Spontaneous Imbibition of Brine and Oil in Gas Shales - American

May 6, 2013 - Spontaneous Imbibition of Brine and Oil in Gas Shales: Effect of. Water Adsorption and Resulting Microfractures. H. Dehghanpour,* Q. Lan...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/EF

Spontaneous Imbibition of Brine and Oil in Gas Shales: Effect of Water Adsorption and Resulting Microfractures H. Dehghanpour,* Q. Lan, Y. Saeed, H. Fei, and Z. Qi Department of Civil and Environmental Engineering, School of Mining and Petroleum Engineering, University of Alberta, Edmonton, Alberta T6G 2W2, Canada ABSTRACT: We measure spontaneous imbibition of aqueous (deionized water and KCl solutions of various concentrations) and oleic (kerosene and iso-octane) phases in several dry organic shale samples selected from two wells drilled in the Horn River basin. We find that the imbibition rate of aqueous phases is much higher than that of oleic phases even when plotted versus scaling dimensionless time, contrary to capillary-driven imbibition models. The observed difference is more pronounced in samples with higher clay content. The results suggest that the excess water intake is due to (1) the additional driving force provided through water adsorption by clay minerals, and (2) the enhancement of sample permeability through adsorptioninduced microfractures. engineering context.31,33,34 However, the role of this adsorption force on brine intake and wettability of organic shales is poorly understood. In a previous paper,35 we showed that fresh water intake of dry shale samples from the Horn River basin is much higher than their oil (kerosene) intake. This Article extends the previous work by comparing the imbibition rate of brine of various salinities and two oil samples in the similar shale samples. The rest of this Article is divided into four sections: Section II describes the materials and methodology used for each set of experiments. Section III presents and describes various plots of normalized imbibed mass versus time. Section IV uses an analytical scaling method and plots the imbibition data versus dimensionless time for consistent comparisons. Section V discusses the results and concludes the Article.

I. INTRODUCTION Recent advances in horizontal drilling and multistage hydraulic fracturing have unlocked the North American organic-rich shales for hydrocarbon recovery. During a hydraulic fracturing operation, a mixture of water and sand with a few chemical additives is injected under high pressure into the tight reservoir to create fractures (the pathways for oil and gas flow). The shale reservoir matrix is in contact with the fracturing fluid through the extensive surface area created after the fracturing operation. Imbibition of fracturing fluid into the rock matrix has been identified as a mechanism for fracturing fluid loss and reservoir damage.1−5 On the other hand, recent studies6 show that effective imbibition can increase initial gas production rate after extended shut-in of the well. Furthermore, spontaneous imbibition has been considered as an enhanced recovery method in shale oil reservoirs.7,8 Understanding and modeling the fluid− shale interaction is critical for optimizing the chemical formulation of fracturing and treatment fluids and for estimating the field-scale imbibition rate. Capillary pressure, which is a function of interfacial tension, rock wettability, and pore radius, controls spontaneous imbibition in conventional reservoir rocks9−14 and in low permeability rocks.15,16 Therefore, spontaneous imbibition has been used as a reliable technique to quantify the wettability of reservoir rocks such as sandstones and carbonates. Shale wettability has been recently measured by various techniques such as nuclear magnetic resonance (NMR), sessile drop, and spontaneous imbibition in shale plugs and crushed shale packs.17−23 However, measuring and modeling spontaneous imbibition in hydrocarbon-bearing shales is challenging because their extremely fine pores have a complex structure24−26 and can be either in organic or in nonorganic material.27−29 Although the organic part of the shale may be hydrophobic, the nonorganic part can be hydrophilic in the presence of clay minerals. Clay minerals can adsorb a considerable amount of water, which is controlled by clay chemistry and water salinity.30−32 Therefore, an organic shale is a mixture of hydrophilic and hydrophobic materials. Swelling of reactive clay minerals in shales results in wellbore instability, which has been well studied in the drilling © XXXX American Chemical Society

II. EXPERIMENTS We measured the spontaneous imbibition of deionized (DI) water, potassium chloride (KCl) solution of various concentrations, and oil (kerosene and iso-octane) in shale samples from Fort Simpson (FS), Muskwa (M), and Otter Park (OP) formations of the Horn River basin.36 A. Material. The experimental materials include shale rock samples and fluids used for imbibition tests.

Table 1. Average Values of Depth, Core Porosity, and TOC of the Five Shale Sections Used in This Study label

formation

depth (m)

ϕcore (%)

TOC (wt %)

FS UM LM UOP LOP

Fort Simpson Upper Muskwa Lower Muskwa Upper Otter Park Lower Otter Park

1755 1771 1792 2632 2640

6.5 8.4 7.1 8.7 8.7

1.73 2.25 3.91 3.01 3.01

Received: February 18, 2013 Revised: May 3, 2013

A

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

Table 2. Average Mineral Concentration (wt %) of the Five Shale Sections Determined by X-ray Diffraction label

