Engineering Geology 172 (2014) 1–11

Contents lists available at ScienceDirect

Engineering Geology

j ourna l homepage: www.e lsev ie r .com/ locate /enggeo

Residual shear strength of unsaturated soils via suction-controlled ringshear testing

Laureano R. Hoyos a,⁎, Claudia L. Velosa b, Anand J. Puppala a

a Dept. of Civil Engineering, Univ. of Texas at Arlington, Arlington, TX 76019, United Statesb Fugro Geoconsulting, Houston, TX 77081, United States

⁎ Corresponding author.E-mail addresses: lhoyos@uta.edu (L.R. Hoyos), clilian

anand@uta.edu (A.J. Puppala).

0013-7952/$ – see front matter © 2014 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.enggeo.2014.01.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 January 2013Received in revised form 8 January 2014Accepted 10 January 2014Available online 24 January 2014

Keywords:Unsaturated soilMatric suctionResidual shear strengthRing shear testing

Results from a comprehensive series of suction-controlled ring shear tests, conducted on statically compactedspecimens of silty clayey sand and silty sand, are presented. The experiments were accomplished in a newlydeveloped servo/suction-controlled ring shear apparatus suitable for testing unsaturated soils under large defor-mations and suction-controlled conditions via the axis-translation technique. The presentwork focuses primarilyon two crucial aspects of compacted unsaturated soil behavior, namely, the behavior of silty clayey sand undersuction-controlled ring shear testing, and the effects of pre-shearing and suction histories on unsaturatedresidual shear strength of compacted silty sand. Test results corroborate the important role played by matricsuction on residual shear strength properties of unsaturated soils. For the range of net normal stresses andsuction states investigated, the increase in residual shear strength with increasing suction was confirmed to bea linear trend for silty sand, but significantly nonlinear for silty clayey sand. Results from multi-stage ring sheartests confirmed that the residual shear strength of unsaturated soils is virtually independent of the pre-shearing and suction histories experienced by the soil.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

A vast majority of the geotechnical infrastructure made of compactedsoil, or resting on unsaturated ground, undergoes a wide range of defor-mations. Calculation of foundation settlement, for instance, requires agood estimation of soil stiffness at relatively small strains. Analysis ofslopes, embankments, and soil bearing capacity, on the other hand,requires good estimations of shear strength from peak to residual. Todate, however, there is very limited experimental evidence of unsaturatedsoil behavior under large deformations, and the corresponding residualshear strength properties, while the soil is being subjected tocontrolled-suction states. This type of research has been deterred in thepast by the lack of suitable testing tools and techniques. It is, therefore,in this context that a suction-controlled ring shear (RS) apparatuswould play a fundamental role in the thorough characterization of thistype of geomaterials.

Only a short handful of researchers have recently begun experimentaltrials with new test methodologies, including Vaunat et al. (2006, 2007),Infante Sedano et al. (2007) andMerchán et al. (2011). Themain focus ofthese pioneering efforts has been on adapting and expanding the capabil-ities of existing Bromhead-type devices for soil testing under controlled-suction states via vapor-transfer or axis-translation techniques. Despitethe crucial findings of these dedicated few, highlighting the key roleplayed by matric suction, a comprehensive experimental effort has yet

av@hotmail.com (C.L. Velosa),

ghts reserved.

to be undertaken toproduce a thorough set of suction-dependent residualfailure envelopes for soils tested at a relatively wide range of low suctionstates (i.e., from0 to100 kPa); and to investigate the effects that bothpre-shearing and suction histories may have on the residual shear strengthproperties of unsaturated soils. The present work is motivated by theseresearch needs.

A comprehensive series of suction-controlled RS tests, performed onstatically compacted specimens of silty clayey sand and silty sand, hasbeen undertaken. The work focuses primarily on two crucial aspects ofcompacted unsaturated soil behavior: (1) behavior of silty clayey sandunder suction-controlled ring shear testing, and (2) effects of pre-shearing and suction histories on unsaturated residual shear strength ofsilty sand. The experiments were conducted in a newly developedservo/suction-controlled RS apparatus that is suitable for testingunsaturated soils under large deformations and suction-controlledconditions via the axis-translation technique. A detailed description ofits full development, including main components and a thoroughperformance verification against the original Bromhead device(Bromhead, 1979), is reported by Hoyos et al. (2011).

The RS unit is suitable for testing both fully-softened shear strengthand residual shear strength parameters that can be used for slopestability assessments of various scenarios (ASTM Standard D6467-06a,2006; ASTM Standard D7608-10, 2010). The main focus of the currentwork is on unsaturated residual shear strength at large displacements,whichwill transpire, for instance, in unsaturated earth slopes experienc-ing seismic events, or shallow and relatively deep foundation soils aboveground water table. On the other hand, accurate assessments of unsatu-rated residual strength parameters is of extreme importance in natural

2 L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

slopes in fissured rocks with unsaturated clayey and silty fills that canundergo significant shear strength changes uponwetting, or shallow fis-sured landslides that can also be activated by wetting. Unsaturatedresidual shear strength can also be used as a macroscopic indicator ofthe nature of micro-structural changes experienced by the soils whensubjected to drying (Merchán et al., 2011).

Silty clayey sand and silty sand were selected as the test materials forthe present work mainly because of their poor gradation and lowplasticity, which minimize the effects of particle size and shape onmenisci formation in the pore-water phase, hence considerablyreducing the time needed for pore-fluid (air andwater) equalization dur-ing suction-controlled RS testing. On the other hand, the relatively signif-icant content of fine-grained soil in both test materials is expected tominimize the potential for wall-friction effects between the specimenand the concentric rings of the RS device. Had the authors chosen toinvestigate residual shear strength of a purely clayey soil, a crucial topicin its own right, the time frame would have been prohibitively long, par-ticularly the multi-stage RS testing program undertaken to study theeffects of pre-shearing and suction histories. In the present work, lowplasticity silty clay has been tested only under saturated conditions withthe sole intent of verifying the suitability of the newly developedRS appa-ratus to reproduce typical residual shear strength properties of clayeysoils under these conditions.

