Académique Documents
Professionnel Documents
Culture Documents
NOTE / NOTE
Abstract: Procedures that allow the prediction of some properties of unsaturated soils or the minimization of the num-
ber of tests needed to measure them are advantageous because the control of suction during testing is a formidable task
that is time consuming and involves a great degree of expertise. A simplified procedure is proposed in this paper to es-
timate the shear strength of an unsaturated soil. The procedure is based on an empirical hyperbolic function that has
been successfully used to fit experimental data. The function requires two input values, namely the shear strength of a
saturated sample and the shear strength of an air-dried sample tested without the need for suction control. Samples
tested under a controlled suction larger than the maximum suction expected in the problem can, alternatively, replace
the air-dried samples. Both alternatives were tested against results for various soils reported in the literature. The good
agreement between the estimates and the experimental data indicates that the proposed procedure is promising and reli-
able for estimating preliminary unsaturated shear strength parameters.
Key words: unsaturated soil, suction, shear strength, prediction, laboratory tests.
Résumé : Les procédures qui permettent de prédire des propriétés de sols non saturés ou de minimiser le nombre
d’essais requis pour les mesurer sont avantageuses parce que le contrôle de la succion durant les essais représente une
tâche importante qui prend du temps et implique un haut niveau d’expertise. Dans cet article, on propose une procé-
dure simplifiée pour estimer la résistance au cisaillement d’un sol non saturé. La procédure est basée sur une fonction
hyperbolique empirique qui a été utilisée avec succès pour lisser des données expérimentales. La fonction requiert deux
valeurs d’entrée : la résistance au cisaillement d’un échantillon saturé et celle d’un échantillon séché à l’air et testé
sans besoin de contrôler la succion. Les échantillons testés sous une succion contrôlée plus grande que la succion
maximale escomptée dans le problème peut alternativement remplacer les échantillons séchés à l’air. Les deux options
ont été testées et comparées aux résultats pour divers sols mentionnés dans la littérature. La bonne concordance entre
les estimations et les données expérimentales indiquent que la procédure proposée est prometteuse et fiable pour
l’estimation préliminaire des paramètres de résistance au cisaillement non saturée.
Mots clés : sol non saturé, succion, résistance au cisaillement, prédiction, essais de laboratoire.
[Traduit par la Rédaction] Vilar 1095
Table 1. Characteristics of the soils tested by Futai (2002), Reis (2004), and Escario (1988).
Soila Soil type φ′ (°) c′ (kPa) cult (kPa) a b R2 Remarksb
1 Sandy clay 27.3 17.0 125.0 1.940 0.0093 0.99 wL = 57%; PI = 29%; γ = 15 kN/m3;
(undisturbed) w = 30%; e = 1.1–1.2; triaxial CD
tests of Futai (2002)
2 Sandy silt 26.4 33.5 77.5 2.016 0.0227 0.99 wL = 42%; PI = 19%; γ = 17 kN/m3;
(undisturbed) w = 25%; e = 0.8–0.9; triaxial CD
tests of Futai (2002)
3 Silty sand 28.0 24.0 115 1.88 0.011 0.98 wL = 38%; PI = 15%; γ = 18 kN/m3;
(undisturbed) w = 17.5%; e = 0.75; triaxial CD tests
of Reis (2004)
4 Madrid gray clay 25.20 170 580 2.126 0.0024 0.98 LL = 71%; PI = 35%; standard Proctor
(statically γdmax = 13.3 kN/m3; wopt = 33.7%; for
compacted) molded specimens, γ = 13.3 kN/m3
and w = 29%; CD direct shear tests
of Escario (1988)
5 Guadalix de la 32.50 93 650 1.570 0.0180 0.98 LL = 33%; PI = 14%; standard Proctor
Sierra red clay γdmax = 18 kN/m3; wopt = 17%; for
(statically molded specimens, γ = 18 kN/m3 and
compacted) w = 13.6%; CD direct shear tests of
Escario (1988)
a
Soil numbers as given in Figs. 1–3.
e, void ratio; LL, liquid limit; PI, plasticity index; R2, coefficient of determination; wL, liquid limit; wopt, optimum moisture content; γ, unit weight;
b
Fig. 1. Shear strength envelopes and experimental and predicted shear strength considering results of tests with air-dried samples (data
from Futai 2002).
