Integrating Functional Resonance and Bayesian Networks for Quantitative Risk Assessment: Application to Adrar's Refinery Pre‐Fractionation

Quantitative risk assessment (QRA) plays a critical role in managing safety in high‐risk industries such as oil and gas refining. Conventional QRA methods, including fault tree analysis (FTA), event tree analysis (ETA), and Petri nets, offer structured modeling frameworks but are limited by their linear and reductionist assumptions, making them less effective in capturing variability propagation and complex interdependencies. To address these limitations, this study proposes a hybrid methodology that integrates the functional resonance analysis method (FRAM) with Bayesian networks (BNs). The approach systematically combines the qualitative strengths of FRAM, which captures functional variability and systemic interdependencies, with the quantitative capabilities of BN, which models probabilistic uncertainty. A structured five‐stage framework is developed, including FRAM modeling, BN mapping, probabilistic parameterization (using data bases and expert elicitation), risk estimation, and sensitivity analysis. The methodology is applied to a refinery pre‐fractionation system, demonstrating its practical value in identifying critical functions and quantifying their influence on overall system risk. Comparative analysis shows that the FRAM‐BN not only overcomes the reductionist limitations of traditional QRA methods but also provides deeper insights into risk variability, resilience, and decision support. Overall, the proposed FRAM‐BN represents a significant step forward in QRA, offering a robust tool for both researchers and practitioners seeking to strengthen safety and resilience in socio‐technical systems.