PSI - Issue 42

5

Andrea Zanichelli et al. / Procedia Structural Integrity 42 (2022) 118–124 Author name / Structural Integrity Procedia 00 (2022) 000 – 000

122

0.5

Experimental Upper curve Mean curve Lower curve

0.4

0.3

0.1 LOAD, P [kN] 0.2

(a)

0.0 0.1 0.2 0.3 0.4 CMOD, [mm] 0.0

Fig. 3. Numerical load-CMOD curves obtained from fracture test simulations, together with experimental scatter band of fracture tests.

As far as the fracture test simulation is concerned, the mesh discretisation is shown in Figure 2 (b) and it is characterised by a symmetric distribution of 650 four-node plate elements. In the same way as for the fracture tests simulations, three values are considered for the ultimate tensile strength: that is the experimental mean value and both the upper and lower values. Accordingly, three numerical load-CMOD curves, named lower, mean and upper curves, have been obtained. In Figure (3), such numerical curves are compared with the corresponding experimental scatter band. It can be noted that the experimental trends are predicted in a quite-satisfactory way by the numerical curves, particularly when the lower value of the ultimate tensile strength is used, that is, with the numerical lower curve in red. In particular, a good correspondence can be observed both in the elastic stage, at the peak-load, and in the inner post-peak softening branch. In Table 1, both the average experimental results (with standard deviations) and the corresponding numerical results are listed in terms of peak load, max P , CMOD at the peak load, peak CMOD , and fracture toughness, ( ) S I II C K + . It can be observed that the peak load and the CMOD at peak are well-predicted when the numerical model is used, since the numerical estimations are in general inside the experimental standard deviation error bars. On the other hand, the numerical fracture toughness is higher with respect to the experimental one. This fact is due to the different unloading compliance computation (that is, such a computation is performed in correspondence of the peak load, instead of using the unloading branch) when the numerical model is used with respect to the experimental curves, as well as to the fact that the numerical model is not able to reproduce the experimentally observed kinked cracks under a far-field Mode I loading. Table 1. Both average experimental results (with standard deviations) and corresponding numerical results in terms of peak load, max P , CMOD at the peak load, peak CMOD , and fracture toughness, ( ) S I II C K + . An example of a column heading max P (kN) peak CMOD (mm) ( ) S I II C K + (MPa . m 0.5 ) Experimental tests 0.397±0.050 0.046±0.009 0.261±0.075 Numerical lower curve 0.400 0.052 0.347 Numerical mean curve 0.442 0.052 0.371 Numerical upper curve 0.482 0.055 0.401