 2. Numerical curve fitting and differentiation for side-grooved C(T) specimen (W ¼ 200 mm; B ¼ 10 mm, Bn ¼ 8 mm,
a0 =W ¼ 0:60) of 20 Mn Mo Ni 5 5 (a) Jpl ðDaÞ-curve (b) RðDaÞ.
132 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

1500 1. 104
C(T) test C(T) test
C(T) FE C(T) FE
M(T) test 8. 103
M(T) test
M(T) FE
1000 M(T) FE
6. 103
Jpl [N/mm]

R [N/mm]
4. 103
500

2. 103

StE 460
0 0
0 2 4 6 0 2 4 6
(a) ∆a [mm] (b) ∆a [mm]

Fig. 3. (a) Jpl ðDaÞ-curves and (b) RðDaÞ-curves for side-grooved C(T) and M(T) specimens (W ¼ 50 mm) of a ferritic steel DIN StE 460;
tests [17,18] and numerical simulations with cohesive model [18,20].

JR -curves and RðDaÞ-curves for side-grooved C(T) and M(T) specimens of a ferritic steel DIN StE 460
are displayed in Fig. 3. The dots represent the test data [17], and the solid lines the results of finite
element (FE) simulations with a cohesive zone model [18–21]. The results confirm a general tendency
observed by Turner [7] or Memhard et al. [8], that RðDaÞ-curves decrease monotonically and after a
non-stationary transition approach a stationary value. Other shapes of RðDaÞ-curves will be discussed
later.

3. Numerical calculation of the dissipation rate

In an elastic–plastic FE analysis of crack extension the dissipation rate can be calculated directly [8,19]
from stresses and increments of strains during one time step, Dt,
Z Z
DUpl 1 p 1
¼ rij Deij dV ¼ re Depe dV ð14Þ
Bn Da Bn Da Vpl Bn Da Vpl

re and epe being von Mises effective stress and effective plastic strain, respectively. The volume integral
may either be performed over the whole body, thus yielding the result of Eq. (1), or just over the plastic
zone at the crack tip, how large it may be. If the FE simulation reflects plastic processes, only, i.e. it is
simply based on the Mises–Prandtl–Reuss constitutive equations and follows experimental records of
either J ðDaÞ or vðDaÞ as, e.g. in [22], dissipated work equals total plastic work as in the experimental
procedure.
Recent approaches to modeling of ductile rupture [18,19,21,23,24] refer to BarenblattÕs idea of intro-
ducing a ‘‘process zone’’ ahead of the crack tip where material degradation and separation occur. This
approach requires a constitutive description of the material behavior in the process zone, which can mirror
the local loss of stress carrying capacity. In general, two alternatives are applied:
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 133

• models based on the micro-mechanisms of ductile failure, namely the nucleation, growth and coalescence
of voids, as e.g. the most commonly used model of Gurson, Tvergaard and Needleman (GTN-model)
[25–28], and
• phenomenological ‘‘cohesive models’’ describing the decohesion process by a traction strength and the
work of separation per unit area, e.g. [18,23,24,29].

An example for the application of this model is shown in Fig. 3a. The JR -curves for side-grooved
C(T) and M(T) specimen have been simulated by plane-strain FE analyses with one set of cohesive
parameters, namely a separation strength of rmax ¼ 3:36 rY ¼ 1579 MPa and a separation energy of Cc ¼
53:3 N/mm [18,20]. The numerical simulations coincide fairly with the experimental data. The M(T) JR -
curve is perfectly met whereas the C(T) JR -curve is slightly underestimated. The model is apparently able to
capture the strongly different R-curve behavior of bend-type and tensile-type specimens with just two
parameters characterizing the local separation process. The experimental RðDaÞ-curves in Fig. 3b are cal-
culated from Eq. (10) and the numerical RðDaÞ-data results from Eq. (14).
For the GTN-model the incremental energy needed for material separation, DUsep , in an incremental
crack advance, DAcr ¼ Bn Da, is given by
Z
DUsep 1
¼ ð1  f Þre Depe dV : ð15Þ
Bn Da Bn Da V
Here, the volume integral is performed over a single row of elements in front of the crack tip being
described by the GTN-model. For the cohesive model, DUsep is calculated from
Z dc
DUsep
¼ Cc ¼ rn ddn ; ð16Þ
Bn Da 0

if the crack has extended over one element length. This approach allows for splitting the total dissipated
work into the (local) work of separation in the process zone and the (global) plastic work in the embedding
material and, thus, solves a classical problem of elastic–plastic fracture mechanics [18,21,30]. The ratio of
the rates of global plastic work, dUpl =Bda, and local separation work, Cc is 130 for M(T) or 20 for C(T)
close to crack initiation (Da ¼ 0:5 mm) and 40 for M(T) or 7 for C(T) at steady-state crack extension,
respectively. This means that the actual separation energy is only less than 1–15% of the plastic work, or in
other words: 85–99% of what is measured in R-curve testing is work of plastic deformation not ‘‘fracture
energy’’.

