Neuber Method For Fatigue
Neuber Method For Fatigue
Neuber Method For Fatigue
The Neuber method is commonly used to estimate the stress field ahead of
a crack tip in an elastic-plastic material and has been incorporated into a
number of failure assessment procedures. In this work a compact tension
specimen is examined under plane strain conditions using the
Ramberg-Osgood power law plasticity model. It is found that existing
procedures provide a reasonable estimate of the equivalent von Mises
stress over a wide range of load levels, but non-conservative estimates of
the maximum principal stress. A method is proposed to determine the
maximum principal stress from the equivalent stress determined by the
Neuber method. Through the use of the proposed techniques, accurate
estimates of the von Mises equivalent and maximum principal stress
ahead of a sharp crack tip can be obtained. The implications of the result
in terms of the lifetime prediction of cracked components are discussed.
INTRODUCTION
In component lifetime assessments a procedure, known as the sigma-d method, has
been proposed to estimate the crack initiation period from pre-existing defects, i.e. the
time required for the onset of crack growth due to creep and/or fatigue. The method,
developed by Moulin et al. [1], predicts that initiation occurs under creep conditions
when the stress at a distance d ahead of the crack tip is equal to the stress which causes
rupture under uniaxial conditions.
The sigma-d method has recently been introduced into the British Energy defect
assessment procedure, R5 [2]. The method is currently based on the elastic-plastic
stress field generated ahead of the crack tip and does not consider the relaxation of
stress due to creep. This stress field may be estimated by a number of methods
including procedures based on Neubers rule [3].
In Davies et al. [4] a comparison of the crack tip stress field predictions obtained
from a number of methods, including Neubers method, with full field finite element
solutions has been performed, for a CT specimen under plane strain conditions. Some
of the key results of [4] are reviewed here. In addition, the sensitivity of the sigma-d
*Department of Mechanical Engineering, Imperial College London, South Kensington Campus, London SW7 2AZ.
Department of Mechanical Engineering, Amirbkabir University of Technology, Hefez Avenue, Tehran, Iran
Contact author email address: catrin.davies@imperial.ac.uk
EMAS 2004
method, and hence lifetime predictions, to the stress estimate will be considered.
STRESS ESTIMATES FROM NEUBERS METHOD
Neuber [3] proposed that, for a material which obeys a non-linear deformation law that
is linear at vanishingly small stresses, the maximum stress in the vicinity of a sharply
curved notch is related to the nominal (remote) stress through a linear elastic stress
concentration factor. Neuber [3] suggests that the method may be generalised to any
arbitrary loading state through the use of the equivalent stress.
Different interpretations of the Neuber method are found in the literature, which
predict either the stress normal to the crack plane (which is also the maximum principal
stress in the plane of the crack under Mode I conditions) or the equivalent von Mises
stress. A brief description of four Neuber methods detailed in [4] follows and a
summary of the key equations are given in Table 1.
TABLE 1
Method
Name
Neuber
Equation
N
N
Equivalent,
0 d
de 0 ref (1)
d
de
+
=
+
d
de
d
E
E ref
E 0
E 0
Max.
Neuber-R5 Principal,
d
Neuber- Equivalent,
R5(A)
d
Max.
Neuber(M) Principal,
(2)
0 d
ref
de
+ 0
= de
E
E 0
E 0
(3)
d d
0 d
ref
de
+ 0
= de
E
E 0
E 0
N
M ME
N
N
0 d
0 ref (4)
de
=
+
de
E
E M 0
E 0
EMAS 2004
For a sharp crack the linear elastic equivalent (von Mises) stress in the crack plane
de, is determined from the K field,
at a distance d directly ahead of the crack tip,
de =
K
where
2 d
(5)
In Eq. (5) K is the linear elastic stress intensity factor. Similarly the linear elastic
maximum principal stress at d, de, is given by
de =
K
2 d
(6)
EMAS 2004
Neuber method. For a given d the value of the ratio d /d is dependent on load,
specimen geometry and strain hardening power law dependence, N. Based on finite
element studies on a CT specimen a representative value of d /d , denoted M, has been
used to determine d from the equivalent Mises stress obtained from the Neuber-R5(A)
method. The value of M has the following dependency on load [4]
M=
ref
d
= 0.22
+ 2.96 .
d
0
(7)
The function M shown in Eq. (7) has been chosen to be conservative over the relevant
crack tip distances, for N between 5 and 10, and to give optimal agreement with the FE
solutions at d = 50 m. The maximum principal stress, d, may be estimated by solving
Eq. (4) using the M value calculated from Eq. (7).
