Clive Pennington discusses Remote Condition Monitoring developments at Manchester Metrolink:
In 2002 I was part of the operations team tasked with supporting the construction and maintenance of Nottingham’s new tramway. At 14km (8.7 miles) long, with 23 stops and 15 trams, it was a small system with around 4km (2.5 miles) of on-street-running.
During the design and construction phase, we spent a lot of time planning our ongoing maintenance strategy, fault reporting and condition monitoring. We had all the same concerns as today’s operators and maintainers in terms of looking after the assets, considering rail and component wear and tear, and the interactions between vehicles and infrastructure such as track and overhead lines. But, given the size of the system and the cost of the available technology at the time, manual inspection was the only real option for monitoring asset condition – and this was necessarily resource- and time-intensive.
‘Automated’ solutions for gathering asset data were available, but these were created for heavy rail applications and capable of gathering data over hundreds of miles at speeds of up to 130km/h (80mph). Although this would have provided useful information about our network, the cost of the technology, either as a service or purchase, was prohibitive. It wasn’t scalable to our requirements and the economies of scale just weren’t present for a system that could be walked in less than a day, or for systems capable of measuring hundreds of wheels in service when the fleet of 15 trams returned to the depot each night.
Remote Condition Monitoring (RCM) and data collection as we know it now was in its infancy. Remember that in 2002, the first BlackBerry smartphone had only just been released, iPods were still fairly radical technology for most people and 64-bit processing power and 3G were still firmly in the ‘coming soon’ category. There was no ‘Internet of Things’ (IoT) and data transmission networks just weren’t up to the job without prohibitive levels of investment.
Technical and engineering opportunities for maintenance efficiencies were undoubtedly limited, particularly for the then-small operations in the UK, so asset inspections were carried out in the traditional manner with track walks, manual measurement and daily checks of the trams in the depot.
Fast forward 15 years to 2017 when the joint venture of Keolis and Amey (KeolisAmey Metrolink – KAM) was awarded the ten-year contract to operate and maintain the Metrolink light rail network that serves Greater Manchester. The Transport for Greater Manchester (TfGM) contractual specification included a number of requirements, including a key commitment for the concessionaire to collect asset condition information. In response KAM gave commitments surrounding asset management systems, processes and the collection of asset data – this would inform TfGM’s ongoing renewals programmes and additionally be invaluable for the operator in ensuring the smooth day-to-day running of the network. It would also allow appropriate resources to be dedicated to maintenance for an ever-expanding light rail network – now the UK’s largest, with almost 100km (62 miles) of route, eight lines (nine in 2020), two depots and 120 trams (with another 27 on order) and carrying around 120 000 passengers each day.
KeolisAmey Metrolink’s obligations included automation of wheel and tyre measurements; rail wear and track geometry; and video capture of strategic assets. Further contractual requirements include monitoring of contact wire height, stagger and thickness, as well as monitoring tram ride quality.
Building a business case
Over the last few years, ‘intelligent infrastructure’ has become an important buzzphrase for the rail industry. Embedded devices, linked by both fixed and wireless infrastructure, now report the condition of network assets through real-time alerts, enabling faster, more effective maintenance. Enabled by the IoT, ‘intelligent devices’ can track assets 24 hours a day, seven days a week by gathering real-time data through sensors, cameras, and other mobile technologies, and feeding this data into analytics engines for further processing. Such systems can monitor temperature, motion, vibration, pressure, and all manner of other criteria. Once this is collected, operators have a holistic view of their networks to identify where maintenance is needed before components show signs of compromise and/or failure.
While new-build networks around the world have some of this functionality ‘built-in’ from construction, operators of established networks – often up to, and beyond a century old – will traditionally have relied on intelligence gathered by personnel out in the field. Such personnel would transfer this knowledge via paper reports detailing the integrity, safety, and compliance of their vehicles and infrastructure. Keeping track of this information and using it to inform investment decisions was often a time-consuming and problematic process.
Whilst Metrolink may now be much larger than Nottingham’s network of 2004 – allowing economies of scale to be realised more easily – the cost of the technology is also now more affordable. These factors combine to make a compelling business case.
With advances in sensor technology, processing power, IoT and cloud computing, RCM solutions have become more scalable and affordable – offering numerous possibilities for the transport industry, and light rail in particular. As part of the concessionaire, Amey’s Consulting division was employed to develop a range of solutions to meet KAM’s committed obligations, drive efficiencies and improve the asset management of the Metrolink network.
At the start of our procurement process, we began with a number of key parameters and considerations:
• The technology and equipment had to be tried and tested. Although technology could be imported from another sector, it was important that we were not buying ‘an experiment’
• The chosen solution must be supported by a rigorous business case and cost-benefit analysis – no-one is going to spend GBP250 000 to solve a GBP20 000 problem
• It must be aligned with the KeolisAmey Metrolink pillars of safety, operational excellence, value for money, partnership, people and leadership and, importantly, ‘thinking like a customer’
• The solution must be easily integrated with the existing asset management software
• The implementation must be easy to manage and communicate to staff members
• What other benefits can be derived from RCM?
