https://www.mdu.se/

The key to system safety is the identification and elimination/mitigation of potential hazards and documentation of evidences for safety cases. This is generally done during the system design and development phase. However, for automated systems, there is also a need to deal with unknowns and uncertainties during operational phase. This paper focuses on virtual boundaries around geographic zones (i.e., geofences) that can serve as an active countermeasure for dynamic management of risks in automated transportation/production contexts. At first, hazard analysis is performed using the Hazard and Operability (HAZOP) and Fault Tree Analysis (FTA) techniques. Based on the hazard analysis, appropriate measures, such as geofences for elimination/mitigation of hazards are defined. Subsequently, they are translated into the safety requirements. We leverage on simulation based digital twins to perform verification and validation of production site by incorporating safety requirements in them. Finally, to manage risks in a dynamic manner, the operational data is gathered, deviations from specified behaviours are tracked, possible implications of control actions are evaluated and necessary adaptations are performed. The risk management is assured in situations, such as communication loss, subsystem failures and unsafe paths. This approach provides a basis to fill the gaps between the safety cases and the actual system safety emanating from system/environment evolution as well as obsolescence of evidences. The applicability of the proposed framework is exemplified in the context of a semi-automated quarry production scenario.