A novel fault injection and monitoring methodology for freight train air brakes: from simulation to field validation

Effective monitoring of the air brake system is of paramount importance, as early fault detection can prevent severe failures. Despite the progress made in Condition Monitoring (CM) and Condition-Based Maintenance (CBM) for freight train brake system, current solutions face several challenges in detecting two major malfunctions: air brake leakages and manual brake activation. Existing methodologies for fault data collection are hindered by key limitations: reliance on simulations or test rigs that do not fully capture real-world operating behaviour, employment of non-optimized monitoring system tailored to the specific application and lack of robust validation under realistic conditions. This paper presents an innovative methodology to address these gaps by integrating model-based fault injection for monitoring system design, custom fault injection devices for controlled fault reproduction on full-scale freight wagons and experimental validation in real-world condition. The latest has been carried out on the closed-circuit test of San Donato where air brake leakages and manual brake activation have been systematically simulated under varied test scenarios including different train speed, braking action, brake mode, and wagon loading. The results indicate that the maximum brake cylinder pressure is a reliable indicator of combined leakages during service and emergency braking, whereas the pressure build-up gradient proves effective in detecting both manual brake activation and combined leakages during service and emergency braking in freight mode. The proposed methodology provides a scalable and reliable foundation for real-world deployment and data collected through the field validation represents a highly valuable fault dataset for the future advancement of diagnostic algorithms.