Lower tail fit of the Weibull distribution for edge-line glass strength for structural use of glass

The increase use of glass from window infill panels to structural glass urged engineers to characterized the mechanical properties of glass, especially the load resistance of glass, precisely. The load resistance of a glass decreases due to the presence of flaws, that acts as stress concentrators and propagation of unstable flaws leads to fracture. Flaws in a glass is inevitable due to manufacturing process of glass, like rapid temperature change (quenching) or contact with tin bath and rollers etc. Additional flaws are introduced due to cutting of glass, abrasions, handling etc. not only in the surface but also in the edge-lines and edge-faces. Weibull distribution incorporates the dependence of strength of glass on flaws shape, size distribution and stress type to describe the strength of glass. The standard to determine the load resistance of glass in the United States, ASTM E1300, clearly states that the standard doesn’t apply to structural glass members. The industry however chooses a very small probability of breakage when dealing with the design of structural glass (in the order of 10-6 ~10-7). Though Weibull distribution is considered to be an apt distribution to measure the probability of breakage of a glass, for a small probability it is evident that the left tail of the Weibull distribution cannot be interpolated by the straight line from the plot of the cumulative probability (ln[(ln(1-Pf)] vs ln(σ)). This paper uses a modified glass failure prediction model for edge-lines put forward by the author, to analyze and compare different edge-lines by fitting a 2-parameter Weibull distribution, and also try to fit the lower tail of the Weibull distribution to calculate the load corresponding to probabilities of the order 10-6~10-7.