4. Shapes of R(Da)-curves

JR -curves display a diversity of appearance, which is difficult to classify. Fig. 4a gives three examples,
namely for:

#1. a side-grooved C(T) specimen of width W ¼ 50 mm, thickness B ¼ 20 mm, net section thickness
Bn ¼ 16 mm, relative initial crack length a0 =W ¼ 0:62, ferritic steel 20 Mn Mo Ni 5 5, experimental
data [15],
#2. an M(T) specimen of width 2W ¼ 100 mm, thickness B ¼ 5 mm, relative initial crack length
a0 =W ¼ 0:51, aluminum alloy Al 2024 FC, experimental data [15],
#3. a thin center cracked panel of width 2W ¼ 508 mm, thickness B ¼ 1 mm, relative initial crack
length a0 =W ¼ 0:2, aluminum alloy Al 2024 T3, numerical simulations [31] of experimental data
[32].
134 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

800 4
specimen type
#1
600 3 #2
specimen type
#3
#1
Jpl [N/mm]

#2
400

R / R∞
2
#3

200 1

0 0
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
(a) ∆a / (W -a 0 ) (b) ∆a / (W -a 0 )

Fig. 4. Shapes of (a) Jpl ðDaÞ-curves and (b) RðDaÞ-curves for (#1) C(T)sg, W ¼ 50 mm, Bn ¼ 16 mm, a0 =W ¼ 0:62, 20 Mn Mo Ni 5 5,
(#2) M(T), 2W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:51, Al 2024 FC, (#3) M(T), 2W ¼ 508 mm, B ¼ 1 mm, a0 =W ¼ 0:2, Al 2024 T3.

A power law as proposed in the standards [1] is obviously inappropriate as curve fit. Of course, the
standard is not meant to be applied to crack extensions larger than Da=ðW  a0 Þ P 0:25. There are ap-
plications however, where larger crack extensions may occur, e.g. in thin panels or pipelines.
Plotting the respective RðDaÞ-curves in Fig. 4b shows three different types, which have also been iden-
tified by Turner and Kolednik [9,33] and Stampfl and Kolednik [13]:

#1. Monotonically decreasing from an initial value and, after a non-stationary transition, approaching a
stationary value, see also [7,8] and Fig. 3b, which is typical for initiation occurring close to or at maxi-
mum load and crack extension remaining normal to the loading (‘‘flat’’ fracture). A similar behavior is
observed for the crack tip opening angle, CTOA, and relations between the two quantities have been
derived by Turner and Kolednik [34]. This curve shape has also been verified in numerical studies
based on test data [18,21,30].
#2. Monotonically decreasing from an initial value and exhibiting a second peak value before approaching
a stationary value, see also Fig. 2b, which is again typical for initiation occurring close to or at max-
imum load but crack extension changing from normal to slant fracture. Stampfl and Kolednik [13] ex-
plain the second peak value in the RðDaÞ-curve, which results in a sudden increase of the slope of the
correspondent JR -curve, by the occurrence of shear lips on the fracture surface, which contribute ad-
ditional plastic work.
#3. Monotonically increasing to a maximum value and decreasing again to a stationary value, which is
typical for initiation occurring well before maximum load.

Crack tip blunting may cause a first peak of RðDaÞ in the beginning as in Fig. 2b, and curve #2 in Fig. 4b,
see [13]. As the emphasis is put upon large ductile crack extension and its simple phenomenological de-
scription, this effect is neglected in the following.
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 135

4 4

eq. (17) eq. (21)

α = 3; λ = 10 α = 3; λ = 10
3 3

α
α
R / R ∞ [-]

R / R∞ [-]
2 2

1 1
1/λ

0 0
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
(a) ∆a / (W -a 0 ) [-] (b) ∆a / (W -a 0 ) [-]

Fig. 5. Curve fitting of RðDaÞ-curves according to Eqs. (17) and (21), respectively; visualization of the parameters.

The monotonical decreasing of RðDaÞ-curves as for type #1 and––partly––#2 of Fig. 4b can be fitted by
the exponential function [8]. 1
  
Da
R ¼ R1 1 þ a exp k ; ð17Þ
W  a0
with three parameters R1 , a, k, see Fig. 5a.

