Professional Documents
Culture Documents
An International Journal
To cite this article: Adel Hanna & Tahar Ayadat (2019): Comparative study of shear strength
characteristics of dry cohesionless sands from triaxial, plane-strain and direct shear tests,
Geomechanics and Geoengineering, DOI: 10.1080/17486025.2019.1648882
Article views: 7
dry sands out using triaxial and plane-strain apparatus. This paper presents a comparative study of shear
The results were used to evaluate Rowe’s stress-dila- strength characteristics of sands as obtained from triaxial,
tancy theory, the relationships between the shear plane-strain and direct shear test results. Furthermore,
strength components due to friction and interlocking, correlations between the angles of shearing resistance
and the volume-change characteristics for cases of obtained by these tests are also presented.
plane-strain and axisymmetric boundary conditions.
Alshibli et al. (2003) conducted an experimental pro-
gram to investigate the effects of grain shape, loading Experimental investigation
condition, and confining pressure on strength proper- In order to perform a comparative study of shear
ties and localization phenomena in sands. The beha- strength characteristics of dry sands from triaxial,
viour of the conventional triaxial compression plane-strain and direct shear tests, the results of the
experiments was compared with the results of plane experimental investigation conducted by Hanna and
strain experiments. Massoud (1981) and Hanna (2001) were used herein.
Likewise, some efforts were also made to correlate The results are analyzed in a different perspective in
these results with those of the direct shear test (among order to investigate the effect of soil relative density or
others, Nash 1953, Rowe 1969, Hanna et al. 1987, soil porosity on the angle of shearing resistance, dila-
Maccarini 1993, Castellanos and Brandon 2013, tancy factor, particles interlocking, and volumetric
Hwang et al. 2016). Castellanos and Brandon (2013) strain and axial strain at failure. Furthermore, the
showed that direct shear tests provide much lower results of this comparison will be used to develop
friction angles in alluvial deposits than triaxial tests. correlations connecting between the angle of shearing
A comparison of direct shear and triaxial test results resistance of triaxial condition test and the direct shear
conducted on undisturbed alluvial soil shows that the test, and between the angle of shearing resistance of
friction angle determined from the direct shear appa- plane-strain and triaxial compression conditions. The
ratus is normally about 2 to 5 degrees lower than that correlations developed to take into account the inter-
determined using the triaxial apparatus. According to action between soil relative density, dilatancy factor
them, this can be attributed to the anisotropic shear and particles interlocking.
strength characteristics of the alluvial soils. The differ- The tests of the experimental investigation were per-
ence in the results of the two test devices is much less formed on dry silica sand to determine the angle of
when remoulded test specimens are used. The shearing resistance for the sand under different loading
remoulding process destroys the anisotropic fabric, and boundary conditions. Table 1 presents the physical
and the shear strength parameters are not as dependent properties of the sand samples tested. Triaxial compres-
on the orientation of the failure plane. As results of sion tests were carried out in the conventional triaxial test
comparing two different testing apparatus, estimated apparatus on cylindrical specimens of 38 mm in diameter
values of shear strength parameters of unsaturated and 76 mm in length. The plane-strain test samples had
soil were slightly larger in the direct shear tests due to nominal dimensions of 92, 38 and 75 mm in width,
the constraint effect of shear box. Hwang et al. (2016) length, and height, respectively. The direct shear tests
compared the shear strength characteristics of an unsa- were performed on samples having dimensions of 60.5,
turated soil using triaxial compression tests and direct 60.5 and 20.5 mm (width, length and thickness, respec-
shear tests. Their results showed that the estimated tively). Test set-up, procedure and results for plane-
values of shear strength parameters of unsaturated strain, triaxial and shear box are documented by Hanna
soil were slightly larger in the direct shear tests due to and Massoud (1981) and Hanna (2001).
the constraint effect of shear box. Tables 2 and 3 summarize the experimental results
However, limited works were reported dealing with described above and Figures 1 and 2 present these
the comparison of shear strength characteristics of results in graphical forms for the angles of shearing
granular soils deduced from these three laboratory resistance for plane-strain (φps), triaxial (φT), and direct
tests conjunctly (e.g. Cioara et al. 2014). Furthermore, shear (φds) tests versus the initial porosity (n) and the
very limited comparative studies focused on dilatancy relative density Rd respectively.
factor, particles interlocking, and volumetric strain and Where:
axial strain at failure. Moreover, no efforts were made
emax e
to correlate the angle of shearing resistance of plane- Rd ¼ (1)
strain and triaxial compression conditions or with emax emin
those of direct shear test by considering the soil relative Where: emax, emin and e are the maximum, the mini-
density, dilatancy factor and particles interlocking. mum and the initial void ratio of the sand.
