Permeability Evolution of Shear Failing Chalk Cores under Thermochemical Influence

Development of petroleum reservoirs, including primary depletion of the pore pressure and repressurization during water injection naturally, leads to changes in effective stresses of the formations. These changes impose mechanical deformation of the rock mass with subsequent altering of its petrophysical properties. Besides mechanical compaction, chalk reservoirs on the Norwegian Continental Shelf also seem susceptible to mineralogical and textural changes as an effect of the injecting fluid’s chemical composition and temperature. Understanding such chemical and thermal effects and how they interplay with the mechanical response to changes in effective stresses could contribute to improved prediction of permeability development during field life. This article presents results from mechanical testing of chalk cores of medium-porosity (32%) outcrop chalk (Niobrara Formation, Kansas) in triaxial cells. The experimental setup allows systematic combinations of fluctuating deviatoric stress, temperature (50 and 130 °C), and injecting fluid (calcite-equilibrated sodium chloride, calcite-equilibrated sodium sulfate, and reactive synthetic seawater) intended to replicate in situ processes, relevant to the North Sea chalk reservoirs. Deviatoric loading above yield resulted in a shear failure with a steeply dipping fracture of the core and a simultaneous increase in permeability. This occurred regardless of the brine composition. The second and third deviatoric loadings above yield did not have the same strong effect on permeability. During creep and unloading, the permeability changes were minor such that the end permeability remained higher than the initial values. However, sodium sulfate-injected cores retained most of the permeability gain after shear fracturing compared to sodium chloride and synthetic seawater series at both temperatures. Synthetic seawater-injected cores registered the most permeability loss compared to the other brines at 130 °C. The results indicate that repulsive forces generated by sulfate adsorption contribute to maintain the fracture permeability.


INTRODUCTION
Water-related improved oil recovery (IOR) techniques have a documented effect on hydrocarbon production in naturally fractured chalk reservoirs, and the success story of seawater injection at the Ekofisk field (North Sea, Norway) is closely linked to fracture permeability. With the matrix permeability evaluated in the range of 1−5 mD, the Ekofisk field chalk accommodates a natural fracture system that enhances the total reservoir permeability by a factor of 50. 1 After several years of primary oil recovery by pore pressure depletion, the effective stress (overburden minus pore pressure) increased, causing the compaction mechanisms of the reservoir chalk to resume. This did not only lead to severe seafloor subsidence and challenges for the production facilities but also lead to significant changes in the physical and mechanical properties of the rock. Yet, despite years of compaction, loss of permeability was not detected. 2,3 Sulak 2 stated that, even though matrix permeability declined with matrix compaction, the effect was of no apparent consequence to the effective permeability. This seeming paradox indicated that fracture permeability, in another order of magnitude compared to the matrix permeability, dominates the effective permeability evolution. Additionally, Teufel 3 pointed out that the deviatoric nature of the reservoir stress state seems to govern the permeability of fractures such that steeply dipping fractures aligned with the maximum horizontal stress, as seen at Ekofisk, will suffer the least permeability loss. The study also underlined the different permeability behaviors under hydrostatic conditions, where permeability declines steadily with increasing hydrostatic stress.
This behavior is clearly demonstrated 4−7 and does not capture permeability changes under deviatoric stresses, typical for reservoirs.
Stress anisotropy was also one of the reasons why field-wide implementation of seawater injection at the Ekofisk field in 1987 was remarkably successful. Teufel and Rhett 8 showed that decreasing effective mean stress by water injection while maintaining a large shear stress leads to increasing fracture density and fracture surface area and to significantly increased reservoir permeability.
