Learning from Carmont

Although the Rail Accident Investigation Branch (RAIB) has yet to produce its final report on the fatal train crash at Carmont on 12 August 2020, there are now clear lessons about earthworks and weather management from this tragic event.

Network Rail’s earthworks assets comprise 70,000 soil cuttings, 20,000 rock cuttings and 100,000 embankments. Most are over 150 years old and were built when there was little understanding of the science of soil mechanics. As a result, cuttings were overly steep, embankments uncompacted and drainage inadequate. Furthermore, earthworks are more vulnerable to changing weather patterns, resulting in longer periods of prolonged, intense rainfall and hotter, drier summers.

Since 2004, there has been an average of 100 earthworks failures per year; yet, in CP6 – up to 30 November 2020 – the annual average was 222. Nevertheless, there has been a significant reduction in high-consequence earthworks failures and derailments since 2004. Unfortunately, the single CP6 earthworks derailment was last year’s fatal Carmont accident.

This reduction follows increasing earthworks expenditure which, in 2010, included work to stabilise the steep Carmont cutting and install a crest drain immediately adjacent to the derailment site, as reported in Rail Engineer (Issue 77, March 2011). This explained how, partly due to an increase in the area of farmland and reduced effectiveness of the field drains, drainage run-off from the fields above had overtopped and was destabilising the cutting.

Interim reports

On 1 September 2020, Network Rail published its interim report on the Carmont derailment which explained the company’s procedures for earthworks management. It noted that, although £1.3 billion is being spent on earthworks in CP6, it is not practicable to rebuild thousands of miles of earthworks to modern standards, so failures are still to be expected. It also detailed immediate actions taken to mitigate this risk including additional precautions for managing earthworks and operating trains during severe weather. 

The report also specified the remits for two task forces led by independent experts. One, under Lord Robert Mair, reviewed the management of earthworks whilst another, led by Dame Julia Slingo, considered weather forecasting. Published in March, these reports aimed to ensure that Network Rail has the expertise, technology and systems to better manage earthworks, and make the best use of weather data.

RAIB’s interim report, published on 19 April 2021, highlighted the sad irony of the derailment being caused by a failed crest drain, installed to protect the cutting. The train had collided with stones washed onto the track from this steeply-sloping gravel-filled drain into which the local topography had directed large amounts of water after 51mm of rain had fallen in three hours, 75% of the area’s average monthly rainfall.

The RAIB investigation had found that the missing gravel had exposed a buried drain pipe for 8 metres upslope of a catchpit where the drain was under a steep gorse-covered slope. RAIB found that this part of the drain was not in Network Rail’s drain maintenance database and was unable to find evidence of it being inspected between its construction and the accident.

RAIB’s ongoing investigation will consider the design and construction of the failed drain. It will also look at the response to severe weather events, decision-making at times of widespread disruption and the mitigation of derailments at such high-risk locations.

Drainage deficiencies

The lack of an asset database record for a drain installed less than ten years ago underscores the conclusion in Lord Mair’s report that drainage is generally regarded by Network Rail as a ‘child’ asset which supports the performance of earthworks and track. As a result, Network Rail has a dated drainage system about which it has little knowledge.

Although his report commends Network Rail for its substantial effort in developing a comprehensive earthworks asset management system, it notes that the Earthworks Technical Strategy does not consider drainage or vegetation management in a meaningful way and that, hence, there are key omissions in the earthworks policy, for example drainage competence.

It notes that earthworks stability is dependent on drainage systems that were installed to default designs that took no account of run-off and water flow. Furthermore, there has been little enhancement of drainage, with replacement over the years being like-for-like.

Rail Engineer’s 2011 feature on the Carmont cutting works shows there was an awareness of the increased run-off from the fields above. Yet the way this run-off overwhelmed the failed drain highlights the importance of drainage having sufficient hydraulic capacity.

The report calls for drainage maintenance and cleaning to be undertaken by dedicated teams with sufficient competent staff, as is the case for earthworks examinations. It considers drainage maintenance to be under-resourced as the off-track teams who do it are often overloaded with drainage inspections or diverted away to respond to incidents.

Reviewing earthworks management

Drainage was just one aspect of Lord Mair’s 543-page report which also considered earthworks vulnerability, as well as earthworks, drainage and vegetation asset management, and monitoring technologies. It reviewed the historic nature of earthworks assets and provided an academic treatise of their soil mechanics and failure mechanisms. In considering changing weather patterns, the report noted the very strong correlation between earthworks failures and rainfall over the past two decades.

