STRENGTH AND DEFORMATION CHARACTERISTICS OF SCORIA …

27

i) Graduate Student, Department of Civil and Environmental Engineering, University of Yamanashi, Japan (shntyan＠yamanashi.ac.jp).ii) Associate Professor, ditto.

The manuscript for this paper was received for review on September 20, 2006; approved on September 20, 2007.Written discussions on this paper should be submitted before September 1, 2008 to the Japanese Geotechnical Society, 4-38-2, Sengoku,Bunkyo-ku, Tokyo 112-0011, Japan. Upon request the closing date may be extended one month.

27

SOILS AND FOUNDATIONS Vol. 48, No. 1, 27–39, Feb. 2008Japanese Geotechnical Society

STRENGTH AND DEFORMATION CHARACTERISTICS OF SCORIA INTRIAXIAL COMPRESSION AT LOW CONFINING STRESS

YANYAN AGUSTIANi) and SATOSHI GOTOii)

ABSTRACT

Scoria deposit can be found around many volcanoes of the Quarternary epoch in the world. In Japan, Mount Fuji isone of volcanoes, where surface area is widely layered by scoria. From a geotechnical engineering point of view, scoriahas many problems because of its physical characteristics by means of its collapsible properties. Since scoria is a non-cohesive material, it is vulnerable to water-induced change and fails easily due to rainfall or snow melting. These arethe primary causes of debris ‰ow which frequently occurs in the Mount Fuji area. Typically, this type of debris ‰ow iscalled ``Yukishiro'' in Japanese. For better understanding of the shear behavior of scoria within low stress level condi-tion, drained triaxial compression tests were performed on loose dry scoria under low conˆning pressures (10–80 kPa)which are assumed as representative of the êld surface stress conditions. The eŠects of grain size, dry density rd andeŠective conˆning stress s?c upon stress-strain behaviour and angle of internal friction qd were investigated. Attentionwas also paid to level of grain breakage Bg and reproducibility of the test. Grain breakage investigations were carriedout on the particle breakage phenomenon, which occurs during testing. Additionally, initial Young's modulus of sco-ria was identiêd. As a result, reproducibility of the test is good in terms of stress-strain curves, however volumetricstrain versus axial strain exhibits less reproducibility than stress-strain curves. At stress level of 10–80 kPa, it wasfound that the dependency of qd on s?c is almost negligible, while there is noticeable dependency of qd on grain size.The eŠect of grain size on stress-strain relationships was observed. It was also found that grain breakage was noticed atstress level of 10¿80 kPa, which describes the dependency scoria parameters on conˆning stress and the eŠect of grainsize upon Bg-s?c relationships was also observed.

Key words: dilatancy, grain breakage, low conˆning stress, scoria, shear strength, triaxial compression test (IGC: D6)

INTRODUCTION

Behind its beauty and popularity, Fuji is a mountain ofmany geological hazards, including volcanic activity andslope failure, i.e.: debris ‰ow, avalanches, rock fall etc.Many researchers predict a new eruption may spew lava,debris and ash over hundreds of square miles threateningnearby villages, cutting oŠ electricity and water supplies,and disrupting road, rail and air travel. Debris ‰ow is oneof the natural hazard phenomena that frequently occur inMount Fuji area. Typically, this type of debris ‰ow isknown as ``Yukishiro'' in Japanese. This slope deforma-tion can be classiêd to a rapid mass movement in whichloose mass (scoria) combine with air, and water to formslurry that then ‰ows down slope, usually associated withsteep slope. Recently, Yukishiro attracted wide public at-tention due to its disaster which aŠected quite a large areanear to the Five Step, an entrance place to climb MountFuji from Yamanashi Prefecture side. Photo 1 shows thedamage caused by Yukishiro occurred on 5th December2004.

The area aŠected by the Yukishiro is mostly underlain

by scoria (on the top surface). This scoria is grouped into``Younger Fuji Tephra'' which is composed of scores ofdark brownish scoriaceous tephras, reaching severalmeters in thickness in the Fuji area (Tsuya et al., 1988).Outcrops around the debris area expose mostly a thickand thin pile layers of scoria and air-fall ash.

The key factors in devising approaches to soil erosionand slope stability evaluation are measurement andcharacterization of the mechanical and structural proper-ties of soil. Drained triaxial compression test is a com-mon method to measure the mechanical properties ofsoil. This test is typically used for stability analyses ofgranular soils, which drainage conditions or path fol-lowed in the triaxial test are models of speciˆc critical de-sign situations required for the analysis of stability ana-lyses in engineering practices.

The physical properties of scoria are characterized bythe peculiar shape and surface texture of grain, highdegree of particle breakage, granular, cohesionless, andfull of vesicles. These physical properties cause an areawith top surface covered by scoria prone to collapse easilyeven though there is a small amount of increase in exter-

28

Photo 1. 5th December 2004 Yukishiro. a) erosion and scouring processes deepening the pre-existing channel and b) covered about 6 meters ofheight a roadway

28 AGUSTIAN AND GOTO

nal load or reduction in the eŠective mean stress. Fieldconditions show that scoria lies at the top surface withdry and loose state so that the element test at laboratoryat dry condition at low stresses considered represent realconditions to measure its mechanical properties.

