Saturated hydraulic conductivity of compacted bentonite-sand mixtures before and after gas migration in artificial seawater

Ysutk Wtne,Shingo Yokoym,Misto Shimshi,Yoichi Ymmoto,Tkhiro Goto

a Sustainable System Research Laboratory,Central Research Institute of Electric Power Industry,Chiba,Japan

b Science and Technology Department,Nuclear Waste Management Organization of Japan,Tokyo,Japan

Keywords: Bentonite Gas migration Hydraulic conductivity Seawater (SW)Self-healing

ABSTRACT To understand the self-healing property of an engineered barrier for radioactive waste disposal,the hydraulic conductivity of compacted bentonite-sand mixtures saturated with artificial seawater (SW)before and after gas migration was examined.Na-and Ca-bentonites were mixed with fine sand at a ratio of 70% bentonite in dry weight.Two aspects were considered during the experiment: the hydraulic conductivity of the specimen that was resaturated after gas migration and the distribution of water content immediately after gas migration to study gas migration pathways.The gas migrated through the entire cross-section of the specimen,and gas breakthrough occurred in the equilibrium swelling pressure range approximately.Subsequently,the gas flow rate reached a sufficient large value when the gas pressure was approximately twice the equilibrium axial pressure (the sum of swelling and confining pressures),which excluded the back pressure.Although the gas migration pathway was not visible when the specimen was observed immediately after gas migration,the water content distribution showed that several parts of the specimen with lower water content were connected in the direction of gas migration.After resaturation,the change in permeability was within a limited range-two to three times larger than that before gas migration for each type of bentonite in SW.This slight change suggests that gas migration creates a pore structure that cannot be sealed via crystalline swelling of montmorillonite in SW,even if highly compacted bentonite is used under a constant-volume condition.

1.Introduction

In the current concept for disposal of radioactive waste,compacted bentonite is used as the buffer material to inhibit nuclide migration owing to its low permeability (JNC,2000;Sellin and Leupin,2013;NUMO,2021).Hydrogen gas can be generated within the buffer material via anaerobic corrosion of the metals used for containers,water radiolysis,and microbial degradation(Miller et al.,2000;Guo and Fall 2021;NUMO,2021).If the gas generation rate exceeds the diffusion rate of the dissolved gas inside the buffer material,gas accumulates within the voids inside the buffer material until the pressure becomes sufficiently large to migrate.Numerous experimental studies have been conducted to investigate gas migration via compacted bentonite saturated with water(Liu et al.,2018,2020;Cui et al.,2019,2021a,b;Guo and Fall 2021;Zhang,2021).The gas transport mechanism of the buffer material is categorized as advection/diffusion of dissolved gas,viscocapillary two-phase flow,dilatancy-controlled flow,and macro-fracture-controlled flow(Marschall et al.,2005).It is known that gas breakthrough occurs when the gas pressure exceeds the sum of the swelling pressure and back pressure of the compacted bentonite(Horseman et al.,1999).

According to the design requirements for the buffer material(NUMO,2021),compacted bentonite must maintain low permeability over a long term,even if gas breakthrough occurs.Gasinduced pathways in the compacted bentonite are expected to heal by swelling and should not become a critical hydraulic path after resaturation.In a previous study,the hydraulic conductivity of compacted Na-bentonite saturated with deionized water (DW)before and after gas migration was measured and compared(Tanaka et al.,2010).It was reported that if the specimen was resaturated with water,there was no change in the hydraulic conductivity owing to the large gas flow rate.The effect of the interface between bentonite blocks,compacted bentonite,and host rocks on gas breakthrough has been investigated(Davy et al.,2009;Gutiérrez-Rodrigo et al.,2021;Villar et al.,2021),and the results suggest that the interfaces could be sealed against gas breakthrough via water saturation of bentonite.Thus,based on experimental studies on the gas migration of the buffer material,it is well known that the self-healing of compacted bentonites can occur under complete water saturation.However,studies have failed to address gas migration and the changes in hydraulic conductivity after gas migration in SW.The groundwater originating from SW may infiltrate the buffer material depending on the repository location.The swelling properties of bentonite and smectite decrease in saline solutions (Pusch and Karnland,1996;Komine et al.,2009;Zhu et al.,2013).Therefore,it is important to investigate the change in the saturated hydraulic conductivity of compacted bentonite after gas migration in SW in terms of self-healing.