None of these requirements were in any way revolutionary and could all be achieved through the application of existing and readily-available technologies. The answers to these questions helped inform the business case and could do the same for authorities and operators of all types of rail infrastructure.
For example, our tram tyre measurement system supplied by SelectraVision utilises lasers to measure the wheel profile and back-to-back tyre distance. This identifies exceedances of two key values as defined by TfGM’s tram maintenance specification and is capable of unattended operation, virtually eliminating manual inspection.
Furthering the aim of improving tram ride quality, monitoring equipment supplied by Donfabs Consilia can be easily installed in the tram’s saloon, employing accelerometers to measure vertical, lateral and horizontal values of perturbations. Algorithms combine this data to produce a ride index in accordance with ISO 2631 (the International Standard related to the evaluation of vibrations and shock on the human body). With minimal installation, this equipment is easily transferred from one vehicle to another and uses GPS location to ‘map’ the ride quality across the network, removing empirical observation and replacing it with objective data that can better inform both vehicle and infrastructure interventions.
For track measurement, SelectraVision equipment is mounted onto the tram bogie to measure geometry, corrugation and rail wear to identify exceedances beyond TfGM’s set parameters. Lasers measure differing rail profiles – both Vignole and grooved – and high-resolution cameras capture video of the rail head. Again, unattended operation is key, and GPS allows for accurate data location.
A similar system from SelectraVision that uses lasers, light beams and high-resolution video mounted on the tram roof is being used to measure the height, stagger and contact wire thickness. These measurements are supported and synchronised with thermal imaging cameras at the pantograph, video of the pantograph contact wire interaction and panoramic video of the location.
Monitoring of in-service pantographs is one of the technologies that I am most excited about and sits above and beyond KAM’s contractual commitment. One of the most disruptive incidents on any tramway is when the overhead line is taken down, often by a damaged pantograph, so this technology is significant if it allows for detailed monitoring of the interaction at the contact wire.
There is no quick and easy fix and repairs can take many hours: services are heavily affected, passengers face significant disruption, valuable revenue is lost and there are also obvious safety concerns.
Our solution uses wayside equipment, Pantobot supplied by Camlin Rail, to provide a regular health check of in-service pantographs, using high-resolution imaging to identify damage, missing components and misalignment. If anomalies are detected, alerts are sent via email and/or SMS to maintenance staff. Detailed visualisation, accessed via a web portal, allows engineers to examine parameters such as section views of the pantograph and analysis of pitch and roll angles, and provide an informed response to the defect, thus preventing a potentially service-disrupting incident.
A functional, yet customisable, interface allows users to easily conduct various analyses (comparative, timescale, histogram etc) with live data, with reports generated either automatically, after extreme events or ad hoc based upon specific templates. Custom alerts can be set when sensors pass set thresholds, and these thresholds can be modified easily.
Taken together, the monitoring and measurement of tram wheels, track, OLE, pantograph and tram ride provide ‘a 3D view’ of the essential interfaces. These are all synchronised by location allowing invaluable insights that were previously unavailable – for example, how track condition affects vehicle ride or the performance of the pantograph.
Aligning these systems to the Metrolink ‘pillars’, we see immediate advantages:
• Zero Harm. We estimate that over 5000 hours/year of track inspections can be saved. This reduction in manual and repetitive tasks may also diminish instances of human error and offer further protection against derailment due to more regular measurement.
• Operational Excellence. Prediction of asset degradation optimises availability and reliability by identifying potential OLE incidents before they occur. This improves overall system resilience.
• Thinking like a Customer. All of the above should result in significant reductions in unplanned disruption.
• People and Leadership. Through technology, KAM is moving from being a ‘reactive maintainer’ to an ‘intelligent operator’. Staff are focused on interpreting data and developing future skills and capability to understand and apply that data to the smooth and cost-effective running of the network.
• Value for Money. The productivity benefits are perhaps the most striking, with an estimated 7000 hours saved in measuring tram tyre profiles, OLE and track equipment.
• Partnership. Technology is an enabler, informing better asset investment choices by TfGM and offering opportunities for improved design of future routes.
Beyond the solutions outlined here, we have already identified a number of opportunities where RCM and data analysis could offer real benefits to both Metrolink and other light rail systems. These include pantograph impacts; substation monitoring (for example, circuit breaker operation); elevator and escalator operation; rail and contract wire temperature – to avoid icing or heat-related damage; weather monitoring and remote CCTV/OTMR downloads.
Work is currently underway to develop these opportunities and investigation has already begun into the remote download of tram fault data to provide instant alerts, and remote access to the full detail of tram defects to drive reliability improvement. These are certainly exciting developments and just the start of a journey which will provide our engineers with better information and TfGM and its customers with a more resilient tramway. It is also one which other tramways would do well to follow.
