TECHNICAL Paper Soil Nail Remediation of Seasonal Slope

TECHNICAL paper Soil nail remediation of seasonal slope movements in an old clay fill embankment dam Andrew Lees, Geofem, Gavin MacDonald, Mouchel, and Andrew Sheerman-Chase, Cares Group

Abstract trees can improve slope stability, it is A curious slope movement pattern 0500m increasingly recognised that they has been recorded during 10 years B5183 can cause summertime settlements of monitoring at one unstable SITE of clay-fill infrastructure section of the downstream slope of LOCATION embankments that do not fully Aldenham embankment dam, in recover in the winter (Andrei, 2000; Hertfordshire, UK. A slip plane in Hilfield Park B5378 Ridley et al, 2004; Russell et al, the upper slope was reactivated each Reservoir 2000). Many of the early summer, which was caused by embankment dams are tree covered shrinkage of the tree-covered toe of (Kennard, 1975; Hoskins and Rice, the embankment due to soil 1992) but, to date, no monitoring Aldenham moisture deficit, leading to loss of Reservoir data has been available to assess the support to and reactivation of the Elstree influence of trees on any settlement. grassed upper slope slip mass. This Soil nailing is suited to firm to led to a seasonal, ratcheting stiff clays, but particular care is A411 accumulation of crest settlement – a required in high-plasticity over- type of mechanism that presents a A41 consolidated clays such as London continual serviceability problem to M1 Clay to avoid a degradation of the infrastructure embankments as well nail system due to cycles of as dams. Figure 1: Location map shrinkage and swelling (Phear et al, A soil nailing scheme was 2005). designed and installed to support outward movement of the slope 2004; Take and Bolton, 2011; This paper presents the results of the upper slope slip mass, and then (Russell et al, 2000). Vaughan et al, 2004). pore pressure and displacement grassed over. Inclinometer and wave Furthermore, the net movements The higher water demand and monitoring from 2000 to 2011, wall settlement data in the seven develop on preferred surfaces where extensive root systems of trees together with wave wall levels for years following nail installation strain softening is occurring, starting increase shrinkage and swelling the period 1978 to 2008, for a showed the scheme to be successful, from the toe of the slope, and, after cycles (Biddle, 1998; Driscoll, 1983), particularly unstable and curious even resulting in some upslope sufficient number of pore water and some trees (eg oak) have a section of the downstream slope of recovery movements. pressure cycles, can lead to higher water demand than others Aldenham embankment dam, progressive failure (Nyambayo et al, (eg elder) (Biddle, 1998). While Hertfordshire, UK. The design and Introduction installation of a nailing scheme in In the late 18th and early 19th 2003 is also described. centuries, many embankment dams were built in the UK for landscaping Rainfall 238m 250m 260m Site description and history purposes and for canal reservoirs. 246m 253m 265m Aldenham embankment dam is 1200 At least about 200 of them are still 100 located within Aldenham Country in use today (Kennard, 1975). These 99.9 Park in Hertfordshire, UK (OS grid early dams were constructed in a 1000 reference TQ167958) and is similar fashion to the railway 99.8 orientated approximately east to embankments of southern England west with the reservoir on the south 800 in the mid-19th century onwards – Elevation (mOD) 99.7 side (Figure 1). It was first completed from thick uncompacted layers of 99.6 in 1795 to supply compensation dumped clay fill (Vaughan et al, 600 water after construction of the 2004) – and have suffered the same 99.5 Grand Union Canal (Faulkner, instability problems. 1972) and in 1802 its height and Vegetated clay soils in temperate 99.4 400 length were increased to today’s climates are subjected to seasonal 99.3 values of 8m and 400m respectively cycles of pore water pressures as high 200 and the reservoir capacity to suctions during peak water demand 99.2 78,000m3. by vegetation in relatively dry (mm) June-August total rainfall Heathrow The contours of the natural 99.1 0 summer months dissipate as water landscape suggest that the original demand lowers and rainfall increases course of the river passed under the