In order to characterize the shear behaviour of scoriawithin low stress level condition, drained triaxial com-pression tests were performed on dry scoria under lowconˆning pressures (10–80 kPa) which are assumed as therepresentative of the êld surface stress conditions.DiŠerent from the normal sand, scoria is quite di‹cult toperform in triaxial test under saturated condition, be-cause of its physical properties. The authors assume thatthere are no signiˆcant diŠerences between air-dry andsaturated samples of scoria in strength and deformationcharacteristics in triaxial compression at low stress condi-tions. Tatsuoka et al. (1986b) found that the diŠerencesin the deformation and strength characteristics of Toy-oura were very small or negligible between air-dry andsaturated samples in triaxial compression. Therefore, inthis study, the tests were held under dry condition withrefers to drained triaxial compression tests. The eŠects ofgrain size and conˆning pressure upon stress-strain be-haviour to angle of internal friction and dilatancy are in-vestigated.

A comprehensive study of stress-strain and strengthproperties of scoria has not been studied adequately, es-pecially at low conˆning stress conditions. The aim ofthis study is to characterize the fundamental scoriaparameters, including its angle of internal friction, anddilatancy angle whose parameters are then basically re-quired for use in the modelling of scoria behaviour, i.eslope stability evaluation and to predict how the scoriamaterial will behave in a larger-scale engineering applica-tion. Moreover, the authors have stressed noticeable par-ticle breakage of scoria which depends on stress level,maximum grain size and aŠect the grain size distribution.

Kusakabe et al. (1991) investigated the deformationcharacteristic of scoria in triaxial compression tests atrange of 20¿4700 kPa of conˆning stress. They reportedthat undisturbed scoria taken from 20¿27 m below

ground surface was highly compacted with relative den-sity of more than 100z and has apparent cohesion ofabout 100 kPa. They also found that the behaviour ofscoria changes from a dilatant brittle material to a plasticmaterial in 700¿1500 kPa of conˆning pressure range,exhibiting volume reduction with particle crushing.Fukushima and Tatsuoka (1984) performed a series ofdrained triaxial compression tests at low pressures onToyoura sand. They found that the angle of internal fric-tion q of saturated Toyoura sand does not change somuch with the change of the minor principal stress s?3when s?3 is lower than around 50 kPa. Tatsuoka et al.(1986a) performed plain strain compression tests on To-youra sand at extremely low pressures and reported thatthe dependency of q on s?3 is very small in plain straincompression for a stress range of s?3＝5¿50 kPa.

EXPERIMENTAL METHOD

Test ApparatusA conventional triaxial compression test apparatus has

been modiêd in our laboratory. The apparatus has beendesigned to test specimens with the initial dimensions h0

(height)＝200 mm and d0 (diameter)＝100 mm. The test isreferred to as vacuum test because the conˆning pressureis applied by partial evacuation of the air inside the dryspecimen. Figure 1 shows the schematic diagram of thetest apparatus used in this study.

Testing ProcedureThe principle of operation of the triaxial test in this

study is as follows: After placing a forming mold over thepedestal in the triaxial apparatus, and pulling the mem-brane through the mould, the membrane is rolled outover the top of the forming mold, and vacuum pressure isapplied in order to tighten the membrane to the formingmold and to eliminate wrinkles on the membrane. At theend of the air lines, a ˆlter paper is constructed to be ‰ushwith the inside of the specimen, so that scoria will not en-ter the line and plug it.

Pluviation of the sample is then carried out with cer-

29

Fig. 1. Schematic diagram of triaxial compression test used in thisstudy

29STRENGTH AND DEFORMATION CHARACTERISTICS OF SCORIA

tain height of pluviation to obtain the desired range ofdry density of the specimen. A thin steel ‰at is used to en-sure the top of the specimen is ‰at and smooth, movingdown the top cap ˆrmly and pulling the membrane uparound it then the membrane is ˆxed to the top cap. Byapplying the desired vacuum conˆnement to the specimenusing the regulator and vacuum gauge, the external at-mospheric pressure acts upon the specimen and make itstiŠ. The split mold is removed carefully and the dimen-sions of the sample are measured.

To measure the local axial and lateral strain, a pair ofLocal Deformation Transducer, LDT (Goto et al., 1991)and three clip gauges (Tatsuoka et al., 1994) are set up onthe specimen. After setting up the LDTs and clip gauges,the triaxial apparatus is placed into the loading machineand carefully centered. The piston is moved downwardsuntil it is in contact with the upper plate which is connect-ed rigidly to the load cell and top cap. The conˆning pres-sure s?c was due partial air evacuation of the latex mem-brane, using a vacuum pump connected to the bottompedestal and the top cap of the specimen. After about 2hours of isotropic consolidation under conˆning pressureof 10¿80 kPa, a series of triaxial compression was car-ried out under drained conditions. Specimen is loaded ad-ditionally by a compressive stress q by moving the load-ing piston downwards at a ˆxed displacement rate with anaxial strain rate of 0.17z/min. mainly, with the conˆn-ing pressure held constant. Another two diŠerences of ax-ial strain rate, 0.50 and 0.80z/min, were performed toinvestigate the eŠect of strain rate upon stress-strain be-havoiur of scoria. All values of stresses and strains wereautomatically recorded by computer.