calcite

quartz

dolomite

chlorite IIb2

Illite 1Mt

plagioclase albite

pyrite

matrix density

FS UM LM UOP LOP

0.5 ± 0.4 0 0.9 ± 0.5 4.4 ± 0.2 12.9 ± 0.4

29 ± 1.3 36.7 ± 1.2 45 ± 1.7 60.8 ± 1.2 43.6 ± 1.1

2.7 ± 0.3 5.2 ± 0.4 1.9 ± 0.3 2.6 ± 0.2 2.2 ± 0.5

6.5 ± 0.8 4.4 ± 0.4 0 0 0

55.4 ± 1.7 48.3 ± 1.5 43.0 ± 1.7 25.7 ± 1.3 33.8 ± 1.2

4.1 ± 0.5 3.6 ± 0.5 5.2 ± 0.5 3.7 ± 0.4 4.4 ± 0.4

1.7 ± 0.2 1.7 ± 0.2 4.0 ± 0.2 2.8 ± 0.1 3.2 ± 0.2

2.747 2.744 2.79 2.748 2.772

Table 3. Properties of Different Fluids Used for Imbibition Experiments at 25 °C

Table 6. Mass, Average Depth, Cross-Sectional Area, and Thickness of the Cylindrical Samples Used in Set 3 Testsa

fluid

density (g/cm3)

viscosity (cP)

surface tension (N/m)

label

kerosene iso-octane DI water 2 wt % KCl 4 wt % KCl 6 wt % KCl

0.80 0.69 1.00 1.00 1.02 1.03

1.32 0.50 0.9 0.89 0.89 0.89

28 18.8 72 72.7 73.3 73.9

UM3 UM4 UM5 UM6 LM6 LM7 LM8 UOP9 UOP10 UOP11 LOP9 LOP10 LOP11

Table 4. Mass, Average Depth, Cross-Sectional Area, and Thickness of the Cylindrical Samples Used in Set 1 Testsa

a

label

mass (g)

area (cm2)

thickness (cm)

diameter (cm)

depth (m)

FS1 FS2 UM1 UM2 LM1 LM2 LM3 LM4 LM5 UOP1 UOP2 UOP3 UOP4 UOP5 LOP1 LOP2 LOP3 LOP4 LOP5

255.2 276.2 360.7 347.2 423 282.7 416.9 341.7 397 529.5 581 335.2 592.4 558.4 335.9 416.5 235 398.3 446.9

78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5 78.5

1.34 1.34 1.86 1.77 2.23 1.59 2.2 1.85 2.04 2.74 3.00 1.7 3.09 2.91 1.21 2.36 1.21 2.29 2.37

10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10

1755 1755 1758 1758 1792 1792 1792 1792 1792 2603 2603 2603 2603 2603 2639 2639 2639 2639 2639

mass (g) area (cm2) thickness (cm) diameter (cm) depth (m) 40.6 9.82 8.79 12.82 16.35 12.84 17.46 17.46 14.73 10.99 15.22 15.07 12.82

7.6 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8 3.8

1.3 0.88 0.79 1.18 1.48 1.25 1.68 1.23 1.02 1.18 1.38 1.85 1.15

3.11 2.2 2.2 2.2 2.2 2.2 2.2 2.2 2.2 2.2 2.2 2.2 2.2

1772 1772 1772 1772 1796 1796 1796 2608 2608 2608 2642 2642 2642

a

The boundary condition in all samples is OEO. The second column gives the total mass of the sample and coating material.

Table 7. Mass, Average Depth, Cross-Sectional Area, Thickness, and Boundary Condition of the Cylindrical Samples Used in Set 4 Testsa

The boundary condition in all samples is AFO.

1. Shale Samples. A total of 54 shale samples were selected from the cores of two wells drilled in Horn River shales. The samples are classified into five sections of Fort Simpson (FS), Upper Muskwa (UM), Lower Muskwa (LM), Upper Otter Park (UOP), and Lower Otter Park (LOP). The average depth, core porosity, and total organic carbon (TOC) content of the five sections are listed in Table 1. The average concentration of different minerals determined by X-ray diffraction (XRD) analysis is given in Table 2. The petrophysical properties of the five sections have been detailed elsewhere.35

label

mass (g)

area (cm2)

thickness (cm)

diameter (cm)

depth (m)

B.C.

UM7 UM8 LM9 LM10 LM11 LM12 LM13 LM14 UOP12 UOP13 LOP12 LOP13 LOP14 LOP15 LOP16 LOP17

9.8 8.78 16.34 12.84 197.23 220.35 166.4 174.22 14.72 13.40 14.81 15.70 320.57 322.29 216.79 184.25

3.8 3.8 3.8 3.8 73.48 73.48 73.48 73.48 3.8 3.8 3.8 3.8 73.48 73.48 73.48 73.48

0.88 0.79 1.48 1.25 1.3 1.3 1.1 1.17 1.23 1.22 1.85 1.85 1.95 2.03 1.15 0.9

2.2 2.2 2.2 2.2 9.67 9.67 9.67 9.67 2.2 2.2 2.2 2.2 9.67 9.67 9.67 9.67

1772 1772 1796 1796 1796 1796 1796 1796 2608 2608 2642 2642 2642 2642 2642 2642

OEO OEO OEO OEO AFO AFO AFO AFO OEO OEO OEO OEO AFO AFO AFO AFO

a

For samples with OEO boundary condition, the second column gives the total mass of the sample and coating material.

