Can. Geotech. J.36: 88–101 (1999) © 1999 NRC Canada

88

The behavior of a compressible silty fine sand

An-Bin Huang, Huai-Houh Hsu, and Jia-Wei Chang

Abstract: Publications associated with sands are often limited to clean (i.e., little fines content), uniform, uncementedsilica or quartz sand. On the other hand, the importance of mineral content, particle shapes, as well as gradation to thebehavior of sand has long been recognized. Although systematic studies of sands other than clean quartz sand havebeen limited, there is increasing attention being paid to sands with an appreciable fines content. Because of a majorconstruction project, extensive field and laboratory experiments were performed on a silty fine sand from Mai-Liao,which is located on the central west coast of Taiwan. Results show that Mai-Liao Sand (MLS), a silty sand, can besignificantly more compressible than clean quartz sand under static load. The particles of MLS have moderate strength,and significant crushing can be induced by triaxial shearing. As a result, MLS has low dilatancy and a relatively smallrange of peak friction angles. Cone penetration tests in MLS were conducted in a calibration chamber. Analyses of thedata indicate that interpreting cone tip resistance in MLS using methods developed based on clean quartz sand withoutconsidering the differences of compressibility can be unrealistic. This paper documents results of the experimentalstudies on MLS.

Key words: silty fine sand, strength, dilatancy, compressibility, crushing, in situ test.

Résumé: Les publications associées aux sables se limitent souvent aux sables quartziques ou siliceux propres (peu defines), uniformes et non cimentés. Par ailleurs on a reconnu depuis longtemps que le comportement de ces matériauxétait fortement tributaire de la partie minérale, de la forme des particules ainsi que de la répartition granulométrique.Bien qu’il n’existe qu’un petit nombre d’études systématiques sur les sables autres que les sables quartziques propres,ceux qui contiennent un part appréciable de fines retiennent de plus en plus l’attention. A l’occasion d’un importantprojet de construction, on a effectué des essais complets, en place et au laboratoire, sur un sable fin silteux de Mai-Liao, sur la côte centre-ouest de Taiwan. Les résultats indiquent que le sable de Mai-Liao (MLS) peut être beaucoupplus compressible, sous charge statique, qu’un sable quartzique propre. Les particules de MLS ont une résistancemodérée et un cisaillement triaxial peut provoquer un broyage significatif des grains. Ainsi le MLS présente-t-il unedilatance faible et un intervalle relativement étroit d’angles de frottement au pic. Des essais de pénétration au cône ontété effectués sur le MLS dans une chambre de calibration. L’analyse des données indique qu’il peut être irréalisted’interpréter la résistance de pointe selon des méthodes développées pour des sables propres, sans considérer lesdifférences de compressibilité. Cet article présente les résultats des essais expérimentaux sur le MLS.

Mots clés: sable fin silteux, résistance, dilatance, compressibilité, broyage, essai in situ.

[Traduit par la Rédaction] Huang et al. 101

IntroductionPublications associated with sands are often limited to

clean (i.e., little fines content), uniform, uncemented silicaor quartz sand. The sands generally have a low compress-ibility under normal, static loading conditions. The relation-ships among confining stress, void ratio, strength, anddilatancy are expected to follow a certain pattern. There areempirical rules or analytical models developed either for testresult interpretation or for design of foundations for suchsand. On the other hand, the importance of mineral content,particle shapes, as well as gradation to the behavior of sandhas long been recognized (Koerner 1970a, 1970b; Joustraand de Gijt 1982; Ishihara 1993). Although systematic stud-ies of sands other than clean quartz sand have been limited,increasing attention is being paid to sands with an apprecia-ble amount of fines. The silty fine sand referred to in this

paper can have significant amounts of fines; it constitutesmost of the natural sand deposits along the west coast ofTaiwan. The sand deposit has depths typically in the rangeof several hundred metres.

This paper compiles available data on the silty fine sandfrom a test site in Mai-Liao, which is located on the centralwest coast of Taiwan. Extensive soil borings and in situ testshave been performed on Mai-Liao Sand (MLS) for variouspurposes related to the project. In addition to data collectedfor engineering purposes, a series of research-quality laboratoryexperiments have been performed to establish a comprehen-sive database. Chamber calibration tests of cone penetrationtests (CPT) were performed to establish an interpretationmethod for CPT in MLS. The data were analyzed and orga-nized using some of the existing frameworks to describe thecharacteristics of this silty sand and to propose a method fordetermining geotechnical engineering parameters using CPT.

Test site

This paper concentrates on data collected at a test sitewhere a total of 50 borings were conducted. Figure 1 shows

Received November 7, 1997. Accepted October 7, 1998.

A.-B. Huang, H.-H. Hsu, and J.-W. Chang.Department ofCivil Engineering, National Chiao-Tung University, Hsin-Chu,Taiwan.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

the general location and dimensions of the test site (markedABCD) and its relationship with a petrochemical construc-tion project area. Reclaimed land in excess of 2400 ha thatextends the original shoreline further into the ocean is underconstruction in this area. The original ground surface was ata maximum of 2–3 m below sea level. Major constructionactivities included dredging sand offshore and transportingand placing it onshore as a hydraulic fill. The placement ofthe hydraulic fill would raise the grade to approximately 3 mabove sea level. The hydraulic fill along with its underlyingnatural deposits are subsequently densified using dynamiccompaction. The in situ tests in this area included standardpenetration tests (SPT), CPT, and down-hole (DH) seismicshear wave velocity measurements. Locations of these bore-holes are shown in Fig. 1. The borehole number is shown inFig. 1 only when data retrieved there are utilized later in thepaper.

Geological origin and physical propertiesof MLS

MLS came from the central mountain range that lies lon-gitudinally on the east side of Taiwan. Sedimentary andmetamorphic rocks were deteriorated and flushed by rain-falls through steep slopes and rapidly flowing streams beforesettling down on the west plain. The process of transporta-tion ground the fractured rock into sand and silt particles.Figure 2 compiles the blow counts per 30 cm of penetration,

N from SPT, cone tip resistance,qc, from CPT, and fines(particles pass No. 200 sieve) content profiles from 13 of theborings shown in Fig. 1. The shear wave velocities shown inFig. 2 were collected from seven test locations. The fines ofMLS are mostly nonplastic; the in situ fines contents variedfrom less than 10 to over 90%. The soil deposit is generallyclassified as SM or ML, depending on the amount of finescontent. The ML material is mostly found below a depth of10 m.

