Descriptive modelling of team troubleshooting in nuclear domain

Even though modern engineering systems are becoming more reliable, there is evidence that main control room (MCR) operators and maintenance technicians have problems in diagnosing complicated events and multiple simultaneous events in nuclear power plants (NPPs). There is a lot of research on finding faults in complex incident situations, but there is quite little knowledge of the cognitive strategies, states and activities of MCR teams in troubleshooting situations. This knowledge is, however, important in modelling collaborative diagnostic reasoning and in developing training interventions and decision support systems to support reasoning and problem solving in NPPs. The present paper will firstly critically review existing methods to model collaborative cognition. Our results suggest that some modelling tools, such as social network analysis and information network analysis are valuable in analysing information sharing and team situation awareness in collaborative troubleshooting, but they do not provide detailed information about the sequential evolution of a team’s knowledge state throughout the diagnostic process. On the other hand, the tools representing an individual troubleshooter’s successive knowledge states, transformed by information processing activities, are ignorant of the distributed nature of complex troubleshooting. Therefore, these approaches have to be tailored to suit better the analysis of team collaboration and co-operation in incident and accident situations, i.e., we have to better understand the collaborative aspects of cognition and team-level collaboration in fault finding and in diagnosing faults. We propose a modelling approach suitable for analysing collaborative diagnostic reasoning and troubleshooting of a NPP MCR crew which is based on existing methods and tools. The approach describes the progress and evolution of a MCR operator crew’s knowledge states throughout the critical sections of a simulator run.