experimental investigation of the effects of suffusion on ... papers...abstract: this paper reports...

Ming Xiao1 and Nathan Shwiyhat2

Experimental Investigation of the Effectsof Suffusion on Physical and GeomechanicCharacteristics of Sandy Soils

REFERENCE: Xiao, Ming and Shwiyhat, Nathan, “Experimental Investigation of the Effects of Suffusion on Physical and GeomechanicCharacteristics of Sandy Soils,” Geotechnical Testing Journal, Vol. 35, No. 6, 2012, pp. 1–11, doi:10.1520/GTJ104594. ISSN 0149-6115.

ABSTRACT: This paper reports the measured effects of suffusion, a type of internal erosion, on the physical and geo-mechanical properties ofsoils including permeability, volume change, compressive strength, and soil grains gradation. A poorly graded river sand with the addition of 10 %kaolinite clay was used to create three gap-graded soils and an unaltered poorly graded “original soil.” Testing was performed using a modifiedtriaxial apparatus that permits seepage through compacted specimens and allows collection of effluent and eroded soils. The specimens, 5.1 cm indiameter and 10.2 cm in length, were monitored for changes in volume and permeability during the suffusion tests. After erosion, the specimenswere compressed using the consolidated-undrained (CU) test. The collected effluent samples were dried to determine the erosion rate and erodedsoil particle gradations with respect to effluent volume. Companion control specimens were tested without erosion. The results revealed that suffu-sion may affect some physical and geo-mechanical properties of soils. Permeability reduction was generally observed in all soils, indicating finegrains migration and clogging within the specimens. The three gap-graded soils each exhibited a greater degree of internal erosion (suffusion), per-meability reduction, and volume change than the original soil. The experimental setup and testing protocol also provided a feasible methodologyfor further research on the effects of suffusion on the physical and geo-mechanical properties of soils. Limitations of this research and futureresearch recommendations on this topic are provided at the end of this paper.

KEYWORDS: suffusion, effects of suffusion, erosion, gap-graded soil, geo-mechanical effect, sand

Introduction

Subsurface erosion has been one of the most prevalent causes ofcatastrophic failures of levees and earthen dams. Such examplesinclude the 1972 failure of the Buffalo Creak dam in WestVirginia (Wahler 1973) and the 1990 collapse of an earthen damin South Carolina (Leonards and Deschamps 1998). Subsurfaceerosion has various forms as shown in Fig. 1. They are: (1) pip-ing—soil grains inside the soil matrix are mobilized and washedout of the matrix by concentrated seepage, resulting in a tubularchannel, or pipe, that progressively forms from downstream toupstream; the pipe can develop into a large tunnel that can causesignificant loss of soil and structural integrity; (2) suffusion—themobilization and transportation of fine grains within a coarser soilmatrix; suffusion may occur in the presence of discontinuity orsegregation of soil grains; and (3) dispersion—a chemicallyinduced erosion. The potential for dispersion depends on thechemistry and mineralogy of the soil, as well as the presence of

dissolved salts in the pore water or eroding water (Sherard andDecker 1977).

Suffusion usually occurs in soil matrices that are sufficientlycoarse to permit the movement of fines through the constrictionsformed by coarse grains. Wan and Fell (2008) defined suffusionas “the process by which finer soil grains are moved through con-strictions between larger soil grains by seepage forces.” The finessusceptible to transport are those entirely contained within thepores of the coarser grains and may not be subjected to effectivestress, as the coarse grains construct the load-supporting skeletonof the soil. The mobilized grains must be sufficiently small topass through constrictions formed by coarser grains that form theskeleton of the soil. Mobilized grains that do not pass throughconstrictions can block the constrictions. Suffusion, though lesscatastrophic in terms of potential failure mechanisms, can bechronically destructive. Suffusion may commonly result in theclogging of soil filters or drainage layers and leads to excess porewater pressure. Suffusion may also result in increased porosity,permeability, seepage, and accentuated consolidation of a soillayer or earthen structure. Experimental research by Wan and Fell(2004) found that “…50 % of the finer fraction as defined by thepoint of inflection of broadly graded soils and the fine limit of thegap in gap-graded soils can be eroded” by suffusion in internallyunstable soils. Previous research has been focused more onthe erosion behaviors of various soils under various hydraulicconditions, and less on the lasting physical and mechanical effects

Manuscript received December 1, 2011; accepted for publication April 6,2012; published online September 2012.

1Associate Professor, Ph.D., Dept. of Civil and Geomatics Engineering,M/S EE94, California State Univ., Fresno, CA 93740, e-mail:[email protected]

2Former Graduate Student, Dept. of Civil and Geomatics Engineering,California State Univ., Fresno, CA 93740, e-mail: [email protected]

Copyright VC 2012 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959. 1

Geotechnical Testing Journal, Vol. 35, No. 6, 2012Available online at www.astm.org

doi:10.1520/GTJ104594

Copyright by ASTM Int'l (all rights reserved); Sun Aug 9 14:52:22 EDT 2015Downloaded/printed byPenn State University (Penn State University) pursuant to License Agreement. No further reproductions authorized.

of internal erosion, particularly the effect of suffusion, on the settle-ment potential, permeability change, and compressive strength of asoil stratum because of the loss of fine grains within the soil matrix.

