Interactions between inclusion particles and the steel–slag interface directly affect the inclusion removal efficiency and thus influence steel cleanliness. Herein, the three‐phase interactions are resolved using the volume of fluid (VOF) method coupled with a dynamic overset mesh. The simulation is able to capture the instantaneous interface deformation and predict the particle motion driven by capillary force. The model validity is first demonstrated by comparison with analytical results. Then, a parameter study is conducted to examine the most influential factors governing the separation process. The results show that the system's wetting condition and the slag viscosity have a decisive effect on particle behavior at the interface (separation or entrapment). From an energy perspective, a better wetting condition generates more energy sources, and the interfacial energy is efficiently transformed into the particle's kinetic energy within a less viscous environment, thus leading to better separation. Besides, a criterion for predicting particle behavior is developed based on a modified Reynolds number (, relevant to fluid properties) and a quantity related to particle dynamics (ζ). The current work brings insights into the interfacial phenomenon during inclusion removal, which can be incorporated into large‐scale simulations to estimate the removal efficiency more accurately.