RailTech Europe 2017 – Day Three review
The third and final day of RailTech Europe 2017 was based on Predictive Maintenance of Rail Infrastructure, and its increasing importance in delivering efficient working practices, cost savings and enhanced safety. Launching proceedings was Marc Antoni, Director of the Rail System Department and Rail System Forum at the International Union of Railways (UIC).
“Solutions are at the end, but first you need to know – what is your real dream?,” he asked. “It is to have a safe and secure railway. We have to remember that the railway is a ‘system of systems’ – with signalling at its heart, and a network too. It is a real challenge to adapt to the technologies. Systems are not the same in Germany, The Netherlands, Switzerland, Russia, and we have to think about this.”
So what, he added, were the common targets? A key aim over the next four years is achieving a reduction of 30 per cent cost of operation; doubling the traffic capacity on an existing network or tracks at the same operational costs; no possibility to use massive modernisation or renewal without acceptable founding rate; and the chance to use ‘digital improvement’ to create the ‘move’.
“Moving digital isn’t a technical issue but a question of skill within the human process – the real difficulty is the change management, and we need to be clever,” said Antoni.
Reaching these goals will require an adapted asset management policy for the entire networks, in which railway’s objectives are shared. “You must think global, what will be the consequences?” he added. In practice, asset management should always look to develop specific methods and tools for the lowest whole life cycle; develop specification and procurement methods; and to ‘use what we already have’.
“We have to think about things in advance,” said Antoni. “If we haven’t given any thought, we will have to pay more to do the same. The battle is won in the early stage of the design. The impact of new design and renewal is huge, and if we have the wrong design, the costs are very high.” Digitalisation, he added, was providing great possibilities but it ‘must be used in the right way’.
There were four steps for modelling and operational asset management: tools for LCC calculation; tools for estimation of maintenance needs of the assets; work of the deterioration and failure laws of each infrastructure, and a database describing the population.
“This gives us our asset strategy,” he said. “Today we have more and more information, more computerised systems, which help us to define our strategy.”
Signalling was ‘a little different’, requiring another type of interface. “Signalling is at the heart of the railway system, and design choices are key for maintenance, safety and security.”
Antoni also emphasised the importance of considering security and safety as being interlinked “Security is safety and safety is security,” he said. “When we think about a network using what is on the shelf, we have to think resilience of the whole system. For example, in Russia, China and Japan you always have a fallback system. It is mandatory there. Security has to be thought about when you talk about safety. This is not working in silos. Resilience needs to be considered from the railway system’s point of view.”
He concluded by posing the question that all stakeholders have to ask themselves – what is acceptable?
“A miracle is never coming alone – it needs to be facilitated. If we don’t think of the future, we will pay for it. It will be very expensive, and maybe not just in money but also in terms of human life.”
Paul Carstens & Adrie Koster, Programme Director and Procurement Manager Maintenance respectively at Dutch rail infrastructure manager ProRail then delivered a joint overview of their work on Performance-based Maintenance (PBM).
Carstens first highlighted some of the key characteristics of the Dutch rail market, which carries 1.1 million passengers on one of the densest networks in the world. “In order to continue this impressive performance, we invest more than 1.64 billion Euros in rail construction and maintenance, and of those 314 million Euros in small scale maintenance,” he said.
“Over about 20 years, the costs for small scale maintenance increased to an unsustainable level, and the nature of the old maintenance contracts contained insufficient incentives for optimisation and innovation, as we focused on prescribed activities rather than processes,” said Carstens.
In utilising PBM, ProRail is striving for safety and better performance, compliance with EU legislation, and a better price-performance ratio. “We want to stimulate sustainability and innovation. We have to predictive and we need to be consistent.”
Compliance pressure, increased costs and lagging maintenance performance requires a new approach with regard to small-scale maintenance, and so PBM has been introduced to tackle this.
All PBM contracts covering the country’s 21 operating regions will have been publicly tendered by the end of 2019. Adrie Koster expanded on this: “We agree on quality and performance levels, and this is different to the previous activity-based approach. We give a lot of freedom to the contractors to organise the contract that works best for them.
