Novel experimental techniques to assess the time-dependent deformations of geosynthetics under soil confinement

Carina Maia Lins Costa, Jorge Gariel Zornerg

a Department of Civil Engineering, Federal University of Rio Grande do Norte, Av.Senador Salgado Filho 3000, Natal, Rio Grande do Norte, 59078-970, Brazil

b Department of Civil Architectural and Environmental Engineering, The University of Texas at Austin, 301 E.Dean Keeton, Austin, TX, 78712-0280, USA

Keywords:Geosynthetics Geotextile Creep Stress relaxation Reinforced soil Long-term deformation

ABSTRACT A new experimental approach to assess the impact of soil confinement on the long-term behavior of geosynthetics is presented in this paper.The experimental technique described herein includes a novel laboratory apparatus and the use of different types of tests that allow generation of experimental data suitable for evaluation of the time-dependent behavior of geosynthetics under soil confinement.The soil-geosynthetic interaction equipment involves a rigid box capable of accommodating a cubic soil mass under plane strain conditions.A geosynthetic specimen placed horizontally at the mid-height of the soil mass is subjected to sustained vertical pressures that, in turn, induce reinforcement axial loads applied from the soil to the geosynthetic.Unlike previously reported studies on geosynthetic behavior under soil confinement,the equipment was found to be particularly versatile.With minor setup modifications,not only interaction tests but also in-isolation geosynthetic stress relaxation tests and soil-only tests under a constant strain rate can be conducted using the same device.Also,the time histories of the reinforcement loads and corresponding strains are generated throughout the test.Results from typical tests conducted using sand and a polypropylene woven geotextile are presented to illustrate the proposed experimental approach.The testing procedure was found to provide adequate measurements during tests, including good repeatability of test results.The soil-geosynthetic interaction tests were found to lead to increasing geotextile strains with time and decreasing reinforcement tension with time.The test results highlighted the importance of measuring not only the time history of displacements but also that of reinforcement loads during testing.The approach of using different types of tests to analyze the soil-geosynthetic interaction behavior is an innovation that provides relevant insight into the impact of soil confinement on the time-dependent deformations of geosynthetics.

1.Introduction

Geosynthetics have become widely used in the construction of reinforced soil retaining structures.Yet, geosynthetic materials exhibit a time-dependent stress-strain behavior that cannot be overlooked when designing structures such as geosyntheticreinforced soil (GRS) walls (e.g.Benjamim et al., 2007; Yang et al.,2009; Costa et al., 2016).Assessment of the time-dependent behavior of geosynthetics is important for design of reinforced soil structures since reinforcements are expected to remain under tension throughout the design life of the structure.

The stress-strain-time behavior of geosynthetics in isolation has been recognized as a complex mechanism, and two complementary viscous behaviors, i.e.creep and stress relaxation, have been typically evaluated.Creep involves the development of timedependent deformations in a geosynthetic under constant axial loading.On the other hand, stress relaxation corresponds to the time-dependent decrease in unit tension in a geosynthetic subjected to imposed constant strains.

Studies on geosynthetic long-term deformation in insolation were conducted by Bueno et al.(2005), Yeo and Hsuan (2010),Kongkitkul et al.(2014), Nuntapanich et al.(2018), Pinho-Lopes et al.(2018), and Dias Filho et al.(2019).Factors such as polymer type, temperature and loading rate are known to affect geosynthetic viscous response.The influence of the soil confinement on the time-dependent deformation of geosynthetics has also been evaluated in the literature(e.g.McGown et al.,1982;Levacher et al.,1994; Wu and Hong, 1994; Sawicki and ′Swidzi′nski, 1999; Becker and Nunes, 2015).However, there has been little consensus on the significance of the impact of the soil confinement on the viscous response of a geosynthetic-reinforced soil mass.While some authors have reported a significant reduction in time-dependent strains due to the soil confinement (e.g.McGown et al., 1982),other researchers have observed only minor effect due to such confinement(e.g.Levacher et al.,1994;Wu and Hong,1994;Becker and Nunes, 2015).