in the winter months. In clay slopes, Jan 78 Jan 80 Jan 82 Jan 84 Jan 86 Jan 88 Jan 90 Jan 92 Jan 94 Jan 96 Jan 98 Jan 00 Jan 02 Jan 04 Jan 06 Jan 08 embankment at about 150m from the resulting cycles of shrinkage and Date the eastern shoulder, and there is swelling of the ground surface leads also a kink in the embankment crest to a net ratcheting downward and Figure 2: Wave wall settlement at unstable section at this location which probably

ground engineering september 2013 23 TECHNICAL paper

Figure 3: In January 1997 a 36m long sheet pile wall was installed in the upper upstream slope

resulted from a misaligned particularly unstable in the past 50 year). dependent on summer rainfall meeting of both sides of the years. From about 1962, settlement These increases coincided with amounts. embankment over the original river of the wave wall at this point was unusually dry summers, as indicated In January 1997 a 36m long sheet course during construction. reported. Wave wall levelling was by the rainfall totals for the 23km pile wall was installed in the upper The embankment fill was sourced undertaken in this area on a regular distant Heathrow weather station, upstream slope around the unstable from the weathered London Clay of basis from 1978 onwards and its and settlements appeared to slow section to address the risk of loss of the basin and slips occurred during results are shown in Figure 2. Prior during comparatively wet summers freeboard here due to the wave wall construction, necessitating to 1988, average annual settlements (eg 1999, 2002), suggesting that the settlements, as shown in Figure 3. re-profiling to a shallower 1:4 slope, between chainage 246m and 260m wave wall settlements were a result Earlier, in 1990, 2m deep gravel- and have continued intermittently to (the chainage is measured from the of suction-induced shrinkage in the filled drainage trenches at 5m the present day, more usually in the west shoulder) were about 5mm, but clay fill. Interestingly, Smethurst et centres were installed in the upper upstream slope. there were marked increases in the al (2012) noted that summer downstream slope of the whole dam The downstream slope around rate of settlement in 1989-90 (to suctions in a London Clay cutting to mitigate general deformations the location of the kink has become 22mm/year) and 1995 (to 48mm/ slope in Newbury were heavily occurring in the crest footpath. As shown in Figure 2, while installation of the sheet pile wall Subsided wave wall Oak and Hawthorn trees successfully restored the freeboard, A1/1 & (all standing vertically) neither this nor the installation of 100 A1/2 the gravel drains in 1990 arrested the wave wall settlements. Only the 98 A2/1 A2/2 installation of soil nailing in the 96 Sheet London upper downstream slope of the pile 94 Clay fill Probable slip surface unstable section in October 2003

Elevation (mOD) wall 92 finally halted the settlements. ? ? An important factor governing 90 ? pore water suctions in the 88 Low masonry wall In situ (undistressed) embankment is the presence of trees 86 London Clay on the lower downstream slope. Historical aerial photographs dating Inclinometer tube Typical back to 1947 (Lees et al, 2013) show piezometer levels: Plate magnet Qualitative that trees were allowed to grow on Spider magnets summertime End of winter Slope height 7.7m all the downstream slope up to 1974 movement End of summer when, following recommendations vector Mean angle 1:5.1 Datum magnet under the Reservoir (Safety Piezometer tip 010m Provisions) Act 1930, the trees and shrubs were removed from the Figure 4: Cross-section of unstable downstream slope upper part of the slope (Kennard,

