fotohansel - Fotolia
Italy’s primary train operator, Trenitalia, is combining the internet of things (IoT), analytics and in-memory computing in a project to make maintenance of its trains more efficient and effective.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Trains have complex mechanical and electrical systems with hundreds of thousands of moving parts. Trenitalia wants the project to enable it to predict failures before they happen, rather than just react to them.
To deliver a reliable service, trains need regular maintenance that can be triggered when a certain mileage is reached, after a particular time period, or by monitoring the condition of parts. Condition monitoring involves checking the operation of the equipment and only changing something if it shows signs of deterioration.
The IoT and the availability of complementary technologies have even made it possible to predict the failure of some items of equipment. This is what Trenitalia has started to do after deploying a dynamic maintenance management system (DMMS) that combines three technologies: the IoT, analytics and in-memory computing.
“SAP Hana is the platform that was selected to build the DMMS,” says Danilo Gismondi, CIO at Trenitalia. “It is now possible to affordably collect huge amounts of data from hundreds of sensors in a single train, analyse that data in real time and detect problems before they actually happen. This approach will be gradually introduced and will lower maintenance costs by up to 8%.
“The project was built according to the requirements expressed by the operation management and represents a paradigm shift. There is no more maintenance by mileage or by time. The goal is to keep costs to a minimum by applying predictive maintenance and data analytics to all essential railcar parts.”
Advances in sensor and communication technologies have enabled continuous data collection from various systems in trains. This means that mechanical and electrical conditions, operational efficiency and other performance indicators can be monitored 24 hours a day.
These capabilities allow maintenance to be planned with the maximum interval between repairs, while minimising the number and cost of unscheduled outages caused by system failure.
“Predictive maintenance is much more than that, however,” says Gismondi. “We can minimise unscheduled breakdowns of all mechanical equipment and ensure that repaired equipment is in an acceptable condition. Most mechanical problems can be minimised if they are detected and repaired early.”
It is also about identifying various business scenarios, he adds. For example, an effective predictive maintenance system should also help the operator to plan an inventory for replacement parts and suggest which systems need a design upgrade because of continued poor performance.
Read more about predictive maintenance
- National Grid is using cloud-based analytics powered by IBM SoftLayer to provide preventative maintenance.
- SAP’s connected manufacturing framework sets its sights on delivering true visibility, unifying business processes through data, delivering predictive maintenance, and more.
- Predictive analytics is being used in the oil and gas industry to better manage asset maintenance on capital equipment.
Gismondi says it was vital to find the right platform to support the project. The company evaluated several technologies and eventually chose SAP Hana.
“One of the things we liked about Hana is that it tightly integrates different layers – the in-memory computing, enabling real or near real-time computing, the database and the analytics,” he says. “In simple words, we apply big data logic to the maintenance process.
“I think this is a new IT frontier, a new way of handling huge volumes of data in a time-responsive fashion that cannot be achieved by traditional technologies.”
Gismondi says Trenitalia was also attracted by the ability to tailor the solution to meet its specific requirements, “and this is of great competitive value”.
Efficient data analysis
When it is completed in two years’ time, the predictive maintenance system will generate 700TB of data per year. An algorithm has been developed to make data analysis more efficient.
The system will be active on most of the company’s long-distance and high-speed trains, and will be progressively installed on the new trains that will replace the old regional ones.
Data will be collected via Wi-Fi, mostly at stations and in depots. “We are also working on trying to transmit data about the most critical parts every 10 minutes via the network provided by our telco provider,” says Gismondi.
“The system will help us to create a unique way of collecting data to improve our efficiency as well as reducing operating costs.”