Mechanical behavior of sandy facies of Opalinus Clay under different load conditions

Chun-Liang Zhang ,Ben Laurich

a Gesellschaft für Anlagen-und Reaktorsicherheit(GRS),Braunschweig,Germany

b Federal Institute for Geosciences and Natural Resources(BGR),Hannover,Germany

Keywords:Clay rock Compression Extension Creep Stiffness Strength

A B S T R A C T The mechanical behavior of sandy facies of Opalinus Clay at the Mont Terri underground rock laboratory(URL)in Switzerland was investigated with drained and undrained triaxial compression and extension,cyclic compression,and creep tests.Samples were taken from boreholes drilled parallel to bedding.Most of the samples were reconditioned to minimize sampling effects of desaturation and micro-cracking.The compression was accomplished by increasing axial stress at constant radial stress.The extension was carried out by increasing radial stress at constant axial stress.Moreover,extension was also achieved by simultaneously increasing radial stress and decreasing axial stress under constant mean stress.The test results showed elastoplastic stress-strain behavior with volumetric compaction until onset of dilatancy at high deviatoric stresses above 80% -90% of the peak failure strength.The strength is dependent upon load path and mean stress.The strength under triaxial compression is higher than that under extension.The respective strength increases with increasing mean stress.Desaturation enhances the stiffness and strength of the claystone.The deformation and strength of the claystone are time-dependent.Under constant deviatoric stress,the claystone crept continuously with time,which can be characterized by a transient phase and a following stationary phase,and even a tertiary phase at high deviatoric stresses to rupture.

1. Introduction

The indurated Opalinus Clay(OPA)formation has been proposed as a potential host rock for deep geological repository for high-level radioactive waste (HLW) in Switzerland (Nagra, 2002, 2014).Recently,part of the OPA formation in Southern Germany has been considered as one of the German generic models for elaborating the methodology of safety demonstration of a HLW repository in clay rock(Jobmann et al.,2017).The selection of OPA as a potential host rock is due to its favorable attributes,such as appropriate formation size and depth,homogeneous and stable rock mass,extremely low hydraulic conductivity,self-sealing potential of fractures,and high sorption capacity for retardation of radionuclides.

Over last two decades,the OPA has been comprehensively investigated at the Mont Terri underground rock laboratory(URL)with some 140 in situ experiments by 16 research institutions from 8 countries(Bossart and Milnes,2017;Bossart et al.,2017;Jaeggi et al.,2017).These experiments aim to characterize the clay rock,to test and demonstrate technologies for repository construction,waste emplacement, backfilling and sealing of the remaining openings,to understand and predict coupled processes and complex interactions in the multi-barrier disposal systems,and to estimate model parameters for safety assessment of the potential repositories.Fig.1 shows the geological structure and horizontal cross-section of the URL.At the URL location,the OPA formation can be divided into three layered lithological facies:(1)clay rich shaly facies-a dark gray silty calcareous shale and argillaceous marl in the lower half of the sequence;(2)sandy-carbonate rich facies-a gray sandy and argillaceous limestone in the middle of the sequence;and(3)sandy facies-silty to sandy marls with sandstone lenses cemented with carbonate in the upper part(Thury and Bossart,1999;Pearson et al.,2003;Bossart and Milnes,2017).Up to now,the clay rich shaly facies has been extensively characterized with laboratory and in situ experiments(e.g.Corkum and Martin,2007;Naumann et al.,2007;Popp and Salzer,2007;Bock et al.,2010;Valente et al.,2012;Zhang et al.,2013;Wild et al.,2015,2017;Minardi et al.,2017;Amann et al.,2017;Bossart and Milnes,2017;Marschall et al.,2017;Favero et al.,2018;Giger et al.,2018;Wild and Amann,2018;B?hm and Gr?sle,2018).Valuable knowledge about the thermo-hydro-mechanical(THM)properties and behavior of the shaly facies has been obtained.In order to characterize the sandy-carbonate rich facies for establishing a robust mechanical database for the entire OPA formation,a new gallery(Gallery 18)was planned in 2018(Swisstopo,2018)and constructed in 2019 within these facies(Fig.1a)for more in situ experiments.Before the URL extension,a laboratory program was launched by the Federal Institute for Geosciences and Natural Resources(BGR)and the Gesellschaft für Anlagen-und Reaktorsicherheit(GRS)to pre-investigate the hydro-mechanical behavior of the sandy facies.

Fig.1.Geological structure map and horizontal cross-section of the Mont Terri underground rock laboratory(a,taken from Swisstopo,2018)and sampling areas in the sandy facies(b).

A major safety concern in underground repositories is the excavation damaged zone(EDZ).It develops around excavated openings and may reduce the barrier function of the host rock against water flow and radionuclide transport.As observed in the OPA at Mont Terri URL(Bossart et al.,2004;Bossart and Milnes,2017)and in the similar Callovo-Oxfordian(COX)claystone at Bure URL in France(Armand et al.,2014;de La Vaissière et al.,2015),fractures are generated in the EDZ and the resulting permeabilities near drift walls are several orders of magnitude higher than that of the rock mass in far field.However,significant selfsealing of fractures was observed in laboratory tests on the clay rich OPA shaly facies and the COX clay rich unit(e.g.Bock et al.,2010;Zhang,2011,2013;Auvray et al.,2015;Guss et al.,2017;Giot et al.,2018)as well as in the in situ EDZ(de La Vaissière et al.,2015;Marschall et al.,2017).

In order to characterize the mechanical properties and the behavior of the OPA sandy facies,we performed different kinds of laboratory experiments to investigate(1)mechanical deformation and damage under various load conditions during repository excavation and operation,(2)moisture effects due to ventilation and water migration,and(3)self-sealing of fractures under hydromechanical impact during a long-term post-closure phase of the repository.This paper presents our investigations on the mechanical behavior only,while the studies of moisture effects and selfsealing of fractures are still ongoing and will be reported in upcoming publications.

2. Characterization of samples

2.1. Sample preparation

Our core samples were taken from boreholes BMB-A0,BDM-B9 and BLT-A8,which were drilled with compressed air in a diameter of 110 mm and lengths of 9-20 m parallel to bedding in the sandy facies(Fig.1b).After coring,they were immediately sealed in vacuum-tight aluminum foil or confined in rubber jackets and cells(Fig.2)to hinder damage from unloading and desaturation.The sealed samples were then stored in room temperature(22°C)for several months to a year until testing.From the cores,cylindrical samples were carefully prepared by cutting,smoothing and planishing the end faces in a lathe to three sizes of diameter/height(D/H)=50 mm/100 mm,70 mm/140 mm and 100 mm/200 mm for respective testing(Fig.2).Even though the sampling procedure was carefully conducted,the samples were inevitably desaturated to some extend and a few visible micro-fissures appeared on the surface of some samples.These fissures occur mostly along the bedding planes.For all investigated samples,the bedding foliation is in orientation parallel to the sample axis.

