Flow of yield stress fluid in a percolating network

✨

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 push a thick, sticky substance—like honey mixed with toothpaste (a "yield-stress fluid")—through a sponge. But this isn't just any sponge; it's a sponge where some of the holes are completely blocked by air bubbles or debris.

This paper is a scientific study of exactly that scenario. The researchers wanted to understand how this thick fluid moves through a messy, partially blocked network of tunnels. They used a computer model to simulate this, treating the fluid like a stubborn traveler who refuses to move unless pushed hard enough, and the blocked holes as dead ends.

Here is the breakdown of their findings using simple analogies:

1. The Setup: The "Traffic Jam" Sponge

Think of the porous medium (the sponge) as a giant city grid with thousands of streets.

The Fluid: Imagine a heavy truck that won't start moving unless the engine is revved past a certain speed (the "yield stress").
The Blockages: Now, imagine that the widest, easiest highways in the city are suddenly closed off by construction barriers (the blocked throats).
The Goal: The truck needs to get from the West side of the city to the East side.

2. The Two Big Scenarios

The researchers discovered that the behavior of the fluid depends entirely on how many roads are open. They found two distinct "worlds":

World A: The "Connected City" (Above the Threshold)

If enough roads are open (even if the widest ones are blocked), the city still has a connected network.

What happens: The fluid finds a way. It might take a slightly longer, winding route, but it flows.
The Result: The flow is predictable. If you push harder, more fluid moves. The "traffic" behaves like a standard fluid, just with a bit more resistance. The specific details of which tiny road is open don't matter much; the overall system averages out to a smooth, predictable flow.

World B: The "Critical Bottleneck" (At the Threshold)

This is the most exciting part of the paper. This happens right at the tipping point where the city is just barely connected. If you block one more road, the city splits in two, and the truck can't get through at all.

What happens: The fluid is forced to squeeze through a very specific, fragile, and winding "backbone" of the network. It's like the truck is forced to drive only on a single, narrow, winding mountain pass while all other roads are closed.
The Result: The flow becomes chaotic and unpredictable.
- No Averages: In the "Connected City," if you ran the simulation 1,000 times, you'd get roughly the same answer every time. In the "Critical Bottleneck," every single time you run the simulation, the result is totally different. One time the truck gets stuck; the next time it finds a lucky shortcut. You cannot predict the flow just by knowing the average size of the holes; you have to know the exact map of that specific moment.
- Fractal Geometry: The path the fluid takes isn't a straight line; it's a "fractal" path—a shape that is infinitely twisty and turns back on itself, much longer than the straight-line distance.

3. The "Goldilocks" Pressure

The fluid needs a specific amount of pressure to start moving.

Too little pressure: The fluid sits still (like the truck with the engine off).
Just enough pressure: The fluid finds the "path of least resistance." In the critical bottleneck scenario, this path is so twisty and long that the pressure needed to push the fluid through is surprisingly high and varies wildly from one simulation to the next.
Huge pressure: Once you push hard enough, every open road in the network starts carrying traffic. The system behaves like a normal fluid again, and the weird, chaotic behavior disappears.

4. Why Does This Matter?

You might wonder, "Who cares about toothpaste in a sponge?"
Actually, this applies to many real-world problems:

Oil Recovery: Heavy oil in the ground often acts like this sticky fluid. If the oil is trapped in rock with some pores blocked, understanding these "critical bottlenecks" helps engineers figure out how much pressure to pump to get the oil out.
Cleaning Pollution: If you inject foam or grout into soil to stop pollution from spreading, you need to know if it will flow through the soil or get stuck.
3D Printing & Construction: Understanding how thick pastes flow through complex structures helps in designing better materials.

The Big Takeaway

The paper teaches us that disorder creates surprise.
When a system is well-connected, it behaves predictably. But when a system is on the edge of falling apart (the percolation threshold), the specific arrangement of the holes matters more than the holes themselves. The fluid stops following the "average" rules and starts following the "weird, twisty" rules of the specific path it finds.

In short: If you are trying to push a thick fluid through a blocked maze, the closer you are to the maze being completely blocked, the more unpredictable and twisty the journey becomes.

1. Problem Statement

The study investigates the flow dynamics of a Bingham yield-stress fluid through an unsaturated porous medium. Unlike previous studies focusing on saturated media, this work considers a scenario where a non-wetting phase (e.g., trapped air bubbles) occupies the largest pore throats, effectively blocking them.

The Challenge: The fluid can only flow through the percolating cluster formed by the remaining unblocked throats.
The Complexity: The flow is governed by two competing factors:
1. Rheology: The fluid requires a critical pressure drop ( $\Delta P_0$ ) to overcome the yield stress before flowing.
2. Geometry: The network is disordered, and the blocking of throats creates a percolation problem where flow paths are tortuous and limited, especially near the percolation threshold ( $p_c$ ).
Key Question: How do the critical pressure threshold, permeability, and flow scaling behave as the fraction of blocked throats approaches the percolation threshold, and how does this differ from the behavior in fully connected (saturated) networks?