A batch of 20 tons of MLS was taken from within a depthof 2 m at the test site to provide specimens for laboratoryconsolidation, triaxial, and CPT calibration chamber tests.The sand was sieved to filter out leaves or gravel-sized par-ticles and then kiln-dried. The drying process also serves tomix and unify the gradation of sand taken from the field.MLS has a dark gray color. Figure 3 shows grain size distri-bution curves from four randomly selected specimens of thekiln-dried MLS. Apparently, the drying and mixing processhas effectively unified the sand gradation. Essentially allparticles pass a No. 60 sieve; the fines contents of the fourkiln-dried specimens have an average value of 15.1%. Forcomparison purposes, grain size distribution curves from fieldsamples obtained in various boreholes are also included inFig. 3 (broken curves). The 15% fines content is more repre-sentative for MLS within top 10 m in the field. The kiln-dried MLS has a maximum dry unit weight (γd max) of16.6 kN/m3 and a minimum dry unit weight (γd min) of12.8 kN/m3. The specific gravity of MLS has an averagevalue of 2.69. X-ray diffraction analysis on MLS showedsignificant contents of muscovite and chlorite in addition toquartz. The grain shapes of particles retained on a No. 200sieve are subangular to angular and flaky, as indicated by thescanning electronic micrographs shown in Fig. 4.

Field characterization of MLS

According to the SPT and CPT profiles shown in Fig. 2,the soil conditions throughout the test site are fairly uniformin the lateral direction. For a given depth, theqc andN val-ues are confined to a relatively narrow range. Theqc, maxi-mum shear modulus,Gmax, from DH shear wave velocitymeasurements, andN values from SPT were collected atcomparable depths from various boreholes to facilitate com-parisons with previously reported relationships. The ratios ofqc, normalized with respect to atmospheric pressurePa, overN as shown in Fig. 5 are erratic but indicate a trend of aslightly negative relationship betweenqc/PaN and fines con-tent, as suggested by Kulhawy and Mayne (1991). The fieldSPT was conducted with various types of hammers, withoutenergy calibration. A significant part of that erratic correla-tion may be due to the variations of test equipment and (or)energy efficiency involved in the available SPT data. ForMLS and the available data, the correlation betweenN andqc is not strong enough to justify the application of the em-pirical equation suggested by Kulhawy and Mayne (1991).

Figure 6 shows a correlation betweenGmax andqc for theMLS and that proposed by Rix and Stokoe (1991) for quartzsands. The majority of the MLS data points fall below theRix and Stokoe correlation, indicating a lower stiffness(lower Gmax for the sameqc), lower peak strength of MLS(lower qc for the same vertical stressσv′), or both.

© 1999 NRC Canada

Huang et al. 89

TaiwanStrait

ConstructionSite

Land

Mai-Liao

N

Taiwan

0

00

CD

BA

1

12

0

0

0

0

m

kmkm

: SPT : SPT+CPT : SPT+DH : SPT+CPT+DHB

A

D

CBorehole No.

4

10

25 26 27

29 32

36

3840 42

48

1

1121 23

41

47

5

Fig. 1. General layout of the Mai-Liao project site and soilboring locations.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

© 1999 NRC Canada

90 Can. Geotech. J. Vol. 36, 1999

Compressibility of MLS

Compressibility of sand under static conditions is usuallylow, at least for quartz sand, except when the stress is ex-tremely high. Under high-stress conditions, the compressionof sand is mainly a result of particle crushing. Extensivestudies on compression and crushing of granular materialsunder elevated stresses have been reported (Lee andFarhoomand 1967; Vesic and Clough 1968; El-Sohby andAndrawes 1972; Hardin 1985; Been et al. 1991; Lade et al.1996; Yamamuro et al. 1996; Konrad 1998). The compress-ibility of MLS was measured isotropically in a triaxial celland one-dimensionally in an oedometer using the kiln-driedmaterial.

The isotropic consolidation tests were performed on spec-imens, 70 mm in diameter and 150 mm high, prepared inmembrane-lined split mould mounted on the lower platen ofthe triaxial cell. Dry sand was first placed in the mould us-ing a spoon and then statically pressed to achieve the desiredvoid ratio in four equal layers. Because of the high com-pressibility of MLS, to be described later, static pressing wasrather effective in densifying the sand specimen withoutcausing particle segregation. To verify the uniformity of gra-dation, a specimen was prepared following this procedureand then saturated in the triaxial cell under an effective con-fining stress of 30 kPa. The confining stress was then re-leased and the specimen removed from the triaxial cell. Thespecimen was sliced into nine equal layers. Table 1 showsthe fines content measured for each of the nine layers. Theyvaried from 14.6 to 16.2% with an average of 15.3%, whichis 0.2% higher than the average of those shown in Fig. 3. Nosign of concentration of fines was noticed.

The specimen was saturated under a back-pressure of98.1 kPa prior to the doubly drained isotropic consolidation

test in the triaxial cell. A burette was connected to the drain-age lines of the triaxial specimen. The water level in the bu-rette was monitored by a differential pressure transducer.The amount of porewater drainage in and out of the speci-men during the isotropic consolidation was monitored bymeasuring the porewater level fluctuation in the burette. Fig-ure 7 shows a set of isotropic consolidation curves in termsof void ratio versus effective mean normal stress (e – log p′).The maximum void ratio (emax = 1.04) and the minimumvoid ratio (emin = 0.57) of MLS are indicated in Fig. 7 toprovide a reference of relative density,Dr, for the consolida-tion curves. The step-loading consolidation tests had a maxi-mum consolidation stress of 588 kPa.

The primary compression index taken as the slope of thee –log p′ curve near the maximum consolidation stress is ap-proximately 0.08 ± 0.01. The swelling index or slope of therebound part of thee – log p′ curve is 0.013 ± 0.002. Thesecompression indices are unusually high for a sand undersuch mild stress conditions. The consolidation stress wasmaintained for 60 min for each load increment to facilitateobservations of secondary consolidation. The time settle-ment record provided a set of void ratio – time (e – log t)curves to determine the secondary compression index. Theratios of primary over the secondary compression indices(equivalent toCα /Cc in one-dimensional compression) takenbeyond the apparent preconsolidation stress varied from 0.02to 0.025, which is within the range of sand and sandy silt inone-dimensional compression, as reported by Mesri et al.(1990). The whole specimen was used for a sieve analysisafter the consolidation test. The change in fines contents waswithin 1% compared with the average fines content of15.1% shown in Fig. 3. The difference is likely to be withinthe accuracy of the experiments, indicating that the volumechange was mainly a result of rearrangement of sand parti-