The effect of soil confinement on the time-dependent response of geosynthetics has been experimentally investigated using two reasonably different approaches, depending on the loading approach adopted to mobilize tension in the confined geosynthetic reinforcement.One type of device described in the literature allows for testing of geosynthetics placed between two layers of soil that are not allowed to deform during testing.After a target confining pressure is applied to the soil,tensile forces are applied directly to the reinforcement.This technique has been used exclusively to evaluate the potential effect of soil confinement to reduce timedependent deformations, with focus on assessing the impact of the type or structure of the geosynthetic under the soil confinement.Reduced time-dependent deformations have been typically reported for the case of nonwoven geotextiles due to the restricted movement of the fibers and their alignment in the direction of loading as a result of the soil confinement.This type of device was reported by McGown et al.(1982), Levacher et al.(1994), Wu and Hong (1994), and França and Bueno (2011).The equipment used to conduct tests following the first approach has some advantages in comparison with the second one.It is less complex to develop,the tests are typically easier to perform, and the results are less difficult to interpret.However, it has the disadvantage of not allowing evaluation of the effect of soil rheologic properties on geosynthetics’ long-term behavior.

The second approach involves a device in which a geosynthetic is confined between two soil layers but;in this case,the soil layers experience deformation.A constant vertical pressure is applied to the soil,which in turn mobilizes the axial load in the reinforcement.This testing approach allows both soil and reinforcement to exhibit time-dependent deformations and accounts for soil-geosynthetic interaction.This type of equipment is deemed to simulate closely the typical load transfer mechanism anticipated in actual geosynthetic-reinforced soil structures, a mechanism involving tensile loads that are induced in the reinforcement by soil stresses.This approach makes it feasible to evaluate the effect of soil rheologic properties on geosynthetics’ long-term behavior.This is especially important considering that some authors have emphasized that the time-dependent deformation of geosynthetics is affected by the rheologic characteristics of the confining soil (e.g.Boyle and Holtz,1996; Wu and Helwany,1996; Liu et al., 2009; Li et al., 2012).Wu and Helwany (1996) reported that if the confining soil exhibits less of a tendency to creep than the geosynthetic, the soil would then impose a restraining effect on the deformation of the geosynthetic.Studies involving centrifuge tests have not shown such restraining effect when geosynthetic is confined with soils of low creep potential, such as sands, in comparison with clayey soils(e.g.Costa et al.,2016).Also,in addition to showing contradicting conclusions, the number of studies that investigated this loading approach was limited.

Nonetheless, equipment based on the second approach aforementioned has provided important contributions to the assessment of long-term soil-geosynthetic interaction behavior(e.g.Wu and Helwany, 1996; Helwany and Shih, 1998; Ketchart and Wu,2002).The main differences amongst different devices in this category are the specimen dimensions and testing instrumentation.Only the equipment described by Wu and Helwany (1996) and Ketchart and Wu(2002)focused on assessing the time-dependent strains developed in the reinforcement.The device proposed by Helwany and Shih (1998) measured lateral displacements at the mid-height of the soil-geosynthetic facing and tensile forces in the reinforcement over time.However, displacements were measured for the facing specimen on one end of the geosynthetic, and axial force (but not strain) was recorded on the other end; measurements of the reinforcement axial tension and corresponding longterm strain could not be obtained.

This paper presents an experimental approach that includes a new apparatus and the use of different types of tests that collectively provide a robust set of data on long-term deformations of geosynthetics.Unlike previous studies, the device can perform soil-geosynthetic interaction tests by providing geosynthetic strains and the corresponding tension with time.The device is also capable of testing the soil and reinforcement separately, which is important to ultimately characterize the soil-reinforcement interaction.The ability of testing the unreinforced soil at a constant strain rate constitutes a unique experimental feature of this study.Typical results obtained using this approach are presented for a sand and a polypropylene woven geotextile.

2.Test apparatus

The equipment developed was used to perform all types of tests presented herein, except unconfined geosynthetic creep tests,which were performed using a traditional apparatus in accordance with ISO 13431 (1999).

2.1.Setup for soil-geosynthetic interaction test

A schematic of the equipment is shown in Fig.1.In developing the equipment, the researchers’ main objective was to design a device capable of simulating the load transfer mechanism anticipated in GRS structures and being able to obtain both tensile forces and corresponding strains in the reinforcement over elapsed time.The objective of the proposed soil-geosynthetic interaction test was not to anticipate strains in any particular GRS wall, but to identify the effect of soil confinement on reinforcement timedependent behavior.

The apparatus involves a rigid box capable of accommodating a cubic soil mass under plane strain condition with side length measuring 200 mm.A geosynthetic layer secured via clamps isplaced horizontally at the mid-height of the soil mass.The device is suitable to study the time-dependent behavior of geotextiles and geogrids.A pressurized air system was used to apply sustained vertical pressure at the top and bottom of the soil mass.The equipment is capable of measuring displacements along the geotextile layer to obtain reinforcement strains.The interfaces between the soil and walls of the box were lubricated to minimize interface friction.

