Вы находитесь на странице: 1из 8

ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

ICSE6-28

Internal stability of granular materials in triaxial tests

Livah ANDRIANATREHINA1, Hanne SOULI1, Jean-Jacques FRY2, Quoc Tuan PHAN3,


Jean-Marie FLEUREAU3

1
Laboratoire de Tribologie et Dynamique des Systmes, CNRS UMR 5513,
Ecole Nationale dIngnieurs de Saint Etienne,
58 rue Jean Parot, 42023 Saint Etienne Cedex - e-mail : milivah@yahoo.fr; hanene.souli@enise.fr

2
Electricit de France, Centre dIngnirie Hydraulique,
73073 Le Bourget du Lac cedex e-mail : jean-jacques.fry@edf.fr

3
Laboratoire de mcanique des Sols, Structures et Matriaux, CNRS UMR 8579, Ecole Centrale Paris,
Grande voie des vignes, 92295 Chtenay-Malabry, France
e-mail : tuandhxd1986@gmail.com; jean-marie.fleureau@ecp.fr

Abstract: Instability of granular materials may result in drastic change in their mechanical properties. This instability
may be triggered by hydraulic or mechanical loading. In dams, it may cause an heterogeneity of the material and leads
to global instability of a part of the work.
The paper presents the results of an experimental research on this phenomenon. The test materials were made from an
alluvial soil, with a maximum grain diameter of 16 mm. Nine grain size distributions, classified as stable or
unstable according to Kenney & Lau (1985) criteria, were defined. Triaxial tests were carried out on samples of
these soils under 3 effective confining stresses and 2 relative densities. After each triaxial test and after the
measurement of the maximum density, the grain size distribution of the specimens was measured, in order to analyse
the movement of particles within the material. An index was defined to characterise the stability of the material, and
compared to several criteria proposed in the literature (Kenney & Lau, Kezdi, Terzaghi, Burenkova..). The results
highlight the limitations of present criteria and the need to better understand the mechanisms involved in the movement
of the particles within the soil, especially when considering severe conditions like those imposed during a triaxial test.

Key words : internal stability, triaxial tests, stability criteria, soil

I INTRODUCTION
Instability of granular materials may result in drastic changes in their mechanical properties. Instability
may be triggered by hydraulic or mechanical loading. In dams, it may cause an heterogeneity of the material
and lead to global instability of a part of the hydraulic structure. An interesting method to assess the stability
of a granular soil was presented by Kenney & Lau in 1985 and slightly modified in 1986, based on the
assumption that each class of grain, characterized by its diameter d must be prevented to move by the grains
whose diameter ranges between d and 4d.
In this paper, we compare the soil stability of nine samples with different grain size distribution curves,
derived from direct grain size measurements carried out after triaxial and vibration tests to that predicted by
several methods (Kenney & Lau, Kezdi, Terzaghi, USACE, Burenkova). The results of the triaxial tests
themselves will not be presented in the paper.

II MATERIALS
The test materials were made from an alluvial deposit from the Rhne river in France, with a maximum
grain diameter of 16 mm. Nine grain size distributions, classified as stable or unstable according to
Kenney & Lau (1985) criteria, were defined to carry out the tests. One of the main parameter of these
distributions is the percentage of grains smaller than 0.315 mm which appears in the name of the samples;

911
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

this diameter is considered as the limit of the fine grains. The properties of the materials are shown in Figs. 1
and 2 and summarized in Table 1. After the determination of the standard maximum density, the samples
were separated into 3 parts corresponding to the lower, intermediate and upper part of the specimen. Grain
size analysis was performed on each part in order to estimate the migration of fines caused by vibration
under stress.