0 10 20 30 40 50SPT - N

0

10

20

30

40

Dep

th,m

0 5 10 15CPT -qc, MPa

0 20 40 60 80 100Fines content, %

0 100 200 300 400

Shear wave velocity, m/s

Fig. 2. Profiles of in situ tests.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

cles and crushing was minimal.In order to facilitate further comparisons with quartz sand

in compressibility and crushing, one-dimensional consolida-tion tests under elevated stress conditions were conducted onMLS. A high-capacity, strain-controlled load frame origi-nally designed for creep tests on rock specimens was usedfor the consolidation tests. The sand specimen was preparedin a thick, oil-impregnated bronze ring 70 mm in diameterand 25 mm high. The specimen was inundated in water as in

the case of a step-loading consolidation test. The verticaldeformation of the specimen was monitored using a lineardisplacement transducer and the vertical stress measuredwith a load cell. A strain rate of 0.065%/min was used inthese tests. This strain rate is approximately 10 times that ofthe isotropic step-loading consolidation tests describedabove. Figure 8 showse versus vertical consolidation stress,σv′, from the one-dimensional consolidation tests with threedifferent initial void ratios. The specimens were consoli-dated to a maximum consolidation stress of 10 MPa. A com-parison with quartz sand (see Fig. 8) under similar loadingconditions shows that MLS is at least five times as com-pressible as quartz sand, depending on the stress level anddensity. Results show a further increase of compressibilitybeyond the maximum stress level applied in the isotropicconsolidation (i.e., 588 kPa). The one-dimensional compres-sion index according to thee – log σv′ curves atσv′ valuesbetween 3 and 10 MPa, where three curves essentially

© 1999 NRC Canada

Huang et al. 91

Fig. 4. Scanning electronic micrographs of MLS. (a) Particlesretained on a No. 200 sieve; (b) particles passing a No. 200 sieve.

(a)

0.0010.0100.1001.00010.000Sieve opening, mm

0

10

20

30

40

50

60

70

80

90

100

Fin

es b

yw

eigh

t,%

Fin

es b

yw

eigh

t,%

0

10

20

30

40

50

60

70

80

90

100

MEDIUMSAND

COARSESAND

GRA-VEL FINE SAND SILT SIZE CLAY SIZE

Field (0–10m)

Kiln-dried

0.0010.0100.1001.00010.000Sieve opening, mm

0

10

20

30

40

50

60

70

80

90

100

0

10

20

30

40

50

60

70

80

90

100

MEDIUMSAND

COARSESAND

GRA-VEL FINE SAND SILT SIZE CLAY SIZE

Field (10–40m)

Kiln-dried

(b)

Fig. 3. Grain size distribution of MLS.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

merge together, is approximately 0.29. This compression in-dex value is comparable with that of a lean clay. The wholespecimen was used in the measurement of fines content fol-lowing the consolidation test. All three specimens showedan increase of fines content of approximately 3% after thetest. Judging from the moderate increase of fines content, itappears that the compression of MLS under one-dimensionalloading is again mostly a result of rearrangement of soil par-ticles.

Hardin (1985) indicated that most crushing occurs to par-ticles with diameters larger than silt size (D ≥ 0.074 mm).

© 1999 NRC Canada

92 Can. Geotech. J. Vol. 36, 1999

10 100 1000log p', kPa

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05

e

Initial Dr, %

50

60

70

85

emin

emax

Fig. 7. e – log p′ curves from isotropic consolidation tests.

10 100 1000 10000 100000logσ'v, kPa

0.5

0.6

0.7

0.8

0.9

1.0

e Yamamuro et al. ( 1996 )

MLSInitial Dr = 24 %

Initial Dr = 38 %

Initial Dr = 62 %Quartz sand

Loose

Medium

Fig. 8. e – log σv ′ curves from one-dimensional consolidationtests.

Layer No. Height (mm) Fines content (%)

1 0–17 15.32 17–34 15.73 34–51 16.24 51–68 15.45 68–86 14.66 86–103 14.77 103–120 15.38 120–137 15.49 137–154 15.4

Table 1. Vertical distributions of fines content of the triaxialspecimen.

100 1000qc / (σ'v)

1/2

10

30

50

0

20

40

60

Gm

ax/q

c Rix and Stokoe ( 1991 )

2000

4 25 29 38 48

5 26 32 40

10 27 36 42

Borehole No.

Fig. 6. Correlation betweenGmax and qc of MLS.

10 30 50 70 900 20 40 60 80 100Fines content, %

1

3

5

7

9

0

2

4

6

8

10

q c/P

aN

Kulhawy and Mayne (1991)

4 25 29 38 48

5 26 32 40

10 27 36 42

Borehole No.

Fig. 5. Variation of qc/PaN with fines content.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

Following this concept, a breakage potentialBp can be com-puted as the area between the line defining the upper limit ofsilt size,D = 0.074 mm, and the part of the particle size dis-tribution curve for whichD > 0.074 mm. According to Hardin’s(1985) definition and the grain size distribution curves shownin Fig. 3, the MLS has aBp of approximately 0.21. ThisBpis less than the lowest value reported by Hardin (1985) forvarious sands. The lowBp values are consistent with thesmall amounts of crushing observed in consolidation tests.

Shear strength, dilatancy, and crushing ofMLS

A series of isotropically consolidated drained (CID) triaxialtests with volume change measurements were performed onthe kiln-dried MLS. The main objective of these tests is toevaluate the strength and dilatancy characteristics of MLS. Atotal of 20 CID triaxial tests were performed with combina-tions of four levels of initial relative densities and five levelsof effective confining stress. Variables applied in this seriesof triaxial tests are summarized in Table 2.

The triaxial specimens were prepared following the sameprocedure as previously described for isotropic consolidationtests. Upon preparation, the triaxial specimen was flushedwith carbon dioxide and then deaired water. A back-pressureof 98.1 kPa was used in all tests to assure saturation and thespecimens had a minimumB value of 0.97. The variation ofvoid ratio during specimen consolidation was measured fol-lowing the same procedure as used in isotropic consolidationtests. The change of void ratio resulting from consolidationwas deducted in the analysis. Triaxial shearing was appliedthrough strain-controlled axial compression at a rate of0.1%/min. No lubrication was applied to the end platens, asthe specimen height over diameter ratio exceeded 2; conse-quently, the effect of end friction was expected to be mini-mal (Bishop and Green 1965). Because of the fine nature ofthe sand particles, membrane penetration was not expectedto be significant. Figure 9 shows a set of the CID triaxialtests on specimens with initialDr of 50%. The data are pre-sented in terms of deviator stress (σ1′ – σ3′) and volumetricstrain (εv) versus axial strain (εa).

