Towards High-Order CFD-DEM: Development and Validation

TL;DR

This paper introduces a high-order, load-balanced CFD-DEM solver that improves accuracy, stability, and efficiency in simulating complex solid-fluid systems through advanced numerical schemes and parallelization.

Contribution

It develops a monolithic finite element CFD-DEM solver supporting high-order schemes and dynamic load balancing, enhancing simulation accuracy and computational efficiency.

Findings

The solver accurately simulates Rayleigh Taylor instability.

It effectively models particle sedimentation and fluidized beds.

The approach achieves better stability and efficiency in large-scale tests.

Abstract

CFD-DEM is used to simulate solid-fluid systems. DEM models the motion of discrete particles while CFD models the fluid phase. Coupling both necessitates the calculation of the void fraction and the solid-fluid forces resulting in a computationally expensive method. Additionally, evaluating volume-averaged quantities locally restricts particle to cell size ratios limiting the accuracy of the CFD. To mitigate these limitations, we develop a monolithic finite element CFD-DEM solver which supports dynamically load-balanced parallelization. This allows for more stable, accurate and time efficient simulations as load balancing ensures the even distribution of workloads among processors; thus, exploiting available resources efficiently. Our solver also supports high order schemes; thus, allowing the use of larger elements enhancing the validity and stability of the void fraction schemes while achieving better accuracy. We verify and validate our CFD-DEM solver with a large array of test cases: the Rayleigh Taylor instability, particle sedimentation, a fluidized bed, and a spouted bed.