Inference for Strength Distributions of Brittle Fibers under Increasing Failure Rate

Abstract

A different approach to the problem of estimating the strength distribution of brittle fibers is proposed. The procedure assumes a Poisson flaw model, but rather than presum ing a particular functional form, it is assumed that the probability distribution of strength has an increasing failure rate (IFR). The maximum likelihood estimate of the fiber strength distribution is obtained within the class of all distributions with the IFR property, yielding a continuous distribution estimate which is a good fit to fiber tensile strength data, typically fitting much better than the usual Weibull model. The method is illustrated with carbon fiber data.