GEOMECHANICS AND GEOENGINEERING 3
Table 1. Physical properties of the sand used in the present investigation (Hanna 2001).
Void ratio, e Particle size, Dx (mm)
Sand type Grading/Particle shape Specific gravity, Gs Max. Min. D10 D15 D50 D60 D85
S1 Uniform/Rounded 2.65 0.80 0.40 0.11 0.135 0.23 0.265 0.36
S2 Uniform/Angular 2.63 0.90 0.50 0.30 0.40 0.51 0.7 0.75
S3 Well graded/Angular 2.64 0.95 0.40 0.40 0.45 0.63 0.8 0.9
Table 2. Summary of direct shear box test results (Hanna and Massoud 1981).
Relative den- Normal Shear stress at fail- Stress ratio at fail- Change of thickness at fail- Angle of shearing resis-
Sand Porosity, sity, stress, ure, ure ure, tance,
type n (%) Rd (%) σn (kPa) τf (kPa) τf/σn Δy (mm) φ (°)
S1 44.4 00 172 116 0.674 0.0312 31.7
37.9 47.5 172 123 0.700 0.0366 36.0
35.9 60 172 128 0.726 0.0406 37.6
32.4 85 172 148 0.839 0.0610 39.8
28.5 100 172 162 0.942 0.0780 42.1
44.4 00 344 241 0.700 0.0221 31.0
37.9 47.5 344 258 0.751 0.0279 33.7
35.9 60 344 267 0.775 0.0366 35.4
32.4 85 344 295 0.857 0.0406 37.9
28.5 100 344 333 0.969 0.0452 40.1
S2 47.3 00 172 118 0.686 0.0548 32.8
41.5 47.5 172 123 0.700 0.0508 36.4
39.0 65 172 148 0.839 0.0610 39.7
35.0 90 172 158 0.900 0.0801 42.0
33.3 100 172 163 0.945 0.0879 43.4
47.3 00 344 254 0.740 0.0327 32.1
41.5 47.5 344 263 0.764 0.0406 35.2
39.0 65 344 280 0.795 0.0508 38.0
35.0 90 344 300 0.872 0.0610 40.3
33.3 100 344 325 0.944 0.0687 43.5
S3 48.7 00 172 120 0.700 0.0356 34.4
40.3 50 172 148 0.839 0.0406 40.0
32.8 84 172 168 0.976 0.0584 44.3
28.5 100 172 176 1.028 0.0667 46.0
48.7 00 344 255 0.740 0.0308 34.2
40.3 50 344 285 0.810 0.0356 40.0
32.8 84 344 352 1.023 0.0610 44.0
28.5 100 344 358 1.043 0.0689 45.8
It can be noted that the values of the angle of shearing the present investigation for the plane-strain, triaxial test
resistance obtained from the plane-strain tests for dense results and direct shear tests were equal (φcv = 30°) (with
sand were about 5% to 15% higher than the values error < 1%).
obtained from triaxial compression tests, and 5% to Furthermore, the higher the normal stress the lower
10% higher than the corresponding value obtained from the secant friction obtained by these tests. This can be
direct shear tests. While for loose sand, the angles of explained by the fact that at higher normal stresses the
shearing resistance obtained from the plane-strain were dilatancy of the sand decreases. In the present investi-
slightly higher than those of the direct shear and triaxial gation, sands S1 and S2 exhibited lower angles of shear-
compression tests by 2% to 3% and 1% to 2% respectively. ing resistance than sand S3. This is mainly due to its
Furthermore, the angles of shearing resistance from angular particles in shape, which provide higher inter-
direct shear tests were about 2.5–10% higher than those locking component.