2. Servo/suction-controlled RS apparatus: basic features

2.1. General assembly

TheRS apparatus allows for the applicationof vertical loads up to 8 kN,monotonic torque up to 113 N-m, and unlimited angular rotation. It con-sists of three main modules: (1) Main cell with rotational shear system,including pneumatic actuator for application of normal loads and an elec-tromechanical rotary actuator for application of torque loads; (2) Dataacquisition and process control (DA/PC) system, with performance anddata reduction software for real-time calculation of normal and shear

Filter paper5-bar ceramic

Air-pressure conduits

(a)

(c)

Fig. 1. General RS assembly: (a) lower platen, (b)

stresses and average linear and angular displacements; and (3) PCP-15Usuction control panel for implementation of axis-translation technique(Hoyos et al., 2011). In the present work, an orderly step-by-stepsetting-up procedure was established as follows:

1. All actuators and the DA/PC system are switched on to allow theinstruments to come into equilibrium and minimize the influence oftemperature offsets.

2. A small piece ofwet filter paper is placed over the top of each high-air-entry (HAE) ceramic disk, prior to specimen compaction, to ensurephase continuity between the pore-water in the soil and the water inthe saturated disk: Fig. 1(a).

3. The 15 mm (0.59 in. thick specimen is then statically compacted intothe bottom annular platen, having 152.4 mm (6 in.) OD and 96.5mm (3.8 in.) ID: Fig. 1(b). The specimen is quickly transferred to theRS frame and the platen tightly fixed onto the bottom base plate:Fig. 1(c).

4. The vertical load shaft is brought up through a servo controller and theupper annular platen affixed to the top of the piston shaft: Fig. 1(c). Alight vertical seating load of 25 N is applied in order to bring the upperannular platen in full contact with the specimen.

5. All drainage and flushing lines are filled with de-aired water andflushed several times to avoid any trapped air in the whole system.

6. The main RS cell is installed and the top cover plate affixed to the cell:Fig. 1(d). A pore-air pressure line from the PCP-15U panel is connectedto the cover plate via a quick connector.

7. Readings of the load-torque transducer are reset, and the LVDT andangular deformation sensors are re-zeroed, prior to RS testing.

8. The specimen is then subject either to a suction-controlled single-stage or multi-stage RS test using the s = ua testing concept(i.e., uw = 0).

9. When the test is finished, all pressures are gradually reduced back toatmospheric pressure, the main cell and top annular platen gentlyremoved, and the soil failure surface thoroughly examined viamicroscopic digital imaging (Velosa, 2011).

Compacted soil

ua

(b)

(d)

uaua

test specimen, (c) top platen, (d) main cell.

0

5

10

15

20

25

30

35

100001000100101

Gra

vim

etric

wat

er c

onte

nt (

%)

Matric suction (kPa)

SC-SM

SM

CL

Fredlund & Xing (1994)

Fig. 3. Soil–water characteristic curves from SC-SM, SM, and CL soils.

3L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

2.2. Saturation of HAE ceramics

In order to ensure phase continuity between the water compartmentbeneath the HAE ceramics and the pore-water in the soil, all 5-barceramics are to be saturated prior to RS testing. An in-placesaturation procedure was adopted, which can be summarized asfollows. The ceramics are first immersed in de-aired distilled water ina beaker for at least 24 h. A 100 kPa (14.5 psi) vacuum is then appliedfor 48 h to remove the occluded air bubbles in the ceramics: Fig. 2(a).The ceramics are then mounted and epoxy-sealed into a stainless steelring as part of the bottom annular platen assembly: Fig. 1(a). Theannular cavity, reserved for the soil specimen, is filled with de-aireddistilled water to a height of about 25 mm (1 in.): Fig. 2(b). Once themain cell and top cover plate are set into place, the water film is subjectto an air pressure of 200 kPa for at least 48 h. During this period, anyremaining air in the ceramics is expected to dissolve in water. A similarprocedure has been successfully adapted by Hoyos and Macari (2001)and Hoyos et al. (2012) to the working conditions of HAE ceramicsin true triaxial cells. After saturation of the ceramics, the main cell isremoved and the remaining water film is eliminated using a suctionpipette. The ceramics, however, remain covered with water until thesoil is ready to be compacted.

3. Test materials and compaction properties

The testmaterials used in this investigation classify as silty clayey sand(SC-SM), silty sand (SM) and silty clay (CL), according to the USCS. (Thelatterwas tested only under saturated conditions to verify typical residualstrength properties.) SC-SM soil has 60% sand, 34.05% silt, and 5.95% clay;optimummoisture content, OMC=26%, andmaximum Proctor dry den-sity, γd-max = 1.33 g/cm3. Fine-grained fraction yields liquid limit, LL =26.4%, and plasticity index, PI = 6.2%. SM soil has 83.6% sand, 9.8% silt,and 6.6% clay; optimummoisture content, OMC= 10.5%, and maximumProctor dry density, γd-max= 1.84 g/cm3. The soil does not exhibit plasticcharacteristics. CL soil has 18% sand, 50% silt, and 32% clay; optimummoisture content, OMC = 17%, and maximum Proctor dry density, γd-

max = 1.77 g/cm3. Fine-grained fraction yields liquid limit, LL = 37%,and plasticity index, PI = 20%. The specific gravity for SC-SM, SM andCL soils are 2.71, 2.68 and 2.72, respectively. Fig. 3 shows the dryingloop of the corresponding soil–water characteristic curves (SWCCs),assessed via pressure plate testing, as well as the best-fit model curvesas per Fredlund and Xing (1994). The SWCCs were obtained to assistthe authors in defining the appropriate range of suction states (beyondthe air-entry value of each test soil) to be induced on compacted speci-mens of SC-SM and SM soils via the axis-translation technique.

All RS test specimens were prepared directly into the lower annularplaten via in-place static compaction: Fig. 1(b). The upper annularplaten is used to compress one single lift of the loose soil–water mix toa target dry unit weight of 95% of the corresponding γd-max. Amonotonic force is applied by means of a triaxial loading frame at aconstant compaction displacement rate of 1.25 mm/min. Specimens ofSC-SM and SM soils were prepared at water contents corresponding to

Fig. 2. Saturation of HAE ceramics: (a) vac

suction values slightly less than 25, 50, 75, or 100 kPa, according to theirrespective SWCC shown in Fig. 3. Two additional specimens of SC-SMand CL soils were prepared at water content 6% greater than the corre-sponding Proctor optimum, and subsequently soaked in distilled waterin the main RS cell for testing under saturated state (s = 0).