100

80
Passing (%)

60

C5
40
C10
C20
20 C30
C7
0
100 1000 10000 100000
Grain diameter (m)

Fig. 1: Continuous grain size distributions

100

80
Passing (%)

60
D5
D10
40
D20
D30
20

0
100 1000 10000 100000
Grain diameter (m)

Fig. 2: Discontinuous grain size distributions

< 0.315 mm d60 d10


Name dmax CU emax emin
% mm mm
C5 5 8.5 0.62 13.7 0.510 0.332
C7 7 7.5 0.8 9.4 0.499 0.331
C10 continuous 10 6.2 0.3 20.7 0.400 0.226
C20 20 4.2 0.22 19.1 0.450 0.271
C30 30 1.7 0.2 8.5 0.501 0.346
16 mm
D5 5 9 3.5 2.6 0.582 0.430
0.315
D10 dis- 10 9 2.9 - 29 0.519 0.355
3.15
continuous
D20 20 8.2 0.23 35.7 0.406 0.244
D30 30 7.1 0.2 35.5 0.373 0.248

Table 1: Properties of the tested grain size distributions

912
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

III EXPERIMENTAL PROCEDURE OF THE TESTS


Triaxial tests were performed on the nine soil samples, compacted to 3 different initial relative densities
(dense: Dr = 90%, medium: Dr = 50% and loose: Dr = 30%), under 3 effective confining stresses (50, 100
and 200 kPa). The procedure for the preparation of the specimens is as follows:
(a) the dry soil was mixed with a variable percentage of water (from 0 to 4%), according to its fine
content and the density to which it was to be compacted; the role of water was to prevent the soil to
reach too high density during its laying out;
(b) a precise mass of wet soil was then carefully deposited in a cylindrical mold placed on the pedestal of
the triaxial cell, on which a rubber membrane was tended, with 10 cm inner diameter and 17.5 cm
height and compacted to a definite volume corresponding to the chosen dry relative density. The
same operation was repeated 5 times in order to carry out compaction in 5 homogenous layers with
the same dry density.
(c) once the mold was filled with compacted soil, the upper cap was placed on the specimen and the
membrane was closed and sealed around the lower and upper ends using torical seals.
The triaxial tests procedure consists in :
(d) saturating the sample in two steps by: (1) circulating water from the bottom to the top under 1.5 m
hydraulic head to eliminate the main air bubbles and saturating the circuits; (2) simultaneously and
continuously increasing the confining stress and the pore-water pressure from initial values 3 = 35
kPa and u = 15 kPa to final values 3 = 620 kPa and u = 600 kPa, respectively. The objective of this
second step is to dissolve the air according to Henry law;
(e) checking the value of Skempton coefficient B (= u/3) by increasing the confining pressure from
620 to 670 kPa under undrained conditions. In all the tests, B values larger than 0.95 were obtained,
confirming the good saturation of the samples;
(f) consolidating the sample under effective confining stresses ranging between 50 and 200 kPa. This is
done by increasing the confining stress while maintaining the pore pressure constant. This conditions
were chosen to shorten the tests and to be near to the in situ conditions.
(g) shearing the sample under constant strain rate (generally lower than 1% per hour), constant total
confining stress and mostly undrained conditions, up to 20 to 25% axial strain; the aim is to obtain
the failure of he sample.
(h) in the same way as for vibration tests, after each test, the samples were separated in 3 parts
corresponding to the lower, intermediate and upper third of the sample. Grain size analysis was
carried out on each part in order to estimate the migration of the fine particles during the triaxial test.

IV INTERNAL STABILITY OF THE MATERIALS ACCORDING TO VARIOUS CRITERIA


The studied soils, in which the coarser grains can be considered as supporting the structure and the finer
grains as filling the voids to a different extent, may be internally unstable if the granular material is gap-
graded or if the finer soil particles are smaller than constrictions in the coarser particles [Rnnqvist, 2010].
Various criteria have been proposed in the literature to estimate the internal stability of a soil, based on its
grain size distribution curve, several of which will be examined in this paper and confronted to the
experimental results.
The method of [Kenney & Lau, 1985, 1986] consists in plotting the grain size distribution curve of a soil in
a diagram with the mass fraction F of particles whose diameter is smaller than d on the X-axis and the mass
fraction H of the particles whose diameter ranges between d and 4d on the Y-axis. The soil will be
considered as stable if, for F < 20 or 30%, the curve thus obtained is located below the line H = F. On the
contrary, if some portion of this curve passes above the line H = F, the soil will be considered as unstable.
According to Kenney & Lau, the values of 20% applies to the widely graded (WG) soils in the range 0.2 < F
< 1 and the value of 30%, to the narrowly graded (NG) soils in the range 0.3 < F < 1. Kenney & Lau
criterion also includes threshold values of uniformity coefficient Cu, not evocated here. According to the
classification of Kenney & Lau, the 5 samples with continuous grading (samples C5 to C30) are stable and
all the 4 samples with discontinuous grading (samples D5 to D30) are unstable. The results are shown in Fig.
3.

