Geological and Geotechnical Characteristics of London Clay from the Isle of Sheppey

Since 1960s, the Isle of Sheppey along with the Isle of Grain, Cliffe, and Foulness Island have been labelled as potential sites for construction of a new airport as an alternative to the Heathrow airport to reduce the aircraft noise pollution in south east of England (Helsey and Codd 2014). The proposals of new airport in the Isle of Sheppey, the birthplace of British aviation, although rejected several times by the Department for Transport, implies the significance of this site which may in future be considered for wider infrastructure investments and developments. The ground profile of the Isle of Sheppey consists of deep London clay (LC) deposits. Very few works can be found in the literature on characterisation of the Sheppey LC (e.g. Burnett and Fookes 1974; Huggett 1994), despite the necessity for developing a clear understanding of the geological and hydro-mechanical properties of this soil. In one of the few studies on Sheppey soils, Jardine et al. (2003) reported characterisation of a soft peaty clay deposit found at the Queenborough bypass viaduct site on the north of the Isle of Sheppey. This temperate estuarine marsh clay is known as Queensborough clay (QC) and has been the subject of several studies (e.g. Smith 1992). The QC was reported to be mixed with material from surrounding LC hills and cliffs as a result of mass movement and sea erosion (Jardine et al. 2003). The reconstituted QC was found to have identical Atterberg limits, grading curve, and intrinsic compression parameters to the reconstituted LC samples investigated and reported by Jardine (1985). Hence, still there appears to be a lack of technical knowledge on characteristics of the Sheppey LC. An experimental program was, therefore, devised to fill this gap and add valuable information to the existing database. Undisturbed block samples of 350 mm cube were taken at 4 m depth below non-quarried ground level at New Hook farm reservoir site located on the outskirts of Sheerness town on the Isle of Sheppey. This site is owned by SW Attwood & Partners Ltd. The principal operation of the site is a fishing pond and water reservoir and at the time of sampling, it was in its initial stages of excavation. Figure 1 presents the site map and location of block sampling.

Commercial laboratory testing was carried out by CET Structures Ltd and included contamination analysis, permeability and in-situ water content measurements. Preliminary laboratory testing for microstructural analysis and determination of the index properties, chemical composition, and one-dimensional compressibility were carried out at the University of Nottingham and The University of Warwick. The obtained results are compared with the corresponding characteristics of the shallow depth LC from Terminal 5 (T5) site at Heathrow airport, reported in the literature.

King (1981) studied the biostratigraphy and lithological variations of LC formation and suggested a division of the formation into five principal units, named Divisions A to E. These divisions are subdivided into several members within the LC formation, named as; Walton, Claygate, Ockendon, Aveley, Sheppey, and Bracknell Members. The LC formation in Sheppey cliffs, Isle of Sheppey, which exposes the upper part of the LC formation is named Sheppey Member and corresponds approximately to the upper part of Division C, Division D and most of Division E (King 1981). The thickness of the Sheppey Member is approximately 55 m and is classified as Thames Group. According to Aldiss (2014), the Sheppey Member mainly comprises silty clay diffusely interbedded with sandy clayey silt and contains several layers of calcareous concretions.

Burnett and Fookes (1974) reported that the Sheppey LC comprises of 58% of particles < 2 µm size and 42% particles > 2 µm size. The particles < 2 µm size were approximately 90% clay minerals which were found to comprise 57% montmorillonite (a member of smectite group), 24% illite, 12% kaolinite, and 7% chlorite. The particles > 2 µm size were mainly silt which was found to comprise 70% quartz carbonates, 15% pyrite, and 15% clay minerals. Similar compositions were reported by Gasparre (2005) for Unit C (7 m depth) T5 LC, which comprises 53% of particles < 2 µm size formed mainly of 60% illite–smectite, 22% illite, 15% kaolinite, and 4% chlorite.

The macro-fabric of LC is characterised by the presence of discontinuities which play a significant role in its mechanical behaviour (Burland 1990; Hight and Jardine 1993; Hight et al. 2003). These discontinuities are in the form of fissures, joints (or backs), bedding surfaces (or laminations), tectonic shear surfaces, recently formed shear surfaces, and minor faults (Skempton et al. 1969; Hight et al. 2003). Figure 2 depicts the fissured and jointed pattern of the LC at shallow depth at New Hook farm and T5 sites. The occurrence of fissures was reported to be the dominant feature of the block samples retrieved from Unit C (7 m depth) at T5 site (Gasparre 2005).

The brown, brownish grey, or mottled brown and grey coloured LC, typically found in the top 5–10 m depths, has been recognised as weathered LC (colour changes from blue-grey to brown) and contains selenite crystals, a form of gypsum-hydrated calcium sulphate (Hight et al. 1993; Aldiss 2014). Figure 3 shows the presence of selenite crystals in retrieved samples from New Hook farm.

