Unsupported browser

For a better experience, please update your browser to its latest version.

Your browser appears to have cookies disabled. For the best experience of this website, please enable cookies in your browser

We'll assume we have your consent to use cookies, for example so you won't need to log in each time you visit our site.
Learn more

Technical paper: Railway earthworks with particular emphasis on the behaviour of clay embankments in the South east.

G P Birch and E R Evans, Network Rail

Arup Tech Paper logo (NB)

1.0 Abstract

Commencing with a geological setting and historical background, this paper explores the causal factors for earthwork failures – earthwork forensics – and the effects of extreme weather and vegetation management on earthwork performance.

figure 1

Figure 1 - Track geometry defects over an underbridge at Coalpit Heath, near Bristol

The paper then focuses on the relationship between soil moisture deficit (SMD) and behaviour of clay cored embankments in the south east, introducing the system called Earthworks Watch which provides both earthworks and track teams with an enhanced understanding of track geometry issues and aiding pro-active asset management (figure 1).

The paper concludes with predictions on earthwork deformation due to changes in SMD caused by climate change.


2.0 Introduction

Earthworks comprise the cuttings and embankments required to convey the railway on a level path, or at acceptable gradients, through undulating terrain. Earthworks make up around 60% of the British rail network, the remainder being at grade, on structures or in tunnels. Nationally, the balance between embankments and cuttings tends more towards 60/40, respectively.

The differentiation between cuttings and embankments is fundamental in appreciating the geotechnical properties and, therefore, engineering behaviour of the earthworks (figure 2). For example, the behaviour of material exposed in a cutting face reflects the period since placement, which can be many millions of years. However, the behaviour of embanked materials reflects the period since construction – around 170 years.

figure 2

In the case of embankments, the formation on which the railway and its associated ballast are set is assumed to be a stable and solid foundation. However, in many parts of the UK, more especially in the south east, this formation is constantly shifting in response to the tendency to creep towards a more natural angle of repose.

3.0 Geological setting

The topographic landscape (physiography) of the UK represents the visual expression of the underlying geology. Mountainous regions are associated with older, harder rock formations while valleys and low-lying areas are typically a result of erosion of weaker, younger rocks.

In contrast to the good engineering properties of the stronger geological formations, those made in part or entirely of clay, or mudstones which weather down to clays, do not behave well when excavated for cuttings and, even worse, heaped up to form embankments.

It should be recognised that, while one imagines that the formation of the earth is complete, the fact is that the various processes shaping the landscape are on-going, which is why we continue to experience soil creep, landslips, rock falls and erosion.

The geological map of the UK reveals the predominance of the older, harder, rocks in the north and west of England, Scotland and Wales, where the rock formations are aged around 350 to 500M years in contrast to the softer rock sequences of the south east with ages of around only 100M years.

The south east in particular, suffers from the presence of expansive clays which have the propensity to swell as their overburden is removed by erosion or by excavation, a condition known as over-consolidation. These troublesome clays occur to the south and east of a line joining the Humber and Severn Estuaries (figure 3).

figure 3

4.0 Historical background

The 1830s saw a proliferation of parliamentary submissions for new railway schemes (peaking in 1841), many of which were poorly engineered and stood little chance of gaining Royal Assent. Competition was fierce and a large body of experts was established by the government to filter out those schemes which had little potential. Competing promoters were keen to find fault with each others schemes in order for them to be thrown out (Biddle, 1990).

Railway promoters also faced hostilities on the ground from both the wealthy estate owners, who did not want lines passing through their land, and the canal operators who could foresee the adverse impact of the new lines on their trade. The former led to some peculiar deviations to skirt around estate boundaries, such as at Hatfield House, and the latter led to the survey teams having to hire protection, in the form of prize fighters, and in some cases having to carry out surveys at night to avoid the angry mobs sent by the canal companies.

As a consequence, the early railways were built to less than desirable standards with tight curves and, more significantly, minimal land-take. This has left a legacy of over-steepened cuttings and embankments, which are gradually attempting to revert to a more natural angle of repose, undermining, or encroaching onto, neighbours’ property.

Historical research into railway earthworks has highlighted the limitations of understanding of geotechnical behaviour at time of construction such that cuttings slopes were often set at the maximum angle possible without regard for the material variability or long-term stability. Sub-aerial weathering over the ensuing 170 years has led to differential erosion leaving uneven cutting faces with unstable crests, often regressing beyond the railway boundary (Skempton, 1996).

Likewise, embankments were made by end-tipping of materials derived from the nearest cutting without regard for suitability or longer term material behaviour (Vaughan et al, 2004). The fact that many embankments failed during construction is evidenced by the widening of land-take necessitated at the time of construction.

The importance of adequate drainage was recognised at the time of construction and the formation/ballast interface was designed with a camber or cross-fall from the centre to the sides so that water passing through the ballast was directed into cutting cess drains or shed off down the embankment sides (Gardner, 1921). However, the loss of this cross-fall on clay-cored embankments has led to the development of water pockets, which can lead to surficial slips or total collapse of the embankment. Compensation for loss of formation used to be achieved by the addition of, by then, widely available boiler ash, hence the common reference to many clay embankments as “ash banks”.

5.0 Earthwork “forensics”

figure 4

Figure 4 - Common signs of a slip developing in a clay-cored embankment.

When attempting to understand the causes of failures (figure 4) it is helpful to consider the various factors and processes under two groups; preparatory factors and triggering factors.

Preparatory factors are existing characteristics or long term processes which reduce the condition of the earthwork to point of quasi-stability. Triggering factors are events of short duration, which initiate failure, typically extreme weather events.

Preparatory factors for embankment failures include construction on soft basal layers, such as peat and alluvium; over-steepened construction angle rather than a design; unsuitable embankment fill, such as over-consolidated clays; saturation from water trapped beneath the track ballast; repeated tamping; compression and failure of culverts built at an angle to the embankment; flooding of the embankment toe; damage from burrowing animals and water accumulation; 170 years of weathering; and seasonal creep.

Triggering factors for embankment failure can include high transient water pressures from extreme weather; rapid drawdown following flooding; scour from floodwater; and loss of support from retaining wall or culvert collapse.

The observation of recent and historical failures can provide key insight into the actual failure mechanisms at each site building up a catalogue of scenarios demonstrating some degree of commonality. Typically, embankments fail where the preparatory factors exist or at transitions between cutting and embankment; on-off under-structure transitions (figure 5); embankment low points; and locations of previous settlement or landslips.

figure 5Where failures have been driven by water passing along the embankment the shape of the back-scar, or failure surface, in plan is very often hinged, with the maximum displacement at the upstream end.

An understanding of the history of the original construction of the railway line forms an essential part of the “forensic research” as the ravages of time and absence of proper maintenance can be a significant contribution to preparatory factors.

For example, the railway engineers sought to optimise the balance between construction costs and operational costs. Approaches to higher ground, such as escarpments, were aligned along natural depressions in the landscape, such as existing or former river valleys, in order to reduce the length of more-costly tunnelling methods. In doing so the builders recognised the need to make provision for the collection and safe conveyance of water entering the works, not just from within the limits of the purchased land, but from the entire catchment through which the railway was aligned.

It is the failure to maintain such water catchment systems through loss of records and obliteration by unfettered vegetation growth which leads to inundation of the clogged system during storm conditions. The consequences of drainage failures within tunnel approach cuttings (figure 6) can be exacerbated by the difficulty in accessing train derailments especially should they end up in the tunnel with approaching traffic leading to the most severe consequences.

figure 6

6.0 Effects of weather

There are many ways in which the different types of weather can impact on railway operations. However, what is important is the deviation from normal expectations. All infrastructure operators are geared towards coping with normal conditions and, to some extent, with abnormal conditions, but few can continue unaffected by extreme conditions.

Extreme events are often expressed in terms of return period for a given locality or area rather than actual magnitude. In this way, a storm event can be described in context, since a bout of very heavy rainfall may be classed as extreme in the south east but quite normal in north Wales.

Of greatest significance to earthwork stability is rainfall as water is the primary agent in earthwork destruction, not just through physical mobilisation by erosion or scour, but through the “lubrication” of soil particles subjected to excessive pore water pressures.

In this respect, the winter of 2000/01 provided fresh insight into the behaviour of earthworks, both embankments and cuttings, in response to extremes of rainfall (Birch & Dewar, 2002).