The concepts of relative dilatancy index,IR (Bolton1986), and state (Been and Jefferies 1985) are used to evalu-ate the triaxial test data as they relate to strength anddilatancy. Bolton (1986) suggested that the drained peakfriction angle, ′φmax, should be obtained by dropping a tan-gent from the origin to a single Mohr circle (i.e., a secantfriction angle) that represents the maximum deviator stressconditions from a triaxial test. The secant drained frictionangle under the critical state (i.e., a condition where shearingcontinues without volume change),′φcrit , is mainly a functionof mineral content of the sand. The′φcrit of MLS measured

from triaxial tests on loose specimens under critical statevaried from 30.8 to 32.4° with an average value of 31.6°.The ′φcrit is comparable with the lower end of a quartz sand.This is apparently due to the contents of softer minerals suchas muscovite and chlorite.

Figure 10 shows a plot of′φmax – ′φcrit versus ′pp fromtriaxial tests on MLS in which ′pp is the mean effective stressat peak deviator stress. According to Bolton (1986),IR com-bines the effects of sand density and confining stress; empir-ically:

[1] ID

Qp

PR

r p

a

= −′

−100

1001ln

where Q is an empirical constant that increases with thecrushing strength of sand grains andPa is the atmosphericpressure. For quartz or feldspar,Q = 10, and for chalk,Q =5.5. IR controls ′φmax – ′φcrit and the maximum rate of dilation(dεv/dεa)max as

[2] ′ − ′ =φ φmax crit Ro3I

[3] −

=d

dv

a max

Rεε

03. I

© 1999 NRC Canada

Huang et al. 93

250

750

1250

0

500

1000

1500

σ' 1- σ

' 3,k

Pa

σ'c= 588.4 kPa

392.3 kPa

245.2 kPa

147.1 kPa

49.0 kPa

Initial Dr = 50 %

-8

-6

-4

-2

2

4

6

0ε v,%

5 15 250 10 20 30εa, %

49.0 kPa

147.1 kPa

245.2 kPa

392.3 kPa

588.4 kPa

Fig. 9. CID triaxial test results with an initialDr of 50%.

Initial voidratio, e0

Initial Dr

(%)Effective confining stress,

σc ′ (kPa)

0.80 50 49.0 147.1 245.2 392.3 588.40.76 60 49.0 147.1 245.2 392.3 588.40.71 70 49.0 147.1 245.2 392.3 588.40.64 85 49.0 147.1 245.2 392.3 588.4

Table 2. Variables applied in the MLS triaxial tests.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

The range of ′φmax – ′φcrit shown in Fig. 10 is at the lowerend of similar values for quartz sands reported by Bolton(1986). The maximum rates of dilation (dεv/dεa)max of MLSin triaxial tests are significantly less than those for quartzsand reported by Bolton (1986) under comparable′pp, asshown in Fig. 11. If eq. [2] is used to fit the available datapoints of MLS,Q should fall in the range between 7 and 9.The Q values correspond to those of limestone or anthraciteaccording to Bolton (1986), indicating a mildly crushablenature of the MLS grains. No singleQ value can provide areasonable fit to the data points of′φmax – ′φcrit (Fig. 10) and(dεv/dεa)max (Fig. 11) simultaneously. Sieve analyses per-formed on MLSspecimens within a thickness of 1 cm alongthe shear plane after the triaxial tests showed an increase infines contents from 2 to 8% as′pp varied from 50 to 800 kPa.The initial relative density of the triaxial specimen does notappear to have a significant effect on the increase of finescontent induced by triaxial shearing. The low dilatancy andshearing-induced fines content increase indicate that′φmax –′φcrit for MLS is likely a result of shearing and crushing of

sand particles as well as dilation caused by particles overrid-ing one another.

Figure 12 shows a comparison of critical state lines be-tween MLS and Ticino sand, a clean, uniformly gradedquartz sand. The combination of effective mean normalstress (p′) and void ratio towards the end of triaxial testsshown in Fig. 9 (i.e., initialDr of 50%) was used to establishthe critical state line for MLS. The selection of critical stateis approximate, as volume change continues in some of thetriaxial tests, even at 25% of axial strain. The critical stateline for Ticino sand is according to results reported byKonrad (1998). Been et al. (1991) have indicated that thecritical state line is generally bilinear. Depending on the soiltype, grain crushing occurs during shear when′p reaches 1–2 MPa according to earlier reports (Konrad 1998). The slopeof the critical state line beyond crushing stress (λc) is muchsteeper due to changes in gradation and grain shape as a re-

sult of grain crushing. For MLS, the bilinear nature of thecritical state line is not as clearly defined as for Ticino sand.The increase in fines after triaxial shearing, as describedabove, is an indication that crushing of sand grains orchange of the slope of the critical state line occurs in MLSat a much lower stress.

Laboratory CPT calibration in the siltysand

There have been reports (McNeilan and Bugno 1984) in-dicating that cone penetration in silt or silty sand may not bea fully drained test. Due to positive excess pore pressure in-duced by the cone penetration, the cone tip resistance,qc, insilty material may be lower as compared with tests in a cleansand with similar relative density. Dilatancy is another factorbelieved to have strong influence onqc values (Yu andHoulsby 1991; Huang and Ma 1994; Salgado et al. 1997).The presence of fines and low dilatancy of MLS raised aquestion as to the validity of interpreting CPT using methodsdeveloped for clean, uniform quartz sand. A series of CPTchamber calibration tests were performed to address theseissues.

Calibration chamber system and testing programFigure 13 shows a schematic view of the calibration

chamber. It was designed to house a 525 mm diameter and760–815 mm high specimen. Chamber systems with similardimensions have been reported (Bellotti and Pedroni 1991;Huang et al. 1991) and applied by others (Voyiadjis et al.1993). The design makes a compromise between practicalityin operation and the idea of having a large diameter ratio be-tween chamber specimen and the cone. The relatively smallchamber size was utilized for ease of saturation, back-pressuring, and handling. The specimen diameter is 15 timesthat of a standard cone penetrometer (i.e., 35.6 mm). Appar-ently, CPT performed in a chamber with this diameter ratiowill be subject to boundary effects (Parkin 1988) for clean

© 1999 NRC Canada

94 Can. Geotech. J. Vol. 36, 1999

0 20 40 60 80 100Dr, %

0.0

0.2

0.4

0.6

0.8

1.0

(dε v

/dε a

) max

Quartz sandp'p = 300 kPa( Q = 10 )

MLSp'p = 300 kPa( Q = 7 )

p'p , kPa

250

300

400

500

Fig. 11. Dilatancy of MLS during triaxial tests.

10 100 1000 10000p'p , kPa

0

2

4

6

8

10

12

φ' max

- φ' cr

it,d

eg.

