Subsurface environments like shale gas reservoirs, carbonate rocks, and geothermal systems present a fundamental physical paradox. They consist of two radically different domains: a dense rock matrix that stores the vast majority of the fluid but has extremely low permeability, and a network of fractures that stores very little fluid but acts as a high-conductivity highway system.
If you try to simulate this using a standard single-porosity grid, you are forced to average these properties together. This numerical smoothing destroys the underlying physics: it severely underestimates how fast fluids channel through fractures and misrepresents how slowly fluids drain from the matrix. To solve this, researchers have developed methods that couple explicit geometric fracture representations with dual-porosity continuum models.
The Dual-Continuum Paradigm
To resolve the scale conflict between matrix and fractures, traditional reservoir simulation relies on the dual-porosity or dual-permeability concept. As explained by Mingzhou Zhang in Research on Water Influx Issues in Carbonate Reservoir Development, this framework treats the reservoir as two distinct, overlapping mathematical continua occupying the same physical space. One continuum models the high-storage matrix, while the other models the high-flow fracture network.
Standard Grid (Averaged) Dual-Continuum (Overlapping)
┌───────────────────────┐ ┌───────────────────────┐
│ │ │ ┌═════════════════┐ │
│ K_effective │ │ │ Fracture Flow │ │
│ (Smoothed Physics) │ │ └═════════════════┘ │
│ │ │ │ (Transfer │
│ │ │ ┌─────────────────┐ │
│ │ │ │ Matrix Storage │ │
└───────────────────────┘ └───────────────────────┘
The magic that makes this work is the inter-domain mass transfer. The two systems are coupled via a transfer function or mass exchange term that dictates how fast fluid moves across the interface between the matrix and the fractures. Dung Ngoc Tien Tran et al. demonstrated in A model of solute diffusion in unsaturated double-porosity medium by homogenization that these dual-equation systems can be rigorously derived using asymptotic homogenization. This mathematical framework provides a fast alternative to fine-scale simulations by capturing macroscopic behavior without needing to model every microscopic pore.
However, classical dual-porosity models assume the fracture network is uniform and pervasive. In reality, fractures have explicit, highly irregular geometries—ranging from massive regional faults to intricate, tree-like hydraulic networks.
The Solution: Embedded Discrete Fracture Models
To capture real-world fracture geometries without causing computational gridlock, modern approaches use the Embedded Discrete Fracture Model (EDFM).
Traditionally, to simulate an explicit fracture, engineers had to use unstructured grids that conformed perfectly to the fracture's complex surfaces. This required massive local grid refinement, leading to millions of tiny cells and prohibitive computational costs.
As Yong Wang et al. demonstrate in Numerical Simulation of a Horizontal Well With Multi-Stage Oval Hydraulic Fractures in Tight Oil Reservoir Based on an Embedded Discrete Fracture Model, EDFM bypasses this by decoupling the fracture geometry from the matrix grid. The rock matrix is discretized using a simple, computationally efficient structured grid. The fractures are then treated as lower-dimensional geometric entities (such as 2D planes cutting through a 3D space) that are mathematically "embedded" directly into the matrix cells.
Conforming Mesh (Expensive) Embedded Fracture (EDFM)
┌─────┬─────┬─∕───┬─────┐ ┌───────────┬───────────┐
│ │ │∕ │ │ │ │ ∕ │
├─────┼─────🗲─────┼─────┤ ├───────────┼────∕──────┤
│ │ ∕│ │ │ │ │ ∕ │
├─────┼───∕─┼─────┼─────┤ ├───────────┼──∕────────┤
│ │ ∕ │ │ │ │ │ ∕ │
└─────┴─∕───┴─────┴─────┘ └───────────┴───────────┘
Why does this work so well? Instead of forcing the grid to conform to the fracture, the simulator calculates the physical intersections between the embedded fracture planes and the background matrix blocks. It then introduces specialized non-neighboring connections (NNCs) to handle the fluid exchange between the matrix cell and the fracture segment passing through it.
Yalchin Efendiev et al. show in Constrained energy minimization based upscaling for coupled flow and mechanics that this dual-continuum approach can be combined with advanced upscaling techniques. This allows researchers to capture both fluid flow and geomechanical interactions efficiently on a coarse grid, drastically reducing computational overhead while maintaining high fidelity.
Integrating Discrete Networks with Dual Continua
The true power of modern subsurface modeling comes from marrying these two concepts: using explicit geometric representations for large, dominant fractures (like hydraulic fractures or faults) and dual-porosity continua for the background micro-fractures.
In Integration of Geological Model and Numerical Simulation Technique to Characterize the Remaining Oil of Fractured Biogenic Limestone Reservoirs, Xintong Yu et al. show how integrating Discrete Fracture Networks (DFNs) with dual-porosity dual-permeability simulations allows for independent tracking of fluid pressure and saturation in both domains. This hybrid approach yields highly accurate reserve calculations and production history matching, making it possible to pinpoint bypassed oil pockets.
Similarly, Dong Chen et al. utilize an EDFM approach in Research on Water Invasion Law and Control Measures for Ultradeep, Fractured, and Low-Porosity Sandstone Gas Reservoirs: A Case Study of Kelasu Gas Reservoirs in Tarim Basin to model a "pore-fracture-fault" triple medium. This enables the analysis of fluid flow and water invasion mechanisms within highly complex, interconnected geological structures.
Modeling Complex Branching and Data Assimilation
To make these models match physical reality, researchers have advanced how these explicit geometries are constructed and calibrated. For instance, hydraulic fracturing rarely creates simple planar structures; instead, it forms intricate, branching networks.
To capture this behavior, Xin Liu et al. in Productivity Model for Multi-Fractured Horizontal Wells with Complex Fracture Networks in Shale Oil Reservoirs Considering Fluid Desorption and Two-Phase Behavior and Dengji Tang et al. in The Inversion Method of Shale Gas Effective Fracture Network Volume Based on Flow Back Data—A Case Study of Southern Sichuan Basin Shale leverage fractal theory. By representing fracture networks as tree-shaped fractals, their models simulate two-phase flow, fracture compressibility, and fluid desorption with much higher fidelity than simplified geometric assumptions allow.
Because we cannot see directly underground, these explicit networks must be inferred from indirect data. To address this challenge, Maruti Kumar Mudunuru et al. developed a sequential inversion framework in Sequential geophysical and flow inversion to characterize fracture networks in subsurface systems. Their method uses microseismic data to constrain fracture orientations and lengths before running flow simulations, significantly reducing subsurface uncertainty.
By combining explicit geometry modeling with data assimilation and local grid refinements—such as the probabilistic workflows and logarithmically spaced dual-permeability grids discussed by Jie Nie et al. and Bo Wang et al.—engineers can rapidly predict well productivity and optimize fracturing treatments in highly complex, interference-heavy environments.
Key Foundational Work
Sequential geophysical and flow inversion to characterize fracture networks in subsurface systems (2017) — contributed 1 concept elements to this direction.
A model of solute diffusion in unsaturated double-porosity medium by homogenization (2018) — contributed 1 concept elements to this direction.