A comparative study of statistical methods for critical parameters at tempered glass fallout in an enclosure fire

Abstract

Tempered glass is substituting float glass and extensively employed in glass façades and some structural members. However, previous studies primarily focus on float glazing, little is known about the statistical characteristics of tempered glazing fallout in a compartment fire. In particular, the correlation between critical parameters and fallout time of tempered glass is still unknown. Based on the previous experimental data in open literature of 11 tempered glass panels in a compartment model, a comparative analysis was performed to optimize the statistical methods of the breakage behaviour for tempered glazing. The parameters from the previous Wang et al.’s research, such as tempered glass fallout time, the temperature at the center of tempered glass surface, the temperature at the edge of tempered glass surface, the temperature difference between fire exposed and frame covered areas, and total heat flux at tempered glass fallout, were analyzed by 7 different statistical functions, including the Normal, Lognormal, 2-parameter Weibull, 3-parameter Weibull, Exponential, Gamma and Logistic distributions. The results suggested that the most appropriate distributions for tempered glass fallout time, the temperature at the center of tempered glass surface, the temperature at the edge of tempered glass surface, the temperature difference between fire exposed and frame covered areas, and total heat flux at tempered glass fallout are respectively Lognormal, 3-parameter Weibull, 2-parameter Weibull, 3-parameter Weibull, and 3-parameter Weibull distributions. Overall, the 3-parameter Weibull distribution is more suitable for the statistical analysis of fallout behavior of tempered glazing in fires. For critical parameters of tempered glass fallout, the 50th percentile is close to the most probable percentile. To avoid the fallout of 6 mm thick tempered glazing, the temperature difference between fire exposed and frame covered areas as well as heat flux should be controlled under 319 °C and 38.5 kW/m2. From the correlation analysis between fallout time and each critical parameter, it was established that the surface temperature may give more reliable prediction than total heat flux for tempered glass fallout.