A network-based approach to modeling safety accidents and causations within the context of subway construction project management

It is widely accepted that the construction industry is dangerous, and subway construction projects are more inherently dangerous than general construction projects. On the basis of identifying intra-relationships between causations, intra-relationships between accidents, and inter-relationships between causations and accidents, this research aims to develop a subway construction safety risk network (SCSRN) with two layers of 56 causations and 44 accidents. Topological features were identified to denote the roles of individual nodes of causation or accident in the total network of SCSRN. Robustness assessment was conducted in the consideration of three strategies of node failure, namely, random attack, deliberate attack according to the decreasing order of node's total degree, and deliberate attack according to the decreasing order of node's betweenness centrality. The results indicated that the second strategy was more efficient in reducing the connectivity of SCSRN. It was suggested that managers should not pay equal attention to causations and accidents with unequal roles in the SCSRN. This research was beneficial to capturing complex cause-and-effect relationships of accident occurrence at subway construction sites. It can provide the opportunities for optimizing safety-related decisions on the elimination of causations and controlling accidents with high efficiency on site.