Cumulative damage models for system failure with application to carbon fibers and composites

To estimate the probability distribution of the tensile strength of materials, specimens; tested are typically of a variety of sizes, or ''gauge lengths.'' Here, a cumulative damage argument is used to, develop nets models for the strength of a general system that incorporate the size as a variable and are strong competitors to the commonly used Weibull model. These strength distributions can be represented as three-parameter versions of the Birnbaum-Saunders distribution and the inverse Gaussian distribution. Estimation is discussed and illustrations are presented for experimental strength data for carbon fibres and small composite specimens.