Risk-based Event Tree Analysis

Fault tree analysis (FTA) works backward from an undesired top event to identify combinations of failures that cause it — it is deductive and asks 'what could cause this failure?' Event tree analysis (ETA) works forward from an initiating event through subsequent barriers to identify possible consequences — it is inductive and asks 'what happens next?' In a full PRA the two are typically combined: fault trees quantify the probabilities of branches in the event tree, and the event tree organizes the accident sequences.

Include every safety function that can materially change the consequence if it succeeds or fails, given the specific initiating event. Functions that are irrelevant to the scenario (not challenged or with near-certain success) can be excluded, but document the justification. Industry guidance such as NUREG/CR-2300 (nuclear) or CCPS Guidelines for Chemical Process Quantitative Risk Analysis provides discipline-specific checklists of typical safety functions.

Small trees (fewer than ten branch points) can be quantified manually in a spreadsheet by multiplying frequencies and probabilities along each path. Larger trees require dedicated PRA tools (e.g., CAFTA, RiskSpectrum, OpenPSA) that automate sequence quantification, importance measures, and uncertainty propagation. For regulatory submissions in nuclear or major hazards industries, software tools with quality-assurance pedigrees are generally required.

Criteria vary by industry and jurisdiction. Nuclear regulators (NRC, ONR) use core damage frequency benchmarks (typically less than 1E-4 per reactor-year) and large early release frequency limits. The UK HSE uses individual risk per year thresholds (broadly acceptable below 1E-6/year, intolerable above 1E-4/year) under ALARP. Chemical industry often uses societal risk F-N curves. Always verify which criteria apply in your regulatory context before interpreting results.

Standard ETA may be performed qualitatively — identifying accident sequences and consequence categories without assigning numerical probabilities. Risk-based ETA requires quantification: branch probabilities and initiating event frequencies are assigned so that each sequence can be expressed as a risk metric (frequency per year times consequence severity). This quantified output enables direct comparison against numerical risk acceptance criteria and supports importance ranking of accident sequences.