Dynamic Refinement Evaluation and Criteria Optimization for Flow Features in Preparation for LES of Turbulent Hydrogen-Air Deflagration Experiment

Abstract One of the worst case scenarios in a nuclear power plant is a hydrogen explosion. The explosion strength depends on turbulence, as it may stretch and accelerate the flame, therefore enhanced understanding of flame-turbulence interaction is crucial for accurate explosion risk assessment. LES is required to numerically study flame-turbulence interaction in a practically relevant scales. Unfortunately, larger scale LES even with dynamic mesh refinement could require substantial computational resources. To determine optimal sufficient mesh and dynamic refinement criteria to resolve both flow structures and flame without performing full-scale calculations, it is desirable to first carry out small-scale simulations. In this study mesh requirements sufficient to resolve flow features corresponding to the conditions more extreme than in a reference large-scale experiment were investigated. Simulations were carried out using open-source CFD toolkit OpenFOAM-11. Several static and dynamic meshes were evaluated by comparing both fraction of resolved kinetic energy and maximum mean downward velocity, as coarse meshes failed to correctly resolve transitions to supersonic flows. For refinement criterion, velocity gradient-dependent multiplier from WALE (Wall-Adapting Local Eddy-viscosity) SGS (Subgrid-scale) model was chosen. It was determined that 8 mm base mesh with 2 levels of dynamic refinement with criterion value 2500 s−0.5 resembles static 2 mm mesh results well and is enough to resolve dominant flow features in the domain. The study thus serves as a preparation for the full-scale LES simulations.