Deciding with Thresholds: Importance Measures and Value of Information

Area of Science:

• Risk Analysis and Management
• Probabilistic Safety Assessment
• Decision Science

Background:

• Risk-informed decision making relies on acceptable risk levels compared to risk metrics.
• Importance measures assess event significance but don't indicate decision impact.
• Current methods lack a unified approach to decision alteration by events.

Purpose of the Study:

• Develop a value-of-information-based importance measure for probabilistic risk assessment.
• Integrate risk metrics, importance measures, and risk thresholds into a single expression.
• Enhance the understanding of how specific events influence overarching risk decisions.

Main Methods:

• Linking value of information with importance measures in probabilistic risk assessment models.
• Calculating the new importance measure from existing risk achievement worth and risk reduction worth.
• Analyzing properties including the joint decision worth of basic event groups.

Main Results:

• A novel importance measure is derived, unifying risk metric, importance measures, and risk threshold.
• The measure offers additional insights without increasing computational load.
• Demonstrated application to the Advanced Test Reactor's loss of coolant accident sequence.

Conclusions:

• The new importance measure provides a comprehensive view of event significance in risk-informed decision making.
• It complements existing importance measures by directly addressing decision alteration.
• The approach enhances the utility of probabilistic safety assessment for critical infrastructure and space missions.