Table 5. Mass, Average Depth, Cross-Sectional Area, and Thickness of the Cylindrical Samples Used in Set 2 Testsa

a

label

mass1 (g)

mass2 (g)

mass3 (g)

area (cm2)

thickness (cm)

diameter (cm)

depth (m)

UOP6 UOP7 UOP8 LOP6 LOP7 LOP8

476.2 433.6 465.4 416.2 390.5 394

475.3 433.4 465.2 416.2 390.4 393.4

474.5 433.4 464.4 416.3 390.7 393.7

70.66 71.83 73.48 62.46 71.83 62.46

2.5 2.4 2.4 2.6 2.5 2.5

9.48 9.56 9.67 8.91 8.81 8.91

2608 2608 2608 2642 2642 2642

The boundary condition in all samples is AFO. Subscripts 1, 2, and 3 represent the first, second, and third imbibition tests conducted on each sample. B

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

Figure 1. Brine and oil normalized imbibed mass versus time in (a) UM, (b) LM, (c) UOP, (d) LOP, and (e) FS shale samples (set 1). The clay-rich samples from UM (a) and FS (e) completely break after exposure to 2 wt % KCl solution, and therefore the corresponding imbibition profiles are incomplete. Pc ratio is (σwa cos(θw))/(σo1a cos(θo1)) that is given in Table 8 for each section. 2. Fluids. Kerosense, iso-octane, DI water, and KCl solutions of various concentrations were used for the imbibition tests. Density, viscosity, and surface tension of fluids are listed in Table 3. B. Test Procedure. A total of 69 imbibition tests are conducted, which can be categorized into four sets. In sets 1 and 2, large samples with core-size diameter (∼10 cm) are immersed in the fluid, and all sample faces are open for imbibition. This boundary condition is referred here as AFO (all faces open). In set 3, small cylindrical plugs cut from the cores are immersed in the fluid, and only one sample end is open for imbibition. This open end face is parallel to the lamination direction in all tests. The other faces are coated by epoxy. This boundary condition is referred to here as OEO (one end open). In set 4, both large and small samples are immersed in two different oils. The general test procedure includes the following steps: (1) Heat the selected sample at 100 °C in the oven for 24 h to ensure moisture evaporation.

(2) Measure the dried sample mass and bulk volume. (3) Place the sample in the imbibition cell and measure the weight gain at selected time intervals. 1. Set 1 (Parallel Imbibition Tests). Set 1 compares the mass of oil and brine of various concentrations imbibed in different shales. A total of 19 samples, with the physical properties listed in Table 4, were selected from the three formations. Each sample was immersed in oil, brine, or DI water, and the mass gain was frequently measured for 72 h that resulted in 19 imbibition curves. The pictures of the samples were taken before and after each test to observe possible physical alterations. 2. Set 2 (Sequential Imbibition Tests). Set 2 compares the mass of DI water and brines of different concentrations sequentially imbibed in identical samples. The basic test procedure is similar to that of set 1. However, for this set, three imbibition tests were conducted on each sample in the following steps: (1) Dry the selected sample and measure DI water imbibition profile. C

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

The test procedure is similar to that of set 2. One sample, with the properties given in Table 6, is sequentially immersed in 6, 4, and 2 wt % KCl brine and DI water, and the imbibition rate is measured. In contrast to set 2 tests, here we first immerse the sample in brine with high salinity (6 wt %) and finally in DI water to minimize the chance for microfracture induction in the early tests. b. Part 2 (Parallel Imbibition Tests). This part investigates the effect of KCl concentration on the imbibition rate in separate shale samples of similar size, mass, and physical properties. Ternary plugs with similar size were cut from LM, UM, UOP, and LOP core sections that resulted in the total of 12 samples with the properties given in Table 6. The three samples selected from each section were immersed in 6 wt % KCl, 4 wt % KCl, and DI water, and the mass gain was measured that resulted in a total of 12 (4 × 3) imbibition curves. 4. Set 4 (Parallel Oil Imbibition Tests). Set 4 compares the imbibition rate of kerosene and iso-octane into similar shale samples in two parts. In part 1, binary samples, with core-size diameter and similar thickness, from LM and LOP sections, with the properties given in Table 7, were immersed in kerosene and iso-octane, and the imbibition profiles were measured. In part 2, small and coated binary samples from LM, UM, LOP, and UOP, with the properties given in Table 7, were immersed in iso-octane and kerosene, and the imbibtion profiles were measured. Set 4 resulted in a total of 16 imbibition profiles. The samples boundary conditions in parts 1 and 2 were AFO and OEO, respectively.

(2) Dry the same sample and measure brine imbibition profile. (3) Dry the same sample and again measure DI water imbibition profile. We applied the above procedure on three UOP and three LOP samples for brine concentrations of 2, 4, and 6 wt %. Therefore, a total of 18 (6 × 3) spontaneous imbibition profiles were produced. The sample properties are given in Table 5. It was not possible to conduct the above test on Muskwa samples because they break after the first imbibition test as observed in set 1 results. 3. Set 3 (Comparing Brine and DI Water Imbibition Rate in Small Samples). Set 3 compares the mass of DI water and KCl brines of various concentrations in small samples. In sets 1 and 2, all faces of relatively large samples are open for imbibition, and therefore the chance of microfracture induction is relatively high. To address this problem, in set 3, we use small cylindrical plugs with only one end open for imbibition. The other faces are sealed with epoxy. Using small coated samples has the following advantages over using large and uncoated samples: (1) It is easier to select homogeneous samples of similar size and mass without initial microfractures for more accurate comparative experiments. (2) Sealing the small sample with epoxy reduces the chance for microfracture induction during the test. (3) The imbibition can be assumed one-dimensional and countercurrent, which can be modeled mathematically. The experiments in set 3 are divided into two parts: a. Part 1 (Sequential Imbibition Test). This part investigates the effect of KCl concentration on the imbibition rate in a Muskwa sample.