• The parameter a characterizes the initial value R0 ¼ RðDa ¼ 0Þ according to


R0
a¼  1: ð18Þ
R1
The determination of the initial value, R0 , is subject to the general uncertainty of determining the point of
physical crack initiation after crack tip blunting. As the emphasis is put upon large ductile crack ex-
tension and its simple phenomenological description, these uncertainties are not further considered, here.
The two parameters a and k determine the initial transient phase of the JR -curve.
• The final stationary value, R1 , is the so-called ‘‘crack propagation energy’’ as introduced by Klemm [6]
for dynamic crack extension in pipes. It allows for a physically founded extrapolation of resistance
curves beyond the measured range for long crack extensions [8].
• The parameter k describes the rate of decay from the initial value, R0 to the final value, R1 ,
  
W  a0 dR 
k¼ ð19Þ
R1  R0 da Da¼0

1
Note that in the fit curve given in [8], Da has been normalized by the specimen width, W , instead of the ligament width, W  a0 , so
that the values given for k will differ by the respective factor, ðW  a0 Þ=W . The normalization by the ligament width appears to be more
plausible.
136 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

and its reciprocal, 1=k, is a relative characteristic transition length in the sense, that for a crack extension
of DaðW  a0 Þ ¼ n=k, the respective value of R is
     
nðW  a0 Þ R0
R Da ¼ ¼ R1 1 þ en : ð20Þ
k R1

The analysis of JR -curves of thin aluminum specimens [15] showed that a better fitting could be obtained
in some cases by using
" sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!#
Da
R ¼ R1 1 þ a exp k ð21Þ
W  a0

instead of Eq. (17), see Fig. 5b. The parameters R1 and a keep the same meaning as before, whereas no
simple physical interpretation exists for k in this case, as the initial slope of the function (21) is infinite. The
descent of the curve is much steeper, meaning that the transition regime is much smaller, which is essential
for its better fitting properties in the respective cases.
Curves of type #2, exhibiting a second peak value of R, which results in a sudden increase of the
slope of the correspondent JR -curve, would require different fit curves for smaller and larger crack
extension. This would include additional parameters describing location and magnitude of the second
peak, i.e. the point where crack extension has changed from normal to slant fracture [13]. As will be shown
in the next section, some satisfactory overall representation by Eq. (17) may be found in these cases nev-
ertheless.
Curves of type #3, starting at lower R-values, R0 < R1 , running through a maximum, Rm , and then
decreasing to a stationary value, R1 , again, would need a different kind of fit function. This problem is not
further discussed here, as the respective cases occur for large thin panels mainly, where crack initiation
occurs well before maximum load.

5. Integration of R(Da)-curves

The differential equations (10) and (13) for bend- or tensile-type specimens, respectively, can be inte-
grated for a given RðDaÞ-curve to obtain a JR -curve [8]. By introducing the normalized quantities,
a a0 Da Da Jpl R
x¼ ¼ þ ¼ x0 þ ; y¼ ; r¼ ð22Þ
W W W W R1 R1
and the abbreviations,
9
cðxÞ gðxÞrðxÞ rðxÞ =
pðxÞ ¼ ; qB ðxÞ ¼ ; qT ðxÞ ¼
1x 1x 1x ; ð23Þ
;
gðxÞ ¼ 2 þ g1 ð1  xÞ; cðxÞ ¼ 1 þ c1 ð1  xÞ
the differential equations take the simple forms,

y 0 þ pðxÞy ¼ qB ðxÞ for CðTÞ and SEðBÞ
; ð24Þ
y 0 ¼ qT ðxÞ for MðTÞ and DEðTÞ

with the general solutions,


 R R R 
expð pðxÞ dxÞ½ qB ðxÞ expð pðxÞ dxÞ dx þ C0  for CðTÞ and SEðBÞ
y¼ R : ð25Þ
qT ðxÞ dx þ C0 for MðTÞ and DEðTÞ
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 137

2.5
a 0 /W = 0.5
a 0 /W = 0.5
J pl,0 / R∞ = 0.2 J pl,0 / R∞ = 0.2
1.5
2.0
α = 3; λ = 10

1.5
1.0
Jpl / R∞

Jpl / R∞
1.0

0.5
C(T) 0.5 C(T)
SE(B) SE(B)
M(T) M(T)
0.0 0.0
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
(a) ∆a / (W -a 0 ) (b) ∆a / (W -a 0 )

Fig. 6. Shapes of Jpl ðDaÞ-curves for (a) steady-state crack growth and (b) transient behavior according to Eq. (17).

