Products & Technology, Research & Development, Technology and IT

Data-driven maintenance: taking rail profiling to the cloud

Cloud

Lifting data from the digital grave and into the cloud has opened up possibilities for rail maintenance. Autech explains how.

Twenty years ago, Swiss rail maintenance machine manufacturer Autech began providing its customers with an innovative way to measure their tracks. Using electronic measurement data collected by maintenance and measurement machines, rail infrastructure owners and operators could see the cross-sections of their rails, enabling an understanding of the wear and tear of this critical infrastructure.

Despite having this data on hand, CTO of Autech, Peter Merz found that it was not being put to use.

“What we saw is then they piled up the data, they printed it out and put it in the archive, and basically this data was lost.”

While some aggregated data was put into enterprise resource planning (ERP) systems, the fine-grain measurements that could provide a maintenance engineer with insights were unavailable.

“The individual measurements were deleted or put in a storage system and were buried in the digital grave,” said Merz.

Having had this experience, Merz and the team at Autech began working on creating a cloud-based solution that would enable rail engineers to easily make use of the data they were collecting. The software system they developed has been named RailCloud.

“RailCloud really plots the view of the maintenance field engineers, so they can see their track, the overall condition of the track, but also the data on the individual section, even a single cross-section measurement,” said Merz.

RailCloud takes measurements collected in the field and combines them in a single, analysable database that is presented based on the geography of the rail track. The software’s base layer is a map of the system, and asset data stored in the cloud is overlaid on that map.

“It starts with the topography, the mapping, so the field engineer can go to this crossing, this intersection and so on. This is connected to the measurement systems, so the measurement systems automatically upload data, located by GPS,” said Merz.

“You can connect your measurement equipment to your network environment, so the data is automatically sorted, assigned, and allocated.”

The cloud-based software can then assign work orders and maintenance tasks based on thresholds set by the operator. In addition, having the data collected together, operators can now begin to predict rates of wear and trends, enabling predictive maintenance regimes.

“Of course, it’s a continuous thing – every year you make the measurements, every year you plan your maintenance. But with RailCloud we kept it quite light weight to make it simple and smart. You really can work on a daily basis with it, collecting measurement data, network, topology, workflows. Then you get data driven maintenance.”

DATA FROM THE SOURCE
To collect data on track condition and wear rates, Autech have recently developed RailXS, bringing together 30 years of rail measurement knowledge.

“The big advantage is it is very lightweight, it’s about 60-70kg and it can be mounted on any suitable rollingstock equipment,” said Merz. “This can be a dedicated equipment, it can be a small trolley, it can be an existing maintenance rollingstock, but it also can be a regular rollingstock.”

By mounting on regular rollingstock, measurement does not have to wait for track maintenance periods or shutdowns and can be done many times in one day.

The data is collected through laser optical sensors, which can record track parameters and the rail profile. Data is then automatically uploaded to the cloud platform RailCloud either via WiFi or a mobile internet connection. If this is not available, the data is stored and then uploaded once the vehicle returns to the depot or an area of internet connectivity. Before uploading, the measurement data is tagged with a location, either through GPS locating or RFID readers. Having these automatic systems means the data is ready to be utilised by the rail maintenance engineer, rather than having to be sorted or allocated.

“By transferring the data into the RailCloud it’s automatically allocated, you don’t have to work again. You can introduce filters to smoothen, aggregate, or transfer the data, or to do additional calculations, but the real key is to automatically map the data to your network and then there is no manual interaction needed again,” said Merz.

THE KEY TO PREDICTIVE MAINTENANCE
During the development process, the focus for RailCloud was to keep the software as lightweight as the measurement systems that supported it. This has enabled the software to be adopted by smaller operators, without the need for expensive experts and consultants to set up the system. Already, the system is in use on the tram networks of Zürich and Amsterdam where it has driven smarter maintenance practices.

“In Zürich, one of the departments wanted to do a replacement and the maintenance department said no we don’t need this replacement yet,” said Merz. “Using the RailCloud data they could prove that instead of a replacement being due every 5 years, it’s only in 12 years. RailCloud is driving fact- based decisions.”

Due to its flexibility, and the lack of a need for scheduled measurements by specialised vehicles, RailCloud can help operators take the next step to predictive maintenance.

“The big advantage is that you don’t measure every five years or every three years, you can regularly measure four times a year or even once a month,” said Merz. “You can set your intervals according to your needs, but in fact if you measure five times a year or 12 times a year, you have much better prognosis points of your wear rates.”

As wear rates are not linear, having more data points can enable a clearer picture of the wear curve to appear than what would be possible if measurements are only conducted every few years, said Merz.

“If you measure once a month you really see the trend or the curve, of your wear rate, and you see also deviation or if it changes in behaviour. That’s a big advantage, not just to know the state the track is in but what will happen.

“It’s the key to go into predictive maintenance.”

Send this to a friend