Spurious currents suppression by accurate difference schemes in multiphase lattice Boltzmann method

Zhangrong Qin, Wenbo Chen, Chun‐Yan Qin, Xin Xu, Binghai WenarXiv 2022 · DOI · arXiv

TL;DR

Researchers have significantly reduced "spurious currents"—unphysical velocity artifacts in multiphase fluid simulations—by replacing traditional isotropic gradient operators with high-order compact finite difference schemes. This approach allows for stable, highly accurate simulations of liquid-gas interfaces even at extreme density ratios and low temperatures.

What Problem Does This Paper Address?

In the world of computational fluid dynamics, the Lattice Boltzmann Method (LBM) is a favorite for simulating multiphase flows (like droplets or bubbles) because it handles complex interfaces naturally. However, it has a persistent "ghost": spurious currents.

These are tiny, unphysical circulating flows that appear near curved interfaces. Even when a droplet should be perfectly still, these parasitic velocities persist, draining energy, reducing accuracy, and sometimes causing the entire simulation to crash. They stem from a fundamental "thermodynamic inconsistency"—essentially, the math used to calculate forces at the interface doesn't perfectly balance on the discrete grid.

Methods

The authors investigate the origin of these currents using a multiphase LBM driven by chemical potential. Their core strategy is to improve the accuracy of the discrete gradient calculations used to determine the interaction forces between phases.

They move away from standard "isotropic" difference schemes and implement two main types of high-order schemes:

  1. Explicit Finite Difference Schemes (EFDS): Standard high-order approximations (up to 6th order).
  2. Compact Finite Difference Schemes (CFDS): These involve solving a tridiagonal system of equations to get derivatives, offering much higher "resolution" (accuracy across different spatial scales) for the same number of grid points.

They also introduce an Optimized Compact Finite Difference Scheme (OCFDS) designed specifically for high resolution. To test these, they used the Multiple-Relaxation-Time (MRT) collision operator and the Exact Difference Method (EDM) for force incorporation.

Datasets

As is standard in computational physics, the "datasets" are established numerical benchmarks rather than external data:

  • Flat Interface Simulations: Used to analyze numerical errors and verify coexistence densities against theoretical Maxwell constructions.
  • 2D Circular Droplets: The primary testbed for measuring the magnitude and pattern of spurious currents.
  • 3D Circular Droplets: Used to confirm that the 2D findings scale correctly to three dimensions.
  • Equations of State (EOS): The model was tested across several EOS, including Van der Waals (VDW), Redlich-Kwong-Soave (RKS), Peng-Robinson (PR), and Carnahan-Starling (CS).

Key Insights and Ideas

  • Isotropy vs. Accuracy: While many researchers focus on making gradient operators "isotropic" (the same in all directions), this paper argues that formal accuracy and resolution are actually more critical for suppressing spurious currents.
  • Resolution Matters for Steep Slopes: The authors found that when an interface is "steep" (narrow transition region), the resolution of the scheme (how well it handles high-frequency changes) is a better predictor of accuracy than its formal order of accuracy.
  • The $k$ Factor: They utilized a proportional coefficient ($k$) to decouple the momentum space from the mesh space. This allows researchers to adjust the interface thickness on the grid without changing the underlying physics, which is a powerful tool for balancing stability and resolution.
  • Angular Deviation: They proved that spurious currents are directly caused by the "angular deviation" of the calculated force—if the math says the force points slightly away from the center of a droplet, a vortex is born.

Key Results

  • Dramatic Suppression: Using high-order CFDS and OCFDS reduced spurious currents to the order of $10^{-5}$, even when the density ratio between liquid and gas was as high as 1000.
  • Low-Temperature Stability: The high-order schemes maintained excellent agreement with theoretical predictions (Maxwell equal-area construction) down to a reduced temperature of 0.2. Standard schemes often fail or become highly inaccurate much sooner.
  • 3D Consistency: The 3D simulations showed results highly consistent with 2D, proving the method is robust for real-world spatial complexities.
  • OCFDS Superiority: The optimized 4th-order compact scheme often outperformed even 8th-order standard compact schemes when the interface was sharp.

Limitations and Future Work

  • Boundary Complexity: High-order schemes require information from more neighboring nodes. This makes implementing complex boundary conditions (like a droplet hitting a textured wall) significantly more difficult.
  • Computational Cost: While more accurate, compact schemes require solving a linear system (the Thomas algorithm), which adds a bit of computational overhead compared to simple explicit schemes.
  • Future Directions: The authors suggest that future work should focus on even higher-order schemes and applying these accurate difference methods to more complex, dynamic simulations like droplet splashing or boiling.

Reference: Qin, Z., et al. (2022). Spurious currents suppression by accurate difference schemes in multiphase lattice Boltzmann method. arXiv:2205.09186v1 [physics.comp-ph].

Summary updated

Abstract

Spurious currents, which are often observed near a curved interface in the multiphase simulations by diffuse interface methods, are unphysical phenomena and usually damage the computational accuracy and stability. In this paper, the origination and suppression of spurious currents are investigated by using the multiphase lattice Boltzmann method driven by chemical potential. Both the difference error and insufficient isotropy of discrete gradient operator give rise to the directional deviations of nonideal force and then originate the spurious currents. Nevertheless, the high-order finite difference produces far more accurate results than the high-order isotropic difference. We compare several finite difference schemes which have different formal accuracy and resolution. When a large proportional coefficient is used, the transition region is narrow and steep, and the resolution of finite difference indicates the computational accuracy more exactly than the formal accuracy. On the contrary, for a small proportional coefficient, the transition region is wide and gentle, and the formal accuracy of finite difference indicates the computational accuracy better than the resolution. Furthermore, numerical simulations show that the spurious currents calculated in the 3D situation are highly consistent with those in 2D simulations; especially, the two-phase coexistence densities calculated by the high-order accuracy finite difference are in excellent agreement with the theoretical predictions of the Maxwell equal-area construction till the reduced temperature 0.2.
BibTeX
@misc{qin2022spurious,
  author = {Zhangrong Qin and Wenbo Chen and Chun‐Yan Qin and Xin Xu and Binghai Wen},
  title = {{Spurious currents suppression by accurate difference schemes in multiphase lattice Boltzmann method}},
  year = {2022},
  eprint = {2205.09186},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2205.09186},
}

Loading related content...

Figure