PSI - Issue 47

Teresa Morgado et al. / Procedia Structural Integrity 47 (2023) 882–887 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

886

5

response during damage evolution. This viscous stabilisation component can be specified in a sub-option of the Maxps damage method to facilitate the convergence of the simulation. Due to this factor, it is necessary to perform at the end of each simulation a comparison between two parameters, the Viscous Dissipation (ALLVD) and the Strain Energy (ALLSE), to understand the influence of viscous stabilisation on the results. If the ALLVD value is not sufficiently smaller than the ALLSE value, viscous stabilisation will likely influence the results obtained. Therefore, several simulations are needed to find a balance between convergence and the influence of viscous stabilisation on the results to obtain an acceptable outcome in a reasonable computational time. The same data from its experimental test were used in the fatigue study. The choice of the conventional method over the XFEM method was because this last one requires more computational time, and the fatigue simulation is particularly demanding in terms of computational time. The type of loading present in this study was performed using a periodic wave with a frequency of f= 20 Hz, a maximum force of 7500 N and a stress ratio of R= 0.1 (Morgado, 2015). Due to the need for more extensive computational time for fatigue simulations, this work was considered fatigue analyses with a maximum number of 1000 cycles. All 3D simulation used reduced integration linear hexahedral elements C3D8R (Abaqus, 2014) for the specimen and linear quadrilateral elements R3D4 in the components which carry out the loading and the support of the test piece. It was used a crack with 10 mm. Fig 5 presents the 3D model used with .

Fig. 5. Models used in 3DSimulation (a) 3D Model – Conventional Method; (b) 3D Model – XFEM Method.

3. Results and Discussion Fig. 6 presents the relation between the SIF Num and SIF Ref versus the crack length; the first and second orders elements and the contours are also represented. From this Fig. 6 is observed that the first order elements, CPS4, presents values similars to the experimental ones and the crack with 10 mm presents the best results.

Fig. 6. SIF and crack length for different contours.