Positivity-preserving third order DG schemes for Poisson-Nernst-Planck equations

Hailiang Liu, Zhongming Wang, Peimeng Yin, Hui Yu

TL;DR

This paper introduces novel third-order discontinuous Galerkin (DG) schemes for solving the Poisson-Nernst-Planck (PNP) equations, which are crucial for modeling ion transport in various devices. The key innovation is a provably positivity-preserving property for ion concentrations, achieved through a combination of a specialized numerical flux, a scaling limiter, and careful time discretization, ensuring physically realistic results.

What Problem Does This Paper Address?

The Poisson-Nernst-Planck (PNP) system is fundamental for understanding ion transport in systems like biological ion channels and semiconductor devices. A critical challenge in numerically solving the PNP system is ensuring that the computed ion concentrations remain non-negative, as negative concentrations are physically impossible. Existing numerical methods often struggle to maintain this positivity, especially at higher orders of accuracy, leading to unphysical results. This paper aims to develop a third-order accurate numerical scheme that rigorously preserves the positivity of ion densities throughout the simulation.

Methods

The authors propose a direct discontinuous Galerkin (DDG) method for the PNP system. The core of their approach involves:

  • Reformulation of the PNP Equations: The original PNP system is reformulated using a nonlogarithmic Landau transformation, which simplifies the drift-diffusion terms.
  • Direct Discontinuous Galerkin (DDG) Scheme: This method uses a piecewise polynomial approximation space and a specific choice of numerical fluxes at element interfaces. The DDG flux is designed to incorporate interface jumps of both the solution and its second-order derivatives, which is key for stability and accuracy.
  • Positivity-Preserving Techniques:
    • Weighted Numerical Integration: A test set of three points is used for weighted numerical integration to ensure a positive decomposition, which helps stabilize the scheme in L∞ norm.
  • Scaling Limiter: A novel scaling limiter is introduced, based on weighted cell averages, to restore positivity when numerical solutions exhibit negative values. This limiter is designed to maintain accuracy.
  • Time Discretization: The schemes are analyzed with first-order Euler forward time discretization, and it's noted that higher-order SSP Runge-Kutta methods can also be employed.

Datasets

Not discussed in the paper. The paper focuses on developing and testing numerical methods for a well-defined mathematical system, rather than training on or evaluating against empirical datasets.

Key Insights and Ideas

  • Provable Positivity Preservation: The most significant insight is the development of a third-order accurate scheme that provably preserves the positivity of ion concentrations. This is a crucial step towards reliable simulations of physical phenomena governed by the PNP equations.
  • DDG for Positivity: The DDG framework, with its specific interface flux formulation, proves to be a powerful tool for achieving positivity preservation, especially when combined with the proposed limiter.
  • Limiter Design: The scaling limiter is a key component, effectively correcting non-physical negative values without sacrificing the scheme's accuracy. Its design is inspired by limiters used for conservation laws.
  • Independence of Poisson Solver: The positivity-preserving property of the density solver is shown to be independent of the Poisson solver, allowing for flexibility in choosing the Poisson solver's parameters.

Key Results

  • Third-Order Accuracy: Numerical examples demonstrate that the proposed DDG schemes achieve third-order spatial accuracy for both one and two-dimensional problems.
  • Positivity Preservation: Simulations show that the cell averages of ion concentrations remain positive throughout the simulation, even for long time integrations, and the limiter effectively restores positivity when needed.
  • Conservation of Mass and Energy Dissipation: The schemes are shown to conserve total mass and exhibit free energy decay, which are intrinsic properties of the PNP system.
  • Robustness: The method is tested on various examples, including those with exact solutions, demonstrating its reliability and effectiveness.

Limitations and Future Work

  • Time Step Restriction: The explicit time discretization (Euler forward) leads to a time step restriction that is dependent on the mesh size (e.g., $\Delta t \sim O(h^2)$). The authors mention that IMEX (Implicit-Explicit) methods could relax this, but the current formulation (2.2) doesn't directly support such a separation.
  • CFL Conditions: The derived CFL conditions for positivity preservation are sufficient but not always necessary. In practice, they are enforced only when positivity is violated.
  • Future Work: The paper focuses on third-order schemes. Extending this framework to even higher orders or exploring more efficient time-stepping strategies could be future directions. Investigating the application of these schemes to more complex geometries or boundary conditions could also be a natural extension.

Summary updated

Abstract

In this paper, we design and analyze third order positivity-preserving discontinuous Galerkin (DG) schemes for solving the time-dependent system of Poisson--Nernst--Planck (PNP) equations, which has found much use in diverse applications. Our DG method with Euler forward time discretization is shown to preserve the positivity of cell averages at all time steps. The positivity of numerical solutions is then restored by a scaling limiter in reference to positive weighted cell averages. The method is also shown to preserve steady states. Numerical examples are presented to demonstrate the third order accuracy and illustrate the positivity-preserving property in both one and two dimensions.
BibTeX
@article{liu2022positivity,
  author = {Hailiang Liu and Zhongming Wang and Peimeng Yin and Hui Yu},
  title = {{Positivity-preserving third order DG schemes for Poisson-Nernst-Planck equations}},
  journal = {Journal of Computational Physics},
  year = {2022},
  doi = {10.1016/j.jcp.2021.110777},
}
Figure