PHAST - Physics-informed High-speed AI Suite for Two-phase boiling heat transfer

Objective

PHAST delivers a physics-informed AI suite for cryogenic boiling heat transfer providing rapid, reliable predictions. Cryogenic fluids underpin critical technologies including medical imaging, cryosurgery, hydrogen infrastructure, and quantum computing, yet prediction inaccuracies prevent full potential exploitation. Current empirical models exhibit errors exceeding 50%, forcing operation at 70% of critical heat flux capacity. PHAST establishes the first multiscale modelling suite integrating atomic-to-macroscopic phenomena using physics-informed neural networks. The objective is developing an open-source platform combining molecular dynamics, continuum-scale solvers, and experimental data for accurate heat transfer predictions. Three work packages structure the methodology: nanoscale molecular dynamics simulations for realistic parameters and closure models; continuum-scale simulations generating validated databases mapping nitrogen boiling across operating regimes; and PHAST-Suite development where neural networks incorporate conservation laws for rapid predictions. Results deliver cross-sector impact: improved hydrogen and LNG efficiency reducing energy consumption; optimized helium management in MRI systems and cryopreservation; accelerated quantum computer development. The open-source design enables extensions to other fluids, fostering scientific-industrial collaboration. PHAST addresses critical thermal science gaps, benefiting energy, healthcare, and high-tech industries while advancing research excellence, interdisciplinary training, international mobility, open science, and EU clean energy priorities.