Geotechnical Hazards Induced by Montmorillonite Presence

Abstract

Often referred to as expansive soils in international literature, soils with large swelling and shrinking potential, are defined as natural materials that exhibit volume variations related to variations of moisture. Most professionals consider that the expansive manifestations are related to the mineralogic composition, especially with the presence of the smectite class of minerals, more specifically with the montmorillonite. Despite sustained worldwide studies over decades, the direct and firm correlation between the number of clayey minerals and expansive properties has not reached a conclusive form, and the behavior of expansive soils remains still unrevealed and exhibits unexpected features under moistening or drying conditions in natural habit, or in relation with infrastructure works. This chapter presents the results of an extended and complex geotechnical investigation of expansive soils which concludes with the validation of a simple procedure of identification of the areas where the swelling behavior of soils may exceed the equilibrium of the geological structure and produce a variety of effects such as lumps or landslides in areas with sloping terrain.

1. Introduction

Expansive soils are implicitly pointed out associated with large damage to onshore or offshore, over, or underground infrastructures projects in more than 60 countries, on all continents except the Antarctic one, often after long periods of drought, heavy rains, freezing, or unfreezing.

The broadening of expansive soils is usually incompletely presented in geologic or pedologic maps, only for the superficial parts of the ground associated with highly over-consolidated claystone and clay shale [1].

Despite worldwide efforts to depict, understand, and predict these peculiarities, annual damages related to geomechanical processes produced by variations of water content in expansive soils are reported all over the world.

For instance, in UK-£400 million, [2], or in the USA- $15 billion, [3]. In the UK, or in China where expansive soil covers more than one hundred thousand square kilometers, expansive soil behavior is stated as one of the most dangerous geological hazards that affect large and diverse engineering projects [4, 5, 6].

Even if the specific influence of these soils is not yet quantified in Romania, large areas and important infrastructures are also affected mainly by landslides.

This generally unfavorable situation is the result of the cumulation of various factors related to (i) incomplete knowledge of these soils in the various conditions of occurrence, both as mineralogical nature and sedimentologic structure; (ii) poor understanding of geotechnical behavior and its limits; (iii) insufficient sampling and specific testing in large projects which may put in evidence the presence and the specific properties of these soils, and finally (iv) the inconsistence bond between geotechnical and hydrogeological investigations.

2. Characterization of expansive soils

2.1 The mineralogical nature

The swelling processes are always associated with the presence of clay minerals. Also named hydrous phyllosilicate, the clay minerals are divided into several classes based on the specific structure composed of silica, alumina, and water with variable content of inorganic ions like Mg²⁺, Na⁺, and Ca²⁺, [7].

There are four major groups of clay minerals that are present in soils. Among, them the layer group is the most important one which comprises: type 1:1 (kaolinite, halloysite, serpentine); type 2:1 (smectite, vermiculite, illite, mica), and type 2:1:1 (chlorite).

The largest swelling potentials are related to the presence of minerals from the smectite family, which contains saponites, montmorillonite, and bentonite, and absorb the largest quantities of water between clay sheets [7, 8].

Physico-chemical characteristics of minerals that dictate the amplitude of swelling potential are in order of impact: average specific surface area, particle thickness, interlayer space, and cation exchange capacity. Table 1 presents these characteristics specific to some usually encountered clay minerals.

Type	Clay mineral	Particle thickness	Basal spacing	Cation exchange capacity at pH 7	Specific; area	Swelling potential
		(nm)	(Å)	(milliequivalent /100 g)	(m^2/g)
2:1	Smectite-Montmorillonite	2	9.8–20	80–120	40–800	High
2:1	Vermiculite	—	10–15	100–150	760	High
2:1	Illite	10	10–13	10–10	80–120	High
1:1	Kaolinite	100	7.2	3–15	5–40	Almost none
2:1:1	Chlorite	—	14	10–40	10–55	None

Table 1.

Physico-chemical characteristics of minerals [7, 9].