ANALYSIS METHODS AND MODELS
Specimen Geometry, Material and Finite Element Model
A compact tension (CT) specimen has been examined with a crack length to specimen
width ratio, a/W, equal to 0.5. The Ramberg-Osgood material model is used with
= 0.1, E/0 = 912, representative of austenitic stainless steel at around 600oC. Two
values of N have been used, N = 5 and 10. Plane strain conditions have been examined
and small displacement theory used (i.e. effects of crack blunting have been ignored).
Finite element calculations have been performed using the commercial software
package ABAQUS [9]. Only one half of each specimen has been modelled due to
symmetry and a sharp crack has been represented with a focused mesh at the crack tip.
The mesh for the CT specimen consists of 2005 nodes and 1895 plane strain, linear
hybrid elements. Further details of the finite element analysis are provided in [4].
Stress Estimates and Initiation Time Predictions
The four Neuber equations, shown in Table 1, have been solved iteratively to give the
stress d or d . The accuracy of the Neuber stress estimates has been assessed by
comparison with finite element solutions. The sensitivity of the initiation time
predictions given by sigma-d method to the accuracy of the various stress estimates has
also been examined. The sensitivity is assessed by calculating the ratio between the
initiation time predicted from the Neuber stress estimate to that obtained from the finite
element stress solution. For this purpose, a stress to rupture time relation has been
assumed such that the initiation time, ti, is estimated from (see e.g. [5])
ti = B p .
EMAS 2004
(8)
where p and B are the rupture stress exponent and constant, respectively and is the
relevant stress estimate (i.e. d or d ). A typical value of p equal to 10 has been
assumed. The results are expressed here in normalised form and are independent of B.
RESULTS
Maximum Principal Stress PredictionsNeuber-R5 Method
A comparison of the maximum principal stress distributions, d, over a range of crack
tip distances d, obtained from the Neuber-R5 method (Eq. 2) and the finite element
(FE) results is presented in Fig. 1 for N = 10. All stress values are normalised by 0 and
the distance, d, is normalised by crack length, a. The results for ref /0 = 0.1 and 1.8
are shown in Fig. 1(a) and (b), which correspond to small scale and large scale yielding
conditions, respectively. At ref /0 = 0.1 the stress predicted by the Neuber-R5 method
falls below the FE solution at d/a < 0.004. Very poor agreement between the
Neuber-R5 estimate and FE solution is seen in Fig. 1(b) for ref /0 = 1.8.
Based on a standard CT specimen (W = 50 mm), the ratio between the d stress
from the Neuber-R5 method and the FE solution at d = 50 m is shown in Fig. 2(a) for
N = 5 and 10 (A distance d = 50 m corresponds to d/a = 0.002 for a standard CT
specimen). The Neuber stress estimate and the finite element solution are represented in
the figure by the subscripts N and FE, respectively. It is seen in Fig 2(a) that a
non-conservative prediction of d (indicated by a stress ratio less than 1) is obtained by
the Neuber-R5 method over the entire load range. The ratio of the respective initiation
time predictions from Eq. (8) is shown in Fig. 2(b). It is seen that the values of
initiation time, ti, predicted from the Neuber-R5 stress estimate for ref /0 > 0.2 are
non-conservative and up to four orders of magnitude greater than the predictions based
on the FE stress solution. Note that the implicit assumption here is that initiation is
controlled by the maximum principal stress and hence in Eq. (8) is equal to d.
Equivalent von Mises Stress PredictionsNeuber and Neuber-R5(A)
A comparison of the d stress from the Neuber method, Eq. (1), the Neuber-R5(A)
method, Eq. (3), and the FE solution is presented in Fig. 3 for N = 10. For ref /0 = 0.1
(Fig. 3a) the solutions are almost indistinguishable from each other. At ref /0 = 1.8
(Fig. 3b) it can be seen that both methods provide a conservative prediction of the
equivalent von Mises stress distribution, relative to the FE distribution, over the
distances of interest.
