Stochastic micromechanical modeling of cleavage fracture in the ductile-brittle transition region

Abstract

The large scatter in cleavage fracture toughness values of steels in the ductile to brittle transition region has made characterization and design difficult. Current weakest link statistical models predict that cleavage toughness data will follow a two-parameter Weibull distribution with known shape parameter. This dissertation first showed the current models to inadequately describe the data, then revised the model and applied the revised model to the problem of design. Because studies indicated that crack propagation may also be important, the model was revised to include modified weakest link behavior in which the effects of propagation were included. Through the vehicle of statistics, micromechanics were used to predict the macroscopic properties. The revised model successfully predicted the trend toward higher than expected Weibull shape parameters. A three-parameter Weibull distribution was found to be an empirically effective approximation to the model and the data. The revised model was applied to the problems of nuclear pressure vessel design in the transition region, and to the problem of local brittle zones in weldments. A Master Curve was developed that was found to be a useful substitution for or complement to the ASME design curves. It successfully characterizes the distribution of cleavage fracture toughness data in the transition and lower shelf regions aim provides statistical tools for designers. An order statistics approach was developed for small censored data sets for which the master curve method was inappropriate. Predicting the toughness of weldments with local brittle zones was also made more difficult by the breakdown of weakest link assumptions. Results were consistent with model predictions that arrest toughness of the surrounding material plays a major role in the importance of the local brittle zone.