Date of Award

January 2020

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Chemical Engineering

First Advisor

Gautham Krishnamoorthy

Abstract

Fluidization process technology has been widely studied since its first commercial uses during the 1920s. Initially, developmental progress was slow as fluidized beds are complex in design and often hard to scale up in size. However, with the emergence of computational fluid dynamics (CFD), researchers were able to simulate wide ranges of fluidization and create new fluidization designs to help solve complex problems. This study aims to determine if CFD can be effective in designing spout-fluid beds with draft tubes for advanced energy applications. CFD has been employed for studying the fluidization characteristics, design, and optimization of spouted beds. However, they have mainly been used in conjunction with large (Geldart D) particles where the drag forces dominate the wall frictional and collisional losses. Subsequently, identifying the correct drag law with the Geldart D particles has often deemed to be sufficient to get an accurate fluidization pattern. Additionally, the more accurate Discrete Element Method (DEM) methodology where the forces and motion of the individual particles are tracked has also been employed when the numbers of particles are computationally manageable (between 104 and 105). In this project, the feasibility of employing Geldart B particles in spouted bed reactors are explored. In lieu of the fact that the number of resulting particles make the use of the DEM framework unfeasible we employ the Two-Fluid Modeling (TFM) methodology in this scenario. Further, frictional and collisional losses become more important here (due to smaller particle sizes) which makes the selection of appropriate interaction terms in the TFM framework all the more important. The first CFD software used in this study was Multiphase Flow with Interphase eXchanges (MFiX), an open-source CFD software created and maintained by the National Energy and Technology Laboratory (NETL). The purpose of using the MFiX software was to design a spout-fluid bed with a draft tube that may be used in Chemical Looping Combustion (CLC) applications with the goals of minimizing the pressure drop and maximizing the residence time. Specifically, the drag law, coordinate system, and grid/mesh resolution were studied to determine the optimal design settings. The optimal design settings were determined by comparing the simulation results to previous experimental work conducted. For the drag law comparison, three commonly used drag laws (Syamlal-O’Brien, Gidaspow, Wen-Yu) in fluidized bed simulations were employed to select the law that gives the optimal performance results in the model. Results of the drag law comparisons showed that the Syamlal-O’Brien model (with appropriate parameters representative of the fluidizing particle) is the best drag law for the spouted fluid bed with a draft tube. Therefore, it was employed in the comparisons of the co-ordinate system and mesh resolution The next parameter studied was the co-ordinate system set for the model. Due to the high computational cost associated with 3D simulations, it is traditional to simulate the fluidization behavior using 2D axisymmetric grids or even 2D planar grids. While 2D axisymmetric grids are an accurate representation of a cylindrical reactor, the imposition of symmetric boundary conditions along the axis can lead to unrealistic jetting behavior (which will be exacerbated in a spouted bed unit due to high centerline velocity). While many studies have shown that 2D planar and 2D axisymmetric result in similar fluidization characteristics in bubbling beds, the importance of this assumption in spouted bed units needs to be ascertained. The results of our model showed that the cylindrical 2-D axisymmetric simulation predictions were qualitatively similar to the fluidization behavior that was observed experimentally. The final MFiX parameter studied was the grid/mesh resolution of the model. In bubbling beds, a cell size of 9 diameters has been deemed adequate to capture pressure drop. However, we need to ascertain if this is applicable in spouted beds with draft tubes. However, a comparison of pressure drop predictions from employing cell sizes of 11 particle diameters and 6 particle diameters showed that grid convergence was not yet achieved. Therefore, further research would be required to determine the optimal cell size for grid independence in spouted bed systems. In the second part of this study, the CFD software ANSYS Fluent was utilized to determine the effectiveness of a spout-fluid bed with a draft tube for the density separation of particles. This specific design function may be desirable for the continuous removal of a product such as carbon black separation in the pyrolysis of tires. ANSYS Fluent was selected for this portion of the study due to its ability to handle complex geometry and enable the specification of phase-specific interaction terms (drag laws and collision models). The different particles investigated were of 300 and 100 µm diameters and 1200 and 4000 kg/m3 densities respectively. The results of the CFD simulations showed that a stable spout-fluid bed with a draft tube design could be created. However, no significant separation of particles was achieved. Additionally, CFD simulations and experimental trials were conducted on conical bottom spouted beds to determine if the separation of pyrite-rich minerals from coal. The results of the CFD and experiment both concluded that no separation was achievable. The recovery of coal product was only 5% of the original mass and the ash content of the product coal decreased to 25-28% when the target separation criteria was less than 20% ash. Therefore, we concluded for effective density separation of Geldart B particles design modifications to the spouted bed reactor should be investigated in future work.

Share

COinS