1195

Undrained anisotropy of Hostun RF loose sand:

new experimental investigations
Zeina Finge, Thiep Doanh, and Philippe Dubujet

Abstract: The undrained behaviour of loose and overconsolidated Hostun RF sand in triaxial compression and extension
tests is described. The samples are isotropically or anisotropically overconsolidated along several constant effective
stress ratio paths with various overconsolidation ratios (OCR), up to 24. To minimize the effect of variation of density
on the observed undrained behaviour, all tested samples are required to have a nearly identical void ratio before the
final monotonic undrained shearing. Isotropically overconsolidated and normally consolidated samples exhibit the same
phenomenon of partial static liquefaction, but anisotropically overconsolidated specimens reveal a completely different
undrained behaviour. A common pseudoelastic response is observed for a given overconsolidation history. This response
is induced by recent stress history in terms of effective stress paths, independent of the OCR during overconsolidation.
The initial gradient of the effective stress paths seems to depend solely on the direction of the previous linear stress
path history. This paper offers a comprehensive understanding of the mechanism of the induced anisotropy of loose
sand created by simple linear stress paths from three different initial stress states in the classical triaxial plane. The
pseudoelastic response can be adequately modelled by a simple hyperelastic component of the elastoplastic framework.
Key words: induced anisotropy, overconsolidation, instability, laboratory undrained tests, sand, hyperelasticity.

Résumé : Cet étudie le comportement non drainé de sable Hostun RF très lâche et surconsolidé à l’appareil triaxial.
Les échantillons sont surconsolidés isotropiquement ou anisotropiquement le long des chemins linéaires de rapport de
contraintes constant avec plusieurs valeurs du rapport de surconsolidation, jusqu’à 24. L’indice des vides de tous les
échantillons testés est pratiquement identique dans le but de minimiser son influence éventuelle sur le comportement
observé. Tandis que la surconsolidation isotrope montre le phénomène de liquéfaction statique qui est un comportement
non drainé très semblable à celui de la consolidation isotrope vierge, la surconsolidation anisotrope indique un com-
portement complètement différent. Pour chaque chemin de surconsolidation, on observe une réponse pseudo-élastique
indépendamment du rapport de surconsolidation. Il semble que l’inclinaison initiale des chemins de contraintes effectives
ne dépend que de la direction des préchargements. Cet article a permis une compréhension détaillée du mécanisme de
l’anisotropie induite des sables lâches créée par les chemins simples de contraintes linéaires à partir de plusieurs états
initiaux de contrainte dans le plan triaxial classique. La réponse pseudo-élastique peut ètre simuler par un simple modèle
hyperélastique de l’élastoplasticité.
Mots clés : anisotropie induite, surconsolidation, instabilité, essais non drainés, hyperélasticité.

1. Introduction Many experimental studies have been devoted to the induced

anisotropy since the early days of soil mechanics, including
Ko -consolidated soil is widely studied in the geotechnical
those of Poorooshasb et al. (1966, 1967), Tatsuoka and Ishi-
literature because it simulates the sedimentation of natural soil
hara (1974), Ishihara and Okada (1978a, 1978b, 1982), and
under the vertical action of gravity. The microstructural changes
others. Due to experimental difficulties, these works are of-
developed at the end of the Ko consolidation are commonly re-
ten restricted first to isotropic overconsolidated materials. The
ferred to as the inherent anisotropy. The induced anisotropy is standard drained triaxial compression preshear is largely used
another form of anisotropy resulting from application of stress to study the effects of induced anisotropy on the subsequent
after the completion of the Ko consolidation process. The usual drained and undrained behaviour.
overconsolidation ratio (OCR), a scalar measure of structural By taking advantage of recent advances in measurements and
changes, is adopted to characterize the stress-induced compo- control techniques, true Ko consolidation is performed to better
nent of anisotropic soil.
simulate the anisotropic character of gravitational forces during
deposition of most natural soils. The dependence of soil char-
Received 29 October 2004. Accepted 21 April 2006. Published on acteristics (frictional angle, dilatancy angle, effective cohesion)
the NRC Research Press Web site at http://cgj.nrc.ca/ on has been investigated using the newly designed simple shear de-
15 December 2006. vice (Arthur and Menzies 1972), the hollow cylinder apparatus
Z. Finge and T. Doanh.1 École Nationale des Travaux Publics (Hight et al. 1983), and the plane strain machine (Tatsuoka et
de l’État, Laboratoire Géomatériaux, URA 1652, 69518 Vaulx en al. 1986). For example, induced anisotropy and principal stress
Velin Cedex, France. rotation have been explored in a series of pioneer papers by
Ph. Dubujet. ENISE, Laboratoire de Tribologie et Dynamique des Arthur (Arthur and Menzies 1972; Arthur et al. 1977, 1980).
Systèmes, UMR 5513, 42023 Saint-Etienne Cedex, France. These relevant experimental studies showed the key role of the
1
Corresponding author (e-mail: [email protected]). major axis of consolidation and the first stress loading in mod-
ifying greatly the mechanical properties of soil samples.

Can. Geotech. J. 43: 1195–1212 (2006) doi: 10.1139/T06-068 © 2006 NRC Canada
1196 Can. Geotech. J. Vol. 43, 2006

Most of the past experimental studies mentioned previously Fig. 1. Different preloading stress paths.
concern medium to dense materials. Loosely deposited soil re-
ceived considerably less attention, despite the associated devas-
tating liquefaction phenomenon. A completely different
undrained behaviour was revealed in the early works on loose
soil by Casagrande (1975), Castro (1969), and Poulos (1981).
Since then, experimental works on this topic continued to ap-
pear, advancing progressively on the study of some influenc-
ing factors: the effects of undrained preshear (Ishihara and
Okada 1978b); the influence of an initial static deviatoric stress
(Kramer and Seed 1988; Lade et al. 1988; Canou et al. 1991);
the paramount importance of initial density (Konrad et al. 1991;
Verdugo and Ishihara 1996) and sample preparation methods
(Vaid et al. 1999); the dependence on the preloading history
(Di Prisco et al. 1995); in particular, the anisotropic consolida-
tion (Doanh et al. 1997; Kato et al. 2001). New concepts were
introduced to explain the newly observed behaviour: the steady
state of deformation by Castro (1969), the concept of state pa- compression and extension triaxial tests were performed on
rameter by Been and Jefferies (1985), the Lade instability lines isotropically and anisotropically overconsolidated samples.
(Lade et al. 1988), the Sladen collapse surface (Sladen et al. In the first part of the laboratory program, a detailed de-
1985), and the Konrad conceptual framework (Konrad 1993) scription of the static liquefaction of loose and isotropically
on the steady state of deformation based on triaxial compres- overconsolidated sand samples is obtained. The isotropic over-
sion tests on virgin isotropic consolidated materials. consolidation is chosen first to isolate the effects of the effective
The triaxial extension tests were used recently to extend the mean pressure p
. The static liquefaction phenomenon of loose
previous observations on the compression side. Doanh et al. sand is corroborated and extended to isotropically overconsol-
(1997) reported a comprehensive set of experiments on the idated sand in triaxial compression and extension.
very loose Hostun RF sand subject to several constant stress The second part of the laboratory program shows the influ-
ratio paths. The instability concept of Lade (Lade et al. 1988) ence of two other preloading histories plotted in Fig. 1, adding
was examined in the case of extension tests and extended to the effects of the previous deviatoric stress q on the mechani-
anisotropic sand. An extension of Konrad’s (1993) conceptual cal behaviour. The preloading history consists of an anisotropic
framework to anisotropic sand is proposed in Doanh and Ibraim K consolidation along a constant effective stress ratio path in
(2000). compression, from an initial isotropic stress state at fabrication
Studies on Ko overconsolidation represent the next logical stage point A of 20 kPa of p
, up to a desired value of OCR at
step and include the following: Georgiannou et al. (1990) on point C. The consolidation ratio K = σr
/σa
is defined as the ratio
anisotropically consolidated clayey sands; Matiotti et al. (1995) of the effective radial stress, σr
, to the effective axial stress, σa
.
and Gajo and Piffer (1999) on lightly isotropic overconsolidated Then the test is unloaded back along the same constant stress
Hostun RF sand; Kurukulasuriya et al. (1999) on kaolin clay; ratio path to a stress point D having a pc
of 50 kPa. This final pc