Measurement DevicesThe initial dry density of scoria is given by:

rd＝ms

(pd0h0/4)(1)

In which ms, d0 and h0 are weight of dry scoria, initialdiameter and initial height of the specimen, respectively.The measurements of the initial dimension of the speci-mens were held at 10 kPa of vacuum applied. A load cell

was connected rigidly to the rod which was further con-nected to the loading piston. Its capacity is 5000 N to anaccuracy of ±0.5 N. The applied partial vacuum acts aseŠective conˆning pressure s?c or eŠective minor principlestress s?3 during the triaxial compression test, with an ac-curacy of s?3＝±0.5 kPa. Stress and strain parameters aredeˆned as positive when representing a compressive state,in this paper. Before any barreling of the specimen oc-curs, the principal stresses are assumed uniform insidethe scoria specimen and have the following expressions:

s?1＝s?3＋q (2)

where s?1 is the eŠective major principle stress, s?3 is theeŠective minor principle stress and q is the deviatoricstress.

And the principle minor stress was calculated as:

s?3＝sVC＋Dsm (3)

where, sVC is the vacuum conˆnement and Dsm is thestress correction for the forces working in the membrane.The stress correction was made by using the followingequation (Henkel and Gilbert, 1952):

Dsm＝－2tEm

de3 (4)

The measurement results show that the latex membraneused in the present tests has Young's modulus average ofEm＝1285 kPa. In the initial state, the diameter d andthickness t of the membrane is 97.00 mm and 0.2 mm, re-spectively.

The lateral strain of the specimens were measured di-rectly by three clips that are mounted, which contact thesample in the upper, middle, and the lower parts, respec-tively (Tatsuoka et al., 1994). These bronze clips areequipped with electric strain gauges, that are called ``clipgauges''. An increase in the sample diameter causes achange of curvature of the clip gauges, which results inlocal radial strain being measured. The lateral strain is theaverage of those measured using the three clip gauges.Since only up to about 4z of lateral strain can be meas-ured by the clip gauges, the radial strain calculationshould be done for all the tests where the clips were re-moved immediately before it reaches 4z of lateral strain.

The results are evaluated following the Rowe's stress-dilatancy theory as follows:

R＝s?1s?3＝K Ø1－dev

de1» (5)

where R is the stress ratio, dev is the volumetric strain in-crement, de1 is the major principal strain increment and K＝tan2 (45＋qf/2), in which qf is angle of internal frictionat failure. And the volumetric strain increment is ex-pressed as:

dev＝de1＋2de3 (6)

where de3 is the minor principal strain increment. By sub-stituting Eq. (6) into Eq. (5) one can obtain

30

Fig. 2. Plot of stress ratio R and minor principle compressive strain e3

versus major principle compressive strain e1 at each time step.Dashed line of R and e3 are denoted the calculation results by usingRowe's stress-dilatancy theory. Inset is two examples of data ˆlter-ing at 1st and 2nd data reading

Photo 2. Triaxial compression test apparatus with LDTs and clipgauges mounted on the specimen

Fig. 3. Grain size distribution of scoria of each series before triaxialtest


R＝s?1s?3＝K Ø－2de3

de1» (7)

The calculation of the rate of dilatancy (2de3/de1) isderived from major and minor principle compressivestrain data as a function of elapsed time t, e3(t) and e1(t).To limit scatter on (de3/de1), primarily e3 and e1 areˆltered by averaged over the 0.085z axial strain (60 datapoints, where each test comprises about 4000 data pointsbefore the clip gauges unmounted from the specimen).Figure 2 shows the plot of the major principle compres-sive strain e1 and the minor principle compressive strain e3

at each time step. The number of ˆltering points waschosen, 0.085z axial strain in this paper, so that theshapes of initial data are respected and ˆltered. The (de3

/de1) obtained then plotted against stress ratio R. Afterthat, K is directly derived from this plot as the parameterof the linear ˆt between R and the dilatancy rate (－2de3

/de1).K obtained from this method is then used for the calcu-

lation of the minor principle strain increment de3 wherethe data is blank since the clip gauges are removed fromthe specimen. From Eq. (7) de3 right after clip gauges areremoved, say at i th reading, is calculated as:

(de3)i＝－1

2K Øs?1s?3»i－1

(de1)i (8)

This value of de3 at i th reading then is used for the calcula-tion of diameter A of specimen and major principle stresss?1 at i th reading. And repeated calculation is done for thenext readings so that the values of ev and s?1 are obtaineduntil the last value where the test is terminated.

One may pay attention to the fact that volumetricstrains are measured more directly in the triaxial testingof normal sands, for example, by ensuring that samplesare saturated with water prior to testing, and by accurate-ly recording in‰ow and out‰ow of pore water during atest. In the current case, it is di‹cult to achieve the satu-rated condition for scoria sample, because of its physicalproperties i.e. scoria is full of bubble holes make air

trapped on it which is di‹cult to escape when scoria issubjected into saturated condition. Instead, a partialvacuum was applied to follow the direct approach on ele-ment test on the scoria on dry condition.

LDT, a simple gauge developed by Goto et al. (1991)was used for measuring the axial strain locally. LDT con-sists of a thin plate of phosphor bronze on both sideswhere a strain gauge is glued. A pair of LDTs was set uparound the sample so that the average of local strain overthe central part of the sample could be measured. Photo2 shows the set up of the clip gauges and the LDTs ontriaxial specimen.