To identify the basic structural properties (soil fabric) and chemical composition of the retrieved samples, a microstructure and mineralogy analysis was carried out using a Philips XL30 FEG environmental scanning electron microscope (ESEM) equipped with an Energy Dispersive X-ray (EDX) analyser. For ESEM analysis, cubic subsamples of approximately 1 cm³ size were excavated from the central portions of the intact samples.

The micrographs of Figs. 4 and 5 reveal the large proportions of randomly oriented silt with irregular angular fine sand-sized grains and calcite crystals imbedded in the clay matrix.

The clay structure appears to be flocculated and composed of proportionally equal large flocs and single plate-shaped particles with edge-to-edge, edge-to-face and face-to-face contacts (Fig. 6). As can be seen, the edge-to-edge contacts of the clay domains have resulted in development of large pores in the clay cluster. This soil fabric is comparable with flocculated cardhouse fabric reported by Gasparre (2005) for T5 LC at shallow depths. The micrograph of Fig. 7 depicts a pyrite particle surrounded by clay domains and single particles.

Figure 8 shows the steps of drying a flooded specimen in the ESEM. As can be seen, the drying process does not noticeably affect the macrostructure and arrangement of the clay flocs. Also shown in the figure, is the removal of the water menisci developed at inter-floc pores (throats) during sequential drying of the specimen.

The diagenesis analysis carried out by Huggett (1994) on LC samples recovered from Warden Point on the Isle of Sheppey, revealed the similarity in compositions of Sheppey LC with Ockenden LC, with differences being only in the relative proportions of some diagenetic minerals. The principal authigenic minerals in Sheppey LC are carbonates, apatite and pyrite. The fibrous calcite, present in Sheppey concretions, comprises a maximum of 5% total Magnesium (Mg), Manganese (Mn), and Iron (Fe). About 50% of the concretions comprise predominantly detrital clay minerals of illite and smectite, with the presence of a slight amount of coarsely crystalline authigenic kaolinite. Silt grains are predominantly quartz and K-feldspar. A clay/quartz ratio of 3.2:1 was also reported. Similar results were obtained from the EDX analysis carried out in this research, as shown in Fig. 9. The key petrographical characteristics of the sample are summarised in Table 1. It is observed that Silicon dioxide (SiO₂), Aluminium oxide (Al₂O₃), Calcium carbonate (CaCO₃), and Iron disulphide (FeS₂) are predominant constituents of the samples. Calcium (Ca) and Iron are the products of weathering of silicate minerals. Calcium carbonate precipitates are formed from the chemical interaction of Calcium and Carbon Dioxide (CO₂) molecules dissolved in sea water. Similarly, Iron reacts with Sulphur (S), resulting from rotting organic matter, and forms pyrite (FeS₂) (Gasparre 2005).

The waste acceptance criteria (WAC) testing was carried out by CET Structures Ltd on representative samples taken from the reservoir site to determine if the soil (waste) has any hazardous properties as per the Environment Agency’s Hazardous Waste Technical Guidance WM2. The likely behaviour of waste in a landfill is assessed against WAC limits for granular waste acceptance at inert waste landfills, stable non-reactive hazardous landfills, and hazardous waste landfills. These limits are based on total concentration limits for the organic components and pH, as well as leachable components (Environment Agency 2013). The obtained results are summarised in Tables 2 and 3. For all of the evaluated parameters, the obtained results from chemical analysis are below the set limits, except the values of pH and total dissolved solids (TDS) that lay above the WAC limits for inert waste landfill.

The particle size distribution (PSD) curve for natural samples was obtained from wet sieving and hydrometer sedimentation tests according to BS1377:Part2:1990. As shown in Fig. 10 and Table 4, the percentage by weight of fine grains (clay and very fine silt) appears to be larger for Sheppey LC in comparison with shallow depth T5 LC. Using GRADISTAT statistical package (Blott and Pye 2001), further grain size parameters such as sand, silt, and clay contents, texture, mean, mode, median, skewness, sorting, D₁₀, D₂₅, D₅₀, D₇₅, and D₉₀ values were derived as shown in Table 4.

Laboratory determination of index parameters carried out according to BS1377:Part2:1990, confirmed a range of 19–24% and 70–78% respectively for plastic limit (w_p) and liquid limit (w_L) indices. The upper band values are closely matched with w_p = 28% and w_L = 77% reported by Jardine et al. (2003) for Sheppey LC. Overall, the plastic limit of Sheppey LC at shallow depths appears to be less than shallow depth T5 LC with w_p = 29% for Unit C, whereas, the plasticity index for Sheppey LC (I_p = 52) is higher than Unit C T5 LC (I_p = 37). Based on the unified soil classification system (USCS), Sheppey LC is classified as clay of high plasticity (CH). Based on BS1377:Part2:1990, the specific gravity (G_s) was measured as 2.67 which is lower than the average value of 2.74 reported for Unit C T5 LC (Gasparre 2005).

