In this paper, the composite variability model as a stochastic model of the fatigue crack propagation rate has been examined, and compared with the intra-specimen variability model examined in the previous papers. The composite variability model assumes that the parameter
C in the Paris's law
da/dN=
C(Δ
K)
m is a random variable of which randomness is composed of the variability within a specimen and that between specimens, whereas the intra-specimen variability model assumes only the former component. First, equations for the mean value, μ
N, and the coefficient of variation, η
N, of the crack propagation life
N were derived based on the composite variability model. These equations were found to be in good agreement with our experimental results. With respect to η
N, the composite variability model was in better agreement with the experimental results than the intra-specimen variability model was. The correlation distance estimated by the composite variability model was of the same order as that estimated by the intra-specimen variability model, although the former is a little smaller than the latter. It was also shown that the scatters of
CS and
ms observed experimentally when the equation
da/dN=
Cs(Δ
K)
ms is applied to each specimen could be explained by the composite variability model as well as by the intra-specimen variability model. As a whole, the composite variability model was found to be in a slightly better agreement with our experiment than the intra-specimen variability model was, but more experimental examinations are necessary before reaching a definite conclusion.
抄録全体を表示