24 ground engineering september 2013 1975), probably due to concerns 2004). From direct shear testing, a with tree roots penetrating the dam residual shear strength of = 12° φ′ or erosion resulting from uprooting and c = 0kPa was determined for the Horizontal displacement (mm) ′ of trees in high winds (Hoskins and London Clay fill clods, with -10 0 10 20 30 40 50 60 Rice, 1992). somewhat higher values obtained for 0 Since 1974 to the present day, the clods containing foreign material. upper slope has been grassed while only the lower slope has been Monitoring instrumentation covered by trees, as shown in the and data Depth (m) 2 diagram of the unstable section in Vibrating wire piezometers Figure 4. This resulted in a connected to a data logger were differential between upper and lower installed in two boreholes at the slope soil suctions that likely played depths shown in Figure 4. In two 4 a significant role in the initiation of other boreholes, inclinometer tubes a curious movement pattern at the combined with extensometer spider unstable section. magnets to measure vertical 6 settlement were installed at the Ground investigation depths shown in Figure 4. In order to design a stabilisation Measurements from these were scheme at the unstable section (and taken at various intervals as short as 8 to address smaller deformations in two weeks at the start to as much as other sections of the embankment 12 months near the end, depending dam – refer to Lees et al (2013)), a on the availability of funds, using Datum reset to 14 April 2003 sub-surface investigation and manual probes. Occasional gross 10 28 May 2003 monitoring of the embankment dam errors in the data were corrected but 27 Jun 2003 was undertaken in February 2001. otherwise the data appeared 7 Jul 2003 Five boreholes were sunk in the sufficiently accurate (estimated 4 Aug 2003 upper downstream slope at the uncertainty ±2 mm), since it was 12 unstable section generally to possible to identify consistent trends 19 Aug 2003 between 1m and 4m below dam in all the measurements. 1 Sep 2003 foundation level. The London Clay 22 Sep 2003 fill encountered was generally a firm Deformation monitoring 14 brown/grey sandy gravelly clay with results decomposed roots and plant Inclinometer A1/1 profiles for remains, similar to the London Clay Summer 2003 with values reset to Figure 5: Inclinometer A1/1 – sample profiles from Summer 2003 fills encountered in old railway zero in April 2003 are shown in embankments that comprise clods Figure 5. In spite of this being a of intact clay surrounded by a particularly dry summer, it is clear matrix of remoulded clay and that, remarkably, slippage occurred Horizontal movement (northwards positive): mm foreign material (eg silt and sand) along a 5.5m deep plane. A 6m deep picked up during handling and slip plane was also recorded in -20 0 20 40 60 80 100 120 20 transportation (O’Brien et al, 2004). inclinometer A2/1 which, combined The underlying London Clay was a with the wave wall settlement, firm grey/brown gravelly clay allowed the slip plane shown in becoming stiff with depth. Figure 4 to be drawn. 0 Average liquid limit and plasticity The inclinometers recorded Winter 2001/02 index were 70% and 42% similar movements in the summers Winter 2002/03 respectively with clay fraction of 2001 and 2002. This would -20 recorded, using the sedimentation account for the continual settlement rtical movement: mm Ve method, at 50%, which places the of the wave wall and, since it is Post fill in the high plasticity category located well behind the slip, the nailing -40 recovery (BSI, 1999). ineffectiveness of the sheet pile wall Summer 2001 Gravimetric moisture content at halting the settlement. However, largely ranged between 30% to 40%, there is no clear evidence of the toe so at the lower end of the plasticity of the slip emerging further down -60 Summer 2002 range and about half the liquid limit the slope and the trees on the lower value which, according to Driscoll’s slope all stand vertically. (1983) crude estimate that The extensometer magnet -80 desiccation of clay commences elevation readings at 0.8m nominal when the moisture content falls depth have been combined with the Summer 2003 below about half the liquid limit horizontal deflection reading in the -100 value, suggests that at the time of the corresponding inclinometer A2/1 ground investigation – after a very for the whole monitored period in wet winter – the clay fill of the upper Figure 6 to produce a vector plot of downstream slope was at field ground deformations at this point in -120 capacity. the plane perpendicular to the dam. Nailing works From routine isotropically The overall approximate seasonal consolidated undrained triaxial tests movements are indicated by the -140 on nine samples of the London Clay arrows. fill clods, mean peak shear strength There were cycles of downslope values of = 23° and c = 5kPa movement in each summer of 2001 φʹ ʹ -160 were determined, which are similar to 2003, with by far the largest to other values determined for slippage occurring during the Figure 6: Measured deformations at 0.8 m depth in borehole A2/1 London Clay fill (eg O’Brien et al, particularly dry summer of 2003, (April 2001 to March 2011)

