The Socio-Technical Risk Analysis (SoTeRiA) Research Laboratory at the University of Illinois at Urbana-Champaign supports the development of next-generation Probabilistic Risk Assessment (PRA) via advanced simulation-informed dependency modeling, uncertainty-based validation, and treatment of socio-technical interactions involving humans, physical systems, AI-driven automation, and organizational factors. This paper reports recent progress from two sponsored research efforts that focus on the advanced simulation-informed, socio-technical, and AI-Driven approaches.

First, the paper discusses progress from an ongoing DOE-sponsored and industry-cost-shared project on simulation-informed dependency treatment to enhance operational flexibility, avoid production loss, and maintain safety of operating plants and advanced reactors. This project develops a new methodological framework, Simulation-Informed Probabilistic Dependency Analysis (SIP-DA), for treatment of dependencies in nuclear plant PRA, using hardware common-cause failures (CCFs) as the primary example. SIP-DA addresses a key limitation of conventional CCF treatment, which relies heavily on sparse and sometimes outdated historical event data and therefore cannot adequately capture plant-specific conditions, updated maintenance strategies, or newer reactor designs. The framework extends the conventional PRA hierarchy downward from PRA events to functional failure modes, physical failure modes, and key performance measures, and then to coupled physical and maintenance process simulations. Results are then propagated upward through uncertainty treatment, probabilistic validation, and integration with empirical CCF parameters. Recent progress shows how SIP-DA can improve causal resolution, enable more realistic updating under maintenance or design changes, and better support maintenance planning, design tradeoffs, and risk-informed licensing and operational decisions.

Second, the paper reports on progress from an ongoing NSF-sponsored project on modeling and analyzing interactions among human performance, physical failure mechanisms, and organizational factors in the socio-technical risk analysis of complex technological systems. This work advances the theoretical and methodological bases for explicitly incorporating these interactions into PRA and Generation Risk Assessment (GRA), with emphasis on AI-driven predictive maintenance systems in nuclear power plants. The research focuses on interactions among human operators, AI-based automation, relevant degradation and failure phenomena, organizational and managerial factors, and regulatory considerations. A central challenge is limited trust in AI-based automation by regulators, operators, and managers, which can delay licensing, constrain adoption, and reduce potential safety and economic benefits. Recent progress highlights approaches for characterizing how these interactions influence safety and financial risks and for identifying the most risk-significant socio-technical contributors.

Taken together, these efforts advance PRA beyond static, data-limited formulations toward more mechanistic, uncertainty-aware, and socio-technical representations of risk, thereby strengthening risk-informed decision making while preserving safety and reducing operational, financial, and regulatory burdens.