Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries

arXiv:2605.30878 · physics.flu-dyn, physics.comp-ph · Submitted 2026-05-29 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries".

Mira: Here's the gist of the paper. Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries introduces a wetting boundary condition for curved solid boundaries within a diffuse interface framework for lattice Boltzmann method.

Kai: Numerical simulations performed on an NVIDIA A100 GPU demonstrate that the wetting scheme retains the model’s ability to handle large density and viscosity contrasts while producing relatively small spurious currents.

Mira: The present scheme agrees well with analytical solutions/other numerical works for both static and dynamic contact lines on curved solid boundaries.

Kai: Model Formulation The color-gradient model is used to simulate incompressible immiscible Newtonian two-phase flow, governed by the equations of fluid dynamics and the phase-field evolution equation.

Title and authors: Kai: We've been talking about the paper "Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries" and how it sets up a new way to handle wetting conditions on curved surfaces. This segment focuses specifically on what the title and authors tell us about this research area.

Mira: The authors are bringing their specific method—the color-gradient lattice Boltzmann modeling—to solve the problem of applying a wetting boundary condition when the solid surface has a curve. It’s about bridging two different computational worlds: standard fluid dynamics simulation methods and the constraints imposed by physical wetting phenomena on curved geometry.

Lev: The core idea is that they're using this color-gradient model to simulate two immiscible Newtonian fluid flow, which means we're dealing with fluids that don't mix, and they are governed by the standard equations of fluid dynamics plus a phase field evolution equation. That’s the setup we need to be aware of.

Kai: The key thing is how they introduce this wetting boundary condition into that framework. It isn't just slapping on a simple mathematical rule; it relies on updating the order parameter, or the color/phase field values, specifically at ghost nodes that sit inside the solid phase.

Mira: That update rule is what allows them to extend the equilibrium color profile—that's their reference to phi eq of eta—into the solid phase itself. This extension into a region where there should ideally be no fluid is where they are making their main physical contribution.

Lev: And this whole process of updating the ghost nodes is what feeds into solving the fluid dynamical equations, which we need to consider because that’s how you actually get motion and flow in the simulation.

Kai: The paper claims that they use a specific equilibrium color profile solution for this, which is given by one half plus or minus a hyperbolic tangent multiplied by two times eta over W. That's the mathematical expression they start with.

Mira: And then they define the wetting boundary condition itself as something like the normal vector dot product of the unit normal and that color field equals negative cosine theta. That’s how they translate the physical contact angle into a mathematical constraint for their simulation.

Lev: I'm always thinking about where this rule fits in terms of implementation. If you're running this on actual hardware, you need to know exactly what kind of stencil or approximation you are using for those gradient evaluations, because that’s where the accuracy can really drop off quickly.

Kai: And they address that by saying their method relies on a color-gradient model that employs a velocity-based equilibrium distribution function, which is a specific way to structure the lattice Boltzmann equations themselves.

Mira: That choice of distribution function is what allows them to solve the fluid dynamics equations for these two immiscible fluid phases in this manner, and it’s crucial because it dictates how well the system behaves under different flow conditions.

Lev: So, when we look at the numbers, they are using a grid where equal spacing in the x and y directions is defined as one. That's important because that sets the scale for all their measurements of things like distance and normal vectors.

Kai: And they also define a specific relationship between that grid spacing, which we call delta eta s, and the physical distance between the ghost node and a point in the fluid domain, which is exactly how they set up their measurement points to test things.

The paper's summary: Kai: So to wrap up what we just covered, this paper is essentially saying they introduced a wetting boundary condition for curved solid boundaries inside a diffuse interface framework for lattice Boltzmann methods. This means they are solving the problem of getting the physical wetting behavior right on non-flat surfaces using this specific mathematical modeling technique.

Mira: In simple terms, the summary says they've developed this new way to handle things by modifying ghost nodes to extend the equilibrium color profile into the solid phase so that it respects those curved boundary conditions. They are showing that this works for both static and dynamic contact lines on curved surfaces while matching analytical solutions.

Lev: The practical implication is that they’re tackling a known issue in simulations where you have to enforce a no-slip or wetting condition on a surface, but now they have a specific prescription for how to do it using color fields instead of some other, less reliable method.

