Residual Saturation under Pressure-Controlled Drainage

Technical Summary: Residual Saturation under Pressure-Controlled Drainage

Problem Statement
The paper addresses a fundamental unresolved question in the physics of porous media: the behavior of residual saturation under pressure-controlled drainage in the infinite-size limit. While standard invasion percolation (IP) models predict that the invaded-phase saturation vanishes as system size approaches infinity (leaving a finite fraction of defending fluid trapped only due to the fractal nature of the breakthrough cluster), it remains unclear whether a pressure-controlled process—which allows the invading phase to access all pores connected to the inlet at a given capillary threshold—results in a finite residual saturation of the defending phase in the thermodynamic limit. This distinction is critical for understanding field-scale phenomena such as waterflooding, where the question arises whether residual oil saturation remains finite as the invaded region approaches reservoir dimensions.

Methodology
The authors formulate pressure-controlled drainage as Bond Percolation with Trapping (BPT) on a pore-network graph.

• Graph Representation: The porous medium is modeled as an undirected graph where nodes represent pore regions and edges represent throats with assigned entry radii (capillary thresholds).
• Process Definition: Unlike standard IP, which advances throat-by-throat via the locally most accessible path, BPT operates under a global pressure constraint. As capillary pressure increases, the "active graph" (comprising throats with entry thresholds below the current pressure) expands.
• Trapping Mechanism: Invasion is defined by connectivity to the inlet on the active graph. Trapping occurs when a portion of the defending phase loses connectivity to the outlet set. The algorithm updates the system by: (1) increasing pressure to activate new throats; (2) identifying the invaded region (inlet-connected active pores); (3) identifying the defending region (outlet-connected non-invaded pores); and (4) classifying all remaining disconnected pores as "trapped."
• Simulations: The study compares BPT against standard Invasion Percolation (IP) on various 2D lattices (square, triangular, Voronoi) and 3D lattices (diamond, simple-cubic with varying coordination numbers). Simulations utilize lognormally distributed throat radii and ensemble averaging to analyze finite-size effects.

Key Contributions and Results

1. Finite Residual Saturation in the Infinite Limit:
   The primary finding is that pressure-controlled drainage leads to a space-filling invaded state where the residual saturation of the defending phase approaches a non-zero, finite limit ($0 < S_\infty < 1$) as system size $L \to \infty$. This contrasts sharply with standard IP, where the invaded saturation vanishes at breakthrough ($S_{IP} \to 0$) because the invading cluster is fractal and bypasses large regions of the defending fluid.

2. Finite-Size Scaling in Two Dimensions:
   In 2D, the deviation of the residual saturation from its thermodynamic limit follows a finite-size scaling law:
   $$S_r(L) \approx S_\infty - \left(\frac{L_0}{L}\right)^\delta$$
   where the exponent $\delta \approx 0.25$. This exponent is universal across different lattice types (square, triangular, Voronoi) and is independent of microscopic details. The slow convergence ($\Delta S_r \sim L^{-0.25}$) implies that achieving representative elementary volumes in 2D requires extremely large system sizes (e.g., doubling the system size reduces the finite-size bias by only a factor of $\approx 1/16$).

3. Dimensionality and Coordination Number Effects in Three Dimensions:
   In 3D, finite-size corrections decay significantly faster, with an exponent $\delta \approx 0.75$. Consequently, the representative volume required to approximate the infinite-size limit is much smaller in 3D than in 2D. Furthermore, while the asymptotic residual saturation remains finite in 3D, its value depends on the coordination number of the lattice; higher connectivity provides more alternative invasion pathways, reducing the trapped fraction, though it does not eliminate it entirely.

4. Distinction from Invasion Percolation:
   The paper clarifies that the difference between BPT and IP is not merely a matter of curve shape or burst dynamics but a fundamental difference in asymptotic states. IP yields a fractal invading cluster at breakthrough, whereas BPT yields a space-filling invaded region with a finite trapped fraction coexisting in the infinite limit.

Significance
The paper establishes a direct theoretical connection between percolation theory and pressure-saturation relations by demonstrating that connectivity and trapping are the determining factors for residual saturation under pressure-controlled drainage. By showing that a finite fraction of the pore space remains trapped even in the infinite-system limit, the work extends the standard invasion-percolation picture beyond the breakthrough state. This provides a rigorous framework for interpreting macroscopic constitutive laws and suggests that the "representative elementary volume" problem is highly sensitive to dimensionality, being far more restrictive in 2D than in 3D. The results imply that in pressure-controlled scenarios, a finite fraction of the defending fluid is irrecoverable regardless of system size, a finding with direct implications for subsurface fluid redistribution and recovery processes.