On the Difficulty to Implement the Coupled Criterion to Predict Failure in Tempered Glass

Defining the tensile strength of a brittle material like glass is not easy and the presence of residual stresses further complicates the problem. Indeed, measuring the tensile strength by a standard bending test in glass specimens is more a measurement of the presence of surface defects than the determination of a material parameter. This explains that it has not a deterministic value but is defined through a probabilistic distribution which makes it difficult to use the Coupled Criterion (CC) which specifically requires the knowledge of the tensile strength together with the material toughness to be implemented so as to predict crack nucleation. In addition, the residual stresses move the stress peak inside the specimen while failure occurs at the surface due to a critical defect, which further hinders the implementation of the CC. We show that it is necessary to ensure that crack initiation does occur in the vicinity of a stress peak to obtain reliable fracture predictions thanks to the CC. The data of the tensile strength then provides a reliable critical failure load without it being necessary to finely model the defects which are at the origin of the rupture.