Every particular soil layer is a mixture of clay minerals with other particles of inert mineral type (quartz, feldspar, etc.), a fact that triggers a unique swelling and shrinking behavior. It is considered that clays which are geological materials of great mineralogical variety, may exhibit swelling properties if containing mineral particles more than 50% with less than 2-micron size.

It is concluded [9] that usually, the presence of montmorillonite indicates large probabilities of producing reversible processes, swelling or shrinking, under moisture variation, while soils with a predominance of kaolinite or illite exhibit at drying, in the first instance, an initial large volume decrease and subsequently limited swelling on rewetting.

2.2 Micro and macroscale “symptoms” of expansive soils

In the microscale register, swelling and shrinkage mechanisms of expansive soils may be explained by two major theories: crystal swelling of clay mineralogy in which the water molecules get combined with the cations associated with mineral flakes to form hydrated ones and a diffuse double layer of colloidal chemistry in which around negatively charged clay minerals hydrated ions and polar water molecules are firmly adsorbed and form a fixed layer and a thick diffusion layer thus widening the spacing between mineral particles [10]. During the swelling process, the inter-clay bonds are weakened or broken, the layer structure of clay minerals is deformed, and in the macroscopic field, the strength of the soil decreases.

The magnitude of the swelling-shrinking potential is strictly dependent on the types and the percents of clay minerals and is also directly related to the initial saturation degree, the initial void ratio, or the geologic vertical stress [11].

Over decades, the specific geomechanical behavior of expansive soils has been characterized as dramatic either by:

1. the total collapse of soil structure after a single swell/shrink cycle [12, 13];

2. the decrease of shear strength from 3 to 5.5 times [14];

3. the decline with 70–90% of shear resistance, mostly by the reduction of effective cohesion, sometimes very close to or equal to zero [15, 16].

Moreover, the shrinking processes should not be considered reversible, since cracks created after shrinkage may not always completely close up after rewetting, often only deepening the watering front together with the sediment deposition [17].

The area of manifestation of these processes is predominately located at the surface of the terrains, wherever is taking place the seasonal variation of underground water [4], on depths less than 5–6 m [6], according to climate particularities, drying/rehydration regime and mostly to variation of temperature.

Usually, swelling of expansive soils is considered only in the vertical direction, but it has been proved that lateral swelling pressure develops additional stress to the lateral earth pressure on retaining walls which may increase at the bottom of the wall equal to 1.3–4 times the overburden [18], and reduce the bearing capacity of piles [1, 19].

Nonetheless, similar processes may be present at larger depths, by the water flow through more permeable layers (in contact with expansive soils) charged from distant front-loading sections [20].

2.3 Geotechnical identification tests

Various procedures for the identification of expansive soils may be established following the international literature, national codes, or professional guides, which are based on a large variety of parameters, indices, or techniques, however, none of those are comprehensive and worldwide accepted. Thus, in countries with predominantly arid climates, the geotechnical methodology for expansive soil characterization is based on shrinkage tests, while in temperate climates swelling tests prevail.

Jones and Jefferson, [4] differentiate the following laboratory testing classes: (i) swelling tests, which may be divided into swelling strain and swelling pressure tests; (ii) index tests based on basic parameters such as liquid limit, plastic limit, plasticity index (usual or modified); (iii) oedometer-based methods, which may be free swell tests or constant volume tests; (iv) suction-based tests which use soil-water characteristic curves; (v) mineralogical tests.

Nelson et al. [1], define three classes of identification methods: (i) based on physical properties – plasticity, free swell test, potential volume change, expansion index, linear extensibility, and standard absorption of moisture content; (ii) based on mineralogical composition–X-ray diffraction, differential thermal analysis, and electron microscopy; (iii) based on chemical analyses – cation exchange capacity, specific surface area, and total potassium content.

Based on the above-defined tests, various schemes of characterization and classification have been developed worldwide, which may be generally categorized into four groups [4]: free swell, heave potential, degree of expansiveness, and shrinkage potential.