Sample PreparationThe scoria used in this study was obtained from the

north ‰ank of Mount Fuji, Yamanashi Prefecture. Fourseries of grain size distributions of scoria used in thisstudy are shown in Fig. 3 Particles above 19.5 mm in di-ameter were removed from the original scoria obtainedfrom the source to keep the ratio of the specimen di-ameter (100 mm) to the maximum particle above ˆve.Particles above 9.5, 4.5 and 2 mm in maximum particlesize of this original scoria have also been cut in order to

31

Table 1. List and some results of the tests

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

Dmax

(mm)Series Test code

s?c(kPa)

rd

(g/cm3)·ea

(z/minute)(s?1/s?3)max

(ev)max

(z)qd

(degree)E0

(MPa)Bg

(z)

19.00

A loose

A1 10.51 1.280

0.17

5.530 0.047 43.92 19.51 1.15

A2 20.37 1.230 5.446 0.139 43.61 51.56 1.32

A3 49.76 1.230 5.654 0.255 44.38 85.87 4.80

A4 80.00 1.210 5.295 0.682 43.02 126.04 9.12

A5 79.77 1.270 5.330 0.970 43.16 104.13 9.89

A6 79.97 1.270 5.325 0.863 43.14 129.54 10.49

B dense

B1 10.54 1.350 5.490 0.053 43.78 26.69 1.17

B2 20.16 1.320 5.350 0.151 43.24 36.79 1.45

B3 49.83 1.300 5.460 0.601 43.66 85.30 4.45

B4 80.01 1.320 5.272 0.785 42.93 102.91 9.44

9.50

C loose

C1 10.50 1.231 4.908 0.164 41.41 18.44 0.90

C2 20.46 1.214 4.893 0.144 41.35 39.76 1.21

C3 50.17 1.218 5.065 0.398 42.08 55.49 4.18

C4 76.12 1.165 4.750 0.899 40.71 91.01 7.83

D debse

D1 10.41 1.305 4.902 0.254 41.39 22.72 0.97

D2 20.48 1.297 4.873 0.413 41.26 30.91 1.07

D3 50.75 1.284 5.057 0.856 42.05 58.89 4.10

D4 79.12 1.293 4.628 1.117 40.14 79.03 8.89

4.75

E loose

E1 10.71 1.281 4.699 0.155 40.47 17.96 0.97

E2 20.28 1.291 4.704 0.404 40.49 31.45 0.87

E3 50.61 1.287 4.785 0.828 40.87 53.49 3.86

E4 80.30 1.271 4.751 0.975 40.71 65.46 7.23

F

dense

F1 50.65 1.368 4.631 0.531 40.15 52.31 3.58

F2 50.51 1.343 4.564 0.733 39.83 53.28 4.21

F3 50.43 1.350 0.50 4.480 1.008 39.42 42.65 3.39

F4 50.44 1.340 0.80 4.678 0.825 40.37 48.58 4.21

G

G1 10.49 1.315

0.17

4.576 0.387 39.89 16.88 1.52

G2 20.94 1.312 4.614 0.612 40.07 36.07 0.92

G3 50.45 1.302 4.645 0.825 40.22 66.62 3.95

G4 72.57 1.325 4.650 1.037 40.24 76.69 5.78

2.00

H loose

H1 10.14 1.339 4.384 0.099 38.94 20.20 0.89

H2 20.59 1.342 4.197 0.464 37.97 27.32 1.12

H3 50.50 1.318 4.510 0.475 39.57 53.36 3.56

H4 76.47 1.343 4.471 0.70 39.38 68.94 6.23

I dense

I1 10.39 1.440 4.377 0.173 38.90 18.23 1.23

I2 20.32 1.410 4.416 0.408 39.10 22.91 1.15

I3 50.41 1.419 4.579 0.629 39.90 44.41 3.55

I4 80.76 1.425 4.472 0.829 39.38 50.00 5.40

NOTE(1) maximum grain size, (4) eŠective conˆning pressure, (5) dry density at s?c＝10 kPa, (6) strain rate, (7) stress ratio, (8) maximum volumetric strain,(9) angle of internal friction, (10) initial Young' modulus, (11) grain breakage calculated based on Marshal's method


32

Fig. 4. Reproducibility of the test. Test reproducibility is good in termof stress-strain relation

Fig. 5. Stress-strain relationships at diŠerent axial strain rate

Fig. 6. Externally and locally measured axial strain from various grainsize at 20 kPa of stress level


investigate the strength and deformation characteristicsof scoria in various maximum particle sizes. The types ofparticle size distributions were chosen to keep the originalparticle size distribution of the êld. Mainly, grains con-sisted of angular fragments with the solid density scoria rs

＝2.72 g/cm3. The samples were prepared by air pluvia-tion. A pluviation apparatus was set up so that during theparticles falling into mould, the scoria container wasmoved upwards to keep the drop height between the con-tainer and the mould constant. Densiˆcation of the sam-ples was accomplished by increasing the pluviationheight. The samples used in the tests were divided into 4series of diŠerent grain size and each series was dividedinto two diŠerent ranges of densities. The details areshown in Table 1.

TEST RESULTS AND DISCUSSION

The reproducibility of the test is ˆrstly investigated fol-lowed by the stress-strain behaviour. The tests have beenconducted to explore the eŠect of strain rate, scoria grainsize, initial density and eŠective conˆning stress upon theangle of internal friction, Young's modulus and stress-dilatancy response. Finally grain breakage of scoria parti-cle was also investigated.

