Statistical Inferences on Cleavage Fracture Toughness Data

Abstract

A method to predict lower bound cleavage fracture toughness values from small data sets is presented. The method is derived from a Weibull statistic based micromechanical analysis and can be applied in the ductile-brittle transition region. Lower bound predictions can compare well with the ASME KIR and KIC design curves. Difficulties in comparing Weibull shape parameters calculated by different methods are also discussed and several calculation methods are compared with a Monte Carlo simulation. The preferred methods are the maximum likelihood estimators (MLEs), least squares with the (i−.5)/n centralizing approach, or a simple estimation scheme which gives a “Good Linear Unbiased Estimator.”