John Beal, Ph.D. Candidate
Prof. Zahra Mohaghegh, Director of Research
May 29, 2025| 10:00am - 12:30pm CST
This final examination will be held in 101A Talbot Laboratory
Temporal Coupling of Maintenance Human Performance, Physical Degradation, and Digital Twin Models for Probabilistic Risk Assessment in Nuclear Power Plants
ABSTRACT: Probabilistic Risk Assessment (PRA) has played a critical role in improving the safety and performance of existing Nuclear Power Plants (NPPs) and is a key element of the U.S. Nuclear Regulatory Commission’s Risk-Informed Performance-Based (RIPB) regulatory framework. PRA methodologies have evolved over the years, and their continued advancement is increasingly important in light of the Accelerating Deployment of Versatile, Advanced Nuclear for Clean Energy (ADVANCE) Act of 2024. This research leverages modeling and simulation (M&S) to estimate the reliability of repairable components in data-scarce situations, such as new reactor designs or aging plants undergoing operational changes, including the adoption of digital twins (DT). The methodological contributions of this thesis are demonstrated using the Extremely Low Probability of Rupture (xLPR) Probabilistic Fracture Mechanics (PFM) code, applied to reactor coolant system piping susceptible to stress corrosion cracking (SCC). However, the approach is broadly applicable to a wide range of component types and degradation mechanisms. This thesis makes three key scientific contributions:
- A novel method for estimating component reliability by integrating Human Reliability Analysis (HRA)-based maintenance models with physical degradation models. Utilizing dynamic PRA techniques, the method captures the temporal, bi-directional interactions between human actions and component degradation.
- An uncertainty-based validation methodology for coupled maintenance and physical degradation models, designed for contexts lacking empirical maintenance performance data. This approach involves constructing a graphical causal model to represent sources of uncertainty and their interrelationships, followed by quantification using a Bayesian Belief Network (BBN). This supports a comprehensive treatment of epistemic uncertainties, including model-form uncertainty.
- An Integrated PRA (I-PRA) methodological framework tailored for DT-enabled NPPs. This framework simulates the dynamic, bidirectional interactions among a high-fidelity physical twin (PT), human interventions (including maintenance decision-making and HRA-based maintenance performance models), and the DT. It establishes a two-way integration between the human-PT-DT system and PRA, allowing for forward-looking plant-level risk estimates based on projected component conditions. These insights inform both near-term maintenance actions and long-term strategic and regulatory decision-making in alignment with RIPB principles.