Crack propagation under mixed-mode loading and associated modelling

VÉRONIQUE DOQUET; GRACIELA BERTOLINO

Reino Unido

Conferencia; nternational Conference on Multiaxial Fatigue and Fracture, I CMFF8; 2007

ICMFF

Cracks submitted to cyclic shear-mode loading, at low K, generally bifurcate to grow in mode I, but when KII or KIII exceed a threshold value, macroscopic shear-mode crack growth may occur. This threshold value, above which classical bifurcation criteria fail, is quite variable from one metal to the other. The characteristics (importance of crack faces friction, cyclic behaviour) which make a metal more or less prone to shear-mode cracking are discussed, based on the results of Mode II fatigue crack growth tests performed -partly in a SEM- on maraging steel, ferritic-pearlitic steel and TA6V, as well as results from the literature. For high enough KII, decelerating mode II propagation takes place -along a distance that increases with KII- before bifurcation occurs. This deceleration, also reported for mode III in the literature, has been attributed to a friction-induced decrease in effective KII or KIII The large quantity of fretting debris continuously coming out from the crack testifies to the importance of contact forces. Elastic-plastic finite element simulations of mode II crack growth through periodic node release were however performed to check if friction-induced decrease in effective KII is the only reason for the deceleration of cracks or if stress/strain redistributions as the crack propagates through its plastic zone also play a role. Shielding of the main crack tip by aborted branches and transverse microcracks ahead of the crack tip is also envisaged. The limitations of the method generally used in the literature to derive the effective stress intensity factor in mode II from measured crack faces displacement are discussed. When crack tip plasticity is moderate (high yield stress, low KII), crack tip shielding by friction can be evaluated from the comparison of measured displacements with those predicted by LEFM, but when plasticity is important, it tends to make displacements larger than predicted by LEFM and to “hide” friction effects. To evaluate friction effects, measured displacements have to be compared with those obtained by elastic-plastic finite element simulations. 3D effects can also complicate the analysis of measured displacements. The correlation of measured crack growth rates with the effective stress intensity factor shows that for high effective KII, mode II crack growth is faster than mode I. The crack paths observed in the materials investigated are not predicted by classical bifurcation criteria. Neither is the influence of the test frequency and environment on the crack paths in maraging steel nor the differences in crack paths observed in different materials for homothetic loading paths, since crack paths are usually supposed to depend only on the mode-mixity ratio. These can be predicted by the maximum growth rate criterion i.e. the comparison, at any cycle, of the current growth rate of the main crack, in mode II with the potential growth rate of an incipient branch, in mode I, to decide which is fastest and thus preferred. To use this criterion, kinetic data on mode II crack growth, as well as mode I have to be used. But to predict crack path and crack growth rate under biaxial non proportional loadings this approach and kinetic data on pure mode I and pure II crack growth may not be sufficient since the crack path and growth rate depend on the loading path. For example, when mode I and mode II are applied sequentially crack growth is coplanar provided KI/KII is high enough and the growth rate is much higher than the sum of mode I and mode II contributions, while an overlap of the mode I and mode II cycle favours bifurcation. That is why a “local approach” of bifurcation based on elastic-plastic finite element computations and the application at the crack tip of two critical-plane fatigue criteria predicting either tensile-stress-dominated or shear-dominated fracture and on the choice of the fastest damage mode is proposed. In pure mode II, this approach successfully predicts a transition from bifurcation, as predicted by usual criteria, to coplanar crack growth when KII increases. It takes into account the redistribution of stresses and strains due to plasticity at the crack tip and it can be applied even when small-scale yielding conditions are no more fulfilled. It explains the formation of an unexpected “anti-branch” at high KII and R = 0, at an angle where the normal stress computed by LEFM is compressive. The proposed approach may also be used under non-proportional loading. It captures the synergy of mode I and mode II when they are applied sequentially