PSI - Issue 28

Mohamed Ali Bouaziz et al. / Procedia Structural Integrity 28 (2020) 393–402 M.A. BOUAZIZ et al/ Structural Integrity Procedia 00 (2019) 000–000

399

7

 i i u u u   1

1

  

  



  

  

  

  

u







(4)

 2 x

x

x





1

1

1

5. Results and discussions The J -integral was computed by contour integration using experimentally measured strains and displacements. The crack extension history (Figure 4) was taken into account. For each loading step, the contour was located in such a way that it contained the crack tip while ensuring that most of the surface it covered was in front of the crack tip. The finite element model established to follow crack growth calculated, at each step, the J -integral with the built-in procedure of the FE code (Brocks and Scheider 2001). The J -integral variation is compared in Figure 6 to that obtained by the experimental method based on DIC measurements. Similar results were provided by both methods. Very low levels of J were detected and the values of J gradually increased as a function of the applied load. Both methods yielded almost the same levels of J (Root Mean Squared Error = 0.46 kJ/m²). It was noted that starting at a relatively high level of loading (~ step 660), the value of J calculated experimentally was no longer continuous. The experimental values were still close to the FE values but presented some drops. This non-continuous variation is probably due to the location of the crack tip and in particular the effect of the discretization discussed in Section 3.

Fig. 6. J -integral variation in the studied part of the test. This effect is shown on the crack resistance curve in Figure 7. Pinning/depinning steps were detected for a constant interval of 12 µm, which corresponds to the discretization of the crack path adopted in the procedure for locating the crack tip. The enhanced procedure proposed in Section 3 made it possible to minimize the discretization effect on the J-R curve (Figure 8).