Table 1 shows the resume of the lists and alsohighlights some results from the tests.

Reproducibility of the TestsIn the present tests, reproducibility was also studied by

performing tests under the same initial density and thesame conˆning stress. For each test, repeated tests werecarried out in the present study. If a large deviation is ob-served then a further test is conducted. Figure 4 showsthree repeated tests on dry-loose scoria for Dmax＝19.00and 4.75 mm. It can be seen that the reproducibility ofthe test is good in terms of stress-strain curve, contrarilyto volumetric strain which is less reproducible. Theseresults are typical of those obtained for loose sample ofsand. Further, evaluating volumetric strain from localsample diameters add to the scatter, especially where thedeviator stress is reaching its ultimate value, make the

result of volumetric strain evolution vary from test totest.

To investigate the eŠect of strain rate on stress-strainrelationships, triaxial compression test on scoria has alsopreformed at 3 diŠerences of axial strain rate. Based onthe results shown in Fig. 5, it can be considered that thestress-strain relations are not aŠected by strain rate forstrain rate of 0.17, 0.50 and 0.80z/min.

Stress and Strain RelationsIn the evaluation of deformation behaviour of soil

specimen in triaxial test, it is important to investigate thesource of error involved in external axial strain, such asthe bedding error of the top and bottom of the specimen.A pair of LDT s was set up on the specimen, so that theinternal axial strain can be measured and the overallbedding error at the top and bottom of specimen can beestimated. However, stress strain plot results of scoria inthis study show that there is no signiˆcant diŠerence be-tween internal and external axial measurement. Figure 6shows the results of four typical tests of those performedin the present study, showing stress-strain relationship ofscoria at 20 kPa of stress level, both externally and locallymeasured axial strain. It may be seen from Fig. 6, thatthere is almost no signiˆcant diŠerence in the stress strain

33

Fig. 7. Relationships between stress ratio, axial strain and volumetricstrain of dry scoria with Dmax＝19.00 m. (a) rd＝1.210¿1.280g/cm3 and (b) rd＝1.300¿1.350 g/cm3

Fig. 8. Relationships between stress ratio, axial strain and volumetricstrain of dry scoria with Dmax＝9.50 mm. (a) rd＝1.165¿1.231g/cm3 and (b) rd＝1.284¿1.305 g/cm3


relationship between the internal and external measure-ment. A large degree of divergence of internal and axialstrain was not observed in all data. They exhibit the samemanner as shown by Fig. 6, that is to explain the overallbedding error at the top and the bottom of the specimensis very small and negligible.

Figures 7 through 10 show the relationships betweenstress ratio s?1/s?3, axial strain e1 and volumetric strain ev

of all series of drained triaxial compression tests on sam-ples having similar densities. Stresses have been plottedby normalizing the eŠective principle stress ratio s?1/s?3and axial strains are plotted up to 16z. On all stress-strain data plot, hardening behaviour with peak stresswas observed. When considering the evolution of volu-metric strain versus axial strain, an initial compressivestage was observed, followed by gradual dilatation of thespecimen. The following trends of behaviour may be seenfrom the ˆgures:

The eŠect of the direction of applied major principalstress on the stress-strain behaviour is dependent on theconˆning stress, as can be conˆrmed from the Figs.7¿10. The behaviour of volumetric strain curves cor-responds, well to that of stress ratio-axial strain curves.That is, the stress ratio becomes larger, the larger thevolume change due to dilatancy is. Concerning the defor-mation characteristics, it may be seen from these ˆguresthat for all series, dense or loose, the slope d(s?1/s?3)/de1

at a same stress ratio level decreases with the increase ins?c before peak condition, while the peak stress ratioremains relatively independent on the eŠective conˆningstress s?c and initial density rd. The axial strain e1 at maxi-mum stress ratio increase with the increase in s?c, especial-ly at s?c≧20 kPa. After the maximum is passed, the prin-cipal stress ratio decreases slightly with an increase in axi-al strain and appears to have a tendency ˆnally to becomeconstant. Rather than stress-strain curves, volumetricstrain evolution with axial strain exhibits noticeablevariability.

Unsmooth stress strain relations seen in the lower con-ˆning stress (s?c≦20 kPa) may be due to the stick-slip be-haviors at the lower conˆning stresses, which is usuallyaccompanied by an abrupt decrement of axial stress.

The eŠect of grain size on the stress strain relationshipsat similar stress levels was also investigated. Figures 11through 14 are plots of the stress strain relationships ofscoria in various grain sizes at similar stress levels. Thefollowing eŠects of grain size on the stress-strain relation-ships maybe seen from the ˆgures as trends of behaviour.Firstly, before the maximum stress ratio, the slope d(s?1/s?3)/de1 at a similar stress ratio level decreases with thedecrease in grain size, while the axial strain e1 at the maxi-mum stress ratio is independent of grain size. Secondly,larger grain size gives higher maximum stress ratio atstress-strain relationships and from these ˆgures it can be

34

Fig. 9. Relationships between stress ratio, axial strain and volumetricstrain of dry scoria with Dmax＝4.75 m. (a) rd＝1.281¿1.293 g/cm3

and (b) rd＝1.302¿1.325 g/cm3

Fig. 10. Relationships between stress ratio, axial strain and volumetricstrain of dry scoria with Dmax＝2.00 m. (a) rd＝1.318¿1.343 g/cm3

and (b) rd＝1.410¿1.440 g/cm3


considered that the maximum stress ratio is strongly de-pendent on grain size.