Lord Mair concluded that the dominant reason for continuing failures is the exposure of over-steep and previously failed slopes to rainfall patterns not previously experienced. His report expressed reservations about Network Rail’s use of soil moisture index to monitor earthworks instead of the more important parameter of pore water pressure.

The threats from climate change were considered to be:

• longer periods of prolonged rainfall in winter months leading to rising groundwater levels and higher slope pore pressures
• more frequent periods of more intense rainfall triggering washouts and debris flows
• hotter, drier summers increasing the amplitude of cyclic slope pore pressure changes, particularly clay embankments
• increased demand on drainage capacity and the risk of it being overwhelmed.

The report considered that “predicting exactly where failures will occur is like looking for a needle in a haystack” and noted that a more practical approach is to “search for the haystacks”, i.e. vulnerable lengths of slope. It noted that a localised failure strongly indicates that the remainder of the similar slope is vulnerable to future failures.

It also considered how vegetation can have both a beneficial effect – reducing surface erosion, providing root reinforcement, avoiding channelling of flows, maintaining surface pore water suctions – and a detrimental one with blocked ditches and pipes, leaf fall, tree fall and desiccation by the track. Although Network Rail has done work to improve vegetation management, the report considers a more integrated approach to the management of earthworks, drainage and vegetation is needed.

Asset management

The report commends Network Rail for its substantial work in developing a comprehensive earthworks asset management system. It notes that, unlike track or rolling stock assets, earthworks are inherently variable and that this is further affected by uncertain environmental conditions. Furthermore, the earthworks failures trend has significantly worsened since the start of CP6. The 251 failures in the first year of this control period is about double the number in each of the previous three five-year control periods.

Recommended changes in examination regime include extending the season to include April and undertaking examinations during or shortly after heavy rainfall. The use of drones and helicopters is recommended both for this and identifying any changes that could adversely affect earthwork stability between routine examinations.

The report considers the effectiveness of the Earthworks Hazard Category (EHC). It notes that 59% of the 2019/20 earthwork failures were classified in lowest risk ‘A to C’ pre-failure categories. Furthermore, around two-thirds of the failures in the early part of CP6 were at sites where no work was planned during the control period. This suggests that many vulnerable earthworks are not included in the investment plan. Hence, it recommends a review of the EHC process.

The difficulty faced by Network Rail’s geotechnical engineers in making effective asset management decisions on the basis of multiple and disparate data sources is also considered by the report which concluded that an improved earthworks asset management system that uses data from intelligent infrastructure is needed.

Monitoring to mitigate

As it is currently not possible to detect or prevent all earthwork failures, the report considers their mitigation. This needs reliable monitoring to inform Network Rail’s engineers of the condition of the more critical geotechnical assets.

There are two objectives for such monitoring: detection of failures affecting the safety of the line and collecting data to predict possible failures.

Techniques used or trialled by Network Rail for rapid and instantaneous detection of failures include distributed acoustic sensing by optical fibres surface-mounted tiltmeters and inclinometers and instrumented flexible barriers for rock and soil slopes.

Those with a slow failure response that are suitable for the collection of condition data include:

• distributed acoustic sensing by optical fibres
• the promising application of wireless tiltmeter systems
• satellite InSAR (Interferometric Synthetic Aperture Radar) which compares radar images over time to detect ground deformation to millimetre accuracy
• aerial and land-based LiDAR and photogrammetry which international experience indicates to be promising surveillance technologies for slope and landslide management, and the need for train-mounted LiDAR systems to update the geometry and features of cutting slopes
• Electrical Resistivity Tomography (ERT)
• Shape Acceleration Arrays (SAA)
• the continuation of acoustic sensing as part of Network Rail’s R&D programme in view of its potential to detect instability of soil and rock slopes.

More widespread use of helicopter flights to inspect earthworks was considered necessary, especially in hilly or mountainous terrain and after an extreme weather event. In Scotland, five such flights are made each year, specifically for earthworks inspections.

The identification of potential embankment failure sites by Network Rail’s track geometry data collection and analysis workstream was commended, as was the company’s Intelligent Infrastructure programmes and impressive R&D portfolio of novel earthworks monitoring technologies.