Romanian normative [21] propose a scheme of characterization based on a particular collection of parameters: colloidal fraction (A2μ), plasticity index (Ip), activity index (I_A), plasticity criterion (Cp), free swelling (U_L), shrinkage limit (w_s), volumetric shrinkage (Cv), maximum wetting heat (q_max), moisture content at 15 bar suction (w₁₅), and swelling pressure (p_u).

3. Geotechnical hazard induced by montmorillonite presence

3.1 Geologic and hydrogeologic regional frame

As we underline above, swelling processes may be active at larger depths, by the water flow through more permeable layers (in contact with expansive soils) charged from distant front-loading sections.

This is the case of a large structure attributed to the Upper Pliocene-Lower Pleistocene named “Cândeşti Layers”, in which expansive soil strata are inconsistently wetted at large and variable pressures.

The geological unit we refer to is deposed over the Getic Depression and Moesian Platform units, on the outer side of the Carpathian Arch (Oriental, Curvature, and Meridional) [22].

It has a thickness that may exceed 250 m and an irregular shape with widths starting from 6 to 8 km in the Curvature area, to more than 80 km in the western Meridional Carpathian as it is presented in Figure 1 [23].

Figure 1.

Extent of the Artesian Dacic Basin [23].

The “Cândeşti Layers” is considered a multilayer-aquifer (with terms from Jurassic to Upper Pleistocene) with a feeding front in the North end at the contact with the Carpathian Chain.

The main underground flowing directions are from Nord to South, and the hydrogeological characteristics are hydraulic conductivities in the order of about 100 m/day and transmissivity of less than 1000 m²/day.

The hydrodynamic regime is defined by high pressures (up to 40 Bars), generated by deeper regional aquifers, with which it is in connection. Often in places, the package of multilayer-aquifer, defined also as an “Artesian Dacic Basin”, may function under a pressure regime that surpasses the surface of the terrain, causing the watering from the bottom up of the clayey aquitard layers [23].

In this geological and hydrogeological context, the swelling hazard emerges from the rhythmic sedimentation regime which is materialized in alternant sequential layers separated in two main terms: the course one (gravels and sands) and the fine one (expansive clays and fine sand), disposed in the peculiar geologic structure which is presented in Figure 2.

Figure 2.

Exemplification of the mechanism of ascensional watering of expansive soils. Specific geologic cross section of Artesian Dacic Basin [24].

3.2 Geotechnical and mineralogical characterization of “Cândeşti layers”

3.2.1 Geotechnical identification

The most common identification tests performed worldwide are grain size distribution and plasticity (Atterberg) limits, which have been executed on the entire set of samples with expansive behavior.

The results of identification tests, presented in Table 2, have been divided into three classes according to the plasticity index: class C1- of medium plasticity; class C2- of high plasticity, and C3 of very high plasticity.

Class of plasti-city	C1 (Ip < 20%)		C2 (20% ≤ Ip < 35%)		C3 (Ip ≥ 35%)		Total no of samples	[%]
	No of samples	[%]	No of samples	[%]	No of samples	[%]
Cl	—	—	5	2	311	75	316	48
si Cl	1	4	91	41	65	16	157	24
sasi Cl	21	75	73	33	4	1	98	15
sa Cl	3	11	50	22	32	8	85	13
cl Si	—	—	4	2	—	—	4	<1
cl Sa	3	11	—	—	—	—	3	<1
Total	28	100	223	100	412	100	663	100

Table 2.

Results of geotechnical identification of soils.

3.2.2 Expansive properties

Among the many geotechnical properties supposed to characterize the expansivity of soils, we chose to analyze the most representative ones: free swelling (U_L) and swelling pressure (p_u) in relation to colloidal fraction (A2μ) and plasticity index (Ip).

3.2.2.1 Free swelling (U_L)

It is defined as the increase in the volume of a soil sample, without any external constraints, when submerged in water, the range of values of this parameter may point to soils less active (U_L < 70%), with medium activity (70% < U_L < 100%), active (100% < U_L < 140%), or very active (U_L > 140%), in relation with water, according to Romanian normative [21].