4. Experimental variables and procedures

4.1. Pore-fluid equalization

The first stage of a suction-controlled multi-stage RS test requiresbringing the soil to an initial net normal stress, (σn − ua) = 25 kPa, anda corresponding matric suction state, s = ua = 25, 50, 75, or 100 kPa.To accomplish this, a vertical load is first monotonically applied via theupper annular platen to induce a normal stress 25 kPa greater than thetarget value of suction. The soil is then allowed to consolidate under thisload for at least 2 h. Pore-air pressure ua is then increased via compressedair in the main RS cell, as shown in Fig. 1(d), until reaching the desiredsuction state, s = ua (uw = 0). The vertical load is adjusted accordinglyto keep the net normal stress constant at 25 kPa.

Fig. 4 shows the change of vertical soil displacementwith time duringa typical consolidation–equalization stage of SM soil under net normalstress of 25 kPa and target suction value of 50 kPa. The test starts withan immediate increase in vertical displacement due to the monotonicapplication of vertical load (path AB). Once the vertical displacementreaches a constant value, pore-air pressure ua is increased to the targetlevel. Because the initial suction of all compacted specimens was slightlyless than the target value, thus following a drying path, the increase in uainduces further vertical displacement (shrinkage) in the soil (path BC).All test specimens exhibited this typical behavior. Pore-fluid equalizationwas considered accomplished when no further change both in water

uum beaker, (b) in-place water bath.

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.060 4 8 12 16 20 24

Ver

tical

soi

l dis

plac

emen

t (m

m)

Elapsed time (hr)

A

B

C

Fig. 4. Typical vertical soil displacements induced by normal stress (AB) and suction (BC)increases.

Table 1Experimental variables for suction-controlled multi-stage shearing.

Test soil Test no. s = (ua − uw): kPa (σn − ua): kPa Stages

CL 1 0 25, 50, 100, 200 4SC-SM 1 0 25, 50, 100, 200 4

2 25 25, 50, 75, 100 43 50 25, 50, 75, 100, 200 54 100 25, 50, 75, 100, 200 5

SM 1 50 25 12 100 25, 50, 100 33 25 25, 50, 75 34 25, 50, 75, 100 25 45 25 25, 50, 75, 50, 25 5

100 25 16 100 25 1

140

4 L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

volume expelled from the soil (less than 0.035 ml/day) and vertical soildisplacement was observed.

Typical pore-fluid equalization records for SC-SM soil under netnormal stress of 25 kPa and sustained suction states, s = 25, 50, and100 kPa, are shown in Fig. 5. The patterns confirm that in each casethe initial suction was below target. Times required for consolidation–equalization ranged from 70 to 110 h for SC-SM soil, and from 50 to90 h for SM soil, depending on the target values of suction and netnormal stress. After complete equalization, a shearing stage(s) wasinitiated following a constant-suction multi-stage RS testing scheme.

4.2. Multi-stage shearing

In this work, a total of 11 suction-controlled RS tests were performedon an equal number of statically compacted specimens of CL, SC-SMor SM soil under either saturated (s = 0) or constant-suction states(s = 25, 50, 75, or 100 kPa). The tests were performed by followinga multi-stage scheme in which residual strength assessments weremade at one or more net normal stress values, (σn − ua) = 25, 50,75, 100, or 200 kPa. The series was comprised of 1 test on CL soil, 4tests on SC-SM soil, and 6 tests on SM soil. The correspondingexperimental variables are given in Table 1. The maximum horizontaldisplacement induced by shearing ranged from 30 mm to 100 mm.Shearing was stopped when it was readily apparent that a residualstress had been reached on each stage. Results from previousperformance verification tests reported by Hoyos et al. (2011) showedthat the apparatus is capable of producing repeatable results via axis-translation technique. In this work, all suction-controlled RS tests wereconducted at an equivalent horizontal displacement rate of 0.025 mm/min,which corresponds to a rotational speed of 0.023°/min. This shearingrate is slightly lower than that used in previous works where

0

2

4

6

8

10806040200

Exp

elle

d w

ater

vol

ume

(cc)

Elapsed time (hr)

s = 25 kPa

s = 50 kPa

s = 100 kPa

Fig. 5. Assessment of appropriate pore-fluid equalization times for SC-SM soil.

significantly higher suction values (up to 100 MPa) were induced viarelative humidity based technique (Vaunat et al., 2006; Merchán et al.,2011).

5. General soil response under saturated state

Multi-stage RS tests were first conducted on two saturated speci-mens of CL and SC-SM soils in order to verify their saturated residualstrength properties: Tests CL-1 and SC-SM-1 in Table 1. As previouslymentioned, the specimens were compacted to a target dry unit weightof 95% of γd-max, withmoisture content 6% greater than the correspond-ing optimum. The specimens were then soaked in distilled water in themain RS cell for at least 24 h, attaining a final degree of saturationbetween 97% and 99% (Velosa, 2011). RS test results for SC-SM soil areshown in Fig. 6, which are presented in terms of shear stress (kPa) vs.equivalent horizontal displacement (mm). Expectedly, the residualstrength increases with normal stress. Fig. 7 shows the residual failureenvelopes obtained for all thee soils. The envelope for SM soil isreproduced from Hoyos et al. (2011).

It has been shown that for most clayey soils the residual failureenvelopes are rather nonlinear (Skempton, 1985; Stark and Eid, 1997).However, for soils with low clay fraction (particle size b 2 μ), or soilswith liquid limit less than 60, the relatively spherical particles and/orstiff clay plates dominate their shear behavior (Stark and Vettel,1992). Under drained conditions, these particles are able to establishedge-to-face interaction; consequently, both the initial contact areaand the increase in contact area during shearing are small, for anyrange of normal stresses. This leads to an approximately linear residualfailure envelope. Hence, given that all three soils in this work have arelatively low clay fraction (b32%), the best-fit residual failure

0

20

40

60

80

100

120

0 20 40 60 80 100

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

( n - ua) = 200 kPa

( n - ua) = 100 kPa

( n - ua) = 50 kPa

( n - ua) = 25 kPa

σ

σ

σ

σ

Fig. 6. Results from RS tests on SC-SM soil under saturated (s = 0) condition.

