Introduction

Expansive soils are quite unstable and cause damage to the structures overlying them due to their volume change behavior. The swell shrink behavior of expansive soils causes damages to structures that are much higher than the damages caused by other natural disasters like earthquakes and floods [9, 12]. The behavior of these fine grained soils is better understood with their type of constituent mineral and unsaturated behavior arising from seasonal moisture variation of the field [12]. Previous studies show that the soil–water interaction induced by the climatic changes is very complex, and involves the coupled effects among the changes in water content, suction, stress, deformation and shear strength [13, 18].

The moisture absorption characteristic of a soil is mainly dependent on the pore size distribution, suction and mineralogy of a soil [11]. This expansive behavior of the clays is mainly attributed to the water absorption characteristic of minerals like Montmorillonite (MM) [11]. Hence, it is important to study both unsaturated soil response and clay mineralogy in better understanding the soil behavior. Soil suction concept and prediction procedures provide a better characterization of the behavior of expansive soils and a more reliable estimate of anticipated volume change for selected conditions based on comparisons with measured field behavior [15].

The soil water characteristic curve (SWCC) is a fundamental property governing soil–water interaction and is the relationship between matric suction (ψ) and water content. The matric suction corresponding to the sudden drop in the curve is referred to as air entry suction (ψa). The water content corresponding to the bend in the curve at low degrees of saturation is known as the residual water content (θr). Both these parameters define the shape of the curve along with the type of soil. Soils with smaller pores have high ψa. Soils with very large pore sizes exhibit greater changes in water content with matric suction [6]. Table 1 presents the two most commonly used SWCC models for expansive soils available in the literature. These models showed reliable fitting when tested for expansive soils and are being adopted in the present research.

Table 1 Soil water characteristic curve (SWCC) models
Full size table

Many researchers including Bernier et al. [2], Alonso et al. [1] and Delage et al. [4] studied the relationship between matric suction and swelling behavior of expansive clays. Leong et al. [10] conducted suction measurements using the WP4 Dew-point Potentiameter and measured total suction for two expansive clays. It was found that there was a close match between the total suction measured when compared with filter paper technique. Hoyos et al. [8] conducted SWCC tests on stabilized expansive clays and analyzed using Fredlund and Xing’s [7] SWCC model equation. SWCC test and modeling results show a relatively minor influence of cement dosage on the soil’s entry values and SWCC model parameters.

Pedarla et al. [14] conducted suction studies on two expansive clays with different mineralogy and found that mineralogy plays an important role in the swell-shrink and suction characteristics of the clay. Also, soil with higher percentage Montmorillonite (MM) showed higher swell-shrink characteristic. In a study conducted by Villar and Lloret [17] on sand bentonite mixtures, it was found out that the swelling of bentonite is directly related to the vertical pressure applied during saturation. The current study examines the influence of Montmorillonite (MM) presence in natural expansive clays. Soil water characteristic curve modeling parameters are also studied for the effect of MM presence.

Experimental Study

Four high plasticity expansive clays were selected for the present research study. These surficial disturbed soils were collected from the different regions across United States having different geological formations and distinct behavior. Standard laboratory tests were conducted on both the soil specimens in order to identify their behavior. The optimum moisture content and dry density properties were determined using Standard Proctor Test and these results are summarized in Table 2.

Table 2 Basic soil characterization
Full size table

In this research, clay mineralogy and individual mineral constituents are identified from the Chittoori and Puppala’s [3] technique. Chittoori and Puppala [3] used basic soil properties like total potassium (TP), specific surface area (SSA) and cation exchange capacity (CEC) to predict the expansive clay mineral contents in a soil. This method has been adopted in the current study to evaluate the mineralogy of the clays. Based on the USCS soil classification system all the soils were classified as high plasticity clays (CH). Grayson soil exhibited the maximum plasticity index value compared to the other soils. Also, the Montmorillonite content of the soil is high compared to the other soils.

Swell Behavioral Studies

One dimensional swell strain and swell pressure tests were performed in a conventional consolidometer setup following ASTM D-4546 method. After the compacted soil specimen was placed in the consolidometer an initial seating load of 7 kPa was applied. Once proper loading contact was achieved, the test setup was inundated with water and the specimen was allowed to undergo swelling in the vertical direction. The soil specimens were allowed to swell for a period of 24 h or until the swell readings reached plateau conditions. Swell deformation measurements were taken at regular time intervals and used to determine vertical swell strains. Load back swell pressure test is conducted on all soils. The total load applied to the specimen to bring back to its original position was then used to calculate its swell pressure. Appliance correction factors were applied to each of the swell pressure tests for the determination of actual swell pressure as suggested in [5]. Figure 1 presents the 1-D swell strains exhibited by Colorado soil specimen at two different compaction dry densities, maximum dry density (MDD) and 95 % of maximum dry density (95 % MDD). The specimen exhibited a maximum swell of 12 % at maximum dry density (MDD) and 8.2 % swell at 95 % MDD. Figures 2 and 3 present the load-back swell pressures obtained from Colorado soil at two different densities. Soil specimen at 95 % MDD exhibited a swell pressure of 137.7 kPa (20 ψ) and 194 kPa (28 ψ) at MDD condition. The particle and mineral density was very high in case of MDD condition which resulted in higher swelling behavior.