Using gas migration and permeability tests,this study investigated the hydraulic conductivity of compacted Na-and Cabentonite-sand mixtures saturated with SW before and after gas migration.In SW,the swelling and breakthrough pressures were firstly clarified,and the breakthrough conditions were compared with those in the case of DW.Secondly,an experiment was conducted to determine the change in the hydraulic conductivity of the specimen that was resaturated after gas migration.The distribution of water content immediately after gas migration was also measured.Finally,based on these experimental results,the gas migration mechanism and the reason for the change in saturated hydraulic conductivity after gas migration in SW were discussed.

2.Test materials and specimen production

2.1.Materials and methods

The specimen used in this study included Na-bentonite (Kunigel-V1,Kunimine Industries Co.,Ltd.) mined from Yamagata Prefecture,Japan,and Ca-bentonite (Mikawa,Hojun Co.,Ltd.) mined from Niigata Prefecture,Japan (physicochemical properties are listed in Table 1).The particle density of the soil was measured using a pycnometer based on the Japanese Industrial Standard(JIS A 1202,2009),with a sample weighing 3-8 g,and a vacuum pump for degassing.The particle densities of the Na-and Ca-bentonite were 2.768 g/cm3and 2.574 g/cm3,respectively.This difference was presumably due to the crystalline/non-crystalline mineral composition.The amount of methylene blue adsorbed onto the specimen was measured using the spot method (JIS Z 2451,2019).The montmorillonite content of each bentonite material,evaluated by the ratio of exchangeable cation capacity of montmorillonite and bentonite,was 51% for Na-bentonite and 41% for Ca-bentonite(Shimbashi et al.,2020).The leachable cations,which correspond approximately to the exchangeable cations,were identified by leaching with 6%benzyl trimethylammonium solution based on the modified Steel Founders’Society of America(SFSA Designation 13T-86,1986) method (Shinoki et al.,2009).Sand was mixed with bentonite to produce the specimen.The grain size distribution of the bentonites and sand is shown in Fig.1.The laser diffraction method with the apparatus SALD-2300 (Shimadzu Co.Ltd.) was applied to the bentonites,and the sieving method (JIS A 1204,2009) was applied to the sand.

Table 1 Physicochemical properties of the bentonites and sand used in this study.

2.2.Specimen production

In a safety case regarding geological disposal in Japan (NUMO,2021),the buffer material was examined as a compacted bentonite-sand mixture,with a mixing ratio of 70%bentonite and a dry density of 1.6 g/cm3.Sand,in the mixture,improves the compressibility and manufacturability of the bentonite blocks.With the same specifications as those for the specimen in the aforementioned safety case,the bentonite mixture and the dry density were set for the experiment in this study.To compare the results of different types of bentonites,Ca-bentonite specimens were produced using the same specifications as those for Nabentonite.Based on the optimum water content of the materials,the water content was 15%for the Na-bentonite-sand mixture and 21% for the Ca-bentonite-sand mixture.

A cylindrical specimen with a diameter of 60 mm and a height of 20 mm was produced by static compaction in the test ring.The material was placed in a rigid mold and vertical pressure was gradually increased up to 2 MPa for Na-bentonite and 5 MPa for Cabentonite under compaction pressure,and the loads were maintained for 10 min.The height of the specimen reached 20 mm after the pressure was unloaded,and rebound occurred.The rebound,increase in the specimen height after unloading,was approximately 0.22-0.56 mm.

Fig.1.Grain size distribution of bentonites and sand.