0

20

40

60

80

100

120

140

0 50 100 150 200 250

Res

idua

l she

ar s

tres

s (k

Pa)

Net normal stress (kPa)

SM

CL

SC-SM

SC-SM'r = 22.08o

c'r = 15.56 kPa

CL'r = 28.01o

c'r = 11.97 kPa

SM'r = 30.28o

c'r = 9 kPaφ

φ

φ

Fig. 7. Residual failure envelopes from SC-SM, SM, and CL soils under saturated (s = 0)condition.

0

40

80

120

160

0 20 40 60 80 100

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

( n - ua) = 200 kPa

( n - ua) = 50 kPa

( n - ua) = 25 kPa

( n - ua) = 100 kPa *

* Test stopped due to power outage

( n - ua) = 75 kPa

σ

σ

σ

σ

σ

Fig. 9. Results from RS tests on SC-SM soil under constant suction, s = 50 kPa.

5L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

envelopes are expected to be linear, as shown in Fig. 7, with coefficientsof determination, R2 = 0.994 to 0.999. Also expectedly, SM soil yieldsthe higher effective residual friction angle, ϕ'r = 30.28°.

Residual friction angles for CL and SC-SM soils are 28.01° and 22.08°,respectively. Since CL soil has a higher clay fraction, it was initiallyexpected to yield a lower residual friction angle. Lupini et al. (1981), how-ever, have demonstrated that the residual strength in the slidingmode offailure is strongly dependent on soil mineralogy. The SC-SM soil presentsa large amount of platy mica (Velosa, 2011), producing a more polishedfailure surface and, hence, a lower residual friction angle: Fig. 7. (Featuresof a typical failure plane are discussed in the last section.) Residualstrength parameters in Fig. 7 are also typical of each corresponding testsoil (Skempton, 1985).

6. Response of compacted SC-SM soil under suction-controlledRS testing

Results from 3multi-stage RS tests, conducted on an equal number ofcompacted specimens of SC-SM soil at constant-suction values, s = 25,50, and 100 kPa, are shown in Figs. 8–10, respectively: Tests SC-SM-2, SC-SM-3, and SC-SM-4 in Table 1. For each suction state, residualstrength assessments were made at four or more levels of net normalstress, (σn − ua) = 25, 50, 75, 100, or 200 kPa. It can be readily

0

20

40

60

80

100

0 20 40 60 80 100

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

( n- ua) = 100 kPa

( n - ua) = 75 kPa

( n - ua) = 50 kPa

( n - ua) = 25 kPa

σ

σ

σ

σ

Fig. 8. Results from RS tests on SC-SM soil under constant suction, s = 25 kPa.

observed that in all cases the residual strength increases with netnormal stress. Matric suction, on the other hand, has a significant effecton residual strength, with a considerable increase for s = 100 kPa. Thiscan be readily attributed to a process of clay and/or silt aggregation(aggregation stiffening) during suction-induced drying, which rendersa more frictional material in its response during shearing (Merchánet al., 2011). Residual stress values are reached at horizontal displace-ments close to 60 mm for lower net normal stresses, and beyond to100 mm for higher levels.

Multi-stageRS testing allows for assessment of peak stress only duringthe first shearing stage, i.e., under lowest net normal stress of 25 kPa.Fig. 11 shows the shear stress vs. horizontal displacement response ofSC-SM soil, during first shearing stage, at suction states, s = 0, 25,50, and 100 kPa; including the corresponding change in vertical soil dis-placement during suction-controlled shearing. All specimens exhibit anincrease in shear stress up to a peak value, followed by a gradualdecrease until a readily apparent residual state is attained; the peak isclearlymore pronounced at higher suctions. Although all specimens ini-tially exhibit a slight dilatancy, the general tendency is to a compressivebehavior under sustained shearing; vertical soil displacements, howev-er, are significantly lower for suction, s = 100 kPa, indicating a morefrictional material at this level of suction (most probably caused bysuction-induced stiffening of particle aggregation).

0

40

80

120

160

200

0 15 30 45 60 75 90

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

( n - ua) = 200 kPa

( n - ua) = 100 kPa

( n - ua) = 50 kPa

σ

σ

σ

Fig. 10. Results from RS tests on SC-SM soil under constant suction, s = 100 kPa.

-0.2

0.0

0.2

0.4

0.6

0.8Ver

tical

soi

l dis

plac

emen

t (m

m)

Equivalent horizontal displacement (mm)

s = 100 kPa

s = 50 kPa

s = 0 kPa

s = 25 kPa

0

20

40

60

80

100

120

140

0 20 40 60 80 100

0 20 40 60 80 100

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

s = 100 kPa

s = 50 kPa

s = 0 kPa

s = 25 kPa

Fig. 11. Results from s-controlled RS tests on SC-SM soil under net normal stress,(σn − ua) = 25 kPa.

-0.4

-0.2

0.0

0.2403020100V

ertic

al s

oil d

ispl

acem

ent (

mm

)

Equivalent horizontal displacement (mm)

s = 100 kPa

s = 0 kPa

s = 50 kPa

0

20

40

60

80

100

120

140

403020100

Sh

ear s

tress

(kP

a)

Equivalen horizontal displacement (mm)

s = 100 kPa

s = 50 kPa

s = 0 kPa

s = 50 kPa s = 50 kPa

Hoyos et al. (2011)

Hoyos et al. (2011)

s = 100 kPa

Fig. 12. Results from s-controlled RS tests on SM soil under net normal stress,(σn − ua) = 25 kPa.

( n– ua) : kPa

Shear stress, (kPa)

25

50

75

A

CD

E

Constant suction plane:s = 25 kPa

s = ua (kPa)

F

s = u (kPa)σ

τ

B

Fig. 13.Multi-stage RS test path (AF) at constant suction, s = 25 kPa.

