Risk assessment of chemical release accident triggered by landslide using Bayesian network

Area of Science:

• Environmental Science
• Risk Assessment
• Chemical Engineering

Background:

• Landslides increasingly trigger industrial chemical release accidents, posing significant risks.
• Limited research exists on the impact of landslide-triggered chemical release accidents.
• Bayesian networks (BNs) are emerging tools for assessing natural hazard-triggered technological accidents (Natech).

Purpose of the Study:

• To extend BN-based risk analysis for landslide-triggered chemical releases.
• To evaluate human health risks from n-hexane release.
• To assess the effectiveness of countermeasures for specific facilities.

Main Methods:

• Developed a methodology for human health risk assessment using Bayesian networks.
• Simulated 461,260,800 scenarios of chemical release accidents.
• Evaluated societal risk against international safety criteria.

Main Results:

• Societal risk for a storage tank near a slope exceeded Netherlands' safety criteria.
• Reducing storage rate decreased fatality risk by up to 40%.
• Tank-to-slope distance was the primary risk factor; catch basins significantly reduced result variance.

Conclusions:

• Limiting storage rates and enhancing physical measures like catch basins are crucial for mitigating landslide-related chemical release risks.
• The developed methodology can be adapted for other natural disasters and multiple scenarios.
• BNs offer a robust framework for quantifying uncertainties in Natech risk assessments.