and Zdraković and Jardine (2000) on silt with local strain sen- is chosen to have a large value of OCR, up to 24, with the cur-
sors. rent equipment. The K overconsolidation is different from the
The purpose of this paper is to present a set of new exper- usual Ko overconsolidation in which no radial deformation is
allowed. The OCR is defined as pmax
/ pc
; pmax

is the maximum
imental observations obtained from an extensive program on
loosely compacted and anisotropically overconsolidated Hos- effective mean pressure attained on the K-consolidation line,
tun RF sand. The overconsolidation is achieved along several and pc
is the effective mean pressure before the final monotonic
constant stress ratio paths (K overconsolidation). A special ex- undrained shearing. Figure 1 illustrates the three different stress
perimental program is designed to study the effects of the stress- paths performed in this paper.
induced anisotropy created by simple linear stress paths from Recent advances in measurements and control techniques
various initial static stress states in the classical triaxial plane. permit the modifications of a Farnell triaxial compression ma-
A detailed description of the undrained instability of the nor- chine to accurately apply the bidirectional loadings (axial and
mally consolidated Hostun sand has been reported previously radial) to a cylindrical soil sample. The modifications, sam-
(Doanh et al. 1997). This paper highlights the effectiveness of ple preparation, saturation, and verification of the computer-
using simple hyperelasticy to model the observed pseudoelastic controlled stress path can be found in Doanh et al. (1997) or
response of overconsolidated sand. Ibraim (1998). The initial void ratio determination, the repeata-
bility of the undrained test as a consequence of measurement
errors induced by the imperfections of testing method, and the
2. Experimental program verification of the antifrictional system are shown in Doanh and
Ibraim (2000).
An extensive laboratory program was conducted to study Very loose and contractive sand samples were prepared with
some effects of the induced anisotropy on the undrained be- a modified moist tamping and undercompaction method (Bjer-
haviour of loosely compacted Hostun RF sand from several rum et al. 1961; Ladd 1978). Although the moist-tamped sand
initial stationary stress states in the triaxial plane. Undrained may not be representative of natural sand (Frost and Park 2003),

© 2006 NRC Canada

Finge et al. 1197

Table 1. Undrained tests on loose and overconsolidated Hostun sand.

Testsa OCRe0 ec pmax qpeak ppeak εpeak φpeak ηpeak
(kPa) (kPa) (kPa) % (◦ )
Isotropic overconsolidation K = 1.00
I-CO1 1 1.000 0.956 50 16.40 32.81 0.144 13.34 0.50
I-CO1.4 1.4 1.027 0.920 70 17.43 34.77 0.235 13.37 0.50
I-CO2 2 1.033 0.946 100 25.99 41.27 0.214 16.56 0.63
I-CO4 4 1.017 0.951 200 35.91 46.72 0.259 19.92 0.77
I-CO8 8 1.041 0.923 400 43.99 48.38 0.367 23.25 0.91
I-EO1 1 1.058 0.948 50 –15.95 41.56 –0.124 –11.83 –0.38
I-EO2 2 1.027 0.919 100 –18.45 38.48 –0.132 –15.10 –0.48
I-EO4 4 1.021 0.918 200 –25.90 42.15 –0.182 –20.02 –0.61
I-EO8 8 1.028 0.933 400 –30.50 46.55 –0.210 –21.58 –0.66
Anisotropic overconsolidation K = 0.50
KB-CO1 1 1.042 0.913 50 25.57 45.11 0.069 15.00 0.57
KB-CO2 2 1.022 0.948 100 43.76 54.23 0.100 20.83 0.81
KB-CO4 4 1.049 0.912 200 59.39 59.79 0.245 25.22 0.99
KB-CO8 8 1.035 0.890 400 80.40 71.38 0.426 28.31 1.13
KB-CO12 12 1.036 0.883 600 101.51 83.04 1.023 30.40 1.22
KB-CO24b 24 1.021 0.926 600 148.77 113.99 1.404 32.41 1.31
KB-EO1 1 1.035 0.941 50 –10.43 33.29 –0.186 –9.51 –0.31
KB-EO2 2 1.029 0.933 100 –14.70 30.44 –0.295 –15.22 –0.48
KB-EO4 4 1.031 0.912 200 –16.69 30.52 –0.357 –17.51 –0.55
KB-EO8 8 1.045 0.927 400 –14.48 30.87 –0.388 –14.74 –0.47
KB-EO12 12 1.038 0.909 600 –16.10 31.95 –0.466 –15.76 –0.50
Anisotropic overconsolidation K = 0.35
KC-CO1 1 1.034 0.981 50 37.54 49.65 0.017 19.62 0.76
KC-CO2 2 1.025 0.893 100 62.11 58.11 0.108 26.98 1.07
KC-CO4 4 1.029 0.893 200 93.69 74.85 0.386 31.19 1.25
KC-CO8 8 1.034 0.923 400 172.21 123.12 2.269 34.55 1.40
KC-CO12 12 1.025 0.898 600 — — — — —
KC-EO1 1 1.014 0.955 50 –7.22 26.26 –0.310 –8.28 –0.27
KC-EO4 4 1.027 0.939 200 –7.14 19.88 –0.369 –11.01 –0.36
KC-EO8 8 1.024 0.924 400 –7.32 25.46 –0.466 –8.68 –0.29
Note: εc , void ratio at the end of the preloading history before undrained shearing.
a I, isotropic consolidation; KB and KC, anisotropic K consolidation with stress ratio K of 0.50 and
0.35, respectively; C and E, monotonic final undrained shearing more in compression and extension,
respectively; O, overconsolidation.
b Test conducted with p
= 25 kPa.
c

it constitutes a useful laboratory step towards a better under- extension and terminated with a large strain reversal from com-
standing of the latter. Although the water-pluviation method pression to extension or vice-versa. The axial deformation rate
may better simulate natural deposited sand, the moist-tamping is about 0.21%/min. Table 1 summarizes all tests in this study
method is the most useful method of obtaining very loose sand with their initial conditions and results.
samples in the laboratory. Verification of the isotropic hypothe-
sis on samples obtained by this method is done in Ibraim (1998) 2.1. K consolidation and void ratio before undrained
and Finge (2004). All samples in this study have the same ini- shearing
tial void ratio at fabrication state, e0 of around 1.03. Special Figures 2a and 2b present the total volumetric strain and the
care was taken to have samples with similar void ratios at the void ratio reduction developed in isotropic overconsolidation.
beginning of the undrained shearing to minimize the effect of All samples follow a nearly identical normal consolidation line
variation of density on the observed undrained behaviour. The and different swelling lines. These swelling lines are almost par-
average relative density, Dr , before undrained shearing is less allel. The isotropic hypothesis (volumetric strain εv = 3.0εa )
than 15%. The physical properties of the well-known Hostun over the studied stress range is nearly verified. The average
RF uniform and angular quartz sand are as follows: maximum gradient is 2.97 ± 0.30 during loading and 3.32 ± 0.36 during
void ratio emax = 0.961, minimum void ratio emin = 0.624, me- unloading, indicating a steeper line in the vertical direction after
dian particle size D50 = 0.38 cm, the uniformity coefficient isotropic overconsolidation. For the largest OCR value of 8, a
Uc = 2.0, and the specific gravity Gs = 2.65. Other properties residual contractancy of 1.8% of εv is obtained, meaning that a
can be found in Flavigny et al. (1990). very loose sample with e0 around 1.00 at the fabrication stage
All tests were sheared axially in a strain-controlled mode remains loose after being compressed to 400 kPa and unloaded
up to +15% of axial strain, εa , in compression or −10% in to 50 kPa. The void ratio at the end of the preloading history

