Comparison of Monte Carlo and bootstrap analyses for residual life and confidence interval

Abstract

Failure starts with creation of a crack, then the propagation of the crack and eventually the fracture of the material. Furthermore, material selection, geometry, processing and residual stresses are critical factors that may contribute to uncertainty and prospective failure mechanisms in engineering. These issues may also arise in computational analysis, a problematic model, for instance, a three-dimensional surface fracture that may necessitate numerous degrees of freedom during analysis. However, considering the multiple incidents of material failure, detailed analysis and efforts to prevent premature material failure for safety and engineering integrity can be carried out. Thus, the objective of this study is to model crack growth in a surface-cracked structure. Aluminium alloy 7075-T6 was the material of interest in this study. The S-version finite element method (SFEM) was used to study fracture propagation. The numerical approach developed in this research was the probabilistic SFEM. Instead of mesh rebuilding, a typical finite element approach, the SFEM uses global-local element overlay method to create a fatigue crack growth model, which was then used for crack research. Empirical computation and previous experimental data were used to evaluate the stress intensity factor (SIF), surface crack growth and fatigue life. The SIF was determined using a virtual crack closure method (VCCM). In addition, the probabilistic approach is also a critical method to generate random parameters, such as Monte Carlo and bootstrap methods. The SIF, fatigue life and surface crack growth were validated and deemed to be within the acceptable range.