Fig. 2. Shear strength envelopes and experimental and predicted Fig. 3. Experimental and predicted shear strength envelopes con-
variation of shear strength considering results of tests with air- sidering one set of tests with dry samples: (a) Madrid gray clay;
dried samples (data from Reis 2004). (b) Guadalix red clay (data from Escario 1988).
Most test results presented in the literature are from soils the proposed method make this procedure a practical and re-
that were tested under some limited values of soil suction at liable method of estimating the shear strength of unsaturated
which shear strength is increasing. Consequently, it is proba- soils. The simplicity of the procedure lies in the fact that it
ble that the ultimate shear strength was not reached. This requires only one set of tests on air-dried samples or at a
feature can be considered by changing the way parameter b known suction, together with the saturated effective shear
is determined. As the point corresponding to the maximum strength parameters.
test suction belongs to the curve that represents the shear Some peculiarities of the method rest with the equation
strength envelope, defining the maximum measured cohe- used to describe the influence of matric suction on unsatu-
sion cm (or τm , the maximum shear strength) at the maxi- rated shear strength. As shown, the proposed method does
mum value of matric suction, ψm, it can be shown that not take into account the air-entry value or any other quanti-
tative data from the SWCC, and thus it is different from
[6] b = 1/(cm − c′ ) − 1/(ψ m tan φ′ )
other methods such as those of Vanapalli et al. (1996) and
The use of eqs. [2], [3], and [6] is illustrated considering Fredlund et al. (1996).
the data shown in Table 2. Table 3 shows some additional The method of Fredlund et al. (1996) yields a linear rela-
data for soils listed in Table 2 together with the derived soil tionship between shear strength and suction up to the air-entry
parameters obtained and the coefficients of determination value. In eq. [2], as long as the product of the parameter b and
(R2) for the experimental and predicted data. matric suction is negligible when compared with a, the rela-
Figure 4 shows a comparison between the predicted and tionship is almost linear and represented by
experimental results. As can be seen, there is good agree-
ment, with R2 values greater than 0.95 for most of the data [7] c = c′ + ψ tan φ′
tested.
Under this circumstance, the contribution of matric suc-
Discussion tion to shear strength is equivalent to the contribution of net
normal stress, as must be expected in a saturated soil. Evalu-
The good agreements obtained between the calculated and ation of whether the proposed model yields such a linear re-
measured values and the simplicity of the two alternatives of lationship up to the air-entry value of the soil could not be
© 2006 NRC Canada
Vilar 1093
Table 3. Parameters used to validate the alternative procedure of calculating unsaturated shear strength.
Input parameters Output parameters
Soil a
φ′ (°) c′ (kPa) cm (kPa) ψ m (kPa) a b R2
6 30.0 30.0 103.8 200 1.733 0.0049 0.97
7 38.7–40.1 9.2 62.0 150 1.218 0.0108 0.99
8 21.2 35.0 92.3 200 2.579 0.0045 0.99
9 29.0 7.8 104b 394 1.805 0.0058 0.96
10 25.5 44.4 173b 498 2.098 0.0035 0.94
11c
σ – ua = 25 kPa 23.0 10.0 73.0 500 2.356 0.0112 0.97
σ – ua = 100 kPa 23.0 38.0 113.0 500 2.356 0.0086 0.99
σ – ua = 200 kPa 23.0 84.0 164.0 500 2.356 0.0078 0.90
12c
σ – ua = 100 kPa 40.0 150.0 183.7 307 1.193 0.0258 0.94
σ – ua = 20 kPa 40.0 53.0 68.0 330 1.193 0.0630 0.83
a
Soil numbers as given in Fig. 4.
b
Average value near the maximum suction.
c
c′ and cm are the whole shear strength (cohesion intercept plus friction component).
