An approach to estimating the individual risk for toxic-gas releases using the load-resistance model

• Load-resistance model was incorporated into toxic gas risk assessment.
• Directional effect by wind direction must be considered for dispersion modeling.
• Separable Monte Carlo takes advantage of improving the computing efficiency.
• Exposure distribution tends to have long right-tail behavior.
• Risk estimates corresponding to a percentile can be directed to wrong decision.

A new approach to quantify the uncertainty of the individual risk for toxic releases is presented in this paper. The individual risk is defined as the probability of fatality per year. The probability of fatality is calculated by a classical load-resistance model based on reliability (survivability) theory. The load effect is defined as the concentration intensity to which a human is exposed. Furthermore, the resistance is defined as the human tolerance to a certain concentration load in this study. The Monte Carlo method is used to obtain the probability distributions of outputs (the load effect and resistance) propagated from the uncertainties of the input variables. The fatality probability exceeding a limit state can then be obtained by comparing pairs of samples from the load effect and the resistance distributions. The separation of sampling from the load and resistance distributions is also proposed to allow more efficient calculation than that achieved by the classical Monte Carlo method. The analytical risk estimates computed by the load-resistance model are compared to conventional risk estimates that correspond to the upper-end percentile of the load-effect distribution. A case study shows that the conventional risk estimates can be directed to wrong decisions when the load-effect distribution has upper-end tail heaviness.