III. RESULTS This section presents and discusses the results of the four sets of experiments. In each subsection, the imbibition profiles are compared to identify the key factors controlling spontaneous liquid intake of dry shales. A. Set 1. Figure 1 compares the imbibition profiles of oil and brine of different salinities. Brine intake of all samples is significantly higher than their oil intake. This is more pronounced in FS and μM samples. It is well-known that the primary driving force for spontaneous imbibition is capillary pressure, which at the pore scale is approximated by ((σia cos(θi))/(Dp)), where σia, θi, and Dp are the surface tension, contact angle, and average pore diameter of the porous medium, respectively. One may argue that the observed difference is due to the lower surface tension of oil as compared to water. Therefore, we multiply the oil imbibition profiles by (σwa cos(θw))/(σoa cos(θo)), which is shown by the dashed curves in Figure 1. The approximate values of (σwa cos(θw))/(σoa cos(θo)) for each section are listed in Table 8. If capillarity is the only factor controlling the water intake, the maximum water intake should be around the dashed curve. However, the measured brine imbibition is still much above the dashed curve for all of the samples. The difference between oil and water imbibition rate is more pronounced in the dimensionless plots presented in section IV. The excess water intake relative to oil intake, which is more pronounced in clay-rich FS and UM samples, can be due to water adsorption on the surface of dehydrated clay minerals. Figure 2 shows the pictures of FS and M samples before and after 3 days of exposure to brine and kerosene. In contrast to oil,

Table 8. Measured Values of Water and Oil Contact Angles (θw and θo) and the Ratio between Water and Oil Capillary Pressures Based on Young Laplace Equationa UOP

LOP

θw (deg) θo1 (deg)

parameter

27 0

38 0

45 0

46 0

50 0

θo2 (deg)

0

0

0

0

0

(σwa cos(θw))/(σo1a cos(θo1))

2.14

1.89

1.70

1.67

1.54

(σwa cos(θw))/(σo2a cos(θo2))

3.42

3.02

2.71

2.66

2.47

a

FS

UM

LM

Subscripts o1 and o2 represent kerosense and iso-octane, respectively.

Figure 2. Pictures of Fort Simpson and Upper Muskwa samples before and after exposure to oil and 2 wt % KCl brine.

Figure 3. Pictures of Lower Muskwa and Otter Park samples before and after exposure to brine of different salinities. D

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

Figure 4. The results of sequential imbibition experiments (set 2) conducted on three Upper Otter Park samples (a, c, and e) and three Lower Otter Park samples (b, d, and f).

adsorption can still be responsible for the excess brine intake, but it is not strong enough to break some of the samples. Although high brine intake of the shale samples shown in Figure 1 indicates water adsorption by clay mineral, we do not observe a consistent dependence of imbibition rate on KCl concentration. This might be due to the difference in size and mass of the samples used in set 1 tests. To address this problem, we conducted set 2 tests where the imbibition masses of water and brine in the identical samples are measured and compared. B. Set 2. In set 2, each sample is sequentially immersed in DI water, KCl solution of a certain salinity, and again DI water. The test was conducted on three samples from UOP and three samples from LOP, which resulted in the total of 18 imbibition profiles. Each subfigure in Figure 4 compares the imbibition profiles of DI water, brine, and again DI water sequentially

brine alters the two samples physically, which indicates the water adsorption by clay minerals present in the two samples. Figure 3 shows the pictures of M and OP samples before the imbibition test and after 3 days of exposure to 2 and 6 wt % brines. Brine does not destroy the OP samples, although some microfractures are induced, which will be discussed later. Increasing the brine salinity from 2 to 6 wt % reduces the degree of physical alteration of M samples, which indicates the role of KCl in swelling inhibition.37 Interestingly, the imbibiton mass of 6 wt % KCl in LM, LOP, and UOP samples is still unexpectedly higher than that of oil in the same samples (Figure 1b−d), although the pictures of these samples shown in Figure 3 do not show a strong swelling. However, we observe some induced microfractures after brine imbibition tests that will be discussed later. Therefore, water E

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

imbibed in each sample. In each plot, circles and triangles represent the first and last test where DI water is used. Squares represent the second imbibiton test, where brine with concentration of 2, 4, or 6 wt % is used. Figure 4a, c, and e compares the mass of DI water and KCl solution with respective concentrations of 2, 4, and 6 wt %, imbibed in UOP samples. Figure 4b, d, and f compares the mass of similar fluids imbibed in LOP samples. In general, the data do not show a strong salinity effect on aqueous phase imbibition in any of the samples. Interestingly, the early time mass intake measured in the first test, where the DI water is used, is relatively lower than that measured in the other two subsequent tests. This trend indicates that in general the imbibition conductance is higher in the second and third tests. One may argue that the drying process does not completely dehydrate the shale samples. Even if some water remains in the sample, one should expect less imbibition rate in the second and third tests because hydrated clay minerals have less affinity for water adsorption. Furthermore, the equilibrium total imbibed mass generally slightly increases from the first immersion to the third immersion. This trend indicates that the accessible pore volume slightly increases after the first immersion. Although OP samples do not swell and break when exposed to water, we observed microfractures induced after the first imbibition test that is shown for example in Figure 5. Therefore, the higher

Figure 6. Comparison between the imbibition rate of DI water and KCl brine of different concentrations in one Muskwa sample (related to part 1 of set 3).

