Lagrangian Averaging for Compressible Fluids

Harish S. Bhat, Razvan C. Fetecau, Jerrold E. Marsden, Kamran Mohseni, Matthew West

TL;DR

This paper extends the Lagrangian Averaged Euler (LAE-$\alpha$) equations to compressible fluids, offering a new way to regularize fluid dynamics by introducing dispersion instead of artificial viscosity. This approach provides a more physically grounded method for modeling turbulence and shocks in compressible flows.

What Problem Does This Paper Address?

The paper tackles a fundamental challenge in fluid dynamics: accurately modeling compressible flows, especially when dealing with turbulence and shock waves. Traditional methods often rely on adding artificial viscosity to smooth out these phenomena, which can introduce unphysical behavior. The authors propose Lagrangian averaging as an alternative, aiming to regularize the Euler equations by incorporating dispersion, which naturally arises from averaging over small-scale fluid fluctuations. This method seeks to capture the essential physics of turbulence and shocks without the drawbacks of artificial viscosity.

Methods

The core of the paper lies in extending the Lagrangian Averaged Euler (LAE-$\alpha$) framework, previously developed for incompressible fluids, to barotropic compressible flows. The methodology involves:

  • Variational Principle: Starting with a given action for the fluid system.
  • Taylor Expansion: Expanding the action in terms of small-scale fluid fluctuations ($\xi$).
  • Truncation and Averaging: Truncating the expansion and then averaging over these fluctuations. This averaging is performed using a novel "tube of trajectories" approach, where fluctuations are centered around a mean flow.
  • Flow Rules for Closure: Modeling the evolution of the fluctuations using "flow rules." These rules are crucial for closing the system of equations, providing a way to determine how the fluctuations evolve over time.
  • Isotropic and Anisotropic Models: Deriving both isotropic (assuming statistical uniformity in all directions) and anisotropic (allowing for directional variations) models for the averaged equations.

The paper also revisits the incompressible case to clarify the derivation and then applies the generalized procedure to compressible flows, carefully separating the averaging process from the modeling assumptions.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Dispersion as a Regularizer: The most significant insight is the use of Lagrangian averaging to introduce dispersion, rather than artificial viscosity, as a means to regularize the compressible Euler equations. This offers a more physically motivated approach to handling turbulence and shocks.
  • Unified Treatment of Shocks and Turbulence: The Lagrangian averaging method provides a single framework to simultaneously model the effects of turbulence and strong shocks in compressible flows, which is a significant advantage over separate modeling approaches.
  • Averaging at the Variational Level: The paper emphasizes that averaging is performed at the level of the variational principle, not at the level of the final PDEs. This is a key distinction from traditional averaging techniques like RANS and LES.
  • Flow Rules for Closure: The introduction of "flow rules" as a mechanism for achieving closure in the averaged equations is a critical methodological contribution. These rules dictate the evolution of fluid fluctuations and are essential for obtaining a solvable system.
  • Simplified Derivation: The authors present a simplified and clarified derivation of both isotropic and anisotropic LAE-$\alpha$ equations for compressible fluids, making the methodology more accessible.

Key Results

The paper successfully derives the Lagrangian averaged Euler (LAE-$\alpha$) equations for barotropic compressible fluids, presenting both anisotropic and isotropic models. These equations are derived by extending the Lagrangian averaging technique to account for density fluctuations alongside velocity fluctuations. The paper also provides an improved derivation of the incompressible LAE-$\alpha$ equations. The derived equations are shown to reduce to the standard Euler equations when the averaging parameter $\alpha$ is zero.

Limitations and Future Work

The authors acknowledge several limitations and suggest avenues for future research:

  • Initialization Problem: A significant open problem is how to obtain initial data for the LAE-$\alpha$ equations from initial data for the original Euler equations. The paper highlights the lack of a theoretical foundation for current initialization methods.
  • Barotropic Assumption: The current derivation is limited to barotropic flows, meaning the potential energy is a function only of density. Future work aims to remove this assumption to derive models valid in regimes with strong physical shocks.
  • One-Dimensional Models: The current framework does not readily yield a one-dimensional compressible averaged model. The authors are investigating a filtered Lagrangian approach for this purpose.
  • Numerical Implementation: While the paper focuses on the theoretical derivation, the practical implementation and numerical simulation of these equations, especially for complex flows and wall-bounded scenarios, remain areas for further exploration.
  • Relationship with Other Models: Further investigation is needed to understand the relationship between LAE-$\alpha$ models and other turbulence modeling approaches like RANS and LES, particularly concerning continuity equations for mean density.
  • Thermodynamics and Equation of State: The interaction of Lagrangian averaging with the thermodynamics of the system and the derivation of a corresponding equation of state for averaged variables is an open question.

Summary updated

Abstract

This paper extends the derivation of the Lagrangian averaged Euler (LAE-$\alpha$) equations to the case of barotropic compressible flows. The aim of Lagrangian averaging is to regularize the compressible Euler equations by adding dispersion instead of artificial viscosity. Along the way, the derivation of the isotropic and anisotropic LAE-$\alpha$ equations is simplified and clarified. The derivation in this paper involves averaging over a tube of trajectories $\eta^\epsilon$ centered around a given Lagrangian flow $\eta$. With this tube framework, the LAE-$\alpha$ equations are derived by following a simple procedure: start with a given action, Taylor expand in terms of small-scale fluid fluctuations $\xi$, truncate, average, and then model those terms that are nonlinear functions of $\xi$. Closure of the equations is provided through the use of flow rules, which prescribe the evolution of the fluctuations along the mean flow.
BibTeX
@inproceedings{bhat2005lagrangian,
  author = {Harish S. Bhat and Razvan C. Fetecau and Jerrold E. Marsden and Kamran Mohseni and Matthew West},
  title = {{Lagrangian Averaging for Compressible Fluids}},
  booktitle = {Multiscale},
  year = {2005},
  doi = {10.1137/030601739},
}
Figure