(A) new approach to accident analysis based on activity system model

Abstract

This study proposes a new approach for modelling and analysing human-related accidents based on activity theory (AT). Most of the existing methods seem to be insufficient for comprehensive analysis of human activity-related contextual aspects of accidents when in-vestigating the causes of human errors. Additionally, they identify causal factors and their interrelationships with a weak theoretical basis. More seriously, the existing methods have some drawbacks which have been strongly pointed out in the introduction of a new para-digm of system safety called Safety-II as follows: (1) an over-simplified approach that iden-tifies linear, simple cause-effect relationships, (2) looking for plausible causal factors only based on a causation model assumed in a method for analysing accidents, and (3) neglecting other possible causes by over-relying on a root cause. We argue that activity theory offers useful concepts and insights to overcome (or at least lessening) the drawbacks of the existing accident analyses. Furthermore, AT is not a causation model for accident analysis, but can be used as a method for describing human activities with their contextual back-grounds in a system. In this study, we proposed a new approach to human-related accident analysis based on activity theory. The AT-based approach encourages analysts to look at human-related accident situations with a more broad perspective and to consider as many different possible causal factors as possible, which can be meaningful to an accident under investiga-tion. Moreover, the AT-based method can be effectively used to mitigate the limitations of traditional approach to accident analysis, such as over-relying on a causality model and sticking too much to a root-cause, by making analysts look at human-related accidents from a range of perspectives. The case studies in Korean nuclear power plants were conducted to demonstrate the applicability of the proposed method. These case studies showed that the proposed approach could produce a meaningful set of human activity-related contextual factors, which cannot easily be obtained by using existing methods. It can be especially ef-fective when analysts think it is important to diagnose accident situations with human activ-ity-related contextual factors derived from a theoretically sound model and to identify acci-dent-related contextual factors systematically. We also proposed an integrated approach for analysing human-related accidents by combining AT-based approach and HFACS (Human Factors Analysis and Classification System) which offers a comprehensive viewpoint and failure taxonomy for analysing human errors. By combining them, the proposed method can help safety professionals consider a range of failure types and their relevant failure modes that can be regarded as causes for a human-related accident and examine contextual factors and their interrelationships with a sound theoretical basis. In order to demonstrate the usefulness of the proposed method, we conducted a case study in nuclear power plants. Through the case study, we could confirm that the proposed method would be a useful method for modelling and analysing human-related accidents, enabling safety professionals to identify a plausible set of causal factors efficiently in a methodical consideration of contextual backgrounds surrounding human activities.