Fig.1.Schematic of proposed equipment (three-dimensional view).

Fig.2 presents a schematic side view of the test box,which has two movable sidewalls composed of 13-mm thick steel plates.The movable sidewalls, labeled S1 and S2 in Fig.2, are connected to a lateral steel frame and the linear low friction bearings are used to facilitate movement of sidewalls S1 and S2.The upper and lower portions of the movable sidewall S1 displace outward together.Sidewall S2 has the same characteristics as sidewall S1 but it moves in the opposite direction.The other sidewalls are stationary and consist of 30 mm-thick glass plates attached to a rigid steel frame.Transparent sidewalls are used to allow visualization of the specimen during testing.As shown in Fig.2, a load cell is installed to record the force in the geosynthetic.

The loading mechanism implemented in the test apparatus is illustrated in Fig.3.The vertical pressure(σv) is initially applied to the soil through the pressurized air system, resulting in the development of horizontal stresses within the soil mass and consequently horizontal pressure (σh) on the sidewalls.The horizontal pressure acting on the movable sidewall S1 is indicated in Fig.3 via the corresponding resultant forces Fh1and Fh2,which are induced by σhon the upper and lower parts of S1, respectively.Forces Fh1and Fh2are then transferred to the geotextile reinforcement through the clamp-frame connection shown in Fig.3, which connects the lateral frame to the clamp(and is the only connection between the movable sidewall and clamp).With the load transfer mechanism facilitated by the equipment design, the geosynthetic deforms but only because of displacement of the soil and movable sidewall S1.Since the previously cited clamp-frame connection is the only connection between the movable sidewall and the reinforcement, the load cell, positioned between that connection and the clamp, is able to record the total resultant force (Fh1＋Fh2)acting on the reinforcement at point P (Fig.3).

Incorporating the ability to measure reinforcement tensile loads over time to the device was the main challenge in equipment development.This capability was realized through the device’s unique design involving movable sidewalls, the clamp-frame connection system and the location and characteristics of the load cell.

Fig.2.Schematic of test box (side view, not to scale).

Fig.3.Load-transfer mechanism for soil-geosynthetic interaction tests showing forces acting on movable sidewall S1 and corresponding reinforcement reaction force(not to scale).

The load cell used for the device is manufactured by Precision Transducers Ltd.,and has a load capacity of 2.5 kN with a resolution of 0.25 N.Calibration tests showed excellent repeatability and no hysteresis.Additional tests carried out with a sustained constant load applied for extended duration demonstrated good performance over time.For example, an applied load of 1 kN that was maintained constant for 48 h showed an oscillation of only 2.5 N.

Instrumentation also included telltales connected to the geosynthetic at four different points, labeled A, B, C and D in Fig.4, to obtain geosynthetic strains.The telltales consisted of stainless steel wires with a diameter of 0.6 mm to minimize soil disturbance.The wires were lubricated with silicone grease and inserted into protective stainless steel tubes of 0.8 mm in diameter.An electrical device with a resolution of 0.01 mm monitored the telltale displacements in order to calculate strains in segments AB,BC,and CD,as shown in Fig.4.

The instrumentation used for the various device setups, with and without reinforcement, was established considering the symmetry of the box and specimen(see the axis of symmetry in Figs.3 and 4).During equipment development,results of calibration tests using load cells on both sidewalls (S1 and S2) were found to be remarkably similar, with equally good results obtained when evaluating the consistency of telltale measurements.

2.2.Setup for tests using soil only

The equipment is capable of performing tests using soil only by making minor modifications to the setup as previously describedfor soil-geosynthetic tests.For tests using soil only,the soil stressstrain behavior is evaluated for short-term conditions separately,without geosynthetic.This is important to gain insight into the behavior of a reinforced soil mass, which depends not only on the reinforcement but also on the soil properties.

Fig.4.Positions of telltale-geosynthetic connections (not to scale).Unit in mm.

Testing of soil only was conducted by applying a sustained vertical pressure to the top and bottom of the soil mass, while allowing movable sidewalls S1 and S2 to displace outward.Fig.5 presents the device setup for tests using soil only.To allow movement of sidewalls S1 and S2 at a constant displacement rate, each movable sidewall was coupled to an individual shaft with roller bearings,as shown in Fig.5.The movement directions of S1 and S2,used in the soil-geosynthetic interaction test(see Fig.2),have been adopted in soil-only tests.Belt drives connect the individual shafts to a main shaft supported by roller bearings.A low-speed motor(100 resolutions per minute (rpm)) with a set of gears rotates the main shaft to move sidewalls S1 and S2 simultaneously at a constant speed.Displacement rates of 0.1 mm/min,0.7 mm/min,1 mm/min and 4 mm/min were considered depending on the selected set of gears.A dial gage was used to measure the displacements of the movable sidewalls with a resolution of 0.01 mm.As the movable sidewalls displaced, the total resultant load acting on the movable sidewall S1 was recorded by the load cell.

