Probabilistic Safety Assessment and Risk Management
This domain tests how fault trees, event trees and reliability data are used to estimate core damage frequency and to support risk-informed decisions.
Core rules
- An event tree starts from an initiating event and follows accident sequences through success or failure of safety functions. A fault tree works downward from a top event to basic events.
- Minimal cut sets are the smallest combinations of basic events that cause the top event. The probability of a cut set is the product of its basic-event probabilities if they are independent.
- Common cause failure is modelled with beta factors, multiple Greek letters or similar methods. It often dominates redundant safety systems.
- Importance measures include Fussell–Vesely, risk achievement worth and risk reduction worth. They rank contributors to risk and to risk change.
- PSA levels: Level 1 estimates core damage frequency; Level 2 estimates containment failure and release; Level 3 estimates offsite consequences. Human reliability analysis is part of the model.
Worked example
Two redundant pumps are required to operate in parallel. Each has an independent failure probability of 0.01. The common cause beta factor is 0.1. Estimate the system failure probability.
Independent failure of both pumps is (0.01)² = 1×10^-4, reduced by the non-common-cause fraction: about (0.9×0.01)² = 8.1×10^-5. Common cause failure is approximated by β × q = 0.1 × 0.01 = 1×10^-3. Total is about 1.08×10^-3. The reasoning is that the common cause term is more than ten times the independent term, so the redundant system is dominated by common cause failure.
Traps
Do not multiply dependent failures as if they were independent. Do not truncate cut sets so aggressively that dominant contributors disappear. Do not treat PSA as a precise absolute number; it is a decision-support tool. Do not ignore human actions and recovery. Do not use importance measures without stating the base case.
Key points
- Event trees model sequences; fault trees model causes.
- Minimal cut sets are the smallest failure combinations.
- Common cause failure often dominates redundant systems.
- Beta factor is a common-cause model parameter.
- PSA Level 1 addresses core damage frequency.
- Importance measures rank risk contributors.