Maintaining high fracture permeability, on the other hand, is not without challenges as production-related changes such as primary depletion of the pore pressure and repressurization during water injection continuously alter the effective stresses of the successions. 9,10 Such stress history affects the physical properties of porous rocks and their mechanical strength, 11,12 generating a complex porosity−permeability behavior in porous rocks. Studies of fracture permeability response to fluctuating effective stress 13−15 in porous rocks show less permeability changes with each stress cycle, and in any case, the loss of permeability seems irreversible. However, these studies only focus on stress-related permeability and do not incorporate other permeability-altering parameters, specific to carbonates. Ekofisk field observations of continued high compaction rates in certain areas even after repressurization and at constant pressure indicated the activity of other compaction mechanisms than only increasing effective stress and identified the chalk contact with nonequilibrium cold seawater as a likely cause. 16 Field observations and laboratory studies remark that the temperature of the injecting fluid also contributes to continuous changes in the reservoir stress state; cold injecting fluid cools down the rock, causing it to contract and leading to a decrease in effective stress and increased permeability. 8,17 In a study of mechanical behavior of chalk cores exposed to temperature cycling, Voake et al. 18 observed more irreversible strain in chalk cores that experienced temperature fluctuations compared to cores tested at constant temperature.
Additionally, the fluid temperature plays an important role in how the fluid interacts with the reservoir rock. Chalk reservoirs found on the Norwegian Continental Shelf seem susceptible to mineralogical and textural changes, and their mechanical stability is dependent on the injected fluid's chemical composition and temperature. 7,19−26 Studies show that high temperature enhances the rock−fluid interactions and their effects are more obvious than at lower temperature. Particularly, ions such as Mg 2+ and SO 4 2− , which are present in seawater, are responsible for the main chemical and mineralogical alterations in chalk (adsorption, calcite dissolution, and precipitation of new minerals). While seawater-like brines injected at increasing temperatures have a positive effect on the IOR potential, 27 they also reduce the mechanical strength of chalk, affect its elastic properties, 28 and lead to precipitation of permeability-inhibiting minerals such as anhydrite and gypsum. 23 In the above review, most of the research has been conducted on whole cores to study the effect of systematically changing effective stress, temperature, and brine composition and see the resulting impact in terms of rock−fluid interactions, compaction, porosity, and permeability. Kallesten et al. 29 modeled chemical creep compaction in a fractured core subject to hydrostatic stresses but concluded that more experimental data were needed to characterize the fracture. A limited number of studies have considered fractured cores, deviatoric stress, or doing repeated stress cycles. Unique to the present study is that we generate fractured cores under deviatoric stress conditions and conduct repeated stress cycles while flooding inert or reactive brines at different temperatures, relevant to the chalk fields on the Norwegian Continental Shelf. The main aim is to see how different experimental conditions under stress cycles are able to affect the permeability of the system by investigating the interplay between varying parameters (stress state, temperature, and brine chemistry) and their combined effect on permeability evolution in chalk at such conditions. The test setup is designed to replicate the production-related dynamics of a chalk reservoir by considering shear-fractured outcrop chalk cores exposed to cyclic deviatoric stress states while systematically changing either the test temperature or the injected brine. In this way, different combinations of the experimental setup highlight the individual contribution of temperature, brine chemistry, and cyclic deviatoric stress on permeability evolution. The results will help answer the following: • How do deviatoric stress cycles affect permeability?
• Is there any difference in permeability evolution when superimposing temperature or chemistry changes to the deviatoric stress cycles? • Are these individual effects concurring or competing?
A better understanding of coexisting factors that affect reservoir permeability could bridge the gap toward an improved prediction of permeability development during field life.

EXPERIMENTAL PROCEDURES AND METHODS
This article presents results from mechanical testing of 12 chalk cores in triaxial cells. Such tests allow systematic changes of stress, temperature, and injecting fluid, intended to replicate in situ processes at reservoir conditions.
2.1. Sample Set. The sample set consists of 12 outcrop chalk cores (Niobrara Formation, Kansas/Utah) of medium porosity for a chalk (32%, ±1%). Experimental and theoretical studies demonstrate that mechanical strength of chalk is orientationdependent and that chalk deformation behavior differs when main stress is applied parallel or perpendicular to bedding. 30,31 To ensure comparability between the tests, all cores are drilled from the same block collection and in the same direction. Further, the cores are cut to an average length of 75 mm and lathed to a diameter of 38.1 mm. The length−diameter ratio of approximately 2 should accommodate the steeply dipping shear fracture plane typical for deviatoric loading. 32 2.2. Injecting Brines. The injecting brines define three test series: calcite-equilibrated sodium chloride (CE-NaCl), calciteequilibrated sodium sulfate (CE-Na 2 SO 4 ), and reactive synthetic seawater (SSW). By equilibrating the two former brines with calcite, no mineralogical interactions are expected, although surface reactions would be more likely. CE-NaCl is inert in all practical aspects and serves as a standard in comparison to CE-Na 2 SO 4 and SSW brines. The concentrations of NaCl and Na 2 SO 4 are designed to match the ionic strength of SSW ( Table 1). The injection rate is 2 pore volumes (PVs)/day. After testing, the cores are flooded with 4 PVs distilled water (DW) to avoid salt precipitation.