The integration constant, C0 , is determined from the initial conditions at physical or technical crack
initiation, J0pl ¼ Jipl at Da ¼ 0 or J0pl ¼ J0:2
pl
at Da ¼ 0:2 mm, respectively.
The function Jpl ðDaÞ depends on the right hand sides of Eq. (24), i.e. the function RðDaÞ. The integration
has to be performed numerically, in general. A few analytical solutions for pure steady-state crack ex-
tension, i.e. a constant dissipation rate, rðxÞ ¼ R=R1 ¼ 1, and a transient behavior according to Eq. (17)
were given in [8], evidencing that the shape of JR -curves is intrinsically geometry dependent. Fig. 6 shows a
qualitative comparison of JR -curves of common fracture mechanics specimens, normalized by R1 . They
demonstrate that a unique curve fit as, e.g., in ASTM E1820 [1] is not physically based. Due to Eq. (13), the
slopes of JR -curves of M(T) specimens may even rise for steady-state crack extension. An increasing slope
of JR -curves of M(T) specimens is hence partly due to the integration procedure for calculating the J in-
tegral and does not necessarily indicate any ‘‘increase’’ of crack extension resistance. The latter can only be
concluded from the RðDaÞ-curve (type #2).
The re-integration of JR -curves from fits of RðDaÞ data provides a verification of the quality of the fit.
Three examples for Al 2024 T351 [15,16] are presented in the following figures.
Fig. 7: The RðDaÞ-curve of an M(T) specimen, W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:5, is of type #1, which
can be easily fitted by Eq. (17). Integrating the RðDaÞ fit curve according to Eq. (25) results in a rather
‘‘perfect’’ JR -curve.
Fig. 8: The RðDaÞ-curve of an M(T) specimen, W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:7, is of type #2 with a
second peak value. Two ranges of Da can hence be chosen for the fitting procedure, small and large crack
extension, which result in two parameter sets. The respective integrated curves approximate the experiment
JR -curve over Da < 3 mm (fit #1) or the whole range data (fit #2), respectively.
Fig. 9: The RðDaÞ-curve of an M(T) specimen, W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:7, is also of type #2
with two plateau values. Two sets of fit parameters for either of large crack extension can be calculated. Fit
#3, which pays no heed to the transition regime for small crack extension at all, appears to yield the best
overall approximation of the experimental JR -curve.
138 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

5000 400

Al 2024 T351 Al 2024 T351


4000 test
fit eq. (16) 300

3000

J pl [N/mm]
R [N/mm]

200

2000

100
1000
test
integrated fit
0 0
0.00 0.05 0.10 0.15 0.00 0.05 0.10 0.15
(a) ∆a / (W -a 0 ) [-] (b) ∆a / (W -a 0 ) [-]

Fig. 7. (a) RðDaÞ-curve and (b) Jpl ðDaÞ-curve for an M(T) specimen (W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:5).

4000 400

Al 2024 T351 Al 2024 T351


test
test
3000 300 int. fit #1
fit #1
int. fit #2
fit #2
J pl [N/mm]
R [N/mm]

2000 200

1000 100

0 0
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
(a ) ∆a / (W -a 0 ) [-] (b) ∆a / (W -a 0 ) [-]

Fig. 8. (a) RðDaÞ-curve and (b) JR -curve for an M(T) specimen (W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:7).

6. Effects of specimen geometry

All RðDaÞ-curves which can be fitted by Eq. (17) allow for discussing the effects of specimen geometry and
loading mode in terms of the three fitting parameters, R1 , a, k, having some physical significance, as de-
scribed in Section 4. This is done for side-grooved C(T) specimens of the ferritic steel 20 Mn Mo Ni 5 5
[15,16] by plotting the respective parameters in dependence on the ligament width, W  a0 , and the relative
thickness, Bn =ðW  a0 Þ, see Figs. 10–12. Some general tendencies are observed:
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 139

800 200

Al 2024 T351 Al 2024 T351


test
600 150
fit #1
fit #2
R [N/mm]

J pl [N/mm]
fit #3
400 100

test
200 50
int. fit #1
int. fit #2
int. fit #3
0 0
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
(a) ∆a / (W -a 0 ) [-] (b) ∆a / (W -a 0 ) [-]

Fig. 9. (a) RðDaÞ-curve and (b) Jpl ðDaÞ-curve for a C(T) specimen (W ¼ 100 mm, B ¼ 5 mm, a0 =W ¼ 0:7).

20 Mn Mo Ni 55 sg
3000

2000
R ∞ [N/mm]

B / (W -a 0 )
≤ 0.1
1000 0.2
0.4-0.5
0.8-1.0
2
0
0 20 40 60 80 100
W - a 0 [mm]

Fig. 10. Dependence of crack propagation energy, R1 , on ligament dimensions.

• The crack propagation energy, R1 , increases with the ligament length, W  a0 , see Fig. 10. Lower bound
values are obtained for specimens with square ligament, Bn =ðW  a0 Þ  1.
• The ratio of the initial to the final value of the energy dissipation rate, R0 =R1 ¼ a þ 1, increases with
the ligament length, W  a0 , see Fig. 11, displaying some scatter, however, which cannot be related to the
relative thickness, Bn =ðW  a0 Þ. The value for one specimen with W ¼ 50 mm, Bn ¼ 16 mm, a0 =W ¼ 0:62
falls completely out of the scatter band of the rest.
140 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

20 Mn Mo Ni 5 5 sg

R0 / R∞ [-]
4
B / (W -a 0 )
≤ 0.1
0.2
2 0.4-0.5
0.8-1.0
2
0
0 20 40 60 80 100
W - a 0 [mm]

Fig. 11. Dependence of normalized dissipation rate at initiation, R0 =R1 , on ligament dimensions.