Initial Dr, %

50

60

70

85

Dr = 85 %( Q = 9 )

Dr = 50 %(Q = 9 )

Fig. 10. ′φmax – ′φcrit versus ′pp for MLS.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

quartz sand. The chamber was designed to provide constant-stress lateral boundary conditions only.

The piston at the bottom of the specimen controls verticalstress. A sheet of 6.4 mm thick porous plastic is attached tothe end plates to facilitate double drainage for the specimen.A total of six open-ended piezometers are installed on thebottom plate, as shown in Fig. 13. Two of the piezometersare located at a distance of dL = 25.4 mm from the center ofthe specimen, three at dL = 50.8 mm, and one at dL =101.6 mm. The piezometers are made of 1.65 mm outsidediameter, 0.23 mm wall stainless steel tubing filtered with apiece of nonwoven geotextile. The piezometer tips are at ap-proximately 260 mm above the bottom of the specimen.Each piezometer is connected to a pressure transducer, asshown in Fig. 13. The voids within the piezometers andpressure transducer ports were filled with deaired glycerin.During saturation of the specimen, the piezocone is insertedin a sealed container at the top of the specimen. The con-tainer, made of delrin, also serves as a bushing.

Specimen preparation and testing procedureThe chamber specimen is prepared in a membrane-lined

split mould, similar to the concept used in triaxial tests. Thekiln-dried MLS was placed in a 500-kg bag initially, whichin turn was lifted by a crane. A load cell was attached be-tween the bag and the crane hook to monitor the weight ofthe bag. The specimen was created in four equal layers. Thesand was slowly released from the bottom of the bag fromimmediately above the surface of deposition. This processminimizes the height of deposition, and thus the chance ofsegregation. The initial relative density upon deposition wasapproximately 50%. A 30 mm diameter and 500 mm longsteel rod was inserted in the sand and rocked sideways todensify the sand. The locations of steel rod insertion wereevenly distributed. Since the weight of each specimen layerwas known, the density was monitored by the height of thespecimen. The densification stopped when the desired speci-

men height was reached. This sand placement anddensification process was repeated four times to create aspecimen. Upon completion of sand deposition, a vacuumwas applied to the sealed specimen and the membrane jacketremoved. The vacuum was connected to the top of thepiezocone container, the highest position of the chamber as-semblage to be saturated. The chamber cell was then assem-bled and 30 kPa confining stress applied to the specimen.

The piezocone tip was placed in a sealed container filledwith glycerin and deaired under vacuum prior to installationin the chamber. The chamber and piezocone assemblage wasbolted onto a hydraulic frame where cone penetration wouldtake place.

The saturation process started by infiltrating CO2 slowlyfrom the bottom of the specimen under vacuum. The CO2circulation continued for approximately 30 min. Deaired wa-ter was then allowed to permeate from the bottom of speci-men under vacuum. The permeation took approximately 3 h,until the piezocone container was flooded with water. A

© 1999 NRC Canada

Huang et al. 95

1500 mm

piezometer

specimen

top plate

bottom plate

pressure transducer

pistonring piston

wall

pistonshaft

linear ballbearing

membrane

tierod

cellwall

drainageport

conecontainer

cone

Fig. 13. Schematic view of the calibration chamber.

10 100 1000 10000log p', kPa

0.5

0.7

0.9

0.6

0.8

1.0

e

Critical State Line

MLS

Ticino (Konrad 1998)

CPT initial state

Fig. 12. Critical state lines and CPT initial states.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

back-pressure and the desired confining stress were then ap-plied. The status of saturation was monitored through the sixpiezometers installed in the specimen. In the six saturatedchamber specimens shown in Table 3, the averageB param-eter from the piezometer readings after 2 h of saturationranged from 0.95 to 0.98. Considering the compressibility ofMLS, theseB values should correspond to a degree of satu-ration in excess of 99.8% according to Black and Lee(1973).

Variables applied in the calibration tests include initialDrof the specimen, effective vertical stressσv′, and the ratio ofeffective horizontal stressσh′/σv′ (K). Table 3 summarizesthese applied variables. All tests used stress-controlled verti-cal and horizontal boundary conditions. Dynamic compac-tion was an important part of the construction activities in

Mai-Liao. The upperK values shown in Table 3 were se-lected to reflect a moderately to highly overconsolidatedstress history due to dynamic compaction.

CPT chamber test resultsThere were a total of 40 chamber CPT performed in this

series of experiments, as shown in Table 3. Six of them wereperformed in a saturated specimen using a standard(35.6 mm diameter) piezocone to evaluate the developmentof excess pore pressure during cone penetration. The porouselement was located immediately behind the cone tip. Theremaining CPT were conducted in a dry specimen using ahalf-size (17.8 mm diameter) cone. Figure 14 depicts theqc,friction ratio FR, and excess pore pressure∆u obtained fromTest 4. The corresponding piezometer readings obtained dur-

© 1999 NRC Canada

96 Can. Geotech. J. Vol. 36, 1999

TestNo.

Initial Dr

(%)Type ofspecimen K

σv ′(kPa)

p′(kPa)

Back-pressure(kPa)

qc

(MPa)

1 70 Saturated 1 49.0 49.0 98.1 5.992 70 Saturated 1 49.0 49.0 343.4 5.533 70 Saturated 1 98.1 98.1 98.1 8.414 70 Saturated 1 147.1 147.1 294.2 10.165 70 Saturated 1 196.1 196.1 294.2 11.506 70 Saturated 1 294.2 294.2 490.3 14.897 70 Dry 1 294.2 294.2 — 14.448 70 Dry 0.5 98.1 65.4 — 6.299 70 Dry 2 98.1 163.4 — 12.70

10 70 Dry 3 98.1 228.8 — 14.2611 70 Dry 3 147.1 343.2 — 19.6712 70 Dry 2 196.1 326.9 — 16.3813 70 Dry 3 196.1 457.6 — 20.8514 70 Dry 0.5 294.2 196.2 — 9.1815 70 Dry 2 294.2 490.3 — 22.1416 70 Dry 3 294.2 686.5 — 30.4417 70 Dry 2 392.3 653.8 — 25.9418 70 Dry 0.5 490.3 326.9 — 19.6119 70 Dry 1 490.3 490.3 — 25.5020 85 Dry 0.5 98.1 65.4 — 9.1821 85 Dry 1 98.1 98.1 — 9.3522 85 Dry 2 98.1 163.4 — 12.2023 85 Dry 3 98.1 228.8 — 14.8324 85 Dry 1 196.1 196.1 — 13.1925 85 Dry 2 196.1 326.9 — 18.1826 85 Dry 0.5 294.2 196.2 — 14.0727 85 Dry 1 294.2 294.2 — 16.1228 85 Dry 2 294.2 490.3 — 28.8929 85 Dry 0.5 490.3 326.9 — 21.7430 85 Dry 1 490.3 490.3 — 25.5031 50 Dry 0.5 98.1 65.4 — 3.2232 50 Dry 1 98.1 98.1 — 5.5833 50 Dry 2 98.1 163.4 — 9.2634 50 Dry 1 196.1 196.1 — 8.7735 50 Dry 2 196.1 326.9 — 14.6636 50 Dry 0.5 294.2 196.2 — 9.0337 50 Dry 1 294.2 294.2 — 11.3438 50 Dry 2 294.2 490.3 — 18.6139 50 Dry 0.5 490.3 326.9 — 15.0740 50 Dry 1 490.3 490.3 — 18.62

