A probabilistic model for the design of building glazing plates of monolithic glass in buildings is presented here. The model is implemented for practical use in Windesign, a computer program developed by the authors. A novel contribution of the model is the consideration of maximal crack sizes as a reference parameter to predict material fracture instead of the critical strengths of the materials, as is usually the case. A sensitivity analysis is also included to assess the relative influence of the different parameters on the model.

A central point in a probabilistic model for designing brittle materials is how to analytically determine the statistical distribution of the critical strength from experimental data. The cutting process used in the preparation of specimens is frequently the source of edge defects causing failures at stresses generally lower than those that have originated from micro-cracks at the surface. Since this kind of failure occurs almost exclusively in certain testing procedures but is not present in the real, edge-polished glazing elements, adequate treatment of this data is needed to derive the cumulative distribution function (cdf) characterizing the surface strength of the material. Neglecting data associated with the edge failures results in an erroneous estimation of the cdf. This paper describes the statistical procedure for the evaluation of this kind of data (known as confounded data) that permits the correct estimation of the cdf of the surface strength taking into account all the results obtained in the tests, irrespective of their origin. The theoretical model is applied to experimental data resulting from 4-point bending tests on glass specimens.