Model of incompressible turbulent flows via a kinetic theory

This paper presents an extended kinetic theory model for incompressible turbulent flows that improves consistency with established turbulence theory, enables accurate simulation of wall-bounded flows through low-Reynolds number damping, and successfully captures non-Newtonian effects by demonstrating that averaged turbulent flows behave similarly to rarefied gas flows.

The Gist

Imagine you are trying to predict how a crowd of people moves through a busy train station.

The Old Way (The "Traffic Cop" Approach):
For decades, engineers have tried to model fluid turbulence (like air around a plane or water in a pipe) by treating it like a smooth, thick fluid. They use rules of thumb, like "if the wind blows this hard, the friction will be that much." It works okay for simple things, but when the flow gets chaotic, swirling, and unpredictable, these rules break down. They rely on "magic numbers" (empirical parameters) that scientists just have to guess or tune based on experiments. It's like trying to predict a traffic jam by only looking at the average speed of cars, ignoring that some drivers are speeding, some are braking, and some are changing lanes.

The New Way (The "Kinetic Theory" Approach):
This paper introduces a fresh perspective. Instead of treating the fluid as a smooth blob, the authors treat it like a gas of billions of tiny, invisible particles (eddies) that are constantly bumping into each other. This is called Kinetic Theory.

Think of it this way:

• The Old Model looks at the crowd from a helicopter and says, "They are moving generally north."
• This New Model puts on a pair of super-glasses and looks at every single person in the crowd, tracking how they jostle, bump, and change direction. It realizes that the "chaos" isn't just noise; it has its own physics.

The Two Big Improvements

The authors took a previous version of this "particle" model and fixed two major bugs:

1. Tuning the "Relaxation Time" (The Bouncing Ball Analogy)
Imagine a ball bouncing on a trampoline. How long it takes to settle down after a bounce is the "relaxation time."

• In the old model, the math for how long the ball bounces was slightly off, leading to weird predictions (like the fluid acting much more "sticky" than it should).
• The authors recalibrated this timing. They found a new "sweet spot" for how long these turbulent eddies interact. This made the math line up perfectly with established laws of physics without needing to cheat with made-up numbers. It's like tuning a guitar string so it hits the perfect note naturally, rather than forcing it.

2. Handling the Walls (The "Crowd at the Door" Analogy)
The old model worked great in the middle of a wide-open field (unbounded flow) but failed miserably near walls (like the side of a pipe).

• Near a wall, the fluid slows down and gets "sticky" due to friction. The old model didn't know how to handle this transition.
• The authors built a special "Low-Reynolds" version of the model. They added a "damping function"—think of it as a traffic warden near the wall who tells the particles to slow down and behave more orderly as they get close to the edge. They also created special rules for how particles bounce off the wall, ensuring the model works from the very center of the flow all the way to the wall.

What Did They Prove?

They tested their new model on a classic problem: Couette Flow.

• The Setup: Imagine two giant parallel plates (like the floor and ceiling of a room). The floor is stationary, and the ceiling is sliding sideways at high speed. The air in between gets dragged along, creating a swirling, turbulent mess.
• The Result: They ran simulations and compared them to real-world experiments and super-accurate computer models (DNS).
  • Speed: Their model predicted the speed of the air perfectly, matching the "logarithmic law" (a famous rule for how speed changes near a wall).
  • Friction: It calculated the drag (friction) on the walls with high accuracy.
  • The Catch: While it was great at predicting the average speed and the main "shear" stress (the sliding friction), it slightly underestimated the complex, chaotic "jiggling" of the particles right next to the wall. It's like the model knows the crowd is moving north, but it's a little unsure about exactly how many people are bumping into the wall.

Why Does This Matter?

The biggest win here is physics over guessing.

• Old Models: "Let's add a coefficient of 0.09 here because it worked in 1980."
• This Model: "The math of how these particles collide naturally leads to a coefficient of 0.0816. We don't need to guess; the physics tells us the answer."

This approach bridges the gap between the chaotic world of turbulence and the clean, predictable world of gas physics. It suggests that even in the wildest storms and fastest jets, there is an underlying order that can be described by tracking the "dance" of these invisible particles, rather than just guessing the average steps.

In a nutshell: The authors built a better "particle simulator" for turbulence. They fixed the timing, added a "wall rule," and proved it works. It's a step toward understanding turbulence not as a messy problem to be approximated, but as a complex dance of particles that follows strict, discoverable laws.

Technical Summary

Here is a detailed technical summary of the paper "Model of incompressible turbulent flows via a kinetic theory."

1. Problem Statement

Turbulent flow modeling remains a significant challenge in fluid dynamics. Traditional approaches, such as Reynolds-Averaged Navier-Stokes (RANS) equations, rely on closure models (e.g., linear eddy viscosity) that often depend on ad hoc empirical parameters and assume a separation of scales between mean and fluctuating flows. These assumptions limit their ability to capture non-equilibrium physics, non-Newtonian effects, and complex anisotropic behaviors, particularly in wall-bounded flows. While kinetic theory offers a mesoscopic perspective that naturally handles non-equilibrium states (analogous to rarefied gas dynamics), previous attempts to apply it to turbulence have suffered from:

• Unrealistic transport coefficients (e.g., excessively high turbulent Prandtl numbers).
• Lack of proper treatment for wall-bounded flows (specifically near-wall damping and viscous sublayer resolution).
• Incomplete theoretical analysis of higher-order transport terms (e.g., turbulent kinetic energy flux).

