Time filtering methods for electrohydrodynamics models

Conghui LiarXiv 2024 · DOI · arXiv

Abstract

Electrohydrodynamics (EHD) involves the complex coupling of electric fields and fluid flow, presenting significant challenges for numerical stability and accuracy. This paper introduces a robust numerical framework that integrates a "time filtering" step into a standard Finite Element Method (FEM). By applying this post-processing correction, the proposed algorithm achieves second-order temporal convergence for variable density models. The framework is validated using manufactured solutions for both variable density and temperature-dependent EHD models, demonstrating its potential for applications in inkjet printing, electrostatic spraying, and microfluidic cooling.

Introduction

Simulating Electrohydrodynamics (EHD) is notoriously difficult due to the tight coupling between fluid mechanics (Navier-Stokes equations), charge transport, and electrostatics (Poisson’s equation). Small numerical errors in charge distribution can propagate into massive instabilities in fluid motion predictions. While standard numerical methods exist, achieving high-order temporal accuracy typically requires complex and computationally expensive algorithms.

This research addresses the need for a calculation format that is both easy to implement and mathematically well-behaved. The core objective is to ensure that numerical simulations reliably converge toward the true physical solution as time steps are refined.

Methodology

The proposed approach utilizes the Finite Element Method (FEM) for spatial discretization, partitioning the fluid domain into a triangular mesh. The primary innovation lies in the temporal treatment through a "time filtering" mechanism.

Dual Model Approach

The study establishes numerical formats for two distinct EHD scenarios:

  1. Variable Density Model: Focuses on the interaction between charge density and fluid velocity.
  2. Temperature-Dependent Model: Incorporates thermal effects, adding temperature as a coupled variable.

Time Filtering and Variational Formulation

Instead of a standard one-step update, the algorithm employs a time-filter post-processing step. After computing an intermediate solution for the current time step, the algorithm "corrects" it using values from the two previous time steps. This method allows a scheme that might otherwise be first-order accurate to reach second-order convergence. The paper derives the weak forms (integral equations) for the coupled system to ensure the conservation of mass, momentum, and charge across the mesh.

Numerical Experiments

To validate the framework, the author uses Manufactured Solutions implemented in the FreeFEM++ software package. By defining complex trigonometric functions as "true" solutions for velocity, pressure, and charge density, the exact numerical error can be quantified.

Key Results

The effectiveness of the time-filtering method is demonstrated through convergence analysis. For the Variable Density Model, the error reduction follows a second-order pattern. As the number of time steps ($N$) increases, the convergence order for all variables—including density, velocity, and potential—consistently approaches 2.0.

Table 1: Convergence results for the variable density model, showing second-order accuracy (Order ≈ 2.0) for various physical parameters.

For the more complex Temperature-Dependent Model, the method maintains stability. While the specific implementation tested shows convergence orders hovering near 1.0, the framework remains robust despite the added physical complexity of the thermal variable ($\theta$).

Table 2: Convergence results for the temperature-dependent model, demonstrating stable first-order convergence (Order ≈ 1.0) as complexity increases.

Discussion and Conclusion

The "time filter" approach provides a significant advantage by decoupling the complexity of the EHD equations. It allows for solving charge density, fluid velocity, and electric potential in a sequence that maintains stability without requiring the solution of a single massive matrix at every step. This modularity makes the method highly applicable to various industrial processes, from pesticide spraying to microfluidic cooling.

Future work may address the computational overhead associated with storing data from previous time steps, particularly for large-scale 3D simulations. Additionally, exploring adaptive time-stepping could further optimize performance during transient flow events. Ultimately, validating these second-order results against physical laboratory data from EHD experiments will be the next step in bridging the gap between mathematical theory and industrial application.

Summary updated

BibTeX
@misc{li2024time,
  author = {Conghui Li},
  title = {{Time filtering methods for electrohydrodynamics models}},
  year = {2024},
  eprint = {2403.14308},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2403.14308},
}
Figure