Scaling of phase-change relaxation in VOF simulations of convection-driven melting

Abstract

Phase-change relaxation is commonly introduced in Volume-of-Fluid (VOF) simulations of melting through a relaxation rate that controls the conversion between phases. Although this parameter strongly influences interface evolution and temperature fields, its selection remains largely empirical and often varies from one application to another. The present study examines the relationship between phase-change relaxation and the characteristic transport processes governing convection-driven melting. A VOF-based phase-change model with a mushy-zone treatment is applied to n-octadecane (a phase change material (PCM)) melting experiments and to the OECD MCCI-2 and MCCI-3 corium–concrete interaction (CCI) experiments, covering a wide range of thermal conditions and material properties. The simulations reproduce the measured melting behavior, temperature evolution and concrete ablation trends with good agreement. A parametric investigation shows that the relaxation rate has only a limited effect on global melting and ablation rates but significantly influences local interface evolution and transient temperature distributions. Comparison of the characteristic time scales reveals a consistent trend across both PCM and CCI configurations: accurate solutions are obtained when the phase-change relaxation time remains much smaller than the thermal diffusion time scale while remaining linked to the convective turnover time of the melt region. For the cases considered in this work, the ratio between relaxation and convective time scales is found to be of O ( 10 − 2 ) . Based on this observation, a transport-scale formulation is proposed in which the relaxation rate is determined from local conductive and convective time scales. The resulting model reproduces the behavior of optimally calibrated constant-relaxation simulations without case-specific tuning. These results suggest that phase-change relaxation in convection-driven melting is governed primarily by transport time scales rather than by a fixed empirical coefficient.