Previous work on suffusion studied the performance ofgranular filters, effects of pore constriction sizes on suffusion,and methods of predicting soils internal stability. Suffusion is acomplex process that is dictated by the size distributions of thegrains and pore constrictions as well as the internal stress condi-tions. Many researchers recognized the importance of the size ofpore channels to the resistance to suffusion. Kenney and Lau(1985) described soil pores as a network of interconnected porechannels and that the minimum constriction size along a flowpath is the controlling constriction size. Another important factordictating suffusion is the presence of fine grains located withinthe voids constructed by larger grains that form the skeleton ofthe soil. Bendahmane et al. (2008) had found that the ability ofthese grains to be removed from the coarse structure is highlydependent on the state of confining stresses, hydraulic gradient,and initial porosity. They performed 30 tests using clay contentsvarying from 10 % to 30 %, confining pressures ranging from100 to 250 kPa (14.5 to 36.3 psi), and hydraulic gradients rang-ing from 5 to 140. Their tests were carried out using a triaxialapparatus similar to that used by Sanchez et al. (1983). Theyfound that for a given clay content, an increase in the confiningstresses generally yielded a decrease in the maximum suffusionrate. However, they also found that an increase in the confiningstresses yielded an increase in the erosion of coarser sand grains.This confirms the previous work by Papamichos et al. (2001),who found that an increase in the axial load applied to a sandysoil yielded an increase in erosion rate. Few researchers havestudied the post-suffusion characteristics of soils, especially thetriaxial compressive strength and volume change. Arnisimov and

Ter-Martirosyan (2009) performed a series of tests to study theeffect of suffusion on the compressive Young’s modulus. Theyfound that when the hydraulic gradient was increased from 0.8to 4, the suffusion reduced the compressive Young’s modulusby ten times and additional suffusion-induced settlementoccurred.

To model the in situ stress conditions, triaxial tests can beused to study suffusion and piping. Sanchez et al. (1983) werethe first researchers to evaluate the erosion potential of embank-ment core materials using triaxial erosion tests. The recentexperiments by Bendahmane et al. (2008) revealed the complexeffects of confining pressure on suffusion. The true triaxial testsby Richards and Reddy (2008) preliminarily indicated the con-fining stress and pore pressure are critical to piping initiation.The research presented in this paper employed a similar experi-mental design of triaxial erosion tests and focused on the effectof suffusion on the physical and mechanical behaviors of asandy soil.

The hypotheses of this research are that in the presence ofsuffusion porosity increases, settlement occurs, and triaxialcompressive strength decreases. The main purposes of thisresearch are to (1) study, under laboratory-controlled condi-tions, the mechanisms of suffusion in internally unstable soils,such as gap-graded soils, and (2) reveal the quantitative effectsof suffusion on the soils physical and mechanical properties,such as permeability, volume, and strength variation. Thisresearch focuses on the laboratory study of the initiation ofsuch processes and utilized some exaggerated hydraulic condi-tions to speed the erosion development. Further understandingof the suffusion initiation and the physical changes within thesoil can allow for more accurate estimations of the potentiallong-term effects.

FIG. 1—Subsurface erosion in earthen embankment.

2 GEOTECHNICAL TESTING JOURNAL


Materials and Methodology

Experimental Setup

The suffusion tests were conducted using a triaxial apparatus,which was modified to allow effluent water and eroded soil grainsto exit the cell and be captured in an effluent tank or container.The effluent tank can be pressurized to simulate any reasonabledownstream pressure or be open to atmosphere. In this research,the downstream pressure was maintained at atmospheric pressure.During the suffusion process, a volume change unit (VCU) wasinstalled to monitor the total movement of the confinement waterinto and out of the triaxial cell, thereby monitoring the total vol-ume change of the specimen. Figure 2 provides a photo of the testsetup and Fig. 3 shows the modified triaxial base pedestal. Thetop portion of the pedestal is an inverted cone and it funnels into alarge port that connects to a control valve. The diameter of thepedestal is 5.1 cm (2 in.). A U.S. #4 screen (4.75 mm) was setinto a depressed platform on top of the pedestal to support theremolded specimen while permitting relatively unrestricted seep-age. The screen was intended to support the specimen during theundrained compression test that followed the internal erosion testto examine the strength variation because of suffusion. The triax-ial cell is equipped with pressure transducers that monitor theback-pressure at the top of the sample, the pore-pressure at thebottom of the sample, and the cell confining pressure. All trans-ducers communicate with an automatic data logger that connectsto a laptop computer.