0.040
B / (W -a 0 )
≤ 0.1
0.030 0.2
0.4-0.5
0.8-1.0
2
1 / λ [-]

0.020

0.010

20 Mn Mo Ni 55 sg
0.000
0 20 40 60 80 100
W - a 0 [mm]

Fig. 12. Dependence of transition length, 1=k, on ligament dimensions.

• The characteristic transition length, 1=k, decreases with W  a0 , see Fig. 12, displaying some scatter,
again, which cannot be related to the relative thickness, Bn =ðW  a0 Þ. The value for the specimen with
Bn =ðW  a0 Þ  2 falls completely out of the scatter band of the rest.
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 141

Several parameters have been proposed to quantify ‘‘constraint effects’’ on crack extension resistance
[35], mainly the biaxiality factor b [36], the T stress [37] and the Q stress [38,39]. Neither of them was able to
satisfactorily scale the geometry effects on the resistance data in the present case [16]. This is because T
stress [37] and Q stress characterize crack tip fields whereas the work of plastic strain, DUpl , per crack
extension increment, Da, is the dominant contribution to the dissipation rate. The latter is––like J ––an
overall structural property, which depends on the amount of plastification, namely the size of the plastic
zone and the average magnitude of plastic strain. A normalization of RðDaÞ-curves by some appropriately
defined plastic limit-load factor,

r Y a0

Rref ¼ rY ðW  a0 Þ fY ; ð26Þ
E W

can hence be expected to reduce the geometry dependence, at least for fully yielded specimens, as was
shown in [40] for side-grooved M(T) and C(T) specimens of a DIN StE 460 steel.
The geometry factor, fY , is the ratio of the plastic limit-load of the cracked structure, FY , to the yield load
of an uncracked structure with same net section, F0 ,
8 9
> FY >
> for MðTÞ and DEðTÞ >
FY < 2rY BðW  a0 Þ =
fY ¼ ¼ ; ð27Þ
F0 > > FY >
: for CðTÞ and SEðBÞ > ;
rY BðW  a0 Þ

where rY is the yield strength. Plastic limit-loads of plane specimen geometries can be determined from slip
line solutions for the limiting cases of plane stress and plane strain. Numerous solutions can be found in the
EFAM handbook ETM97 [41].
For M(T) specimens,

2
fY-MT ¼ pffiffiffi ; ð28Þ
3

holds for plane strain [41,42]. For C(T) specimens, a plane-strain solution for single edge cracked, SE,
specimens under combined bending and tension can be taken [42,43] which has been approximated by a
polynomial for an easier handling [44],
a
a
2
0 0
fY-CT ¼ 0:74  1:20 þ 0:46 : ð29Þ
W W
Fig. 13 in comparison to Fig. 10 illustrates that some beneficial effects with respect to geometry inde-
pendence can be obtained by this normalization for the side-grooved C(T) specimens of 20 Mn Mo Ni 5 5
for W  a0 P 40 mm and Bn =ðW  a0 Þ P 0:2. Normalization works also well for the specimen with W ¼ 50
mm, Bn ¼ 16 mm and a0 =W ¼ 0:6, i.e. approximately square ligament, Bn =ðW  a0 Þ ¼ 0:8. For smaller
specimens, W  a0 < 20 mm, normalization does not capture the dependence on the ligament width,
W  a0 , correctly, and in particular 3D effects of specimen thickness, Bn are not described for W  a0 < 40
mm. This is probably due to the assumption of plane-strain condition in all cases and could be overcome by
a thickness dependent limit-load factor, fY .
Based on this normalization, the transfer of JR -curves between two different specimen geometries, ‘‘A’’
and ‘‘B’’, is obtained by

Step 1: convert the Jpl ðDaÞ-curve of specimen ‘‘A’’ into an RðDaÞ-curve by Eq. (10) or (13) and fit it by
Eq. (17);
142 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

1000 B n / (W -a 0 )
≤ 0.1
0.2
800
0.4-0.5
0.8-1.0

R∞ / Rref
600
2.0

400

200

0
0 20 40 60 80 100
W - a 0 [mm]

Fig. 13. Dependence of normalized crack propagation energy on ligament dimensions.

Step 2: convert the RðDaÞ-curve of specimen ‘‘A’’ into an RðDaÞ-curve of specimen ‘‘B’’ by multiplying it
with the factor RBref =RA
ref calculated from Eq. (26),

RBref A
RB1 ¼ R1 ; ð30Þ
RAref

Step 3: integrate the RB ðDaÞ-curve by Eq. (25) into a ‘‘predicted’’ Jpl ðDaÞ-curve for ‘‘B’’.

