Issue34

M. Ševčík et al, Frattura ed Integrità Strutturale, 34 (2015) 216-225; DOI: 10.3221/IGF-ESIS.34.23 223 Fig. 9 also contains overall prediction (green curve) that represents crack initiation in path I, crack propagation in path II and final propagation in path III. Unfortunately, such overall prediction is not possible to construct due to the fact that the crack propagation path is often changed during the crack propagation and sometimes multiple growing cracks in various crack propagation paths can be recorded. On the other hand it is possible to predict upper and lower bounds, see Fig. 10. Figure 10 : Comparison of experimental data from [21] with upper and lower predictions of P - a diagram for specimens MMB-06 to MMB-11 ( c = 150 mm, c g = 38 mm). Fig. 10 shows comparison of experimental data for 6 MMB specimens. The critical values of the strain energy release rate for crack initiation and crack propagation was taken from Tab. 3. The upper (lower) bound was obtained in such a way that the highest (lowest) critical strain energy release rates of the group of specimens were taken as the fracture criterion. These values are highlighted in Tab. 3. It can be seen that that the upper bound prediction for path II well fits with experimental results for crack lengths longer than 80 mm whereas lower bound prediction for path I is valid for crack lengths between 50 – 100 mm. For longer cracks the lower bound prediction for path III is nearest to the lowest experimental results obtained in the MMB test. Similar behavior was found also for other groups of specimens. Mixed-mode fracture criterion The predictions described in the previous section were based on the total strain energy release rate fracture criterion. The critical strain energy release rate was experimentally measured for every single specimen and served as the input data for the predictions. However, the fracture criterion based on the total strain energy release rate does not consider the mixed- mode nature on the crack tip and it is therefore limited to adhesively bonded joints where the crack tip mode-mixity is the same as the experimentally tested MMB specimen. To obtain more general fracture criterion it is important to perform number of mixed-mode delamination tests and experimentally obtain the mixed-mode fracture criterion. Such fracture criterion can be generally written in the following form: ) G, G,a(G )G,G,a(G crit _II crit _I crit II I  (11) where G I_crit , G II_crit are critical strain energy release rates for pure mode I and pure mode II. The criterion can have various shapes e.g. linear, semi-linear, elliptical, polynomial etc. Preview of the fracture criterion for crack propagation in path II is shown in Fig. 11. Similarly to the fracture criterion shown in Fig. 11, fracture criteria for crack initiation and crack propagation in various crack propagation paths can be experimentally found. These mixed-mode fracture criteria are necessary for the prediction of P - a diagram using the present analytical model. For the purpose of this paper the exponential power law fracture criterion from [14] was used for the prediction of P - a diagram for various MMB configurations. The components of strain energy release rate in the fracture criterion were calculated using extended global method that was developed for the asymmetrical mixed-mode specimens [21]. The