Abstract
This study investigates the flow of fluid in porous media across two and three dimensions, modeled via bond percolation on a lattice. By analyzing tracer particles driven by pressure differences between two fixed sites ("wells") separated by a Euclidean distance $r$, we characterize the distribution functions for both the shortest geometric path and the minimal traveling time. We propose a universal scaling ansatz that accounts for system size $L$ and bond occupancy probability $p$. Extensive simulations in 2D and 3D confirm that while the shortest path and fastest path are distinct, they follow a similar mathematical distribution governed by specific critical exponents. Our results provide a comprehensive set of parameters, including fractal dimensions and power-law exponents, that characterize flow behavior in disordered media.
Introduction
Predicting how fluids like water or oil migrate from an injection point to a production well is a central challenge in geophysics and reservoir engineering. Because the internal "plumbing" of underground aquifers and oil fields is largely inaccessible, researchers rely on statistical mechanics—specifically percolation theory—to model these complex, heterogeneous environments.
This research addresses the limitations of previous models that focused primarily on infinite systems at the critical point (the threshold where a connected path first forms). By examining finite-sized systems and "off-critical" scenarios (where the medium is more or less connected than the bare minimum), this paper provides a more realistic framework for understanding fluid transport in disordered materials.
Modeling Fluid Flow in Disordered Media
The researchers utilized bond percolation on $L^d$ lattices ($d=2, 3$) where bonds are occupied with a probability $p$. To simulate realistic flow, the occupied bonds were treated as a network of resistors. The study employed a multi-step computational approach:
- Cluster Growth: The Leath algorithm was used to generate percolation clusters.
- Path Identification: A "burning" algorithm identified the shortest geometric paths between two sites.
- Flow Dynamics: By solving Kirchhoff’s laws for pressure at every site and applying Darcy’s law, the researchers determined the velocity of tracer particles, allowing them to identify the "fastest" (minimal time) path.
The Scaling Ansatz
The core contribution of this work is a proposed distribution function, $P'(x|r)$, which acts as a mathematical template for both path length and travel time. This ansatz incorporates three distinct exponential "cutoff" functions to account for physical constraints:
- The Euclidean Limit: A path cannot be shorter than the straight-line distance ($r$) between two points.
- Finite Size Effects: The boundaries of the system ($L$) influence the available paths.
- Correlation Length: The scale of disorder ($\xi$) determines when the medium stops looking like a fractal and starts looking homogeneous.
Key Insights: Geometry vs. Dynamics
A significant finding of the study is the distinction between the shortest path (the path with the fewest bonds) and the fastest path (the path with the minimal travel time). Because fluid flow is driven by local pressure gradients, the path with the fewest steps is not always the one with the highest velocity.
In 2D simulations, the researchers found these two quantities are strongly correlated, following the relationship $t_{min} \sim \ell^{1.17}$. Despite these physical differences, both the shortest path and the minimal travel time distributions fit the same scaling template, differing only in their specific critical exponents.
Scaling Exponents and Data Collapse
The authors calculated a definitive set of exponents that characterize these distributions:
- Fractal Dimensions: The fractal dimension of the shortest path ($d_{min}$) was confirmed at $\approx 1.13$ (2D) and $\approx 1.37$ (3D). The dimension for minimal travel time ($d_{tm}$) is higher, at $\approx 1.33$ (2D) and $\approx 1.45$ (3D).
- Power-Law Slopes: The slope for travel time distributions ($g'_{tm}$) was found to be $\approx 2.0$ in 2D and $\approx 2.1$ in 3D.
The robustness of the model was proven through "data collapse," where results from varying system sizes and distances converged onto a single curve when scaled according to the proposed ansatz.
The Homogeneity Crossover
The paper illustrates the transition of a system from fractal behavior to homogeneous behavior. When the distance between wells ($r$) is significantly larger than the correlation length ($\xi$), the distribution of travel times shifts from a skewed power law to a standard Gaussian (Normal) distribution. This crossover is vital for engineers to understand when local disorder ceases to dominate the overall flow rate of a reservoir.
Limitations and Future Directions
While the 2D results are highly precise, 3D simulations for systems significantly above the percolation threshold remain computationally expensive, leaving some large-scale parameters inconclusive. Furthermore, the model assumes a constant total flux and a binary (on/off) bond system. Future research could improve realism by incorporating a continuous distribution of bond conductances, better mimicking the varied pore sizes found in natural rock and soil.
Summary updated