The ability of the equipment to conduct tests using soil only allowed evaluation of the soil response under the same conditions used for soil-geosynthetic tests, that is, the same specimen dimensions, the same system for applying vertical pressure and similar boundary conditions.

2.3.Setup for stress relaxation tests

For unconfined stress relaxation tests, the reinforcement was placed horizontally inside the box and attached to the clamps.The target load was then achieved by displacing the movable sidewalls S1 and S2,after which the strain is kept constant.The load cell was used to record the force in the reinforcement over time.The load cell in tests conducted using this configuration was positioned as indicated in Fig.3, similar to the case of soil-geosynthetic tests.

3.Description of the experiments

3.1.Geotextile and sand used in specimen preparation

Fig.5.Schematic of device setup for testing soil only (top view, not to scale).

Testing was carried out using a woven geotextile and sand as the confining soil.The geotextile was a polypropylene woven geotextile with a mass per unit area of 126 g/m2.Table 1 presents the average unconfined ultimate tensile strength for the geotextile, the corresponding strain at failure and the geotextile secant tensile stiffness at 2% strain.These values were obtained from wide-width strip tensile tests conducted in the machine direction in accordance with ASTM D4595-17 (2017).

Fig.6 presents the results of a series of conventional creep tests conducted without soil confinement,in accordance with ISO 13431(1999).The applied load levels corresponded to 5%,12%and 20%of the geotextile ultimate tensile strength (Tult).The in-isolation geotextile creep can be evaluated by fitting the logarithmic function indicated in Eq.(1).The regression line obtained for each load level is also presented in Fig.6.The coefficient of determination(R2)was equal to 0.91, 0.99 and 0.99 for 5%, 12% and 20% of Tult,respectively.As expected, larger load levels led to higher creep rates,as shown by the values of α in Fig.6.

where εris the reinforcement time-dependent strain, ε1is the reinforcement strain for t ＝ 1 h,α is the creep rate,and t is the time.

The soil is poorly graded quartz silica sand with rounded to sub-rounded particles, and is classified as SP according to the unified soil classification system (USCS).The sand has an average particle size of 0.23 mm, a coefficient of uniformity of 2.7, and a coefficient of curvature of 1.09.Fig.7 presents the gradation curve for the sand.The unit weight of the sand is 26.5 kN/m3,and the minimum and maximum dry unit weights are 14.2 kN/m3and 17.7 kN/m3, respectively.The sand maximum void ratio(emax) equals 0.87 and its minimum void ratio (emin) is 0.5.Shear strength properties of the sand were obtained from a triaxial testing program.For the selected relative density used in the tests presented herein (Dr＝50%), the sand shows peak friction angles (φp) of 36°.

3.2.Testing program

3.2.1.Tests using sand only and sand-geotextile interaction tests

Table 2 provides a summary of several tests conducted using the proposed equipment.The initial character of the test designations shown in Table 2 corresponds to the type of test (“S”stands for the tests with soil only and “SG” for the sandgeotextile tests), followed by the values of applied vertical pressure.Three different values of applied vertical pressure (100 kPa,150 kPa and 200 kPa) were adopted in the testing program and some tests were conducted specifically to evaluate the repeatability of the results.To denote repeat tests, the letter “R" was appended to the test designation.For tests using sand only, most tests were performed using a displacement rate equal to 0.7 mm/min.To indicate a different speed value, the letter “V” was added to the test designation, followed by No.2 (V ＝0.1 mm/min) or No.3 (V ＝4 mm/min).The specimen was prepared with a soil relative density (Dr) of 50% for tests using sand only and for sand-geotextile interaction tests.

Table 1Wide-width tensile test results for geotextile reinforcement.

Fig.6.Unconfined creep results for geotextile at different levels of ultimate tensile strength (Tult).

Fig.7.Gradation curve for the tested sand.

3.2.2.Geotextile stress relaxation tests

The scope of the stress relaxation tests is presented in Table 3.Test designations denote the type of test(“R”for stress relaxation)and testing sequence.