2. Methodology

The authors employ a 2D pore network model to simulate the system:

Network Structure: A diamond lattice of size $L \times W$ (where $W=L$ ). Nodes represent pores, and links represent cylindrical throats with radii ( $r_{ij}$ ) drawn from a uniform distribution.
Blocking Mechanism: A fraction $1-p$ of the largest radii throats are blocked (removed), simulating trapped non-wetting fluid. The system is analyzed as a function of the open fraction $p$ .
Governing Equations:
- Mass Conservation: Kirchhoff's law is applied at each node ( $\sum q_{ij} = 0$ ).
- Constitutive Law: A piecewise-linear approximation of the Bingham fluid flow is used for each link. Flow occurs only if the local pressure drop exceeds a local yield threshold $\tau_{ij} = 2\tau_c l_{ij}/r_{ij}$ .
- Generalized Darcy's Law: The total flow rate $Q$ is related to the pressure drop $\Delta P$ via an effective permeability $\kappa_{eff}$ and an apparent critical pressure $\Delta P^*$ .
Numerical Approach:
- Percolation: The Newman-Ziff algorithm determines the percolation threshold $p_c$ and the spanning cluster for each realization.
- Flow Solution: The Augmented Lagrangian method is used to solve the coupled non-linear equations for mass conservation and rheology.
- Threshold Calculation: Dijkstra's algorithm (undirected) is used to find the "critical path" (the path of least resistance) to determine the onset pressure $\Delta P_0$ .
System Parameters: Simulations range from $L=32$ to $L=1024$ , with extensive averaging over 10,000 disorder realizations.

3. Key Contributions

Non-Directed Flow Analysis: Unlike previous "directed" models that assume flow follows a specific gradient, this study uses a non-directed network model. This is crucial for high non-wetting saturation where flow paths become highly tortuous and cannot be pre-assigned a direction.
Identification of Two Distinct Regimes: The paper rigorously distinguishes between the behavior above the percolation threshold ( $p > p_c$ ) and at the threshold ( $p = p_c$ ).
Loss of Self-Averaging: A major finding is that at the percolation threshold, macroscopic quantities (permeability, critical pressure) become non-self-averaging. Their standard deviations scale identically to their means, meaning they do not converge to a deterministic value even in large systems.
Scaling Laws: The authors derive specific scaling exponents for the critical pressure and permeability in both the low-flow and high-flow limits, linking them to percolation theory exponents (fractal dimensions, conductivity exponents).

4. Key Results

A. Low Flow Rate Regime ( $Q \to 0$ )

Above Threshold ( $p > p_c$ ):
- The critical pressure $\Delta P_0$ scales linearly with system size $L$ ( $\Delta P_0 \sim L$ ).
- Fluctuations scale as $L^{1/3}$ , consistent with the KPZ universality class (directed polymers in random media).
- The system is self-averaging.
At Threshold ( $p = p_c$ ):
- The critical pressure becomes super-extensive: $\Delta P_0 \sim L^{D_{min}}$ , where $D_{min} \approx 1.13$ is the fractal dimension of the chemical path (shortest path on the backbone).
- The flow is constrained to the chemical path of the percolation backbone.
- Non-self-averaging: Both the mean and the standard deviation of $\Delta P_0$ scale as $L^{1.13}$ .
- The effective permeability $\kappa_0$ scales as $L^{-D_{min}}$ .

B. High Flow Rate Regime ( $Q \to \infty$ )

Above Threshold ( $p > p_c$ ):
- The system behaves like a Newtonian fluid with a finite, size-independent permeability.
- The apparent critical pressure offset $\Delta P^*$ scales linearly with $L$ .
At Threshold ( $p = p_c$ ):
- The effective permeability $\kappa_\infty$ decays to zero as $L^{-t/\nu}$ (where $t \approx 1.31$ is the conductivity exponent and $\nu = 4/3$ ).
- The apparent critical pressure offset $\Delta P_\infty$ scales as $L^{D_\tau}$ , where $D_\tau \approx 1.2$ relates to the tortuosity of streamlines.
- Non-self-averaging: Similar to the low-flow case, fluctuations scale identically to the mean, preventing convergence to a deterministic value.

C. Intermediate Regime and Effective Medium

Crossover: The transition between the critical regime and the bulk regime is described by scaling functions involving the distance from the threshold $\delta p = p - p_c$ .
Effective Homogeneous Medium: The authors test an approximation where the disordered radii are replaced by their average values.
- Result: This approximation fails for $p > p_c$ but becomes valid at $p = p_c$ for large systems. This implies that at the critical point, the flow is governed almost entirely by the geometry of the percolation backbone, rendering the specific disorder of the radii irrelevant.
Quadratic Regime Vanishing: In fully connected networks ( $p=1$ ), there is an intermediate regime where $Q \sim (\Delta P - \Delta P_0)^2$ . The paper shows this quadratic regime vanishes as $p \to p_c$ , leaving a purely linear relationship over the entire flow curve due to the scarcity of available paths.

5. Significance and Implications

Theoretical Advancement: The work bridges yield-stress rheology with percolation theory and statistical physics (KPZ universality, depinning). It establishes that near the percolation threshold, the flow is dominated by the fractal geometry of the backbone rather than the specific distribution of pore sizes.
Industrial Relevance: The findings are critical for applications involving unsaturated porous media, such as:
- Grouting: Injecting stabilizing fluids into soils where air pockets block large pores.
- Environmental Remediation: Using foams to displace pollutants in contaminated ground.
- Oil Recovery: Understanding the flow of heavy oils (which exhibit yield stress) in reservoirs with trapped gas.
Predictive Power: The identification of non-self-averaging behavior suggests that for systems near the percolation threshold, a single "representative elementary volume" (REV) may not exist. Instead, the flow properties are inherently stochastic and depend on the specific realization of the network geometry.

In summary, the paper provides a comprehensive framework for understanding how structural disorder (percolation) and material non-linearity (yield stress) interact, revealing that at the critical point, the system's behavior is dictated by universal geometric scaling laws rather than local material properties.