2. Part 2 (Parallel Imbibition Test). In part 2 of this set, we cut ternary samples with almost equal size from each of LM, UM, UOP, and LOP sections that resulted in the total of 12 samples. The samples selected from each section are immersed in DI water, 4 wt %, and 6 wt % KCl brines, and the imbibition profiles are measured. Each subfigure of Figure 7 compares the imbibition profiles of DI water and the two brines. Figure 7a, b, and d shows that increasing KCl concentration generally slightly decreases the imbibition rate. However, this trend is not observed in Figure 7c. One should note that the observed differences between the curves may also be due to slight differences in samples size and small-scale heterogeneity. D. Set 4. The results of kerosense and iso-octane imbibition tests will be presented in the dimensionless plots of section IV.

IV. SCALING THE IMBIBITION DATA The imbibition results presented by plotting normalized imbibed mass (w) versus time (t) are affected by (1) rock properties such as porosity (ϕ), permeability (k), and pore structure; (2) fluid properties such as viscosity (μ), surface tension (σ), and density (ρ); and (3) geometrical parameters such as size, shape, and boundary condition of the samples. Our objective in this Article is to compare consistently the imbibition rate of oil and brine of different salinities in similar shales. Various dimensionless time groups have been proposed on the basis of the diffusivity equation that governs spontaneous imbibition in porous media.9,10,38−41 The most frequently used dimensionless time (tD) for scaling spontaneous imbibition data is given by σ 1 t D = t k /ϕ μm Lc2 (1)

Figure 5. Pictures of LOP6 (a) and UOP6 (b) after imbibition tests (set 2) show water-induced cracks.

imbibition mass observed at the early time scales in the second and third tests can be explained by the presence of microfractures induced in the first test. C. Set 3. In the experiments of set 1, we observed that brine breaks the Muskwa samples. Therefore, it was impossible to conduct sequential imbibition tests (set 2) on large Muskwa samples. To address the problem of microfracture induction, and also to minimize the heterogeneity effect, small samples were used in set 3. Here, only one end of the sample is open for imbibition, and the other faces are sealed by epoxy. 1. Part 1 (Sequential Imbibition Test). In part 1 of this set, we study the effect of salinity on brine imbibition in a Muskwa sample by conducting sequential imbibition tests similar to set 2 tests. One Muskwa sample was sequentially immersed in 6, 4, and 2 wt % KCl brines and DI water, and the imbibition profiles were measured, which are compared in Figure 6. The data do not suggest any salinity effect, which confirms the results of set 2 experiments. However, in contrast to set 2 results, here the first imbibition curve is not lower than other curves at early time scales. This observation indicates the absence of microfractures in this test that can be explained by two reasons: (1) The sides and one end of the small sample are sealed by epoxy that reduces the chance for microfracture induction. (2) The sample is first exposed to 6 wt % brine as opposed to set 2 where the samples are first exposed to DI water.

where Lc is the charactristic length, which depends on sample shape, size, and boundary condition.9 μm is the geometric mean of water and oil viscosities.9,10 In the dimensionless time proposed for gas/ water/rock systems,39,40 only water viscosity is honored. Similarly, in our analyses, we use water or oil viscosity in eq 1. In Figure 8, we plot the normalized brine and oil mass imbibed in the large uncoated samples of sets 1 and 2 and 4 versus the corresponding dimensionless time. On the basis of the existing theories of the spontaneous imbibition, all of the data should sit around one curve. Interestingly, we observe that the brine data are much above the oil data. This result confirms the results of set 1 in the dimensional plots of Figure 1. The observed difference between brine and oil data can be explained by (1) water adsorption and (2) increase in sample permeability and effective fluid−rock interface due to induced microcracks that are not accounted for in the F

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

Figure 7. Comparison between imbibition rate of DI water and KCl brine of different salinities in ternary samples selected from (a) UM, (b) LM, (c) UOP, and (d) LOP sections.

dimensionless time given by eq 1 because fixed original values of ϕ, k, and Lc are used. To confirm the observations of Figure 8, we similarly plot the normalized brine and oil mass imbibed in the small coated samples of sets 3 and 4 versus the corresponding dimensionless time in Figure 9. Again, we observe that brine data are above oil data, although the separation is relatively smaller in this case. For a more consistent comparison, we plot all of the imbibition data versus dimensionless time in Figure 10. The data can be classified into three groups: The maximum values relate to brine and DI water imbibition in large uncoated samples with the boundary condition of AFO. The intermediate values relate to brine and DI water imbibition in small coated samples with the boundary condition of OEO. The minimum values relate to oil imbibition in both small and large samples. The unexpected high values of brine and DI water data relative to oil data can be explained through water adsorption by clay minerals and the resulting changes in sample properties that will be discussed later. The separation between normalized brine intake of small and large samples can be explained by two reasons: (1) The chance of microfracture generation in uncoated large samples is higher than that in small coated samples. (2) For large (core-size diameter) samples, water can imbibe through horizontal direction that is parallel to the lamination, and vertical direction that is perpendicular to the lamination. For small samples (plugs), water can only imbibe through the vertical direction that is perpendicular to the lamination direction. We have previously shown that the imbibition parallel to the lamination is faster than that perpendicular to the lamination.42

