PREVENTING interruptions to passenger and freight services during construction works taking place near track is essential to maintaining a safe and efficient railway. Normally during the design phase of a construction project, any potential impact on the supporting ground around rail infrastructure is carefully considered and appropriate construction techniques selected to minimise any impact. Nevertheless, it is still vitally important to monitor track during construction to check that any movement is within tolerance in order to continue operating rail services safely.

 

There are several methods used by surveyors and/or monitoring engineers to measure track movement. These are usually either contacting sensors like tilt meters, or survey devices such as a manual survey with a precise level or a robotic total station. However, track engineers in the Netherlands are utilising another potential solution which could offer an alternative.

Engineers at Kouwenberg Infra, working for Dutch infrastructure manager Prorail, were looking for a means of minimising track access to improve safety and to reduce disruption to services and project lead times as well as limiting the costs and delays associated with obtaining a possession.

As a result they selected a non-contact system, the Dynamic Monitoring Station (DMS) developed by Imetrum, Britain, and delivered in Benelux by RailAssist. The system utilises a high resolution, non-contact video system and is able to continuously monitor track displacement during installation of a new under-track crossing by open front drilling.

ImetrumDMS uses patented image processing algorithms embedded within video gauge software. During operation, the user identifies "virtual targets" in the video image and these are then tracked in real time, outputting measurements frame-by-frame. The DMS is camera-based so it is able to keep a visual record of all load events. It also offers the possibility of recording measurements at distances from less than 1m to more than 1km from the object while recording up to 2000 measurements/s across multiple synchronised cameras, enabling coverage of larger areas or offering different views of the same object. Real-time output is possible at speeds ranging from 1000 measurement/s to one per day, while the DMS has an optical measurement resolution of better than 0.1mm for bridge and track monitoring.

To enable a greater understanding of key events, it is possible to take a video of the test either as a simple visual record, or to reanalyse additional data points offline. For short-term monitoring, the system is battery-powered and can use the natural appearance of an object as a "virtual target," so there is no need to gain access to track infastructure. All this simplifies project planning, saving both time and money.

Part of a benchmarking trial conducted at the National Physical Laboratory, London, where side-by-side measurements were recorded. From these tests, Kouwenberg Infra and ProRail were looking for the best value method to obtain high-quality track displacement data as they were about to specify continuous monitoring of pipe drillings for all pipes with a diameter of more than 1.2m.

Dutch project

For the Dutch project, RailAssist continuously measured the displacement of two rails while a 1.2m-diameter concrete pipe was driven through a railway embankment using open front drilling with normal rail services continuing throughout construction. The conventional Total Station measurement method was also conducted in parallel to compare the technique used and data produced.

The Imetrum system was set up on the embankment at a point of safety and outside the anticipated zone of influence of the pipe drilling. Setup was 3.5m from the nearest rail and fenced by a tactile chain to comply with ProRail's safety regulations.

In this scenario a single tripod-mounted camera system focused on a section of track directly on top of the line of the concrete pipe at a shallow elevation. The DMS can run on 12V batteries, but in this instance was powered by an on-site generator. The DMS' sample frequency was set at 2.5Hz, which on this occasion was considered the preferred sampling speed for monitoring, while the field of view was defined as 1m on each side of the line of drilling.

Here the DMS system operated using the natural pattern of the rail clips as virtual targets for monitoring displacement at each of the four sleepers. However, this relies on the type of rail clips used on the line and the required field of view. As an alternative marks can be applied using spray paint, self-adhesive or magnetic strips.

As the works progressed, real-time measurement data was captured, recorded and displayed using video gauge software from a safe point at the foot of the embankment. This allowed the contractor to instantly monitor any impact their work was having on track alignment.

For this project it was agreed that as long as the track deflection was within ±2mm, drilling could continue. If the track deflection was exceeded, the speed of the drilling would be lowered or stopped. The actual safety value (ascertained as a safe minimum prior to any risk of derailment) for this particular track is a vertical displacement of the sleeper of 20mm. However, displacements of less than this have been shown to have an impact on ride quality.

Four sleepers were tracked in real time using their natural pattern as the 'target' for tracking by the DMA, with the data displayed live on screen. The data obtained returned calculated resolutions of 0.03mm, much higher accuracy than the ±1mm required by the project. Overall track movement was within a band width of +1 and -2mm, well within the safety limits. Elastic track displacements caused by trains running along the line were also noted at approximately 4mm and did not alter throughout the duration of the test.

Following the day on site, an assessment of capability was undertaken by RailAssist, comparing the DMS against the current preferred monitoring method for similar works, a total station.

Key benefits

The three key benefits identified by the engineers were:

  • reduced cost: As the DMS is able to use the structure's natural patterns to monitor movement, there was no need to obtain a track possession to add reflective targets on the rails, saving money by eliminating the approval process, and also saving time in arranging the track possession. This cost saving was recently used by another contractor in the Netherlands to partly justify purchasing the system
  • improved safety: The project was safer as there was no need for survey staff to access the track, and
  • improved data: As the DMS was able to monitor multiple points simultaneously, it was possible to monitor several areas of interest under train loading, giving a more accurate picture of the impact of the works.

While this project involved a relatively simple single camera setup, monitoring in 2D at 2.5Hz, other projects undertaken in Britain have involved much higher speeds and many more cameras.

Work to characterise the trackbed before and after remedial ground stabilisation works has been undertaken in a number of locations, including near Manchester. Here a multi-camera DMS system was used, with cameras running at 30Hz positioned at 50m intervals in the 150m area of concern.

The ease of taking measurements from distance has been generating considerable interest among those involved in understanding wheel-rail and track component interaction. With current or new designs or devices typically using strain gauges, linear variable differential transformers (LVDTs), geophones and accelerometers to generate measurements, in a track environment, this is a time-consuming and costly process, meaning it is rarely done.

As a result, a project designed in partnership with LB Foster Rail Technologies and undertaken at the Rail Alliance's Long Marston site in central England, attempted to establish whether there was an easier way of recording measurements such as strain, displacement or rotation of a similar quality to using contact sensors, without the problems inherent in attaching them to track.

Using pre-calibrated, synchronised stereo cameras to generate live measurements in 3D, a measurement resolution of 0.01mm was achieved at 6m from track. As a detailed understanding of track behaviour and multiple points on the track component were required, self-adhesive targets were used to pinpoint measurement locations. The system was operated for two days, capturing around 100 train passes, and providing sufficient information to make decisions on whether to adopt the innovation.

Adoption of this technology is well established for bridge monitoring, but it is still in the early days for rail track and components, with work now underway to increase the understanding of end user requirements.

The technology has proven to be accurate and capable, but further development of best practice engineering is required to ensure consistent implementation within survey and monitoring teams. At the same time, engineers at Imetrum are continuing to evolve the user interface and measurement capabilities of the DMS to further tailor the system for rail applications. Avenues for development include optimising the technique for long term (multi-year) monitoring as an alternative to laser based systems.

Already sampling speeds of around 300Hz are possible in real time, which are more than enough for all but the fastest railway lines around the world. But with processor speeds and camera specifications continuing to develop, speeds of 1000Hz and more are just around the corner, which could open up a range of new monitoring opportunities.