“The real stuff is preventive maintenance – this is what we want to achieve, to keep looking for options to avoid failure. The core of the PBM is a mechanism that includes fault analysis – whatever the contractor can think of to avoid failures.”
The result has been that PBM regions have a lower yearly average in train hindrance and deviations affecting train services than pre-PBM regions. Failures were significantly more with the older contractual arrangements.
“The question is – is this an easy process? It is not,” said Koster. “We have had all kinds of struggles. But there has been one critical success factor, and that is how the contractors are working together. New rules changed the relationship, but a good cooperation and understanding means it is much better than it was in the past.
“PBM is not just about contracting,” added Koster. “The contracting and tendering is a tool, but you need to keep up good relationships. We also need to think about our own organisations. We have set up a great training programme, together with the contractors, to ensure everyone knows what the intentions of the contracting arrangements are. It is much easier if you are on the same level.”
He concluded: “We are looking at our internal cultures. It is really a broad range of activities to make sure we have a better performance on the digital railway.”
Maurice Feijen, Senior Service Architect at Thales Transportation Systems, shared some of their work on creating a railway where ‘asset failures are not service affecting and all maintenance interventions are planned’.
“We do this by making the railway infrastructure smart,” he said. This meant capability for managing assets such as point machines and axle counters, and then connecting it with centralised IT solutions.
On the remote condition monitoring of network assets, sometimes it can be ‘too late’, so the aim is to predict when the asset is failing, he said.
“From insight to prediction,” said Feijen. “Using the algorithms, you can do some future prediction. If you put this in one holistic picture, we gather all of the data, including from other databases, such as weather. We include this in a cloud, we filter the information and then we have a database.
“The kind of capabilities, the skills we need, are different to what we need for engineering systems. But for this service solution, we need software engineering and experts on statistics. Ideally you would like a person who has all the skills – this is the mythical unicorn.
“We believe it is more efficient and more effective to provide this as a service. The customer always remains the owner of the point machines. We will do the analysis and we provide the results back. This is done in a secure way, through the cloud. Another way is perhaps for some customers who are maybe hesitant using cloud services, they may have the key.”
An additional advantage of this, he explained, was that it also meant Thales could keep improving on the algorithms, and introduce new products to replace old ones.
“What is important to realise is that this is not ‘future stuff’ – it is happening today. We invest in these technologies as it is a way to improve efficiency and reduce costs for maintenance engineers,” he added.
Professor Rolf Dollevoet, from the Technical University of Delft in The Netherlands talked about his ideas for what he described as ‘the Maintenance of the Future’.
He began with a theme voiced by other Dutch speakers: “We have an ambition here in Holland – everything must be done better but with 20 per cent less budget. So we have to change and think different and act different. Innovation is not done alone, we do it as a team, it is done with partners. So, we need a system jump. My experience is that innovation and new things takes time, maybe five or six years, but it will come.”
He briefly outlined the student and course composition at the TU Delft, explaining that it was offering opportunities for rail careers based on the premise of the sharing of information between stakeholders. “You need knowledge…you need education. If you look at Holland, is there education about the railway business? There is but there are not many courses and it is not that well organised,” said Dollevoet.
These include people from small businesses in the rail sector who are attracted to courses such as the MSc annotation Rail. “I think this is really important. In Delft it is possible that students can study now for a Masters in railway engineering, and that is huge. We now have 11 courses, so you can really choose for the rail business.
“We also have the professional education. It is possible now to give education to people in the business. (As people) we never have time because we are busy with the operational business. What I see is people from the smaller companies are really willing to go for this,” he said. “I think they feel the pain of a lack of tactical information, and a system approach is what you need to get innovation together.”
Dollevoet then outlined his visions for what he called the ‘the real world: “I have two dreams – predictive monitoring assets ‘on board’. I think everything must be on board. Every day, we have a train every six minutes, so when can we do the maintenance? Why not do this monitoring on board, and also the maintenance of the track on board? When you look at predictive monitoring, we have a couple of ways. We have remote sensing, for example on the Betuweroute freight line we can see within millimetres if it is sinking or moving, and how it moves.
