Multiphase Flow and Coalescence Filtration in Fibrous Filters: A Review of Numerical and Machine Learning Approaches

Coalescence filtration in fibrous filters underpins critical processes in various industrial applications, including liquid–liquid separation and aerosol particles removal from gas streams. Accurate prediction and optimization of these systems rely on numerical modeling to simulate multiphase flow and filtration mechanisms, including droplet capture, coalescence, and drainage. This review critically evaluates current computational models for coalescence filtration in fibrous filters, ranging from fiber-scale methods - such as Volume of Fluid, phase-field, Lattice–Boltzmann, and pore-network modeling-to macro-scale approaches. For each method, the underlying physics, literature findings, advantages, and limitations are examined. Machine learning techniques are also assessed for their potential to accelerate and augment traditional simulation models. By integrating insights from process-based and data-driven modeling, this work outlines future directions and key challenges aimed at improving prediction accuracy and computational efficiency.