2.2. Mineralogical composition

Fig.2.Pictures of the confinement,storage and preparation of test samples.

Table 1 lists the major mineralogical phases in samples from borehole BDM-B9,as recorded by X-ray diffraction with Rietveld quantification.The mineralogical composition displays a strong spatial variability:within a 4 m interval,clays range from 24% to 41% ,quartz from 33% to 39% ,carbonates from 14% to 33% ,and feldspar from 8% to 9% .This variability is in accordance with that of the samples from borehole BLT-A2(Kaufhold et al.,2011),which shows relative lower contents of clay minerals(18% -23% ),and similar contents of quartz(31% -44% )and carbonates(22% -42% ).Compared to the shaly facies with clay minerals of 60% -80% ,quartz of 10% -27% and carbonates of 4% -35% (Thury and Bossart,1999;Bossart and Milnes,2017),the sandy facies generally has much lower clay contents but more quartz and carbonates.Kaufhold et al.(2013,2016)and Houben et al.(2014)demonstrated that the sandy facies is more heterogeneous on millimeter to centimeter scale than the shaly facies:The sandy facies comprises coarse-grained carbonate rich lenses and exposes a smaller contribution to clay matrix(down to 32% compared to the maximum of 85% in the shaly facies).In general,a material with high clay content is weaker than that with a high abundance of hard grains(e.g.Lupini et al.,1981;Bakker et al.,2017).This holds for the shaly and the sandy facies:they expose a drastic difference in bedding parallel uniaxial strength of about 10 MPa versus 18 MPa,respectively,and in bedding perpendicular uniaxial strength of 7 MPa versus 16 MPa(Jaeggi et al.,2017).Similarly,stiffness of the sandy facies exceeds that of the shaly facies,exhibiting Young’s modulus of 12.6 GPa versus 7.4 GPa in the direction parallel to the bedding,and 3.6 GPa versus 2.4 GPa in perpendicular direction(V?bel et al.,2014).

2.3. Petrophysical properties

The petrophysical properties of the samples,such as grain density,bulk density,dry density,porosity,water content,and degree of water saturation,were determined before testing.According to Brown(1981),these properties are defined as follows.

Grain density:

where Msis the mass of solids,and Vsis the volume of solids.

Table 1 Main mineralogical components of the sandy facies from borehole BDM-B9.

Bulk density:

where M is the mass of bulk sample,V is the volume of bulk sample,and Mwis the mass of water.

Dry density:

Porosity:

where Vvis the volume of voids.

Water content:

Degree of water saturation:

where ρwis the density of pore water.

The bulk density was determined by measuring volume and weight of the sample after preparation.The water content was measured on 20 mm thick disks that were left from the sample preparation.They were dried in oven at a temperature of 105°C for 2 d and subsequently checked for weight loss.The grain density was measured with a helium gas pycnometer on powder samples produced during sample preparation and dried at 105°C.Dry density,porosity and degree of water saturation are derived from the above measurements.The initial characteristics of all samples are summarized in Tables 2-5 for respective tests.The data of grain density,dry density,water content,and saturation degree are depicted in Figs.3-5 along borehole length.The grain density is relatively constant at ρs=(2.69±0.012)g/cm3(Fig.3),which is consistent with the data obtained on the OPA sandy facies from literature(e.g.Yu et al.,2017).The results from Yu et al.(2017)also show marginal differences of the grain densities between the sandy and shaly facies in a range of 2.69-2.71 g/cm3,which are also comparable to that of COX claystone(Zhang et al.,2019).The dry density and porosity along borehole BMB-A0 are relatively constant at ρd=(2.37±0.027)g/cm3and φ=11.7% ±1.3% (Fig.4a),but vary significantly along borehole BDM-B9(Fig.4b).The dry densityincreases dramatically from ρd=2.33 g/cm3to 2.53 g/cm3in the small distance of 6-8 m and the corresponding porosity drops from φ=13.5% down to 6.5% .This illustrates a significant heterogeneity of the claystone in this borehole.Similarly,the water contentsmeasured on the samples from borehole BMB-A0 vary in a small range around w=3.12% ±0.45% and the corresponding saturation degrees are Sw=65% ±10% (Fig.5a).The data obtained on the samples from borehole BDM-B9 indicate a significant decrease in water content from w ≈4% at L ≈6 m down to w ≈1.8% at L ≈8 m(Fig.5b),in correlation to the decrease in porosity(Fig.4b).The higher porosity samples contain more water.The corresponding degrees of water saturation are quite similar at Sw=65% ±10% along the borehole.The results indicate a remarkable desaturation of the samples due to coring,storage and preparation.

Table 2 Initial characteristics of the samples from borehole BMB-A0 for triaxial compression tests(TCS)and important results(sample size:D=70 mm,and H=140 mm).

Table 3 Initial characteristics of the samples from borehole BMB-A0 for triaxial extension tests(TEM and TEA)and important results(sample size:D=50 mm,and H=100 mm).

Table 4 Initial characteristics of the samples from borehole BDM-B9 for triaxial cyclic compression tests(TCD)and important results(sample size:D=70 mm,and H=140 mm).

Table 5 Initial characteristics of the samples from boreholes BLT-A8 and BDM-B9 for triaxial creep tests(TCC).

Fig.3.Distributions of grain density along boreholes BMB-A0 and BDM-B9.

In order to recover the natural water content in the claystone,resaturation was carried out by wetting some samples in humid air at relative humidity(RH)of ～100% for 1.5-2.5 months.Some other samples were tested as received in order to compare with the reconditioned samples and to highlight the effects of water saturation on the mechanical properties of the claystone.Fig.6 shows the wetting results of the samples BMB-A0 in group B with a diameter of 70 mm and a height of 140 mm(Table 2).The wetting over 1.5 months increased the water content from the initial mean value of wo=3.1% to a high mean level of wr=4.1% and the corresponding saturation degree from Swo=68% to Swr=85% .Higher saturation degrees of 92% -99% were achieved for a longer wetting duration of 2.5 months(Tables 4 and 5).It is noted that wetting resulted in some swelling of the unconfined samples and hence in changes of the measured volumes.For instance,sample BDM-B9-12 with an initial size of D/H=70 mm/140 mm and an initial water content of 2.23% was rewetted at RH=100% over 2.5 months,resulting in an increase in water content by 1.33% and swelling to an axial strain of 0.3% and a radial strain of 0.4% .The radial swelling normal to bedding is relatively larger due to more opening of the bedding planes,which can even generate visible cracks along bedding planes,especially when the sample was oversaturated for a longer time,as observed on another unconfined sample rewetted at RH=100% over 6 months in Fig.7.We therefore propose a new fracture type,where bedding parallel cracks form due to locally inhomogeneous swelling,instead of the well-known fracture generation by desiccation and unloading(e.g.Houben et al.,2014).In order to minimize the swelling effects,claystone samples have to be hydrostatically recompressed to approach the natural intact state for mechanical testing(Conil et al.,2018;Zhang et al.,2019).A few samples that were significantly damaged due to swelling were not used for the mechanical testing here.