ground engineering september 2013 25 TECHNICAL paper

interspersed with partial recovery was required in order to arrest Piezometer A1/2 Piezometer A2/2 Inclinometer A2/1 1m depth in approximately November to annual downslope movements of at February each year. The A2/1 70 0 least 20-40 mm. This would negate instrument suffered some the risk of eventual uncontrolled disturbance, but no damage, during release of water from the reservoir if 60 the nailing works and continued to 10 the movements continued and the provide data up to the last reading in slip grew beyond the extent of the March 2011. The plot shows that 50 sheet pile wall, as well as saving not only were downslope 20 continual maintenance to restore movements halted by the nailing but the crest footpath. However, the the low pre-stress imposed on the 40 embankment dam forms an installed nails resulted in partial (kPa) water pressure Pore 30 important part of the popular

upslope recovery movements. 30 Horizontal displacement (mm) Aldenham Country Park and it was important that the stabilisation Pore water pressure 40 scheme preserved the park monitoring results 20 environment. This ruled out Two piezometers were installed removal of the trees (which also had near to where potential slip planes 10 50 preservation orders on them) and were anticipated, as shown in Figure heavy structures, such as an 4. At such depths, they were unlikely embedded wall in the downstream

to record significant variations in 1 Mar 01 1 Mar 02 1 Mar 03 1 Mar 04 1 Mar 05 1 Mar 06 1 Mar 07 1 Mar 08 1 Mar 09 slope, that may have been ineffective pore water pressure due to 30 Aug 01 30 Aug 02 30 Aug 03 30 Aug 04 30 Aug 05 30 Aug 06 30 Aug 07 30 Aug 08 in any case. vegetation-induced suctions near Date The preferred option was soil the surface. Nevertheless, since Figure 7: Piezometer data nailing because the nails could be records show that the reservoir level spatially distributed to support the varied by no more than 30mm above thereby removing support for the either initiated progressive entire slip mass to prevent crest or 15mm below the overflow weir slope. Andrei (2000) also noted formation of a slip plane or caused settlement and because the nails level of 98.18 mOD, it can be significant accumulations of track reactivation of one of the many could be buried and grassed over to assumed that variation in pore settlements during the summer reported slip failures that occurred restore the park environment. pressure would have been caused by months on a clay fill railway during construction when the slope There were concerns regarding climatic changes. embankment wherever trees were was steeper. deterioration of the nailing scheme The measurements from both located on its slopes. The clay fill at the unstable over time due to the repeated piezometers are shown in Figure 7. It is likely also at Aldenham dam section – on the probable former shrinkage and swelling of the clay A2/2 took longer to settle down, but that clay shrinkage caused by trees river course – may have been placed fill, as described subsequently by both eventually recorded a small but on the lower slope removed support in difficult, wet conditions, leading Phear et al (2005). To address these regular approximate 5kPa seasonal to the treeless upper slope, resulting to more slip failures during or soon concerns, the following measures variation reaching minima in late in ratcheting downslope movements. after construction, prior to were taken: summer and maxima in spring. The But why did a deep-seated slip re-profiling. Positive pore water n nail centres were quite small at lowest pore water pressure was failure occur at this unstable section pressure along much of the slip 1.5m on a triangular grid; recorded in the dry summer of and not in other parts of the dam in plane even during the summer n quite long nails (12m) were used 2003. A2/2 was damaged beyond recent years? Higher magnitude months, due to seepage from the to extend well beyond the active repair during the nailing works but pore water pressure cycles were reservoir, also contributed to the zone in the fill and into the intact A1/2 continued to function and likely here due to the large number movement mechanism. London Clay a 0.35m thick granular recorded about a 25kPa excess pore of high water demand English oak layer stabilised by a single biaxial pressure due to the nail pre-stress trees located on the lower slope at Soil nailing scheme geogrid (Tensar SS40) at the base that then dissipated over the the unstable section. The resulting Clearly, stabilisation of the unstable below the nail plates was placed following two years. loss of toe support each summer section of the downstream slope over the nail heads to help distribute Also plotted on Figure 7 is a temporal plot of horizontal displacement recorded by inclinometer A2/1 at 1m depth. It confirms the paradox that significant downslope movements in the summer (prior to the nailing works 100 in September 2003) coincided with reduced pore water pressure. 98 London Clay fill Discussion of monitoring data 96 Sheet