Kai: They are also emphasizing that this scheme retains the model's ability to handle large density and viscosity contrasts present in the fluid dynamics simulation while keeping the resulting spurious currents relatively small. That’s a big win for physical accuracy.

Mira: And they put their results side-by-side with Fakhari et al.'s work, showing that their scheme agrees well with both analytical solutions and other numerical works for both static and dynamic contact lines on curved solid boundaries.

Lev: So the summary boils down to this: they fixed a known problem in lattice Boltzmann methods by using a color gradient model to enforce the wetting boundary condition on curved surfaces effectively, and it performs better numerically than existing methods tested against them.

Kai: It’s about taking a complex physical setup—curved boundaries, immiscible fluids with high contrast—and finding a stable way to model the interface interaction that keeps errors under control.

The paper's improvements: Mira: Now let's talk about what the paper suggests are the specific improvements they made to this modeling approach, beyond just stating what they did. This is where we look at how their methodology itself is actually better than what came before.

Kai: One major improvement is the way they handle those ghost nodes. Instead of just relying on some simpler extrapolation methods, they introduce a specific color-field update rule that extends the equilibrium profile into the solid phase based on the contact angle theta and a point in the fluid domain.

Mira: That rule is superior because it directly enforces the physics at that boundary, which means it avoids issues like introducing artificial mass sources or sinks at solid boundaries that other approaches have run into, which is mentioned in reference twenty one.

Lev: From an implementation standpoint, this direct enforcement is good because it tackles the non-locality issues too; they mention that using higher-order stencils for gradient evaluation can introduce extra complexity and errors when you try to enforce these conditions.

Kai: They also show that the way they handle the order parameter profile near a solid boundary can be sensitive, but their scheme manages to keep those errors manageable by controlling the grid resolution parameter epsilon.

Mira: And they showed that in the limit where epsilon approaches zero, which means increasing the interface grid resolution, their proposed method converges toward Fakhari et al.'s prescription as well. That’s a strong convergence proof showing it's not just a lucky coincidence but mathematically sound behavior.

Lev: That convergence to an existing scheme is a nice piece of validation because it shows they aren't inventing something completely new and weird; they are refining an existing concept by showing how their parameter epsilon controls the transition between different behaviors.

Kai: And on the performance side, they’ve shown that this approach keeps the spurious currents in check at a level of one thousand cubic lattice units to the sixth power, which is significantly lower than what popular pseudopotential models achieve. That's a tangible reduction in simulation error right there.

Conclusion: Kai: So wrapping up the paper "Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries," this work gives us a comprehensive framework for handling wetting conditions on curved solid boundaries in both static and dynamic situations with the lattice Boltzmann method.

Mira: Essentially, they’ve shown that by using their color gradient approach to modify ghost nodes, you can get better accuracy than other numerical methods while keeping errors relatively small when dealing with large density and viscosity contrasts.

Lev: I think what this paper really contributes is providing a robust framework that lets researchers study systems involving solid particles at liquid gas interfaces in a way that's scalable, especially since they’ve shown it can be extended to three dimensions.

Kai: It confirms the scheme handles the geometry and the physics well, and it agrees with analytical solutions for both static and dynamic contact lines on curved surfaces, which is a very strong indicator of correctness.

Mira: The key point is that their wetting prescription depends on that interface grid resolution parameter epsilon, and as epsilon goes to zero, it approaches Fakhari et al.'s prescription in the high resolution limit.

Lev: One final thought from my side is that if we could get this framework running reliably on real hardware with the guaranteed error control they've shown, it would be a huge step forward for applying these ideas where fidelity matters most.

Kai: So that’s what we got today with this paper, and it provides a solid foundation for anyone looking to simulate complex wetting phenomena on curved surfaces using lattice Boltzmann methods.