3.3.Testing procedure

3.3.1.Soil-geosynthetic interaction tests

This section describes different stages during testing, including lubrication of the soil specimen interfaces, placement of soil and geosynthetic,loading application and time-dependent monitoring.

(1) Lubrication of the soil specimen interfaces

Lubrication between the soil specimen and the device sidewalls was achieved using latex membrane and silicon grease to minimize interface friction.The same silicone grease used by Tatsuoka et al.(1984) was adopted in this study.The interface friction angle was estimated from the results of interface shear tests conducted using a direct shear test device (100 mm × 100 mm in plane) typically employed to evaluate strength properties of soils.

For a displacement rate of 1 mm/min,the friction angle between the lubricated latex membrane and glass was 0.4°for a vertical pressure of 50 kPa.To minimize the increase in friction angle after keeping the vertical pressure constant for extended periods of time,polytetrafluoroethylene(PTFE)powder(also known as Teflon)was added to improve grease performance over time.The use of PTFE powder was reported to reduce the flow of grease and consequently the changes in friction angle over elapsed time (Tatsuoka et al.,1984).

DuPont Teflon powder composed of microspheres of 20 μm in diameter was selected and the mixtures containing 20%, 30%, 40%and 50% PTFE power were tested.The percentage of PTFE corresponds to the weight of the powder in relation to the total weight of the blend (grease ＋PTFE).Table 4 summarizes the results of the interface friction angle for silicone grease with different percentages of PTFE powder.The same setup previously described was used for testing silicon grease with PTFE powder added.A sustained vertical pressure equal to 50 kPa was applied.

As shown in Table 4,the use of PTFE powder increased the value of the friction angle in relation to pure silicon grease.However,increasing PTFE dosage was found to decrease friction angle variations over time.Upon evaluation of the interface test results,20%PTFE dosage was selected as it exhibited a significantly smaller variation in friction angle over time than pure grease while presenting the lowest friction angle among the tested dosages.

(2) Placement conditions for soil and geosynthetic

Preparation of the soil-geosynthetic specimen after lubrication involved three main stages: placement of the first soil layer,placement of the geosynthetic with telltales attached, and placement of the top soil layer.The displacement of the movable sidewalls was restrained throughout specimen preparation.

To achieve the soil target density, the layers were placed and compacted by pluviating dry sand at controlled combinations of flow rate and drop height using a device constructed for sand pluviation(raining).This device was designed following recommendations reported by Rad and Tumay (1987).The sand raining apparatus consists of a steel frame with a container at the top and two sieves(with holes measuring 6.3 mm in diameter)located underneath.The set of sieves, or diffuser, is fixed to the steel frame to facilitate achieving specimens that are homogeneous.The sand raining procedure involves releasing the sand stored in the container,which then passes through the diffuser and into the test box.

The bottom of the container consists of a perforated cover(shutter)that releases the sand through the opening of a trapdoor.The relative density is mainly determined by the flow rate and drop height of the sand.Soil specimens with different densities can be prepared for soil-geosynthetic interaction tests by maintaining a constant distance between the diffuser and top of the test box(drop height) and changing the flow rate of the material.Flow rate variation was accomplished by using shutters with varying hole diameters.For the tests presented in this paper, shutter holes measuring 13 mm in diameter were used to achieve a target relative density of 50%.The shutter holes had a uniform pattern and spacing of 25 mm.The distance from the diffuser to the top of the box was 650 mm.A margin of error of less than 3%relative density was quantified for the target relative density using the device constructed for sand pluviation.

(3) Loading application and time-dependent monitoring

Testing was initiated by applying pressure through the air bladders to induce a preload in the reinforcement of about 1%of its tensile strength, and the corresponding displacements in the telltales and movable sidewall S1 were recorded.After application ofthe preload,the vertical pressure was gradually increased until the target value was reached, after 1 min approximately.Once the loading process was completed, the displacement data (telltales and movable sidewall)were continually collected for up to 1000 h of testing duration.

Table 2Tests carried out using soil only and using sand-geotextile specimens.

Table 3Stress relaxation tests.

Table 4Interface friction angle for silicone grease with PTFE powder.

This procedure is consistent with guidelines provided for geosynthetics creep test standards (e.g.ISO 13431,1999).For the tests conducted as part of the study presented in this paper, displacement data were recorded at 1 min, 2 min, 4 min, 8 min, 15 min,30 min, 1 h, 2 h, 4 h, 8 h and 10 h after the prescribed vertical pressure had been applied.The tensile force in the geotextile at point P was recorded at the same intervals.