© 2006 NRC Canada

1198 Can. Geotech. J. Vol. 43, 2006

Fig. 2. Results on isotropically overconsolidated samples. The numbers on the curves in Figs. 2–8 indicate the OCR value.
(a) Volumetric behaviour. (b) Void ratio reduction.

Table 2. Average gradient δεv /δεa for different K strain of about 7.00%.
overconsolidation.
K Loading Unloading
3. Experimental results
1.00 2.97 3.32
0.50 1.02 2.33 The test results were interpreted in terms of the deviatoric
0.35 0.62 1.85 stress q = σa
− σr
and the effective mean pressure p
=
(σa
+ 2σr
)/3. The stress paths were examined in the q–p
plane,
ec is 0.935 ± 0.015 or Dr = 7.79 ± 7.40% for all tests. Fig- and the response behaviour mainly in the deviatoric – axial
ures 3a and 3b give an example of the total volumetric strain strain plane and in the normalized excess pore pressure – axial
and the void ratio reduction developed in the K consolidation strain plane. The measured excess pore pressure was normal-
with KB = 0.50. The void ratio ec before undrained shearing is ized by the effective mean pressure before the final monotonic
0.918 ± 0.020 for the KB series and 0.926 ± 0.032 for the KC undrained shearing, and the origin of the axial strain εa was
series. Linear stress paths create almost linear strain paths after a taken at the end of the anisotropic K overconsolidation.
small nonlinearity in the first part. The average value of the gra-
dient δεv /δεa for different K consolidation is shown in Table 2. 3.1. Isotropic overconsolidation
A complete description of this analysis can be found in Finge The first monotonic undrained test series on isotropic over-
(2004). A maximum value of 3.77% of εa before undrained consolidated samples was conducted to extend the undrained
shearing is obtained for the largest OCR value of 24 of the KB behaviour of very loose Hostun sand previously studied by
series in Fig. 3a. Shear bands have never been observed, and the Meghachou (1992), Konrad (1993), and Doanh et al. (1997) to
shape of the sample remains almost cylindrical. These visual normally consolidated samples. This series tends to corroborate
observations suggest homogeneous deformations up to an axial the compression behaviour of lightly overconsolidated Hostun

© 2006 NRC Canada

Finge et al. 1199

Fig. 3. Results on KB-overconsolidated (KB = 0.50) samples. (a) Volumetric behaviour. (b) Void ratio reduction.

sand investigated by Di Prisco et al. (1995), Matiotti (1996), and of the yield surface, however, and thus isotropic preconsolida-
Gajo and Piffer (1999) and Toyoura sand investigated by Ishi- tion markedly enhances the stability in compression. Neverthe-
hara and Okada (1978a). It explores the undrained behaviour at less, many specific questions are still unresolved.
higher overconsolidation levels in compression and extension. In this study, all isotropically overconsolidated tests, up to
The undrained behaviour established in previous investigations OCR = 8, exhibit a behaviour similar to that of normally con-
(Doanh et al. 1997) on normally consolidated sand serves as solidated tests with some important differences. Globally, a con-
background for evaluating the new results presented here. tinuous contractant behaviour in compression and extension is
Previous published results on virgin isotropically consoli- observed during the monotonic shearing. At large strains, some
dated samples give a fairly accurate image of the global extension tests exhibit a small dilatant behaviour like that of
behaviour of very loose Hostun sand in compression and exten- their counterpart of normally consolidated samples.
sion, i.e., the partial static liquefaction, the applicability of the The stress–strain relationship of overconsolidated loose sand
instability concept, the deviatoric stress peak at relatively low is also characterized by a sharp decrease in deviatoric stress af-
axial strain, the post-peak reduction in strength and the contrac- ter a peak, with a maximum in compression and a minimum in
tive behaviour, the normalized behaviour, the increasing pattern extension, reached rapidly at small strains after the start of the
of axial strain at peak with increasing confining pressure, the undrained test (Fig. 4a). The numbers shown in Fig. 4a indicate
continuous response at isotropic stress state with small dilative the OCR value. This softening characteristic is followed by a
behaviour in extension, and the perpendicular intersection of steady stress state in which the residual strength remains con-
the effective stress paths with the hydrostatic axis. stant under a large range of axial strains. The deviatoric stress
Despite the small number of lightly overconsolidated sam- sometimes increases slightly after +8% in compression tests or
ples under isotropic conditions with OCR < 2, earlier published decreases after −6% in extension tests. The deviatoric stress at
experiments show qualitatively the same partial static liquefac- peak, qpeak , and the corresponding axial strain at peak, εpeak ,
tion behaviour. The experiments indicate an isotropic expansion depend on the OCR. In Fig. 5a, the deviatoric stress peaks, in-

© 2006 NRC Canada

1200 Can. Geotech. J. Vol. 43, 2006

Fig. 4. Results on isotropically overconsolidated (K = 1.00) samples. (a) Stress–strain relationships. (b) Pore pressure generation.
(c) Effective stress paths.

© 2006 NRC Canada

Finge et al. 1201

Fig. 5. Detailed results on isotropically overconsolidated (K = 1.00) samples. (a) Stress–strain relationships. (b) Pore pressure
generation. (c) Effective stress paths.