“The data is free because the Government in Holland pays for it until 2025. We can also zoom in on the axle box acceleration, and this is another way to do asset management. So, an integral system approach is necessary.”
He likened the current process of how failing assets are dealt with as akin to a ‘funeral’. He said it was clear that the damaged asset had ‘passed away’, and that we were now only ‘planning the funeral’ rather than visiting the doctor.
“I think that is the message,” he added. “We only look at the problems now, and how to solve them now. That’s what we have to do. I cannot do it alone, ProRail cannot do it alone, the university cannot do it alone – we can only do this together. It is a system approach and then we take these steps.”
“The ideas are there but who is doing this? We have contracts, we have to be compliant and it is not easy. But you can also ‘do something’. Do a pilot and then everyone sees how good it is and people will ask for it, and then you have your accelerator for innovation.”
A different perspective to the day was offered by Mantas Vaskela, the CEO of drone-piloting business Laserpas, with his ideas on the commercial use of drones for data-driven maintenance. In his presentation, The Jungle of Infrastructure, he highlighted the potential for drones in assisting infrastructure managers with inspecting rail tracks and surrounding areas.
One example is that it can enable the monitoring of trees’ condition – and the likelihood of them falling – close to the rail line, as well as ensuring any encroaching vegetation is kept in check with regular trimming.
All risk factors, said Vaskela, such as trees, buildings and erosion are matched against infrastructure, while dangerous areas are marked. “The biggest problem with drones is that people want to sell you drones, but when you ask them if they can operate for me, they say ‘no, you do it yourself’.
“We are specifically focused on bringing solutions to the railways,” he said. “We are trying to learn what we need do for our customers. You will most likely be able to see ground erosion from 5-10 metres away, so we are trying to maximise our technology to give you some answers. Railways are very safe but require a lot of specialised training and a lot of time to learn about,” he added.
Maarten Zanen, Business Leader Asset Analytics at global design consultancy Arcadis, focused on how the railway sector could make best use of analytics as a driver for data-driven maintenance. He used sport as an example of how better results were being achieved through data-driven decisions, namely that of Dutch F1 racing driver Max Verstappen.
“Max Verstappen had the right genes to start with but he also had a great team which was analysing all of the elements,” said Zanen. “In doing so, they invented a new aspect into F1, and that was how to save your tyre. If this is the case for sports, how is it with asset management?
“Assets have a vast amount of information, and that is because sensors are getting cheaper, algorithms are getting better, and there is more and more open data. Government departments are sharing their data, and the more we do this, the more we can find the relevant connections.”
Key to enabling this, he added, was the ‘capturing of knowledge’: “You can use asset analytics as a way to minimise disruption, but also to optimise your maintenance projects. For an asset manager, it’s good to have an understanding of the conditions. if you extend that timelines to 30 years, you are able to see the condition and see what is the required efforts in terms of activity and money.
Zanen then introduced a problem commonly suffered by railway undertakings – that of ‘slippery lines’ caused by fallen leaves. “We always start with the business question,” he said. “What are the factors that might influence this? This is typically between the wheel and the rail, but also external factors are taken into consideration. We use data from tracks and trains but also externally, from the weather.”
For the business, the problem causes delays, leads to train cancellations, results in potentially hazardous situations and creates the need for increased maintenance.
“How can we reduce the delays as much as possible? How can we reduce the wear on wheels and rails,” he asked.
By efficiently reusing existing data such as slippery tracks records, the data from numerous weather stations and positioning of numbered trees along the line, the issue could be tackled. They could also deduce a clear relation between windy conditions and slippery tracks, with the highest hourly wind gust being the most ‘significant parameter’.
A predictive model using a day-ahead approach could be created, delivering insights into which factors have the most contribution to slippery tracks, and this can ultimately be utilised for preventive and mitigating measures.