Materials and Soils Characterization

Wan and Fell (2008) stated that broadly graded soils and gap-graded soils are susceptible to suffusion. The authors performed aseries of exploratory testing of soils with various gradations andfound that poorly graded and gap-graded soils, in fact, yield moresuffusion than well-graded soils, which experienced only minoramounts of suffusion. Therefore poorly graded and gap-graded soilswere studied to observe the effect of suffusion on their physical and

mechanical characteristics. Gap-graded soils are usually avoided inconstruction. However, improper handling methods during con-struction can cause segregation of materials that may lead to local-ized gap-graded soils. A poorly graded sandy soil was obtainedfrom an aggregate mining company that operates on the San Joa-quin River in Fresno, CA. Kaolinite, a non-expansive clay mineral,was combined with the poorly graded river sand at a ratio of 10 %by dry weight. In addition to the poorly graded sand, three gap-graded soils were tested. The gap-graded soils were manufacturedby re-grading the same poorly graded river sand by removingselected grain sizes corresponding with certain sieves, and thencombining with 10 % kaolinite clay. By using the same source ma-terial to manufacture the three different gap-graded soils, the effectof aggregate angularity on suffusion was reduced. The first gap-graded soil was created by removing the grains passing the #16 andretained on the #50 sieve (1.18 mm to 0.3 mm); the second byremoving the grains passing the #30 and retained on the #50 sieve(0.6 mm to 0.3 mm); the third soil by removing the grains passingthe #30 and retained on the #100 sieve (0.6 mm to 0.15 mm). Thethree types of gap-graded soils are denoted as #16�#50, #30�#50,and #30�#100. The original soil that was used to produce the threegap-graded soils is denoted as “original soil.” The proposed and theactual soil gradations are presented in Figs. 4(a) and 4(b), respec-tively. The variance between the proposed gradations and the actualgradations of the manufactured soils is attributed to the difficulty inachieving complete removal of the selected grain sizes duringthe sieve-shaking process. The characteristics of the soils andthe remolded specimens used in the erosion tests are summarized inTable 1.

To better characterize the physical characteristics of the fourtypes of soils and develop understanding of their different

FIG. 2—Suffusion experimental setup.

FIG. 3—Modification details of triaxial base pedestal: (a) photo, and (b)design schematic.

XIAO AND SHWIYHAT ON EXPERIMENTAL INVESTIGATION 3


behaviors in the presence of erosion, the pore size distributionswere experimentally derived. Pores in soils are interconnected andcontinuous; they are extremely intricate and topologically variabledepending on the soil particle shape, size distribution, and com-paction. The pore geometry can be idealized as a pore body withconstrictions that are referred to as pore throats. The pore-size dis-tributions are represented using the pore throat distributions(PTDs) in this study, because it is the pore throats that govern thesuffusion. The mercury intrusion porosimetry (MIP) test was usedto determine the PTDs. This test involves compacting a specimeninto a small sample container of 2.5 cm� 2.5 cm� 2.5 cm. Amercury source is connected to the sample container and is incre-mentally injected into the voids within the specimen. Injection

pressures up to 11.03 MPa (1600 psi) are utilized, therebydefining the pore structure of the specimens down to a pore throatdiameter of approximately 0.1 lm. Table 2 provides the character-istics of the samples used in the MIP. Figure 5 provides the PTDsof the soils in terms of pore throat diameters versus volume per-centages of the pores associated with each pore throat diameter. Itis assumed that complete saturation of the voids was obtained dur-ing the test. Figure 5 shows that there are generally two primarygroups of pore sizes for each soil. The larger-sized group isbetween approximately 0.05 mm and 0.4 mm, which is believedto be constructed by the coarse soil skeleton, and the smaller-sizedgroup is between approximately 0.00015 mm and 0.005 mm,because of the fine grains located within the larger pores. Betweenthe two ranges of pore throat sizes, the pore throats are almostnonexistent.

Test Procedure

Suffusion tests were performed on specimens compacted to a tar-get relative compaction of 85 % of the maximum dry densitydetermined by the modified proctor test (ASTM D-1557) and at amoisture content slightly wet of the optimum moisture content.Remolded erosion specimens were 5.1 cm (2.0 in.) in diameterand 10.2 cm (4.0 in.) in height. The control specimens were notsubjected to suffusion and were only tested for the physical andgeo-mechanical characteristics. The soils were compacted in asplit compaction-mold in three lifts of equal height, and the com-pacted surface of each layer was scarified prior to placement ofsubsequent lifts. Remolded specimens were set up in the modifiedtriaxial equipment.

FIG. 4—Gradations of poorly graded and gap-graded soils: (a) proposed gra-dations, and (b) actual gradations.

TABLE 1—Erosion specimen details.

Original Soil #16�#50 #30�#50 #30�#100

Maximum dry density, g/cm3 2.08 2.19 2.10 2.10

Optimum moisture content 7.8 % 6.9 % 8.8 % 6.7 %

Relative compaction, % 84.8 85.1 85.2 85.1

Dry density, g/cm3 1.76 1.86 1.79 1.79

Porosity 0.34 0.31 0.33 0.33

Void ratio 0.52 0.44 0.50 0.50

Total pore volume, cm3 71 64 69 69

TABLE 2—Summary of MIP test specimen characteristics.

Original Soil #16�#50 #30�#50 #30�#100

Dry density, g/cm3 1.90 1.88 1.95 1.95

Relative compaction,a % 91 86 93 93

Porosity 0.29 0.30 0.27 0.27

Void ratio 0.41 0.43 0.37 0.37

aThe compaction ratio of the specimens used in the MIP (86 to 93 %) is differ-ent from that used in the erosion tests (85 %).

