Characterizing Spatial Heterogeneity in Subsurface Reservoirs

45 concepts38 papers1 proposal

Subsurface formations—whether they are oil and gas reservoirs, saline aquifers for carbon sequestration, or foundational soil layers—are rarely uniform blocks of rock. Instead, they are highly irregular, multi-scale "patchworks" shaped by millions of years of depositional environments, tectonic stresses, and chemical transformations.

Understanding the ground beneath us requires moving away from idealized, homogeneous models. Spatial heterogeneity dictates how fluids travel, where resources accumulate, and how the subsurface responds to human engineering.

The Physics of Anomalous Transport

In a perfectly uniform porous medium, fluid flows predictably, spreading out in a classic Gaussian distribution. However, real-world geological formations break these simplified assumptions. Carbonate and shale reservoirs often feature complex dual-pore networks consisting of both microscopic matrix pores and large, irregular cavities known as vugs or fractures (Xu et al.).

When fluid encounters these extreme variations in hydraulic conductivity, transport becomes "anomalous." Rather than moving at a uniform front, the fluid finds preferential high-permeability pathways, leaving behind stagnant zones. As Kaili Wang demonstrates in Impact of Hydraulic Conductivity on Solute Transport in Highly Heterogeneous Aquifer, modeling these highly heterogeneous aquifers requires stochastic frameworks like fractional Lévy motion. These models capture the real-world reality of fluid transport: sharp leading edges where fluid shoots forward along fractures, and long lingering tails where fluid remains trapped in tight pores.

To understand these pathways without a granular map of every internal connection, researchers rely on statistical physics. Andrade et al. mapped complex fluid dynamics onto a universal scaling template derived from percolation theory. This framework explains how fluid travel times and path lengths scale across disordered media, providing a mathematical foundation for predicting flow when direct sampling is impossible.

Homogeneous Flow (Gaussian)        Heterogeneous Flow (Anomalous / Preferential)
[ = = = = = = = = = > ]            [ =====>                      ]  <- Fast track (Fracture)
[ = = = = = = = = = > ]            [ =>                          ]  <- Trapped zone (Matrix)
[ = = = = = = = = = > ]            [ ========>                   ]  <- Vug network

Bridging the Scale Gap

The primary challenge in characterizing subsurface heterogeneity is a severe data mismatch. Engineers collect high-resolution, microscopic data from drill cuttings and core samples, but they must use this information to predict macroscopic behavior across kilometers of a reservoir field.

When lateral variations in rock types (facies) and mineral compositions occur rapidly across short distances, assuming uniform properties leads to failed extraction or storage strategies (Amer et al.; Eid et al.). For example, adjacent wells drilled into the same submarine fan formation can exhibit vastly different productivity levels simply because one hit a localized sand channel while the other missed it (Li et al.).

To bridge this gap, modern workflows integrate micro-scale petrophysical measurements with macro-scale geophysical data:

  1. Micro-to-Macro Scoring: Xu et al. developed a scoring system that uses Principal Component Analysis (PCA) to weight 14 parameters, combining microscale drill cutting data with macroscale well logs to establish clear production thresholds.
  2. Pore Architecture Mapping: By pairing Nuclear Magnetic Resonance (NMR) with X-ray Diffraction (XRD), Liao et al. showed how microscopic mineral variations dictate the specific pressure differentials required to displace fluids in tight oil reservoirs.
  3. 3D Structural Modeling: Researchers combine 2D and 3D seismic data with well logs to map complex fault networks and rapid lateral rock variations, turning fragmented borehole data into continuous spatial models (Eid et al.; Amer et al.).

Dynamic Heterogeneity and Reservoir Degradation

Spatial heterogeneity is not a static property; it changes dynamically when humans alter the subsurface environment. This is particularly evident in geological carbon sequestration, where $CO_2$ is injected into deep saline aquifers.

The introduction of $CO_2$ disrupts the chemical equilibrium between the native water and the surrounding rock. In Effects of CO2–water–rock cycling cycles on sandstone pore structure, Sun et al. uncover a complex chemical tug-of-war triggered by cyclic fluid injection. The acidic fluid dissolves minerals, creating secondary pore spaces, but it simultaneously causes salt precipitation elsewhere.

Cyclic CO2 Injection -> Chemical Disruption -> Localized Dissolution (Low Density)
                                            -> Mineral Precipitation (High Density Clogging)
                                            => Induced Structural Heterogeneity

Crucially, this process does not happen uniformly. It creates localized zones of high-density clogging alongside low-density dissolution channels. This induced heterogeneity progressively degrades the structural uniformity of the reservoir rock, causing fluid flow patterns to become increasingly erratic and compromising the long-term predictability of carbon storage sites.

Computational Frameworks for Uncertainty

Simulating these highly variable systems introduces a massive computational bottleneck. To capture fine-scale fracture networks or localized clogging, traditional numerical methods require ultra-fine grids, which demand impractical amounts of computing power. Conversely, smoothing these variations out over a coarse grid erases the very heterogeneity that drives anomalous flow.

To resolve this computational gridlock, Sergei Stepanov introduces a hybrid approach in Prediction of discretization of online GMsFEM using deep learning for Richards equation. By combining the Generalized Multiscale Finite Element Method (GMsFEM) with deep learning, the author trains neural networks to rapidly predict local multiscale basis functions. This allows the simulation to embed fine-scale permeability variations directly into a coarser, computationally efficient grid, capturing complex flow dynamics without the heavy computational toll.

Ultimately, because we can never fully see underground, characterizing the subsurface remains an inverse problem: we must infer hidden material properties from sparse, noisy observations at the surface or within wells. Framing this challenge within a Bayesian framework allows researchers to treat permeability not as a single fixed value, but as a spatial probability distribution (Dashti et al.). By quantifying this inherent uncertainty, engineers can build risk-aware models that accept subsurface heterogeneity as a fundamental reality rather than a mathematical inconvenience.

History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_localized_reservoir_degradation_b0020972,
  title = {localized reservoir degradation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/localized-reservoir-degradation/rdr_b0020972f48e458a9c480c7dc04f2cf8},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 1, frozen text):

@misc{lacuna_localized_reservoir_degradation_b0020972_v1,
  title = {localized reservoir degradation},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/localized-reservoir-degradation/rdr_b0020972f48e458a9c480c7dc04f2cf8/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake