During the recent wet winter of 2013/14, as continual storms battered the south and west of the UK, geotechnical engineering featured rather more in the news than usual. Reports of landslips, sinkholes, groundwater flooding and coastal failures showed the public how much the infrastructure that supports their daily life, is itself supported by that which is hidden below the ground surface. Dramatic landslips affected the railway network, for example, several failures occurred on the Tonbridge to Hastings line in Kent which closed it for a month.

Given the age of the UK railway network, much of it dating from Victorian times, it is not surprising that minor landslips are regularly seen. The original cuttings and embankments were constructed without modern engineering knowledge, design processes and construction methods. Many of them have mixed geologies, are over-steep and poorly compacted by today’s standards; few have site investigations or as-built records which usually help engineers to understand the ground’s hidden risks.

Managing this ageing earthworks asset to keep the railway operational is a real challenge, particularly in periods of exceptional rainfall such as we have seen this winter in the UK, and also recently in the very wet summer of 2012. This sustained wet weather coincided with an increase in earthworks failures and associated disruption. A robust balance of proactive and reactive measures is needed to keep the railway operational.

Both new technologies and an approach that focuses on risk help to make the most of limited information and resources. Arup geotechnical engineers, asset managers and risk specialists have great confidence that the use of riskbased prioritisation methods will add significant value to the complex and challenging decision-making process; this is a view shared by Network Rail.

In this digital age, new tools are becoming available to support the earthworks asset manager’s role. LiDAR topographical surveying, GPS referencing of video and photography, and wireless technologies improve coverage of remote parts of the network. GIS databases help create a joined-up digital model, containing a wealth of data that can be overlaid and analysed to understand the risk profile of a railway network. Formal risk management techniques help to compare and prioritise the most significant risks across different asset types, including earthworks, even where there is a degree of uncertainty.

Railway earthworks and their behaviour do have inherent uncertainties, some of which are listed below:

  • natural ground is typically variable with mixed geology in many earthworks.
  • earth fill was often end-tipped into railway embankments in an age when the importance of good compaction wasn’t fully understood, and many embankments have since been topped up or widened with ash and other poorly controlled materials
  • the variability of the ‘trigger’, for example the exact amount and location of heavy rainfall
  • a lack of information that would help to assess an old asset, such as geotechnical desk studies, ground investigations or as-built drawings
  • the interaction between earthworks and other features such as drains and culverts: these are vital for removing harmful water but are not always under the railway operator’s control.

The result of this is that earthwork failures will occur in unexpected places, not only at the usual problem ‘hot spots’. Given the size of the UK rail network, it is impractical to inspect every earthwork during and after heavy rain or to impose blanket speed restrictions. On the other hand, it is recognised that the likelihood of a train colliding with a failed earthwork and causing a safety incident is higher during periods of heavy rain.

Individual slope failures cannot all be predicted, due to the uncertainties listed above. However, a network-wide system of prioritising the highest risk sites has the advantage of highlighting not only the earthworks most likely to fail due to heavy rainfall, but also those busy routes or vulnerable tracks with the worst potential consequences if a landslip should occur. Figure 1 shows a simple framework for assessing an earthworks, asset portfolio for the risk of rainfallrelated failure. Matching the highest risk sites up to the rainfall forecast allows targeted control measures to be applied for each storm event.

Figure 1: Risk assessment framework showing the interaction of likelihood and consequence factors

Likelihood of an earthwork failing

In theory, the likelihood of an earthwork failing can be assessed using statistical analysis, modelling, or expert judgment. In reality, and certainly in the case of the UK railway network, detailed modelling of each and every slope is impractical. Therefore, expert judgment can be combined with analysis of the limited statistical data available, to rank embankments and cuttings in terms of their relative susceptibility to failure.

The likelihood assessment starts from the available information about the nature of the earthworks. For Network Rail, data about its 17,500km of cuttings and embankments is gathered by periodic walk-over examinations by geotechnical engineers, and uploaded into a networkwide database. Each 100m (five-chain) length of earthwork is scored by its overall condition, depending on both ‘static’ features such as geology and geometry and ‘dynamic’ (i.e. changing over time) observations of movement or landslip potential.

A key indicator for likelihood is an earthwork’s condition score. Other inspected features that could make an asset more vulnerable in the particular scenario of heavy rainfall, such as poorly maintained drainage, nearby watercourses, natural seepages and water concentration features, can also be used to increase an individual asset’s ranking in terms of failure likelihood.

The consequence side of the risk equation is less about the geotechnical factors and more about the rail environment and the layout of the track. Assuming that any earthwork failure could lead to a derailment, due to collision with debris from a cutting or loss of support to the track from a slipped embankment, the events that might follow could range from derailment only to subsequent collisions with oncoming trains, obstacles such as bridges, part of the train falling down an embankment or even into water.

All of the above scenarios are likely to be more severe where higher speed trains are involved, and the safety consequences would increase for heavily occupied trains. Clearly, different locations around the network will have very different severity of potential consequences, and the risk-based decision-making framework should take these into account.

Figure 2 illustrates a risk matrix combining likelihood and consequence scores and hence placing all earthworks into one of four risk categories from ‘Low’ to ‘Very High’. Using the logic described above, a typical ‘Very High’ risk rated site might comprise sidelong ground: with a cutting with a high failure likelihood (poor condition, blocked drains) on one site, and a rail vehicle drop-off hazard on the downhill side of the track. Another high-risk scenario might be a cutting in poor condition with drainage problems and high linespeed/ multiple tracks, leading towards a high collision hazard such as overbridge abutments.

Figure 2: Risk matrix showing the interaction of likelihood and consequence components to assess overall risk

Testing an ageing infrastructure to new degrees

In summary, recent weather events as well as long-term changes in weather patterns are testing an ageing and stretched infrastructure asset portfolio to new degrees. Changes are required to ensure that reactive measures are in place to continue to manage earthworks safely as well as proactive measures to provide a long-term sustainable programme of works that will ensure that railway earthworks remain fit for purpose in the future.

This article represents the personal views of the authors and not Arup or Network Rail.