Page 310 - Handbook of Materials Failure Analysis

306 CHAPTER 12 A nonlocal damage-mechanics-based approach

1000
s8
Exp.
Analysis
800
s7
J-integral (N/mm) 600 s4 s5
s6

400

s2 s3
200
s1
0
0 1 2 3 4
Crack growth (mm)
FIGURE 12.22
J-resistance behavior of the SEB specimen with initial crack at the ferrite-buttering interface
(experiment vs. analysis). s1-s8 are the points (star symbols) at which snapshots were
taken at different instances of loading.

parameters of individual materials could be successfully used in the weld joint frac-
ture mechanics specimens and hence these are transferable. This method can be used
to design a dissimilar weld joint where the parameters like width of buttering and
weld layer, material properties of electrodes, etc., can be optimized to obtain a weld
with desired mechanical strength and fracture properties (Table 12.7).

7 CONCLUSION
Though local damage models for ductile crack growth simulation have been very
successful in predicting the load-deformation and fracture resistance behavior of dif-
ferent types of specimens and components, it is not able to simulate the conditions of
large stress gradients at the crack-tip. This is due to the mesh-dependence nature
of the solutions. In addition, the size effect cannot be predicted (by scaling the mesh
of the reference specimen) unless one changes the mesh design in the analysis. Sym-
metric boundary conditions are widely used in FE analysis to save computational
effort. In certain situations, the symmetric boundary conditions cannot be used
due to presence of several types of nonsymmetries in loading, material and boundary
conditions. In these situations, the results of the nonlocal model are more realistic
when compared to those of local models. Due to the above observation, one must
be careful about the use of local damage models for prediction of crack path when
the geometry, crack configurations, and loading conditions are complex. The local
formulation of the Rousselier model is able to predict the failure behavior of most