Modelling Volume Change Potential in the London Clay

The London Clay Formation is particularly susceptible to shrink–swell behaviour that has resulted in a long history of foundation damage due to ground movement across the outcrop. Damage has cost up to £500 million in a single year. Underlying most of the Greater London area, the London Clay Formation is of major engineering importance as it is on and within this formation that the majority of the city’s infrastructure, buildings and underground services are constructed. The Volume Change Potential of a soil is the relative change in volume to be expected with changes in soil moisture content and the subsequent shrinkage or swelling can cause major damage to structures above or below ground. Detailed statistical and spatial analyses of data across the London Clay outcrop has revealed a significant geographical trend in the volume change potential of this deposit, confirming an overall increase from west to east, but also showing subtle trends with depth. This paper describes how this analysis was carried out and shows how such assessments can yield valuable information about shrink–swell behaviour not only of the London Clay but of similar shrink– swell prone argillaceous formations elsewhere. Introduction The effects of shrinkage and swelling of clay soils, with respect to foundation and building damage, were first recognised by geotechnical specialists following the dry summer of 1947. After the drought of 1975–76 insurance claims in the UK came to over £50 million. In 1991, after a preceding drought, claims peaked at over £500 million. The Association of British Insurers has estimated that the average cost of shrink–swell related subsidence to the insurance industry stands at over £400 million a year. If the UK were to experience an increase in extended periods of dry weather, prior to future rainfall events, costs could rise significantly. London Clay – Geological Outline The London Clay Formation is of major importance in the fields of geotechnical engineering and engineering geology. This is because it has hosted a large proportion of sub-surface engineering works in London over the last 150 years. It has also been the subject of internationally recognised research in soil mechanics over the last 50 years (Skempton & Delory, 1957; Chandler & Apted, 1988 and Takahashi et al., 2005). The London Clay is subject to shrinkage and swelling behaviour, which has resulted in a long history of foundation damage within the outcrop. Although it has been the most commonly used soils testing material (assumed to have a uniform consistency throughout) since the early 1960s, few publications have covered the Volume Change Potential of the formation, except as a subsidiary topic in Burnett & Fookes (1974) and Forster (1997) and, more recently, the geotechnical studies of Hight et al. (2002) and Pantelidou & Simpson (2007). The term ‘London Clay’, used in this paper, refers to the dominantly clay lithologies within the London Clay Formation. C:\Documents and Settings\adixon\Local Settings\Temporary Internet Files\Content.Outlook\7NFVY2G5\08-112 revised2.doc The London Clay Formation outcrop, and subcrop, in southern England is confined to the London and Hampshire Basins. It has an extensive outcrop in the London Basin, due to the erosion of much of the overlying deposits. In the Hampshire Basin the outcrop is a narrow strip, less than 3km in width. The total, onshore, sub-crop area of the London Clay Formation is approximately 6,400km and the total outcrop is approximately 2,300km, about 1% of the area of Great Britain (based on the British Geological Survey 1:250k digital geology map – DigMap 250k. Typical unweathered London Clay is a firm to very stiff, fissured, dark grey or blue-grey silty clay deposited in a marine shelf environment in a tropical climate during the Eocene (Murray & Wright, 1974 and Sellwood & Sladen, 1981). The western part of the sequence is interpreted to have been deposited near to land and the eastern part in the deeper water of the North Sea Basin (Curry, 1992). When weathered, the London Clay is a soft to firm brown clay with ochreous staining due to oxidation of iron compounds. The clay itself may be pale grey or grey-green due to leaching (Smith, 1989). In the east of the London Basin Clay dominates the sequence with inter-bedded fine sand and silt; in the west it generally contains more sand and silt. In the Hampshire Basin the London Clay is generally more sandy and silty throughout. Thickness In the London Basin the total London Clay Formation thickness is between 90m in the west thickening to 150m in the east (Sellwood & Sladen, 1981). In the Hampshire Basin the London Clay thickens from Alum Bay in the west (75m) to Whitecliff Bay in the east (142m). Much of the outcrop