2.3. Porosity and Permeability Calculation. For porosity calculation, the dry shaped cores were saturated with distilled water under vacuum conditions. The initial porosity (φ) is given by the ratio between pore volume (the difference in initial saturated and dry masses (M sat and M dry , respectively) divided by the density of distilled water (ρ dw )) and initial bulk volume (V bulk, i ) The effective permeability (k) calculation is based on Darcy's law, 33 assuming a steady laminar fluid flow and symmetric axial and radial deformations where μ is the fluid viscosity as a function of salinity and temperature (cP; after El-Dessouky and Ettouny 34 ), L i is the initial length of the core (cm), ΔL is the change in core length (cm), Q is the flow rate (cm 3 /s), D i is the initial diameter of the core (cm), ΔD is the mean change in core diameter (cm), and ΔP is the pressure drop over the core during flooding (atm). Uncertainties involved in the permeability calculation include fluid viscosity μ ± 2%, change in core length ΔL ± 0.7%, change in diameter ΔD ± 1%, flooding rate Q ± 2%, and differential pressure ΔP ± 0.075%. Permeability uncertainty Δk was estimated at ±3% by applying error estimation method shown in eq 3 where x, ..., z are the measured values of the parameters with their respective Δx, ..., Δz uncertainty. 6 A summary of the core properties before the mechanical testing (length, diameter, dry mass, saturated mass, porosity, permeability, and specific surface area) is listed in Table 2 and marked with index i (initial).

Triaxial Cell
Tests. The triaxial cell is equipped with an outer heating jacket and a regulating system (Omron E5CN) with precise proportional integral derivative (PID) temperature control (±1.0°C). The system includes two Quizix QX-2000HC pumps that control the axial and confining pressures independently and a fluid injection pump (Gilson 307HPLC) as well as a backpressure regulator that controls the pore pressure.
The cores are saturated with distilled water prior to cell mounting; they are isolated from the oil bath in the confining chamber by a heat shrinkage sleeve (fluorinated ethylene propylene, FEP; 0.5 mm in wall thickness). An extensometer surrounds the core at mid-length and measures the changes in diameter throughout the test (radial strain). Changes in the cores' length (axial strain) are monitored by an external axial linear variable displacement transducer (LVDT) placed on top of the cell piston.
Each series is performed at 50°C, below the threshold for sulfate effects on chalk 35 and at 130°C (reservoir temperature at Ekofisk) to trace the thermochemical effects on shear failing chalk cores injected with various brines. The low temperature tests require an initial heating to 130°C to make sure that the shrinking sleeve is tight and uniform around the core, comparable to the tests at high temperature. During this stage, the injecting fluid is distilled water and brine injection only starting after the system had cooled down to 50°C to avoid rock−brine interactions at high temperature.
Throughout the test, the confining pressure and pore pressure are constant at 1.2 and 0.7 MPa, respectively. In this way, the changes in effective stresses are a function of the increasing or decreasing axial stress. The cell piston is lowered carefully, and after core contact, it applies a pressure (0.5 MPa) that slightly overcomes the piston friction but causes negligible core deformation. The deviatoric tests start only when the triaxial cell has reached the right test temperature, the cores are brineflooded with at least 2 PVs, the differential pressure is stable, and the system is in equilibrium (axial, radial strain rates close to 0).