3.Experimental procedure

3.1.Gas migration test with measurement of hydraulic conductivity

In this study,a gas migration test with pressure-controlled gas injection was conducted on saturated compacted bentonite-sand mixtures.The experimental apparatus for water injection and subsequent gas injection tests is shown in Fig.2.It includes a constant-volume specimen cell,a displacement gauge to confirm no change in the specimen height,two sets of load cells to measure the axial pressure,input lines for high-purity helium gas and solutions,and an output line to separately measure the gas and solution flows using burettes(Fig.2).The axial pressures in the upper and bottom sides were measured,considering that the axial pressure might differ owing to the interfacial friction and heterogeneity induced by the reorganization of clay fabric and the associated porosity (Massat et al.,2016;Rawat et al.,2019).Hydraulic conductivity was measured using an apparatus (Fig.2a).After determining the initial value of hydraulic conductivity of the specimen,the gas pressure can be applied in steps,and both the influent and effluent flow rates for the gas and solutions can be measured individually.

The apparatus was assembled as shown in Fig,2a.Thin filters with a pore size of 0.25 μm×0.075 μm were set on both faces of the specimen to prevent specimen loss during the experiment.The upper porous end piece with the piston was brought in contact with the specimen by adjusting the load frame,and a contact pressure of approximately 20 kPa was measured using the upper load cell.Subsequently,the bottom load cell responded 1-6 kPa,which corresponded to the contact pressure at the bottom.The contact pressures act as a confining pressure to the specimen.Displacement gauges were set on the piston to measure the change in specimen height during the experiment.DW or artificial SW(chemical composition listed in Table 2) was supplied from the bottom of the specimen.

Table 2 Chemical composition (mmol/L) of artificial SW used in this study.

The axial pressure,which comprised the swelling and contact pressures,was measured by the load cells at both sides of the specimen,with a precision of 1.7 kPa.The swelling pressure is defined as the pressure for restraining changes in the specimen volume during hydration and swelling.Because the axial strain was restrained,as indicated by the displacement gauge,the increment in axial pressure during hydration under atmospheric pressure can be interpreted as the swelling pressure.

Fig.2.(a) Apparatus of gas migration test and (b) Details of the porous end piece.

The upper porous end piece was separated into inner and outer parts,thus,the flow inside and outside the specimen was measured individually(Fig.2b).The interface between the specimen and the mold might be hydraulic pathways as the side wall leakage if the permeability of the specimen is significantly low.Clarifying whether the interface is a dominant gas migration pathway is important to discuss the gas breakthrough and change in hydraulic conductivity.Therefore,the outflow from the outside of the specimen was measured individually.During the experiment,the saturation degree of the specimen was estimated using the following equation based on Boyle-Charles’ law while neglecting the solubility of air (Kohno and Nishigaki,1982;Watanabe and Tanaka,2016):

whereSrdenotes the saturation degree of the specimen and is estimated by the compressibility of the pore fluid.P0is the initial absolute pressure(kPa),Vvis the volume of the void(m3),and ΔVis the absorbed water volume (m3) when the backpressure is increased by ΔP.If the specimen was partially saturated,the backpressure increased the amount of absorbed water.In this study,the estimated saturation degree was required to be more than 99%.After saturation,the equilibrium axial pressure which excluded the back pressure was measured on both sides of the specimen.

Hydraulic conductivity was measured using the falling-head test method (ASTM D5084-10,2010).The hydraulic conductivity was corrected with respect to the value at 15°C using the temperature-dependent viscosity of water for ease of data comparison (JIS A 1218,2009),because the room temperature was approximately 22°C,with a slight variation.The equivalent hydraulic conductivity(keq) was calculated as

wherekinandkoutare the hydraulic conductivities(m/s)measured by the flow rates inside and outside the specimen,respectively;andAinandAoutare the inner and outer cross-sectional areas(m2)of the specimen,respectively.Hydraulic pressure was applied to increase the hydraulic gradient without exceeding the equilibrium axial pressure.The hydraulic conductivity was considered steady if four or more consecutive hydraulic conductivity determinations fell within±50%of the mean value,based on the requirements(ASTM D5084-10,2010).

