Analysis of the behaviour of a natural expansive soil under cyclic drying and wetting

Abstract

Expansive soils swell and shrink regularly when subjected to moisture changes. Clayey soils are available worldwide and are a continual source of concern causing substantial damage to civil engineering structures. Cyclic expansion and shrinkage of clays and associated movements of foundations may result in cracking and fatigue to structures. In France, the damage caused by this phenomenon was estimated to be more than 3.3 billion euros in 2002 (Vincent in 3eme conference SIRNAT-Forum des journees pour la Prevention des Risques Naturels, Orleans, janv. 2003) and the Paris region is one of the most affected. The objective of this study is to investigate the swell–shrink behaviour of a natural clayey soil considered to be responsible for a lot of damage observed on buildings in the Paris region, and thus contributing to the characterisation and understanding of expansive clayey soils. The studied soil, Argile verte de Romainville, is a lagoonal-marine deposit and is part of the Paris Basin Tertiary (Oligocene) formations (Fig. 1). It is a clayey soil sampled in the eastern region of Paris. The mineralogical and geotechnical properties of the soil are presented in Table 1. The soil contains quartz (15–20%), carbonates (12–20%) and traces of mica and feldspars. X-ray diffraction showed that carbonates are essentially dolomite and the clay minerals are dominantly illite, kaolinite and a small amount of smectite (Fig. 2). A grain size analysis shows that the clay content (<2 μm) varies between 78 and 80%. The study of its microstructure by means of the scanning electron microscope indicates that the clayey soil has structural elements oriented in the direction of bedding. The structure of the sample generally consisted of dense and continuous clay matrices with very limited visible pore spaces (Fig. 3). At its natural water content (w = 25%), the soil shows mainly a unimodal pore size distribution with an average pore radius of 0.07 μm and a very limited porosity with radii larger than 10 μm (Fig. 4). To assess the effect of suction on the simultaneous changes in void ratio and degree of saturation under zero external stresses, drying–wetting tests are performed on the natural samples. The osmotic technique (Polyethylene glycol solutions) and various salt solutions are used to control the suction values ranging from 1 to 300 MPa. Once equilibrium is reached at the given suction, the samples are weighed and their volume is measured. A synthesis of the drying–wetting paths is given on Fig. 5. The swelling potential of the soil is evaluated using both indirect (or empirical methods Tables 2 and 3) and direct methods. Swell percentage and swell pressure of the soil are measured in a conventional oedometer apparatus according to ASTM (D 4546-85). The test specimens are 70 mm in diameter and the height varies between 12 and 24 mm. The swell percentage is measured under a nominal pressure of 0.7, 2.0 and 6.3 kPa. Swelling pressure of the soil is measured by the conventional consolidation test method (free swell and load, ASTM D 4546-85 method A) and by a constant volume method (ASTM D 4546-85 method C). The test parameters and results for each specimen are given in Tables 4 and 5, and on Fig. 7. Cyclic swell–shrink tests are carried out on similar samples taken from the same monolith. A scheme that permits the study of the clayey soil behaviour at the extreme states of wetting and drying is chosen. The test begins by wetting the samples at their natural moisture content and density. When swelling is stabilized, the water is removed from around the samples and they are dried in an oven maintained at 45°C until the vertical deformation (shrinkage) is stabilised and are then rewetted and so on. Some experiments are stopped at different swelling phases for microstructural study of the soil. The test parameters of the specimens are given in Table 9 and the results are shown in Figs. 9 and 10. The evolution of the microstructure during wetting and drying cycles is investigated using scanning electron microscope and mercury intrusion porosimetry. Observations are made only on soil specimens taken at the end of the swelling phase of the selected cycles. In order to preserve the microstructure, the specimens are cut in small pieces, frozen by liquid nitrogen and finally sublimated. The results of the drying–wetting path including the water retention curve are shown on Fig. 5. The results show that on the drying path (in the void ratio versus water content plane) the soil first follows nearly the saturation line and then, as the water content decreases, the void ratio tends towards a constant value. A shrinkage limit of w = 14.5 % and a corresponding suction value of 15 MPa is deduced from this path. An air entry value of 10 MPa is obtained from degree of saturation versus suction curve. The wetting path shows that the wetting–drying path is reversible for suction values higher than 60 MPa. The different indirect methods used to assess the swelling potential of the Argile verte de Romainville show a general agreement with respect to its swelling potential ranging from high to very high (Table 3). Examination of the free swell test results shows that the Argile verte de Romainville exhibits swell percentage in the range of 15–26% and that its degree of swelling depends on the initial conditions (water content, dry density) and the applied load (Table 4). The higher the water content and the applied load, the lower the swell percentage. A specimen taken parallel to the bedding plane shows similar values of swell percentage with a steep volume change versus time curve indicating an anisotropy of permeability. The two direct methods used to assess the swelling pressure of the Argile verte de Romainville give different values (Table 5). The values obtained by the constant volume method are relatively close and are about 700 kPa. Lower values varying between 360 and 540 kPa are obtained by the conventional consolidation test (free swell-consolidation). This indicates that besides the initial conditions, the swelling pressure is strongly dependent on the stress path followed. The results obtained from the wetting–drying cycle tests show that the magnitude of the first swell cycle is controlled by the initial water content, the maximum deformation occurring on the second cycle and the stabilization of swelling deformation from the third cycle (Figs. 9, 10). Furthermore, the experimental data indicate that upon repeated wetting and drying, the swelling rate of the soil becomes faster, which is explained by an increase in permeability of the soil due to the development of preferential flow paths (micro cracks) on drying. With an increasing number of cycles, a permanent increase in the volume of the samples is observed. This suggests that the swelling–shrinkage behaviour of expansive soils is not completely reversible. Mercury intrusion porosimetry analysis and SEM observations before and after different numbers of cyclic swelling indicate that the swelling–shrinkage cycles are accompanied by a continual reconstruction of the soil structure (Figs. 11, 12). The mercury intrusion porosimetry results show that with an increasing number of wetting–drying cycles the pore volume and the average diameter of the pores increase progressively (Fig. 11). Larger modifications are observed in the pores with radius in the range of 0.1–5 μm. SEM observations also show further destruction of large aggregates and disorientation of structural elements as the number of cycles increases (Fig. 12). After the fifth cycle, the soil original structure is totally lost and a disoriented homogeneous and loose structure with more homogeneous pore spaces is observed (Fig. 12d).