By applying this procedure to the JR -curve of side-grooved C(T) specimens for a ferritic steel, DIN
StE 460, the JR -curve of side-grooved M(T) specimens could be perfectly predicted [40], thus indicat-
ing that the normalization will also work between different loading configurations. The transfer proce-
dure worked perfectly in both directions, i.e. from C(T) to M(T) as well as from M(T) to C(T), see
[16].
The same normalization and prediction procedure is also applied for the numerically simulated JR -
curves [18] shown in Fig. 3, which are based on the experimental data of [17]. Though the tested material
is nominally the same StE 460 and the specimen geometries are identical, the test data analyzed in [40]
differ from the data in [17]. Fig. 14 shows the curve fits of the numerical RðDaÞ-data by Eq. (17). The
steep descent of the C(T) curve in the beginning could not be sufficiently fitted, but as is shown later in
Fig. 15a, this does not affect the integrated JR -curves significantly. The normalization of the experimental
RðDaÞ data by the limit-load factor of Eq. (26) shifts the two curves closely together, thus indicating that
it captures the geometry effects. The transfer between the two specimen geometries can now be performed
by scaling the crack propagation energy according to Eq. (30), whereas the two other parameters, a and
k, are taken from specimen ‘‘A’’, and re-integrating the RB ðDaÞ-curve to obtain a predicted JR -curve for
specimen ‘‘B’’. The results of this transfer into both directions, i.e. from C(T) to M(T) and from M(T) to
C(T) are shown in Fig. 15a and b, respectively. The predicted M(T) JR -curve overestimates the measured
one, slightly, and inversely, the predicted C(T) JR -curve underestimates the measured one. All in all, the
transfer works satisfactorily in reconciling the large difference in R-curve behavior between the two
specimens.
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 143

8000 400
StE 460 StE 460
C(T) FE C(T) FE
6000 eq. (17) 300 eq. (17)
M(T) FE M(T) FE
eq. (17) eq. (17)

R / Rref [-]
R [N/mm]

4000 200

2000 100

0 0
0 2 4 6 0 2 4 6
(a) ∆a [mm] (b) ∆a [mm]

Fig. 14. (a) Fitting of RðDaÞ-curves by Eq. (16), M(T): R1 ¼ 1400 N/mm, a ¼ 5:0, k ¼ 10:2; C(T): R1 ¼ 356 N/mm, a ¼ 3:6, k ¼ 17:4;
(b) normalization of RðDaÞ-curves by limit-load factor, Eq. (26) M(T): Rref ¼ 29:66 N/mm, Eq. (28); C(T): Rref ¼ 3:97 N/mm, Eq. (29).

1500 1500
M(T) FE
C(T) FE
M(T) test
C(T) test
M(T) int.fit
C(T) int.fit
C(T) FE
M(T) FE
C(T) test
M(T) test
1000 1000 C(T) pred
M(T) pred
Jpl [N/mm]
Jpl [N/mm]

500 500

StE 460 StE 460


0 0
0 2 4 6 0 2 4 6
(a) ∆a [mm] (b) ∆a [mm]

Fig. 15. Transfer of a Jpl ðDaÞ-curve from one specimen geometry to another, (a) Prediction of an M(T) Jpl ðDaÞ-curve from a C(T)
Jpl ðDaÞ-curve and (b) prediction of a C(T) Jpl ðDaÞ-curve from an M(T) Jpl ðDaÞ-curve.

7. Conclusions and outlook

Turner [7] pointed out that using a rate quantity as driving force in elasto-plastic fracture mechanics is
preferable to the cumulative J -integral, because it is consistent with incremental plasticity. He proposed the
144 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

energy dissipation rate, R, as a measure of resistance to crack extension. The analysis of experimental data
revealed three types of RðDaÞ-curves. If crack initiation starts close to or at maximum load, the respective
RðDaÞ-curves can be characterized by three parameters, namely the initial value, R0 ¼ RðDa ¼ 0Þ, the final
stationary value or crack propagation energy, R1 , and a characteristic transition length, 1=k. As JR ðDaÞ-
curves can be obtained from RðDaÞ-curves by integration, these three parameters together with an inte-
gration constant, the initiation value, Ji , characterize ductile fracture resistance both quantitatively and
physically understandable way. The geometry dependence of fracture resistance can hence be characterized
quantitatively in terms of these parameters.
Geometry functions derived from plastic limit-load expressions have been defined for normalization
of RðDaÞ-curves, which cover some of the geometry effects, primarily for materials of low strength but
high toughness. The normalization affects the stationary value, R1 , only. Rules for transferring the
parameters of the transition regime, R0 and k, which are also subject to geometry effects, are still
missing.
The problem of geometry dependence of R-curves, which is crucial for fracture mechanics assessments
of components, has been intensively discussed for decades, and the number of experimentally deter-
mined JR -curves as well as the number of recognized constraint effects has ever increased. A convincing
and physically based solution has never been found, however. The concept of the dissipation rate does
not give a final answer either, but it provides a sound concept for a better physical understanding of:
(i) what JR -curves actually are; (ii) how they can be parametrized; and (iii) which are the reasons for
‘‘geometry effects’’ and how the latter can be quantified under certain conditions. If further JR ðDaÞ-
data were to be re-analyzed according to this concept, this might finally lead to a systematic and quanti-
tative treatment of geometry effects resulting in an engineering concept of fracture mechanics assessment for
components.

