Riding Network Rail’s New Measurement Train

Network Rail, as the infrastructure owner of Britain’s railways, constantly monitors the condition of its assets, particularly its 20,000 miles of track. One of the ways it does this is by using a fleet of inspection trains and vehicles, usually retired passenger stock fitted out with gauges, monitors and sensors of various sorts.

Flagship of this fleet is the New Measurement Train (NMT), although, since it has been in service for the last 15 years, it’s hardly new.

Affectionately known as the Flying Banana, due to its distinctive yellow livery, the NMT is equipped with the newest equipment, high-tech measurement systems, track scanners, and a high-resolution camera. A converted Intercity High Speed Train, the NMT covers 115,000 miles in a year and will capture around 10TB of image data every 440 miles.

Travelling at 125mph, it identifies faults quickly and accurately, helping Network Rail to keep the railway safe because it can discover problems at an early stage. Engineers can then make repairs or plan maintenance to prevent serious incidents, such as derailments.

To see what it can do, Rail Engineer was invited to join the NMT at Birmingham International station for a run to Northampton and back to Birmingham New Street.

Collecting and processing data

Steve Quinby, Network Rail’s head of delivery for data collection, and his team operate a variety of vehicles that examine the railway’s infrastructure in a number of different ways. In all, there are currently some 64 vehicles, making up three different trains.

The responsibilities of the Data Collection team begin, perhaps rather obviously, with data collection, “just as it says on the tin”, making use of a series of different infrastructure monitoring systems on the vehicles. These systems are all linked to highly accurate locational positioning systems which utilise GPS, inertial navigation and other methods, to ensure that each and every set of data collected can be identified to a precise location on the infrastructure.

Once the data has been collected, it is processed appropriately, to turn it into useful information for the management of the infrastructure. This is a crucial step in the process. The systems used generate enormous volumes of data but this is of little use until it is translated into information that can be understood and acted upon either by humans or by other intelligent systems.

The NMT’s predecessor, the High-Speed Track Recording Coach (HSTRC), had a bit of a reputation amongst track maintenance people for generating reams of computer paper, covered in numbers, much of which were meaningless. The useful outputs it generated were those that summarised this data in meaningful reports which directed maintenance staff to the important defects in the track, and the places where new defects were beginning to develop.

The next and vital responsibility of the team is the planning of train runs. This becomes quite a tricky process. It means balancing the requirements of carrying out infrastructure inspections at set frequencies on the one hand with, on the other, the demands of customers running increasing numbers of trains on the ever-busier network.

Occupying train paths with infrastructure monitoring trains might perhaps seem a waste, but standards rightly demand that inspections occur at appropriate intervals in order to ensure the safety of the network. If these requirements cannot be met, consequences follow, according to the level of risk implied by the failure. They might mean speed restrictions, or complete line closures, for example, until the missed inspection can be carried out. The use of trains for these inspection and monitoring activities is often the only realistic, safe and economic method.

As an example, the systems for track monitoring used on the NMT replace the old track patrols carried out by staff on foot. These were hugely time consuming, a significant safety risk for the staff, and less effective than the modern systems on the train.

In consequence the trains very definitely need to run, and to cover each route comprehensively, every time they are designated to do so. Steve’s team works very closely with train planners and operators to ensure that this happens. Of course, things do occasionally still go wrong, and so inspections get missed or are only partially completed. For instance, a monitoring train may be routed onto a Slow line when it was due to examine the Fast, or be sent along Platform 1 at a station when it was Platform 3 that was due for inspection.

Alternatively, there might be a failure associated with the monitoring train or its equipment. When this happens, the team has to generate a recovery plan that will get the missed infrastructure covered and, wherever possible, covered before some protective restriction, such as a speed restriction, has to be applied. Consequently, the plan for the trains is a critical responsibility for Steve and the team.

Finally, the team is responsible for ensuring that the monitoring vehicles and their systems are correctly maintained and calibrated. This is no small task, given the number of vehicles and the number and complexity of the systems on board them. In many instances, the systems are supplied and maintained by external specialist suppliers, who alone have the expertise required.

