Immiscible two-phase flow in porous media: a statistical mechanics approach

This paper presents a review of a statistical mechanics approach, based on Jaynes' information-theoretical entropy, which provides a manageable continuum-scale description of immiscible two-phase flow in porous media by deriving thermodynamics-like relations and identifying new emergent variables such as co-moving velocity.

The Gist

Here is an explanation of the paper, translated from complex physics jargon into everyday language using analogies.

The Big Picture: The "Scale-Up" Problem

Imagine you are looking at a sponge. If you zoom in with a microscope, you see a chaotic maze of tiny tunnels (pores) filled with water and oil. The oil and water are fighting for space, pushing against each other, and getting stuck on the rough walls of the tunnels. This is the pore scale.

Now, zoom out to look at the whole sponge. You can't see the individual tunnels anymore. It just looks like a solid, wet block. This is the Darcy scale (the scale engineers use to design oil wells or water filters).

The Problem: How do we describe the messy, chaotic fighting of the oil and water in the tiny tunnels using simple rules for the whole block?

• The Old Way: Scientists have used "Relative Permeability Theory" for 90 years. It's like a rulebook that says, "If the sponge is 50% wet, the water flows at speed X." It works okay, but it's a bit of a black box. It doesn't explain why the flow behaves that way, and it gets very messy when things get complicated.
• The New Way: This paper proposes a new method. It treats the flow of fluids through a sponge like a thermodynamic system (like steam or gas), even though it's not hot and there are no molecules bouncing around in the traditional sense. They use a mathematical tool called Information Theory to build a new "physics of flow."

---

The Core Idea: "Agitation Temperature" (Agiture)

In normal thermodynamics (like a pot of boiling water), Temperature measures how much the molecules are jiggling. Hotter = more jiggling.

In this paper, the authors introduce a new concept called Agiture (short for "Agitation Temperature").

• The Analogy: Imagine a busy highway.
  • Normal Temperature: How fast the cars are driving.
  • Agiture: How chaotic the traffic is. If the cars are just cruising smoothly, the "agiture" is low. If the cars are swerving, merging, braking, and changing lanes constantly, the "agiture" is high.
• In the sponge, the "cars" are the blobs of oil and water. The "agitation" comes from the pressure pushing them through the tiny holes. The higher the pressure pushing the fluids, the more the fluid blobs are jiggling and rearranging themselves. This "jiggle" is the Agiture.

The "Glassy" Traffic Jam

The paper discovers that the flow of fluids in a sponge isn't always smooth. It has different "phases," similar to how water can be ice, liquid, or steam.

1. The Frozen Phase (Phase Ia): The pressure is too low. The oil and water are stuck. They are like cars in a parking lot that won't move.
2. The Glassy Phase (Phase Ib): This is the most interesting part. The fluids are moving, but they are jammed.
   • The Analogy: Think of a crowded subway station during rush hour. People are moving, but they are constantly bumping into each other, stopping, starting, and shifting. It's a "glassy" state—fluid but disordered.
   • The authors found that when the flow enters this "glassy" state, it behaves strangely. It creates hysteresis (history matters). If you push the fluids one way and then stop, the flow doesn't go back to exactly where it started. It's like a memory effect.
3. The Smooth Phase (Phase III): The pressure is very high. The fluids are moving so fast that they stop jamming and flow smoothly, like water in a clear pipe.

The Secret Ingredient: The "Co-moving Velocity"

This is the paper's biggest discovery. In the old theories, scientists assumed that if you know how fast the average fluid is moving, you know everything.

The authors found that there is a hidden variable called the Co-moving Velocity.

• The Analogy: Imagine a river with a strong current (the average flow).
  • Some leaves (water) are floating right in the middle, moving fast.
  • Some leaves (oil) are stuck near the muddy banks, moving slow.
  • The Old View: "The river moves at 5 mph."
  • The New View: "The river moves at 5 mph, BUT the leaves near the bank are actually drifting backward relative to the center stream, while the center leaves are drifting forward."
• The Co-moving Velocity measures how much the two fluids are "slipping" past each other or swapping places. It turns out this "slip" is crucial for predicting how the fluids actually behave. Without measuring this "slip," your predictions will be wrong.

