Risk analysis of dust explosion scenarios using Bayesian networks

Area of Science:

• Industrial safety
• Risk assessment methodologies
• Explosion analysis

Background:

• Dust explosions pose significant industrial hazards.
• Existing risk assessment methods may not fully capture complex dependencies.
• Dynamic risk management requires continuous learning from incidents.

Purpose of the Study:

• To propose a novel methodology for dust explosion risk analysis using Bayesian networks and bow-tie diagrams.
• To evaluate risks considering common cause failures and event dependencies.
• To identify critical root events and vulnerable system components.

Main Methods:

• Bayesian network modeling for causal relationships.
• Bow-tie diagrams for event-consequence representation.
• Diagnostic analysis to pinpoint critical root events.
• Probability adaptation for sequential updating and learning from historical data.

Main Results:

• Identification of dust particle properties, oxygen concentration, and staff safety training as key risk factors.
• Quantification of risks associated with dust explosion scenarios.
• Demonstration of the methodology's applicability through a case study.
• Highlighting the importance of dependencies and common cause failures in risk assessment.

Conclusions:

• The proposed Bayesian network methodology provides a robust framework for dust explosion risk assessment.
• Diagnostic analysis effectively identifies critical contributing factors.
• The probability adaptation concept enhances dynamic risk management capabilities.
• The approach aids in identifying system vulnerabilities for targeted safety improvements.