Figure 11a presents a comparison of the normalised compression curves for Sheppey LC and the natural LC from Unit C of T5 site. As the plasticity of the LC deposits increases from west to east of the London basin (Hight et al. 2007), the compression curve for Sheppey specimen was expected to lay below the compression curve for T5 specimen. It is observed that the T5 specimen is less compressible than the Sheppey specimen, mainly due to its lower plasticity index (I_p = 37) and initial water content (w₀ = 24%). Average compression index (C_c) and swelling index (C_s) values of respectively 0.22 and 0.10 were obtained from the compression tests on intact Sheppey specimens having an average initial void ratio of e₀ = 0.85. Similar values are derived from the compression curve of the intact T5 specimen.

It is worth mentioning here that soil structure can be disturbed during the process of sampling, transport, and storage, modifying initial properties of the soil sample. Among various soils sampling techniques, block sampling (employed in this study) is considered as a suitable method, which if performed adequately can produce ‘class 1′ quality samples of natural clays with minimum to negligible level of disturbance to the soil structure. Using this method, soil properties such as particle size, water content, density, density index, permeability, compressibility, and shear strength can remain unchanged (BS-EN-1997:22007). Furthermore, block samples allow for reliable determination of porosity. Accordingly, the laboratory void ratio values obtained from block samples are, with good accuracy, comparable to the field void ratio.

Figure 11b presents a comparison of the normalised compression curves for reconstituted Sheppey LC and the reconstituted T5 LC. The reconstituted specimens were obtained from consolidation of soil slurry in a Perspex consolidometer under a vertical stress of 80 kPa for a duration of 5 days. Average C^*_c and C^*_s values of respectively 0.38 and 0.13 were obtained from reconstituted specimens with e₀ = 0.93. The reconstituted T5 specimen, with average C^*_c and C^*_s values of respectively 0.39 and 0.15 (Gasparre, 2005), exhibits higher compressibility in comparison with the reconstituted Sheppey specimen.

As reported in Rezania et al. (2019), the maximum value of creep index (C_α) for Sheppey LC falls approximately in the range of 0.007–0.008 and 0.012–0.013 respectively for intact and reconstituted specimens, the latter being comparable with the average value of C_α = 0.016 reported by Sorensen (2006) for reconstituted T5 LC. The ratio α = C_α/C_c for Sheppey LC falls approximately in the range of 0.015–0.046 and 0.023–0.037 respectively for intact and reconstituted specimens. The maximum values of α occur at stress levels in a range of 1–2 times yield vertical net stress (σ_p), for both intact and reconstituted specimens.

A series of drained constant rate of strain (CRS) compression-relaxation tests was performed on reconstituted Sheppey LC using an innovative CRS oedometer cell (Bagheri et al. 2020) equipped with two high-capacity tensiometers (Bagheri et al. 2018) for monitoring pore-water pressure evolutions, in order to evaluate the rate-dependent and stress-relaxation responses. Specimens of different initial water contents (and hence different initial suction s₀) were loaded to a certain stress level (σ₀ = 3450 kPa), then allowed for stress relaxation at zero rate of axial displacement for a minimum time period of 210 h. Table 5 outlines the details of these tests and Fig. 12 presents the soil water retention curve (SWRC) of the reconstituted soil sample. The SWRC was obtained using HCT technique and following the dynamic method as described in Bagheri (2018). Figure 13 presents the 1D compression behaviour of the specimens loaded at two different strain rates of \(\mathop {\varepsilon_{v1} }\limits^{\cdot}\) = 4.8 \(\times\) 10^–7 s⁻¹and \(\mathop {\varepsilon_{v2} }\limits^{\cdot}\) = 2.4 \(\times\) 10^–6 s⁻¹, where \(\mathop {\varepsilon_{v2} }\limits^{\cdot}\) = 5 \(\mathop {\varepsilon_{v1} }\limits^{\cdot}\), in axial strain (ε_a) versus vertical net stress (σ_vnet) space. The values of the yield vertical net stress (σ_p) were found to decrease exponentially with increase of w₀ for both selected strain rates as shown in Fig. 14.

The values of the coefficient of stress relaxation (R_α), being the rate of change of σ_vnet with time (t) at zero strain rate in log–log space, fell within a range of 0.011–0.019 and 0.017–0.029, respectively for fast and slow strain rates (Bagheri et al. 2019), the latter corresponds well with the values of α ratio of 0.023–0.030 obtained from multi-staged loading oedometer tests on reconstituted Sheppey LC at different initial water contents reported in Bagheri (2018).

Some characteristics of LC from New Hook farm in the Isle of Sheppey were discussed and compared with shallow depth (Unit C) T5 LC. The following conclusions can be drawn;