Applicability of statistical models for risk analyses - a probabilistic model of support systems for prevention of right-turn collisions

摘要

In quantitative risk analyses, the top event of fault tree (FT) is originated from basic events through logical causation mechanisms. The occurrences of basic events are often modeled with failure and repair rates with exponential distributions. However, the exponential distribution could not always reflect the reality in some cases where normal or other types of distribution should be adapted. In this paper, Monte Carlo simulations are practiced for the risk estimation of turning to the right collisions caused by automobiles. Comparisons are made between two statistic models, exponential and normal distributions. The result shows the deviation rates are within 10% for those cases. It is concluded that the exponential distribution is applicable to the probabilistic models of FTA where normal distribution would be more suitable.