Chris Stinchcombe, Engineering Director at KeolisAmey Metrolink, says “We are very fortunate to have a client with such a commitment to investing in industry-leading practices, and through a collaborative effort in researching available technologies, have now agreed the best solutions to apply to the Metrolink network. This enables us to improve performance for our customers, and create value for money by informing the best renewals and enhancements into the future.”
To create a truly ‘intelligent asset’ network it is necessary to combine RCM data with other sources and convert it into actionable insights.
Olga Evstafyeva explains how she is working with Clive to explore how to maximise the value from the data they are collecting at Metrolink:
There are a number of steps in the journey towards intelligent infrastructure. First the data (both real-time and historic) needs to be aggregated and collated in one place, then it needs to be processed and analysed using appropriate algorithms. Finally, by applying suitable detection flags it can be used to pre-emptively identify a change in condition and recommend intervention. This will drive a step-change in maintenance cycles. We don’t envisage technology replacing expert engineering knowledge, purely allowing such skilled resource to concentrate on interpreting data and implementing decisions more rapidly based upon it.
Integrating data sources to allow real-time or automated reporting will also allow valuable engineering staff to focus on value-added tasks, rather than data gathering activities. Technical teams will still be co-located close to the assets in question, with a central analytics team crunching the ‘Big Data’ to provide engineers with the most up-to-date and accurate information. This provides a system-wide view and a ‘single source of truth’ to better inform decision-making and investment choices.
To enable this transition to intelligent infrastructure, our team developed a Big Data analytics solution called Mercury – a sensor-agnostic IoT platform for remote condition monitoring and enhanced maintenance of a wide range of smart assets. The platform has been deployed on the Forth Road Bridge, where it is used to generate real-time and historical insights on asset performance, allowing engineers to make robust, data-driven decisions. Its state-of-the-art machine learning and smart alarm capabilities provide an extra layer of reassurance, alerting engineers to potential problems before they become major problems.
For Transport Scotland, Mercury provides assurance that the bridge, a complex asset with over 300 sensors, is safe and fit for purpose. By aggregating all of the data in one place, analysing it in real-time, and producing automatic reports that can be distributed to the engineering team and stakeholders, Mercury allowed the FRB team to significantly streamline engineering decisions and processes.
For example, following Storm Ali on 19 September 2018, engineers were able to evaluate in real-time how the bridge was performing and ensure it was fit for purpose – confirming that the storm had not damaged key components. After the event an automatic report was generated, highlighting how key bearings performed before, during and after the storm. This information was then easily distributed across all FRB teams.
This example of breaking down information silos is easily transferable. Metro and light rail systems already generate vast amounts of useful data and Metrolink will soon have detailed asset condition information that can drive significant improvements. What we often find with our clients is that the information is not always used to the best advantage, data comes in through disparate systems, in different formats and at different intervals. For example, in a tramway this could be from onboard the trams, sensors, control systems and from the supervisory control and data acquisition systems (SCADA).
Through this approach we have been able to transition our clients from a ‘react and remedy’ approach to a far more effective ‘predict and prevent’ model. Being able to interrogate their asset base at any time, and from anywhere, allows for greater collaboration between teams, contributes to reducing risks, and improves the ability to make data-driven and timely decisions. For light rail systems, having access to these insights could provide a greater understanding of asset condition, degradation and the relationship to the drivers of performance such as delays, safety and cost. Ultimately this will provide a better tram system for the customer.
Data is useless unless it allows us to make better decisions. Capitalising on modern capabilities in sensing technologies, cloud computing and programming is what allows us to join all of this together in platforms like Mercury.
Looking wider, with the increasing availability of intelligent tools – such as drills that record the location, angle and depth that the bit travels, to wrenches that record the torque, time and place of their use – we can now create a full audit trail of work being done, improving safety, enabling regulatory assurance and demonstrating long-term value for money.
Combining the data on one platform allows us to easily analyse and model the data to, for example, enable engineers to identify potential issues early on, allowing advice on operational interventions to be provided to tram drivers and control room staff. It also enables trend analysis which may indicate emerging issues with individual components, or highlight inconsistencies due to unforeseen effects of maintenance interventions.
Similarly, having all this data for interrogation in one place allows for modelling of extreme events by creating a digital twin of the asset. If it is detected that a particular weather event is responsible for certain asset failures, analysis engines can replicate the effects of similar occurrences going forward in a safe environment. Linking in future weather forecasts, engineers can predict ‘hot spots’, providing assurance during and after major events and using predictive models to inform operational and maintenance decisions.
Big data is no longer a buzzword. It is becoming part of the ‘business as usual’ in transport and is already a major part of our everyday lives, from weather forecasts to advice on travel modes, traffic conditions and the prediction of arrival times. For light rail, the access, analysis and the use of data will result in more efficient, reliable, and sustainable systems contributing to making light rail more affordable and supporting the case for future investment.
This article is adapted from a presentation given at the 2019 EU Light Rail congress in Brussels.
Article appeared originally in TAUT 985 (January 2020).