In Figure 3 the values of U_L are plotted in different colors, according to the class of plasticity to which they are attributed. It presents the variation in depth of U_L and reveals that according to this criterion, 35% of samples have medium activity, 28% are active and 24% are very active. Concerning the last category of very active samples, we mention that most part of those (79%) have U_L < 300%, and the highest recorded value is U_L = 603%.

Figure 3.

Variation of U_L in-depth.

Figure 4a and c reveals poor correlations between U_L and Ip or A2μ unfulfilled by unexpected great values of U_L registered in all plasticity classes. After excluding the highest values of free swelling, (U_L > 190%), from both ranges, the variabilities UL = U_L(Ip) and U_L = U_L(A2μ) have settled in good linear correspondences.

Figure 4.

(a and b) Variability between U_L and Ip. (c and d) Variability between U_L and A2μ.

3.2.2.2 Swelling pressure

It is defined as the external pressure required to consolidate a sample to its initial void ratio, after the expansion due to water submersion, or the pressure required to hold the soil at constant volume when water is added, the swelling pressure may be tested by two [1] or three methods [25].

The most common are the conventional consolidation test which usually provides upper bound values and the constant volume method. Less used is the method of equilibrium of void ratios which gives lower values. In this research, the conventional consolidation test has been used.

For smooth processing of the data, an additional classification of samples has been made according to the range of values of this important parameter. The new system overlaps the precedent one and is presented in Table 3.

Class of swelling pressures		C1		C2		C3		Total no of samples	[%]
		No	[%]	No	[%]	No	[%]
(I)	pu < 200 kPa	28	100	203	91	254	62	485	73
(II)	200 ≤ pu < 400 kPa	—	—	16	7	114	28	130	20
(III)	pu ≥400 kPa	—	—	4	2	44	11	48	7

Table 3.

Classification system according to swelling pressure results.

As one can visually observe from Figure 5, about three-quarters of the total amount of samples (73%) present swelling pressures under 200 kPa (class I). Samples of medium and high plasticity (C1-C2) are entirely enclosed in this pressure class (100–91%), while samples of very high plasticity (C3) present this range of swelling pressures at 62%.

Figure 5.

Variation of p_u in depth.

The highest values of swelling pressure grouped in class (III) represent 7% of the samples, which belongs mainly to class (C3). Concerning the class (III) of swelling pressure, we underline that most part of these samples (80%) have p_u < 600 kPa, and the highest value recorded exceeds 1000 kPa.

Figure 6a and c presents the correspondence between p_u and Ip, respectively p_u and A2μ, depicted according to both systems of classification. Once again unpredictable, high values of swelling pressure, encountered in classes C2-II, C2-III, and C3-III make the correlations weak, and it still remains poor even after excluding some unusual values (p_u > 400 kPa).

Figure 6.

(a and b) Correspondence between p_u and Ip. (c and d) Correspondence between p_u and A2μ.

Finally, for some of the samples that have been tested both for free swelling and for swelling pressure (Figure 7a), a medium-strong correlation has been established after excluding values with p_u > 400 kPa (Figure 7b).

Figure 7.

Correspondence between p_u and U_L.

3.2.3 Mineralogical identification

As part of the specific characterization of Cândeşti Layers, several mineralogical analyses have been executed which specified that the swelling minerals prevailed in all tested samples as is presented in Table 4.

Class	Swelling pressure classes pu (kPa)	Samples (10)	Percents of minerals (%) X-ray diffraction
		Number	Smectite	Illit	Kaolinit	Chlorit
II	200 < pu < 400	4	70–77	18–25	4–6	—
III	pu > 400	6	63–79	17–32	2–4	3–8

Table 4.

Mineralogic characterization of Cândeşti Layers samples.

The result of these mineralogical analyses confirms the presumption of the presence of hydrous phyllosilicate in expansive soils (pressure classes II and III). More than that, the fact that smectite class of clay minerals – which includes the montmorillonite - prevails in all tested samples explains the unusual values of geotechnical parameters U_L and p_u recorded in this geologic unit.

