Contour integral evaluation: three-dimensional case
This example illustrates contour integral evaluation in a fully
three-dimensional crack configuration.
The example provides validation of the method (for linear
elastic response), because comparative results are available for this geometry.
Abaqus
provides values of the J-integral; stress intensity
factor, ;
and T-stress as a function of position along a crack front
in three-dimensional geometries. Several contours can be used; and, since the
integral should be path independent, the scatter in the values obtained with
different contours can be used as an indicator of the quality of the results.
The domain integral method used to calculate the contour integral in
Abaqus
generally gives accurate results even with rather coarse models, as is shown in
this case.
Abaqus
offers the evaluation of these parameters for fracture mechanics studies based
on either the conventional finite element method or the extended finite element
method (XFEM).
Problem description
Two geometries are analyzed in this example. In addition, both the
conventional finite element method and the extended finite element method are
used.
Semi-elliptic crack in a half-space
The first geometry analyzed is a semi-elliptic crack in a half-space and is shown in Figure 1. The crack is loaded in Mode I by far-field tension. When used with the
conventional finite element method, because of symmetry, only one-quarter of the
body needs to be analyzed. The mesh is shown in Figure 2. Reduced-integration elements
(C3D20R) are used, with the
midside nodes moved to the quarter-point position on those element edges that
focus onto the crack tip nodes. This quarter-point method provides a strain
singularity and, thus, improves the modeling of the strain field adjacent to the
crack tip (see Contour integral evaluation: two-dimensional case for a discussion of this
technique).
The normal to the crack front is used to specify the crack extension direction,
as shown in Figure 3. The mesh extends out far enough to cause the boundary conditions on the far
faces of the model to have negligible effect on the solution. Three rings of
elements surrounding the crack tip are used to evaluate the contour integrals.
A model meshed with second-order tetrahedral elements
(C3D10) is also created using the
conventional finite element method. This model is shown in Figure 4 with a mesh that is unstructured (as is typical for
tetrahedral meshes) and, therefore, is not focused about the crack front. This
mesh does not have midside nodes adjacent to the crack front moved to the
quarter-point position. A refined mesh around the crack front is required to
obtain accurate results from contour integral evaluations in these
conditions.
When used with the extended finite element method, the mesh is not required
to match the cracked geometry. The presence of a crack is ensured by the
special enriched functions in conjunction with additional degrees of freedom.
This approach also removes the requirement to define the crack front explicitly
or to specify the virtual crack extension direction when evaluating the contour
integral. The data required for the contour integral are determined
automatically based on the level set signed distance functions at the nodes in
an element. The mesh with first-order brick elements (C3D8) for the first geometry is shown in
Figure 5;
and the crack front, which is represented by the level set contour plot of
output variable PSILSM, is shown in
Figure 6.
Semi-elliptic crack in a rectangular plate
The second geometry analyzed is a semi-elliptic crack in a rectangular plate, as shown in Figure 7. The plate is subjected to uniform tension. Due to symmetry only one-quarter
of the body needs to be analyzed when used with the conventional finite element
method. The dimensions of the plate relative to the plate thickness,
t, are as follows: the half-height 16; the half-width 8; the midplane crack depth 0.6; and the surface crack half-length 2.5, which results in a surface crack aspect ratio of 0.24.
The mesh for the plate using the conventional finite element method is shown in
Figure 8, with its profile on the crack plane for brick elements and tetrahedral
elements shown in Figure 9 and Figure 10, respectively. A more refined mesh is required around the
crack front when using tetrahedral elements. The model uses
C3D8 and
C3D10 elements for the
conventional finite element method and both
C3D8 and
C3D4 elements for the extended
finite element method.
Results and discussion
The results are presented for each of the geometries.
Semi-elliptic crack in a half-space
The J-integral values computed by
Abaqus
using the conventional finite element method for the first geometry are given
in
Table 1
as functions of angular position
along the crack front, where
is defined by ,
The values show a rather smooth variation along the crack front and are
reasonably path independent; that is, the values provided by the three contours
are almost the same. There is some loss of path independence and, hence,
presumably, of accuracy as the crack approaches the free surface (at
0).
