Paradigm shift: condition based maintenance solutions at InnoTrans 2018
Carrying out maintenance tasks only when there’s a requirement for – instead of when specified by time-based cycle-models: this becomes possible by harnessing the potential of digitalisation. At this year’s InnoTrans in Berlin, Frauscher will demonstrate how.
The company uses its distributed sensor solution: The Frauscher Tracking Solutions FTS. They enable a continuous monitoring of the wheel-rail interaction. This allows for insights into the changes of the condition of various infrastructure components whenever a train passes them. This includes rails, rail fastenings, sleepers or the track bed. In addition, defects on rolling stock-components, such as wheels, can also be detected.
Learning by doing
“Long story short: it’s going to change a lot”, Martin Rosenberger, CTO Frauscher Sensor Technology, summarises the evolution, which has happened throughout the last years. “When we started using Distributed Acoustic Sensing (DAS) in the railway industry, we were immediately aware of the high potential of this technology. Based on practical experience, we now know, there’s even more potential than we initially thought.” Today, the DAS-based Frauscher Tracking Solutions FTS are in use in numerous installations. Depending on each operator’s requirements, various applications are tested. “In the area of condition monitoring, we initially focused on acute rail defects and environmental influences – such as rock falls, mud slides and so on. During that phase we saw that the system detected a wide range of other events – as well as those, we were interested in”, Rosenberger remembers. “But instead of just filtering these influences, we decided to have a closer look at them.”
From a good idea – to a greater one
The local inspection at positions that generated conspicuous signatures revealed a broad range of root causes. “This confirms our theory that it is possible to detect more than just single events via DAS. Being able to consider such a high range of sources revealed an immense potential: As disparate as the appearance of track defects can be, the development history is equally as varied and diverse. And that’s what we have to monitor – with the ultimate goal of identifying defects as soon as they appear. Then it is possible to observe their deterioration and apply analysis, generate information and send reports. On that basis, maintenance tasks can be better planned - with foresight and efficiency.”
Not too much and not too little
However, just calibrating the system with an intensity that enables a detection of even the smallest defects is not the entire solution, as numerous positions would be indicated which are not relevant for maintenance teams at all. “This is where it becomes more complex”, Rosenberger explains. “In a first step, we must store the basic status of the infrastructure in the FTS: bridges, rail joints as well as switches – as all of these generate signatures, which are similar to those generated by rail defects. Ambient noises, for example caused by pumps or generators, can be filtered. At the moment we do this when installing and commissioning the system, but in parallel we are working on tools to automate this process.”
One glance is enough
Against the backdrop of this defined basic status, the operator receives periodic reports or warnings and alarms via defined interfaces. Therefore, the system determines indicators from the signatures it detects. Based on this, trend analysis are formed for individual, ten-meter long infrastructure segments. Notifications can now be created and transmitted using defined, user-configured threshold values. These indicate the type of change as well as the location. To present the information in visual form, Frauscher provides a graphical user interface with a map view and diagrams.
A system that learns
However, detecting changes alone is not enough. One and the same technical anomaly can in fact produce indicators that vary in frequency and intensity depending on whether the train passing over the section of track in question is a long, heavy freight train or a short, lighter passenger train. The DAS systems that are currently available are not yet able to classify the changes detected with sufficient accuracy. Maintenance personnel therefore evaluate items that stand out during the development and installation phase. A visual inspection is often enough to identify the cause of the warning, classify the change and take the necessary action within a defined period. Thereby, the system gets appropriate feedback and is continuously learning.
“It is a joint process”
It is very important that this appraisal and classification is understood and grasped as a fixed part of the maintenance process. “One key step is the definition of threshold values as well as the determination of the sensitivity of the system. Experienced specialists, who are processing notifications can thereby co-design the system according to their actual requirements. Even in times of increasing digitalisation, the knowhow of these employees will form an indispensable basis in the railway industry. Systems harnessing condition based maintenance (CBM) strategies must be developed in a close cooperation between manufacturers and operators and be calibrated individually. That way they’ll become supportive tools which can increase the efficiency whilst decreasing running costs”, Martin Rosenberger emphasizes the importance of a cooperation of operators, service contractors and system manufacturers.