Kaylee Smith

Date of Award

January 2020

Document Type


Degree Name

Master of Science (MS)


Chemical Engineering

First Advisor

Gautham Krishnamoorthy


The modeling work and simulation results contained within this thesis come from two different applications that together emphasize multiple of the challenges currently faced by researchers in the field of numerical modeling of gas-solids flows with computational fluid dynamics (CFD) tools, and highlight avenues of potential resolutions to these challenges.

In the first body of work, the MFiX CFD suite, developed by the Department of Energy’s National Energy Technology Laboratory (DOE NETL), was utilized to model and simulate several experimental conditions of a fluidized bed (in 2D), and a hopper (in 3D), where solid-solid collision effects play a dominant role. Of concern in these studies are the physical prediction capabilities and associated computational costs of the three different multiphase frameworks available in MFiX: the Discrete Element Model (DEM), the Two Fluid Model (TFM), and the newer hybrid Multiphase Particle in Cell Model (MPIC). Initial selection of an appropriate multiphase modeling framework that satisfies the level of detail and computational cost constraints associated with the problem at hand, is crucial to the successful use of CFD tools in industry. The DEM and TFM frameworks were deemed to be the most accurate in simulating transient pressure profiles in the fluidized bed scenario compared to experimental measurements. TFM framework also proved to be 35% faster. While the MPIC framework was on average 90% faster than the DEM framework, it failed to produce reasonable predictions of physical flow behaviors. An additional motivation behind this research was to test and explore further reductions in computational costs offered by a recently developed interface of MFiX with the linear solver library, PETSc. Using previously identified numerical strategies in PETSc to solve the pressure equations, a more robust solver convergence behavior than the native pressure solver package was achieved across all three frameworks. Most notably, it enabled the use of larger and fewer time steps in the DEM framework, resulting in a 4-20% reduction in overall solve time to simulate 20 seconds of fluidized bed flow. Despite the significant reduction in computational time, simulation accuracy in terms of predicting the average pressure drop was slightly diminished using the PETSc solver in the DEM framework. Simulations of pure granular flow in a hopper revealed that while the TFM framework experienced difficulties converging the solids pressure term, it was still capable of predicting mass discharge rates that were very similar to those of the DEM framework, but at a comparatively lower computational cost. Again, the MPIC framework predictions differed significantly from the DEM results which are considered the benchmark/gold standard for modeling granular multiphase flows. Thus, despite the significant computational advantages of the MPIC framework over the other two, proper caution needs to be exercised when utilizing it to simulate densely packed solid flows.

In the second body of work, a collection of CFD models and simulations were developed using the ANSYS Fluent DPM framework to simulate air combustion of three different coal types (Powder River Basin (PRB), Illinois #6, and Sufco 2) from select experiments conducted on a pilot-scale combustor from the University of Utah. The objective of this study was to investigate the sensitivity of ash deposition behaviors to select modeling parameters, with the aim of formulating a particle capture model. Ash formation and deposition is a complex physio-chemical process that negatively affects boiler operation and predicting ash deposit growth rates with CFD modeling techniques is extremely challenging. Many previous attempts by others neglect the importance of adequately resolving the particle size distribution and using an adequate spatial resolution near the heat transfer surface. Combustion modeling methodologies were validated against experimental measurements of flue gas ash concentrations and reactor profiles of temperature, and estimates of velocity. Simulation predictions were deemed to be in satisfactory agreement with experimental measurements. Temperature and velocity profiles were only mildly influenced by the resolutions of both the particle size distribution model and the near-boundary spatial mesh. Simulation predictions for impaction rates on a collector probe boundary were large in comparison to measured values of deposition rates, enforcing the importance of capture efficiency in the effort to accurately predict ash deposit growth rates. Impaction rates also proved to be moderately sensitive to the number of bins used to resolve the particle size distribution, and the degree of this sensitivity was unique to each coal type further emphasizing the challenges in universally modeling combustion and ash deposition across fuel sources. Impaction rates increased significantly when employing a more refined near-boundary mesh which highlights the importance of spatial resolution modeling parameters in successful ash deposition simulation efforts. Additionally, a Weber number criteria capture method was tested across all three coal types and critical Weber numbers were identified in each case which were significantly different between coal types. These values were found to be much smaller than 1 which signifies the importance of considering attraction forces between the particles and the deposition surface. Predicted deposition rates when applying the critical Weber numbers as capture criteria agreed well with measured values, but demonstrated sensitivity to the number of bins in the particle size distribution model. In the PRB and Illinois coal cases, a mere 10% adjustment in the critical Weber cutoff value resulted in a roughly 30% difference in predicted deposition rates, demonstrating that this method of modeling particle capture is not universal and should be used with caution. Results from this work demonstrate that ash deposition processes are still not fully understood, and the reinforces the need for more collaborative efforts between CFD modelers and experimentalists.