Probabilistic Fracture Mechanics Analyses Based on Parallel Monte Carlo Method.

Abstract

This paper describes a probabilistic fracture mechanics ( PFM ) program based on the parallel Monte Carlo (MC) algorithm. In the present parallel algorithm, a sampling space of probabilistic variables such as fracture toughness values and crack depth and aspect ratio of initial semi-elliptical surface cracks is divided into a number of small cells. Fatigue crack growth simulations and failure judgements of samples are performed cell by cell on a parallel computing system consisting of multiple micro-processors, TRANSPUTERs (T800). As an example, the developed PFM computer program was applied to the analyses of PFM problems of aged RPV material. The results showed that break probabilities of the analyzed model were of an order of 10-7 and that the performance of parallel processing was over 90%. It was also demonstrated from these analyses that degradation of fracture toughness values due to neutron iraddiation influences significantly influences failure probabilities.