Incorporation of Engineering Model Uncertainties in Probabilistic Analysis of Explosive Effects on Tunnel Facilities

Probabilistic methods to predict explosive effects against underground structures have traditionally been concerned primarily with site characterization uncertainties such as rock layering and rock property variability. However, uncertainty in the predictive models themselves can have an equal or greater impact on the probability of structural collapse. Uncertainties were quantified for equations used to predict the effectiveness of explosive penetrators, by comparing the engineering models to test data in which the rock properties were measured. The engineering model uncertainties were incorporated into probabilistic explosive effects software along with a confidence-based convergence algorithm that determines when sufficient simulations have been performed in order to confidently know the distribution of possible outcomes. The influence of engineering model uncertainties is demonstrated by simulating an explosive event against a highway tunnel facility. The predicted outcome (including expected rubble volume and the probability of catastrophic failure) is compared for simulations that incorporate only uncertainties in rock layering and rock properties, and simulations that incorporate engineering model uncertainties in addition to site characterization uncertainties.