Measurements were made of interfacial velocities and transient and steady-state temperature profiles during the freezing and melting of water in a 0.5 × 0.5 × 2-in. high test chamber. Heat flow was one-dimensional, up and down. Tests both included and excluded the density-inversion temperature of 4 deg C. Freezing at the top and at the bottom, melting at the top, and at the bottom, all were achieved by selection of cold-end temperatures between −50 and −50 deg C and hot-end temperatures between 3 and 97 deg C. Runs included conditions with buoyancy forces in the liquid, with buoyancy forces existing but insufficient to cause convection, and with natural convection occurring at all times. With no natural convection the results agreed with predictions found by use of the numerical technique of Murray and Landis developed originally for cases with no convection. The onset of natural convection was found to be at a Rayleigh number of about 1700. Proper selection of the significant length, the ΔT, and the coefficient of expansion for the Rayleigh number is described. The effective thermal conductivity for Rayleigh numbers up to 107 agreed with prior correlations obtained with free convection but with no phase change. The numerical calculation procedure was modified successfully by use of the effective k. At the highest Rayleigh number, an unusual case of oscillations in the interface velocity is reported.

This content is only available via PDF.
You do not currently have access to this content.