|
Abstract:
|
The explosive PETN (pentaerythritol tetranitrate) is widely used in detonators and boosters for commercial blasting as well as munitions. Safety is a prime concern, since blasting materials or munitions may be involved in accidents causing exposure to mechanical shock, electrical discharge, or fire. Safety requirements may demand that the probability of accidental detonation be as low as 10^-6 in such situations, a level that is difficult to justify based solely on test data. We present a case study in which a physical model for PETN decomposition caused by heat is used to develop a prior distribution for the mean temperature at which detonation will no longer occur. The prior is updated in a probit analysis with a small dataset of experimental data to justify the conclusion that a safety criterion is satisfied. Our methodology is generally applicable to a variety of safety scenarios in which physical models are combined with empirical evidence to estimate probabilities of rare events.
|
