(267f) A Very, Very Small-Scale Experiment of Fluidized Particle Segregation: A Prerequisite for the Uncertainty Quantification of CFD-DEM Simulations

Authors: 
LaMarche, C. Q. - Presenter, Particulate Solid Research, Inc.
Dahl, S. R., University of Colorado at Boulder
Fullmer, W., National Energy Technology Laboratory
Hrenya, C. M., University of Colorado at Boulder
Systems involving fluidized particles are ubiquitous in industrial operations for which thorough mixing of particulate material is often a requirement. Computational fluid dynamics coupled with the discrete element method (CFD-DEM) allows for a deeper physical understanding of the how the fluid-particle and particle-particle interactions within these operations affect the overall behavior. Just as the uncertainty in experimental results require quantification, the predictions obtained by CFD-DEM should also include bounds associated with their uncertainty. Such uncertainty quantification requires thousands of simulations or more. The already high computational cost associated with simulating the large number of particles in a typical lab-scale experiment with CFD-DEM makes performing such uncertainty quantification prohibitive. In this work, a very, very small scale (< 7,000 particles), segregation experiment was designed such that CFD-DEM simulation could be performed in a reasonable time – i.e., less than 8 hours with a single CPU. The details of the segregation experiment and system conditions, along with the careful characterization of particle properties, is explained. Additionally, the sensitivity of the experimental results (intruder segregation time) to various system conditions and intruder properties, e.g., intruder particle size, initial position, number of intruders, and superficial velocity, on the segregation rate are discussed. Preliminary CFD-DEM simulations of one of the experimental conditions are overviewed. Finally, the usefulness of a very, very-small-scale device to perform fast experiments with a small number of particles – to assess brute-force uncertainty quantification and simplified methods - is discussed.