Introduction of the human factor in the estimation of accident frequencies through fuzzy logic

Abstract

The frequency of occurrence of an accident scenario is one of the key aspects to take into consideration in the field of risk assessment. This frequency is commonly assessed by a generic failure frequency approach. Although every data source takes into account different variables, aspects such as the human factor are not explicitly detailed, mainly because this factor is laborious to quantify. In the present work, the generic failure frequencies are modified using fuzzy logic. This theory allows the inclusion of qualitative variables that are not considered by traditional methods and to deal with the uncertainty involved. This methodology seems to be a suitable tool to integrate the human factor in risk assessment since it is specially oriented to rationalize the uncertainty related to imprecision or vagueness. A fuzzy modifier has been developed in order to introduce the human factor in the failure frequency estimation.

In order to design the proposed model, it is necessary to consider the opinion of the experts. Therefore, a questionnaire on the variables was designed and replied by forty international experts. To test the model, it was applied to two real case studies of chemical plants. New frequency values were obtained and together with the consequence assessment, new iso-risk curves were plotted allowing to compare them to the ones resulting from a quantitative risk analysis (QRA). Since the human factor is now reflected in the failure frequency estimation, the results are more realistic and accurate, and consequently they improve the final risk assessment