of triaxial tests for dense sand, whereas they were slightly Figures 3 and 4 present the relationships between
lower or equal for the case of loose sands. the dilatancy rate at failure, and the interlocking versus
It is of interest to note that the variation of the angles the porosity in plane-strain and triaxial compression
φps, φT and φds with the relative densities are almost linear tests. The interlocking between the particles, for the
(the coefficients R2 obtained for this correlation are situ- sand tested, was determined from the expression
ated between 0.948 and 0.998 with a mean value of 0.981); given by Rowe (1962) as follows:
and converge to a value of φps = φT = φds = φcv at the
loosest state (Rd = 0). This finding agreed well with the σ1 dv 1=2
tan α ¼ 1 (2)
results of Rowe (1969). The values of φcv deduced from σ3 dε1
4 A. HANNA AND T. AYADAT
48
44
A ngl e of s heari ng resi s tanc e, φ (o)
42
40
38
36
34
32
30
28.5 30.5 32.5 34.5 36.5 38.5 40.5 42.5 44.5 46.5
Porosity, n (%)
Figure 1. Angle of shearing resistance in plane strain, triaxial compression and direct shear tests versus porosity (Sand A).
GEOMECHANICS AND GEOENGINEERING 5
50
44
A ngl e of s heari ng resi s tanc e,φ (o)
42
40
38
36
34
32
30
0 10 20 30 40 50 60 70 80 90 100
Relative density, R d (%)
Figure 2. Angle of shearing resistance in plane strain, triaxial compression and direct shear tests versus relative density (sand A).
Where important to note that the higher density (i.e. the lower
α = interlocking parameter, porosity), the lesser axial strain required to achieve
dv = volumetric strain failure for triaxial and plain strain testing. While, for
σ1 and σ3 = major and minor principal stresses, the lower porosity, the higher volumetric strain
respectively. required to achieve failure for three types of testing.
It is of interest to note that the dilatancy in plane- Worthwhile to note that the volumetric strain for
strain tests is relatively low as compared to triaxial triaxial and plain strain testing was taken as the ratio
results, while interlocking in plane-strain tests is higher between the measured change in volume during shear
than for triaxial compression tests under the same (positive for expansion) and the volume of the sample at
condition. The difference tends to be less at the lower the beginning of shear. Whereas for the direct shear test,
densities. since the volume change is an invariant, the normal
Figures 5 and 6 present the volumetric strain and strain was used in its evaluation. In general, the normal
the axial strain at failure versus the porosity, respec- strains induce volume changes, whereas shear strains
tively. It can be noted that the volumetric strain curves induce a change of shape but no volume change.
tend to converge at the lowest densities, and as
expected the triaxial specimens had greater positive
volume change at failure than those of the correspond- Analytical models
ing plane-strain and direct shear tests. This can be
As mentioned in the previous section, new analytical
explained by the fact that in case of triaxial tests the
expression was developed to predict the angle of shear-
particles have more freedom to move around each
ing resistance of triaxial condition test from the direct
other, and the sample is free to expand against the
shear test data. In addition, new correlation between
confining stress, which increases of the volumetric
the angle of shearing resistance of plane-strain and
strain and decrease the interlocking. Furthermore, the
triaxial compression conditions was proposed.
axial strain at failure for the sand samples tested in
The deduced angle of shearing resistance from the
triaxial compression tests is higher than that given by
triaxial testing was determined from Mohr-Column
the plane-strain compression tests. The difference is
criterion as follows (Bolton, 1986):
increasing with the increase of the porosity. It is also
6 A. HANNA AND T. AYADAT
2.4
Sand S1
____ Plane-strain
Sand S2
2.2 ------ Triaxial Sand S3
1.8
1.6
1.4
1.2
1
28.5 32.5 36.5 40.5 44.5 48.5 52.5
Porosity, n (%)
Figure 3. Dilation factor at failure in plane-strain and triaxial compression tests versus porosity (cell pressure = 172 kPa).
80
Sand S1 (172 kPa)
78 Sand S1 (344 kPa)
Sand S2 (172 kPa)
Sand S2 (344 kPa)
76
Sand S3 (172 kPa)
Sand S3 (344 kPa)
74
Interl oc ki ng parameter,α (o)
72
70
68
66
64
62
60
28.5 32.5 36.5 40.5 44.5 48.5 52.5
Porosity, n (%)
4.5
DS = Direct Shear Sand S1 (DS)
T = Triaxial Sand S1(T)
4
PS = Plane Strain
Sand S1 (PS)
Sand S2 (DS)
3.5
SandS2 (T)
Sand S2 (PS)
3 Sand S3 (DS)
Vo lumetric strain , ΔV/V (%)
Sand S3 (T)
Sand S3 (PS)
2.5
1.5
0.5
0
28.5 33.5 38.5 43.5 48.5 53.5
-0.5
Porosity, n (%)
Figure 5. Volumetric strain at failure in plane -strain, triaxial compression and direct shear tests versus porosity (normal stress = 172
kPa).
