Railway: Digital within analogue

Digitalisation is the dominant topic in the railway industry too. All available data must be used intelligently for this purpose. This is emphasised by Stefan Marschnig, Associate Professor at TU Graz.

Has the railway industry missed the digital revolution? This is the impression it gives sometimes with the constant talk of the “rise in digitalisation” in the field of science and the railway industry. However, the rail sector is anything but an analogue specialist field that has been left behind. Instead, digitalisation arrived decades ago – much earlier than in other sectors. For a long time now, railways have been controlling operations digitally, collecting large volumes of data and making databased decisions. The challenge today therefore does not lie in digitalisation itself, but in making use of its technical possibilities to convert raw data into valuable information.

Vast quantities of data are recorded in the railway sector, but only purposeful networking can produce information.Vast quantities of data are recorded in the railway sector, but only purposeful networking can produce information.

Do not confuse data with information

Data is commonly used once only to answer questions like “what?”, “where?” and “how often?”. The following examples, however, show how additional information can be generated by analysing existing data. This means that data can also provide answers to the question “why?”, thereby contributing to system optimisation through a more essential, targeted use.

The recording of incidents using measurement technology is without a doubt crucial to everything safety-related and to the railway. For example, unauthorised access to the tracks, external manipulation of systems and system components, or unsafe operating states. Combined real-time data is also an undisputed added value to the operational aspect, for instance when passing on accurate information to passengers. The technical progress of the rail system cannot however be achieved by merely collecting data or through data networking; what is needed instead is data analysis. Only then can we understand signs of wear and tear inherent to the system and optimise hardware and any associated processes.

Data with different levels of detail can complement each other.Networking different datasets: Data with different levels of detail can complement each other. With alternative evaluations – in this case, by means of fractal analysis – new information can be obtained from existing data.

Data analysis based on a substructure assessment

The Institute of Railway Engineering and Transport Economy of the Graz University of Technology has spent more than 15 years focusing on the description of the track superstructure behaviour. Since the quality of the track superstructure cannot be described based on data from a single measurement run – but its current condition can – work began on analysing the track position data of consecutive measurement runs and describing it using quality trends.

The data analysis does not take into consideration the linear expansion of a measurement signal with regard to a current condition, but instead its chronological sequence. In order to investigate the reasons for different changes in condition, the boundary conditions of the track being examined must be taken into account during the next step. Data from the test carriage of ÖBB-Infrastruktur AG is therefore combined with other datasets in this example. The condition trends can thereby be analysed dependent on the curvature, track superstructure, track load in total gross tonnes per day and the age of the system. Maintenance work that has been carried out, such as track tamping, rail grinding or cleaning of the ballast bed can also be taken into consideration together with the relevant date and processing time. This database (TUG-DB) allows detailed searches for the detailed analysis of the course of the track position to be carried out.

This approach, which can be read in detail in Dr Matthias Landgraf's dissertation, shows that data with different levels of detail can complement each other and that, depending on the objective of the statement, different data should be collected and evaluated.

Considering the change in signal within a temporal context provides a description of condition trends.Measurement data analysis in time series: Considering the change in signal within a temporal context provides a description of condition trends. Based on corresponding databases, it is then possible to generate a variety of information.

Data analysis – the outlook

In the next few years, additional research and application fields will emerge against the backdrop of ever-increasing and more cost-effective technical possibilities for data collection and storage. The networking of different datasets shown in the previous example can be enhanced by (at least) two additional essential aspects: the prompt recording of local errors and further specification of the load collective. The Institute of Railway Engineering and Transport Economy has already requested and started some research projects, which are dedicated to both of these aspects. As far as measurement technology is concerned, the focus is on four technical options:

  • Onboard Measuring (OBM): Simple and inexpensive measurement equipment can in the future be fitted onto standard trains (this is already happening in part today). Very consolidated data series, for example of accelerations, can promptly detect a sudden system failure. The detection of breakages of rail and frog parts or of rail joint failures is significantly improved by this trend.
  • Fixed measurement points (Wayside Train Monitoring Systems, WTMS): As already mentioned, the load collective is crucial to the damage mechanisms along the route. The carriage-specific presentation of this load collective already provides a much more detailed picture of the cumulative load. This data model is nevertheless based on major simplifications, since the vehicle forces are derived from the as-new condition of the vehicles. Recording interaction forces that actually arise at fixed measurement points involves an additional stage of detail since wear conditions on the vehicle (for example flat wheels) are received, which in turn significantly affect the forces.
  • Fibre optic measurement systems (Distributed Acoustic Sensing, DAS): The acoustic evaluation of light waves, which are transported in the trackside fibre optic cables, is a promising technology. Intelligent evaluations of these light waves could give infrastructure operators tremendous added value. On the one hand, this allows for safetyrelated aspects (unauthorised track access, protection of work crews) to be presented metrologically. On the other hand, the technical monitoring of the wheel-rail contact is at least theoretically feasible. This would mean that both trackside and on-board points of failure (for example broken rails and flat wheels) could be identified. If, however, evaluations like these are combined with already existing databases, this could provide answers to outstanding questions regarding the system reaction and ultimately the system configuration and coordination.
  • Local metrological arrangement of track/point components (smart assets): To fully understand which forces actually affect the infrastructure and to what extent, it may be useful to take local measurements, for instance on point components or individual sleepers. A continuous, network-wide arrangement of all assets will not be considered, not only for economic reasons, but also because having a large number of sensors installed would lead to an increased number of error messages.

Besides a basic understanding of the data generated, the correlations between the data are also of particular interest. If, for example, flat wheels are detected on WTMS, then these can be identified as DAS samples. Equally, force peaks would have to occur on local track components, and by analogy, on the vehicle, and thus be identified in the OBM. The DAS and OBM technologies could then use information again from single-point observations in linear extension. Research will show how this additional information can be used over time in terms of the quality behaviour of the route.

Fixed measurement point ARGOS of ÖBB-Infrastruktur AGFixed measurement point ARGOS of ÖBB-Infrastruktur AG

Future measures

With all of these technical possibilities, one question remains to be answered first: How much should be measured? The more data to be compiled, the higher the probability that
a. this data is not utilised further and
b. that data and information derived from it contradict each other.

The validity of measured data and therewith its reliability is of paramount importance to the infrastructure operator. When taking measures and making decisions, digital stops here – the measures are and remain analogue. Being able to monitor infrastructure systems as best as possible and based on data is one thing, but making decisions about measures and implementing them is an entirely different matter.

The derivation and implementation of measures is influenced by numerous boundary conditions, and not the system itself. Valid state-of-the-art technical evaluations hereby form an essential basis for decisionmaking, but cannot however result directly in an analogue measure. Economic considerations, available budget resources, resource planning and – particularly when it comes to the railway – operational aspects must be included in the planning and implementation of measures in order to ultimately guarantee having the right measure, at the right time and in the right place. This analogue world is not likely to change much in the future, whilst on the other hand the digital possibilities will continue to make rapid progress when it comes to evaluating the condition of systems. With regard to the basis of decision-making, it is essential to draw on the best available know-how.

Stefan Marschnig

30.06.2017

Technologies

1646 words

11 minutes reading time

Railway: Digital within analogue Railway: Digital within analogue Railway: Digital within analogue Railway: Digital within analogue Railway: Digital within analogue

Related articles

Mixed technology for more information

Technologies

Mixed technology for more information

Mayank Tripathi | 27.06.2017 | 1618 words | 11 minutes reading time

Mixing wisely for real added value: When Distributed Acoustic Sensing (DAS) is linked to axle counters and inductive wheel sensors, valuable information can be generated for railway applications.

Read more
How systems communicate

Technologies

How systems communicate

Stefan Lugschitz | 29.06.2017 | 1467 words | 10 minutes reading time

No function without communication: interfaces make sure that components of the railway infrastructure are networked effectively.

Read more
The question remains: Analogue or digital

Technologies

The question remains: Analogue or digital

Manfred Sommergruber | 26.06.2017 | 796 words | 6 minutes reading time

Signals from wheel sensors can be analogue or digital. Manfred Sommergruber uses the example of the Frauscher RSR110 wheel sensor to explain the respective features.

Read more
“Strictly according to protocol”

Applications

“Strictly according to protocol”

Fabian Schwarz | 28.06.2017 | 1143 words | 8 minutes reading time

Railway applications are becoming more and more complex. Melanie Kleinpötzl, Frauscher Product Management, explains in an interview how high-performance software protocols guarantee the essential problem-free communication between systems.

Read more