2.5. Stress Cycles. The cores undergo the same stress sequence three times. The first step is axial loading at a constant injection rate (0.01 mL/min) above yield (defined by initiation of nonlinear stress−strain relationship) until shear failure (fracture formation) at which the value of axial stress drops. Immediately after, the axial pump is set to apply a constant

ACS Omega
http://pubs.acs.org/journal/acsodf Article pressure, slightly below the failure point, allowing the core to deform (creep) over 3 days. The axial pressure then returns to the starting point (0.5 MPa) at the same rate as loading. The second and third stress cycles, following the same procedure as the first, start after the system had reached equilibrium. In the second and third cycles, the creep phases are shorter (1 day).

RESULTS
Deviatoric loading above yield resulted without exception in a steeply dipping shear fracture along the core (Figure 1, left). This is typical for compressive triaxial tests at low confining pressure, which allows the core to expand radially and facilitates well-organized shear failure between grains. 36 Figure 1 shows experiment KE20 (CE-NaCl series, 50°C) to exemplify the typical permeability response (green curve) to first deviatoric loading above yield (red curve). Deviatoric loading within the yield curve (where the stress−strain relationship is linear) generally caused permeability decreases in all cores, in average, of 10% in the first cycle and 5% during second and third cycles compared to permeability values at the beginning of the respective cycles. The shear fracture event, recognized by a drop in axial stress, initiated an abrupt increase in permeability, and generally, this occurred during the first loading. However, although all cores experienced the same deformation during the first loading, the mechanical strength of the cores and the degree of permeability rise differed. These differences should indicate the roles of injecting brine composition or temperature, the only two differing parameters in the test setup.
3.1. CE-NaCl Injection. Figure 2 shows permeability versus axial stress during deviatoric loading (left column) and permeability evolution in time during creep (right column) for the CE-NaCl test series. A clear distinction is seen between low temperature (blue curves) and high temperature (red curves) tests. As seen in Figure 2a, both cores tested at 50°C (KE20 and KE4, blue lines) fractured at lower axial stress (11)(12) than those tested at 130°C (red lines, KE44 and KE45, 15−17 MPa). The shear fracture caused a drop in axial stress together with a clear rise in permeability (45% in average) and occurred mainly in the first loading. Test KE4 (50°C), however, experienced a brief, unexpected pressure fluctuation in the beginning of the first loading at 5 MPa (Figure 2a), which might be related to an instrument error. This may have temporarily stabilized the core so that, at the end of the first deviatoric loading, the drop in axial stress from 11 to 9 MPa was an effect of microcracking, a precursor to fracturing. 37 A more pronounced shear fracture developed during the second loading cycle, leading to 1.75% axial deformation and 0.63% radial expansion, together with permeability rise by 36% (Figure 3b, blue triangles).
The mechanical strength decreased in all tests in the second loading (Figure 2b,c). The maximum axial stress was 3−4 MPa less compared to the first loading at 130°C, while at 50°C, the difference is approximately 5 MPa. Permeability increased in all tests after second loading but, unlike the first loading, by only less than 10% (excluding KE4).
Further, after the third loading, permeability decreased by 10% at 50°C. Figure 2c (blue lines) shows that, after an initial decline during loading, the measured permeability remained constant. At 130°C, on the other hand, it remained almost unchanged throughout the loading (Figure 2c, red lines).
The axial strain was the highest when the fracture occurred and it varied between 0.75 and 1.75% (Table 3), after which the cores experienced an average of 0.40% axial deformation at both temperatures. Radial strain does not correlate with the change in permeability so that more radial expansion does not necessarily lead to higher permeability.
Permeability seems unaffected by the degree of creep. Axial deformation during creep is a function of the applied axial stress, a value that was subjectively chosen for each core at each creep phase (Table 4). For example, in the KE44 test, the axial stress was 30% higher than KE45 at the same temperature, and it deformed three times more axially and six times more radially than KE45 during the first creep phase. Yet, permeability in both these tests is almost unchanged throughout the entire creep time (Figure 2d, red lines). The same trend is observed also in the creep phases 2 and 3, where the permeability does not register any significant changes. The two low temperature tests (KE4 and KE20), however, seem to decrease slightly during the first creep (Figure 2b, blue lines), but a direct comparison between the two tests is difficult because of the permeability fluctuations in KE4, which had not yet fractured properly at this point. During the second and third creep, on the other hand, the permeability only decreased somewhat in KE4 following the delayed clear shear fracture, but KE20 permeability stayed constant, similar to the high temperature tests (Figure 2e,f).