Fig. 1
figure 1

One dimensional swell strains for Colorado soil

Full size image
Fig. 2
figure 2

Load-back swell pressure test on Colorado soil at 95 % MDD

Full size image
Fig. 3
figure 3

Load-back swell pressure test on Colorado soil at MDD

Full size image

Specimens compacted at lower dry unit weight had experienced lower swell strains and exhibited lower swelling pressures. This reduction in swell strains were expected as compaction void ratio is larger at low dry unit weight condition and hence this void space accommodates a partial amount of swelling in the soil specimen. Table 3 below presents the 1-D swell strains and load back swell pressure result summary for all the selected soils.

Table 3 Summary of the 1-D swell strains and swell pressure test results
Full size table

Test results showed that the soil exhibiting the maximum swell strains was Grayson soil. The soils presented in Table 3 are in the order of their maximum swell strains exhibited in the present swell testing. Chemical activity derived from clay mineralogy influences the swell behavior of expansive clays. The soils exhibiting maximum swell strains (Grayson) also showed a higher swell pressure due to the presence of high amounts of Montmorillonite content present in this soil when compared to the same of the remaining soils. Soils compacted at maximum dry density (MDD) condition showed higher swell strains and swell pressures due to the presence of more soil and mineral content than at 95 % MDD.

Influence of Mineral Montmorillonite on swell behavior

Montmorillonite mineral due to its high cation exchange capacity and large specific surface area, plays an important role in the swell behavior of expansive clays. The presence of this mineral allows the clay to have higher swelling/shrinking capabilities. Figure 4 shown above presents the variation of 1 Dimensional swell (1-D swell) with Montmorillonite content. From the test results on four expansive clays it was observed that, with an increase in Montmorillonite content the soils exhibited more 1-D swell strains. Fitting models were plotted both for MDD and 95 % MDD condition. These fitting models were based on the test soils exhibit a coefficient of determination (R2) around 0.8. These models can be used to predict the one dimensional swell strains from the pre-determined mineral Montmorillonite content in a soil.

Fig. 4
figure 4

Variation of 1-D Vertical Swell with Montmorillonite content for expansive clays

Full size image

Figure shown below presents the variation of swell pressures of expansive clays with percent Montmorillonite content in a soil. The load back swell pressure tests were conducted on soil specimens compacted at MDD and 95 % MDD. From the test results it was observed that, with an increase in Montmorillonite content the soils exhibited more swell pressures (Fig. 5).

Fig. 5
figure 5

Variation of swell pressures with Montmorillonite content for expansive clays

Full size image

These fitting models were based on four soils and represent a coefficient of determination (R2) value around 0.9. From the above figures it can be noticed that the variation of swelling strains and swelling pressures are proportional to the percent Montmorillonite (%MM) content in soil. The curves are assumed to start from origin as the degree of expansion reduces tremendously with no %MM content in soil. However, Illite and Kaoline minerals minimally contribute towards soil swell behavior and are currently neglected.

Unsaturated Soil Behavior

Soil–water characteristic curves (SWCC) were measured using a pressure cell apparatus at low matric suction range and filter paper technique at high matric suction range. The pressure cell apparatus with a 500 kPa ceramic plate was used for measuring the SWCC following the drying path in the suction range of 0–500 kPa (ASTM D6836-02). The filter paper technique (ASTM D5298-10) was adopted for the matric suction for a range of 500–10,000 kPa. Samples were air dried to attain target moisture levels and once equilibrated and tested using filter paper technique. Figures 6, 7, 8 and 9, present the SWCCs measured for the suction range of 0–10,000 kPa. At select suction cycles, the specimen dimensions were measured. The volumetric change measured was then estimated and applied to the calculations of volumetric water contents corresponding to the respective values of suction pressures. Two models, Van Genuchten [16] and Fredlund and Xing [7] were used to fit the experimental data. The change in volume is predominant in expansive clays during application of suction conditions which alters the volumetric water content in the compacted soil specimens. Hence forth, care is taken to measure the volumetric changes during SWCC studies.

