Development of the Microstructure Based Stochastic Life Prediction Models

摘要

The goal of this program was to develop a methodology for predicting fatiguelives of structural alloys based on their microstructural characteristics and mechanical properties. Such methodology has been successfully implemented and verified by predicting fatigue lives of the four different variants of the aluminum airframe 7050-T7451 plate alloy and, with modifications, of the butt welds of the ship hull HSLA-80 steels. The key features of the method is the assumption that the incipient fatigue crack size distribution is related to the size distribution of the bulk material flaws through the statistics of extreme. When combined with the Monte-Carlo (MC) crack growth model the extreme value estimates of the initial crack sizes gave excellent predictions of the fatigue lives of the 7050-T7451 alloy for samples both with and without stress concentrators. The specially for this purpose developed MC model utilized initial crack size distribution, crack location, crystallographic texture on the crack path and crack deflections as the random variables. A modified, closed-form three-parameter version of the model has been proposed for the butt welds of the HSLA-80 steels. This version, with parameters obtained based on the constant amplitude data, showed outstanding predictive capabilities for the samples with welds subjected to the variable amplitude loading conditions. Both versions of the model represent very useful and economical alternative to the lengthy fatigue testing programs. They allow for rapid differentiation between fatigue qualities of different material variants and on the parametric studies of the effects of the microstructural variables on fatigue lives. The methodology should instrumental in aiding alloy designers and process engineers in optimizing alloy microstructures for fatigue performance.