Generation and Analysis of Late Stage Accident Debris Bed Cooling Simulations for Varying Porosities Using Al Methods

This thesis project examines the influence of debris bed porosity on its coolability during reactor accidents. In such scenarios, debris beds can form with varied geometries and porosity levels, making effective cooling crucial for mitigating severe consequences. To explore this, after familiarization with the COCOMO (Corium Coolability Model) code, simulations of varying porosity will be conducted using available computing clusters to model the thermal-hydraulic behavior of debris beds across a spectrum of porosity values.

The generated simulation data will undergo thorough analysis to uncover correlations between porosity and cooling performance. Both classical and AI-driven methods, such as decision tree algorithms and regression models, will be employed to predict coolability and determine the porosities within the bed that facilitate effective cooling. Visualization software like VisIt will be utilized to represent the simulation results graphically, enabling better interpretation and communication of findings. These approaches aim to deepen our understanding of the relationship between porosity and debris bed behavior, offering valuable insights for managing reactor accidents.

This research is expected to enhance predictive modeling capabilities for reactor accident scenarios and contribute to the development of robust strategies to ensure reactor safety during critical incidents.

Examiners: Prof. Dr. Mathias Niepert; Prof. Dr.-Ing. Jörg Starflinger (IKE)

Supervisor: Jasmin Joshi-Thompson (IKE).