(33c) Particle Tracking Velocimetry Measurements for CFD-DEM Validation

Authors: 
LaMarche, C. Q., Particulate Solid Research, Inc.
Liu, P., University of Colorado at Boulder
Fullmer, W., National Energy Technology Laboratory
Cocco, R., Particulate Solid Research, Inc.
Hrenya, C. M., University of Colorado at Boulder
The coupled Eulerian-Lagrangian method known as CFD-DEM is a powerful modeling tool that provides particle-scale, detailed insight into gas-solid multiphase flow devices. However, since CFD-DEM tracks the motion of every particle in a given system, many of the validation-grade experiments conducted previously for Eulerian-Eulerian (TFM) methods are too large (particle count too high) for CFD-DEM. By validation-grade, we mean high-quality experimental data where all particle properties and flow boundary conditions relevant to the computational model have either been well-characterized or the uncertainties quantified. Some more recent works, see e.g., Gopalan et al. (2016), have collected validation-grade experimental data specifically targeted at CFD-DEM. Previous works focused on simple systems and simple particle physics, most commonly in rectangular fluidized beds with a single gas inlet, leaving a gap between the existing CFD-DEM validation database and the TFM validation database and more realistic, engineering data. We aim to help bridge this gap by providing validation-grade data of increasingly more complex problems and physics which are still amenable to CFD-DEM--both current capabilities and projected short-term future capabilities.

In this presentation, we summarize recent experiments conducted at Particulate Solid Research, Inc. and the University of Colorado which utilize particle tracking velocimetry (PTV). At one end of the Geldart spectrum, large Group D particles are studied in a semicircular fluidized bubbling bed that is operated slightly above and slightly below minimum fluidization. Additional fluidization is provided near the flat, front face of the unit via four horizontal air jets. The two pairs of jets are submerged, subsonic but high-velocity and set at the right and left hand sides of the bed directed inwards towards each other at elevations of approximately 5 and 15 cm above the distributor. High speed video (HSV) imaging is taken of the face of the semicircular bed to capture the dynamics of the opposing jets. PTV is used to collect data from the HSV images which is then averaged into bins (grid) to create 2-D time averaged solids velocity profiles and ultimately used to calculate momentum-based jet penetration depth measurements. At the other end of the Geldart spectrum, experiments have also been performed for mildly cohesive Group A and A/C particles in the dilute riser section of a circulating fluidized bed. The entrained fines tend to form agglomerates: groupings of a few particles in size held together by interparticle forces. Local regions of the riser flow can be imaged with HSV and a new PTV method has been developed that can distinguish between primary particles and agglomerates. With the new PTV method, the hydrodynamic behavior of particles and agglomerates are extracted, such as the size, concentration, mean velocity, solids volume fraction and kinetic energy associated with relative particle and agglomerate motion (e.g. granular temperature).

CFD-DEM simulations using the open source MFiX code have also been conducted for some of the experimental conditions and we will briefly provide preliminary results, both successes and challenges. In general, the numerical simulations of the air jets in the semicircular bed compare quite favorably to the experimental data. The most noticeable bias that exists is that the simulated jets tend to penetrate further into the bed than observed experimentally, perhaps a consequence of the (momentum) point source modeling of the jets. For simulations of the cohesive particles (Geldart Group A), the CFD-DEM predictions are complicated by the fact that interparticle forces need to be modeled. We briefly report on simulations which utilize a recently developed adhesion model (LaMarche et al. 2017). Corresponding with the increase in complexity of the physics of the problem, the numerical results are degraded compared to the more idealistic (spherical, monodisperese and non-cohesive) Group D particle results, although some good agreement is still observed. Notably, when the uncertainty in local superficial gas velocity is considered, the agreement in the entrainment flux improves substantially.

Gopalan, B., M. Shahnam, R. Panday, J. Tucker, F. Shaffer, L. Shadle, J. Mei, W. Rogers, C. Guenther, M. Syamlal, Measurements of pressure drop and particle velocity in a pseudo 2-D rectangular bed with Geldart Group D particles, Powder Technology, 291, 299-310 (2016).
LaMarche, C.Q., S. Leadley, P. Liu, K. M. Kellogg, C. M. Hrenya, "Method of quantifying surface roughness for accurate adhesive force predictions," Chemical Engineering Science, 158, 140-153 (2017).