Shear Strength of Tropical Soils and Bentonite Mixtures for Barrier Design

Compacted clayey tropical soils have great potential to be used as barriers in waste disposal facilities, considering that some technical requirements are fulfilled (e.g. hydraulic conductivity; compatibility after disposal; shear strength; swelling/cracking, etc). In turn, the bentonite addition is relevant for the cases where the hydraulic conductivity must be lowered, and therefore investigations on the changes of the mechanical parameters of tropical soils induced by the bentonite addition are of great interest. This paper presents the results of a laboratory investigation on the shear strength of different samples of tropical soils and their mixtures with bentonite in the proportions of 0, 3, 6, 9 and 12% (dry weight basis). The tropical soil samples were collected considering the lateritic, transitional and non-lateritic behavior according to the MCT-M (Modified-Miniature-Compacted-Tropical) classification. The laboratory tests consisted of CU (ConsolidatedUndrained) triaxial tests under the confining stresses of 100, 200 and 400 kPa. The results have showed that the addition of bentonite produced a significant increase in the plasticity of the tropical soil samples (PI increases 4 to 6 times), considerably reduced their friction angle (by as much as 11) and stiffness at peak (modulus reductions of 2.5 to 4 MPa) and gave rise to a slight increase in their cohesion (up to 12 kPa). These facts must be considered in the barrier stability analyses when heavy loads are applied. The important role of the shear strength on barrier design is highlighted.


Introduction
To be used as liners on barrier design for waste disposal facilities, soils must suit geotechnical criteria such as low hydraulic conductivity, long-term compatibility with the disposed liquid, adequate swelling/cracking properties and adequate shear strength (Shackelford, 1994).
In fact, the shear strength was highlighted once the weight on designs of solid waste landfill barriers (urban and industrial) has been steadily increasing as a response to the lack of sites for waste disposal facilities.
The hydraulic adequacy of tropical soils and their admixtures to be used on barrier design was evaluated by Anderson & Hee (1995), Osinubi & Nwaiwu (2002) and Amadi & Eberemu (2012).
In turn, the shear strength of geosynthetic clay liners (GCL) was assessed by Stark & Eid (1996), Chiu & Fox (2004), Fox et al. (2006), Fox & Kim (2008), Chen et al. (2010) and Eid (2011).Some other papers report studies of the shear strength of soils for barrier purposes such as Heineck (2002), Hueckel & Pellegrini (2002), Santamarina (2003), Sunil et al. (2009) and Batista & Leite (2010).In these papers, the addition of bentonite led to a decrease in the resistance to shear of the soil as a whole.Particularly, a reduction of the friction angle (f) and an increase of the cohesion (c) was observed.
Considering the absence of research on the matters of using tropical soils and their admixtures with bentonite, this study was proposed to cover most of the geotechnical criteria referred in the first paragraph, such as hydraulic conductivity, shear strength and long-term compatibility.The tropical soil samples investigated were classified as lateritic, transitional and non-lateritic according to the MCT-M Classification as proposed by Vertamatti (1998).Morandini &Leite (2013 and2015) report part of the results of this research, focused on the hydraulic behavior of admixtures of tropical soils and 0, 3, 6, 9, and 12% of bentonite (dry weight basis).
In particular, the present paper focuses on the shear strength of admixtures of tropical soils with bentonite in the same proportions investigated by Morandini &Leite (2013 and2015).Laboratory works involved triaxial shear strength tests under different confining stresses.The longterm compatibility issue will be reported elsewhere.

Soil Characterization
The tropical soil samples were collected as an attempt to represent lateritic (SL sample), transitional (ST sample) and non-lateritic (SN sample) behavior according to the MCT-M Classification (Vertamatti, 1998).The sampling locations in the state of Minas Gerais, Brazil can be seen on Fig. 1.
The evaluated tropical soils were collected from the geological formation denominated "Iron Quadrangle".This type of soil is formed from igneous rocks (mostly basalt) or metasedimentary rock from Precambrian orogenetic cycles.It is characterized as residual soil formed by pedogenetic process known as laterization, in which the hot and humid climate and the profile of high draining favors the processes of hydrolysis, hydration and oxidation.After the formation of sediments, the silica particles are highly leached to lower layers and consequently the iron and aluminum oxides remain in the surface layers.
The main factor responsible for the contrast between the classes of the studied tropical soils is their position on the pedological layer: the laterite soil was collected at the top of a slope (theoretically with high concentration of oxides); the non-lateritic soil was collected at the foot of a cut slope (with characteristics of the parent rock and a high concentration of silica); the transitional soil was collected at the foot of a hill, in a spot with process of erosion and transport of different pedological layers.
The mixtures were prepared by hand and the proportions of bentonite Brasgel-PA® (X-ray diffraction presented by Morandini & Leite, 2015) were 0, 3, 6, 9 and 12% (dry weight basis), which received the designations expressed in Table 1.Recent studies indicate the use of bentonite in content between 2 and 20% mixed with sand and other soils, such as Chapuis (1990), Shackelford (1994), Anderson & Hee (1995), Batista & Leite (2010) and Amadi and Eberemu (2012).The complete characterization procedures and results for all these samples were firstly presented by Morandini & Leite (2015).Table 2 shows only a summary of their geotechnical properties.The parameters presented in Table 2 were tested according to the procedures of the standards ABNT NBR 6459 (1984), ABNT NBR 7180 (1984), ABNT NBR 7181 (1984) and ABNT NBR 7182 (1986).The SL, SL and SN samples refer to the original soils and BB sample refers to Brasgel Bentonite.
The mineralogy of the SL sample is mainly composed of kaolinite, quartz, gibbsite and hematite.The ST sample presents kaolinite, illite, chlorite and smectite, while the SN sample is composed of kaolinite, quartz and goethite.The chemical composition of these samples indicates high contents of Fe and Al oxides in the following order: sample SL > sample ST > sample SN.