and iso-octane into three shale formations. The data were presented by plotting the normalized imbibed mass versus time in section III and dimensionless time in section IV. The key results are as follows: (1) Brine intake alters some of the shale samples, and the alteration degree depends on sample mineralogy, and to some extent on KCl concentration. (2) Brine intake of all samples is considerably higher than their oil intake. (3) Brine intake can induce microfractures, which enhance the sample permeability and, in turn, the imbibition rate. (4) Brine intake is not consistently correlated to KCl concentration. (5) Brine and oil data form two separate clusters on the plot of normalized imbibed mass versus dimensionless time. Result 1 complements our previous results on alteration of similar samples by fresh water.35 The significant brine intake of Fort Simpson and Muskwa samples results in microfracture induction and sample disintegration. However, brine intake does not result in disintegration of the Otter Park samples, although some microfractures are observed. The physical alteration degree is correlated to the amount of brine intake. By increasing the sample depth, from Fort Simpson to Muskwa, and Otter Park formations, the degree of physical alteration, and brine intake decrease. This trend can be explained by the change in clay content by increasing the depth. FS has the highest clay concentration followed by UM and LM, while OP samples have the least clay content. Adsorption of water on the surface of negatively charged clay platelets develops internal expansive stresses, and in turn expands and disintegrates the unconfined shale samples.30,37 However, the XRD results show that Illite is

V. DISCUSSION OF EXPERIMENTAL RESULTS The total of 69 tests were conducted to measure spontaneous imbibition of DI water, KCl brines of various salinities, kerosene, G

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

Figure 8. Normalized imbibed mass versus dimensionless time for large samples with AFO boundary condition used in sets 1, 2, and 4.

Figure 9. Normalized imbibed mass versus dimensionless time for small samples with OEO boundary condition used in sets 3 and 4. H

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

Figure 10. Normalized imbibed mass versus dimensionless time for both small and large samples.

the dry organic shales studied here. This contradicts the observation that oil completely spreads, while brine shows a measurable contact angle on the clean surface of all the samples. The excess driving force can be explained by the adsorption potential of the compacted clay platelets present in the dry shale samples. Simply, the negatively charged clay platelets strongly attract polar water molecules.30−32 This driving force is absent in the case of oil imbibition. Therefore, the total water flux during spontaneous imbibition in a dry organic shale consists of capillary-driven (qcap) and adsorption-driven (qads) components:

the dominant clay mineral in all of the samples, while the swelling clays such as montmorrillonite are negligible. It is generally believed that Illite having a 2:1 structure with the potassium as the interlayer cation does not swell. However, previous experiments show that water adsorption can even alter illitic shales.31 In addition, the presence of a small amount of mixed layer clay (interlayered or interstratified mixtures of Illite and montmorrillonite) may be the possible reason for the observed alteration. The hydration tendency of mixed layer clays is greater than that of Illite.30 We also observed some dependence of sample alteration on KCl concentration. For example, physical alteration of UM samples exposed to 2 wt % KCl brine is greater than that of the similar samples exposed to 6 wt % KCl. This observation can be explained by the swelling inhibition characteristics of KCl.37 Simply, increasing the salt concentration reduces the activity difference between shale water and external water, which in turn reduces the water adsorption and shale expansion.31,43 In addition, pottasium ions in brine can replace cations already present in the shale. Cations with low hydration energy such as K+ remain unhydrated and interact more strongly with the surface of clay minerals as compared to smaller cations such as Na+. The reluctance of potassium ions to hydration reduces the swelling tendency of shale and results in better screening of negatively charged clay layers.37,44 However, clay-rich FS samples completely disintegrated even when exposed to 6 wt % KCl brine. Furthermore, changing the KCl concentration does not consistently affect the imbibition rate (result 4). Result 2 indicates that the driving force and also the effective permeability for water intake are higher than that for oil intake in

qcap = D

qads = c

∂Sw k ∂P , whereD = w c ∂x μw ∂Sw

∂w ∂x

(2)

(3)

Here, qcap is related to the gradient of water saturation in space (((∂Sw)/(∂x))), while qads is related to the gradient of water mass fraction in space ((∂w)/(∂x)). D is a capillary diffusion parameter, which depends on water saturation and rock-fluid multiphase properties, while c is adsorption constant, which depends on rock structure, clay mineralogy, and fluid chemistry. For the samples used in this study, qads dominates the water intake, while qcap dominates the oil intake. The excess water intake can also be attributed to the enhancement of sample permeability due to water-induced microfracture. The adsorption of water by clay platelets induces large internal stresses in confined samples or expands the unconfined samples.31 I

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

On one hand, the imbibed water can damage the reservoir through various mechanisms.3−5 On the other hand, we observe that strong water adsorption by the shale samples can induce microfractures that enhance the total rock permeability. This spontaneous permeability enhancement can enhance gas recovery rate from low permeability shale reservoirs.7,52 In addition, rapid imbibition of water into the matrix helps to clean up water in fractures that in turn enhances gas effective permeability. Therefore, extended shut-in can reduce and increase initial production rates of water and gas, respectively.6

