Predictive maintenance in retrofit

Selection of data suppliers for the retrofit

The Siemens plant in Cham. © Siemens

In Cham, minimizing the effort required was the decisive factor when selecting the sensor technology. The aim was to precisely monitor critical components such as the drive motors for chain and step conveyors. Two equivalent solutions were used for the retrofit:

Sirius 3RC7 (Intelligent Link Module, ILM): Twelve circuit-breaker/contactor combinations were equipped with 3RC7 modules. These provide relevant operating data directly from the field (e.g. phase currents and unbalance and active power)

Sirius 3RC7 © Siemens

Additional sensors have also been integrated, for example to monitor the temperature in the control cabinet and the compressed air flow rate. For the linear direct drives, Sinamics frequency inverters also provide additional data such as torque, current consumption, action time, position lag and speed.

AI-supported analysis platform

Senseye Predictive Maintenance, a cloud-based solution from Siemens that is based on machine learning and offers scalability for large enterprise environments, was chosen as the AI-supported analysis platform. It analyzes machine data (vibrations, current, torque, temperature) for condition monitoring and enables users to identify the causes of faults. Senseye also uses generative AI to create contextualized advice and suggested solutions from user comments, sharing expertise across sites and continuously improving the system. Senseye thus covers several aspects of the process shown in Figure 1 for the introduction of predictive maintenance concepts.

In total, more than 70 so-called assets from the production line in Cham supply data to Senseye. An asset is the digital model of a real device or machine that is created in the system. For example, a motor, a pump, but also a combination of components can be considered an asset.

Simatic ET 200SP e-Starter. © Siemens

A look at the system structure of the solution in Cham shows: The load feeders are equipped with either the Sirius 3RC7 or the ET 200SP e-Starter and connected to the automation system via the Simatic ET 200SP. Both devices record current and power directly in the feeder and report deviations immediately. When parameterized switch-off thresholds are reached, they take over the contactor control. An industrial edge device forwards the data to Senseye. Additional sensors such as Sinamics frequency converters are connected directly to the Simatic S7 controller.

Implementation during ongoing operations

The challenge for the pilot project was to retrofit the production system with the necessary functions and components during operation. Nevertheless, integrating the additional devices into the switchgear was particularly easy. Compared to the 3RT2 contactor, the 3RC7 does not take up any additional space in the height of the control cabinet. The Simatic ET 200SP e-Starter also has a compact design with dimensions of 151 mm × 30 mm × 167 mm.

Thanks to the space-saving design and the use of the standalone adapter for a double-row setup, no redesign of the control cabinet was necessary. The structure only had to be adapted slightly. The existing circuit-breakers could continue to be used.

It was precisely at this step that the team in Cham was able to record its first success. After adapting the Step7 project planning in the TIA portal, a linear direct drive installed in the system was put back into operation. During the recordings in the Simatic S7 controller, it became apparent that the Sinamics frequency inverter used for track recording was recording unusually high differences in the target and actual position of the axes at a certain point in time. A comparison with the raw data in Senseye confirmed this anomaly. The team then examined the axes and discovered that the automatic lubrication was defective on one side. The axle was running dry on one side and the bearings had already been damaged.

This example clearly shows the potential of predictive maintenance. Linear direct drives are expensive, customer-specific components whose failure can lead to considerable production downtime and financial losses. The drive control compensated for the anomaly and did not yet issue a warning, as the control loop was still operating within the tolerances. If these tolerances are exceeded, the wear is at best in the critical range. In the worst case, the deficit is only recognized when the axle fails. According to the team's estimates, the early intervention in Cham prevented a production stoppage of seven days.

Validation and scaling of the potential

The pilot project in Cham demonstrates the feasibility of predictive maintenance in retrofitting. The early detection of bearing damage demonstrates the potential for avoiding downtime and increasing efficiency.

Looking to the future, the pilot project should show whether it makes economic sense to equip all systems across the board or whether the focus should be on particularly critical assets. In addition to the aforementioned linear direct drives, these critical systems can also include plant infrastructure components such as refrigeration systems or compressed air units, the failure of which has far-reaching effects on the entire operating process.

Carsten Moye, Siemens Smart Infrastructure - Electrical Products

Hanover Fair, Hall 27, Stand A48