Accuracy of stochastic finite element analyses for the safety assessment of unreinforced masonry shear walls

To examine structural safety and reliability, an accurate prediction of the variability of structural resistance must first be determined. This may be achieved through extensive physical testing or, more commonly in modern research, synthetic data generation, such as stochastic finite element analyses (SFEAs). Due to the prevalence and versatility of such techniques, and the need for a high level of confidence when performing a safety assessment, an understanding of the accuracy of data derived from SFEAs is essential. In this paper, SFEA models have been developed to predict the responses of 16 unreinforced masonry walls tested in a laboratory under cyclic in-plane lateral loading. Each SFEA has been developed to reflect the as-built conditions of these experimental specimens and focus on predicting the shear capacity of the failure mechanisms that unreinforced masonry shear walls are susceptible to. From the results of these SFEAs, the accuracy (quantified as model error) of this modelling strategy has been estimated by comparing the peak in-plane shear resistances of the laboratory and numerical models.