FIG. 5—Pore throat distributions.



Specimens were confined at an effective stress of 13.8 kPa(2.0 psi) during the saturation process. To speed the saturationprocess the specimens were first evacuated of air by pumping car-bon dioxide (CO2) gas through the specimen. The CO2 wasinjected at the base of the specimen and allowed to permeatethrough the specimen to replace air for approximately 15 mins at apressure of approximately 13.8 kPa (2.0 psi). The CO2 dissolvesmuch more readily into water than air. After the CO2 injection pe-riod, de-aired water was applied at the base of the sample throughthe pore-pressure application port with a slight upward hydraulicgradient of 0.5 to 1.0. The use of de-aired water allows some ofthe gas within the specimen to dissolve into the water to speed thesaturation process. After saturation was obtained (B value >0.95),the specimens were consolidated at an effective stress of 34.5 kPa(5.0 psi). Because of the pressure fluctuation, the actual effectivestresses for all the erosion samples were in a narrow range of 27.1to 34.4 kPa (3.93 to 4.99 psi). Samples were allowed to consoli-date for a minimum of 24 h. Then the samples were subjected toseepage and erosion. The seepage was induced by applying adownward hydraulic gradient of 20.8, which was used in all theerosion tests. This hydraulic gradient was induced by applying a20.7 kPa (3.0 psi) water pressure to the top of the specimen, andleaving the bottom of the specimen open to atmosphere (0 kPa).Our exploratory experiments, which were not reported in this pa-per, showed that at a certain critical gradient, an internal channel(piping) was formed in the soil specimen; the piping caused thesoil to collapse, and the experiment had to be terminated. Thisphenomenon was repeatedly observed in the exploratory experi-ments. Because this paper focuses on suffusion, critical gradientfor piping was not used. But the hydraulic gradient of 20.8, underan effective confinement stress of 34.5 kPa (5 psi), was less thanthe critical gradient for piping, meanwhile, it is high enough toinduce suffusion, so the hydraulic gradient of 20.8 was usedthroughout the erosion tests.

In this test, the bottom of the pedestal was open to the atmos-phere, so that collection of eroded soils at different seepage vol-umes can be accurately accomplished. The permeating fluidflowed downward through the sample and the inverted cone of themodified base pedestal, and then out of the nozzle at the base ofthe cell. The effluent was collected over predetermined effluentvolume intervals with shorter intervals at first and longer intervalsas the test progressed. The effluent was allowed to drip from thenozzle into graduated beakers. At the end of each volume interval,the elapsed time was recorded, and the specimen volume was cal-culated from the VCU reading, and the permeability of the speci-men was measured. The tests were continued until the effluentvisually cleared and it typically occurred within 4 h from the startof the test. The beakers with effluent were transferred into a dryingoven and then dried to a constant mass. The beakers were weighedto determine the mass of eroded solids. The eroded solids col-lected in the beakers were combined into larger samples and thenwet sieved to determine the approximate gradations of the erodedmaterial with respect to seepage volume. The wet sieving processused the following decremented sieve opening sizes: 3.363 mm,1.679 mm, 0.841 mm, 0.419 mm, 0.211 mm, 0.104 mm, 0.038mm, and 0.020 mm. At the end of the erosion, an undrained com-pression test was conducted. The thickness of the membrane that

was used to encase the soil specimen is 0.30 mm, its Young’smodulus is 1868.5 kN/m2. Both parameters were input into the tri-axial test data acquisition software, so the effect of membranestiffness was automatically accounted for during the compression.

Test Program

The following behaviors and physical and geo-mechanical charac-teristics were measured in each of the four test specimens duringthe application of seepage and the seepage-induced suffusion:

• cumulative erosion versus seepage,• erosion rate versus seepage,• permeability versus seepage,• specimen total volume change versus seepage,• eroded particle gradations versus seepage,• post-erosion undrained compression test,• triaxial consolidated-undrained compression test on control

samples, and

A duplicate test on soil #16�#50 was conducted to check therepeatability of this testing. The same measurements were con-ducted on the duplicate specimen.

Results and Discussion

Suffusion Characteristics

The suffusion test results were based on the dry mass of erodedsolids captured in the effluent collection containers. Figure 6 pro-vides the eroded soil mass concentrations in the effluent (mg/ml)versus the seepage volume. The seepage volume was normalizedusing the number of pore volumes of the sample to account forvariations in sample volume and porosity, so that comparisonbetween samples of different sizes can be made. The results showthat the rate of erosion typically decreased as the test progressed.Each of the gap-graded soils had a larger erosion rate than thepoorly graded original soil within the duration of the tests. Aduplicate test was performed for the #16�#50 gap-graded soil.

FIG. 6—Suffusion rate versus seepage volume.



The duplicate test result showed a rather large variance from thefirst test, even though the duplicate specimen was prepared at thesame soil characteristics as the first specimen. The duplicate testinitially followed a similar trend to that of the first test but then alarge mass of coarse soil was eroded from the soil. The sandgrains that were eroded, although fewer in number, weighed muchmore than the fine silt and clay grains that were eroded because ofsuffusion and therefore skewed the results of the repeat test. Theerosion of larger sand grains eventually stopped and the suffusionagain was the main erosive mechanism. Afterward, the resultsbegan to follow the same downward asymptotic trend as in thefirst test. The coarse sand erosion in this soil type was likelybecause of the removal of the coarser soil grains (#16) that madethe soil skeleton more susceptible to erosion.

