When compared to observations from historical data, statistical risks often reflect significant discrepancies. The basic problem concerns the way objective statistical risks are represented. Risk is defined as the product of the probability of an incident and its consequent damages, averaged over all possible outcomes. Statistical risk can be expressed more completely as a relationship between the frequency of occurrence of an event (f) and the number of people affected (n). Because f-n curves represent the entire risk spectrum associated with each incident, they can also be used to estimate anticipated interval times between consequent events for the transport of a given class of dangerous commodity. A number of f-n curves for the road and rail transport of chlorine and liquefied petroleum gas were established. These results were compared with f-n curves reported elsewhere in the literature. Discrepancies between statistical fatalities from the risk model estimates and observed fatalities from the data are explained in terms of the expected interval times between designated events for each mode and type of material.
Abstract