Explaining the discrepancy between design expectations and experiments on a tensile bolted joint for laminated glass

We discuss a probabilistic methodology to align theoretical strength predictions with experimental results, by establishing quantitative relationships between failure probabilities and applied test loads. This enables reliable safety margin verification for structural designs even when only limited experimental data is available on full-scale prototypes. The methodology is demonstrated through the case study of a structural bolted joints connecting large, non-standard four-ply laminated heat-tempered glass plates. The research combines experimental testing with a theoretical interpretation grounded in probabilistic mechanics. Observing that the expected strength of the joint, derived from semi-probabilistic design, is much lower than the actual test results, our study examines how key factors, primarily lamination effects on redundancy and statistical interference between intrinsic glass strength and heat-induced surface prestress (tempering), can influence structural capacity. While the application is specific, the underlying approach remains applicable to other structural systems. The analysis demonstrates that accounting for the statistical interaction of multiple variables, overlooked in semi-probabilistic methods, can lead to strength predictions up to 125% higher than those obtained through the semi-probabilistic approach, in agreement with experimental findings. This improved approach forms the basis for enhanced verification methodologies, promoting optimized material utilization and mitigating excessive conservatism in structural glass design. • Structural capacity of a bolted joint in a laminated-glass stringer was experimentally assessed. • The capacity highly surpassed the theoretical estimates derived from the semi-probabilistic approach. • The semi-probabilistic method neglects statistical interference and the redundancy of laminates. • A full-probabilistic model indicates a tensile capacity nearly double the semi-probabilistic estimate. • Standardized design rules should be refined to improve the predictions of joint capacity.