Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications

arXiv:2611.16873 · physics.flu-dyn · Submitted 2026-11-12 · 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: "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications".

Mira: Here's the gist of the paper. Pseudopotential models are well established for multiphase and multicomponent simulations using the lattice Boltzmann method, but they suffer from spurious currents at interfaces and solid walls which introduce numerical artifacts.

Kai: An improved lattice Boltzmann pseudopotential model is presented that reduces spurious currents by computing discrete fluid density gradients with rotationally symmetric gradient operators formulated as a convolution.

Mira: Model Formulation and Discretization The paper interprets the pseudopotential method as a regular spatial gradient on discrete density fields, rewriting the theory in the form of discrete convolutions for convenience and comparability.

Kai: The macroscopic velocity of component kappa is modified by an interaction force term proportional to the gradient of other components' density fields.

Title and authors: Kai: We started with the title of "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications". It sounds very technical, but essentially they are taking an established tool and trying to clean up its flaws in simulations involving porous media.

Mira: Right. The title immediately tells you the focus is on making the lattice Boltzmann method work better when you’re simulating things like rocks or soil where interfaces are everywhere. They are addressing the known issue of spurious currents that plague these multiphase simulations, which we talked about earlier.

Lev: So, what they're really saying is that existing pseudopotential models aren't robust enough for real-world engineering problems where you need accuracy at those boundaries. If the results are noisy, you can’t trust them for anything practical.

Kai: They’re proposing a specific fix: they are rewriting the theory as a regular spatial gradient on discrete density fields and using discrete convolutions to make the math easier to compare with other methods. This is how they introduce an interaction force term related to other components' density fields.

Mira: And then, they add something else by introducing a force term based on the gradient of a solid indicator field specifically for adhesion at those fluid-solid interfaces, which was previously missing from the models.

Lev: That makes sense because standard boundary conditions can be very delicate when you start adding these complex interaction forces in there, so this explicit adhesion force is a practical way to manage that complexity.

Kai: Moving on to the actual improvement they highlight, they focus on making sure the gradient operator itself is rotationally symmetric, which is what fixes those spurious currents when solid walls aren't aligned with the lattice directions.

Mira: They achieve this by using a convolution scheme where the force is computed as a sum over spatial directions using specific kernels. They even derive something called the Scharr operator, which they claim to be maximally rotationally symmetric by minimizing error in Fourier space for the twenty-two point five degree case.

Lev: Minimizing error in Fourier space sounds mathematically heavy, but if it actually fixes that rotational asymmetry issue that causes the artifacts, then for real hardware testing, that could translate into a much more reliable simulation setup for complex geometries.

Kai: They tested this new model using droplets on walls as a common test case and also used analytically solvable capillary rise experiments described by Jurin’s law and porous media imbibition as practical applications to see how it behaves in the real world.

Mira: The results from those tests show that the Scharr operator actually performs better than the classical LBM gradient operator when you look at various ways of solid interaction, and for surface tension measurements in a two-component system, they yielded comparable results to the old method.

Lev: Comparable surface tension is fine, but for me it matters if the underlying physics is captured correctly. The paper shows that this rotationally symmetric approach gives better results for simulating porous media applications compared to the classical approach, and that’s where I want to see gains in terms of predictive accuracy for physical systems.

Kai: So to wrap up, this paper on "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications" gives us a better tool to handle those interface artifacts by using a rotationally symmetric gradient operator, which they derived by minimizing error in Fourier space.

The paper's summary: Kai: Now that we’ve covered the title and authors, let’s look at what the actual paper is summarizing in "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications". It boils down to taking an older technique and fixing those numerical errors that creep in at interfaces.

Mira: Right. They are pointing out that those spurious currents are a major problem, which just creates noise when you’re trying to model real fluid flow around a wall or through porous rock. The paper essentially says the key is using rotationally symmetric operators to handle the density gradients better.

Lev: So, what they're saying is that this method should give you results that are more trustworthy for anything you try to build physically, not just neat simulations on a screen. If we get those errors, the data isn't useful for engineering applications.

Kai: They lay out the logic: they interpret the pseudopotential method as a regular spatial gradient on density fields and rewrite it using discrete convolutions to make it easier to compare with other methods. Then they add an extra piece by introducing a force term based on the gradient of a solid indicator field specifically for adhesion at those fluid-solid interfaces.

Mira: That force term makes sense because standard boundary conditions can be very tricky when you start adding these complex interaction forces in there, so this explicit adhesion force is a practical way to manage that complexity.

Lev: It’s good to see them address the need for an explicit adhesion force because it shows they aren't just patching things up without understanding how those forces actually interact with the fluid dynamics.

Kai: The real improvement they highlight is optimizing that gradient operator itself, because the classical one based on lattice weights isn't rotationally symmetric, which causes those errors when the solid walls aren't lined up perfectly with the lattice directions.

