VR Group strives for punctuality through analytics
VR Group is constructing, by means of data on maintenance and traffic events, a system that improves the punctuality and reliability of trains by enabling better anticipation. This also benefits VR Group by reducing maintenance costs. VR Group offers a concrete example of the possibilities Big Data and anticipatory analytics give to organisations and of how the industrial Internet (Internet of Things) changes and develops established operations models.
Safety and Quality Manager Juha Artukka and Technology Manager Juha Ohvo of VR Group say that the company accumulates an enormous quantity of information on the maintenance of trains and traffic events daily.
- Data is accumulated, for example, on equipment technology, faults, servicing, running of trains, timetables, costs and weather. Sensors on the trains monitor their different functions, and new data is constantly obtained from them as the equipment runs in the traffic. This means tens of millions of data rows per month, says Artukka.
Data is accumulated, for example, on equipment technology, faults, servicing, running of trains, timetables, costs and weather. Sensors on the trains monitor their different functions, and new data is constantly obtained from them as the equipment runs in the traffic. This means tens of millions of data rows per month.
About four years ago, VR Group became interested in how the servicing of trains can be anticipated and optimised, and how real-time information on the condition of the equipment can be obtained by means of the anticipatory analytics tools of SAS Institute.
The project started in 2013 with the definition of the indicators associated with analytics and anticipation. Artukka, who was initially the project leader, says that it was quickly discovered that it is sensible for the company to focus on analytics primarily in the optimisation of servicing and expensive components. The opportunity to conduct root cause analyses of failures was also seen as important.
- VR Group has 1,500 trains a day running on the rails, and as we operate in a demanding logistical traffic system, each technical malfunction causes great secondary costs. Maintenance costs consist mainly of the servicing and repair of large power transmission components. These costs can be influenced by service life analyses and anticipating service breaks. With the help of analytics, it is also possible to develop work planning, repair operations and the anticipation of servicing needs, thereby improving the punctuality of trains and reducing costs, Artukka explains.
The wheel and axle set indicates the condition of the equipment
The largest expense items of maintenance particularly accrue from the replacement and turning of wheel and axle sets. The trains have a total of more than 30,000 wheel and axle sets. Their economic value can reach tens of thousands of Euros per set. The wheel and axle sets must be turned regularly because of their wear and tear, and each turning incurs an expense.
According to Artukka, the assessment of the condition of the wheel and axle sets has been challenging thus far. The assessment is influenced by a number of variables, such as the weather, the speed of the trains, the utilisation rate of the equipment, material choices and the number of turning times.
- If the dates of turning can be optimised, the wheel lasts longer. The ordering of materials must always be anticipated as well as possible, because it is difficult to obtain small material consignments from suppliers. Up to now, anticipation has been manual, Artukka says.
Now, thanks to the predictive analytics provided by the SAS Institute, a mathematical model has been created by means of which the condition of wear of the wheel and axle sets can be established, and it can be estimated more precisely when they must be turned or fully replaced.
Passenger trains run over an optical measuring station. The measuring station sends to VR 100 megabytes of data as the train passes the measuring device.
A Pendolino train has several hundred microprocessors that send data on faults and changes of status between different functions or subsystems. By means of this sensor data, it is possible to access the so-called root causes of the breakdown of equipment and to monitor the optimal functioning of duplicated devices. This prevents the concatenation of faults.
- The wheel and axle set is a good indicator of the condition of the equipment, because it affects passenger comfort and also, for example, the condition of the bogies and engine of the train. This also facilitates ordering spare parts and ensures that they are always available when needed, Ohvo notes.
Servicing intervals will be redefined
Wheel and axle sets are only one example of the possibilities opening to VR Group with analytics. Ohvo, who is currently in charge of the implementation of the project, says that the aim is to modernise the equipment maintenance system and to anticipate servicing procedures. Currently, servicing intervals are based, for example, on the assessment that a train must be serviced approximately after every 60,000 kilometres.
- It is possible for us to specify and optimise the scheduling of servicing by type of equipment. By optimising servicing intervals, we can improve the reliability of the equipment and reduce its maintenance costs. Finding the root causes of faults also increases the savings and improves the reliability of trains, Ohvo says.
- In the future, we will get information on the condition of the different components of a train and can find out, for example, whether their replacement intervals are too short or too long. If we identify a typical fault in the equipment, we can also contact our system supplier, for example, says Artukka.
The objective is to improve the quality of data
In order for analytics to produce accurate results, it is important that the data is reliable. Therefore, VR Group also strives to improve the quality of the accumulating data.
- It is important that we continue to develop the quality of our operational processes. VR Group has several source systems, which means a large quantity of data. Although some of the necessary data accrues from systems controlling transport operations and from the information systems of trains, data is also collected manually. The data is not yet of sufficiently good quality throughout, says Artukka.
A large part of the collected data is process data accrued directly from the systems of trains, such as notices of defect given by the staff. However, the staff enter the data obtained from repairs of faults and servicing procedures manually into the system. Therefore, it is important that everyone makes data entries with care.
Some of the staff have already been able to familiarise themselves with the servicing systems, and according to Ohvo and Artukka, the reception has been positive.
- At the moment, there are six or seven system end users, but the objective is that in the future 20−30 maintenance experts and production planners will use the system actively.
How can VR Group improve the punctuality and reliability of trains and reduce servicing costs
Analytics enables access to the root causes of equipment failure, specifying the servicing intervals of trains, components and subsystems and preventive or need-based servicing operations.
Read more here about how you can benefit from the industrial internet using analytics