Statistical lifetime predictions for aramid fibers

This paper investigates the statistical procedures that can be used to analyze stress rupture data for aramid yarns, with a view to making reasonable predictions for the allowable prestress levels in parallel-lay ropes or fiber-reinforced polymer tendons for use as prestressing tendons in concrete, structures. Two existing data sets are combined and used to illustrate the principles that are being discussed. Arrhenius, exponential, and inverse power physical models are used with Weibull and lognormal statistical distributions. A wide range of variations of the scale and shape parameters are analyzed. Various statistical criteria are used to reject models that are not statistically secure, and the Kullback-Liebler and Akaike information criteria are used to further limit and rank the possible models. The results show that there is still a large range of possible predictions for the long-term stress-rupture lifetime of aramid yarns, but also that statistical models are available for investigating further test work that needs to be carried out.