3. Experimental methods

Laboratory testing to characterize the mechanical behavior of low-permeability claystone is a great challenge due to a high sensitivity of the material to unloading and humidity change by coring and sampling(which induced desaturation,shrinkage and micro-fissures),chemical interactions with water and resulting effects (swelling, weakness and damage), and strong hydromechanical coupling effects (variations in pore pressure and effective stress).Specific testing techniques and procedures are required with careful sampling and reconditioning,measurements of stress,strain and pore pressure,reconsolidation to minimize effects of sampling-induced micro-fissures, and others (Zhang et al., 2013, 2019; Menaceur et al., 2015; Wild et al., 2017;Belmokhtar et al.,2018;Conil et al.,2018;Ewy,2018;Giger et al.,2018;Wild and Amann,2018).

One of the technical challenges is to measure or control pore water pressure during mechanical loading.Usually,the pore water pressure of a sample is measured by the pressure in external reservoirs at the sample ends.However,it is still questionable whether the water pressure measured in the reservoirs is representative of the pore water pressure inside the sample and how the internal structure of the claystone is affected by contact with the external water.For the pore water pressure measurement,test conditions must be well controlled.For instance,claystone samples must be fully saturated;the pressure in the reservoirs must be steadily in equilibrium with the pore water pressure during testing;swelling of the samples is not allowed to avoid alteration of the internal structure of the material;and the volume of the reservoir must be fully saturated and maintain constant.These conditions are hardly met in laboratory tests.As mentioned earlier,some desaturation of claystone samples is unavoidable due to coring,storage and preparation.The samples usually need to be resaturated in humid environment and by injecting formation water.As contacting water,however,the claystone is going to swell,even under high confining stresses as observed on COX claystone under the lithostatic stresses of 12-15 MPa and pore water pressure of 4.5 MPa at the sampling location of the Bure URL(Zhang et al.,2013).Therefore,any water injection to claystone samples shall be avoided for mechanical testing due to swelling-induced alteration of the internal structure(Conil et al.,2018;Zhang et al.,2019).

Fig.4.Distributions of dry density and porosity along boreholes BMB-A0 and BDM-B9.

Fig.5.Distributions of water content and saturation degree along boreholes BMB-A0 and BDM-B9.

For this among other reasons,our tests were conducted in drained and undrained conditions without introducing water to the samples and without monitoring pore water pressure.As a consequence,the tests can only be analyzed by adopting total stress instead of effective one.It is to be pointed out here that the effective stress in indurated clay rock is partly transferred by bound pore water in interlayers and interparticle narrow pore space between clay particles(Horseman et al.,1996).This was evidenced by Zhang(2017a,b)with various kinds of experiments on COX and OPA claystones,which demonstrated that if not total,most of the bound pore water in these claystones is supporting the lithostatic stresses and even bearing deviatoric stresses up to the material strength.Hence,the total stress applied to the claystone samples might be equivalent to the effective one.For the drained tests,we regard the bound pore water pressure to be equivalent to the effective stress and the pressure of mobile water in relatively large pores to be at atmospheric state.For the undrained tests on the partly saturated samples,we speculate that some water in saturated pores could be expelled into unsaturated space during mechanical loading.The pressure of the mobilizing pore water may be very low before fully saturated.

Fig.6.Results of wetting the samples in vapor over 1.5 months from the initial low saturation(wo and Swo)to the high saturation(wr and Swr).

The main purpose of our tests is to investigate the effects of loading paths on the deformation and strength behavior of the OPA sandy facies with a large number of samples(see Tables 2-5).Additionally,effects of water saturation on the mechanical properties are also examined with two groups of samples with different saturation degrees(Table 2):group A as received with average saturation degree of Sw=68% and group B rewetted to Sw=85% .

Fig.8 shows the schematic layouts of undrained and drained triaxial tests.For the samples in Tables 2,3 and 5,the so-called undrained condition was applied by sealing the samples in rubber jackets and covering them at the top and bottom with load pistons(Fig.8b).The samples in Table 4 were tested in drained condition(Fig.8c),where the sample ends were covered with metallic porous discs connected to long tubes to keep atmospheric pressure.The pore water can be expelled by compression into the porous disc without rising back pressure at the sample ends.All the samples with different sizes were loaded in three triaxial cells with controlled strain or stress rates.During loading,axial strain(εa)is recorded by linear variable displacement transducers (LVDTs)installed inside the cell along the sample length,while radial strain(εr)is measured by a circumferential extensometer chain mounted around the sample outside the jacket.The volumetric strain is obtained approximately by εv=εa+2εrfor small strains.

Fig. 7.Wetting-induced cracks along bedding planes in an unconfined sample(D=100 mm,and H=100 mm)at RH=100% over 6 months.

Fig.8.Assemble of sample for triaxial testing(a)under undrained(b)and drained(c)conditions.

As mentioned above,this paper focuses on the effects of loading paths on the mechanical behavior of the OPA sandy facies. Common triaxial compression tests determine the deformation and strength of a rock sample in such a way that the sample becomes axially shorter.However,an underground gallery excavation results in various kinds of stress paths around the openings,mostly dominated by extension near the gallery walls(Cristescu and Hunsche,1998;Brady and Brown,2006).In order to simulate the prevailing in situ stress conditions using a triaxial testing apparatus,several different loading paths were applied to the samples,as illustrated in Fig.9.The tests were generally performed in two steps with hydrostatic precompaction followed by deviatoric stressing. The precompaction aimed to minimize the effects of micro-fissures induced by sampling,which was performed at a loading rate of 7×10-3MPa/s up to an isostatic stress of 15 MPa and kept for a duration of 15-20 h.Based on the maximum burial of approximately 1350 m(Mazurek et al.,2006,2008),the precompaction stress corresponds approximately to the maximum effective stress (the maximum overburden pressure of about 30 MPa minus the mobile pore water pressure of about 13.5 MPa). Thus, we regard the pre-compaction to have not caused drastic(if any)over-consolidation of the claystone matrix but the closure of micro-fissures. After unloading to a desired level,subsequent deviatoric loading was conducted with regard to five respective loading paths and their scientific objectives:

(1)Investigation of the effects of different loading paths on short-term deformation and strength with respect to excavation:

i)Compression by increasing axial stress σaat constant radial stress σr(TCS),

ii)Extension at constant in situ stress resembling mean stress σm= (σa+2σr)/3 (TEM) by simultaneously increasing radial stress σrand decreasing axial stress σato determine extensional deformation and strength of the rock,and

Fig.9.Loading paths applied in mechanical testing of the OPA sandy claystone.

iii)Extension by increasing radial stress σrat constant axial stress σa(TEA)to resemble conditions at the opening wall;

(2)Investigation of elastic parameters under different loads by cyclic axial compression εaat constant radial stress σr(TCD);and

(3)Investigation of time effects on deformation and strength by creep under multistep constant stress states,i.e.σaand σrare kept constant(TCC).