3.3 Quantification of geotechnical hazard of “Cândeşti Layers”

Starting from the hydrodynamic characterization of the multi-layer aquifer “Cândeşti Layers” exposed above (in §3.1.), the assessment of the geotechnical risk of swelling of aquitards expansive terms submitted to pressures of underground waters raised at the regional scale, up to 40 Bars, conducted to postulate that risk of moistening by vertical ascendant drainage is high and affects the whole area of expansion of this unit.

By considering the wetting zone in acceptance of [26], as the zone in which water contents may increase by supply from external sources, including by capillarity rise, the peculiar sedimentologic and hydrogeologic architecture described above, conduct in geotechnical terms to the extension in depth of it, up to the thickness of Cândeşti Aquifer.

In consequence, we began to evaluate the depth of the potential heave as the maximum depth at which the swelling pressure of the soil equals or exceeds overburden vertical stress [26], a quantity that we will note ${\Delta }_{sw}$, according to (Eq. 1).

$${\Delta }_{sw}={\sigma }_g-p_u$$E1

Figure 8a presents the repartition in depth of the values of $\left({\Delta }_{sw}\right)$, which allows us to observe that for a large part of tested samples, this difference is positive $\left({\Delta }_{sw}>0\right)$, and thus, the presence of expansive soils in foundation terrains of shallow or deep foundations is outside of this geotechnical hazard. Nevertheless, 9% of C2 and 25% of C3 (summing up 19% of all expansive samples, regardless of swelling pressure class), reveal negative values of $\left({\Delta }_{sw}<0\right)$, which are presented in detail in Figure 8b.

Figure 8.

(a) Variation in depth of the potential heave of Cândeşti Layers. (b) Detail of negative values of the potential heave.

One can observe that the main part of these negative differences (77%) are less than -200 kPa while 9% of samples present values between -800 kPa and -400 kPa and are located in the first 10 m depth from the surface of the terrain. The whole range of negative values is extended in depth up to almost 29 m, which is in consequence the thickness of the potential heave.

In cases of special interest and limited extension, a detailed design must be based furthermore on more specific laboratory trials of every expansive layer situated in the active zone, such as consolidation-swell tests and constant volume tests, which will allow a proper settlement calculation [1, 20].

In cases of broad infrastructure projects that run through geological structures exposed to swelling hazards on large depths, such as the one presented above, the calculations of $\left({\Delta }_{sw}\right)$ values must be considered as a qualitative evaluation and a fundamental first step in the process of assessment of this geotechnical risk. The assessment process once triggered in specified areas, must be continued with denser and specific tests in order to reveal the precise extension of these types of soils.

This simple procedure based on the most common geotechnical tests, executed in the preliminary stages of investigations, may point not only to the extension of expansive soils but also to the areas where the swelling behavior of soils may exceed the equilibrium of the geological structure and produce a variety of effects such as lumps or even landslides in areas with sloping terrain.

As an exemplification of this procedure, Figure 9a presents a vertical section alongside an infrastructure route that crosses a zone located in the Artesian Dacic Basin, with the distribution of isobars of swelling pressures recorded at different depths. Details represented in Figure 9b, allow the visualization of the specific extension in depth of zones of potential heave, and consequently the extension of the swelling geotechnical risk.

Figure 9.

(a) Chart of the distribution of zones of potential heave alongside an infrastructure route. (b) Detail with extension in depth of zone of potential heave.

4. Conclusions and outlook

In the worldwide context of poor mineralogical testing associated with geotechnical investigations required for large infrastructure objectives, the random presence of montmorillonite inside sedimentary layers represents a strong geotechnical hazard, that may produce large damages of shallow or deep foundations.

The paper presents such a study case in which exceptional swelling pressures have been revealed by numerous specific geotechnical investigations (663 expansive soil samples) to whom they have added only several mineralogical analyses by X-ray diffraction.