6 L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

The presentwork is one of the first attempts at investigating the effectof matric suction on volume change response (compressible and/or dilat-ant nature) of compacted soils via suction-controlled RS testing. Ideally,residual condition should be defined when no further change both inshear stress and in specimen height are observed. Previous works onintermediate and clayey soils, however, show virtually no change inshear stress from about 50 mm (or even less) to 350 mm of equivalenthorizontal deformation under suction-controlled RS testing (InfanteSedano et al., 2007; Merchán et al., 2011), as has been consistentlyobserved in the present work.

7. Response of compacted SM soil under suction-controlledRS testing

Fig. 12 shows the shear stress vs. horizontal displacement response ofcompacted SM soil, during first shearing stage, under suction states, s =50 and 100 kPa; including the corresponding change in vertical soildisplacement: Tests SM-1 and SM-2 in Table 1. The curve for saturatedcondition (s = 0) is reproduced fromHoyos et al. (2011). Matric suction,again, is observed to have a significant effect on residual strength, with aconsiderable increase for s = 100 kPa. Because of the predominantlycoarse-grained nature of the SM soil, all specimens exhibit a certaindegree of dilatancy toward a critical state, after an initial compression atlower horizontal displacements. The results evidence the enhancementof soil brittleness and dilatancy with increasing suction.

Results from the first shearing stage of the test conducted at suction, s= 100 kPa, are remarkably similar to those reported by Hoyos et al.(2011) from a test conducted under same suction and net normal stressstates, which further demonstrates the reliability of the newlydeveloped RS apparatus. Assessing the extent to which these results arerepeatable was clearly warranted within the scope of the present work,

which focuses primarily on the effects of pre-shearing and suction histo-ries on unsaturated residual strength of SM soil, as documented in thefollowing sections.

7.1. Effect of pre-shearing history

In order to assess the dependency of the residual shear strength ofSM soil over its pre-shearing history, a new statically compacted

25

50

100

75

s = ua (kPa)( n – ua) : kPa

Shear stress, (kPa)

25

50

100

A

Constant net normal stressplane: ( n – ua) = 25 kPa

75

B

C D

E F

GH

σ

σ

τ

Fig. 15.Multi-stage RS test path (AH) at constant net normal stress, (σn − ua) = 25 kPa.

7L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

specimenwas subjected to amulti-stage RS test under net normal stressvalues, (σn − ua) = 25, 50, and 75 kPa, and constant-suction, s =25 kPa, as depicted by path AF in Fig. 13: Test SM-3 in Table 1. Resultsare compared with those reported by Hoyos et al. (2011) from asingle-stage test conducted under net normal stress, (σn − ua) =75 kPa, and same suction, s = 25 kPa. Results from both tests areshown in Fig. 14, including the corresponding change in vertical soildisplacement. It can be observed that when the last-stage shearing(path EF) is applied on a pre-sheared specimen (paths AB and CD),both the peak and the dilatant behaviors are virtually suppressed.However, the residual stress is reasonably close to that induced by asingle-stage shearing under same stress state. This confirms previouslyreported observations that the unsaturated residual shear strength ofcompacted soils depends only on the applied level of net normal stress,and not on the pre-shearing history experienced by the soil (Bishopet al., 1971; Bromhead and Curtis, 1983; Vaunat et al., 2006).

7.2. Effect of suction history

In order to assess the dependency of the residual shear strength of SMsoil over its past suction history, a new statically compacted specimenwas subjected to a multi-stage RS test under constant net normal stress,(σn − ua) = 25 kPa, and suction states, s = 25, 50, 75, and 100 kPa, asdepicted by path AH in Fig. 15: Test SM-4 in Table 1. Results are comparedwith those generated during the first shearing stage of the two tests con-ducted under same net normal stress and suction states, s = 50 and100 kPa: Tests SM-1 and SM-2 in Table 1. Results from all three tests areshown in Fig. 16. It can be noticed that when shearing is applied on apre-sheared specimen, i.e., path CD for s = 50 kPa, or path GH for s =100 kPa, the residual stress response of the soil is virtually the same asthat from a single-stage shearing under same stress state. These results

-0.2

-0.1

0.0

0.1

0.2Ver

tical

soi

l dis

plac

emen

t (m

m)

Equivalent horizontal displacement (mm)

0

20

40

60

80

100

120

140

0 10 20 30 40 50 60

0 10 20 30 40 50 60

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

Single-stage

Multi-stage (EF)

( n - ua) = 75 kPa

Hoyos et al. (2011)

σ

Fig. 14. Results from single- and multi-stage RS tests on SM soil at constant suction,s = 25 kPa.

corroborate that the unsaturated residual shear strength of compactedsoils is virtually independent of their past suction history (Vaunat et al.,2006, 2007).

7.3. Simultaneous effect of pre-shearing and suction histories

Finally, with the aim of studying the simultaneous influence of pre-shearing and suction histories over the residual strength of SM soil, anew statically compacted specimen was subjected to a multi-stage RStest that involved a constant-suction load–unload net normal stresspath, followed by a suction-increase path, as depicted by path AL inFig. 17: Test SM-5 in Table 1. The soil was initially sheared (path AJ)under constant suction, s = 25 kPa, following a load–unload sequenceof net normal stress values, (σn − ua) = 25, 50, 75, 50, and 25 kPa. Thesoil was then dried via axis-translation (path JK) to a suction state, s =100 kPa, and re-sheared (path KL) under constant net normal stress,(σn − ua) = 25 kPa. Results are compared with those from a newsingle-stage test conducted on a separate specimen under net normalstress, (σn − ua) = 25 kPa, and matric suction, s = 100 kPa: Test SM-6in Table 1. Results are shown in Fig. 18, including the correspondingchange in vertical soil displacement.

