ABSTRACT
This study involves the probabilistic assessment of various industrial accidents in terms of the damage caused by them. The distribution of damage due to industrial accidents can be described using a Pareto distribution that appears a log-log straight curve, which is referred to as a risk curve. The parameter of the damage distribution can be estimated as a general posterior distribution by the Bayesian estimation method. By compounding the damage distribution with the estimated parameter distribution, the data-based predictive distribution is determined. The probabilistic assessment scheme that can analyze damage magnitudes by this predictive distribution using only observed accident data without estimating a distribution parameter is presented.
