13th International Conference on Fracture June 16–21, 2013, Beijing, China -6- a) b) Figure 4. Horizontal (a) and vertical (b) component of the displacement field expressed in pixels as projected onto the basis of suited functions in post-processing of the measured displacement (1 pixel → 44.2 µm). From the amplitudes ωn and νn, it is possible to plot the SIF and T-stress as function of crack length. In figure 5a and 5b, the SIF obtained by DIC are shown together with the SIF obtained by Abaqus. In average the differences is 9%. For the DIC method, an error bars was estimated by extracting extreme positions of crack tip and measuring the corresponding change in K (average error bars were 11 %). For both methods, we found that KI is constant from 40 mm to the end of sample, demonstrating that the crack propagation is stable (ie. The cracks propagates with a constant velocity), whereas KII fluctuates around zero (Fig 5b). Despite an average difference of 22 %, a good agreement is found in the case of the T-stress component when compared to the FE results. This difference is compensated by error bars (an average error bar were 15%) (Fig 5c). In both case, the T-stress varies from negative value for short cracks to positive value for crack lengths above 5 mm, contrary to the compact tension geometry, where T values are always positive [14]. Compared to the stress intensity factor, the error bar in the case of T-stress is higher because the T-stress is the coefficient of higher order terms of the displacement field expansions. Without surprise, those measures confirms Irwin’s criterion KI=Kc during stable propagation and the Principle of Local Symmetry KII=0. It confirms also Cotterell and Rice crack path stability criterion. As soon the T-stress becomes positive, the crack path doesn’t remain straight (fig. 2b) and find another, not straight, path which satisfies nevertheless Irwin’s criterion and the Principle of Local Symmetry. This is one more [16] experimental demonstration that this double criterion doesn’t lead to a unique solution: the observed crack path is not the trivial straight one but another more stable bifurcated one.
RkJQdWJsaXNoZXIy MjM0NDE=