Network topological approach to modeling accident causations and characteristics: Analysis of railway incidents in Japan

Abstract

Railway networks are a critical infrastructure in today's society, and many efforts have been made to promote their safety. Even when safety measures are introduced, railway incidents still occur. Incident investigation can acquire facts about the causation factors and effects of incidents. In this study, we propose a network analytical framework to identify the factors and effects of railway incidents. This framework captures the complexity of interdependencies and delineates the causation of railway incidents in incident chains; the incident chains are then remodeled to study the topological characteristics of factors and effects in a network model. Reported railway incident cases in Japan are selected for analysis from three perspectives: local view analysis, global view analysis, and contextual view analysis. Statistics and network diagrams are developed from each perspective to analyze the relational railway incident data. The results show that most railway incidents are caused by more than one factor, and one factor tends to cascade into others. Each factor has its own role in the occurrence of an effect, but interdependencies can still be found among most factors. The significant factors in railway incidents are revealed in this study, and it is suggested that tackling the occurrence of the significant factors and their interdependent factors is effective for controlling and reducing the occurrence of railway incidents. The network analytical framework can be used to forward estimate the effect specifically caused by a factor and its propagation exposure, as well as to backward determine the possible causes of an effect from its interdependent intermediates and factors.