Fig. 4. Experimental and predicted shear strength of soils from Table 1, considering the effective shear strength parameters of saturated
soil and the shear strength at the largest suction used in the tests.
completely assessed because information about the SWCC is 70 kPa for soil number 9, whose air-entry value is 70 kPa.
unavailable for many of the soils listed in Table 2. In Fig. 4b More precise evaluation of the air-entry value is available
a linear relationship is observed up to 15 kPa of suction for for soil number 11 (Fig. 4c). For the samples tested under
soil number 10, which presents an air-entry value of about 25 kPa of net normal stress, the air-entry value was 20 kPa,
40 kPa. Also in Fig. 4b, the linear relationship holds up to and for 100 and 200 kPa of net normal stress, the air-entry
© 2006 NRC Canada
1094 Can. Geotech. J. Vol. 43, 2006
values were 30 and 45 kPa, respectively (Vanapalli et al. 8 September 1995. A.A. Balkema, Rotterdam, The Netherlands.
1996). Figure 4c shows that the matric suction at which the Vol. 1, pp. 31–38.
nonlinearity commences appears to be close to the measured Drumright, E.E., and Nelson, J.D. 1995. The shear strength of un-
air-entry values for the three net normal stresses used in the saturated tailings sand. In Proceedings of the 1st International
tests. These observations suggest that the envelope predicted Conference on Unsaturated Soils, Paris, 6–8 September 1995.
by the proposed model begins to separate from the almost Edited by E.E. Alonso and P. Delage. A.A. Balkema, Rotterdam,
linear relationship at a suction value that is below or close to The Netherlands. Vol. 1, pp. 45–50.
the air-entry value, but not beyond it. Escario, V. 1988. Formulae for the shear strength envelope of par-
tially saturated soils. Report 68, Ingenieria Civil, CEDEX, Ma-
Figure 4d shows a typical set of data that show a decrease
drid, Spain. [In Spanish.]
in shear strength after a maximum. The soil is coarse silty Escario, V., and Saez, J. 1986. The shear strength of partly satu-
sand, and the decrease is more noticeable for the lower net rated soils. Géotechnique, 36(3): 453–456.
normal stress used in the tests when the soil showed dilative Fredlund, D.G., and Rahardjo, H. 1993. Soil mechanics for unsatu-
behavior. In predominantly granular soils it is expected that rated soils. John Wiley & Sons, New York.
the main contribution to soil suction is that from capillarity, Fredlund, D.G., Morgenstern, N.R., and Widger, R.A. 1978. The
as the effect of adsorptive forces will be less pronounced in shear strength of unsaturated soils. Canadian Geotechnical Jour-
these soils. It is therefore reasonable to admit that the effect nal, 15(3): 313–321.
of suction on shear strength will reach a maximum and will Fredlund, D.G., Rahardjo, H., and Gan, J.K.M. 1987. Non-linearity
reduce as strain and dilation induce a perturbation on the of strength envelope for unsaturated soils. In Proceedings of the
capillary meniscus, thus causing a reduction in shear 6th International Conference on Expansive Soils, New Delhi, In-
strength and other mechanical properties that depend on suc- dia, 1–4 December 1987. A.A. Balkema, Rotterdam, The Neth-
tion. erlands. Vol. 1, pp. 49–54.
In this case, depending on the largest suction value used Fredlund, D.G., Xing, A., Fredlund, M.D., and Barbour, S.L. 1996.
in the tests, the calculated values can be lower than the mea- The relationship of the unsaturated soil shear strength to the
sured values. As the model considers that shear strength as- soil-water characteristic curve. Canadian Geotechnical Journal,
sociated with the largest suction used in the tests is the 33(3): 440–448.
maximum, the difference between measured and predicted Futai, M.M. 2002. Theoretical–experimental study of the behavior
values is as large as the decrease in shear strength after the of non-saturated tropical soils applied to a gully erosion. Ph.D.
maximum. Even in the case of the test results shown in thesis, COPPE-Federal University of Rio de Janeiro, Rio de Ja-
neiro, Brazil. 559 pp.
Fig. 4d, however, the coefficient of determination is still
Gan, J.K.-M., and Fredlund, D.G. 1996. Shear strength characteris-
high, but one must be aware that in granular soils the model
tics of two saprolitic soils. Canadian Geotechnical Journal,
can yield conservative values.
