Application of Filtered Two-Fluid Models to Industrial-Scale Fluidized Beds

Supported by recent developments in high performance computing capabilities, computational fluid dynamics (CFD) has become an invaluable tool in the study, design and scale-up of gas-solid fluidized beds. In an industrial context, CFD of commercial-scale fluidized beds is used either in the development stages of new processes and equipment, or as a way of troubleshooting and improving existing production assets.

Two approaches are typically considered for modeling gas-solid flows. While the Eulerian-Lagrangian approach (e.g., CFD-DEM) is increasingly used for fundamental studies in academic research, it is still too computationally expensive for industrially relevant scales in refining and chemicals applications. The preferred method for industrial applications is therefore the Eulerian-Eulerian approach (i.e., two-fluid model). The two-fluid model with classical drag laws is able to capture the complex multi-scale flow phenomena occurring in gas-solid flows, but to preserve accuracy, the mesh resolution should be limited to ten particle diameters at most [1], which is impractical for most full-scale applications. By replacing the classical micro-scale drag laws with a meso-scale ‘filtered’ model, the mesh resolution requirements and hence computational cost can be significantly reduced [2]. These filtered models are derived from highly resolved two-fluid simulations in a periodic domain, which are then filtered using various filter sizes, with the ultimate objective of producing an accurate drag force in cells that are larger than 100 particle diameters.

The CFD group at ExxonMobil has been using filtered drag models for over a decade. In this presentation, industrial successes in applying these models to refining and chemicals processes will be discussed. Furthermore, we will highlight some challenges that had to be tackled when using the existing filtered models outside their original validation range.

[1] Agrawal, K. et al. Journal of Fluid Mechanics 445 (2001) 151-185.

[2] Igci, Y., et al. AIChE Journal 54 (2008) 1431-1448.