The TCSs were conducted on 18 samples in two groups from borehole BMB-A0(Table 2).Group A consisted of 9 samples without reconditioning and with relatively low saturation degrees of Sw=68% ±9% ,while the other 9 samples in group B had been rewetted to higher saturation degrees of Sw=85% ±5% .All the samples were loaded at an axial strain rate of 7×10-7s-1under different radial stresses of σr=0.5-10 MPa.This strain rate is in the same order of 10-6-10-8s-1,mostly applied in the undrained tests on the OPA and COX claystones(e.g.Menaceur et al.,2015;Wild et al.,2017;Ewy,2018;Giger et al.,2018).The low strain rates allow homogeneous dissipation of pore water pressure throughout the sample.The identical load conditions applied to the samples with and without rewetting allow identification of effects of water saturation on the mechanical properties of the claystone.

The triaxial extension tests(TEM and TEA)were carried out on 9 samples from the same borehole BMB-A0 (Table 3). Without reconditioning,their saturation degrees of Sw=61% ±7% were relatively low because they had been exposed to air in a longer preparation phase.Extensional loading(TEM)was conducted by simultaneously increasing radial compression at a rate of 3.5×10-3MPa/s and axial tension load at 7×10-3MPa/s and at respective mean stress σm=15 MPa,16 MPa,18 MPa and 20 MPa.On the contrary,the TEA loads were applied by increasing the radial compressive stress at a rate of 7×10-3MPa/s during various constant axial stresses of σa=1 MPa,2 MPa,3 MPa,5 MPa and 8 MPa,respectively.The loading rate is comparable to the strain rate in the TCS tests.

The cyclic compression tests(TCD)are designed to examine variations of rock elasticity with loading and deformation in drained condition.The tests were conducted on 6 samples from borehole BDM-B9(Table 4).Four samples were rewetted to high saturation degrees of Sw=91% -100% ,whereas the other two without wetting remained at relatively low saturation degrees of Sw=66% -86% .The deviatoric loading was conducted by cyclically increasing axial compression at constant radial stress of σr=1-5 MPa.The multiple loading-unloading cycles were performed at a constant axial strain rate of 7×10-7s-1,which is the same order of the low strain rates of 10-6-10-8s-1applied in the drained tests on claystones(e.g.Belmokhtar et al.,2018;Ewy,2018;Giger et al.,2018;Wild and Amann, 2018), in order to avoid excess pore pressure during compression.Each cycle was limited in a range of Δσa=-4 MPa for unloading and Δσa=6 MPa for reloading,corresponding to stress increments of 2 MPa.From the load cycles,one can determine elastic parameters at different load conditions,namely Young’s modulus E and Poisson’s ratio ν,by

where Δεrand Δεarepresent the radial and axial strain changes in each loading cycle, respectively. The Poisson’s ratio is an average of the response in the direction perpendicular and parallel to the bedding.For a complete description of the elastic behavior of the transversely isotropic material,five independent elastic parameters are needed (Zhang et al., 2019) but not determined in this program.

In order to examine the time effects on deformation and strength of the claystone,triaxial creep tests(TCC)were carried out on two highly saturated and one dried sample in undrained conditions over months to years.Their initial characteristics are given in Table 5.Before testing,sample BLT-A8(D/H=99 mm/190 mm)and BDM-B9-12 (D/H = 70 mm/140 mm) were rewetted at RH=100% over 2.5 months to high saturation degrees of Sw=93% and Sw=98.5% ,respectively,while BDM-B9-11(D/H=70 mm/140 mm)was dried to a low saturation degree of Sw=27% .Multistep deviatoric stresses were applied by keeping a mean stress of σm=(σa+2σr)/3=13 MPa.Both axial and radial strains were measured on the large sample BLT-A8,but radial strain measurement on the other two samples in another creep rig could not be accomplished.The tests on both samples BDM-B9-11 and BDM-B9-12 with different saturation degrees under identical load conditions aimed at examining the influence of water saturation on creep of the claystone.

4. Results and discussion

4.1. Pre-compaction

Fig.10 shows the typical results of the pre-compaction obtained on a non-wetted sample with saturation degree Sw=80% (Fig.10a)and another rewetted one with Sw=96% (Fig.10b)under isostatic stress of 15 MPa for 15-18 h,respectively.In all samples,the bedding foliation is parallel to the axial direction.Yet surprisingly,some of them showed quasi-isotropic compaction with similar axial and radial strains,i.e.εa≈εr,and some others showed even slightly larger axial strains. This observation on cylindrical samples cannot be reasonably interpreted yet. The anisotropy of the sedimentary rock shall be better studied with cubic samples under hydrostatic compression in a true triaxial apparatus(Popp and Salzer,2007).In contrast to the non-wetted samples,the rewetted samples showed clear anisotropy with a relatively larger radial strain normal to bedding compared to the axial one parallel to bedding, i.e. εr＞ εa. We attribute this anisotropy to the closure of bedding parallel fissures developed by a somewhat inhomogeneous swelling during the foregoing rewetting in unconfined conditions(Fig.7).In consequence,the compaction of the rewetted samples with more micro-fissures is larger compared to the non-wetted samples. This finding is demonstrated in Fig.11 by the volumetric compaction curves vs.time for the non-wetted samples with Sw=68% ±9% (Fig.11a)and the rewetted ones with Sw=85% ±5% (Fig.11b),respectively.The volume reduction of the rewetted samples reached εv=0.45% -0.7% ,which is higher than that of the non-wetted samples(εv=0.35% -0.55% ).However,the remaining volumetric strains after unloading varied from 0.2% to 0.4% for both nonwetted and rewetted samples,revealing the permanent plastic closure of the micro-fissures by the pre-compaction.Additionally,the pore water in the samples could be expelled into unsaturated pores and fissures during mechanical compression.The compaction increased the saturation degree by ΔSw≈1% -3% for the samples,depending on the magnitude of the porosity reduction.Obviously,full saturation could not be reached at most of the samples with initial Swvalues below 97% ,but only at a few highly saturated samples with Sw≥97% (Tables 4 and 5).For samples with low saturation degrees(Sw＜97% ),air-filled pores might not be sufficiently water-filled during the compression and the resulting pore pressure increase might be negligible.Thus,the applied total stress might be considered equivalent to the effective stress.

