抄録
In this paper, examinations have been made on the intra-specimen variability model as a stochastic model of the fatigue crack propagation rate. This model assumes that the parameter C in the Paris' law da/dN=C(ΔK)m is a spatial random variable within a specimen. The examinations were made with respect to the following two points. The first examination was concerned with the relation between the experimentally observed inter-specimen variability of the parameters in the Paris' law and this model. When the equation da/dN=Cs(ΔK)ms is applied to each specimen, inter-specimen variability of Cs and ms is observed experimentally. It is known that log Cs and ms are normally distributed approximately and there exists a strong correlation between log Cs and ms. This paper first examined if these experimental characteristics can be interpreted by the model. From the analysis, it was shown that the above mentioned experimental characteristics could be interpreted by the intra-specimen variability model both qualitatively and quantitatively. The second examination was concerned with the correlation distance which appears in this model. It was attempted to estimate the value of the correlation distance, and a rather large value was obtained. Thus, it was pointed out that a detailed consideration is necessary on the relation between the correlation distance and the microstructure of the material. Finally, a possibility of a modified model was suggested in which the parameter C was assumed to be a random variable among specimens as well as within a specimen.
