Deformation-assisted fluid percolation in rock salt

See allHide authors and affiliations

Science  27 Nov 2015:
Vol. 350, Issue 6264, pp. 1069-1072
DOI: 10.1126/science.aac8747

Salted away no longer?

Rock salt deposits are thought to be impermeable to fluid flow and so are candidates for nuclear waste repositories. Ghanbarzadeh et al. found that some salt deposits in the Gulf of Mexico are infiltrated by oil and other hydrocarbons. If these salt domes are not completely isolated from the surrounding environment, they will not be suitable for deep geological waste storage sites.

Science, this issue p. 1069


Deep geological storage sites for nuclear waste are commonly located in rock salt to ensure hydrological isolation from groundwater. The low permeability of static rock salt is due to a percolation threshold. However, deformation may be able to overcome this threshold and allow fluid flow. We confirm the percolation threshold in static experiments on synthetic salt samples with x-ray microtomography. We then analyze wells penetrating salt deposits in the Gulf of Mexico. The observed hydrocarbon distributions in rock salt require that percolation occurred at porosities considerably below the static threshold due to deformation-assisted percolation. Therefore, the design of nuclear waste repositories in salt should guard against deformation-driven fluid percolation. In general, static percolation thresholds may not always limit fluid flow in deforming environments.

Rock salt in sedimentary basins has long been considered to be impermeable and provides a seal for hydrocarbon accumulations in geological structures (1, 2). The low permeability of rock salt also has the potential to isolate nuclear waste from ambient groundwater and may provide a suitable deep geological waste repository (3, 4). This option is currently being reconsidered in the United States after the closure of the Yucca Mountain repository in Nevada (3). However, field observations of oil-impregnated rock salt (5), geochemical evidence for the replacement of the in situ brines (6), and the drainage of brine from mining-induced fractures and dilatant microcracking (3, 7) demonstrate that the permeability of natural rock salt may not be negligible.

Brine-filled pore networks in rock salt approach textural equilibrium due to fast reaction kinetics of salt dissolution and reprecipitation (8). Percolation in these networks is controlled by the dihedral angle θ at the solid-solid-liquid triple junctionsEmbedded Image (1)where γss and γsl are the solid-solid and solid-liquid surface energies (912). The dihedral angle is therefore a thermodynamic property that changes with pressure P and temperature T. The static pore-scale theory shows that texturally equilibrated pore networks percolate at any porosity if θ ≤ 60°, whereas a finite porosity is required for percolation if θ > 60° (1012).

The experimentally measured θ in salt-brine systems decreases with both increasing P and T (Fig. 1), suggesting that fluids at shallow depth must overcome a percolation threshold, whereas fluids at greater depth are likely to percolate at any porosity. The PT trajectory of multiple petroleum wells in the Gulf of Mexico crosses this transition and therefore provides an opportunity to test the static pore-scale theory in a realistic field setting.

Fig. 1 Brine percolation in rock salt.

PT trajectories of multiple subsalt petroleum wells are shown together with experimentally measured dihedral angles θ for the salt-brine system (8). The static theory predicts that fluid must overcome a percolation threshold in the gray area, whereas fluids are predicted to percolate at any porosity in the white area. The light gray area highlights the transition zone, 60° < θ < 65°, between percolating and disconnected pore space (8). The segment of each well that is located within the salt has a lower geothermal gradient due to the high conductivity of salt and is shown as a dashed line. The depth axis is only for illustration and assumes an overburden with constant density, ρ = 2300 kg/m3.