10
Sand S1 (172 kPa)
9 Sand S1 (344 kPa)
Sand S2 (172 kPa)
Sand S2 (344 kPa)
Sand S3 (172 kPa)
8 Sand S3 (344 kPa)
____ Triaxial
7
------ Plane strain
A x i al S train,ε1 (%)
0
28.5 32.5 36.5 40.5 44.5 48.5 52.5
Porosity, n (%)
Figure 6. Axial strain in plane-strain and triaxial compression tests versus porosity.
8 A. HANNA AND T. AYADAT
1 þ sin ϕT 1 þ sin ϕT
K¼ (16) RT ¼ (26)
Dð1 sin ϕT Þ 1 sin ϕT
Equating Equations (14) and (16), the following equa- By substituting Equations (12) and (26) in (11), it can
tion can be written be written,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi h ipffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ sin ϕT 3D: tan ϕds þ 12D 3D2 1þsin ϕT
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1sin ϕT 1 12D 3D2
Dð1 sin ϕT Þ D 12D 3D2 þ D2 : tan ϕds 4D: tan ϕds tan ϕps cos ϕcv ¼ h i h i (27)
ϕT ϕT
(17) 4 1þsin
1sin ϕ D 1þsin
1sin ϕ þ 3D
T T
2.4
Sand S1 (present w ork)
Sand S2 (present w ork)
2.2 Sand S3 (present w ork)
Hanna 2001
Cornforth 1961
Dil atanc y factor, D 2
1.8
1.6
D = Rd + 1
1.4
1.2
1
0 10 20 30 40 50 60 70 80 90 100
Relative density, Rd (%)
Figure 7. Variation of the dilatancy factor (D) with the relative density (Rd).
Table 4. Comparison of the predicted and measured triaxial angles of shearing resistance reported in the literature (normally
consolidated sands).
Predicted
Measured (Equation 18) Predicted (modified Coefficients)
(3) The deduced angle of shearing resistance from direct (7) Simple expressions (Equations 18 and 30)
shear test for the case of dense sands was about 2.5% were proposed to predict the angle of shearing
to 10% higher than the corresponding values for the resistance for any type of the three listed test-
triaxial test. It was slightly lower or equal for the case ing methods using the results of the two
of loose sand. others.
(4) In plane-strain tests, sand particles have less freedom (8) A good agreement has been obtained between
to move around, which results in an increase of the predicted results values using the proposed
particle interlocking. While, in triaxial tests, the par- methods and the results of the present experi-
ticles possess the freedom to move around each mental investigation and those available in the
other, and the sample is free to expand against the literature.
confining pressure, which results in an increase of
the volumetric strain and a decrease of interlocking.
For the better graded and angular particles, the NOTATION
interlocking effect is greater than with the rounded
Ai ; Bi ; Ci and Di are coefficients, which depend mainly on
and uniform particles. D and OCR
(5) The axial strain at failure in triaxial tests is D: the dilatancy factor
greater than the corresponding value in plane- Dx: the diameter of the sand particles for which (X)% of
strain tests at the same confining pressure and particles are finer
density. dε1: plastic component of the axial strain
dv: plastic component of the volumetric strain
(6) The volumetric strain at failure in triaxial com-
Gs:
pression tests is greater than the corresponding K: the material parameter;
value obtained from plane-strain and direct ko: coefficient of earth pressure at rest
shear tests. OCR: overconsolidated ratio
12 A. HANNA AND T. AYADAT
Table 5. Comparison of the predicted and measured plane-strain angles of shearing resistance reported in the literature (normally
consolidated sands).
