A safety risk assessment method considering epistemic uncertainty

Abstract

In the areas of safety risk assessment, how to consider uncertainty properly becomes a challenge. To resolve this problem, a new system state description method was presented firstly in the DET, that is system state sequence which can record the evolvement process of accident, and then the arithmetic of state sequence probability was designed. The probability of accident can be calculated easily in the DET based on those probabilities. Because the uncertainty of input parameters is ubiquitous in the safety risk assessment model, the QMU (Quantification of Margins and Uncertainty) method was introduced. Using LHS (Latin Hypercube Sample) to research the uncertainty of risk index was adopted, at the same time, the margins and uncertainty of risk index can be obtained which is also the information of safety risk assessment. To demonstrate the performance gains of our model, a cooling liquid tank system was analyzed as an example. The result shows that these methodologies can be used to assess risk more conveniently and credibly. Furthermore, the conclusions can be used to guide the risk control and improve system safety.