The residual stress response of SM soil is, once again, virtually thesame. It can also be noted that the drying process (path JK) stiffens thepre-existing soil structure, which is manifested by the peak and dilatantbehaviors during last-stage shearing (path KL). This dilatancy, however,is still less pronounced than during single-stage shearing. It can therebybe concluded that the effect of suction on residual strength of compactedsoil is larger than that of net normal stress, since the peak and dilatantbehaviors apparently are not fully suppressed if the soil is subject to a

0

20

40

60

80

100

120

140

0 5 10 15 20 25 30

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

Single-stage

Multi-stage (CD, GH)

s = 100 kPa

s = 50 kPa

Fig. 16. Results from single- and multi-stage RS tests on SM soil at constant net normalstress, (σn − ua) = 25 kPa.

J

( n – ua) : kPa

Shear stress, (kPa)

25

50

75

Constant suction plane:s = 25 kPa

s = ua (kPa)

25

50

75 Constant suction plane:s = 100 kPa

A B

CD

E F

GH

I J

K L

σ

τ

Fig. 17.Multi-stage RS test path (AL) to assess the effect of pre-shearing/suction histories.

8 L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

drying process after pre-shearing. An increase in dilatancy with suctionwas also reported by Vaunat et al. (2006). Finally, first shearing stagesof tests SM-2 and SM-6 (Table 1), both conducted at s = 100 kPa, yield

-0.2

-0.1

0.0

0.1

0.2403020100

Ver

tical

soi

l dis

plac

emen

t (m

m)

Equivalent horizontal displacement (mm)

0

20

40

60

80

100

120

140

403020100

She

ar s

tres

s (k

Pa)

Equivalent horizontal displacement (mm)

Single-stage

Multi-stage (KL)

s = 100 kPa

Fig. 18. Effect of pre-shearing/suction histories on SM soil response under RS testing.

virtually identical results, as shown in Figs. 16 and18;which further dem-onstrates the reliability of the newly developed RS apparatus.

8. A residual shear strength framework for unsaturated soils

8.1. Compacted SM soil

Fig. 19(a) shows the effect ofmatric suction on both the position andslope of all residual failure envelopes assessed from suction-controlledmulti-stage RS testing on compacted SM soil. It can be readily observedthat the final positioning of the envelopes is greatly influenced bysuction, with a considerably higher position for s = 100 kPa. Suctionalso has a significant effect on the residual apparent cohesion,with low-est value close to 9 kPa for s=0. The envelopes for suction values, s=0,25, and 75 kPa, are essentially those reported by Hoyos et al. (2011);however, the authors have now added the new residual strength dataobtained from the additional series of tests accomplished in the presentwork, averaging all of the previous and current results and re-assessingthe best-fitting envelopes in each case: This is the reason the residualfriction angles are slightly different (and more reliable) than thosereported by Hoyos et al. (2011). A whole new residual failure envelope,for s = 100 kPa, has been generated from the present work: Tests SM-2and SM-6 in Table 1.

For the range of matric suction states considered in this work, theresidual failure envelopes are confirmed to be essentially linear, withcoefficients of determination, R2 = 0.994 to 1.0. However, the residualfriction angle ϕ'r tends to increase with increasing matric suction.Evidence of an increase in residual friction angle with increasing suctionhas also been reported by Salman (1995), via triaxial testing;Feuerharmel et al. (2006), Zhan and Ng (2006) and Hossain and Yin(2010), via direct shear testing; and Vaunat et al. (2007), via ring sheartesting.

Fig. 19(b) shows the effect of net normal stress on both the positionand slope of all residual failure envelopes projected onto the residualshear stress vs. matric suction plane. The envelopes are also confirmedto be virtually linear, with coefficients of determination, R2 = 0.906 to1.0, although essentially parallel, which implies a constant residualbeta angle ϕb

r with respect to matric suction. This parameter is

0

50

100

150

200

200150100500

Res

idua

l she

ar s

tres

s (k

Pa)

Net normal stress (kPa)

s = 100 kPa

s = 75 kPa

s = 50 kPa

s = 25 kPa

s = 0 kPa

'r = 30.28

'r = 47.20

'r = 42.98

'r = 46.1

'r = 35.53

(a)

0

50

100

150

200

0 25 50 75 100 125

Res

idua

l she

ar s

tres

s (k

Pa)

Matric suction (kPa)

Net normal stress = 100 kPa

Net normal stress = 75 kPa

Net normal stress = 50 kPa

Net normal stress = 25 kPa

br = 41.21

br = 41.88

br = 41.71

br = 41.98

(b)

φ ° φ °

φ °

φ °φ °

φ °

φ °

φ °

φ °

Fig. 19. Residual failure envelopes from SM soil: (a) effect of matric suction, (b) effect ofnet normal stress.

0

40

80

120

160

0 25 50 75 100 125

Res

idu

al s

hea

r stre

ss (k

Pa)

Matric suction (kPa)

Net normal stress = 200 kPa

Net normal stress = 100 kPa

Net normal stress = 75 kPa

Net normal stress = 50 kPa

Net normal stress = 25 kPa

(b)

0

40

80

120

160

200150100500

Res

idu

al s

hea

r stre

ss (k

Pa)

Net normal stress (kPa)

s = 100 kPa

s = 50 kPa

s = 25 kPa

s = 0 kPa

'r = 22.09

'r = 26.1

'r = 24.70

'r = 22.28

(a) φ °

φ °

φ °

φ °

Fig. 20.Residual failure envelopes fromSC-SM soil: (a) effect ofmatric suction, (b) effect ofnet normal stress.

R² = 0.8831

R² = 0.9626

R² = 0.9861

R² = 0.9947

-30

-20

-10

0

10

20

30

40

0 25 50 75 100 125

Res

idua

l bet

a an

gle,

b r

(deg

.)

Matric suction (kPa)

Net normal stress = 25 kPa

Net normal stress = 50 kPa

Net normal stress = 75 kPa

Net normal stress = 100 kPa

φ

Fig. 21. Effect of matric suction on residual beta angle (ϕbr) of compacted SC-SM soil.

9L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

thoroughly defined in the original framework postulated by Fredlundet al. (1978) for “peak” shear strength of unsaturated soils. The notationadopted in this work simply reflects the “residual” nature of this sameshear strength parameter. Results shown in Fig. 19 corroborates that aconceptual residual shear strength framework for compacted soils, sim-ilar to that postulated for peak shear strength (Fredlund et al., 1978),could eventually be devised as more experimental evidence of thiskind is made available.

