🔬 physics

Monolithic kinetic algorithm for heterogeneous porous media systems using a continuous one-domain approach

This paper proposes a monolithic lattice Boltzmann model that utilizes a continuous one-domain approach to accurately simulate weakly compressible, isothermal flows in heterogeneous porous media by recovering generalized volume-averaged equations with correct porosity scaling, Galilean invariance, and enhanced numerical stability for device-scale design applications.

Original authors: Nikita Gusev, Ilya Karlin

Published 2026-06-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Nikita Gusev, Ilya Karlin

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to predict how water flows through a complex landscape. Sometimes the water flows freely through a wide, open river (free fluid). Other times, it has to squeeze through a dense forest of trees or a thick sponge (porous media). In the real world, these two worlds often meet: a river might flow into a marsh, or a filter might sit inside a pipe.

For a long time, computer scientists had to use two different rulebooks to simulate this. One rulebook for the open river, and a completely different, complicated set of rules for the sponge. When the water moved from the river to the sponge, the computer had to stop, switch rulebooks, and try to glue the two simulations together at the boundary. This was messy, prone to errors, and computationally expensive.

The Big Idea: One Rulebook for the Whole World

The researchers in this paper, Nikita Gusev and Ilya Karlin, have built a new "monolithic" (all-in-one) computer algorithm. Think of it like a universal translator that speaks the language of both the open river and the sponge simultaneously.

Instead of treating the river and the sponge as two separate places, their model treats the entire computer screen as one continuous landscape. In this landscape, the "sponginess" (porosity) can change smoothly from 0% (solid rock) to 100% (empty space). The computer doesn't need to know where the river ends and the sponge begins; it just calculates the flow based on how "spongy" the local area is.

How It Works: The "Particle Crowd" Analogy

To do this, they use a method called the Lattice Boltzmann Method (LBM). Imagine the fluid isn't a continuous river, but a massive crowd of tiny, invisible particles hopping from one grid square to the next.

In an open area, the particles hop freely.
In a porous area, the particles bump into invisible "obstacles" (the solid parts of the sponge), which slows them down and changes their direction.

The paper's innovation is a new way to program these particles so they automatically know how to behave when they move from a free-flowing area into a porous one, without the computer needing to draw a hard line between them.

The "Traffic Cop" Problem

A major headache in previous simulations was that the math didn't add up correctly when the "sponginess" changed.

The Old Way: Imagine a traffic cop trying to manage cars. In the open highway, the cop counts cars normally. In the construction zone (porous media), the cop tries to count them, but the math gets messy. The pressure and speed calculations would get distorted depending on how many construction barriers were there. This meant the simulation would break if the construction zone wasn't perfectly uniform.
The New Way: The authors fixed the "traffic cop's" math. They created a new set of rules that ensures the pressure and speed calculations remain accurate even if the density of the construction barriers changes from one spot to the next. They also fixed how the "friction" (viscosity) is calculated, ensuring it behaves correctly whether the fluid is moving fast or slow, and whether the area is a sponge or an open pipe.

What They Tested

To prove their new "universal rulebook" works, they ran several tests:

The Flowing Channel: They simulated fluid moving through a pipe where the "sponginess" changed smoothly from one side to the other. Their model matched the known physics perfectly.
The Porous Cylinder: They simulated wind blowing past a cylinder made of a porous material (like a mesh ball). They showed that if the ball is very porous, the wind goes right through it, smoothing out the turbulence behind it. If it's less porous, it acts more like a solid rock, creating a wake of swirling air behind it. Their model captured all these subtle changes.
The Cavity: They simulated a box with a moving lid (like a sliding door) and a porous obstacle inside. They showed how the swirling vortex of air inside the box changes shape depending on how "spongy" the obstacle is.

Why This Matters

The paper claims this method is a powerful tool for early-stage design. Because the model is so flexible and handles changing porosity so well, engineers can use it to design things like:

Filters and Separators: Designing materials where the density of the filter changes to optimize flow.
Fuel Cells: Understanding how gas moves through the complex, porous structures inside a battery or fuel cell.
Heat Exchangers: Designing structures where fluid flows through packed beds to transfer heat efficiently.

In short, they built a single, robust computer engine that can simulate fluid flowing through anything from an empty pipe to a dense sponge, and anywhere in between, without needing to switch gears or draw artificial lines. This allows for faster, more accurate designs of devices that rely on complex fluid flow.

Technical Summary: Monolithic Kinetic Algorithm for Heterogeneous Porous Media Systems

Problem Statement
The simulation of fluid flow in heterogeneous porous media, particularly in systems coupling free-fluid regions with porous structures, presents significant challenges. Traditional approaches often rely on "two-domain" formulations, which solve distinct governing equations for free-fluid and porous regions and require explicit interfacial boundary conditions (e.g., Beavers-Joseph conditions) to couple them. Alternatively, "one-domain" approaches use a single set of equations but have historically struggled with consistency in heterogeneous media.

Existing Lattice Boltzmann Method (LBM) formulations for porous media often suffer from inconsistent porosity scaling in the recovered macroscopic equations. Specifically, previous models frequently fail to correctly scale the pressure gradient and viscous stress terms with porosity when porosity varies spatially. Furthermore, many formulations express the Brinkman viscous stress in terms of gradients of the intrinsic velocity rather than the superficial (Darcy) velocity, leading to inaccuracies in heterogeneous domains. Additionally, standard LBM models often lack the flexibility to independently tune the effective bulk viscosity or the speed of sound, limiting computational efficiency and stability.