In Fig. 4(a), the ratios between the d stress from the Neuber and Neuber-R5(A)
methods and the FE solution are shown over the load range considered, for d = 50 m
(based on a standard CT specimen). For ref /0 < 1.0 results from both solutions are
virtually indistinguishable, and are slightly non-conservative for ref /0 < 0.2. At
higher loads a conservative prediction is achieved by both methods. It is also seen in
EMAS 2004
the figure that the prediction from the Neuber-R5(A) method (the solid line)
consistently falls below that of the Neuber method (the dash line) and is significantly
less conservative, though more accurate, at high loads (ref /0 > 0.8). Figure 4(b)
shows the ratios of the predicted initiation times, assuming that initiation is controlled
by the equivalent stress. It is seen that the Neuber stress estimate (Eq. 1), in conjunction
with Eq. (8), leads to a prediction of ti that is more conservative than that from the
Neuber-R5(A) stress estimate and is in fact on the order of four times more
conservative for N = 10 at high loads.
Maximum Principal Stress Predictions Modified Neuber Method
A comparison of the d stress determined by the modified Neuber method (Neuber(M))
is shown in Fig. 5 for the CT specimen with N = 10. At ref /0 = 0.1 (Fig. 5a), the
Neuber(M) stress estimate is almost indistinguishable from the FE solution. At the
higher load of ref /0 = 1.8 (Fig. 5b) the scaling factor used provides conservative
results for d/a > 0.002. In this region the Neuber(M) stress is approximately 20%
higher than the finite element solution.
The ratio between the maximum principal stress in the plane of the crack, d,
estimated by the modified Neuber method and that obtained from the FE solution is
shown in Fig. 6(a) for N = 5 and 10. Results are in all cases conservative, indicated by a
ratio greater than 1. Consequently, conservative initiation time predictions are obtained
by the Neuber(M) stress estimate in relation to the FE solution, indicated by a ratio less
than 1 in Fig. 6(b). It is seen in this figure that initiation time predictions from
Neuber(M) are very close to those predicted from the FE stress at intermediate loads,
and within an order of magnitude of the FE prediction over the load range.
Comparison of Neuber R5 Maximum Principal Stress Estimates to the FE von Mises
Stress Solution
The ratio between the maximum principal stress estimate at d = 50 m from the
Neuber-R5 method, (d)N , and the FE solution for the equivalent Mises stress, (d )FE
is shown in Fig. 7(a). If it is assumed that initiation time is controlled by the equivalent
d, as in e.g. [5], and not by the maximum principal stress, d, as proposed in the
stress
sigma-d method, Fig. 7(a) indicates that the use of the Neuber- R5 method (giving d)
will provide a conservative prediction for the cases considered, under plane strain
conditions. Within the load range (0.2 ref /0 1.4) for N = 10, the ratio
(d )N / (d )FE is generally constant at around 1.3 (see Fig. 7a). Therefore using Eq. (8)
and assuming that crack initiation under creep conditions is controlled by the equivalent
stress; the resultant ti predicted from the Neuber-R5 method would be conservative
(underpredict the initiation time) by a factor of around fourteen for this load range,
under plane strain conditions, as shown in Fig 7(b) (i.e. considerably more conservative
than the Neuber(M) method).
EMAS 2004
EMAS 2004
E, E
K
, 0, N
d
de
de
ref
v
(1)
Moulin, D., Drubay, B., Acker, D., and Laiarinandrasana, L., A Practical Method
Based on Stress Evaluation (d Criterion) to Predict Initiation of Crack Under
Creep and Creep-Fatigue Conditions, J. Press. Vess-T ASME, 1995, Vol. 117, pp.
16.
(2)
British Energy Generation Ltd., R5: Assessment Procedure for the High
Temperature Response of Structures. BEGL., Issue 3, 2003.
(3)
Neuber, H., Theory of Stress Concentration for Shear Strained Prismatic Bodies
with Arbitrary Non-Linear Stress Strain Law. J. App. Mech-T ASME, 1961, pp.
544550.
(4)
Davies, C. M., ODowd, N. P., Nikbin, K. M., Webster, G. A., Biglari, F.,
Comparison of Methods for Obtaining Crack Tip Stress Distributions in an
Elastic-Plastic Material, Submitted for publication, 2004.