Methods
Ordinary triaxial soil tests were used to determine the shear strength, axial deformation and pore pressure in-   crease of the compacted soil samples of Table 2. Consolidated-Undrained (CU) test was chosen, as an attempt to represent a fast loading condition over low-permeability soils, as usually is the case of clayey barriers.
According to the procedures indicated by ASTM-D4767 (1995), soil specimens were firstly compacted under Normal Proctor energy (moisture content 2% above the optimum moisture).The sample was then trimmed to a final geometry of 50 ± 2 mm diameter (D) and 100 ± 4 mm height (H).These dimensions are close to the specifications indicated by the standard ASTM-D4767 (1995): H = 2.0 to 2.5D.Three soil specimens were used for each triaxial test, leading to a final number of 45.
The confining stresses (s 3 ) of 100, 200 and 400 kPa have been applied to represent the field reality of bottom liners loaded with Municipal Solid Waste (MSW).For exemple, a landfill 100 m high and a density of 1.2 kN/m 3 , would be a confining pressure of about 60 kPa.General procedures involve saturation, consolidation and shearing of the soil specimens.Since this is a current geotechnical laboratory test, some brief description of these methods is given next: • Saturation.Counter pressure method, as suggested by Head (1986).Small pressure gradients are applied according to the following premise: confining stress > pore pressure at the base of the soil specimen > counter pressure at the top of the soil specimen.Counter pressure difference between the top and bottom of the sample never exceeds 20 kPa.Confining pressure was applied by adding the counter pressure without the necessity of retained counter pressure, resulting in effective stress of 100, 200, and 400 kPa.Samples were considered saturated when the parameter B of Skempton was above 95%.• Consolidation.Confining stress application and free drainage of the soil specimens until the pore pressure decreases until equalization.The consolidation was measured by evaluating the volume variation vs. the square root of time.• Shearing.The axial load (deviator stress) was applied through controlled deformation at a constant velocity of 0.0025 mm/s during 65 min, leading to a maximum axial deformation of 20%.This rate of axial loading was established according to the criteria suggested by Head (1986) for fine soil, with 5% of the axial strain being reached at 50% of the estimated time for soil consolidation.
The friction angle (j) and cohesion (c) were determined by the stress path approach (Lambe, 1964), using the diagram of mean total stress (p) or mean effective stress (p') vs. the deviator total stress (q) or deviator effective stress (q'), as expressed by Eqs. 1 through 4.
where s 1 is the axial stress, s 3 the confining stress and u the pore pressure.
The shear strength envelope in the p-q or p'-q' space is represented by the Eqs. 5 and 6, respectively.q a p = + tan a (5) where p, p' and q, q' are parameters that represent the average stress and deviator total stress, respectively.And a, a' and a, a' are modified parameters obtained from the strength envelope at failure.In turn, the parameters j and c (Mohr-Coulomb envelope) are related to the referred modified parameters by Eqs. 7 through 10, in terms of total and effective stresses.
Additionally, Eqs.11 and 12 were used to estimate the modulus of elasticity at peak (taken across the line connecting the origin to the peak resistance) and the A parameter (Skempton, 1954), respectively.The A parameter was measured to represent the effect of the axial load on the generation of pore pressure, a fundamental parameter for the study of low permeability soils.
where s d is the deviator stress at peak (Ds 1 -Ds 3 ), e a is the axial strain at peak and Du the pore pressure variation.

