Assessing spill fire characteristics through machine learning analysis

A fuel spill can be a source of risk-significant fire scenarios in nuclear power plants (NPPs). Fuels can leak from main components, such as pumps, turbine lubrification lines, hydraulic valves, and diesel generators. Once ignited, a spill fire has a complex, dynamic changing burning area due to the fuel spread and burning. To quantify spill-fire consequences in NPPs, it is essential to analyze the spread of the liquid fuel and the fuelburning behavior. In the current study, past spill fire experimental data are collected to analyze the spreading and burning behaviors for different spill sizes and fuels. Random forest machine learning (ML) models are used to evaluate the most important parameters that impact the fire scenario conditions and provide a computationally efficient model to support fire probabilistic risk assessment. The results show the importance of the fuel leakage rate or quantity, the spill surface slope, substrate material, and fuel properties.