We confirm the static pore-scale theory in undrained laboratory experiments on synthetic salt samples that have been imaged with nondestructive x-ray microtomography after quenching to ambient conditions (13). We present the results of two representative experiments (Fig. 2) performed at P = 20 MPa and T = 100°C (Exp-I) and P = 100 MPa and T = 275°C (Exp-II). The three-dimensional (3D) reconstruction (Fig. 2, A and B) and medial axis representation of the pore space (Fig. 2, C and D) show that the brine network is disconnected in Exp-I and is connected in Exp-II. This is confirmed by statistical analysis of the coordination number distributions that show that almost all nodes in Exp-I have coordination number 1, whereas the coordination numbers of 3 and 4 are most abundant in Exp-II (fig. S4). The distribution of the apparent dihedral angles has a median of 67 ± 5° for Exp-I and 52 ± 6° for Exp-II (Fig. 2E). Distributions with a single narrow peak, as well as similarity to previously reported values of dihedral angle (8), indicate that the experiments are approaching textural equilibrium. Comparison of experiments with the regime diagram for fluid percolation show that static pore-scale theory successfully predicts the connectivity of the pore space (Fig. 2F). These experimental results confirm the first-order control of the dihedral angle on brine percolation and serve as a baseline for the field observations of fluid distributions in deformed rock salt.

Fig. 2 Pore networks in rock salt.

Hydrostatic experiments on synthetic rock salt have been performed at P = 20 MPa and T = 100°C (Exp-I) and P = 100 MPa and T = 275°C (Exp-II). (A and B) 3D reconstruction of the pore network at textural equilibrium; all edges of the 3D volumes correspond to 660 μm. (C and D) The skeletonized pore network extracted from the reconstructed 3D volume; colored according to local pore-space-inscribed radius, with warmer colors indicating larger radius. (E) Distribution of apparent dihedral angles in the experiments. (F) Exp-I and Exp-II in the θϕ space regime diagram with the percolation threshold obtained from the static pore-scale theory (10, 12). Inserted images show the details of automated dihedral angle extraction from 2D images (13). We report the median value of dihedral angles and the estimated errors based on the 95% confidence interval. (G) Porosity of natural rock salt inferred from resistivity logs (Fig. 3B).

Commercial interest in the large hydrocarbon accumulations below extensive bodies of allochthonous salt in the deepwater Gulf of Mexico provides an opportunity to test the static pore-scale theory in slowly moving natural rock salt. We studied field data from the salt section of 48 wells crossing the predicted transition zone from disconnected to percolating pore space (Fig. 1) to constrain the brine and hydrocarbon connectivity. Typically, no intact core is recovered from the salt section of wells, and the available data sets consist of wireline well logs and mud logs (13). Wireline well logs, obtained by lowering a measurement tool into the well, characterize different properties of the formation rock and fluids (Fig. 3, A and B). Mud logs, which record the hydrocarbon gas content and observations from the drill cuttings brought to the surface, provide direct constraints on the presence of hydrocarbons in salt (Fig. 3, C to E). Hydrocarbon signs reported in mud logs include fluorescence, oil staining, oil cut, and dead oil embedded in the salt.

Fig. 3 Petrophysical observations.

Wireline well logs and mud logs data constraining the fluid distribution and connectivity in the well GC8 from the deep water Gulf of Mexico (13). (A) Gamma-ray log, (B) electrical resistivity, (C) total hydrocarbons gas, (D) gas chromatography, (E) hydrocarbon signs (FL, fluorescence; OS, oil stain; DO, dead oil; and OC, oil cut) in mud logs, and (F) the dihedral angle inferred from experimental data (Fig. 1). Shading around each curve shows the measurement error and average fluctuations in data. The gray background corresponds to shaded areas in the experimental data (Fig. 1).

We chose only those salt sections for analysis that were free of other rock fragments, as indicated by low values of naturally occurring gamma radiation (Fig. 3A). In contrast to the uniform gamma-ray signature, all other logs (Fig. 3, B to E) show a distinct change in the bottom third of the salt. The very high electrical resistivity in the upper two-thirds of the salt section implies that the conductive brine is not connected (Fig. 3B) (14). In this region, the porosity calculated from Archie’s law is below 0.4% (Fig. 2G) (13). The reduction of electrical resistivity by an order of magnitude in the bottom third suggests that brine is connected at porosities below 0.8% (Fig. 2G). The salt-brine dihedral angle inferred from the PT trajectory of the well (fig. S6) (13) drops below 60° in the bottom third of the salt (Fig. 3F), consistent with the static pore-scale theory.