8.2. Compacted SC-SM soil

Fig. 20(a) shows the effect ofmatric suction on both the position andslope of all residual failure envelopes obtained for compacted SC-SMsoil. The patterns are similar to those of SM soil: envelopes essentiallylinear, with coefficients of determination, R2 = 0.954 to 0.999; higherposition for s = 100 kPa; and lowest residual apparent cohesion closeto 17.5 kPa for s = 0. In this case, however, the residual friction angleϕ'r remains reasonably constant, suggesting that the envelopes becomemore parallel as the plasticity of the soil increases. Fig. 20(b) shows theeffect of net normal stress on residual failure envelopes projected ontothe residual shear stress vs. matric suction plane. Results show a highnonlinearity with respect to suction. The data have been fitted bysecond-degree polynomial functionswith coefficients of determination,R2= 0.985 to 1.0. It can also be noted that the nonlinearity of the enve-lopes increases with increasing net normal stress.

To better appreciate the effect of suction on residual shear strength ofcompacted SC-SM soil, the change of residual beta angle ϕb

r with matric

suction has been plotted in Fig. 21 for different net normal stresses. Thedata have also been fitted by second-degree polynomial functions. Theresidual beta angles start at a constant value of 22.09° (ϕb

r = ϕ'r) undersaturated state, s = 0; then increase (ϕb

r N ϕ'r) until a suction stateclose to the air-entry value of the soil, s = 28 kPa, according to theSWCC in Fig. 3; and finally decrease gradually (ϕb

r b ϕ'r) to as low aszero, or even negative values, for higher suction values. This decrease is

10 L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

more dramatic at higher levels of net normal stress, with a considerablereduction for (σn − ua) = 100 kPa.

Vanapalli et al. (1996) and Lu and Likos (2004), among severalothers, have thoroughly documented that there is a direct correspon-dence between the nonlinear nature of the peak shear strength enve-lope, with respect to increasing matric suction, and the SWCC. Withinthe regime of relatively low suction, and prior to the air-entry pressure,the soil pores remain essentially saturated, the shear strength envelopeis reasonably linear, and the beta angle ϕb is effectively equal to thefriction angle ϕ'. As the soil becomes unsaturated, the reduction in thevolume of pore-water within this regime effectively reduces the contri-bution of matric suction toward increasing shear strength. This effect ismore noticeable as the net normal stress increases. This particular fea-ture of the hydro-mechanical behavior of unsaturated soils might pro-vide the most rational basis for understanding the patterns observedin Fig. 21.

9. Typical failure plane: SC-SM soil

One of themost crucial aspects of ring shear testing is ensuring a goodgrip between the soil and the top and bottom annular platens to preventshearing at the platen–soil interfaces. Although the objective is to shearthe specimen close to the platen itself, the failure plane is to be inducedwithin the thickness of the specimen being tested. The short handful ofprevious works accomplished on the present subject, however, includingInfante Sedano et al. (2007) andMerchán et al. (2011), do not dwell suf-ficiently on this crucial aspect of suction-controlled ring shear testing.

Somedistinctive features of a typical failure plane induced by shearingon compacted SC-SM soil are shown in Fig. 22. A microscopic digitalimage of the top surface of the compacted soil is shown in Fig. 22(a) asgeneral reference. (The photo was taken after the compacted specimenwas briefly exposed to air drying.) A similar image of the induced failureplane is shown in Fig. 22(b), which is characterized by a polished andbright surface, clearly indicating a reorientation of plately clayeyminerals(mica) along the plane; hence the development of slickensides alongwhich residual strength is measured (Velosa, 2011). This may explainthe acute drop in shear stress (brittleness), from peak to residual,observed during the first shearing stage of most RS tests; and the more

(c)

(a)

Fig. 22. Typical failure plane in SC-SM soil: (a) reference top surface prior to shearing, (b) failure

ductile response during subsequent shearing stages under larger net nor-mal stresses.

Furthermore, a thorough inspection of the top and bottom annularplatens shows that, in most cases, the induced failure plane was withinthe thickness of the sheared specimen, about 2.0 mm deep on average:Fig. 22(c)–(d). It is worth noting that the rough-surfaced disk adaptedto the top platen is a custom-made, sintered-bronze porous diskmanufactured by Geotest Instrument Corp. Although the actualroughness index of the disk was not provided, its roughness was madeconsiderably higher than that of the top disk used in a conventionalBromhead device. As shown in Fig. 22(d), failure occurs by rupture ofthe specimen along its upper half thickness, with a considerable amountof soil adhered to the roughened upper platen after being displaced rela-tive to the soil below. Typical failure planes observed in SM soil were alsoabout 1.90 mm deep, on average (Hoyos et al., 2011).

Although amore in-depth analysiswas beyond the scope of thiswork,the presence of a failure plane within the thickness of SM soil was also“qualitatively” verifiedvia SEM imaging. Close examinationof SEMmicro-graphs prior to shearing indicates that the sandy particles seem to be gen-erally sub-angular in shape, with a relatively homogeneous size. Aftershearing, the particles are generally angular in shape, exhibiting a widerrange of sizes (Velosa, 2011). This can be attributed to the breaking of par-ticles during shearing and, in some cases,may explain the slight reductionin shear strength, from peak to residual, when the soil is subject to rela-tively large deformations. These observations provided reasonably strongevidence of a failure plane also being induced within the thickness of SMspecimens.

10. Concluding remarks

Results from a comprehensive series of suction-controlled RS tests onstatically compacted specimens of silty clayey sand (SC-SM) and siltysand (SM) soils were presented. The experiments were conducted in anewly developed servo/suction-controlled RS apparatus that is suitablefor testing unsaturated soils under large deformations via axis-translation technique. The results reflect the important role played bymatric suction on residual shear strength properties of unsaturated soils.For the range of net normal stresses and suction states investigated, theincrease in residual shear strength with increasing suction was found to

2.0-mm deepfailure surface

(b)

(d)

surface after shearing, (c) bottomplatenwith sheared soil, (d) top platenwith sheared soil.