“This shows the power of analytics,” said Zanen. “We all know that data has its limitations. It’s good to know that data is never 100 per cent, so the more sets you combine, the better you can handle with limited quality and reach the outcomes you need.”
The afternoon session began with Fred Beilhack, Managing Director of construction firm Swietelsky, who looked at ‘maintenance with the trains running’. The main considerations for railway companies, he said, were four-fold: maximum track availability for the customer; minimum locking times for the railway’s operation; maximising safety for track workers; and intelligent planning for implementation.
Highlighting the ‘enormous’ growth of rail freight in Europe, he said nearly 3.1 billion tonnes went through European seaports Europe every year, about a quarter of which is in the Netherlands. In addition, passenger transport – in stark contrast to the USA and Australia – is very important to Europe, which is why railways require short locking times and track availability around the clock.
Safety for track workers, he said, required a complete track lock; working with fixed barriers; working with automatic warning system; while maintaining the possibility for work on the track
Possible ‘intelligent projects’ include a complete track lock for ballast cleaning and change of rails, which he claimed could save ‘nearly 40 per cent’ of time over traditional methods. With a fixed barrier it is possible to work on the other side while existing services continue to run.
The theme of track workers’ safety continued in the next presentation, a joint venture by Inge Thys of Belgian rail infrastructure manager Infrabel, and Lex van der Poel of railway safety innovators Dual Inventive.
“Maintaining infrastructure needs to be done, and workers need to be protected to do their job”, said Thys, who then outlined Infrabel’s five priorities: safety first; trains on time; a network for all the trains of tomorrow; a financially sound company; and being ‘in tune with society’.
“Safety is first, but you cannot make sure that our well-being is covered if you don’t do something about operational safety – only two years ago these concepts were completely separate,” she said. “We also need to do maintenance but not at the cost that trains aren’t running.”
Being in tune with society, added Thys, meant being sympathetic towards the needs of people living near the railways in terms of noise and dust, and also everyone who uses public roads: “Accidents mean interventions that you cannot make the best plans for,” she said.
The industry was now facing key challenges around staffing and the need to make the best use of experience which has been built up by long-time employees.
“Safety first – it is our logo and our motive,” she said. “But we have to practice it and that is not so easy. We always say: ‘we have to keep safe but the trains have to keep running’. How do we keep the right balance between those two? We have an increasing number of passenger trains, an increasing number of freight trains. The more maintenance we need, the more interaction we need from our people on the track, when one or more trains are running – so how do we do that? We do maintenance between two trains, we do maintenance when the line is out of service. How do we handle this?
“Another concern is that budgets are more and more limited, which does not give us the possibility to invest every time in infrastructure. That means, if we cannot invest in more infrastructure, we need to do more maintenance.
“Another issue that we have is that our people are old. Approximately 30 per cent are retiring within these five years. We have to replace them but it is very difficult to find workers with the right experience, the right skills. But we don’t have a policy within Infrabel to engage people before someone is leaving. That means experience is leaving – we have do something about it, because if experience is leaving, that means that security becomes an issue.”
These were the reasons why, she added, Infrabel was looking at new technology to help tackle these issues. She then highlighted particular cases where ‘technical solutions’ were being sought, such as the issue around safety guards.
“In Belgium, when people work on tracks, most of the time we use safety guards. What do these people do? Normally these are track workers, people who have the skills to maintain the track. What do they have to do? They have to watch in order to see if a train is coming, and if a train is coming they have to warn the team on the track that they must evacuate.
“But, what we always say, is this is a human being. He can be unwell, he can be distracted, he can make a fault. And if he does that, a train can enter into the team and you can hurt the people you should be protecting. It is not logical today, with all the technology that we have, to put that kind of responsibility on one person, that’s not normal and therefore we have to find a solution.
“It is also frustrating knowing that we have less and less people to do the job. The skilled people are leaving and you have to put a skilled person on the track only to watch. But he cannot do anything operational. He does not do maintenance – he is only watching.”
She then referred to the work being done with Dual Inventive, which enables them to guard the tracks while they are being worked on.