Relationships between qd, s?c and Grain SizeThe angle of internal friction of scoria, qd, determined

by the Eq. (9) at the peak value of stress ratio, is plottedagainst eŠective conˆning stress for various grain sizes asshown in Fig. 15.

qd＝arcsinØs?1s?3»max

－1

Øs?1s?3»max

＋1(9)

The results show that the dependency of qd on s?c areconsidered negligible for the range of s?c＝10¿80 kPa forall various grain sizes. The dependency is very small ifany; these are for larger grain size, say Dmax＝19.00 and9.50, at stress level of s?c＝50¿80 kPa. It may be due tograin breakage of scoria particles, where larger grain sizeof scoria particles exhibit higher crushability than smallerones.

To show the eŠect of grain size on the angle of internalfriction in these specimens, the curve of Fig. 15 are cross-plotted in Fig. 16 to show the angle of internal frictionangle as a function of the maximum grain size of eachspecimen. These curves show that the angle of internal

friction increases as the maximum grain size increases.The angle of internal friction of samples with a maximumgrain size of 19.5 mm were in general 4 to 5 degrees morethan those of the samples with a maximum grain size of 2mm, about 2.5 to 3.5 degrees more than those of the sam-ples with 4.75 mm maximum grain size, and 1.7 to 2.8degrees more than those of the samples with 9.5 mm max-imum grain size.

The same tendency is also seen in Takase dam materialinvestigation results (JGS, 1986) for well graded particledistribution samples. The increase in maximum grain sizewas followed by the increase in angle of internal friction.However, the opposite tendency is seen from sampleswith geometrically similar particle distribution. Whilescattering of the data lower than 40 mm of maximumgrain size are seen, the curves show the angle of internalfriction decreases as the maximum grain size increases.These tendencies are directly related to uniformity of par-ticles, maximum grain size and particle breakage. It iswell known that uniform particle size distribution sam-ples exhibit more crushing than well graded samples hav-ing the same maximum size. This may be due to particlesinterlocking with well graded samples that is higher thanuniform graded samples make well graded samples moreresistant to failure or crush compared to samples havinguniform graded particles distribution. And uniformgranular material composed of large particles crush more

35

Fig. 11. Relationships between stress ratio, axial strain and volumetricstrain of dry scoria on diŠerent grain size at 10 kPa of stress level.(a) rd＝1.280¿1.339 g/cm3 and (b) rd＝1.309¿1.440 g/cm3



readily than one composed of smaller particles of thesame material.

Previous researchers have also focused on investigatingthe eŠects of maximum grain size on the shear strengthbehavior. Marachi et al. (1969) have investigated thestrength of Pyramid Dam material by testing a series ofsamples with geometrically similar particle size distribu-tions in triaxial compression. For modeled specimens ofthe Pyramid Dam material, the results indicated that theangle of internal friction is aŠected by the size of particlesin test specimens. They found that the angle of internalfriction decreases as the maximum grain size increases.And this trend seemed to be unaŠected by the conˆningpressure or the material type.

Nakata et al. (2001) concluded that in the case of thewell graded sand, the number of particle contact for largeparticles is very high while the opposite is true for thesmaller particles. In this case the tensile splitting stress forthe large particles is relatively low while that of the smallparticles with a low particle contact is much higher. Eventhough larger particles are statistically weaker, theirprobability of survival in the matrix is much higher be-cause their coordination number is much higher. Largestrain soil behavior involves particle rotation and contactslippage.

Naturally the larger grain size of scoria used in thisstudy has a larger angularity of particles; as a result, larg-

er grain size particle has a higher possibility in the break-ing of its asperities, when a loading is applied on it. It canbe considered that higher degree of crushability for largergrain size of the scoria in this study is the result of an in-crease in the breaking of its asperities and surface grind-ing, rather than particles splitting. At low density condi-tion, the interparticle coordination is low, shear deforma-tion causes particle rotation and chain buckling, and thepacking gradually densiês. However, rotation is foiledin dense states with high interparticle coordination; there-fore, energy applied during shear loading is consumedeither in dilation or in frictional slippage at contacts. Wi-thin this particle-level mechanical framework, angularityand roughness may add di‹culty to particle rotation, en-hance dilatancy and the evolution of anisotropy.

Stress-dilatancy RelationshipsRowe (1969) proposed a stress-dilatancy theory for the

shearing of a granular aggregate accounting for inter-locking on what would have been a preferred plane ofsliding. At low stress ratios, particles rearrange them-selves in an organized manner attaining a denser packing.From that stage, particles can move no more except byoverriding one of their neighbouring particles at a micro-scope level, the aggregate tends to dilate. Rowe relatesprincipal stress ratio s?1/s?3 to the rate of dilatancy dev/dea

as shown in Eq. (5).