Table 3. Variables applied in the calibration chamber tests andqc results.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

© 1999 NRC Canada

Huang et al. 97

ing cone penetration are shown in Fig. 15. CPT is known forits capability of sensing thin layers of soils with differentproperties. The relatively smoothqc profiles are an indica-tion of reasonable uniformity of the specimens. The excesspore pressure obtained in the piezometers during Test 4 hada maximum value of approximately 50 kPa. The excess porepressure measurement on the cone was negative, after a sta-ble qc was reached (i.e., beyond a depth of 200 mm). Thereis no apparent relationship between the excess pore pressureand distance between the pizometer and center of the speci-

men, as shown in Fig. 15. In Test 4, the cone penetrationwas interrupted at 300 mm and the excess pore pressure wasallowed to dissipate before further penetration. The full dis-sipation of excess pore pressure took approximately 25 s.According to McNeilan and Bugno (1984), theqc values im-mediately after restart of the penetration can be at least 20%higher than that just prior to the interruption for CPT in siltor sandy silt. The discrepancy inqc is a result of high porepressure development during cone penetration in silt. For thetest shown in Fig. 14, the peakqc after restart was11.55 MPa, which is 4% higher than theqc recorded justprior to the interruption of penetration, and much lower thanthose reported by McNeilan and Bugno (1984).

The major differences between Tests 6 and 7 (see Table 3)are the cone diameter and saturation of the chamber speci-men. The diameter ratio for the half-size cone used in Test 7is 30, which is twice that of Test 6. If the effect of excesspore pressure development or change of diameter ratio issignificant, theqc of the half-size cone (i.e., Test 7) in a dryspecimen should be significantly larger than that of the stan-dard cone (i.e., Test 6) in a saturated specimen. Figure 16shows theqc profiles from these two tests. The averageqctaken at depths between 200 and 500 mm, where they havereached a plateau, agree within 4%. Observation in Tests 4,6, and 7 indicate that CPT in MLS is essentially a drainedtest. The following analysis of CPT data makes no distinc-tion between CPT in dry and saturated specimens. Also, be-cause the excess pore pressure development was minimal,no pore pressure correction was made to convertqc to qT.

The insensitivity ofqc to the change in diameter ratio,comparable with CPT calibration tests in very loose uniformquartz sand (Parkin 1988), is a direct reflection of the com-pressible (or lack of dilatancy) nature of MLS. An importantadditional implication is that the chamber specimen is largeenough even for a standard cone penetrometer and thereshould be insignificant boundary effects. The use of a half-size cone is an additional assurance of minimal boundaryeffects.

0 5 10 15qc, MPa

0.0 0.2 0.4 0.6 0.8 1.0FR, %

100

300

500

700

0

200

400

600

800

-100 -50 0 50 100∆u, kPa

Initial Dr = 70 %,σ'v = 147.1 kPa ,σ'h = 147.1 kPa , Back-pressure = 294.2 kPa

Dep

thof

pene

tratio

n,m

m

Fig. 14. CPT results of Test 4.

-20 0 20 40 60 80∆u, kPa

100

300

500

700

0

200

400

600

800

Dep

thof

pene

tratio

n,m

m

Initial Dr = 70 %σ'v = 147.1 kPa ,σ'h = 147.1 kPa , Back-pressure = 294.2 kPa

dL, mm

25.4

25.4

50.8

50.8

50.8

101.6

Fig. 15. Excess pore pressure measurements from piezometers inTest 4.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

Figure 17 shows the averagedqc values and their relation-ship with initial σh′ in the specimen. The averageqc is deter-mined based on readings taken in calibration tests at depthsfrom 200 to 500 mm whereqc has reached more or less asteady value. Theqc in MLS has a clear and positive rela-tionship with σh′. The influence ofσh′ appears to be muchstronger than the initialDr. The relationship betweenqc andσv′ is not nearly as obvious as shown in Fig. 18. For low-compressibility sand, there have been conflicting reports onthe relationship betweenqc andσh′ (Houlsby and Hitchman1988; Huang and Ma 1994). In practice at least,σh′ in sandis rarely determined based onqc, as the noise ofqc may bemore significant than the effects ofσh′. The trend shown inFig. 17 is an indication that for a high-compressibility sandsuch as MLS, there is the potential to determineσh′ usingqc.

Comparisons with CPT in clean, uniformlygraded quartz sand

Because of the high compressibility of MLS, the changeof void ratio orDr after the specimen is subjected to a con-fining stress is significant. Monitoring the change of void ra-tio was difficult for the chamber specimens, especially whenthey were dry. Four sets of compression tests on saturatedspecimens in a triaxial cell were performed to determine thevoid ratio under a variety of initial states. The fourK valuesshown in Table 3 were used in each of the four sets of com-pression tests. The confining stress was increased in stepsfollowing a given K value. The same initialDr and stressconditions used in chamber tests were duplicated in thesecompression tests. The same specimen preparation, satura-tion, and testing procedures as described previously for iso-tropic compression tests were followed. The results of thecompression tests are plotted in Fig. 12 in terms of void ra-tio versus effective mean normal stress and referred to as theCPT initial states. These initial states, which consider the ef-

fects of sand compression due to consolidation, are used inthe following analysis for the CPT data in MLS.

Grain crushing of MLS occurs even under mild stressconditions, as demonstrated in Fig. 12. The framework proposedby Konrad (1998) provides a system where the effects ofgrain crushing are integrated in the interpretation of CPT insand. The interpretation is based on normalizedqc, normal-ized state parameter (ψN = ψ/(emax – emin)), initial state stress( ′p0), and the stress at critical state (′pcs). The main purposeof normalization is to minimize the effects from differencesin p′0, range of void ratios (i.e.,emax – emin), and ′pcs for vari-

© 1999 NRC Canada

98 Can. Geotech. J. Vol. 36, 1999

0 200 400 600 800 1000σ'h, kPa

5

15

25

35

0

10

20

30

40

q c,M

Pa

Initial Dr, %

50

70

85

Fig. 17. qc versusσh ′.