“Today, it’s our safety guard or traffic control who decides whether a train can pass by. But if traffic control decides to let the train pass by without making sure that the tracks are clear, we have an issue.
Infrabel is now utilising Dual Inventive’s ZKL3000 system, a self-monitoring line blockage system. It protects a section of track by creating and monitoring a short circuit in the track. By simulating a train in the section, it causes the track circuit to show ‘OCCUPIED’, thus activating the protection signals. The section is immediately blocked, enabling workers to perform work on the track safely
“ZKL helps us to avoid (the issue) because it takes control over the signals,” said Thys. “Unless the track workers and equipment are out of the way, the driver does not get a green sign. And that, for our people, is a real comfort, because now they can control their own safety. Before they were depending on the safety guard.”
Lex van der Poel then went on to explain how Dual Inventive was also employing the ‘Internet of Things’ (IoT) approach to helping Infrabel ensure trains were detected with an alarm system.
“Innovation is only innovation when it is implemented,” he said. “We are always looking for partnerships to bring to the market what is needed.”
“The Internet of Things, I know it sounds a little bit strange and some people are still afraid of it, but what we mean is to use the 3G, 4G and 5G communication protocols, (and) not just wirelessly of course. It is focused, you get a signal where you want to have it. And especially with 5G, you are talking delay times of only milliseconds. It brings all the capabilities you want for a fast response, especially for safety.
Third party technologies
“Also, you will not close off the technology to link to it – it is easy to link to third party technologies. If you are dealing with information from train drivers, you can bring it directly to people in the field. What we have built, together with our partners, we have all the focus to make the rail world more reliable, more sustainable and more robust for the future – IT can help you in many ways,” he said.
“We think about engagement. Engagement, in short term, you bring information to the people. If you do it with this technology, gives you real time capability to see where the things are broken. Its the same with IoT. That makes it a very flexible and responsive system for planned responses. That is what we are trying to bring into the market, that for us the future.”
Inge Thys rounded off their presentation by emphasising the importance for Infrabel of engaging with its employees as it introduces new technologies.
“It’s very important to us, because we have a generation of employees who have been here 30 or 35 years, and now all of a sudden we are going to change our way of protecting them. We have to convince them, and that is why ZKL will be used for everyone at Infrabel,” she said.
Further reflections on the ‘human factor’ came from Lynn Chamberlain-Clark, Change Lead at Network Rail. She shared some of challenges in taking staff ‘on the journey’ of technological innovations, particularly those rooted in working practices of the past.
“I make no bones about coming and talking somewhere where we talk lot about digitalisation and technology and all the opportunities that innovation can give to us, but I remind you that there are other people in this as well, and that you need to think about how you can work with them,” she said.
“When I had a look at what RailTech was trying to do, there was something about new and innovative solutions, there was something about making sure we can maintain the railway, even when trains are still running, and there was also something about predictive maintenance. For me, all of those involve people,” said Lynn.
“We are human beings and human beings will be involved as well as machines and technology, and we need to somehow help them see that the solutions that we are coming up with will really help them, and that they are better than their ‘man with the flag’.
“We believed that if we actually put in no trains around when people were working that they would be safer, and actually we haven’t found that. We have found that the shift has just happened, that the lack of safety is with the machines they are using, not the trains. So we need to have to look at how we can help people with that.”
She talked about how staff liked being ‘superheroes’, possessing the capacity to do things that needed to be done in a hurry and reacting to emergency situations.
“Therefore, to predict things, and plan things, doing things as a ‘slow burner’ is not what they want to do, so you have to work on that culture and make that more appealing,” she said.
Network Rail had had a problem whereby there was ‘no belief and no trust’ in senior managers, and that was part of a wider ‘blame culture and non-reporting culture’.
“But what does that have to do with technology and innovation?” said Lynn. “People didn’t want to trust what we were going to bring in because they didn’t believe we were going to look after them.”