36



Fig. 15. Relationship between angle of internal friction qd and eŠec-tive conˆning stress s?c


Figure 17(a) shows the plot of stress-dilatancy relationscalculation results of series A. K was obtained as a linearˆt between stress ratio and dilatancy, directly derivedfrom this plot. It may be seen that the dependency of thestress-dilatancy relation on s?c is found from this ˆgure.While the data are not presented here, similar trends wereseen for other series examined. Figure 17(b) is the plot ofK-s?c relationships of all series at various grain sizes.While scattering of the data is not small, the results showthat dependency of K on Dmax are seen for Dmax＝2¿9.5mm. The results showing the eŠects of maximum grainsize Dmax on dilatancy rate K are somewhat inconclusive.Although the general trend of the results seems to indi-cate that the value of K increases as the maximum particlesize increases as shown in Fig. 17(c).

The data points of stress dilatancy relations at peakstress condition of scoria in triaxial test of all series areshown in Fig. 18. It is indicated in Fig. 18 that the trenddata points of scoria in this study is similar with the trendof scoria by Kusakabe et al. (1991) at stress level19.6¿4704 kPa, as well as Toyoura sand by Fukushimaand Tatsuoka (1984) at low stress level. The results showclearly that the stress dilatancy relation at maximumstress ratio Rmax of scoria in triaxial compression is de-pendent on eŠective conˆning stress s?c. The results showclearly that the stress dilatancy relation at maximumstress ratio Rmax of scoria in triaxial compression is aŠect-

ed by eŠective conˆning stress s?c. Scattering of the data isseen from the ˆgure, especially for 80 kPa of stress level.However, it can be considered that the slope of the linearˆt between Rmax and D is large for the larger value of s?c.

Deformation CharacteristicsIt may be seen from Figs. 7 through 10, that the volu-

metric strain-axial strain curves show diŠerences at diŠer-ent conˆning stress. At higher stresses this diŠerence in-crease with the increase in conˆning pressure. Figures

37

Fig. 16. The dependency of internal friction angle qd on grain size atstress level of 10¿80 kPa of scoria compared to Takase dammaterial

Fig. 17. Stress-dilatancy calculation results from Row's theory. (a) ex-ample plots of principle stress ratio R versus dilatancy rate D ofdata series A. K is directly derived from the plots as the parameterof the linear ˆt between R and D. (b) The dependency of K on s?c.(c) Relationship between K and Dmax


19(a) and (b) show the relationships between volumetricstrain ev and conˆning stress s?c. The results show clearlythat larger s?c gives larger value of ev, which means thelarger the s?c applied, the smaller the volume change ofscoria sample is. Larger grain size exhibits more deforma-ble than smaller grain size consistently, both for ev at ea＝2z and ev at maximum, while smaller grain size display anoticeable variability of volumetric strain.

The initial Young's modulus was determined for alltests as the slope of the best linear ˆt in a least squaresmethod for data points obtained from q-ea (locally meas-ured axial strain by LDT) relations for axial strain wassmaller by up to 0.01z (see Fig. 20(a)). A linear ˆt wasperformed in a least squares method between E0 and s?c.The dependency of initial Young' modulus E0 on eŠectiveconˆning stress s?c for each series of tests performed, isshown by Fig. 20(b). While some scatters appear on thedata, it can be considered that the relationship of E0 ands?c are aŠected by grain size.

Evaluation of Scoria Grain BreakageThe grain size distribution has been evaluated before

and after triaxial compression tests by traditional tech-niques. Measurements have been conducted on threesamples of average mass of each specimen. The resultsshow that the grain size distribution is noticeably aŠectedby triaxial testing. Curves in Fig. 21 show the diŠerencein grain size distribution before and after the tests. It isconcluded that the scoria particles exhibit plastic defor-mation or crushing at particle contacts during triaxialcompression test.

The degree of crushability was analyzed by using Mar-sal grain breakage calculation method (1973). Principal-ly, grain breakage Bg is obtained by calculating the diŠer-ence between percentage of the total sample contained ineach grain size fraction before and after the test. Graphi-cally, the calculation is explained by Fig. 22(a). The totalsum of the diŠerences DWk must be zero. Then, grainbreakage Bg is equal to the sum of the positive DWk, andexpressed in percent.

Figure 22(b) shows the dependencies of grain breakage

on stress level, while the increase of crushing degree withthe increase of grain size can also be seen from the ˆgure.It may be seen from Fig. 22(b) that grain breakage Bg at10–20 kPa of stress level does not change so much, allother things being equal. In contrast, the change of Bg isfound to be very clear at s?c higher than 20 kPa. The eŠectof grain size upon Bg-s?c relationships can be observedfrom this ˆgure. The diŠerence of Bg between the smallergrain size and the larger grain size which increase with theincrease in s?c, is also remarkable. It is well known thatthe degree of crushing is directly related to the particlestrength. For the same materials, the granular material

38

Fig. 18. Stress-dilatancy relations at peak of scoria in triaxial com-pression compared with previous researches on scoria (Kusakabe etal., 1991) and Toyoura sand (Tatsuoka et al., 1986)

Fig. 19. Relationships between: (a) ev-s?c at ea＝2% and (b) mximumvolumetric strain (ev)max and s?c

Fig. 20. Determination of initial Young's modulus E0: (a) four exam-ples of series A and (b) grain size eŠect upon E0-s?c relationship

Fig. 21. Grain size distribution of scoria before and after triaxial test


composed of large particles with a high angularity crush-es more readily than one composed of smaller particles,because of stress concentrations at their angularities, asshown by scoria in this study.