A back pressure of 300 kPa was applied,and helium gas was injected from the bottom of the specimen.The effective gas pressure (Pge) is written as follows:

wherePgis the injected gas pressure (kPa),andPbis the back pressure(kPa).The gas pressure increased at a rate of 1/3 kPa/min during the daytime,with a pressurization of 80-160 kPa/d,which is comparable to the value of 192 kPa/d applied by Tanai et al.(1999) for compacted bentonite with a height of 20 mm.The effective gas pressure at the first appearance of bubbles in the semitransparent drainage tube is called the breakthrough pressure.When the effluent gas was first observed during pressuring,the gas breakthrough might have occurred by the gas pressure,which was lower than the observed value,owing to the time required for gas to flow across the specimen(Graham et al.,2002).The value of the gas breakthrough pressure should be the value between two gas pressure intervals.Therefore,the range of possible breakthrough was considered in data analysis.If effluent gas is observed during the gas pressure is maintained,the abovementioned range will decrease.

Fig.3.Measurement of gas and solution flows in effluent side.

The effluent flow rates of the solution and gas were measured individually using burettes,as depicted in Fig.3.A constant back pressure of 300 kPa was applied at the end side,and the upper side of the single-tube burette was undrained.Thus,the effluent gas accumulated in the single-tube burette at 300 kPa.Because its inner diameter was small at the top,the small volume of gas flow immediately after the breakthrough could be measured precisely.The flow rate was measured on the inner and outer sides of the specimen.The flow rate measurement was terminated when the total outflow gradually increased to 50 cm3/min (normal condition),which was the same condition for end pressurizing used by Tanaka et al.(2010).Hoch et al.(2004) summarized the gas migration mechanism step by step: (i) gas flow was governed by conventional concepts of capillary pressure and relative permeability;(ii) microfissures were created,which provided pathways for the gas to enter the clay;and(iii)macroscopic fracture occurred,and it differed from the microfissures by the fracturing scale.The case where the effective gas pressure at the rate of total outflow reached a sufficient large value,i.e.50 cm3/min,was likely to correspond to macroscopic fracturing.However,a quantitative analysis of the fracturing size was difficult.Therefore,it was defined as the progressed breakthrough pressure in this study.After reaching the flow rate,the gas pressure gradually decreased until a range of 100-200 kPa,maintaining 50 cm3/min to expand the gas migration pathway,and 3-10 d were required for this to occur.

3.2.Investigation of change in hydraulic conductivity and gas migration pathway

Two identical bentonite specimens were tested to study the hydraulic conductivity of the compacted bentonite-sand mixture that was resaturated after gas migration and the water content distribution immediately after gas migration.One specimen was used to investigate the change in the saturated hydraulic conductivity,and the other was used to investigate the gas migration pathway.When the specimen was resaturated after gas migration,prior to supplying the solution,the specimen and connected lines were vacuumed to remove the helium gas accumulated at the bottom side of the specimen to prevent excess gas entry into the specimen by immediate permeation,although slight drying of the porous end pieces,filter,and specimen might occur.Subsequently,the saturation degree (>99%) was determined using Eq.(1).The hydraulic conductivity was measured immediately after resaturation to prevent cation exchange in bentonite via extra permeation.

To measure the water content distribution in terms of the gas migration pathway,the specimen was divided into 27 parts,as shown in Fig.4.The water content was measured based on the Japanese Industrial Standard(JIS A 1203,2009).However,when SW was used,the water content was corrected using the mass of salt precipitation after oven-drying as follows:

Fig.4.Separation of specimen to measure the water content.

wherewcis the water content corrected by the mass of the salt precipitation(%);msalt,maandmbare the masses(g)of the wet and dried samples with the vessel,respectively;andmcis the vessel mass(g).ma-mbincludes only the evaporated water,while the salt precipitation is measured as a solid mass.Because the salt is dissolved in the pore solution,the mass ratio of residual salt by ovendrying the solution,Csalt,was preliminarily measured,and the net quantity of solid and liquid mass was calculated.Approximately,1 g of SW was placed in an evaporating dish and dried at 110°C for 72 h.Solid mass was measured with a precision of 0.001 g.Csaltwas calculated as the mass ratio of the salt to the water.

