(147d) Investigation of Heat Transfer in Fluidized Bed Heat Exchangers

Fluidized bed heat exchangers play an important role in various industrial processes, where efficient heat transfer is required. In this study, we look at the heat transfer phenomena within fluidized bed heat exchangers using two different modeling approaches: high-fidelity discrete element modeling (DEM) and computationally inexpensive two-fluid modeling (TFM). Both models are validated against experimental data for heat transfer coefficients in a parallel-plate narrow channel fluidized bed heat exchanger. In TFM, since the solids are not individually tracked, the model is computationally inexpensive but requires closures for the constitutive relations that dictate the solid stresses and heat flux. To obtain the effective solid conductivity in TFM, a continuum conduction model developed by Morris et al. is implemented. The conduction model uses chute flow DEM data to obtain a Nusselt number correlation which is a function of solid concentration. Our findings show that DEM outperforms TFM in capturing heat transfer trends observed experimentally. To understand the disparities between the two models, we decouple heat transfer and hydrodynamic differences, examining these factors individually. Specifically, for studying the heat transfer, we validate the Nusselt number correlation, originally developed for chute flows, for its applicability in fluidized beds using DEM. Scaled Nusselt number is plotted against the solid fraction and the effect of particle roughness as well as conduction lens radius on the curve is studied. It is observed that these parameters have a significant effect on the wall to particle heat transfer. Hence, the Nusselt number correlations need to be modified to include the effects of these parameters. Furthermore, we validate the closures used in TFM for predicting heat transfer by applying them to DEM hydrodynamic flow fields. To enhance the accuracy of predictions, histograms are used to study the noise in the Nusselt number when plotted against solid fraction. Notably, curve fits to the means of these histograms prove to be more accurate predictors when applied to DEM flow fields. Additionally, we also observe consistent trends in the histograms for a given particle surface roughness across different solid fraction values. Upon plotting the standard deviation of each histogram against the solid concentration, we observe consistent curves across different fluidizing velocities for the same surface roughness. This consistency further emphasizes the importance of parameter considerations and accurate modeling techniques for reliable predictions in fluidized beds. Overall, our study contributes to a deeper understanding of heat transfer mechanisms in fluidized bed heat exchangers. Since we have validated our heat transfer models, in order to obtain a comprehensive method to improve TFM predictions, next steps would include studying the hydrodynamic predictions of the models.