The effluent samples that were collected during the erosion testswere combined into larger-volume samples and were tested for par-ticle size distributions using wet sieving. Figures 7 through 11show the eroded soil particle gradations with seepage of each ofthe five erosion tests. The progressions of particle gradations withseepage do not show a consistent trend. For the original soilshown in Fig. 7, finer grains were generally eroded first, with anobservable progression toward coarser particle erosion. For thegap-graded soil #16�#50 (Figs. 8 and 9), coarse grains were

eroded first, and the eroded soil gradations generally became fineras the test progressed. For the soil types #30�#50 (Fig. 10) and#30�#100 (Fig. 11), no clear trend is shown. The eroded particlegradations may depend on the grain size distribution and porethroat distribution of the soil. But the authors cannot offer anexplanation of the different gradation behaviors of the erodedgrains of the four soils using the measured grain size distributionsin Fig. 4 and the pore throat distributions in Fig. 5. Another obser-vation from the gradations of the eroded grains is that, for the gap-graded soils, the majority (90 %) of the grains that were erodedand exited the specimen were generally finer than the fine limit ofeach gap.

After the completion of the duplicate erosion test on the#16�#50 specimen, the sample was dissected into three verticallifts, each approximately equal to one-third of the final specimenheight. Each lift was tested for fines content by performing aminus #200 wash and oven drying in accordance with the ASTMD-1140. The results of the tests are provided in Table 3. Consider-ing that the initial fines content of the specimen prior to erosionwas determined to be 16.3 %, the results indicate that some of thevery fine silt and clay-sized grains were able to be mobilized alongthe entire length of the specimen and possibly exit. The resultsalso suggest that the least amount of suffusion occurred within the

FIG. 7—Eroded particle size distributions with effluent for poorly graded orig-inal soil.

FIG. 8—Eroded particle size distributions with effluent for gap-graded soil#16�#50.

FIG. 9—Eroded particle size distributions with effluent for gap-graded soil#16�#50 (duplicate test).




top portion of the specimen and that some fine grains that weretransported from the upstream portion were deposited in thedownstream portion of the specimen, which was deduced by themiddle section having fines content that was less than the bottomportion. The clogging at the downstream section was further evi-denced by the permeability test results presented in the nextsection.

Permeability Test Results

Permeability test results were computed based on the volumeof the collected seepage, the time interval for each effluentcollection, the average hydraulic gradient calculated using thedifference between the specimen’s upstream hydraulic pressureand the atmospheric pressure at the base of the specimen, andthe initial specimen length prior to saturation and consolidation.Figure 12(a) provides the permeability measured in the specimensversus the seepage volume and demonstrates the reduction ofpermeability as erosion and subsequent clogging progressed.Figure 12(b) provides the permeability variation versus total timeelapsed. The #16�#50 soil, with the coarsest grains (and conse-quently largest pores) removed, was found to have the smallestinitial permeability. The #30�#100 soil, with the finest grainsremoved (leaving relatively larger pores in the specimen), had thelargest initial permeability. The permeability fluctuation for the#16�#50 rerun specimen may be because of the washing out ofrelatively larger grains, resulting in a local void ratio increase. Theother specimens generally show a decreasing trend in permeabil-ity. At the beginning of the erosion tests, the permeability wasgenerally constant or decreased slightly. As the tests progressed,the permeability decreased faster for the #16�#50 specimens and

the #30�#100 specimen, likely because of particle deposition andclogging at the downstream section. As the sample clogged, theaverage pore constriction size decreased; progressively smallergrains became entrapped, accelerating the clogging process, untilsuffusion and particle migration stopped.

Volume Change

The change in sample volume was monitored during the erosiontesting. A volume change unit (VCU) that was connected to thetriaxial cell recorded the changes in the flow of water into and outof the cell with a precision of 0.0016 ml (0.0001 in.3). The resultsobtained from the VCU were converted into percentages of theinitial sample volume, and the change in sample volume versusseepage and time are reported in Fig. 13. The results show that thegap-graded soils each produced a larger settlement than the poorlygraded original soil. The repeat test of the #16�#50 soil is notshown in the figure because of the initial erosion of the sand thatcaused a much larger volume change (approximately 3.6 %) as aresult of larger grains falling out of the specimen.

Triaxial Compressive Strength Test Results

Undrained triaxial compression of the samples was performed onthe erosion specimens immediately following the erosion testingand also on a separate control specimen for each of the soil types.

TABLE 3—Results of minus #200 wash test results on #16�#50 rerun (initialfines content of specimen before erosion¼ 16.3 %).

Vertical Portion Fines Content (% passing #200 sieve)

Top third 14.7

Middle third 13.0

Bottom third 13.3

FIG. 12—Permeability variation because of suffusion: (a) permeability varia-tion versus seepage volume, and (b) permeability variation versus elapsedtime.




