Modeling tunnel construction risk dynamics: Addressing the production versus protection problem

Accidents remain a pervasive problem in tunnel construction. A major contributor to these accidents is a construction contractor's inability to determine an appropriate trade-off between production and protection goals. Thus, to examine this issue, a systemic model, which integrates System Dynamics (SD), Bayesian Belief Networks (BBN) and smooth Relevance Vector Machines (sRVM) (referred to as 'Organizational Risk Dynamics Observer' (ORDO)), is developed to investigate the mechanism of risk migration that resulted from the interactions between a contractor's organizational and technical systems. The model is demonstrated on an urban metro tunnel project that was constructed in Wuhan, China. It is revealed that when attention focused upon production, the propensity for minor accidents to occur increased, which triggered management to focus on protection. This increasing emphasis on protection may have muted the safe systems of working as incidents may be unreported, which can inhibit the motivation for safety awareness. When coupled with an increase in production pressure, the tunneling project could become prone to experiencing a major accident. Based on the results, it is suggested that the whole organization must continue to foster a sound safety culture by resisting production pressure at the expense of compromising safety.