2. Methodology

The authors extend and analyze a kinetic model for incompressible turbulence originally proposed by Chen et al. (2023), which is based on a Klimontovich-type kinetic equation for fluid elements. The study employs the following methodological steps:

• Kinetic Framework: The model utilizes a velocity distribution function (VDF), $F(\mathbf{x}, \boldsymbol{\xi}, t)$, representing the statistical state of fluid elements. The evolution is governed by a Boltzmann-BGK (Bhatnagar-Gross-Krook) equation with a collision term modeled as a relaxation toward a local Gaussian equilibrium distribution.
• Relaxation Time Re-selection: A critical modification involves redefining the relaxation time $\tau$. The authors derive $\tau = K / (7\epsilon)$ (where $K$ is turbulent kinetic energy and $\epsilon$ is the dissipation rate) to ensure consistency with established turbulence theory, correcting the previous formulation ($\tau = 6K/7\epsilon$) which yielded unrealistic transport coefficients.
• Chapman-Enskog (CE) Expansion: The authors perform a rigorous CE expansion of the kinetic equation. This mathematical technique expands the VDF in powers of a turbulence-based Knudsen number ($\mathcal{K}$) to derive macroscopic transport equations.
  • First Order: Recovers linear eddy viscosity and gradient diffusion models.
  • Second Order: Derives nonlinear eddy viscosity and closure models, including previously unreported expressions for the turbulent kinetic energy (TKE) flux and higher-order derivatives.
• Wall-Bounded Extensions:
  • High-Reynolds-number (HR-BGK) Model: Uses wall functions and non-equilibrium extrapolation boundary conditions for the log-layer.
  • Low-Reynolds-number (LR-BGK) Model: Incorporates a van Driest damping function into the relaxation time and adds an explicit molecular viscous diffusion source term to the kinetic equation. This allows for the direct resolution of the viscous sublayer using a fully diffuse Maxwell boundary condition.
• Numerical Validation: The models are validated against experimental data and Direct Numerical Simulation (DNS) results for turbulent plane Couette flow at Reynolds numbers $Re = 1300$ and $Re = 10,133$. A reduced distribution function approach is used to solve the quasi-one-dimensional problem efficiently.

3. Key Contributions

• Theoretical Consistency: The paper provides a rigorous theoretical derivation showing that the kinetic model naturally yields standard turbulence closures (linear eddy viscosity) at the first order and nonlinear corrections at the second order without empirical tuning.
• Improved Transport Coefficients: By selecting $\tau = K/7\epsilon$, the model achieves a turbulent Prandtl number ($Pr_T \approx 0.7$) and a model constant $C_\mu \approx 0.0816$, which align quantitatively with standard $K-\epsilon$ models, unlike the previous version which predicted $Pr_T \approx 4.2$.
• Novel TKE Flux Derivation: The study derives a closed-form, second-order CE solution for the turbulent kinetic energy flux, capturing counter-gradient diffusion phenomena and non-Fourier heat transfer analogies, which are rarely modeled in conventional RANS.
• Unified Wall Treatment: The development of both HR-BGK and LR-BGK models within a single kinetic framework allows for flexible application: using wall functions for efficiency or resolving the viscous sublayer directly for accuracy.
• Non-Newtonian Insights: The analysis demonstrates that averaged turbulent flow behaves similarly to rarefied gas flow at finite Knudsen numbers, capturing non-Newtonian effects (anisotropy and memory effects) that linear eddy viscosity models miss.

4. Results

• Mean Velocity Profiles: Both the LR-BGK and HR-BGK models show excellent agreement with experimental and DNS data for mean velocity profiles in both outer and wall units across the tested Reynolds numbers. The HR-BGK model performs exceptionally well in the log-layer, while the LR-BGK model accurately captures the viscous sublayer.
• Skin Friction: The models predict skin friction coefficients ($C_f$) with high accuracy (deviations < 5%) across a wide range of Reynolds numbers ($10^3$to$10^4$).
• Reynolds Stresses:
  • Shear Stress: The models accurately predict the Reynolds shear stress distribution and peak locations.
  • Normal Stresses: The models capture the anisotropy of normal stresses (non-Newtonian effects). However, the separation between normal stress components in the near-wall/buffer region ($1 \lesssim y^+ \lesssim 10$) is underestimated compared to DNS, indicating a limitation in the single-relaxation-time approximation for this specific region.
• Velocity Distribution Functions (VDFs): Analysis of the reduced VDFs reveals significant departures from local Gaussian equilibrium, particularly in the streamwise momentum and energy transport components. These non-equilibrium signatures confirm the model's ability to capture physics beyond the Navier-Stokes level.

5. Significance

This work represents a significant step toward a physics-based turbulence modeling framework that reduces reliance on empirical ad hoc parameters.

• Foundational Physics: It bridges the gap between kinetic theory and macroscopic turbulence, demonstrating that turbulence can be modeled via a mesoscopic kinetic equation derived from first principles of the Navier-Stokes equations.
• Beyond Linear Models: The model naturally incorporates nonlinear and non-equilibrium effects (memory effects, anisotropy) that are difficult to capture with traditional linear eddy viscosity models.
• Future Potential: The framework provides a robust foundation for extending turbulence modeling to complex 3D flows, unsteady phenomena, and compressible turbulence, potentially offering a unified approach to multi-scale fluid dynamics without the need for scale-separation assumptions.

In summary, the paper successfully refines a kinetic theory-based turbulence model, validates its theoretical consistency through Chapman-Enskog analysis, and demonstrates its practical efficacy in predicting complex wall-bounded turbulent flows with reduced empirical dependence.