Dynamic Risk Assessment Of Chemical Process Industries Using Bayesian Model Simulation

With the rapid development of chemical process industries, chemical production load has continuously increased, resulting in more attention paid on the safety issues. At present, most related researches mainly focus on how to realize the detection and fault diagnosis of chemical process, while few experts establish the evaluation model of chemical process from the risk dissemination mechanism. Therefore, this paper aims to establish dynamic failure assessment method with Bayesian theory to evaluate the failure probabilities of shifts’ operation in chemical process industries.This paper takes a specific chemical production as an example. Bayesian model based on the accident sequence precursor data is used to quantitative analyze the influence of shift worker’s on chemical process risk. The operational ability of shifts of operators and the correlation between different teams have significant impact on the performance of the chemical process. Meanwhile, copula function can reflect the complex nonlinear relationship between different variables. Copula selection can be important, especially when data are limited. In order to eliminate the influence of expert prior parameters, this paper uses the maximum entropy method, with less information, to choose the accurate categories of copula. Therefore, considering the sequential and coupling properties of different teams operation, this paper uses Bayesian theory combined with copula to obtain the conditional probabilities model and the Markov Chain Monte Carlo methods to estimate the risk of various teams’ operation.In addition, after abnormal events occurred, the end states of equipment operation are affected by the shifts operation. By studying the shifts operation leading to the three categories of operational end-state, the difference of four shift’s operational ability is a quantitative assessment. Therefore, based on the high temperature abnormal event tree model, analyzing and comparing different reliability of each team operation during the shifts’ skills contributes to the failure probabilities of each end-state.