The results of the triaxial testing are shown in Figs. 14 through 17for each of the four soil types. The corrected deviator stress (withrespect to the actual sample cross-sectional area) and the measuredpore pressure with respect to the axial strain for both the controland the erosion specimens are provided in the figures. It is notedthat each figure shows two types of curves: deviator stress andpore water pressure, both have the same unit; so the y axis hastwo labels. The legend of each figure identifies each curve.Tables 4 through 7, which correspond to Figs. 14 through 17,respectively, provide the shear and failure conditions applicable toeach of the test specimens. The triaxial test results do not offer,

with certainty, an explanation of the strength performance of thepost-suffusion soil specimens: the three gap-graded specimens af-ter erosion consistently had higher compressive strength than thecontrol specimens without erosion. The authors think it wasbecause of the loss of saturation during the erosion process whilethe bottom of the erosion specimens were subjected to atmos-pheric pressure. This loss of saturation consequently resulted inreduced pore pressure and higher compressive strength. The con-trol specimens were not subjected to erosion and thereby main-tained saturation until the time of compressive strength testing.On the repeat specimen of soil #16�#50, the B-value was checkedimmediately after erosion and was found to be at 0.86, demon-strating that loss of saturation had occurred during the erosion pro-cess. The specimen was then re-saturated to bring the B-valueback above 0.95 and then compressed. It was still found that thepore pressure during the compression was lower than the controlspecimen during compression. The authors speculate that the re-saturation process may have changed the grain and pore structurearrangements. However, for the poorly graded original soil, thecompressive strength of the control specimen is higher than thespecimen that experienced erosion. This is mostly likely becauseof the different effective confining pressures: 41.0 kPa (or 6.0 psi)for the control specimen, and 27.1 kPa (or 3.9 psi) for the erosion

FIG. 13—Volume reduction because of suffusion: (a) erosion-induced volumereduction with seepage, and (b) erosion-induced volume reduction withelapsed time.

FIG. 16—Corrected deviator stress and pore pressure versus axial strain for#30�#50 gap-graded soil, consolidated-undrained (CU) test, initial effectiveconfining stress¼ 34.6 kPa (control sample) and 33.1 kPa (erosion sample).

FIG. 14—Corrected deviator stress and pore pressure versus axial strain forthe original soil, consolidated-undrained (CU) test, initial effective confiningstress¼ 41.0 kPa (control sample) and 27.1 kPa (erosion sample).

FIG. 15—Corrected deviator stress and pore pressure versus axial strain for#16�#50 gap-graded soil, consolidated-undrained (CU) test, initial effectiveconfining stress¼ 36.6 kPa (control sample) and 36.5 kPa (erosion sample),and 31.2 kPa (erosion rerun sample).



specimen. Higher effective confining stress results in higher com-pressive strength. Therefore, the different confining stresses usedin the original soil in Fig. 14 made the compressive strengths ofthe two specimens incomparable.

Further Discussion of the Test Results

As various test results in the literature seem to indicate, and alsoin agreement with Reddi et al. (2000), the particle redeposition orinternal clogging of specimens by transported grains is a signifi-cant factor in the suffusion of soils. The erosion rates, permeabil-ity variations, and the volume change characteristics are allindicative of clogging phenomenon occurring within the samplesas erosion progressed. As other researchers have suggested (Wanand Fell 2004; Rönnqvist 2010), the gap-graded soils, when com-pared with the poorly graded sample, were more susceptible tosuffusion as demonstrated by the larger erosion rates and volumechange that occurred within these soils. As revealed in this paperand also by other researchers (Arnisimov and Ter-Martirosyan2009; Petrukin and Presnov 1991), volume change during suffu-sion can occur. Research presented herein shows soils exhibitingup to 0.75 % volume reduction. The amount of volume changecorrelates well with the amount of eroded mass, i.e., more erosionmass results in more volume change.

The amounts of volume change, erosion, and permeabilitychange may seem minor. The testing in this research was only per-formed over a short duration; but in field conditions, the flow andassociated changes in soil structure can continue for many yearsand can result in more severe changes to the soil structure. Addi-tionally, the boundary conditions in this laboratory testing may beprohibitive of the full extent to which the suffusion can occur. Inthe case of downward flow along a cutoff wall that penetrates asand layer, continued seepage along this path can result in suffu-sion and volume change within the sand layer along the wall.Localized volume changes along the wall at various depths canresult in vertical displacement, which can cause differential settle-ment and potential failure along an earthen structure. This furtherindicates the importance of this research.

Limitations of This Research

There is an inability in this testing, and testing by others in labora-tory bench scale, to adequately address the effect of the boundaryconditions on the onset and progression of suffusion. In field con-ditions, seepage follows the path of least resistance and the seep-age path may be tortuous in various directions, thereby exposingeroded grains to more exit paths in the pore structure. In this test-ing, the seepage was restricted solely to the downward direction

TABLE 5—#16�#50 Gap-graded soil triaxial shear tests.

