Effect of suction on the mechanical behaviour of unsaturated compacted clay–sand mixtures

The sand–clay mixtures can be used for hydraulic barriers on slopes, as barriers for nuclear waste repositories or thermohydraulic backfills for geothermal boreholes. Moreover, the unsaturated compacted clay–sand mixtures can be used in civil engineering works such as construction of roads, dams and other types of embankments. They can be used also in raw earth building material. The sand–clay mixture is used generally in compacted fills and in compressed earth bricks (CEB) that require a high dry unit weight, as clay particles can fill the voids between sand particles. Further, mixing sand and clay (bentonite) may lead to a lower swelling pressure than that for clay alone [1]. Generally, the sand fraction is added to clayey soil to have a high shear strength or low compressibility or to reduce its swelling and the clay fraction is added to granular soil to reduce its permeability.

The behaviour of sand–clay mixture is governed by the granular phase when the sand matrix is predominant or by the cohesive phase when the clay matrix is predominant. Different studies were conducted to show the effect of sand or clay content on the behaviour of sand–clay mixtures. They generally showed that when sand was mixed with kaolin clay, the soil changed its behaviour from sand to clay. This clay content is enough to fill the voids of the granular portion at its maximum porosity. For example, Novais-Ferreira [2] described the existence of three zones of behaviour of the given sand–clay mixtures as a function of clay content: the first one, when the corresponding clay content is ≤28%, non-cohesive behaviour, where cohesion is negligible and the angle of friction is high (above 30°); the second one, when the clay content is between 28% and 41% (transition behaviour), where the soil is sensitive both to cohesion and the friction angle; the third one, when the clay content is ≥41%, cohesive behaviour, where cohesion is higher and the angle of friction is lower. However, Skempton [3] postulated for the clays that if the clay fraction is less than approximately 25%, the mixture behaves much like sand or silt rather than clay, but the residual strength is controlled almost entirely by the sliding friction of the clay minerals when the clay fraction is above 50%. Muir Wood and Kumar [4] demonstrated that the predominance of the mechanical behaviour of the clay matrix on the mechanical behaviour of the mixture occurs when the clay content is greater than 40% and the volume fraction of the granular constituent reaches about 0.45.

In recent years, many experimental tests and methods were conducted to study the hydro mechanical behaviour of sand–clay mixtures, such as the shear strength and the compression behaviour [4,5,6,7,8,9,10,11,12,13], the unconfined compressive strength (UCS) [14,15,16,17,18], the compaction characteristics [19,20,21,22] and the hydraulic properties [19, 23, 24]. In addition, the effect of suction and water content on the mechanical behaviour of sand–clay mixtures has also been investigated by many authors [13,14,15,16,17].

For shear strength, Vallejo and Mawby [5] showed that the shear strength is governed by the granular phase when the sand content is greater than 75% and by the cohesive phase when the sand content is lower than 40%. When the sand content is between 40% and 75%, the shear strength of the mixtures is partially controlled by the granular phase. Prakasha and Chandrasekaran [6] found that the inclusion of sand grains in a clay matrix leads to an increase in pore pressure resulting in a decrease in undrained shear strength, while Shafiee et al. [8] reported that the undrained shear strength increases with increasing sand content.

Likewise, Mun et al. [13] studied the impact of clay content on the undrained shear strength and on the compression curves of unsaturated sand–clay mixtures using the triaxial compression tests. The results indicated that sand–clay mixtures have lower undrained shear strength than sand or clay, but there is an intermediate rate of increase in shear strength with increasing strain rate, and the suction in the clay matrix has an important effect on the preconsolidation stress, which decreases as the percentage of clay increases. Pakbaz and Moqaddam [9] conducted consolidated-drained direct shear tests to study the effect of increasing clay content and the effect of gradation of sand on the behaviour of overconsolidated sand–clay mixtures and on their shear strength properties. The result of tests indicated that at a particular sand gradation, with an increase in clay content, the shear strength decreased. Also, at a constant clay content, the shear strength decreased with a decrease in sand grains’ size. Cabalar and Mustapha [11] also studied the effect of particle gradation of sands with distinct shapes (rounded and angular) on the variation of the liquid limit (LL) and on the shear strength of clay mixed with different percentages of these sands. The results of the conducted tests indicated a decrease of undrained shear strength with an increase in the amount of sand. The use of rounded sands in a clay matrix leads to the development of higher undrained shear strength values, on which gradation of the sands has no effect.

