Fluid Dynamics Face-Off: Benchmarking the Best Computational Methods
"Lattice-Boltzmann, Coupled Lagrangian-Eulerian, and Smoothed Particle Hydrodynamics go head-to-head in resolving shear-driven flow fields."
The interplay between fluids and structures—known as fluid-structure interaction (FSI)—is critical in numerous engineering applications, from designing aircraft engines that can safely ingest soft bodies to understanding how blood flows through arteries affected by plaque. Accurately modeling FSI requires capturing how fluids influence structures and vice versa; neglecting either aspect can lead to flawed predictions and designs.
Currently, two main strategies tackle FSI: monolithic approaches, where fluid and structural equations are solved simultaneously, and partitioned approaches, which couple individual fluid and structural solvers. Monolithic methods tend to be more stable, while partitioned methods can be more computationally efficient. This article compares three methods: two monolithic (Coupled Lagrangian-Eulerian (CLE) and Smoothed Particle Hydrodynamics (SPH)) and one partitioned (lattice-Boltzmann methods (LBM)).
CLE and SPH have proven their ability to model significant structural deformation and domain separation. LBM has demonstrated its ability as an explicit fluid solver. This study aims to validate fluid domain responses for future FSI applications. By comparing and validating fluid modeling capabilities of monolithic and partitioned methods, this work sets the stage for more complex FSI simulations.
The Methodology Throwdown
This study investigates the accuracy and efficiency of three computational fluid dynamics (CFD) methods: Lattice-Boltzmann Method (LBM), Coupled Lagrangian-Eulerian (CLE), and Smoothed Particle Hydrodynamics (SPH). These methods are evaluated by simulating a lid-driven cavity flow, a classical problem in fluid dynamics. The results are compared to an implicit Navier-Stokes solution and established literature to determine the strengths and weaknesses of each approach.
- Navier-Stokes (Baseline): Solved using ANSYS FLUENT to provide a steady-state, incompressible baseline solution.
- Lattice-Boltzmann Method (LBM): A partitioned method using a simplified, Boolean gas particle motion on a lattice structure.
- Coupled Lagrangian-Eulerian (CLE): A monolithic method using an FE formulation designed for FSI and metal forming.
- Smoothed Particle Hydrodynamics (SPH): A monolithic method that discretizes the fluid domain using a set of volumeless particles.
The Verdict
The study concludes that LBM and CLE show the most promise for modeling complex fluid flows, demonstrating good accuracy and computational efficiency. SPH, while versatile, requires further development to improve its accuracy and stability in commercial implementations. These findings provide valuable insights for researchers and engineers selecting CFD methods for FSI simulations and other fluid dynamics applications.