913
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

100
100
C5
D5
C10
80 D10
80 C20
D20
C30
D30
Stable C7 60
60 materials Line H=F (K&L)

H
Line H=F (K&L)
Stable
H

materials
40 40 NG
NG
WG Unstable
WG Unstable materials (K&L)
materials (K&L) 20 H=15 (Kezdi)
20
Unstable Unstable
materials (Kezdi) materials (Kezdi)
0 0
0 20 40 60 80 100 0 20 40 60 80 100
F F

Fig. 3: Kenney & Lau representation of the grain size distribution curves to estimate the soil stability (left :
continuous gradings ; right : discontinuous gradings). K&L stands for Kenney & Lau (1986)

[Li and Fanning, 2008] suggested that combining the methods of [Kenney & Lau, 1986] and [Kezdi, 1979]
yielded a more reliable estimation of soil stability. In Kezdi method, a soil is considered as stable if the
parameter H is smaller than 15, unstable above this value. In fact, Kezdi method consists in dividing the
material gradation curve into a coarse fraction and a fine fraction at a point on the curve; in effect, the fine
fraction works as the soil, and the coarse fraction works as the filter for the soil. Adding this condition to that
of Kenney & Lau (Fig. 3) shows that 3 out of the 5 potentially stable soils of Kenney & Lau should be in fact
unstable: these are the soils with continuous grading curves and the smallest percentages of grains smaller
than 0.315 mm (i.e. soils C5, C7 and C10).
Another way of analyzing the internal stability of a material is to divide it into 2 fractions, as in Kezdi
method and checking if Terzaghi filter rule, or the rule proposed by [USACE, 1953], are verified. This was
done for all the studied materials, considering 1 mm as the boundary diameter between coarse and fine
grains, as shown in Fig. 4.

100

80
Passing (%)

60

40

C10
20 Fraction > 1 mm
Fraction < 1 mm

0
0.1 1 10 100
Diameter of grains (mm)

Fig. 4: Division of the initial material into a fine soil and a coarse soil

Terzaghi rule is expressed by the double inequation :


d15(f) d
4 or 5 15(f) 20
d 85(s) d15(s)
where the index (s) is related to the fine soil and the index (f), to the coarser material of the filter.

914
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

30 D20
D30 D10
D5
Unstable
C7
20

d15(f)/d15(s)
C5
C10
Stable C30
10
C20

0
0 4 8 12 16
d15(f)/d85(s)

Fig. 5: Stability of the different soils according to Terzaghis criterion

d 50( f )
On the other hand, the rule of the US Corps of Engineers states that : 25
d 50(s )
The results for the different studied soils, in relation to Terzaghi and USACE criteria, are shown in Fig. 5.
According to Terzaghi criterion, most of the soils with continuous grading are stable, except C7 which is on
the boundary of the stable domain, and all the soils with discontinuous grading are unstable. The same
conclusion can be derived from the USACE criterion: for most of the soils with continuous grading (C5,
C10, C20 and C30), the value of the d50(f)/d50(s) ratio is smaller than 25, its value is equal to 27 in the case of
C7 and, for all the soils with discontinuous grading, the value of the ratio exceeds 37.
Another criteria, which is based on 2 ratios of grain diameters, is that of [Burenkova, 1993] where a soil is
considered as internally stable if the following double inequality is verified
0.76 log(h) + 1<h<1.86 log(h)+1, where h= d90/d60 and h = d90/d15
The results for the different soils is shown in Fig. 6. Considering this criterion, 4 of the soils are considered
as unstable (C7, C30, D20, D30), 4 others are uncertain as their representative point lies on the boundary of
the stable domain (D5, D10, C5, C10) and only 1 is stable (C20). This result is quite different from those
given by previous criteria.