(5)
(6)
(7)
Chen, G. X., Wang, C. H., and Rose, L. R. F., A Perturbation Solution for a Crack
in a Power Law Material Under Gross Yielding. Fatigue Frac. Eng. M, 2002,
Vol. 25, pp. 231242.
(8)
RCC-MR, Guide for Leak Before Break Analysis and Defect Assessment,
Appendix A16, 2002, AFCEN.
(9)
ABAQUS Users Manual, Version 6.2, Hibbit, Karlsson & Sorenson Inc., 2001.
EMAS 2004
FIGURES
5.0
10.0
ref / 0 = 0.1
N = 10
4.0
ref / 0 = 1.8
3.0
2.0
1.0
4.0
2.0
0.0
0.00
(b)
0.01
0.02
0.03
0.0
0.00
0.04
0.01
d /a
0.04
1.E+05
0.7
1.E+03
N =5
0.6
N = 10
0.5
0.4
1.E+04
ti (d )N / ti (d )FE
0.8
(d )N / (d )FE
0.03
d /a
d = 50 m
a = 25 mm
0.9
N = 10
1.E+02
1.E+01
(a)
(b)
0.2
N = 5
d = 50 m
a = 25 mm
1.E+00
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
Figure 2
0.02
1.0
0.3
Neuber - R5
6.0
(a)
Figure 1
FE
N = 10
8.0
Neuber - R5
d /0
d /0
FE
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
Ratio between (a) d stress and (b) ti prediction from Neuber-R5 and FE.
EMAS 2004
2.5
2.0
1.5
1.0
1.5
1.0
0.5
0.5
0.01
0.02
0.03
0.0
0.00
0.04
0.01
1.6
1.E+01
Neuber
Neuber - R5(A)
1.4
N =5
1.3
1.2
1.1
1.0
d = 50 m
a = 25 mm
N = 10
(a)
1.E-01
Neuber
(b)
Neuber R5(A)
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
ref / 0
Ratio between the (a) von Mises stress and (b) initiation time prediction
from the Neuber method, the adapted R5 Neuber method and FE.
10.0
ref / 0 = 0.1
ref / 0 = 1.8
Neuber (M)
N = 10
4.0
d /0
2.0
N = 10
8.0
FE
3.0
1.0
6.0
4.0
Neuber (M)
2.0
(a)
0.0
0.00
FE
(b)
0.01
0.02
0.03
0.04
d/a
Figure 5
N = 10
1.E-03
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Figure 4
N = 5
1.E-02
1.E-04
0.8
5.0
0.04
d = 50 m
a = 25 mm
1.E+00
ti (d )N / ti (d )FE
1.5
0.9
0.03
Figure 3
(d )N / (d )FE
0.02
d/a
d/a
d / 0
Neuber
Neuber - R5(A)
FE
N = 10
(b)
(a)
0.0
0.00
ref / 0 = 1.8
2.0
d / 0
d / 0
2.5
FE
ref / 0 = 0.1
Neuber
N = 10
Neuber - R5(A)
0.0
0.00
0.01
0.02
0.03
0.04
d/a
EMAS 2004
10
1.6
1.E+01
d = 50 m
a = 25 mm
1.E+00
(d )N / (d )FE
1.4
1.3
N = 10
1.2
N =5
1.1
1.0
ti (d )N / ti (d )FE
1.5
1.E-01
N = 10
1.E-03
0.9
0.8
1.E-04
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
Figure 6
2.6
Ratio between (a) d stress and (b) ti prediction from Neuber(M) and FE.
2.0
1.8
N = 10
1.6
1.4
1.2
1.E-01
1.E-02
N = 10
1.E-03
(b)
(a)
1.0
d = 50 m
a = 25 mm
1.E+00
N =5
ti (d )N / ti (d )FE
2.2
(d )N / (d )FE
1.E+01
d = 50 m
a = 25 mm
2.4
d = 50 m
a = 25 mm
(b)
(a)
N = 5
1.E-02
N = 5
1.E-04
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
Figure 7
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
ref / 0
Ratio between (a) d from Neuber-R5 and d from FE and (b) their
respective initiation time predictions.
EMAS 2004
11