The phenomenon of significant Elevation (mOD) pile 94 downslope movement during wall Probable periods of high soil suction, with the 92 slip surface greatest movements being observed ? in the driest summers, is curious but ? ? 90 40˚ has been observed before. Smethurst 40˚40˚ and Powrie (2007) recorded 40˚ 88 40˚ increasing bending moments in 45˚ discrete piles used to stabilise an 50˚ unvegetated Weald clay fill railway 86 55˚ In situ 60˚ embankment during the dry London Clay to horizontal summer months of 2001 and 2003. 010m The authors suggested that suctions caused by mature trees at the toe of the slope caused the clay to crack, Figure 8: Nailing scheme

26 ground engineering september 2013 the nail loads evenly over the surface surcharge of equivalent self-weight, failure loads (two at 130kN, one at to and reactivation of the grassed of the slope; thereby removing the lateral support 160kN). Thirteen (or 5.3%) of the upper slope slip mass. It could not n effective stress, residual shear to the upper slope but maintaining 243 working nails were test loaded be determined whether the slip strength parameters ( = 12°, c = the weight at the toe. This was to 150% working load and all passed plane developed progressively from ϕ′ ′ 0 for the upper and lower fill; = equivalent to forming a 3m deep with no significant deflection. All the pore suction cycles or whether it ϕ′ 14°, c = 0 for the remainder tension crack immediately upslope the working nails were tensioned to was pre-existing from the reported ′ containing a higher proportion of of the trees and its success at back- about 10kN when locked off. post-construction failures. foreign material) were adopted for analysing the slip failure lends As shown in the displacement The toe of the embankment is the clay fill in the slip circle analyses weight to the argument that measurements at borehole A1/1 in stable due to reinforcing effect of and bond strength design, and shrinkage of the tree-covered toe Figure 6 as well as the wave wall the tree roots and probable persistent constant volume strength ( = 22°, each summer was the trigger settlement data in Figure 2, the soil soil moisture deficit, so the toe of ϕ′ c = 0) for the intact London Clay; mechanism for the slope nailing scheme was successful in the slip plane did not reach the ′ n monitoring of slope movements movements. halting the accumulation of ground surface, resulting in a would continue for at least five years The nailing arrangement was downslope movements. This was bulging of the mid-slope profile after completion of works, and wave designed in Talren using the same also borne out by deflection evident from topographical surveys. wall level monitoring in perpetuity. method of modelling the toe measurements at other depths and An unobtrusive slope stabilisation The nailing scheme was designed shrinkage, with a target overall in another borehole (refer to Lees et scheme was required on this country using Talren 97 version 2.2 (Bishop factor of safety of 1.5. The shear al (2013)). park site, so a soil nailing scheme method slip circle analysis) and in resistance of the nails was ignored. Indeed, the pre-stress in the was designed and installed to accordance with Clouterre (1991). The nailing arrangement shown in installed nails resulted in an upslope support the upper slope slip mass, Hydrostatic conditions were Figure 8 was found to be the most recovery (Figure 6) of 35mm in the and then grassed over. Inclinometer assumed below a groundwater level efficient (factor of safety 1.57, first year and a further 20mm in the and wave wall settlement data in the that was based on the reservoir level, maximum nail load 65kN), although following six years. No net annual seven years following nail the average piezometer readings and the upper nails were constrained by downslope movement was recorded installation have shown the scheme assuming groundwater level at the sheet pile wall and could not be by any of the monitoring to be successful, even resulting in ground surface downstream of the installed at their most efficient instruments in the 7.5 year some upslope recovery movements. dam. Zero pore pressure was angles. monitoring period following nail assumed above groundwater level. The nails comprised 32mm installation. Acknowledgements When analysed as a whole, the diameter hollow manganese steel The site investigation, monitoring downstream slope had adequate bars, installed with sacrificial drill Conclusions and soil nailing scheme were funded stability, even with residual shear bits on their ends and with a water- A curious slope movement pattern by Hertfordshire Country Council, strength. However, when shrinkage cement flush fed through the hollow has been recorded at one unstable the leaseholders of Aldenham of the toe was included, a slip circle bar (to cause less swelling and section of the downstream slope of Embankment Dam. The authors are with 0.98 factor of safety whose softening of the borehole walls an old clay fill embankment dam. A grateful to Norman Fernades and geometry matched the profile compared with water flush). slip plane in the upper slope was Zia Qadeer of Mouchel for determined from the inclinometers Once each nail had been installed, reactivated each summer and the organising and undertaking the was obtained only by constraining a Portland cement grout was drier the summer, the greater the recording of instrument readings on slip circles to pass at depth to avoid injected down the hollow bar until downslope displacement. This was the site, and to the staff of shallow failures. all drill flush and debris had exited caused by shrinkage of the tree- Cementation Skanska for their Toe shrinkage was modelled by the borehole. Three test nails were covered toe of the embankment due advice and cooperation during replacing the tree-covered part of installed on site with 4m long bond to soil moisture deficit in the planning and execution of the the slope for up to 3m depth by a lengths and achieved satisfactory summer, leading to loss of support nailing works.