Mira: They introduce this convolution-based scheme where they compute the force as a sum over spatial directions using specific gradient kernels. This allows them to move towards something more physically accurate.

Kai: They benchmarked this new model using droplets on walls as a common test case. And they also used analytically solvable capillary rise experiments and porous media imbibition as practical applications to see how it behaves in the real world.

Mira: The results from those tests show that the Scharr operator, which is derived by minimizing error in Fourier space for the twenty-two point five degree case, actually performs better than the classical LBM gradient operator when you look at various ways of solid interaction.

Lev: Comparable surface tension measurements are fine, but for me it matters if the underlying physics is captured correctly. The paper shows that this rotationally symmetric approach gives better results for simulating porous media applications compared to the classical approach, and that’s where I want to see gains in terms of predictive accuracy for physical systems.

Kai: They also mentioned a way you can simplify the solid interaction by only introducing it through one fluid component. This helps avoid creating density gradients for the main fluid right at those walls, which is useful when studying just one wetting phase within a two-component system.

Kai: So to wrap up, this method allows us to model wetting behavior more accurately with a rotationally symmetric gradient operator without adding extra computational costs.

The paper's improvements: Kai: We’ve covered the title and authors and the summary, but now let's get into the specific improvements they suggest in "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications". They are focusing on how they actually fixed those numerical issues.

Mira: Right. They are pointing out that a big improvement is using a rotationally symmetric gradient operator instead of the classical one, which directly addresses the spurious currents when solid walls aren't aligned with the lattice directions.

Lev: That means if you’re running simulations on hardware, you can get better fidelity right at those interfaces without needing to worry about alignment issues causing numerical noise in your results.

Kai: They also detail how they compute this new force using a convolution-based scheme, which is a sum over spatial directions using gradient kernels. This allows them to move towards something more physically accurate by factoring the computation into lower dimensional convolutions.

Mira: And they derived the Scharr operator, which is described as maximally rotationally symmetric by minimizing error in Fourier space for the twenty-two point five degree case. This operator is supposed to be much better at capturing that required symmetry than what’s available classically.

Kai: They benchmarked this new model using droplets on walls as a common test case and also used analytically solvable capillary rise experiments described by Jurin’s law and porous media imbibition as practical applications to see how it behaves in the real world.

Kai: So to wrap up, this method allows us to model wetting behavior more accurately just by introducing this rotationally symmetric gradient operator without adding extra computational costs to the simulation.

Mira: That’s a big deal because it means we get better physics with less overhead than what we might expect from a full rewrite of the method.

Lev: And they did flag a limitation, though, which is that they need to look into whether the difference between these two operator sets stays constant over time when the fluid has enough time to spread out.

Kai: That’s fair; if those differences are transient and settle down quickly, then this is a viable path forward for using this improved lattice Boltzmann pseudopotential method.

Mira: If they change over time, then we need a much more complex simulation scheme than just optimizing a single gradient operator.

Conclusion: Kai: So we've looked at how they fixed those spurious currents in "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications". They essentially showed that using a rotationally symmetric gradient operator, like the Scharr operator, makes modeling wetting behavior much more accurate without adding extra computational cost.

Mira: That’s the big picture, right? It means we can get better physics for things like droplet behavior on slanted walls or in porous media simulations with a method that is already fairly efficient.

Lev: I just want to reiterate that this hinges on those assumptions about the operator staying consistent over time when the fluid spreads out. If it drifts, then you’re back to square one, and that could mean a lot of extra work for real hardware testing.

Kai: Exactly. So, while the results look strong in benchmarks like Jurin’s law tests and droplet experiments, we still need those long-term stability checks before we can fully trust it for heavy engineering work.

Mira: It’s a solid improvement because it gives us a way to model wetting behavior more accurately with a rotationally symmetric gradient operator without adding extra computational costs.

Lev: So, the implication is that we have a more robust boundary condition tool here, which is exactly what we need when building reliable simulators for complex geometries.

Kai: That’s right. The paper shows that the choice of solid interaction model really depends on the specific simulation setup you are running.

Mira: And in radially symmetric pressure boundary conditions, the Scharr operator consistently outperformed that classical gradient operator, which is a pretty strong piece of evidence for its superiority in those specific setups.

Lev: For us, it means we can start using this refined approach for our core fluid dynamics calculations without having to spend all our time debugging numerical artifacts caused by wall alignment issues.

Kai: So, to wrap up on "Rotationally Symmetric Lattice Boltzmann Pseudopotential Methods for Porous Media Applications", the main point is that you can model wetting behavior more accurately with a rotationally symmetric gradient operator without adding extra computational costs.

Mira: And while it’s better, the authors also cautioned that whether these two operator sets stay different over time when the fluid has enough time to spread out needs further investigation before we can fully trust it for long-term simulations.