3.3.Experimental condition

The experimental program is summarized in Table 3.The compacted Ca-bentonite-sand mixture was mixed with two solutions,i.e.SW and DW,while the compacted Na-bentonite-sand mixture was only mixed with SW,as reported by Tanaka et al.(2010),who investigated it with DW.Three test cases were conducted using the six specimens.The Na-and Ca-bentonite-sand mixtures with each solution are denoted as Na-bent-SW,Ca-bent-SW,and Ca-bent-DW,respectively.The effective clay dry density (ρdb) is defined by the following equation (JNC,2000):

wheremsbis the dry mass of bentonite (mg);VbandVvare the volumes of bentonite and void (m3),respectively;ρdis the dry density (g/cm3);Rsis the ratio of sand mass (%);and ρssis the particle density of sand (g/cm3).If there is no sand mixture,ρdbequals ρd.

4.Results and discussion

4.1.Equilibrium axial pressure and gas breakthrough

The experimental results before gas migration are presented in Table 4.The estimated saturation degree of the specimen exceeded 99%,indicating saturation of the specimen before the permeability and gas migration tests.The different equilibrium axial pressures excluding the back pressure were measured in two load cells,where the higher value was obtained in the upper side.The equilibrium axial pressures of the two specimens based on Na-bent-SW were 247-322 kPa and 302-361 kPa,respectively.Considering the contact pressure,the equilibrium swelling pressure was lower than those measured using Kunigel-V1 under DW conditions (Tanaka et al.,2010),i.e.637?698 kPa at an effective clay dry density of 1.392-1.407 g/cm3.The equilibrium axial pressure of Ca-bent-SW was lower than that of Ca-bent-DW.The reduction in equilibrium axial pressure by SW was 48%-59%for Na-bent and 26%-38% for Ca-bent.The influence of SW on the axial pressure was reduced when a compacted Ca-bentonitesand mixture was used.

Table 3 Program of gas migration test.

Table 4 Summary of experimental results before gas migration.

The relationship between the cumulative gas flow upon pressurization and elapsed time is shown in Fig.5.A breakthrough was observed inside the specimen by increasing the effective gas pressure in all the cases.Sequentially,a breakthrough outside the specimen occurred when a more effective gas pressure was applied.If the gas breakthrough proceeded evenly inside specimen,the inner side of specimen will indicate the breakthrough first owing to the difference in the cross-sectional area of the porous end piece(Fig.2b).A time lag occurs when the effluent porous end piece and the line are filled by the effluent gas between the inner and outer sides of the specimen.This indicates that the side wall leakage,which is a local flow between the specimen and test ring,is not dominant and that a relatively uniform gas flow occurs in the specimen.The relationship between breakthrough pressure and equilibrium axial pressure is shown in Fig.6.The breakthrough pressure was recorded at the inside of the specimen,because of prior occurrence of breakthrough.A higher equilibrium axial pressure resulted in a higher breakthrough pressure.The breakthrough pressure in the SW condition was lower than that in the DW condition,and the reduction rate was almost the same as the equilibrium axial pressure.In a previous study(Tanaka et al.,2010),a linear relationship was observed under DW conditions.The equilibrium axial pressure,which was higher at the top,showed an approximate linear relationship with the breakthrough pressure,ranging from approximately 0.7 to 1.2 times the equilibrium axial pressure,with no systematic changes in the type of bentonite andsolution.The breakthrough pressure was approximately equal to or exceeded the higher equilibrium axial pressure of each specimen,which was measured using the top load cell.The difference in the axial pressure,which presumably occurred by the interfacial friction and the heterogeneity during hydration,was determined by the location of load cell (Massat et al.,2016;Rawat et al.,2019).However,the maximum value of the equilibrium axial pressure or the effective stress in the specimen was significant in terms of gas migration.Therefore,breakthrough could occur under the following condition even using SW as the effective stress equals the sum of the swelling and contact pressures:

Fig.5.Cumulative gas flow (normal condition) and effective gas pressure in the gas migration test for (a) Na-bent-SW with dry density of 1.592 g/cm3;(b) Na-bent-SW with dry density of 1.599 g/cm3;(c)Ca-bent-SW with dry density of 1.59 g/cm3;(d)Ca-bent-SW with dry density of 1.591 g/cm3;(e)Ca-bent-DW with dry density of 1.607 g/cm3;and(f)Cabent-DW with dry density of 1.602 g/cm3.