3.2. CE-Na 2 SO 4 Fluid Injection. The results from the CE-Na 2 SO 4 tests ( Figure 3) show a similar permeability pattern as the CE-NaCl series. Permeability declined during the axial loading and increased with fracture formation in the first cycle (Figure 3a). Unlike the CE-NaCl tests, there is no clear maximum axial stress correlation between the high and low test temperatures. During the first loading cycle, the maximum axial stress was similar in three of the four tests, between 10 and 12 MPa, while KE22 withheld a maximum axial stress of 14 MPa. However, the high temperature tests registered the highest permeability increase when the shear fracture occurred, approximately double compared to the intact core permeability. At lower temperature, permeability increased with 50−80%. When comparing the first loading of the duplicate tests at each temperature, the magnitude of permeability rise correlates with the axial and radial strain: the higher the strain, the higher the permeability rise (Table 5). Left: Typical example of axial stress−strain relationship (red curve) and permeability evolution (green curve) during first deviatoric loading above yield, here, data from the KE20 experiment (CE-NaCl series, 50°C ). Permeability generally decreases within the yield curve; the shear fracture event causes a concomitant drop in axial stress and an increase in permeability.
The mechanical strength decreased in all tests from the first cycle to the second and third cycles, regardless of test temperature (Figure 3b,c). Permeability increased in all experiments after the second loading cycle, this time by a factor of 2−4 more in the low temperature tests compared to the high temperature. This also corresponds to higher axial strain (0.62− 0.98%) and radial strain (−0.2% in average) at low temperature compared to high temperature tests, which deformed in an average of 0.25% axially and below −0.1% radially at the same stage ( Table 5).
The stress conditions in all three creep phases were similar (Table 6), and the cores sustained approximately 7 MPa constant axial stress and registered minor radial strain in all tests (between −0.02 and −0.06%). The axial strain varied somewhat with the temperature in the first creep phase so that, at low temperature, the axial strain was approximately double compared to the axial strain at 130°C, but this had no clear effect on permeability.
The permeability generally declined after the first creep, mostly during the first creep day, after which it stabilized until the end of the creep phase (Figure 3d). Creep phases 2 and 3 (Figure 3e,f) induced minimal changes in permeability so that most of the permeability gain during the first loading cycle was retained throughout the test.
3.3. Synthetic Seawater Injection. The SSW-injected cores show similar permeability behavior as the CE-NaCl and CE-Na 2 SO 4 series. As in the CE-Na 2 SO 4 series, temperature did not play a distinguishing role in the maximum axial stress during the first loading cycle. Three of the tests fractured after a peak of approximately 12 MPa (Figure 4a), although performed at a different temperature. Core KE73 (130°C, red diamonds) fractured after a maximum axial stress of 14 MPa. Although, generally, permeability increased after the first loading in all tests, there is no clear pattern in the rise magnitude ( Table 7).
The strain−permeability correlation is seen here as well, where tests that showed higher permeability increase also experienced higher radial strain (Table 7). Core KE51 stands out with the highest permeability rise after the first loading (126%).
SSW flooding weakened the chalk cores similarly at both temperatures, showing an approximately 3−5 MPa decline in maximum axial stresses between the first loading and the subsequent loading phases. As in the CE-NaCl and CE-Na 2 SO 4 Figure 2. Permeability response to (a−c) axial stress and (d−f) creep in CE-NaCl experiments at 50°C (blue lines) and 130°C (red lines); the highest permeability increase is associated with core fracturing (a, and for KE4, b) after which no notable permeability change was observed. Creep under deviatoric conditions had a minimal effect on permeability evolution. The test temperature affected the mechanical strength during loading but did not play a decisive role in permeability during creep.