ParametersControl

SpecimenErosion

Specimen

RerunErosion

Specimen

Initial shear conditions

Cell pressure, kPa (psi) 53.7 (7.8) 54.7 (7.9) 54.7 (7.9)

Initial pore pressure, kPa (psi) 17.1 (2.5) 18.1 (2.6) 23.5 (3.4)

Initial effective stress, kPa (psi) 36.6 (5.3) 36.5 (5.3) 31.2 (4.5)

Conditions at failure

Axial strain, % 0.6 2.7 1.2

Deviator stress, kPa (psi) 17.2 (2.5) 49.2 (7.1) 32.6 (4.7)

Pore pressure, kPa (psi) 37.4 (5.4) 36.3 (5.3) 34.5 (5.0)

Major effective principal stress, kPa (psi) 33.5 (4.9) 67.6 (9.8) 52.8 (7.7)

Minor effective principal stress, kPa (psi) 16.3 (2.4) 18.3 (2.7) 20.2 (2.9)FIG. 17—Corrected deviator stress and pore pressure versus axial strain for#30�#100 gap-graded soil, consolidated-undrained (CU) test, initial effectiveconfining stress¼ 33.9 kPa (control sample) and 34.4 kPa (erosion sample).

TABLE 4—Original soil triaxial shear tests.

ParametersControl

SpecimenErosion

Specimen


Cell pressure, kPa (psi) 82.7 (12.0) 56.8 (8.2)

Initial pore pressure, kPa (psi) 41.7 (6.1) 29.7 (4.3)

Initial effective stress, kPa (psi) 41.0 (6.0) 27.1 (3.9)


Axial strain, % 11.4 4.3

Deviator stress, kPa (psi) 124.7 (18.1) 82.0 (11.9)

Pore pressure, kPa (psi) 34.2 (5.0) 28.6 (4.2)

Major effective principal stress, kPa (psi) 173.2 (25.1) 110.2 (16.0)

Minor effective principal stress, kPa (psi) 48.5 (7.0) 28.2 (4.1)


ParametersControl

SpecimenErosion

Specimen













and only within the area of 5.1 cm in diameter. The boundary con-dition in this test can be prohibitive of realistic seepage flow andsuffusion. Better simulation of the boundary condition to allowmore realistic flow condition may induce a larger degree of suffu-sion. Next, the application of the hydraulic gradient across thespecimen creates an increased effective stress toward the base ofthe specimen and perhaps localized consolidation. This can causeminor volume change and permeability reduction, which were notaccounted for in the presented test results. It is also noted that thisresearch, as well as research by others, does not consider effectsof scaling for geometric or dynamic similitude. Considering theseitems, the actual field behaviors may be different from those foundin this research.

Conclusions and Recommendations

Conclusions

The experimental findings presented herein provide evidence thatsuffusion may affect the physical and geo-mechanical propertiesof soils. Though the effects are seemingly minor in the benchscale testing, the length of time over which the seepage and suffu-sion occur can be much greater in the field. The following conclu-sions are drawn from this research.

(1) In the presence of suffusion, changes in permeability, com-pressive strength characteristics, and volume (i.e., settle-ment) can occur. Permeability decreases as suffusionprogresses and the permeability reduction ranges fromnearly one to two orders of magnitude in the gap-gradedsoils, to almost no change in the poorly graded originalsoil. The degree of permeability reduction is highly depend-ent on the internal clogging that occurs. The changes incompressive strength cannot be accurately explained in thispaper: the gap-graded soils show that compressive strengthwas greater after erosion, whereas the original soil showsthat the compressive strength was less after erosion. Loss ofsaturation during erosion is presumed to account for the dif-ferences in strength. The changes in volume show approxi-mately one-third to three-quarter % change in volume forthe soil types tested.

(2) The progressions of particle gradations with seepage do notshow a consistent trend. The eroded particle gradations maydepend on the grain size distribution and pore throat distribu-tion of the soil, but the different gradation behaviors of theeroded grains of the four soils cannot be explained.

(3) Gap-graded soils tend to produce more pronounced physicaland geo-mechanical changes than poorly graded soils, as sug-gested by the greater erosion rates, permeability reduction,and volume reduction.

(4) Eroded grains beyond the fine limit of the gap in gap-gradedsoils are susceptible to being transported by suffusion.

Recommendations for Continued Research

To futher the understanding of the physical and geomechaniccharacteristics of erodible soils by suffusion, the followingresearch recommendations are provided to future researchers tosupplement the findings of this research and address some of theaforementioned limitations.

(1) Perform tests on a larger variety of poorly graded and gap-graded soils, both stable and unstable in nature, as well as avariety of clay or fines contents. Physically manufacturing thesoils from the same original soil is recommended to neglectthe effects of soil aggregate angularity.

(2) Modify test procedure to lengthen the duration of the erosionand define a threshold for ending a test. This may involvedesigning an automatic influent and effluent collectionsystem.

(3) Modify test procedure to first identify the critical hydraulicgradient for each soil type by trial tests. After identifying thecritical hydraulic gradient, the soils should be tested with sim-ilar ratios of applied gradient to the critical hydraulic gradient.By holding this ratio constant for different soil types, theeffects of the pore structure of the specimens on the totaldegree of suffusion may become more evident.