7
C30
6

5
h' = d90/d60

Unstable
4
C20
3
Internally stable
C5
2 D5
C10 D30
D10 D20
1 C7
Unstable
0
1 10 100
h" = d90/d15

Fig. 6: Burenkova stability criterion for the 9 studied soils

915
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

V DIRECT ASSESSMENT OF GRAIN STABILITY AFTER TRIAXIAL AND VIBRATION


TESTS
As an example, the results of the measurements after the triaxial tests, for the tests carried out at a relative
density of 50% and a confining stress of 100 kPa, are shown in Fig. 7 for the 9 grain size distributions. The
tests performed at relative densities of 30 and 90%, as well as those done under confining stresses of 50 and
200 kPa are not shown here. Neither are shown the results of the tests after a vibration phase.

100 100 100


initial initial initial
triax_upper triax_upper triax_upper
80 triax_medium 80 triax_medium 80 triax_medium
triax_lower triax_lower triax_lower
triax_mean triax_mean triax_mean

Passing (%)
Passing (%)

Passing (%)
60 60 60

40 40 40

20 20 20
C5 C20 D10
0 0 0
0.1 1 10 100 0.1 1 10 100 0.1 1 10 100
Grain diameter (mm) Grain diameter (mm) Grain diameter (mm)

100 100 100


initial initial
initial
triax_upper triax_upper
triax_upper
80 triax_medium 80 triax_medium 80
triax_medium
triax_lower triax_lower
triax_lower
triax_mean triax_mean
Passing (%)

Passing (%)

Passing (%)
60 60 60

40 40 40

20 20 20
C7 C30 D20
0 0 0
0.1 1 10 100 0.1 1 10 100 0.1 1 10 100
Grain diameter (mm) Grain diameter (mm) Grain diameter (mm)

100 100 100


initial initial
triax_upper triax_upper initial
triax_upper
80 triax_medium 80 triax_medium 80
triax_lower triax_lower triax_medium
triax_mean triax_mean triax_lower
Passing (%)

Passing (%)

Passing (%)

60 60 60

40 40 40

20 20 20
C10 D5 D30
0 0 0
0.1 1 10 100 0.1 1 10 100 0.1 1 10 100
Grain diameter (mm) Grain diameter (mm) Grain diameter (mm)

Fig. 7: Grain size distributions measured after the triaxial test at 3 = 100 kPa and Dr = 50% for the samples
with continuous grading (C5 to C30) and for the samples with discontinuous grading (D5 to D30)

Considering these results, there are in some cases a slight difference between the initial grain size
distribution and the mean distribution at the end of the test, which might be attributed to grain breaking.
To characterize the evolution of the grain size distribution for the 8 to 12 conditions considered for each
sample, a simple parameter was defined, which is the maximum distance (P, in %) between the upper and
lower grain size distribution curves obtained after the triaxial and vibration tests. Then, we considered the
mean value of this parameter for all the tests carried out on a given soil. The results are shown in Fig. 8.

916
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

Stability of soil is characterized by a low value of the parameter P (typically P < 5%) and instability by a
value of P larger than 10%. Intermediary values of this parameter, between 5 and 10%, may correspond to
soils with limit stability. Finally, 5 soils can be considered as unstable and 4 as stable, but the boundary
between stable and unstable samples appears quite different from the continuous/discontinuous grading
boundary.