Designing a safety culture within Network Rail centred on five elements: create the environment through leadership; a ‘bottom-up’ engagement policy, with staff encouraged to report issues and be accountable; middle management engagement, which is essentially about taking responsibility; systematic risk management, which means identifying the early barriers; and finally continuous improvement, whereby review is always ongoing.
Ultimately, said Lynn, people did not want, nor should they need to, ‘rely on the man with the flag’.
“People often believe they are being safe when they are, in fact, being lucky. People will work hard to do a job but they can cut corners. We need to challenge this with people, and then bring in a learning culture – that is fundamental.”
It was also vital, she said, for management – described at times as a ‘permafrost’ – to set the right example.
“A safety vision promotes the importance of safety leadership and behaviours – that is then authentic to the people,” she added.
Jeroen van Ravenzwaaij, Account Manager at building and design consultancy BREM, then brought proceedings back into the technical arena by looking at new permanent optical monitoring techniques of constructions. This includes innovations such as liquid-level system monitoring, a predictive tool which, in one example shown to delegates, was used to detect an imminent roof collapse, thereby enabling a safe evacuation.
Using fibre optic-based technology enables the development of sensors which could be utilised in the rail sector, such as static monitoring measurements of bridges and trains.
“You can build it in as part of your asset management, both before and after the work,” said van Ravenzwaaji. “It is theoretically proven, and you have control of the total construction.”
He highlighted several track-based projects utilising the technology, including sensors on the bottom of a TGV train, which sends the information to the desktop, enabling operatives to measure excess.
“More remote monitoring creates networks for the full asset management programme of constructions,” he concluded.
Johannes Neuhold of TU Graz, then moved things track-side to explain about his methodology for assessing track ballast condition within the context of a ‘maintenance versus total renewal approach’.
He explained the process of using fractal track geometry to ascertain the condition of the ballast. This is a mathematical process that splits up a measuring signal – here the signal of the track geometry – into different wavelength ranges. By analysing which wavelength range dominates the irregularities within the signal, it is possible to evaluate the reason for track geometry problems.
This, for example, this can indicate ‘fouled’, or clogged, ballast. Neuhold concluded that it was possible to describe the condition of ballast across the network through the fractal analysis approach, and that ballast cleaning as an additional maintenance was ‘economically reasonable’ – at least in the Austrian example presented – until about two-thirds of the track’s service life had been reached.
It was possible, he added, to develop a ballast strategy based on three key parameters – the age of the track, the fractal figure and strategic service life. Furthermore, ballast cleaning might be a ‘reasonable option’ for around 11 per cent of the analysed network, and that ballast cleaning is, and will become increasingly so, a more relevant topic for tracks with concrete sleepers.
“The problem is that ballast fouling implies negative consequences for the entire track construction – ballast cleaning is a very complex measure,” he said.
Rounding off the day was Andreas Seim, Research Manager at Sintef, one of Scandinavia’s largest independent research bodies. His presentation, the ‘Benefits of real-time condition monitoring of switches: an empirical study’, examined the effectiveness of real-time condition monitoring, and how it could be best deployed to help ensure train delays are kept to a minimum.
Effectively answering such questions requires new analytics based decision support that combines infrastructure data with train operations data, he said. “Such technology may facilitate more holistic and effective prioritization of infrastructure managers’ use of resources.”
Failure of switches is a major cause of delays in old railway infrastructures, and so the Norwegian Infrastructure Manager Jernbaneverket deployed condition monitoring equipment on selected switches during 2015 and 2016 to enable smarter maintenance practices and reduce failure rates.
New maintenance processes and routines have been developed and implemented to realize benefits from with this technology. To help design further rollout in 2017, Seim said they had set out measure and document realised benefits of the current installations and to predict benefits of condition monitoring on the remaining switches in priority areas.
“We have developed two algorithms that automatically identify and measure actual train delays caused by each failure based on detailed running times recorded in a traffic information management system,” he said.
Combined with data from a maintenance management system to identify switch failures, this capability allows for accurate measuring of effects of switch failures on train traffic. This formed the basis for measuring realised benefits of the implemented condition monitoring, and for predicting the value of implementing the same technology on other specific switches.