After the gas breakthrough occurred in the equilibrium swelling pressure range approximately,the cumulative gas flow gradually increased,and the flow rate rapidly increased.When the total out flow approached 50 cm3/min,both inner and outer outflows occurred.Although a further study necessitates the precise measurement of the outer outflow because it might include the side wall leakage,gas discharge was predominantly observed in the inner cross-sectional area of the specimen prior to the subsequent increase in outflow of outer cross-sectional area when the progressed breakthrough occurred.After the advection/diffusion of dissolved gas or viscocapillary two-phase flow occurs,the flow pathways will dilate accompanied by a significant gas flux,based on the previous studies (Graham et al.,2002;Hoch et al.,2004;Cui et al.,2019).Hence,it is presumed that a larger pathway was presumably developed by a higher gas pressure above the tensile strength and effective stress according to the following equation,based on the hydraulic fracturing mechanism (Harrington and Horseman,2003).

Fig.6.Relationship between breakthrough pressure and equilibrium axial pressure.

whereTis the tensile strength of specimen (kPa);andis the effective stress in the radial direction (kPa),which is orthogonal to that of fracturing.The relationship between the progressed breakthrough pressure and equilibrium axial pressure is shown in Fig.7.The reduction in the progressed breakthrough pressure by SW was 55%-65% for Na-bentonite and 13%-38% for Cabentonite.The influence of SW on the progressed breakthrough pressure was remarkable for a compacted Na-bentonite-sand mixture.The progressed breakthrough pressure was approximately twice the equilibrium axial pressure.Specifically,the progressed gas breakthrough occurred when the effective gas pressure was approximately equal to or exceeded the higher equilibrium axial pressure in each specimen.It was found that the relationship between the progressed breakthrough pressure and equilibrium axial pressure was the same regardless of the solution used in this experiment.A comparison of the first breakthrough with the progressed one revealed that the progressed breakthrough pressure was approximately twice of the breakthrough pressure (Fig.8).However,the progressed breakthrough pressure indicated a better correlation with the equilibrium axial pressure.This is because the breakthrough pressure contained variations due to the time rag of gas transport.Focusing on the measured value of total axial stress,Tanaka et al.(2010),through a simulation,discussed the gas migration mechanism at an effluent gas flow of 50 cm3/min,which could be explained by assuming hydraulic fracturing.Therefore,based on the linear relationship in Figs.6 and 7 corresponding to the data in Tanaka et al.(2010),gas fracturing likely occurred in the specimen tested in this study.

4.2.Water content distribution

Fig.7.Relationship between progressed breakthrough pressure and equilibrium axial pressure.

Fig.8.Relationship between progressed breakthrough pressure and breakthrough pressure.

The water content at each part of the specimen and the average values after gas migration are listed in Table 5.The water content decreased slightly after gas migration,compared to the theoretical water content at saturation (26%-26.3% for Na-bentonite and 23.8-24.4% for Ca-bentonite),based on the void ratio.The saturation degree of Na-bent-SW,Ca-bent-SW,and Ca-bent-DW decreased to 90.1%,89.8%,and 93.9% after gas migration,respectively.Regarding the experimental result shown in Fig.5,both the inner and outer outflows occurred when the progressed breakthrough occurred.If the specimen shrinks in the radial direction,then the side wall leakage will increase significantly,and the inner outflow will decrease.Therefore,the entire volume of specimen was presumably maintained even if the water content slightly decreased owing to the gas migration.

The decrease in water content in all parts of the specimen was consistent with the fact that gas flow was observed inside and outside the specimen.The parts where the water content was below the average value were continuous in the vertical direction,such as locations 9 for Na-bent-SW,7-9 for Ca-bent-SW,and 3 for Ca-bent-DW.Although the change in water content ranged from 0.7% to 1.7%,the water content was presumably lowered in the region where the gas migration pathways were concentrated.Because gas migration seems to have occurred relatively evenly throughout the specimen under a constant specimen volume,one gas migration pathway is expected to be narrower.

