Pore-Resolving Simulations of Biomass Char Particle Combustion

Abstract

Biomass char morphology affects combustion behavior at the particle scale for zone II conditions, in which both heterogeneous reaction and intra-particle diffusion govern the overall rate. Furthermore, particle-scale processes affect reactor-scale outputs, and reactor-scale simulations are sensitive to particle-scale models. However, most char particle combustion models employ coarse-grained, effective-continuum approaches, which treat all porosity at the subgrid-scale. Effective-continuum approaches are not valid or accurate in the presence of large, irregular pores which can approach the size of the particle. A 3-D, pore-resolving CFD simulation approach using real biomass char particle geometries obtained from X-ray micro-computed tomography (micro-CT) is therefore used to examine the impact of morphology on zone II combustion for pulverized (∼100 µm) biomass char particles for the first time. In contrast to larger, millimeter to centimeter sized particles, the sub-millimeter, high aspect ratio biomass char particles exhibited localized reactant penetration into the innermost regions of the particles, facilitated by the presence of large pores connected to the external surface. The oxygen mole fraction distributions were governed by the large pore morphology, were non-monotonic with distance from the surface, and achieved minima in thick microporous char regions surrounding the large pores. A comparison between the pore-resolving simulation and an equivalent, spatially resolved, effective-continuum simulation revealed that even in the microporous char, the effective-continuum model underpredicted reactant penetration. A careful comparison was then performed between 30 pore-resolving particle simulations and several effectiveness factor models that employed particle-specific parameters. Commonly used uniform cylinder models significantly underpredicted effectiveness factors for these real pulverized pine char particles, while accessible hollow cylinder models achieved less than 10% relative error when averaged over all 30 particles.