Fig. 8: Direct measurement of the stability of the studied soils after triaxial and vibration tests

VI DISCUSSION AND CONCLUSION


The comparison between the results of the direct measurements and those of the stability criteria presented
in this paper is shown in Table 2. None of the criteria gives a correct answer for all the soils. The reason may
be that most of these criteria were established in conditions (e.g. downward or upward flow of water, or
small vibrations) in which the coarser fraction of the soil practically did not move and the only movement
was the transport by water of the finer particles through the large pores. The conditions of our tests were
much more severe as they involved the displacement of all the grains, which made it easier for the smaller
particles to move through the larger ones. Moreover, in these tests, two mechanisms were involved: the
effect of a water flow during saturation and consolidation of the samples and that of soil deformation during
shearing. These conditions are representative of those which may be found during the lifetime of a dam.
However, it appears that the association of Kenney & Lau and Kezdi criteria give a fairly good picture of the
internal stability of the soil. Indeed, there are only 2 differences between the direct measurements and those
of these two criteria, which are for the C10 and D30 samples. In the last case, even if the soil should be
unstable, the fact that the percentage of fine particles is large results in the fact that it is rather difficult for
them to move.

C5 C7 C10 C20 C30 D5 D10 D20 D30


Direct measurement U U S S S U U U S
Kenney & Lau (1985, 1986) S S S S S U U U U
Kenney & Lau (1985,1986)
U U U S S U U U U
+ Kezdi (1979)
Terzaghi, USACE (1953) S LS S S S U U U U
Burenkova (1993) LS U LS S U LS LS U U

Table 2: Comparison between the direct measurements of the internal stability of the studied samples and the
results derived from several stability criteria (U: unstable; S: stable; LS: limit stability)

917
ICSE6 Paris - August 27-31, 2012 - Andrianatrehina et al.

The main conclusion of this study is that the analysis of stability of a soil must take into account many
parameters which are often not considered in the criteria found in the literature. Among the parameters which
are liable to influence the stability, the percentage of fine particles (here, the percentage of grains smaller
than 0.315 mm) plays an important role. Generally, fine particles are defined as particles whose diameter is
smaller than 0.075 mm. In the studied dam materials with maximum particle diameters up to 10 cm or more,
the very fine particles (< 0.075 mm) are not plastic and their percentage is very small (always smaller than
2%), so that the value of 0.315 mm appears to be more significant in relation to grain size distribution and
soil behavior.

VII NOMENCLATURE
Cx, Dx: Name of sample with continuous (resp. discontinuous) grading, where x represents the percentage of
particles smaller than 0.315 mm
CU: uniformity coefficient (= d60/d10)
dy: diameter of grains corresponding to y% in mass of particles smaller than this diameter on the grain size
distribution curve; (f) related to filter, i.e. the coarser material; (s) related to soil, i.e. the finer material.
dmax: maximum grain diameter of a sample
emin, emax: minimum and maximum void ratios derived from standard ASTM tests
F: Percentage in mass of particles whose diameter is smaller than the diameter d
H: Percentage in mass of particles whose diameter is comprised between d and 4d
3, u: Total confining stress and pore pressure in a triaxial test

VIII REFERENCES AND CITATIONS


Burenkova V.V. (1993). - Assessment of suffusion in non-cohesive and graded soils. Proc. of the First
International Conference Geo-Filters, Filters in Geotechnical Engineering. Brauns, Heibum & Schuler
(eds), Balkema. p. 357-360
Kenney T.C. & Lau .D. (1985). - Internal stability of granular filters, Canadian Geotechnical Journal, 22:
215-225.
Kenney T.C. & Lau D. (1986). - Internal stability of granular filters: reply, Canadian Geotechnical
Journal, 23: 420-423.
Kezdi A. (1979). - Soil physics selected topics. Elsevier Scientific Publishing Co., Amsterdam, 160 p.
Li M. & Fannin R.J. (2008). - Comparison of two criteria for internal stability of granular soil. Canadian
Geotechnical Journal. 45: 1303-1309.
Rnnqvist H. (2010). - Predicting surfacing internal erosion in moraine core dams. Licentiate thesis,
Royal Institute of Technology KTH, Stockholm, Sweden, TRITA-LWR.
USACE. (1953). - Filter Experiments and Design Criteria. Technical Memorandum No. 3-360. Waterways
Experiment Station, Vicksburg.

918