3.3.2.Tests using soil only

A single layer of sand 200 mm thick was prepared for testing soil without reinforcement and specimen preparation was similar to that for soil-geosynthetic tests.After applying a sustained vertical pressure,the motor system began displacing the movable sidewalls at a constant rate.The horizontal resultant force acting on the movable sidewall S1 was also monitored during testing.

3.3.3.Stress relaxation tests

For the unconfined stress relaxation test,the reinforcement was loaded for a period of 1 min to achieve the target strain.The strain was then kept constant as the load cell recorded the reinforcement force over elapsed time.For the tests presented herein,the load was recorded at 1 min,2 min,4 min,8 min,15 min,30 min,1 h,2 h,4 h,8 h and 10 h after the prescribed strain had been achieved.

4.Typical test results and discussion

Results from tests conducted using the developed apparatus are presented in this section.The room temperature was controlled,and monitored throughout testing,presenting only slight variations within the limit of fluctuations recommended for unconfined creep tests with geosynthetics (e.g.ISO 13431, 1999; ASTM D5262-07(2016), 2016).

Fig.8 shows the results recorded by the load cell for the tests with soil only.The load values in Fig.8 were calculated per unit width of movable sidewall.This approach was adopted to allow comparison of the results with those from the tests with geosynthetics, whose results are typically presented in units of force per unit width of reinforcement.Strains were calculated using the displacement of movable sidewall S1 and the corresponding initial specimen length (equal to 100 mm, considering the axis of symmetry shown in Fig.5).

As shown in Fig.8,the load initially decreased to a strain of 1%-2% approximately.For tests with the lowest vertical pressure(σv＝ 100 kPa), the force tended to remain constant after a minimum value was reached.For the other tests, after reaching the minimum load value,there was an increase in the load with strain,followed by a tendency to stabilize as the specimen deformed.This behavior is consistent with that anticipated for the mobilization of soil shear strength.Initially,the mobilized shear strength increases with the specimen deformation until its peak value is reached (as indicated in Fig.8 by the minimum value recorded by the load cell).As strain increases, the mobilized friction angle decreases toward the value corresponding to the critical state condition and consequently the load cell records an increase in the load acting on the movable sidewall S1, as indicated in Fig.8 for tests S-200 and S-200-R.Test results showed good repeatability, demonstrated by comparing results of tests S-100 with S-100-R and S-200 with S-200-R, as shown in Fig.8.

Fig.8 also displays the effect of displacement rate on load results,showing tests conducted at rates of 0.7 mm/min(test S-100),0.1 mm/min (test S-100-V2) and 4 mm/min (test S-100-V3).Despite substantial variation in displacement rate, only minor effects were observed in the test results.Nevertheless,the ability to perform tests at different speeds was important for analysis ofthe data from soil-geotextile interaction, as it was particularly relevant to identifying the sources of time-dependent response in those tests.

Fig.8.Lateral load on the movable sidewall S1 versus soil strain for different levels of vertical pressure and displacement rates (tests with soil only).

The soil-only tests showed negligible impact of displacement rate on the test results,suggesting a low creep potential of the sand used in this study for the tested conditions.However, it should be noted that granular materials have been reported to be capable of presenting creep(Wang et al.,2011;Karimpour and Lade,2013;Lv et al., 2017; Levin et al., 2019).Nevertheless, sand creep is not as significant as that usually anticipated for other soils such as clays.

Fig.9 presents the changes in strain over time for soilgeotextile interaction tests (SG-100 and SG-200) obtained using external and telltale measurements.The term “external” in the figure’s caption refers to as specimen deformations defined using the displacements of movable sidewall S1 as registered by the dial gage.Telltale displacements were used to calculate the strains,taking into account the deformation of segments AB, BC and CD(Fig.4).For example,the strain values for segment AB correspond to the relative displacement of telltales positioned at points A and B,divided by the length of this segment.

The data in Fig.9 exhibit a uniform strain pattern along the reinforcement length,which is consistent with the good agreement of strains for external and internal displacement measurements.The coefficient of variation (COV) of the strain at 1 min, obtained from external and internal measurements, was approximately 3%and 3.6% for tests SG-100 and SG-200, respectively.These COV values are of the same order of magnitude as those obtained for strain at failure shown in the wide-width tensile tests(see Table 1).

Fig.10 displays the strain over time for all soil-geotextile interaction tests, and illustrates the good repeatability of the results.These strains correspond to the mean values considering the previously cited different approaches (external and telltale measurements).Regression lines for experimental data using the logarithmic function presented in Eq.(1)are also shown in Fig.10.The maximum difference in time-dependent strain rate is about 5%between tests SG-100 and SG-100-R (see values of α in Fig.10).