Lev: If they change over time, then we need a much more complex simulation scheme than just optimizing a single gradient operator, which is something I’m always thinking about for real hardware constraints.

German Aerospace Center (DLR) · Helmholtz Institute Ulm Electrochemical Energy Storage (HIU) · Ulm University

physics.flu-dyn

Submitted: 2026-11-12

Updated: 2026-11-12

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 79/100

The gist: Pseudopotential models are well established for multiphase and multicomponent simulations using the lattice Boltzmann method…

Key concepts

Spurious Currents
These are numerical artifacts that appear in simulations, especially near interfaces like solid walls. They occur because the classical gradient operators used in the model are not rotationally symmetric, causing incorrect fluid flow patterns and inaccurate predictions.
Convolution-Based Gradient Scheme
Instead of using simple lattice weights for gradients, this method computes forces by summing over spatial directions using special kernels (gradient kernels). This convolution approach allows for a more flexible and accurate calculation of density gradients, leading to better symmetry in the model.
Scharr Operator
This is a specific, maximally rotationally symmetric gradient operator derived by minimizing errors when simulating a 22.5-degree case in Fourier space. It is superior to classical lattice weight operators because it ensures that fluid interactions are handled consistently regardless of the orientation of the solid boundary.

Terminology

Summary

Pseudopotential models are well established for multiphase and multicomponent simulations using the lattice Boltzmann method, but they suffer from spurious currents at interfaces and solid walls which introduce numerical artifacts. p. 0 An improved lattice Boltzmann pseudopotential model is presented that reduces spurious currents by computing discrete fluid density gradients with rotationally symmetric gradient operators formulated as a convolution. p. 0

Model Formulation and Discretization

The paper interprets the pseudopotential method as a regular spatial gradient on discrete density fields, rewriting the theory in the form of discrete convolutions for convenience and comparability. The macroscopic velocity of component kappa is modified by an interaction force term proportional to the gradient of other components' density fields. To account for adhesion at fluid-solid interfaces, a force term based on the gradient of a solid indicator field is introduced. p. 3

Gradient Operator Optimization

The classical gradient operator based on lattice weights is not rotationally symmetric and leads to spurious currents when solid walls are unaligned with the lattice directions. The authors introduce a convolution-based gradient scheme where the force is computed as a sum over spatial directions using gradient kernels, which can be factored into lower-dimensional convolutions. p. 6 A maximally rotationally symmetric gradient operator, referred to as the Scharr operator, is derived by minimizing the error of the 22. p. 75° case in Fourier space. p. 7

Benchmarking and Results

The model variants are benchmarked using droplets on walls as a common test, analytically solvable capillary rise experiments (Jurin’s law), and porous media imbibition as a practical application. Contact angle measurements show that the Scharr operator performs better than the classical LBM gradient operator based on lattice weights across various solid interaction variants. p. 9 Surface tension measurements using both operators yield comparable results for surface tension in a two-component system. p. 9

Symmetry and Application Insights

The comparison of simulation results in a porous medium application shows that the gradient operator optimized for rotational symmetry performs better than the classical approach. The choice of solid interaction model depends on the simulation setup, with introducing solid interaction only through one fluid potentially avoiding density gradients at walls for the main fluid. p. 0 Future work will involve applying the 3D implementation to more realistic inhomogeneous and anisotropic porous media. p. 0

Outlook

The study demonstrates that wetting behavior can be modeled more accurately by introducing a rotationally symmetric gradient operator without additional computational costs. The choice of the solid interaction model is problem-dependent, and the Scharr operator consistently outperformed the classical gradient operator in radially symmetric pressure boundary conditions. p. 0 Future research will investigate whether deviations between the two operator sets remain constant over time in cases where fluid has enough time to spread.

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. Rotationally symmetric gradient operators can be used to compute interaction forces, which reduces spurious currents in simulations where solid walls are not aligned with lattice directions. This allows for more accurate modeling of fluid-solid interactions compared to classical lattice weights approaches.

  2. The model can incorporate solid interaction only through one fluid component to avoid numerical artifacts like density gradients at walls for the main fluid. This delegation of solid interaction can be used when studying the behavior of a specific fluid, such as in Jurin’s law tests.

  3. Using Scharr operators, which are derived by minimizing the error of the 22.5 degree case, provides a more accurate representation of rotationally symmetric gradient operators than classical lattice weights. This leads to better results in simulations involving droplets on slanted walls and improved surface tension measurements for capillary rise experiments.

  4. The use of interaction through fluid B only avoids creating gradients for fluid A on fluid-solid interfaces, which can be beneficial when studying the behavior of a single wetting phase within a two-component system. This prevents numerical artifacts that would otherwise modify contact angles or throat diameters in porous media invasions.

Related papers