Fig.10.Pre-compaction of a non-wetted sample(a)and a rewetted one(b)under an isostatic stress of 15 MPa.

Fig.11.Volumetric compaction of the(a)non-wetted and(b)rewetted samples under an isostatic stress of 15 MPa.

Fig.12.Typical stress-strain curves(Δσ-εa,Δσ-εr,and εv-εa)with dilatancy threshold σD,peak strength σF and residual strength σR obtained during triaxial compression at a radial stress of 1 MPa.

4.2. Compression and extension

Three loading paths were applied to simulating the prevailing stress conditions during excavation of underground openings,namely triaxial compression(TCS)and triaxial extension at constant mean stress(TEM)as well as triaxial extension at constant axial stress(TEA).

Fig.12 shows the typical results from a TCS test at a radial stress of 1 MPa in terms of deviatoric stress(Δσ=σa-σr)vs.axial strain(εa)/radial strain(εr)and volumetric strain vs.axial strain(εv-εa).The stress-strain curves show axial compression(εa＞0)and radial extension(εr＜0),resulting in overall volumetric compaction(εv＞0)with loading.As the stress reaches a critical value σD,the sample compaction reaches its maximum and then turns over to increase in volume.The stress at the onset of volume increase is called dilatancy threshold.From this very point,the space accommodated by the formation of sample internal voids(i.e.micro-cracks)overcomes the volume reduction caused by loading.Further increase of the deviatoric stress leads to the formation of more and larger micro-cracks until sample failure at its peak strength σF.Subsequently,the stress drops,contemporaneously with crack and shear zone evolution until a residual strength σRis reached,which keeps relatively constant during the post-failure stage.

Fig.13.Stress-strain curves(Δσ-εa,Δσ-εr,and εv-εa)obtained on the(a)dried and(b)wetted samples during triaxial compression at various radial confining stresses.

The stress-strain curves obtained at different radial stresses are presented in Fig. 13a for the non-wetted samples with Sw=68% ±9% and in Fig.13b for the rewetted samples with Sw=85% ±5% ,respectively.The results show that the claystone deforms gradually from brittle with rapid rupture to ductile with slow rupture as the radial stress increases.This phenomenon may be attributed to different degrees of sample compaction,where high confining stresses result in less favorable conditions for rapid particle rotations and hence hinder brittle fracturing.The stressstrain behavior is qualitatively independent of water saturation(Sw), but the quantities of the deformation and strength are different,as compared in Fig.14 for the tests on two non-wetted and two rewetted samples at a radial stress of 1 MPa.It can be seen that the slopes of the stress-strain curves of the non-wetted samples are steeper(indicating higher stiffness)and the peak stresses are larger than those of the rewetted samples.The higher stiffness and strength of the claystone with lower saturation degrees are mainly attributed to an increase in cohesion and friction resistance between clay particles due to reduced thickness of interparticle bound water-films between clay particles and more solid-solid contacts.Contrarily,increasing water content enlarges the thickness of the bound water-films and decreases solid-solid contacts,which reduces the inherent cohesion and friction resistance.Hence,the stiffness and strength decrease with increasing water content.Similar conclusions on the impact of water content on the mechanical properties were also drawn for the OPA shaly facies(e.g.Jaeggi et al.,2017)and the COX claystone(e.g.Zhang,2017b).

Fig.14.Comparison of the stress-strain curves of the dried and wetted samples at radial stress of 1 MPa.

Fig.15 presents the stress-strain curves obtained from the extension tests TEM at constant mean stresses of σm=15-20 MPa and TEA at constant axial stresses of σa=1-5 MPa on the nonwetted samples with saturation degree of Sw=61% ±7% .The stress difference is defined here by Δσ=σr-σa.It can be seen that the samples deformed nonlinearly in radial compression(εr＞0)and axial tension(εa＜0)by simultaneously increasing σrand decreasing σa(TEM)and by increasing σrat constant σa(TEA).The obtained volumetric strains evolved over a transition phase from compaction to dilatancy at high deviatoric stresses.Whereas the dilatancy thresholds observed during TEM testing are about 90% of the peak strengths(σD≈0.9σF),the dilatancy during TEA testing occurred nearly at the peak stresses(σD≈σF).The failure occurred suddenly without post-failure stage under the stress-controlled loading conditions.The data of extensional strength will be evaluated later in Section 4.4 in comparison with the compressive strength.

4.3. Elastic parameters

Fig.15.Stress-strain curves obtained from the triaxial extension tests TEM at different mean stresses(a)and TEA at different axial stresses(b)on the non-wetted samples with saturation degrees of 61% ±7% .

Fig.16.Deviatoric stress-strain curves(Δσ-εa,Δσ-εr,εv-εa)with dilatancy threshold σD and peak strength σF during multiple axial cyclic compression at a radial stress of 4 MPa.

The elastic parameters(Young’s modulus E and Poisson’s ratio ν)of the OPA sandy claystone were determined under drained cyclic load conditions(TCD).Multiple axial loading-unloading cycles were performed on 6 samples(Table 4)at radial stresses of σr=1-5 MPa and at atmospheric back pressure at the sample ends.Typically,Fig.16 shows the deviatoric stress vs.axial/radial strain(Δσ-εaand Δσ-εr)and volumetric strain vs.axial strain(εv-εa)curves for sample BDM-B9-15,tested at σr=4 MPa.The envelopes of the stress-strain curves show a similar behavior to the monotonic compression(see Fig.12)with axial compression(εa＞0),radial extension(εr＜0)and volumetric compaction(εr＞0)until onset of dilatancy at stress σD.The dilatation evolved to the peak stress σFand beyond.

Based on the data,the elastic parameters can be obtained by Eqs.(7)and(8)at different load and deformation conditions.Fig.17 illustrates the evolution of E and ν with increase of axial strain and in correlation with the evolution of volumetric strain εv.The values of E and ν are averaged for each loading-unloading cycle.The elastic stiffness E increases with volume compaction until the dilatancy point σDis reached.Subsequently,the stiffness decreases with increasing dilatancy or damage.However,the Poisson’s ratio ν maintained relatively constant in the beginning,and then increased slightly and rapidly from the dilatancy point σDto the peak failure σF,which indicates the aperture opening of micro-cracks in the minor principal stress(σr)direction(σ1=σrfor extension).During the post-failure stage,the relatively larger axial deformation leads to a decrease of ν.