© 2006 NRC Canada

1202 Can. Geotech. J. Vol. 43, 2006

dicated by small arrows, appear to follow the increase in the paths of a series of 15 repeatability tests on normally consoli-
OCR. The greater the OCR, the larger the deviatoric stress and dated samples in a previous work by Doanh and Ibraim (2000)
the larger the axial strain at peak. An OCR of 8 significantly are also shown, after normalization, in boldface broken lines
increases the undrained strength 168% in compression and only to emphasize this strange feature. The light-face broken lines
80% in extension. This means almost 9.9◦ and −9.7◦ of differ- show the two new normally consolidated tests from this study.
ences in the mobilized frictional angle at peak, φpeak , in com- Although the same sample preparation method was used, the re-
pression and extension, respectively. A regular increase in axial peatability tests may experience some initial bedding problems
strain at peak is also observed, for example, in compression, at small strains, as shown in Fig. 5c.
the peak stress is reached rapidly at 0.144% axial strain for The small circles indicating the stress states at peak are rep-
the isotropic normally consolidated sample, 0.204% at OCR = resented in the q–p
plane in Fig. 4c. In compression and exten-
2, and 0.259% and 0.367% at OCR = 4 and 8. In extension, sion, the effective stress peaks are clearly a function of OCR. In
the same regular progression of axial strain is also observed, other words, the value of the mobilized friction angle at peak,
ranging from −0.132% at OCR = 2 to −0.210% at OCR = 8. φpeak , is about 13.3◦ in compression and −11.8◦ in extension
The undrained shear strength is mobilized rapidly, even for ex- for normally consolidated samples. For an OCR of 8, the cor-
tension tests. These results are expected from previous stud- responding friction angles at peak jump to 23.3◦ and −21.6◦ .
ies with compression tests, and they are confirmed by these Clearly, the Lade instability lines (Lade et al. 1988) and the
new extension tests. Typical excess pore pressure generations Sladen collapse surface (Sladen et al. 1985), based on the de-
at different OCRs are shown in Fig. 4b. The static liquefaction, viatoric stress peak, depend on the previous preloading history.
defined as the ability to reach the origin of the q–p
plane un- These concepts are not intrinsic like the failure envelope.
der monotonic shearing conditions, occurs in compression only
for lightly overconsolidated samples I-CO1 and I-CO1.4 under 3.2. Anisotropic overconsolidation
an initial effective confining pressure of 50 kPa. The partial After the first series of nine tests on isotropically overconsol-
static liquefaction phenomenon, with small residual values of idated samples (K = 1.0), two other series were performed on
q and p
, is attainable with the rest of these tests. Compared to anisotropically overconsolidated samples with a medium value
the quick buildup in normally consolidated samples, the excess of the consolidation ratio KA of 0.50 and a small value KB of
pore pressure of overconsolidated sand developed progressively 0.35.
at a slower pace with the increasse in axial strain and reached The KB-overconsolidation line is above the Lade instabil-
a constant value at about 5% of εa in compression and exten- ity line defined for isotropic sand. Previously published data
sion. The final pore pressure seems to depend on the OCR. (Doanh et al. 1997) have demonstrated that the undrained be-
The greater the OCR, the smaller the final pore pressure. With haviour generated by a higher consolidation ratio, at KA = 0.66,
OCR = 8 (tests I-CO8 and I-EO8), the pore pressure peak value which is below the isotropic instability line, is very similar to
is about 87% of pc
in compression and only 76% in extension. that of virgin isotropic consolidated specimens. For this reason,
Figure 5b examines in detail the first part of all undrained the series KA is discarded in this study. The consolidation ratio
monotonic tests and reveals the tendency to dilation in the ini- KC is chosen because it is close to the failure line. Therefore,
tial phase of the extension tests. This initial dilatant behaviour the third series allows us to check the validity of the previously
clearly depends on the OCR. A large OCR implies a large am- observed undrained anisotropic behaviour.
plitude of dilatancy. For example, the pore pressure decreases
slightly but continuously to about only −5.7% of the initial ef- 3.2.1. KB overconsolidation
fective confining pressure for test I-EO1, and down to −16.6% In contrast to the isotropic overconsolidated series in which
for test I-EO8 with the largest OCR tested. This characteristic the expected partial static liquefaction occurs for all tested
changes after about −0.1% of εa , and overconsolidated loose OCRs, the series with a medium consolidation ratio, KB = 0.50,
sand samples contract again. From an initial isotropic stress reveals a completely different undrained behaviour. The stress–
state, the responses of the material show a regular pattern re- strain relationships, the effective stress paths, and even the ex-
lated to the OCR. The transition from compression to extension cess pore pressure generations change notably with changes in
appears to be smooth in both stress–strain relationships and nor- the OCRs and K (anticipating the results presented in the next
malized pore pressure responses. section).
The effective stress paths in the q–p
plane are illustrated in The increment of the undrained mobilized strength, qu , is
Fig. 4c. From an initial isotropic stress state and for a given defined as
OCR, the two effective compression and extension stress paths
established a “close” surface pointing towards the origin of the [1] qu = qpeak − qc
q–p
plane, with a clearly asymmetric shape about the hydro- where qc is the deviatoric stress at the beginning of the final
static axis. A striking feature emerges from this figure. Contrary undrained shearing. This mobilized undrained strength incre-
to virgin isotropic normally consolidated samples (broken lines ment, qu , in compression, when it exists, increases with an
in this figure), the stress paths of all isotropic overconsolidated increase in the OCR, for the same consolidation stress ratio K,
samples do not intersect perpendicularly the isotropic consoli- and equally increases with a decrease in K for the same OCR.
dation axis. In Fig. 5c, a unique effective stress path is observed In extension, the qu seems to be only slightly affected by the
for all overconsolidated samples in compression. A unique ini- variations in K or OCR.
tial gradient dq/dp
of 10.0 is obtained. This sharp change In Fig. 6a, the expected partial static liquefaction of loose
of p
indicates a surprising anisotropic elastic behaviour for a sand is still observed for samples below an OCR of 8, with
supposed isotropic material. The usual vertical effective stress the usual sharp drop in deviatoric stress immediately after a

© 2006 NRC Canada

Finge et al. 1203

Fig. 6. Results on anisotropically overconsolidated (K = 0.50) samples. (a) Stress–strain relationships. PT, phase transformation. (b) Pore
pressure generation. umax , maximum pore pressure generation. (c) Effective stress paths.

small peak. Above this OCR level, an apparent increase in de- the same trend after a clearly identified minimum following the
viatoric stress at large strains is noticed. As OCR increases, peak. The tests with OCR greater than 12 exhibit a temporary
with a gap between 8 and 12, the stress–strain curves follow softening behaviour. The final increase in deviatoric stress at

© 2006 NRC Canada

1204 Can. Geotech. J. Vol. 43, 2006

Fig. 7. Detailed results on anisotropically overconsolidated (K = 0.50) samples. (a) Stress–strain relationships. (b) Pore pressure
generation. (c) Effective stress paths.

large strains indicates dilative behaviour. In compression, up to reversed trend occurs. A regular increase in axial strain at peak
OCR = 4, the axial strain at peak εpeak is much smaller than that with an increase in OCR is also observed as in the isotropic
of isotropic overconsolidated samples (Figs. 5a, 7a), and then a overconsolidated series: from 0.07% at OCR = 1 to 1.02% at