The load response over time obtained from the soil-geotextile interaction tests can be seen in Fig.11.The initial load (t ＝1 min)recorded in tests with reinforcement was remarkably similar to that recorded on the movable sidewall S1 in tests with soil only,for the same vertical pressure and magnitude of horizontal strain.As a reference,the initial load shown in Fig.11 for test SG-100 was found to be 2.4 kN/m, corresponding to a horizontal strain of approximately 3%.The load for test S-100,involving the same strain level,was found to be 2.35 kN/m (see Fig.8).

Fig.9.Reinforcement strains over time using external and internal measurements(sand-geotextile interaction tests SG-100 and SG-200).

Fig.10.Mean values of time-dependent reinforcement strains with time from sandgeotextile interaction tests.

A similar behavior was observed for the tests with higher vertical pressure (SG-200 and S-200).The initial force in the reinforcement and the expected load based on the test with sand only were analogous.The good agreement of the load data between the tests with soil only and the soil-geotextile interaction tests provides additional evidence on the consistency of the data presented herein.

In Fig.11,a clear decrease in the load over time can be observed.For the tests conducted using σv＝ 100 kPa, an approximate reduction of 10% in the load occurred after 10 h.For the test conducted using higher vertical pressures (SG-200), the load value decreased by 20% over 10 h of testing.This indicates that maintaining a constant vertical pressure over time to simulate the field conditions presented in GRS walls does not guarantee a constant load in the reinforcement.A time-dependent load reduction may occur in the geosynthetic as specimen deforms.Since conventional in-isolation geosynthetic creep test involves the development of deformations over time under a constant axial loading, the reinforcement strains using the soil-geosynthetic interaction device should not be referred to as creep strains.

Figs.10 and 11 show that neither the strain nor the load remains constant over time,which means that an intermediate viscoelastic behavior between creep and stress relaxation occurred.This finding suggests that not only in-isolation creep tests but also stress relaxation tests are relevant for an accurate interpretation of the confined tests in order to broaden our understanding of sandgeotextile interaction behavior.

The results of the geotextile stress relaxation tests presented in Fig.12 demonstrate the significant potential to decrease the load in the reinforcement as its deformation is restrained.The results for test R1, for example, show that the force induced in the reinforcement after 10 h of testing decreased 50% in relation to the value recorded at 1 min.

Fig.13 depicts the geotextile strains over time for interaction tests SG-100 and SG-200,both performed under soil confinement,in comparison to those from conventional unconfined creep tests.The initial load(t ＝1 min)of the interaction test SG-100 is similar to the applied load of the unconfined creep test performed with 12% of the geotextile ultimate tensile strength (Tult).Similarly, theinitial load of the interaction test SG-200 corresponds to the applied load of the unconfined creep test performed with 20% of Tult.

Fig.11.Reinforcement load over time from sand-geotextile interaction tests.

Fig.12.Reinforcement load with time from stress relaxation tests.

The geotextile strains for the confined test SG-100 shown in Fig.13 were very similar to those without soil confinement.For test SG-200, the reinforcement strain at 10 h of testing was approximately 10% smaller under soil confinement as compared with the conventional creep test.This is found to be a consistent behavior,since a comparatively larger decrease in the load (approximately 20%) is observed throughout the test(Fig.11).

The behavior of reinforcement load with time obtained from sand-geotextile interaction tests and unconfined stress relaxation tests is displayed in Fig.14.The initial load (t ＝1 min) of the interaction test SG-100 is similar to the load of the unconfined stress relaxation test R2.Tests SG-200 and R1 also present similar values of initial load.Although both types of tests (soil-geosynthetic interaction tests and stress relaxation tests) show decrease in load over elapsed time, the load paths were significantly different.This behavior indicates that load reduction in interaction tests was not as high as anticipated if soil strain had been fully prevented as in the stress relaxation tests.This is consistent with the better agreement between soil-geosynthetic interaction tests and unconfined creep tests.

Fig.13.Time-dependent strains obtained from sand-geotextile interaction tests and unconfined creep tests.

Fig.14.Reinforcement load over time from sand-geotextile interaction tests and unconfined stress relaxation tests.

