Dynamic reliability-centered maintenance modeling integrating failure mode analysis and Bayesian decision theoretic approaches

As industries face increasing pressure to maintain complex systems with minimal downtime and optimized cost structures, traditional static maintenance strategies fall short of addressing real-time uncertainty and evolving operational conditions. Reliability-Centered Maintenance (RCM), while foundational, must adapt to incorporate probabilistic reasoning and continuous decision-making to remain effective. This paper presents a dynamic RCM modeling framework that integrates Failure Mode and Effects Analysis (FMEA) with Bayesian decision-theoretic approaches to enable real-time, risk-informed maintenance interventions. The model begins with a comprehensive failure mode mapping using FMEA to identify critical assets, failure causes, and effects. Each failure mode is assigned dynamic risk priority numbers (RPNs) that evolve based on operational data, sensor inputs, and environmental variability. A Bayesian belief network is layered onto this framework to update prior failure probabilities as new data becomes available, capturing the stochastic nature of degradation. Decision nodes within the Bayesian structure enable cost-risk trade-offs to be evaluated in real time, ensuring that optimal maintenance actions are selected under uncertainty. Furthermore, the model accommodates adaptive learning through posterior updates, refining both failure predictions and policy recommendations over time. A case study involving an energy generation plant demonstrates a 31% improvement in mean time between failures (MTBF) and a 24% reduction in maintenance costs. The dynamic fusion of qualitative failure analysis with quantitative Bayesian inference allows for smarter, context-aware decision-making and predictive readiness. This study provides a robust framework for implementing intelligent RCM strategies in Industry 4.0 environments where responsiveness, resilience, and safety are paramount.