M. Bhattacharya, S. Dash, M. Sutar, J. Ravinder, M. Nimalan, A. Subhedar

Department of Chemical Engineering, Indian Institute of Technology Bombay · Department of Chemical Engineering, Indian Institute of Technology Delhi · Engineering for Research (e4r), Thoughtworks

physics.flu-dyn, physics.comp-ph

Submitted: 2026-05-29

Updated: 2026-05-29

Comments: Will be submitted to: Industrial & Engineering Chemistry Research

DOI: 10.1021/acs.iecr.6c02798

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 79/100

The gist: Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries introduces a wetting boundary condition for curved solid boundaries within a diffuse interface…

Key concepts

Color-gradient model
This is a simulation technique where the fluid's state is represented by colors that change smoothly across the domain. It uses an equilibrium color profile to describe how different fluid phases interact, allowing the model to capture complex flow behaviors in two-phase immiscible fluids.
Wetting boundary condition
This specific rule dictates how a liquid phase interacts with a curved solid surface. The scheme updates 'ghost node' colors based on the contact angle and a point nearby in the fluid, ensuring the simulation correctly models the wetting behavior on non-flat surfaces.
Spurious currents
These are unwanted, artificial flows or numerical errors that appear in simulations, especially near sharp interfaces. The proposed scheme keeps these currents very small (order of $10^{-6}$), meaning the results are physically accurate and not dominated by numerical noise.
Interface grid resolution ($\epsilon$)
This refers to how finely the simulation grid is divided near the fluid interface. The accuracy of the wetting scheme depends on this resolution; it approaches a known, simpler prescription in the limit where $\epsilon$ becomes very small.

Terminology

Summary

Color-gradient lattice Boltzmann modeling of wetting boundary condition on curved solid boundaries introduces a wetting boundary condition for curved solid boundaries within a diffuse interface framework for lattice Boltzmann method. p. 0

Where the paper says it

Subhedar1∗ 1Department of Chemical Engineering, Indian Institute of Technology Bombay, Maharashtra-400076, India 2Department of Chemical Engineering, Indian Institute of Technology Delhi, Delhi-110016, India and 3Engineering for Research (e4r), Thoughtworks, Pune, India (Dated: June 1, 2026) We introduce a wetting boundary condition for curved solid boundaries within a diffuse interface framework for lattice Boltzmann method. The boundary condition relies on updating the order parameter (color/phase-field) values on ghost nodes inside the solid phase.

Numerical simulations performed on an NVIDIA A100 GPU demonstrate that the wetting scheme retains the model’s ability to handle large density and viscosity contrasts while producing relatively small spurious currents. p. 0
Where the paper says it

The ghost node color modification rule, in turn, extends the equilibrium color profile into the solid phase. Numerical simulations performed on an NVIDIA A100 GPU demonstrate that the wetting scheme retains the model’s ability to handle large density and viscosity contrasts while producing relatively small spurious currents.

The present scheme agrees well with analytical solutions/other numerical works for both static and dynamic contact lines on curved solid boundaries. p. 0
Where the paper says it

Numerical simulations performed on an NVIDIA A100 GPU demonstrate that the wetting scheme retains the model’s ability to handle large density and viscosity contrasts while producing relatively small spurious currents. The present scheme agrees well with analytical solutions/other numerical works for both static and dynamic contact lines on curved solid boundaries.

Model Formulation

The color-gradient model is used to simulate incompressible immiscible Newtonian two-phase flow, governed by the equations of fluid dynamics and the phase-field evolution equation. The equilibrium color profile φeq(η) solution of Eq. p. 1

Where the paper says it

(2) is modeled as Fs = −σ 48 W φ(1 − φ) φ − 1 2 + 3W 2 ∇2φ ∇φ, (4) where σ is the liquid-gas surface tension. In the absence of fluid velocity u, equilibrium color profile φ eq, solution of Eq.

(3) is given by φeq(η) = 1/2 ± tanh 2η/W. p. 1
Where the paper says it

In the absence of fluid velocity u, equilibrium color profile φ eq, solution of Eq. (3), is given by φ eq(η) = 1 2 1 ± tanh 2η W , (5) where η is a local curvilinear coordinate that is normal to the liquid-gas interface, such that η = 0 identifies the sharp interface.

