Storage tank uncertainty dynamic risk assessment based on fuzzy dynamic Bayesian network

Abstract

In order to overcome the static nature of traditional risk assessment and analyze the uncertainty propagation problem in risk assessment, this paper proposes a new dynamic risk assessment (DRA) methodology to quantify the input data uncertainty propagation in the tank risk analysis. In this paper, the bow-tie (BT) model is developed to identify the key factors and possible consequences of tank failure. The hierarchical Bayesian analysis (HBA) model is utilized to calculate the failure probabilities of the basic events and protective layers. The triangular fuzzy number in fuzzy set theory (FST) is utilized to preserve the uncertainty of the results, whereby fuzzy prior probabilities and fuzzy transfer probabilities are calculated, and then risk diagnosis and probability updating are performed by dynamic Bayesian network (DBN). Taking the floating-top storage tank leakage accident as an example, the uncertainty in the risk analysis is reduced by calculating the possibility range of the risk occurrence, and the possible accidents are accurately predicted. The results show that the method is able to objectively assess the risk level of most events and quantify the process of uncertainty propagation in the input data.