Abstract
Experiments were conducted to observe the solidification sequence and final shrinkage void distribution more precisely than was previously possible for liquids frozen from above in a rigid, constant-volume container. Physical models have been developed to give quantitative predictions about the different regimes of void initiation and why voids ultimately cease to grow downward with the freezing front. In both models, natural convection is of paramount importance for understanding the heat flows within the solidification chamber. Agreement of the results with experimental data is encouraging. We feel that these results will eventually lead to better void management in phase-change materials for storing thermal energy.