Volume filtered FEM-DEM framework for simulating particle-laden flows in complex geometries

TL;DR

This paper introduces a volume-filtered FEM-DEM computational framework for simulating large-scale particle-laden flows in complex geometries, such as patient-specific medical domains, with improved coupling and collision algorithms.

Contribution

It develops an efficient, conservative coupling scheme between FEM and DEM for particle-fluid interactions in complex geometries, enabling convergence and scalability.

Findings

Successfully applied to various test cases demonstrating the method's capabilities.

Shows potential for large-scale, patient-specific flow simulations.

Provides qualitative analysis of particle-laden flow behaviors.

Abstract

We present a computational framework for modeling large-scale particle-laden flows in complex domains with the goal of enabling simulations in medical-image derived patient specific geometries. The framework is based on a volume-filtered Eulerian-Lagrangian method that uses a finite element method (FEM) to solve for the fluid phase coupled with a discrete element method (DEM) for the particle phase, with varying levels of coupling between the phases. The fluid phase is solved on a three-dimensional unstructured grid using a stabilized FEM. The particle phase is modeled as rigid spheres and their motion is calculated according to Newton's second law for translation and rotation. We propose an efficient and conservative particle-fluid coupling scheme compatible with the FEM basis that enables convergence under grid refinement of the two-way coupling terms. Efficient algorithms for neighbor detection for particle-particle collision and particle-wall collisions are adopted. The method is applied to a few different test cases and the results are analyzed qualitatively. The results demonstrate the capabilities of the implementation and the potential of the method for simulating large-scale particle-laden flows in complex geometries.