© 2006 NRC Canada

Finge et al. 1205

OCR = 12 in compression and −0.19% at OCR = 1 to −0.47% strain at the phase transformation, εPT . The reason behind the
at OCR = 12 in extension. The behaviour of partial static liq- appearance of the phase transformation line with large OCRs
uefaction is observed for all anisotropic K-overconsolidated is not yet fully understood, however.
series. A more striking feature is shown in Fig. 6c. In compres- The observed undrained response may be due to time effects,
sion, all effective stress paths are strongly inclined towards the especially the initial gradient of the effective stress paths. Since
positive p
axis and follow a common highly nonlinear curve up the elapsed time corresponding to the overconsolidation phase
to the deviatoric stress at peak, qpeak , independent of the OCR before undrained shearing ranges from 1 h for test KB-CO2
attained during the K-consolidation process before the final to 17 h for test KB-CO12, the same initial response shown in
undrained shearing. A pronounced bulge in the effective stress Fig. 7c suggests a time-independent behaviour of the effective
path with a very sharp bend at qpeak is observed in compression stress paths.
(the direction of consolidation), whereas that in extension is
left nearly unchanged. The descending part after qpeak moves Lastly, it is worth noting the continuous transition from com-
equally upwards with an increase in OCR. This part charac- pression to extension in terms of stress–strain relationships, ef-
terizes the undrained instability behaviour of loose sand in the fective stress paths and even excess pore pressure generations
sense of Hill (1958). In the case of monotonic solicitation in tri- for this KB series.
axial compression, this is equivalent to a decrease in deviatoric
stress. The length of this descending part seems to be linked 3.2.2. KC overconsolidation
with the OCR or the induced anisotropy. The smaller the OCR,
the longer the softening part of the stress. As the OCR increases, Consider now the undrained tests of the third series with a low
the stress state at peak moves closer to the phase transformation consolidation ratio of KC = 0.35. Regarding the previous series,
(PT) line (see later in the paper). The effective stress paths then the question is as follows: given the knowledge of two previous
move up again after the appearance of the dilatancy. Each test series with K = 1.00 and 0.50, can we anticipate experimentally
reaches its arbitrary final stage of about 15.0% of axial strain. the outcome of this low consolidation ratio? In other words, can
In extension, all samples follow a roughly common curve until we expect more surprising results?
the steady state of deformation, and highly overconsolidated Returning to experiments with the lowest consolidation ratio
samples exhibit dilatant behaviour at large strains. in this study, the stress–strain relationships in Fig. 8a give an im-
As for the isotropic overconsolidated series, Fig. 7c gives a pression of déjà vu. Nevertheless, additional details contribute
unique initial gradient of the effective stress paths of only 4.6 to building a more complete view of the global undrained be-
at a confining pressure of 50 kPa. haviour of anisotropic overconsolidated sand. For a fixed OCR,
Typical excess pore pressure generations, U , at different as the consolidation ratio K decreases, the deviatoric stress
OCRs are given in Fig. 6b. As noted in the observations for increases continuously without a peak and without reaching a
stress–strain relationships, a different kind of pore pressure de- stable value at large strains in compression, as in the case of test
velopment appears, compared to Fig. 4b of isotropic overcon- KC-CO12 at OCR = 12. A stationary value of deviatoric stress
solidated series. In compression, normal consolidation indicates is temporarily reached with test KC-CO8 at OCR = 8, within a
a quick buildup of excess pore pressure. Low overconsolidation range of 2.00% of axial strain, from 0.10% to 0.30%, before the
levels show a small plateau in the region of 0.05%–0.15% of final increase. Below this level of OCR, as usual, partial static
axial strain before a continuous increase to a constant value at liquefaction occurs with the exception of a very small qpeak
large strains. High overconsolidation levels have a new scheme but clearly identified for normally consolidated samples. The
of pore pressure development as shown in Fig. 7b. The pore asymptotic stabilization of the undrained strength of normally
pressure attains a first peak at about 0.10% of axial strain, then consolidated samples with decreasing ratio K was reported in
it decreases to a minimum corresponding to an axial strain in the a previous work (Doanh et al. 1997). An even more striking
region of 0.50%–1.00%. Then a second peak of pore pressure feature emerges from Fig. 8c representing the effective stress
after 7.00% of axial strain is distinguished. The last peak usu- paths. The highly nonlinear boundary stretches over a much
ally identifies the transition from contractancy to the dilatancy larger range of p
, covering almost 140 kPa from the initial sta-
phenomenon, beginning from OCR = 2. tionary stress state to the final running up along the failure line
A continuous response of excess pore pressure generation like a dense sand. For the initial gradient of the effective stress
U is noticed between compression and extension tests in paths, a single value of 4.6 is measured. This value is identical
Fig. 7b. The U in extension shows dilative behaviour until to that of the series with K = 0.50.
a minimum is reached at around −0.05% of εa , and then it The position of the PT points for each test is easily identified
increases continuously to a maximum at about –5.00% of εa by the last peak of the excess pore pressure generation, which
before the appearance of the final dilative behaviour. is the transition from contractancy to dilatancy at large strains
In Fig. 6c, the mobilized friction angle at the PT state, φPT , (Fig. 8b), except in the test with the largest OCR. As indicated
does not vary with variation in OCR. The PT points even con- earlier, these PT points are close to the PT line for the series
verge convincingly towards a single line passing through the with KB = 0.50.
origin of the q–p
plane. The slope of this PT line is about 1.45,
The following observations are still valid: a smooth transition
or φPT of 35.6◦ ; thus the concept of the phase transformation
from compression to extension, a regular pattern of increasing
line (Tatsuoka and Ishihara 1974; Luong 1978) developed for
(decreasing) values of qpeak and εpeak (εPT ) with an increase
dense sand seems to apply to loosely and anisotropically over-
in OCR, the existence of a possible temporary softening be-
consolidated sand. For this consolidation level, the larger the
haviour, and a nearly common response in extension.
OCR, the greater the qpeak and εpeak and the smaller the axial

© 2006 NRC Canada

1206 Can. Geotech. J. Vol. 43, 2006

Fig. 8. Results on anisotropically overconsolidated (K = 0.35) samples. (a) Stress–strain relationships. (b) Pore pressure generation.
(c) Effective stress paths.

3.3. Undrained anisotropic behaviour analysis elastoplasticity framework: the unexpected initial gradient, the
The rest of the paper is devoted to a reexamination of some common highly nonlinear part of the effective stress paths, and
surprising experimental features with respect to the classic the complex stress–strain behaviour of overconsolidated sand.

© 2006 NRC Canada

Finge et al. 1207

Table 3. Initial gradient of effective stress paths in compression.

Mode Initial gradient, dq/dp Ea /Er
Virgin isotropic consolidation Vertical 1.00
Isotropic overconsolidation, K = 1.00 10.0 1.40
Anisotropic overconsolidation, K = 0.50 4.6 2.27
K = 0.35 4.6 2.27

