Skip to Content

The journey to better Rail infrastructure management passes via digitalization

Capgemini Engineering
2 Dec 2022
capgemini-engineering

Digitalization of asset monitoring and processes key for Artificial Intelligence

As reported by the International Energy Agency in 2019, and Statista in 2022, the global demand for urban and mainline rail will grow in the coming years due to urbanization, attention on environmental impact, and an increasing population. This growth will require adjusted maintenance levels, to ensure an efficient lifecycle for incident management and improve the regularity of traffic flows.

The goal for railway organizations is to manage their infrastructure better; to analyze it regularly and on a large scale in certain territories, to detect any technical failures. The main objective is to ensure appropriate maintenance (to determine what & why), at the right place (where), at the right moment (when) and with the right people (who).

To achieve this, railway organizations are looking into other industries that previously made a digital transition, and into new disruptive solutions, such as Predictive Maintenance (PdM). PdM is a type of condition-based maintenance that monitors the condition of assets remotely through sensor devices. These sensor devices supply data in real-time, which is then used to predict when the asset will require maintenance and thus prevent equipment failure. However, implementing PdM can be a long process, which requires understanding what each maintenance type offers, which solutions can be built for which benefits and, finally, the essential steps needed to pave the way to success.

Why do we need Predictive Maintenance?

Corrective maintenance: unavoidable, but sometimes inefficient

Maintenance applied to railway assets is mainly corrective. This is due to the nature of the network, which is concerned with physical systems, whether they are dedicated to production or supervision. However, the various origins and purposes of these assets means Corrective Maintenance (CM) puts a toll on organizations which invest in numerous competencies and specialties, either internally, by training its employees, or externally, through costly maintenance programs with manufacturers or third parties. This results in:

  • Unpredictability, in the long run. Moreover, the root causes of problems are not identified
  • Interruption to production, leading to downtime and unproductive employees
  • Shortened asset lifespan, by not taking care of equipment and only performing maintenance when components break

Although it requires very little in terms of labor, cost, and planning to implement, solely relying on CM can result in expensive and potentially dangerous long-term problems concerning production, employee safety, and environmental impact, among others. A better solution is to prevent the problem from occurring rather than dealing with its consequences. For this reason companies are shifting towards preventive or predictive maintenance.

Preventive maintenance: digitalizing the process

Preventive Maintenance (PM) is the proactive monitoring and maintenance of assets, such as adjustments, cleaning, and repairs, and is based around either a calendar, with a recurring and time-based maintenance schedule, or around usage. Most industries work with PM plans for their production lines and/or operations, as they want to keep the number of breakdowns to a minimum.

Implementing PM could be challenging, though, as it implies:

  • Upfront costs and ongoing investments, because solutions are dedicated, and must be fed with reliable data from different sources
  • Additional labor, hiring staff members skilled in asset maintenance
  • Possibility of over-maintenance due to too much preventive work

To ease the implementation of PM, companies can invest in digital solutions like Enterprise Asset Management (EAM), Computerized Maintenance Management Systems (CMMS) or Asset Operation Management (AOM). These types of software help track, organize, and document the processes, streamlining them to keep the need for extra labor to a minimum.

The success of implementing PM rests on an organization’s ability to make its first step into the digital world, by modernizing the way it deals with its assets’ maintenance lifecycle, and the data sources available or that need to be upgraded. But, as PM is based around “experience” and “routines”, it carries the risk of doing too much – or too little – maintenance on an organization’s infrastructure.

Predictive maintenance: just in time, just enough

PdM addresses the issues raised by CM and PM, by scheduling maintenance when specific conditions are met, and of course, before the asset breaks down. By establishing nominal work conditions for assets, PdMcan compare the actual infrastructure state to baselines, thus detecting any drift in real-time, and then plan targeted maintenance to address the issue. In the next chapter we will dig deeper into this topic and define the essential steps to unlock its potential.

How to unlock Predictive Maintenance

Connect my Things to the Internet

The first step is to be able to collect useful data: a PdM program relies on the ability of systems to acquire historical data and to use live data to analyze failure patterns. But legacy systems in industries such as Railway often lack the means to communicate information to determine their health state. This is where interoperability and Internet of Things (IoT) technologies shine. The former helps systems to communicate altogether, using a common language thanks to standards and Service Oriented Architectures like OPC UA (Open Platform Communications Unified Architecture) . The latter helps by filling the time gap between old and new sensors or monitoring systems, adding a component at the infrastructure’s edge which can supervise or serve as a gateway for assets.

Together, IoT and interoperability systems empower the connectivity of our global architecture and allow data to be gathered seamlessly.

Data: The Good, the Bad and the Big

Information is at the core of maintenance processes. The better information we have, the better the maintenance will be. Digital solutions must improve the way we gather, aggregate, classify, and validate our data, to:

  • Identify the data that indicates the system’s health state, to minimize the overall amount of information needed for the business. More is not always merrier!
  • Verify the data in terms of security, integrity, and validity. It is important to be able to always rely on the correct information, avoiding corruption from communication or even hacking, or false positive effects.
  • Store the data in a Single Source of Truth (SSOT). It is used to ensure every business within an organization relies on the same data all the time.
  • Transform the data into a useable asset for maintenance, by using diagnosis, for example, considering what’s happening and what to do now, and by and prognosis methods which ask what will happened in a near future.

Therefore, we use the word “data-driven” for architectures that can do this. An interesting solution would be the use of digital twins: virtual replicas of our physical assets, able to aggregate all information and display the most relevant to users. One must be careful with this solution though, as it implies some risks that must be mitigated.

Decisions, decisions …

PdM relies on business intelligence from the maintenance department workforce. Field experience is tantamount to a database of trained analysis models and is the primary source of information to unlock the potential of PdM. To do so, we must digitalize the whole maintenance process, from fault detection at the beginning to the return on experience at the end) and aggregate this intelligence into our solutions. PdM programs should change the workflow of the maintenance department: not only to add new technologies or tools, but to provide a different approach on how to maintain your assets with the business as the core of the strategy. Change management must highlight the vision and the legitimacy of such a transformation throughout the whole organization to be successful.

An example of a large and extended infrastructure

One of our railway clients is conducting an extensive transformation program, co-constructed with Capgemini. The first phase is to improve its legacy supervision capabilities and use the data it has to hand to improve its maintenance and intervention process. The client’s goals are to control the infrastructure lifecycle and to control and facilitate maintenance processes. To do so, it developed a unique application capable of interoperating with all the sub-systems in charge of controlling the infrastructure.

This application lets the business have a real-time view of the infrastructure, mapped geographically. It offers a streamlined intervention process between its supervision centers and field workers. This application is an excellent example of a first step taken by connecting all systems, using and generating maintenance data, and helping people make decisions in-field, even if the app is not a PdM solution.

Conclusion

The most important aspects of a PdM program are patience and commitment. Many programs fail because of a lack of vision, experience, or methodology for applying PdM principles within an organization. It is a steep path that can be difficult to climb without the right resources or the knowledge. Therefore, many organizations are inclined to outsource PdM programs, either to develop this activity more quickly or to benefit from leveraging the cross-sector activities the service provider can bring. Maintenance is a subject addressed across all industries, and we are seeing common ground everywhere. These include new data sources, the ability to access our assets remotely, having a 360° view of our infrastructure, and IoT as Supervision. Digital technologies are key to these different use cases and must be as standard and as interoperable as possible to better meet the challenges faced by industries today.

Further reading:

Sources:

Author

Eric Jarde

Supervision and Monitoring Rail Offer leader
‘’Data-centric and interoperability strategies are increasingly key for Rail Supervision Systems.’’