13th International Conference on Fracture June 16–21, 2013, Beijing, China -4- The twist angle is neither uniform along the shear lip, nor constant as the crack grows. It is substantially less than 45°. In some cases, the peak value is not observed at the free surface but up to 1mm in depth. The evolutions of the shear lips width in air and NaCl solution under the same mechanical conditions are compared on Fig 2. Fig 2: Evolution of shear lips width in air and NaCl solution under similar loading conditions a) specimens CCP7a and CCP7Na and b) steel specimens CC2a and CCP2Na Shear lips development was delayed and less complete in the NaCl solution, in agreement with some observations from the literature [2, 6]. The evolution of crack front tunnelling, ∆a, is plotted on Fig 3. Fig 3: Evolution of crack front tunnelling, in steel specimens, in air or NaCl solution In air, tunnelling was progressively reduced and cancelled by the development of shear lips, in spite of a drop in nominal ∆KI in the slanted zones and of a probable local increase in closure effects, generally more important at low ∆KI. Both factors should reduce the crack growth rate near the free surfaces and thus increase the convexity of the front. A reverse effect is observed. The reduction of tunnelling provides evidence for an effective contribution of shear modes to the crack driving force. However, fractographic observations revealed fatigue striations in normal zones as well as in slanted areas, where a few isolated rubbing marks were also observed near the side surfaces. Hence mode I still contributed to crack growth in the slanted zones and it would be incorrect to describe crack twisting as a transition from tension to shear-mode, as sometimes done in the literature. 4. Numerical study The numerical study is constituted of two parts: 1) an elastic analysis in an extended finite element code (X-FEM, implemented in Code_Aster at EDF R&D, www.code-aster.org/) well fitted to
RkJQdWJsaXNoZXIy MjM0NDE=