ACS Omega
http://pubs.acs.org/journal/acsodf Article series, permeability did not change significantly during deviatoric loadings 2 and 3. Creep permeability trends were nearly parallel in all SSW experiments (Figure 4d−f), generally decreasing. End perme-ability was lowest in core KE51 (130°C, Figure 4f, red bullets), which, throughout the test, lost more than half of the permeability gain after fracturing in loading 1. Core KE47 (50°C ) deformed at a high rate in all three creep cycles (Table 8),

ACS Omega
http://pubs.acs.org/journal/acsodf Article and the test was ultimately interrupted as the core collapsed. Yet, despite the accelerating creep conditions, the permeability decline rate is comparable to the other tests in this series.

DISCUSSION
The main drive in permeability change is the shear fracturing.
Occurring mainly in the first deviatoric loading, the fracture serves as a "highway" for brine flow that induces a simultaneous permeability rise. The fracture dip angle is high and close to parallel to fluid flow due to the low confining pressure. The magnitude of permeability increase at this stage generally determines the cores' end permeability. Figure 5 graphically displays the cumulative permeability evolution in all experiments, taking into account the unloading sequence as well. The first data point is the beginning of the first deviatoric cycle, and each line segment represents the permeability behavior during consecutive stress states. At both temperatures, end permeability of CE-Na 2 SO 4 -flooded cores is the highest. Megawati et al. 35 suggests that sodium sulfate brine injection at 130°C does not cause new mineral precipitation such as anhydrite, but sulfate adsorption on chalk grains will change the chalk's surface charge and cause a disjoining force close to granular contacts. The disjoining force following sulfate adsorption may in fact explain why all CE-Na 2 SO 4 -flooded cores registered the highest permeability increase ( Figure 5, blue lines).
On the other hand, SSW-flooded cores (yellow lines) gained least permeability at 130°C, approximately half compared to CE-Na 2 SO 4 series at the same temperature. That is most likely because SSW poses a different and more complex thermochemical scenario. According to previously mentioned studies, 23,30 SSW injection at 130°C leads to both sulfate adsorption and mineralogical changes. Ca 2+ substitution with Mg 2+ and anhydrite (CaSO 4 ) precipitation contribute further to chalk weakening and alter fracture permeability. Particularly, anhydrite precipitation is a common permeability inhibitor, and Figure 5 shows that end permeability of both SSW-injected cores at 130°C (yellow lines) lies below the other two series (CE-Na 2 SO 4 , blue lines, and CE-NaCl, green lines).
Another notable aspect from Figure 5 is that the degree of permeability rise related to fracturing (load 1) correlates with test temperature: overall, permeability increased 70%, in average, at the beginning of the tests at high temperature and only 40% average at low temperature. There was also higher variation in permeability rise among high temperature tests, compared to low temperature tests, where four out of six tests had an almost identical permeability increase. Additionally, end

ACS Omega
http://pubs.acs.org/journal/acsodf Article permeability at 130°C increased, in average, by a factor of 1.7, while at 50°C, the average increasing factor was 1.3. Both stress and thermochemical test conditions impact the cores' mechanical strength to different degrees, generally decreasing throughout the test. Repetitive deviatoric loadings following the fracture formation do not seem to affect permeability as much as the shear fracturing event. This is in agreement with other studies 13,14 that, although performed under different thermochemical conditions than the present study, report that a second deviatoric stress cycle had a less effect on permeability evolution compared to the first cycle. The cycles naturally weaken the cores, with each loading−unloading adding more fatigue and deformation to the rock and altering its elastic properties. 11 The effect of temperature and injecting brine on the cores' mechanical strength is shown in Table 9.
The CE-Na 2 SO 4 -injected cores at 50°C registered the least mechanical strength decline (2−3 MPa) compared to the SSW (5 MPa) and CE-NaCl (4 MPa) series. At 130°C, however, the CE-NaCl-injected cores retain most of their initial strength, while CE-Na 2 SO 4 and, particularly, SSW series become weaker. This observation is in agreement with the previous studies 23, 30 on SSW influence on the mechanical stability of chalk.
Additionally, higher temperature intensifies the rock interactions with the injection brine. 7,19,23,24,30,35 Creep conditions did not seem to override the permeability gain during fracturing. This is in agreement with other studies 3,6 that observed a low permeability decline in deviatoric conditions. Table 10 summarizes the key changes recorded during each experiment. The net axial and radial strains are calculated as the percentage axial compaction and radial expansion, respectively, at the end of the test relative to the initial length and diameter respectively shown in Table 2.
