| dc.description.abstract | This study presents a comprehensive investigation of the dynamic characteristics of the charging process in a direct-contact thermal energy storage (TES) system utilizing phase change materials (PCM). Both numerical and experimental methods are adopted for investigating two distinct cases of a system that consists of a rectangular unit (2D & 3D) that initially contains a solid PCM block.
In the numerical analysis utilizing computational fluid dynamics (CFD), time-dependent continuity, momentum, and energy equations were solved with the aid of a “one-fluid” model for handling three phases, whereas the single-domain enthalpy-porosity approach for modeling phase transition was used. The first idealized no-inlet case concerned a stagnant hot heat transfer fluid (HTF being water) layer resting above a solid cold PCM (n-octadecane) block, with no inlet or outlet ports in the unit. For this case, a slower melting process was observed, with complete melting achieved at 890 seconds. The no-inlet case reached a thermal efficiency of 74.324%, highlighting the potential of the system in energy storage even in stagnant conditions.
For the more realistic second case (inlet case), injection of the HTF at a temperature higher than the PCM’s melting point through three inlet ports at the top of the unit was studied. The impinging HTF spread over the PCM where a thickening hot water layer formed, causing the PCM to melt progressively and concurrently replacing the air at the top of the unit. Shut-off of the HTF flow was applied to realize an equivalent system to the first case in terms of storage capacity. The inlet case exhibited a faster melting process, with complete melting achieved at 785 seconds, an 11.8% improvement over the no-inlet case. The system's thermal efficiency reached 73.79%. Initiation of the Rayleigh-Taylor instability phenomenon (due to density difference between the solid PCM and water) and its progression for both cases were observed, leading to detachment and rise of molten PCM droplets from peaks (humps) of the evolving wavy PCM/HTF interface. Propagation of any density-driven waviness at the HTF/PCM interface was verified through established analytical expressions involving surface tension effects.
The rise of droplets through the HTF was analyzed in detail using a process called binarization, whereby instantaneous contours of the PCM liquid phase fraction being 0.5 were digitized and image processing of a pixelized domain was conducted. Instantaneous droplet shape evolution was evaluated in terms of the number of droplets, area, hydraulic diameter, center of gravity, velocity, and the droplet Reynolds number. As many as three simultaneously rising droplets within the continuous phase were observed for both cases. The inlet case exhibited a faster, more dynamic droplet formation process due to the initial active flow of HTF, whereas for the no-inlet case, with initial stagnant HTF and an air layer above the PCM, droplet formation was slower and more gradual. A total of 304 droplets were generated for the no-inlet case, while for the inlet case, 261 droplets were formed after 46 seconds of HTF injection. Wavelengths of the water/PCM interface ranged between 24 and 80 mm.
To investigate the effect of the longitudinal dimension on the Rayleigh-Taylor instability of the PCM/HTF interface, three unit lengths were simulated. The findings confirmed that the observed instabilities are inherent to the system, and the evaluated wavelengths remained generally independent of the longitudinal length of the TES system. This result indicates that the longitudinal dimension does not play a role in the current simulations.
In the experimental set-up, with its dimensions exactly matching those of computational domain, solid olive oil was used as PCM to replicate conditions from the CFD simulations. For the first experimental case, the HTF was simply placed above the solid PCM without any inlets or outlets, resulting in a slower melting process. This experiment ran continuously for 146.92 minutes until complete melting. In the second experimental case, hot HTF (water) was injected for 71.5 seconds, with the experiment running for 161.12 minutes until complete melting of the olive oil. Results of both experimental cases validated the numerical models, confirming the physics observed in the simulations. Compared to 445 droplets in the no-inlet case, droplet formation was more frequent in the HTF injection case, with a total of 604 droplets observed. To track temperature variations, eleven T-type thermocouples were installed along the inner surface of the back vertical transparent panel, with an additional thermocouple used to measure ambient temperature. Experimental temperature measurements generally supported the CFD-based monitored temperatures, but their frequency response was limited. Through proper non-dimensionalization of time, the maximum-minimum temperature envelope of both CFD and experimental data exhibited a unified agreement. Additionally, the comparison of observed wavelengths between CFD simulations and experiments showed good agreement, with the simulations accurately capturing the essence of the melting process despite differences in material properties between n-octadecane and olive oil. | en_US |
