Skip to content

Cooperation streamlines train traffic

Published: 29 September 2020

As there are many different actors involved in train traffic, it has been difficult to remedy certain problems.
– But now, thanks to the projects that Luleå University of Technology is implementing, it has become possible to collaborate. It is thanks to this that Norrtåg no longer has any major problems with the wheels, says Håkan Jarl, fleet controller.

The AI Factory for Railways, AIFR, project at Luleå University of Technology gathers stakeholders from the train industry to create an AI engine where information can be gathered and shared for planning and troubleshooting.
According to the Executive director, and also professor Ramin Karim, the goal is for the trains to run on time.

Håkan Jarl, who works with train traffic in the four northernmost regions in the country, believes that he and the others from the industry can contribute to the research from a business perspective.
– We were also part of the precusor to AIFR, which Luleå University of Technology had, Epilot. Then we wanted to develop our working methods to access the wheel problems on Norrtåg, says Håkan Jarl.
In addition to Norrtåg, it is also about Tåg i Bergslagen, X-trafic and Region Värmlands traintrafic. A major problem has been the wheels of the vehicles purchased for the traffic of Norrtåg.
– The wheels wear out too quickly, especially in winter when the wheel damage develops much faster. There were many articles in the newspapers in the winters of 2017-2018. We are several parties, vehicle manufacturers, vehicle owners, operators, maintenance suppliers and we as tenants of the vehicles.

The same picture of reality

Håkan Jarl says that it was important that everyone had the same picture of reality, based on facts. He believes that it has not been possible to gather all actors in any other way than through this type of research project that is carried out at Luleå University of Technology.
– There was a great need to see what we could do with the wheels, what maintenance strategy would we have, so that the logistics in the workshop also work, what maintenance intervals did we need to have? The important thing was that the trains could run on time.

Håkan Jarl says that in the previous project it was possible to test a digital system that produced specific data. It was also possible to arrange alarms based on the parameters set by the participants. In the project, a system was developed with preventive turning of the wheels at fixed intervals.
– Now we monitor via the system that emerged in the project. Today it is a commercial system. For a year and a half now, we have had no delays due to wheel problems.
Håkan Jarl is satisfied. The client wants to know how big the profit is before a system of this kind is purchased and it can be difficult to convince all parties involved, now there was the answer.
– Now we have extended the life of the wheels by more than 50 percent on the current vehicle type thanks to the prevention program, which means huge cost savings.

Examines automatic image analysis

In the current, follow-up project AI for Railways, he believes that there are opportunities for further development.
– Like finding the exact time when it is ultimate to turn the wheels. We do not have that answer. That is what I want to see if it can be developed.
As part of AIFR's project, an investigation is also underway into whether it is possible to use automatic image analysis to detect faults on the train wheels. The images could be taken in the workshop and then processed in the AI engine.

– It can be a complement to the manual inspection that is done.
He is convinced that when the project is completed, it will have resulted in new useful results.
– There is an incredible competence in this project. If it does not happen here, it will happen nowhere. There are big gains to be made. You do not have to experience the acute events, but you can see things in advance. It's about getting it planned instead of it being a surprise, says Håkan Jarl.

Tags