100 300 5000 200 400 600σ'v, kPa

5

15

25

35

0

10

20

30

40

q c,M

Pa

Initial Dr, %

50

70

85

Fig. 18. qc versusσv ′.

2 6 10 14 180 4 8 12 16 20qc, MPa

100

300

500

0

200

400

600

Dep

thof

pene

tratio

n,m

m

Initial Dr = 70%σ'v = σ'h = 294.3 kPaCone diameter, mm

35.6

17.8

Fig. 16. Comparison ofqc profiles from tests with two differentcone diameters.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

ous soil types. Using Ticino sand and a′p0 of 100 kPa as areference, it is possible to obtain a unique normalizedqc forseveral types of sand, as reported by Konrad (1998). Thisframework offers an ideal platform whereqc in MLS can becompared with a clean, uniformly graded sand, such asTicino sand, considering the differences in dilatancy, grada-tion, and grain crushing characteristics.

Figure 19 compiles MLS chamber CPT data along withTicino sand reported by Baldi et al. (1986) and Konrad(1998). Theqc value is normalized following an empiricalequation by Konrad (1998) for a reference′p0 of 100 kPa as

[4]q p

pq p

p

p

p

c

kPa

c−′

= −′

′

′ =

0

0 100

0

0

00

0

100

.373

where p0 is the total mean stress at initial state. For therange ofψN and ′p0 applied, the data points of MLS are clus-tered within a rather narrow range of normalizedqc. This isan indication that for a given initial stress state,qc in MLS isconsiderably less sensitive toψN (or relative density) incomparison with Ticino sand. The phenomenon is consistentwith the result depicted in Fig. 17 where initialσh′ shows amuch stronger influence thanDr.

The framework by Konrad (1998) calls for a further nor-malization to consider the differences in′pcs that relate tograin crushing. In this case, the normalizedqc obtained ineq. [4] is multiplied by the ratio of (′ ′p pcsTicino csMLS/ ) thatcorresponds to the sameψN. The last normalization shouldbring MLS data points closer to those of the Ticino sand, ac-cording to Konrad (1998). However, due to the drastic dif-ferences inλc between MLS and Ticino sand, as shown inFig. 12, ( ′ ′p pcsTicino csMLS/ ) increases significantly withψN.As a result, the ′pcs normalization would shift MLS datapoints into a trend that is opposite to that of the Ticino sand(i.e., normalizedqc increases withψN), which is apparentlyunrealistic. The ′pcs normalization implicitly assumes that fora given ′p0 andψN, sand of different types near the cone tip

during penetration should have similar dilative orcontractive behavior. Or, the state paths should follow thesame direction to reach critical state where grain crushingoccurs under high-stress conditions. This assumption wouldbe valid for most of the clean, uniformly graded quartz sandwith comparable compressibility. As reported in Fig. 8, thecompressibility of MLS can be many times higher, espe-cially under high stresses. It is thus expected that MLS neara penetrating cone tip is much more contractive than Ticinosand under the same′p0 andψN. For Konrad’s (1998) frame-work to be valid, it is necessary to account for such differ-ences in sand compressibility. For MLS, it may require thechoice of ′pcs that corresponds to a lowerψN than that inTicino sand. The current framework, however, does not pro-vide a basis for an appropriate selection of thatψN.

The above analysis provides convincing evidence that aseparate method is called for when interpreting CPT inMLS. The CPT calibration data in MLS were compiled andan empirical equation that relatesqc to Dr and state of stresswas developed. This equation follows the pattern ofFioravante et al. (1991) that considers the effects ofσv′ andσh′ as follows:

[5] q Dc v h rexp= ′ ′230 1 450108 0 425( ) ( ) ( . ). .σ σ

Note thatDr used in eq. [5] is calculated based on void ra-tios after the application of confining stress. Figure 20shows a comparison betweenqc values obtained in the avail-able chamber CPT and those estimated using eq. [5]. Thecoefficient of correlation is 0.966.

Concluding remarks

The paper compiled and analyzed a series of in situ testdata on MLS, a relatively compressible silty fine sand. Lab-oratory experiments were performed to determine the com-pressibility, shear strength, and dilatancy of MLS that has a

© 1999 NRC Canada

Huang et al. 99

0 10 20 30 40Computed qc, MPa

0

10

20

30

40

Mea

sure

dq c,

MP

a

MLS

Fig. 20. Comparison between measuredqc and qc computedusing eq. [5].

-1.0 -0.5 0.0 0.5Normalized state parameter,ΨN

10

100

1000

(qc

-p0

)/p'

0at

p' 0=

100

kPa

Sand type

MLS

Ticino (Baldi et al. 1986; Konrad 1998)

Fig. 19. Normalizedqc versusψN.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

© 1999 NRC Canada

100 Can. Geotech. J. Vol. 36, 1999

fines content of approximately 15%. A series of CPT wereconducted in a calibration chamber to evaluate the possibledifferences between tests in MLS and those in clean, uni-formly graded quartz sand. Based on the findings of thesestudies, the following conclusions are made.

(1) For a compressible sand such as MLS, the change invoid ratio as a result of static loading can be many times thatof a clean, uniformly graded quartz sand. The effects ofcompressibility should not be ignored for MLS in the inter-pretation of test data.

(2) The ′φcrit of MLS is comparable with the lower end ofa quartz sand. However, the range of′φmax – ′φcrit for MLS isrelatively small compared with quartz sand. This is a resultof the lack of dilatancy of MLS.

(3) For a given initial stress state,qc in MLS is consider-ably less sensitive toψN (or relative density) in comparisonwith clean quartz sand.

(4) Normalizing CPT results in MLS following the frame-work proposed by Konrad (1998), without considering thatthe differences of compressibility may be unrealistic. Theanalysis of available data provides convincing evidence thata separate method is called for when interpreting CPT inMLS.

(5) For MLS, there appears to be a consistent relationshipbetweenqc and σh′. The qc values obtained from availablechamber CPT show a coefficient of correlation of 0.966 withthose estimated using the empirical equation developedherein.

(6) The laboratory experiments described in this paper arerestricted to MLS with a fines content of approximately15%. Further studies are necessary to determine the effectsof different fines contents on the behavior of MLS.

Acknowledgments

This research was conducted under a contract with theConstruction Engineering Division of the Formosa PlasticsGroup, Taipei, Taiwan. The National Science Council of theRepublic of China provided additional funding under con-tracts 82-0115-E-009-373 and 83-0410-E-009-026.

