Abstract

Non-equilibrium multiphase processes occur in numerous engineering applications, ranging from fuel injection to semiconductor manufacturing. In non-equilibrium multiphase systems, two or more phases (e.g., liquid–gas, liquid–solid) exchange momentum, heat, and mass more rapidly than they can relax to thermodynamic equilibrium. Steep gradients and metastable states drive rapid interfacial events, including nucleation, cavitation, flash boiling, coalescence/breakup, and crystallization. In these scenarios, interface dynamics, phase change and eventually reactivity are tightly coupled. Because equilibrium assumptions fail, modelling must resolve evolving interfaces and phase change with thermodynamically consistent frameworks (e.g., diffuse-interface/phase-field or high-fidelity Volume of Fluid, VoF), realistic equations of state, and an appropriate evaluation of surface tension. While an accurate description of the physics of multiphase processes is possible in ideal setups or simplified systems, simulating large-scale devices requires a high degree of approximation due to the sheer amount of characteristic times and lengths involved. The above often renders computational fluid dynamics unsuitable for designing new applications and even less applicable for control.  This talk outlines the theoretical foundations of multiphase flow modelling, emphasising the distinction between actual two-phase flow and dense mixtures, and how density-gradient theory provides a practical discriminator. Building on this, two frameworks were proposed, one based on the geometric/algebraic VoF and the other on phase field modelling (Navier–Stokes–Korteweg), including equations of state capable of representing metastable liquids under tension. Following the theoretical description, the talk details three applications: thermally induced secondary atomization of multi-component sustainable aviation fuels (SAF), ammonia injection, and a cavitation-resolving model for ultrasonically enhanced reactors. Following this, it will be illustrated how the latter has directly enabled the design of two patented processes, culminating in the successful development of KAUST spin-offs. The final part of the presentation will focus on how neural operators (e.g., Fourier Neural Operators) accelerate these simulations to real-time, enabling the creation of digital twins and model-predictive controls and how, with these tools and the collaboration of other research groups, we are building a control framework for high-intensity focused ultrasound-based cancer treatment.

Biography

Dr. Paolo Guida is a Research Scientist in the Physical Science and Engineering Division at KAUST, specializing in numerical modelling of non-equilibrium multiphase flows and ultrasound-enabled process intensification. He earned his MSc and BSc in Chemical Engineering from Politecnico di Milano (Milan). After earning his PhD in Mechanical Engineering from KAUST in 2021, he has led translational projects with industry partners, including the development of ultrasonically enhanced lithium-ion battery recycling (CirCoLi) and oxidative desulfurization (uODS). The two inventions were awarded Gold Medals at the 2025 International Exhibition of Inventions in Geneva. He is now working on the development of tools based on neural operators to enable real-time digital twins and model-predictive control across recycling, refining, and cancer treatment, as well as on further improving and scaling up new processes. His work has led to multiple patents and spin-offs (CirCoLi, uODS/Pelican Technologies, MolEvo) and first-author publications in venues such as Journal of Fluid Mechanics, Physics of Fluids and Chemical Engineering Journal.