Result 3 indicates that brine can spontaneously induce microfracture in organic shales. The induced microfractures, which are possibly the result of water adsorption, enhance the shale permeability, and in turn the imbibition rate. This phenomenon that does not occur during the oil imbibition can partly explain the high ratio of water to oil intake observed in all samples studied here. Result 4 indicates that the presence of KCl does not consistently influence the imbibition rate in organic shales. However, previous experiments reporting counter-current imbibition of brine into oil-saturated reservoir rocks45,46 show that reducing brine salinity increases water imbibition rate, and in turn oil recovery rate. Although sufficient increase of KCl concentration reduced the degree of clay swelling and imbibition rate in some of the Muskwa samples, in general we did not observe a strong dependence of imbibition rate on KCl concentration. The lack of consistent correlation between imbibition rate and KCl concentration observed in our experiments is partly because the shale samples are initially dry, and initial water saturation is zero. Furthermore, the dry samples may contain some salt from the initial formation water, and the KCl concentrations used in this study are not high enough to observe a strong salinity effect. Result 5 indicates that the observed difference between brine and oil imbibition rate is not due to the difference in viscosity, surface tension, and sample geometry. On the basis of the existing theory of capillary-dominated spontaneous imbibition, oil and brine imbibed mass should sit around one curve when plotted versus dimensionless time. The dimensionless time should to some extent account for the change in viscosity, surface tension, sample size/shape, and boundary condition. Therefore, the observed difference can be explained by the physical phenomena discussed above (results 1, 2, and 3). In other words, oil and brine imbibition data do not scale because the dimensionless time used in this study does not account for water adsorption by clay minerals and the resulting induced microfractures that increase the sample effective permeability. Furthermore, the observed scatter in brine data can be explained by the presence of induced microfractures in some samples and also small-scale heterogeneity. In conclusion, the existing scaling groups, developed for clean reservoir rocks such as sandstones and carbonates, should be modified for application in organic shales, which remains the subject of future study. A. Significance of Results. We observed significant brine intake of various dry organic shale samples. It can be argued that in situ shales may have some initial water saturation that has been shown to influence the spontaneous imbibition rate in gas/ water/rock systems.47 However, the irreducible water saturation of some gas shale27 and tight gas reservoirs4,48 is abnormally low. In other words, such reservoirs are in a state known as “subirreducible initial water saturation”, created by the excessive drying at high paleo temperatures and pressures, and the lack of sufficient water for increasing the irreducible water saturation.27 Our measurements indicate that the rock matrix of such reservoirs can adsorb the water in the surrounding fractures, and the adsorbed mass can be more than 1% of the exposed rock mass. The water imbibed into the complex and fine pore structure of shales24−26 can hardly be drained by gas due to the strong capillary pressure effect. Furthermore, it has been reported that less than 50% of the injected fracturing water is recovered during the flowback operations.27,49−51 We conclude that spontaneous imbibtion of fracturing water into the shale matrix can partly explain poor load recovery reported in the field.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Dr. Ergun Kuru for useful discussions, Dr. Dipo Omotoso for analyzing the XRD data, and the British Columbia Oil and Gas Commission for providing the core samples. This research was partly supported by Natural Science and Engineering Council of Canada, FMC Technologies, Trican Well Services, and Nexen.



REFERENCES

(1) Taylor, R. S.; Fyten, G.; O’Brien, C.; Hoch, O. Paper presented at the Petroleum Society’s 9th Canadian International Petroleum Conference, Calgary, Alberta, Canada, 2008. (2) Paktinat, J.; Pinkhouse, J.; Johnson, N.; Williams, C.; Lash, G.; Penny, G.; Goff, D. Paper SPE 104306 presented at SPE Eastern Regional Meeting, Canton, OH, 2006. (3) Bahrami, H.; Rezaee, R.; Clennell, B. J. Pet. Sci. Eng. 2012, 88−89, 100. (4) Bennion, D. B.; Thomas, F. B. J. Energy Resour. Technol. 2005, 127, 240. (5) Shaoul, J.; van Zelm, L.; de Pater, C. SPE Prod. Oper. 2011, 26, 388. (6) Cheng, Y. J. Can. Pet. Technol. 2012, 51, 143. (7) Wang, D.; Butler, R.; Liu, H.; Ahmed, S. SPE Reservoir Eval. Eng. 2011, 14, 485. (8) Wang, D.; Butler, R.; Zhang, J.; Seright, R. SPE Reservoir Eval. Eng. 2012, 15, 695. (9) Zhang, X.; Morrow, N. R.; Shouxiang Ma, S. SPE Reservoir Eng. 1996, 11, 280. (10) Ma, S.; Morrow, N. R.; Zhang, X. J. Pet. Sci. Eng. 1997, 18, 165. (11) Babadagli, T. SPE J. 2001, 6, 465. (12) Morrow, N. R.; Masonb, G. Curr. Opin. Colloid Interface Sci. 2001, 6, 321. (13) Cai, J.; Yu, B.; Zou, M.; Luo, L. Energy Fuels 2010, 24, 1860. (14) Cai, J.; Hu, X.; Standnes, D. C.; You, L. Colloids Surf., A 2012, 414, 228. (15) Zhou, D.; Jia, L.; Kamath, J.; Kovscek, A. J. Pet. Sci. Eng. 2002, 33, 61. (16) Takahashi, S.; Kovscek, A. J. Pet. Sci. Eng. 2010, 71, 47. (17) Pagels, M.; Hinkel, J. J.; Willberg, D. Paper SPE 151832 presented at SPE International Symposium and Exhibition on Formation Damage Control, Lafayette, LA, 2012. (18) Roychaudhuri, B.; Tsotsis, T.; Jessen, K. Paper SPE 147652 presented at SPE Annual Technical Conference and Exhibition, Denver, CO, 2011. (19) Borysenkoa, A.; Clennellb, B.; Sedeva, R.; Ralstona, J.; Ravenc, M.; Dewhurstb, D.; Paper presented at the International Symposium of the Society of Core Analysts, Trondheim, Norway, 2006. (20) Byrnes, A. P. Paper presented at the Technical Workshops for the Hydraulic Fracturing Study: Fate and Transport, United States Environmental Protection Agency, Washington, DC, 2011. J

