Probabilistic failure analysis for real glass components under general loading conditions

In this paper, the generalized local model (GLM) is applied to derive the primary failure cumulative distribution function (PFCDF) of annealed glass in order to achieve the failure prediction of structural glass. The uniqueness of the glass characterization is demonstrated irrespective of the test, specimen size, and geometry used. Consequently, the strength of glass is unequivocally derived in a probabilistic way as a material property, so that the definition of normalized testing specimens in international standards might be put under question. Furthermore, the application of the GLM to the results assessment allows to ensure a correct transferability of the laboratory data from simple specimens to the practical design of real glass components and vice versa. The feasibility of the GLM to characterize the strength of annealed glass from different test types is illustrated by means of an extensive experimental program.