Results and Discussion
The influence of the bentonite addition on the stiffness of the mixtures is illustrated on Figs. 2, 3 and 4, which show the deviator stress (s d = Ds 1 -Ds 3 ) vs. axial strain (e a ) curves.It is quite clear that the bentonite addition reduced the peak deviator and residual stresses of the samples.
Higher values for the peak deviator stresses were found for the SL sample compared to ST and SN samples for all the confining stresses used in the tests.This fact is due in part to the coarser grain size of the SL sample compared to ST and SN samples, and also to the cementation (micro concretions) often found in lateritic soils, as mentioned by Nogami & Villibor (1985).
As expected, peak axial deformations increased with increasing confining stresses for all samples.However, that peak axial deformation did not vary significantly among the soils and bentonite contents.
The moduli at peak (E p ) were estimated from the peak axial stresses and strains, and are presented in Fig. 5 as a function of the bentonite content for all samples.As also expected, there was a reduction on the modulus at peak with the addition of bentonite.For instance, 12% of bentonite caused a reduction of 4 MPa on E p of the SL sample and 2.5 MPa for the ST and SN sample.
As expected, for all the soil samples there was a significant decrease in the friction angle as a response to the increase of the bentonite content, as shown on Figs. 6 and 7.However, the rate of this decrease was very similar for the SL and ST samples, and more accentuated for the SN sample.For instance, the value of j' = 34°for the natural  lateritic sample (sample SL) reduced to j' = 29°when 12% of bentonite was considered (sample SL12).Mesri & Olson (1971) reported a significant reduction of the friction angle of pure montmorillonite clays by increasing the effective stress.Therefore, it would be expected that besides the addition of bentonite, the variation of the confining stresses (s 3 ) from 100 to 400 kPa could also exert some influence on the friction angles, which was not confirmed, considering the linearity observed in the curves of Fig. 7.
As also expected, cohesion experimented linear increase with the bentonite addition (Fig. 7) for all samples.For instance, cohesion increased from 6 to 12 kPa considering the addition of 12% of bentonite for all samples.
As general remark about the influence of the addition of bentonite on the strength parameters (cc'and jj'), it can be said that the decrease in the friction angle can be compensated by the increase in the cohesion, depending on the loading conditions.Under light loading (e.g. at the beginning of operation of a landfill site), the mobilization of the friction angle is small and the increase in the cohesion may be sufficient to compensate for the loss in the global  shear strength.On the other hand, under heavy loading (e.g. at the end of operation of a landfill site), the mobilization of the friction angle is high and so is the loss in shear strength.However, friction angle values and cohesion of the SL and of the ST + 12% of bentonite would not be likely to lead to soil failure.
Figure 8 shows that the parameter A at failure increases as a function of the bentonite content for all samples and all confining pressures, especially for the confining stress of 400 kPa.Morandini & Leite (2015) report the sig-nificant reduction of the hydraulic conductivity caused by the addition of bentonite in the same soil samples and mixtures, which might be responsible for this increase in the pore pressure during the application of the deviator stress.
A higher parameter A means that the barrier would be subject to higher pore pressure generation during the loading phase.Under such circumstances, and if liners are installed deep in the landfill, special care must be taken in the rising rate of the landfill.

Conclusions
As a general remark, for the evaluated soils in this study, it is concluded that the lateritic character has minor influence on the mechanical behavior when bentonite is applied.
In turn, the changes in the mechanical behavior of the tropical soil samples (in different magnitudes according to the class SL, SN and ST) induced by the addition of bentonite are summarized next: • A significant increase in the plasticity; • A decrease in the peak stress at failure, with no influence on the subsequent deformations; • A significant reduction in the stiffness; • The pore pressure considerably increased during the shear stress application, as suggested by successive increases in the parameter A; • Significant reduction of the friction angle and a slight increase in the cohesion.At first, the above conclusions seem to be obvious, considering the high plasticity of the bentonite.Neverthe-less, the high magnitude of the mechanical changes imposed by the bentonite addition was relevant and must be taken into consideration for barrier design.In terms of effective friction angle, there was a decrease of approximately 14%, 20% and 40% respectively for the SL, ST and SN samples and these samples with addition of 12% of bentonite.The stiffness declines up to 25% compared to pure soils and soils with 12% of bentonite.Therefore, bentonite plays an important role in reducing the hydraulic conductivity of tropical soils, but caution should be exercised about the quantities of this material to be used.

Figure 5 -
Figure 5 -Modulus at peak (E P = maximum deviator stress / maximum axial strain) as a function of the bentonite content: (a) SL sample; (b) ST sample and (c) SN sample.

Figure 6 -
Figure 6 -Shear strength envelopes for total (a) and effective (b) stresses.

Figure 7 -
Figure 7 -Variation of the friction angle (a) and cohesion (b) with bentonite content (total and effective stresses).

Figure 8 -
Figure 8 -Parameter A (Skempton, 1954) at failure as a function of the bentonite content: (a) Sl sample; (b) ST sample and (c) SN sample.

Table 1 -
Sample designation and bentonite proportions (dry weight basis).