The wetting boundary condition is written as ˆns · ˆnφ = − cos θ. p. 1
Where the paper says it

∂φeq ∂η = 4 W φ eq(1 − φ eq). (6) The wetting boundary condition is written as ˆns · ˆnφ = − cos θ, (7) where ˆns is a unit normal to the solid boundary pointing away from the surface.

The boundary condition is validated by updating ghost node color-field φG at xG based upon the contact angle θ and color-field φF at a point xF in the fluid domain that is distance ∆x away from xG in the direction of normal to the solid surface ˆns. p. 2
Where the paper says it

The wetting boundary condition, Eq. (9), is enforced by updating ghost node color-field φG at xG based upon the contact angle θ and color-field φF at a point xF in the fluid domain that is distance ∆x away from xG in the direction of normal to the solid surface ˆns.

Scheme Comparison and Accuracy

The proposed wetting scheme, Eq. (9), is compared with Fakhari et al. p. 7

Where the paper says it

’s [20] scheme. At extreme contact angles 30° and 150°, the error in vertical displacement from the proposed scheme is ∼ 9%, compared to ∼ 14% from Fakhari et al.

’s [20] scheme, Eq. p. 0
Where the paper says it

The ghost node color modification rule, in turn, extends the equilibrium color profile into the solid phase. Numerical simulations performed on an NVIDIA A100 GPU demonstrate that the wetting scheme retains the model’s ability to handle large density and viscosity contrasts while producing relatively small spurious currents.

(10). The difference between these two series gives φG − φFakhari G = ε3/12 [1 − 3(1 − φF)φF] φF(1 − φF) + O(ε4). p. 3
Where the paper says it

(10)), can be expressed as a series in the small parameter ε for a given value of color-field φF in the fluid domain. The difference between these two series gives, φG − φ Fakhari G = ε 3 12 [1 − 3(1 − φF)φF] φF(1 − φF) + O(ε 4 ).

In the bulk fluid phases, where φF(1 − φF) = 0, two prescriptions are the same. p. 3
Where the paper says it

The difference between these two series gives, φG − φ Fakhari G = ε 3 12 [1 − 3(1 − φF)φF] φF(1 − φF) + O(ε 4 ). (11) Equation (11) suggests that in the bulk fluid phases φF(1 − φF) = 0, two prescriptions are the same.

The limit of ε = -4∆x/W cos θ → 0 corresponds to increasing interface grid resolution, and the two prescriptions should approach each other. p. 3
Where the paper says it

Further, differences in the two prescriptions appear at third order in ε. For a given contact angle θ, limit of ε = − 4∆x W cos θ → 0 corresponds to increasing interface grid resolution, and the two prescriptions should approach each other.

Numerical Benchmarks

The scheme is tested in several benchmarks, including a drop on a solid plate where the proposed wetting boundary scheme does consistently better over that of Fakhari et al [20] for the range of prescribed contact angles. For a stationary drop on a solid cylinder, the proposed scheme produces −0. p. 4

Where the paper says it

For the remaining benchmarks, interface width is kept constant as W = 4. Drop on solid cylinder Consider a stationary drop sitting on a solid cylinder of radius R under negligible gravity as shown in Fig.

35% and 0.26% errors, respectively, compared to −0. p. 4
Where the paper says it

’s [20] scheme makes −0. 7% errors in scaled height hmax/R respectively.

76% and 0.7% from Fakhari et al. p. 1
Where the paper says it

A color-gradient model [26, 27] that employs a velocity-based equilibrium distribution function is used to solve fluid dynamical equations for two immiscible fluid phases. Accuracy of the proposed scheme is assessed against analytical solutions and the scheme of Fakhari et al.

’s [20] scheme for contact angles of 30° and 150°. p. 4
Where the paper says it

We perform simulations for various contact angles in the range 30° − 150°, and compare hmax measured from simulations with the analytical solution. Figure (9) compares the height of the drop measured from the cylinder center obtained using the proposed wetting scheme against Fakhari et al.

For a fixed particle at the liquid-gas interface, the proposed wetting scheme produces less error compared to Fakhari et al. p. 7
Where the paper says it

’s [20] scheme. Thus, for all contact angles considered, the proposed wetting scheme produces lesser error compared to Fakhari et al.

