A Study on the Decision Tree Approach for Predicting EO/EG Facility Leakage and Diffusion Hazards.

Area of Science:

• Petrochemical safety engineering
• Chemical process hazard analysis
• Environmental risk assessment

Background:

• Ethylene oxide (EO) and ethylene glycol (EG) are vital petrochemicals.
• EO/EG facilities face significant fire and explosion risks, leading to severe consequences.
• Accurate prediction of wide-area accident consequences is crucial for risk management.

Purpose of the Study:

• To develop a rapid prediction model for wide-area accidents involving ethylene oxide and ethylene glycol.
• To assess the consequences of leakage scenarios in EO/EG facilities.
• To enhance risk prevention and control strategies for the petrochemical industry.

Main Methods:

• Decision tree theory was applied to develop a prediction model.
• Ninety leakage scenarios for EO/EG facilities were defined and simulated using FLACS software.
• Eighty datasets of gas-phase concentration distribution data were used for training and validation (60/20 split).

Main Results:

• The decision tree model accurately predicted EO/EG leakage dispersion over areas exceeding 1 km².
• Prediction accuracy of the developed model surpassed 90%.
• The model demonstrated rapid prediction capabilities for dispersion patterns and concentration distributions.

Conclusions:

• The study successfully achieved rapid prediction of EO/EG leakage dispersion using limited simulation data.
• The developed model provides critical support for risk prevention and control in EO/EG facilities.
• This approach enhances the ability to manage potential wide-area accidents in the petrochemical sector.