is covered by Quaternary deposits. Presently, the succession considered to be definitive is that described by King (1981), though the upper boundary is still disputed (Daley, 1999). Lithology The London Clay Formation varies in lithology, being dominantly argillaceous but containing layers with significant proportions of fine sand at the top and bottom of the sequence, and to the west. The basement beds of the London Clay Formation are characterised by glauconitic sandy silts whilst the uppermost beds comprise a flint pebble bed and sandy clayey silts with bands of sideritic ironstone nodules. Gypsum and selenite are found as individual crystals, or small clusters of crystals, formed by the oxidation of pyrite during weathering reacting with calcium carbonate released from shelly material (Lake et. al., 1986). Concretions of various kinds are found in the London Clay Formation; they include septarian nodules that form large spheroids (typically 30 to 50cm in diameter) of cemented mudstone, containing ferroan calcite. Based on major revisions of the London Clay Formation carried out by King (1981), and later by Ellison et al. (1994 & 2004), four informal lithostratigraphical units (Units A to D) and the formally named Claygate Member have now been recognised (Figure 1). Mineralogy The dominant clay mineral in the London Clay is illite. Subsidiary clay minerals are mixed layer clays (mainly smectite) and kaolinite (particularly in the west Hampshire basin and in Dorset). Discrete smectite horizons are common at the base of the London Clay Formation, in the central and eastern parts of the London Basin. Like similar deposits in the underlying Thanet Sand Formation these probably represent altered volcanic ash (Knox, 1979). Acidic groundwater results from the weathering of pyrite, which is widely dispersed throughout the London Clay Formation. Corrosion of iron tunnel linings has been attributed to pyrite oxidation within sand lenses at the base of the London Clay Formation (Rainey & C:\Documents and Settings\adixon\Local Settings\Temporary Internet Files\Content.Outlook\7NFVY2G5\08-112 revised2.doc Rosenbaum, 1989) and acid groundwater attack has been reported at Old Street Station, London by Mather (1985) and is associated with rising groundwater levels (Chisholm, 1984). Crystals of gypsum and selenite are found throughout both fresh and weathered London Clay. Zeolites are found in association with smectite bands and traces of chlorite are also found within the London Clay Formation. In the London area, the London Clay Formation is associated with rising groundwater levels as a consequence of diminished abstraction from the underlying Chalk aquifer. This has implications for both increasing the flooding hazard, in water sources and in to basement areas, and problems for engineering development including swelling associated with existing foundations (Lerner & Barrett, 1996 and Simpson et al., 1989). Figure 1 – Lithostratigraphical sequence of the London Clay Formation for central London and Essex (Ellison et al., 2004) Shrinkage & Swelling All clays are subject to shrinkage and swelling, but those with a higher proportion of expansive clay minerals, such as smectite, are more prone. The clay within the London Clay Formation is dominantly illite/smectite and is particularly susceptible to shrink–swell behaviour; it is generally accepted to be highly plastic (Driscoll, 1983; Cripps & Taylor, 1986 and Reeves et. al., 2006), gradually increasing in plasticity from west to east in the London Basin by approximately 30% (Burnett & Fookes, 1974). The depth to which shrinkage and swelling occurs is usually confined to the active zone (upper 1.5m) where moisture change is most likely to occur, unless this zone is extended by the presence of tree roots (Driscoll, 1983). In a study of subsidence claims, related to shrink–swell clays, the London Clay is described as “the most commonly encountered problem soil” (Crilly, 2001). Volume Change Potential The Volume Change Potential (VCP) of a soil is the relative change in volume to be expected with changes in soil moisture content and is reflected by shrinking and swelling of the ground. That is, the extent to which the soil shrinks as it dries out, or swells when it gets wet. However, despite the various available test methods for determining these two phenomena (e.g. BS 1377, 1990: Part 2, tests 6.3 & 6.4, Shrinkage Limit and 6.5, Linear Shrinkage and Part 5, test 4, Swelling Pressure) (British Standards Institution, 1990), they are rarely employed in the course of routine site investigations in the UK. Hence, few data are available for data-basing the directly measured shrink–swell properties of the major clay formations. Consequently, reliance is placed on estimates based on index parameters, namely, liquid