4.3.Hydraulic conductivity

Before the gas migration,the hydraulic conductivity of Na-bent-SW,Ca-bent-SW and Ca-bent-DW was 1.9×10-11-2.0×10-11m/s,5.7 × 10-12-6.6 × 10-12m/s,and 1.5 × 10-12-1.6 × 10-12m/s,respectively(Table 4).The results of the hydraulic conductivity for each bentonite-sand mixture were highly reproducible.A comparison of the hydraulic conductivity of the saturated specimen before and after gas migration is shown in Fig.9.The saturated hydraulic conductivity after gas migration was approximately equal to that measured before gas migration,and the specimen saturated by SW showed higher hydraulic conductivity than the specimen saturated with DW.Strictly,the hydraulic conductivity measured under DW conditions was somewhat smaller,while that measured under SW conditions was 2-3 times larger than that before gas migration for each type of bentonite.The estimated saturation degree was 99.1%-99.4%,which was almost in the same range as that before gas migration,and partial saturation did not significantly influence the difference in permeability.Although the change in the hydraulic conductivity before and after gas migration was very small,the experimental results under SW conditions suggest that a change in the pore structure contributing to permeability occurred due to gas migration,as described in the following.

Table 5 Water content distribution after gas migration.

5.Reason for low permeability and factor to change permeability by gas migration

Various studies have indicated that the swelling and hydraulic characteristics of bentonite and montmorillonite decline under saline water conditions (Pusch and Karnland,1996;Komine et al.,2009;Zhu et al.,2013;Chen et al.,2015).The low permeability observed in this study could be related to the crystalline swelling of montmorillonite in the compacted bentonite,even with the diminished osmotic effect.Crystalline swelling is driven by the hydration of interlayer cations adsorbed on the montmorillonite.Diffuse ion layer growth is curtailed in the presence of saline solutions,and the expansive soil could develop swell potentials on inundation with 0.4-4 mol/L-NaCl;it was surmised that only crystalline swelling was operative on wetting for relatively saline solutions(Rao et al.,2013).Therefore,crystalline swelling was most likely dominant in the specimen saturated with SW.

Fig.9.Hydraulic conductivity of saturated specimen before and after gas migration.

For the specimen used in this study,the montmorillonite void ratioem,defined by the following equation(Ito and Mihara,2005),was 2.02-2.04 for Na-bentonite and 2.31-2.38 for Ca-bentonite:

whereVmis the volume of the montmorillonite (m3),eis the void ratio,Cmis the montmorillonite content,Cbis the bentonite content,ρsmis the particle density of montmorillonite (g/cm3),and ρsnmis the particle density of accessory minerals in bentonite (g/cm3).Based on the work of Watanabe and Yokoyama (2021),ρsnmwas calculated to be 2.76 g/cm3for Na-bentonite and 2.45 g/cm3for Ca-bentonite.Regarding the microstructure of compacted bentonite,the montmorillonite void ratio can be determined by the following equation,assuming uniform distribution and an array of montmorillonite in the specimen and using the thickness of montmorillonite,t(0.96 nm) and interlayer distance,d(m):

Thus,in the case of free swelling,the interlayer distance of each bentonite can be estimated as 1.9-2 nm for Na-bentonite and 2.2-2.3 nm for Ca-bentonite,respectively.Because the thickness of a single water hydrate is approximately 0.3 nm (Kozaki et al.,1998),the interlayer distance calculated corresponds to the 6-8 hydrated layers at the void ratio of the specimen used in the gas migration test.Regarding crystalline swelling,the interlayer distance of Naand Ca-montmorillonite is limited to the 3-layer hydrate state(Fukushima,1984;Likos,2004).For the specimen used in this study,40%-50% of the voids were occupied by hydrated water in the crystalline swelling state.As depicted in Fig.10a,the interlayer cannot expand fully under the crystalline swelling state;hence,large pores must exist in the external layer space.Based on the external water viscosity (Ichikawa et al.,1999),the water in the larger pores is remarkably related to the permeability of the compacted bentonite(Fig.10b).For the specimen with a relatively lower void ratio,the hydrated montmorillonite reduced larger pores through crystalline swelling and contributed to the low permeability of the specimen under SW conditions.Thus,the low permeability (e.g.10-11m/s) could be measured under SW conditions for Na-and Ca-bentonite.