Influences of the radial confining stress on the elastic parameters are illustrated in Fig.18.Logically,the elastic stiffness E increases with increasing radial(or mean)confining stress,which enhances the density of the porous medium.In contrast,the Poisson’s ratio ν is almost independent of the radial stress(except for a sample at σr=2 MPa).

4.4. Dilatancy,peak and residual strengths

Fig.17.Elastic parameters in correlation with volume change as a function of axial compression at a radial stress of 4 MPa loading.

Fig.18.Variations of the elastic parameters with axial strain under different radial confining stresses.

Fig.19.Comparison of the dilatancy,peak and residual strengths obtained from the triaxial compression tests on the non-wetted samples with saturation degrees of Sw=68% ±9% and rewetted samples with Sw=85% ±5% as a function of radial stress.

Tables 2-4 summarize the identified dilatancy threshold σD,the peak strength σF,and the residual strength σRas well as the fracture angle β oblique to the direction of the major principal stress.All stresses therein were obtained on the samples in the triaxial compression(TCS),extension(TEM and TEA)and cyclic compression(TCD)tests.Fig.19 depicts the data obtained on the samples from borehole BMB-A0 with average saturation degree of Sw=68% (non-wetted)and Sw=85% (rewetted)in the TCS tests.The linear regressions are obtained for the non-wetted and rewetted samples separately.It is obvious that the dilatancy threshold, peak failure and residual strengths increase with increasing radial stress.At a given radial stress,the dilatancy and peak strengths of the rewetted samples(Sw=85% )are about 10% lower than those of the non-wetted samples(Sw=68% ),but the residual strength values fit a unique line and seem to be independent of the water saturation degree.We explain this parity by the rheological properties of the fabric of a developed shear zone.Such a zone,once formed,exhibits face-to-face aligned clay particles with a rather low angle of friction(e.g.Laurich et al.,2014).The dilatancy or damage thresholds determined on the nonwetted and rewetted samples(blue points and lines)are quite close to the respective peak strengths(red points and lines).The stress ratio of dilatancy to peak stress(σD/σF)varies in a range of 80% -98% for the non-wetted samples and 78% -95% for the rewetted ones,on average σD/σF=90% for the non-wetted samples and σD/σF=85% for the rewetted samples.The high ratio of dilatancy to peak stress is also true for the claystone under other stress conditions:σD/σF=88% -100% from the cyclic compression TCD tests(Table 4)and σD/σF=88% -99% for the extensional TEM and TEA tests(Table 3,except for an abnormally low value).After the peak,the fractured claystone is still capable of bearing certain loads, for instance, about 50% of the peak stress(σR/σF≈0.5)under triaxial compression(see Fig.19).As discussed earlier,the strength of the claystone depends on water content and saturation degree. Higher water saturation means more wetted surfaces between clay particles,in which the inherent cohesion and friction resistance are relatively lower compared to unlubricated solid-solid contacts.More importantly,capillary action in unsaturated samples can lead to a drastically high suction that drives more solid-solid contacts between clay particles,and hence yields a heavy increase in effective stress(Zhang et al.,2007;Amann and Vogelhuber,2015;Minardi et al.,2017;Zhang,2015). Thus, the strength decreases with increasing water saturation.

In order to uniformly evaluate the strength data obtained from different loading paths,a more complex Mohr-Coulomb model is adopted,which considers the third stress invariant in terms of Lode’s angle as an additional factor influencing the rock strength(UPC,2015)as

where q is the shear stress defined as

The mean stress σmis express as

The Lode’s angle θ is defined as follows:

In Eqs.(9)-(12),σ1,σ2and σ3are the major,intermediate and minor principal stresses,respectively,and σ1＞σ2＞σ3;c is the cohesion;and φ is the friction angle of the material.The Lode’s angle represents the stress geometry or loading path.In case of triaxial stress conditions,we have q=|σ1-σ3|=Δσ，θ=-π/6 for compression,where σa=σ1≥σr=σ2=σ3；and θ=π/6 for extension,where.σa=σ1≤σr=σ2=σ3.

Fig.20 summarizes the peak shear strengths as a function of mean stress obtained from the compression tests(TCS)on the samples with average saturation degrees of Sw=68% and 85% ,respectively,the cyclic compression tests (TCD) on the highly saturated samples with Sw=95% ,and the extension tests(TEA and TEM)on the relatively dry samples with Sw=61% and 63% ,respectively.Fitting the data of the compression tests TCS and TCD（θ=-π/6）yields the strength parameters for the samples with different saturation degrees.The cohesion is determined as c=4.5 MPa,3.5 MPa and 2.5 MPa for Sw=68% ,85% and 95% ,respectively,whereas the friction angle determined is φ=45°,independent of the saturation degree.It seems that the water saturation affects the cohesion of the claystone but not-or only to a smallamount-theinnerfriction.Thecohesiondecreaseslinearlywith increasing water saturation,as shown in Fig.21.An extrapolationyields the parameter c=2.3 MPa for full saturation.The extrapolated shear strength for the fully saturated sandy claystone is depicted in Fig.20,which is only slightly lower than that at Sw=95% .

It is also obvious that at a given mean stress,the strength obtained by extension testing(TEA and TEM,θ=π/6)is lower than that by compression(TCS and TCD,θ=-π/6).Using the strength parameters of c = 4.5 MPa and φ = 45°obtained from the compression tests on the samples with Sw=68% ,the extensional strength of the samples can be well predicted by the Mohr-Coulomb model(Eq.(9)).Some underestimation of the data is probably due to the relatively lower saturation degrees of Sw=61% -63% of the TEM and TEA samples.Low water saturation enhances the strength of claystone.

After testing, failure modes of the samples were visually inspected.Fig.22 shows pictures of some failed samples during the triaxial compression(TCS and TCD)and the extension(TEA and TEM).The fracture angle is defined as the angle β of the fracture surface to the direction of the major principal stress σ1(σ1=σafor triaxial compression,and σ1=σrfor extension).The measured β values are given in Tables 2-5.The Mohr-Coulomb model can predict a fracture angle by β=45°-φ/2=22.5°.This value represents the mean value of the measured data of β=19°-27°under compression(Tables 2 and 4)and β=15°-25°under extension(Table 5),except for small values of β=13°-15°at low confining stresses.Fig.23 illustrates the model prediction for the fracture angle under compression at different radial stresses.The model overestimates the fracture angle at low radial stresses(σr＜2 MPa),which might be attributed to aperture opening effect of the axial bedding planes in radial direction.The failure modes of the sandy claystone with bedding planes parallel to the major principal stress are similar to those observed on the OPA shaly facies(Naumann et al.,2007;Popp and Salzer,2007;Valente et al.,2012).

