Multi-state safety analysis of civil Aero-engine: From functional block diagram to bayesian network

This paper presents a multi-state safety-analysis framework for civil aero-engines to support compliance with CCAR §33.75 (Safety Analysis). Functional block diagrams are mapped to Bayesian networks, enabling component and function behaviors to be modeled beyond binary states and capturing causal dependencies among failure modes and hazardous engine effects (like loss of thrust control, inability to shut down, and overspeed, etc.). Occurrence probabilities are computed via junction-tree propagation, allowing forward prediction and evidence-based diagnosis under varied operating conditions. Two case studies evaluate the approach against traditional methods (fault-tree analysis and failure modes and effects analysis), demonstrating improved expressiveness for multi-state dependencies, transparent probability updates, and consistent quantitative estimates for §33.75 hazard categories. The results indicate that the proposed framework provides a rigorous and traceable basis for quantitative safety substantiation and can complement existing certification workflows in civil aviation airworthiness.