References Andrei, A (2000) “Embankment Junction Canal. David & Charles, Conference. Thomas Telford, Smethurst, J A; and Powrie, W stabilisation works between Newton Abbot. London, 898-910. (2007) “Monitoring and analysis of Rayners Lane and South Harrow Hoskins, C G and Rice, P R (1992) O’Brien, A S; Ellis, E A; Russell, D the bending behaviour of discrete Underground stations”. Cooling Vegetation and Embankment (2004) “Old railway embankment piles used to stabilise a railway Prize paper. Ground Engineering, Dams. Water Resources and clay fill – laboratory experiments, embankment”, Géotechnique, 57, 33(1), January 2000. Reservoir Engineering. Thomas numerical modelling and No.8, 663-677. Biddle, P G (1998) Tree Root Telford, London. field behaviour”, Advances in Smethurst, J A; Clarke, D; and Damage to Buildings, Vol. 1 Kennard, M F (1975) “18th Century Geotechnical Engineering. The Powrie, W (2012) “Factors Causes, Diagnosis and Remedy. Dams in England”. Proceedings Skempton Conference. Thomas controlling the seasonal variation in Willowmead, Wantage. of first British Dam Society Telford, London, 911-921. soil water content and pore water British Standards Institution Conference, Newcastle. Phear, A; Dew, C; Ozsoy, B; pressures within a lightly vegetated (1999) Code of Practice for Site Wharmby, N J; Judge, J; and clay slope”, Géotechnique, 62, Lees, A S; MacDonald, G No.5, 429-446. Investigations. BS 5930:1999. J; Sheerman-Chase, A; and Barley, A D (2005) Soil nailing – BSI, Milton Keynes. Schmidt, F (2013) “Seasonal Best Practice Guide, CIRIA C637, Take, W A; and Bolton, M D (2011) Clouterre (1991) French National slope movements in an old clay London. “Seasonal ratcheting and softening Research Project Clouterre – fill embankment dam”,Canadian Ridley, A; McGinnity, B; and in clay slopes, leading to first-time Recommendations Clouterre Geotechnical Journal, 50, No.5, Vaughan, P (2004) “Role of pore failure”, Géotechnique, 61, No.9, (English translation). Federal 503-520. water pressures in embankment 757 – 769. Highway Administration FHWA- Nyambayo, V P; Potts, D M; stability”. Proc. ICE Geotechnical Vaughan, P R; Kovacevic, N; and SA-93-026, Washington, USA. and Addenbrooke, T I (2004) Engineering, 157, No.4, 193-198. Potts, D M (2004) “Then and Driscoll, R (1983) “The influence “The Influence of permeability Russell, D; Ellis, E; O’Brien, now: some comments on the of vegetation on the swelling and on the stability of embankments A; and McGinnity, B (2000) design and analysis of slopes shrinking of clay soils in Britain”. experiencing seasonal cyclic “Role of vegetation in the and embankments”. Advances in Geotechnique, 33, No.2, 93-105. pore water pressure changes”. stability and serviceability of Geotechnical Engineering. The Skempton Conference. Thomas Faulkner, A H (1972) The Grand Advances in Geotechnical railway embankments”. Railway Engineering. The Skempton Engineering 2000. Telford, London, 241-290.

ground engineering september 2013 27