Simulation-based structural reliability applied to fracture mechanics

Resumo

Monte Carlo simulation techniques, applied to structural reliability analysis, have always al-
lowed the solution of complex, large and non-linear problems. However, these techniques demands a

high computational cost in problems presenting small probability of failure. In this context, intelligent

sampling techniques are used to reduce the number of simulations needed to solve structural prob-
lems, reducing the processing time. This paper addresses a study on different combinations of sampling

strategies, such as the Latin Hypercube Sampling, Antithetic Variates Sampling, Asymptotic Sampling
and Enhanced Sampling, all applied within the Monte Carlo technique. The models are applied to

benchmark problems in Linear Elastic Fracture Mechanics, more specifically those ones presenting ana-
lytical solution. The advantages and limitations of each method are discussed based on the accuracy of

the probabilistic response and the associated computational cost.