From primitive beginnings

Infrastructure measurement has a long history on British railways. Devices that measured the ride quality in passenger vehicles go back to the original private railway companies in pre-grouping days. One such was the Hallade system, which used pendulums and a paper reel recording system to record the displacements experienced on board a coach where it was placed.

Then there was the “Porcupine”. This was a primitive means of gauging the infrastructure around the track and was, in fact, a converted brake van to which were attached a series of poles. These stuck out all around the profile of the vehicle, which was hauled along a stretch of track to “gauge” it. Typically, it would be used at a tight bridge or in a tunnel. The poles would be shifted inwards when they struck an obstacle, and the theory was that, after the journey was completed, each pole would indicate the worst clearance along the measured route at the point on the vehicle’s profile where it was attached. A composite drawing representing each of the various pole positions would then be taken to be the worst-case profile of the route, and the limit of the allowable vehicle profile.

Having used such a vehicle early in my railway career, I can say that this was a pretty crude and potentially inaccurate process!

British Rail went through several stages of improved infrastructure monitoring systems, including Neptune track geometry measurement vehicles that used contact-based methods to record track at up to 20mph, and culminated, as far as track was concerned, in the HSTRC. This used similar track geometry recording systems to the NMT and ran at similar speeds, but it lacked the digitisation systems and did not have the other capabilities of the NMT.

It may not seem necessary to explain why Network Rail uses the modern technologies it now has, but there is more to it than just accuracy. First of all, the old methods, such as track patrolling, required large numbers of people to spend a lot of time out on the live railway, so anything that removes that requirement and the associated safety risks has to be worthwhile.

Modern technology is more accurate, as already suggested, but it is also far faster and more efficient. This means that it is now possible to carry out monitoring vastly more frequently. This is enabling a switch from finding and fixing faults to predicting and preventing them, which significantly improves both safety and performance.

The Hatfield train crash on 17 October 2000, caused by rolling contact fatigue failure of a rail, was the catalyst for a radical approach to track inspection which led to the introduction of the NMT and the other systems that are now operated by Network Rail.

New monitoring systems come about in response to business needs. When a new problem is identified, Network Rail will approach a range of suppliers of relevant technology. These will then propose possible approaches, and Network Rail will select the most promising for joint development with the supplier concerned. Once tried and tested, the new system will be implemented on the relevant monitoring trains.

Between them all, Network Rail’s inspection fleet works 24/7 and covers some 750,000 track miles in 2,000 recording shifts each year. The volume of data generated is enormous, since the NMT alone produces about 7TB for every 350 miles covered. Data is recorded to hard drives and these are delivered to the data centres in cases carrying 21TB.

On-board the development coach

The NMT, which has two monitoring coaches and seven monitoring systems, is manned by on-train technicians (OTTs) who control the infrastructure monitoring processes. The so-called ‘development’ coach contains two main systems – the PLPR, or plain-line pattern recognition system, and the Fraunhofer system.

The first mentioned utilises extremely high-speed digital photography to capture detailed images of the track as the train passes over it. At up to 125mph, this takes an image every 8mm along the track. Sitting alongside this, a LiDAR system scans the track simultaneously. The digital images are analysed by algorithms that identify anomalies which may be track defects. These, known as “candidates”, might be anything from a missing clip to a rail surface defect. The data and the candidate defect information are all saved onto hard-drives that are transferred later to the data analysis centre at Derby.

At Derby the candidate defects are reviewed by inspectors who confirm whether or not they represent real threats. The LiDAR information is used in this process when the photographic images are unclear – for example, if a rail clip is obscured by debris, the algorithm checking the photographic data would throw up a potential defect, but the inspector would use the LiDAR data to see that the clip was actually present and then delete the defect record.