Forecasting chaotic weather

Dame Julia Slingo’s report considered how Network Rail could obtain the best possible weather forecasts and make best use of them.

It looked at the latest advances in weather forecasting to show what could be made available to Network Rail and described how forecasts start with a global atmospheric assessment requiring 100 million observations to be processed. Global forecasts are needed as the UK’s weather often has its roots from the other side of the globe. They provide the boundary conditions for finer-scale UK regional forecasts that are then undertaken.

This process takes two hours and is repeated every six hours. It requires 20 quadrillion calculations and generates 10,000GB of data. Such forecasting is one of the most complex computing applications, involving over a million lines of code and the use of dedicated supercomputers.

Due to the chaotic nature of climate systems, current practice is to produce an ensemble of forecasts to assess the probabilities of a range of outcomes. These are continually reviewed to provide increasingly narrower spread closer to the time of the forecast.

Kilometre-scale forecasting 1-3 days in advance is now possible due to the recent development of models that accurately represent the landscape and a better understanding of the physics of thunderstorms and convection. Nowcasting is a technique that forecasts the next 1-2 hours using optical flow techniques that extrapolate weather radar images to detect the severe convective storms of the type seen at Carmont.

At the time of the derailment, Network Rail’s weather advice used a 10km weather model that could not capture local extremes. However, since then there has been a rapid development of the company’s weather services.

Harnessing forecasts

Using weather data to best manage the risks to the operational railway requires an understanding of how rainfall translates into geohazards and the use of forecasts to take timely operational decisions.

The rail network is particularly susceptible to hazards such as surface flooding, washouts and earthwork slides. The weather report notes that Carmont showed how hourly rainfall intensity may be a critical factor driving earthworks failures.

Before the derailment, Network Rail used Extreme Weather Action Teleconferences to advise routes of forthcoming adverse weather. This was considered to be a static process, with limited capability to adjust alerts in an evolving weather situation.

Action to be taken was in accordance with thresholds which, according to the report, needed a major overhaul to reflect variations in exposure across the network, particularly in respect of rainfall. It noted that, after Carmont, the company acted swiftly to improve preparedness for extreme weather events and their impact on earthworks, with the development of a Convective Alert Tool.

Dame Slingo’s report recommends use of the following weather management framework:

• Awareness – possible regional red weather alerts are recognised 4-5 days out
• Preparation – route controls assign red weather alerts two days out using of kilometre-scale forecasts and begins to take preparatory action
• Response – monitoring and alerting by nowcasting during extreme weather events
• Recovery – establish priorities and provide weather forecasts for recovery.

Implementing this framework requires both competent personnel and effective systems that clearly present relevant data to support effective decision making. The weather task force found that there was a gulf in expertise between those creating weather information and those receiving it. Hence it considered that Network Rail should have a ‘weather academy’ to ensure its staff are well-informed users of weather services. This includes scale and predictability awareness. For example, a 10km scale thunderstorm has an average predictability limit of around three hours whilst a 100km frontal rain system can be forecast two days ahead.

The weather task force concluded that a suitable digital platform to present relevant weather data is needed and that this should use Network Rail’s Geographic Information System to provide this data on the network map to best aid effective decision making.

Such a prototype system, together with Met Office data, was used in a ‘Sandpit’ trial in December. This correctly predicted localised events and generated positive feedback from those involved. One Route Operations Manager, who was particularly impressed by the detail and granularity provided, was confident that such systems would enable mitigation measures to be applied in the right place and also could see the end of miles of unnecessary blanket speed restrictions.

Reducing the risk

Modern standards require a significant amount of earthwork retention when double-tracking a single line or reopening a previously closed railway. However, as noted in Network Rail’s interim report following the Carmont derailment, rebuilding thousands of miles of earthworks to such standards is not practicable in the short-term, either from a funding or delivery perspective.

Prior to Carmont, the last fatality from an earthworks failure was in 1995. Yet, with various earthworks derailments since then, there was a recognition that this was a significant area of risk. Hence £1.3 billion is to be invested in earthworks and drainage in CP6, nearly double that of CP4. Furthermore, Network Rail had improved its earthworks management for which it was commended in the task force reports. Yet Carmont showed that more needed to be done.

The depth and detail of the reports led by Lord Mair and Dame Slingo reflect the expertise of their members and offer many best-practice solutions. Whilst it is not reasonably practicable to detect or prevent all earthworks failures, the recommendations should significantly reduce the risks involved.