The south east experienced some 160 earthwork failures, commencing with the weak clay-cored embankments, many of which had been moving towards a condition of quazi-stability and were pushed into failure by the 280% excess of rainfall during the months leading up to Christmas. A second peak of excess rainfall followed in the early spring and before the ground condition had recovered from the pre-Christmas rainfall.

What is interesting is that, while the pre-Christmas failures were of embankments within the clay areas, the post Christmas failures were of cuttings in all material types. This is attributed to the high state of saturation of the ground, such that further rainfall provided the trigger to mobilisation of saturated materials in the cutting faces. The period was effectively one of destructive testing and provided the opportunity to witness some of the fundamental / pervading causal factors in earthwork failure.

One of these factors is illustrated by the embankment slip at Dutch House, near Mottingham in south east London (figure 7) where the failure backscar revealed a V-shaped trough of water-filled clean ballast running beneath the cess rail. This trough is attributed to the regular ballast addition and machine tamping, necessitated to rectify the deterioration in track geometry, brought about by traffic loading and seasonal creep of the embankment. The water trapped high up in the embankment ran for some hours after the failure as the water drained away right back to the nearest cutting, 250m to the east.

figure 7

The winter of 2013/14 brought another spate of landslips and flooding across the south east. The nature of the weather patterns was unusual in that the rain storms passing from south west towards north east across the region comprised very narrow bands of intensive focused deluge, locally exceeding 25mm/hr.

7.0 Effects of vegetation

Vegetation has a significant influence on earthwork behaviour, both within cuttings and on embankments, but it is important to recognise the difference between woody growth, such as trees, and non-woody growth, such as scrubby bushes and grass.

Contrary to popular belief, that “trees hold up the banks”, trees are overall detrimental to railway earthworks because both embankments and cuttings slopes are unnaturally steep and unsuitable for tree growth. In the case of cuttings, in particular, trees struggle to gain adequate footing and can collapse unexpectedly with devastating consequences. The growth of tree roots can physically damage a cutting face by the action of root jacking, which can release rocks and cause substantial wedges of rock to fall on the line.

Bushes, shrubs and grass, on the other hand, can help protect the cutting faces from sub-aerial weathering and provide a buffer from extremes of temperature.

Up until the 1960s the lineside was maintained by local workers whose responsibilities included the control of vegetation on the embankment and cutting faces, partly to reduce the risk of fires ignited by sparks from the steam engines. However, this maintenance regime was relaxed as steam traction was replaced by electric or diesel.

Repeated cycles of cutting back and re-growth of woody vegetation has a coppicing effect, creating a hazard of dangerously large root balls.

This can be exacerbated by the development of an overhang, or cornice, of loose surface material bound together by a tangle of roots along the crest of steep cutting faces, typically in chalk cuttings. Such cornices can overhang by in excess of a metre, creating a considerable hazard to the safety of the trains passing beneath.

Field studies, commissioned by Network Rail, into the effects of tree cover versus grass cover on the integrity or serviceability of clay-cored embankments has demonstrated that there is a 10-fold increase in disturbance to slope materials where trees are present (Scott et al, 2007). Each moisture cycle pushes soil particles outwards from the slope during the wetting phase, but during the drying phase the effect of gravity prevents the particles returning along the same path, there being a net down-slope shuffling of individual particles (figure 8). This process of surface creep is, thus, exacerbated in the presence of trees.

figure 8

Any stabilising benefit from the binding effect of tree roots is more than offset by the increased dis-benefit of greater ground disturbance. Removal of vegetation is likely to improve the serviceability, however, this may induce problems with the ultimate stability. Numerical modelling has demonstrated the potential benefits from removing woody vegetation from the upper half, or two thirds, of the embankment slope in order to reduce seasonal shrink/swell movements while retaining sufficient suctions on the lower parts to ensure stability of the earthwork (Powrie, 2013). However, the benefit to stability from soil suction is needed during the wetter winter months when the trees are not transpiring.

Extremes of desiccation can run well into autumn leading to irregular settlements on clay-cored embankments. The locations of track defects, or “rough rides”, are often directly aligned with the locations of high water demand oak trees, and the Autumn leaf fall data verifies that oak trees continue transpiration well after other species.