Fig.10.Schematic of interlayer distance and macroscopic pore related to crystalline swelling: (a) Interlayer distance for unit volume;(b) Volumetric composition of compacted bentonite;and (c) Restructuring macroscopic pore due to gas migration.

Meanwhile,there is a possibility that the hydraulic conductivity increases if the gas migration changes the pore structure and generates larger connected pores that cannot be sealed by crystalline swelling.The gas migration pathway was invisible on the surface of the specimen after the experiment,specifically after the pore gas pressure was released,although white dots appeared to be dried around a relatively large particle such as the sand particle(Fig.11).The dried area indicates that the interface between the large particle and clay matrix contributed to gas migration.Meanwhile,considering the ratio of sand mixture (30%),the sand connectivity in the specimen was not dominant.The gas migration path would be “interface-clay matrix-interface” in the case of sand mixture.It is suggested that visible fractures did not exist after the gas pressure was released owing to the elastic behavior of the soil for highly compacted bentonite under constant-volume conditions.The opening of the developed pathways due to the increase of gas pressure in the pathways results from the consolidation of the clay matrix surrounding the pathway and/or the compression of the pore water in the clay matrix (Guo and Fall 2021).Pathway dilatancy was a possible condition in this study,namely,the effective gas pressure at the progressed breakthrough was larger than the sum of the minimum principal stress and the tensile strength of the specimen(Tanaka et al.,2010).As shown in Fig.10c,a minor change in the pore structure due to gas breakthrough might increase the number of connected pores that cannot be sealed by crystalline swelling,even with a small amount,potentially inducing a slight increase in hydraulic conductivity under SW conditions.

Consequently,the experimental results indicated that the nature of the low permeability of highly compacted bentonite does not substantially change owing to the gas breakthrough and sequential gas migration under SW conditions,irrespective of the type of bentonite.However,the SW condition included a two-to three-fold increase in the saturated hydraulic conductivity after gas migration.It should be considered that gas migration affects the larger pore structure,which cannot be sealed by crystalline swelling under SW conditions,even when using highly compacted bentonite under constant-volume conditions.Further investigation of the changes in the pore structure of compacted bentonite after gas migration and resaturation by SW is required.From the engineering perspective,it would be effective to increase the dry density or effective clay dry density of the bentonite-sand mixture to fully seal the void via crystalline swelling,if the change in permeability under SW conditions is not allowed for the barrier design.

Fig.11.Effluent face of specimen after gas migration.

6.Conclusions

Gas migration and permeability tests using compacted Na-and Ca-bentonite-sand mixtures,with a mixing ratio of 70%bentonite,were conducted to investigate the change in saturated hydraulic conductivity after gas migration under SW conditions.For the constant specimen volume condition,both the breakthrough pressure and progressed breakthrough pressure increased with increasing equilibrium axial pressure (the sum of swelling and contact pressures) when artificial SW was used,and their linear relationships were commonly obtained in DW and SW conditions in this experiment.The saturation degree decreased by 6%-10%after gas migration,and the water content distribution suggested a reduction in proximity to the gas migration pathway,although the pathway was invisible on the specimen surface.Based on the hydraulic conductivity measurements,the change in permeability was in a limited range,two to three times larger than that before gas migration for each type of bentonite in SW.Under SW conditions,crystalline swelling is more significant in sealing the hydraulic path and demonstrates the low permeability of the compacted bentonite,even with a diminishing osmotic effect.However,because of the slight increase in saturated hydraulic conductivity after gas migration in SW,the microstructural change in gas migration might induce a slight increase in the larger connected pores and permeability.In the experiment,the dilation of the compacted bentonite-sand mixture was considerably inhibited under the constant-volume condition.To examine the self-healing of the engineered barrier in saline water in practice,it is important to consider site-specific restraining conditions of the compacted bentonite,and the possibility that larger connected pores cannot be sealed by crystalline swelling of montmorillonite.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We thank Mr.Atsushi Mori for his assistance in the gas migration test.