Table 5 presents the reinforcement secant stiffness (J10) at 10 h calculated by Eq.(2) for different types of tests (unconfined creep tests,stress relaxation tests and sand-geotextile interaction tests).The same equation can be used for different types of tests,although the magnitude of creep stiffness, for example, is not necessarily equal to the stress relaxation value.The initial load(t ＝1 min)for all tests presented in Table 5 is approximately the same.

where J10is the reinforcement stiffness for t ＝10 h, T10is the reinforcement load for t ＝ 10 h,and ε10is the reinforcement strain for t ＝10 h.

Table 5Reinforcement stiffness (J10)at 10 h of testing.

The effect of soil confinement due to the textile structure of geotextiles (woven, nonwoven, knitted) is expected to cause an increase in geosynthetic stiffness.However, despite the smaller strains over time under the soil confinement, the magnitude of secant stiffness(J10)obtained from the sand-geotextile interaction test was not increased.In fact,the stiffness value obtained from the confined tests lies between the values obtained from the unconfined creep test and that from the unconfined stress relaxation test.Thus, the smaller strain in test SG-200 for the geotextile under confinement after 10 h of testing is due to the load reduction throughout the test, as shown in Fig.11, but not because of the confinement effect due to the geotextile structure.The decrease in the load is believed to occur because of the effect of the soil rheologic properties on the geosynthetics’ long-term behavior.This finding from the data presented in Table 5 illustrates the importance of using different types of tests to analyze the effect of soil confinement on time-dependent geosynthetic strains.

Reinforcement strain at 10 h in test SG-200 was approximately 10%smaller under soil confinement as compared with the result of the corresponding conventional creep test.However,the reduction in the long-term strain was also not as high as anticipated,considering that the soil used in this study belongs to a group of soils frequently considered to have negligible creep.Since the use of different types of geotextiles and soils may lead to different behaviors, this finding is limited to the materials and test conditions used in the present study.Overall, it should be noted that the geosynthetic time-dependent response obtained from different types of tests, as presented herein, is capable of providing significant insight into the long-term response of confined geosynthetics.

5.Conclusions

This paper described a unique test device and the integration of results from different types of tests to investigate geosynthetic long-term deformation under soil confinement.The following conclusions can be drawn:

(1) The device was found to provide results consistent with the expected trends for tests conducted with or without reinforcement.The results of tests with soil only were consistent with the behavior expected from conventional soil shear strength tests.Duplicate tests showed good repeatability,providing additional indication of the adequacy of the specimen preparation and testing procedures.

(2) The soil-geosynthetic interaction tests showed that the initial tension in the reinforcement did not remain constant with time for a constant applied vertical pressure.This result highlights the importance of measuring the reinforcement load over time in order to properly interpret the results of geosynthetic tests under soil confinement.

(3) All tests conducted under soil confinement indicated timedependent reinforcement strains and the tension in the reinforcement decreased during the interaction tests.The geosynthetic ultimate response was neither creep nor stress relaxation, but an intermediate viscoelastic behavior.Thus,not only in-isolation creep tests but also stress relaxation tests are relevant for interpretation of the confined tests in order to further understand the soil-reinforcement interaction response.

(4) The approach of using different types of tests was found to be truly relevant to analyzing experimental data on the complex soil-geosynthetic interaction behavior.For the geotextile tested in this study, a reduction of time-dependent strains under soil confinement occurred in comparison with inisolation creep tests.The integration of various types of tests revealed that long-term strains were not reduced because of the effect of soil confinement due to the geotextile structure.Smaller strains over time obtained under soil confinement occurred as a result of the load decrease throughout the test.

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

This paper is dedicated to Benedito Bueno (in memoriam), a brilliant researcher and former professor at University of São Paulo,to whom the authors are indebted for his vision and enormous contribution during the development of this work.This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil(finance code 001).

List of notations

COV Coefficient of variation (%)

DrSoil relative density (%)

emaxMaximum void ratio of sand

eminMinimum void ratio of sand

Fh1Resultant force on the upper part of the movable sidewall

Fh2Resultant force on the lower part of the movable sidewall

GRS Geosynthetic-reinforced soil

J10Reinforcement stiffness for t ＝10 h (kN/m)

PTFE Polytetrafluoroethylene

SP Poorly graded sand

USCS Unified soil classification system

t Time (in min or h)

TultUltimate geotextile tensile strength (kN/m)

T10Reinforcement load for t ＝10 h (kN/m)

α Creep strain rate, or time-dependent strain rate(%/h)

εrReinforcement strain (%)

ε1Reinforcement strain for t ＝ 1 h (%)

ε10Reinforcement strain for t ＝ 10 h (%)

φpPeak friction angle of soil (°)

σhHorizontal pressure(kPa)

σvVertical pressure (kPa)