3.3.1. Initial gradient of effective stress path Fig. 9. Different interpretations of effective stress paths of
Table 3 summarizes the initial slopes of the effective stress overconsolidated loose sand.
paths for three overconsolidated series. These measured ini-
tial gradients are reasonably beyond the experimental errors.
They are the average values of at least four tests, and 15 tests
in the case of the isotropic normally consolidated series. The
measured inclinations contradict the results of Gajo and Pif-
fer (1999), which indicate that the effective stress path of an
isotropic overconsolidated sample (OCR = 2) is perpendicular
to the p
axis at the beginning of undrained shearing. The results
confirm the experimental results of Di Prisco et al. (1995) for the
same OCR, which apparently gives a slope towards the positive
p
axis. It is nearly impossible to obtain a reliable initial gradi-
ent in these works because of possible measurement errors in
the small-strain range. Close scrutiny of some previously pub-
lished data for Ko -overconsolidated soils (Georgiannou et al.
1990; Zdravković and Jardine 2000) indicates a strong proba- anisotropic structure created by isotropic overconsolidation. In
bility of similar initial gradients, although the method of sample other words, it reveals the initial anisotropic structure of sand
fabrication is different. Some questions arise because of these samples unaltered by isotropic overconsolidation. The moist
initial strongly inclined gradients of effective stress paths of tamping and undercompaction process logically creates an ini-
K-overconsolidated sand: tial structure with a strong vertical direction. Isotropic consoli-
(1) For the isotropic overconsolidated series of loose sand, dation of sufficient magnitude seems to successfully erase this
what are the real reasons for the appearance of these in- inherent anisotropy, giving a vertical effective stress path for
clined initial slopes of effective stress paths instead of the normally consolidated samples. The key question is as follows:
expected vertical stress paths? why the reappearance of this inherent anisotropy by isotropic
overconsolidation? Further tests need to be done to elucidate
(2) What is the explanation for the single highly nonlinear this strange feature. Advanced imaging techniques can be used
curve followed by all effective stress paths for a given con- to follow the microstructure evolution at different stages of over-
solidation ratio K, irrespective of the OCR? Can a physical consolidation (Frost et al. 1999).
meaning be attributed to this curve?
This previous analysis, based on elasticity theory, can be ap-
Concerning the unexpected initial inclination of the effective plied to anisotropically overconsolidated samples. A positive
stress paths of the isotropic overconsolidated series, contrasting deviatoric stress of an anisotropic consolidation gives a ver-
the initial vertical slope of the isotropic normally consolidated tical stress greater than the horizontal stress. Consequently, a
series, as reported in previous work (Doanh et al. 1997), the larger positive deviatoric stress or a lower consolidation ra-
current elastoplasticity theory can offer a plausible explana- tio K creates a smaller initial gradient or a larger ratio of axial
tion. In Fig. 9, isotropic consolidation pushes the yield surface Young’s modulus to radialYoung’s modulus Ea /Er , confirming
f outward, and an isotropic unloading left the actual yield sur- qualitatively the measured results in Table 3. A single constant
face fmax associated with the maximum consolidation pressure elastic stiffness, however, cannot correctly describe the highly
unchanged. Inside fmax , the strain during the first part of the nonlinear common curve of experimental results. A nonlinear
reloading is considered to be elastic, and it can be conveniently anisotropic elasticity is required, at least to simulate the strong
modelled by isotropic elasticity. As a result, from an initial curvature of the observed effective stress paths.
stationary isotropic stress point, the effective stress path rises
For each consolidation ratio K, a single initial gradient is
almost vertically with respect to the effective mean pressure,
measured, irrespective of the applied OCR, up to 24. Thus the
since this pressure remains constant during the initial increment
inherent anisotropy developed at the end of the K consolidation
of deformation. This probable evolution of the effective stress
is left unchanged during the unloading, and the common non-
path is shown in Fig. 9 as path A. Anisotropic elasticity offers
linear curve followed by all effective stress paths for a given K
another possible interpretation of these unexpected initial gradi-
is probably the pseudolinear elastic response of the overconsol-
ents. Path B in Fig. 9 portrays this alternative interpretation with
idated sand. A low consolidation ratio K accentuates the non-
greater stiffness in the vertical direction (Graham and Houlsby
linearity and results in the response curving strongly towards
1983). For the first series, the initial slope with a strong incli-
the positive p
axis.
nation towards the positive p
axis simply means a surprising

© 2006 NRC Canada

1208 Can. Geotech. J. Vol. 43, 2006

3.3.2. Undrained strength without a peak in the stress–strain relationship. Due to the di-
Table 1 gives the undrained strength, mobilized frictional latancy effect beyond the PT line, the effective stress path of a
angle φpeak , and effective stress ratio at peak ηpeak for all over- loosely overconsolidated sample moves up the failure surface
consolidated tests and permits the analysis of the influences of like that for a dense sand. Consequently, the void ratio alone is
the drained preshear history on the undrained behaviour. As not sufficient to control the undrained behaviour of sand.
noticed earlier, the undrained strength is remarkably enhanced The previous experimental results suggest that negative K
by overconsolidation. To facilitate interpretation, the undrained consolidation can produce the same three stages of undrained
strength is plotted in Fig. 10. A simple linear correlation be- anisotropy in extension and an inclined initial gradient of the
tween the undrained strength and the OCR is found: effective stress path towards the negative p
axis. Since it is
more difficult to conduct the K consolidation on the extension
[2] qpeak = a + b log(c × OCR) side, no further attempt has been conducted.
Best-fit curves with all experimental data points from each K-
3.3.4. Pseudoelastic response
consolidation series are constructed with excellent coefficients
Experimental investigations show a common pseudoelastic
of correlation (greater than 0.97 in compression for all cases,
response up to qpeak for each K-overconsolidation history, in-
and only 0.70 in extension). The analytical responses are plot-
dependent of the OCR attained. The observed pseudoelastic do-
ted using broken lines in Fig. 10, indicating a dependence of the
main can be as large as the entire past maximum yield surface.
coefficients a, b, and c on the consolidation ratio K. An extrap-
Inside the current mainstream of the nonassociated elastoplas-
olation of this model gives an undrained compressive strength
ticity framework, the usual incremental isotropic hypoelastic-
of about 149 kPa for a sample associated with an OCR of 24
ity cannot be used to describe the progressive increase in pore
on the consolidation line of K = 0.50. A complementary test
pressure during the elastic phase, since the theoretical effective
(as indicated in Fig. 10 by the symbol on the K = 0.50 line at
mean pressure increment remains unchanged in undrained load-
25 kPa) corresponding to the case with a consolidation pressure
ing.An alternate isotropic hyperelastic component, proposed by
of only 25 kPa is performed to verify this hypothesis. Thus even
Lade and Nelson (1987), is chosen for its capability to generate
in the absence of additional data, the present results suggest the
the needed pore pressure during the elastic phase. The analytical
use of this correlation equation (i.e., eq. [2]) to approximate
equation of the pseudoelastic response in the (q, p
) plane can
roughly the qpeak with OCR for practical purposes.
be found in Doanh et al. 2 using the incompressibility condition:
3.3.3. Three stages of undrained anisotropy   2  2 m
3p q
As discussed earlier, the undrained response of anisotropic + R
 pa 3 pa 
overconsolidated sand changes with changes in K and OCR. [3] p
= po
  2  2 
3po q
Three different zones, in terms of K and OCR, of undrained pa + R3 poa
anisotropic behaviour can be distinguished in compression. The
first zone is defined by complete unstable behaviour with de- where p
is the effective mean pressure, po
is the initial pres-
creasing undrained strength from a pronounced peak of the de- sure, R is a dimensionless intermediate constitutive parame-
viatoric stress, and continuous increase of pore pressure under ter, R = 6(1 + ν)/(1 − 2ν), m is a dimensionless constitutive
undrained shearing. Liquefaction or partial liquefaction occurs parameter, pa is the reference atmospheric pressure, and ν is
in this zone. The undrained behaviour of the isotropic overcon- Poisson’s ratio assumed constant. The hyperelastic response
solidated series is in this first zone. Inside the second zone, the (eq. [3]) is completely determined by the initial effective stress
undrained behaviour temporarily becomes rather unstable. A state (qo , po
) before undrained shearing.
temporary drop in the undrained strength accompanied by rising Equation [3] represents a symmetric conical surface with
pore pressure is observed, and then a tendency to dilate begins respect to the hydrostatic axis. For a fixed value of K other than
to appear. A complex combination of contractive and dilative 1, eq. [3] gives an initial inclined slope independent of the OCR
behaviour is observed in this intermediate zone. The undrained level and a common curved elastic response for all effective
behaviour of medium anisotropic overconsolidated series KB stress paths. As expected, this isotropic hyperelastic component
represents this zone. The third zone indicates a continuous in- always generates a vertical initial gradient of the effective stress
crease in the undrained strength accompanied by a decrease in paths for all initial isotropic stress states, regardless of the OCR
pore pressure. This stable behaviour is usually associated with level.
dense sand with a strong dilatancy without softening. A small This simple isotropic hyperelastic component replaces the
initial contractancy is also observed. High consolidation ratio isotropic elasticity of a updated nonassociated elastoplastic-
K and OCR create this behaviour. ity model with a mixed kinematic hardening mechanism. The
This scheme links the well-known undrained behaviour of original model with isotropic elasticity is described in Dubujet
loose, medium, and dense sand. A similar change of dense sand and Doanh (1997). In the updated model, an isotropic mecha-
evolving from stable behaviour at low confining pressures to nism with a large range of the pseudoelastic domain is associ-
unstable behaviour at very high pressures through temporary ated with a kinematic deviatoric mechanism taking into account
instability at medium confining pressures is observed by Lade the presheared history. Figures 11a and 11b give examples of
et al. (1988), and a refined version is reported by Yamamuro the theoretical response in the case of KB (OCR = 2) and KC
and Lade (1997). K overconsolidation can progressively trans- (OCR = 4) overconsolidation.Although this simple hyperelastic
form the undrained response of a loose sand sample to that
of a dense sand while in a relatively loose state. The last test, 2
Doanh, T., Finge, Z., Boucq, S., and Dubujet, Ph. Induced anisotropy
KC-CO12, shows a strong tendency to dilate at large strains of Hostun RF loose sand: another missing parameter. In preparation.