Why This Matters (The "Thermodynamics" Connection)

The authors used a brilliant trick from the 1950s by a man named Edwin Jaynes. Jaynes said: "You don't need heat to do thermodynamics. You just need information."

• The Metaphor: Imagine you are a detective trying to guess what a suspect did.
  • If you know nothing, you guess randomly (Maximum Ignorance/Entropy).
  • If you know the suspect was at the bank at 2 PM, you narrow down your guess.
  • The authors applied this to the sponge. They said, "We don't know exactly where every drop of oil is, but we know the average flow and the average pressure."
  • By maximizing their "ignorance" (using math to find the most likely arrangement of fluids given what they know), they derived a set of rules that look exactly like the laws of thermodynamics, but for fluids in a sponge.

The "Aha!" Moment: It Works for Mixtures Too

The paper ends with a fun twist. They found that this "Co-moving Velocity" concept isn't just for oil and water in rocks. It actually exists in ordinary chemistry too!

• When you mix water and alcohol, the total volume shrinks. The authors showed that the math describing how the fluids "slip" past each other in the sponge is mathematically identical to how water and alcohol molecules "slip" past each other in a glass.
• This means their new theory isn't just a weird trick for sponges; it's a fundamental law of how mixtures behave.

Summary

This paper is a bridge. It connects the messy, chaotic world of tiny fluid blobs fighting in a rock to the clean, orderly world of thermodynamics.

1. Old Way: Guess the flow based on simple rules.
2. New Way: Treat the flow like a "temperature" of chaos (Agiture).
3. Discovery: Fluids can get "glassy" and jammed, and they have a hidden "slip" speed (Co-moving velocity) that we must measure to understand them.
4. Result: A new, more accurate way to predict how oil, water, and gas move underground, which is vital for cleaning up oil spills, storing carbon, or finding new energy sources.

Technical Summary

Here is a detailed technical summary of the paper "Immiscible two-phase flow in porous media: a statistical mechanics approach" by Alex Hansen and Santanu Sinha.

1. Problem Statement

The central challenge in the physics of immiscible two-phase flow in porous media is the scale-up problem: bridging the gap between pore-scale hydrodynamics (where fluid interfaces, capillary forces, and viscous forces compete) and the continuum scale (Darcy scale), where the medium is treated as a homogeneous continuum.

• Current State: The industry standard is Relative Permeability Theory (originating from Muskat, Wyckoff, Botset, and Leverett in the 1930s-40s). This phenomenological approach splits the total pressure into two fluid-specific pressures ($p_w, p_n$) and defines relative permeabilities ($k_{rw}, k_{rn}$) and capillary pressure ($p_c$) as functions solely of saturation ($S_w$).
• Limitations: This theory relies on strong, often unverified assumptions (e.g., that parameters depend only on saturation). It fails to capture complex emergent behaviors like hysteresis, non-linear flow regimes, and glassy dynamics. Attempts to derive this from first principles have historically led to "exploding complexity."

2. Methodology: Non-Thermal Statistical Mechanics

The authors propose a novel framework based on Edwin T. Jaynes' generalization of statistical mechanics (1950s). Instead of thermal equilibrium, this approach applies Shannon information entropy to non-thermal, steady-state hydrodynamic systems.

• The Approach:
  1. Representative Elementary Area (REA): The system is modeled as a cylindrical porous sample with a cross-sectional area $A$. A Representative Elementary Area (REA) is defined containing wetting fluid flow ($Q_w$), non-wetting fluid flow ($Q_n$), wetting area ($A_w$), and non-wetting area ($A_n$).
  2. Configurational Entropy: The system's state $X$ is defined by the fluid configuration and velocity field. The authors define a configurational entropy $S = -\int p(X) \ln p(X) dX$.
  3. Maximum Entropy Principle: The probability distribution $p(X)$ is determined by maximizing entropy subject to constraints on the average flow rates and areas.
  4. Emergent Variables: This maximization yields a partition function and introduces three new intensive variables (Lagrange multipliers):
     • Agiture ($\theta$): Analogous to temperature, related to the pressure gradient ($\theta \propto |\nabla p|$).
     • Flow Derivative ($\mu$): Analogous to chemical potential, conjugate to wetting area/saturation.
     • Flow Pressure ($\pi$): Conjugate to pore area (related to quenched disorder).

