A Finite Analytic Method for Two-Dimensional Power-Law Fluid Flow in Heterogeneous Porous Media Using Unstructured Grids

Deep underground, in the complex network of rocks that hold oil, gas, and water, fluids do not always move in simple, predictable lines. The rock itself is rarely uniform; it is a patchwork of materials with vastly different abilities to let fluid pass through. Some layers are like open highways, while others are like dense walls. When engineers try to model how these fluids flow, they must account for these sudden changes in the rock's properties. Furthermore, the fluids themselves can be tricky. While water flows easily, many industrial fluids—such as the slurries used in paper manufacturing or the drilling muds used in oil wells—behave differently. They are "non-Newtonian," meaning their thickness and resistance to flow change depending on how hard they are pushed. When these complex fluids encounter a rock that changes its character abruptly, the math describing their movement becomes incredibly difficult, often breaking down at the very points where the rock properties shift.

For decades, scientists have relied on a standard way of calculating these flows, a method that essentially averages the rock's properties between two points. This approach works well enough when the rock is relatively consistent, but it fails dramatically when the rock is highly varied. In those cases, the standard method consistently underestimates how much fluid can actually move, leading to errors that can be massive. To solve this, researchers at the University of Science and Technology of China have developed a new computational technique. Instead of relying on simple averages, this new method uses a precise mathematical description of how fluid behaves right at the edge where two different rock types meet. By capturing the true, sharp behavior of the fluid at these critical boundaries, the new method provides a much clearer and more accurate picture of underground flow, even when the rock is extremely uneven and the fluid is complex.

The researchers focused on a specific type of fluid known as a power-law fluid. These are fluids where the relationship between the force pushing them and the speed they move follows a specific, non-linear rule. In the real world, this covers a wide range of substances, from thick polymer solutions to certain types of crude oil. The challenge arises when these fluids flow through porous media—rocks full of tiny holes—where the permeability, or the ease of flow, changes suddenly from one spot to the next. In such environments, the pressure of the fluid does not change smoothly; instead, it creates a sharp spike or a "singularity" right at the boundary between the easy-flowing and hard-flowing rock. Traditional computer models, which divide the rock into a grid of small triangles, struggle with these spikes. They assume the pressure changes in a straight line between points, which is physically incorrect at these boundaries. As a result, the models produce flow rates that are far too low, and they require an enormous number of tiny grid points to get even a rough approximation of the truth.

To fix this, the team developed a Finite Analytic Method. The core idea is to stop guessing how the fluid behaves at the boundary and instead use a known, exact mathematical description of that behavior. The researchers realized that near the point where the rock properties change, the fluid pressure follows a specific curved pattern, rising or falling in a way that depends on the power-law nature of the fluid. They calculated the exact shape of this curve, known as an eigenfunction, for the most critical part of the flow. This calculation involved solving a complex set of equations that describe how the fluid moves in a wedge-shaped area where the rock properties differ. By finding the smallest and most significant solution to these equations, they captured the dominant way the fluid behaves right at the trouble spot.

Once this local behavior was understood, the researchers built a new numerical scheme around it. Instead of using the old, flawed averaging method to connect the grid points, they used their new analytical solution to calculate the flow across the boundaries of each triangular cell in their grid. This allowed the computer to "know" exactly how the fluid should behave as it squeezed through the narrow neck between high and low permeability zones. The result is a simulation that respects the physics of the problem at the most difficult points, rather than smoothing them over with an approximation.

The team tested this new method against several challenging scenarios to see how well it performed. They began with a classic "checkerboard" pattern, where high-permeability and low-permeability rocks were arranged in alternating squares. In this setup, the rock properties changed by a factor of one hundred. When they compared their new method to the traditional one, the difference was stark. The traditional method underestimated the flow by huge margins, sometimes missing the true value by nearly ninety percent on a coarse grid. In contrast, the new method stayed remarkably close to the correct answer, even when the grid was not very fine. As they made the grid finer, the new method converged to the correct answer very quickly, while the traditional method struggled to catch up, even with extremely fine grids.

They also tested the method on a regular hexagonal domain and on a domain with a random, log-normal distribution of permeability, which mimics the chaotic nature of real underground reservoirs. In every case, the new method proved superior. It handled the extreme variations in rock properties with ease and remained accurate regardless of the specific type of power-law fluid being simulated. One of the most striking findings was that the accuracy of the new method did not depend heavily on the specific properties of the fluid or the rock. Whether the fluid was very thick or very thin, or whether the rock contrast was mild or extreme, the method delivered consistent results. The researchers found that a moderate level of grid refinement was sufficient to achieve high accuracy, with errors dropping to less than eight percent even in the worst-case scenarios.

This work represents a significant step forward in how we simulate fluid flow in complex underground environments. By replacing a flawed averaging technique with a solution that respects the true physics of the boundary, the researchers have created a tool that is both more accurate and more efficient. For industries that rely on understanding how fluids move through the earth, from energy production to environmental management, this means simulations can be trusted more and run faster. The method does not require the massive computational power that was previously needed to get a decent answer, making high-fidelity modeling of complex, heterogeneous systems much more accessible. The findings suggest that by looking closely at the singular points where the world changes, we can build a better understanding of the whole.