8.0 Earthworks Watch

The behaviour of earthworks is determined by five key variables (figure 9). The first two, asset type and geology, are fixed. The remaining three – condition, moisture and vegetation – are subject to constant change in response to weather, remedial intervention and vegetation management. Normal variations in moisture can be accommodated by the infrastructure until extremes of saturation, such as low soil moisture deficit (SMD) or high SMD, develop in response to prolonged periods of wet or dry weather, respectively.

figure 9

Put simply; low SMD results in a safety concern, high SMD results in serviceability issues.

The relationships between soil moisture and the behaviour of clay-cored embankments were highlighted by researchers at Imperial College working with London Underground to understand the causes of the failure of embankments comprising London Clay (Ridley et al, 2004). The key parameters of rainfall, SMD and hydrologically effective rainfall (HER) were originally developed for the agricultural industry to determine the optimum conditions for crop planting and harvesting but have been shown to be useful for understanding clay embankment behaviour.

Network Rail has developed these relationships to inform asset managers, maintainers and emergency response teams of the current condition of earthworks and the propensity for change in response to extremes of moisture or desiccation (figure 10). The system, called Earthworks Watch, uses both weekly and monthly data on rainfall and ground moisture purchased from the Met Office in order to monitor trends in earthwork behaviour.

figure 10

The analysis is presented in both map (weekly) and graphical (monthly) form and circulated widely within the organisation as a system which provides forewarning of trends away from normal conditions, be it too wet or too dry (figures 11 and 12). The monthly plots are accompanied by a brief narrative, or prognosis, to assist users in their application.

figure 11

figure 12

It is also important to recognise that seasonal variations in SMD depend on both precipitation and temperature, as a warm atmosphere can absorb more moisture than cold, thus increasing evaporation during the summer months.

The system cannot predict the exact location and timing of landslips or poor track geometry. However, the system provides an educated interpretation of trends which could lead to an incident and facilitate appropriate actions to be taken when, for example, responding to a train driver’s “rough ride” report. This avoids the tendency to over-react by imposing an operational damaging speed restriction for fear of imminent collapse when the SMD trend is reporting desiccation conditions rather than saturated conditions, which could result in catastrophic collapse.

9.0 Impacts on track geometry

Working closely with the track asset performance and maintenance teams within Network Rail it became apparent that the rainfall/SMD plots were a “light bulb moment” when overlaid with their track geometry plots (figure 13). The national plot for Good Track Geometry was showing reversals in an otherwise improving trend which correlated convincingly with periods of desiccation, high SMD, calculated for the south east routes.

figure 13

10.0 Mitigation

The system enables managers to carry out pre-emptive maintenance to track susceptible to extremes, such as early tamping to avoid speed restrictions or retrospective maintenance complicated by temperature – stressing issues. Bi-directional daytime tamping has proven successful in some routes. (This involves the running of both up and down off-peak services on the same track, freeing up the other line for pre-emptive tamping.) The three basic aspects drainage, vegetation and diggings (vermin) are most easily addressed.

While the simplest technical solution to failing earthwork is full re-construction, this is rarely selected as the lines in the south east are too heavily used to facilitate the disruption. Consequently, engineers are challenged with selecting retrofit options which can be carried out with minimal interference to railway operations. Well-proven and robustly engineered techniques are generally favoured but innovation is encouraged and some new techniques which are showing promise include discrete auger driven piles (Smethurst & Powrie, 2007) and the less invasive Electrokinetics (Lamont-Black & Weltman, 2010).

The continuous monitoring of earthwork performance is an aspiration of asset managers. However, in-ground instrumentation is expensive and too localised in extent. The wide deployment of low-cost, live surface movement monitors is achievable and, being real-time, provides the operators with immediate knowledge of a potential problem.

Train-mounted ground penetrating radar has potential to identify possible future embankment slip localities by interrogating routes for the typical ‘signature’ patterns identified from known slip sites. There are many areas of active research and one which shows particular promise is the development of electrical resistance tomography, akin to the computed tomography scans used in medicine (Gunn et al, 2009).