3. Key Contributions and Theoretical Framework

A. Thermodynamics-Like Formalism

The method successfully constructs a "non-thermal thermodynamics" at the Darcy scale.

• Legendre Transforms: The framework establishes relationships between extensive variables (flow rates, areas) and intensive variables ($\theta, \mu, \pi$) similar to standard thermodynamics.
• Constitutive Equations: It derives a constitutive equation relating average fluid velocity ($v_p$) to the pressure gradient, saturation, and porosity without relying on the phenomenological relative permeability curves.

B. The Co-Moving Velocity ($v_m$)

A critical theoretical breakthrough is the identification of the co-moving velocity ($v_m$).

• Distinction: The authors distinguish between seepage velocities ($v_w, v_n$), which are the actual average velocities of each fluid phase measured in the lab, and thermodynamic velocities ($\hat{v}_w, \hat{v}_n$), which are derivatives of the total flow rate with respect to area.
• The Relation: The seepage velocities are related to thermodynamic velocities via the co-moving velocity:
  $$v_w = \hat{v}_w - S_n v_m$$
  $$v_n = \hat{v}_n + S_w v_m$$
• Significance: $v_m$ represents the relative motion of the two fluids that is not captured by simple saturation changes. It is a necessary variable to close the system of equations at the continuum scale.

C. Phase Diagram and Glassy Dynamics

By mapping the flow problem to a spin model and using Boltzmann machine learning, the authors identify a phase diagram for steady-state flow based on the Capillary number ($Ca$) and saturation ($S_n$):

• Phase Ia (Frozen): Interfaces are pinned by capillary forces; no flow occurs (kinetic arrest).
• Phase Ib (Glassy): Interfaces move intermittently. The system exhibits glassy behavior characterized by a non-zero Edwards-Anderson order parameter and large fluctuations/hysteresis. This phase corresponds to the power-law regime ($v \propto |\nabla p|^\beta$ with $\beta \approx 1.85$).
• Phase II/III (Non-Glassy): At high $Ca$, the system transitions to a linear Darcy-like regime where interfaces move freely.
• Critical Line: The transition between the glassy (power-law) and non-glassy (linear) regimes is identified as a true critical line (phase transition), explaining the hysteresis observed in porous media.

D. Analogy to Ordinary Thermodynamics

The authors demonstrate a profound link between their non-thermal theory and ordinary thermodynamics of fluid mixtures.

• They show that the co-moving velocity ($v_m$) in porous media is mathematically analogous to the co-molar volume ($\nu_m$) in binary fluid mixtures (e.g., water and ethanol).
• Just as mixing ethanol and water results in a volume contraction ($V < V_1 + V_2$), the flow of two immiscible phases in porous media involves a "co-moving" effect that cannot be described by simple additive velocities. This validates the universality of the statistical mechanics approach.

4. Results

• Validation: The theoretical framework explains experimental observations of power-law flow regimes and hysteresis that Relative Permeability Theory cannot.
• Data Collapse: Analysis of relative permeability data and simulations shows that the co-moving velocity follows a simple linear functional form with respect to the flow derivative ($\mu$), similar to how co-molar volume depends on partial molar volume differences in standard mixtures.
• Unified Description: The theory unifies pore-scale physics (interfaces, ganglia dynamics) with Darcy-scale continuum descriptions without arbitrary phenomenological fitting parameters.

5. Significance and Future Directions

• Paradigm Shift: This work challenges the dominance of Relative Permeability Theory, offering a rigorous, physics-based alternative rooted in information theory and statistical mechanics.
• Practical Applications: Understanding the glassy phase is crucial for practical applications (e.g., oil recovery, groundwater remediation) where hysteresis and non-linear flow dominate.
• Future Work: The authors suggest extending this framework to non-equilibrium thermodynamics to handle transient flows (displacement processes) rather than just steady states. This would eliminate the need for the problematic capillary pressure function ($p_c$) by using gradients in agiture and flow derivative as driving forces.

In conclusion, Hansen and Sinha provide a robust mathematical structure that treats immiscible two-phase flow as a non-thermal thermodynamic system, successfully identifying new emergent variables (agiture, flow derivative, co-moving velocity) and explaining complex phenomena like glassy flow and hysteresis through a unified statistical mechanics lens.