WHILE demand for transport is widely predicted to increase, the capability to expand networks and purchase additional fleets is more limited. Operators and manufacturers are instead looking for ways to increase the efficiency of existing fleets and networks.

Siemens’ answer to the need to improve efficiency and increase fleet and network availability is the Alliance of Availability, which was launched at InnoTrans in 2018 and now includes around 30 organisations.

“The idea with the Alliance for Availability is really to bring everyone together for one goal: the high availability of systems,” says Siemens Mobility CEO - customer service, Mr Johannes Emmelheinz.

He points to the airline industry’s Star Alliance as the most well-known example of how the Alliance works, with each member retaining their autonomy, identity, core competencies, regional areas, and fleets, “but they joined with the goal to have a better network globally to secure a smoother, more comfortable ride from A to B for passengers.”

Maintenance on the railway in the past has largely been undertaken through interval-based regimes, with vehicles taken out of service depending on distance travelled or time in service. Parts such as brake pads and filters were then changed regardless of how worn they were.

This is changing with the introduction of multiple onboard sensors that can monitor almost any component, providing maintenance teams with up-to-date data on the condition of the components.

Emmelheinz says the next step is predictive maintenance, combining data from numerous sensors to predict ahead of time when components are expected to fail. An example is the wear on wheels, which could be calculated by combining data from the train, the track, driver behaviour including the speed of acceleration and braking, and weather conditions including snow and ice, which is then processed by algorithms using artificial intelligence (AI).

“The idea with the Alliance for Availability is really to bring everyone together for one goal: the high availability of systems.”

Mr Johannes Emmelheinz, Siemens Mobility CEO - customer service

“If you then add this to the condition based [maintenance regime], you have a very, very smart way of dealing with maintenance and servicing,” Emmelheinz says.

The same is also being applied to infrastructure, with equipment such as point motors now able to detect when they may fail ahead of time. For example, Siemens is already monitoring platform screen doors, power supplies, signalling, lifts and escalators on the Singapore metro.

Siemens is now able to predict door failures on trains four weeks before they develop, allowing maintenance to take place during scheduled periods, reducing the time the vehicle spends in the depot, or, even more importantly, avoiding a critical fault during operation. The end goal for the alliance is to achieve this for all components and elements across the network.

Artificial intelligence

Fleet maintenance is carried out by vastly experienced staff who can identify issues and problems as they develop. But as systems become more complex, and other factors are considered, this increasingly requires more in-depth analyses with the help of AI.

“If you combine passenger occupancy, train driving behaviour, track condition, braking condition, the change in wear and tear of brake pads in summer and in winter, no brain is able to do this.”

Purchase a book on Amazon, and the AI-based algorithm will suggest 10 others that you may also want to buy, and you might buy a couple more. But making a correct prediction 20% of the time is not enough when it comes to maintenance systems. “We need a quality level of 99.9% of how we predict the future,” Emmelheinz says.

The AI system is used to “clean” the data, sorting false positives and other discrepancies from actual faults. For example, a bag getting stuck in a door may register as a fault, but would need to be identified as a false positive and filtered out from the alerts sent to the maintenance teams to avoid the unnecessary removal of the train from service.

“When you combine all the data in a meaningful way then you can do this prediction and really come up with modelling which describes your future,” Emmelheinz says. “Because it’s so complex, we believe not one company can do this all by themself. You need to have the different players with [expertise] based in their discipline, who know their part of the puzzle best, contributing together and making sure that this comes together with a clear goal to have the highest availability going forward.”

Since launching in 2018, the alliance has expanded to include sub-suppliers, operators, cyber security experts and universities.

“For me, the power is also that we have these different people or different parties as partners, and it’s not all the same companies doing something together,” Emmelheinz says. “It’s really people joining with different skill sets, focuses and experience.”


The Alliance is also looking to improve the connection between maintenance and operation, to optimise fleet availability in line with peak periods of demand. Ideally, the maintenance team can identify when maintenance is required and how long it will take to rectify, with the operation team then able to identify the optimal time to remove the vehicle from service. This includes being aware of special events such as sports matches or trade fairs, when passenger numbers are higher than usual.

