Skip to content

NUREG-1792 — Good Practices for HRA

NRC guidance on implementing human reliability analysis in PSAs. Covers method selection, HFE definition, HEP quantification, and event tree integration.

Key Good Practices

  • Scenario-based: HFEs defined in context of specific accident scenarios, not generically
  • Method selection: ATHEANA for commission errors, SPAR-H for screening, IDHEAS-ECA for cognitive analysis
  • Dependency assessment: Sequential actions assessed for dependency
  • Documentation: Full traceability from scenario through HFE to quantification

Extension Needs for AI-Assisted Operations

Report 5 identifies needs for: new HFE types (follow incorrect AI, override correct AI), new dependency structures (three-way conditional), new performance-shaping-factors (AI reliability, trust calibration), and acknowledgment that no empirical data exists for AI-related HEPs.