limit, plastic limit, plasticity index, and density (Reeve et al., 1980; Holtz & Kovacs, 1981; Oloo et al., 1987). No consideration has been given to the saturation state of the soil and therefore to the effective stress or pore pressures within it. The most widely used parameter for determining the shrinkage and swelling potential of a soil is the Plasticity Index (IP). Such plasticity parameters, being based on remoulded specimens, cannot precisely predict the shrink–swell behaviour of an in-situ soil. However, they do follow properly laid down procedures, being performed under reproducible conditions to internationally recognised standards (Jones, 1999). A ‘Modified Plasticity Index’ (IP’) is proposed in the Building Research Establishment Digest 240 (1993) for use C:\Documents and Settings\adixon\Local Settings\Temporary Internet Files\Content.Outlook\7NFVY2G5\08-112 revised2.doc where the particle size data, specifically the fraction passing a 425μm sieve, is known or can be assumed as 100% passing (BRE, 1993). The Modified IP’ takes into account the whole sample and not just the fines fraction, it therefore gives a better indication of the ‘real’ plasticity value of an engineering soil: 100 425 % ' m Ip Ip μ < × = Where: Ip’ is ‘Modified’ Plasticity Index Ip is ‘Standard’ Plasticity Index Data source A meaningful assessment of the shrink–swell potential of the London Clay across its entire outcrop requires a considerable amount of high-quality and well-distributed spatial data, of a consistent standard throughout. This was achieved by utilising the large body of index test data stored in the British Geological Survey’s National Geotechnical Properties Database. At the time of this study, the national database contained data from more then 61,000 boreholes, comprising nearly 280,000 geotechnical samples, with over 84,000 containing relevant plasticity (Ip) data. Of these, 11,879 were determined for London Clay. Figure 2 shows the distribution of these data across England and Wales. Figure 2 – Plasticity Index data distribution across England and Wales contained in the BGS National Geotechnical Properties Database Statistical analyses The data were validated by removing negative values, duplicate values, data-entry error values and those with no ground level, or inaccurate co-ordinate values before statisticallybased evaluations were made. This reduced the number of ‘acceptable’ plasticity values to 11,366. To quantify the Volume Change Potential of the London Clay a preliminary statistical evaluation of the IP’ values was carried out to determine the overall range of the data values with respect to their locations across the outcrop. This was carried out by plotting IP’ values against their Easting and Northing positions (Figures 3a and 3b) to determine if a west-east trend of increasing plasticity, as reported by Burnett & Fookes (1974), or any northsouth trend was evident. By comparing these plots with a distribution plot of the plasticity data over the London Clay outcrop (Figure 4) it was possible to assess the overall variations of IP’ within the London Clay, and, in addition, to recognise five distinct areas within the outcrop as reflected by the number and value of the data points. Of the five recognised areas, Area 1 corresponds to the Hampshire Basin (Hampshire, Dorset, West Sussex and the Isle of Wight), Area 2 (west of London) covers most of Berkshire and Surrey, Area 3 covers Greater London, Hertfordshire, south Essex and Kent and Area 4 covers the rest of Essex and Suffolk, an area largely overlain by Glacial Till deposits. Area 3 was split into 3a (central London) and 3b (east of London) to deal with the data in this area more effectively. C:\Documents and Settings\adixon\Local Settings\Temporary Internet Files\Content.Outlook\7NFVY2G5\08-112 revised2.doc Figures 3a and 3b – Spread of data samples in West-East and North-South directions Figure 4 – Distribution of plasticity data (Ip') across London Clay outcrop Summary statistics for all 11,366 data samples across the London Clay outcrop and for the data within each of the five sub-areas are presented in Table 1. These include a count of the number of IP’ data points in each area, the minimum, maximum, mean, mode and median values and a series of inter-quartile values, including the upper quartile, the lower quartile values. These statistics are illustrated in the form of extended box and whisker plots (Figures 5 and 6) and histograms (Figure 7). Extended box plots are constructed from the 0.5, 2.5, 10, 25, 50, 75, 90, 97.5 and 99.5 percentiles of the data sets. The selected percentiles have been chosen as a compromise between practical geotechnics and statistical rigour (Hallam, 1990). Table 1 – Statistical analysis of IP’ ALL AREA1 AREA2 AREA3a AREA3b AREA4 count 11366 825 1857 746