This research focusing on fatigue crack growth experiment and assessments of fatigue crack growth model. Fatigue has been considered as the most important phenomena in the engineering problems. It has been found to occur in every engineering field. Many methods were introduced to overcome fatigue problem. One of it is fatigue crack growth prediction models. Fatigue crack growth model prediction can determine the residual life of the mechanical component. Fatigue crack growth is an important parameter in engineering. It may begin from undamaged area and propagate afterwards. In this case, a mechanical component condition before an unstable crack still can be used until fatigue. With the study of fatigue crack growth, it can predict the crack life of the component and can reduce cost to repair the component. In this thesis, primary objective is to develop, test and provide computational fracture mechanics model that emphasized an uncertainties quantification for surface crack. Fracture and failure study that has been promoted by S-version finite element method is focus on the analysis of the surface-crack problem. In this project, a few limitations of the study have been implemented as to present a specific scope of the investigation. Both fatigue crack growth model is considered; Paris law model and modified Forman model. The specimen in this analysis is aluminium al 2024 T3. The effect of load on the fatigue crack growth rate is investigated on aluminium 2024 T3. It is concluded that modified Forman has a low exponent compared to the Paris law. Further the crack growth prediction based on modified Forman is lower than Paris law. Paris law only can predict the crack in the region II while modified Forman can predict the crack in region II and III. After finish setup the project based on the instruction, the project is proceeded and the data is collected. The result is analysed based on the objective by using data collected from the project. After that, conclusion is made based on the result of the project.
Crack increment is a phenomenon in fracture mechanics. It is occurred due to the stress concentration at the imperfection material. Thus, it leads a crack to growth. Then, the crack will reach to a critical crack length before catastrophic failure could occur. Before the catastrophic failure occur, the cracked structure can be fully utilised until the crack reach to the critical crack length. Thus, it is crucial to investigate the behaviour of the crack increment in fracture mechanics. The main purpose of this paper is to model the crack increment in fracture mechanics analysis via Kolmogorov-Smirnov test. The modelling requires a collection of crack data through experimental work. Then, the data was evaluated based on Kolmogorov-Smirnov test. The results show that the crack increment can be modelled as a Gaussian distribution.
Stress intensity factor (SIF) is one of the most fundamental and useful parameters in all of fracture mechanics. The SIF describes the stress state at a crack tip, is related to the rate of crack growth, and used to establish failure criteria due to fracture. The SIF is determined to define whether the crack will grow or not. The aims of this paper is to examine the best sampling statistical distributions in SIF analysis along the crack front of a structure. Box-Muller transformation is used to generate the statistical distributions which is in normal and lognormal distributions. This method transformed from the random number of the variables within range zero and one. The SIFs are computed using the virtual crack-closure method (VCCM) in bootstrap S-version finite element model (BootstrapS-FEM). The normal and lognormal distributions are represented in 95% of confidence bounds from the one hundred of random samples. The prediction of SIFs are verified with Newman-Raju solution and deterministic S-FEM in 95% of confidence bounds. The prediction of SIFs by BootstrapS-FEM in different statistical distribution are accepted because of the Newman-Raju solution is located in between the 95% confidence bounds. Thus, the lognormal distribution for SIFs prediction is more acceptable between normal distributions.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.