Safety improvement in a gas refinery based on resilience engineering and macro-ergonomics indicators: a Bayesian network-artificial neural network approach

摘要

The risk of accidents at workplaces, particularly in the sensitive locations with unsafe behaviors, have increased substantially, needing to be managed accurately. To ameliorate the safety in such systems, enhancing the integrated resilience engineering and macro-ergonomics concepts is of pivotal importance. In this sense, this paper unveils a novel method based on Bayesian network and artificial neural network models to enhance safety of such systems considering both mentioned concepts. Exploiting the Bayesian network, the effects of the indicators on the system safety efficiency is evaluated according to the expert's opinions. The Artificial neural network examines these effects based on the operator's opinions. Thereinafter, to decrease the uncertainty and bias of results and also augment the robustness and accuracy of them, the combination of the results of these models is considered as the final criterion. For analyzing the efficacy of the proposed method, a case study in a gas refinery in Ilam, Iran is conducted. The results corroborate the validity and efficacy of the proposed method and draw outstanding managerial insights.