In terms of UCS, Anuchit [15] studied the effect of sand content for different applied suctions and concluded that the UCS increased with increasing sand content for all the ranges of applied suction. For the effect of suction on UCS of clayey soils with different sand contents, the author found that the UCS increased with increasing matric suction for the matric suction range less than 50 kPa, but it decreased with increasing matric suction for the suction range greater than 50 kPa, whatever was the sand content in the mixture. Khan et al. [16] studied the compressive strength of compacted natural clay of high plasticity mixed with 20% and 40% of sand. The authors found that the compressive strength decreased with an increase in sand content because of increased material heterogeneity and loss of sand grains from the sides during shearing tests. The authors also showed that the compressive strength increased with decreasing water content of the material and for each sand–clay mixture, the compressive strength increased with an increase in density of the material. Sun et al. [14] carried out series of one-dimensional compression tests on the unsaturated compacted sand–bentonite mixture using suction-controlled oedometer. The results indicated that the yield stress increases and the compression index decreases with an increase of suction.

For compacted sand–clay mixture, some of the researchers investigated the effect of the percentages of sand and clay on the soil structure and the compaction properties (optimum water content and dry density values). For example, Kenney et al. [19] observed that there may be an optimal clay content that would lead to an increase in the dry density of a compacted sand–bentonite mixture. However, Howell et al. [20] concluded that the addition of clay may lead to an increase or decrease in the optimal water content and maximum dry density under the standard Proctor compaction effort according to the type of processed clay soil (water sorptivity and swelling potential), curing period and the mixing procedure. Cabalar and Mustafa [21] carried out California bearing ratio (CBR), unconfined compression strength (UCS) and compaction tests on various contents of sand and clay mixture. The results showed a decrease in UCS and an increase in CBR values with an increase in sand content.

The mixture with an appropriate sand content is generally used to valorise the fine raw earth to an eco-building material like compressed earth block (CEB) and to improve its hydromechanical behaviour defined by the UCS and the strain modulus. These parameters are measured generally when the CEB is made at a water content near saturation to reach the maximum density. However, the CEB dries before being implemented in the building. Therefore, the main objective of this paper is to study the behaviour of the compacted clay–sand mixtures on the drying–-wetting paths, and then to investigate the influence of suction, water content and the added sand content on UCS and on the strain modulus of the compacted clay–sand mixture. To reach these purposes, series of unconfined compression tests at different water contents are performed on the kaolinite–sand mixtures compacted in standard Proctor properties. Furthermore, to link the water content to suction, the soil–water characteristic curves (SWCCs) of the compacted materials are highlighted by investigation of the effect of added sand percentage on the drying–wetting paths.

The clayey soils used in manufacturing of the building materials, dams and road embankments need to be amended generally with sand to meet the standards in strength and to reach the recommended behaviour. In this study, a commercial kaolinite (Sibelco, Hostun, France) noted 100K (100% kaolinite) is taken as a reference clayey soil. Two mixtures are prepared using this kaolinite and a French building commercial sand 0/4 provided by ULTIBAT and referenced ULTIBAT-91. These mixtures are noted as 75K25S (75% of kaolinite and 25% of sand) and 50K50S (50% of kaolinite and 50% of sand), respectively.

Many researchers have worked on the kaolinite 100K, both in terms of characterisation and mechanical behaviour [24,25,26,27,28,29]. Therefore, in this study, the lab experiments were focused on the soil properties of the mixture kaolinite–sand, such as the grain size distribution analysis, Atterberg limits and compaction. In accordance with the NF EN ISO 17892-4 [30] standard, the grain size distribution curves of the kaolinite 100K, sand and the mixtures 75K25S and 50K50S are shown in Fig. 1. The grain size disribution characteristic parameters are summarised in Table 1. The used sand is well-graded with a uniformity coefficient Cu of 3.15. The Atterberg limit tests were carried out in accordance with the NF EN ISO 17892-12 [31] standard to determine the liquid and plastic limits (LL, PL) of the soils. The results are plotted in Table 1. Fig. 2 presents the location of the kaolinite 100K and the mixtures 75K25S and 50K50S in Casagrande plasticity chart. These materials are classified as low plasticity clay. The LL and the plasticity index (PI) of the soil mixtures decrease with increasing sand percentage (SP) following a linear function as illustrated in Fig. 3. A similar linear variation was also reported previously [7, 11, 32]. Monkul and Ozden [7] found the same slope of the line PI (SP) which is equal to about 0.19 (%/%).

Figure 1

Figure 2

Figure 3

Standard Proctor tests were conducted in accordance with Afnor NF P 94 093 [34] standard. The curves highlighted the relationship between water content and dry density curves of the kaolinite 100K, and the mixtures are presented in Fig. 4. The Proctor optimum water contents and the maximum dry densities (wSPO, ρdmax) of the soils were determined. The optimum Proctor line and the saturation curves are also plotted in Fig. 4.