Acknowledgements

The authors gratefully acknowledge financial support for the analysis and interpretation of R-curve data
by the Deutsche Forschungsgemeinschaft under contract number Br 521/7-1.

References

[1] ASTM E 1820-01: Standard test method for measurement of fracture toughness. Annual Book of ASTM Standards. vol. 03.01.
West Conshohocken: American Society for Testing and Materials; 2002.
[2] Garwood SJ. Effect of specimen geometry on crack growth resistance. In: ASTM STP 677. American Society for Testing and
Materials; 1979. p. 511–32.
[3] Brocks W, K€ unecke G, Noack H-D, Veith H. On the transferability of fracture mechanics parameters to structures using FEM.
Nucl Engng Des 1989;112:1–14.
[4] Link RE, Landes JD, Herrera R, Zhou Z. Something new in size and geometry effects for J –R curves. In: Blauel JG, Schwalbe
K-H, editors. Defect Assessment in Components, Fundamentals and Applications, ESIS/EGF 9. London: Mechanical
Engineering Publications; 1991. p. 707–21.
[5] Holmes B, Priest AH, Walker EF. Prediction of linepiping fracture behaviour from laboratory tests. Int J Pres Ves Piping
1983;12:1–27.
[6] Klemm W. Material resistance against fast ductile fracture in pipeline steels. Dynamic Fracture Mechanics for the 1990s.
In: Homma H, Shockey DA, Yagawa G, editors. Proc of the 1989 ASME Pressure Vessels and Piping Conference. 1989.
p. 99–104.
[7] Turner CE. A re-assessment of ductile tearing resistance (Part I and II). In: Firrao D, editor. Fracture Behaviour and Design of
Materials and Structures. Proc ECF 8, vol. II. 1990. p. 933–49. and p. 951–68.
[8] Memhard D, Brocks W, Fricke S. Characterization of ductile tearing resistance by energy dissipation rate. Fatigue Fract Engng
Mater Struct 1993;16:1109–24.
W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146 145