This accuracy loss is assumed to be attributable to the coarse and rather
distorted mesh in that region.
The stress intensity factors
obtained using the conventional finite element method along the crack line are
compared in
Table 2
and in
Figure 11
with those obtained by Newman and Raju (1979), who used a nodal force method to
compute
from a finite element model that had 3078 degrees of freedom. The
J-integral values calculated by
Abaqus
are converted to
using
where
is the Poisson's ratio and E is the Young's modulus. The
fourth column of
Table 2
presents the stress intensity factors, ,
that are calculated directly by
Abaqus
using the conventional finite element method. For comparison, the stress
intensity factors
obtained using the extended finite element method are also included in
Figure 11.
The comparisons show good agreement with the results published by Newman and
Raju (1979).
Semi-elliptic crack in a rectangular plate
The stress intensity factor values, , computed by Abaqus based on the conventional finite element method for the second geometry are
compared with the results calculated by Nakamura and Parks (1991) in Figure 12, and the agreement between them is excellent. The difference is less than 4%
when comparing the results obtained with brick elements with those obtained with
second-order tetrahedral elements using the conventional finite element method.
The results obtained using the extended finite element method with linear brick
or linear tetrahedral elements are also included in Figure 12 for comparison.
Figure 13
presents the T-stresses calculated by
Abaqus
and those obtained by Nakamura and Parks (1991) and Wang and Parks (1992).
Comparison shows good agreement between them, especially near the middle of the
crack line.
Second model with linear tetrahedral elements using the extended finite element method.
References
Nakamura, T., and D. M. Parks, “Determination
of Elastic T-Stress along Three-Dimensional Crack Fronts
Using an Interaction Integral,” International
Journal of Solids and
Structures, vol. 28, pp. 1597–1611, 1991.
Newman, J.C., and I. S. Raju, “Stress-Intensity
Factors for a Wide Range of Semi-Elliptical Surface Cracks in Finite Thickness
Plates,” Engineering Fracture
Mechanics, vol. 11, pp. 817–829, 1979.
Wang, Y-Y., and D. M. Parks, “Evaluation
of the Elastic T-Stress in Surface-Cracked Plate Using the
Line-Spring Method,” International Journal of
Fracture, vol. 56, pp. 25–40, 1992.
Tables
Table 1. J-integral estimates for semi-elliptic crack using the conventional
method with brick elements (× 10−3 N/mm (top); × 10−3
lb/in (bottom)).
Crack Front Location,
(deg)
Contour
Average Value
1
2
3
0.00
0.8081
0.8232
0.8222
0.8178
4.6099
4.6964
4.6907
4.6656
11.25
0.7817
0.7818
0.7840
0.7825
4.4597
4.4599
4.4727
4.4641
22.50
0.8703
0.8814
0.8834
0.8783
4.9647
5.028
5.0397
5.0108
33.75
1.0300
1.0458
1.0485
1.0415
5.8761
5.9662
5.9817
5.9413
45.00
1.2236
1.2229
1.2261
1.2242
6.9801
6.9762
6.9947
6.9836
56.25
1.3808
1.3800
1.3836
1.3815
7.8771
7.8725
7.8933
7.8809
67.50
1.4488
1.4723
1.4762
1.4658
8.2649
8.3991
8.4213
8.3617
78.75
1.5746
1.5745
1.5786
1.5759
8.9827
8.9818
9.0053
8.8989
90.00
1.5572
1.5783
1.5825
1.5727
8.8832
9.
9.0275
8.9715
Table 2. Comparison of computed Mode I stress intensity factors using the conventional method with
brick elements (N/mm2–mm1/2 (top);
lb/in2–in1/2 (bottom)).
Crack Front Location,
(deg)
Newman and Raju
Average Value with Conventional
Method (calculated from J-integral)
Average Value with Conventional
Method (calculated directly by
Abaqus)