Measured (Hanna 2001) Predicted (Equation 30)
Soil type σ3 (kPa) Rd φT (°) sinφT φps (°) tanφps tanφps φps (°)
S1 172 0 32.5 0.537 33.0 0.645 0.640 32.6
0.25 34.0 0.559 36.5 0.740 0.720 35.8
0.38 35.0 0.573 38.0 0.781 0.764 37.4
0.56 37.0 0.601 41.0 0.869 0.839 40.9
0.69 38.5 0.622 42.5 0.916 0.896 41.8
1.0 41.0 0.656 47.0 1.072 1.007 45.2
344 0 32.0 0.5299 32.5 0.637 0.631 32.2
0.44 34.0 0.559 37.0 0.753 0.751 36.9
0.56 35.0 0.573 39.0 0.809 0.791 38.3
0.75 37.0 0.601 42.0 0.900 0.865 40.8
1.0 39.0 0.629 45.0 1.000 0.947 43.4
S2 172 0 33.5 0.551 34.0 0.674 0.657 33.3
0.31 35.5 0.580 39.0 0.809 0.762 37.3
0.46 37.0 0.601 40.5 0.854 0.823 39.4
0.62 39.0 0.629 42.5 0.916 0.898 41.9
0.85 40.5 0.649 46.5 1.053 0.973 44.2
1.0 42.0 0.669 47.5 1.091 1.037 46.1
344 0 33.0 0.544 33.5 0.661 0.649 33.0
0.31 34.5 0.566 36.0 0.726 0.741 36.5
0.62 37.0 0.649 40.5 0.854 0.848 40.3
0.85 38.0 0.615 41.0 0.869 0.903 42.0
1.0 42.0 0.669 46.0 1.035 1.037 46.1
S3 172 0 34.5 0.566 35.0 0.700 0.675 34.0
0.24 38.0 0.615 38.5 0.795 0.798 38.6
0.59 41.0 0.656 45.0 1.000 0.941 43.3
0.82 44.0 0.694 47.5 1.091 1.065 46.8
1.0 45.5 0.713 48.0 1.111 1.143 48.8
344 0 34.0 0.5591 34.5 0.687 0.666 33.7
0.24 36.0 0.587 40.0 0.839 0.758 37.2
0.59 41.0 0.656 43.0 0.932 0.941 43.2
0.82 43.5 0.688 45.5 1.017 1.051 46.4
1.0 45.5 0.713 49.5 1.170 1.144 48.8
Glass ballotini 138 1.0 34.5 0.566 37.5 0.767 0.815 39.1
Rowe (1962) 0.8 33.0 0.544 36.0 0.726 0.763 37.3
0.6 30.5 0.507 32.5 0.637 0.686 34.4
0.4 27.6 0.463 28.4 0.541 0.602 31.0
0.2 26.6 0.447 27.0 0.510 0.560 29.2
0.0 24.0 0.406 24.0 0.445 0.484 25.8
Mersey river quartz 28 1.0 42.0 0.669 46.5 1.053 1.037 46.0
Rowe (1962) 0.8 40.6 0.719 43.5 0.949 0.968 44.0
0.6 38.0 0.615 40.0 0.839 0.870 41.0
0.4 36.9 0.600 38.0 0.781 0.809 38.9
0.2 34.5 0.566 35.0 0.700 0.720 35.7
0.0 32.0 0.530 32.0 0.625 0.631 32.3
Brasted sand 276 1.0 42.0 0.669 46.0 1.035 1.037 46.1
Cornforth (1964) 0.67 40.0 0.642 43.0 0.933 0.931 42.9
0.33 35.5 0.580 37.0 0.754 0.766 37.4
0.17 34.0 0.559 35.0 0.700 0.704 35.1
0.0 33.0 0.544 33.5 0.662 0.648 33.0
e: the initial void ratio of the sand. σ: Normal stress applied on the plane of failure
emax.: maximum void ratio of sand τ: shear stress on the plane of failure
emin.: minimum void ratio of the sand
R: the principal stress ratio
R d: the relative porosity of the sand
Acknowledgments
Rps: plane-strain principal stress ratio
RT: triaxial principal stress ratio The financial support from the Natural Science and
Si: sand ‘i’ Engineering Research Council of Canada (NSERC) and
β: the angle of dilatancy Concordia University are acknowledged.
Δy: Change of thickness at failure
φcv: the angle of shear resistance at constant volume
φps: angle of shearing resistance of plane-strain test,
φT: angle of shearing resistance of triaxial test, Disclosure statement
σ 1: major principal stress,
No potential conflict of interest was reported by the authors.
σ 3: minor principal stress
GEOMECHANICS AND GEOENGINEERING 13