CONCLUSIONS

The strength and deformation characteristics of scoriahave been studied in triaxial compression. While the

presented data is limited, this experimental study hashelped to explain the behaviour of dry scoria aggregatessheared at low conˆning stress (10 to 80 kPa) in drainedconditions. The results show that there is no strikingdiŠerence in shear strength behaviour from that reportedfor sand. The eŠects of maximum grain size on shearstrength behaviour and on grain breakage has been eval-uated. While stress level (10 to 80 kPa,) is not aŠectingthat much the shear strength behaviour of scoria, maxi-mum grain size and gradation should be taken into ac-count in evaluating the shear strength behaviour of scor-ia. The results are also conˆrmed by comparing scoria to

39

Fig. 22. Evaluation of grain breakage: (a) determination of grainbreakage Bg (%), Marsal (1973) and (b) calculation result for dryscoria in various grain size


that reported for rock ˆll dam material.From the results of the drained triaxial compression

tests on disturbed specimens of scoria at low conˆningstress, the following conclusions may be drawn:(1) In terms of stress-strain curves, test reproducibility is

quite good, contrarily to volumetric strain which ex-hibit less reproducible.

(2) The eŠect of grain size on stress-strain relationshipsat similar stress level was investigated, while the axialstrain e1 at maximum stress ratio is independent ongrain size.

(3) The dependency of internal friction of dry scoria qd

on eŠective conˆning stress s?c are considered negligi-ble for the range of s?c＝10¿80 kPa for all variousgrain size. The eŠect of grain size upon qd-s?c relation-ships were also observed.

(4) The stress-dilatancy relation of scoria at the maxi-mum stress condition in triaxial compression is in-‰uenced by s?c. Relatively, dense samples haveparameter K at peak stress ratio higher than that forloose one. The parameter K has its minimum value4.25 for dense sample and 4.21 for loose one at 10

kPa of s?c.(5) The increase of initial Young's modulus E0 of devia-

toric stress-axial strain was observed with the increaseof conˆning stress s?c. While some scatters appear onthe data, it can be considered that E0 and s?c relationsare aŠected by grain size.

(6) Grain breakage Bg was noticed at stress level of10¿80 kPa, which describes the dependency of sever-al scoria parameters on conˆning stress. The eŠectsof grain size upon Bg-s?c relationships were also ob-served. This may due to a high angularity of largeparticles which crush more readily than one com-posed of smaller particles, because of stress concen-trations at their angularities.

REFERENCES

1) Fukushima, S. and Tatsuoka, F. (1984): Strength and deformationcharacteristics of saturated sand at extremely low pressures, Soilsand Foundations, 24(4), 30–48.

2) Goto, S., Tatsuoka, F., Shibuya, S., Kim, Y. S. and Sato, T.(1991): A simple gauge for local small strain measurement in thelaboratory, Soils and Foundations, 31(1), 169–180.

3) Henkel, D. J. and Gilbert, G .C. (1952): The eŠect of rubber on themeasured triaxial compression strength of clay samples, G áeotech-nique, 3, 20–29.

4) Kusakabe, O., Maeda, Y., Ohuchi, M. and Hagiwara, T. (1991):Strength-deformation characteristics of an undisturbed scoria andeŠects of sample disturbance, Proc. Japan Society of Civil Eng.,(439/III–17), 69–77 (in Japanese).

5) Marachi, N. D., Chan, C. K., Seed, H. B. and Duncan, J. M.(1969): Strength and deformation characteristics of rockˆll materi-als, Dept., of Civil Eng., University of California, Berkeley, ReportNo. TE–69–5.

6) Marsal, R. J. (1973): Mechanical properties of rockˆll, In Embank-ment Dam Engineering, Casagrande Volume, 109–199.

7) Nakata, Y., Hyodo, M., Hyde, A. F. L., Kato, Y. and Murata, H.(2001): Microscopic particle crushing of sand subjected to highpressure on one-dimensional compression, Soils and Foundations,41(1), 69–82.

8) Rowe, P. W. (1969): The relation between the shear strength ofsands in triaxial compression, plane strain and direct shear,G áeotechnique, 19(1), 75–86.

9) Tatsuoka, F., Sakamoto, M., Kawamura, T. and Fukushima, S.(1986a): Strength and deformation characteristics of sand in planestrain compression at extremely low pressures, Soils and Founda-tions, 26(1), 65–84.

10) Tatsuoka, F., Goto, S. and Sakamoto, M. (1986b): EŠects of somefactors on strength and deformation characteristics of sand at lowpressures, Soils and Foundations, 26(1), 105–114.

11) Tatsuoka, F, Teachavorasinskun, S., Dong, J., Kohata, Y. andSato, T. (1994): Importance of measuring local strains in cyclictiraxial tests on granular materials, Proc. ASTM Symposium Dy-namic Geotechnical Testing II, ASTM, STP 1213, 288–302.

12) Technical Committee of Deformation and strength characteristicsof granular material, (1986): Deformation and Strength Character-istics of Granular Material, JGS (in Japanese).

13) Tsuya, H., Machida, H. and Shimozuru, D. (1988): Explanatorynote for geologic map of Mt. Fuji, Second Printing, GeologicalSurvey of Japan.

Documents

STRENGTH AND DEFORMATION CHARACTERISTICS OF SCORIA …