Emmelheinz says this whole-system approach is required for the system to be truly effective. “You can optimise the brake system, the air condition, a switch point, but it doesn’t help you if the switch point is optimised but the balise is not working anymore, or the door is working nicely, but the brake system doesn’t work,” he says. “All these puzzle pieces need to come together, and that’s when it really gets complex.”

The Alliance makes use of two systems developed by Siemens: Railigent and MoBase. Railigent is Siemens’ open ecosystem designed to use algorithms and data analysis to provide predictive maintenance, while MoBase is a spare part management system designed to allow the easy identification of required spare parts through a photo, additive manufacturing solution and easy and quick online order and delivery.

Developing system-wide predictive maintenance is easier for metros than for mainline railways, due to the more complex nature of the latter where multiple operators and various fleets tend to operate over the same infrastructure. This poses challenges in gathering data from the locomotives built by different manufacturers, which may not be involved in creating the predictive maintenance system.

“When you deal with one operator, he doesn’t want to have five different systems, he wants to have one system,” Emmelheinz says. “Normally it is driven by the operators, the end customer. Singapore has decided to have one system for the entire Singapore transport system, and they support us because we are delivering the system and we ask Bombardier to provide the data which are needed. If your customer tells you that you need to deliver this data so that you get your next order, then you’re going to do this.”

“That’s why it’s important that you have your architecture, your structure right, and the governance and the protection of the system is right.”

Johannes Emmelheinz

This can pose challenges when competitors join the Alliance, with processes required to ensure that the correct data is shared with the right people, without oversharing confidential information.

“It’s really how you structure the entire setup, how you deal with confidential information, how you deal with cyber security, because what we don’t want is someone accessing such a system and doing something very bad to a metro system,” Emmelheinz says. “That’s why it’s important that you have your architecture, your structure right, and the governance and the protection of the system is right. It’s not like Google where you don’t know what is happening with your data.”

Emmelheinz stresses that the industry is looking to work together to continue to grow rail’s market share, instead of simply competing for the same slice of the pie. “The question is how can we jointly make more transport happen, and not fighting against each other,” he says.

An even bigger challenge is in the age of many fleets, with sensors only regularly installed on locomotives manufactured over the last five years.

As well as improving fleet reliability, improved collection and processing of data also improves the level of information provided to passengers.

“If you have this reliable information, then you also can communicate to the passengers in such a way where you’re saying ‘please use platform A instead of platform B,’ or use a different elevator, because the other one is already heavily occupied,” Emmelheinz says. “Or when you come to the platform, please move on to section B of the platform because the train will arrive in this section with empty cars.”

This has been the result on the Thameslink line in London, where data on passenger numbers has allowed the operator to provide information to allow social distancing.

“This was heavily appreciated, that we made sure that people were spread out over the platforms and in the trains,” Emmelheinz says. “People feel safe enough to ride the train… because we can provide this additional value with the additional information.

“For the passenger, I believe it’s a more convenient ride end to end. And we need to make sure that the system is technically delivering the availability of the infrastructure and the rolling stock to deliver this.”

The partners involved in Railigent are:

  • Knorr-Bremse
  • Sopra Steria
  • Boom
  • Igus
  • Witronic
  • Sensonic
  • Voestalpine Signalling
  • Ginkgo
  • Dellner
  • Hacon
  • Voith
  • M79
  • Strukton Rail
  • DMA
  • Uptime Engineering
  • Perpetuum
  • SKF
  • X-Plus
  • Televic Rail

The partners involved in MoBase are:

  • Kroschke Deutschland
  • Kroschke
  • Schwan
  • One4Business Solutions powered by Sonepar
  • Black Box
  • Huber & Suhner
  • SKF
  • ime Elektrotechnik
  • Bürklin
  • Hoffmann Qualitätswerkzeuge
  • TGS Spezialwerkzeuge
  • Hahn & Kolb Werkzeuge
  • Harting Deutschland
  • Hirschmann
  • CIR
  • Peiner Umformtechnik