In addition to a connected brine phase, the total gas hydrocarbons and gas chromatography logs indicate a substantial increase in the amount of natural gas in the lower third of the salt (Fig. 3, C and D). We observe this general pattern also in the mud logs that contain no indications of hydrocarbons in the top two-thirds but show multiple signs of hydrocarbons in the bottom third (Fig. 3E). In the presence of brine, hydrocarbons are the nonwetting phase, so that the textural equilibration of the pore network occurs through brine-mediated dissolution and reprecipitation of the salt. The dihedral angle of the brine-salt system governs the connectivity of pore space, consistent with observations in wireline well logs and mud logs. Once hydrocarbons overcome the capillary entry pressure (5), they can enter the salt in regions where the brine network is connected. Subsequent imbibition of the brine can trap the hydrocarbons in the pore space (fig. S7). The presence of hydrocarbons therefore indicates that a connected pore space existed during the entry of the hydrocarbons into the rock salt. This interpretation is consistent with previous work reporting direct observations of oil-stained salt cores recovered from conditions where θ < 60° (5).

High-quality resistivity logs (Fig. 3B) are only available in two wells due to technical difficulties and lack of commercial interest in the salt section of wells. Therefore, we rely on the logs that detect hydrocarbons to infer the connectivity of the brine in the remaining 46 wells. We group spatially associated wells to look at the distribution of hydrocarbons in salt sections (Fig. 4). The abundance of hydrocarbons is affected by the distance of the nearest hydrocarbon source from the bottom of the salt. For example, the first oil source is more than 2000 m below the base of salt in the wells of group WR13, justifying the sparsity of hydrocarbon signs.

Fig. 4 Fluid distributions in salt wells.

Hydrocarbons signs from mud logs of all 48 wells covering 150,000 m of salt are shown as a function of dihedral angle (13). Wells are divided into 14 groups based on spatial proximity. Salt extent is shown by an arrow in each region. Theoretical fluid connectivity is indicated by gray scale (Fig. 1). Abbreviations denote the following protraction areas in the Gulf of Mexico: AT, Atwater Valley; GC, Green Canyon; KC, Keathley Canyon; MC, Mississippi Canyon; and WR, Walker Ridge.

We converted the depth to dihedral angle using available experimental data (Fig. 1 and fig. S6). All the wells that we considered show signs of connected pore space at depths where the dihedral angle is below 60°, except the shallow wells of group MC11. Using the two electrical resistivity logs and Archie’s law, we estimate that the porosity of these connected regions is less than 1% (Fig. 2G). This provides direct field evidence that dihedral angles below 60° allow the percolation of texturally equilibrated pore networks at porosities below the transport limit in more typical porous media that originated as clastic sediments (15).

Nonetheless, field data also show evidence of percolating pore space at shallower depths, where the dihedral angle is substantially above 60° (Fig. 4). Under these conditions, the porosity must increase above a threshold to allow percolation. Static pore-scale theory requires porosities between 2 and 3% to allow percolation at dihedral angles between 65° and 70° (Fig. 2F). However, none of the porosities inferred from the available resistivity logs exceed 1%, and most are substantially lower (Fig. 2G), which is consistent with direct measurements of rock salt porosity (16, 17). The observation of percolating fluids at high dihedral angles and low porosities is not consistent with the static theory.

Viscous flow of rock salt due to the density contrast with the surrounding sediments may explain the failure of the static pore-scale theory to predict the percolation of pore space at high dihedral angles. At low effective mean stress, deformation-induced microcracking can lead to the formation of a percolating pore space (5). This microcracking-induced percolation is commonly observed in the zone of disturbed rock around openings in salt mines or nuclear waste repositories and under high overpressures in nature (3, 5). At the depth of petroleum wells considered here, the effective mean stress is sufficient that deformation occurs in the compaction regime, where existing microcracks close and heal (18, 19).

However, deformation may induce permeability even in the absence of microcracking. At high effective mean stress and in the presence of small amounts of brine, the dislocation creep of salt is accompanied by fluid-assisted dynamic recrystallization and pressure solution creep (2022). Both static and dynamic recrystallization are associated with transformation of the isolated grain-boundary fluid inclusions into grain-boundary fluid films (23, 24). The dynamic wetting of the grain boundaries and compaction have been observed in deformation experiments under conditions where Embedded Image (22). This suggests that dynamic grain-boundary wetting induced fluid percolation and drainage at porosities below the percolation threshold.