11.0 Climate change

Standing out among the consequences of climate change is the prediction that what we now consider to be extreme events will become increasingly frequent. Challenged with addressing future resilience to increasing propensity for weather extremes, Network Rail has worked with TRL to investigate the cost benefits of increasing the resilience of geotechnical assets to climate change (Reeves S et al, 2012).

TRL obtained data from the UK Climate Projections (UKCP09) weather generator on the typical weather patterns likely in the 2020s and 2050s and from this it was possible to calculate estimates of future SMD. East Anglia suffers from both expansive soils and extended periods of dry weather and so three sites on the Anglian route were selected for the case studies (figures 14 and 15).

figure 14

figure 15

The costing model found that preventative engineering would be less cost-effective than maintenance and delay costs over both 20 year and 50 year lifespans.

While this outcome goes against the aspiration to raise resiliance of the railway infrastructure in the longer term (Doherty et al, 2012), it emphasises the need to remain alert to the unpredictability of weather extremes and to maintaining reactive resources to attend to incidents as and when they occur.

12.0 Discussion


The Earthworks Watch system described here takes a fairly high level analysis of the precursory factors from existing spatial (geological and physiographic) and transient (rainfall and soil moisture) data sets which provides tangible guidance to the busy track and geotechnical management teams.

Further development is possible, in particular with regard to the effects of HER and a deeper analysis of the newly available data, in terms of improved spatial accuracy and sampling frequency.

In this context, Network Rail is developing an evolution of the system which replaces SMD with soil moisture index (SMI). This change improves both the correlation with earthwork failure data and also has improved granularity as SMI is reported on 12km by 12km grid, and is reported twice daily, whereas SMD is reported on 40km by 40km grid, and is reported weekly.


The authors are indebted to their colleagues, both within Network Rail and the wider industry, for their contributions and encouragements to write up the initiatives described.


Biddle G, 1990. The Railway Surveyors. British Railways Property Board/Ian Allan. London.

Birch G P & Dewar A L, 2002. Earthwork Failures in Response to Extreme Weather. Keynote Paper in Proc. 5th Int. Conf. on Railway Engineering 2002, London.

Doherty A, Dora J & Newsome C, 2012. Enhanced resilience in Britain’s railway infrastructure. ICE proceedings Vol 165 Issue CE6. ICE Publishing, London.

Gardner J W F, 1921. Earthworks in Railway Engineering. The Glasgow Text Books. Constable & Company Ltd, London.

Gunn D A, Haslam E, Kirkham M, Chambers J E, Lacinska A, Milodowski A, Reeves H, Ghataora G, Burrow M, Weston P, Thomas A, Dixon N, Sellers R & Dijkstra T, 2009. Moisture measurements in an end-tipped embankment: Application for studying long term stability and ageing. Proc. 10th Int. Conf. Railway Engineering, London.

Lamont-Black J & Weltman A, 2010. Electrokinetics strengthening and repair of slopes. Ground Engineering, April 2010.

Powrie W, 2013. Geotechnical Engineering in the Railway Industry. Rail Professional April 2013.

Reeves S, Reid M, Sharpe J & Bradbury T, 2012. The costs and the benefits of inceasing the resilience of rail geotechnical assets to climate change. Transport Research Laboratory Report RPN2409.

Ridley, A M, Vaughan, P R, McGinnity, B and Brady, K, 2004. Pore pressure measurements in infrastructure embankments. Proc. Advances in Geotechnical Engineering. The Skempton Conf., London 2004, London: Thomas Telford. Vol. 2: 922-932.

Scott J M, Loveridge F & O’Brien A S, 2007. Influence of Climate Change and Vegitation on Railway Embankments. Proc. 14th Int. Conf. Soil Mechanics and Geotechnical Engineering, Madrid, 2, 659-664.

Skempton, A W, 1996. Embankments and cuttings on the early railways. Proc. Inst. Civ. Eng: Construction History, 11, 33-49.

Smethurst, J A & Powrie W, 2007. Monitoring and analysis of bending behaviour of discrete piles used to stabilise a railway embankment. Geotechnique, 57(8) 663-677.








Have your say

You must sign in to make a comment

Please remember that the submission of any material is governed by our Terms and Conditions and by submitting material you confirm your agreement to these Terms and Conditions. Links may be included in your comments but HTML is not permitted.