© 2006 NRC Canada

Finge et al. 1209

Fig. 10. Undrained strength of loosely overconsolidated samples.

Fig. 11. Theoretical and experimental responses. (a) KB overconsolidation. (b) KC overconsolidation.

© 2006 NRC Canada

1210 Can. Geotech. J. Vol. 43, 2006

model without any reference to anisotropy gives a correct pseu- static liquefaction for a low OCR and high K ratio to com-
doelastic response when compared with that from experimental plete stability for a high OCR and low K, while being in a
data, the complete elastoplastic model fails to simulate many relatively loose state before the final monotonic undrained
other experimental features in this study, such as the plastic shearing. Temporary instability was observed for a medium
deformation, the progressive appearance of dilatancy, and the OCR and medium K.
evolution of undrained behaviour.
(5) The observed common pseudoelastic response can be
adequately simulated by simple isotropic hyperelasticy,
4. Conclusions without reference to anisotropy, within the nonassociated
elastoplasticity framework.
The undrained anisotropic behaviour of loose Hostun RF
sand was investigated in a series of triaxial compression and ex-
tension tests from three initial stress states, isotropic and highly 5. Acknowledgement
anisotropic, in the triaxial plane. Loose Hostun RF sand sam-
ples prepared by the moist-tamping technique were overcon- Financial support to the first author provided by the French
solidated either isotropically or anisotropically along several Educational Ministry of Research and Technology (MENRT)
constant effective stress ratio paths with various overconsoli- is fully acknowledged.
dation ratios, up to OCR = 24, to study the effects of induced
anisotropy created by simple linear stress path history on the
undrained anisotropic behaviour. From the observed results, a References
few conclusions concerning the effects of previous deviatoric Arthur, J.R.F., and Menzies, B. 1972. Inherent anisotropy in a sand.
strain histories are summarized as follows: Géotechnique, 22(2): 115–128.
Arthur, J.R.F., Chua, K.S., and Dunstan, T. 1977. Induced anisotropy
(1) Partial static liquefaction, usually associated with loose in a sand. Géotechnique, 27(1): 13–30.
sand, occurs for all isotropic overconsolidated samples in Arthur, J.R.F., Chua, K.S., Dunstan, T., and Rodriguez Del, J.I. 1980.
compression and extension in this study. Up to OCR = 8, Principal stress rotation: a missing parameter. Journal of Geotech-
isotropic overconsolidation markedly enhances the nical Engineering, ASCE, 106(4): 419–433.
undrained shear strength. Deviatoric stress peaks qpeak oc- Been, K., and Jefferies, M.G. 1985. A state parameter for sands.
cur at relatively small axial strains and, for the same OCR, Géotechnique, 35(2): 99–112.
the deviatoric stress peak in compression is considerably Bjerrum, L., Krimgstad, S., and Kummeneje, O. 1961. The shear
larger than that in extension. A linear correlation between strength of a fine sand. In Proceedings of the 5th International
qpeak and OCR on a logarithmic scale was found, with dif- Conference on Soil Mechanics and Foundation Engineering, Paris,
ferent coefficients for each anisotropic consolidation level France. Vol. 1, pp. 29–37.
K. This correlation can be used to roughly estimate the Canou, J., Thorel, L., and De Laure, E. 1991. Influence d’un déviateur
undrained strength. de contrainte initial sur les caractéristiques de liquéfaction statique
du sable. In Proceedings of the 10th European Conference on Soil
(2) As the OCR increases, the effective stress paths of
Mechanics and Foundation Engineering, Firenze, Italy, 26–30 May
isotropic overconsolidated samples enlarge with an asym-
1991. A.A. Balkema, Rotterdam, The Netherlands. Vol. 1, pp. 49–
metric shape with respect to the isotropic axis. In contrast,
52.
anisotropic overconsolidation shifts the effective stress
Casagrande, A. 1975. Liquefaction and cyclic deformation of sands: a
paths towards the direction of preloading. A pronounced
critical review. Harvard Soil Mechanics Series, Harvard University,
bulge in the effective stress path with very sharp bend is
Cambridge, Mass.
observed in the same direction of preloading (compres-
Castro G. 1969. Liquefaction of sands. Ph.D. thesis, Harvard Univer-
sion), whereas that in the opposite direction of preloading
sity, Cambridge, Mass.
(extension) is left nearly unchanged.
Di Prisco, C., Matiotti, R., and Nova, R. 1995. Theorical investigation
(3) In compression and in extension, the initial part of the ef- of the undrained stability of shallow submerged slopes. Géotech-
fective stress paths of all anisotropic overconsolidated sam- nique, 45(3): 479–496.
ples follows a common pseudoelastic response up to qpeak . Doanh, T., and Ibraim, E. 2000. Minimum undrained strength of Hos-
The initial gradient of the effective stress path seems to tun RF. Géotechnique, 50(4): 377–392.
relate solely to the direction of the previous stress history. Doanh, T., Ibraim, E., and Matiotti, R. 1997. Undrained instability of
Using anisotropic elasticity, these initial gradients can be very loose Hostun sand in triaxial compression and extension. Part 1:
explained in part because of induced anisotropy created by Experimental observations. Mechanics of Cohesive–Frictional Ma-
the overconsolidation. For isotropic overconsolidation, the terials, 2(1): 47–70.
initial slope of the effective stress paths towards the posi- Dubujet, Ph., and Doanh, T. 1997. Undrained instability of very loose
tive p
axis contradicts the popular concept of increasing Hostun sand in triaxial compression and extension. Part 2: Theoreti-
initial isotropy by isotropic overconsolidation. cal analysis using an elastoplasticity model. Mechanics of Cohesive-
Frictional Materials, 2(1): 71–92.
(4) Depending on the anisotropic consolidation ratio K and the Finge, Z. 2004. Contribution à l’étude du comportement non drainé
OCR reached during the overconsolidation, the des sables lâches surconsolidés et anisotropes. Thèse de doctorat,
undrained compression behaviour changes from partial Institut National des Sciences Appliquées de Lyon, Lyon, France.