4.5. Time-dependent deformation

Time-dependent deformability of the OPA sandy claystone was studied on three samples(BLT-A8,BDM-B9-11 and BDM-B9-12,Table 5)under undrained multistep triaxial loads by keeping a mean stress of σm=(σa+2σr)/3=13 MPa over time periods of months to years.Samples BLT-A8 and BDM-B9-12 were rewetted to high saturation degrees of Sw=93% and Sw=98.5% ,respectively,while BDM-B9-11 was dried to a low saturation degree of Sw=27% .The tests on samples BDM-B9-11 and BDM-B9-12 with different saturation degrees aimed at examining the influence of water saturation on creep of the claystone.Again,the bedding of both samples was parallel to their axes.

4.5.1. Creep behavior

Fig.20.Shear strength of the OPA sandy claystone under triaxial compression at constant radial stress(TCS and TCD,σr=c)and extension at constant mean(TEM,σm=c)or axial stress(TEA,σa=c)as a function of mean stress and saturation degree(Sw).

Fig.21.Cohesion related to the saturation degree for the OPA sandy claystone.

Sample BLT-A8 had been pre-compacted at isostatic stress of 15 MPa for 7 d,resulting in a decrease of porosity from 8.3% to 8% and thus an increase of saturation degree from 93% to 96% .It was then followed by creep testing under multistep deviatoric stresses of Δσ=σa-σr=3 MPa for 1 month,9 MPa for 3.5 months,15 MPa for 5.5 months,and 24 MPa for 5 months.The mean stress was kept at σm=13 MPa for the first three phases and then slightly increased to 14 MPa for the last phase.The temperature was controlled at 27.5°C±0.1°C.The applied stresses and temperature,the resulting axial,and radial and volumetric strains are illustrated in Fig.24a.It can be found that:(1)strains could be hardly detected during the first phase at Δσ=3 MPa;(2)each sudden increase in deviatoric stress yielded a rapid deformation with axial compression,radial extension, and volume dilatancy due to the relatively larger extension in radial direction normal to the bedding;(3)under each constant stress state,the axial and radial strains continued with time from a transient phase to a steady state(Fig.24b);and(4)the volume did not change much with time at deviatoric stresses of Δσ ≤15 MPa,but at a high deviatoric stress of Δσ=24 MPa at σr=6 MPa,a dilation process evolved with time indicating a damage evolution.This dilatancy stress(black point in Fig.25)is slightly lower than that from the short-term compression tests on the samples with nearly the same saturation degrees of 95% ±5% .It seems that the damage of the claystone is little dependent on load duration.

Fig. 23.Shear fracture angles observed on the failed samples under triaxial compression at different radial stresses and model prediction.

From the strain-time curves,the axial and radial strain rates at each load stage are derived and depicted in Fig.24b against time.The data show that the creep rates decreased with time during the first 2-3 months and then tended to steady state.The stationary creep rates were quite low in a range of 10-12-10-11s-1,which increased with increasing deviatoric stress, i.e. the creep was accelerated by increasing the deviatoric stress.

Inordertoexaminetheinfluenceofwatersaturationoncreepofthe OPA sandy claystone,a dried sample BDM-B9-11 with Sw=27% and a rewetted sample BDM-B9-12 with Sw=98.5% were tested under identicalmultisteploads,i.e.thesampleswere firstlypre-compactedat isostatic stress of σm=13 MPa for 0.5 month,and then the deviatoric stress was increased to Δσ=3 MPa for 0.5 month,9 MPa for 1 month,and 15 MPa for another month at σm=13 MPa.The temperature was 26°C±0.2°C.Unfortunately,radial strain could not be monitored in the used creep rig.The applied stresses and the resulting axial strains are illustrated in Fig.26a for both samples.The following lists the sequence of observations.

Fig.22.Pictures of some samples fractured under triaxial compression(a)and extension(b).

First, the sudden hydrostatic compression resulted in rapid compaction.The wetted sample was more compacted to an axial strain of εa=0.33% than the dried one to εa=0.19% ,again indicating more micro-cracks induced during sample resaturation under unconfined conditions.During the pre-compaction phase at the high load of σm=13 MPa,the compaction of the dried sample terminated several days later,while the wetted sample exhibited a gradual expansion to Δεa≈-0.01% rather than a compaction.It might be attributed to swelling effects from clay minerals in the sample.In the previous experiments(Zhang,2017a,b),some swelling was already observed on COX claystone even under the in situ rock stresses(σ1=15 MPa,and σ2=σ3=12 MPa)and pore water pressure of 4.5 MPa at Bure URL(Armand et al.,2014).The pre-compaction might have led to a decrease in the thickness of water-films in interlayers in clay particles and to a decrease of distances between clay particles,in which the disjoining(swelling)pressures acting in the intraparticle and interparticle bound water might become higher than the external confining stress and in turn enlarge the thickness of the narrowed interlayers and interparticle distances until equilibrium.Additionally,the mobile water in relatively large pores could be expelled into unsaturated pores, increasing the saturation degree by 1% -3% as observed on the other samples(see Section 4.1),and thus the highly saturated sample with the initial Swvalue of 98.5% could be fully saturated by the pre-compaction.

Fig.24.Results of a multistep triaxial creep test on a highly saturated sample.

Fig.25.Comparison of the dilatancy and failure strengths observed on the OPA sandy claystone during the short-and long-term tests.

Following the pre-compaction phase,a low deviatoric stress phase of Δσ=3 MPa was applied,under which the gradual axial expansion of the saturated sample continued over the entire time span.This result implies again that the internal swelling pressure acting in the intraparticle and interparticle bound water must exceed the external(axial and radial)stress.As the axial stress was increased to 19 MPa and the radial stress was decreased to 10 MPa(Δσ=9 MPa),a very slow axial compression appeared at the saturated sample,but not at the dried one.A significant creep took place at both saturated and dried samples at further increased deviatoric stress of Δσ=15 MPa(σa=23 MPa,and σr=8 MPa).After about a month,the creep of the saturated sample then evolved with increased rates into rupture(Fig.26b).Before failure,the creep rates of both saturated and dried samples were quite similar with a transient phase in the beginning and the following stationary phase,which lasted for a short time at the saturated sample.

