Grid independence behaviour of fluidized bed reactor simulations using the Two Fluid Model: Effect of particle size
Peer reviewed, Journal article
Accepted version
Date
2014Metadata
Show full item recordCollections
- Publikasjoner fra CRIStin - SINTEF AS [5912]
- SINTEF Industri [1593]
Abstract
It is well known that particle size has a significant influence on the grid independence behaviour of fluidized bed reactor simulations carried out using the Two Fluid Model (TFM) approach. The general rule of thumb states that the cell size should scale linearly with the particle size so that the cell size is always at most a factor of 10 larger than the particle size. In this study, however, the effect of particle size on grid independence behaviour was shown to be unexpectedly large. In particular, a five-fold increase in particle size permitted the use of a 63 times larger cell size, implying a 633 ≈ 250,000 times speedup for resolved simulations in the planar 2D domain considered in this study. Thus, the general rule of thumb was found to be overly cautious, especially for larger particles. Closer investigation revealed the particle relaxation time to be a very good predictor of the grid independent cell size. Although this finding needs to be confirmed for parameters other than only the particle size, this relation can theoretically be used to greatly shorten the time-consuming grid independence studies that are required before any fluidized bed simulation campaign. In general, the rapid increase in cell size allowed by larger particle sizes showed that reasonably accurate industrial scale simulations (5 m inner diameter reactor) are already possible in 2D for large particles (~ 600 μm). If the 2D grid independence behaviour assessed in this study is extendible to 3D, larger particle sizes in the range of 500–1000 μm can already be simulated in full 3D for reactor sizes ranging from 1–4 m. Simulation of smaller particle sizes (< 200 μm) will remain out of reach for many decades to come, however, and a filtered coarse grid approach will definitely be required to make such simulations possible.