33(4): 595–609.
Gan, J.K.M., Fredlund, D.G., and Rahardjo, H. 1988. Determina-
tion of the shear strength parameters of an unsaturated soil using
Conclusion the direct shear testing. Canadian Geotechnical Journal, 25(3):
The shear strength envelope for unsaturated soils as a 500–510.
Ho, D.Y.F., and Fredlund, D.G. 1982. A multistage triaxial test for
function of suction is generally nonlinear, especially if a
unsaturated soils. Geotechnical Testing Journal, ASTM, 5(1):
large range of soil suctions is considered. Based on the ob-
18–25.
servation that an empirical hyperbolic relationship can pro-
Khalili, N., and Khabbaz, M.H. 1998. A unique relationship for c
vide a good fit to the experimental data for many soils, a for the determination of the shear strength of unsaturated soils.
simplified procedure is proposed to estimate the shear Géotechnique, 48(5): 681–687.
strength envelope for unsaturated soils. The proposed Machado, S.L., and Vilar, O.M. 1998. Shear strength of unsatu-
method uses effective stress parameters from saturated sam- rated soil: lab tests and predicting equations. Solos e Rochas,
ples and the results of only one set of tests performed on air- 21(2): 65–78. [In Portuguese.]
dried specimens or on specimens tested under a controlled Miao, L., Yin, Z., and Liu, S. 2001. Empirical function represent-
suction, larger than the maximum suction expected in the ing the shear strength of unsaturated soils. Geotechnical Testing
problem under analysis. Generally good agreement between Journal, ASTM, 24(2): 220–223.
measured and predicted values was observed. The use of air- Oberg, A., and Salfors, G. 1997. Determination of shear strength
dry samples could replace the more sophisticated suction parameters of unsaturated silts and sands based on the water re-
control tests, as some data available in the literature show tention curve. Geotechnical Testing Journal, ASTM, 20(1): 40–
that this method could be promising. Of course, it is not in- 48.
tended that this simplified procedure replace a more compre- Rassam, D.W., and Cook, F. 2002. Predicting the shear strength en-
hensive investigation of soil properties, but rather it is velope of unsaturated soils. Geotechnical Testing Journal,
intended to provide a tool to make preliminary estimates of ASTM, 25(2): 215–220.
the shear strength parameters of unsaturated soils. Reis, R.M. 2004. Stress–strain behaviour of two horizons of a re-
sidual sol from gneiss. Ph.D. thesis, University of São Paulo,
São Carlos, Brazil. 190 pp. [In Portuguese.]
References Röhm, S.A., and Vilar, O.M. 1995. Shear strength of an unsatu-
rated sandy soil. In Proceedings of the 1st International Confer-
de Campos, T.M.P., and Carrillo, C.W. 1995. Direct shear testing ence on Unsaturated Soils, Paris, 6–8 September 1995. Edited
on an unsaturated soil from Rio de Janeiro. In Proceedings of by E.E. Alonso and P. Delage. A.A. Balkema, Rotterdam, The
the 1st International Conference on Unsaturated Soils, Paris, 6– Netherlands. Vol. 1, pp. 31–38.
Satija, B.S. 1978. Shear behavior of partly saturated soils. Ph.D. van Genutchen, M.T. 1980. A closed-form equation for predicting
thesis, Indian Institute of Technology, Delhi, India. the hydraulic conductivity of unsaturated soils. Soil Science So-
Teixeira, R.S., and Vilar, O.M. 1997. Shear strength of an unsatu- ciety of America Journal, 44: 892–898.
rated compacted soil. In Proceedings of the 3rd Brazilian Sym- Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E., and Clifton, A.W.
posium on Unsaturated Soils, Rio de Janeiro, 21–25 April 1997. 1996. Model for the prediction of shear strength with respect to
Edited by T.M.P. de Campos and E.A.Vargas. Freitas Bastos soil suction. Canadian Geotechnical Journal, 33(3): 379–392.
Editora, Rio de Janeiro, Brazil. Vol. 1, pp. 161–169. [In Portu-
guese.]