Methodology
The authors propose a monolithic, single-domain Lattice Boltzmann model based on a continuous one-domain approach. The methodology is built upon the following pillars:

Macroscopic Model: The target hydrodynamics are derived using the method of volume averaging, resulting in a generalized Darcy-Forchheimer-Brinkman model. This model is valid across the entire computational domain, treating the transition between free-fluid and porous regions as a continuous variation of physical properties (porosity $\epsilon$ ). The governing equations include:
- A continuity equation for volume-averaged density ( $\epsilon\rho$ ).
- A momentum equation incorporating Darcy drag, Forchheimer inertial drag, and an effective viscous stress tensor ( $\sigma_{eff}$ ).
- Crucially, the effective viscous stress is expressed in terms of gradients of the superficial velocity ( $v$ ), not the intrinsic velocity ( $u = v/\epsilon$ ), ensuring consistency with the Brinkman formulation for variable porosity.
Kinetic Solver: The model utilizes a single generalized kinetic equation on standard lattices (e.g., D2Q9, D3Q19).
- Equilibrium Distribution: A product-form discrete equilibrium is employed to ensure correct moments and Galilean invariance. This includes specific correction terms ( $\Phi_{ext}$ and $\Phi_{\eta}$ ) to restore isotropy in the viscous stress tensor and allow for independent tuning of the bulk viscosity.
- Forcing Scheme: A consistent generalized forcing scheme (Relax-to-Force method) is used to incorporate Darcy-Forchheimer drag, body forces, and necessary corrections for porosity gradients.
- Porosity Scaling: The locally conserved fields are defined as volume-averaged density and momentum, naturally incorporating porosity scaling into the kinetic moments.
- Velocity Evaluation: An iterative algorithm is proposed to solve the implicit relationship between the macroscopic velocity and the forcing terms, handling the nonlinear Darcy-Forchheimer drag and viscous corrections.
Boundary Conditions: The authors adapt standard wet-node boundary conditions (equilibrium scheme, non-equilibrium extrapolation, and Tamm-Mott-Smith scheme) to the porous-media formulation. A specific procedure is derived to calculate the intrinsic fluid density at boundaries using mass continuity, accounting for the presence of the forcing term and porosity variations.

Key Contributions

Consistent One-Domain Formulation: The paper presents the first LBM formulation that recovers a Brinkman viscous stress tensor with fully consistent porosity scaling, expressed in terms of the superficial velocity. This allows the model to handle spatially varying porosity without artificial interfacial conditions.
Correction of Pressure and Viscous Terms: The authors derive and implement specific forcing corrections to rectify inconsistencies in the pressure gradient and viscous stress terms that arise in heterogeneous media, ensuring the model recovers the correct macroscopic limit.
Independent Parameter Tuning: The model allows the speed of sound and the effective bulk viscosity to be specified independently of the shear viscosity. This enhances computational efficiency (by allowing higher sound speeds) and numerical stability (by tuning bulk viscosity).
Monolithic Algorithm: The approach uses a single kinetic equation and solver for the entire domain, simplifying the implementation for complex geometries involving both free-fluid and porous regions.

Results and Validation
The model was validated through a series of benchmarks and numerical studies:

Consistency Analysis: Linear temporal spectral analysis confirmed that the model recovers the correct effective speed of sound and dissipation rates for shear and acoustic modes. The analysis demonstrated that the product-form equilibrium with the proposed corrections preserves Galilean invariance and isotropy, unlike previous second-order equilibrium formulations.
Benchmark Flows:
- Couette Flow: Simulations in a homogeneous porous channel under gravity showed excellent agreement with reference numerical solutions for both velocity and density profiles across various Reynolds and Darcy numbers.
- Poiseuille Flow: The model accurately captured flow in homogeneous channels and, crucially, in channels with heterogeneous porosity profiles. It correctly predicted flow channeling effects where fluid preferentially flows through high-porosity regions.
Numerical Demonstrations:
- Flow Past a Permeable Cylinder: The model successfully simulated confined flow past a porous cylinder, identifying five distinct flow regimes based on Reynolds number and core porosity. It captured phenomena such as wake stabilization, detached recirculation, and far-wake instabilities, showing good qualitative agreement with literature.
- Lid-Driven Cavity with Porous Obstacle: The model simulated a cavity with an internal porous obstacle, demonstrating how permeability affects vortex formation, flow penetration, and recirculation patterns. The results highlighted the transition from solid-like behavior to permeable behavior as porosity increased.

Significance and Claims
The authors claim that this work provides a robust, unified framework for simulating transport in heterogeneous porous media. By eliminating the need for explicit interfacial boundary conditions and ensuring consistent scaling of all hydrodynamic terms, the model is particularly suited for:

Early-stage, device-scale design studies of engineered porous structures with spatially varying porosity.
Simulating complex interactions between free fluids and permeable bodies (e.g., fuel cells, filtration systems, porous combustion).
Capturing transport phenomena in systems where the transition between free flow and porous media is continuous or where sharp interfaces are difficult to resolve.

The paper concludes that while the current study focuses on 2D validation, the formulation is inherently multidimensional and ready for extension to 3D geometries. Future work is identified as necessary for including energy balance (heat transfer), multicomponent mixtures, and multiphase flows to fully address coupled transport in realistic engineering systems.

Technical Summary: Monolithic Kinetic Algorithm for Heterogeneous Porous Media Systems

More like this