(4) Modify test procedure and enhance test setup to maintain satu-ration during the erosion process so that the undrained com-pression tests can be correctly performed.

(5) For gap-graded soils, use a base screen that better fit the sizeof the fine limit of the gap to better restrict the coarse skeletonfrom being eroded.

(6) Investigate the effects of the boundary conditions on the testor modify test setup to reduce the effect.

References

Arnisimov, V. V. and Ter-Martirosyan, Z. G., 2009, “Effect ofMechanical Suffusion on Additional Settlements of FoundationBeds,” Soil Mech. Found. Eng., Vol. 46, No. 4, pp. 129–135;Osnovaniya, Fundamenty i mekhanika gruntov, No. 4, pp.7–11 (in Russian).

ASTM D-1140, 2006, “Standard Test Method for Amount of Ma-terial in Soils Finer Than No. 200 (75 lm) sieve,” AnnualBook of ASTM Standards, Vol. 04.08, ASTM International,West Conshohocken, PA, pp. 103–106.


ParametersControl

SpecimenErosion

Specimen













ASTM D-1557, 2009, “Standard Test Method for LaboratoryCompaction Characteristics of Soil Using Modified Effort(56000 ft-lbf/ft3 (2700 kN-m/m3)),” Annual Book of ASTMStandards, Vol. 04.08, ASTM International, West Consho-hocken, PA, pp. 124–133.

Bendahmane, F., Marot, D., and Alexis, A., 2008, “ExperimentalParametric Study of Suffusion and Backward Erosion,” J. Geo-tech. Geoenviron. Eng., Vol. 134, No. 1, pp. 57–67.

Kenney, T. C. and Lau, D., 1985, “Internal Stability of GranularFilters,” Can. Geotech. J., Vol. 22, pp. 215–225.

Leonards, G. A. and Deschamps, R. J., 1998, “Failure of CyanideOverflow Pond Dam,” J. Perform. Constr. Facil., Vol. 12, No.1, pp. 3–11.

Papamichos, E., Vardoulakis, I., Tronvoll, J., and Skæjrstein, A.,2001, “Volumetric Sand Production Model and Experiment,”Int. J. Numer. Analyt. Meth. Geomech., Vol. 25, pp. 789–808.

Petrukin, V. P. and Presnov, O. M., 1991, “Construction Proper-ties of Soils. Suffusion Compressibility of Cemented SiltySands,” Osnovaniya, Fundamenty i mekhanika gruntov, No. 5,pp. 14–16 (in Russian).

Reddi, L. N., Lee, I., and Bonala, M. V. S., 2000, “Comparison ofInternal and Surface Erosion Using Flow Pump Test on aSand-Kaolinite Mixture,” Geotech. Testing J., Vol. 23, No. 1,pp. 116–122.

Richards, K. S. and Reddy, K.R., 2008, “Experimental Investiga-tion of Piping Potential in Earthen Structures, Geosustainability

and Geohazard Mitigation,” Proceedings of GeoCongress 2008,ASCE Geotechnical Special Publication No. 178, Reddy et al.,Eds., ASCE Geo-Institute, pp. 367–376.

Rönnqvist, H., 2010, “Predicting Surfacing Internal Erosion inMoraine Core Dams,” Master’s thesis. Hydraulic EngineeringResearch Group, Department of Land and Resources Engineer-ing, Royal Institute of Technology, Stockholm, Sweden.

Sanchez, R. L., Strutynsky, A. I., and Silver, M. L., 1983,“Evaluation of the Erosion Potential of Embankment CoreMaterials Using the Laboratory Triaxial Erosion TestProcedure,” Technical Report GL-83-4, Army Engineer Water-ways Experiment Station, Geotechnical Lab, Vicksburg, MS.

Sherard, J. L. and Decker, R. S., Eds., 1977, “Dispersive Clays,Related Piping, and Erosion in Geotechnical Projects,” STP623, ASTM International, West Conshohocken, PA.

Wahler, W. A., 1973, “Analysis of Coal Refuse Dam Failure,Middle Fork Buffalo Creek, Saunders West Virginia,” ReportNo. PB-215, National Technical Service, Washington, D.C.,pp. 142–143.

Wan, C. F. and Fell, R., 2004, “Investigation of Rate of Erosionof Soils in Embankment Dams,” J. Geotech. Geoenviron.Eng., Vol. 130, No. 4, pp. 373–380.

Wan, C. F. and Fell, R., 2008, “Assessing the Potential of InternalInstability and Suffusion in Embankment Dams and TheirFoundations,” J. Geotech. Eng., ASCE, Vol. 134, No. 3, pp.401–407.



Experimental Investigation of the Effects of Suffusion on Physical and Geomechanic Characteristics of Sandy SoilsIntroductionMaterials and MethodologyExperimental SetupMaterials and Soils CharacterizationTest ProcedureTest Program

Results and DiscussionSuffusion CharacteristicsPermeability Test ResultsVolume ChangeTriaxial Compressive Strength Test ResultsFurther Discussion of the Test ResultsLimitations of This Research

Conclusions and RecommendationsConclusions

References

experimental investigation of the effects of suffusion on ... papers...abstract: this paper reports...

Documents