Exploring sloshing dynamics through experimental validation and machine learning-driven computational fluid dynamics

Abstract

In this study, sloshing behavior in partially filled tanks is investigated using experimental, computational, and data-driven approaches. Sloshing is a critical phenomenon in aerospace applications, as it can affect system stability and induce time-dependent variations in the center of gravity (CG). An experimental setup equipped with load cells and pressure sensors was developed to measure the dynamic response of the liquid. Based on these measurements, time-resolved CG variations were reconstructed and analyzed at the system level. In parallel, computational fluid dynamics (CFD) simulations were performed using OpenFOAM to model two-phase flow behavior. A strong agreement between experimental and numerical results was observed in both time and frequency domains. In addition, a Long Short-Term Memory (LSTM) model was employed to predict CG motion, demonstrating good predictive capability. Furthermore, the influence of static magnetic fields on sloshing dynamics was examined, revealing frequency-dependent effects. The results provide a comprehensive framework for understanding and predicting sloshing behavior and contribute to the development of reliable modeling approaches for liquid-containing engineering systems.