Blast Load Variability and Accuracy of Blast Load Prediction Models

Abstract

A statistical analysis of explosive blast loading field (test) data has revealed a high level of variability of peak reflected pressure, impulse and time of positive phase duration for repeatable tests where variability would be expected to be a minimum. The model error (accuracy) of a widely used predictive blast load model is also assessed. A probabilistic model of blast loading is then developed that considers variability and/or uncertainty of explosive mass, net equivalent quantity of an explosive in terms of TNT mass, stand-off distance, air temperature, air pressure, inherent variability and model error. Two widely used explosives are considered: Tritonal (military) and ANFO (terrorism). This type of statistical and probabilistic analysis is essential for structural reliability analysis of structures subject to explosive blast loading where load variability is an important contributor to damage and safety risks. It was found that the TM5–1300 design values for peak reflected pressure and time of positive phase duration adequately represent median values of the probability distribution of blast loads. The TM5–1300 design values for peak reflected impulse were 40% higher than median values with probabilities of exceedance of only 4% to 23%. This over-estimation of actual blast loads on a structure may lead to conservative design outcomes.