Blast Design-Basis Threat Uncertainty and Its Effects on Probabilistic Risk Assessment: Fid-Bau Portal

Blast Design-Basis Threat Uncertainty and Its Effects on Probabilistic Risk Assessment

Campidelli, M. / El-Dakhakhni, W. W. / Tait, M. J. / Mekky, W.

At present, the deterministic approach to blast-resistant design appears to be the prevalent philosophy in practice, as suggested by the notion of design-basis threat (DBT) outlined by the ASCE/SEI 59-11. In this study, it is argued that even in the case of a clearly defined blast scenario, the resulting loading parameters can show a remarkably high degree of uncertainty, as corroborated by data from arena tests. Therefore, future standards should integrate a DBT definition guided by probabilistic risk assessment (PRA), in order to extend the general principles of limit state design philosophy to blast-resistant design. This paper investigates the sources of the variability observed in key blast wavefront metrics, including the peak pressure, specific impulse, positive phase duration, and decay coefficient. This objective is accomplished by quantifying and propagating the uncertainty associated with several input parameters, such as the mass of explosive, equivalent TNT–mass factor, pressure transducer and specimen positions relative to the explosive charge, and atmospheric conditions. All input parameters are modeled as random variables and the wavefront metrics are subsequently calculated via the Kingery and Bulmash model and corrected for nonstandard ambient conditions. The input uncertainties are propagated through the computations using the Monte Carlo method and the resulting confidence intervals, estimated for each selected DBT, are compared with corresponding arena test measurements. Within the context of the test program presented herein, it is found that the considered input parameters partially account for the discrepancies between test data and predicted wavefront metrics, whose uncertainty is largely controlled by the equivalent TNT–mass factor. In addition, it is found that the input uncertainty is significantly amplified by the structural response model, which reveals a significant variability in the response predicted for the specimens considered in this study.