Modeling Heat Transfer in Granular Flows for TES and CSP Applications

As renewable energy systems improve and produce a larger share of the global energy, they still face issues with reliability and intermittency of the energy generated. As a solution, energy storage technologies and systems compatible with energy storage are being studied. Thermal energy storage (TES) has been shown to be a reliable solution to make concentrated solar power (CSP) and nuclear energy systems more dispatchable. Some next-generation renewable energy systems utilize thermally inert particles as the thermal energy storage medium. In the charging phase, fluidized particle beds can act as heat exchangers to heat the particles, while in the discharging phase, they can be used to extract stored heat from the particles. Predicting the heat transfer in fluidized particle systems remains an ongoing challenge, where the thermal performance can exhibit counter-intuitive behavior and scaling between a laboratory and pilot scale systems is not trivial. Although Lagrangian models such as the discrete element model (DEM) have high accuracy, they are very expensive, making them unsuitable to be used to model complex large-scale systems. Eulerian models like the two fluid model (TFM) are comparatively inexpensive but need closures to improve prediction accuracy. We use DEM to model a fluidized bed representing the discharging phase of a TES system. The model is validated against experiments, where it predicts the minimum fluidization velocity within 10% of the experimental values obtained using a defluidization curve. This model is used to study the experimentally observed phenomena of incomplete mixing in a fluidized bed at superficial velocities close to the minimum fluidization velocity. To further understand heat transfer in fluidized beds, we model a narrow channel fluidized bed heat exchanger which is representative of the receiver in a CSP system or the charging phase in a combined CSP-TES system. A DEM model is calibrated and validated by comparing the predicted heat transfer coefficients with experimental values. This model is used to verify a Nusselt Number correlation developed for chute flows by Morris et al., for fluidized beds. The correlation is used to obtain an effective thermal conductivity in TFM that depends on particle properties and solid concentration. The TFM is validated against experimental values and predicts within a 30% error. It is observed that thermal contact with the wall has a significant impact on the wall-to-bed heat transfer and the contact may be affected due to preferential paths taken by the fluidizing gas. To further reduce the discrepancy between the two models, an in-depth study is conducted into the hydrodynamics and heat transfer. Studying the DEM model, we see that the Nusselt number correlation is affected by the calibration and due to the rapid fluctuations in solid concentration in a fluidized bed. The correlation is modified using instantaneous data and incorporating the effects of model parameters, namely the surface roughness and the lens radius. The modified correlation is verified using TFM closures on DEM flow fields for both fluidized beds and chute flows. It is observed that there is a difference in the hydrodynamic predictions made by TFM and DEM. The stress model and the granular temperature calculation are modified and Ansys fluent is used as the solver which predicts closer to the DEM field. The modified correlation is used as an effective thermal conductivity model and the heat transfer predictions are within 15% of the DEM predictions. These insights are integrated to model a prototype scale 100 KWth Gen-3 enclosed particle solar receiver. Next steps include validating the model with experimental data, after which the model can be used to study counterflow receiver systems accounting for irregular shape, air and particle flow rates, and effect of the solar field and thermal losses to the atmosphere. This work aims to help develop a better understanding of heat transfer in fluidized bed systems.