Machine learning for predictive modeling of Leidenfrost droplet dynamics using operator learning and Physics-Informed Neural Networks (PINNs).
ETH ZürichNumerical analysis of packed beds for fusion reactor blanket concepts using CFD-DEM and porous media approaches.
IIT BombayAgentic AI framework for OpenFOAM using Retrieval-Augmented Generation and vector databases.
iteraSim
This work investigates the complex rebound dynamics of viscoplastic droplets on Leidenfrost surfaces. By implementing a custom VOF solver in OpenFOAM, we captured the intricate vapour-liquid interactions and non-Newtonian rheological effects during high-temperature impacts.
Adaptive mesh refinement for complex interfaces using snappyHexMesh.
Leveraging high-performance computing clusters (Euler @ ETH) for parallel solving.
Training PINNs and Operator Learning models on simulation datasets.
Automated feature extraction and hydro-thermal performance mapping.
