A network-based approach to model-based safety analysis integrating physical and topological representations for complex aerothermal-mechanical systems

Safety analysis of complex aerothermal-mechanical systems like aero-engines is crucial for risk reduction and failure prevention. Model-based safety analysis (MBSA) offers an effective framework for this purpose. This study proposes a novel MBSA approach integrating physical simulation with complex network theory for critical risk component identification. First, a component risk correlation network (CRCN) is constructed as a directed weighted network with node attributes. Node risk attributes are quantified using an improved risk priority number (RPN), in which severity is evaluated through a combined physical-topological approach. Physical severity is obtained via system-level modeling and fault injection simulations, while topological severity reflects the structural importance of components. The network edges represent inter-component risk correlations, and their weights are determined using the analytic hierarchy process (AHP) to reflect risk propagation differences. Second, a novel gravity-inspired indicator is proposed to identify critical components by incorporating both local and global risk characteristics. Finally, the analysis results support the development of targeted safety control strategies. The proposed method is applied to a CD-135 aviation piston engine, identifying critical risk components that are often overlooked by conventional approaches. The findings are also validated against the manufacturer’s sensor placement, providing preliminary support for its practical relevance.