Compared to the short-term strength of 30 MPa at σr=8 MPa determined on the highly saturated samples with Sw=95% ±5% (Fig.25),the creep rupture stress of 15 MPa seems to be quite low.Similar low creep rupture stresses were also observed on the highly saturated COX samples in undrained conditions(Zhang et al.,2019).The low creep rupture stresses observed might be caused(1)by high internal swelling pressures that built up in the interlayers in clay particles and in the interparticle narrow pore space and(2)by high water pressures increased in the large pores under the previously applied high radial stress.The internal pore pressures,particularly the swelling pressures,could exceed the reduced external radial stress,resulting in a decrease in effective stress down to zero and even possibly to negative values(tensile state).At the lowered effective radial stress,the applied deviatoric stress could yield failure as illustrated in Fig.25(red point)by assuming zero effective radial stress.This creep failure stress is slightly lower than the short-term uniaxial strength.However,this hypothesis needs to be confirmed by undrained triaxial creep testing with monitoring pore water pressure.

4.5.2. Stationary creep rate

The stationary shear creep rates,˙ε=2|˙εa-˙εr|/3，obtained on the OPA sandy facies(sample BLT-A8),are compared in Fig.27 with those of the clay rich OPA shaly facies and the COX claystone at-490 m level of the Bure URL(Zhang et al.,2013;Zhang,2015).The COX claystone contains clay minerals of 50% -55% ,carbonates of 20% -25% ,quartz of ～20% and others of ～3% (Armand et al.,2014).All the tested samples were highly saturated with degrees of Sw≥95% and the testing temperatures were comparable in a range of 25°C-28°C.The comparison shows that the OPA sandy claystone crept more slowly than the clay rich ones at a given deviatoric stress,i.e.the creep rate increases with increasing clay content.

Mitchell(1992)derived a constitutive equation for creep of clay soils through thermo-dynamical analysis of soil flow by application of the absolute reaction-rate theory. This model was slightly modified and applied by Zhang(2015)for the stationary creep of clay rocks:

where T is the absolute temperature(K);R is the universal gas constant,and R=8.31433×10-3kJ/(mol K);Q is the apparent activation energy(kJ/mol);A is a parameter in s-1;and α is a parameter in MPa-1.These parameters were established as follows:A=2×10-4s-1,α=0.2 MPa-1,and Q=45 kJ/mol for the COX claystone(Zhang,2015);and A=6×10-4s-1,α=2 MPa-1,and Q=45 kJ/mol for the OPA sandy claystone.Fig.27 shows a reasonable agreement between the model(solid line)and the data for the respective claystone.However,it is noted that the observed volumetric deformation(consolidation and/or dilatancy)with time is not yet taken into account in the creep equation.Particularly,dilatancy or damage induced by excavation accelerates drift convergence as observed in the URLs of Mont Terri(Bossart and Milnes,2017)and Bure(Armand et al.,2014).The damage effect on the long-term deformation needs to be involved in the creep models for predicting long-term performance of the potential repositories in the OPA and COX formations.

At low stresses of Δσ ≤24 MPa for the OPA sandy claystone and Δσ ≤10 MPa for the COX claystone,the creep rate is almost linearly related to the stress.According to Rutter(1983)and Liu et al.(2018),the linear stress/strain rate relation implicates that the timedependent deformation of water-saturated claystone is controlled by diffusive mass transfer or pressure solution processes in interfaces between grains.One of the required conditions for diffusive transport is the existence of interparticle water-films,which must be so strongly adsorbed onto grain surfaces that they are able to support shear stress without being squeezed out.As demonstrated in Horseman et al.(1996)and Zhang(2017a,b),large amounts of the pore water in the studied claystones are strongly adsorbed between clay particles and capable of sustaining high shear stresses up to the material strength.At high stresses,the creep rate increases exponentially,as shown by the COX claystone(Fig.27).This might be contributed by more rapid slips of waterfilms at interparticle contacts and additional micro-cracking along the particle boundaries.Accumulation of the micro-cracks could lead to macro-fractures and ultimately to creep failure.This micro-structural hypothesis requests to be validated by microscopic observations.

Laboratory investigation of the long-term deformability of claystones is a great challenge due to the slow strain rates,high sensitivity of the material to environment variations,and strong thermo-hydro-mechanical-chemical coupling effects.In addition to the requirements on the short-term testing with careful sampling and reconditioning, specific methodologies are required in creep tests.They need more precisely boundarycontrolled conditions, precise measurements of deformation,pore pressure and other parameters during long time periods of months to years at each load step.Maintaining such accurate and precise measurement equipment for these long runtimes is ambitious. Micro-structural observation and analysis of the creeping material might give relevant insights to understand the micro-processes and identify the creep mechanisms.However,even after long runtimes, the sample deformations are marginally small and it is uncertain if those strains can be identified,even using electron microscopy.Based on the creep tests and micro-structure observations,constitutive models shall be developed,validated by comparison with qualified laboratory and in situ test data, and then applied for the (numerical)prediction of the long-term deformation of the clay host rock under repository conditions.

Fig.26.Creep of the saturated and dried samples under multistep triaxial loads.

Fig.27.Comparison of the stationary creep rates between the OPA sandy claystone and the clay rich OPA and COX claystones.

5. Conclusions

The mechanical behavior of the sandy facies of OPA at the Mont Terri URL was experimentally investigated with drained and undrained triaxial compression and extension,cyclic compression,and creep tests on samples parallel to bedding.In order to minimize the effects of sampling-induced desaturation and micro-fissures,most of the samples were rewetted in humid air to high saturation degrees of 85% -100% and recompacted at hydrostatic stress of 15 MPa corresponding to the maximum effective consolidation pressure experienced during the geological history.

The OPA sandy claystone under triaxial compression and extension showed the elastoplastic stress-strain behavior with volumetric compaction until onset of dilatancy at high deviatoric stresses above 80% -90% of the peak failure strength.The volume compaction before dilatancy leads to an increase in elastic stiffness.In contrast,the dilatancy results in decrease of the stiffness.The strength is dependent on the loading path in terms of Lode’s angle and mean stress.The strength under triaxial compression is higher than that under extension.The respective strength increases with increasing mean stress.The strength behavior of the claystone under triaxial compression and extension can be revealed by the Mohr-Coulomb failure model.Moreover,the claystone is highly sensitive to humidity change.The mechanical stiffness and the strength of the claystone increase with decreasing water saturation.

Time-dependences of deformation and strength of the claystone were observed.Under constant deviatoric stress,it creeps continuously with time,which can be characterized by a transient phase and a following stationary phase,and even a tertiary phase at high deviatoric stresses to rupture.The creep of the OPA sandy claystone is slower than that of the clay rich shaly facies of OPA and COX claystones.Even though some hypotheses have been proposed for creep and failure mechanisms in claystone,clear evidence is still missing and needs further precisely controlled long-term creep experiments and micro-structural analysis.

Declaration of Competing Interest

The authors wish to confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.

Acknowledgments

This work was funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) under contract number 02E11304.We would like to thank the staff of Swisstopo for their valuable on-site support.Anonymous reviewers are also acknowledged for their many helpful suggestions on the manuscript.