’s [20] scheme across all contact angles considered. p. 7
Where the paper says it

’s [20] scheme. Thus, for all contact angles considered, the proposed wetting scheme produces lesser error compared to Fakhari et al.

Computational Performance

GPU simulations using JAX on NVIDIA A100 GPUs show that small grid sizes underutilize GPU resources, while increasing the grid size improves performance until saturation is reached. On average, single-precision FP32 computations are at least two times faster than the double precision FP64 ones for most benchmarks. p. 9

Where the paper says it

On average, single precision FP32 computations are at least two times faster than the double precision FP64 ones as shown in Table I. For the first three benchmarks, FP32 and FP64 produced same results.

The maximum magnitude of spurious currents for the proposed wetting scheme remains of the order of O(10−6) in lattice units, which is a couple of orders of magnitude lower than the popular Pseudopotential lattice Boltzmann models. p. 4
Where the paper says it

The maximum magnitude of the spurious currents for these two contact angles are 5. 74 × 10−7 and 7 × 10−7 , respectively for the proposed wetting scheme.

Conclusion

The wetting scheme accurately handles a wide range of contact angles on curved solid boundaries in both static and dynamic situations. The wetting prescription depends on the interface grid resolution ε, and it approaches Fakhari et al. p. 6

Where the paper says it

For a given contact angle θ, increasing W ∆x effectively decreases magnitude of the small parameter |ε| = 4∆x W | cos θ|. For a moderate grid resolution W ∆x = 4, the present wetting scheme performs better than the scheme of Fakhari et al.

’s prescription [20] in the high grid resolution limit (ε → 0). p. 3
Where the paper says it

Further, differences in the two prescriptions appear at third order in ε. For a given contact angle θ, limit of ε = − 4∆x W cos θ → 0 corresponds to increasing interface grid resolution, and the two prescriptions should approach each other.

The scheme can be directly extended to three dimensions. p. 1
Where the paper says it

We introduce a wetting boundary condition for curved solid boundaries in two dimensions. The scheme relies on adjusting the order parameter on ghost nodes such that the equilibrium order parameter profile is extended in the solid phase.

The proposed method provides a robust framework for simulating systems involving solid particles at liquid–gas interfaces. p. 4
Where the paper says it

Fixed particle at liquid-gas interface Particle arrangements at fluid-fluid interfaces are common in multi-phase suspensions [36]. In these systems, wettability controls both the capillary forces acting on the particles and the equilibrium shape of the fluid-fluid

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved wetting boundary condition for diffuse interfaces allows lattice Boltzmann methods to accurately model wetting phenomena on curved solid boundaries by updating ghost node color-field values to extend equilibrium profiles into the solid phase. This enables reliable simulation of static and dynamic contact lines on curved surfaces while maintaining accuracy against analytical solutions.

  2. The system can now accurately handle large density and viscosity contrasts in fluid dynamics simulations due to the scheme's ability to retain model fidelity while producing relatively small spurious currents near interfaces, which is crucial for physical accuracy.

  3. The AI system can perform high-performance numerical simulations on NVIDIA A100 GPUs using a JAX-based framework, achieving high performance metrics like million lattice updates per second (MLUPS) with optimized kernels.

  4. The model can be extended to three dimensions, allowing the simulation of more complex geometries beyond two dimensions, which is necessary for simulating realistic systems involving solid particles at liquid-gas interfaces.

  5. The proposed method provides a robust framework for simulating systems involving solid particles at liquid–gas interfaces, enabling the study of phenomena such as drop impact on cylinders and fixed particle arrangements under various contact angles.

Abstract

We introduce a wetting boundary condition for curved solid boundaries within a diffuse interface framework for lattice Boltzmann method. The boundary condition relies on updating the order parameter (color/phase-field) values on ghost nodes inside the solid phase. The ghost node color modification rule, in turn, extends the equilibrium color profile into the solid phase. Numerical simulations performed on an NVIDIA A100 GPU demonstrate that the wetting scheme retains the model's ability to handle large density and viscosity contrasts while producing relatively small spurious currents. The present scheme agrees well with analytical solutions/other numerical works for both static and dynamic contact lines on curved solid boundaries.

Related papers