In order to characterize the swelling properties of soils, two specific parameters have been analyzed (U_L and p_u) in correlations with two general parameters (Ip and A2μ), in the frames of two classification systems. The main results may be synthesized as follows:

• concerning the free swelling parameter:

  • the highest recorded value is U_L = 603%;

  • the most part of samples (76%) presented values U_L < 140%;

  • variabilities U_L = U_L(Ip) and U_L = U_L(A2μ) reveal weak correlations due to unexpectedly great values of U_L registered in all plasticity classes, which have settled in good linear correspondences (R² = 0.71–0.75) after excluding from both ranges the highest values of free swelling (U_L > 190%);

• concerning the swelling pressure parameter:

  • the highest value recorded is p_u = 1080 kPa, followed by 7% of samples with p_u > 400 kPa;

  • variabilities p_u = p_u (Ip) and p_u = p_u (A2μ) reveal weak correlations due to unexpected great values of p_u, which remains in poor correspondences (R² = 0.28–0.31) even after excluding from both ranges the highest values of swelling pressures (p_u > 400 kPa);

• for some of the samples that have been tested both for free swelling and for swelling pressure (133 samples), a medium-strong correlation p_u = p_u (U_L) with R² = 0.57, has been established after excluding values with p_u > 400 kPa;

• the lack of strong and “natural” correspondence between swelling parameters (U_L, p_u) and general parameters (Ip, A2μ) is due to the randomness presence of hydrous phyllosilicate minerals, namely smectite class which includes the montmorillonite, a well-known responsible for the expansion of clayey soils;

• for nine samples collected from swelling pressure classes II and III the percentual amount of smectite class was in great amount, between 63% and 79%;

• given the fact that the distribution of montmorillonite across the entire sedimentary structure of various thicknesses is randomly, and in the context of peculiar hydrogeologic architecture of Cândeşti Aquifer (multistate-aquifer with high pressiometric regime), the probability of wetting of expansive soils at various depth is high; all these findings drive to a general characterization of these soils in natural conditions or in relation with great infrastructure projects, as having a hazardous geotechnical behavior;

• finally, the evaluation of the depth of the potential heave according to [26], offers quantitative spatial indications regarding the active zone in which the terrains may be affected by swelling/shrinking processes.

Based on the previous conclusions we suggest several directions to conduct proper investigations in the area of occurrence of this particular swelling formation, in order to reduce the degree of uncertainty of knowledge, and thus to reduce the geotechnical hazard related to the presence of montmorillonite in geological structures by:

• intensive boosting the research activities of swelling potential mapping at the level of every country involved in large infrastructure projects, regardless of the source of funding;

• the swelling potential mapping activities should:

  • be executed by interdisciplinary teams composed of geologists specialized in clay mineralogy, engineering geologists, hydrogeologists, and civil engineers;

  • start from an up-to-date geologic map, combined with a regional hydrogeologic map containing the regional underground water corps at convenient scales (1: 50000 to 1:15000) in order to be of practical use, not just scientific;

  • be based on geological and geotechnical investigation reports executed in the concern area, which must be accessed based on the principle of public interest;

  • be focused not only on the extent of expansive soils in 3D directions but also on the limits of variation of superficial underground water table levels and of piezometric regime of deeper aquifers in hydraulic contact with expansive layers;

  • be accompanied either by a digital 3D model with the extent of expansive soil layers combined with upper and lower limits of water table/piezometric levels, or by a dense grid of cross and longitudinal sections extended in depth up to the level of the deeper potential heave reported in the area;

  • contain also delimitations of the areas of landslides (regardless of the stage of evolution) and superficial damages to buildings (civil, industrial, or infrastructures);

• community standardization of types, numbers, and depth of geological and geotechnical investigations regardless of the purpose, in areas of expansive clay occurrences;

• community standardization of types, numbers, and density of laboratory geotechnical and mineralogical analyses that must lead to a correct and complete characterization of expansive layers, according to the stage of further investigations (preliminary, design, detail, or monitoring).

Swelling potential maps should be exposed and promoted in an open access regime, but in the meantime must be enclosed and endorsed by national standard normative and become mandatory to consider in order to prevent further damage to individual or large infrastructure investments.

Acknowledgments

The authors are grateful to the company Geotesting CI, Bucharest, Romania, for the permission to use the laboratory data included in this paper.