[9] Turner CE, Kolednik O. Application of energy dissipation rate arguments to stable crack growth. Fatigue Fract Engng Mater
Struct 1994;20:1109–27.
[10] Kolednik O, Shan G, Fischer FD. The energy dissipation rate––a new tool to interpret geometry and size effects. In: Piascik RS,
Newman JC, Dowling NE, editors. Fatigue and Fracture Mechanics. ASTM STP 1296, vol. 27. Philadelphia: American Society
for Testing and Materials; 1997. p. 126–51.
[11] Sumpter JDG. An alternative view of R-curve testing. Engng Fract Mech 1999;64:161–76.
[12] Kolednik O. A simple model to explain the geometry dependence of J –Da-curves. Int J Fract 1993;63:263–74.
[13] Stampfl J, Kolednik O. The separation of the fracture energy in metallic materials. Int J Fract 2000;101:321–45.
[14] Garwood SJ, Robinson JN, Turner CE. The measurement of crack growth resistance curves using the J -integral. Int J Fract
1975;11:528–30.
[15] Schwalbe K-H, Hellmann D. Correlation of stable crack growth with the J -integral and the crack tip opening displacement, effects
of geometry, size, and material. Report GKSS 84/E/37, GKSS Research Centre Geesthacht, 1984.
[16] Anuschewski P, Brocks W, Hellmann D. Characterisation of ductile tearing resistance by the energy dissipation rate: effects of
material, specimen shape and size. Report GKSS 2002/13, GKSS Research Centre Geesthacht, 2002.
[17] H€acker R, Gerwien P, Thiemich K-D, Wossidlo P, Baer W. Gegen€ uberstellung von Ergebnissen bruchmechanischer
Untersuchungen an C(T)-Proben and M(T)-Proben aus dem Werkstoff StE 460. Bericht BAM-V.33 98/2, Bundesanstalt f€ ur
Materialforschung und-pr€ ufung, Berlin, 1990.
[18] Siegmund Th, Bernauer G, Brocks W. Two models of ductile fracture in contest: porous metal plasticity and cohesive elements. In:
Brown MW, de los Rios ER, Miller KJ, editors. Proc ECF 12, Sheffield, 14–18 September 1998, Fracture from Defects, vol. II.
Engineering Materials Advisory Services Ltd.; 1998. p. 933–8.
[19] Siegmund Th, Brocks W. Prediction of the work of separation and implications to modeling. Int J Fract 1999;99:97–116.
[20] Brocks W, Siegmund Th. Effects of geometry and material on the energy dissipation rate. Fracture Mechanics: Application and
Challenges, Proc ECF 13, San Sebastian, 6–9 September 2000.
[21] Siegmund Th, Brocks W. The role of cohesive strength and separation energy for modeling of ductile fracture. In: Jerina KL, Paris
PC, editors. Fatigue and Fracture Mechanics. ASTM STP 1360, vol. 30. American Society for Testing and Materials; 2000.
p. 139–51.
[22] Siegele D, Schmitt W. Determination and simulation of stable crack growth in ADINA. Comput Struct 1983;17:697–703.
[23] Yuan H, Lin G, Cornec A. Verification of a cohesive zone model for ductile fracture. J Mech Phys Solids 1994;42:
1397–434.
[24] Lin G, Kim Y-J, Cornec A. Fracture toughness of a constrained metal layer. Comput Mater Sci 1997:36–47.
[25] Gurson AL. Continuum theory of ductile rupture by void nucleation and growth: Part I––yield criteria and flow rules for porous
ductile media. J Engng Mater Technol 1977;99:2–15.
[26] Tvergaard V. Ductile fracture by cavity nucleation between larger voids. J Mech Phys Solids 1982;30.
[27] Needleman A, Tvergaard V. An analysis of ductile rupture at a crack tip. J Mech Phys Solids 1987;35:151–83.
[28] Xia L, Shih CF, Hutchinson JW. A computational approach to ductile crack growth under large scale yielding. J Mech Phys
Solids 1995;43:389–413.
[29] Needleman A. An analysis of decohesion along an imperfect interface. Int J Fract 1990;42:21–40.
[30] Siegmund Th, Brocks W. A numerical study on the correlation between the work of separation and the dissipation rate in ductile
fracture. Engng Fract Mech 2000;67:139–54.
[31] Siegmund Th, Brocks W. Modeling crack growth in thin sheet aluminum alloys. In: Halford GL, Gallagher JP, editors. Fatigue
and Fracture Mechanics. ASTM STP 1389, vol. 31. American Society for Testing and Materials; 2000. p. 475–85.
[32] Broek D, Thomson D, Jeong DY. Testing and analysis of flat and curved panels with multiple cracks. FAA-NASA Int Symp
Advanced Structural Integrity Methods for Airframe Durability and Damage Tolerance, NASA Conf Publ 3274, Part I, 1994.
p. 85–98.
[33] Turner CE, Kolednik O. A micro and macro approach to the energy dissipation rate model of stable ductile crack growth. Fatigue
Fract Engng Mater Struct 1994;17:1089–107.
[34] Turner CE, Kolednik O. Simple test method for energy dissipation rate, CTOA and the study of size and transferability effects for
large amounts of ductile crack growth. Fatigue Fract Engng Mater Struct 1994;20:1507–28.
[35] Brocks W, Schmitt W. The second parameter in J –R curves: constraint or triaxiality? In: Kirk M, Bakker A, editors. 2nd Symp
Constraint Effects, ASTM STP 1224, Philadelphia. 1995. p. 209–31.
[36] Leevers PS, Radon JC. Inherent stress biaxiality in various fracture specimen geometries. Int J Fract 1982;19:311–25.
[37] Rice JR. Limitations of small scale yielding approximations for crack tip plasticity. J Mech Phys Solids 1974;22:
17–26.
[38] OÕDowd NP, Shih CF. Family of crack-tip fields characterized by a triaxiality parameter. J Mech Phys Solids 1991;39:
989–1015.
[39] Shih CF, OÕDowd NP, Kirk MT. A framework for quantifying crack tip constraint to surface flaws. ASTM-Symposium on
Constraint Effects in Fracture, Indianapolis, ASTM STP 1171; 1991. p. 2–20.
146 W. Brocks, P. Anuschewski / Engineering Fracture Mechanics 71 (2004) 127–146

[40] Memhard D, Brocks W, Fricke S. Characterization of ductile tearing resistance by energy dissipation rate. In: Schwalbe K-H,
Berger C, editors. Structural Integrity: Experiments–Models–Applications. Proc ECF 10, vol. I. Warley: Engineering Materials
Advisory Services Ltd.; 1994. p. 149–58.
[41] Schwalbe K-H, et al. EFAM ETM 97––the ETM method for assessing the significance of crack like defects in engineering
structures. Report GKSS 98/E/6, GKSS Research Centre Geesthacht, 1998.
[42] Miller AG. Review of limit loads of structures containing defects. Int J Pres Ves Piping 1988;32:197–327.
[43] Ewing DJF, Richards CE. The yield-point loads of single notched pin-loaded tensile strips. J Mech Phys Solids 1974;22:27–36.
[44] Kim Y-J, McClintock FA, Parks DM. Yield locus in deep, single face cracked specimens under combined bending and tension.
J Appl Mech 1996;63:1045–6.

You might also like