Under Railway Group Standards, Network Rail is allowed 72 hours after the train run which collected the data to get defect information to the maintainers. Of course, there may be defects that need higher priority than that, and there are appropriate measures in place to manage these. For the most serious cases, which need traffic to be stopped, the system would identify these to the OTT by audible and visual alarms. The train would then be stopped as quickly as possible, thus blocking the affected track. Action would then be taken to get the line blocked by normal procedures, have the defect corrected, and get the line reopened.

The Fraunhofer system is a contactless overhead line monitoring system. It employs lasers to measure the position of the contact wire. Like the PLPR system, this one is highly accurate, even at 125mph. It identifies the height and stagger of the wire and identifies any locations where the wire is outside the allowable tolerances. The system includes a digital camera system that looks at the contact wire to measure its wear. The hard-drives with this data also go off to an analysis centre, in this case at Milton Keynes.

Track inspection

The track geometry measuring system, with which many readers will already be familiar, and which would be recognisable to someone familiar with the old HSTRC, is in the ‘production’ vehicle of the train. This system measures the track geometry and calculates, for each eighth of a mile, a standard deviation (SD) for each of the parameters measured. Each SD gives a measure of the extent to which the track deviates from the ideal geometry for the particular parameter to which it relates. Track quality standards lay down threshold SD values for each parameter according to the track category of the line (a measure that takes account of line-speed and traffic intensity). There are planning thresholds and immediate action thresholds.

In addition to the actual geometry data and the SDs, the system incorporates a six-foot laser scanner, that checks the distance to the adjoining track, and analysis that looks out for cyclic top and dynamic gauge defects (issue 157, November 2017), both of which are derailment risks that are not easy to identify.

Like the PLPR system, the data recording and analysis system on the NMT is linked to the highly accurate positioning system of the train, and issues reports to maintainers for action. The data is also used for more sophisticated analysis of the assets and trends in their condition, assisting the development of long-term plans for renewals and more.

The inspection fleet

In addition to the NMT, other units in the fleet include the ultrasonic test trains (UTUs), the structure gauging train (SGT), the radio survey coach and Mentor.

The UTUs use ultrasonic rail examination systems provided by Sperry to check for internal defects in the rails. These can run at up to 35mph, a maximum determined by the speed of sound in the steel.

Laser scanning enables the SGT to capture structure gauging data vastly more quickly and accurately than the old “porcupine” wagons, to facilitate the clearance of vehicles to run on the network and to ensure that the clearances around the tracks have not been reduced in any way, such as through the planned or unintentional movement of the track or through distortion in a tunnel lining.

As one would expect, the radio survey coach monitors the strength of radio signals around the network, to ensure that safety critical and operational railway radio signals are available and reliable as required.

Mentor is a coach fitted with overhead line monitoring equipment, this time employing contact-based systems rather than the laser system of the Fraunhofer units on the NMT.

There are secondary systems in use on the trains as well. Many are fitted with forward-facing video recording. On the UTUs, there is a laser system manufactured by KLD Labs that measures rail profiles to detect wear and unsafe profiles, and GPR (ground proving radar) that allows users to “see” under the track to a depth of between 2½ feet and 6 feet below the surface. GPR results show where there are interfaces between different materials, such as the ballast/formation boundary. They can also show if ballast is clogged with clay or other contaminants and where it is waterlogged.

In order to ensure that the railway network is fully covered by track geometry measuring systems, MPVs (multi-purpose vehicles) are also fitted with the necessary equipment. These are able to cover short sections of track that cannot be reached by the NMT or the Track Recording Unit (TRU), a two-car unit based on a Class 150/1 train, typically in stations or other complex areas which need covering at night. Trials have also been undertaken with track geometry systems fitted to service trains.

The implementation of all this technology has been a real success. The UTUs, with their Sperry nine-sensor ultrasonic wheel probes, have been a major part of Network Rail’s dramatic reduction in rail failure numbers from over 1,000 each year in the late 1990s to only 125 in 2016. This alone has had major benefits in safety, performance and financial terms. It has also attracted the attention of other railways around the world, who wish to understand how it has been achieved and copy Network Rail’s example.

The ability to predict and plan is further improving safety, performance and availability, and is reducing costs.