11L.R. Hoyos et al. / Engineering Geology 172 (2014) 1–11

be virtually linear for SM soil, but significantly nonlinear for SC-SM soil.Multi-stage test results also appear to confirm that the residual shearstrength of compacted soils is independent of both the pre-shearing andthe suction histories undergone by the soil. Results, in general, suggestthat a conceptual residual shear strength framework for unsaturatedsoils, similar to that postulated for peak shear strength, could eventuallybe devised as more experimental evidence of this kind is made available.Further investigations, including Environmental Scanning ElectronMicroscopy (ESEM) technology, are necessary to correlate and furthersubstantiate the observedphenomenawith changes in soil particle grada-tion, aggregation and plasticity as the soil is subjected to suction-controlled ring shearing.

Acknowledgments

The core system of the RS apparatus used in this experimental effortwas developed under the U.S. National Science Foundation Award #CMS-0626090 (Program Director: Dr. Richard J. Fragaszy). This supportis gratefully acknowledged. Any findings, conclusions, or recommenda-tions expressed in this material are those of the authors and do notnecessarily reflect the views of the National Science Foundation.

References

ASTM Standard D6467-06a, 2006. Standard Test Method for Torsional Ring Shear Test toDetermine Drained Residual Shear Strength of Cohesive Soils. ASTM International,West Conshohocken, PA.

ASTM Standard D7608-10, 2010. Standard Test Method for Torsional Ring Shear Test toDetermine Drained Fully Softened Shear Strength and Nonlinear Strength Envelopeof Cohesive Soils (Using Normally Consolidated Specimen) for Slopes with No Pre-existing Shear Surfaces. ASTM International, West Conshohocken, PA.

Bishop, A.W., Green, G.E., Garga, V.K., Andresen, A., Brown, J.D., 1971. A new ring shearapparatus and its application to the measurement of residual strength. Geotechnique21 (2), 273–328.

Bromhead, E.N., 1979. A simple ring shear apparatus. Ground Eng. 12 (5), 40–44.Bromhead, E.N., Curtis, R.D., 1983. A comparison of alternative methods of measuring the

residual strength of London clay. Ground Eng. 16 (4), 39–41.

Feuerharmel, C., Pereira, A., Gehling, W.Y.Y., Bica, A.V.D., 2006. Determination of the shearstrength parameters of two unsaturated colluvium soils using the direct shear test.Proceedings of the Fourth International Conference of Unsaturated Soils, Carefree,Arizona, vol. 1, pp. 1181–1190.

Fredlund, D.G., Xing, A., 1994. Equations for the soil–water characteristic curve. Can.Geotech. J. 31, 521–532.

Fredlund, D.G., Morgenstern, N.R., Widger, R.A., 1978. The shear strength of unsaturatedsoils. Can. Geotech. J. 15 (3), 313–321.

Hossain, Md.A., Yin, J.H., 2010. Behavior of a compacted completely decomposed granite soilfrom suction controlled direct shear tests. J. Geotech. Geoenviron. Eng. 136 (1), 189–198.

Hoyos, L.R., Macari, E.J., 2001. Development of a stress/suction-controlled true triaxialtesting device for unsaturated soils. Geotech. Test. J. ASTM 24 (1), 5–13.

Hoyos, L.R., Velosa, C.L., Puppala, A.J., 2011. A servo/suction-controlled ring shear apparatusfor unsaturated soils: development, performance, and preliminary results. Geotech.Test. J. ASTM 34 (5), 413–423.

Hoyos, L.R., Pérez-Ruiz, D.D., Puppala, A.J., 2012. Refined true triaxial apparatus for testingunsaturated soils under suction-controlled stress paths. Int. J. Geomech. ASCE 12 (3),281–291.

Infante Sedano, J.A., Vanapalli, S.K., Garga, V.K., 2007. Modified ring shear apparatus forunsaturated soil testing. Geotech. Test. J. ASTM 30 (1), 1–9.

Lu, N., Likos, W.J., 2004. Unsaturated Soil Mechanics. John Wiley & Sons, Hoboken, N.J.Lupini, J.F., Skinner, A.E., Vaughan, P.R., 1981. The drained residual strength of cohesive

soils. Geotechnique 31 (2), 181–213.Merchán, V., Romero, E., Vaunat, J., 2011. An adapted ring shear apparatus for testing partly

saturated soils in the high suction range. Geotech. Test. J. ASTM 34 (5), 433–444.Salman, T.H., 1995. Triaxial behavior of partially saturated granular soils. (Ph.D. disserta-

tion) University of Sheffield, U.K.Skempton, A.W., 1985. Residual strength of clays in landslides, folded strata, and the

laboratory. Geotechnique 35 (1), 3–18.Stark, T.D., Eid, H.T., 1997. Slope stability analyses in stiff fissured clays. J. Geotech.

Geoenviron. Eng. 123 (4), 335–343.Stark, T.D., Vettel, J.J., 1992. Bromhead ring shear test procedure. Geotech. Test. J. 15 (1),

24–32.Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E., Clifton, A.W., 1996. Model for the prediction of

shear strength with respect to soil suction. Can. Geotech. J. 33 (3), 379–392.Vaunat, J., Amador, C., Romero, E., Djeren-Maigre, I., 2006. Residual strength of a low

plasticity clay at high suctions. Proceedings of Fourth International Conference onUnsaturated Soils, Carefree, Arizona, vol. 1, pp. 1279–1289.

Vaunat, J., Merchán, V., Romero, E., Pineda, J., 2007. Residual strength of clays at highsuctions. Proceedings of the Second International Conference on Mechanics ofUnsaturated Soils, Weimar, Germany, vol. 2, pp. 151–162.

Velosa, C.L., 2011. Unsaturated soil behavior under large deformations using a fullyservo/suction-controlled ring shear apparatus. (Ph.D. dissertation) University ofTexas at Arlington, Texas (192 pp.).

Zhan, T.L.T., Ng, C.W.W., 2006. Shear strength characteristics of an unsaturated expansiveclay. Can. Geotech. J. 43 (7), 751–763.