References

Baldi, G., Belloti, R., Ghiona, V., Jamiolkowski, M., andPasqualini, E. 1986. Interpretation of CPTs and CPTUs. Part 2.Drained penetration of sands.In Proceedings of the Fourth In-ternational Geotechnical Seminar on Field Instrumentation andin situ Measurements, Nanyang Technological Institute, Singa-pore, pp. 129–156.

Been, K., and Jefferies, M.G. 1985. A state parameter for sands.Géotechnique,35(2): 99–112.

Been, K., Jefferies, M.G., and Hachey, J. 1991. The critical state ofsands. Géotechnique,41(3): 365–381.

Bellotti, R., and Pedroni, S. 1991. Design and development of asmall calibration chamber for compressible sands.In Proceed-ings of the First International Symposium on Calibration Cham-ber Testing, Potsdam, N.Y.Edited by A.-B. Huang. Elsevier,New York, pp. 91–100.

Bishop, A.W., and Green, G.E. 1965. The influence of end restrainton the compression strength of a cohesionless soil.Géotechnique,15(3): 243–266.

Black, D.K., and Lee, K.L. 1973. Saturating laboratory samples byback pressure. Journal of the Soil Mechanics and Foundation

Engineering Division, ASCE,99(1): 75–93.Bolton, M.D. 1986. Strength and dilatancy of sands. Géotechnique,

36(1): 65–78.El-Sohby, M.A., and Andrawes, K.Z. 1972. Deformation character-

istics of granular materials under hydrostatic compression. Ca-nadian Geotechnical Journal,9(3): 338–350.

Fioravante, V., Jamiolkowski, M., Tanizawa, F., and Tatsuoka, F.1991. Result of CPTs in Toyoura quartz sand.In Proceedings ofthe First International Symposium on Calibration ChamberTesting, Potsdam, N.Y.Edited byA.-B. Huang. Elsevier, NewYork, pp. 135–146.

Hardin, B.O. 1985. Crushing of soil particles. Journal of theGeotechnical Engineering Division, ASCE,111(10): 1177–1192.

Houlsby, G.T., and Hitchman, R. 1988. Calibration chamber testsof a cone penetrometer in sands. Géotechnique,38(1): 39–44.

Huang, A.B., and Ma, M.Y. 1994. An analytical study of cone pen-etration tests in granular material. Canadian Geotechnical Jour-nal, 31(1): 91–103.

Huang, A.B., Bunting, R.D., and Carney, T.C. 1991. Piezobladetests in a clay chamber.In Proceedings of the First InternationalSymposium on Calibration Chamber Testing, Potsdam, N.Y.Edited byA.-B. Huang. Elsevier, New York, pp. 161–174.

Ishihara, K. 1993. Liquefaction and flow failure during earth-quakes. The 33rd Rankine Lecture. Géotechnique,43(3): 351–415.

Joustra, K., and de Gijt, J.F. 1982. Results and interpretation ofcone penetration tests in soils of different mineralogical compo-sition. In Proceedings of the Second European Symposium onPenetration Testing, Amsterdam, The Netherlands, Vol. 2,pp. 615–626.

Koerner, R.M. 1970a. Effect of particle characteristics on soilstrength. Journal of the Soil Mechanics and Foundation Engi-neering Division, ASCE,96(4): 1221–1234.

Koerner, R.M. 1970b. Behavior of single mineral soils in triaxialshear. Journal of the Soil Mechanics and Foundation Engi-neering Division, ASCE,96(4): 1373–1390.

Konrad, J.M. 1998. Sand state from cone penetrometer tests: aframework considering grain crushing stress. Géotechnique,48(2): 201–215.

Kulhawy, F.H., and Mayne, P.W. 1991. Relative density, SPT, andCPT interrelationships.In Proceedings of the First InternationalSymposium on Calibration Chamber Testing, Potsdam, N.Y.Edited byA.-B. Huang. Elsevier, New York, pp. 197–211.

Lade, P.V., Yamamuro, J.A., and Bopp, P.A. 1996. Significance ofparticle crushing in granular materials. Journal of theGeotechnical Engineering Division, ASCE,122(4): 309–316.

Lee, K.L., and Farhoomand, I. 1967. Compressibility and crushingof granular soils in anisotropic triaxial compression. CanadianGeotechnical Journal,4(1): 68–86.

McNeilan, T.W., and Bugno, W.T. 1984. Cone penetration test re-sults in offshore california silts.In Proceedings of the Sympo-sium on Strength Testing of Marine Sediments: Laboratory andin-situ Measurements, San Diego, Calif. ASTM STP 883,pp. 55–71.

Mesri, G., Feng, T.W., and Benak, J.M. 1990. Postdensificationpenetration resistance of clean sands. Journal of theGeotechnical Engineering Division, ASCE,116(7): 1095–1115.

Parkin, A.K. 1988. The calibration of cone penetrometers.In Pro-ceedings of the First International Symposium on PenetrationTesting, ISOPT-1, Orlando, Fla.Edited byDe Ruiter. Balkema,Rotterdam, The Netherlands, pp. 221–243.

Rix, G.J., and Stokoe, K.H. 1991. Correlation of initial tangentmodulus and cone penetration resistance.In Proceedings of theFirst International Symposium on Calibration Chamber Testing,

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.

© 1999 NRC Canada

Huang et al. 101

Potsdam, N.Y.Edited by A.-B. Huang. Elsevier, New York,pp. 351–362.

Salgado, R., Mitchell, J.K., and Jamiolkowski, M. 1997. Cavity ex-pansion and penetration resistance in sand. Journal ofGeotechnical and Geoenvironmental Engineering, ASCE,123(4): 344–354.

Vesic, A.S., and Clough, G.W. 1968. Behavior of granular materi-als under high stresses. Journal of the Soil Mechanics and Foun-dation Engineering Division, ASCE,94(3): 661–688.

Voyiadjis, G.Z., Kurup, P.U., and Tumay, M.T. 1993. Preparationof large-size cohesive specimens for calibration chamber testing.Geotechnical Testing Journal, ASTM,16(3): 339–349.

Yamamuro, J.A., Bopp, P.A., and Lade, P.V. 1996. One dimen-sional compression of sands at high pressures. Journal of theGeotechnical Engineering Division, ASCE,122(2): 147–154.

Yu, H.S., and Houlsby, G.T. 1991. Finite cavity expansion indilatant soils: loading analysis. Géotechnique,41(1): 173–183.

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

atio

nal C

hiao

Tun

g U

nive

rsity

on

04/2

8/14

For

pers

onal

use

onl

y.