These laboratory results must be extrapolated to natural conditions using appropriate microphysical models and suggest that fluid-assisted dynamic recrystallization becomes important at strain rates below 10–10 s–1 (21). This is consistent with the recrystallized microstructures and x-ray microtomography of grain-boundary brine films in natural rock salt, as well as estimated natural strain rates between 10–15 and 10–11 s–1 (5, 25, 26). This confirms earlier suggestions that dynamic grain-boundary wetting associated with grain-boundary migration is a plausible mechanism in natural rock salt.

This conclusion is also supported by the comparison of the relative magnitude of shear stresses, Δσ, and the capillary pressure introduced by surface tension forces, Δp, given by capillary numberEmbedded Image(2)where r is the mean radius of disconnected pores. Microstructural evidence preserves records of differential stresses up to 1 MPa in subhorizontal bedded salts (27) and 2 MPa in salt domes (5, 28). In comparison, the capillary pressure for r = 10–4 m and γsl = 0.1 N/m is on the order of 103 Pa (29). Therefore, Embedded Image and the shear stresses in rock salt may exceed capillary pressures and hence facilitate deformation-assisted percolation. This provides an explanation for the penetration of hydrocarbons into shallow regions of the salt, where θ > 60° and porosity is below the static percolation threshold (Fig. 2G and Fig. 4).

These field observations have implications for ensuring hydrological isolation of nuclear waste in rock salt. At the relatively shallow depth typically considered for geological storage, the dihedral angle is between 65° and 72° and should prevent brine percolation in rock salt, based on static pore-scale theory and experiments. However, field observations reported here show that such moderate dihedral angles do not guarantee hydrological isolation in deformed rock salt. The deformation-assisted percolation observed in salt sections of petroleum wells is not associated with human-made excavations, suggesting that this mechanism is not limited to the vicinity of the repository site and the duration of room closure around the waste. Lower differential stresses recorded in shallow bedded rock salt suggest that it is more likely to provide an impermeable barrier. However, tectonic forces and excavations can result in high stresses in shallow cold salt. Therefore, it is important to characterize the salt microstructure of potential repositories to determine the stress history, state of grain boundaries, and fluid distribution. Future work should also constrain the permeability that can be generated by deformation-assisted percolation and its persistence.

Beyond the direct application to salt-brine systems, the field observations reported here also provide an important test of a general theory that underlies our understanding of fluid percolation and flow in ductile regions of Earth. This is of particular interest to the debate about whether moderate dihedral angles can prevent the segregation of core-forming melts in the deforming lower mantle (3032). The inaccessibility of Earth’s mantle to field observations has prevented the resolution of this debate. The observations of fluid distribution in rock salt reported here show that deformation-assisted percolation is possible and suggest that core formation by percolation may be a viable mechanism, even if the dihedral angle is above 60°.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

Database S1

References (3343)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. W. B. Lindquist et al., 3DMA-Rock: A Software Package for Automated Analysis of Rock Pore Structure in 3-D Computed Microtomography Images (2005).
  3. Acknowledgments: S.G. is supported by the Statoil Fellows Program at The University of Texas at Austin. M.A.H. and J.E.G. were partially supported by NSF grants EAR CMG-1025321 and EAR-1348050, respectively. Imaging was performed at the High-Resolution X-ray Computed Tomography Facility at the Department of Geological Science, The University of Texas at Austin, which is partially supported by NSF grant EAR-1258878. Parts of image analysis was done on high-performance computing resources at Texas Advanced Computing Center. We are thankful to D. Ebrom, R. Hunsdale, and T. Løseth for providing field data and guiding their analysis. The manuscript also benefited from constructive comments by M. P. A. Jackson and R. A. Ketcham, as well as reviews from J. L. Urai and two anonymous reviewers. The authors are grateful to the Statoil Gulf Services LLC for granting permission to publish the field data. Other data are available in the manuscript as well as in the supplementary materials. The authors claim no conflicts of interest.
View Abstract

Navigate This Article