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX

Energy & Fuels

Article

(21) Odusina, E.; Sondergeld, C.; Rai, C. Paper SPE 147371 presented at the Canadian Unconventional Resources Conference, Calgary, Alberta, Canada, 2011. (22) Sulucarnain, I.; Sondergeld, C. H.; Rai, C. S. Paper SPE 162236 presented at the SPE Canadian Unconventional Resources Conference, Calgary, Alberta, Canada, 2012. (23) Borysenko, A.; Clennell, B.; Sedev, R.; Burgar, I.; Ralston, J.; Raven, M.; Dewhurst, D.; Liu, K. J. Geophys. Res. 2009, 114, B07202. (24) Bai, B.; Elgmati, M.; Zhang, H.; Weia, M. Fuel 2012, 105, 645. (25) Clarkson, C.; Wood, J.; Burgis, S.; Aquino, S.; Freeman, M. SPE Reservoir Eval. Eng. 2012, 15. (26) Clarkson, C.; Solano, N.; Bustin, R.; Bustin, A.; Chalmers, G.; He, L.; Melnichenko, Y.; Radliński, A.; Blachd, T. Fuel 2013, 103, 606. (27) Wang, F. P.; Reed, R. M. Paper SPE 124253 presented at the Annual Technical Conference and Exhibition, New Orleans, LA, 2009. (28) Sondergeld, C.; Ambrose, R.; Rai, C.; Moncrie, J. Paper SPE 131771 presented at SPE Unconventional Gas Conference, Pittsburgh, PA, 2010. (29) Curtis, M.; Ambrose, R.; Sondergeld, C.; Rai, C. Paper SPE 137693 presented at the Canadian Unconventional Resources and International Petroleum Conference, Calgary, Alberta, Canada, 2010. (30) Hensen, E. J. M.; Smit, B. J. Phys. Chem. 2002, 106, 12664. (31) Chenevert, M. E. J. Pet. Technol. 1970, 22, 1141. (32) Fripiat, J.; Letellier, M.; Levitz, P. Philos. Trans. R. Soc., A 1984, 311, 287. (33) Osisanya, S. O.; Chenevert, M. E. Paper 94-20 presented at SPE Annual Technical Meeting, Calgary, Alberta, 1994. (34) Osisanya, S. O. J. Can. Pet. Technol. 1995, 34, 50. (35) Dehghanpour, H.; Zubair, H. A.; Chhabra, A.; Ullah, A. Energy Fuels 2012, 26, 5750. (36) Ross, D. J. K.; Bustin, R. M. AAPG Bull. 2008, 92, 87. (37) Steiger, R. P. J. Pet. Technol. 1982, 34, 1661. (38) Schmid, K. S.; Geiger, S. Water Resour. Res. 2012, 48, W03507. (39) Handy, L. Pet. Trans. AIME 1960, 219, 75. (40) Li, K.; Horne, R. N. SPE J. 2004, 9, 322. (41) Li, K.; Horne, R. N. SPE Reservoir Eval. Eng. 2006, 9, 251−258. (42) Makhanov, K.; Dehghanpour, H.; Kuru, E. Paper SPE 162650 presented at the SPE Canadian Unconventional Resources Conference, Calgary, Alberta, Canada, 2012. (43) Chenvert, M. E. Paper presented at ISRM International Symposium, Pau, France, 1989. (44) Boek, E. S.; Coveney, P. V.; Skipper, N. T. J. Am. Chem. Soc. 1995, 117, 12608. (45) Morrow, N. R.; Tang, G.; Valat, M.; Xie, X. J. Pet. Sci. Eng. 1998, 20, 267. (46) Tang, G.; Morrow, N. SPE Reservoir Eng. 1997, 12, 269. (47) Li, K.; Chow, K.; Horne, R. N. SPE Reservoir Eval. Eng. 2006, 9, 295. (48) Newsham, K. E.; Rushing, J. A. Paper SPE 75710 presented at the 2002 SPE Gas Technology Symposium, Calgary, Alberta, Canada, 2002. (49) Soeder, D. J. Phys. Today 2011, 64, 8. (50) Asadi, M.; Woodroof, R. A.; Himes, R. E. SPE Prod. Oper. 2008, 23, 147. (51) Pagels, M.; Hinkel, J. J.; Willberg, D. M. Paper SPE 151729 presented at SPE International Symposium and Exhibition on Formation Damage Control, Lafayette, LA, 2012. (52) King, G. E. Paper SPE 152596 presented at the SPE Hydraulic Fracturing Technology Conference held in The Woodlands, TX, 2012.

K

dx.doi.org/10.1021/ef4002814 | Energy Fuels XXXX, XXX, XXX−XXX