PSI - Issue 13

Yuri Petrov et al. / Procedia Structural Integrity 13 (2018) 1620–1625 Yuri Petrov/ Structural Integrity Procedia 00 (2018) 000–000

1623

4

parameter – the incubation time. 3. Numerical investigation of stress intensity factor – crack velocity relation 3.1. Experiments by Kalthoff (1983). Dependence of the − dependence on specimen shape. Experimental results on dynamic crack propagation in Araldite B material is presented in work by Kalthoff (1983). Three types of specimens were investigated in this work: double cantilever beam (DCB), single edge notched specimen (SEN) and specimen of a mixed type. In order to measure stress intensity factor for a moving crack, caustics method was used. In addition to this, the crack tip position was registered and thus the crack velocity could be assessed. All the specimens had an artificial initial crack, which started to grow due to quasistatic loading. Different − curves were observed for different specimens. These curves were obtained numerically using the developed scheme and Araldite B material properties. The incubation time is not known for this material and 1.1 µ s value was used. Experimental points and simulation results are presented in figure 2. Two branches of the − curve were obtained using a single set of material properties.

Figure 2. Stress intensity factor – crack velocity dependence for DCB and SEN specimens; experimental points by Kalthoff (1983) and simulation results. 3.2. Dependence of the − relation on loading rate. In work by Ravi-Chandar and Knauss (1984 a,b,c) it was pointed out that the stress intensity factor can vary considerably even when crack propagates with a constant velocity. Thus, existence of the − dependence was doubted. On the contrary, well-established curves were observed for quasistatically loaded samples (Dally (1979)). Equally shaped samples of the same hypothetical brittle material (with properties close to PMMA) were investigated in order to study influence of loading conditions on the − curve. In the first case samples were quasistatically loaded in a way similar to the Kalthoff’s experiments, while in the second case faces of the initial crack were loaded with a short pressure pulse, representing loading scheme used in works by Ravi-Chandar and Knauss (1984a). Results of the numerical simulations are shown in figure 3. Additionally, load types are shown schematically. Figure 3a shows both scatter data for the values and the fitting line for the case of quasistatic loading. Figure 3b shows only scatter data – two rows of data points corresponding to two crack velocities. Since the scatter of the values appeared to be small for the case of quasistatic loading, well-known − could be constructed using fitting. On the contrary, pulse