抄録
The target of statistical analyses for industrial accident occurrence is to provide various basic information for the decision making on the policy of loss prevention or risk management. For this purpose, it is required that the characterizing and evaluating of the potential risk of accident occurrence could be performed from the results of statistical analyses. This paper proposes a new statistical technique which can give a quantitative index to characterize the potential risk of industrial accident occurrence for many engineering systems. An approach taken in this study is based on the concept of the risk curve, in which the relationship between exceedance cumulative0frequencies and their corresponding magnitude of consequence for hazardous events is described in log-log scale. The representation of the risk for accident occurrence has been examined with the definition of risks considered by Kalpan and Garrick, and the result showed that the risk curves have been clarified to be of characterizing the risk of hazardous events. Furthermore, using a lot of collected actual accident data it was found that the risk curve can be essentially expressed with a mathematical form uniquely determined by one distribution parameter. This parameter characterizing the risk curve means the index as a measure of risk for accident occurrences. It was concluded that a possible method for statistically assessing risks of accident occurrence adopts this index (safety index). Finally, the application of the proposed methodolgy to risk evaluation for damages of equipments and materials will be presented