TL;DR
This paper introduces a novel numerical scheme for simulating fluid dynamics with variable density, specifically the Navier-Stokes equations. The key innovation is a method that rigorously preserves fundamental physical properties like mass conservation, density positivity, and energy conservation, which are often lost in traditional numerical approaches. This is achieved through a combination of a square transformation for density positivity and a new recovery technique to maintain mass and energy balance, leading to more accurate and reliable simulations.
What Problem Does This Paper Address?
Simulating fluid flow governed by the Navier-Stokes equations is crucial in many scientific and engineering fields. However, when the fluid's density changes (variable density), these equations become significantly more complex. Standard numerical methods often struggle to accurately capture essential physical properties of the fluid. Specifically, maintaining:
- Mass Conservation: The total mass of the fluid should remain constant over time.
- Density Positivity: The density of the fluid must always be a positive value.
- Energy Identical-Relation: The way energy is conserved or dissipated should accurately reflect the underlying physics.
These properties are vital for the stability and physical realism of simulations. Existing methods often sacrifice one or more of these properties, leading to inaccuracies, especially in long-term simulations or when dealing with complex flow phenomena. This paper aims to bridge this gap by developing a numerical scheme that preserves all these critical physical properties simultaneously for variable density flows.
Methods
The authors propose a fully discrete, first-order finite element method. The core of their approach involves several key techniques:
- Square Transformation for Density Positivity: To ensure the numerical density remains positive, they employ a transformation where the density $\rho$ is represented as $\rho = \sigma^2$. This guarantees $\rho \ge 0$ regardless of the numerical scheme's behavior.
- New Recovery Technique for Mass and Energy Conservation: While the square transformation helps with positivity, it can lead to a loss of mass conservation and numerical dissipation of energy when approximating the transformed equations. To counteract this, they introduce a novel "recovery technique." This technique is applied in later steps of the scheme to recover the discrete system's mass and eliminate unwanted numerical energy dissipation, thereby preserving the energy identical-relation.
- Equivalent Formulation: The momentum equation is reformulated to facilitate the derivation of the energy identical-relation.
- Finite Element Discretization: The spatial domain is discretized using finite element spaces, with specific choices for approximating density, velocity, and pressure.
- Time Stepping: A first-order time-stepping scheme is used.
The novelty lies in the combination of these elements, particularly the specific recovery technique designed to simultaneously address mass and energy conservation issues arising from the density transformation.
Datasets
Not discussed in the paper. The paper focuses on developing and analyzing a numerical method, and its evaluation is done through numerical experiments with prescribed analytical solutions and benchmark flow problems (back-step flow, flow around a cylinder).
Key Insights and Ideas
- Unified Preservation of Physical Properties: The most significant insight is the successful development of a single numerical scheme that simultaneously preserves mass conservation, density positivity, and the energy identical-relation for variable density Navier-Stokes equations. This is a challenging feat in numerical fluid dynamics.
- Form-Invariant Positivity: The square transformation for density positivity is form-invariant and independent of the specific discrete scheme, making it a robust tool.
- Recovery as a Key Strategy: The paper highlights the effectiveness of a carefully designed "recovery technique" to correct for numerical artifacts introduced by transformations or approximations, particularly in maintaining mass and energy balance.
- First-of-its-Kind Energy Identical-Relation Preservation: The authors claim this is the first work to preserve the original energy identical-relation for variable density Navier-Stokes equations, distinguishing it from methods that might preserve a modified energy balance.
Key Results
- Theoretical Guarantees: The proposed scheme is proven to inherit mass conservation, density positivity, and the energy identical-relation from the continuous equations.
- Error Estimates: The paper derives error estimates for the numerical solution, showing a convergence rate of $O(\tau^2 + h^2)$ in the $L^2$-norm for the density and velocity, where $\tau$ is the time step and $h$ is the spatial mesh size.
- Numerical Verification: Numerical examples demonstrate the correctness and efficiency of the scheme.
- Convergence Orders: Tables 1 and 2 show that the numerical results achieve the expected convergence orders for the scheme and recovery factors.
- Property Preservation: Figures 1 and 2 illustrate the conservation of mass (after recovery), positivity of density, and the energy identical-relation (differences between the two sides of the relation are close to zero) under different conditions (e.g., with and without body force).
- Flow Simulations: Simulations of back-step flow and flow around a cylinder (Figures 4-13) show physically plausible results, capturing phenomena like vortex shedding and the impact of variable density on flow patterns.
Limitations and Future Work
- First-Order Scheme: The proposed scheme is first-order accurate in time. While it preserves key properties, higher-order accuracy might be desirable for some applications.
- Numerical Oscillations in Density: The authors acknowledge that due to the hyperbolic nature of the density equation, small numerical oscillations in the density can still appear. They suggest that methods like the least squares finite element method or discontinuous Galerkin methods might mitigate this, though these approaches have their own technical challenges.
- Future Directions: The authors suggest extending their work to:
- Higher-order schemes that preserve properties.
- Coupled problems involving other physical fields, such as electrohydrodynamics or magnetohydrodynamics.
Summary updated