Efficient simulation of gas-liquid multiphase flow in rushton tanks through Integrated VOF and ROM methods

Multiphase flows in stirred-tank reactors are common in industrial processes such as chemical reactions, crystallization, and fermentation. These flows (gas-liquid, liquid-liquid, and solid-liquid) are governed by fluid properties (e.g., viscosity and density), impeller design, and operating conditions. For example, low-viscosity and low-density fluids generate greater turbulence during high-speed mixing. A better understanding of these factors is key to improving mixing efficiency and product quality. Computational fluid dynamics (CFD) simulations are widely used to analyse flow patterns and mixing. However, high computational demands and modelling limitations limit their effectiveness, highlighting the need for more efficient predictive models. To address these needs, this study reviews recent advances in CFD and reduced-order models (ROMs) for stirred-tank multiphase flows, comparing their advantages and limitations. It then presents a high-fidelity ROM developed using ANSYS and validated against full-order CFD simulations. Results show that ROM accuracy depends on the fidelity of the full-order model and the number of snapshots used for ROM training. Improvements to the full-order model were made to generate a richer snapshot dataset, improving ROM performance. Finally, scale-up experiments confirmed that the optimised ROM can accurately and effectively simulate the multiphase flow behaviour in the stirred tank.