Physics-Informed & Data-Driven Algorithms for Multiphase Flows: OpenFOAM VOF Spurious-Current Source Term and TALE Ligament-Length Modeling

Accurate numerical simulation of multiphase flows, such as spray atomisation, is critical for engineering design but remains a significant computational challenge. A primary obstacle is the persistence of spurious vorticity, or spurious currents, which are non-physical velocity fields arising from numerical imbalances in surface tension modelling. These artefacts distort fluid interfaces, particularly at high density ratios, compromising the fidelity of predictions. This thesis first addresses this problem by systematically diagnosing spurious vorticity in the Volume-of-Fluid (VOF) method within OpenFOAM. It introduces and validates a novel, physics-based mitigation strategy: the Vorticity-Sourced Diffusion (VSD) method. Unlike conventional corrective forces, the VSD method introduces a selective, proactive dissipation mechanism. It is derived from the vorticity transport equation to target the numerical source of spurious rotation, and it is selectively applied only to the non-physical, imbalance-driven velocity component, preserving physical flow dynamics. Validation on canonical static and dynamic benchmarks, such as the static bubble and Hysing rising bubble cases, demonstrates that the VSD method achieves a multi-order-of-magnitude reduction in spurious currents and dramatically improves geometric fidelity without the trade-offs in dynamic accuracy seen in prior corrective approaches. In a complementary investigation, this work develops a reduced-order model for a key process in atomisation: shear-driven ligament elongation. The Taylor-Analogy Ligament Elongation (TALE) model, a one-dimensional ordinary differential equation, is formulated. The model is calibrated against a custom high-fidelity VOF simulation dataset using a direct data-driven parameter identification method. The calibrated TALE model shows excellent predictive agreement, and its coefficients are correlated to flow parameters, such as the Weber number, via simple parametric laws. This research delivers two key contributions: the robust VSD method, a significant advancement for achieving high-fidelity multiphase simulations, and the computationally efficient TALE model, which provides a framework for novel hybrid Euler-Lagrangian models to bridge a critical gap in atomisation simulation.