An incompressible multi-phase SPH method

Xiangyu Hu, Nikolaus A. AdamsJournal of Computational Physics 2007 · DOI

TL;DR

This paper introduces a novel incompressible multi-phase Smoothed Particle Hydrodynamics (SPH) method that accurately handles sharp density and viscosity discontinuities. By employing a fractional time-step approach and a new projection formulation, the method enforces both zero-density variation and velocity divergence-free conditions, leading to more stable and accurate simulations of complex fluid flows.

What Problem Does This Paper Address?

Simulating incompressible multi-phase flows, especially those with sharp interfaces between fluids of different densities and viscosities, presents a significant challenge for traditional SPH methods. Existing approaches often struggle with:

  • Smearing of Interfaces: Standard SPH formulations tend to smooth out sharp discontinuities in density and viscosity, leading to inaccurate representation of interfaces and artificial surface tension effects.
  • Particle Clustering: Some methods designed to enforce incompressibility can lead to undesirable particle clustering, which further degrades accuracy.
  • Computational Cost: While weakly compressible SPH methods are simpler, they are less accurate for high Reynolds number flows. Strictly incompressible methods often involve solving complex Poisson equations, increasing computational expense.

This paper aims to bridge this gap by developing an SPH method that can accurately capture these sharp interfaces without sacrificing computational efficiency or introducing numerical artifacts.

Methods

The core of the proposed method lies in a novel fractional time-step integration algorithm and a multi-phase projection formulation:

  • Fractional Time-Step Integration: The algorithm splits the time-stepping process into half-steps and full-steps. Intermediate velocities and positions are calculated, and then corrected iteratively to satisfy the incompressibility constraints.
  • Multi-Phase Projection Formulation: This is the key innovation. The authors introduce new gradient and divergence operators that do not require the density or viscosity fields to be differentiable. This allows for the direct handling of sharp discontinuities without smearing.
  • Simultaneous Enforcement of Constraints: The method simultaneously enforces two crucial incompressibility conditions:
  • Zero-Density-Variation Condition: This is achieved by adjusting particle positions to maintain constant particle density.
  • Velocity-Divergence-Free Condition: This is enforced by modifying particle velocities to ensure zero divergence.
  • Iterative Correction Scheme: The density and velocity corrections are performed iteratively, ensuring that viscous and surface forces are calculated with constrained particle positions and velocities, minimizing errors introduced by these forces.
  • Reference Pressure: To avoid stability issues associated with negative pressures, a constant positive reference pressure is superimposed onto the computed pressure.

Datasets

Not discussed in the paper. The paper focuses on validating the method with established benchmark problems.

Key Insights and Ideas

  • Sharp Interfaces without Smearing: The most significant insight is the development of gradient and divergence operators that can handle discontinuous fields. This is crucial for accurately modeling multi-phase flows where interfaces are inherently sharp.
  • Unified Approach for Incompressibility: By enforcing both density and velocity constraints within a single fractional time-step framework, the method provides a robust and consistent way to handle incompressibility in multi-phase flows.
  • Reduced Artificial Surface Tension: By avoiding the smearing of density discontinuities, the method naturally eliminates spurious pressure effects that can manifest as artificial surface tension in other SPH formulations.
  • Efficiency Gains: The method aims to be computationally more efficient than weakly compressible SPH for high Reynolds number flows and offers a more direct approach than some other incompressible methods.

Key Results

The paper presents numerical examples that demonstrate the method's accuracy and convergence properties:

  • Taylor-Green Flow: The method accurately reproduces the decay of velocity and maintains a uniform particle distribution, even with a large number of particles. The constrained density correction significantly reduces errors compared to unconstrained simulations.
  • Droplet Oscillation: The simulations show good agreement with existing results for both single-phase and different-density multi-phase droplet oscillations, with the present method eliminating noise from artificial sound waves observed in weakly compressible SPH.
  • Drop Deformation in Shear Flow: The method accurately predicts the deformation of drops in shear flows, showing good agreement with theoretical predictions and other numerical methods across various capillary and viscosity ratios.
  • Rayleigh-Taylor Instability: The simulations capture the asymmetric evolution of the interface and predict a stronger roll-up of plumes compared to previous SPH methods, attributed to the direct handling of density discontinuities. The method also avoids "particle clumping" issues.

Limitations and Future Work

  • Computational Overhead: While aiming for efficiency, the iterative nature of the projection method and the enforcement of multiple constraints do introduce some computational overhead compared to simpler methods.
  • Time Step Limitations: The paper outlines several time-step criteria (CFL, viscous-diffusion, surface tension) that must be satisfied for stability, which can limit the time step size in certain scenarios.
  • Future Work: The authors suggest that the method's ability to handle sharp interfaces directly opens avenues for studying phenomena like secondary instabilities and complex interface dynamics with greater fidelity. The paper also implies that further optimization of the iterative solvers could enhance computational efficiency.

Summary updated

BibTeX
@article{hu2007incompressible,
  author = {Xiangyu Hu and Nikolaus A. Adams},
  title = {{An incompressible multi-phase SPH method}},
  journal = {Journal of Computational Physics},
  year = {2007},
  doi = {10.1016/j.jcp.2007.07.013},
}

Loading related content...

Figure