Methodology to determine probits for glazing failure and hazard

Risk analysis tools use probit functions to quantify the risk of lethality due to glass breakage. These probit functions are often based on data for an average sized or standard sized window. In reality however, there will be a range of window sizes and thicknesses. As a result, the current probits have the following shortcomings: - The applicability of the probits to a range of windows can be debated. - The probits are not applicable for different regions or different countries, when typical glazingdimensions differ. - It is hard to exchange and compare the probits used by different countries. TNO has developed a method to determine the probit for glazing failure and hazard for any predefined range of windows. This method is based on strength and hazard models for glazing. A Bayesian Belief Network approach is used to introduce the uncertainties and the scatter regarding the range of glazing dimensions. The paper describes the method and some examples and findings are given. With this method TNO has developed two different sets of probits; one for the range of glazing in NL-domestic houses, and one for the range of glazing in Out-of-Area structures. These probits have been implemented in the NLQRA model ‘Risk-NL’.