Fig. 6
figure 6

Soil water characteristic curve of Burleson soil

Full size image
Fig. 7
figure 7

Soil water characteristic curve of Colorado soil

Full size image
Fig. 8
figure 8

Soil water characteristic curve of Grayson soil

Full size image
Fig. 9
figure 9

Soil water characteristic curve of San Antonio soil

Full size image

All the high plasticity clayey specimens were statically compacted to 95 % MDD as it represents a near field compaction state. The variation of volumetric moisture content with matric suction for Burleson, Colorado, Grayson and San Antonio soil specimens are presented in Figs. 6, 7, 8 and 9, respectively. As stated in literature, experimental data fitted with standard Van Genuchten [16] and Fredlund and Xing [7] models and are presented along test results.

Burleson soil absorbed maximum saturated volumetric moisture content (θsat) of 50 % and showed an air entry value (ψa) of 74.5 kPa (10.8 ψ). The residual moisture content retained in the specimen (θres) was 5 %. Similar results were observed for the rest of high plasticity soils selected for this research. Unique soil suction parameters obtained from the measured soil water characteristic curves are presented in Table 4. The experimental SWCC data obtained from the laboratory measurements were plotted against the commonly used curve fitting models.

Table 4 Soil suction parameters obtained from measured SWCCs
Full size table

Table 5 presents the curve fitting parameters obtained from the Van Genuchten [16] and Fredlund and Xing [7] models plotted against all the four soils. The coefficient of determination (R2) for all the models are greater than or equal to 0.97. These models have proven to be reliable in modeling the soil water characteristic curves of present soils.

Table 5 Curve fitting parameters for SWCC models
Full size table

Analysis of Modeling Parameters

SWCC model parameters represent the inherent characteristics of the soil specimen. The ‘a’ parameter from Van Genuchten and ‘a1’ from Fredlund and Xing models represent the air entry head or pressure of a soil. Parameters ‘n’ and ‘n1’ govern the pore size distribution and ‘m’ and ‘m1’ represent the overall symmetry of the curve. Hence, for the four soils in the present research, variation of air entry head and pore distribution are studied.

Figures 10 and 11 present the variation of ‘a’ and ‘a1’ parameters with swell strain. ‘a’ and ‘a1’ experimental results were fitted with best fit models and a coefficient of determination (R2) is obtained for these models.

Fig. 10
figure 10

Variation of Van Genuchten’s SWCC modeling parameters with percent mineral Montmorillonite

Full size image
Fig. 11
figure 11

Variation of Fredlund and Xing’s SWCC modeling parameters with percent mineral Montmorillonite

Full size image

The ‘a’ and ‘a1’ parameters showed a consistent trend with expansive mineral content of the soil. Grayson soil exhibited the highest ‘a’ parameter from Van Genuchten model whereas Burleson soil has the least value. Parameters ‘n’ and ‘m’ from Van Genuchten model were inconsistent with percent mineral Montmorillonite. For the Fredlund and Xing fitting model, ‘a1’ parameter was highest for Burleson and least for Grayson soil. This inverse relationship with mineral content is due to the fact that ‘a1’ is an inverse function of pore size distribution for a soil.

From the above plots, it was concluded that the fitting parameters ‘n, m, n1, m1’ from both the models did not show a consistent trend in variation with expansive behavior of soils. Parameters ‘n, m, n1, m1’ did not exhibit any variation with change in clay fraction or percent mineral Montmorillonite. This could be due to the dependency of SWCC parameters solely on grain size distribution. Hence, it is premature at this point to draw conclusive trend between soil mineralogy parameters with the corresponding SWCC model parameters ‘n, m, n1, m1’.

Summary

Four natural expansive clays were selected for the present study and studied for basic soil characterization. Statically compacted soil specimens were tested for swelling strains and swelling pressure tests to understand their response to moisture imbibition. Grayson soil with dominant mineral Montmorillonite exhibited high swelling strains and swelling pressure compared with other soils. Swelling strains and swelling pressure results determined in the laboratory were proportional to the %MM (Montmorillonite) present in the soils. This re-emphasizes the dependency of swelling behavior on mineral Montmorillonite. However, swelling behavior caused from Illite fraction of the clay content is currently neglected in this study.

It is found that, air entry values and the residual moisture contents determined from SWCC are independent of the degree of expansiveness of a soil. From the analysis of SWCC modeling parameters, it is observed that the air entry or pressure head parameters showed direct relationship with %MM. Pore size distribution and gradation parameters n, m, n1, m1 did not exhibit any variation with change in clay fraction or percent mineral Montmorillonite. This could be due to the dependency of SWCC on grain size distribution, rather than the degree of expansion of a soil or the presence of mineral Montmorillonite. Future studies are needed on diverse soil types to develop trends with more than one soil composition related parameter.