Dynamic Risk Analysis Of Chemical Process Based On Alarm Database

Chemical accidents are of low frequency, but high consequence. The analysis of chemical process is conducive to enhancing the risk management in the process of chemical production, thus to reduce the potential chemical accident. The traditional methods of risk analysis are always based on the major chemical accidents. However, the shortage of accident data has weakened their performance. Near-miss is the precursor of accident. In chemical process, a large amount of near-miss data are recorded in distributed control system (DCS) and emergency shutdown system (ESD) alarm databases. Therefore, much risk information id hidden in the alarm databases.In this paper, a dynamic risk analysis method is put forward by utilizing the alarm databases. Based on the theory of Bayesian theory and copula theory, the method established two models to perform dynamic risk analysis to describe the chemical risk from the perspective of frequency and probability respectively. In frequency perspective, taking the operator teams into consideration, a model is established to predict the near-miss accident frequency. A high near-miss frequency indicates the high accident probability. In probability perspective, three models are built to assess different incidents probability (e.g. High/Low alarm, High-High/Low-Low alarm) of key variables considering the correlation between variables. The near-miss frequency predicted and the probability assessed reflect the risk level of chemical process, which can be used as risk reference index, improving risk management in the chemical process...