This presentation outlines recent developments in advancing multiphase fluid dynamics simulations using Flash-X, with a focus on notable performance improvements resulting from its integration with AMReX. The talk provides an overview of Flash-X’s composable software architecture designed for modeling diverse simulations, including solid-liquid-gas interactions, phase transitions, and chemical transport, and compares its capabilities with those of existing open-source tools and commercial products. Additionally, this presentation outlines computational workflows centered around Flash-X, emphasizing its integration with scientific machine learning models. It identifies potential research directions that leverage Flash-X to establish a robust infrastructure for composability and performance portability in incompressible multiphase flows. The ultimate goal is to apply these advancements to address real-world engineering problems.
Speaker Bio: Akash Dhruv is a Postdoc in the Mathematics and Computer Science Division at Argonne National Laboratory where he performs research in algorithm development, DevOps, and performance engineering for scientific computing applications, along with building computational pipelines to integrate simulation with machine learning workflows. He received his Ph.D. in mechanical and aerospace engineering from George Washington University, Washington D.C., and a B.Tech in mechanical engineering from National Institute of Technology, Surat.
See all upcoming talks at https://www.anl.gov/mcs/lans-seminars