Figure 4

The optimum water contents and maximum dry densities of the soils are presented in Table 1. The kaolinite 100K showed a maximum dry density of 1.7 g/cm³ at an optimum water content wSPO equal to 17.2%, which is close to PL = 20%. These data are similar to those reported by Khan et al. [16], Azam and Chowdhury [35] and Marinho and Oliveira [36]. Marinho and Oliveira [36] reported that for cohesive soils, the optimum water content is within ±5% of the PL. It can be observed from Fig. 4 that an increase in sand content decreases the void ratio of the mixture, and thus increases its dry density from 1.7 g/cm³ (case 100K) to 1.77 and 1.93 g/cm3 for 75K25S and 50K50S, respectively. The same trend was observed by Khan et al. [16].

Fig. 4 shows that this increase in density due to the increase of sand content needs lower water content (wSPO). It also shows that, the optimum Proctor line and the 80% saturation curve are superposed, which reveals that the three soils are quasi-saturated at this state. Based on the analysis of dozens of soils around the world, Fleureau et al. [26] proposed correlations of some parameters of fine soils compacted at the optimum Proctor. These correlations link the optimal dry unit weights (γdmax) and the water contents to the LL as presented by Eqs 1 and 2. (1) wSPO= 1.99+0.46 LL−0.0012 LL2 {{\rm{w}}_{{\rm{SPO}}}} = \,1.99 + 0.46\,{\rm{LL}} - 0.0012\,{\rm{L}}{{\rm{L}}^2} (2) γdmax= 21−0.113 LL+0.00024 LL2 {\gamma _{{\rm{dmax}}}} = \,\,21 - 0.113\,{\rm{LL}} + 0.00024\,{\rm{L}}{{\rm{L}}^2} Therefore, in this section, the measured maximum dry unit weights and the optimum water contents in SPO of the studied soils were compared to the correlated values as shown in Table 2, to verify the obtained values. A good agreement was observed between the measured and the correlated values.

To determine the hydromechanical properties of sand–kaolinite mixtures, two different types of tests were conducted. The first one concerned the UCS test carried out on samples prepared in standard Proctor conditions (ρdmax and wSPO). The second one concerned the study of drying–wetting paths and of SWCC using different techniques.

3.1

The UCS test

The UCS tests were carried out according to the standard NF EN ISO 17892-7 [37]. The Kaolinite 100K and the mixtures 75K25S and 50K50S were first air dried and then wetted to the standard optimum water content wSPO and kept in a sealed bag for 24 h to homogenise the water content. Afterwards, the soils were compacted under quasi-static loading to reach ρdmax in a specific mould using top and bottom pistons simultaneously. This technique produces uniform and homogeneous samples and limits the density gradient along the z axis of the samples (Fig. 5). The specimens were cylindrical with a diameter of 50 mm and a height of 100 mm (Fig. 5), so a slenderness ratio equal to 2 was obtained, which was within the range (2.0–2.5) specified by the standard NF EN ISO 17892-7 [37]. After compaction, the specimens were wrapped in a waterproof plastic film and stored for 24 h. This allows the water content to homogenise within the compacted sample. Then, the UCS tests were carried out at a slow constant displacement rate of 0.0203 mm/min until failure, using an electromechanical press to avoid any dynamic effect. Finally, the water content was measured after failing.

Figure 5

From the stress–strain curve (q(ɛ1)) of kaolinite and mixture samples, a strain modulus noted ‘E50’, presented in Fig. 6, is then defined according to the standard NF EN ISO 17892-9 [38] as follows: –

The point corresponding to 50% of the maximum strength of the material is located on the stress–strain curve.

–

Then the secant line through the origin of the axes and through this point is drawn.

–

The slope of this secant line represents the large strain secant modulus called E50 modulus [39].

Figure 6

In this study, the effect of suction and granular fraction content on the mechanical properties of the unsaturated fine material was studied. The unconfined axial compressive loading tests were performed on different clay–sand mixtures prepared under standard Proctor conditions (ρdmax and wSPO), and then, the soil samples were air-dried to the required value of water content to characterise the evolution of the UCS and the strain secant modulus with the percentage of the added sand and the water content. In addition, the relationship between the water content and suction (SWCC) has been highlighted through the development of drying–wetting curves, which allows the effects of suction on UCS and on strain secant modulus to be described.

The compaction tests reveal that an increase in sand content contributes to a decrease in the void ratio, and hence to an increase in the density of the mixture and decrease in the corresponding optimum water content. The drying–wetting curves show that the shrinkage limit decreases with the increase in percentage of the added sand, and consequently increases with the LL of mixtures.

The main results of the added SP effect at a given state (water content, density and suction) highlight the following: –

When clay is mixed with 25% of sand, the UCS and E50 modulus of 75K25S samples increase to 37% and 70%, respectively, compared to those of clayey samples. The increase in strength is due to both the effective cohesion of the kaolinite and the friction generated by the added sand fraction.