Physics-Informed Neural Networks for Enhanced Interface Preservation in Lattice Boltzmann Multiphase Simulations

Yue Li, Lihong Zhang · 2025

Read Physics-Informed Neural Networks for Enhanced Interface Preservation in Lattice Boltzmann Multiphase Simulations by Yue Li, Lihong Zhang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-10539.txt.

Description

Yue Li and Lihong Zhang sharpen multiphase flow simulation with neural networks in this 2025 paper. Lattice Boltzmann methods simulate fluids on a grid of particles. In two phase flows their interfaces blur over time, which ruins accuracy. A physics informed network is coupled to the solver to fight that blur.

Droplet tests measure interface width, steepest gradient, and phase separation. The coupled method kept interfaces thin and stable far longer. Visual comparisons make the gain plain to see. The network respects the underlying fluid equations, so physics stays honest. Interface energy and effective width metrics confirm the numbers. Sharp interfaces matter for sprays, bubbles, and oil water flows alike.

Catalog record

CREATORYue Li, Lihong Zhang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE26 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-10539

Browse the full library · About the Tower of Babel