Comparison of Drag Models for Modeling Fluidized Beds with the Coarse Grain DEM-CFD Approach

CFD coupled with DEM is widely used for accurate simulations of fluidized beds. However, the CFD-DEM approach is computationally very demanding. As for today, it is restricted to lab-scale reactors with only a limited number of particles. Coarse grain approaches with coupled CFD-DEM have become popular over the last years to address the aforementioned issue. Particles of the original system are summarized into so-called grains. Thus, the number of particles can be reduced drastically, allowing the calculation of industrial-scale reactors in a realistic time frame. An important parameter is the coarse grain factor, which is defined as the ratio of grain diameter to particle diameter:

l=d_g/d_p

The goal of the coarse-grain approach is to mimic the original system precisely. However, with growing grain size, accurate representation of the interaction forces becomes challenging. Coarse simulations can fail to resolute the sub-grid phenomena, such as bubble and cluster formation, which can lead to overprediction of drag force.

As a solution to this, sub-grid drag models can be applied. Sub-grid drag models are developed by regressing finely resolved simulations, minimizing the energy for transporting particles in the flow, or correlating experimental results. The models, however, tend to be ad-hoc, and a thorough analysis of the limitations of the range of process and system parameters is often missing. In this work, common sub-grid drag models are compared against finely resolved DEM-CFD simulations. In the first step, the limitations of using the coarse grain approach with a homogeneous drag law are investigated. In the second second step, inaccurate simulations are repeated with selected sub-grid models and their applicability is evaluated for fluidized bed configurations with different Geldart-type particles, gas inlet velocity, and reactor width.