A Theoretical and Experimental Investigation of Unidirectional Freezing of Nanoparticle-Enhanced Phase Change Materials

Abstract

Highly-conductive nanostructures may be dispersed into phase change materials (PCM) to improve their effective thermal conductivity, thus leading to colloidal systems that are referred to as nanostructure-enhanced PCM (NePCM). Results of a theoretical and experimental investigation on freezing of NePCM in comparison to the base PCM are presented. A one-dimensional Stefan model was developed to study the unidirectional freezing of NePCM in a finite slab. Only the thermal energy equation was considered and the presence of static dispersed nanoparticles was modeled using effective media relations. A combination of analytical and integral methods was used to solve this moving boundary problem. The elapsed time to form a given thickness of frozen layer was therefore predicted numerically. A cooled-from-bottom unidirectional freezing experimental setup was designed, constructed, and tested. Thermocouple readings were recorded at several equally spaced locations along the freezing direction in order to monitor the progress of the freezing front. As an example, cyclohexane (C6 H12 ) and copper oxide (CuO) nanoparticles were chosen to prepare the NePCM samples. The effective thermophysical and transport properties of these samples for various particle loadings (0.5/3.8, 1/7.5, and 2/14.7 vol. %/wt. %) were determined using the mixture and Maxwell models. Due to utilization of the Maxwell model for thermal conductivity of both phases, the numerical predictions showed that the freezing time is shortened linearly with increasing particle loading, whereas nonmonotonic expediting was observed experimentally. The maximum expediting was found to be nearly 8.23% for the 0.5 vol. % sample. In the absence of a nanoparticle transport model, the mismatch of the cold plate boundary conditions, lack of accurate thermophysical properties, especially in the solid phase of NePCM samples and precipitation issues with 2 vol. % samples were addressed by improving the experimental setup. Through adopting a copper cold plate, utilizing measured thermal conductivity data for both phases and using 1, 2, and 4 wt. % samples, good agreement between the experimental and numerical results were realized. Specifically, adoption of measured thermal conductivity values for the solid phase in the Stefan model that were originally underestimated proved to be a major cause of harmony between the experiments and predictions.