Turbulent mixing of a stably stratified fluid

Antoine Venaille, Joël SommeriaSPIRE 2008

Abstract

We present a phenomenological model for the turbulent mixing of a stratified fluid, where density variations are driven by a scalar such as temperature or salinity. The model tracks the temporal evolution of the scalar's Probability Density Function (PDF) across a vertical profile, incorporating the effects of turbulent diffusion, sedimentation, and the dissipation of scalar fluctuations through a turbulent cascade. By analyzing limit cases and applying the model to restratification, we demonstrate that the turbulent cascade is the critical mechanism ensuring the physical irreversibility of the mixing process.

Introduction

In geophysical environments like the atmosphere and oceans, fluids are typically stably stratified, with denser layers residing beneath lighter ones. Turbulence, induced by wind or heating, disrupts this equilibrium and mixes these layers. Traditional models often focus solely on mean density profiles; however, turbulence creates a heterogeneous "soup" of densities at any given altitude. Accurately modeling these processes—essential for predicting heat and nutrient transport—requires understanding the full Probability Distribution Function (PDF) of the fluid's properties. A key challenge addressed in this work is ensuring the model respects the physical reality that mixing is irreversible.

Modeling Framework

The proposed one-dimensional model focuses on vertical height $z$ and couples the evolution of the buoyancy PDF with turbulent kinetic energy. The model integrates three distinct physical mechanisms:

  1. Turbulent Diffusion: This acts to spread density patches and homogenize the fluid.
  2. Sedimentation: This represents gravity's restorative force, modeling fluid particles moving toward their equilibrium altitude in a "reference state"—the configuration with minimum potential energy.
  3. The Turbulent Cascade (Dissipation): This core innovation uses a "self-convolution" approach to capture how large-scale fluctuations are broken down into smaller scales until they vanish. Mathematically, Laplace transforms describe this "folding" process, providing a physical basis for why mixing increases the minimum potential energy of the system.

The model achieves statistical closure by defining a characteristic lengthscale ($l$) that transitions between the buoyancy length (dominant in high stratification) and the integral scale (dominant in high turbulence).

Results and Discussion

The model successfully predicts the restratification phase following a burst of turbulence. A central insight is the necessity of the dissipation term for physical realism. Without dissipation, the system tends to return to its original state, failing to capture the permanent increase in potential energy associated with irreversible mixing.

Evolution of the mean buoyancy profile over height $z$ at different time intervals, illustrating the transition from initial stratification to a mixed state and subsequent restratification.

As shown in the mean profile evolution, the fluid initially mixes and then settles into a new stratified state. However, the nature of this final state depends heavily on the dissipation of fluctuations. The model demonstrates that during mixing, the PDF broadens as diverse densities are brought to the same height, and then narrows as the cascade dissipates those fluctuations.

Comparison of mean buoyancy profiles and 3D PDFs with and without the dissipation of scalar fluctuations. The dissipation term is essential for capturing the irreversible nature of mixing and the resulting energy penalty.

The comparison in Figure 6 highlights that the turbulent cascade is what makes the mixing "stick." When dissipation is active, the final state possesses higher potential energy than the initial state, representing the quantitative "energy penalty" of mixing. This irreversibility is a hallmark of the model's physical accuracy.

Conclusion

This PDF-based model provides a robust framework for understanding turbulent mixing in stratified fluids. While it currently assumes a smooth cascade rate and ignores small-scale spatial correlations, it offers a significant step forward in capturing the irreversibility of mixing. Future work may extend this 1D approach to complex 2D or 3D scenarios, such as gravity currents, and incorporate internal intermittency to better reflect the sporadic nature of real-world turbulence.

Summary updated

Abstract

Nous proposons un modèle phénoménologique pour le mélange turbulent d'un fluide stratifié. La stratification dépend d'un scalaire, qui peut être la température, ou la concentration en sel du fluide. Le model donne l'évolution temporelle de la forme de la PDF du scalaire. Il prend en compte les effets de diffusion turbulente, de sédimentation et de dissipation des fluctuations du scalaire par cascade turbulente. Nous discutons la paramétrisation du problème. Des cas limites ``test'' sont étudiés pour comprendre l'effet de chaque terme. Le modèle est ensuite appliqué au phénomène de la restratification
BibTeX
@inproceedings{venaille2008turbulent,
  author = {Antoine Venaille and Joël Sommeria},
  title = {{Turbulent mixing of a stably stratified fluid}},
  booktitle = {SPIRE},
  year = {2008},
}

Loading related content...

Figure