© 2006 NRC Canada

Finge et al. 1211

Flavigny, E., Desrues, J., and Palayer, B. 1990. Le sable Hostun RF. Matiotti, R. 1996. Analisi sperimentale del fenomeno della lique-
Revue Française de Géotechnique, 53: 67–69. fazione statica delle sabbie sciolte e sue applicazioni numeriche.
Frost, J.D., and Park, J.Y. 2003. A critical assessment of the moist- Ph.D. thesis, Politechnico di Milano, Milan, Italy.
tamping technique. Geotechnical Testing Journal, 26(1): 57–70. Matiotti, R., Ibraim, E., and Doanh, T. 1995. Undrained behavior of
Frost, J.D., Chen, C.C., Park, J.Y., and Jang, D.J. 1999. Quantitative very loose RF sand in compression and extension tests. In Proceed-
characterisation of microstructure evolution. In Physics and me- ings of the 5th International Symposium on Numerical Models in
chanics of soil liquefaction. Edited by Y. Lade and J. Yamamuro. Geomechanics. Edited by G.N. Pande and S. Pietruszczak. A.A.
A.A. Balkema, Rotterdam, The Netherlands. pp. 169–177. Balkema, Rotterdam, The Netherlands. pp. 119–124.
Gajo, A., and Piffer, L. 1999. The effects of preloading history on the Meghachou, M. 1992. Stabilité des sables lâches : essais et modéli-
undrained behaviour of saturated loose sand. Soils and Foundations, sation. Thèse de doctorat, Université scientifique et médicale de
39(6): 43–54. Grenoble, Grenoble, France.
Georgiannou, V.N., Burland, J.B., and Hight, D.W. 1990. The Poorooshasb, H.B., Holubec, I., and Sherbourne, A.N. 1966. Yielding
undrained behaviour of clayey sands in triaxial compression and and flow of sand in triaxial compression: Part 1. Canadian Geotech-
extension. Géotechnique, 40(3): 431–449. nical Journal, 3: 179–190.
Graham, J., and Houlsby, G. T. 1983. Anisotropic elasticity of a natural Poorooshasb, H.B., Holubec, I., and Sherbourne, A.N. 1967. Yielding
clay. Géotechnique, 33(2): 165–180. and flow of sand in triaxial compression: Parts 2 and 3. Canadian
Geotechnical Journal, 4: 376–397.
Hight, D.W., Gens, A., and Symes, M.J. 1983. The development of a
Poulos, S.J. 1981. The steady state of deformation. Journal of Geotech-
new hollow cylinder for investigating the effects of principal stress
nical Engineering, ASCE, 107(5): 553–562.
rotation in soils. Géotechnique, 33(4): 355–384.
Sladen, J.A., D’Hollander, R.D., and Krahn, J. 1985. The liquefac-
Hill, J. 1958. A general theory of uniqueness and stability in elas-
tion of sands, a collapse surface approach. Canadian Geotechnical
tic plastic solids. Journal of the Mechanics and Physics of Solids,
Journal, 22: 564–578.
6: 236–249.
Tatsuoka, F., and Ishihara, K. 1974. Yielding of sand in triaxial com-
Ibraim, E. 1998. Différents aspects du comportement des sables à partir pression. Soils and Foundations, 14(2): 63–76.
des essais triaxiaux: des petites déformations à la liquéfaction sta- Tatsuoka, F., Sonoda, S., Hara, K., Fukushima, S., and Pradhan, T.B.S.
tique. Thèse de doctorat, Institut National des Sciences Appliquées 1986. Failure and deformation of sand in torsional shear. Soils and
de Lyon, Lyon, France. Foundations, 26(4): 79–97.
Ishihara, K., and Okada, S. 1978a. Effects of stress history on cyclic Vaid, Y.P., Sivathayalan, S., and Stedman, D. 1999. Influences of
behaviour of sand. Soils and Foundations, 18(4): 31–45. specimen-reconstituting method on the undrained response of sand.
Ishihara, K., and Okada, S. 1978b. Yielding of overconsolidated sand Geotechnical Testing Journal, 22(3): 187–195.
and liquefaction model under cyclic stresses. Soils and Foundations, Verdugo, R., and Ishihara, K. 1996. The steady state of sandy soils.
18(1): 57–72. Soils and Foundations, 36(2): 81–91.
Ishihara, K., and Okada, S. 1982. Effects of large preshearing on cyclic Yamamuro, J.A., and Lade P.V. 1997. Static liquefaction of very loose
behaviour of sand. Soils and Foundations, 22(3): 109–125. sands. Canadian Geotechnical Journal, 34(6): 905–917.
Kato, S., Ishihara, K., and Towhata, I. 2001. Undrained shear char- Zdravković, L., and Jardine, R.J. 2000. Undrained anisotropy of Ko -
acteristics of saturated sand under anisotropic consolidation. Soils consolidated silt. Canadian Geotechnical Journal, 37(1): 178–200.
and Foundations, 41(1): 1–11.
Konrad, J. M. 1993. Undrained response of loosely compacted sands
during monotonic and cyclic compression tests. Géotechnique, List of symbols
43(1): 69–89. Dr relative density
Konrad, J. M., Flavigny, E., and Megachou, M. 1991. Comportement D50 median particle size
non drainé du sable d’Hostun lâche. Revue Française de Géotech- e void ratio
nique, 54: 53–63. ec void ratio before undrained shearing
Kramer, S.L., and Seed, H.B. 1988. Initiation of soil liquefaction un- emax maximum void ratio
der static loading conditions. Journal of Geotechnical Engineering, emin minimum void ratio
ASCE, 114(4): 412–430. e0 initial void ratio at fabrication stage
Kurukulasuriya, L.C., Oda, M., and Kazama, H. 1999. Anisotropy of Ea axial Young’s modulus
undrained shear strength of an overconsolidated soil by triaxial and Er radial Young’s modulus
plane strain tests. Soils and Foundations, 39(1): 21–29. f yield surface
Ladd, R.S. 1978. Preparing test specimens using undercompaction. fmax yield surface assocaiated with the maximum
Geotechnical Testing Journal, 1(1): 16–23. consolidation pressure
Lade, P.V. and Nelson, R.B. 1987. Modelling the elastic behaviour of Gs specific gravity
granular materials. International Journal for Numerical and Analyt- K consolidation stress ratio σr
/σa

ical Methods in Geomechanics. 11(5): 521–542. Ko lateral earth pressure at rest
Lade, P.V., Nelson, R.B., and Ito,Y.M. 1988. Instability of granular ma- m dimensionless constitutive parameter


terials with nonassociated flow. Journal of Engineering Mechanics, OCR overconsolidation ratio pmax /pc

ASCE, 114(12): 2173–2191. p
effective mean pressure (σa + 2σr
)/3

Luong, M.P. 1978. Etat caractéristiques du sol. Compte Rendu de pa reference atmospheric pressure
l’Académie des Sciences, Série B, 287: 305–307.

© 2006 NRC Canada

1212 Can. Geotech. J. Vol. 43, 2006

pc
effective mean pressure before undrained uc uniformity coefficient

shearing umax maximum pore pressure generation


pmax maximum effective mean pressure during the εa axial strain
isotropic overconsolidation εmax maximum axial strain attained during the
po
initial pressure drained preshear


pPT effective mean pressure at phase εpeak axial strain at peak
transformation εPT axial strain at phase transformation
q deviatoric stress σa
− σr
εv volumetric strain
qc deviatoric stress at the beginning of the final φpeak mobilized friction angle at deviatoric stress
undrained shearing peak
qmax maximum deviatoric stress attained during the φPT mobilized friction angle at phase
drained preshear transformation
qpeak deviatoric stress at peak ηpeak effective stress ratio q/p
at deviatoric stress
qu mobilized undrained shear strength of peak
anisotropic specimens ν Poisson’s ratio
R dimensionless intermediate constitutive σa
effective axial stress
parameter σr
effective radial stress
u pore pressure generation

© 2006 NRC Canada