In CE-NaCl and CE-Na 2 SO 4 test series, the low temperature tests suffered more axial compaction than the high temperature tests. This correlates well with the end permeability: permeability increased only slightly in CE-NaCl-flooded cores at 50°C (0 and 21%) but more clearly at higher temperature (66%). A similar pattern appears in CE-Na 2 SO 4 tests. Low temperature cores experienced more compaction (2.5 and 2.8%) and less permeability increase (41 and 73%) than the cores flooded with CE-Na 2 SO 4 at 130°C (1.6% axial strain and a permeability increase by 95 and 122%). This indicates that, when flooding equilibrium brines that do not alter chalk mineralogy, brine temperature can play a significant role in mechanical compaction (strain), a key permeability-controlling factor.
This correlation is not clear in SSW series. Experiment KE47 failed before all stress sequences were complete so that the high compaction (4.3%) and high end permeability (+80%) are irrelevant in this argument. At 130°C, KE51 deforms more axially (2.9%) than the second test at the same temperature (KE73, 1.3%) and end permeability is consequently lower with higher compaction, as in the previous two series.
There is little variation in net radial strain among the tests (between −0.2 and −0.8%, excluding KE47), indicating that flooding brine and temperature do not decisively affect fracture aperture, and any radial strain is rather stress-induced. Ultimately, there is no immediate correlation between the net radial strain and end permeability.

CONCLUSIONS
This study focuses on permeability evolution in fractured chalk cores exposed to cyclic deviatoric stress and thermochemical influence relevant to reservoir conditions. The test setup included three series, each with specific injecting brine: two calcite-equilibrated brines (CE-NaCl and Na 2 SO 4 ) and one

ACS Omega
http://pubs.acs.org/journal/acsodf Article nonequilibrium brine (SSW). Each test series was performed at 50 and 130°C. Such flooding experiments highlight the interplay between the parameters. The results showed that stress-induced fracturing is the main permeability drive as permeability changed decisively only as a response to fracturing of the core. The deviatoric stress state together with low confining pressure induced shear fracturing at a steep angle (over 70°), close to the flooding direction. Subsequent deviatoric loadings had a little effect on permeability in all tests, regardless of injecting brine and test temperature. During creep, permeability generally declined slightly or remained unchanged. Our results indicate that, once chalk has fractured, the effective permeability is insensitive to compaction cycles and reactive flow, both at high and low temperatures. The CE-Na 2 SO 4 test series stand out with the highest final permeability at both temperatures. This is likely a result of sulfate adsorption on the chalk grain surface creating enough disjoining force at granular contact to preserve permeability. Additionally, calcite equilibration of the brine prevented calcium displacement and, consequently, anhydrite precipitation.
If flooding CE-Na 2 SO 4 through fractured chalk seems to sustain the permeability gain related to shear fracturing, then flooding reactive brine such as SSW at a high temperature has a competing effect on permeability evolution. SSW test series registered most permeability loss at 130°C, most likely due to chemical alteration, possibly precipitation of anhydrite.
However, all cores had a positive net permeability change. Despite fracturing and exposure to different stress states, temperatures, and brine conditions, core permeability at the end of the test seems to remain within the same order of magnitude as the original value, ranging between the initial value and double of the initial value. This indicates strong insensitivity to changes in reservoir conditions.
The results are repeatable, confirming permeability behavior for this experiment setup. Future studies should investigate the effects of longer-term stress cycles or several stress cycles at actual reservoir stress conditions on permeability evolution in fractured chalk cores. Additional analyses such as scanning electron microscopy can then verify the chemical alteration caused by reactive fluid flow in chalk, as suggested by Minde et al., 38 and determine mineral alteration along the fracture or in the matrix. Although challenging to obtain, similar flooding experiments on actual reservoir chalk from the North Sea would provide important data for validating outcrop chalk test results and refining permeability models for the North Sea reservoir chalk. Especially, the importance of the adsorption mechanism to maintain fracture permeability in the field should be investigated further.